2023-04-27 12:19:53,924 INFO [train.py:976] (6/8) Training started 2023-04-27 12:19:53,924 INFO [train.py:986] (6/8) Device: cuda:6 2023-04-27 12:19:53,926 INFO [train.py:995] (6/8) {'frame_shift_ms': 10.0, 'allowed_excess_duration_ratio': 0.1, 'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'warm_step': 2000, 'env_info': {'k2-version': '1.23.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'a23383c5a381713b51e9014f3f05d096f8aceec3', 'k2-git-date': 'Wed Apr 26 15:33:33 2023', 'lhotse-version': '1.14.0.dev+git.b61b917.dirty', 'torch-version': '1.13.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.1', 'icefall-git-branch': 'master', 'icefall-git-sha1': '45c13e9-dirty', 'icefall-git-date': 'Mon Apr 24 15:00:02 2023', 'icefall-path': '/k2-dev/yangyifan/icefall-master', 'k2-path': '/k2-dev/yangyifan/anaconda3/envs/icefall/lib/python3.10/site-packages/k2-1.23.4.dev20230427+cuda11.6.torch1.13.1-py3.10-linux-x86_64.egg/k2/__init__.py', 'lhotse-path': '/k2-dev/yangyifan/anaconda3/envs/icefall/lib/python3.10/site-packages/lhotse-1.14.0.dev0+git.b61b917.dirty-py3.10.egg/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-3-0423201227-84b4557756-8lx4n', 'IP address': '10.177.6.147'}, 'world_size': 8, 'master_port': 12354, 'tensorboard': True, 'num_epochs': 30, 'start_epoch': 1, 'start_batch': 0, 'exp_dir': PosixPath('pruned_transducer_stateless7/exp_multidataset'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'base_lr': 0.05, 'lr_batches': 5000, 'lr_epochs': 3.5, 'context_size': 2, 'prune_range': 5, 'lm_scale': 0.25, 'am_scale': 0.0, 'simple_loss_scale': 0.5, 'seed': 42, 'print_diagnostics': False, 'inf_check': False, 'save_every_n': 2000, 'keep_last_k': 30, 'average_period': 200, 'use_fp16': True, 'use_multidataset': True, 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'full_libri': True, 'manifest_dir': PosixPath('data/fbank'), 'cv_manifest_dir': PosixPath('data/en/fbank'), 'max_duration': 700, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'blank_id': 0, 'vocab_size': 500} 2023-04-27 12:19:53,927 INFO [train.py:997] (6/8) About to create model 2023-04-27 12:19:54,602 INFO [zipformer.py:178] (6/8) At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8. 2023-04-27 12:19:54,618 INFO [train.py:1001] (6/8) Number of model parameters: 70369391 2023-04-27 12:19:57,215 INFO [train.py:1016] (6/8) Using DDP 2023-04-27 12:19:58,267 INFO [multidataset.py:46] (6/8) About to get multidataset train cuts 2023-04-27 12:19:58,268 INFO [multidataset.py:49] (6/8) Loading LibriSpeech in lazy mode 2023-04-27 12:19:58,288 INFO [multidataset.py:65] (6/8) Loading GigaSpeech 1998 splits in lazy mode 2023-04-27 12:20:00,752 INFO [multidataset.py:72] (6/8) Loading CommonVoice in lazy mode 2023-04-27 12:20:00,755 INFO [asr_datamodule.py:230] (6/8) Enable MUSAN 2023-04-27 12:20:00,755 INFO [asr_datamodule.py:231] (6/8) About to get Musan cuts 2023-04-27 12:20:03,037 INFO [asr_datamodule.py:255] (6/8) Enable SpecAugment 2023-04-27 12:20:03,037 INFO [asr_datamodule.py:256] (6/8) Time warp factor: 80 2023-04-27 12:20:03,037 INFO [asr_datamodule.py:266] (6/8) Num frame mask: 10 2023-04-27 12:20:03,038 INFO [asr_datamodule.py:279] (6/8) About to create train dataset 2023-04-27 12:20:03,038 INFO [asr_datamodule.py:306] (6/8) Using DynamicBucketingSampler. 2023-04-27 12:20:07,561 INFO [asr_datamodule.py:321] (6/8) About to create train dataloader 2023-04-27 12:20:07,562 INFO [asr_datamodule.py:435] (6/8) About to get dev-clean cuts 2023-04-27 12:20:07,563 INFO [asr_datamodule.py:442] (6/8) About to get dev-other cuts 2023-04-27 12:20:07,564 INFO [asr_datamodule.py:352] (6/8) About to create dev dataset 2023-04-27 12:20:07,798 INFO [asr_datamodule.py:369] (6/8) About to create dev dataloader 2023-04-27 12:20:25,619 INFO [train.py:904] (6/8) Epoch 1, batch 0, loss[loss=7.431, simple_loss=6.742, pruned_loss=6.869, over 16783.00 frames. ], tot_loss[loss=7.431, simple_loss=6.742, pruned_loss=6.869, over 16783.00 frames. ], batch size: 102, lr: 2.50e-02, grad_scale: 2.0 2023-04-27 12:20:25,620 INFO [train.py:929] (6/8) Computing validation loss 2023-04-27 12:20:32,881 INFO [train.py:938] (6/8) Epoch 1, validation: loss=6.911, simple_loss=6.238, pruned_loss=6.721, over 944034.00 frames. 2023-04-27 12:20:32,881 INFO [train.py:939] (6/8) Maximum memory allocated so far is 11328MB 2023-04-27 12:20:36,291 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:20:52,240 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:20:55,589 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=100.78 vs. limit=5.0 2023-04-27 12:21:03,478 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=8.82 vs. limit=2.0 2023-04-27 12:21:17,132 INFO [train.py:904] (6/8) Epoch 1, batch 50, loss[loss=1.287, simple_loss=1.132, pruned_loss=1.375, over 17158.00 frames. ], tot_loss[loss=2.165, simple_loss=1.96, pruned_loss=1.97, over 757620.93 frames. ], batch size: 46, lr: 2.75e-02, grad_scale: 2.0 2023-04-27 12:21:21,675 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=16.83 vs. limit=2.0 2023-04-27 12:21:39,522 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=7.51 vs. limit=2.0 2023-04-27 12:21:46,495 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:22:02,791 WARNING [train.py:894] (6/8) Grad scale is small: 0.001953125 2023-04-27 12:22:02,791 INFO [train.py:904] (6/8) Epoch 1, batch 100, loss[loss=1.089, simple_loss=0.9232, pruned_loss=1.3, over 17191.00 frames. ], tot_loss[loss=1.638, simple_loss=1.457, pruned_loss=1.622, over 1329422.44 frames. ], batch size: 44, lr: 3.00e-02, grad_scale: 0.00390625 2023-04-27 12:22:13,668 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 5.700e+01 2.326e+02 5.095e+02 1.135e+03 3.099e+06, threshold=1.019e+03, percent-clipped=0.0 2023-04-27 12:22:20,962 WARNING [optim.py:388] (6/8) Scaling gradients by 0.0112030990421772, model_norm_threshold=1019.0284423828125 2023-04-27 12:22:21,067 INFO [optim.py:450] (6/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.24, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=1.974e+09, grad_sumsq = 5.165e+10, orig_rms_sq=3.822e-02 2023-04-27 12:22:24,837 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=57.08 vs. limit=5.0 2023-04-27 12:22:34,165 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=5.75 vs. limit=2.0 2023-04-27 12:22:43,841 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=144.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:22:46,127 WARNING [optim.py:388] (6/8) Scaling gradients by 0.0022801109589636326, model_norm_threshold=1019.0284423828125 2023-04-27 12:22:46,230 INFO [optim.py:450] (6/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.68, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=1.358e+11, grad_sumsq = 3.286e+12, orig_rms_sq=4.131e-02 2023-04-27 12:22:49,488 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=8.90 vs. limit=2.0 2023-04-27 12:22:49,648 WARNING [optim.py:388] (6/8) Scaling gradients by 0.04246773198246956, model_norm_threshold=1019.0284423828125 2023-04-27 12:22:49,753 INFO [optim.py:450] (6/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.92, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=5.309e+08, grad_sumsq = 1.285e+10, orig_rms_sq=4.131e-02 2023-04-27 12:22:51,348 WARNING [optim.py:388] (6/8) Scaling gradients by 0.000716241542249918, model_norm_threshold=1019.0284423828125 2023-04-27 12:22:52,145 INFO [optim.py:450] (6/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.93, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=1.876e+12, grad_sumsq = 4.542e+13, orig_rms_sq=4.131e-02 2023-04-27 12:22:52,994 INFO [train.py:904] (6/8) Epoch 1, batch 150, loss[loss=1.081, simple_loss=0.9087, pruned_loss=1.23, over 17054.00 frames. ], tot_loss[loss=1.406, simple_loss=1.232, pruned_loss=1.459, over 1772145.58 frames. ], batch size: 55, lr: 3.25e-02, grad_scale: 0.00390625 2023-04-27 12:22:53,728 WARNING [optim.py:388] (6/8) Scaling gradients by 0.049951765686273575, model_norm_threshold=1019.0284423828125 2023-04-27 12:22:53,836 INFO [optim.py:450] (6/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.92, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=3.845e+08, grad_sumsq = 8.968e+09, orig_rms_sq=4.287e-02 2023-04-27 12:22:58,541 WARNING [optim.py:388] (6/8) Scaling gradients by 0.00609818659722805, model_norm_threshold=1019.0284423828125 2023-04-27 12:22:58,644 INFO [optim.py:450] (6/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.49, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=1.379e+10, grad_sumsq = 3.140e+11, orig_rms_sq=4.392e-02 2023-04-27 12:23:16,872 WARNING [optim.py:388] (6/8) Scaling gradients by 0.059935860335826874, model_norm_threshold=1019.0284423828125 2023-04-27 12:23:16,973 INFO [optim.py:450] (6/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.63, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=1.808e+08, grad_sumsq = 3.916e+09, orig_rms_sq=4.617e-02 2023-04-27 12:23:27,199 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=7.22 vs. limit=2.0 2023-04-27 12:23:28,341 WARNING [optim.py:388] (6/8) Scaling gradients by 0.060559310019016266, model_norm_threshold=1019.0284423828125 2023-04-27 12:23:28,444 INFO [optim.py:450] (6/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.90, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=2.556e+08, grad_sumsq = 6.221e+09, orig_rms_sq=4.108e-02 2023-04-27 12:23:42,816 WARNING [train.py:894] (6/8) Grad scale is small: 0.00390625 2023-04-27 12:23:42,816 INFO [train.py:904] (6/8) Epoch 1, batch 200, loss[loss=0.9063, simple_loss=0.7615, pruned_loss=0.9612, over 17000.00 frames. ], tot_loss[loss=1.263, simple_loss=1.095, pruned_loss=1.323, over 2113519.56 frames. ], batch size: 41, lr: 3.50e-02, grad_scale: 0.0078125 2023-04-27 12:23:51,003 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.164e+02 2.076e+02 2.707e+02 4.762e+02 1.423e+06, threshold=5.415e+02, percent-clipped=11.0 2023-04-27 12:23:51,003 WARNING [optim.py:388] (6/8) Scaling gradients by 0.002041660714894533, model_norm_threshold=541.4743041992188 2023-04-27 12:23:51,106 INFO [optim.py:450] (6/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.86, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=6.060e+10, grad_sumsq = 1.544e+12, orig_rms_sq=3.924e-02 2023-04-27 12:24:00,575 WARNING [optim.py:388] (6/8) Scaling gradients by 0.02974529005587101, model_norm_threshold=541.4743041992188 2023-04-27 12:24:00,710 INFO [optim.py:450] (6/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.85, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=2.829e+08, grad_sumsq = 7.259e+09, orig_rms_sq=3.897e-02 2023-04-27 12:24:01,478 WARNING [optim.py:388] (6/8) Scaling gradients by 0.01955481991171837, model_norm_threshold=541.4743041992188 2023-04-27 12:24:01,614 INFO [optim.py:450] (6/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.84, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=6.426e+08, grad_sumsq = 1.649e+10, orig_rms_sq=3.897e-02 2023-04-27 12:24:05,544 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=11.70 vs. limit=2.0 2023-04-27 12:24:30,812 INFO [train.py:904] (6/8) Epoch 1, batch 250, loss[loss=0.8907, simple_loss=0.7547, pruned_loss=0.8613, over 16209.00 frames. ], tot_loss[loss=1.169, simple_loss=1.007, pruned_loss=1.215, over 2359717.44 frames. ], batch size: 165, lr: 3.75e-02, grad_scale: 0.0078125 2023-04-27 12:24:33,593 WARNING [optim.py:388] (6/8) Scaling gradients by 0.057925041764974594, model_norm_threshold=541.4743041992188 2023-04-27 12:24:33,696 INFO [optim.py:450] (6/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.59, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=5.145e+07, grad_sumsq = 1.327e+09, orig_rms_sq=3.876e-02 2023-04-27 12:24:47,393 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=61.26 vs. limit=5.0 2023-04-27 12:24:48,307 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=11.17 vs. limit=2.0 2023-04-27 12:25:01,422 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=12.38 vs. limit=2.0 2023-04-27 12:25:07,674 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=55.04 vs. limit=5.0 2023-04-27 12:25:16,110 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:25:20,574 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 12:25:21,111 WARNING [train.py:894] (6/8) Grad scale is small: 0.0078125 2023-04-27 12:25:21,111 INFO [train.py:904] (6/8) Epoch 1, batch 300, loss[loss=0.9446, simple_loss=0.786, pruned_loss=0.9246, over 17054.00 frames. ], tot_loss[loss=1.098, simple_loss=0.9399, pruned_loss=1.126, over 2565869.44 frames. ], batch size: 55, lr: 4.00e-02, grad_scale: 0.015625 2023-04-27 12:25:30,078 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 9.163e+01 1.282e+02 1.731e+02 2.635e+02 2.769e+04, threshold=3.463e+02, percent-clipped=7.0 2023-04-27 12:25:37,331 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=44.43 vs. limit=5.0 2023-04-27 12:25:51,594 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=48.40 vs. limit=5.0 2023-04-27 12:26:03,736 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=13.93 vs. limit=2.0 2023-04-27 12:26:12,599 INFO [train.py:904] (6/8) Epoch 1, batch 350, loss[loss=1.011, simple_loss=0.8352, pruned_loss=0.9667, over 17080.00 frames. ], tot_loss[loss=1.049, simple_loss=0.8919, pruned_loss=1.06, over 2728497.77 frames. ], batch size: 53, lr: 4.25e-02, grad_scale: 0.015625 2023-04-27 12:26:18,617 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=357.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 12:26:24,282 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=37.37 vs. limit=5.0 2023-04-27 12:26:28,462 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.7875, 5.7853, 5.7886, 5.7797, 5.7837, 5.7883, 5.7864, 5.7822], device='cuda:6'), covar=tensor([0.0032, 0.0070, 0.0034, 0.0069, 0.0028, 0.0017, 0.0053, 0.0057], device='cuda:6'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0008, 0.0008, 0.0009, 0.0009], device='cuda:6'), out_proj_covar=tensor([8.7899e-06, 8.9363e-06, 8.7742e-06, 8.5590e-06, 8.9231e-06, 8.5268e-06, 8.5101e-06, 8.7099e-06], device='cuda:6') 2023-04-27 12:26:52,608 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=387.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:27:03,476 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=62.84 vs. limit=5.0 2023-04-27 12:27:06,007 INFO [train.py:904] (6/8) Epoch 1, batch 400, loss[loss=0.874, simple_loss=0.7214, pruned_loss=0.8021, over 16882.00 frames. ], tot_loss[loss=1.013, simple_loss=0.8548, pruned_loss=1.006, over 2863870.31 frames. ], batch size: 116, lr: 4.50e-02, grad_scale: 0.03125 2023-04-27 12:27:17,878 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 7.937e+01 1.110e+02 1.413e+02 1.837e+02 3.136e+02, threshold=2.826e+02, percent-clipped=0.0 2023-04-27 12:27:44,702 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=10.25 vs. limit=2.0 2023-04-27 12:27:46,076 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=439.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:27:56,299 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=448.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 12:27:59,554 INFO [train.py:904] (6/8) Epoch 1, batch 450, loss[loss=0.9918, simple_loss=0.8061, pruned_loss=0.913, over 16689.00 frames. ], tot_loss[loss=0.9856, simple_loss=0.8258, pruned_loss=0.9635, over 2963734.54 frames. ], batch size: 57, lr: 4.75e-02, grad_scale: 0.03125 2023-04-27 12:28:51,188 INFO [train.py:904] (6/8) Epoch 1, batch 500, loss[loss=0.879, simple_loss=0.7192, pruned_loss=0.7677, over 15485.00 frames. ], tot_loss[loss=0.9699, simple_loss=0.8069, pruned_loss=0.9325, over 3034712.30 frames. ], batch size: 190, lr: 4.99e-02, grad_scale: 0.0625 2023-04-27 12:29:01,368 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 7.996e+01 1.277e+02 1.626e+02 2.121e+02 5.144e+02, threshold=3.251e+02, percent-clipped=17.0 2023-04-27 12:29:25,729 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=7.30 vs. limit=2.0 2023-04-27 12:29:44,925 INFO [train.py:904] (6/8) Epoch 1, batch 550, loss[loss=0.9737, simple_loss=0.7801, pruned_loss=0.8654, over 16676.00 frames. ], tot_loss[loss=0.9562, simple_loss=0.7899, pruned_loss=0.9035, over 3103501.12 frames. ], batch size: 57, lr: 4.98e-02, grad_scale: 0.0625 2023-04-27 12:29:58,107 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=562.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:30:03,851 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=568.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:30:15,779 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=10.10 vs. limit=5.0 2023-04-27 12:30:27,535 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=590.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:30:38,230 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=600.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 12:30:38,745 INFO [train.py:904] (6/8) Epoch 1, batch 600, loss[loss=0.8709, simple_loss=0.699, pruned_loss=0.7476, over 17047.00 frames. ], tot_loss[loss=0.9423, simple_loss=0.7739, pruned_loss=0.8739, over 3154579.09 frames. ], batch size: 41, lr: 4.98e-02, grad_scale: 0.125 2023-04-27 12:30:48,385 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 8.862e+01 1.447e+02 1.955e+02 2.870e+02 6.309e+02, threshold=3.911e+02, percent-clipped=20.0 2023-04-27 12:30:56,619 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=3.98 vs. limit=2.0 2023-04-27 12:31:02,541 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=623.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:31:07,507 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=629.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 12:31:28,001 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=648.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:31:30,995 INFO [train.py:904] (6/8) Epoch 1, batch 650, loss[loss=0.9534, simple_loss=0.7682, pruned_loss=0.7895, over 16723.00 frames. ], tot_loss[loss=0.9324, simple_loss=0.7623, pruned_loss=0.8467, over 3194656.23 frames. ], batch size: 62, lr: 4.98e-02, grad_scale: 0.125 2023-04-27 12:31:31,365 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=651.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-27 12:31:32,142 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=652.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 12:31:44,191 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=4.71 vs. limit=2.0 2023-04-27 12:32:22,394 INFO [train.py:904] (6/8) Epoch 1, batch 700, loss[loss=0.8087, simple_loss=0.6585, pruned_loss=0.6392, over 16834.00 frames. ], tot_loss[loss=0.92, simple_loss=0.7509, pruned_loss=0.8149, over 3223433.15 frames. ], batch size: 42, lr: 4.98e-02, grad_scale: 0.25 2023-04-27 12:32:31,774 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.097e+02 2.100e+02 3.098e+02 3.988e+02 9.500e+02, threshold=6.196e+02, percent-clipped=26.0 2023-04-27 12:32:34,805 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.11 vs. limit=5.0 2023-04-27 12:32:50,525 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2023-04-27 12:33:00,844 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=739.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 12:33:05,984 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=743.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:33:13,736 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2023-04-27 12:33:14,006 INFO [train.py:904] (6/8) Epoch 1, batch 750, loss[loss=0.7887, simple_loss=0.6517, pruned_loss=0.5921, over 16534.00 frames. ], tot_loss[loss=0.899, simple_loss=0.7349, pruned_loss=0.7739, over 3246565.03 frames. ], batch size: 75, lr: 4.97e-02, grad_scale: 0.25 2023-04-27 12:33:51,950 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=787.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:34:03,301 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.48 vs. limit=5.0 2023-04-27 12:34:06,256 INFO [train.py:904] (6/8) Epoch 1, batch 800, loss[loss=0.8035, simple_loss=0.68, pruned_loss=0.5641, over 17111.00 frames. ], tot_loss[loss=0.8731, simple_loss=0.7171, pruned_loss=0.7276, over 3261516.43 frames. ], batch size: 48, lr: 4.97e-02, grad_scale: 0.5 2023-04-27 12:34:15,021 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2023-04-27 12:34:17,530 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.716e+02 2.855e+02 4.131e+02 5.648e+02 8.889e+02, threshold=8.261e+02, percent-clipped=19.0 2023-04-27 12:34:54,114 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=847.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:34:58,583 INFO [train.py:904] (6/8) Epoch 1, batch 850, loss[loss=0.6814, simple_loss=0.5812, pruned_loss=0.4635, over 16795.00 frames. ], tot_loss[loss=0.8405, simple_loss=0.6952, pruned_loss=0.6764, over 3284724.27 frames. ], batch size: 83, lr: 4.96e-02, grad_scale: 0.5 2023-04-27 12:35:50,495 INFO [train.py:904] (6/8) Epoch 1, batch 900, loss[loss=0.7146, simple_loss=0.5957, pruned_loss=0.5007, over 12313.00 frames. ], tot_loss[loss=0.8052, simple_loss=0.672, pruned_loss=0.625, over 3296698.22 frames. ], batch size: 246, lr: 4.96e-02, grad_scale: 1.0 2023-04-27 12:35:57,655 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=908.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:35:58,637 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.40 vs. limit=5.0 2023-04-27 12:36:00,422 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.021e+02 3.478e+02 3.989e+02 5.953e+02 1.563e+03, threshold=7.979e+02, percent-clipped=10.0 2023-04-27 12:36:08,844 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=918.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 12:36:14,440 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=924.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 12:36:19,921 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-27 12:36:37,786 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=946.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 12:36:42,340 INFO [train.py:904] (6/8) Epoch 1, batch 950, loss[loss=0.6454, simple_loss=0.5749, pruned_loss=0.3925, over 17108.00 frames. ], tot_loss[loss=0.7751, simple_loss=0.6526, pruned_loss=0.5808, over 3301879.83 frames. ], batch size: 47, lr: 4.96e-02, grad_scale: 1.0 2023-04-27 12:36:44,104 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=952.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 12:37:25,327 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.98 vs. limit=2.0 2023-04-27 12:37:35,473 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1000.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:37:35,994 INFO [train.py:904] (6/8) Epoch 1, batch 1000, loss[loss=0.5985, simple_loss=0.5282, pruned_loss=0.3678, over 16292.00 frames. ], tot_loss[loss=0.7401, simple_loss=0.6288, pruned_loss=0.5363, over 3307489.33 frames. ], batch size: 36, lr: 4.95e-02, grad_scale: 1.0 2023-04-27 12:37:46,289 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.400e+02 3.913e+02 4.932e+02 6.154e+02 1.349e+03, threshold=9.864e+02, percent-clipped=6.0 2023-04-27 12:38:21,637 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1043.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 12:38:29,480 INFO [train.py:904] (6/8) Epoch 1, batch 1050, loss[loss=0.5744, simple_loss=0.5098, pruned_loss=0.3466, over 16688.00 frames. ], tot_loss[loss=0.7085, simple_loss=0.6083, pruned_loss=0.4958, over 3312009.43 frames. ], batch size: 134, lr: 4.95e-02, grad_scale: 1.0 2023-04-27 12:38:40,741 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9412, 4.5989, 4.7288, 4.7511, 4.4393, 4.8402, 4.7712, 4.8525], device='cuda:6'), covar=tensor([0.2697, 0.2043, 0.2351, 0.2580, 0.2359, 0.2013, 0.1793, 0.1966], device='cuda:6'), in_proj_covar=tensor([0.0052, 0.0048, 0.0058, 0.0052, 0.0053, 0.0056, 0.0045, 0.0055], device='cuda:6'), out_proj_covar=tensor([4.7032e-05, 4.1976e-05, 5.1671e-05, 4.7956e-05, 4.8165e-05, 4.7877e-05, 4.2047e-05, 5.2502e-05], device='cuda:6') 2023-04-27 12:39:12,770 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1091.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:39:21,796 INFO [train.py:904] (6/8) Epoch 1, batch 1100, loss[loss=0.6253, simple_loss=0.5679, pruned_loss=0.3589, over 17041.00 frames. ], tot_loss[loss=0.6804, simple_loss=0.5898, pruned_loss=0.4611, over 3306998.19 frames. ], batch size: 53, lr: 4.94e-02, grad_scale: 1.0 2023-04-27 12:39:33,202 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.538e+02 4.173e+02 5.189e+02 6.744e+02 1.137e+03, threshold=1.038e+03, percent-clipped=1.0 2023-04-27 12:40:17,536 INFO [train.py:904] (6/8) Epoch 1, batch 1150, loss[loss=0.5104, simple_loss=0.4805, pruned_loss=0.2721, over 17165.00 frames. ], tot_loss[loss=0.6512, simple_loss=0.5709, pruned_loss=0.4268, over 3317852.63 frames. ], batch size: 46, lr: 4.94e-02, grad_scale: 1.0 2023-04-27 12:40:19,588 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1153.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:41:10,398 INFO [train.py:904] (6/8) Epoch 1, batch 1200, loss[loss=0.568, simple_loss=0.5023, pruned_loss=0.338, over 15653.00 frames. ], tot_loss[loss=0.6266, simple_loss=0.5542, pruned_loss=0.3993, over 3313285.68 frames. ], batch size: 190, lr: 4.93e-02, grad_scale: 2.0 2023-04-27 12:41:12,650 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1203.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 12:41:21,557 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.465e+02 4.824e+02 5.887e+02 7.169e+02 1.569e+03, threshold=1.177e+03, percent-clipped=1.0 2023-04-27 12:41:25,802 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1214.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:41:29,970 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1218.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 12:41:34,803 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1224.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 12:41:43,987 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.09 vs. limit=5.0 2023-04-27 12:41:57,793 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1246.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:42:03,009 INFO [train.py:904] (6/8) Epoch 1, batch 1250, loss[loss=0.5278, simple_loss=0.4842, pruned_loss=0.2942, over 16803.00 frames. ], tot_loss[loss=0.6085, simple_loss=0.5423, pruned_loss=0.3781, over 3309054.75 frames. ], batch size: 102, lr: 4.92e-02, grad_scale: 2.0 2023-04-27 12:42:17,894 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1266.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:42:25,411 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1272.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 12:42:49,932 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1294.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 12:42:56,462 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2023-04-27 12:42:56,579 INFO [train.py:904] (6/8) Epoch 1, batch 1300, loss[loss=0.5094, simple_loss=0.4773, pruned_loss=0.2733, over 16828.00 frames. ], tot_loss[loss=0.5922, simple_loss=0.5323, pruned_loss=0.3588, over 3310245.63 frames. ], batch size: 83, lr: 4.92e-02, grad_scale: 2.0 2023-04-27 12:43:07,776 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.547e+02 4.841e+02 5.972e+02 7.377e+02 1.990e+03, threshold=1.194e+03, percent-clipped=4.0 2023-04-27 12:43:51,780 INFO [train.py:904] (6/8) Epoch 1, batch 1350, loss[loss=0.5009, simple_loss=0.4801, pruned_loss=0.2583, over 17139.00 frames. ], tot_loss[loss=0.5751, simple_loss=0.5214, pruned_loss=0.3404, over 3296499.36 frames. ], batch size: 47, lr: 4.91e-02, grad_scale: 2.0 2023-04-27 12:44:49,663 INFO [train.py:904] (6/8) Epoch 1, batch 1400, loss[loss=0.5157, simple_loss=0.4976, pruned_loss=0.2633, over 16861.00 frames. ], tot_loss[loss=0.5606, simple_loss=0.5122, pruned_loss=0.325, over 3305183.52 frames. ], batch size: 62, lr: 4.91e-02, grad_scale: 2.0 2023-04-27 12:45:00,101 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.619e+02 5.087e+02 6.130e+02 7.329e+02 1.559e+03, threshold=1.226e+03, percent-clipped=4.0 2023-04-27 12:45:27,771 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1434.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:45:46,438 INFO [train.py:904] (6/8) Epoch 1, batch 1450, loss[loss=0.5178, simple_loss=0.4783, pruned_loss=0.2829, over 16892.00 frames. ], tot_loss[loss=0.545, simple_loss=0.5026, pruned_loss=0.3093, over 3309239.80 frames. ], batch size: 116, lr: 4.90e-02, grad_scale: 2.0 2023-04-27 12:46:34,582 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1495.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 12:46:41,351 INFO [train.py:904] (6/8) Epoch 1, batch 1500, loss[loss=0.5036, simple_loss=0.4848, pruned_loss=0.2589, over 16609.00 frames. ], tot_loss[loss=0.5334, simple_loss=0.4956, pruned_loss=0.2974, over 3313819.41 frames. ], batch size: 62, lr: 4.89e-02, grad_scale: 2.0 2023-04-27 12:46:44,313 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1503.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:46:51,086 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1509.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 12:46:52,703 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.422e+02 5.181e+02 6.478e+02 8.944e+02 1.260e+03, threshold=1.296e+03, percent-clipped=1.0 2023-04-27 12:47:38,770 INFO [train.py:904] (6/8) Epoch 1, batch 1550, loss[loss=0.4459, simple_loss=0.4444, pruned_loss=0.2178, over 17210.00 frames. ], tot_loss[loss=0.525, simple_loss=0.4908, pruned_loss=0.2886, over 3311672.30 frames. ], batch size: 44, lr: 4.89e-02, grad_scale: 2.0 2023-04-27 12:47:39,059 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1551.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 12:47:56,155 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1566.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:48:35,336 INFO [train.py:904] (6/8) Epoch 1, batch 1600, loss[loss=0.581, simple_loss=0.5393, pruned_loss=0.314, over 16926.00 frames. ], tot_loss[loss=0.5203, simple_loss=0.4891, pruned_loss=0.2824, over 3319588.37 frames. ], batch size: 51, lr: 4.88e-02, grad_scale: 4.0 2023-04-27 12:48:47,163 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.141e+02 5.212e+02 6.871e+02 8.557e+02 2.137e+03, threshold=1.374e+03, percent-clipped=9.0 2023-04-27 12:48:47,542 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1611.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:49:05,678 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1627.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 12:49:32,287 INFO [train.py:904] (6/8) Epoch 1, batch 1650, loss[loss=0.464, simple_loss=0.4728, pruned_loss=0.221, over 17003.00 frames. ], tot_loss[loss=0.5152, simple_loss=0.4872, pruned_loss=0.2763, over 3310402.76 frames. ], batch size: 50, lr: 4.87e-02, grad_scale: 4.0 2023-04-27 12:49:40,110 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-04-27 12:49:57,810 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1672.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:50:02,577 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2023-04-27 12:50:26,591 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8999, 4.8766, 4.6958, 4.7400, 4.5730, 5.0149, 5.0397, 4.5049], device='cuda:6'), covar=tensor([0.0448, 0.0564, 0.0657, 0.0776, 0.0901, 0.0445, 0.0495, 0.1037], device='cuda:6'), in_proj_covar=tensor([0.0104, 0.0122, 0.0117, 0.0131, 0.0134, 0.0103, 0.0095, 0.0128], device='cuda:6'), out_proj_covar=tensor([9.0347e-05, 1.1156e-04, 9.9810e-05, 1.1397e-04, 1.2488e-04, 9.1567e-05, 8.5029e-05, 1.1964e-04], device='cuda:6') 2023-04-27 12:50:29,818 INFO [train.py:904] (6/8) Epoch 1, batch 1700, loss[loss=0.5303, simple_loss=0.5187, pruned_loss=0.2683, over 17120.00 frames. ], tot_loss[loss=0.5068, simple_loss=0.4833, pruned_loss=0.2682, over 3316740.25 frames. ], batch size: 47, lr: 4.86e-02, grad_scale: 4.0 2023-04-27 12:50:41,826 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.189e+02 5.218e+02 6.714e+02 8.049e+02 1.427e+03, threshold=1.343e+03, percent-clipped=1.0 2023-04-27 12:51:05,647 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.08 vs. limit=5.0 2023-04-27 12:51:28,764 INFO [train.py:904] (6/8) Epoch 1, batch 1750, loss[loss=0.4158, simple_loss=0.4291, pruned_loss=0.1965, over 17212.00 frames. ], tot_loss[loss=0.4953, simple_loss=0.4775, pruned_loss=0.2581, over 3323619.01 frames. ], batch size: 44, lr: 4.86e-02, grad_scale: 4.0 2023-04-27 12:52:15,510 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1790.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 12:52:27,926 INFO [train.py:904] (6/8) Epoch 1, batch 1800, loss[loss=0.5043, simple_loss=0.5147, pruned_loss=0.2432, over 16676.00 frames. ], tot_loss[loss=0.4893, simple_loss=0.4753, pruned_loss=0.2521, over 3327484.18 frames. ], batch size: 57, lr: 4.85e-02, grad_scale: 4.0 2023-04-27 12:52:37,328 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1809.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:52:38,890 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.321e+02 5.485e+02 6.556e+02 7.752e+02 2.000e+03, threshold=1.311e+03, percent-clipped=5.0 2023-04-27 12:53:05,929 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2617, 3.2041, 2.8984, 2.9602, 3.3310, 2.8056, 3.0825, 2.9436], device='cuda:6'), covar=tensor([0.0544, 0.0506, 0.0819, 0.0634, 0.0465, 0.1083, 0.0978, 0.0766], device='cuda:6'), in_proj_covar=tensor([0.0068, 0.0066, 0.0062, 0.0065, 0.0068, 0.0070, 0.0074, 0.0067], device='cuda:6'), out_proj_covar=tensor([6.5789e-05, 6.1502e-05, 6.2163e-05, 6.1049e-05, 6.3591e-05, 6.8238e-05, 6.9816e-05, 6.2260e-05], device='cuda:6') 2023-04-27 12:53:10,964 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1838.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:53:18,447 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2023-04-27 12:53:22,967 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-27 12:53:25,086 INFO [train.py:904] (6/8) Epoch 1, batch 1850, loss[loss=0.4521, simple_loss=0.4419, pruned_loss=0.2302, over 16873.00 frames. ], tot_loss[loss=0.4839, simple_loss=0.4729, pruned_loss=0.2474, over 3324834.11 frames. ], batch size: 109, lr: 4.84e-02, grad_scale: 4.0 2023-04-27 12:53:27,540 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2023-04-27 12:53:32,840 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1857.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 12:53:32,931 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1857.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:53:48,657 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1870.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:54:22,085 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1899.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-27 12:54:23,724 INFO [train.py:904] (6/8) Epoch 1, batch 1900, loss[loss=0.446, simple_loss=0.4488, pruned_loss=0.2203, over 16453.00 frames. ], tot_loss[loss=0.4727, simple_loss=0.4666, pruned_loss=0.2389, over 3324241.85 frames. ], batch size: 68, lr: 4.83e-02, grad_scale: 4.0 2023-04-27 12:54:36,243 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.249e+02 5.423e+02 6.685e+02 8.506e+02 1.861e+03, threshold=1.337e+03, percent-clipped=8.0 2023-04-27 12:54:44,603 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1918.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 12:54:48,424 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.08 vs. limit=2.0 2023-04-27 12:54:49,157 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1922.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 12:54:58,016 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3261, 4.2239, 4.4056, 4.4284, 4.6923, 4.2966, 4.2628, 4.6968], device='cuda:6'), covar=tensor([0.0318, 0.0379, 0.0580, 0.0416, 0.0307, 0.0362, 0.0456, 0.0280], device='cuda:6'), in_proj_covar=tensor([0.0101, 0.0098, 0.0111, 0.0108, 0.0093, 0.0099, 0.0110, 0.0091], device='cuda:6'), out_proj_covar=tensor([8.3768e-05, 9.9151e-05, 1.1469e-04, 9.6510e-05, 9.3174e-05, 9.3619e-05, 1.0240e-04, 8.2561e-05], device='cuda:6') 2023-04-27 12:55:00,051 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1931.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-27 12:55:22,814 INFO [train.py:904] (6/8) Epoch 1, batch 1950, loss[loss=0.4188, simple_loss=0.4475, pruned_loss=0.1938, over 17135.00 frames. ], tot_loss[loss=0.4656, simple_loss=0.463, pruned_loss=0.2335, over 3319054.09 frames. ], batch size: 48, lr: 4.83e-02, grad_scale: 4.0 2023-04-27 12:55:39,908 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6221, 5.6306, 5.2495, 5.2483, 5.0992, 5.2447, 5.5143, 5.2090], device='cuda:6'), covar=tensor([0.0060, 0.0037, 0.0084, 0.0104, 0.0109, 0.0256, 0.0051, 0.0064], device='cuda:6'), in_proj_covar=tensor([0.0022, 0.0023, 0.0031, 0.0030, 0.0026, 0.0029, 0.0022, 0.0024], device='cuda:6'), out_proj_covar=tensor([1.7965e-05, 1.7440e-05, 2.4463e-05, 2.4081e-05, 1.9174e-05, 2.3150e-05, 1.7483e-05, 1.8969e-05], device='cuda:6') 2023-04-27 12:55:42,144 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1967.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 12:55:42,323 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0300, 3.4164, 3.2706, 3.4181, 3.6182, 1.5633, 3.6953, 2.6559], device='cuda:6'), covar=tensor([0.5081, 0.2409, 0.3641, 0.1213, 0.3229, 1.1662, 0.1238, 0.2626], device='cuda:6'), in_proj_covar=tensor([0.0049, 0.0031, 0.0050, 0.0037, 0.0029, 0.0065, 0.0035, 0.0025], device='cuda:6'), out_proj_covar=tensor([4.5300e-05, 2.8185e-05, 4.4898e-05, 2.8722e-05, 2.7507e-05, 5.6761e-05, 2.8559e-05, 2.3829e-05], device='cuda:6') 2023-04-27 12:56:23,561 INFO [train.py:904] (6/8) Epoch 1, batch 2000, loss[loss=0.4006, simple_loss=0.4085, pruned_loss=0.1963, over 16703.00 frames. ], tot_loss[loss=0.4585, simple_loss=0.459, pruned_loss=0.2284, over 3316687.74 frames. ], batch size: 89, lr: 4.82e-02, grad_scale: 8.0 2023-04-27 12:56:36,802 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.491e+02 5.646e+02 7.242e+02 8.659e+02 1.834e+03, threshold=1.448e+03, percent-clipped=2.0 2023-04-27 12:57:27,604 INFO [train.py:904] (6/8) Epoch 1, batch 2050, loss[loss=0.4417, simple_loss=0.4413, pruned_loss=0.2211, over 16923.00 frames. ], tot_loss[loss=0.4484, simple_loss=0.4528, pruned_loss=0.2215, over 3322254.41 frames. ], batch size: 109, lr: 4.81e-02, grad_scale: 8.0 2023-04-27 12:58:19,042 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2090.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 12:58:32,223 INFO [train.py:904] (6/8) Epoch 1, batch 2100, loss[loss=0.3703, simple_loss=0.4012, pruned_loss=0.1697, over 16845.00 frames. ], tot_loss[loss=0.4394, simple_loss=0.4483, pruned_loss=0.2149, over 3322854.59 frames. ], batch size: 42, lr: 4.80e-02, grad_scale: 16.0 2023-04-27 12:58:45,932 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.641e+02 4.108e+02 4.910e+02 5.858e+02 1.001e+03, threshold=9.819e+02, percent-clipped=0.0 2023-04-27 12:59:20,450 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=2138.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 12:59:35,891 INFO [train.py:904] (6/8) Epoch 1, batch 2150, loss[loss=0.4551, simple_loss=0.4535, pruned_loss=0.2284, over 16729.00 frames. ], tot_loss[loss=0.4342, simple_loss=0.4455, pruned_loss=0.2111, over 3321295.92 frames. ], batch size: 134, lr: 4.79e-02, grad_scale: 16.0 2023-04-27 13:00:10,051 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.99 vs. limit=5.0 2023-04-27 13:00:15,472 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5547, 3.5753, 3.4288, 3.2189, 3.4579, 3.8057, 2.8730, 3.2944], device='cuda:6'), covar=tensor([0.0208, 0.0179, 0.0347, 0.0346, 0.0250, 0.0142, 0.0565, 0.0293], device='cuda:6'), in_proj_covar=tensor([0.0054, 0.0052, 0.0053, 0.0057, 0.0047, 0.0052, 0.0057, 0.0059], device='cuda:6'), out_proj_covar=tensor([4.3045e-05, 3.8744e-05, 3.9408e-05, 4.5847e-05, 3.5438e-05, 3.8140e-05, 4.3969e-05, 4.5638e-05], device='cuda:6') 2023-04-27 13:00:32,066 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2194.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:00:40,251 INFO [train.py:904] (6/8) Epoch 1, batch 2200, loss[loss=0.3549, simple_loss=0.4077, pruned_loss=0.1511, over 17218.00 frames. ], tot_loss[loss=0.4277, simple_loss=0.4425, pruned_loss=0.2062, over 3324673.00 frames. ], batch size: 45, lr: 4.78e-02, grad_scale: 16.0 2023-04-27 13:00:45,651 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-04-27 13:00:53,199 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.805e+02 5.075e+02 6.122e+02 8.359e+02 1.658e+03, threshold=1.224e+03, percent-clipped=12.0 2023-04-27 13:00:56,703 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2213.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:01:04,673 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3916, 3.4221, 3.1613, 2.9162, 3.2369, 3.2000, 3.1904, 3.0836], device='cuda:6'), covar=tensor([0.0306, 0.0360, 0.0387, 0.0494, 0.0477, 0.0371, 0.0630, 0.0427], device='cuda:6'), in_proj_covar=tensor([0.0053, 0.0053, 0.0049, 0.0053, 0.0054, 0.0056, 0.0059, 0.0053], device='cuda:6'), out_proj_covar=tensor([4.9605e-05, 5.0031e-05, 4.7460e-05, 4.9285e-05, 4.8277e-05, 5.4967e-05, 5.7940e-05, 5.0326e-05], device='cuda:6') 2023-04-27 13:01:07,645 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2222.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 13:01:11,881 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2226.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:01:44,909 INFO [train.py:904] (6/8) Epoch 1, batch 2250, loss[loss=0.4597, simple_loss=0.4626, pruned_loss=0.2284, over 16427.00 frames. ], tot_loss[loss=0.4232, simple_loss=0.4408, pruned_loss=0.2026, over 3319039.54 frames. ], batch size: 146, lr: 4.77e-02, grad_scale: 16.0 2023-04-27 13:02:05,702 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2267.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:02:08,866 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=2270.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:02:49,153 INFO [train.py:904] (6/8) Epoch 1, batch 2300, loss[loss=0.3851, simple_loss=0.4365, pruned_loss=0.1668, over 16640.00 frames. ], tot_loss[loss=0.4175, simple_loss=0.4382, pruned_loss=0.1983, over 3324941.10 frames. ], batch size: 57, lr: 4.77e-02, grad_scale: 16.0 2023-04-27 13:03:01,864 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.985e+02 4.330e+02 5.680e+02 7.295e+02 1.284e+03, threshold=1.136e+03, percent-clipped=1.0 2023-04-27 13:03:07,886 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=2315.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:03:38,134 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-04-27 13:03:39,722 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2340.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:03:53,136 INFO [train.py:904] (6/8) Epoch 1, batch 2350, loss[loss=0.5265, simple_loss=0.5093, pruned_loss=0.2719, over 12251.00 frames. ], tot_loss[loss=0.4134, simple_loss=0.4356, pruned_loss=0.1955, over 3309323.17 frames. ], batch size: 246, lr: 4.76e-02, grad_scale: 16.0 2023-04-27 13:04:32,212 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2023-04-27 13:04:54,412 INFO [train.py:904] (6/8) Epoch 1, batch 2400, loss[loss=0.4449, simple_loss=0.4519, pruned_loss=0.2189, over 16771.00 frames. ], tot_loss[loss=0.4082, simple_loss=0.4331, pruned_loss=0.1916, over 3309786.67 frames. ], batch size: 124, lr: 4.75e-02, grad_scale: 16.0 2023-04-27 13:04:55,631 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2401.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 13:05:07,293 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.161e+02 4.808e+02 6.214e+02 8.409e+02 1.581e+03, threshold=1.243e+03, percent-clipped=3.0 2023-04-27 13:05:12,222 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.17 vs. limit=2.0 2023-04-27 13:05:31,323 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3318, 5.0235, 5.2844, 5.4348, 4.8730, 5.2248, 5.3148, 4.8735], device='cuda:6'), covar=tensor([0.0216, 0.0254, 0.0155, 0.0085, 0.0565, 0.0185, 0.0153, 0.0204], device='cuda:6'), in_proj_covar=tensor([0.0085, 0.0074, 0.0102, 0.0078, 0.0110, 0.0086, 0.0077, 0.0083], device='cuda:6'), out_proj_covar=tensor([9.2340e-05, 7.2364e-05, 1.1095e-04, 7.9460e-05, 1.2823e-04, 9.3563e-05, 8.1384e-05, 9.1395e-05], device='cuda:6') 2023-04-27 13:05:55,891 INFO [train.py:904] (6/8) Epoch 1, batch 2450, loss[loss=0.3977, simple_loss=0.4164, pruned_loss=0.1895, over 16680.00 frames. ], tot_loss[loss=0.4055, simple_loss=0.4322, pruned_loss=0.1893, over 3310930.60 frames. ], batch size: 89, lr: 4.74e-02, grad_scale: 16.0 2023-04-27 13:06:31,951 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.23 vs. limit=2.0 2023-04-27 13:06:38,639 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2817, 4.2916, 3.8867, 4.2092, 3.9374, 4.3145, 4.3842, 4.4308], device='cuda:6'), covar=tensor([0.0289, 0.0320, 0.0433, 0.0293, 0.0631, 0.0204, 0.0228, 0.0194], device='cuda:6'), in_proj_covar=tensor([0.0042, 0.0040, 0.0050, 0.0045, 0.0045, 0.0043, 0.0044, 0.0044], device='cuda:6'), out_proj_covar=tensor([3.7088e-05, 3.5200e-05, 4.4043e-05, 3.9020e-05, 4.2907e-05, 3.4249e-05, 4.2285e-05, 3.9100e-05], device='cuda:6') 2023-04-27 13:06:51,183 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2494.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:06:58,990 INFO [train.py:904] (6/8) Epoch 1, batch 2500, loss[loss=0.3737, simple_loss=0.4101, pruned_loss=0.1686, over 16584.00 frames. ], tot_loss[loss=0.3999, simple_loss=0.4291, pruned_loss=0.1853, over 3308956.31 frames. ], batch size: 68, lr: 4.73e-02, grad_scale: 16.0 2023-04-27 13:07:11,499 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.680e+02 4.365e+02 5.111e+02 7.039e+02 1.163e+03, threshold=1.022e+03, percent-clipped=0.0 2023-04-27 13:07:14,743 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2513.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:07:29,888 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2526.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:07:41,811 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9981, 4.9052, 4.2220, 4.5226, 4.9090, 5.0724, 4.3950, 4.9521], device='cuda:6'), covar=tensor([0.0182, 0.0152, 0.0203, 0.0250, 0.0097, 0.0114, 0.0159, 0.0132], device='cuda:6'), in_proj_covar=tensor([0.0050, 0.0047, 0.0068, 0.0063, 0.0047, 0.0052, 0.0063, 0.0060], device='cuda:6'), out_proj_covar=tensor([5.0744e-05, 4.8709e-05, 8.2173e-05, 6.7410e-05, 4.2587e-05, 5.0632e-05, 7.1931e-05, 6.4310e-05], device='cuda:6') 2023-04-27 13:07:50,931 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=2542.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:08:01,505 INFO [train.py:904] (6/8) Epoch 1, batch 2550, loss[loss=0.3877, simple_loss=0.405, pruned_loss=0.1852, over 16923.00 frames. ], tot_loss[loss=0.3991, simple_loss=0.4285, pruned_loss=0.1848, over 3310572.34 frames. ], batch size: 116, lr: 4.72e-02, grad_scale: 16.0 2023-04-27 13:08:09,471 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4654, 3.5206, 3.6237, 3.6690, 3.8950, 3.7355, 3.3835, 3.9514], device='cuda:6'), covar=tensor([0.0781, 0.0644, 0.0939, 0.0867, 0.0867, 0.0586, 0.0938, 0.0472], device='cuda:6'), in_proj_covar=tensor([0.0113, 0.0099, 0.0120, 0.0118, 0.0115, 0.0107, 0.0112, 0.0094], device='cuda:6'), out_proj_covar=tensor([1.1080e-04, 1.1729e-04, 1.3276e-04, 1.2189e-04, 1.2545e-04, 1.1636e-04, 1.1414e-04, 9.2100e-05], device='cuda:6') 2023-04-27 13:08:15,979 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=2561.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:08:32,316 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=2574.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:09:07,692 INFO [train.py:904] (6/8) Epoch 1, batch 2600, loss[loss=0.4057, simple_loss=0.427, pruned_loss=0.1922, over 15551.00 frames. ], tot_loss[loss=0.3962, simple_loss=0.4265, pruned_loss=0.1829, over 3297765.80 frames. ], batch size: 191, lr: 4.71e-02, grad_scale: 16.0 2023-04-27 13:09:20,088 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.935e+02 4.794e+02 6.148e+02 7.560e+02 1.171e+03, threshold=1.230e+03, percent-clipped=4.0 2023-04-27 13:10:11,240 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.8864, 6.1272, 5.5460, 6.2143, 5.7953, 5.5841, 5.8238, 6.3129], device='cuda:6'), covar=tensor([0.0273, 0.0369, 0.0518, 0.0183, 0.0405, 0.0266, 0.0283, 0.0141], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0140, 0.0123, 0.0094, 0.0126, 0.0111, 0.0116, 0.0094], device='cuda:6'), out_proj_covar=tensor([1.1030e-04, 1.2600e-04, 1.0476e-04, 7.1105e-05, 1.0530e-04, 9.0892e-05, 9.9254e-05, 8.3785e-05], device='cuda:6') 2023-04-27 13:10:11,888 INFO [train.py:904] (6/8) Epoch 1, batch 2650, loss[loss=0.308, simple_loss=0.3681, pruned_loss=0.124, over 16771.00 frames. ], tot_loss[loss=0.3918, simple_loss=0.4239, pruned_loss=0.1798, over 3304301.06 frames. ], batch size: 39, lr: 4.70e-02, grad_scale: 16.0 2023-04-27 13:10:55,296 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2685.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:11:09,975 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2696.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 13:11:15,038 INFO [train.py:904] (6/8) Epoch 1, batch 2700, loss[loss=0.4089, simple_loss=0.4319, pruned_loss=0.1929, over 16490.00 frames. ], tot_loss[loss=0.3872, simple_loss=0.422, pruned_loss=0.1762, over 3314232.90 frames. ], batch size: 75, lr: 4.69e-02, grad_scale: 16.0 2023-04-27 13:11:29,642 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.771e+02 4.333e+02 5.264e+02 6.238e+02 1.242e+03, threshold=1.053e+03, percent-clipped=1.0 2023-04-27 13:11:34,919 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6364, 5.8523, 5.4947, 5.9665, 5.3919, 5.5338, 5.6426, 5.9882], device='cuda:6'), covar=tensor([0.0307, 0.0447, 0.0436, 0.0210, 0.0437, 0.0236, 0.0268, 0.0169], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0132, 0.0114, 0.0090, 0.0119, 0.0102, 0.0115, 0.0087], device='cuda:6'), out_proj_covar=tensor([1.0648e-04, 1.2039e-04, 9.6791e-05, 6.8689e-05, 9.9649e-05, 8.1730e-05, 9.9074e-05, 7.7885e-05], device='cuda:6') 2023-04-27 13:12:14,235 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2746.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 13:12:20,232 INFO [train.py:904] (6/8) Epoch 1, batch 2750, loss[loss=0.4337, simple_loss=0.4454, pruned_loss=0.211, over 12399.00 frames. ], tot_loss[loss=0.3812, simple_loss=0.4182, pruned_loss=0.172, over 3317152.86 frames. ], batch size: 246, lr: 4.68e-02, grad_scale: 16.0 2023-04-27 13:13:23,779 INFO [train.py:904] (6/8) Epoch 1, batch 2800, loss[loss=0.3185, simple_loss=0.3748, pruned_loss=0.1311, over 16156.00 frames. ], tot_loss[loss=0.3782, simple_loss=0.4161, pruned_loss=0.1702, over 3315825.14 frames. ], batch size: 35, lr: 4.67e-02, grad_scale: 16.0 2023-04-27 13:13:24,271 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0744, 2.8076, 2.5792, 2.1155, 1.9934, 1.6389, 3.2741, 3.1854], device='cuda:6'), covar=tensor([0.0986, 0.1100, 0.0819, 0.2100, 0.2017, 0.2489, 0.0463, 0.0757], device='cuda:6'), in_proj_covar=tensor([0.0032, 0.0031, 0.0039, 0.0051, 0.0052, 0.0054, 0.0027, 0.0026], device='cuda:6'), out_proj_covar=tensor([3.5157e-05, 3.1637e-05, 3.7278e-05, 4.9826e-05, 4.7985e-05, 4.9902e-05, 2.6267e-05, 2.6809e-05], device='cuda:6') 2023-04-27 13:13:35,358 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.742e+02 4.329e+02 5.433e+02 6.254e+02 2.106e+03, threshold=1.087e+03, percent-clipped=5.0 2023-04-27 13:13:54,333 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2824.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:14:25,999 INFO [train.py:904] (6/8) Epoch 1, batch 2850, loss[loss=0.3625, simple_loss=0.4028, pruned_loss=0.1611, over 16468.00 frames. ], tot_loss[loss=0.3757, simple_loss=0.414, pruned_loss=0.1687, over 3314972.69 frames. ], batch size: 68, lr: 4.66e-02, grad_scale: 16.0 2023-04-27 13:14:50,323 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1025, 5.3995, 5.1604, 5.5021, 4.9691, 5.2500, 5.1602, 5.1671], device='cuda:6'), covar=tensor([0.0343, 0.0525, 0.0418, 0.0228, 0.0489, 0.0260, 0.0356, 0.0444], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0149, 0.0128, 0.0098, 0.0128, 0.0110, 0.0129, 0.0097], device='cuda:6'), out_proj_covar=tensor([1.1560e-04, 1.3702e-04, 1.1051e-04, 7.6873e-05, 1.0909e-04, 9.1998e-05, 1.1276e-04, 8.8855e-05], device='cuda:6') 2023-04-27 13:14:58,653 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2433, 5.1897, 4.6218, 4.6737, 5.2044, 5.2968, 4.7747, 5.2905], device='cuda:6'), covar=tensor([0.0177, 0.0124, 0.0160, 0.0224, 0.0067, 0.0114, 0.0134, 0.0114], device='cuda:6'), in_proj_covar=tensor([0.0051, 0.0047, 0.0067, 0.0065, 0.0047, 0.0054, 0.0062, 0.0058], device='cuda:6'), out_proj_covar=tensor([5.8812e-05, 5.3030e-05, 9.1082e-05, 7.8202e-05, 4.8529e-05, 5.8105e-05, 7.7988e-05, 7.2213e-05], device='cuda:6') 2023-04-27 13:15:09,670 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2885.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 13:15:22,209 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2023-04-27 13:15:27,285 INFO [train.py:904] (6/8) Epoch 1, batch 2900, loss[loss=0.3679, simple_loss=0.4134, pruned_loss=0.1612, over 17099.00 frames. ], tot_loss[loss=0.3762, simple_loss=0.4126, pruned_loss=0.1699, over 3309099.43 frames. ], batch size: 53, lr: 4.65e-02, grad_scale: 16.0 2023-04-27 13:15:40,095 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4280, 4.4291, 4.1054, 4.7024, 4.5590, 4.6681, 4.6872, 4.6076], device='cuda:6'), covar=tensor([0.0286, 0.0280, 0.1005, 0.0356, 0.0376, 0.0262, 0.0252, 0.0295], device='cuda:6'), in_proj_covar=tensor([0.0129, 0.0119, 0.0194, 0.0146, 0.0124, 0.0132, 0.0119, 0.0134], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-04-27 13:15:40,742 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.037e+02 5.015e+02 6.760e+02 8.832e+02 1.641e+03, threshold=1.352e+03, percent-clipped=13.0 2023-04-27 13:16:13,959 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2937.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:16:27,288 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-27 13:16:32,280 INFO [train.py:904] (6/8) Epoch 1, batch 2950, loss[loss=0.4603, simple_loss=0.4505, pruned_loss=0.235, over 12033.00 frames. ], tot_loss[loss=0.3739, simple_loss=0.4104, pruned_loss=0.1687, over 3306527.33 frames. ], batch size: 246, lr: 4.64e-02, grad_scale: 16.0 2023-04-27 13:16:47,514 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.29 vs. limit=5.0 2023-04-27 13:16:51,970 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9164, 3.6623, 3.6200, 3.8470, 2.9839, 3.8931, 2.9645, 3.8588], device='cuda:6'), covar=tensor([0.1921, 0.0117, 0.0260, 0.0124, 0.0209, 0.0192, 0.0322, 0.0087], device='cuda:6'), in_proj_covar=tensor([0.0097, 0.0042, 0.0045, 0.0029, 0.0026, 0.0039, 0.0050, 0.0038], device='cuda:6'), out_proj_covar=tensor([9.5897e-05, 3.9402e-05, 4.7007e-05, 3.1475e-05, 3.0789e-05, 4.0802e-05, 5.1296e-05, 3.5990e-05], device='cuda:6') 2023-04-27 13:17:29,856 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2996.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 13:17:32,186 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2998.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 13:17:35,316 INFO [train.py:904] (6/8) Epoch 1, batch 3000, loss[loss=0.3697, simple_loss=0.3986, pruned_loss=0.1704, over 16844.00 frames. ], tot_loss[loss=0.3724, simple_loss=0.4094, pruned_loss=0.1677, over 3314431.80 frames. ], batch size: 96, lr: 4.63e-02, grad_scale: 16.0 2023-04-27 13:17:35,317 INFO [train.py:929] (6/8) Computing validation loss 2023-04-27 13:17:45,053 INFO [train.py:938] (6/8) Epoch 1, validation: loss=0.2847, simple_loss=0.3895, pruned_loss=0.08992, over 944034.00 frames. 2023-04-27 13:17:45,053 INFO [train.py:939] (6/8) Maximum memory allocated so far is 15980MB 2023-04-27 13:17:59,838 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.008e+02 4.355e+02 5.212e+02 6.501e+02 1.071e+03, threshold=1.042e+03, percent-clipped=0.0 2023-04-27 13:18:09,364 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3019.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:18:36,918 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3041.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:18:41,574 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=3044.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:18:41,736 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5007, 2.5426, 2.2649, 2.6163, 2.3031, 2.5863, 2.4622, 2.6955], device='cuda:6'), covar=tensor([0.0103, 0.0098, 0.0220, 0.0093, 0.0100, 0.0129, 0.0131, 0.0057], device='cuda:6'), in_proj_covar=tensor([0.0022, 0.0020, 0.0023, 0.0021, 0.0023, 0.0020, 0.0024, 0.0018], device='cuda:6'), out_proj_covar=tensor([2.0568e-05, 2.0150e-05, 2.1531e-05, 1.9672e-05, 2.1103e-05, 1.9707e-05, 2.1214e-05, 1.6362e-05], device='cuda:6') 2023-04-27 13:18:50,096 INFO [train.py:904] (6/8) Epoch 1, batch 3050, loss[loss=0.3654, simple_loss=0.3971, pruned_loss=0.1669, over 16368.00 frames. ], tot_loss[loss=0.3678, simple_loss=0.4063, pruned_loss=0.1646, over 3314578.28 frames. ], batch size: 146, lr: 4.62e-02, grad_scale: 16.0 2023-04-27 13:19:27,620 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3080.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 13:19:53,976 INFO [train.py:904] (6/8) Epoch 1, batch 3100, loss[loss=0.3616, simple_loss=0.4063, pruned_loss=0.1585, over 17104.00 frames. ], tot_loss[loss=0.3659, simple_loss=0.4049, pruned_loss=0.1635, over 3323734.70 frames. ], batch size: 47, lr: 4.61e-02, grad_scale: 16.0 2023-04-27 13:20:07,649 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.721e+02 4.392e+02 5.238e+02 7.570e+02 1.450e+03, threshold=1.048e+03, percent-clipped=8.0 2023-04-27 13:21:00,201 INFO [train.py:904] (6/8) Epoch 1, batch 3150, loss[loss=0.3215, simple_loss=0.3779, pruned_loss=0.1325, over 16973.00 frames. ], tot_loss[loss=0.3643, simple_loss=0.4028, pruned_loss=0.1629, over 3315047.37 frames. ], batch size: 41, lr: 4.60e-02, grad_scale: 16.0 2023-04-27 13:21:38,518 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3180.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:22:05,299 INFO [train.py:904] (6/8) Epoch 1, batch 3200, loss[loss=0.3843, simple_loss=0.4092, pruned_loss=0.1797, over 16438.00 frames. ], tot_loss[loss=0.3614, simple_loss=0.4006, pruned_loss=0.1611, over 3311772.05 frames. ], batch size: 146, lr: 4.59e-02, grad_scale: 16.0 2023-04-27 13:22:17,347 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.273e+02 4.612e+02 6.158e+02 7.733e+02 1.150e+03, threshold=1.232e+03, percent-clipped=3.0 2023-04-27 13:23:07,318 INFO [train.py:904] (6/8) Epoch 1, batch 3250, loss[loss=0.3132, simple_loss=0.3774, pruned_loss=0.1245, over 17218.00 frames. ], tot_loss[loss=0.3582, simple_loss=0.399, pruned_loss=0.1587, over 3321484.93 frames. ], batch size: 46, lr: 4.58e-02, grad_scale: 16.0 2023-04-27 13:23:56,366 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3288.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:24:02,463 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3293.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:24:12,251 INFO [train.py:904] (6/8) Epoch 1, batch 3300, loss[loss=0.3858, simple_loss=0.4152, pruned_loss=0.1782, over 16909.00 frames. ], tot_loss[loss=0.3599, simple_loss=0.4004, pruned_loss=0.1597, over 3316964.51 frames. ], batch size: 109, lr: 4.57e-02, grad_scale: 16.0 2023-04-27 13:24:25,122 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.531e+02 4.292e+02 5.268e+02 6.867e+02 1.392e+03, threshold=1.054e+03, percent-clipped=2.0 2023-04-27 13:24:30,856 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-27 13:24:54,860 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2915, 2.4067, 2.3641, 2.2037, 2.4888, 2.6293, 2.3739, 1.8664], device='cuda:6'), covar=tensor([0.0220, 0.0324, 0.0313, 0.0270, 0.0182, 0.0136, 0.0192, 0.0323], device='cuda:6'), in_proj_covar=tensor([0.0027, 0.0034, 0.0034, 0.0033, 0.0030, 0.0033, 0.0034, 0.0031], device='cuda:6'), out_proj_covar=tensor([3.7616e-05, 4.0545e-05, 4.2951e-05, 3.6830e-05, 3.6659e-05, 3.8707e-05, 3.7145e-05, 3.8369e-05], device='cuda:6') 2023-04-27 13:25:03,541 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3341.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:25:16,148 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3349.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 13:25:17,759 INFO [train.py:904] (6/8) Epoch 1, batch 3350, loss[loss=0.2653, simple_loss=0.335, pruned_loss=0.0978, over 16772.00 frames. ], tot_loss[loss=0.3581, simple_loss=0.3998, pruned_loss=0.1582, over 3313620.18 frames. ], batch size: 39, lr: 4.56e-02, grad_scale: 16.0 2023-04-27 13:25:50,431 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3375.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:26:09,263 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=3389.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:26:24,811 INFO [train.py:904] (6/8) Epoch 1, batch 3400, loss[loss=0.343, simple_loss=0.4068, pruned_loss=0.1396, over 16678.00 frames. ], tot_loss[loss=0.3582, simple_loss=0.4001, pruned_loss=0.1582, over 3297855.77 frames. ], batch size: 57, lr: 4.55e-02, grad_scale: 16.0 2023-04-27 13:26:39,219 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.681e+02 4.195e+02 5.225e+02 6.801e+02 1.040e+03, threshold=1.045e+03, percent-clipped=0.0 2023-04-27 13:27:32,304 INFO [train.py:904] (6/8) Epoch 1, batch 3450, loss[loss=0.3694, simple_loss=0.4172, pruned_loss=0.1608, over 17128.00 frames. ], tot_loss[loss=0.3527, simple_loss=0.3959, pruned_loss=0.1547, over 3307402.65 frames. ], batch size: 47, lr: 4.54e-02, grad_scale: 16.0 2023-04-27 13:28:03,169 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-27 13:28:12,587 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3480.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:28:39,711 INFO [train.py:904] (6/8) Epoch 1, batch 3500, loss[loss=0.3737, simple_loss=0.4056, pruned_loss=0.1709, over 16721.00 frames. ], tot_loss[loss=0.3489, simple_loss=0.3926, pruned_loss=0.1526, over 3312004.49 frames. ], batch size: 134, lr: 4.53e-02, grad_scale: 16.0 2023-04-27 13:28:53,813 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.034e+02 4.599e+02 5.593e+02 7.423e+02 2.273e+03, threshold=1.119e+03, percent-clipped=10.0 2023-04-27 13:29:16,288 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=3528.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:29:40,763 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3546.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:29:46,547 INFO [train.py:904] (6/8) Epoch 1, batch 3550, loss[loss=0.3418, simple_loss=0.4037, pruned_loss=0.1399, over 17149.00 frames. ], tot_loss[loss=0.3492, simple_loss=0.3922, pruned_loss=0.1531, over 3309211.48 frames. ], batch size: 49, lr: 4.51e-02, grad_scale: 16.0 2023-04-27 13:30:27,408 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8706, 5.2247, 4.8558, 5.2464, 4.7246, 5.0844, 4.8955, 5.1492], device='cuda:6'), covar=tensor([0.0560, 0.0565, 0.0563, 0.0315, 0.0538, 0.0330, 0.0461, 0.0418], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0172, 0.0141, 0.0112, 0.0143, 0.0125, 0.0151, 0.0108], device='cuda:6'), out_proj_covar=tensor([1.3364e-04, 1.6320e-04, 1.2475e-04, 9.7665e-05, 1.2761e-04, 1.0949e-04, 1.4443e-04, 1.0621e-04], device='cuda:6') 2023-04-27 13:30:44,887 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3593.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:30:54,496 INFO [train.py:904] (6/8) Epoch 1, batch 3600, loss[loss=0.311, simple_loss=0.355, pruned_loss=0.1335, over 16769.00 frames. ], tot_loss[loss=0.3462, simple_loss=0.3902, pruned_loss=0.1511, over 3302146.54 frames. ], batch size: 83, lr: 4.50e-02, grad_scale: 16.0 2023-04-27 13:31:04,159 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3607.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 13:31:08,701 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.446e+02 5.845e+02 7.592e+02 1.096e+03 2.095e+03, threshold=1.518e+03, percent-clipped=22.0 2023-04-27 13:31:52,475 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=3641.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:31:56,486 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3644.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:32:06,054 INFO [train.py:904] (6/8) Epoch 1, batch 3650, loss[loss=0.2897, simple_loss=0.3329, pruned_loss=0.1232, over 16703.00 frames. ], tot_loss[loss=0.3422, simple_loss=0.3863, pruned_loss=0.1491, over 3298031.89 frames. ], batch size: 89, lr: 4.49e-02, grad_scale: 16.0 2023-04-27 13:32:42,343 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3675.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:32:43,575 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3676.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:32:47,038 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-04-27 13:33:20,375 INFO [train.py:904] (6/8) Epoch 1, batch 3700, loss[loss=0.3393, simple_loss=0.3709, pruned_loss=0.1538, over 16860.00 frames. ], tot_loss[loss=0.3387, simple_loss=0.3811, pruned_loss=0.1482, over 3299134.82 frames. ], batch size: 116, lr: 4.48e-02, grad_scale: 16.0 2023-04-27 13:33:35,072 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.010e+02 4.933e+02 6.416e+02 8.029e+02 1.327e+03, threshold=1.283e+03, percent-clipped=0.0 2023-04-27 13:33:54,096 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=3723.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:34:15,092 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3737.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 13:34:33,710 INFO [train.py:904] (6/8) Epoch 1, batch 3750, loss[loss=0.3677, simple_loss=0.3883, pruned_loss=0.1735, over 16790.00 frames. ], tot_loss[loss=0.3376, simple_loss=0.3796, pruned_loss=0.1478, over 3292231.75 frames. ], batch size: 124, lr: 4.47e-02, grad_scale: 16.0 2023-04-27 13:34:34,176 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4163, 4.4198, 4.7234, 4.7552, 4.9842, 4.4897, 4.5617, 4.7104], device='cuda:6'), covar=tensor([0.0340, 0.0284, 0.0432, 0.0507, 0.0432, 0.0436, 0.0497, 0.0298], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0100, 0.0128, 0.0127, 0.0136, 0.0113, 0.0123, 0.0099], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:6') 2023-04-27 13:35:33,805 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2106, 4.0467, 4.3410, 4.3655, 4.6864, 4.2735, 4.2447, 4.5291], device='cuda:6'), covar=tensor([0.0303, 0.0296, 0.0458, 0.0438, 0.0258, 0.0259, 0.0388, 0.0194], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0103, 0.0131, 0.0132, 0.0139, 0.0114, 0.0125, 0.0101], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:6') 2023-04-27 13:35:46,282 INFO [train.py:904] (6/8) Epoch 1, batch 3800, loss[loss=0.3435, simple_loss=0.3701, pruned_loss=0.1585, over 16832.00 frames. ], tot_loss[loss=0.3373, simple_loss=0.3786, pruned_loss=0.148, over 3291910.78 frames. ], batch size: 102, lr: 4.46e-02, grad_scale: 16.0 2023-04-27 13:36:00,501 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.355e+02 5.070e+02 6.208e+02 8.135e+02 1.675e+03, threshold=1.242e+03, percent-clipped=4.0 2023-04-27 13:36:57,572 INFO [train.py:904] (6/8) Epoch 1, batch 3850, loss[loss=0.2959, simple_loss=0.3425, pruned_loss=0.1247, over 16495.00 frames. ], tot_loss[loss=0.3346, simple_loss=0.3759, pruned_loss=0.1466, over 3296834.61 frames. ], batch size: 75, lr: 4.45e-02, grad_scale: 16.0 2023-04-27 13:38:09,799 INFO [train.py:904] (6/8) Epoch 1, batch 3900, loss[loss=0.335, simple_loss=0.3751, pruned_loss=0.1474, over 16298.00 frames. ], tot_loss[loss=0.3321, simple_loss=0.3734, pruned_loss=0.1454, over 3291739.69 frames. ], batch size: 165, lr: 4.44e-02, grad_scale: 16.0 2023-04-27 13:38:11,841 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3902.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:38:24,273 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8428, 3.6748, 3.4010, 2.4444, 3.0636, 1.9441, 3.5673, 4.0805], device='cuda:6'), covar=tensor([0.0277, 0.0396, 0.0309, 0.1496, 0.0757, 0.1508, 0.0361, 0.0201], device='cuda:6'), in_proj_covar=tensor([0.0044, 0.0041, 0.0062, 0.0106, 0.0093, 0.0095, 0.0053, 0.0035], device='cuda:6'), out_proj_covar=tensor([6.1059e-05, 5.4762e-05, 6.9240e-05, 1.1332e-04, 1.0082e-04, 9.9217e-05, 6.4467e-05, 4.8223e-05], device='cuda:6') 2023-04-27 13:38:24,733 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.787e+02 4.914e+02 5.725e+02 7.504e+02 1.784e+03, threshold=1.145e+03, percent-clipped=3.0 2023-04-27 13:38:26,671 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-04-27 13:38:46,384 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3926.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:39:11,977 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3944.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:39:21,370 INFO [train.py:904] (6/8) Epoch 1, batch 3950, loss[loss=0.3159, simple_loss=0.3584, pruned_loss=0.1368, over 16857.00 frames. ], tot_loss[loss=0.3298, simple_loss=0.3712, pruned_loss=0.1442, over 3287775.84 frames. ], batch size: 109, lr: 4.43e-02, grad_scale: 16.0 2023-04-27 13:39:48,107 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2023-04-27 13:40:13,990 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3987.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:40:20,466 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=3992.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:40:37,499 INFO [train.py:904] (6/8) Epoch 1, batch 4000, loss[loss=0.2907, simple_loss=0.348, pruned_loss=0.1166, over 16899.00 frames. ], tot_loss[loss=0.3269, simple_loss=0.3686, pruned_loss=0.1426, over 3287385.83 frames. ], batch size: 96, lr: 4.42e-02, grad_scale: 16.0 2023-04-27 13:40:52,170 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.081e+02 4.378e+02 5.733e+02 7.371e+02 1.613e+03, threshold=1.147e+03, percent-clipped=6.0 2023-04-27 13:41:24,000 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4032.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 13:41:40,132 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-27 13:41:50,490 INFO [train.py:904] (6/8) Epoch 1, batch 4050, loss[loss=0.2376, simple_loss=0.3088, pruned_loss=0.08319, over 16880.00 frames. ], tot_loss[loss=0.3188, simple_loss=0.3645, pruned_loss=0.1365, over 3287717.10 frames. ], batch size: 109, lr: 4.41e-02, grad_scale: 16.0 2023-04-27 13:43:04,570 INFO [train.py:904] (6/8) Epoch 1, batch 4100, loss[loss=0.3048, simple_loss=0.3639, pruned_loss=0.1228, over 16622.00 frames. ], tot_loss[loss=0.3123, simple_loss=0.3615, pruned_loss=0.1315, over 3283122.74 frames. ], batch size: 62, lr: 4.40e-02, grad_scale: 32.0 2023-04-27 13:43:18,962 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.292e+02 4.086e+02 5.427e+02 7.425e+02 1.337e+03, threshold=1.085e+03, percent-clipped=4.0 2023-04-27 13:43:48,044 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1364, 1.4796, 1.5643, 2.0316, 2.3040, 1.9775, 1.8884, 1.6810], device='cuda:6'), covar=tensor([0.0096, 0.0385, 0.0213, 0.0165, 0.0075, 0.0113, 0.0126, 0.0188], device='cuda:6'), in_proj_covar=tensor([0.0023, 0.0037, 0.0027, 0.0027, 0.0022, 0.0026, 0.0026, 0.0030], device='cuda:6'), out_proj_covar=tensor([2.4512e-05, 3.9128e-05, 2.8555e-05, 2.7528e-05, 1.9549e-05, 2.2970e-05, 2.4430e-05, 2.7612e-05], device='cuda:6') 2023-04-27 13:44:05,026 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=4141.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:44:20,296 INFO [train.py:904] (6/8) Epoch 1, batch 4150, loss[loss=0.4223, simple_loss=0.4351, pruned_loss=0.2048, over 11072.00 frames. ], tot_loss[loss=0.3242, simple_loss=0.3723, pruned_loss=0.1381, over 3257945.56 frames. ], batch size: 246, lr: 4.39e-02, grad_scale: 32.0 2023-04-27 13:44:43,495 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8597, 5.0559, 4.7870, 5.0188, 4.5156, 4.8866, 4.6539, 5.1760], device='cuda:6'), covar=tensor([0.0344, 0.0595, 0.0519, 0.0321, 0.0472, 0.0361, 0.0459, 0.0275], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0155, 0.0134, 0.0101, 0.0130, 0.0115, 0.0137, 0.0095], device='cuda:6'), out_proj_covar=tensor([1.2640e-04, 1.4979e-04, 1.1767e-04, 9.1411e-05, 1.1959e-04, 1.0498e-04, 1.3573e-04, 9.5285e-05], device='cuda:6') 2023-04-27 13:45:37,099 INFO [train.py:904] (6/8) Epoch 1, batch 4200, loss[loss=0.3364, simple_loss=0.3991, pruned_loss=0.1369, over 17105.00 frames. ], tot_loss[loss=0.3315, simple_loss=0.3807, pruned_loss=0.1411, over 3229109.30 frames. ], batch size: 48, lr: 4.38e-02, grad_scale: 16.0 2023-04-27 13:45:39,615 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4202.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 13:45:39,697 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=4202.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:45:52,579 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.534e+02 4.783e+02 6.571e+02 8.274e+02 1.863e+03, threshold=1.314e+03, percent-clipped=9.0 2023-04-27 13:45:53,967 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1161, 2.5324, 2.4835, 3.1769, 3.3143, 3.1830, 2.2870, 2.9317], device='cuda:6'), covar=tensor([0.1657, 0.0165, 0.0881, 0.0102, 0.0099, 0.0243, 0.0450, 0.0151], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0063, 0.0109, 0.0040, 0.0039, 0.0059, 0.0083, 0.0060], device='cuda:6'), out_proj_covar=tensor([1.5513e-04, 7.1561e-05, 1.2656e-04, 5.1115e-05, 5.1317e-05, 8.0364e-05, 9.6951e-05, 6.9217e-05], device='cuda:6') 2023-04-27 13:46:32,797 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.13 vs. limit=2.0 2023-04-27 13:46:35,237 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1929, 1.8391, 1.9767, 1.9912, 2.3585, 1.9494, 2.2388, 2.2145], device='cuda:6'), covar=tensor([0.0090, 0.0658, 0.0238, 0.0198, 0.0097, 0.0255, 0.0169, 0.0206], device='cuda:6'), in_proj_covar=tensor([0.0028, 0.0050, 0.0037, 0.0034, 0.0030, 0.0034, 0.0035, 0.0032], device='cuda:6'), out_proj_covar=tensor([3.9012e-05, 8.0621e-05, 5.7848e-05, 4.3657e-05, 4.1779e-05, 4.7990e-05, 4.5832e-05, 4.1724e-05], device='cuda:6') 2023-04-27 13:46:50,185 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=4250.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:46:50,856 INFO [train.py:904] (6/8) Epoch 1, batch 4250, loss[loss=0.3305, simple_loss=0.3869, pruned_loss=0.1371, over 16910.00 frames. ], tot_loss[loss=0.331, simple_loss=0.3819, pruned_loss=0.14, over 3217682.78 frames. ], batch size: 96, lr: 4.36e-02, grad_scale: 16.0 2023-04-27 13:47:05,472 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2023-04-27 13:47:35,518 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1377, 1.7288, 1.9131, 2.6986, 1.7771, 2.9689, 2.6763, 2.6789], device='cuda:6'), covar=tensor([0.0036, 0.0319, 0.0215, 0.0116, 0.0133, 0.0065, 0.0174, 0.0172], device='cuda:6'), in_proj_covar=tensor([0.0022, 0.0037, 0.0027, 0.0027, 0.0023, 0.0026, 0.0026, 0.0030], device='cuda:6'), out_proj_covar=tensor([2.3622e-05, 4.1405e-05, 2.8346e-05, 2.8323e-05, 2.0890e-05, 2.2978e-05, 2.5473e-05, 2.8112e-05], device='cuda:6') 2023-04-27 13:47:37,200 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4282.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:47:45,305 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=4288.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:48:04,724 INFO [train.py:904] (6/8) Epoch 1, batch 4300, loss[loss=0.3428, simple_loss=0.4135, pruned_loss=0.136, over 16828.00 frames. ], tot_loss[loss=0.3285, simple_loss=0.3819, pruned_loss=0.1376, over 3217398.40 frames. ], batch size: 102, lr: 4.35e-02, grad_scale: 16.0 2023-04-27 13:48:21,410 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.968e+02 4.766e+02 5.895e+02 7.865e+02 1.445e+03, threshold=1.179e+03, percent-clipped=2.0 2023-04-27 13:48:52,596 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4332.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:49:06,346 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8439, 2.4467, 2.6012, 3.0845, 3.2821, 2.7655, 3.0659, 2.7096], device='cuda:6'), covar=tensor([0.0087, 0.0570, 0.0222, 0.0101, 0.0072, 0.0165, 0.0129, 0.0204], device='cuda:6'), in_proj_covar=tensor([0.0028, 0.0052, 0.0036, 0.0034, 0.0030, 0.0033, 0.0035, 0.0032], device='cuda:6'), out_proj_covar=tensor([3.9820e-05, 8.4778e-05, 5.7501e-05, 4.3722e-05, 4.1623e-05, 4.7635e-05, 4.5728e-05, 4.3514e-05], device='cuda:6') 2023-04-27 13:49:17,764 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=4349.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:49:20,274 INFO [train.py:904] (6/8) Epoch 1, batch 4350, loss[loss=0.299, simple_loss=0.3705, pruned_loss=0.1137, over 16605.00 frames. ], tot_loss[loss=0.3311, simple_loss=0.3854, pruned_loss=0.1384, over 3226256.17 frames. ], batch size: 62, lr: 4.34e-02, grad_scale: 16.0 2023-04-27 13:49:39,477 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3748, 3.2471, 2.6074, 3.9967, 3.3349, 3.5299, 3.0700, 3.1989], device='cuda:6'), covar=tensor([0.0432, 0.0595, 0.0707, 0.0309, 0.1126, 0.0351, 0.0669, 0.0992], device='cuda:6'), in_proj_covar=tensor([0.0055, 0.0061, 0.0052, 0.0050, 0.0103, 0.0058, 0.0075, 0.0061], device='cuda:6'), out_proj_covar=tensor([5.8344e-05, 6.3588e-05, 5.2913e-05, 5.8096e-05, 1.1173e-04, 6.1608e-05, 7.4440e-05, 7.2833e-05], device='cuda:6') 2023-04-27 13:50:04,949 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=4380.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:50:30,251 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8046, 4.7373, 4.0481, 4.0410, 4.6994, 4.5445, 4.4032, 4.5495], device='cuda:6'), covar=tensor([0.0075, 0.0068, 0.0110, 0.0306, 0.0049, 0.0106, 0.0064, 0.0099], device='cuda:6'), in_proj_covar=tensor([0.0046, 0.0037, 0.0055, 0.0067, 0.0039, 0.0049, 0.0052, 0.0050], device='cuda:6'), out_proj_covar=tensor([7.9607e-05, 6.3198e-05, 1.0535e-04, 1.1575e-04, 5.9936e-05, 8.4657e-05, 9.5190e-05, 9.6132e-05], device='cuda:6') 2023-04-27 13:50:36,405 INFO [train.py:904] (6/8) Epoch 1, batch 4400, loss[loss=0.3072, simple_loss=0.3762, pruned_loss=0.1191, over 17231.00 frames. ], tot_loss[loss=0.3316, simple_loss=0.3862, pruned_loss=0.1385, over 3221003.13 frames. ], batch size: 45, lr: 4.33e-02, grad_scale: 16.0 2023-04-27 13:50:42,268 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2023-04-27 13:50:51,959 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.736e+02 5.030e+02 6.587e+02 8.143e+02 1.430e+03, threshold=1.317e+03, percent-clipped=9.0 2023-04-27 13:51:37,030 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5633, 4.3018, 4.0791, 4.2077, 3.4132, 2.5508, 4.5044, 4.5942], device='cuda:6'), covar=tensor([0.1767, 0.0474, 0.0717, 0.0271, 0.1534, 0.1419, 0.0193, 0.0040], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0091, 0.0131, 0.0075, 0.0107, 0.0114, 0.0073, 0.0039], device='cuda:6'), out_proj_covar=tensor([1.6661e-04, 1.0925e-04, 1.3250e-04, 7.7748e-05, 1.3076e-04, 1.1872e-04, 7.9735e-05, 4.4619e-05], device='cuda:6') 2023-04-27 13:51:48,407 INFO [train.py:904] (6/8) Epoch 1, batch 4450, loss[loss=0.3313, simple_loss=0.3952, pruned_loss=0.1337, over 16871.00 frames. ], tot_loss[loss=0.332, simple_loss=0.3884, pruned_loss=0.1377, over 3227978.04 frames. ], batch size: 116, lr: 4.32e-02, grad_scale: 16.0 2023-04-27 13:52:57,241 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4497.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:53:02,592 INFO [train.py:904] (6/8) Epoch 1, batch 4500, loss[loss=0.284, simple_loss=0.3562, pruned_loss=0.1058, over 16561.00 frames. ], tot_loss[loss=0.3294, simple_loss=0.3866, pruned_loss=0.1361, over 3230950.89 frames. ], batch size: 75, lr: 4.31e-02, grad_scale: 8.0 2023-04-27 13:53:20,027 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.477e+02 4.034e+02 4.961e+02 6.637e+02 1.457e+03, threshold=9.923e+02, percent-clipped=1.0 2023-04-27 13:53:58,151 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5678, 2.0936, 2.1135, 2.5086, 2.6598, 2.5298, 2.5516, 2.2449], device='cuda:6'), covar=tensor([0.0101, 0.0625, 0.0309, 0.0137, 0.0118, 0.0188, 0.0192, 0.0349], device='cuda:6'), in_proj_covar=tensor([0.0029, 0.0054, 0.0039, 0.0035, 0.0031, 0.0035, 0.0036, 0.0032], device='cuda:6'), out_proj_covar=tensor([4.0276e-05, 9.2628e-05, 6.4129e-05, 4.6870e-05, 4.4632e-05, 5.1610e-05, 4.7976e-05, 4.5887e-05], device='cuda:6') 2023-04-27 13:54:14,092 INFO [train.py:904] (6/8) Epoch 1, batch 4550, loss[loss=0.3283, simple_loss=0.3864, pruned_loss=0.1351, over 17127.00 frames. ], tot_loss[loss=0.3295, simple_loss=0.3868, pruned_loss=0.1361, over 3239896.12 frames. ], batch size: 47, lr: 4.30e-02, grad_scale: 8.0 2023-04-27 13:54:50,498 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0492, 3.9978, 3.6919, 4.4680, 4.3351, 4.2611, 4.2670, 4.2044], device='cuda:6'), covar=tensor([0.0384, 0.0309, 0.1364, 0.0313, 0.0398, 0.0287, 0.0348, 0.0297], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0129, 0.0218, 0.0152, 0.0123, 0.0135, 0.0117, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 13:54:57,994 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4582.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:55:12,377 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5298, 3.9217, 3.7754, 3.5829, 2.9808, 2.2460, 4.0689, 4.4952], device='cuda:6'), covar=tensor([0.1476, 0.0544, 0.0812, 0.0421, 0.2022, 0.1704, 0.0219, 0.0048], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0096, 0.0135, 0.0077, 0.0117, 0.0116, 0.0073, 0.0042], device='cuda:6'), out_proj_covar=tensor([1.7434e-04, 1.1561e-04, 1.3798e-04, 8.0854e-05, 1.4219e-04, 1.2191e-04, 8.1404e-05, 4.9436e-05], device='cuda:6') 2023-04-27 13:55:25,714 INFO [train.py:904] (6/8) Epoch 1, batch 4600, loss[loss=0.2885, simple_loss=0.3618, pruned_loss=0.1076, over 16597.00 frames. ], tot_loss[loss=0.3275, simple_loss=0.3863, pruned_loss=0.1343, over 3244901.61 frames. ], batch size: 68, lr: 4.29e-02, grad_scale: 8.0 2023-04-27 13:55:43,311 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.332e+02 3.980e+02 4.876e+02 6.501e+02 1.584e+03, threshold=9.751e+02, percent-clipped=6.0 2023-04-27 13:56:07,462 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=4630.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:56:28,166 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4644.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:56:36,565 INFO [train.py:904] (6/8) Epoch 1, batch 4650, loss[loss=0.3436, simple_loss=0.3833, pruned_loss=0.152, over 11809.00 frames. ], tot_loss[loss=0.3251, simple_loss=0.3842, pruned_loss=0.133, over 3234997.37 frames. ], batch size: 246, lr: 4.28e-02, grad_scale: 8.0 2023-04-27 13:56:51,249 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8468, 4.0807, 3.9643, 2.0370, 3.8071, 4.0645, 3.7982, 3.7890], device='cuda:6'), covar=tensor([0.0086, 0.0082, 0.0180, 0.1482, 0.0168, 0.0086, 0.0107, 0.0126], device='cuda:6'), in_proj_covar=tensor([0.0053, 0.0055, 0.0057, 0.0119, 0.0052, 0.0051, 0.0051, 0.0066], device='cuda:6'), out_proj_covar=tensor([6.9720e-05, 7.0999e-05, 8.0782e-05, 1.5864e-04, 7.6653e-05, 6.6045e-05, 7.8801e-05, 8.3888e-05], device='cuda:6') 2023-04-27 13:57:00,894 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-27 13:57:50,180 INFO [train.py:904] (6/8) Epoch 1, batch 4700, loss[loss=0.2987, simple_loss=0.3619, pruned_loss=0.1178, over 16813.00 frames. ], tot_loss[loss=0.3213, simple_loss=0.3804, pruned_loss=0.1311, over 3229095.68 frames. ], batch size: 39, lr: 4.27e-02, grad_scale: 8.0 2023-04-27 13:58:07,983 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.460e+02 4.210e+02 5.632e+02 6.697e+02 1.082e+03, threshold=1.126e+03, percent-clipped=2.0 2023-04-27 13:59:02,501 INFO [train.py:904] (6/8) Epoch 1, batch 4750, loss[loss=0.3859, simple_loss=0.4066, pruned_loss=0.1826, over 12112.00 frames. ], tot_loss[loss=0.3172, simple_loss=0.3765, pruned_loss=0.129, over 3225800.00 frames. ], batch size: 247, lr: 4.26e-02, grad_scale: 8.0 2023-04-27 13:59:16,861 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2792, 3.1843, 3.1392, 3.3027, 2.7215, 3.2504, 3.1313, 2.9412], device='cuda:6'), covar=tensor([0.0305, 0.0162, 0.0238, 0.0166, 0.0924, 0.0239, 0.0597, 0.0243], device='cuda:6'), in_proj_covar=tensor([0.0080, 0.0061, 0.0105, 0.0079, 0.0133, 0.0086, 0.0074, 0.0082], device='cuda:6'), out_proj_covar=tensor([1.3627e-04, 9.7197e-05, 1.7644e-04, 1.2263e-04, 1.8872e-04, 1.5197e-04, 1.2558e-04, 1.4227e-04], device='cuda:6') 2023-04-27 13:59:51,512 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9478, 3.5690, 3.3548, 3.2015, 2.7398, 2.1851, 3.6059, 4.1204], device='cuda:6'), covar=tensor([0.1429, 0.0530, 0.0870, 0.0429, 0.1556, 0.1458, 0.0214, 0.0063], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0101, 0.0140, 0.0082, 0.0123, 0.0118, 0.0077, 0.0044], device='cuda:6'), out_proj_covar=tensor([1.7686e-04, 1.2171e-04, 1.4536e-04, 8.8693e-05, 1.5089e-04, 1.2673e-04, 8.7011e-05, 5.2606e-05], device='cuda:6') 2023-04-27 14:00:11,423 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4797.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:00:16,558 INFO [train.py:904] (6/8) Epoch 1, batch 4800, loss[loss=0.3268, simple_loss=0.3957, pruned_loss=0.1289, over 16427.00 frames. ], tot_loss[loss=0.3136, simple_loss=0.373, pruned_loss=0.1271, over 3228116.10 frames. ], batch size: 146, lr: 4.25e-02, grad_scale: 8.0 2023-04-27 14:00:34,048 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.251e+02 4.322e+02 5.200e+02 6.651e+02 1.076e+03, threshold=1.040e+03, percent-clipped=0.0 2023-04-27 14:01:22,948 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=4845.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:01:31,748 INFO [train.py:904] (6/8) Epoch 1, batch 4850, loss[loss=0.363, simple_loss=0.4034, pruned_loss=0.1614, over 11974.00 frames. ], tot_loss[loss=0.3137, simple_loss=0.3739, pruned_loss=0.1267, over 3216457.11 frames. ], batch size: 246, lr: 4.24e-02, grad_scale: 8.0 2023-04-27 14:02:47,242 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0248, 3.0784, 1.4666, 3.1366, 2.4474, 2.9800, 1.7592, 2.5904], device='cuda:6'), covar=tensor([0.0099, 0.0116, 0.1122, 0.0090, 0.0378, 0.0149, 0.1038, 0.0363], device='cuda:6'), in_proj_covar=tensor([0.0043, 0.0040, 0.0082, 0.0042, 0.0068, 0.0039, 0.0088, 0.0056], device='cuda:6'), out_proj_covar=tensor([5.3866e-05, 5.4991e-05, 1.1664e-04, 5.3054e-05, 8.6234e-05, 5.9960e-05, 1.1863e-04, 7.8972e-05], device='cuda:6') 2023-04-27 14:02:49,096 INFO [train.py:904] (6/8) Epoch 1, batch 4900, loss[loss=0.3221, simple_loss=0.3916, pruned_loss=0.1263, over 15246.00 frames. ], tot_loss[loss=0.3124, simple_loss=0.3738, pruned_loss=0.1255, over 3196060.24 frames. ], batch size: 190, lr: 4.23e-02, grad_scale: 8.0 2023-04-27 14:03:07,683 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.068e+02 3.780e+02 4.613e+02 5.959e+02 1.211e+03, threshold=9.227e+02, percent-clipped=3.0 2023-04-27 14:03:10,024 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9278, 2.0024, 2.0712, 2.1058, 2.6759, 2.7373, 2.6011, 2.6518], device='cuda:6'), covar=tensor([0.0136, 0.0781, 0.0261, 0.0222, 0.0107, 0.0142, 0.0120, 0.0151], device='cuda:6'), in_proj_covar=tensor([0.0033, 0.0064, 0.0043, 0.0039, 0.0032, 0.0036, 0.0038, 0.0034], device='cuda:6'), out_proj_covar=tensor([4.8572e-05, 1.1137e-04, 7.3079e-05, 5.4978e-05, 4.6579e-05, 5.5494e-05, 5.2656e-05, 5.0658e-05], device='cuda:6') 2023-04-27 14:03:20,848 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8926, 4.5704, 4.8156, 5.0969, 4.0722, 4.8461, 4.4943, 4.6455], device='cuda:6'), covar=tensor([0.0281, 0.0133, 0.0196, 0.0100, 0.0765, 0.0231, 0.0139, 0.0161], device='cuda:6'), in_proj_covar=tensor([0.0081, 0.0060, 0.0108, 0.0081, 0.0131, 0.0086, 0.0074, 0.0085], device='cuda:6'), out_proj_covar=tensor([1.4111e-04, 9.8816e-05, 1.8413e-04, 1.2910e-04, 1.8749e-04, 1.5636e-04, 1.2796e-04, 1.4944e-04], device='cuda:6') 2023-04-27 14:03:47,165 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9649, 3.9634, 3.9003, 4.3803, 4.3190, 4.2359, 4.3123, 4.2026], device='cuda:6'), covar=tensor([0.0412, 0.0240, 0.0996, 0.0269, 0.0342, 0.0246, 0.0214, 0.0250], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0128, 0.0214, 0.0151, 0.0125, 0.0130, 0.0114, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 14:03:49,669 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-27 14:03:54,938 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4944.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:04:04,716 INFO [train.py:904] (6/8) Epoch 1, batch 4950, loss[loss=0.3089, simple_loss=0.3773, pruned_loss=0.1202, over 16705.00 frames. ], tot_loss[loss=0.3129, simple_loss=0.374, pruned_loss=0.1259, over 3204070.82 frames. ], batch size: 134, lr: 4.21e-02, grad_scale: 8.0 2023-04-27 14:05:04,813 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=4992.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:05:17,360 INFO [train.py:904] (6/8) Epoch 1, batch 5000, loss[loss=0.3288, simple_loss=0.3951, pruned_loss=0.1312, over 16477.00 frames. ], tot_loss[loss=0.3141, simple_loss=0.3758, pruned_loss=0.1262, over 3211352.77 frames. ], batch size: 146, lr: 4.20e-02, grad_scale: 8.0 2023-04-27 14:05:35,391 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.744e+02 4.820e+02 5.822e+02 7.344e+02 1.526e+03, threshold=1.164e+03, percent-clipped=12.0 2023-04-27 14:06:31,099 INFO [train.py:904] (6/8) Epoch 1, batch 5050, loss[loss=0.268, simple_loss=0.3418, pruned_loss=0.09715, over 16536.00 frames. ], tot_loss[loss=0.3131, simple_loss=0.3754, pruned_loss=0.1254, over 3209357.51 frames. ], batch size: 75, lr: 4.19e-02, grad_scale: 8.0 2023-04-27 14:06:35,570 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-04-27 14:07:42,590 INFO [train.py:904] (6/8) Epoch 1, batch 5100, loss[loss=0.2919, simple_loss=0.3576, pruned_loss=0.1131, over 16759.00 frames. ], tot_loss[loss=0.3089, simple_loss=0.3715, pruned_loss=0.1232, over 3214763.40 frames. ], batch size: 89, lr: 4.18e-02, grad_scale: 8.0 2023-04-27 14:07:59,829 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.617e+02 4.246e+02 5.535e+02 6.678e+02 1.197e+03, threshold=1.107e+03, percent-clipped=1.0 2023-04-27 14:08:58,115 INFO [train.py:904] (6/8) Epoch 1, batch 5150, loss[loss=0.3263, simple_loss=0.3942, pruned_loss=0.1292, over 15203.00 frames. ], tot_loss[loss=0.3085, simple_loss=0.3724, pruned_loss=0.1223, over 3213289.82 frames. ], batch size: 190, lr: 4.17e-02, grad_scale: 8.0 2023-04-27 14:10:12,903 INFO [train.py:904] (6/8) Epoch 1, batch 5200, loss[loss=0.2919, simple_loss=0.3533, pruned_loss=0.1153, over 16505.00 frames. ], tot_loss[loss=0.3086, simple_loss=0.3714, pruned_loss=0.1229, over 3215415.75 frames. ], batch size: 75, lr: 4.16e-02, grad_scale: 8.0 2023-04-27 14:10:19,702 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3103, 4.5937, 4.3646, 4.4104, 4.5940, 4.9969, 4.8840, 4.4530], device='cuda:6'), covar=tensor([0.0784, 0.1087, 0.0797, 0.1344, 0.1697, 0.0701, 0.0723, 0.1450], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0191, 0.0154, 0.0159, 0.0194, 0.0144, 0.0148, 0.0206], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:6') 2023-04-27 14:10:24,295 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7135, 2.0851, 1.9202, 3.0012, 3.1748, 3.2984, 3.1947, 3.1308], device='cuda:6'), covar=tensor([0.0124, 0.0760, 0.0524, 0.0127, 0.0113, 0.0185, 0.0129, 0.0180], device='cuda:6'), in_proj_covar=tensor([0.0036, 0.0070, 0.0051, 0.0043, 0.0034, 0.0037, 0.0042, 0.0037], device='cuda:6'), out_proj_covar=tensor([5.5917e-05, 1.2439e-04, 8.8826e-05, 6.3259e-05, 5.1657e-05, 5.8171e-05, 5.8605e-05, 5.6789e-05], device='cuda:6') 2023-04-27 14:10:30,246 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.582e+02 4.001e+02 4.794e+02 6.093e+02 1.086e+03, threshold=9.588e+02, percent-clipped=0.0 2023-04-27 14:11:26,080 INFO [train.py:904] (6/8) Epoch 1, batch 5250, loss[loss=0.2759, simple_loss=0.3412, pruned_loss=0.1053, over 16522.00 frames. ], tot_loss[loss=0.3077, simple_loss=0.3693, pruned_loss=0.1231, over 3208349.48 frames. ], batch size: 75, lr: 4.15e-02, grad_scale: 8.0 2023-04-27 14:12:13,933 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2165, 4.5043, 2.3454, 5.2188, 4.8768, 4.9566, 3.2494, 4.6618], device='cuda:6'), covar=tensor([0.2159, 0.0171, 0.1605, 0.0038, 0.0049, 0.0172, 0.0589, 0.0152], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0076, 0.0138, 0.0047, 0.0049, 0.0061, 0.0102, 0.0080], device='cuda:6'), out_proj_covar=tensor([1.8461e-04, 1.0010e-04, 1.6645e-04, 6.6107e-05, 7.2977e-05, 1.0088e-04, 1.3169e-04, 1.0561e-04], device='cuda:6') 2023-04-27 14:12:37,169 INFO [train.py:904] (6/8) Epoch 1, batch 5300, loss[loss=0.2481, simple_loss=0.3191, pruned_loss=0.08855, over 16679.00 frames. ], tot_loss[loss=0.3045, simple_loss=0.3655, pruned_loss=0.1218, over 3193323.13 frames. ], batch size: 134, lr: 4.14e-02, grad_scale: 8.0 2023-04-27 14:12:54,712 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.864e+02 4.537e+02 5.350e+02 6.217e+02 1.130e+03, threshold=1.070e+03, percent-clipped=3.0 2023-04-27 14:12:59,345 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3886, 5.3568, 5.2024, 5.1847, 5.2364, 5.6246, 5.4635, 5.1199], device='cuda:6'), covar=tensor([0.0760, 0.0825, 0.0662, 0.0994, 0.1520, 0.0594, 0.0612, 0.1336], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0189, 0.0150, 0.0156, 0.0195, 0.0145, 0.0145, 0.0204], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:6') 2023-04-27 14:13:49,534 INFO [train.py:904] (6/8) Epoch 1, batch 5350, loss[loss=0.3166, simple_loss=0.3782, pruned_loss=0.1274, over 16773.00 frames. ], tot_loss[loss=0.3011, simple_loss=0.3629, pruned_loss=0.1197, over 3212731.22 frames. ], batch size: 83, lr: 4.13e-02, grad_scale: 8.0 2023-04-27 14:14:05,038 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.30 vs. limit=5.0 2023-04-27 14:14:10,477 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.4502, 1.2334, 1.2875, 1.4152, 1.3770, 1.3430, 1.4306, 1.5999], device='cuda:6'), covar=tensor([0.0111, 0.0292, 0.0156, 0.0155, 0.0174, 0.0184, 0.0162, 0.0120], device='cuda:6'), in_proj_covar=tensor([0.0029, 0.0052, 0.0033, 0.0031, 0.0031, 0.0036, 0.0029, 0.0032], device='cuda:6'), out_proj_covar=tensor([3.1352e-05, 7.0623e-05, 3.8470e-05, 3.8311e-05, 3.3514e-05, 3.8369e-05, 3.4195e-05, 3.6678e-05], device='cuda:6') 2023-04-27 14:14:40,175 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8686, 1.3504, 1.4616, 1.6883, 1.4802, 1.6035, 2.0639, 1.9571], device='cuda:6'), covar=tensor([0.0118, 0.0522, 0.0197, 0.0263, 0.0164, 0.0268, 0.0156, 0.0270], device='cuda:6'), in_proj_covar=tensor([0.0029, 0.0052, 0.0033, 0.0031, 0.0032, 0.0036, 0.0029, 0.0032], device='cuda:6'), out_proj_covar=tensor([3.1174e-05, 7.0668e-05, 3.8648e-05, 3.8638e-05, 3.3952e-05, 3.8905e-05, 3.4292e-05, 3.6737e-05], device='cuda:6') 2023-04-27 14:15:00,987 INFO [train.py:904] (6/8) Epoch 1, batch 5400, loss[loss=0.3126, simple_loss=0.3773, pruned_loss=0.124, over 16794.00 frames. ], tot_loss[loss=0.3059, simple_loss=0.3672, pruned_loss=0.1223, over 3200206.47 frames. ], batch size: 102, lr: 4.12e-02, grad_scale: 8.0 2023-04-27 14:15:18,319 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.226e+02 4.427e+02 5.350e+02 6.399e+02 9.942e+02, threshold=1.070e+03, percent-clipped=0.0 2023-04-27 14:15:46,507 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1727, 1.3405, 1.3901, 1.8729, 2.2173, 1.9548, 2.4676, 2.2680], device='cuda:6'), covar=tensor([0.0098, 0.0591, 0.0209, 0.0241, 0.0125, 0.0222, 0.0127, 0.0223], device='cuda:6'), in_proj_covar=tensor([0.0027, 0.0050, 0.0033, 0.0031, 0.0030, 0.0035, 0.0028, 0.0030], device='cuda:6'), out_proj_covar=tensor([2.9689e-05, 6.8606e-05, 3.8002e-05, 3.8015e-05, 3.1798e-05, 3.8518e-05, 3.2996e-05, 3.5139e-05], device='cuda:6') 2023-04-27 14:16:19,544 INFO [train.py:904] (6/8) Epoch 1, batch 5450, loss[loss=0.3682, simple_loss=0.4207, pruned_loss=0.1578, over 16676.00 frames. ], tot_loss[loss=0.3117, simple_loss=0.3717, pruned_loss=0.1258, over 3190711.15 frames. ], batch size: 83, lr: 4.11e-02, grad_scale: 8.0 2023-04-27 14:17:08,708 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=2.04 vs. limit=2.0 2023-04-27 14:17:37,159 INFO [train.py:904] (6/8) Epoch 1, batch 5500, loss[loss=0.4772, simple_loss=0.4761, pruned_loss=0.2391, over 11680.00 frames. ], tot_loss[loss=0.3281, simple_loss=0.3835, pruned_loss=0.1363, over 3162037.98 frames. ], batch size: 246, lr: 4.10e-02, grad_scale: 8.0 2023-04-27 14:17:56,197 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.545e+02 5.862e+02 7.575e+02 9.403e+02 2.285e+03, threshold=1.515e+03, percent-clipped=16.0 2023-04-27 14:18:57,385 INFO [train.py:904] (6/8) Epoch 1, batch 5550, loss[loss=0.4889, simple_loss=0.4787, pruned_loss=0.2495, over 11473.00 frames. ], tot_loss[loss=0.3437, simple_loss=0.3946, pruned_loss=0.1464, over 3140512.91 frames. ], batch size: 253, lr: 4.09e-02, grad_scale: 8.0 2023-04-27 14:18:59,849 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-27 14:20:01,862 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6574, 2.4849, 2.4757, 2.0435, 2.4960, 2.4726, 2.5574, 2.2197], device='cuda:6'), covar=tensor([0.0819, 0.0120, 0.0120, 0.0263, 0.0103, 0.0112, 0.0121, 0.0237], device='cuda:6'), in_proj_covar=tensor([0.0086, 0.0035, 0.0037, 0.0051, 0.0035, 0.0035, 0.0041, 0.0048], device='cuda:6'), out_proj_covar=tensor([1.5204e-04, 6.6925e-05, 7.0765e-05, 9.1330e-05, 6.6642e-05, 7.1989e-05, 7.8086e-05, 8.7604e-05], device='cuda:6') 2023-04-27 14:20:17,844 INFO [train.py:904] (6/8) Epoch 1, batch 5600, loss[loss=0.4387, simple_loss=0.4599, pruned_loss=0.2087, over 15210.00 frames. ], tot_loss[loss=0.356, simple_loss=0.4025, pruned_loss=0.1548, over 3077043.66 frames. ], batch size: 190, lr: 4.08e-02, grad_scale: 8.0 2023-04-27 14:20:37,626 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.538e+02 5.637e+02 6.821e+02 8.546e+02 1.933e+03, threshold=1.364e+03, percent-clipped=3.0 2023-04-27 14:21:39,911 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5650.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 14:21:40,595 INFO [train.py:904] (6/8) Epoch 1, batch 5650, loss[loss=0.488, simple_loss=0.4775, pruned_loss=0.2492, over 11202.00 frames. ], tot_loss[loss=0.3671, simple_loss=0.4106, pruned_loss=0.1618, over 3069138.71 frames. ], batch size: 247, lr: 4.07e-02, grad_scale: 8.0 2023-04-27 14:22:59,471 INFO [train.py:904] (6/8) Epoch 1, batch 5700, loss[loss=0.3835, simple_loss=0.4412, pruned_loss=0.163, over 16892.00 frames. ], tot_loss[loss=0.3686, simple_loss=0.4114, pruned_loss=0.1628, over 3061254.59 frames. ], batch size: 90, lr: 4.06e-02, grad_scale: 8.0 2023-04-27 14:23:15,449 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5711.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 14:23:17,957 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.964e+02 6.202e+02 7.607e+02 9.700e+02 2.079e+03, threshold=1.521e+03, percent-clipped=5.0 2023-04-27 14:23:22,550 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.66 vs. limit=5.0 2023-04-27 14:23:32,491 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9714, 3.5921, 2.2802, 3.8977, 3.9263, 3.9809, 2.5679, 3.6670], device='cuda:6'), covar=tensor([0.2427, 0.0189, 0.1708, 0.0080, 0.0108, 0.0225, 0.0746, 0.0197], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0077, 0.0145, 0.0049, 0.0049, 0.0063, 0.0108, 0.0084], device='cuda:6'), out_proj_covar=tensor([1.9558e-04, 1.0466e-04, 1.7987e-04, 7.5477e-05, 7.7856e-05, 1.0859e-04, 1.4614e-04, 1.1388e-04], device='cuda:6') 2023-04-27 14:24:21,182 INFO [train.py:904] (6/8) Epoch 1, batch 5750, loss[loss=0.4009, simple_loss=0.4218, pruned_loss=0.19, over 10964.00 frames. ], tot_loss[loss=0.3729, simple_loss=0.4145, pruned_loss=0.1657, over 3005292.73 frames. ], batch size: 248, lr: 4.05e-02, grad_scale: 8.0 2023-04-27 14:24:28,946 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5755.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:25:42,933 INFO [train.py:904] (6/8) Epoch 1, batch 5800, loss[loss=0.4178, simple_loss=0.4356, pruned_loss=0.2, over 12028.00 frames. ], tot_loss[loss=0.3699, simple_loss=0.4136, pruned_loss=0.1631, over 3027607.53 frames. ], batch size: 247, lr: 4.04e-02, grad_scale: 8.0 2023-04-27 14:26:01,838 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.312e+02 5.567e+02 6.621e+02 8.933e+02 1.804e+03, threshold=1.324e+03, percent-clipped=2.0 2023-04-27 14:26:06,955 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5816.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:26:27,901 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-27 14:27:02,345 INFO [train.py:904] (6/8) Epoch 1, batch 5850, loss[loss=0.358, simple_loss=0.4105, pruned_loss=0.1528, over 16634.00 frames. ], tot_loss[loss=0.3642, simple_loss=0.4101, pruned_loss=0.1591, over 3057333.29 frames. ], batch size: 62, lr: 4.03e-02, grad_scale: 8.0 2023-04-27 14:27:09,988 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6662, 1.5107, 1.5324, 1.5696, 1.7472, 1.8787, 1.7976, 1.7894], device='cuda:6'), covar=tensor([0.0113, 0.0595, 0.0243, 0.0197, 0.0127, 0.0148, 0.0156, 0.0152], device='cuda:6'), in_proj_covar=tensor([0.0036, 0.0082, 0.0059, 0.0044, 0.0038, 0.0039, 0.0046, 0.0038], device='cuda:6'), out_proj_covar=tensor([5.8078e-05, 1.4669e-04, 1.0522e-04, 7.2110e-05, 6.1384e-05, 6.3162e-05, 6.8929e-05, 6.3516e-05], device='cuda:6') 2023-04-27 14:27:15,071 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5858.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:27:42,725 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.46 vs. limit=5.0 2023-04-27 14:28:17,481 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6198, 3.4892, 3.2710, 2.4584, 3.4030, 3.4996, 3.6408, 2.2383], device='cuda:6'), covar=tensor([0.1177, 0.0087, 0.0127, 0.0375, 0.0096, 0.0090, 0.0078, 0.0422], device='cuda:6'), in_proj_covar=tensor([0.0090, 0.0037, 0.0039, 0.0057, 0.0035, 0.0036, 0.0042, 0.0053], device='cuda:6'), out_proj_covar=tensor([1.6283e-04, 7.3338e-05, 7.6629e-05, 1.0462e-04, 6.8801e-05, 7.7128e-05, 7.9998e-05, 1.0041e-04], device='cuda:6') 2023-04-27 14:28:25,812 INFO [train.py:904] (6/8) Epoch 1, batch 5900, loss[loss=0.4005, simple_loss=0.4126, pruned_loss=0.1942, over 11023.00 frames. ], tot_loss[loss=0.3607, simple_loss=0.4079, pruned_loss=0.1568, over 3052552.44 frames. ], batch size: 248, lr: 4.02e-02, grad_scale: 8.0 2023-04-27 14:28:48,065 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.490e+02 5.425e+02 6.564e+02 8.517e+02 1.690e+03, threshold=1.313e+03, percent-clipped=2.0 2023-04-27 14:28:58,975 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5919.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:29:49,226 INFO [train.py:904] (6/8) Epoch 1, batch 5950, loss[loss=0.3557, simple_loss=0.4071, pruned_loss=0.1522, over 16352.00 frames. ], tot_loss[loss=0.3578, simple_loss=0.4071, pruned_loss=0.1542, over 3053184.39 frames. ], batch size: 146, lr: 4.01e-02, grad_scale: 8.0 2023-04-27 14:30:05,315 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4501, 2.3486, 1.6357, 2.5840, 2.0887, 2.4106, 1.9202, 2.2025], device='cuda:6'), covar=tensor([0.0131, 0.0243, 0.1255, 0.0123, 0.0585, 0.0221, 0.0964, 0.0474], device='cuda:6'), in_proj_covar=tensor([0.0052, 0.0052, 0.0109, 0.0055, 0.0091, 0.0051, 0.0121, 0.0083], device='cuda:6'), out_proj_covar=tensor([7.4716e-05, 8.2416e-05, 1.5898e-04, 7.6270e-05, 1.2959e-04, 8.5811e-05, 1.7677e-04, 1.3137e-04], device='cuda:6') 2023-04-27 14:30:23,619 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-04-27 14:30:52,720 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5990.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:31:14,068 INFO [train.py:904] (6/8) Epoch 1, batch 6000, loss[loss=0.3374, simple_loss=0.3901, pruned_loss=0.1424, over 16925.00 frames. ], tot_loss[loss=0.3586, simple_loss=0.4073, pruned_loss=0.155, over 3058983.47 frames. ], batch size: 109, lr: 4.00e-02, grad_scale: 8.0 2023-04-27 14:31:14,068 INFO [train.py:929] (6/8) Computing validation loss 2023-04-27 14:31:23,948 INFO [train.py:938] (6/8) Epoch 1, validation: loss=0.2762, simple_loss=0.3752, pruned_loss=0.08863, over 944034.00 frames. 2023-04-27 14:31:23,949 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17377MB 2023-04-27 14:31:31,562 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6006.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 14:31:41,451 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.294e+02 5.611e+02 7.114e+02 9.006e+02 1.900e+03, threshold=1.423e+03, percent-clipped=2.0 2023-04-27 14:32:21,193 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9013, 3.5000, 2.1723, 4.0133, 3.9414, 3.9001, 2.6090, 3.7102], device='cuda:6'), covar=tensor([0.2296, 0.0247, 0.1748, 0.0084, 0.0119, 0.0242, 0.0734, 0.0233], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0082, 0.0152, 0.0051, 0.0053, 0.0066, 0.0114, 0.0090], device='cuda:6'), out_proj_covar=tensor([2.0360e-04, 1.1342e-04, 1.9159e-04, 8.1949e-05, 8.5848e-05, 1.1657e-04, 1.5595e-04, 1.2676e-04], device='cuda:6') 2023-04-27 14:32:30,070 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6043.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:32:42,910 INFO [train.py:904] (6/8) Epoch 1, batch 6050, loss[loss=0.2908, simple_loss=0.3746, pruned_loss=0.1035, over 16758.00 frames. ], tot_loss[loss=0.3567, simple_loss=0.4052, pruned_loss=0.1541, over 3072588.29 frames. ], batch size: 83, lr: 3.99e-02, grad_scale: 8.0 2023-04-27 14:32:44,849 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6051.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:34:03,525 INFO [train.py:904] (6/8) Epoch 1, batch 6100, loss[loss=0.3408, simple_loss=0.397, pruned_loss=0.1423, over 16671.00 frames. ], tot_loss[loss=0.3523, simple_loss=0.4029, pruned_loss=0.1509, over 3093380.62 frames. ], batch size: 62, lr: 3.98e-02, grad_scale: 8.0 2023-04-27 14:34:09,289 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6104.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:34:21,421 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6111.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:34:24,016 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.820e+02 4.604e+02 6.189e+02 8.357e+02 1.892e+03, threshold=1.238e+03, percent-clipped=2.0 2023-04-27 14:34:59,813 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6135.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:35:26,097 INFO [train.py:904] (6/8) Epoch 1, batch 6150, loss[loss=0.3162, simple_loss=0.3746, pruned_loss=0.1289, over 16597.00 frames. ], tot_loss[loss=0.3501, simple_loss=0.4005, pruned_loss=0.1499, over 3085230.04 frames. ], batch size: 57, lr: 3.97e-02, grad_scale: 8.0 2023-04-27 14:35:57,263 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6170.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:36:39,083 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6196.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:36:46,262 INFO [train.py:904] (6/8) Epoch 1, batch 6200, loss[loss=0.3249, simple_loss=0.3753, pruned_loss=0.1372, over 16692.00 frames. ], tot_loss[loss=0.3492, simple_loss=0.3991, pruned_loss=0.1496, over 3089993.02 frames. ], batch size: 57, lr: 3.96e-02, grad_scale: 8.0 2023-04-27 14:36:49,316 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6203.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:36:51,423 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4635, 2.5905, 1.6018, 2.6161, 1.9929, 2.5299, 1.8685, 2.3036], device='cuda:6'), covar=tensor([0.0147, 0.0141, 0.1221, 0.0107, 0.0563, 0.0262, 0.1046, 0.0481], device='cuda:6'), in_proj_covar=tensor([0.0054, 0.0054, 0.0116, 0.0059, 0.0098, 0.0053, 0.0126, 0.0088], device='cuda:6'), out_proj_covar=tensor([8.0895e-05, 8.8921e-05, 1.7128e-04, 8.4057e-05, 1.4228e-04, 9.3389e-05, 1.8541e-04, 1.4156e-04], device='cuda:6') 2023-04-27 14:37:03,883 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6211.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:37:05,931 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.990e+02 5.160e+02 6.786e+02 8.603e+02 1.824e+03, threshold=1.357e+03, percent-clipped=7.0 2023-04-27 14:37:08,080 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6214.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:37:23,201 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-27 14:37:33,355 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6231.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:38:02,369 INFO [train.py:904] (6/8) Epoch 1, batch 6250, loss[loss=0.3766, simple_loss=0.4249, pruned_loss=0.1642, over 15350.00 frames. ], tot_loss[loss=0.3461, simple_loss=0.3974, pruned_loss=0.1474, over 3108211.71 frames. ], batch size: 190, lr: 3.95e-02, grad_scale: 8.0 2023-04-27 14:38:20,954 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6264.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:38:24,230 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6266.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:38:32,048 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6272.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:38:53,870 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.09 vs. limit=5.0 2023-04-27 14:39:17,017 INFO [train.py:904] (6/8) Epoch 1, batch 6300, loss[loss=0.3363, simple_loss=0.3883, pruned_loss=0.1421, over 16694.00 frames. ], tot_loss[loss=0.3437, simple_loss=0.3965, pruned_loss=0.1454, over 3138558.41 frames. ], batch size: 76, lr: 3.94e-02, grad_scale: 8.0 2023-04-27 14:39:25,730 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6306.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 14:39:36,719 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.400e+02 5.487e+02 6.666e+02 8.343e+02 1.856e+03, threshold=1.333e+03, percent-clipped=2.0 2023-04-27 14:39:59,682 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6327.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:40:28,632 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6346.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:40:36,333 INFO [train.py:904] (6/8) Epoch 1, batch 6350, loss[loss=0.2897, simple_loss=0.3559, pruned_loss=0.1118, over 17145.00 frames. ], tot_loss[loss=0.3467, simple_loss=0.3978, pruned_loss=0.1478, over 3134781.26 frames. ], batch size: 48, lr: 3.93e-02, grad_scale: 8.0 2023-04-27 14:40:41,032 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6354.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 14:41:34,009 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-27 14:41:39,627 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5528, 3.4979, 3.0108, 3.1885, 2.6255, 2.0989, 3.6719, 4.1026], device='cuda:6'), covar=tensor([0.1610, 0.0507, 0.0806, 0.0361, 0.1707, 0.1296, 0.0240, 0.0056], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0135, 0.0170, 0.0101, 0.0178, 0.0138, 0.0106, 0.0060], device='cuda:6'), out_proj_covar=tensor([2.1882e-04, 1.6658e-04, 1.8480e-04, 1.1790e-04, 2.1706e-04, 1.6115e-04, 1.2807e-04, 7.4623e-05], device='cuda:6') 2023-04-27 14:41:50,607 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6399.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:41:53,485 INFO [train.py:904] (6/8) Epoch 1, batch 6400, loss[loss=0.4449, simple_loss=0.4555, pruned_loss=0.2171, over 11131.00 frames. ], tot_loss[loss=0.348, simple_loss=0.3981, pruned_loss=0.1489, over 3135478.93 frames. ], batch size: 248, lr: 3.92e-02, grad_scale: 8.0 2023-04-27 14:42:08,525 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6411.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:42:10,410 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 4.039e+02 5.999e+02 7.400e+02 9.043e+02 1.587e+03, threshold=1.480e+03, percent-clipped=2.0 2023-04-27 14:42:28,687 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1049, 4.6134, 4.5417, 3.6823, 4.8323, 3.8750, 4.4534, 5.0254], device='cuda:6'), covar=tensor([0.0097, 0.0120, 0.0094, 0.0532, 0.0064, 0.0426, 0.0100, 0.0115], device='cuda:6'), in_proj_covar=tensor([0.0047, 0.0036, 0.0051, 0.0073, 0.0038, 0.0065, 0.0051, 0.0050], device='cuda:6'), out_proj_covar=tensor([1.1443e-04, 8.5936e-05, 1.3046e-04, 1.6081e-04, 8.8111e-05, 1.4818e-04, 1.3142e-04, 1.3407e-04], device='cuda:6') 2023-04-27 14:42:33,362 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0435, 3.9927, 4.2870, 4.1811, 4.4331, 4.0998, 3.9686, 4.2113], device='cuda:6'), covar=tensor([0.0361, 0.0363, 0.0631, 0.0709, 0.0506, 0.0396, 0.0845, 0.0367], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0109, 0.0134, 0.0132, 0.0144, 0.0116, 0.0152, 0.0111], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 14:42:41,815 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0657, 2.2987, 2.1896, 3.2469, 3.4265, 3.2899, 2.2065, 3.2634], device='cuda:6'), covar=tensor([0.0081, 0.0659, 0.0483, 0.0148, 0.0125, 0.0198, 0.0262, 0.0185], device='cuda:6'), in_proj_covar=tensor([0.0036, 0.0082, 0.0059, 0.0043, 0.0039, 0.0038, 0.0047, 0.0036], device='cuda:6'), out_proj_covar=tensor([6.2741e-05, 1.4458e-04, 1.0909e-04, 7.5092e-05, 6.7094e-05, 6.6657e-05, 7.5616e-05, 6.3344e-05], device='cuda:6') 2023-04-27 14:43:09,618 INFO [train.py:904] (6/8) Epoch 1, batch 6450, loss[loss=0.3119, simple_loss=0.3644, pruned_loss=0.1297, over 15466.00 frames. ], tot_loss[loss=0.3454, simple_loss=0.3959, pruned_loss=0.1474, over 3111083.36 frames. ], batch size: 191, lr: 3.91e-02, grad_scale: 8.0 2023-04-27 14:43:22,744 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6459.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:43:26,589 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6462.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:43:33,791 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6467.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:44:12,818 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6491.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:44:26,971 INFO [train.py:904] (6/8) Epoch 1, batch 6500, loss[loss=0.3444, simple_loss=0.3936, pruned_loss=0.1476, over 16587.00 frames. ], tot_loss[loss=0.3415, simple_loss=0.3924, pruned_loss=0.1453, over 3111641.63 frames. ], batch size: 62, lr: 3.90e-02, grad_scale: 16.0 2023-04-27 14:44:36,392 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3839, 4.5566, 4.3098, 4.4894, 3.9430, 4.4317, 4.2200, 4.5347], device='cuda:6'), covar=tensor([0.0339, 0.0589, 0.0647, 0.0285, 0.0615, 0.0367, 0.0465, 0.0392], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0189, 0.0181, 0.0122, 0.0156, 0.0126, 0.0170, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-27 14:44:45,180 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.861e+02 4.801e+02 5.896e+02 8.038e+02 1.836e+03, threshold=1.179e+03, percent-clipped=2.0 2023-04-27 14:44:46,852 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6514.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:45:00,382 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6523.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 14:45:05,060 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6526.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:45:07,740 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6528.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:45:42,981 INFO [train.py:904] (6/8) Epoch 1, batch 6550, loss[loss=0.3308, simple_loss=0.4037, pruned_loss=0.129, over 16717.00 frames. ], tot_loss[loss=0.344, simple_loss=0.3955, pruned_loss=0.1462, over 3117822.99 frames. ], batch size: 134, lr: 3.89e-02, grad_scale: 16.0 2023-04-27 14:45:55,700 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6559.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:46:00,535 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6562.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:46:08,846 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6567.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:46:59,693 INFO [train.py:904] (6/8) Epoch 1, batch 6600, loss[loss=0.3285, simple_loss=0.3884, pruned_loss=0.1343, over 16879.00 frames. ], tot_loss[loss=0.3459, simple_loss=0.3983, pruned_loss=0.1468, over 3122142.61 frames. ], batch size: 96, lr: 3.89e-02, grad_scale: 16.0 2023-04-27 14:47:18,178 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.822e+02 5.266e+02 6.465e+02 8.044e+02 1.550e+03, threshold=1.293e+03, percent-clipped=3.0 2023-04-27 14:47:32,879 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6622.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:47:43,308 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6629.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:48:10,813 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6646.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:48:18,077 INFO [train.py:904] (6/8) Epoch 1, batch 6650, loss[loss=0.2906, simple_loss=0.357, pruned_loss=0.1121, over 16870.00 frames. ], tot_loss[loss=0.345, simple_loss=0.3973, pruned_loss=0.1464, over 3126242.50 frames. ], batch size: 42, lr: 3.88e-02, grad_scale: 16.0 2023-04-27 14:48:21,719 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8427, 4.0129, 3.2549, 4.1211, 3.1837, 2.2415, 4.2250, 4.8627], device='cuda:6'), covar=tensor([0.1987, 0.0520, 0.1062, 0.0229, 0.1702, 0.1512, 0.0192, 0.0042], device='cuda:6'), in_proj_covar=tensor([0.0191, 0.0148, 0.0182, 0.0107, 0.0190, 0.0145, 0.0112, 0.0063], device='cuda:6'), out_proj_covar=tensor([2.3204e-04, 1.8123e-04, 1.9896e-04, 1.2583e-04, 2.3186e-04, 1.7139e-04, 1.3576e-04, 8.0733e-05], device='cuda:6') 2023-04-27 14:48:31,223 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-27 14:49:18,695 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6690.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:49:24,164 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6694.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:49:24,604 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.07 vs. limit=2.0 2023-04-27 14:49:32,245 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6699.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:49:34,324 INFO [train.py:904] (6/8) Epoch 1, batch 6700, loss[loss=0.3157, simple_loss=0.3711, pruned_loss=0.1302, over 16720.00 frames. ], tot_loss[loss=0.3451, simple_loss=0.3964, pruned_loss=0.1469, over 3108605.02 frames. ], batch size: 62, lr: 3.87e-02, grad_scale: 16.0 2023-04-27 14:49:50,093 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9609, 3.8147, 3.9054, 4.2912, 4.2529, 4.1455, 4.2644, 4.1099], device='cuda:6'), covar=tensor([0.0367, 0.0350, 0.0873, 0.0288, 0.0369, 0.0331, 0.0289, 0.0291], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0160, 0.0250, 0.0178, 0.0149, 0.0163, 0.0134, 0.0149], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 14:49:52,618 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.128e+02 5.771e+02 7.025e+02 8.722e+02 1.711e+03, threshold=1.405e+03, percent-clipped=7.0 2023-04-27 14:50:00,206 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7885, 3.7865, 3.0668, 3.7718, 3.0012, 2.1369, 4.0541, 4.6946], device='cuda:6'), covar=tensor([0.1870, 0.0522, 0.1068, 0.0272, 0.1818, 0.1503, 0.0224, 0.0038], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0149, 0.0185, 0.0108, 0.0192, 0.0147, 0.0115, 0.0063], device='cuda:6'), out_proj_covar=tensor([2.3347e-04, 1.8312e-04, 2.0212e-04, 1.2625e-04, 2.3374e-04, 1.7278e-04, 1.3903e-04, 8.1503e-05], device='cuda:6') 2023-04-27 14:50:45,228 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6747.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:50:50,989 INFO [train.py:904] (6/8) Epoch 1, batch 6750, loss[loss=0.2994, simple_loss=0.3514, pruned_loss=0.1237, over 16698.00 frames. ], tot_loss[loss=0.3418, simple_loss=0.3936, pruned_loss=0.145, over 3140243.04 frames. ], batch size: 57, lr: 3.86e-02, grad_scale: 16.0 2023-04-27 14:51:20,984 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2329, 1.6100, 1.7580, 2.5264, 2.6763, 2.4500, 1.9365, 2.2780], device='cuda:6'), covar=tensor([0.0077, 0.0585, 0.0282, 0.0106, 0.0091, 0.0168, 0.0186, 0.0144], device='cuda:6'), in_proj_covar=tensor([0.0037, 0.0087, 0.0065, 0.0044, 0.0040, 0.0040, 0.0052, 0.0038], device='cuda:6'), out_proj_covar=tensor([6.5956e-05, 1.5264e-04, 1.2053e-04, 7.9822e-05, 7.0093e-05, 7.0648e-05, 8.4705e-05, 6.8230e-05], device='cuda:6') 2023-04-27 14:51:51,467 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6791.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:52:05,969 INFO [train.py:904] (6/8) Epoch 1, batch 6800, loss[loss=0.3426, simple_loss=0.4114, pruned_loss=0.1369, over 16863.00 frames. ], tot_loss[loss=0.3415, simple_loss=0.3938, pruned_loss=0.1446, over 3137923.17 frames. ], batch size: 96, lr: 3.85e-02, grad_scale: 16.0 2023-04-27 14:52:24,953 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.031e+02 5.369e+02 6.197e+02 7.365e+02 1.417e+03, threshold=1.239e+03, percent-clipped=1.0 2023-04-27 14:52:33,900 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6818.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 14:52:41,648 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6823.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:52:46,958 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6826.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:53:06,062 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6839.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:53:12,991 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1144, 4.2875, 4.0180, 1.5064, 3.0672, 2.3398, 3.5204, 4.2313], device='cuda:6'), covar=tensor([0.0210, 0.0206, 0.0240, 0.2482, 0.0935, 0.1343, 0.0683, 0.0139], device='cuda:6'), in_proj_covar=tensor([0.0094, 0.0069, 0.0105, 0.0152, 0.0144, 0.0137, 0.0134, 0.0065], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-27 14:53:15,923 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.57 vs. limit=5.0 2023-04-27 14:53:23,170 INFO [train.py:904] (6/8) Epoch 1, batch 6850, loss[loss=0.2916, simple_loss=0.377, pruned_loss=0.1031, over 16695.00 frames. ], tot_loss[loss=0.343, simple_loss=0.3957, pruned_loss=0.1452, over 3137624.65 frames. ], batch size: 83, lr: 3.84e-02, grad_scale: 16.0 2023-04-27 14:53:35,298 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6859.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:53:46,413 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6867.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:53:57,789 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6874.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:54:03,455 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6878.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 14:54:37,393 INFO [train.py:904] (6/8) Epoch 1, batch 6900, loss[loss=0.3501, simple_loss=0.4085, pruned_loss=0.1458, over 16699.00 frames. ], tot_loss[loss=0.343, simple_loss=0.3984, pruned_loss=0.1438, over 3152315.94 frames. ], batch size: 89, lr: 3.83e-02, grad_scale: 16.0 2023-04-27 14:54:47,071 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6907.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:54:50,406 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6909.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:54:50,697 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-27 14:54:55,474 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.157e+02 5.195e+02 6.018e+02 7.768e+02 1.319e+03, threshold=1.204e+03, percent-clipped=1.0 2023-04-27 14:54:59,969 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6915.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:55:10,005 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6922.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:55:36,581 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6939.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 14:55:54,245 INFO [train.py:904] (6/8) Epoch 1, batch 6950, loss[loss=0.4069, simple_loss=0.4196, pruned_loss=0.1971, over 11169.00 frames. ], tot_loss[loss=0.3483, simple_loss=0.4013, pruned_loss=0.1477, over 3123455.74 frames. ], batch size: 250, lr: 3.82e-02, grad_scale: 16.0 2023-04-27 14:56:25,069 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6970.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:56:25,260 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6970.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:56:47,759 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6985.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:57:12,232 INFO [train.py:904] (6/8) Epoch 1, batch 7000, loss[loss=0.3726, simple_loss=0.4208, pruned_loss=0.1622, over 15407.00 frames. ], tot_loss[loss=0.3469, simple_loss=0.401, pruned_loss=0.1464, over 3114115.04 frames. ], batch size: 190, lr: 3.81e-02, grad_scale: 16.0 2023-04-27 14:57:16,503 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9744, 4.1293, 3.7350, 1.6100, 2.7271, 2.0632, 3.2877, 4.0700], device='cuda:6'), covar=tensor([0.0263, 0.0158, 0.0236, 0.2362, 0.1046, 0.1496, 0.0924, 0.0158], device='cuda:6'), in_proj_covar=tensor([0.0093, 0.0069, 0.0103, 0.0148, 0.0143, 0.0135, 0.0132, 0.0066], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-27 14:57:30,859 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.210e+02 5.287e+02 6.266e+02 8.014e+02 1.368e+03, threshold=1.253e+03, percent-clipped=3.0 2023-04-27 14:58:03,480 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.00 vs. limit=5.0 2023-04-27 14:58:27,032 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7048.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 14:58:31,415 INFO [train.py:904] (6/8) Epoch 1, batch 7050, loss[loss=0.4239, simple_loss=0.435, pruned_loss=0.2064, over 11248.00 frames. ], tot_loss[loss=0.3464, simple_loss=0.4011, pruned_loss=0.1458, over 3122978.72 frames. ], batch size: 247, lr: 3.80e-02, grad_scale: 16.0 2023-04-27 14:59:26,899 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=2.06 vs. limit=2.0 2023-04-27 14:59:36,796 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8551, 3.5854, 3.1502, 3.2243, 2.6940, 2.0398, 3.6639, 4.3154], device='cuda:6'), covar=tensor([0.1627, 0.0581, 0.0938, 0.0400, 0.1973, 0.1345, 0.0299, 0.0069], device='cuda:6'), in_proj_covar=tensor([0.0200, 0.0153, 0.0186, 0.0110, 0.0200, 0.0147, 0.0121, 0.0067], device='cuda:6'), out_proj_covar=tensor([2.4101e-04, 1.8746e-04, 2.0514e-04, 1.3072e-04, 2.4211e-04, 1.7388e-04, 1.4545e-04, 8.5912e-05], device='cuda:6') 2023-04-27 14:59:49,740 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4506, 5.1355, 5.0289, 5.1169, 5.1607, 5.4749, 5.3541, 4.9659], device='cuda:6'), covar=tensor([0.0725, 0.1020, 0.0752, 0.1255, 0.1528, 0.0727, 0.0703, 0.1687], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0195, 0.0164, 0.0171, 0.0206, 0.0163, 0.0152, 0.0226], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-04-27 14:59:51,865 INFO [train.py:904] (6/8) Epoch 1, batch 7100, loss[loss=0.3024, simple_loss=0.3638, pruned_loss=0.1205, over 17022.00 frames. ], tot_loss[loss=0.3463, simple_loss=0.4, pruned_loss=0.1463, over 3103145.23 frames. ], batch size: 50, lr: 3.79e-02, grad_scale: 16.0 2023-04-27 15:00:05,800 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7109.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 15:00:11,239 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.812e+02 5.623e+02 6.703e+02 8.193e+02 2.007e+03, threshold=1.341e+03, percent-clipped=3.0 2023-04-27 15:00:20,537 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7118.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 15:00:27,650 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7123.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:01:01,088 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5560, 1.5225, 1.4916, 1.6224, 1.7054, 1.6865, 1.7001, 1.7395], device='cuda:6'), covar=tensor([0.0117, 0.0411, 0.0211, 0.0185, 0.0118, 0.0144, 0.0179, 0.0096], device='cuda:6'), in_proj_covar=tensor([0.0040, 0.0087, 0.0066, 0.0047, 0.0041, 0.0040, 0.0054, 0.0037], device='cuda:6'), out_proj_covar=tensor([7.3605e-05, 1.5626e-04, 1.2494e-04, 8.7360e-05, 7.1179e-05, 7.3535e-05, 8.9720e-05, 6.7827e-05], device='cuda:6') 2023-04-27 15:01:11,086 INFO [train.py:904] (6/8) Epoch 1, batch 7150, loss[loss=0.3314, simple_loss=0.3922, pruned_loss=0.1353, over 16220.00 frames. ], tot_loss[loss=0.3429, simple_loss=0.3968, pruned_loss=0.1445, over 3129008.18 frames. ], batch size: 165, lr: 3.78e-02, grad_scale: 8.0 2023-04-27 15:01:30,107 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7163.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:01:34,860 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7166.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:01:42,493 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7171.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:02:19,840 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5876, 1.3654, 1.4668, 1.1919, 1.5885, 1.5117, 1.5975, 1.6085], device='cuda:6'), covar=tensor([0.0064, 0.0384, 0.0151, 0.0228, 0.0130, 0.0204, 0.0118, 0.0129], device='cuda:6'), in_proj_covar=tensor([0.0031, 0.0068, 0.0046, 0.0050, 0.0040, 0.0049, 0.0034, 0.0036], device='cuda:6'), out_proj_covar=tensor([3.8454e-05, 9.9704e-05, 6.3718e-05, 7.0210e-05, 5.5570e-05, 6.7693e-05, 5.0232e-05, 5.1226e-05], device='cuda:6') 2023-04-27 15:02:27,356 INFO [train.py:904] (6/8) Epoch 1, batch 7200, loss[loss=0.2901, simple_loss=0.3553, pruned_loss=0.1125, over 16654.00 frames. ], tot_loss[loss=0.3393, simple_loss=0.3944, pruned_loss=0.1421, over 3118944.26 frames. ], batch size: 134, lr: 3.78e-02, grad_scale: 8.0 2023-04-27 15:02:46,730 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.108e+02 4.516e+02 5.501e+02 7.177e+02 1.508e+03, threshold=1.100e+03, percent-clipped=3.0 2023-04-27 15:03:02,602 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7224.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 15:03:19,572 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7234.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 15:03:25,730 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8197, 4.4116, 2.2179, 5.0212, 4.9308, 4.5770, 2.6857, 4.4378], device='cuda:6'), covar=tensor([0.2333, 0.0209, 0.1923, 0.0067, 0.0072, 0.0166, 0.0922, 0.0244], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0086, 0.0155, 0.0056, 0.0060, 0.0072, 0.0124, 0.0103], device='cuda:6'), out_proj_covar=tensor([2.1654e-04, 1.3126e-04, 2.1120e-04, 9.4481e-05, 1.0339e-04, 1.3502e-04, 1.8084e-04, 1.5437e-04], device='cuda:6') 2023-04-27 15:03:43,883 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7040, 3.7866, 3.4230, 1.7482, 2.8152, 2.0901, 3.3127, 3.9911], device='cuda:6'), covar=tensor([0.0362, 0.0397, 0.0299, 0.2164, 0.1005, 0.1434, 0.0917, 0.0197], device='cuda:6'), in_proj_covar=tensor([0.0097, 0.0073, 0.0110, 0.0152, 0.0146, 0.0140, 0.0135, 0.0067], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-27 15:03:46,512 INFO [train.py:904] (6/8) Epoch 1, batch 7250, loss[loss=0.3515, simple_loss=0.3846, pruned_loss=0.1592, over 11312.00 frames. ], tot_loss[loss=0.3363, simple_loss=0.3916, pruned_loss=0.1406, over 3106814.45 frames. ], batch size: 246, lr: 3.77e-02, grad_scale: 8.0 2023-04-27 15:04:06,739 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7265.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:04:21,352 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7275.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:04:35,931 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7285.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:04:58,721 INFO [train.py:904] (6/8) Epoch 1, batch 7300, loss[loss=0.3238, simple_loss=0.3858, pruned_loss=0.1309, over 16658.00 frames. ], tot_loss[loss=0.3346, simple_loss=0.39, pruned_loss=0.1397, over 3113489.01 frames. ], batch size: 76, lr: 3.76e-02, grad_scale: 8.0 2023-04-27 15:05:19,382 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.541e+02 5.301e+02 6.672e+02 8.069e+02 1.507e+03, threshold=1.334e+03, percent-clipped=6.0 2023-04-27 15:05:49,057 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7333.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:05:49,296 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6452, 1.6627, 1.8624, 1.3143, 2.5825, 2.5161, 2.8394, 2.7233], device='cuda:6'), covar=tensor([0.0044, 0.0484, 0.0214, 0.0292, 0.0092, 0.0172, 0.0112, 0.0095], device='cuda:6'), in_proj_covar=tensor([0.0029, 0.0065, 0.0044, 0.0047, 0.0040, 0.0046, 0.0031, 0.0035], device='cuda:6'), out_proj_covar=tensor([3.5833e-05, 9.6573e-05, 6.1795e-05, 6.5829e-05, 5.4763e-05, 6.4074e-05, 4.6712e-05, 5.0078e-05], device='cuda:6') 2023-04-27 15:05:53,663 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7336.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:06:15,327 INFO [train.py:904] (6/8) Epoch 1, batch 7350, loss[loss=0.3012, simple_loss=0.3681, pruned_loss=0.1172, over 16898.00 frames. ], tot_loss[loss=0.3337, simple_loss=0.3889, pruned_loss=0.1393, over 3072589.35 frames. ], batch size: 96, lr: 3.75e-02, grad_scale: 8.0 2023-04-27 15:06:51,027 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.94 vs. limit=5.0 2023-04-27 15:07:30,952 INFO [train.py:904] (6/8) Epoch 1, batch 7400, loss[loss=0.3134, simple_loss=0.3767, pruned_loss=0.125, over 16869.00 frames. ], tot_loss[loss=0.3362, simple_loss=0.3906, pruned_loss=0.1409, over 3081657.50 frames. ], batch size: 109, lr: 3.74e-02, grad_scale: 8.0 2023-04-27 15:07:36,109 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7404.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 15:07:40,533 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7407.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:07:50,857 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.431e+02 4.999e+02 6.259e+02 7.546e+02 1.554e+03, threshold=1.252e+03, percent-clipped=1.0 2023-04-27 15:07:56,523 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2023-04-27 15:08:34,333 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9363, 2.5797, 2.3104, 3.1681, 2.4252, 2.8783, 2.5644, 2.3074], device='cuda:6'), covar=tensor([0.0307, 0.0303, 0.0261, 0.0245, 0.0800, 0.0226, 0.0484, 0.0932], device='cuda:6'), in_proj_covar=tensor([0.0107, 0.0106, 0.0090, 0.0120, 0.0189, 0.0104, 0.0129, 0.0140], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:6') 2023-04-27 15:08:35,571 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0485, 3.9299, 3.0581, 3.7938, 3.0268, 2.2961, 4.4373, 4.7957], device='cuda:6'), covar=tensor([0.1692, 0.0539, 0.1083, 0.0327, 0.1972, 0.1353, 0.0170, 0.0039], device='cuda:6'), in_proj_covar=tensor([0.0204, 0.0158, 0.0188, 0.0114, 0.0205, 0.0149, 0.0123, 0.0067], device='cuda:6'), out_proj_covar=tensor([2.4507e-04, 1.8980e-04, 2.0708e-04, 1.3415e-04, 2.4735e-04, 1.7789e-04, 1.4946e-04, 8.5074e-05], device='cuda:6') 2023-04-27 15:08:48,505 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9981, 3.7922, 3.6590, 1.7099, 3.8181, 3.8520, 3.3713, 3.4466], device='cuda:6'), covar=tensor([0.0338, 0.0097, 0.0176, 0.1819, 0.0098, 0.0083, 0.0239, 0.0169], device='cuda:6'), in_proj_covar=tensor([0.0082, 0.0063, 0.0064, 0.0142, 0.0064, 0.0055, 0.0068, 0.0079], device='cuda:6'), out_proj_covar=tensor([1.3079e-04, 1.0284e-04, 1.0883e-04, 2.1726e-04, 1.0801e-04, 9.3805e-05, 1.2384e-04, 1.2668e-04], device='cuda:6') 2023-04-27 15:08:52,555 INFO [train.py:904] (6/8) Epoch 1, batch 7450, loss[loss=0.4257, simple_loss=0.4328, pruned_loss=0.2093, over 11433.00 frames. ], tot_loss[loss=0.3393, simple_loss=0.3923, pruned_loss=0.1431, over 3059863.66 frames. ], batch size: 247, lr: 3.73e-02, grad_scale: 8.0 2023-04-27 15:09:16,292 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6502, 3.2362, 3.2004, 1.4839, 3.2685, 3.3222, 3.0616, 3.0704], device='cuda:6'), covar=tensor([0.0358, 0.0138, 0.0200, 0.2038, 0.0164, 0.0146, 0.0194, 0.0171], device='cuda:6'), in_proj_covar=tensor([0.0082, 0.0062, 0.0063, 0.0142, 0.0064, 0.0054, 0.0067, 0.0078], device='cuda:6'), out_proj_covar=tensor([1.3021e-04, 1.0246e-04, 1.0642e-04, 2.1797e-04, 1.0771e-04, 9.2943e-05, 1.2215e-04, 1.2588e-04], device='cuda:6') 2023-04-27 15:09:22,455 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7468.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:10:15,196 INFO [train.py:904] (6/8) Epoch 1, batch 7500, loss[loss=0.4118, simple_loss=0.4301, pruned_loss=0.1967, over 11822.00 frames. ], tot_loss[loss=0.3411, simple_loss=0.3943, pruned_loss=0.144, over 3049113.41 frames. ], batch size: 247, lr: 3.72e-02, grad_scale: 8.0 2023-04-27 15:10:35,038 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.564e+02 5.230e+02 6.451e+02 8.258e+02 1.814e+03, threshold=1.290e+03, percent-clipped=2.0 2023-04-27 15:10:43,190 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7519.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 15:10:57,243 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2023-04-27 15:11:00,976 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6789, 4.4422, 4.0930, 1.5720, 4.4637, 4.4863, 3.5628, 3.7829], device='cuda:6'), covar=tensor([0.0478, 0.0064, 0.0150, 0.2177, 0.0075, 0.0069, 0.0240, 0.0161], device='cuda:6'), in_proj_covar=tensor([0.0086, 0.0065, 0.0064, 0.0147, 0.0066, 0.0054, 0.0069, 0.0082], device='cuda:6'), out_proj_covar=tensor([1.3679e-04, 1.0750e-04, 1.0881e-04, 2.2614e-04, 1.1147e-04, 9.4421e-05, 1.2634e-04, 1.3371e-04], device='cuda:6') 2023-04-27 15:11:06,811 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7534.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 15:11:31,890 INFO [train.py:904] (6/8) Epoch 1, batch 7550, loss[loss=0.3368, simple_loss=0.395, pruned_loss=0.1393, over 16754.00 frames. ], tot_loss[loss=0.3412, simple_loss=0.3938, pruned_loss=0.1443, over 3044959.98 frames. ], batch size: 124, lr: 3.72e-02, grad_scale: 4.0 2023-04-27 15:11:54,763 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7565.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:12:21,502 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7582.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 15:12:50,418 INFO [train.py:904] (6/8) Epoch 1, batch 7600, loss[loss=0.3083, simple_loss=0.3759, pruned_loss=0.1203, over 16743.00 frames. ], tot_loss[loss=0.343, simple_loss=0.3939, pruned_loss=0.1461, over 3013543.44 frames. ], batch size: 89, lr: 3.71e-02, grad_scale: 8.0 2023-04-27 15:13:10,074 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7613.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:13:12,790 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.798e+02 5.452e+02 6.510e+02 8.243e+02 1.443e+03, threshold=1.302e+03, percent-clipped=3.0 2023-04-27 15:13:30,352 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7624.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:13:38,805 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4268, 3.0138, 2.7076, 3.7718, 2.7349, 3.6634, 2.9317, 2.6188], device='cuda:6'), covar=tensor([0.0330, 0.0347, 0.0340, 0.0269, 0.1054, 0.0190, 0.0513, 0.1048], device='cuda:6'), in_proj_covar=tensor([0.0112, 0.0109, 0.0091, 0.0124, 0.0190, 0.0108, 0.0131, 0.0140], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-27 15:13:40,335 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7631.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:13:42,133 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-27 15:14:13,730 INFO [train.py:904] (6/8) Epoch 1, batch 7650, loss[loss=0.3102, simple_loss=0.3735, pruned_loss=0.1234, over 16774.00 frames. ], tot_loss[loss=0.3448, simple_loss=0.3954, pruned_loss=0.1471, over 3039432.75 frames. ], batch size: 83, lr: 3.70e-02, grad_scale: 8.0 2023-04-27 15:14:41,353 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.10 vs. limit=5.0 2023-04-27 15:15:07,031 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-27 15:15:14,418 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7685.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:15:39,928 INFO [train.py:904] (6/8) Epoch 1, batch 7700, loss[loss=0.4155, simple_loss=0.4312, pruned_loss=0.1999, over 11303.00 frames. ], tot_loss[loss=0.3459, simple_loss=0.3955, pruned_loss=0.1481, over 3033976.09 frames. ], batch size: 248, lr: 3.69e-02, grad_scale: 8.0 2023-04-27 15:15:42,292 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4276, 3.3755, 1.4778, 3.4241, 2.0463, 3.2818, 1.6922, 2.4714], device='cuda:6'), covar=tensor([0.0067, 0.0133, 0.1719, 0.0052, 0.0924, 0.0257, 0.1446, 0.0676], device='cuda:6'), in_proj_covar=tensor([0.0063, 0.0067, 0.0142, 0.0063, 0.0123, 0.0074, 0.0152, 0.0114], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-27 15:15:45,539 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7704.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 15:16:00,933 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.630e+02 5.474e+02 6.817e+02 8.632e+02 3.010e+03, threshold=1.363e+03, percent-clipped=1.0 2023-04-27 15:16:03,281 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-04-27 15:16:45,720 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7743.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:16:50,217 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.94 vs. limit=5.0 2023-04-27 15:16:57,055 INFO [train.py:904] (6/8) Epoch 1, batch 7750, loss[loss=0.3039, simple_loss=0.38, pruned_loss=0.1139, over 17005.00 frames. ], tot_loss[loss=0.3446, simple_loss=0.395, pruned_loss=0.1471, over 3040428.82 frames. ], batch size: 41, lr: 3.68e-02, grad_scale: 8.0 2023-04-27 15:16:59,264 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7752.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 15:17:16,062 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7763.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:18:14,288 INFO [train.py:904] (6/8) Epoch 1, batch 7800, loss[loss=0.3326, simple_loss=0.3947, pruned_loss=0.1352, over 16817.00 frames. ], tot_loss[loss=0.3455, simple_loss=0.3959, pruned_loss=0.1475, over 3060591.51 frames. ], batch size: 83, lr: 3.67e-02, grad_scale: 8.0 2023-04-27 15:18:16,703 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7802.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:18:19,952 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7804.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:18:36,278 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.181e+02 4.983e+02 6.289e+02 7.590e+02 1.248e+03, threshold=1.258e+03, percent-clipped=0.0 2023-04-27 15:18:43,854 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7819.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 15:19:31,860 INFO [train.py:904] (6/8) Epoch 1, batch 7850, loss[loss=0.3528, simple_loss=0.4068, pruned_loss=0.1494, over 16877.00 frames. ], tot_loss[loss=0.3467, simple_loss=0.3973, pruned_loss=0.1481, over 3050809.20 frames. ], batch size: 109, lr: 3.66e-02, grad_scale: 8.0 2023-04-27 15:19:51,364 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7863.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:19:56,871 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7867.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:20:13,051 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1276, 4.0256, 3.3821, 3.7451, 3.2260, 2.5222, 4.3858, 4.8582], device='cuda:6'), covar=tensor([0.1577, 0.0510, 0.0966, 0.0331, 0.1681, 0.1110, 0.0213, 0.0037], device='cuda:6'), in_proj_covar=tensor([0.0213, 0.0168, 0.0197, 0.0119, 0.0209, 0.0151, 0.0132, 0.0070], device='cuda:6'), out_proj_covar=tensor([2.5413e-04, 2.0111e-04, 2.1664e-04, 1.4033e-04, 2.5189e-04, 1.8094e-04, 1.6045e-04, 8.8627e-05], device='cuda:6') 2023-04-27 15:20:49,083 INFO [train.py:904] (6/8) Epoch 1, batch 7900, loss[loss=0.4175, simple_loss=0.433, pruned_loss=0.2009, over 11518.00 frames. ], tot_loss[loss=0.3434, simple_loss=0.3949, pruned_loss=0.146, over 3046048.77 frames. ], batch size: 246, lr: 3.66e-02, grad_scale: 8.0 2023-04-27 15:21:11,993 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.293e+02 4.952e+02 6.401e+02 7.784e+02 1.850e+03, threshold=1.280e+03, percent-clipped=3.0 2023-04-27 15:21:39,386 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7931.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:22:08,690 INFO [train.py:904] (6/8) Epoch 1, batch 7950, loss[loss=0.3344, simple_loss=0.3904, pruned_loss=0.1392, over 16269.00 frames. ], tot_loss[loss=0.342, simple_loss=0.3938, pruned_loss=0.1452, over 3062431.05 frames. ], batch size: 165, lr: 3.65e-02, grad_scale: 8.0 2023-04-27 15:22:12,352 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.06 vs. limit=5.0 2023-04-27 15:22:32,860 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6057, 4.4394, 4.4885, 5.0103, 4.9963, 4.5102, 5.0022, 4.8295], device='cuda:6'), covar=tensor([0.0362, 0.0355, 0.1019, 0.0286, 0.0318, 0.0330, 0.0273, 0.0297], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0172, 0.0261, 0.0181, 0.0155, 0.0161, 0.0145, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 15:22:54,877 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7979.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:22:56,161 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7980.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:23:30,885 INFO [train.py:904] (6/8) Epoch 1, batch 8000, loss[loss=0.3134, simple_loss=0.3732, pruned_loss=0.1268, over 16690.00 frames. ], tot_loss[loss=0.341, simple_loss=0.3931, pruned_loss=0.1444, over 3076336.44 frames. ], batch size: 124, lr: 3.64e-02, grad_scale: 8.0 2023-04-27 15:23:51,329 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.520e+02 5.080e+02 6.216e+02 7.580e+02 1.182e+03, threshold=1.243e+03, percent-clipped=0.0 2023-04-27 15:24:11,970 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.62 vs. limit=5.0 2023-04-27 15:24:45,916 INFO [train.py:904] (6/8) Epoch 1, batch 8050, loss[loss=0.3578, simple_loss=0.4041, pruned_loss=0.1558, over 15313.00 frames. ], tot_loss[loss=0.3408, simple_loss=0.3931, pruned_loss=0.1443, over 3077892.09 frames. ], batch size: 190, lr: 3.63e-02, grad_scale: 8.0 2023-04-27 15:25:04,617 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8063.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:25:11,467 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2023-04-27 15:25:40,304 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2047, 3.2177, 2.6852, 1.8947, 2.6724, 2.0700, 2.7407, 3.2096], device='cuda:6'), covar=tensor([0.0269, 0.0277, 0.0264, 0.1650, 0.0716, 0.1118, 0.0601, 0.0196], device='cuda:6'), in_proj_covar=tensor([0.0106, 0.0078, 0.0119, 0.0156, 0.0149, 0.0141, 0.0138, 0.0069], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-27 15:25:57,835 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8099.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:26:00,725 INFO [train.py:904] (6/8) Epoch 1, batch 8100, loss[loss=0.3869, simple_loss=0.4163, pruned_loss=0.1788, over 11396.00 frames. ], tot_loss[loss=0.3387, simple_loss=0.3916, pruned_loss=0.1429, over 3074342.16 frames. ], batch size: 246, lr: 3.62e-02, grad_scale: 4.0 2023-04-27 15:26:15,156 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=8111.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:26:22,779 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.844e+02 5.411e+02 6.350e+02 7.571e+02 1.310e+03, threshold=1.270e+03, percent-clipped=1.0 2023-04-27 15:27:16,439 INFO [train.py:904] (6/8) Epoch 1, batch 8150, loss[loss=0.3424, simple_loss=0.3905, pruned_loss=0.1471, over 16457.00 frames. ], tot_loss[loss=0.3382, simple_loss=0.3905, pruned_loss=0.143, over 3055165.94 frames. ], batch size: 35, lr: 3.62e-02, grad_scale: 4.0 2023-04-27 15:27:28,060 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8158.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:27:35,472 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7825, 3.2703, 3.1087, 2.5012, 3.2663, 3.2167, 3.4236, 1.8991], device='cuda:6'), covar=tensor([0.1205, 0.0113, 0.0102, 0.0428, 0.0101, 0.0108, 0.0095, 0.0728], device='cuda:6'), in_proj_covar=tensor([0.0113, 0.0043, 0.0048, 0.0079, 0.0044, 0.0045, 0.0050, 0.0085], device='cuda:6'), out_proj_covar=tensor([2.3009e-04, 9.7242e-05, 1.1000e-04, 1.7014e-04, 9.8612e-05, 1.0775e-04, 1.0829e-04, 1.7946e-04], device='cuda:6') 2023-04-27 15:28:33,138 INFO [train.py:904] (6/8) Epoch 1, batch 8200, loss[loss=0.3587, simple_loss=0.3898, pruned_loss=0.1637, over 11379.00 frames. ], tot_loss[loss=0.3353, simple_loss=0.3871, pruned_loss=0.1417, over 3049313.20 frames. ], batch size: 248, lr: 3.61e-02, grad_scale: 4.0 2023-04-27 15:28:56,620 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.505e+02 5.388e+02 6.507e+02 7.835e+02 3.023e+03, threshold=1.301e+03, percent-clipped=3.0 2023-04-27 15:29:15,706 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.00 vs. limit=5.0 2023-04-27 15:29:53,196 INFO [train.py:904] (6/8) Epoch 1, batch 8250, loss[loss=0.2987, simple_loss=0.3495, pruned_loss=0.1239, over 12152.00 frames. ], tot_loss[loss=0.3312, simple_loss=0.3853, pruned_loss=0.1385, over 3028231.91 frames. ], batch size: 246, lr: 3.60e-02, grad_scale: 4.0 2023-04-27 15:30:15,970 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8265.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:30:41,222 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8280.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:30:45,214 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6444, 3.5734, 3.2451, 1.7950, 2.7078, 2.0233, 2.9753, 3.5639], device='cuda:6'), covar=tensor([0.0233, 0.0344, 0.0300, 0.2140, 0.0880, 0.1553, 0.0742, 0.0345], device='cuda:6'), in_proj_covar=tensor([0.0102, 0.0079, 0.0116, 0.0153, 0.0148, 0.0141, 0.0135, 0.0070], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 15:31:14,606 INFO [train.py:904] (6/8) Epoch 1, batch 8300, loss[loss=0.2755, simple_loss=0.356, pruned_loss=0.09752, over 16901.00 frames. ], tot_loss[loss=0.3218, simple_loss=0.3792, pruned_loss=0.1322, over 3025334.81 frames. ], batch size: 116, lr: 3.59e-02, grad_scale: 4.0 2023-04-27 15:31:40,147 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.807e+02 4.177e+02 5.020e+02 6.033e+02 1.438e+03, threshold=1.004e+03, percent-clipped=1.0 2023-04-27 15:31:56,745 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8326.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:31:59,949 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=8328.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:32:36,034 INFO [train.py:904] (6/8) Epoch 1, batch 8350, loss[loss=0.3246, simple_loss=0.3848, pruned_loss=0.1322, over 15320.00 frames. ], tot_loss[loss=0.3142, simple_loss=0.3753, pruned_loss=0.1266, over 3024020.07 frames. ], batch size: 191, lr: 3.58e-02, grad_scale: 4.0 2023-04-27 15:33:05,834 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6507, 3.7240, 3.2458, 3.5114, 2.6544, 2.0593, 3.7730, 4.2717], device='cuda:6'), covar=tensor([0.1596, 0.0508, 0.0770, 0.0286, 0.1701, 0.1319, 0.0213, 0.0053], device='cuda:6'), in_proj_covar=tensor([0.0201, 0.0166, 0.0190, 0.0116, 0.0186, 0.0152, 0.0126, 0.0068], device='cuda:6'), out_proj_covar=tensor([2.4019e-04, 1.9663e-04, 2.0832e-04, 1.3688e-04, 2.2467e-04, 1.8272e-04, 1.5241e-04, 8.7043e-05], device='cuda:6') 2023-04-27 15:33:55,738 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8399.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:33:57,930 INFO [train.py:904] (6/8) Epoch 1, batch 8400, loss[loss=0.2881, simple_loss=0.3439, pruned_loss=0.1161, over 12364.00 frames. ], tot_loss[loss=0.3069, simple_loss=0.37, pruned_loss=0.1219, over 3035003.54 frames. ], batch size: 247, lr: 3.58e-02, grad_scale: 8.0 2023-04-27 15:34:21,541 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.723e+02 4.252e+02 5.243e+02 6.006e+02 1.213e+03, threshold=1.049e+03, percent-clipped=1.0 2023-04-27 15:34:33,663 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3681, 3.2017, 3.1936, 3.3846, 3.0436, 3.2969, 3.1568, 3.2078], device='cuda:6'), covar=tensor([0.0334, 0.0203, 0.0198, 0.0150, 0.0552, 0.0165, 0.0702, 0.0190], device='cuda:6'), in_proj_covar=tensor([0.0090, 0.0063, 0.0119, 0.0093, 0.0140, 0.0093, 0.0085, 0.0097], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 15:34:37,302 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-27 15:35:12,952 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=8447.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:35:18,229 INFO [train.py:904] (6/8) Epoch 1, batch 8450, loss[loss=0.2916, simple_loss=0.3491, pruned_loss=0.117, over 12284.00 frames. ], tot_loss[loss=0.302, simple_loss=0.3668, pruned_loss=0.1186, over 3045880.10 frames. ], batch size: 248, lr: 3.57e-02, grad_scale: 8.0 2023-04-27 15:35:30,872 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8458.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:36:24,022 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5586, 3.6142, 1.8368, 3.5842, 2.3831, 3.4687, 1.9125, 2.7790], device='cuda:6'), covar=tensor([0.0088, 0.0102, 0.1559, 0.0053, 0.0945, 0.0211, 0.1400, 0.0593], device='cuda:6'), in_proj_covar=tensor([0.0062, 0.0069, 0.0149, 0.0065, 0.0126, 0.0074, 0.0152, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-04-27 15:36:39,471 INFO [train.py:904] (6/8) Epoch 1, batch 8500, loss[loss=0.248, simple_loss=0.31, pruned_loss=0.09301, over 11758.00 frames. ], tot_loss[loss=0.295, simple_loss=0.3613, pruned_loss=0.1144, over 3033669.17 frames. ], batch size: 247, lr: 3.56e-02, grad_scale: 8.0 2023-04-27 15:36:49,163 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=8506.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:37:04,603 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.878e+02 4.840e+02 6.056e+02 7.383e+02 1.593e+03, threshold=1.211e+03, percent-clipped=7.0 2023-04-27 15:37:40,752 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6024, 3.2484, 3.2740, 2.6244, 3.2832, 3.1235, 3.3137, 1.7915], device='cuda:6'), covar=tensor([0.1380, 0.0122, 0.0117, 0.0396, 0.0109, 0.0134, 0.0109, 0.0892], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0045, 0.0051, 0.0084, 0.0047, 0.0049, 0.0049, 0.0093], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:6') 2023-04-27 15:37:48,091 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-27 15:38:03,990 INFO [train.py:904] (6/8) Epoch 1, batch 8550, loss[loss=0.2765, simple_loss=0.3538, pruned_loss=0.09957, over 16455.00 frames. ], tot_loss[loss=0.2908, simple_loss=0.3578, pruned_loss=0.1119, over 3049982.99 frames. ], batch size: 68, lr: 3.55e-02, grad_scale: 8.0 2023-04-27 15:38:31,544 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2576, 4.0563, 3.8524, 3.5810, 4.1562, 1.8621, 3.7767, 4.0490], device='cuda:6'), covar=tensor([0.0077, 0.0065, 0.0081, 0.0244, 0.0052, 0.1186, 0.0094, 0.0094], device='cuda:6'), in_proj_covar=tensor([0.0045, 0.0038, 0.0055, 0.0070, 0.0040, 0.0088, 0.0053, 0.0051], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 15:38:54,344 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2023-04-27 15:39:44,565 INFO [train.py:904] (6/8) Epoch 1, batch 8600, loss[loss=0.2475, simple_loss=0.3295, pruned_loss=0.08273, over 17050.00 frames. ], tot_loss[loss=0.2904, simple_loss=0.3589, pruned_loss=0.1109, over 3062806.82 frames. ], batch size: 53, lr: 3.54e-02, grad_scale: 8.0 2023-04-27 15:40:17,874 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.023e+02 4.189e+02 5.220e+02 6.302e+02 1.296e+03, threshold=1.044e+03, percent-clipped=1.0 2023-04-27 15:40:28,413 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8621.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:40:55,104 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-27 15:41:26,180 INFO [train.py:904] (6/8) Epoch 1, batch 8650, loss[loss=0.2722, simple_loss=0.3505, pruned_loss=0.09695, over 15499.00 frames. ], tot_loss[loss=0.287, simple_loss=0.3563, pruned_loss=0.1088, over 3035618.01 frames. ], batch size: 192, lr: 3.54e-02, grad_scale: 8.0 2023-04-27 15:43:13,564 INFO [train.py:904] (6/8) Epoch 1, batch 8700, loss[loss=0.2666, simple_loss=0.3269, pruned_loss=0.1032, over 12678.00 frames. ], tot_loss[loss=0.2816, simple_loss=0.3515, pruned_loss=0.1058, over 3039524.92 frames. ], batch size: 247, lr: 3.53e-02, grad_scale: 8.0 2023-04-27 15:43:41,220 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.978e+02 4.653e+02 5.603e+02 6.924e+02 1.986e+03, threshold=1.121e+03, percent-clipped=4.0 2023-04-27 15:43:54,075 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9098, 4.1727, 3.9508, 4.0961, 3.6289, 3.9351, 3.9254, 4.0657], device='cuda:6'), covar=tensor([0.0349, 0.0585, 0.0606, 0.0277, 0.0564, 0.0507, 0.0398, 0.0690], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0205, 0.0185, 0.0122, 0.0154, 0.0131, 0.0172, 0.0134], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-27 15:44:40,517 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-04-27 15:44:49,919 INFO [train.py:904] (6/8) Epoch 1, batch 8750, loss[loss=0.3071, simple_loss=0.3781, pruned_loss=0.118, over 16792.00 frames. ], tot_loss[loss=0.2793, simple_loss=0.35, pruned_loss=0.1043, over 3032970.53 frames. ], batch size: 124, lr: 3.52e-02, grad_scale: 8.0 2023-04-27 15:45:14,499 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8760.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:45:53,497 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.60 vs. limit=5.0 2023-04-27 15:46:42,405 INFO [train.py:904] (6/8) Epoch 1, batch 8800, loss[loss=0.2658, simple_loss=0.3335, pruned_loss=0.09905, over 12750.00 frames. ], tot_loss[loss=0.2764, simple_loss=0.3477, pruned_loss=0.1025, over 3034374.05 frames. ], batch size: 250, lr: 3.51e-02, grad_scale: 8.0 2023-04-27 15:47:13,493 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.267e+02 4.433e+02 5.797e+02 7.279e+02 2.542e+03, threshold=1.159e+03, percent-clipped=4.0 2023-04-27 15:47:25,117 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8821.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:48:27,335 INFO [train.py:904] (6/8) Epoch 1, batch 8850, loss[loss=0.2358, simple_loss=0.3275, pruned_loss=0.07205, over 16771.00 frames. ], tot_loss[loss=0.2756, simple_loss=0.349, pruned_loss=0.1011, over 3037658.89 frames. ], batch size: 124, lr: 3.51e-02, grad_scale: 8.0 2023-04-27 15:50:13,541 INFO [train.py:904] (6/8) Epoch 1, batch 8900, loss[loss=0.2668, simple_loss=0.3454, pruned_loss=0.09416, over 16699.00 frames. ], tot_loss[loss=0.2732, simple_loss=0.3478, pruned_loss=0.09933, over 3028080.85 frames. ], batch size: 76, lr: 3.50e-02, grad_scale: 8.0 2023-04-27 15:50:42,917 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.973e+02 4.392e+02 5.224e+02 6.190e+02 1.359e+03, threshold=1.045e+03, percent-clipped=1.0 2023-04-27 15:50:56,861 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8921.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:51:13,716 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8927.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:52:19,949 INFO [train.py:904] (6/8) Epoch 1, batch 8950, loss[loss=0.3097, simple_loss=0.3607, pruned_loss=0.1293, over 12827.00 frames. ], tot_loss[loss=0.2743, simple_loss=0.3485, pruned_loss=0.1001, over 3050128.10 frames. ], batch size: 248, lr: 3.49e-02, grad_scale: 8.0 2023-04-27 15:53:00,567 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=8969.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:53:29,466 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8525, 3.7719, 3.9698, 4.2921, 4.3114, 3.9657, 4.3007, 4.1529], device='cuda:6'), covar=tensor([0.0408, 0.0348, 0.0785, 0.0258, 0.0275, 0.0389, 0.0257, 0.0259], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0160, 0.0242, 0.0167, 0.0143, 0.0149, 0.0133, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 15:53:38,288 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8986.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:53:42,534 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8988.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:54:08,289 INFO [train.py:904] (6/8) Epoch 1, batch 9000, loss[loss=0.2481, simple_loss=0.3252, pruned_loss=0.08546, over 16353.00 frames. ], tot_loss[loss=0.2697, simple_loss=0.3442, pruned_loss=0.09764, over 3042442.14 frames. ], batch size: 146, lr: 3.48e-02, grad_scale: 8.0 2023-04-27 15:54:08,290 INFO [train.py:929] (6/8) Computing validation loss 2023-04-27 15:54:19,192 INFO [train.py:938] (6/8) Epoch 1, validation: loss=0.2299, simple_loss=0.3267, pruned_loss=0.06658, over 944034.00 frames. 2023-04-27 15:54:19,193 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17377MB 2023-04-27 15:54:46,241 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9013.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:54:51,598 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.333e+02 3.931e+02 4.892e+02 5.908e+02 1.148e+03, threshold=9.783e+02, percent-clipped=2.0 2023-04-27 15:55:56,711 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9047.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:56:02,940 INFO [train.py:904] (6/8) Epoch 1, batch 9050, loss[loss=0.2849, simple_loss=0.3555, pruned_loss=0.1071, over 17035.00 frames. ], tot_loss[loss=0.273, simple_loss=0.3467, pruned_loss=0.09964, over 3046508.91 frames. ], batch size: 55, lr: 3.48e-02, grad_scale: 8.0 2023-04-27 15:56:48,973 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9074.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:56:52,404 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2023-04-27 15:57:35,898 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8406, 3.8628, 3.5621, 1.7559, 2.7811, 2.2544, 3.1917, 3.9189], device='cuda:6'), covar=tensor([0.0297, 0.0343, 0.0308, 0.2034, 0.0994, 0.1221, 0.1005, 0.0299], device='cuda:6'), in_proj_covar=tensor([0.0105, 0.0079, 0.0122, 0.0155, 0.0151, 0.0142, 0.0136, 0.0074], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-04-27 15:57:45,071 INFO [train.py:904] (6/8) Epoch 1, batch 9100, loss[loss=0.2955, simple_loss=0.3692, pruned_loss=0.1109, over 16687.00 frames. ], tot_loss[loss=0.2737, simple_loss=0.3468, pruned_loss=0.1003, over 3079629.79 frames. ], batch size: 134, lr: 3.47e-02, grad_scale: 8.0 2023-04-27 15:57:48,717 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-27 15:58:14,291 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.937e+02 4.687e+02 5.823e+02 7.250e+02 1.575e+03, threshold=1.165e+03, percent-clipped=5.0 2023-04-27 15:58:14,955 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9116.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:59:42,649 INFO [train.py:904] (6/8) Epoch 1, batch 9150, loss[loss=0.2739, simple_loss=0.3406, pruned_loss=0.1035, over 16649.00 frames. ], tot_loss[loss=0.2736, simple_loss=0.3473, pruned_loss=0.09991, over 3072087.49 frames. ], batch size: 134, lr: 3.46e-02, grad_scale: 8.0 2023-04-27 16:00:59,693 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2071, 4.1322, 4.5034, 4.6171, 4.7260, 4.2685, 4.4632, 4.5021], device='cuda:6'), covar=tensor([0.0276, 0.0346, 0.0585, 0.0499, 0.0347, 0.0238, 0.0525, 0.0256], device='cuda:6'), in_proj_covar=tensor([0.0118, 0.0109, 0.0124, 0.0125, 0.0140, 0.0115, 0.0163, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:01:08,029 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-27 16:01:13,344 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8681, 1.5127, 1.5789, 1.2917, 1.6676, 1.6517, 1.7495, 1.7956], device='cuda:6'), covar=tensor([0.0034, 0.0228, 0.0145, 0.0181, 0.0077, 0.0151, 0.0071, 0.0064], device='cuda:6'), in_proj_covar=tensor([0.0033, 0.0074, 0.0058, 0.0064, 0.0050, 0.0062, 0.0033, 0.0043], device='cuda:6'), out_proj_covar=tensor([4.5923e-05, 1.1412e-04, 8.6296e-05, 9.7385e-05, 7.7103e-05, 9.2721e-05, 5.1312e-05, 6.9899e-05], device='cuda:6') 2023-04-27 16:01:13,368 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0820, 2.8128, 2.2554, 3.2029, 3.3940, 3.3111, 2.0103, 3.0356], device='cuda:6'), covar=tensor([0.1660, 0.0265, 0.1384, 0.0118, 0.0120, 0.0313, 0.1015, 0.0426], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0093, 0.0164, 0.0061, 0.0069, 0.0084, 0.0138, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:01:27,877 INFO [train.py:904] (6/8) Epoch 1, batch 9200, loss[loss=0.252, simple_loss=0.3154, pruned_loss=0.09432, over 12414.00 frames. ], tot_loss[loss=0.2687, simple_loss=0.3417, pruned_loss=0.09784, over 3085674.68 frames. ], batch size: 247, lr: 3.45e-02, grad_scale: 8.0 2023-04-27 16:01:54,927 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.438e+02 5.266e+02 6.436e+02 8.220e+02 2.129e+03, threshold=1.287e+03, percent-clipped=5.0 2023-04-27 16:03:02,027 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.02 vs. limit=5.0 2023-04-27 16:03:04,218 INFO [train.py:904] (6/8) Epoch 1, batch 9250, loss[loss=0.2469, simple_loss=0.3296, pruned_loss=0.0821, over 16894.00 frames. ], tot_loss[loss=0.2677, simple_loss=0.3406, pruned_loss=0.0974, over 3077658.12 frames. ], batch size: 96, lr: 3.45e-02, grad_scale: 8.0 2023-04-27 16:04:06,013 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.99 vs. limit=5.0 2023-04-27 16:04:17,162 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9283.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:04:58,120 INFO [train.py:904] (6/8) Epoch 1, batch 9300, loss[loss=0.2403, simple_loss=0.3207, pruned_loss=0.07996, over 16847.00 frames. ], tot_loss[loss=0.2637, simple_loss=0.3374, pruned_loss=0.09496, over 3083258.94 frames. ], batch size: 124, lr: 3.44e-02, grad_scale: 8.0 2023-04-27 16:05:33,415 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.238e+02 3.939e+02 4.601e+02 5.465e+02 1.094e+03, threshold=9.201e+02, percent-clipped=0.0 2023-04-27 16:06:27,895 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9342.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:06:44,326 INFO [train.py:904] (6/8) Epoch 1, batch 9350, loss[loss=0.2996, simple_loss=0.3625, pruned_loss=0.1183, over 15282.00 frames. ], tot_loss[loss=0.263, simple_loss=0.3371, pruned_loss=0.09449, over 3085746.13 frames. ], batch size: 190, lr: 3.43e-02, grad_scale: 8.0 2023-04-27 16:07:24,444 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9369.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:08:26,600 INFO [train.py:904] (6/8) Epoch 1, batch 9400, loss[loss=0.2779, simple_loss=0.3535, pruned_loss=0.1012, over 15374.00 frames. ], tot_loss[loss=0.2626, simple_loss=0.3368, pruned_loss=0.09417, over 3077416.15 frames. ], batch size: 191, lr: 3.43e-02, grad_scale: 8.0 2023-04-27 16:08:54,910 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9414.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:08:56,912 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.254e+02 4.024e+02 4.732e+02 5.961e+02 1.474e+03, threshold=9.465e+02, percent-clipped=2.0 2023-04-27 16:08:58,048 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9416.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:09:12,699 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-27 16:09:34,143 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.48 vs. limit=5.0 2023-04-27 16:10:08,171 INFO [train.py:904] (6/8) Epoch 1, batch 9450, loss[loss=0.218, simple_loss=0.3085, pruned_loss=0.06376, over 17144.00 frames. ], tot_loss[loss=0.2646, simple_loss=0.339, pruned_loss=0.09506, over 3061426.15 frames. ], batch size: 48, lr: 3.42e-02, grad_scale: 8.0 2023-04-27 16:10:33,963 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=9464.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:10:59,059 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9475.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:11:50,037 INFO [train.py:904] (6/8) Epoch 1, batch 9500, loss[loss=0.2409, simple_loss=0.3144, pruned_loss=0.08377, over 12543.00 frames. ], tot_loss[loss=0.2624, simple_loss=0.3374, pruned_loss=0.09368, over 3069923.03 frames. ], batch size: 248, lr: 3.41e-02, grad_scale: 8.0 2023-04-27 16:12:21,412 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.089e+02 4.155e+02 5.314e+02 6.779e+02 1.064e+03, threshold=1.063e+03, percent-clipped=4.0 2023-04-27 16:12:24,189 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8561, 4.0074, 3.9433, 1.8064, 4.0420, 4.1416, 3.5181, 3.4625], device='cuda:6'), covar=tensor([0.0580, 0.0079, 0.0179, 0.1913, 0.0071, 0.0049, 0.0274, 0.0284], device='cuda:6'), in_proj_covar=tensor([0.0105, 0.0073, 0.0072, 0.0150, 0.0065, 0.0067, 0.0080, 0.0093], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:12:47,224 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6784, 4.3706, 4.5401, 4.7329, 4.0035, 4.4921, 4.5610, 4.2201], device='cuda:6'), covar=tensor([0.0222, 0.0136, 0.0146, 0.0095, 0.0649, 0.0148, 0.0149, 0.0165], device='cuda:6'), in_proj_covar=tensor([0.0087, 0.0064, 0.0113, 0.0089, 0.0139, 0.0089, 0.0082, 0.0096], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:13:14,480 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2023-04-27 16:13:37,007 INFO [train.py:904] (6/8) Epoch 1, batch 9550, loss[loss=0.2878, simple_loss=0.3637, pruned_loss=0.106, over 16869.00 frames. ], tot_loss[loss=0.2617, simple_loss=0.3365, pruned_loss=0.09349, over 3075999.11 frames. ], batch size: 124, lr: 3.41e-02, grad_scale: 8.0 2023-04-27 16:14:15,566 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2023-04-27 16:14:46,616 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9583.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:14:48,671 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9584.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:15:18,631 INFO [train.py:904] (6/8) Epoch 1, batch 9600, loss[loss=0.2535, simple_loss=0.3409, pruned_loss=0.08303, over 16595.00 frames. ], tot_loss[loss=0.2649, simple_loss=0.3393, pruned_loss=0.09528, over 3081709.18 frames. ], batch size: 68, lr: 3.40e-02, grad_scale: 8.0 2023-04-27 16:15:48,621 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.034e+02 4.688e+02 5.707e+02 6.655e+02 1.542e+03, threshold=1.141e+03, percent-clipped=4.0 2023-04-27 16:16:20,303 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=9631.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:16:45,903 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9642.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:16:53,452 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9645.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:17:08,557 INFO [train.py:904] (6/8) Epoch 1, batch 9650, loss[loss=0.2596, simple_loss=0.3283, pruned_loss=0.0955, over 12169.00 frames. ], tot_loss[loss=0.2673, simple_loss=0.3423, pruned_loss=0.0962, over 3080485.85 frames. ], batch size: 247, lr: 3.39e-02, grad_scale: 8.0 2023-04-27 16:17:15,070 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3696, 3.2654, 3.2437, 3.3636, 2.9351, 3.3231, 3.2564, 3.1108], device='cuda:6'), covar=tensor([0.0322, 0.0198, 0.0218, 0.0162, 0.0589, 0.0161, 0.0609, 0.0248], device='cuda:6'), in_proj_covar=tensor([0.0089, 0.0066, 0.0112, 0.0091, 0.0138, 0.0089, 0.0082, 0.0097], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:17:26,845 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9658.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:17:45,368 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0732, 2.8293, 2.2512, 3.2824, 3.3512, 3.3797, 2.0957, 2.8798], device='cuda:6'), covar=tensor([0.1557, 0.0260, 0.1311, 0.0108, 0.0123, 0.0275, 0.0941, 0.0440], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0094, 0.0158, 0.0060, 0.0070, 0.0083, 0.0136, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:17:52,791 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9669.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:18:35,170 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=9690.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:18:57,045 INFO [train.py:904] (6/8) Epoch 1, batch 9700, loss[loss=0.3103, simple_loss=0.3628, pruned_loss=0.1289, over 12204.00 frames. ], tot_loss[loss=0.2659, simple_loss=0.3404, pruned_loss=0.0957, over 3062375.25 frames. ], batch size: 248, lr: 3.38e-02, grad_scale: 8.0 2023-04-27 16:19:12,615 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5925, 3.1546, 2.4763, 3.8940, 2.4437, 3.8350, 2.6493, 2.5356], device='cuda:6'), covar=tensor([0.0249, 0.0292, 0.0314, 0.0201, 0.1265, 0.0141, 0.0601, 0.1006], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0131, 0.0105, 0.0150, 0.0206, 0.0125, 0.0149, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:19:24,123 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5418, 2.5776, 2.5843, 2.0927, 2.5552, 2.5737, 2.6829, 1.6561], device='cuda:6'), covar=tensor([0.1020, 0.0068, 0.0112, 0.0474, 0.0104, 0.0110, 0.0094, 0.0853], device='cuda:6'), in_proj_covar=tensor([0.0111, 0.0042, 0.0050, 0.0085, 0.0045, 0.0047, 0.0048, 0.0092], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:6') 2023-04-27 16:19:24,736 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.013e+02 4.098e+02 5.059e+02 6.165e+02 1.399e+03, threshold=1.012e+03, percent-clipped=2.0 2023-04-27 16:19:28,142 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=9717.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:19:33,532 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9719.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:20:27,164 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9743.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:20:40,562 INFO [train.py:904] (6/8) Epoch 1, batch 9750, loss[loss=0.2592, simple_loss=0.3483, pruned_loss=0.08511, over 16775.00 frames. ], tot_loss[loss=0.2653, simple_loss=0.3395, pruned_loss=0.09555, over 3071026.61 frames. ], batch size: 124, lr: 3.38e-02, grad_scale: 8.0 2023-04-27 16:21:17,328 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9770.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:22:06,405 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.3870, 3.1143, 3.0963, 2.2054, 2.9786, 3.0301, 3.0760, 1.7014], device='cuda:6'), covar=tensor([0.1315, 0.0055, 0.0082, 0.0456, 0.0063, 0.0092, 0.0063, 0.0912], device='cuda:6'), in_proj_covar=tensor([0.0113, 0.0043, 0.0051, 0.0086, 0.0047, 0.0048, 0.0050, 0.0095], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:6') 2023-04-27 16:22:19,291 INFO [train.py:904] (6/8) Epoch 1, batch 9800, loss[loss=0.2756, simple_loss=0.3606, pruned_loss=0.09536, over 16921.00 frames. ], tot_loss[loss=0.2633, simple_loss=0.339, pruned_loss=0.09376, over 3077054.18 frames. ], batch size: 109, lr: 3.37e-02, grad_scale: 8.0 2023-04-27 16:22:26,939 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9804.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:22:47,902 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.893e+02 4.491e+02 5.387e+02 6.683e+02 1.234e+03, threshold=1.077e+03, percent-clipped=5.0 2023-04-27 16:24:05,877 INFO [train.py:904] (6/8) Epoch 1, batch 9850, loss[loss=0.2698, simple_loss=0.3498, pruned_loss=0.09484, over 16680.00 frames. ], tot_loss[loss=0.2628, simple_loss=0.3397, pruned_loss=0.09293, over 3079517.45 frames. ], batch size: 134, lr: 3.36e-02, grad_scale: 8.0 2023-04-27 16:25:58,729 INFO [train.py:904] (6/8) Epoch 1, batch 9900, loss[loss=0.2625, simple_loss=0.3434, pruned_loss=0.09077, over 15317.00 frames. ], tot_loss[loss=0.2623, simple_loss=0.3394, pruned_loss=0.09261, over 3056330.83 frames. ], batch size: 191, lr: 3.36e-02, grad_scale: 8.0 2023-04-27 16:26:31,897 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.967e+02 4.199e+02 5.185e+02 6.475e+02 1.120e+03, threshold=1.037e+03, percent-clipped=1.0 2023-04-27 16:26:38,075 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6122, 2.6291, 1.4720, 2.6590, 1.9062, 2.7127, 1.7797, 2.3855], device='cuda:6'), covar=tensor([0.0086, 0.0201, 0.1607, 0.0095, 0.0864, 0.0276, 0.1341, 0.0469], device='cuda:6'), in_proj_covar=tensor([0.0064, 0.0076, 0.0156, 0.0069, 0.0137, 0.0087, 0.0162, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-04-27 16:27:32,790 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9940.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:27:57,264 INFO [train.py:904] (6/8) Epoch 1, batch 9950, loss[loss=0.252, simple_loss=0.3305, pruned_loss=0.0868, over 16629.00 frames. ], tot_loss[loss=0.2641, simple_loss=0.3414, pruned_loss=0.09344, over 3052977.56 frames. ], batch size: 57, lr: 3.35e-02, grad_scale: 8.0 2023-04-27 16:27:59,448 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2632, 1.3141, 1.8293, 2.1085, 2.1150, 2.0990, 1.5429, 2.2889], device='cuda:6'), covar=tensor([0.0056, 0.0439, 0.0193, 0.0108, 0.0069, 0.0110, 0.0261, 0.0047], device='cuda:6'), in_proj_covar=tensor([0.0053, 0.0095, 0.0076, 0.0061, 0.0046, 0.0047, 0.0075, 0.0042], device='cuda:6'), out_proj_covar=tensor([9.9227e-05, 1.7229e-04, 1.4340e-04, 1.1381e-04, 8.1092e-05, 8.7315e-05, 1.3144e-04, 7.6391e-05], device='cuda:6') 2023-04-27 16:28:29,745 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-27 16:30:02,393 INFO [train.py:904] (6/8) Epoch 1, batch 10000, loss[loss=0.2503, simple_loss=0.3422, pruned_loss=0.07918, over 15428.00 frames. ], tot_loss[loss=0.2633, simple_loss=0.3401, pruned_loss=0.0933, over 3051584.93 frames. ], batch size: 192, lr: 3.34e-02, grad_scale: 8.0 2023-04-27 16:30:30,632 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10014.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:30:32,719 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.207e+02 4.345e+02 5.531e+02 6.703e+02 1.367e+03, threshold=1.106e+03, percent-clipped=3.0 2023-04-27 16:31:22,568 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10039.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:31:43,671 INFO [train.py:904] (6/8) Epoch 1, batch 10050, loss[loss=0.2815, simple_loss=0.3485, pruned_loss=0.1073, over 11971.00 frames. ], tot_loss[loss=0.2616, simple_loss=0.339, pruned_loss=0.09211, over 3051454.02 frames. ], batch size: 248, lr: 3.34e-02, grad_scale: 8.0 2023-04-27 16:32:06,741 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10062.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 16:32:23,104 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10070.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:32:52,093 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3139, 2.8746, 2.4496, 3.5670, 2.3283, 3.4409, 2.4962, 2.3908], device='cuda:6'), covar=tensor([0.0328, 0.0341, 0.0315, 0.0278, 0.1192, 0.0202, 0.0585, 0.1112], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0133, 0.0105, 0.0152, 0.0205, 0.0126, 0.0150, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:33:16,706 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10099.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:33:18,963 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10100.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 16:33:19,581 INFO [train.py:904] (6/8) Epoch 1, batch 10100, loss[loss=0.2652, simple_loss=0.3319, pruned_loss=0.09921, over 15421.00 frames. ], tot_loss[loss=0.2639, simple_loss=0.3404, pruned_loss=0.09368, over 3029610.31 frames. ], batch size: 191, lr: 3.33e-02, grad_scale: 16.0 2023-04-27 16:33:46,249 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9752, 5.2453, 4.9387, 5.0396, 4.6739, 4.4340, 4.7528, 5.3147], device='cuda:6'), covar=tensor([0.0264, 0.0482, 0.0598, 0.0268, 0.0379, 0.0406, 0.0369, 0.0397], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0210, 0.0186, 0.0130, 0.0155, 0.0128, 0.0175, 0.0139], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:33:49,433 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.152e+02 4.595e+02 5.786e+02 6.822e+02 2.551e+03, threshold=1.157e+03, percent-clipped=1.0 2023-04-27 16:33:53,662 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=10118.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:34:06,069 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10123.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 16:35:05,142 INFO [train.py:904] (6/8) Epoch 2, batch 0, loss[loss=0.4551, simple_loss=0.4595, pruned_loss=0.2253, over 16798.00 frames. ], tot_loss[loss=0.4551, simple_loss=0.4595, pruned_loss=0.2253, over 16798.00 frames. ], batch size: 102, lr: 3.26e-02, grad_scale: 8.0 2023-04-27 16:35:05,143 INFO [train.py:929] (6/8) Computing validation loss 2023-04-27 16:35:12,742 INFO [train.py:938] (6/8) Epoch 2, validation: loss=0.2235, simple_loss=0.3221, pruned_loss=0.06242, over 944034.00 frames. 2023-04-27 16:35:12,742 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17377MB 2023-04-27 16:35:37,615 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-27 16:35:46,440 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2066, 1.8319, 2.2023, 1.5990, 2.6409, 2.7653, 3.1042, 3.1512], device='cuda:6'), covar=tensor([0.0027, 0.0276, 0.0138, 0.0224, 0.0105, 0.0123, 0.0050, 0.0072], device='cuda:6'), in_proj_covar=tensor([0.0036, 0.0077, 0.0065, 0.0071, 0.0062, 0.0069, 0.0039, 0.0046], device='cuda:6'), out_proj_covar=tensor([5.0514e-05, 1.1865e-04, 9.9956e-05, 1.1038e-04, 9.5770e-05, 1.0476e-04, 5.9811e-05, 7.5628e-05], device='cuda:6') 2023-04-27 16:36:22,853 INFO [train.py:904] (6/8) Epoch 2, batch 50, loss[loss=0.32, simple_loss=0.372, pruned_loss=0.134, over 16451.00 frames. ], tot_loss[loss=0.3252, simple_loss=0.3748, pruned_loss=0.1378, over 749298.82 frames. ], batch size: 68, lr: 3.25e-02, grad_scale: 4.0 2023-04-27 16:36:45,678 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.307e+02 5.068e+02 6.475e+02 7.712e+02 1.660e+03, threshold=1.295e+03, percent-clipped=4.0 2023-04-27 16:37:16,879 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10240.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:37:30,888 INFO [train.py:904] (6/8) Epoch 2, batch 100, loss[loss=0.3885, simple_loss=0.42, pruned_loss=0.1785, over 12516.00 frames. ], tot_loss[loss=0.3108, simple_loss=0.3627, pruned_loss=0.1294, over 1318821.21 frames. ], batch size: 246, lr: 3.25e-02, grad_scale: 4.0 2023-04-27 16:37:41,477 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-27 16:38:22,579 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=10288.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:38:38,636 INFO [train.py:904] (6/8) Epoch 2, batch 150, loss[loss=0.3085, simple_loss=0.358, pruned_loss=0.1295, over 16389.00 frames. ], tot_loss[loss=0.3039, simple_loss=0.3574, pruned_loss=0.1251, over 1767412.37 frames. ], batch size: 68, lr: 3.24e-02, grad_scale: 4.0 2023-04-27 16:38:56,736 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10314.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:39:01,494 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.010e+02 4.543e+02 5.695e+02 6.778e+02 1.541e+03, threshold=1.139e+03, percent-clipped=2.0 2023-04-27 16:39:29,551 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2023-04-27 16:39:47,448 INFO [train.py:904] (6/8) Epoch 2, batch 200, loss[loss=0.3341, simple_loss=0.3639, pruned_loss=0.1521, over 16915.00 frames. ], tot_loss[loss=0.3036, simple_loss=0.3581, pruned_loss=0.1246, over 2110588.58 frames. ], batch size: 90, lr: 3.23e-02, grad_scale: 4.0 2023-04-27 16:39:52,173 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4043, 2.2983, 2.6213, 1.9640, 3.0960, 2.8464, 3.7206, 3.3942], device='cuda:6'), covar=tensor([0.0022, 0.0218, 0.0108, 0.0221, 0.0095, 0.0138, 0.0038, 0.0062], device='cuda:6'), in_proj_covar=tensor([0.0041, 0.0080, 0.0067, 0.0074, 0.0065, 0.0073, 0.0040, 0.0049], device='cuda:6'), out_proj_covar=tensor([5.9509e-05, 1.2322e-04, 1.0326e-04, 1.1485e-04, 1.0126e-04, 1.1323e-04, 6.2322e-05, 7.9761e-05], device='cuda:6') 2023-04-27 16:40:03,411 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=10362.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:40:35,474 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8825, 3.0677, 1.5197, 3.1272, 2.0335, 3.1276, 1.7559, 2.4894], device='cuda:6'), covar=tensor([0.0096, 0.0200, 0.1599, 0.0075, 0.0873, 0.0310, 0.1345, 0.0624], device='cuda:6'), in_proj_covar=tensor([0.0070, 0.0082, 0.0160, 0.0072, 0.0138, 0.0098, 0.0167, 0.0137], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0003, 0.0001, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 16:40:45,269 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8575, 3.9047, 1.6005, 4.0364, 2.4986, 3.9356, 1.7876, 2.8733], device='cuda:6'), covar=tensor([0.0049, 0.0173, 0.1524, 0.0041, 0.0794, 0.0212, 0.1403, 0.0575], device='cuda:6'), in_proj_covar=tensor([0.0070, 0.0083, 0.0162, 0.0073, 0.0140, 0.0098, 0.0168, 0.0138], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0003, 0.0001, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 16:40:50,466 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10395.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 16:40:55,765 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10399.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:40:58,054 INFO [train.py:904] (6/8) Epoch 2, batch 250, loss[loss=0.2991, simple_loss=0.3468, pruned_loss=0.1257, over 16861.00 frames. ], tot_loss[loss=0.3005, simple_loss=0.3553, pruned_loss=0.1228, over 2374672.88 frames. ], batch size: 116, lr: 3.23e-02, grad_scale: 4.0 2023-04-27 16:41:10,997 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10409.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:41:22,651 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.115e+02 4.151e+02 5.077e+02 6.310e+02 1.196e+03, threshold=1.015e+03, percent-clipped=1.0 2023-04-27 16:41:22,965 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10418.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 16:42:03,852 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=10447.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:42:04,306 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-04-27 16:42:09,609 INFO [train.py:904] (6/8) Epoch 2, batch 300, loss[loss=0.2383, simple_loss=0.3179, pruned_loss=0.07936, over 17082.00 frames. ], tot_loss[loss=0.2925, simple_loss=0.3494, pruned_loss=0.1178, over 2581604.83 frames. ], batch size: 55, lr: 3.22e-02, grad_scale: 4.0 2023-04-27 16:42:14,784 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10454.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:42:22,993 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-04-27 16:42:24,664 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2536, 1.7442, 2.4528, 2.7866, 3.2113, 3.1645, 1.8233, 3.1356], device='cuda:6'), covar=tensor([0.0064, 0.0452, 0.0204, 0.0171, 0.0062, 0.0115, 0.0314, 0.0082], device='cuda:6'), in_proj_covar=tensor([0.0057, 0.0095, 0.0078, 0.0064, 0.0046, 0.0049, 0.0078, 0.0045], device='cuda:6'), out_proj_covar=tensor([1.0462e-04, 1.7113e-04, 1.4917e-04, 1.2056e-04, 8.1841e-05, 9.4835e-05, 1.3624e-04, 8.3170e-05], device='cuda:6') 2023-04-27 16:42:36,347 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10470.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:43:10,277 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7501, 4.3150, 4.5353, 4.6970, 4.0873, 4.5042, 4.4984, 4.2294], device='cuda:6'), covar=tensor([0.0238, 0.0181, 0.0184, 0.0111, 0.0738, 0.0197, 0.0219, 0.0237], device='cuda:6'), in_proj_covar=tensor([0.0112, 0.0081, 0.0150, 0.0116, 0.0183, 0.0116, 0.0104, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 16:43:19,861 INFO [train.py:904] (6/8) Epoch 2, batch 350, loss[loss=0.2796, simple_loss=0.3257, pruned_loss=0.1167, over 16773.00 frames. ], tot_loss[loss=0.2888, simple_loss=0.3458, pruned_loss=0.1159, over 2750792.36 frames. ], batch size: 83, lr: 3.21e-02, grad_scale: 4.0 2023-04-27 16:43:23,296 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1030, 3.9626, 3.8400, 3.6037, 3.8698, 1.9618, 3.8623, 4.0673], device='cuda:6'), covar=tensor([0.0106, 0.0094, 0.0098, 0.0351, 0.0088, 0.1220, 0.0094, 0.0124], device='cuda:6'), in_proj_covar=tensor([0.0054, 0.0046, 0.0067, 0.0087, 0.0048, 0.0100, 0.0062, 0.0064], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:43:39,959 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10515.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 16:43:42,895 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.580e+02 4.016e+02 4.975e+02 5.908e+02 1.216e+03, threshold=9.949e+02, percent-clipped=2.0 2023-04-27 16:44:27,073 INFO [train.py:904] (6/8) Epoch 2, batch 400, loss[loss=0.3702, simple_loss=0.3943, pruned_loss=0.173, over 16742.00 frames. ], tot_loss[loss=0.2853, simple_loss=0.3431, pruned_loss=0.1137, over 2881640.98 frames. ], batch size: 134, lr: 3.21e-02, grad_scale: 8.0 2023-04-27 16:45:33,472 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7308, 4.8778, 4.2291, 4.9286, 4.6321, 4.4394, 4.7668, 5.0210], device='cuda:6'), covar=tensor([0.0763, 0.1331, 0.2085, 0.0536, 0.0810, 0.0750, 0.0683, 0.0771], device='cuda:6'), in_proj_covar=tensor([0.0205, 0.0287, 0.0257, 0.0169, 0.0200, 0.0162, 0.0228, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:45:36,485 INFO [train.py:904] (6/8) Epoch 2, batch 450, loss[loss=0.2806, simple_loss=0.3302, pruned_loss=0.1155, over 16921.00 frames. ], tot_loss[loss=0.2824, simple_loss=0.3412, pruned_loss=0.1118, over 2978767.81 frames. ], batch size: 96, lr: 3.20e-02, grad_scale: 8.0 2023-04-27 16:45:41,350 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.7877, 6.1033, 5.7499, 5.9986, 5.3256, 5.0643, 5.6171, 6.1891], device='cuda:6'), covar=tensor([0.0385, 0.0500, 0.0782, 0.0260, 0.0458, 0.0324, 0.0430, 0.0422], device='cuda:6'), in_proj_covar=tensor([0.0202, 0.0283, 0.0253, 0.0167, 0.0197, 0.0161, 0.0225, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:45:59,585 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.527e+02 4.134e+02 5.028e+02 5.858e+02 1.204e+03, threshold=1.006e+03, percent-clipped=2.0 2023-04-27 16:46:44,100 INFO [train.py:904] (6/8) Epoch 2, batch 500, loss[loss=0.3032, simple_loss=0.3516, pruned_loss=0.1274, over 16893.00 frames. ], tot_loss[loss=0.2784, simple_loss=0.3382, pruned_loss=0.1093, over 3051436.66 frames. ], batch size: 116, lr: 3.20e-02, grad_scale: 8.0 2023-04-27 16:47:45,017 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10695.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:47:52,720 INFO [train.py:904] (6/8) Epoch 2, batch 550, loss[loss=0.2523, simple_loss=0.3137, pruned_loss=0.0954, over 16834.00 frames. ], tot_loss[loss=0.2766, simple_loss=0.3368, pruned_loss=0.1082, over 3106300.32 frames. ], batch size: 39, lr: 3.19e-02, grad_scale: 8.0 2023-04-27 16:48:07,203 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8121, 4.8480, 4.5402, 1.8753, 3.4021, 2.6082, 3.9746, 4.7927], device='cuda:6'), covar=tensor([0.0315, 0.0314, 0.0262, 0.1975, 0.0735, 0.1154, 0.0789, 0.0493], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0090, 0.0132, 0.0155, 0.0146, 0.0139, 0.0144, 0.0088], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-27 16:48:17,046 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.652e+02 3.972e+02 4.860e+02 6.091e+02 1.021e+03, threshold=9.720e+02, percent-clipped=1.0 2023-04-27 16:48:17,410 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10718.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 16:48:51,498 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=10743.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:48:52,830 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10744.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:49:02,282 INFO [train.py:904] (6/8) Epoch 2, batch 600, loss[loss=0.275, simple_loss=0.318, pruned_loss=0.116, over 16796.00 frames. ], tot_loss[loss=0.2755, simple_loss=0.3354, pruned_loss=0.1078, over 3158545.83 frames. ], batch size: 83, lr: 3.18e-02, grad_scale: 8.0 2023-04-27 16:49:22,655 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10765.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:49:24,353 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=10766.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 16:49:33,811 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10774.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:50:11,093 INFO [train.py:904] (6/8) Epoch 2, batch 650, loss[loss=0.217, simple_loss=0.2909, pruned_loss=0.0715, over 16881.00 frames. ], tot_loss[loss=0.2734, simple_loss=0.3335, pruned_loss=0.1067, over 3203990.35 frames. ], batch size: 42, lr: 3.18e-02, grad_scale: 8.0 2023-04-27 16:50:17,141 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10805.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:50:23,992 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10810.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 16:50:25,448 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1024, 4.1042, 2.5997, 5.3803, 5.0881, 4.4756, 3.1482, 4.4406], device='cuda:6'), covar=tensor([0.1678, 0.0344, 0.1393, 0.0048, 0.0123, 0.0364, 0.0792, 0.0294], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0103, 0.0158, 0.0060, 0.0079, 0.0089, 0.0140, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:50:33,022 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.483e+02 4.152e+02 4.979e+02 6.118e+02 1.196e+03, threshold=9.959e+02, percent-clipped=5.0 2023-04-27 16:50:39,133 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0190, 4.6546, 4.8803, 5.3496, 5.3426, 4.8605, 5.3468, 5.1304], device='cuda:6'), covar=tensor([0.0348, 0.0462, 0.0925, 0.0252, 0.0289, 0.0388, 0.0219, 0.0327], device='cuda:6'), in_proj_covar=tensor([0.0203, 0.0230, 0.0350, 0.0240, 0.0196, 0.0201, 0.0176, 0.0202], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:50:56,105 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2023-04-27 16:50:57,022 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10835.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:51:19,285 INFO [train.py:904] (6/8) Epoch 2, batch 700, loss[loss=0.2248, simple_loss=0.296, pruned_loss=0.07681, over 16829.00 frames. ], tot_loss[loss=0.2716, simple_loss=0.3324, pruned_loss=0.1054, over 3232498.65 frames. ], batch size: 39, lr: 3.17e-02, grad_scale: 8.0 2023-04-27 16:52:05,447 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.48 vs. limit=5.0 2023-04-27 16:52:06,295 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10885.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 16:52:24,846 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4740, 1.8496, 2.7501, 2.7181, 3.3627, 3.2776, 2.2772, 3.1576], device='cuda:6'), covar=tensor([0.0041, 0.0382, 0.0147, 0.0178, 0.0077, 0.0123, 0.0216, 0.0100], device='cuda:6'), in_proj_covar=tensor([0.0063, 0.0102, 0.0084, 0.0073, 0.0049, 0.0054, 0.0085, 0.0049], device='cuda:6'), out_proj_covar=tensor([1.1544e-04, 1.8323e-04, 1.6141e-04, 1.3823e-04, 8.8774e-05, 1.0491e-04, 1.4892e-04, 9.0844e-05], device='cuda:6') 2023-04-27 16:52:26,661 INFO [train.py:904] (6/8) Epoch 2, batch 750, loss[loss=0.2494, simple_loss=0.3305, pruned_loss=0.08415, over 17014.00 frames. ], tot_loss[loss=0.2729, simple_loss=0.3337, pruned_loss=0.106, over 3248369.02 frames. ], batch size: 50, lr: 3.17e-02, grad_scale: 8.0 2023-04-27 16:52:50,206 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.205e+02 3.722e+02 4.836e+02 5.954e+02 8.013e+02, threshold=9.671e+02, percent-clipped=0.0 2023-04-27 16:53:28,017 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10946.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 16:53:33,379 INFO [train.py:904] (6/8) Epoch 2, batch 800, loss[loss=0.2867, simple_loss=0.3335, pruned_loss=0.12, over 16335.00 frames. ], tot_loss[loss=0.2731, simple_loss=0.3337, pruned_loss=0.1063, over 3251355.43 frames. ], batch size: 165, lr: 3.16e-02, grad_scale: 8.0 2023-04-27 16:53:56,107 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0007, 3.6924, 2.4507, 4.6803, 4.5225, 4.3755, 2.0454, 3.4725], device='cuda:6'), covar=tensor([0.1956, 0.0426, 0.1573, 0.0105, 0.0194, 0.0336, 0.1313, 0.0567], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0104, 0.0159, 0.0062, 0.0082, 0.0090, 0.0141, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:54:43,054 INFO [train.py:904] (6/8) Epoch 2, batch 850, loss[loss=0.2532, simple_loss=0.3113, pruned_loss=0.09755, over 16772.00 frames. ], tot_loss[loss=0.2721, simple_loss=0.333, pruned_loss=0.1055, over 3260482.37 frames. ], batch size: 83, lr: 3.15e-02, grad_scale: 8.0 2023-04-27 16:55:06,011 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.379e+02 3.969e+02 4.868e+02 6.050e+02 1.097e+03, threshold=9.736e+02, percent-clipped=1.0 2023-04-27 16:55:49,295 INFO [train.py:904] (6/8) Epoch 2, batch 900, loss[loss=0.2562, simple_loss=0.3318, pruned_loss=0.09028, over 17093.00 frames. ], tot_loss[loss=0.2687, simple_loss=0.3305, pruned_loss=0.1035, over 3275979.33 frames. ], batch size: 53, lr: 3.15e-02, grad_scale: 8.0 2023-04-27 16:56:09,973 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11065.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:56:16,337 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-27 16:56:53,644 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-27 16:56:57,478 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11100.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:56:58,311 INFO [train.py:904] (6/8) Epoch 2, batch 950, loss[loss=0.281, simple_loss=0.3279, pruned_loss=0.1171, over 16892.00 frames. ], tot_loss[loss=0.2683, simple_loss=0.3302, pruned_loss=0.1032, over 3290306.59 frames. ], batch size: 109, lr: 3.14e-02, grad_scale: 8.0 2023-04-27 16:57:10,344 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11110.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 16:57:14,731 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=11113.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:57:21,473 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.464e+02 3.927e+02 4.944e+02 5.832e+02 1.325e+03, threshold=9.889e+02, percent-clipped=4.0 2023-04-27 16:57:38,033 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11130.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:58:06,963 INFO [train.py:904] (6/8) Epoch 2, batch 1000, loss[loss=0.2018, simple_loss=0.2764, pruned_loss=0.06358, over 16996.00 frames. ], tot_loss[loss=0.2682, simple_loss=0.3296, pruned_loss=0.1034, over 3292205.95 frames. ], batch size: 41, lr: 3.14e-02, grad_scale: 8.0 2023-04-27 16:58:16,902 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=11158.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:58:25,145 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2871, 3.7370, 3.1790, 4.7571, 2.8708, 4.4375, 3.2813, 2.7031], device='cuda:6'), covar=tensor([0.0239, 0.0279, 0.0266, 0.0159, 0.1220, 0.0134, 0.0479, 0.1217], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0147, 0.0120, 0.0169, 0.0226, 0.0139, 0.0159, 0.0201], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 16:58:37,538 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11174.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:59:13,979 INFO [train.py:904] (6/8) Epoch 2, batch 1050, loss[loss=0.2571, simple_loss=0.3138, pruned_loss=0.1002, over 16795.00 frames. ], tot_loss[loss=0.2664, simple_loss=0.3282, pruned_loss=0.1023, over 3303248.51 frames. ], batch size: 102, lr: 3.13e-02, grad_scale: 8.0 2023-04-27 16:59:36,186 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.907e+02 3.907e+02 4.719e+02 5.702e+02 1.083e+03, threshold=9.439e+02, percent-clipped=2.0 2023-04-27 17:00:00,099 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11235.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:00:08,460 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11241.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 17:00:21,895 INFO [train.py:904] (6/8) Epoch 2, batch 1100, loss[loss=0.2601, simple_loss=0.3361, pruned_loss=0.09207, over 17135.00 frames. ], tot_loss[loss=0.2659, simple_loss=0.3278, pruned_loss=0.102, over 3306208.72 frames. ], batch size: 47, lr: 3.12e-02, grad_scale: 8.0 2023-04-27 17:00:49,672 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3144, 4.3591, 4.3287, 4.4034, 4.2857, 4.8249, 4.6929, 4.2642], device='cuda:6'), covar=tensor([0.0973, 0.1163, 0.0953, 0.1644, 0.2230, 0.0814, 0.0790, 0.1893], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0268, 0.0232, 0.0233, 0.0290, 0.0235, 0.0201, 0.0293], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:01:28,364 INFO [train.py:904] (6/8) Epoch 2, batch 1150, loss[loss=0.2552, simple_loss=0.3313, pruned_loss=0.08958, over 16650.00 frames. ], tot_loss[loss=0.2637, simple_loss=0.3262, pruned_loss=0.1006, over 3301290.74 frames. ], batch size: 62, lr: 3.12e-02, grad_scale: 8.0 2023-04-27 17:01:52,662 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.647e+02 4.017e+02 4.974e+02 5.619e+02 1.017e+03, threshold=9.949e+02, percent-clipped=1.0 2023-04-27 17:01:55,458 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8965, 4.6065, 4.8505, 5.3479, 5.3085, 4.8684, 5.3637, 5.1047], device='cuda:6'), covar=tensor([0.0408, 0.0526, 0.1034, 0.0279, 0.0366, 0.0369, 0.0262, 0.0317], device='cuda:6'), in_proj_covar=tensor([0.0215, 0.0241, 0.0358, 0.0250, 0.0207, 0.0204, 0.0186, 0.0208], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 17:02:39,330 INFO [train.py:904] (6/8) Epoch 2, batch 1200, loss[loss=0.2236, simple_loss=0.2988, pruned_loss=0.07422, over 16954.00 frames. ], tot_loss[loss=0.2626, simple_loss=0.3254, pruned_loss=0.09989, over 3301214.29 frames. ], batch size: 41, lr: 3.11e-02, grad_scale: 8.0 2023-04-27 17:03:46,803 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11400.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:03:47,486 INFO [train.py:904] (6/8) Epoch 2, batch 1250, loss[loss=0.2854, simple_loss=0.3373, pruned_loss=0.1168, over 17217.00 frames. ], tot_loss[loss=0.265, simple_loss=0.3267, pruned_loss=0.1017, over 3296141.25 frames. ], batch size: 44, lr: 3.11e-02, grad_scale: 8.0 2023-04-27 17:03:54,279 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8860, 4.5709, 4.6928, 4.8113, 4.0960, 4.6191, 4.6007, 4.4031], device='cuda:6'), covar=tensor([0.0283, 0.0194, 0.0196, 0.0154, 0.0862, 0.0207, 0.0258, 0.0236], device='cuda:6'), in_proj_covar=tensor([0.0125, 0.0089, 0.0163, 0.0130, 0.0198, 0.0125, 0.0113, 0.0141], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:04:10,333 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.776e+02 4.393e+02 5.372e+02 6.605e+02 1.355e+03, threshold=1.074e+03, percent-clipped=5.0 2023-04-27 17:04:26,783 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11430.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:04:29,401 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0007, 2.0702, 2.0271, 1.9999, 2.6527, 2.6635, 2.8655, 2.8864], device='cuda:6'), covar=tensor([0.0095, 0.0190, 0.0124, 0.0182, 0.0087, 0.0115, 0.0069, 0.0061], device='cuda:6'), in_proj_covar=tensor([0.0040, 0.0082, 0.0072, 0.0079, 0.0071, 0.0080, 0.0045, 0.0054], device='cuda:6'), out_proj_covar=tensor([6.3026e-05, 1.2626e-04, 1.1194e-04, 1.2419e-04, 1.1284e-04, 1.2843e-04, 7.0223e-05, 9.0512e-05], device='cuda:6') 2023-04-27 17:04:49,877 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=11448.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:04:50,073 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9293, 3.8645, 1.8552, 3.9581, 2.6611, 3.9473, 1.8461, 2.8981], device='cuda:6'), covar=tensor([0.0050, 0.0131, 0.1554, 0.0046, 0.0701, 0.0266, 0.1342, 0.0575], device='cuda:6'), in_proj_covar=tensor([0.0076, 0.0090, 0.0164, 0.0077, 0.0150, 0.0112, 0.0167, 0.0142], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 17:04:53,724 INFO [train.py:904] (6/8) Epoch 2, batch 1300, loss[loss=0.2667, simple_loss=0.343, pruned_loss=0.09515, over 17154.00 frames. ], tot_loss[loss=0.2653, simple_loss=0.3271, pruned_loss=0.1017, over 3308484.69 frames. ], batch size: 47, lr: 3.10e-02, grad_scale: 8.0 2023-04-27 17:05:30,775 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=11478.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:06:01,549 INFO [train.py:904] (6/8) Epoch 2, batch 1350, loss[loss=0.2871, simple_loss=0.3439, pruned_loss=0.1151, over 16887.00 frames. ], tot_loss[loss=0.2635, simple_loss=0.3259, pruned_loss=0.1005, over 3308595.15 frames. ], batch size: 90, lr: 3.10e-02, grad_scale: 8.0 2023-04-27 17:06:23,976 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3651, 5.6855, 5.2552, 5.6172, 4.8184, 4.7727, 5.1568, 5.7974], device='cuda:6'), covar=tensor([0.0416, 0.0684, 0.0815, 0.0325, 0.0594, 0.0429, 0.0465, 0.0488], device='cuda:6'), in_proj_covar=tensor([0.0208, 0.0288, 0.0250, 0.0171, 0.0194, 0.0165, 0.0229, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 17:06:24,753 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.664e+02 4.249e+02 5.319e+02 6.551e+02 1.548e+03, threshold=1.064e+03, percent-clipped=4.0 2023-04-27 17:06:41,816 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11530.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:06:45,044 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11532.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:06:57,555 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11541.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 17:07:09,714 INFO [train.py:904] (6/8) Epoch 2, batch 1400, loss[loss=0.2348, simple_loss=0.3109, pruned_loss=0.07929, over 17200.00 frames. ], tot_loss[loss=0.2633, simple_loss=0.3259, pruned_loss=0.1003, over 3314341.43 frames. ], batch size: 46, lr: 3.09e-02, grad_scale: 8.0 2023-04-27 17:07:44,614 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-27 17:08:03,732 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=11589.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 17:08:09,075 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11593.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:08:19,413 INFO [train.py:904] (6/8) Epoch 2, batch 1450, loss[loss=0.2693, simple_loss=0.3406, pruned_loss=0.09896, over 17122.00 frames. ], tot_loss[loss=0.2642, simple_loss=0.326, pruned_loss=0.1012, over 3319047.66 frames. ], batch size: 49, lr: 3.08e-02, grad_scale: 8.0 2023-04-27 17:08:43,788 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.475e+02 3.985e+02 4.876e+02 5.970e+02 8.943e+02, threshold=9.752e+02, percent-clipped=0.0 2023-04-27 17:09:04,656 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6782, 3.7966, 3.7692, 3.7302, 3.6916, 4.1376, 4.0349, 3.6565], device='cuda:6'), covar=tensor([0.1594, 0.1137, 0.1049, 0.1396, 0.2104, 0.0942, 0.0754, 0.1647], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0263, 0.0237, 0.0230, 0.0288, 0.0238, 0.0206, 0.0299], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:09:06,422 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11635.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:09:26,114 INFO [train.py:904] (6/8) Epoch 2, batch 1500, loss[loss=0.2764, simple_loss=0.3399, pruned_loss=0.1065, over 16563.00 frames. ], tot_loss[loss=0.2641, simple_loss=0.3257, pruned_loss=0.1012, over 3315737.84 frames. ], batch size: 62, lr: 3.08e-02, grad_scale: 8.0 2023-04-27 17:09:33,474 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1923, 4.5527, 3.8699, 3.8615, 3.5005, 2.5232, 5.0852, 5.4687], device='cuda:6'), covar=tensor([0.1525, 0.0457, 0.0803, 0.0367, 0.1705, 0.1241, 0.0149, 0.0034], device='cuda:6'), in_proj_covar=tensor([0.0237, 0.0217, 0.0228, 0.0154, 0.0241, 0.0173, 0.0172, 0.0111], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-27 17:10:09,275 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0146, 3.9761, 2.2021, 4.1330, 2.7969, 4.0548, 2.0404, 2.9713], device='cuda:6'), covar=tensor([0.0057, 0.0130, 0.1204, 0.0036, 0.0654, 0.0248, 0.1237, 0.0514], device='cuda:6'), in_proj_covar=tensor([0.0082, 0.0098, 0.0168, 0.0079, 0.0154, 0.0117, 0.0174, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 17:10:30,131 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11696.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:10:33,619 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11699.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:10:35,627 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4764, 2.7233, 2.4250, 3.7691, 2.3432, 3.5636, 2.3843, 2.3939], device='cuda:6'), covar=tensor([0.0249, 0.0333, 0.0278, 0.0188, 0.1106, 0.0154, 0.0550, 0.1062], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0149, 0.0124, 0.0177, 0.0227, 0.0141, 0.0159, 0.0206], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 17:10:36,155 INFO [train.py:904] (6/8) Epoch 2, batch 1550, loss[loss=0.3257, simple_loss=0.3666, pruned_loss=0.1424, over 16372.00 frames. ], tot_loss[loss=0.2668, simple_loss=0.3275, pruned_loss=0.103, over 3306259.58 frames. ], batch size: 165, lr: 3.07e-02, grad_scale: 8.0 2023-04-27 17:10:58,966 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.956e+02 4.303e+02 5.262e+02 6.472e+02 2.309e+03, threshold=1.052e+03, percent-clipped=9.0 2023-04-27 17:11:07,154 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-27 17:11:27,717 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.57 vs. limit=5.0 2023-04-27 17:11:44,746 INFO [train.py:904] (6/8) Epoch 2, batch 1600, loss[loss=0.3032, simple_loss=0.3491, pruned_loss=0.1286, over 16404.00 frames. ], tot_loss[loss=0.2698, simple_loss=0.3304, pruned_loss=0.1046, over 3309342.60 frames. ], batch size: 146, lr: 3.07e-02, grad_scale: 8.0 2023-04-27 17:11:57,719 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11760.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:12:53,455 INFO [train.py:904] (6/8) Epoch 2, batch 1650, loss[loss=0.2346, simple_loss=0.2984, pruned_loss=0.08541, over 16816.00 frames. ], tot_loss[loss=0.2721, simple_loss=0.3327, pruned_loss=0.1057, over 3303168.10 frames. ], batch size: 39, lr: 3.06e-02, grad_scale: 8.0 2023-04-27 17:13:15,768 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11816.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:13:17,810 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.782e+02 3.949e+02 4.848e+02 6.012e+02 1.253e+03, threshold=9.697e+02, percent-clipped=3.0 2023-04-27 17:13:35,479 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11830.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:13:40,897 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9385, 2.1977, 2.2475, 2.8780, 3.1045, 2.8091, 1.8054, 2.9303], device='cuda:6'), covar=tensor([0.0051, 0.0241, 0.0186, 0.0106, 0.0031, 0.0111, 0.0228, 0.0067], device='cuda:6'), in_proj_covar=tensor([0.0065, 0.0100, 0.0088, 0.0072, 0.0050, 0.0052, 0.0086, 0.0051], device='cuda:6'), out_proj_covar=tensor([1.1768e-04, 1.7981e-04, 1.7062e-04, 1.3621e-04, 8.8714e-05, 1.0004e-04, 1.4920e-04, 9.6698e-05], device='cuda:6') 2023-04-27 17:13:56,980 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11846.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:14:03,720 INFO [train.py:904] (6/8) Epoch 2, batch 1700, loss[loss=0.2325, simple_loss=0.3007, pruned_loss=0.08212, over 17024.00 frames. ], tot_loss[loss=0.2741, simple_loss=0.3352, pruned_loss=0.1065, over 3305619.65 frames. ], batch size: 41, lr: 3.06e-02, grad_scale: 8.0 2023-04-27 17:14:40,655 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11877.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:14:42,251 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=11878.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:14:55,494 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11888.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:15:13,202 INFO [train.py:904] (6/8) Epoch 2, batch 1750, loss[loss=0.2353, simple_loss=0.3157, pruned_loss=0.07746, over 17158.00 frames. ], tot_loss[loss=0.2732, simple_loss=0.3349, pruned_loss=0.1058, over 3314408.00 frames. ], batch size: 46, lr: 3.05e-02, grad_scale: 8.0 2023-04-27 17:15:21,952 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11907.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:15:36,931 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.686e+02 4.328e+02 5.055e+02 5.950e+02 1.641e+03, threshold=1.011e+03, percent-clipped=5.0 2023-04-27 17:15:40,229 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11920.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:16:17,795 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11948.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:16:21,485 INFO [train.py:904] (6/8) Epoch 2, batch 1800, loss[loss=0.2919, simple_loss=0.3485, pruned_loss=0.1176, over 16482.00 frames. ], tot_loss[loss=0.2728, simple_loss=0.3355, pruned_loss=0.1051, over 3317765.95 frames. ], batch size: 75, lr: 3.05e-02, grad_scale: 8.0 2023-04-27 17:17:03,383 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11981.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:17:07,154 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11984.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:17:09,783 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-27 17:17:15,554 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-27 17:17:16,283 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11991.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:17:33,761 INFO [train.py:904] (6/8) Epoch 2, batch 1850, loss[loss=0.3262, simple_loss=0.3786, pruned_loss=0.1369, over 15489.00 frames. ], tot_loss[loss=0.2758, simple_loss=0.338, pruned_loss=0.1068, over 3304539.62 frames. ], batch size: 190, lr: 3.04e-02, grad_scale: 8.0 2023-04-27 17:17:45,593 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12009.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:17:57,887 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.566e+02 4.018e+02 4.936e+02 6.102e+02 1.131e+03, threshold=9.871e+02, percent-clipped=1.0 2023-04-27 17:18:35,118 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12045.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:18:39,397 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4647, 4.3649, 4.4203, 1.6902, 3.1292, 2.5309, 3.8910, 4.3783], device='cuda:6'), covar=tensor([0.0312, 0.0442, 0.0288, 0.1974, 0.0789, 0.1106, 0.0748, 0.0497], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0101, 0.0137, 0.0155, 0.0144, 0.0138, 0.0149, 0.0096], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-27 17:18:43,196 INFO [train.py:904] (6/8) Epoch 2, batch 1900, loss[loss=0.2854, simple_loss=0.3361, pruned_loss=0.1173, over 16348.00 frames. ], tot_loss[loss=0.2726, simple_loss=0.3361, pruned_loss=0.1046, over 3312704.55 frames. ], batch size: 165, lr: 3.04e-02, grad_scale: 8.0 2023-04-27 17:18:49,606 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12055.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:19:17,206 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12076.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:19:51,447 INFO [train.py:904] (6/8) Epoch 2, batch 1950, loss[loss=0.2852, simple_loss=0.34, pruned_loss=0.1152, over 16718.00 frames. ], tot_loss[loss=0.2716, simple_loss=0.3359, pruned_loss=0.1037, over 3311651.93 frames. ], batch size: 124, lr: 3.03e-02, grad_scale: 8.0 2023-04-27 17:20:14,619 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.438e+02 4.259e+02 5.225e+02 6.323e+02 1.109e+03, threshold=1.045e+03, percent-clipped=2.0 2023-04-27 17:20:20,244 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2023-04-27 17:20:40,733 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12137.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:20:59,921 INFO [train.py:904] (6/8) Epoch 2, batch 2000, loss[loss=0.2745, simple_loss=0.3388, pruned_loss=0.1051, over 16522.00 frames. ], tot_loss[loss=0.2716, simple_loss=0.3359, pruned_loss=0.1036, over 3311677.11 frames. ], batch size: 68, lr: 3.02e-02, grad_scale: 8.0 2023-04-27 17:21:15,741 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3277, 4.2964, 4.7795, 4.7717, 4.9116, 4.3634, 4.5344, 4.7223], device='cuda:6'), covar=tensor([0.0299, 0.0319, 0.0414, 0.0426, 0.0321, 0.0295, 0.0610, 0.0230], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0150, 0.0175, 0.0169, 0.0197, 0.0159, 0.0250, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-27 17:21:28,827 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12172.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:21:51,917 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12188.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:22:09,011 INFO [train.py:904] (6/8) Epoch 2, batch 2050, loss[loss=0.2953, simple_loss=0.3373, pruned_loss=0.1267, over 16743.00 frames. ], tot_loss[loss=0.2728, simple_loss=0.3361, pruned_loss=0.1047, over 3320004.22 frames. ], batch size: 124, lr: 3.02e-02, grad_scale: 16.0 2023-04-27 17:22:11,061 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12202.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:22:28,217 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-27 17:22:32,915 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.569e+02 4.228e+02 5.159e+02 6.169e+02 1.128e+03, threshold=1.032e+03, percent-clipped=1.0 2023-04-27 17:22:55,560 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5925, 5.9267, 5.5402, 5.8628, 5.1738, 4.8101, 5.3339, 6.0176], device='cuda:6'), covar=tensor([0.0410, 0.0531, 0.0779, 0.0268, 0.0457, 0.0390, 0.0440, 0.0441], device='cuda:6'), in_proj_covar=tensor([0.0218, 0.0305, 0.0272, 0.0183, 0.0209, 0.0176, 0.0239, 0.0199], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 17:22:58,999 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12236.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:23:19,203 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6175, 2.1250, 2.8480, 3.2400, 3.6571, 3.7771, 2.2319, 3.6784], device='cuda:6'), covar=tensor([0.0040, 0.0263, 0.0127, 0.0089, 0.0073, 0.0076, 0.0202, 0.0046], device='cuda:6'), in_proj_covar=tensor([0.0069, 0.0109, 0.0091, 0.0079, 0.0053, 0.0055, 0.0093, 0.0056], device='cuda:6'), out_proj_covar=tensor([1.2652e-04, 1.9556e-04, 1.7482e-04, 1.5033e-04, 9.4786e-05, 1.0694e-04, 1.6247e-04, 1.0350e-04], device='cuda:6') 2023-04-27 17:23:20,404 INFO [train.py:904] (6/8) Epoch 2, batch 2100, loss[loss=0.2423, simple_loss=0.308, pruned_loss=0.08831, over 15910.00 frames. ], tot_loss[loss=0.2728, simple_loss=0.3364, pruned_loss=0.1046, over 3317490.01 frames. ], batch size: 35, lr: 3.01e-02, grad_scale: 16.0 2023-04-27 17:23:54,373 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12276.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:24:15,622 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12291.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:24:28,733 INFO [train.py:904] (6/8) Epoch 2, batch 2150, loss[loss=0.2471, simple_loss=0.32, pruned_loss=0.0871, over 17168.00 frames. ], tot_loss[loss=0.2738, simple_loss=0.3372, pruned_loss=0.1052, over 3321706.23 frames. ], batch size: 46, lr: 3.01e-02, grad_scale: 16.0 2023-04-27 17:24:33,514 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12304.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:24:51,837 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.440e+02 4.527e+02 5.287e+02 6.148e+02 1.100e+03, threshold=1.057e+03, percent-clipped=4.0 2023-04-27 17:25:00,529 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.94 vs. limit=5.0 2023-04-27 17:25:16,438 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1304, 4.9129, 4.8275, 4.2900, 4.8890, 1.7615, 4.6491, 4.9502], device='cuda:6'), covar=tensor([0.0063, 0.0055, 0.0075, 0.0331, 0.0056, 0.1337, 0.0079, 0.0113], device='cuda:6'), in_proj_covar=tensor([0.0065, 0.0055, 0.0078, 0.0103, 0.0060, 0.0104, 0.0074, 0.0081], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:25:22,226 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12339.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:25:23,436 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12340.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:25:38,418 INFO [train.py:904] (6/8) Epoch 2, batch 2200, loss[loss=0.2396, simple_loss=0.3239, pruned_loss=0.07769, over 17117.00 frames. ], tot_loss[loss=0.2745, simple_loss=0.3377, pruned_loss=0.1056, over 3321483.47 frames. ], batch size: 47, lr: 3.00e-02, grad_scale: 16.0 2023-04-27 17:25:44,415 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12355.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:26:27,527 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2023-04-27 17:26:48,778 INFO [train.py:904] (6/8) Epoch 2, batch 2250, loss[loss=0.2971, simple_loss=0.3417, pruned_loss=0.1262, over 16798.00 frames. ], tot_loss[loss=0.2763, simple_loss=0.3395, pruned_loss=0.1066, over 3317546.25 frames. ], batch size: 135, lr: 3.00e-02, grad_scale: 16.0 2023-04-27 17:26:52,141 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12403.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:27:12,082 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.514e+02 4.351e+02 5.087e+02 6.619e+02 9.905e+02, threshold=1.017e+03, percent-clipped=0.0 2023-04-27 17:27:32,273 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12432.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:27:51,314 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12446.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:27:57,833 INFO [train.py:904] (6/8) Epoch 2, batch 2300, loss[loss=0.2502, simple_loss=0.3107, pruned_loss=0.0949, over 16806.00 frames. ], tot_loss[loss=0.2747, simple_loss=0.3381, pruned_loss=0.1057, over 3318622.61 frames. ], batch size: 96, lr: 2.99e-02, grad_scale: 8.0 2023-04-27 17:28:27,164 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12472.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:29:06,506 INFO [train.py:904] (6/8) Epoch 2, batch 2350, loss[loss=0.2577, simple_loss=0.3373, pruned_loss=0.08907, over 17117.00 frames. ], tot_loss[loss=0.2758, simple_loss=0.3388, pruned_loss=0.1064, over 3320634.31 frames. ], batch size: 47, lr: 2.99e-02, grad_scale: 8.0 2023-04-27 17:29:07,930 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12502.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:29:15,032 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12507.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:29:31,564 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.397e+02 4.650e+02 5.724e+02 7.343e+02 1.924e+03, threshold=1.145e+03, percent-clipped=10.0 2023-04-27 17:29:33,892 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12520.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:29:45,813 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1622, 4.0125, 1.7373, 4.1643, 2.7270, 4.1673, 2.1824, 3.0415], device='cuda:6'), covar=tensor([0.0056, 0.0340, 0.1452, 0.0040, 0.0665, 0.0164, 0.1119, 0.0508], device='cuda:6'), in_proj_covar=tensor([0.0084, 0.0098, 0.0166, 0.0075, 0.0153, 0.0121, 0.0170, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 17:30:14,729 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12550.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:30:15,611 INFO [train.py:904] (6/8) Epoch 2, batch 2400, loss[loss=0.2781, simple_loss=0.3427, pruned_loss=0.1067, over 16301.00 frames. ], tot_loss[loss=0.2792, simple_loss=0.3412, pruned_loss=0.1086, over 3313093.00 frames. ], batch size: 165, lr: 2.98e-02, grad_scale: 8.0 2023-04-27 17:30:52,437 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12576.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:31:18,957 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2023-04-27 17:31:26,499 INFO [train.py:904] (6/8) Epoch 2, batch 2450, loss[loss=0.2897, simple_loss=0.3457, pruned_loss=0.1168, over 16427.00 frames. ], tot_loss[loss=0.2769, simple_loss=0.3402, pruned_loss=0.1068, over 3316503.74 frames. ], batch size: 146, lr: 2.98e-02, grad_scale: 8.0 2023-04-27 17:31:31,802 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12604.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:31:49,331 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0433, 4.4259, 3.6713, 3.8574, 3.3154, 2.3806, 4.8067, 5.3624], device='cuda:6'), covar=tensor([0.1553, 0.0456, 0.0777, 0.0371, 0.1704, 0.1158, 0.0161, 0.0070], device='cuda:6'), in_proj_covar=tensor([0.0245, 0.0226, 0.0229, 0.0160, 0.0256, 0.0177, 0.0174, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 17:31:51,020 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.702e+02 4.143e+02 4.943e+02 6.147e+02 1.117e+03, threshold=9.887e+02, percent-clipped=0.0 2023-04-27 17:31:59,245 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12624.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:32:20,308 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12640.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:32:35,555 INFO [train.py:904] (6/8) Epoch 2, batch 2500, loss[loss=0.2762, simple_loss=0.3298, pruned_loss=0.1113, over 16898.00 frames. ], tot_loss[loss=0.276, simple_loss=0.3403, pruned_loss=0.1058, over 3314328.30 frames. ], batch size: 109, lr: 2.97e-02, grad_scale: 8.0 2023-04-27 17:32:37,514 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12652.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:32:53,248 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2023-04-27 17:33:27,071 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12688.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:33:43,145 INFO [train.py:904] (6/8) Epoch 2, batch 2550, loss[loss=0.2483, simple_loss=0.3245, pruned_loss=0.08606, over 17236.00 frames. ], tot_loss[loss=0.2746, simple_loss=0.3394, pruned_loss=0.1049, over 3315356.86 frames. ], batch size: 52, lr: 2.97e-02, grad_scale: 8.0 2023-04-27 17:34:08,155 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.860e+02 4.439e+02 5.440e+02 6.842e+02 1.130e+03, threshold=1.088e+03, percent-clipped=2.0 2023-04-27 17:34:27,584 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12732.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:34:53,761 INFO [train.py:904] (6/8) Epoch 2, batch 2600, loss[loss=0.2813, simple_loss=0.3364, pruned_loss=0.1131, over 16930.00 frames. ], tot_loss[loss=0.275, simple_loss=0.34, pruned_loss=0.105, over 3306431.71 frames. ], batch size: 96, lr: 2.96e-02, grad_scale: 8.0 2023-04-27 17:35:08,903 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12763.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:35:33,525 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12780.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:36:01,522 INFO [train.py:904] (6/8) Epoch 2, batch 2650, loss[loss=0.28, simple_loss=0.3442, pruned_loss=0.1079, over 16866.00 frames. ], tot_loss[loss=0.275, simple_loss=0.3399, pruned_loss=0.105, over 3313000.93 frames. ], batch size: 83, lr: 2.96e-02, grad_scale: 8.0 2023-04-27 17:36:03,339 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12802.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:36:27,046 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.569e+02 4.221e+02 4.762e+02 5.639e+02 1.002e+03, threshold=9.524e+02, percent-clipped=0.0 2023-04-27 17:36:30,079 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-04-27 17:36:33,961 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12824.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 17:37:09,359 INFO [train.py:904] (6/8) Epoch 2, batch 2700, loss[loss=0.256, simple_loss=0.3289, pruned_loss=0.0916, over 16507.00 frames. ], tot_loss[loss=0.2739, simple_loss=0.3394, pruned_loss=0.1041, over 3318534.84 frames. ], batch size: 68, lr: 2.95e-02, grad_scale: 8.0 2023-04-27 17:37:32,374 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-27 17:38:19,510 INFO [train.py:904] (6/8) Epoch 2, batch 2750, loss[loss=0.3192, simple_loss=0.3719, pruned_loss=0.1333, over 12396.00 frames. ], tot_loss[loss=0.2725, simple_loss=0.3384, pruned_loss=0.1034, over 3319286.14 frames. ], batch size: 246, lr: 2.95e-02, grad_scale: 8.0 2023-04-27 17:38:41,653 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.401e+02 4.125e+02 4.810e+02 5.955e+02 1.093e+03, threshold=9.620e+02, percent-clipped=2.0 2023-04-27 17:38:50,780 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2023-04-27 17:38:55,495 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1564, 4.1440, 3.3246, 3.7651, 3.2155, 2.2198, 4.5256, 5.2197], device='cuda:6'), covar=tensor([0.1613, 0.0571, 0.0951, 0.0445, 0.1928, 0.1391, 0.0241, 0.0071], device='cuda:6'), in_proj_covar=tensor([0.0235, 0.0220, 0.0223, 0.0156, 0.0246, 0.0175, 0.0178, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 17:39:26,262 INFO [train.py:904] (6/8) Epoch 2, batch 2800, loss[loss=0.2896, simple_loss=0.3579, pruned_loss=0.1106, over 17007.00 frames. ], tot_loss[loss=0.2731, simple_loss=0.3391, pruned_loss=0.1036, over 3327355.11 frames. ], batch size: 53, lr: 2.94e-02, grad_scale: 8.0 2023-04-27 17:39:49,833 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-27 17:40:33,619 INFO [train.py:904] (6/8) Epoch 2, batch 2850, loss[loss=0.3103, simple_loss=0.3625, pruned_loss=0.129, over 16831.00 frames. ], tot_loss[loss=0.2726, simple_loss=0.3381, pruned_loss=0.1036, over 3328621.97 frames. ], batch size: 116, lr: 2.94e-02, grad_scale: 8.0 2023-04-27 17:40:57,333 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.754e+02 4.789e+02 5.762e+02 6.825e+02 1.913e+03, threshold=1.152e+03, percent-clipped=8.0 2023-04-27 17:41:41,131 INFO [train.py:904] (6/8) Epoch 2, batch 2900, loss[loss=0.2388, simple_loss=0.3125, pruned_loss=0.08252, over 17250.00 frames. ], tot_loss[loss=0.2718, simple_loss=0.336, pruned_loss=0.1038, over 3331891.58 frames. ], batch size: 45, lr: 2.93e-02, grad_scale: 8.0 2023-04-27 17:41:51,121 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9475, 5.7020, 5.5187, 5.4993, 5.4981, 6.0254, 5.8936, 5.6116], device='cuda:6'), covar=tensor([0.0560, 0.1025, 0.0918, 0.1501, 0.2284, 0.0689, 0.0810, 0.1639], device='cuda:6'), in_proj_covar=tensor([0.0183, 0.0267, 0.0238, 0.0232, 0.0301, 0.0242, 0.0210, 0.0300], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:42:49,006 INFO [train.py:904] (6/8) Epoch 2, batch 2950, loss[loss=0.3061, simple_loss=0.366, pruned_loss=0.1231, over 16459.00 frames. ], tot_loss[loss=0.2718, simple_loss=0.335, pruned_loss=0.1043, over 3334949.96 frames. ], batch size: 75, lr: 2.93e-02, grad_scale: 8.0 2023-04-27 17:42:50,413 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=13102.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:43:04,454 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-27 17:43:13,803 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.070e+02 4.352e+02 4.979e+02 5.894e+02 1.092e+03, threshold=9.958e+02, percent-clipped=0.0 2023-04-27 17:43:14,144 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=13119.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 17:43:38,976 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7954, 3.1477, 2.1533, 3.8456, 3.7478, 3.7773, 1.6943, 2.9556], device='cuda:6'), covar=tensor([0.1525, 0.0353, 0.1456, 0.0087, 0.0191, 0.0249, 0.1138, 0.0537], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0112, 0.0168, 0.0070, 0.0106, 0.0110, 0.0150, 0.0139], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 17:43:53,709 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=13150.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:43:54,563 INFO [train.py:904] (6/8) Epoch 2, batch 3000, loss[loss=0.2737, simple_loss=0.3349, pruned_loss=0.1062, over 16835.00 frames. ], tot_loss[loss=0.2716, simple_loss=0.3346, pruned_loss=0.1043, over 3333720.55 frames. ], batch size: 102, lr: 2.92e-02, grad_scale: 8.0 2023-04-27 17:43:54,563 INFO [train.py:929] (6/8) Computing validation loss 2023-04-27 17:44:03,915 INFO [train.py:938] (6/8) Epoch 2, validation: loss=0.1858, simple_loss=0.2917, pruned_loss=0.04, over 944034.00 frames. 2023-04-27 17:44:03,915 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17391MB 2023-04-27 17:45:05,142 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4546, 5.2861, 5.1950, 5.2769, 4.7476, 5.2098, 5.1949, 4.8346], device='cuda:6'), covar=tensor([0.0226, 0.0113, 0.0170, 0.0100, 0.0641, 0.0183, 0.0107, 0.0180], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0094, 0.0168, 0.0135, 0.0206, 0.0139, 0.0118, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:45:09,406 INFO [train.py:904] (6/8) Epoch 2, batch 3050, loss[loss=0.2407, simple_loss=0.3217, pruned_loss=0.07988, over 17119.00 frames. ], tot_loss[loss=0.27, simple_loss=0.3335, pruned_loss=0.1033, over 3326191.84 frames. ], batch size: 49, lr: 2.92e-02, grad_scale: 8.0 2023-04-27 17:45:33,140 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.792e+02 4.585e+02 5.657e+02 6.871e+02 1.163e+03, threshold=1.131e+03, percent-clipped=2.0 2023-04-27 17:46:15,657 INFO [train.py:904] (6/8) Epoch 2, batch 3100, loss[loss=0.2388, simple_loss=0.3192, pruned_loss=0.07918, over 17123.00 frames. ], tot_loss[loss=0.2702, simple_loss=0.3337, pruned_loss=0.1034, over 3323569.80 frames. ], batch size: 47, lr: 2.91e-02, grad_scale: 8.0 2023-04-27 17:46:52,719 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3666, 4.1544, 4.3457, 4.7486, 4.7757, 4.3664, 4.5485, 4.6321], device='cuda:6'), covar=tensor([0.0550, 0.0592, 0.1170, 0.0327, 0.0425, 0.0605, 0.0558, 0.0354], device='cuda:6'), in_proj_covar=tensor([0.0243, 0.0286, 0.0408, 0.0293, 0.0229, 0.0225, 0.0216, 0.0232], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:47:22,151 INFO [train.py:904] (6/8) Epoch 2, batch 3150, loss[loss=0.2756, simple_loss=0.3525, pruned_loss=0.09937, over 16681.00 frames. ], tot_loss[loss=0.2697, simple_loss=0.3332, pruned_loss=0.1031, over 3331485.42 frames. ], batch size: 57, lr: 2.91e-02, grad_scale: 8.0 2023-04-27 17:47:27,194 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0248, 4.7108, 5.0179, 5.3388, 5.4139, 4.7554, 5.4163, 5.3093], device='cuda:6'), covar=tensor([0.0452, 0.0528, 0.1085, 0.0332, 0.0287, 0.0455, 0.0241, 0.0273], device='cuda:6'), in_proj_covar=tensor([0.0238, 0.0282, 0.0403, 0.0292, 0.0228, 0.0223, 0.0215, 0.0230], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:47:44,588 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.682e+02 3.986e+02 4.862e+02 6.258e+02 1.253e+03, threshold=9.723e+02, percent-clipped=2.0 2023-04-27 17:48:02,007 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2023-04-27 17:48:07,501 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.63 vs. limit=5.0 2023-04-27 17:48:28,283 INFO [train.py:904] (6/8) Epoch 2, batch 3200, loss[loss=0.2741, simple_loss=0.347, pruned_loss=0.1007, over 16645.00 frames. ], tot_loss[loss=0.2665, simple_loss=0.3316, pruned_loss=0.1007, over 3337036.32 frames. ], batch size: 62, lr: 2.90e-02, grad_scale: 8.0 2023-04-27 17:49:34,227 INFO [train.py:904] (6/8) Epoch 2, batch 3250, loss[loss=0.2451, simple_loss=0.3161, pruned_loss=0.08704, over 15823.00 frames. ], tot_loss[loss=0.2679, simple_loss=0.3327, pruned_loss=0.1015, over 3333865.54 frames. ], batch size: 35, lr: 2.90e-02, grad_scale: 8.0 2023-04-27 17:49:50,650 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0772, 4.0387, 4.1776, 3.6412, 4.1454, 3.9990, 4.6108, 2.6265], device='cuda:6'), covar=tensor([0.0758, 0.0052, 0.0067, 0.0264, 0.0047, 0.0166, 0.0038, 0.0486], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0048, 0.0057, 0.0100, 0.0052, 0.0057, 0.0059, 0.0101], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:49:58,154 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.960e+02 4.105e+02 4.918e+02 6.860e+02 1.329e+03, threshold=9.835e+02, percent-clipped=3.0 2023-04-27 17:49:58,536 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=13419.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:50:30,064 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9306, 3.7417, 3.9230, 4.2284, 4.2540, 3.9136, 3.9866, 4.1539], device='cuda:6'), covar=tensor([0.0439, 0.0518, 0.0974, 0.0336, 0.0394, 0.0669, 0.0708, 0.0345], device='cuda:6'), in_proj_covar=tensor([0.0240, 0.0285, 0.0406, 0.0294, 0.0232, 0.0227, 0.0218, 0.0234], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:50:42,664 INFO [train.py:904] (6/8) Epoch 2, batch 3300, loss[loss=0.3256, simple_loss=0.3727, pruned_loss=0.1392, over 16736.00 frames. ], tot_loss[loss=0.2699, simple_loss=0.3346, pruned_loss=0.1026, over 3333583.73 frames. ], batch size: 124, lr: 2.89e-02, grad_scale: 8.0 2023-04-27 17:51:03,262 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=13467.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:51:48,092 INFO [train.py:904] (6/8) Epoch 2, batch 3350, loss[loss=0.3173, simple_loss=0.3728, pruned_loss=0.131, over 16485.00 frames. ], tot_loss[loss=0.2688, simple_loss=0.3341, pruned_loss=0.1018, over 3334597.13 frames. ], batch size: 146, lr: 2.89e-02, grad_scale: 8.0 2023-04-27 17:52:13,283 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.954e+02 4.311e+02 5.188e+02 6.478e+02 1.182e+03, threshold=1.038e+03, percent-clipped=6.0 2023-04-27 17:52:52,325 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5327, 4.8660, 4.7501, 4.8966, 4.7200, 5.2445, 5.0351, 4.7888], device='cuda:6'), covar=tensor([0.1014, 0.1125, 0.0939, 0.1300, 0.2316, 0.0725, 0.0716, 0.1987], device='cuda:6'), in_proj_covar=tensor([0.0200, 0.0288, 0.0262, 0.0245, 0.0327, 0.0270, 0.0216, 0.0325], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:52:56,206 INFO [train.py:904] (6/8) Epoch 2, batch 3400, loss[loss=0.2928, simple_loss=0.3394, pruned_loss=0.1231, over 16477.00 frames. ], tot_loss[loss=0.2688, simple_loss=0.3336, pruned_loss=0.102, over 3333716.34 frames. ], batch size: 146, lr: 2.88e-02, grad_scale: 8.0 2023-04-27 17:53:01,661 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2926, 3.2798, 2.6636, 4.4686, 2.5108, 4.4840, 2.8692, 2.5463], device='cuda:6'), covar=tensor([0.0209, 0.0352, 0.0324, 0.0175, 0.1219, 0.0128, 0.0566, 0.1427], device='cuda:6'), in_proj_covar=tensor([0.0190, 0.0164, 0.0137, 0.0194, 0.0243, 0.0152, 0.0171, 0.0224], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:53:35,151 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0538, 3.1336, 2.7507, 4.5166, 2.3444, 4.2147, 2.7164, 2.5944], device='cuda:6'), covar=tensor([0.0256, 0.0359, 0.0287, 0.0156, 0.1356, 0.0160, 0.0582, 0.1267], device='cuda:6'), in_proj_covar=tensor([0.0189, 0.0161, 0.0135, 0.0191, 0.0241, 0.0151, 0.0168, 0.0221], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:54:05,265 INFO [train.py:904] (6/8) Epoch 2, batch 3450, loss[loss=0.337, simple_loss=0.376, pruned_loss=0.1489, over 16442.00 frames. ], tot_loss[loss=0.2688, simple_loss=0.3331, pruned_loss=0.1022, over 3323575.50 frames. ], batch size: 146, lr: 2.88e-02, grad_scale: 8.0 2023-04-27 17:54:14,326 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1764, 3.3848, 3.2419, 1.6136, 3.4725, 3.5163, 3.0315, 2.9231], device='cuda:6'), covar=tensor([0.0771, 0.0112, 0.0257, 0.1522, 0.0109, 0.0074, 0.0272, 0.0318], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0083, 0.0080, 0.0149, 0.0075, 0.0071, 0.0099, 0.0113], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-27 17:54:29,793 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.037e+02 4.547e+02 5.342e+02 6.587e+02 1.202e+03, threshold=1.068e+03, percent-clipped=2.0 2023-04-27 17:54:36,193 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.18 vs. limit=2.0 2023-04-27 17:55:11,512 INFO [train.py:904] (6/8) Epoch 2, batch 3500, loss[loss=0.2951, simple_loss=0.3523, pruned_loss=0.119, over 15579.00 frames. ], tot_loss[loss=0.2667, simple_loss=0.3311, pruned_loss=0.1011, over 3311770.14 frames. ], batch size: 191, lr: 2.88e-02, grad_scale: 8.0 2023-04-27 17:56:21,450 INFO [train.py:904] (6/8) Epoch 2, batch 3550, loss[loss=0.2557, simple_loss=0.325, pruned_loss=0.09324, over 16495.00 frames. ], tot_loss[loss=0.2661, simple_loss=0.3302, pruned_loss=0.101, over 3307575.70 frames. ], batch size: 75, lr: 2.87e-02, grad_scale: 8.0 2023-04-27 17:56:45,024 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.316e+02 4.227e+02 5.059e+02 5.919e+02 1.365e+03, threshold=1.012e+03, percent-clipped=3.0 2023-04-27 17:57:11,525 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=13737.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:57:19,107 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.64 vs. limit=5.0 2023-04-27 17:57:29,798 INFO [train.py:904] (6/8) Epoch 2, batch 3600, loss[loss=0.328, simple_loss=0.3645, pruned_loss=0.1457, over 16845.00 frames. ], tot_loss[loss=0.2652, simple_loss=0.3293, pruned_loss=0.1006, over 3308288.31 frames. ], batch size: 116, lr: 2.87e-02, grad_scale: 8.0 2023-04-27 17:57:34,339 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-27 17:57:40,117 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-04-27 17:58:37,338 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=13798.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 17:58:40,990 INFO [train.py:904] (6/8) Epoch 2, batch 3650, loss[loss=0.2526, simple_loss=0.3348, pruned_loss=0.08518, over 17014.00 frames. ], tot_loss[loss=0.2647, simple_loss=0.3279, pruned_loss=0.1007, over 3313911.24 frames. ], batch size: 55, lr: 2.86e-02, grad_scale: 8.0 2023-04-27 17:58:57,801 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6522, 5.0225, 5.0502, 4.9854, 4.8278, 5.3956, 5.2395, 4.8312], device='cuda:6'), covar=tensor([0.0765, 0.1047, 0.1001, 0.1289, 0.1955, 0.0779, 0.0786, 0.1659], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0271, 0.0246, 0.0230, 0.0293, 0.0253, 0.0200, 0.0300], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 17:59:08,122 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.959e+02 4.047e+02 4.833e+02 5.657e+02 1.025e+03, threshold=9.667e+02, percent-clipped=2.0 2023-04-27 17:59:46,284 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-27 17:59:54,876 INFO [train.py:904] (6/8) Epoch 2, batch 3700, loss[loss=0.2789, simple_loss=0.3263, pruned_loss=0.1158, over 11263.00 frames. ], tot_loss[loss=0.265, simple_loss=0.3262, pruned_loss=0.1018, over 3294938.93 frames. ], batch size: 246, lr: 2.86e-02, grad_scale: 8.0 2023-04-27 18:00:21,536 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4317, 3.7501, 3.5227, 1.7039, 3.6659, 3.7621, 3.1326, 3.1675], device='cuda:6'), covar=tensor([0.0643, 0.0086, 0.0241, 0.1372, 0.0127, 0.0087, 0.0293, 0.0243], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0081, 0.0077, 0.0147, 0.0075, 0.0069, 0.0100, 0.0111], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-27 18:00:23,949 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4729, 2.7054, 2.3645, 3.7154, 2.0643, 3.4792, 2.3448, 2.1934], device='cuda:6'), covar=tensor([0.0298, 0.0382, 0.0336, 0.0215, 0.1419, 0.0177, 0.0655, 0.1167], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0166, 0.0138, 0.0198, 0.0246, 0.0154, 0.0173, 0.0226], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 18:01:04,703 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0371, 1.4124, 1.9537, 1.9924, 2.2877, 2.1934, 1.6267, 1.9399], device='cuda:6'), covar=tensor([0.0069, 0.0312, 0.0119, 0.0155, 0.0044, 0.0087, 0.0193, 0.0071], device='cuda:6'), in_proj_covar=tensor([0.0071, 0.0112, 0.0093, 0.0084, 0.0057, 0.0060, 0.0093, 0.0056], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-27 18:01:07,709 INFO [train.py:904] (6/8) Epoch 2, batch 3750, loss[loss=0.2814, simple_loss=0.3289, pruned_loss=0.117, over 16665.00 frames. ], tot_loss[loss=0.2672, simple_loss=0.327, pruned_loss=0.1037, over 3287814.07 frames. ], batch size: 134, lr: 2.85e-02, grad_scale: 8.0 2023-04-27 18:01:33,977 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.812e+02 4.240e+02 4.945e+02 6.119e+02 1.020e+03, threshold=9.889e+02, percent-clipped=4.0 2023-04-27 18:02:24,605 INFO [train.py:904] (6/8) Epoch 2, batch 3800, loss[loss=0.3128, simple_loss=0.354, pruned_loss=0.1358, over 16869.00 frames. ], tot_loss[loss=0.2694, simple_loss=0.328, pruned_loss=0.1054, over 3274511.47 frames. ], batch size: 116, lr: 2.85e-02, grad_scale: 8.0 2023-04-27 18:02:55,516 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0499, 4.1493, 1.7203, 4.1232, 2.8741, 4.1247, 2.2161, 2.8470], device='cuda:6'), covar=tensor([0.0062, 0.0149, 0.1738, 0.0062, 0.0616, 0.0191, 0.1321, 0.0606], device='cuda:6'), in_proj_covar=tensor([0.0078, 0.0099, 0.0170, 0.0078, 0.0152, 0.0127, 0.0171, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 18:03:28,892 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2023-04-27 18:03:40,205 INFO [train.py:904] (6/8) Epoch 2, batch 3850, loss[loss=0.2631, simple_loss=0.3064, pruned_loss=0.1099, over 16886.00 frames. ], tot_loss[loss=0.2688, simple_loss=0.3272, pruned_loss=0.1052, over 3265653.16 frames. ], batch size: 90, lr: 2.84e-02, grad_scale: 8.0 2023-04-27 18:04:07,044 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.317e+02 4.007e+02 4.656e+02 5.451e+02 1.276e+03, threshold=9.312e+02, percent-clipped=3.0 2023-04-27 18:04:08,794 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3416, 4.2182, 4.2627, 3.3264, 4.2048, 1.8835, 4.0440, 4.2817], device='cuda:6'), covar=tensor([0.0124, 0.0080, 0.0090, 0.0508, 0.0079, 0.1621, 0.0102, 0.0154], device='cuda:6'), in_proj_covar=tensor([0.0064, 0.0056, 0.0081, 0.0103, 0.0063, 0.0106, 0.0076, 0.0080], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 18:04:53,515 INFO [train.py:904] (6/8) Epoch 2, batch 3900, loss[loss=0.2672, simple_loss=0.326, pruned_loss=0.1042, over 15480.00 frames. ], tot_loss[loss=0.2669, simple_loss=0.3255, pruned_loss=0.1042, over 3269777.25 frames. ], batch size: 190, lr: 2.84e-02, grad_scale: 8.0 2023-04-27 18:05:14,545 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4037, 4.2271, 4.2960, 3.8429, 4.2763, 2.1729, 4.0918, 4.3590], device='cuda:6'), covar=tensor([0.0077, 0.0062, 0.0065, 0.0280, 0.0062, 0.1052, 0.0077, 0.0082], device='cuda:6'), in_proj_covar=tensor([0.0064, 0.0056, 0.0080, 0.0101, 0.0064, 0.0106, 0.0076, 0.0080], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 18:05:16,991 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-27 18:05:25,179 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14072.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:05:55,874 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14093.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 18:06:03,959 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-27 18:06:05,397 INFO [train.py:904] (6/8) Epoch 2, batch 3950, loss[loss=0.2544, simple_loss=0.3077, pruned_loss=0.1005, over 16407.00 frames. ], tot_loss[loss=0.2653, simple_loss=0.3238, pruned_loss=0.1034, over 3272839.89 frames. ], batch size: 75, lr: 2.83e-02, grad_scale: 8.0 2023-04-27 18:06:07,717 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1486, 5.1733, 4.8908, 4.3854, 4.9831, 2.6706, 4.7460, 5.0978], device='cuda:6'), covar=tensor([0.0064, 0.0048, 0.0064, 0.0289, 0.0049, 0.0995, 0.0072, 0.0074], device='cuda:6'), in_proj_covar=tensor([0.0064, 0.0057, 0.0080, 0.0101, 0.0064, 0.0106, 0.0076, 0.0080], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 18:06:09,451 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14103.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:06:31,641 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.483e+02 3.936e+02 4.888e+02 5.915e+02 1.002e+03, threshold=9.776e+02, percent-clipped=1.0 2023-04-27 18:06:34,549 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14120.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 18:06:52,549 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14133.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:06:53,879 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-27 18:07:16,933 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4168, 3.3476, 3.3810, 3.1056, 3.4094, 2.1670, 3.2554, 3.1627], device='cuda:6'), covar=tensor([0.0061, 0.0058, 0.0079, 0.0211, 0.0054, 0.0916, 0.0080, 0.0095], device='cuda:6'), in_proj_covar=tensor([0.0064, 0.0057, 0.0080, 0.0100, 0.0064, 0.0105, 0.0076, 0.0080], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 18:07:17,634 INFO [train.py:904] (6/8) Epoch 2, batch 4000, loss[loss=0.2945, simple_loss=0.3544, pruned_loss=0.1173, over 12157.00 frames. ], tot_loss[loss=0.2644, simple_loss=0.3232, pruned_loss=0.1028, over 3265249.38 frames. ], batch size: 246, lr: 2.83e-02, grad_scale: 8.0 2023-04-27 18:07:37,780 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14164.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:08:02,151 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14181.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 18:08:31,014 INFO [train.py:904] (6/8) Epoch 2, batch 4050, loss[loss=0.236, simple_loss=0.3069, pruned_loss=0.08255, over 16802.00 frames. ], tot_loss[loss=0.2602, simple_loss=0.3213, pruned_loss=0.09956, over 3260425.67 frames. ], batch size: 39, lr: 2.83e-02, grad_scale: 8.0 2023-04-27 18:08:56,314 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.111e+02 3.521e+02 4.461e+02 5.736e+02 1.012e+03, threshold=8.922e+02, percent-clipped=2.0 2023-04-27 18:09:43,227 INFO [train.py:904] (6/8) Epoch 2, batch 4100, loss[loss=0.2741, simple_loss=0.3443, pruned_loss=0.1019, over 16584.00 frames. ], tot_loss[loss=0.2586, simple_loss=0.3215, pruned_loss=0.09781, over 3265704.22 frames. ], batch size: 62, lr: 2.82e-02, grad_scale: 8.0 2023-04-27 18:10:26,369 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-27 18:10:59,418 INFO [train.py:904] (6/8) Epoch 2, batch 4150, loss[loss=0.3079, simple_loss=0.3765, pruned_loss=0.1196, over 15256.00 frames. ], tot_loss[loss=0.2685, simple_loss=0.331, pruned_loss=0.103, over 3224765.23 frames. ], batch size: 191, lr: 2.82e-02, grad_scale: 8.0 2023-04-27 18:11:25,253 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.427e+02 3.836e+02 4.498e+02 5.328e+02 1.203e+03, threshold=8.995e+02, percent-clipped=3.0 2023-04-27 18:12:14,210 INFO [train.py:904] (6/8) Epoch 2, batch 4200, loss[loss=0.2817, simple_loss=0.3626, pruned_loss=0.1004, over 17113.00 frames. ], tot_loss[loss=0.2762, simple_loss=0.3397, pruned_loss=0.1064, over 3182667.51 frames. ], batch size: 48, lr: 2.81e-02, grad_scale: 8.0 2023-04-27 18:13:15,603 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14393.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:13:26,254 INFO [train.py:904] (6/8) Epoch 2, batch 4250, loss[loss=0.2964, simple_loss=0.3475, pruned_loss=0.1226, over 12038.00 frames. ], tot_loss[loss=0.2791, simple_loss=0.3438, pruned_loss=0.1073, over 3166596.79 frames. ], batch size: 246, lr: 2.81e-02, grad_scale: 8.0 2023-04-27 18:13:44,219 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14413.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:13:52,833 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.326e+02 4.073e+02 4.856e+02 5.593e+02 1.302e+03, threshold=9.713e+02, percent-clipped=3.0 2023-04-27 18:14:05,679 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14428.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:14:24,090 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=14441.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:14:38,545 INFO [train.py:904] (6/8) Epoch 2, batch 4300, loss[loss=0.2886, simple_loss=0.3651, pruned_loss=0.1061, over 16734.00 frames. ], tot_loss[loss=0.2766, simple_loss=0.3433, pruned_loss=0.105, over 3166112.14 frames. ], batch size: 89, lr: 2.80e-02, grad_scale: 16.0 2023-04-27 18:14:49,275 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14459.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:15:10,414 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14474.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:15:12,773 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14476.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 18:15:29,932 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14487.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:15:49,381 INFO [train.py:904] (6/8) Epoch 2, batch 4350, loss[loss=0.3243, simple_loss=0.3797, pruned_loss=0.1345, over 12290.00 frames. ], tot_loss[loss=0.2806, simple_loss=0.3477, pruned_loss=0.1068, over 3178844.78 frames. ], batch size: 246, lr: 2.80e-02, grad_scale: 16.0 2023-04-27 18:16:13,768 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.262e+02 4.011e+02 4.650e+02 6.201e+02 1.293e+03, threshold=9.301e+02, percent-clipped=5.0 2023-04-27 18:16:56,172 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14548.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:16:59,398 INFO [train.py:904] (6/8) Epoch 2, batch 4400, loss[loss=0.2521, simple_loss=0.3336, pruned_loss=0.08535, over 16709.00 frames. ], tot_loss[loss=0.2829, simple_loss=0.3499, pruned_loss=0.108, over 3166627.43 frames. ], batch size: 83, lr: 2.79e-02, grad_scale: 16.0 2023-04-27 18:18:02,571 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3829, 4.7415, 4.6412, 4.7098, 4.7678, 5.2001, 4.9319, 4.7087], device='cuda:6'), covar=tensor([0.0629, 0.0868, 0.0651, 0.1003, 0.1483, 0.0586, 0.0500, 0.1370], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0245, 0.0222, 0.0217, 0.0275, 0.0240, 0.0184, 0.0286], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 18:18:09,487 INFO [train.py:904] (6/8) Epoch 2, batch 4450, loss[loss=0.2759, simple_loss=0.3467, pruned_loss=0.1026, over 16392.00 frames. ], tot_loss[loss=0.2833, simple_loss=0.352, pruned_loss=0.1073, over 3186822.69 frames. ], batch size: 35, lr: 2.79e-02, grad_scale: 16.0 2023-04-27 18:18:34,422 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.338e+02 3.415e+02 4.231e+02 5.024e+02 9.561e+02, threshold=8.461e+02, percent-clipped=1.0 2023-04-27 18:18:45,715 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-27 18:19:17,771 INFO [train.py:904] (6/8) Epoch 2, batch 4500, loss[loss=0.2698, simple_loss=0.3383, pruned_loss=0.1007, over 17238.00 frames. ], tot_loss[loss=0.28, simple_loss=0.3496, pruned_loss=0.1052, over 3196983.08 frames. ], batch size: 52, lr: 2.79e-02, grad_scale: 16.0 2023-04-27 18:19:44,121 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0793, 3.7772, 3.5302, 1.6838, 2.8090, 2.2557, 3.4635, 4.0996], device='cuda:6'), covar=tensor([0.0334, 0.0444, 0.0453, 0.2019, 0.0891, 0.1158, 0.0829, 0.0324], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0105, 0.0149, 0.0158, 0.0148, 0.0139, 0.0153, 0.0095], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-27 18:20:20,479 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-27 18:20:25,962 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2023-04-27 18:20:28,462 INFO [train.py:904] (6/8) Epoch 2, batch 4550, loss[loss=0.3089, simple_loss=0.3697, pruned_loss=0.1241, over 16350.00 frames. ], tot_loss[loss=0.28, simple_loss=0.3495, pruned_loss=0.1053, over 3205461.64 frames. ], batch size: 35, lr: 2.78e-02, grad_scale: 16.0 2023-04-27 18:20:54,773 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.252e+02 3.646e+02 4.280e+02 5.239e+02 1.001e+03, threshold=8.560e+02, percent-clipped=1.0 2023-04-27 18:21:08,878 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14728.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:21:25,078 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14740.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:21:28,208 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5450, 4.3328, 3.5337, 1.3825, 2.8243, 2.3441, 3.5787, 4.5801], device='cuda:6'), covar=tensor([0.0217, 0.0358, 0.0527, 0.2303, 0.0902, 0.1264, 0.0715, 0.0324], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0104, 0.0147, 0.0155, 0.0145, 0.0136, 0.0149, 0.0093], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-27 18:21:43,175 INFO [train.py:904] (6/8) Epoch 2, batch 4600, loss[loss=0.2375, simple_loss=0.323, pruned_loss=0.07595, over 16214.00 frames. ], tot_loss[loss=0.2783, simple_loss=0.3489, pruned_loss=0.1038, over 3211204.06 frames. ], batch size: 35, lr: 2.78e-02, grad_scale: 16.0 2023-04-27 18:21:53,314 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14759.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:22:10,203 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14769.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:22:18,597 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=14776.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:22:18,718 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14776.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 18:22:35,894 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-04-27 18:22:44,292 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9187, 3.2696, 2.6398, 4.4563, 2.3954, 4.3097, 2.9058, 2.5365], device='cuda:6'), covar=tensor([0.0242, 0.0321, 0.0316, 0.0135, 0.1356, 0.0122, 0.0500, 0.1198], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0168, 0.0141, 0.0196, 0.0255, 0.0155, 0.0175, 0.0236], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 18:22:54,478 INFO [train.py:904] (6/8) Epoch 2, batch 4650, loss[loss=0.2742, simple_loss=0.3472, pruned_loss=0.1006, over 16765.00 frames. ], tot_loss[loss=0.2765, simple_loss=0.347, pruned_loss=0.103, over 3220644.56 frames. ], batch size: 39, lr: 2.77e-02, grad_scale: 16.0 2023-04-27 18:22:55,600 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14801.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:23:03,204 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=14807.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:23:20,765 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.700e+02 3.221e+02 3.729e+02 4.286e+02 9.218e+02, threshold=7.459e+02, percent-clipped=2.0 2023-04-27 18:23:27,861 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=14824.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 18:23:54,899 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14843.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:24:06,745 INFO [train.py:904] (6/8) Epoch 2, batch 4700, loss[loss=0.2809, simple_loss=0.3504, pruned_loss=0.1057, over 16672.00 frames. ], tot_loss[loss=0.2742, simple_loss=0.3446, pruned_loss=0.1019, over 3204317.23 frames. ], batch size: 124, lr: 2.77e-02, grad_scale: 16.0 2023-04-27 18:24:37,682 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3073, 1.5059, 1.9187, 2.4698, 2.5057, 2.5568, 1.5182, 2.4526], device='cuda:6'), covar=tensor([0.0035, 0.0232, 0.0095, 0.0073, 0.0029, 0.0057, 0.0179, 0.0031], device='cuda:6'), in_proj_covar=tensor([0.0066, 0.0104, 0.0086, 0.0076, 0.0053, 0.0055, 0.0089, 0.0051], device='cuda:6'), out_proj_covar=tensor([1.1980e-04, 1.8777e-04, 1.6150e-04, 1.4252e-04, 9.3257e-05, 1.0085e-04, 1.5516e-04, 9.3557e-05], device='cuda:6') 2023-04-27 18:24:37,745 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3444, 2.5927, 2.1246, 3.7353, 1.9978, 3.5538, 2.3612, 2.2772], device='cuda:6'), covar=tensor([0.0296, 0.0419, 0.0366, 0.0197, 0.1446, 0.0189, 0.0650, 0.1175], device='cuda:6'), in_proj_covar=tensor([0.0201, 0.0173, 0.0144, 0.0200, 0.0258, 0.0159, 0.0178, 0.0239], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 18:25:19,068 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2191, 1.6457, 2.4819, 3.2412, 3.2520, 3.1466, 1.5499, 3.2447], device='cuda:6'), covar=tensor([0.0027, 0.0277, 0.0113, 0.0054, 0.0028, 0.0068, 0.0219, 0.0031], device='cuda:6'), in_proj_covar=tensor([0.0068, 0.0108, 0.0088, 0.0076, 0.0055, 0.0056, 0.0090, 0.0052], device='cuda:6'), out_proj_covar=tensor([1.2369e-04, 1.9403e-04, 1.6428e-04, 1.4304e-04, 9.6849e-05, 1.0396e-04, 1.5663e-04, 9.4605e-05], device='cuda:6') 2023-04-27 18:25:20,206 INFO [train.py:904] (6/8) Epoch 2, batch 4750, loss[loss=0.2423, simple_loss=0.3117, pruned_loss=0.08647, over 16693.00 frames. ], tot_loss[loss=0.2698, simple_loss=0.3404, pruned_loss=0.09962, over 3215929.98 frames. ], batch size: 62, lr: 2.77e-02, grad_scale: 16.0 2023-04-27 18:25:45,829 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.318e+02 3.402e+02 4.045e+02 5.011e+02 1.217e+03, threshold=8.089e+02, percent-clipped=2.0 2023-04-27 18:26:24,196 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-04-27 18:26:30,650 INFO [train.py:904] (6/8) Epoch 2, batch 4800, loss[loss=0.2487, simple_loss=0.3219, pruned_loss=0.08774, over 16446.00 frames. ], tot_loss[loss=0.2659, simple_loss=0.3365, pruned_loss=0.0977, over 3202883.29 frames. ], batch size: 68, lr: 2.76e-02, grad_scale: 16.0 2023-04-27 18:26:59,290 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6920, 4.3728, 4.3553, 4.5230, 3.8801, 4.3879, 4.2820, 4.1324], device='cuda:6'), covar=tensor([0.0192, 0.0114, 0.0171, 0.0111, 0.0732, 0.0144, 0.0196, 0.0202], device='cuda:6'), in_proj_covar=tensor([0.0115, 0.0085, 0.0148, 0.0114, 0.0175, 0.0122, 0.0104, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-27 18:27:08,628 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2023-04-27 18:27:43,242 INFO [train.py:904] (6/8) Epoch 2, batch 4850, loss[loss=0.298, simple_loss=0.3588, pruned_loss=0.1186, over 12247.00 frames. ], tot_loss[loss=0.267, simple_loss=0.3383, pruned_loss=0.0979, over 3191183.38 frames. ], batch size: 247, lr: 2.76e-02, grad_scale: 16.0 2023-04-27 18:28:11,112 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.137e+02 3.566e+02 4.230e+02 5.018e+02 9.509e+02, threshold=8.461e+02, percent-clipped=3.0 2023-04-27 18:28:52,369 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-27 18:28:58,608 INFO [train.py:904] (6/8) Epoch 2, batch 4900, loss[loss=0.2362, simple_loss=0.3198, pruned_loss=0.07624, over 16907.00 frames. ], tot_loss[loss=0.2655, simple_loss=0.3378, pruned_loss=0.09665, over 3190481.54 frames. ], batch size: 96, lr: 2.75e-02, grad_scale: 16.0 2023-04-27 18:29:24,592 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15069.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:30:04,905 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=15096.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:30:10,701 INFO [train.py:904] (6/8) Epoch 2, batch 4950, loss[loss=0.2495, simple_loss=0.3304, pruned_loss=0.0843, over 17141.00 frames. ], tot_loss[loss=0.2664, simple_loss=0.3384, pruned_loss=0.09722, over 3177366.29 frames. ], batch size: 49, lr: 2.75e-02, grad_scale: 16.0 2023-04-27 18:30:34,534 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=15117.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:30:37,410 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.417e+02 3.788e+02 4.518e+02 5.433e+02 9.876e+02, threshold=9.036e+02, percent-clipped=3.0 2023-04-27 18:30:39,353 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2023-04-27 18:30:47,342 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9295, 4.9259, 5.3335, 5.3706, 5.5522, 4.9048, 5.1316, 5.0997], device='cuda:6'), covar=tensor([0.0192, 0.0178, 0.0386, 0.0426, 0.0282, 0.0196, 0.0475, 0.0190], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0141, 0.0165, 0.0158, 0.0185, 0.0151, 0.0229, 0.0146], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-04-27 18:31:11,853 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1335, 3.8737, 1.4704, 3.9522, 2.2568, 3.8862, 1.7725, 2.7774], device='cuda:6'), covar=tensor([0.0022, 0.0110, 0.1542, 0.0049, 0.0797, 0.0169, 0.1269, 0.0504], device='cuda:6'), in_proj_covar=tensor([0.0070, 0.0095, 0.0162, 0.0076, 0.0152, 0.0119, 0.0169, 0.0145], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 18:31:12,921 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15143.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:31:22,555 INFO [train.py:904] (6/8) Epoch 2, batch 5000, loss[loss=0.2808, simple_loss=0.3527, pruned_loss=0.1045, over 15379.00 frames. ], tot_loss[loss=0.2667, simple_loss=0.3395, pruned_loss=0.09693, over 3188733.42 frames. ], batch size: 190, lr: 2.74e-02, grad_scale: 16.0 2023-04-27 18:31:58,930 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2023-04-27 18:31:59,742 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5890, 4.0907, 1.5672, 4.3459, 2.3570, 4.2709, 1.8090, 2.7031], device='cuda:6'), covar=tensor([0.0023, 0.0257, 0.2030, 0.0031, 0.0932, 0.0168, 0.1733, 0.0783], device='cuda:6'), in_proj_covar=tensor([0.0072, 0.0098, 0.0165, 0.0077, 0.0155, 0.0120, 0.0172, 0.0149], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 18:32:21,458 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=15191.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:32:26,387 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=15195.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:32:34,517 INFO [train.py:904] (6/8) Epoch 2, batch 5050, loss[loss=0.2542, simple_loss=0.3325, pruned_loss=0.08797, over 16893.00 frames. ], tot_loss[loss=0.2653, simple_loss=0.3391, pruned_loss=0.09573, over 3204870.38 frames. ], batch size: 109, lr: 2.74e-02, grad_scale: 16.0 2023-04-27 18:33:00,586 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.434e+02 3.699e+02 4.750e+02 5.849e+02 1.293e+03, threshold=9.501e+02, percent-clipped=2.0 2023-04-27 18:33:45,519 INFO [train.py:904] (6/8) Epoch 2, batch 5100, loss[loss=0.2329, simple_loss=0.3144, pruned_loss=0.07571, over 16548.00 frames. ], tot_loss[loss=0.263, simple_loss=0.337, pruned_loss=0.09448, over 3202443.89 frames. ], batch size: 68, lr: 2.74e-02, grad_scale: 16.0 2023-04-27 18:33:53,404 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=15256.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 18:34:52,182 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3721, 4.3739, 4.7674, 4.8014, 4.8526, 4.4141, 4.4992, 4.4338], device='cuda:6'), covar=tensor([0.0208, 0.0232, 0.0357, 0.0344, 0.0304, 0.0201, 0.0520, 0.0225], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0145, 0.0168, 0.0161, 0.0197, 0.0156, 0.0240, 0.0149], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-04-27 18:34:56,468 INFO [train.py:904] (6/8) Epoch 2, batch 5150, loss[loss=0.2826, simple_loss=0.3638, pruned_loss=0.1007, over 16259.00 frames. ], tot_loss[loss=0.2614, simple_loss=0.3364, pruned_loss=0.09317, over 3198732.66 frames. ], batch size: 165, lr: 2.73e-02, grad_scale: 16.0 2023-04-27 18:35:22,257 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.454e+02 3.560e+02 4.122e+02 5.032e+02 1.152e+03, threshold=8.244e+02, percent-clipped=1.0 2023-04-27 18:35:24,030 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-27 18:36:10,018 INFO [train.py:904] (6/8) Epoch 2, batch 5200, loss[loss=0.2561, simple_loss=0.3349, pruned_loss=0.08867, over 16833.00 frames. ], tot_loss[loss=0.2614, simple_loss=0.3359, pruned_loss=0.09349, over 3197705.14 frames. ], batch size: 102, lr: 2.73e-02, grad_scale: 16.0 2023-04-27 18:37:16,362 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15396.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:37:23,465 INFO [train.py:904] (6/8) Epoch 2, batch 5250, loss[loss=0.2386, simple_loss=0.3139, pruned_loss=0.08165, over 16462.00 frames. ], tot_loss[loss=0.2603, simple_loss=0.3335, pruned_loss=0.09355, over 3205993.88 frames. ], batch size: 75, lr: 2.72e-02, grad_scale: 16.0 2023-04-27 18:37:48,866 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.244e+02 3.713e+02 4.444e+02 5.638e+02 1.103e+03, threshold=8.887e+02, percent-clipped=2.0 2023-04-27 18:38:23,202 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=15444.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:38:32,783 INFO [train.py:904] (6/8) Epoch 2, batch 5300, loss[loss=0.2351, simple_loss=0.3107, pruned_loss=0.07972, over 16455.00 frames. ], tot_loss[loss=0.2564, simple_loss=0.329, pruned_loss=0.09191, over 3204371.79 frames. ], batch size: 146, lr: 2.72e-02, grad_scale: 16.0 2023-04-27 18:39:43,114 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-27 18:39:43,334 INFO [train.py:904] (6/8) Epoch 2, batch 5350, loss[loss=0.2626, simple_loss=0.3432, pruned_loss=0.09097, over 16749.00 frames. ], tot_loss[loss=0.2558, simple_loss=0.3284, pruned_loss=0.09157, over 3201193.15 frames. ], batch size: 83, lr: 2.72e-02, grad_scale: 16.0 2023-04-27 18:39:47,768 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4244, 5.6657, 5.3603, 5.5466, 4.9494, 4.9745, 5.2501, 5.8083], device='cuda:6'), covar=tensor([0.0346, 0.0486, 0.0661, 0.0298, 0.0409, 0.0305, 0.0319, 0.0481], device='cuda:6'), in_proj_covar=tensor([0.0207, 0.0291, 0.0268, 0.0188, 0.0198, 0.0180, 0.0238, 0.0203], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 18:40:09,981 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.455e+02 3.880e+02 4.934e+02 5.714e+02 1.287e+03, threshold=9.868e+02, percent-clipped=1.0 2023-04-27 18:40:31,125 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2023-04-27 18:40:42,569 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8505, 3.7424, 3.6744, 3.2073, 3.6720, 1.7890, 3.5115, 3.6887], device='cuda:6'), covar=tensor([0.0067, 0.0056, 0.0067, 0.0284, 0.0058, 0.1180, 0.0080, 0.0109], device='cuda:6'), in_proj_covar=tensor([0.0063, 0.0051, 0.0075, 0.0099, 0.0060, 0.0108, 0.0070, 0.0076], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-27 18:40:56,835 INFO [train.py:904] (6/8) Epoch 2, batch 5400, loss[loss=0.3433, simple_loss=0.3928, pruned_loss=0.1469, over 12106.00 frames. ], tot_loss[loss=0.2598, simple_loss=0.3322, pruned_loss=0.09368, over 3192914.05 frames. ], batch size: 248, lr: 2.71e-02, grad_scale: 16.0 2023-04-27 18:40:57,211 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=15551.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 18:42:14,657 INFO [train.py:904] (6/8) Epoch 2, batch 5450, loss[loss=0.2871, simple_loss=0.3532, pruned_loss=0.1105, over 16611.00 frames. ], tot_loss[loss=0.2648, simple_loss=0.3364, pruned_loss=0.09656, over 3199192.14 frames. ], batch size: 62, lr: 2.71e-02, grad_scale: 16.0 2023-04-27 18:42:43,047 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.524e+02 4.170e+02 4.882e+02 6.060e+02 1.199e+03, threshold=9.765e+02, percent-clipped=3.0 2023-04-27 18:43:34,025 INFO [train.py:904] (6/8) Epoch 2, batch 5500, loss[loss=0.3211, simple_loss=0.3811, pruned_loss=0.1306, over 16356.00 frames. ], tot_loss[loss=0.2782, simple_loss=0.3471, pruned_loss=0.1047, over 3171543.33 frames. ], batch size: 146, lr: 2.70e-02, grad_scale: 16.0 2023-04-27 18:44:09,174 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.59 vs. limit=5.0 2023-04-27 18:44:50,702 INFO [train.py:904] (6/8) Epoch 2, batch 5550, loss[loss=0.3955, simple_loss=0.4197, pruned_loss=0.1856, over 11279.00 frames. ], tot_loss[loss=0.2931, simple_loss=0.3576, pruned_loss=0.1143, over 3120536.56 frames. ], batch size: 247, lr: 2.70e-02, grad_scale: 16.0 2023-04-27 18:45:19,403 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.692e+02 5.827e+02 6.984e+02 8.601e+02 1.757e+03, threshold=1.397e+03, percent-clipped=15.0 2023-04-27 18:46:01,148 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1747, 3.5035, 3.6452, 1.4683, 3.8656, 3.8279, 2.9903, 3.0222], device='cuda:6'), covar=tensor([0.0941, 0.0122, 0.0156, 0.1807, 0.0067, 0.0059, 0.0339, 0.0381], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0082, 0.0078, 0.0152, 0.0073, 0.0071, 0.0106, 0.0114], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 18:46:11,044 INFO [train.py:904] (6/8) Epoch 2, batch 5600, loss[loss=0.4221, simple_loss=0.4388, pruned_loss=0.2027, over 11254.00 frames. ], tot_loss[loss=0.3063, simple_loss=0.3666, pruned_loss=0.123, over 3065015.42 frames. ], batch size: 248, lr: 2.70e-02, grad_scale: 16.0 2023-04-27 18:46:25,253 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-04-27 18:47:34,289 INFO [train.py:904] (6/8) Epoch 2, batch 5650, loss[loss=0.3183, simple_loss=0.383, pruned_loss=0.1268, over 16800.00 frames. ], tot_loss[loss=0.3171, simple_loss=0.3739, pruned_loss=0.1302, over 3029887.05 frames. ], batch size: 124, lr: 2.69e-02, grad_scale: 16.0 2023-04-27 18:48:01,990 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.886e+02 5.674e+02 6.824e+02 8.519e+02 2.118e+03, threshold=1.365e+03, percent-clipped=2.0 2023-04-27 18:48:53,333 INFO [train.py:904] (6/8) Epoch 2, batch 5700, loss[loss=0.3015, simple_loss=0.3778, pruned_loss=0.1126, over 16952.00 frames. ], tot_loss[loss=0.3208, simple_loss=0.3763, pruned_loss=0.1326, over 3013821.84 frames. ], batch size: 109, lr: 2.69e-02, grad_scale: 16.0 2023-04-27 18:48:53,782 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15851.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:49:44,698 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=15883.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:50:09,229 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=15899.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:50:12,115 INFO [train.py:904] (6/8) Epoch 2, batch 5750, loss[loss=0.3769, simple_loss=0.4006, pruned_loss=0.1766, over 11368.00 frames. ], tot_loss[loss=0.3233, simple_loss=0.3788, pruned_loss=0.1339, over 3003422.54 frames. ], batch size: 248, lr: 2.69e-02, grad_scale: 8.0 2023-04-27 18:50:42,009 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.285e+02 4.919e+02 6.542e+02 8.153e+02 1.932e+03, threshold=1.308e+03, percent-clipped=2.0 2023-04-27 18:51:20,718 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.12 vs. limit=5.0 2023-04-27 18:51:22,879 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8678, 3.6337, 3.7532, 3.8032, 3.3325, 3.6796, 3.6043, 3.5463], device='cuda:6'), covar=tensor([0.0332, 0.0215, 0.0203, 0.0133, 0.0709, 0.0194, 0.0475, 0.0264], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0089, 0.0147, 0.0115, 0.0179, 0.0123, 0.0105, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 18:51:22,915 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=15944.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:51:33,514 INFO [train.py:904] (6/8) Epoch 2, batch 5800, loss[loss=0.3084, simple_loss=0.3736, pruned_loss=0.1216, over 15419.00 frames. ], tot_loss[loss=0.3208, simple_loss=0.378, pruned_loss=0.1318, over 3007834.26 frames. ], batch size: 190, lr: 2.68e-02, grad_scale: 8.0 2023-04-27 18:51:46,808 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1267, 5.4601, 5.0921, 5.3605, 4.7914, 4.5976, 4.9625, 5.5144], device='cuda:6'), covar=tensor([0.0421, 0.0491, 0.0806, 0.0304, 0.0370, 0.0389, 0.0392, 0.0484], device='cuda:6'), in_proj_covar=tensor([0.0211, 0.0290, 0.0269, 0.0188, 0.0199, 0.0182, 0.0237, 0.0204], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 18:52:49,569 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-04-27 18:52:56,020 INFO [train.py:904] (6/8) Epoch 2, batch 5850, loss[loss=0.3193, simple_loss=0.3822, pruned_loss=0.1282, over 16365.00 frames. ], tot_loss[loss=0.3167, simple_loss=0.375, pruned_loss=0.1292, over 3000469.33 frames. ], batch size: 146, lr: 2.68e-02, grad_scale: 8.0 2023-04-27 18:52:57,407 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16001.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:53:11,372 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0392, 2.6887, 2.2816, 3.2835, 3.1829, 3.1942, 2.0083, 2.6610], device='cuda:6'), covar=tensor([0.1309, 0.0361, 0.1137, 0.0081, 0.0189, 0.0253, 0.0956, 0.0646], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0118, 0.0166, 0.0067, 0.0103, 0.0113, 0.0151, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 18:53:25,313 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.722e+02 4.866e+02 5.967e+02 7.152e+02 1.262e+03, threshold=1.193e+03, percent-clipped=0.0 2023-04-27 18:54:18,253 INFO [train.py:904] (6/8) Epoch 2, batch 5900, loss[loss=0.2907, simple_loss=0.353, pruned_loss=0.1142, over 15374.00 frames. ], tot_loss[loss=0.3168, simple_loss=0.3748, pruned_loss=0.1294, over 2986283.68 frames. ], batch size: 190, lr: 2.67e-02, grad_scale: 8.0 2023-04-27 18:54:40,579 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16062.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:55:17,986 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16085.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:55:42,173 INFO [train.py:904] (6/8) Epoch 2, batch 5950, loss[loss=0.2836, simple_loss=0.3527, pruned_loss=0.1073, over 16506.00 frames. ], tot_loss[loss=0.3132, simple_loss=0.3741, pruned_loss=0.1262, over 3012050.05 frames. ], batch size: 68, lr: 2.67e-02, grad_scale: 8.0 2023-04-27 18:56:13,605 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.088e+02 4.747e+02 6.406e+02 8.221e+02 1.979e+03, threshold=1.281e+03, percent-clipped=5.0 2023-04-27 18:56:56,629 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16146.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:57:01,975 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16149.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:57:04,556 INFO [train.py:904] (6/8) Epoch 2, batch 6000, loss[loss=0.3052, simple_loss=0.3681, pruned_loss=0.1211, over 16440.00 frames. ], tot_loss[loss=0.3113, simple_loss=0.3726, pruned_loss=0.125, over 3032208.18 frames. ], batch size: 146, lr: 2.67e-02, grad_scale: 8.0 2023-04-27 18:57:04,556 INFO [train.py:929] (6/8) Computing validation loss 2023-04-27 18:57:15,927 INFO [train.py:938] (6/8) Epoch 2, validation: loss=0.2302, simple_loss=0.3372, pruned_loss=0.06166, over 944034.00 frames. 2023-04-27 18:57:15,928 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17404MB 2023-04-27 18:58:09,118 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6296, 2.7132, 2.3312, 4.0596, 2.0271, 3.8689, 2.4304, 2.3479], device='cuda:6'), covar=tensor([0.0277, 0.0394, 0.0348, 0.0172, 0.1396, 0.0170, 0.0571, 0.1083], device='cuda:6'), in_proj_covar=tensor([0.0204, 0.0176, 0.0149, 0.0205, 0.0257, 0.0161, 0.0180, 0.0235], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 18:58:34,798 INFO [train.py:904] (6/8) Epoch 2, batch 6050, loss[loss=0.3004, simple_loss=0.359, pruned_loss=0.1209, over 15463.00 frames. ], tot_loss[loss=0.3079, simple_loss=0.3703, pruned_loss=0.1227, over 3073486.50 frames. ], batch size: 191, lr: 2.66e-02, grad_scale: 8.0 2023-04-27 18:58:48,637 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16210.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:58:53,867 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4276, 5.6484, 5.3160, 5.4872, 4.9514, 4.8313, 5.1645, 5.7218], device='cuda:6'), covar=tensor([0.0349, 0.0485, 0.0700, 0.0295, 0.0454, 0.0360, 0.0439, 0.0424], device='cuda:6'), in_proj_covar=tensor([0.0221, 0.0302, 0.0283, 0.0195, 0.0209, 0.0188, 0.0248, 0.0218], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 18:58:57,656 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9631, 2.7239, 2.7135, 1.7704, 2.8161, 2.7988, 2.4916, 2.4538], device='cuda:6'), covar=tensor([0.0721, 0.0120, 0.0160, 0.1098, 0.0108, 0.0087, 0.0297, 0.0335], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0082, 0.0077, 0.0153, 0.0076, 0.0074, 0.0105, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 18:59:00,587 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5133, 4.7878, 4.7148, 4.7375, 4.6560, 5.2551, 4.9758, 4.6882], device='cuda:6'), covar=tensor([0.0613, 0.1217, 0.1093, 0.1344, 0.2418, 0.0855, 0.0792, 0.1722], device='cuda:6'), in_proj_covar=tensor([0.0186, 0.0264, 0.0246, 0.0228, 0.0299, 0.0255, 0.0205, 0.0314], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 18:59:04,137 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.028e+02 4.577e+02 5.718e+02 7.421e+02 1.323e+03, threshold=1.144e+03, percent-clipped=1.0 2023-04-27 18:59:33,659 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16239.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:59:52,052 INFO [train.py:904] (6/8) Epoch 2, batch 6100, loss[loss=0.4144, simple_loss=0.4316, pruned_loss=0.1986, over 11568.00 frames. ], tot_loss[loss=0.3042, simple_loss=0.3684, pruned_loss=0.12, over 3102162.14 frames. ], batch size: 247, lr: 2.66e-02, grad_scale: 8.0 2023-04-27 19:00:20,312 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2418, 1.3871, 1.8748, 2.3456, 2.3761, 2.2006, 1.4115, 2.1413], device='cuda:6'), covar=tensor([0.0051, 0.0230, 0.0115, 0.0076, 0.0037, 0.0077, 0.0158, 0.0036], device='cuda:6'), in_proj_covar=tensor([0.0068, 0.0106, 0.0091, 0.0078, 0.0058, 0.0057, 0.0090, 0.0052], device='cuda:6'), out_proj_covar=tensor([1.2007e-04, 1.8746e-04, 1.6690e-04, 1.4587e-04, 9.9206e-05, 1.0311e-04, 1.5428e-04, 9.1494e-05], device='cuda:6') 2023-04-27 19:00:39,248 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-27 19:01:00,947 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4661, 4.5179, 4.5038, 1.8204, 4.7958, 4.8859, 3.7849, 3.7401], device='cuda:6'), covar=tensor([0.0854, 0.0078, 0.0111, 0.1539, 0.0044, 0.0030, 0.0194, 0.0296], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0079, 0.0075, 0.0145, 0.0075, 0.0070, 0.0101, 0.0114], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:01:17,143 INFO [train.py:904] (6/8) Epoch 2, batch 6150, loss[loss=0.2798, simple_loss=0.3542, pruned_loss=0.1027, over 16767.00 frames. ], tot_loss[loss=0.3006, simple_loss=0.3647, pruned_loss=0.1183, over 3103471.38 frames. ], batch size: 124, lr: 2.66e-02, grad_scale: 8.0 2023-04-27 19:01:45,849 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.137e+02 4.737e+02 5.827e+02 7.174e+02 1.739e+03, threshold=1.165e+03, percent-clipped=5.0 2023-04-27 19:02:34,337 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-27 19:02:34,532 INFO [train.py:904] (6/8) Epoch 2, batch 6200, loss[loss=0.262, simple_loss=0.3365, pruned_loss=0.09375, over 16564.00 frames. ], tot_loss[loss=0.2999, simple_loss=0.3634, pruned_loss=0.1182, over 3112595.47 frames. ], batch size: 57, lr: 2.65e-02, grad_scale: 8.0 2023-04-27 19:02:45,549 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16357.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:03:06,797 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16371.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:03:32,991 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-27 19:03:51,376 INFO [train.py:904] (6/8) Epoch 2, batch 6250, loss[loss=0.2812, simple_loss=0.3504, pruned_loss=0.106, over 17011.00 frames. ], tot_loss[loss=0.2997, simple_loss=0.3633, pruned_loss=0.1181, over 3113266.22 frames. ], batch size: 55, lr: 2.65e-02, grad_scale: 8.0 2023-04-27 19:04:18,705 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.383e+02 4.905e+02 5.910e+02 7.685e+02 1.899e+03, threshold=1.182e+03, percent-clipped=6.0 2023-04-27 19:04:37,381 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16432.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:04:51,246 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16441.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:05:05,086 INFO [train.py:904] (6/8) Epoch 2, batch 6300, loss[loss=0.3614, simple_loss=0.4069, pruned_loss=0.158, over 16175.00 frames. ], tot_loss[loss=0.3002, simple_loss=0.364, pruned_loss=0.1183, over 3106036.95 frames. ], batch size: 165, lr: 2.64e-02, grad_scale: 8.0 2023-04-27 19:06:13,116 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2023-04-27 19:06:22,078 INFO [train.py:904] (6/8) Epoch 2, batch 6350, loss[loss=0.308, simple_loss=0.3767, pruned_loss=0.1196, over 16868.00 frames. ], tot_loss[loss=0.3025, simple_loss=0.365, pruned_loss=0.12, over 3106875.91 frames. ], batch size: 116, lr: 2.64e-02, grad_scale: 8.0 2023-04-27 19:06:29,788 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16505.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:06:39,766 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0982, 2.7701, 2.3240, 3.3319, 3.2112, 3.2956, 1.9436, 2.6897], device='cuda:6'), covar=tensor([0.1053, 0.0277, 0.0915, 0.0082, 0.0274, 0.0227, 0.0881, 0.0547], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0122, 0.0166, 0.0069, 0.0109, 0.0116, 0.0152, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 19:06:52,065 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.677e+02 5.217e+02 6.547e+02 8.022e+02 1.954e+03, threshold=1.309e+03, percent-clipped=7.0 2023-04-27 19:06:56,749 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9742, 3.7459, 3.8613, 4.1893, 4.2125, 3.8910, 4.1423, 4.1343], device='cuda:6'), covar=tensor([0.0486, 0.0485, 0.0965, 0.0395, 0.0405, 0.0481, 0.0439, 0.0341], device='cuda:6'), in_proj_covar=tensor([0.0229, 0.0261, 0.0366, 0.0276, 0.0210, 0.0194, 0.0211, 0.0212], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:07:08,017 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-27 19:07:21,374 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16539.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:07:39,360 INFO [train.py:904] (6/8) Epoch 2, batch 6400, loss[loss=0.2678, simple_loss=0.3357, pruned_loss=0.09995, over 16710.00 frames. ], tot_loss[loss=0.3034, simple_loss=0.3653, pruned_loss=0.1208, over 3118067.89 frames. ], batch size: 76, lr: 2.64e-02, grad_scale: 8.0 2023-04-27 19:08:34,558 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=16587.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:08:55,789 INFO [train.py:904] (6/8) Epoch 2, batch 6450, loss[loss=0.2849, simple_loss=0.3364, pruned_loss=0.1166, over 11812.00 frames. ], tot_loss[loss=0.3009, simple_loss=0.3638, pruned_loss=0.119, over 3108535.24 frames. ], batch size: 247, lr: 2.63e-02, grad_scale: 8.0 2023-04-27 19:09:20,969 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.17 vs. limit=2.0 2023-04-27 19:09:25,707 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.052e+02 4.859e+02 5.740e+02 6.998e+02 1.216e+03, threshold=1.148e+03, percent-clipped=0.0 2023-04-27 19:09:40,255 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7349, 2.6746, 2.6713, 2.0839, 2.5563, 2.5238, 2.6865, 1.7817], device='cuda:6'), covar=tensor([0.0430, 0.0045, 0.0056, 0.0258, 0.0061, 0.0086, 0.0041, 0.0414], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0051, 0.0056, 0.0108, 0.0051, 0.0056, 0.0058, 0.0104], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 19:09:57,043 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7411, 4.5410, 4.6020, 4.6422, 4.0990, 4.5956, 4.5415, 4.2952], device='cuda:6'), covar=tensor([0.0342, 0.0199, 0.0147, 0.0104, 0.0650, 0.0161, 0.0168, 0.0210], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0093, 0.0150, 0.0116, 0.0178, 0.0124, 0.0106, 0.0138], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 19:10:13,782 INFO [train.py:904] (6/8) Epoch 2, batch 6500, loss[loss=0.2693, simple_loss=0.3291, pruned_loss=0.1048, over 16783.00 frames. ], tot_loss[loss=0.2973, simple_loss=0.3603, pruned_loss=0.1172, over 3109962.67 frames. ], batch size: 39, lr: 2.63e-02, grad_scale: 8.0 2023-04-27 19:10:24,340 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16657.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:11:11,397 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1611, 3.8858, 4.0582, 4.0765, 3.5621, 4.0583, 3.9198, 3.7309], device='cuda:6'), covar=tensor([0.0376, 0.0207, 0.0178, 0.0127, 0.0716, 0.0200, 0.0292, 0.0252], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0091, 0.0148, 0.0115, 0.0175, 0.0124, 0.0105, 0.0136], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 19:11:32,128 INFO [train.py:904] (6/8) Epoch 2, batch 6550, loss[loss=0.3321, simple_loss=0.414, pruned_loss=0.1251, over 16559.00 frames. ], tot_loss[loss=0.3019, simple_loss=0.3644, pruned_loss=0.1196, over 3093571.97 frames. ], batch size: 57, lr: 2.63e-02, grad_scale: 8.0 2023-04-27 19:11:37,738 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=16705.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:11:59,595 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.469e+02 4.692e+02 5.928e+02 7.455e+02 1.589e+03, threshold=1.186e+03, percent-clipped=4.0 2023-04-27 19:12:11,256 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16727.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:12:30,176 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.83 vs. limit=5.0 2023-04-27 19:12:33,050 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16741.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:12:47,616 INFO [train.py:904] (6/8) Epoch 2, batch 6600, loss[loss=0.303, simple_loss=0.3731, pruned_loss=0.1164, over 16713.00 frames. ], tot_loss[loss=0.3036, simple_loss=0.3673, pruned_loss=0.1199, over 3104032.30 frames. ], batch size: 89, lr: 2.62e-02, grad_scale: 8.0 2023-04-27 19:13:08,924 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.39 vs. limit=5.0 2023-04-27 19:13:47,418 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=16789.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:14:06,684 INFO [train.py:904] (6/8) Epoch 2, batch 6650, loss[loss=0.3663, simple_loss=0.4059, pruned_loss=0.1633, over 11547.00 frames. ], tot_loss[loss=0.3061, simple_loss=0.3684, pruned_loss=0.1219, over 3096438.68 frames. ], batch size: 248, lr: 2.62e-02, grad_scale: 8.0 2023-04-27 19:14:13,248 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16805.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:14:35,746 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.480e+02 5.041e+02 6.003e+02 7.722e+02 1.259e+03, threshold=1.201e+03, percent-clipped=2.0 2023-04-27 19:15:23,513 INFO [train.py:904] (6/8) Epoch 2, batch 6700, loss[loss=0.3546, simple_loss=0.387, pruned_loss=0.1611, over 11470.00 frames. ], tot_loss[loss=0.3046, simple_loss=0.3663, pruned_loss=0.1214, over 3098040.48 frames. ], batch size: 247, lr: 2.61e-02, grad_scale: 8.0 2023-04-27 19:15:26,815 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=16853.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:15:50,666 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16868.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 19:16:38,378 INFO [train.py:904] (6/8) Epoch 2, batch 6750, loss[loss=0.2718, simple_loss=0.3465, pruned_loss=0.09855, over 16177.00 frames. ], tot_loss[loss=0.3035, simple_loss=0.3651, pruned_loss=0.121, over 3104156.88 frames. ], batch size: 165, lr: 2.61e-02, grad_scale: 8.0 2023-04-27 19:16:47,153 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16906.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:17:03,839 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3089, 4.5054, 4.2036, 4.3487, 3.9733, 3.9199, 4.1428, 4.5629], device='cuda:6'), covar=tensor([0.0400, 0.0616, 0.0884, 0.0385, 0.0471, 0.0807, 0.0505, 0.0509], device='cuda:6'), in_proj_covar=tensor([0.0221, 0.0307, 0.0283, 0.0192, 0.0205, 0.0190, 0.0250, 0.0211], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:17:07,558 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.622e+02 5.251e+02 6.489e+02 8.459e+02 1.507e+03, threshold=1.298e+03, percent-clipped=4.0 2023-04-27 19:17:16,709 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8190, 4.0273, 3.6954, 3.9040, 3.5775, 3.6667, 3.7909, 3.9583], device='cuda:6'), covar=tensor([0.0452, 0.0690, 0.0956, 0.0408, 0.0485, 0.0758, 0.0488, 0.0726], device='cuda:6'), in_proj_covar=tensor([0.0221, 0.0307, 0.0283, 0.0193, 0.0205, 0.0190, 0.0250, 0.0211], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:17:22,266 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16929.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 19:17:53,312 INFO [train.py:904] (6/8) Epoch 2, batch 6800, loss[loss=0.2874, simple_loss=0.3597, pruned_loss=0.1076, over 16995.00 frames. ], tot_loss[loss=0.3021, simple_loss=0.3645, pruned_loss=0.1198, over 3110319.22 frames. ], batch size: 55, lr: 2.61e-02, grad_scale: 8.0 2023-04-27 19:18:18,710 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16967.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 19:19:10,224 INFO [train.py:904] (6/8) Epoch 2, batch 6850, loss[loss=0.2869, simple_loss=0.3695, pruned_loss=0.1021, over 16420.00 frames. ], tot_loss[loss=0.3044, simple_loss=0.3666, pruned_loss=0.1211, over 3091214.54 frames. ], batch size: 68, lr: 2.60e-02, grad_scale: 8.0 2023-04-27 19:19:13,759 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9586, 4.0080, 4.4203, 4.3936, 4.4794, 4.0008, 4.0910, 4.1274], device='cuda:6'), covar=tensor([0.0257, 0.0267, 0.0381, 0.0414, 0.0335, 0.0283, 0.0811, 0.0309], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0151, 0.0165, 0.0168, 0.0197, 0.0168, 0.0255, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-27 19:19:21,814 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1987, 5.0296, 4.9237, 5.0108, 4.4723, 5.0287, 4.9316, 4.7098], device='cuda:6'), covar=tensor([0.0340, 0.0168, 0.0178, 0.0108, 0.0733, 0.0227, 0.0167, 0.0321], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0096, 0.0154, 0.0121, 0.0180, 0.0130, 0.0109, 0.0143], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 19:19:38,485 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.045e+02 4.695e+02 5.738e+02 7.185e+02 1.728e+03, threshold=1.148e+03, percent-clipped=4.0 2023-04-27 19:19:49,714 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17027.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:20:24,230 INFO [train.py:904] (6/8) Epoch 2, batch 6900, loss[loss=0.3146, simple_loss=0.3776, pruned_loss=0.1258, over 16471.00 frames. ], tot_loss[loss=0.3033, simple_loss=0.3683, pruned_loss=0.1192, over 3114988.94 frames. ], batch size: 146, lr: 2.60e-02, grad_scale: 8.0 2023-04-27 19:21:01,217 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=17075.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:21:40,735 INFO [train.py:904] (6/8) Epoch 2, batch 6950, loss[loss=0.3152, simple_loss=0.3783, pruned_loss=0.126, over 16743.00 frames. ], tot_loss[loss=0.3073, simple_loss=0.3709, pruned_loss=0.1218, over 3112021.46 frames. ], batch size: 124, lr: 2.60e-02, grad_scale: 8.0 2023-04-27 19:22:09,951 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.451e+02 5.662e+02 6.855e+02 8.747e+02 1.724e+03, threshold=1.371e+03, percent-clipped=6.0 2023-04-27 19:22:27,105 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2253, 3.1205, 3.1988, 3.4458, 3.3960, 3.1333, 3.3732, 3.4063], device='cuda:6'), covar=tensor([0.0491, 0.0445, 0.0981, 0.0388, 0.0457, 0.1214, 0.0564, 0.0395], device='cuda:6'), in_proj_covar=tensor([0.0242, 0.0278, 0.0382, 0.0280, 0.0224, 0.0201, 0.0221, 0.0225], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:22:54,967 INFO [train.py:904] (6/8) Epoch 2, batch 7000, loss[loss=0.2671, simple_loss=0.3593, pruned_loss=0.08747, over 16895.00 frames. ], tot_loss[loss=0.3071, simple_loss=0.3713, pruned_loss=0.1215, over 3097736.72 frames. ], batch size: 96, lr: 2.59e-02, grad_scale: 8.0 2023-04-27 19:23:46,959 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-27 19:23:53,757 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17189.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:24:10,869 INFO [train.py:904] (6/8) Epoch 2, batch 7050, loss[loss=0.3682, simple_loss=0.3942, pruned_loss=0.1711, over 11155.00 frames. ], tot_loss[loss=0.3057, simple_loss=0.3711, pruned_loss=0.1201, over 3123950.37 frames. ], batch size: 247, lr: 2.59e-02, grad_scale: 8.0 2023-04-27 19:24:37,722 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.062e+02 5.191e+02 6.345e+02 7.855e+02 1.482e+03, threshold=1.269e+03, percent-clipped=3.0 2023-04-27 19:24:44,024 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17224.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 19:24:47,352 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8815, 3.7884, 3.7238, 3.1630, 3.7128, 1.6197, 3.4953, 3.7085], device='cuda:6'), covar=tensor([0.0070, 0.0056, 0.0079, 0.0282, 0.0063, 0.1480, 0.0081, 0.0105], device='cuda:6'), in_proj_covar=tensor([0.0064, 0.0052, 0.0078, 0.0098, 0.0058, 0.0112, 0.0070, 0.0077], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:25:22,968 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17250.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:25:23,662 INFO [train.py:904] (6/8) Epoch 2, batch 7100, loss[loss=0.3097, simple_loss=0.3731, pruned_loss=0.1232, over 16225.00 frames. ], tot_loss[loss=0.3047, simple_loss=0.3694, pruned_loss=0.12, over 3124412.85 frames. ], batch size: 165, lr: 2.59e-02, grad_scale: 8.0 2023-04-27 19:25:41,136 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17262.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 19:25:44,896 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0694, 2.4023, 2.0482, 3.4763, 1.9394, 3.1882, 2.3038, 2.0164], device='cuda:6'), covar=tensor([0.0367, 0.0527, 0.0402, 0.0219, 0.1609, 0.0220, 0.0743, 0.1285], device='cuda:6'), in_proj_covar=tensor([0.0213, 0.0184, 0.0158, 0.0214, 0.0266, 0.0171, 0.0188, 0.0243], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:26:38,657 INFO [train.py:904] (6/8) Epoch 2, batch 7150, loss[loss=0.2828, simple_loss=0.3605, pruned_loss=0.1026, over 16321.00 frames. ], tot_loss[loss=0.3026, simple_loss=0.3668, pruned_loss=0.1191, over 3115379.03 frames. ], batch size: 35, lr: 2.58e-02, grad_scale: 8.0 2023-04-27 19:27:07,391 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.740e+02 5.248e+02 6.301e+02 7.700e+02 1.780e+03, threshold=1.260e+03, percent-clipped=1.0 2023-04-27 19:27:43,248 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4398, 3.4854, 3.9262, 3.9123, 3.8862, 3.4997, 3.6647, 3.7209], device='cuda:6'), covar=tensor([0.0304, 0.0296, 0.0297, 0.0349, 0.0375, 0.0300, 0.0593, 0.0302], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0151, 0.0169, 0.0165, 0.0201, 0.0169, 0.0250, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-27 19:27:52,977 INFO [train.py:904] (6/8) Epoch 2, batch 7200, loss[loss=0.2719, simple_loss=0.3506, pruned_loss=0.09658, over 16715.00 frames. ], tot_loss[loss=0.2987, simple_loss=0.3641, pruned_loss=0.1166, over 3112967.16 frames. ], batch size: 134, lr: 2.58e-02, grad_scale: 8.0 2023-04-27 19:28:10,366 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17362.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:28:45,036 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17383.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 19:29:13,593 INFO [train.py:904] (6/8) Epoch 2, batch 7250, loss[loss=0.2331, simple_loss=0.3101, pruned_loss=0.07802, over 16771.00 frames. ], tot_loss[loss=0.2951, simple_loss=0.361, pruned_loss=0.1146, over 3101763.62 frames. ], batch size: 83, lr: 2.58e-02, grad_scale: 8.0 2023-04-27 19:29:31,022 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4367, 4.6838, 4.3126, 4.4590, 4.0192, 3.9155, 4.1685, 4.7266], device='cuda:6'), covar=tensor([0.0436, 0.0646, 0.0852, 0.0354, 0.0479, 0.0711, 0.0501, 0.0532], device='cuda:6'), in_proj_covar=tensor([0.0224, 0.0309, 0.0287, 0.0193, 0.0207, 0.0195, 0.0254, 0.0212], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:29:42,539 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.618e+02 4.777e+02 5.726e+02 7.027e+02 1.873e+03, threshold=1.145e+03, percent-clipped=4.0 2023-04-27 19:29:47,314 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17423.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:30:16,556 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3570, 3.3128, 3.2275, 2.7511, 3.3022, 2.1082, 3.0767, 3.1480], device='cuda:6'), covar=tensor([0.0076, 0.0057, 0.0087, 0.0269, 0.0065, 0.1012, 0.0087, 0.0124], device='cuda:6'), in_proj_covar=tensor([0.0063, 0.0051, 0.0077, 0.0097, 0.0057, 0.0108, 0.0069, 0.0075], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:30:19,115 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17444.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 19:30:29,429 INFO [train.py:904] (6/8) Epoch 2, batch 7300, loss[loss=0.3794, simple_loss=0.4044, pruned_loss=0.1772, over 11085.00 frames. ], tot_loss[loss=0.295, simple_loss=0.3606, pruned_loss=0.1147, over 3096196.92 frames. ], batch size: 247, lr: 2.57e-02, grad_scale: 8.0 2023-04-27 19:30:42,017 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6390, 2.7642, 2.0316, 3.9922, 3.8457, 3.6319, 1.8739, 2.7279], device='cuda:6'), covar=tensor([0.1540, 0.0561, 0.1484, 0.0079, 0.0149, 0.0303, 0.1214, 0.0781], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0122, 0.0164, 0.0069, 0.0107, 0.0120, 0.0153, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 19:30:51,789 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4134, 3.3834, 3.2902, 2.9863, 3.3640, 2.1220, 3.1899, 3.1995], device='cuda:6'), covar=tensor([0.0054, 0.0041, 0.0066, 0.0202, 0.0046, 0.1024, 0.0065, 0.0079], device='cuda:6'), in_proj_covar=tensor([0.0062, 0.0050, 0.0076, 0.0095, 0.0056, 0.0106, 0.0068, 0.0074], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:31:46,483 INFO [train.py:904] (6/8) Epoch 2, batch 7350, loss[loss=0.2914, simple_loss=0.3569, pruned_loss=0.113, over 16472.00 frames. ], tot_loss[loss=0.2948, simple_loss=0.3598, pruned_loss=0.1149, over 3068048.38 frames. ], batch size: 146, lr: 2.57e-02, grad_scale: 8.0 2023-04-27 19:32:16,423 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.647e+02 4.932e+02 5.634e+02 7.018e+02 1.459e+03, threshold=1.127e+03, percent-clipped=2.0 2023-04-27 19:32:22,506 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17524.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 19:32:57,423 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17545.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:33:01,696 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9814, 2.4619, 2.0744, 3.1111, 2.2114, 2.9273, 2.3395, 2.0065], device='cuda:6'), covar=tensor([0.0279, 0.0417, 0.0316, 0.0208, 0.1085, 0.0210, 0.0591, 0.1001], device='cuda:6'), in_proj_covar=tensor([0.0217, 0.0187, 0.0156, 0.0218, 0.0268, 0.0170, 0.0190, 0.0248], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:33:06,151 INFO [train.py:904] (6/8) Epoch 2, batch 7400, loss[loss=0.278, simple_loss=0.361, pruned_loss=0.09746, over 16985.00 frames. ], tot_loss[loss=0.2961, simple_loss=0.361, pruned_loss=0.1157, over 3060038.62 frames. ], batch size: 41, lr: 2.57e-02, grad_scale: 8.0 2023-04-27 19:33:25,297 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17562.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 19:33:41,414 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=17572.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 19:34:27,189 INFO [train.py:904] (6/8) Epoch 2, batch 7450, loss[loss=0.321, simple_loss=0.3689, pruned_loss=0.1366, over 11978.00 frames. ], tot_loss[loss=0.2969, simple_loss=0.3616, pruned_loss=0.1161, over 3083223.91 frames. ], batch size: 246, lr: 2.56e-02, grad_scale: 8.0 2023-04-27 19:34:43,809 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=17610.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:34:52,181 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2387, 3.0739, 2.6532, 2.6633, 2.4349, 1.9868, 3.3778, 3.6130], device='cuda:6'), covar=tensor([0.1939, 0.0756, 0.1102, 0.0654, 0.1542, 0.1340, 0.0373, 0.0217], device='cuda:6'), in_proj_covar=tensor([0.0257, 0.0228, 0.0239, 0.0183, 0.0279, 0.0183, 0.0201, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:34:59,865 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.824e+02 5.406e+02 6.493e+02 7.703e+02 1.630e+03, threshold=1.299e+03, percent-clipped=5.0 2023-04-27 19:35:10,341 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2023-04-27 19:35:49,112 INFO [train.py:904] (6/8) Epoch 2, batch 7500, loss[loss=0.2875, simple_loss=0.3562, pruned_loss=0.1094, over 16530.00 frames. ], tot_loss[loss=0.2999, simple_loss=0.3642, pruned_loss=0.1178, over 3063116.20 frames. ], batch size: 62, lr: 2.56e-02, grad_scale: 8.0 2023-04-27 19:35:50,313 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2023-04-27 19:36:21,476 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.28 vs. limit=5.0 2023-04-27 19:36:57,813 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8598, 4.0237, 3.7750, 3.9331, 3.4759, 3.6319, 3.7736, 3.9538], device='cuda:6'), covar=tensor([0.0427, 0.0559, 0.0697, 0.0313, 0.0434, 0.0627, 0.0420, 0.0592], device='cuda:6'), in_proj_covar=tensor([0.0215, 0.0302, 0.0276, 0.0191, 0.0200, 0.0189, 0.0240, 0.0203], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:37:05,089 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17700.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:37:05,673 INFO [train.py:904] (6/8) Epoch 2, batch 7550, loss[loss=0.2862, simple_loss=0.3578, pruned_loss=0.1073, over 16840.00 frames. ], tot_loss[loss=0.302, simple_loss=0.3647, pruned_loss=0.1196, over 3030654.93 frames. ], batch size: 116, lr: 2.56e-02, grad_scale: 8.0 2023-04-27 19:37:25,288 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8162, 4.4729, 4.5712, 4.6099, 4.0393, 4.5655, 4.5436, 4.3152], device='cuda:6'), covar=tensor([0.0214, 0.0148, 0.0152, 0.0110, 0.0642, 0.0185, 0.0159, 0.0229], device='cuda:6'), in_proj_covar=tensor([0.0123, 0.0094, 0.0146, 0.0118, 0.0174, 0.0126, 0.0106, 0.0137], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 19:37:31,533 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9258, 4.0199, 3.7879, 2.9314, 3.4040, 2.5668, 4.4349, 5.1587], device='cuda:6'), covar=tensor([0.1917, 0.0687, 0.0782, 0.0694, 0.1769, 0.1075, 0.0288, 0.0141], device='cuda:6'), in_proj_covar=tensor([0.0255, 0.0224, 0.0238, 0.0181, 0.0274, 0.0180, 0.0198, 0.0145], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:37:32,654 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17718.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:37:34,648 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.677e+02 5.137e+02 6.469e+02 8.298e+02 1.927e+03, threshold=1.294e+03, percent-clipped=3.0 2023-04-27 19:37:42,008 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6081, 3.3293, 2.2905, 4.6238, 4.5343, 4.0878, 1.8996, 2.9905], device='cuda:6'), covar=tensor([0.1597, 0.0407, 0.1343, 0.0056, 0.0110, 0.0239, 0.1248, 0.0652], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0121, 0.0162, 0.0067, 0.0106, 0.0118, 0.0151, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 19:37:51,152 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17730.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:38:05,273 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17739.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 19:38:22,207 INFO [train.py:904] (6/8) Epoch 2, batch 7600, loss[loss=0.2826, simple_loss=0.3411, pruned_loss=0.1121, over 16974.00 frames. ], tot_loss[loss=0.3008, simple_loss=0.3633, pruned_loss=0.1192, over 3041197.81 frames. ], batch size: 55, lr: 2.55e-02, grad_scale: 8.0 2023-04-27 19:38:37,861 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17761.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:39:25,119 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17791.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:39:39,705 INFO [train.py:904] (6/8) Epoch 2, batch 7650, loss[loss=0.3155, simple_loss=0.3876, pruned_loss=0.1216, over 16385.00 frames. ], tot_loss[loss=0.3018, simple_loss=0.3641, pruned_loss=0.1198, over 3057010.89 frames. ], batch size: 146, lr: 2.55e-02, grad_scale: 8.0 2023-04-27 19:40:10,724 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.673e+02 5.378e+02 6.384e+02 7.645e+02 1.933e+03, threshold=1.277e+03, percent-clipped=1.0 2023-04-27 19:40:17,622 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7586, 4.0702, 3.4107, 2.9631, 3.1336, 2.3658, 4.1252, 4.8188], device='cuda:6'), covar=tensor([0.1789, 0.0532, 0.0900, 0.0605, 0.1703, 0.1082, 0.0309, 0.0141], device='cuda:6'), in_proj_covar=tensor([0.0257, 0.0225, 0.0242, 0.0182, 0.0276, 0.0181, 0.0202, 0.0149], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:40:50,314 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17845.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:40:50,420 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6816, 2.5327, 2.6308, 1.9781, 2.5148, 2.4764, 2.6007, 1.8005], device='cuda:6'), covar=tensor([0.0391, 0.0045, 0.0066, 0.0260, 0.0055, 0.0100, 0.0037, 0.0343], device='cuda:6'), in_proj_covar=tensor([0.0117, 0.0054, 0.0059, 0.0110, 0.0054, 0.0061, 0.0057, 0.0106], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 19:40:58,861 INFO [train.py:904] (6/8) Epoch 2, batch 7700, loss[loss=0.3114, simple_loss=0.3777, pruned_loss=0.1225, over 16770.00 frames. ], tot_loss[loss=0.3018, simple_loss=0.364, pruned_loss=0.1198, over 3074771.35 frames. ], batch size: 62, lr: 2.55e-02, grad_scale: 8.0 2023-04-27 19:41:15,740 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-27 19:42:04,329 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=17893.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:42:16,829 INFO [train.py:904] (6/8) Epoch 2, batch 7750, loss[loss=0.3651, simple_loss=0.3955, pruned_loss=0.1673, over 11632.00 frames. ], tot_loss[loss=0.3014, simple_loss=0.3641, pruned_loss=0.1194, over 3086409.03 frames. ], batch size: 246, lr: 2.54e-02, grad_scale: 16.0 2023-04-27 19:42:46,604 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.308e+02 4.989e+02 6.319e+02 7.239e+02 1.280e+03, threshold=1.264e+03, percent-clipped=2.0 2023-04-27 19:43:31,901 INFO [train.py:904] (6/8) Epoch 2, batch 7800, loss[loss=0.3904, simple_loss=0.4186, pruned_loss=0.1811, over 11381.00 frames. ], tot_loss[loss=0.3047, simple_loss=0.3664, pruned_loss=0.1215, over 3083879.51 frames. ], batch size: 246, lr: 2.54e-02, grad_scale: 16.0 2023-04-27 19:43:47,101 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17960.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:44:52,420 INFO [train.py:904] (6/8) Epoch 2, batch 7850, loss[loss=0.2562, simple_loss=0.3332, pruned_loss=0.08956, over 16549.00 frames. ], tot_loss[loss=0.3037, simple_loss=0.3661, pruned_loss=0.1206, over 3065271.82 frames. ], batch size: 57, lr: 2.54e-02, grad_scale: 16.0 2023-04-27 19:45:18,565 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18018.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:45:21,252 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.805e+02 5.114e+02 6.121e+02 7.560e+02 1.271e+03, threshold=1.224e+03, percent-clipped=1.0 2023-04-27 19:45:22,879 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18021.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:45:46,039 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4572, 4.3111, 4.3494, 1.6860, 4.6328, 4.5569, 3.4897, 3.6995], device='cuda:6'), covar=tensor([0.0778, 0.0113, 0.0158, 0.1486, 0.0039, 0.0031, 0.0238, 0.0272], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0086, 0.0084, 0.0148, 0.0074, 0.0069, 0.0106, 0.0117], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:45:48,884 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18039.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 19:46:05,621 INFO [train.py:904] (6/8) Epoch 2, batch 7900, loss[loss=0.3593, simple_loss=0.3977, pruned_loss=0.1605, over 11626.00 frames. ], tot_loss[loss=0.3017, simple_loss=0.3645, pruned_loss=0.1195, over 3062637.09 frames. ], batch size: 246, lr: 2.53e-02, grad_scale: 8.0 2023-04-27 19:46:14,078 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18056.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:46:19,012 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3547, 4.0804, 4.0421, 3.4202, 4.0766, 1.5245, 3.8791, 4.1024], device='cuda:6'), covar=tensor([0.0051, 0.0053, 0.0065, 0.0325, 0.0051, 0.1561, 0.0072, 0.0097], device='cuda:6'), in_proj_covar=tensor([0.0064, 0.0052, 0.0079, 0.0102, 0.0060, 0.0113, 0.0073, 0.0077], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:46:29,415 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18066.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:47:03,301 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18086.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:47:05,057 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18087.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 19:47:25,403 INFO [train.py:904] (6/8) Epoch 2, batch 7950, loss[loss=0.2431, simple_loss=0.3142, pruned_loss=0.08603, over 16857.00 frames. ], tot_loss[loss=0.3019, simple_loss=0.3646, pruned_loss=0.1196, over 3063250.07 frames. ], batch size: 42, lr: 2.53e-02, grad_scale: 8.0 2023-04-27 19:47:56,300 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.877e+02 5.212e+02 6.244e+02 7.914e+02 2.379e+03, threshold=1.249e+03, percent-clipped=3.0 2023-04-27 19:48:31,629 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8262, 1.2965, 1.6000, 1.6721, 1.7378, 1.7179, 1.4963, 1.7650], device='cuda:6'), covar=tensor([0.0047, 0.0143, 0.0078, 0.0087, 0.0036, 0.0046, 0.0110, 0.0039], device='cuda:6'), in_proj_covar=tensor([0.0069, 0.0108, 0.0094, 0.0080, 0.0061, 0.0057, 0.0096, 0.0055], device='cuda:6'), out_proj_covar=tensor([1.1755e-04, 1.8661e-04, 1.6831e-04, 1.4449e-04, 1.0379e-04, 9.7655e-05, 1.6346e-04, 9.3680e-05], device='cuda:6') 2023-04-27 19:48:42,377 INFO [train.py:904] (6/8) Epoch 2, batch 8000, loss[loss=0.298, simple_loss=0.3696, pruned_loss=0.1132, over 16728.00 frames. ], tot_loss[loss=0.3005, simple_loss=0.3635, pruned_loss=0.1188, over 3085338.67 frames. ], batch size: 89, lr: 2.53e-02, grad_scale: 8.0 2023-04-27 19:49:16,800 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18174.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:49:18,165 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18175.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:49:56,233 INFO [train.py:904] (6/8) Epoch 2, batch 8050, loss[loss=0.3, simple_loss=0.3701, pruned_loss=0.115, over 16237.00 frames. ], tot_loss[loss=0.299, simple_loss=0.3628, pruned_loss=0.1176, over 3100794.25 frames. ], batch size: 165, lr: 2.52e-02, grad_scale: 8.0 2023-04-27 19:50:05,433 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-27 19:50:24,924 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.258e+02 4.729e+02 5.978e+02 7.201e+02 1.227e+03, threshold=1.196e+03, percent-clipped=0.0 2023-04-27 19:50:47,404 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18235.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:50:48,729 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18236.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:51:11,140 INFO [train.py:904] (6/8) Epoch 2, batch 8100, loss[loss=0.2975, simple_loss=0.363, pruned_loss=0.116, over 16396.00 frames. ], tot_loss[loss=0.2964, simple_loss=0.3613, pruned_loss=0.1157, over 3133382.43 frames. ], batch size: 68, lr: 2.52e-02, grad_scale: 8.0 2023-04-27 19:51:54,495 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7158, 3.5871, 3.7068, 3.6756, 3.7176, 4.1225, 3.9821, 3.6569], device='cuda:6'), covar=tensor([0.1490, 0.1223, 0.1090, 0.1669, 0.2270, 0.0934, 0.0840, 0.1776], device='cuda:6'), in_proj_covar=tensor([0.0197, 0.0271, 0.0258, 0.0250, 0.0313, 0.0278, 0.0212, 0.0327], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 19:52:29,017 INFO [train.py:904] (6/8) Epoch 2, batch 8150, loss[loss=0.2562, simple_loss=0.3359, pruned_loss=0.08822, over 16494.00 frames. ], tot_loss[loss=0.295, simple_loss=0.3594, pruned_loss=0.1153, over 3133644.22 frames. ], batch size: 68, lr: 2.52e-02, grad_scale: 8.0 2023-04-27 19:52:53,558 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18316.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:53:00,708 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.225e+02 5.223e+02 6.693e+02 9.034e+02 1.543e+03, threshold=1.339e+03, percent-clipped=7.0 2023-04-27 19:53:08,248 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-27 19:53:11,870 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8612, 3.9222, 4.3913, 4.3952, 4.4283, 3.9449, 4.1273, 4.1129], device='cuda:6'), covar=tensor([0.0217, 0.0296, 0.0272, 0.0322, 0.0292, 0.0239, 0.0557, 0.0260], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0149, 0.0163, 0.0160, 0.0195, 0.0166, 0.0251, 0.0149], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-27 19:53:34,745 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.53 vs. limit=5.0 2023-04-27 19:53:42,715 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8501, 3.6842, 3.6767, 3.0999, 3.7109, 1.6716, 3.4560, 3.6618], device='cuda:6'), covar=tensor([0.0076, 0.0067, 0.0080, 0.0305, 0.0062, 0.1441, 0.0086, 0.0123], device='cuda:6'), in_proj_covar=tensor([0.0064, 0.0053, 0.0080, 0.0101, 0.0060, 0.0112, 0.0072, 0.0079], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:53:48,015 INFO [train.py:904] (6/8) Epoch 2, batch 8200, loss[loss=0.3019, simple_loss=0.3685, pruned_loss=0.1177, over 16262.00 frames. ], tot_loss[loss=0.2933, simple_loss=0.3573, pruned_loss=0.1146, over 3133097.66 frames. ], batch size: 165, lr: 2.51e-02, grad_scale: 4.0 2023-04-27 19:53:57,191 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18356.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:54:26,080 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-27 19:54:39,759 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4869, 4.7807, 4.4673, 4.6079, 4.1125, 4.2073, 4.3206, 4.7396], device='cuda:6'), covar=tensor([0.0414, 0.0575, 0.0783, 0.0314, 0.0522, 0.0580, 0.0458, 0.0635], device='cuda:6'), in_proj_covar=tensor([0.0229, 0.0315, 0.0293, 0.0203, 0.0211, 0.0203, 0.0258, 0.0224], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 19:54:46,778 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18386.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:55:10,284 INFO [train.py:904] (6/8) Epoch 2, batch 8250, loss[loss=0.2641, simple_loss=0.3439, pruned_loss=0.09216, over 16637.00 frames. ], tot_loss[loss=0.2915, simple_loss=0.3568, pruned_loss=0.1131, over 3109422.33 frames. ], batch size: 57, lr: 2.51e-02, grad_scale: 4.0 2023-04-27 19:55:15,797 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18404.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:55:44,640 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.019e+02 4.548e+02 5.368e+02 6.842e+02 2.128e+03, threshold=1.074e+03, percent-clipped=3.0 2023-04-27 19:56:05,978 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18434.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:56:33,573 INFO [train.py:904] (6/8) Epoch 2, batch 8300, loss[loss=0.2541, simple_loss=0.3341, pruned_loss=0.08709, over 16893.00 frames. ], tot_loss[loss=0.2842, simple_loss=0.3518, pruned_loss=0.1083, over 3074645.91 frames. ], batch size: 116, lr: 2.51e-02, grad_scale: 4.0 2023-04-27 19:57:02,978 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.46 vs. limit=5.0 2023-04-27 19:57:37,631 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18490.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:57:54,223 INFO [train.py:904] (6/8) Epoch 2, batch 8350, loss[loss=0.2446, simple_loss=0.33, pruned_loss=0.07959, over 16752.00 frames. ], tot_loss[loss=0.2791, simple_loss=0.3491, pruned_loss=0.1046, over 3065733.11 frames. ], batch size: 89, lr: 2.50e-02, grad_scale: 4.0 2023-04-27 19:58:06,782 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18508.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 19:58:16,694 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.69 vs. limit=5.0 2023-04-27 19:58:28,816 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.204e+02 3.865e+02 4.910e+02 6.166e+02 1.131e+03, threshold=9.820e+02, percent-clipped=1.0 2023-04-27 19:58:42,167 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18530.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:58:44,083 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18531.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:59:15,562 INFO [train.py:904] (6/8) Epoch 2, batch 8400, loss[loss=0.2405, simple_loss=0.3223, pruned_loss=0.07931, over 16493.00 frames. ], tot_loss[loss=0.2741, simple_loss=0.3451, pruned_loss=0.1016, over 3052654.83 frames. ], batch size: 75, lr: 2.50e-02, grad_scale: 8.0 2023-04-27 19:59:16,160 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18551.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:59:45,569 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18569.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 20:00:35,049 INFO [train.py:904] (6/8) Epoch 2, batch 8450, loss[loss=0.2163, simple_loss=0.3026, pruned_loss=0.06495, over 16583.00 frames. ], tot_loss[loss=0.2703, simple_loss=0.3423, pruned_loss=0.09913, over 3047494.61 frames. ], batch size: 68, lr: 2.50e-02, grad_scale: 8.0 2023-04-27 20:00:41,882 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8913, 1.1986, 1.5570, 1.7452, 1.8415, 1.7943, 1.4724, 1.8183], device='cuda:6'), covar=tensor([0.0059, 0.0186, 0.0105, 0.0117, 0.0041, 0.0057, 0.0149, 0.0066], device='cuda:6'), in_proj_covar=tensor([0.0069, 0.0110, 0.0093, 0.0083, 0.0061, 0.0057, 0.0096, 0.0056], device='cuda:6'), out_proj_covar=tensor([1.1897e-04, 1.9022e-04, 1.6669e-04, 1.4711e-04, 1.0192e-04, 9.7477e-05, 1.6318e-04, 9.3062e-05], device='cuda:6') 2023-04-27 20:00:52,822 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7834, 3.8302, 3.4595, 2.5458, 2.6970, 2.1090, 4.1314, 4.5453], device='cuda:6'), covar=tensor([0.1849, 0.0528, 0.0716, 0.0772, 0.1753, 0.1406, 0.0214, 0.0118], device='cuda:6'), in_proj_covar=tensor([0.0244, 0.0213, 0.0228, 0.0174, 0.0238, 0.0182, 0.0191, 0.0139], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 20:01:00,300 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18616.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:01:09,359 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.749e+02 4.009e+02 5.012e+02 6.279e+02 1.629e+03, threshold=1.002e+03, percent-clipped=8.0 2023-04-27 20:01:55,557 INFO [train.py:904] (6/8) Epoch 2, batch 8500, loss[loss=0.2677, simple_loss=0.325, pruned_loss=0.1052, over 11995.00 frames. ], tot_loss[loss=0.2632, simple_loss=0.3362, pruned_loss=0.09507, over 3012753.86 frames. ], batch size: 246, lr: 2.49e-02, grad_scale: 8.0 2023-04-27 20:02:17,839 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18664.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:02:34,636 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18674.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:03:19,433 INFO [train.py:904] (6/8) Epoch 2, batch 8550, loss[loss=0.2423, simple_loss=0.3301, pruned_loss=0.07726, over 16862.00 frames. ], tot_loss[loss=0.2601, simple_loss=0.3338, pruned_loss=0.09321, over 3021372.84 frames. ], batch size: 96, lr: 2.49e-02, grad_scale: 4.0 2023-04-27 20:04:03,004 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.327e+02 4.014e+02 4.987e+02 6.688e+02 1.627e+03, threshold=9.973e+02, percent-clipped=3.0 2023-04-27 20:04:26,455 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18735.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:04:58,635 INFO [train.py:904] (6/8) Epoch 2, batch 8600, loss[loss=0.2775, simple_loss=0.3461, pruned_loss=0.1044, over 15419.00 frames. ], tot_loss[loss=0.2602, simple_loss=0.3346, pruned_loss=0.09286, over 3009042.48 frames. ], batch size: 191, lr: 2.49e-02, grad_scale: 2.0 2023-04-27 20:06:37,153 INFO [train.py:904] (6/8) Epoch 2, batch 8650, loss[loss=0.2259, simple_loss=0.3171, pruned_loss=0.06731, over 16805.00 frames. ], tot_loss[loss=0.2552, simple_loss=0.3316, pruned_loss=0.08945, over 3033370.67 frames. ], batch size: 124, lr: 2.49e-02, grad_scale: 2.0 2023-04-27 20:07:29,096 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.199e+02 3.846e+02 4.801e+02 6.827e+02 6.621e+03, threshold=9.601e+02, percent-clipped=16.0 2023-04-27 20:07:44,470 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18830.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:07:46,554 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18831.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:08:14,641 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18846.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:08:23,077 INFO [train.py:904] (6/8) Epoch 2, batch 8700, loss[loss=0.2325, simple_loss=0.3186, pruned_loss=0.07321, over 16661.00 frames. ], tot_loss[loss=0.251, simple_loss=0.3281, pruned_loss=0.08699, over 3044209.77 frames. ], batch size: 134, lr: 2.48e-02, grad_scale: 2.0 2023-04-27 20:08:49,493 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18864.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 20:09:13,490 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9285, 4.7853, 4.6012, 4.7196, 4.2687, 4.5924, 4.5912, 4.4731], device='cuda:6'), covar=tensor([0.0228, 0.0125, 0.0154, 0.0095, 0.0529, 0.0166, 0.0162, 0.0193], device='cuda:6'), in_proj_covar=tensor([0.0117, 0.0092, 0.0143, 0.0115, 0.0168, 0.0123, 0.0102, 0.0136], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 20:09:14,853 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18878.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:09:17,249 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18879.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:10:01,022 INFO [train.py:904] (6/8) Epoch 2, batch 8750, loss[loss=0.2079, simple_loss=0.2924, pruned_loss=0.06167, over 12451.00 frames. ], tot_loss[loss=0.2489, simple_loss=0.3268, pruned_loss=0.08554, over 3040978.61 frames. ], batch size: 250, lr: 2.48e-02, grad_scale: 2.0 2023-04-27 20:10:57,516 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.382e+02 3.853e+02 4.840e+02 6.218e+02 1.090e+03, threshold=9.680e+02, percent-clipped=4.0 2023-04-27 20:11:10,245 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18930.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:11:53,566 INFO [train.py:904] (6/8) Epoch 2, batch 8800, loss[loss=0.2674, simple_loss=0.3356, pruned_loss=0.09965, over 13002.00 frames. ], tot_loss[loss=0.2471, simple_loss=0.3252, pruned_loss=0.0845, over 3036146.35 frames. ], batch size: 250, lr: 2.48e-02, grad_scale: 4.0 2023-04-27 20:13:18,251 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18991.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:13:38,729 INFO [train.py:904] (6/8) Epoch 2, batch 8850, loss[loss=0.2247, simple_loss=0.3154, pruned_loss=0.06698, over 15281.00 frames. ], tot_loss[loss=0.2465, simple_loss=0.3267, pruned_loss=0.08314, over 3024728.65 frames. ], batch size: 191, lr: 2.47e-02, grad_scale: 4.0 2023-04-27 20:14:28,277 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.418e+02 3.744e+02 4.791e+02 5.972e+02 1.134e+03, threshold=9.582e+02, percent-clipped=2.0 2023-04-27 20:14:43,699 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19030.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:15:27,232 INFO [train.py:904] (6/8) Epoch 2, batch 8900, loss[loss=0.2492, simple_loss=0.3309, pruned_loss=0.08377, over 15340.00 frames. ], tot_loss[loss=0.2454, simple_loss=0.3266, pruned_loss=0.08213, over 3040404.09 frames. ], batch size: 191, lr: 2.47e-02, grad_scale: 4.0 2023-04-27 20:17:32,215 INFO [train.py:904] (6/8) Epoch 2, batch 8950, loss[loss=0.281, simple_loss=0.345, pruned_loss=0.1085, over 12598.00 frames. ], tot_loss[loss=0.2454, simple_loss=0.3261, pruned_loss=0.08236, over 3043678.64 frames. ], batch size: 248, lr: 2.47e-02, grad_scale: 4.0 2023-04-27 20:18:01,917 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6846, 3.4673, 3.6808, 3.9982, 3.9968, 3.5699, 3.9697, 3.9667], device='cuda:6'), covar=tensor([0.0501, 0.0543, 0.1006, 0.0295, 0.0334, 0.0810, 0.0331, 0.0307], device='cuda:6'), in_proj_covar=tensor([0.0232, 0.0280, 0.0369, 0.0271, 0.0210, 0.0192, 0.0219, 0.0223], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 20:18:20,869 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.564e+02 4.048e+02 4.879e+02 5.848e+02 1.298e+03, threshold=9.758e+02, percent-clipped=2.0 2023-04-27 20:18:32,648 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19129.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 20:19:11,801 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19146.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:19:16,518 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5985, 2.5520, 2.5265, 3.9660, 1.8665, 3.7057, 2.3318, 2.3690], device='cuda:6'), covar=tensor([0.0290, 0.0548, 0.0349, 0.0176, 0.1597, 0.0179, 0.0745, 0.1214], device='cuda:6'), in_proj_covar=tensor([0.0214, 0.0193, 0.0161, 0.0214, 0.0266, 0.0172, 0.0195, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 20:19:21,316 INFO [train.py:904] (6/8) Epoch 2, batch 9000, loss[loss=0.236, simple_loss=0.3098, pruned_loss=0.08112, over 16674.00 frames. ], tot_loss[loss=0.2419, simple_loss=0.3227, pruned_loss=0.08056, over 3059081.08 frames. ], batch size: 134, lr: 2.46e-02, grad_scale: 2.0 2023-04-27 20:19:21,316 INFO [train.py:929] (6/8) Computing validation loss 2023-04-27 20:19:31,144 INFO [train.py:938] (6/8) Epoch 2, validation: loss=0.2044, simple_loss=0.3047, pruned_loss=0.05205, over 944034.00 frames. 2023-04-27 20:19:31,145 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-27 20:20:00,849 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19164.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 20:20:52,973 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19190.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 20:21:02,822 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=19194.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:21:13,856 INFO [train.py:904] (6/8) Epoch 2, batch 9050, loss[loss=0.2325, simple_loss=0.3115, pruned_loss=0.07676, over 16380.00 frames. ], tot_loss[loss=0.2439, simple_loss=0.3246, pruned_loss=0.08159, over 3056982.32 frames. ], batch size: 146, lr: 2.46e-02, grad_scale: 2.0 2023-04-27 20:21:37,536 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=19212.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 20:22:01,198 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.755e+02 4.212e+02 4.597e+02 5.989e+02 9.578e+02, threshold=9.193e+02, percent-clipped=0.0 2023-04-27 20:22:07,308 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8840, 3.7907, 3.7552, 3.3101, 3.8038, 1.6953, 3.5335, 3.6748], device='cuda:6'), covar=tensor([0.0075, 0.0064, 0.0081, 0.0216, 0.0054, 0.1467, 0.0089, 0.0117], device='cuda:6'), in_proj_covar=tensor([0.0060, 0.0049, 0.0076, 0.0084, 0.0056, 0.0109, 0.0070, 0.0074], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-27 20:22:58,686 INFO [train.py:904] (6/8) Epoch 2, batch 9100, loss[loss=0.2702, simple_loss=0.3321, pruned_loss=0.1042, over 12277.00 frames. ], tot_loss[loss=0.2447, simple_loss=0.3245, pruned_loss=0.08244, over 3051995.39 frames. ], batch size: 250, lr: 2.46e-02, grad_scale: 2.0 2023-04-27 20:24:23,844 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19286.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:24:55,898 INFO [train.py:904] (6/8) Epoch 2, batch 9150, loss[loss=0.2285, simple_loss=0.314, pruned_loss=0.07154, over 16694.00 frames. ], tot_loss[loss=0.2446, simple_loss=0.325, pruned_loss=0.08211, over 3041113.45 frames. ], batch size: 134, lr: 2.46e-02, grad_scale: 2.0 2023-04-27 20:25:05,795 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1526, 3.3141, 3.3384, 1.4518, 3.5023, 3.5486, 2.7975, 2.8315], device='cuda:6'), covar=tensor([0.0900, 0.0138, 0.0174, 0.1500, 0.0066, 0.0052, 0.0427, 0.0360], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0082, 0.0078, 0.0145, 0.0069, 0.0068, 0.0105, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0001, 0.0002, 0.0003], device='cuda:6') 2023-04-27 20:25:14,928 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3864, 1.7166, 2.4043, 3.0709, 3.0198, 3.2064, 1.9048, 3.1087], device='cuda:6'), covar=tensor([0.0031, 0.0214, 0.0101, 0.0073, 0.0041, 0.0074, 0.0160, 0.0048], device='cuda:6'), in_proj_covar=tensor([0.0072, 0.0110, 0.0097, 0.0084, 0.0065, 0.0060, 0.0100, 0.0056], device='cuda:6'), out_proj_covar=tensor([1.2302e-04, 1.8742e-04, 1.7163e-04, 1.4748e-04, 1.0831e-04, 1.0070e-04, 1.6760e-04, 9.1337e-05], device='cuda:6') 2023-04-27 20:25:49,332 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.714e+02 4.631e+02 5.471e+02 6.935e+02 2.154e+03, threshold=1.094e+03, percent-clipped=5.0 2023-04-27 20:25:59,077 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19330.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:26:40,138 INFO [train.py:904] (6/8) Epoch 2, batch 9200, loss[loss=0.2018, simple_loss=0.2807, pruned_loss=0.06147, over 16596.00 frames. ], tot_loss[loss=0.2393, simple_loss=0.3189, pruned_loss=0.07988, over 3039754.99 frames. ], batch size: 57, lr: 2.45e-02, grad_scale: 4.0 2023-04-27 20:27:30,641 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=19378.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:28:03,622 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19394.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:28:16,032 INFO [train.py:904] (6/8) Epoch 2, batch 9250, loss[loss=0.2038, simple_loss=0.2782, pruned_loss=0.06472, over 11928.00 frames. ], tot_loss[loss=0.2389, simple_loss=0.3181, pruned_loss=0.07983, over 3026544.37 frames. ], batch size: 246, lr: 2.45e-02, grad_scale: 4.0 2023-04-27 20:29:05,882 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.723e+02 4.011e+02 4.767e+02 6.782e+02 2.707e+03, threshold=9.534e+02, percent-clipped=7.0 2023-04-27 20:30:06,013 INFO [train.py:904] (6/8) Epoch 2, batch 9300, loss[loss=0.2357, simple_loss=0.3065, pruned_loss=0.08244, over 12372.00 frames. ], tot_loss[loss=0.2365, simple_loss=0.316, pruned_loss=0.0785, over 3032125.54 frames. ], batch size: 248, lr: 2.45e-02, grad_scale: 4.0 2023-04-27 20:30:14,338 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19455.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:31:20,841 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19485.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 20:31:31,780 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1010, 3.8842, 3.8817, 3.4839, 3.8560, 1.4882, 3.6767, 3.8636], device='cuda:6'), covar=tensor([0.0062, 0.0058, 0.0079, 0.0198, 0.0055, 0.1458, 0.0068, 0.0107], device='cuda:6'), in_proj_covar=tensor([0.0060, 0.0050, 0.0074, 0.0083, 0.0057, 0.0109, 0.0068, 0.0074], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-27 20:31:49,019 INFO [train.py:904] (6/8) Epoch 2, batch 9350, loss[loss=0.2404, simple_loss=0.3082, pruned_loss=0.08629, over 12485.00 frames. ], tot_loss[loss=0.2372, simple_loss=0.3165, pruned_loss=0.07891, over 3027370.11 frames. ], batch size: 246, lr: 2.44e-02, grad_scale: 4.0 2023-04-27 20:32:37,318 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.597e+02 3.883e+02 4.622e+02 6.098e+02 2.142e+03, threshold=9.245e+02, percent-clipped=3.0 2023-04-27 20:33:28,460 INFO [train.py:904] (6/8) Epoch 2, batch 9400, loss[loss=0.1976, simple_loss=0.2732, pruned_loss=0.06106, over 12506.00 frames. ], tot_loss[loss=0.2358, simple_loss=0.3157, pruned_loss=0.07791, over 3027870.69 frames. ], batch size: 250, lr: 2.44e-02, grad_scale: 4.0 2023-04-27 20:34:39,310 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19586.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:35:08,265 INFO [train.py:904] (6/8) Epoch 2, batch 9450, loss[loss=0.2143, simple_loss=0.2984, pruned_loss=0.06506, over 17142.00 frames. ], tot_loss[loss=0.2373, simple_loss=0.3179, pruned_loss=0.07834, over 3040126.84 frames. ], batch size: 48, lr: 2.44e-02, grad_scale: 4.0 2023-04-27 20:35:42,249 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19618.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:35:56,417 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.451e+02 4.122e+02 5.164e+02 6.390e+02 1.240e+03, threshold=1.033e+03, percent-clipped=6.0 2023-04-27 20:36:16,202 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=19634.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:36:48,218 INFO [train.py:904] (6/8) Epoch 2, batch 9500, loss[loss=0.2451, simple_loss=0.3216, pruned_loss=0.08426, over 15393.00 frames. ], tot_loss[loss=0.2361, simple_loss=0.317, pruned_loss=0.07761, over 3040173.41 frames. ], batch size: 191, lr: 2.43e-02, grad_scale: 4.0 2023-04-27 20:37:20,809 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-04-27 20:37:46,313 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19679.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:38:34,553 INFO [train.py:904] (6/8) Epoch 2, batch 9550, loss[loss=0.241, simple_loss=0.3149, pruned_loss=0.08352, over 12359.00 frames. ], tot_loss[loss=0.2371, simple_loss=0.3177, pruned_loss=0.07823, over 3044563.99 frames. ], batch size: 246, lr: 2.43e-02, grad_scale: 4.0 2023-04-27 20:38:35,731 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5941, 4.9525, 4.7217, 4.8170, 4.2945, 4.2182, 4.4836, 4.9785], device='cuda:6'), covar=tensor([0.0381, 0.0686, 0.0808, 0.0292, 0.0548, 0.0547, 0.0411, 0.0565], device='cuda:6'), in_proj_covar=tensor([0.0218, 0.0305, 0.0259, 0.0189, 0.0203, 0.0190, 0.0239, 0.0208], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 20:39:08,558 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2023-04-27 20:39:23,642 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.975e+02 3.908e+02 4.753e+02 5.956e+02 1.328e+03, threshold=9.507e+02, percent-clipped=2.0 2023-04-27 20:40:12,704 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19750.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:40:13,477 INFO [train.py:904] (6/8) Epoch 2, batch 9600, loss[loss=0.2597, simple_loss=0.3479, pruned_loss=0.08571, over 15483.00 frames. ], tot_loss[loss=0.2393, simple_loss=0.3197, pruned_loss=0.07941, over 3054639.67 frames. ], batch size: 191, lr: 2.43e-02, grad_scale: 8.0 2023-04-27 20:41:15,665 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-27 20:41:22,113 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19785.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 20:41:54,465 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1179, 2.9550, 2.6285, 4.5580, 1.9986, 4.3897, 2.6907, 2.5350], device='cuda:6'), covar=tensor([0.0224, 0.0473, 0.0338, 0.0149, 0.1587, 0.0120, 0.0619, 0.1195], device='cuda:6'), in_proj_covar=tensor([0.0215, 0.0196, 0.0159, 0.0218, 0.0264, 0.0172, 0.0193, 0.0240], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 20:42:00,140 INFO [train.py:904] (6/8) Epoch 2, batch 9650, loss[loss=0.2246, simple_loss=0.3016, pruned_loss=0.07378, over 12211.00 frames. ], tot_loss[loss=0.2407, simple_loss=0.3218, pruned_loss=0.07984, over 3054811.83 frames. ], batch size: 250, lr: 2.43e-02, grad_scale: 8.0 2023-04-27 20:42:54,373 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.395e+02 4.451e+02 5.149e+02 6.499e+02 1.556e+03, threshold=1.030e+03, percent-clipped=6.0 2023-04-27 20:43:10,774 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=19833.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 20:43:48,656 INFO [train.py:904] (6/8) Epoch 2, batch 9700, loss[loss=0.2426, simple_loss=0.3224, pruned_loss=0.08136, over 16176.00 frames. ], tot_loss[loss=0.2396, simple_loss=0.3203, pruned_loss=0.07944, over 3067595.06 frames. ], batch size: 165, lr: 2.42e-02, grad_scale: 8.0 2023-04-27 20:44:29,094 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2023-04-27 20:44:56,987 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9101, 3.6615, 3.7999, 4.1025, 4.1009, 3.6949, 4.1885, 4.1094], device='cuda:6'), covar=tensor([0.0417, 0.0517, 0.0929, 0.0367, 0.0364, 0.0685, 0.0302, 0.0291], device='cuda:6'), in_proj_covar=tensor([0.0242, 0.0278, 0.0366, 0.0274, 0.0216, 0.0191, 0.0218, 0.0222], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 20:45:33,548 INFO [train.py:904] (6/8) Epoch 2, batch 9750, loss[loss=0.2351, simple_loss=0.3017, pruned_loss=0.08425, over 12108.00 frames. ], tot_loss[loss=0.2389, simple_loss=0.3188, pruned_loss=0.07943, over 3083297.66 frames. ], batch size: 248, lr: 2.42e-02, grad_scale: 8.0 2023-04-27 20:46:21,255 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.379e+02 3.927e+02 4.879e+02 5.766e+02 1.284e+03, threshold=9.758e+02, percent-clipped=1.0 2023-04-27 20:47:15,375 INFO [train.py:904] (6/8) Epoch 2, batch 9800, loss[loss=0.2489, simple_loss=0.3419, pruned_loss=0.07802, over 16398.00 frames. ], tot_loss[loss=0.2356, simple_loss=0.3173, pruned_loss=0.0769, over 3086531.77 frames. ], batch size: 146, lr: 2.42e-02, grad_scale: 8.0 2023-04-27 20:47:35,374 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-27 20:48:00,423 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19974.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:49:05,346 INFO [train.py:904] (6/8) Epoch 2, batch 9850, loss[loss=0.2472, simple_loss=0.3241, pruned_loss=0.08518, over 16803.00 frames. ], tot_loss[loss=0.2358, simple_loss=0.3184, pruned_loss=0.07663, over 3103417.37 frames. ], batch size: 124, lr: 2.41e-02, grad_scale: 8.0 2023-04-27 20:49:35,823 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.53 vs. limit=5.0 2023-04-27 20:49:46,063 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20020.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:49:46,216 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3057, 2.9803, 2.6471, 2.3358, 2.1177, 1.9896, 2.8872, 3.1746], device='cuda:6'), covar=tensor([0.1215, 0.0551, 0.0828, 0.0639, 0.1537, 0.1214, 0.0336, 0.0210], device='cuda:6'), in_proj_covar=tensor([0.0245, 0.0222, 0.0236, 0.0180, 0.0204, 0.0178, 0.0196, 0.0145], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 20:49:55,601 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.468e+02 4.069e+02 4.622e+02 5.694e+02 1.429e+03, threshold=9.244e+02, percent-clipped=4.0 2023-04-27 20:50:58,204 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20050.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:50:58,949 INFO [train.py:904] (6/8) Epoch 2, batch 9900, loss[loss=0.2382, simple_loss=0.3081, pruned_loss=0.0842, over 12320.00 frames. ], tot_loss[loss=0.2361, simple_loss=0.3186, pruned_loss=0.07675, over 3077285.37 frames. ], batch size: 248, lr: 2.41e-02, grad_scale: 8.0 2023-04-27 20:52:07,419 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4183, 3.4428, 3.2142, 3.3137, 3.0093, 3.3502, 3.2273, 3.1273], device='cuda:6'), covar=tensor([0.0271, 0.0141, 0.0181, 0.0147, 0.0479, 0.0163, 0.0596, 0.0279], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0091, 0.0139, 0.0116, 0.0164, 0.0123, 0.0099, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 20:52:12,594 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20081.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:52:31,317 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20089.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:52:51,146 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=20098.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:52:56,915 INFO [train.py:904] (6/8) Epoch 2, batch 9950, loss[loss=0.2247, simple_loss=0.3048, pruned_loss=0.07232, over 16665.00 frames. ], tot_loss[loss=0.2375, simple_loss=0.3207, pruned_loss=0.07711, over 3077408.64 frames. ], batch size: 57, lr: 2.41e-02, grad_scale: 8.0 2023-04-27 20:53:12,427 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-27 20:53:24,155 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20112.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:53:55,568 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.288e+02 4.201e+02 4.972e+02 6.369e+02 1.989e+03, threshold=9.943e+02, percent-clipped=8.0 2023-04-27 20:54:56,016 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20150.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:54:56,740 INFO [train.py:904] (6/8) Epoch 2, batch 10000, loss[loss=0.2227, simple_loss=0.3159, pruned_loss=0.06476, over 15518.00 frames. ], tot_loss[loss=0.2344, simple_loss=0.3181, pruned_loss=0.07536, over 3097220.18 frames. ], batch size: 192, lr: 2.41e-02, grad_scale: 8.0 2023-04-27 20:55:09,537 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8476, 2.7632, 1.6300, 2.8781, 2.0810, 2.7933, 1.8891, 2.4171], device='cuda:6'), covar=tensor([0.0084, 0.0211, 0.1225, 0.0067, 0.0683, 0.0406, 0.1088, 0.0513], device='cuda:6'), in_proj_covar=tensor([0.0076, 0.0110, 0.0168, 0.0073, 0.0153, 0.0134, 0.0173, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-27 20:55:25,532 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-27 20:55:41,386 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20173.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:56:11,875 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9384, 3.7930, 3.2986, 1.5719, 2.6533, 2.2003, 3.1951, 3.8651], device='cuda:6'), covar=tensor([0.0273, 0.0387, 0.0503, 0.1808, 0.0784, 0.1006, 0.0848, 0.0394], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0099, 0.0152, 0.0147, 0.0137, 0.0132, 0.0142, 0.0103], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 20:56:37,810 INFO [train.py:904] (6/8) Epoch 2, batch 10050, loss[loss=0.2426, simple_loss=0.3197, pruned_loss=0.08276, over 12257.00 frames. ], tot_loss[loss=0.2344, simple_loss=0.318, pruned_loss=0.07539, over 3076779.74 frames. ], batch size: 248, lr: 2.40e-02, grad_scale: 8.0 2023-04-27 20:56:44,603 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20204.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:57:24,817 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.701e+02 4.201e+02 4.991e+02 6.585e+02 1.392e+03, threshold=9.982e+02, percent-clipped=3.0 2023-04-27 20:58:10,189 INFO [train.py:904] (6/8) Epoch 2, batch 10100, loss[loss=0.2282, simple_loss=0.3101, pruned_loss=0.07309, over 16744.00 frames. ], tot_loss[loss=0.2357, simple_loss=0.3191, pruned_loss=0.07617, over 3093132.31 frames. ], batch size: 134, lr: 2.40e-02, grad_scale: 8.0 2023-04-27 20:58:37,994 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20265.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:58:58,929 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20274.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:59:54,874 INFO [train.py:904] (6/8) Epoch 3, batch 0, loss[loss=0.2414, simple_loss=0.3081, pruned_loss=0.08736, over 16880.00 frames. ], tot_loss[loss=0.2414, simple_loss=0.3081, pruned_loss=0.08736, over 16880.00 frames. ], batch size: 42, lr: 2.28e-02, grad_scale: 8.0 2023-04-27 20:59:54,874 INFO [train.py:929] (6/8) Computing validation loss 2023-04-27 21:00:02,299 INFO [train.py:938] (6/8) Epoch 3, validation: loss=0.2012, simple_loss=0.3019, pruned_loss=0.05024, over 944034.00 frames. 2023-04-27 21:00:02,299 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-27 21:00:17,807 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-27 21:00:31,873 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=20322.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:00:38,802 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.937e+02 4.805e+02 5.776e+02 7.144e+02 1.042e+03, threshold=1.155e+03, percent-clipped=3.0 2023-04-27 21:01:12,906 INFO [train.py:904] (6/8) Epoch 3, batch 50, loss[loss=0.3018, simple_loss=0.3506, pruned_loss=0.1265, over 16891.00 frames. ], tot_loss[loss=0.2916, simple_loss=0.3484, pruned_loss=0.1174, over 745644.20 frames. ], batch size: 96, lr: 2.28e-02, grad_scale: 2.0 2023-04-27 21:01:41,120 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7796, 5.3803, 5.2822, 5.3045, 5.2254, 5.8023, 5.4119, 5.1649], device='cuda:6'), covar=tensor([0.0663, 0.1014, 0.0911, 0.1310, 0.1996, 0.0773, 0.0899, 0.2071], device='cuda:6'), in_proj_covar=tensor([0.0197, 0.0294, 0.0270, 0.0262, 0.0324, 0.0285, 0.0220, 0.0337], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 21:01:45,732 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20376.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:02:02,792 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20388.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:02:07,383 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-27 21:02:19,621 INFO [train.py:904] (6/8) Epoch 3, batch 100, loss[loss=0.2522, simple_loss=0.3331, pruned_loss=0.08561, over 17240.00 frames. ], tot_loss[loss=0.2835, simple_loss=0.3434, pruned_loss=0.1118, over 1311304.17 frames. ], batch size: 52, lr: 2.27e-02, grad_scale: 2.0 2023-04-27 21:02:43,271 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20418.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:02:54,676 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.729e+02 4.335e+02 5.443e+02 6.351e+02 1.481e+03, threshold=1.089e+03, percent-clipped=3.0 2023-04-27 21:03:06,817 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2338, 4.7109, 4.7190, 2.1922, 4.9013, 5.0297, 3.7530, 3.8336], device='cuda:6'), covar=tensor([0.0581, 0.0091, 0.0141, 0.1222, 0.0044, 0.0032, 0.0249, 0.0295], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0081, 0.0078, 0.0145, 0.0073, 0.0068, 0.0108, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 21:03:18,756 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20445.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:03:24,415 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20449.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:03:27,023 INFO [train.py:904] (6/8) Epoch 3, batch 150, loss[loss=0.3067, simple_loss=0.3541, pruned_loss=0.1297, over 16709.00 frames. ], tot_loss[loss=0.2782, simple_loss=0.34, pruned_loss=0.1082, over 1751398.62 frames. ], batch size: 134, lr: 2.27e-02, grad_scale: 2.0 2023-04-27 21:03:49,809 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20468.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:04:05,908 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20479.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:04:10,417 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6751, 4.6802, 5.3301, 5.3293, 5.3493, 4.8382, 4.9118, 4.6918], device='cuda:6'), covar=tensor([0.0211, 0.0217, 0.0314, 0.0290, 0.0331, 0.0206, 0.0548, 0.0259], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0161, 0.0182, 0.0180, 0.0210, 0.0185, 0.0273, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-27 21:04:35,430 INFO [train.py:904] (6/8) Epoch 3, batch 200, loss[loss=0.2825, simple_loss=0.3384, pruned_loss=0.1133, over 16461.00 frames. ], tot_loss[loss=0.2759, simple_loss=0.3373, pruned_loss=0.1073, over 2103034.51 frames. ], batch size: 75, lr: 2.27e-02, grad_scale: 2.0 2023-04-27 21:05:09,760 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.165e+02 3.891e+02 4.616e+02 5.990e+02 1.220e+03, threshold=9.233e+02, percent-clipped=2.0 2023-04-27 21:05:40,089 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2209, 2.0072, 1.7448, 1.9643, 2.5982, 2.5412, 2.5192, 2.7923], device='cuda:6'), covar=tensor([0.0042, 0.0148, 0.0152, 0.0165, 0.0075, 0.0114, 0.0058, 0.0065], device='cuda:6'), in_proj_covar=tensor([0.0055, 0.0110, 0.0108, 0.0109, 0.0101, 0.0109, 0.0062, 0.0080], device='cuda:6'), out_proj_covar=tensor([7.5675e-05, 1.6621e-04, 1.5775e-04, 1.6373e-04, 1.5462e-04, 1.6689e-04, 9.2119e-05, 1.2529e-04], device='cuda:6') 2023-04-27 21:05:43,636 INFO [train.py:904] (6/8) Epoch 3, batch 250, loss[loss=0.2139, simple_loss=0.2884, pruned_loss=0.06971, over 16780.00 frames. ], tot_loss[loss=0.2706, simple_loss=0.3336, pruned_loss=0.1038, over 2364945.30 frames. ], batch size: 39, lr: 2.27e-02, grad_scale: 2.0 2023-04-27 21:05:48,012 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3712, 4.2243, 4.2794, 4.6521, 4.7217, 4.1836, 4.6776, 4.6262], device='cuda:6'), covar=tensor([0.0478, 0.0480, 0.1084, 0.0328, 0.0376, 0.0624, 0.0290, 0.0310], device='cuda:6'), in_proj_covar=tensor([0.0281, 0.0323, 0.0443, 0.0329, 0.0250, 0.0230, 0.0245, 0.0258], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 21:05:58,364 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20560.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:06:00,926 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7074, 3.2491, 3.5031, 2.5462, 3.4230, 3.3972, 3.5066, 1.8038], device='cuda:6'), covar=tensor([0.0449, 0.0040, 0.0042, 0.0238, 0.0048, 0.0074, 0.0030, 0.0382], device='cuda:6'), in_proj_covar=tensor([0.0113, 0.0056, 0.0060, 0.0109, 0.0055, 0.0062, 0.0061, 0.0107], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 21:06:28,642 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2614, 5.0883, 5.0587, 5.1295, 4.3477, 5.0664, 5.1963, 4.7090], device='cuda:6'), covar=tensor([0.0292, 0.0163, 0.0156, 0.0099, 0.0907, 0.0195, 0.0137, 0.0245], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0109, 0.0165, 0.0137, 0.0204, 0.0148, 0.0117, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 21:06:35,591 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20588.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:06:54,936 INFO [train.py:904] (6/8) Epoch 3, batch 300, loss[loss=0.2262, simple_loss=0.3018, pruned_loss=0.07534, over 17119.00 frames. ], tot_loss[loss=0.2637, simple_loss=0.3286, pruned_loss=0.09935, over 2578867.66 frames. ], batch size: 47, lr: 2.26e-02, grad_scale: 2.0 2023-04-27 21:07:28,557 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-27 21:07:29,013 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.643e+02 3.919e+02 4.746e+02 5.726e+02 1.058e+03, threshold=9.491e+02, percent-clipped=1.0 2023-04-27 21:07:59,147 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20649.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:08:01,102 INFO [train.py:904] (6/8) Epoch 3, batch 350, loss[loss=0.2404, simple_loss=0.3073, pruned_loss=0.08677, over 15566.00 frames. ], tot_loss[loss=0.2581, simple_loss=0.3244, pruned_loss=0.09589, over 2741049.71 frames. ], batch size: 191, lr: 2.26e-02, grad_scale: 2.0 2023-04-27 21:08:36,925 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20676.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:09:09,380 INFO [train.py:904] (6/8) Epoch 3, batch 400, loss[loss=0.2903, simple_loss=0.3405, pruned_loss=0.1201, over 12290.00 frames. ], tot_loss[loss=0.2557, simple_loss=0.3223, pruned_loss=0.09457, over 2861220.93 frames. ], batch size: 247, lr: 2.26e-02, grad_scale: 4.0 2023-04-27 21:09:41,233 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=20724.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:09:45,245 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.989e+02 4.408e+02 5.242e+02 6.457e+02 1.269e+03, threshold=1.048e+03, percent-clipped=5.0 2023-04-27 21:09:58,623 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5461, 4.1636, 3.8550, 1.9818, 2.9565, 2.3970, 3.6841, 4.1593], device='cuda:6'), covar=tensor([0.0201, 0.0446, 0.0376, 0.1566, 0.0699, 0.0945, 0.0609, 0.0528], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0107, 0.0150, 0.0145, 0.0136, 0.0131, 0.0142, 0.0109], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 21:10:08,712 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20744.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:10:10,652 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20745.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:10:17,850 INFO [train.py:904] (6/8) Epoch 3, batch 450, loss[loss=0.2477, simple_loss=0.3068, pruned_loss=0.09424, over 16861.00 frames. ], tot_loss[loss=0.2517, simple_loss=0.3188, pruned_loss=0.09229, over 2951612.92 frames. ], batch size: 96, lr: 2.25e-02, grad_scale: 4.0 2023-04-27 21:10:30,466 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-04-27 21:10:42,543 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20768.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:10:49,445 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20774.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:11:14,838 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=20793.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:11:24,055 INFO [train.py:904] (6/8) Epoch 3, batch 500, loss[loss=0.2495, simple_loss=0.3271, pruned_loss=0.08594, over 17043.00 frames. ], tot_loss[loss=0.2503, simple_loss=0.3174, pruned_loss=0.09165, over 3028545.69 frames. ], batch size: 55, lr: 2.25e-02, grad_scale: 4.0 2023-04-27 21:11:46,162 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=20816.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:12:00,585 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.032e+02 4.223e+02 4.982e+02 5.898e+02 1.253e+03, threshold=9.964e+02, percent-clipped=1.0 2023-04-27 21:12:34,373 INFO [train.py:904] (6/8) Epoch 3, batch 550, loss[loss=0.2621, simple_loss=0.3375, pruned_loss=0.09331, over 17032.00 frames. ], tot_loss[loss=0.2493, simple_loss=0.3161, pruned_loss=0.09124, over 3100283.03 frames. ], batch size: 55, lr: 2.25e-02, grad_scale: 4.0 2023-04-27 21:12:45,299 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20860.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:13:40,832 INFO [train.py:904] (6/8) Epoch 3, batch 600, loss[loss=0.256, simple_loss=0.3369, pruned_loss=0.08752, over 17124.00 frames. ], tot_loss[loss=0.2486, simple_loss=0.3152, pruned_loss=0.09097, over 3144774.65 frames. ], batch size: 49, lr: 2.25e-02, grad_scale: 4.0 2023-04-27 21:13:50,650 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=20908.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:14:15,688 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.677e+02 4.207e+02 5.314e+02 6.523e+02 1.813e+03, threshold=1.063e+03, percent-clipped=4.0 2023-04-27 21:14:38,951 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20944.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:14:48,031 INFO [train.py:904] (6/8) Epoch 3, batch 650, loss[loss=0.1998, simple_loss=0.2751, pruned_loss=0.06224, over 16799.00 frames. ], tot_loss[loss=0.2459, simple_loss=0.3128, pruned_loss=0.08951, over 3180802.17 frames. ], batch size: 39, lr: 2.24e-02, grad_scale: 4.0 2023-04-27 21:14:56,042 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-27 21:15:18,349 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8493, 2.6820, 2.6806, 4.2886, 2.1258, 3.9159, 2.4563, 2.5517], device='cuda:6'), covar=tensor([0.0279, 0.0556, 0.0358, 0.0164, 0.1427, 0.0201, 0.0735, 0.1038], device='cuda:6'), in_proj_covar=tensor([0.0234, 0.0210, 0.0175, 0.0234, 0.0281, 0.0183, 0.0207, 0.0272], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 21:15:53,469 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.69 vs. limit=5.0 2023-04-27 21:15:57,311 INFO [train.py:904] (6/8) Epoch 3, batch 700, loss[loss=0.2237, simple_loss=0.3059, pruned_loss=0.07074, over 17039.00 frames. ], tot_loss[loss=0.2455, simple_loss=0.3124, pruned_loss=0.08936, over 3219721.56 frames. ], batch size: 50, lr: 2.24e-02, grad_scale: 4.0 2023-04-27 21:15:58,106 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.74 vs. limit=5.0 2023-04-27 21:16:02,563 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-27 21:16:12,656 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6434, 4.3497, 4.5214, 4.8957, 5.0012, 4.3116, 4.9801, 4.8857], device='cuda:6'), covar=tensor([0.0515, 0.0646, 0.1145, 0.0383, 0.0334, 0.0575, 0.0296, 0.0317], device='cuda:6'), in_proj_covar=tensor([0.0289, 0.0338, 0.0456, 0.0344, 0.0262, 0.0239, 0.0256, 0.0272], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 21:16:30,868 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.136e+02 3.749e+02 4.303e+02 5.331e+02 1.033e+03, threshold=8.606e+02, percent-clipped=0.0 2023-04-27 21:16:38,531 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5055, 5.8627, 5.5307, 5.6507, 4.9855, 4.7266, 5.3182, 5.9704], device='cuda:6'), covar=tensor([0.0423, 0.0510, 0.0758, 0.0358, 0.0537, 0.0493, 0.0406, 0.0483], device='cuda:6'), in_proj_covar=tensor([0.0259, 0.0373, 0.0325, 0.0227, 0.0245, 0.0225, 0.0291, 0.0247], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 21:16:40,912 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9696, 2.8224, 2.8123, 1.6652, 2.8746, 2.9053, 2.5199, 2.5090], device='cuda:6'), covar=tensor([0.0672, 0.0102, 0.0141, 0.1022, 0.0085, 0.0079, 0.0300, 0.0328], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0080, 0.0078, 0.0147, 0.0074, 0.0072, 0.0112, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 21:16:55,531 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21044.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:17:03,827 INFO [train.py:904] (6/8) Epoch 3, batch 750, loss[loss=0.2368, simple_loss=0.311, pruned_loss=0.08126, over 16691.00 frames. ], tot_loss[loss=0.2455, simple_loss=0.3128, pruned_loss=0.08905, over 3242516.14 frames. ], batch size: 57, lr: 2.24e-02, grad_scale: 4.0 2023-04-27 21:17:06,023 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21052.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:17:16,710 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-27 21:17:35,031 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21074.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:17:55,680 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3251, 3.3130, 3.3689, 2.7561, 3.3326, 2.1129, 3.0457, 3.0184], device='cuda:6'), covar=tensor([0.0103, 0.0081, 0.0101, 0.0340, 0.0092, 0.1369, 0.0101, 0.0165], device='cuda:6'), in_proj_covar=tensor([0.0075, 0.0062, 0.0095, 0.0109, 0.0069, 0.0120, 0.0082, 0.0095], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 21:17:58,573 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=21092.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:18:10,815 INFO [train.py:904] (6/8) Epoch 3, batch 800, loss[loss=0.2201, simple_loss=0.3001, pruned_loss=0.07007, over 16882.00 frames. ], tot_loss[loss=0.2443, simple_loss=0.3125, pruned_loss=0.08804, over 3250226.24 frames. ], batch size: 42, lr: 2.24e-02, grad_scale: 8.0 2023-04-27 21:18:27,458 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21113.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:18:34,907 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21118.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:18:40,096 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=21122.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:18:47,522 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.745e+02 4.034e+02 4.669e+02 5.861e+02 1.175e+03, threshold=9.338e+02, percent-clipped=5.0 2023-04-27 21:19:20,165 INFO [train.py:904] (6/8) Epoch 3, batch 850, loss[loss=0.2428, simple_loss=0.3082, pruned_loss=0.08867, over 16707.00 frames. ], tot_loss[loss=0.2426, simple_loss=0.3113, pruned_loss=0.08697, over 3260924.38 frames. ], batch size: 134, lr: 2.23e-02, grad_scale: 8.0 2023-04-27 21:19:58,152 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21179.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:20:09,821 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-04-27 21:20:27,602 INFO [train.py:904] (6/8) Epoch 3, batch 900, loss[loss=0.2203, simple_loss=0.2896, pruned_loss=0.07546, over 16986.00 frames. ], tot_loss[loss=0.242, simple_loss=0.3111, pruned_loss=0.08649, over 3279502.32 frames. ], batch size: 41, lr: 2.23e-02, grad_scale: 8.0 2023-04-27 21:21:03,013 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.660e+02 3.821e+02 4.761e+02 5.619e+02 1.173e+03, threshold=9.521e+02, percent-clipped=3.0 2023-04-27 21:21:26,708 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21244.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:21:35,848 INFO [train.py:904] (6/8) Epoch 3, batch 950, loss[loss=0.2166, simple_loss=0.2921, pruned_loss=0.07058, over 17014.00 frames. ], tot_loss[loss=0.2428, simple_loss=0.3114, pruned_loss=0.08709, over 3293015.28 frames. ], batch size: 41, lr: 2.23e-02, grad_scale: 8.0 2023-04-27 21:22:34,707 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=21292.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:22:45,620 INFO [train.py:904] (6/8) Epoch 3, batch 1000, loss[loss=0.2794, simple_loss=0.328, pruned_loss=0.1154, over 12289.00 frames. ], tot_loss[loss=0.2421, simple_loss=0.3103, pruned_loss=0.08698, over 3289292.67 frames. ], batch size: 246, lr: 2.23e-02, grad_scale: 8.0 2023-04-27 21:23:01,293 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21312.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:23:21,765 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.770e+02 3.937e+02 4.742e+02 6.159e+02 9.838e+02, threshold=9.485e+02, percent-clipped=2.0 2023-04-27 21:23:23,994 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21328.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:23:55,379 INFO [train.py:904] (6/8) Epoch 3, batch 1050, loss[loss=0.2073, simple_loss=0.2883, pruned_loss=0.06312, over 17213.00 frames. ], tot_loss[loss=0.2414, simple_loss=0.3098, pruned_loss=0.08647, over 3297075.05 frames. ], batch size: 45, lr: 2.22e-02, grad_scale: 8.0 2023-04-27 21:24:25,359 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21373.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:24:46,835 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21389.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:25:02,819 INFO [train.py:904] (6/8) Epoch 3, batch 1100, loss[loss=0.2167, simple_loss=0.2948, pruned_loss=0.06926, over 17159.00 frames. ], tot_loss[loss=0.2399, simple_loss=0.3086, pruned_loss=0.08559, over 3297967.64 frames. ], batch size: 43, lr: 2.22e-02, grad_scale: 4.0 2023-04-27 21:25:12,307 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21408.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:25:31,121 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4408, 3.3803, 1.4723, 3.4445, 2.1378, 3.4234, 1.5561, 2.7121], device='cuda:6'), covar=tensor([0.0061, 0.0214, 0.1587, 0.0072, 0.0805, 0.0318, 0.1450, 0.0519], device='cuda:6'), in_proj_covar=tensor([0.0086, 0.0131, 0.0171, 0.0082, 0.0158, 0.0162, 0.0181, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-27 21:25:38,414 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.639e+02 4.233e+02 5.080e+02 6.558e+02 1.311e+03, threshold=1.016e+03, percent-clipped=3.0 2023-04-27 21:25:43,576 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3249, 4.5350, 4.8798, 4.9146, 4.9130, 4.5314, 4.2487, 4.5122], device='cuda:6'), covar=tensor([0.0398, 0.0372, 0.0517, 0.0555, 0.0569, 0.0409, 0.1166, 0.0351], device='cuda:6'), in_proj_covar=tensor([0.0191, 0.0177, 0.0199, 0.0195, 0.0229, 0.0202, 0.0294, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-27 21:26:08,909 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2023-04-27 21:26:10,402 INFO [train.py:904] (6/8) Epoch 3, batch 1150, loss[loss=0.2163, simple_loss=0.3065, pruned_loss=0.06304, over 17137.00 frames. ], tot_loss[loss=0.2399, simple_loss=0.3082, pruned_loss=0.08584, over 3298117.68 frames. ], batch size: 48, lr: 2.22e-02, grad_scale: 4.0 2023-04-27 21:26:26,197 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-27 21:26:42,691 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21474.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:27:19,790 INFO [train.py:904] (6/8) Epoch 3, batch 1200, loss[loss=0.2539, simple_loss=0.3292, pruned_loss=0.08929, over 16610.00 frames. ], tot_loss[loss=0.2387, simple_loss=0.3072, pruned_loss=0.08513, over 3296624.50 frames. ], batch size: 57, lr: 2.22e-02, grad_scale: 8.0 2023-04-27 21:27:56,780 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.205e+02 3.899e+02 4.622e+02 5.754e+02 9.760e+02, threshold=9.243e+02, percent-clipped=0.0 2023-04-27 21:28:27,750 INFO [train.py:904] (6/8) Epoch 3, batch 1250, loss[loss=0.2595, simple_loss=0.309, pruned_loss=0.105, over 16502.00 frames. ], tot_loss[loss=0.2402, simple_loss=0.3078, pruned_loss=0.08628, over 3306845.83 frames. ], batch size: 68, lr: 2.21e-02, grad_scale: 8.0 2023-04-27 21:29:09,327 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1855, 2.8519, 2.6586, 4.4125, 1.9507, 4.2619, 2.3526, 2.6132], device='cuda:6'), covar=tensor([0.0272, 0.0653, 0.0423, 0.0204, 0.1773, 0.0198, 0.0889, 0.1493], device='cuda:6'), in_proj_covar=tensor([0.0242, 0.0217, 0.0180, 0.0244, 0.0284, 0.0188, 0.0208, 0.0276], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 21:29:30,675 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21595.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:29:39,775 INFO [train.py:904] (6/8) Epoch 3, batch 1300, loss[loss=0.2468, simple_loss=0.3032, pruned_loss=0.0952, over 15581.00 frames. ], tot_loss[loss=0.2407, simple_loss=0.3078, pruned_loss=0.08682, over 3316303.95 frames. ], batch size: 190, lr: 2.21e-02, grad_scale: 8.0 2023-04-27 21:29:56,107 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21613.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:30:17,377 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.341e+02 3.774e+02 4.586e+02 5.366e+02 8.286e+02, threshold=9.173e+02, percent-clipped=0.0 2023-04-27 21:30:49,956 INFO [train.py:904] (6/8) Epoch 3, batch 1350, loss[loss=0.2215, simple_loss=0.295, pruned_loss=0.07397, over 16729.00 frames. ], tot_loss[loss=0.2405, simple_loss=0.3081, pruned_loss=0.08647, over 3319353.11 frames. ], batch size: 89, lr: 2.21e-02, grad_scale: 8.0 2023-04-27 21:30:56,239 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21656.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:31:13,018 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21668.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:31:21,003 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21674.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:31:35,888 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21684.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:31:58,965 INFO [train.py:904] (6/8) Epoch 3, batch 1400, loss[loss=0.2071, simple_loss=0.2797, pruned_loss=0.06727, over 17212.00 frames. ], tot_loss[loss=0.2406, simple_loss=0.308, pruned_loss=0.08665, over 3310436.05 frames. ], batch size: 44, lr: 2.21e-02, grad_scale: 8.0 2023-04-27 21:32:09,230 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21708.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:32:09,351 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7610, 4.4728, 4.4082, 1.7464, 4.6260, 4.5698, 3.2369, 3.4244], device='cuda:6'), covar=tensor([0.0666, 0.0062, 0.0096, 0.1290, 0.0047, 0.0031, 0.0260, 0.0298], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0079, 0.0079, 0.0145, 0.0076, 0.0071, 0.0111, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 21:32:35,074 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.932e+02 4.012e+02 4.714e+02 6.047e+02 1.220e+03, threshold=9.428e+02, percent-clipped=2.0 2023-04-27 21:33:07,161 INFO [train.py:904] (6/8) Epoch 3, batch 1450, loss[loss=0.2453, simple_loss=0.296, pruned_loss=0.09726, over 16724.00 frames. ], tot_loss[loss=0.2393, simple_loss=0.3066, pruned_loss=0.08602, over 3313302.09 frames. ], batch size: 89, lr: 2.21e-02, grad_scale: 8.0 2023-04-27 21:33:13,876 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=21756.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:33:23,980 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21763.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:33:39,406 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21774.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:34:14,647 INFO [train.py:904] (6/8) Epoch 3, batch 1500, loss[loss=0.2172, simple_loss=0.2906, pruned_loss=0.07188, over 16783.00 frames. ], tot_loss[loss=0.2392, simple_loss=0.3069, pruned_loss=0.08571, over 3321346.03 frames. ], batch size: 39, lr: 2.20e-02, grad_scale: 8.0 2023-04-27 21:34:38,326 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7825, 3.2328, 2.4742, 4.3373, 4.1862, 4.0280, 1.6027, 3.0663], device='cuda:6'), covar=tensor([0.1247, 0.0377, 0.1038, 0.0054, 0.0189, 0.0260, 0.1163, 0.0548], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0127, 0.0162, 0.0073, 0.0141, 0.0134, 0.0152, 0.0153], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 21:34:43,998 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=21822.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:34:47,217 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21824.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:34:52,715 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.241e+02 3.626e+02 4.234e+02 5.507e+02 9.675e+02, threshold=8.468e+02, percent-clipped=1.0 2023-04-27 21:35:23,714 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21850.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:35:24,400 INFO [train.py:904] (6/8) Epoch 3, batch 1550, loss[loss=0.3179, simple_loss=0.3587, pruned_loss=0.1386, over 15502.00 frames. ], tot_loss[loss=0.2421, simple_loss=0.3088, pruned_loss=0.08769, over 3312466.01 frames. ], batch size: 190, lr: 2.20e-02, grad_scale: 8.0 2023-04-27 21:36:31,484 INFO [train.py:904] (6/8) Epoch 3, batch 1600, loss[loss=0.271, simple_loss=0.3323, pruned_loss=0.1049, over 16393.00 frames. ], tot_loss[loss=0.2451, simple_loss=0.3114, pruned_loss=0.08943, over 3305538.54 frames. ], batch size: 146, lr: 2.20e-02, grad_scale: 8.0 2023-04-27 21:36:45,081 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.49 vs. limit=5.0 2023-04-27 21:36:45,215 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.20 vs. limit=5.0 2023-04-27 21:36:46,742 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21911.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:37:07,920 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.625e+02 3.946e+02 4.752e+02 6.203e+02 1.123e+03, threshold=9.504e+02, percent-clipped=2.0 2023-04-27 21:37:38,900 INFO [train.py:904] (6/8) Epoch 3, batch 1650, loss[loss=0.2477, simple_loss=0.3276, pruned_loss=0.08392, over 16745.00 frames. ], tot_loss[loss=0.2459, simple_loss=0.3124, pruned_loss=0.08966, over 3313975.59 frames. ], batch size: 62, lr: 2.20e-02, grad_scale: 8.0 2023-04-27 21:37:39,884 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21951.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:38:02,140 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21968.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:38:03,295 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21969.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:38:04,709 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21970.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:38:24,372 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21984.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:38:34,344 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2367, 3.5018, 3.5366, 1.6876, 3.6838, 3.6259, 3.0453, 2.8469], device='cuda:6'), covar=tensor([0.0741, 0.0116, 0.0137, 0.1249, 0.0060, 0.0060, 0.0297, 0.0358], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0083, 0.0081, 0.0149, 0.0075, 0.0075, 0.0113, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 21:38:50,891 INFO [train.py:904] (6/8) Epoch 3, batch 1700, loss[loss=0.2297, simple_loss=0.3097, pruned_loss=0.07481, over 17167.00 frames. ], tot_loss[loss=0.2474, simple_loss=0.3146, pruned_loss=0.09014, over 3305429.29 frames. ], batch size: 46, lr: 2.19e-02, grad_scale: 4.0 2023-04-27 21:38:59,110 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5419, 3.9629, 3.9711, 1.8974, 4.0710, 3.9525, 3.3513, 3.1330], device='cuda:6'), covar=tensor([0.0693, 0.0086, 0.0098, 0.1119, 0.0049, 0.0048, 0.0230, 0.0328], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0083, 0.0081, 0.0148, 0.0075, 0.0075, 0.0112, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 21:39:01,207 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5385, 4.4624, 4.4978, 4.6025, 4.4985, 5.0854, 4.7588, 4.4305], device='cuda:6'), covar=tensor([0.0988, 0.1407, 0.1135, 0.1394, 0.2368, 0.0839, 0.0989, 0.2090], device='cuda:6'), in_proj_covar=tensor([0.0222, 0.0319, 0.0296, 0.0274, 0.0363, 0.0315, 0.0249, 0.0371], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 21:39:05,872 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-27 21:39:11,658 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22016.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:39:28,009 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.555e+02 4.242e+02 5.121e+02 6.013e+02 1.262e+03, threshold=1.024e+03, percent-clipped=2.0 2023-04-27 21:39:28,486 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0729, 4.8275, 5.1000, 5.4647, 5.5167, 4.6223, 5.5404, 5.4557], device='cuda:6'), covar=tensor([0.0553, 0.0591, 0.0997, 0.0361, 0.0282, 0.0459, 0.0191, 0.0244], device='cuda:6'), in_proj_covar=tensor([0.0298, 0.0353, 0.0483, 0.0364, 0.0271, 0.0251, 0.0278, 0.0289], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 21:39:31,645 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22031.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:39:33,270 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22032.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:39:57,284 INFO [train.py:904] (6/8) Epoch 3, batch 1750, loss[loss=0.1984, simple_loss=0.2734, pruned_loss=0.06166, over 16758.00 frames. ], tot_loss[loss=0.2479, simple_loss=0.3152, pruned_loss=0.09032, over 3304313.48 frames. ], batch size: 39, lr: 2.19e-02, grad_scale: 4.0 2023-04-27 21:40:26,550 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-27 21:41:06,922 INFO [train.py:904] (6/8) Epoch 3, batch 1800, loss[loss=0.2275, simple_loss=0.3064, pruned_loss=0.07429, over 17082.00 frames. ], tot_loss[loss=0.2463, simple_loss=0.3145, pruned_loss=0.08908, over 3308565.93 frames. ], batch size: 47, lr: 2.19e-02, grad_scale: 4.0 2023-04-27 21:41:30,480 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22119.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:41:44,204 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.299e+02 3.837e+02 4.558e+02 5.944e+02 1.243e+03, threshold=9.115e+02, percent-clipped=4.0 2023-04-27 21:41:45,864 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22130.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:41:50,477 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4271, 3.4904, 3.1503, 3.3125, 3.0405, 3.2155, 3.1581, 3.1308], device='cuda:6'), covar=tensor([0.0368, 0.0172, 0.0254, 0.0191, 0.0524, 0.0190, 0.1127, 0.0308], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0129, 0.0191, 0.0159, 0.0229, 0.0172, 0.0140, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 21:42:05,648 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0169, 2.0673, 1.6554, 1.9237, 2.7435, 2.6095, 2.8787, 2.8827], device='cuda:6'), covar=tensor([0.0037, 0.0122, 0.0137, 0.0136, 0.0061, 0.0097, 0.0061, 0.0063], device='cuda:6'), in_proj_covar=tensor([0.0060, 0.0117, 0.0114, 0.0118, 0.0110, 0.0118, 0.0074, 0.0093], device='cuda:6'), out_proj_covar=tensor([8.6502e-05, 1.7351e-04, 1.6266e-04, 1.7365e-04, 1.6737e-04, 1.7761e-04, 1.0986e-04, 1.4390e-04], device='cuda:6') 2023-04-27 21:42:14,095 INFO [train.py:904] (6/8) Epoch 3, batch 1850, loss[loss=0.2562, simple_loss=0.3331, pruned_loss=0.08962, over 17042.00 frames. ], tot_loss[loss=0.248, simple_loss=0.3166, pruned_loss=0.08971, over 3296108.84 frames. ], batch size: 55, lr: 2.19e-02, grad_scale: 4.0 2023-04-27 21:42:37,930 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4056, 2.2887, 1.7973, 2.1735, 2.8694, 2.7478, 3.9422, 3.3285], device='cuda:6'), covar=tensor([0.0023, 0.0142, 0.0190, 0.0155, 0.0081, 0.0124, 0.0031, 0.0066], device='cuda:6'), in_proj_covar=tensor([0.0061, 0.0116, 0.0114, 0.0117, 0.0110, 0.0118, 0.0073, 0.0093], device='cuda:6'), out_proj_covar=tensor([8.7503e-05, 1.7224e-04, 1.6223e-04, 1.7317e-04, 1.6725e-04, 1.7767e-04, 1.0903e-04, 1.4369e-04], device='cuda:6') 2023-04-27 21:42:47,477 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22176.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:43:09,338 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22191.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:43:18,354 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0996, 3.0338, 2.7294, 2.0394, 2.5205, 2.0955, 2.7553, 3.1115], device='cuda:6'), covar=tensor([0.0236, 0.0412, 0.0442, 0.1236, 0.0628, 0.0833, 0.0559, 0.0385], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0122, 0.0155, 0.0147, 0.0137, 0.0132, 0.0146, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-27 21:43:22,345 INFO [train.py:904] (6/8) Epoch 3, batch 1900, loss[loss=0.2361, simple_loss=0.2957, pruned_loss=0.08831, over 16442.00 frames. ], tot_loss[loss=0.2466, simple_loss=0.316, pruned_loss=0.08858, over 3295607.83 frames. ], batch size: 75, lr: 2.18e-02, grad_scale: 4.0 2023-04-27 21:43:31,115 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22206.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:43:42,736 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22215.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:44:02,290 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.402e+02 3.901e+02 4.998e+02 5.919e+02 1.840e+03, threshold=9.997e+02, percent-clipped=6.0 2023-04-27 21:44:12,186 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22237.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:44:12,284 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22237.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:44:32,423 INFO [train.py:904] (6/8) Epoch 3, batch 1950, loss[loss=0.2752, simple_loss=0.3369, pruned_loss=0.1067, over 16798.00 frames. ], tot_loss[loss=0.2462, simple_loss=0.3162, pruned_loss=0.08809, over 3297340.37 frames. ], batch size: 124, lr: 2.18e-02, grad_scale: 4.0 2023-04-27 21:44:32,726 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22251.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:44:58,060 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22269.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:45:06,131 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22276.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:45:35,327 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22298.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:45:36,281 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22299.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:45:38,311 INFO [train.py:904] (6/8) Epoch 3, batch 2000, loss[loss=0.2141, simple_loss=0.2832, pruned_loss=0.07248, over 16663.00 frames. ], tot_loss[loss=0.2457, simple_loss=0.3161, pruned_loss=0.08769, over 3304724.19 frames. ], batch size: 89, lr: 2.18e-02, grad_scale: 8.0 2023-04-27 21:46:01,054 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22317.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:46:13,085 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22326.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:46:17,541 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.887e+02 3.938e+02 4.847e+02 6.098e+02 1.059e+03, threshold=9.693e+02, percent-clipped=1.0 2023-04-27 21:46:20,942 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8331, 2.6648, 2.4787, 4.2679, 1.9630, 4.0433, 2.3166, 2.4649], device='cuda:6'), covar=tensor([0.0288, 0.0616, 0.0394, 0.0181, 0.1662, 0.0194, 0.0770, 0.1202], device='cuda:6'), in_proj_covar=tensor([0.0241, 0.0216, 0.0183, 0.0242, 0.0284, 0.0187, 0.0205, 0.0277], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 21:46:46,960 INFO [train.py:904] (6/8) Epoch 3, batch 2050, loss[loss=0.2216, simple_loss=0.2989, pruned_loss=0.07219, over 17190.00 frames. ], tot_loss[loss=0.2459, simple_loss=0.3161, pruned_loss=0.08781, over 3304799.25 frames. ], batch size: 46, lr: 2.18e-02, grad_scale: 8.0 2023-04-27 21:47:33,615 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22385.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:47:55,914 INFO [train.py:904] (6/8) Epoch 3, batch 2100, loss[loss=0.2837, simple_loss=0.3381, pruned_loss=0.1146, over 16701.00 frames. ], tot_loss[loss=0.2471, simple_loss=0.3171, pruned_loss=0.08854, over 3309177.00 frames. ], batch size: 134, lr: 2.17e-02, grad_scale: 8.0 2023-04-27 21:48:20,755 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22419.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:48:31,785 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.621e+02 3.997e+02 4.789e+02 6.119e+02 1.821e+03, threshold=9.579e+02, percent-clipped=3.0 2023-04-27 21:48:55,082 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22446.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:49:00,597 INFO [train.py:904] (6/8) Epoch 3, batch 2150, loss[loss=0.2394, simple_loss=0.3179, pruned_loss=0.0804, over 17021.00 frames. ], tot_loss[loss=0.2487, simple_loss=0.3183, pruned_loss=0.08953, over 3318386.93 frames. ], batch size: 55, lr: 2.17e-02, grad_scale: 8.0 2023-04-27 21:49:07,008 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.79 vs. limit=5.0 2023-04-27 21:49:16,038 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2023-04-27 21:49:16,048 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-27 21:49:22,529 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22467.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:49:29,645 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1067, 4.9549, 4.8633, 4.3012, 4.9106, 1.9665, 4.6068, 5.0429], device='cuda:6'), covar=tensor([0.0069, 0.0062, 0.0068, 0.0322, 0.0058, 0.1342, 0.0078, 0.0100], device='cuda:6'), in_proj_covar=tensor([0.0078, 0.0064, 0.0101, 0.0117, 0.0074, 0.0118, 0.0090, 0.0102], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 21:49:47,666 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22486.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:50:08,035 INFO [train.py:904] (6/8) Epoch 3, batch 2200, loss[loss=0.2739, simple_loss=0.3257, pruned_loss=0.1111, over 16708.00 frames. ], tot_loss[loss=0.2473, simple_loss=0.3175, pruned_loss=0.08858, over 3329287.47 frames. ], batch size: 134, lr: 2.17e-02, grad_scale: 4.0 2023-04-27 21:50:14,159 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22506.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:50:46,034 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.515e+02 4.185e+02 4.921e+02 5.938e+02 1.054e+03, threshold=9.842e+02, percent-clipped=2.0 2023-04-27 21:50:48,907 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22532.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:51:15,029 INFO [train.py:904] (6/8) Epoch 3, batch 2250, loss[loss=0.3222, simple_loss=0.3749, pruned_loss=0.1348, over 11811.00 frames. ], tot_loss[loss=0.2488, simple_loss=0.319, pruned_loss=0.08936, over 3323583.83 frames. ], batch size: 246, lr: 2.17e-02, grad_scale: 4.0 2023-04-27 21:51:18,569 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22554.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:51:42,423 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22571.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:52:09,942 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22593.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:52:17,851 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2701, 2.1765, 1.9585, 2.0994, 2.7230, 2.6842, 3.0582, 2.8278], device='cuda:6'), covar=tensor([0.0035, 0.0117, 0.0128, 0.0133, 0.0063, 0.0104, 0.0039, 0.0070], device='cuda:6'), in_proj_covar=tensor([0.0062, 0.0119, 0.0116, 0.0118, 0.0110, 0.0117, 0.0076, 0.0095], device='cuda:6'), out_proj_covar=tensor([8.9519e-05, 1.7642e-04, 1.6536e-04, 1.7321e-04, 1.6713e-04, 1.7682e-04, 1.1265e-04, 1.4585e-04], device='cuda:6') 2023-04-27 21:52:20,009 INFO [train.py:904] (6/8) Epoch 3, batch 2300, loss[loss=0.2179, simple_loss=0.2962, pruned_loss=0.06983, over 17256.00 frames. ], tot_loss[loss=0.2467, simple_loss=0.3172, pruned_loss=0.08814, over 3330700.07 frames. ], batch size: 45, lr: 2.17e-02, grad_scale: 4.0 2023-04-27 21:52:57,361 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22626.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:52:58,540 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3187, 5.3505, 4.9990, 5.1645, 4.3348, 5.0946, 5.1089, 4.6979], device='cuda:6'), covar=tensor([0.0332, 0.0128, 0.0226, 0.0167, 0.0978, 0.0211, 0.0184, 0.0276], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0129, 0.0192, 0.0158, 0.0222, 0.0169, 0.0138, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 21:53:01,631 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.100e+02 3.931e+02 4.776e+02 5.724e+02 1.077e+03, threshold=9.553e+02, percent-clipped=1.0 2023-04-27 21:53:12,105 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22639.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:53:30,352 INFO [train.py:904] (6/8) Epoch 3, batch 2350, loss[loss=0.2127, simple_loss=0.2853, pruned_loss=0.0701, over 16809.00 frames. ], tot_loss[loss=0.2477, simple_loss=0.3177, pruned_loss=0.08884, over 3325791.97 frames. ], batch size: 39, lr: 2.16e-02, grad_scale: 4.0 2023-04-27 21:53:38,217 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7311, 3.3245, 2.5314, 4.4528, 4.2492, 4.2025, 1.6615, 2.9429], device='cuda:6'), covar=tensor([0.1364, 0.0404, 0.1115, 0.0051, 0.0211, 0.0239, 0.1203, 0.0638], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0132, 0.0164, 0.0076, 0.0144, 0.0137, 0.0154, 0.0153], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-27 21:53:47,100 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-27 21:54:01,266 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22674.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:54:35,105 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22700.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 21:54:35,724 INFO [train.py:904] (6/8) Epoch 3, batch 2400, loss[loss=0.2161, simple_loss=0.2892, pruned_loss=0.07149, over 16872.00 frames. ], tot_loss[loss=0.2485, simple_loss=0.3189, pruned_loss=0.08901, over 3328516.44 frames. ], batch size: 42, lr: 2.16e-02, grad_scale: 8.0 2023-04-27 21:54:51,060 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-27 21:55:17,391 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.811e+02 3.861e+02 4.414e+02 5.506e+02 1.152e+03, threshold=8.829e+02, percent-clipped=5.0 2023-04-27 21:55:32,304 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22741.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:55:46,915 INFO [train.py:904] (6/8) Epoch 3, batch 2450, loss[loss=0.2136, simple_loss=0.2912, pruned_loss=0.06801, over 16863.00 frames. ], tot_loss[loss=0.2493, simple_loss=0.3201, pruned_loss=0.08923, over 3328924.42 frames. ], batch size: 42, lr: 2.16e-02, grad_scale: 8.0 2023-04-27 21:56:33,456 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22786.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:56:46,185 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22794.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:56:55,075 INFO [train.py:904] (6/8) Epoch 3, batch 2500, loss[loss=0.2348, simple_loss=0.3149, pruned_loss=0.07736, over 17056.00 frames. ], tot_loss[loss=0.249, simple_loss=0.3201, pruned_loss=0.08898, over 3328171.32 frames. ], batch size: 53, lr: 2.16e-02, grad_scale: 8.0 2023-04-27 21:57:27,676 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22825.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:57:32,108 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-27 21:57:33,675 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.850e+02 4.203e+02 4.842e+02 6.402e+02 1.699e+03, threshold=9.683e+02, percent-clipped=7.0 2023-04-27 21:57:37,718 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22832.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:57:41,132 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22834.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:58:03,439 INFO [train.py:904] (6/8) Epoch 3, batch 2550, loss[loss=0.2548, simple_loss=0.3285, pruned_loss=0.09056, over 16833.00 frames. ], tot_loss[loss=0.2501, simple_loss=0.3208, pruned_loss=0.08971, over 3325477.55 frames. ], batch size: 42, lr: 2.15e-02, grad_scale: 8.0 2023-04-27 21:58:10,966 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22855.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:58:17,074 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7712, 2.6663, 2.5250, 4.2572, 2.1575, 3.7331, 2.3886, 2.3976], device='cuda:6'), covar=tensor([0.0313, 0.0650, 0.0410, 0.0208, 0.1513, 0.0246, 0.0856, 0.1233], device='cuda:6'), in_proj_covar=tensor([0.0243, 0.0218, 0.0184, 0.0246, 0.0286, 0.0191, 0.0207, 0.0279], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 21:58:31,403 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22871.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:58:44,551 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22880.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:58:52,575 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22886.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:59:01,457 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22893.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:59:13,835 INFO [train.py:904] (6/8) Epoch 3, batch 2600, loss[loss=0.2446, simple_loss=0.3091, pruned_loss=0.09001, over 16729.00 frames. ], tot_loss[loss=0.2489, simple_loss=0.3202, pruned_loss=0.08887, over 3328980.71 frames. ], batch size: 89, lr: 2.15e-02, grad_scale: 8.0 2023-04-27 21:59:39,353 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22919.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:59:53,532 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.483e+02 3.680e+02 4.418e+02 5.171e+02 1.031e+03, threshold=8.837e+02, percent-clipped=1.0 2023-04-27 22:00:09,151 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22941.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:00:22,556 INFO [train.py:904] (6/8) Epoch 3, batch 2650, loss[loss=0.2527, simple_loss=0.3347, pruned_loss=0.08538, over 17251.00 frames. ], tot_loss[loss=0.2489, simple_loss=0.32, pruned_loss=0.08889, over 3327170.38 frames. ], batch size: 44, lr: 2.15e-02, grad_scale: 8.0 2023-04-27 22:00:48,816 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-27 22:01:22,673 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22995.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 22:01:30,281 INFO [train.py:904] (6/8) Epoch 3, batch 2700, loss[loss=0.248, simple_loss=0.3183, pruned_loss=0.0888, over 17176.00 frames. ], tot_loss[loss=0.248, simple_loss=0.3205, pruned_loss=0.08772, over 3334429.54 frames. ], batch size: 46, lr: 2.15e-02, grad_scale: 8.0 2023-04-27 22:02:09,772 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.756e+02 4.226e+02 4.976e+02 6.032e+02 3.495e+03, threshold=9.952e+02, percent-clipped=7.0 2023-04-27 22:02:24,001 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23041.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:02:34,871 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-27 22:02:37,521 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.55 vs. limit=5.0 2023-04-27 22:02:37,954 INFO [train.py:904] (6/8) Epoch 3, batch 2750, loss[loss=0.2528, simple_loss=0.3279, pruned_loss=0.08888, over 16639.00 frames. ], tot_loss[loss=0.2476, simple_loss=0.3205, pruned_loss=0.08731, over 3330804.06 frames. ], batch size: 57, lr: 2.15e-02, grad_scale: 8.0 2023-04-27 22:03:28,908 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23089.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:03:45,507 INFO [train.py:904] (6/8) Epoch 3, batch 2800, loss[loss=0.2357, simple_loss=0.3124, pruned_loss=0.07956, over 15973.00 frames. ], tot_loss[loss=0.2471, simple_loss=0.32, pruned_loss=0.08714, over 3333489.66 frames. ], batch size: 35, lr: 2.14e-02, grad_scale: 8.0 2023-04-27 22:03:55,514 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23108.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:04:25,544 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.361e+02 3.403e+02 4.548e+02 5.612e+02 1.011e+03, threshold=9.095e+02, percent-clipped=1.0 2023-04-27 22:04:33,958 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6853, 4.6446, 4.2911, 2.0976, 3.3854, 2.7322, 3.9514, 4.3980], device='cuda:6'), covar=tensor([0.0239, 0.0408, 0.0379, 0.1491, 0.0586, 0.0841, 0.0753, 0.0762], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0121, 0.0155, 0.0144, 0.0135, 0.0130, 0.0145, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-27 22:04:54,740 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23150.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:04:55,605 INFO [train.py:904] (6/8) Epoch 3, batch 2850, loss[loss=0.263, simple_loss=0.323, pruned_loss=0.1015, over 16407.00 frames. ], tot_loss[loss=0.2479, simple_loss=0.3193, pruned_loss=0.08828, over 3331361.58 frames. ], batch size: 146, lr: 2.14e-02, grad_scale: 8.0 2023-04-27 22:05:20,362 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23169.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:05:36,715 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23181.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:06:03,310 INFO [train.py:904] (6/8) Epoch 3, batch 2900, loss[loss=0.269, simple_loss=0.3251, pruned_loss=0.1065, over 16417.00 frames. ], tot_loss[loss=0.2473, simple_loss=0.3181, pruned_loss=0.0883, over 3332867.55 frames. ], batch size: 146, lr: 2.14e-02, grad_scale: 8.0 2023-04-27 22:06:25,789 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23217.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:06:40,293 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.42 vs. limit=5.0 2023-04-27 22:06:42,395 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23229.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:06:43,091 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.672e+02 4.009e+02 4.959e+02 5.927e+02 1.397e+03, threshold=9.918e+02, percent-clipped=5.0 2023-04-27 22:07:00,816 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1778, 4.5305, 3.3743, 2.8249, 3.3962, 2.3139, 4.6184, 5.0812], device='cuda:6'), covar=tensor([0.1685, 0.0506, 0.1094, 0.0968, 0.1924, 0.1334, 0.0291, 0.0236], device='cuda:6'), in_proj_covar=tensor([0.0259, 0.0241, 0.0249, 0.0201, 0.0280, 0.0188, 0.0213, 0.0206], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 22:07:05,684 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.39 vs. limit=5.0 2023-04-27 22:07:12,202 INFO [train.py:904] (6/8) Epoch 3, batch 2950, loss[loss=0.2024, simple_loss=0.2762, pruned_loss=0.06428, over 16745.00 frames. ], tot_loss[loss=0.2473, simple_loss=0.3173, pruned_loss=0.08865, over 3331557.67 frames. ], batch size: 39, lr: 2.14e-02, grad_scale: 8.0 2023-04-27 22:07:26,442 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-27 22:07:49,188 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23278.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:08:07,148 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23290.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:08:12,633 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23295.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:08:19,857 INFO [train.py:904] (6/8) Epoch 3, batch 3000, loss[loss=0.2196, simple_loss=0.2874, pruned_loss=0.07596, over 16965.00 frames. ], tot_loss[loss=0.2479, simple_loss=0.3175, pruned_loss=0.0892, over 3321066.18 frames. ], batch size: 41, lr: 2.13e-02, grad_scale: 8.0 2023-04-27 22:08:19,857 INFO [train.py:929] (6/8) Computing validation loss 2023-04-27 22:08:30,494 INFO [train.py:938] (6/8) Epoch 3, validation: loss=0.1721, simple_loss=0.279, pruned_loss=0.03262, over 944034.00 frames. 2023-04-27 22:08:30,495 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-27 22:09:10,133 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.558e+02 3.966e+02 4.777e+02 5.754e+02 1.756e+03, threshold=9.554e+02, percent-clipped=1.0 2023-04-27 22:09:26,140 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23343.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:09:30,080 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-27 22:09:37,531 INFO [train.py:904] (6/8) Epoch 3, batch 3050, loss[loss=0.2937, simple_loss=0.3497, pruned_loss=0.1188, over 12358.00 frames. ], tot_loss[loss=0.248, simple_loss=0.3176, pruned_loss=0.08917, over 3315332.45 frames. ], batch size: 246, lr: 2.13e-02, grad_scale: 4.0 2023-04-27 22:10:00,062 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4335, 4.3644, 4.4253, 3.6236, 4.3238, 1.7544, 4.1122, 4.3933], device='cuda:6'), covar=tensor([0.0083, 0.0069, 0.0074, 0.0363, 0.0062, 0.1407, 0.0097, 0.0112], device='cuda:6'), in_proj_covar=tensor([0.0077, 0.0066, 0.0102, 0.0118, 0.0074, 0.0114, 0.0093, 0.0101], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 22:10:44,544 INFO [train.py:904] (6/8) Epoch 3, batch 3100, loss[loss=0.2341, simple_loss=0.3116, pruned_loss=0.07828, over 16565.00 frames. ], tot_loss[loss=0.2482, simple_loss=0.3174, pruned_loss=0.08951, over 3314280.03 frames. ], batch size: 62, lr: 2.13e-02, grad_scale: 4.0 2023-04-27 22:11:28,255 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.500e+02 3.880e+02 4.394e+02 5.487e+02 1.419e+03, threshold=8.788e+02, percent-clipped=5.0 2023-04-27 22:11:50,472 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7287, 4.5716, 1.8398, 4.8090, 2.9652, 4.7782, 2.4137, 3.3872], device='cuda:6'), covar=tensor([0.0035, 0.0174, 0.1636, 0.0024, 0.0795, 0.0238, 0.1326, 0.0523], device='cuda:6'), in_proj_covar=tensor([0.0089, 0.0134, 0.0168, 0.0084, 0.0158, 0.0170, 0.0179, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-27 22:11:53,305 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23450.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:11:54,095 INFO [train.py:904] (6/8) Epoch 3, batch 3150, loss[loss=0.2044, simple_loss=0.2901, pruned_loss=0.0593, over 17245.00 frames. ], tot_loss[loss=0.2458, simple_loss=0.3157, pruned_loss=0.08793, over 3322188.86 frames. ], batch size: 45, lr: 2.13e-02, grad_scale: 4.0 2023-04-27 22:12:12,236 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23464.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:12:35,051 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23481.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:12:44,153 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23488.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:12:58,215 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23498.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:13:01,351 INFO [train.py:904] (6/8) Epoch 3, batch 3200, loss[loss=0.1986, simple_loss=0.2814, pruned_loss=0.05793, over 17167.00 frames. ], tot_loss[loss=0.2431, simple_loss=0.3139, pruned_loss=0.0861, over 3331707.84 frames. ], batch size: 46, lr: 2.13e-02, grad_scale: 8.0 2023-04-27 22:13:39,247 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23529.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:13:42,225 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.397e+02 3.601e+02 4.370e+02 5.311e+02 9.274e+02, threshold=8.739e+02, percent-clipped=1.0 2023-04-27 22:13:53,347 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4894, 4.6236, 3.8703, 1.6978, 3.0308, 2.4770, 3.7179, 4.2914], device='cuda:6'), covar=tensor([0.0263, 0.0365, 0.0430, 0.1698, 0.0689, 0.0974, 0.0692, 0.0550], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0124, 0.0153, 0.0144, 0.0135, 0.0130, 0.0148, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-27 22:14:06,499 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23549.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 22:14:08,943 INFO [train.py:904] (6/8) Epoch 3, batch 3250, loss[loss=0.2716, simple_loss=0.3317, pruned_loss=0.1058, over 16394.00 frames. ], tot_loss[loss=0.2419, simple_loss=0.3127, pruned_loss=0.08557, over 3330563.74 frames. ], batch size: 146, lr: 2.12e-02, grad_scale: 8.0 2023-04-27 22:14:38,901 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23573.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:14:56,668 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23585.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:15:18,373 INFO [train.py:904] (6/8) Epoch 3, batch 3300, loss[loss=0.2186, simple_loss=0.2898, pruned_loss=0.07372, over 16776.00 frames. ], tot_loss[loss=0.2441, simple_loss=0.3147, pruned_loss=0.08673, over 3326477.81 frames. ], batch size: 83, lr: 2.12e-02, grad_scale: 8.0 2023-04-27 22:15:36,407 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7686, 2.7748, 1.6035, 2.8176, 2.1643, 2.7699, 1.9792, 2.4549], device='cuda:6'), covar=tensor([0.0087, 0.0215, 0.1158, 0.0069, 0.0568, 0.0418, 0.0863, 0.0454], device='cuda:6'), in_proj_covar=tensor([0.0090, 0.0134, 0.0171, 0.0084, 0.0156, 0.0169, 0.0176, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-27 22:15:44,413 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.59 vs. limit=5.0 2023-04-27 22:15:57,197 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.549e+02 3.983e+02 4.910e+02 5.801e+02 1.123e+03, threshold=9.819e+02, percent-clipped=5.0 2023-04-27 22:16:07,758 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6030, 2.4163, 1.9986, 2.0370, 3.0209, 2.6910, 4.0153, 3.3221], device='cuda:6'), covar=tensor([0.0020, 0.0144, 0.0193, 0.0186, 0.0080, 0.0132, 0.0029, 0.0083], device='cuda:6'), in_proj_covar=tensor([0.0065, 0.0122, 0.0119, 0.0120, 0.0112, 0.0122, 0.0083, 0.0099], device='cuda:6'), out_proj_covar=tensor([9.4903e-05, 1.7733e-04, 1.6765e-04, 1.7487e-04, 1.6806e-04, 1.7990e-04, 1.2338e-04, 1.5256e-04], device='cuda:6') 2023-04-27 22:16:08,151 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-27 22:16:24,695 INFO [train.py:904] (6/8) Epoch 3, batch 3350, loss[loss=0.2012, simple_loss=0.2872, pruned_loss=0.05765, over 17163.00 frames. ], tot_loss[loss=0.2444, simple_loss=0.3153, pruned_loss=0.08681, over 3318634.95 frames. ], batch size: 46, lr: 2.12e-02, grad_scale: 8.0 2023-04-27 22:17:13,940 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-27 22:17:33,465 INFO [train.py:904] (6/8) Epoch 3, batch 3400, loss[loss=0.2577, simple_loss=0.3097, pruned_loss=0.1028, over 16858.00 frames. ], tot_loss[loss=0.2448, simple_loss=0.3153, pruned_loss=0.08712, over 3309761.05 frames. ], batch size: 116, lr: 2.12e-02, grad_scale: 8.0 2023-04-27 22:18:13,370 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.692e+02 3.881e+02 4.662e+02 5.457e+02 1.013e+03, threshold=9.324e+02, percent-clipped=1.0 2023-04-27 22:18:32,217 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1445, 4.0586, 4.0836, 3.1130, 4.0469, 1.7680, 3.8226, 4.0288], device='cuda:6'), covar=tensor([0.0102, 0.0095, 0.0099, 0.0516, 0.0080, 0.1576, 0.0108, 0.0149], device='cuda:6'), in_proj_covar=tensor([0.0078, 0.0069, 0.0104, 0.0121, 0.0077, 0.0116, 0.0094, 0.0105], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 22:18:40,209 INFO [train.py:904] (6/8) Epoch 3, batch 3450, loss[loss=0.2316, simple_loss=0.3011, pruned_loss=0.08106, over 16821.00 frames. ], tot_loss[loss=0.2432, simple_loss=0.3139, pruned_loss=0.08629, over 3310370.81 frames. ], batch size: 42, lr: 2.11e-02, grad_scale: 8.0 2023-04-27 22:18:59,370 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23764.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:19:45,408 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-27 22:19:47,198 INFO [train.py:904] (6/8) Epoch 3, batch 3500, loss[loss=0.2599, simple_loss=0.3216, pruned_loss=0.09907, over 16285.00 frames. ], tot_loss[loss=0.2429, simple_loss=0.3129, pruned_loss=0.08642, over 3308422.85 frames. ], batch size: 165, lr: 2.11e-02, grad_scale: 8.0 2023-04-27 22:20:04,594 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23812.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:20:31,310 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.165e+02 3.876e+02 4.693e+02 5.719e+02 1.171e+03, threshold=9.385e+02, percent-clipped=5.0 2023-04-27 22:20:45,497 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8121, 2.7142, 2.6801, 4.3903, 2.0151, 3.9836, 2.4479, 2.5958], device='cuda:6'), covar=tensor([0.0322, 0.0658, 0.0399, 0.0143, 0.1659, 0.0193, 0.0862, 0.1092], device='cuda:6'), in_proj_covar=tensor([0.0251, 0.0225, 0.0189, 0.0253, 0.0297, 0.0201, 0.0218, 0.0288], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 22:20:48,797 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23844.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 22:20:59,081 INFO [train.py:904] (6/8) Epoch 3, batch 3550, loss[loss=0.2198, simple_loss=0.2913, pruned_loss=0.07415, over 16816.00 frames. ], tot_loss[loss=0.2406, simple_loss=0.3112, pruned_loss=0.08502, over 3316329.28 frames. ], batch size: 42, lr: 2.11e-02, grad_scale: 8.0 2023-04-27 22:21:29,304 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23873.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:21:30,552 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4738, 4.8147, 4.4095, 4.6091, 4.2534, 4.1125, 4.3851, 4.8049], device='cuda:6'), covar=tensor([0.0544, 0.0612, 0.0919, 0.0380, 0.0503, 0.0781, 0.0540, 0.0652], device='cuda:6'), in_proj_covar=tensor([0.0278, 0.0400, 0.0344, 0.0245, 0.0259, 0.0245, 0.0313, 0.0273], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 22:21:43,541 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5947, 1.9507, 2.9156, 3.3345, 3.3121, 3.5805, 2.1766, 3.4690], device='cuda:6'), covar=tensor([0.0035, 0.0191, 0.0093, 0.0092, 0.0049, 0.0056, 0.0148, 0.0056], device='cuda:6'), in_proj_covar=tensor([0.0089, 0.0125, 0.0113, 0.0106, 0.0087, 0.0070, 0.0114, 0.0067], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-27 22:21:45,216 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23885.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:21:56,686 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2023-04-27 22:22:05,753 INFO [train.py:904] (6/8) Epoch 3, batch 3600, loss[loss=0.2001, simple_loss=0.2755, pruned_loss=0.06232, over 16780.00 frames. ], tot_loss[loss=0.239, simple_loss=0.3096, pruned_loss=0.08414, over 3306409.65 frames. ], batch size: 39, lr: 2.11e-02, grad_scale: 8.0 2023-04-27 22:22:33,417 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23921.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:22:38,261 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1058, 4.0926, 2.0924, 4.0853, 2.7601, 4.2113, 2.3867, 3.1201], device='cuda:6'), covar=tensor([0.0058, 0.0154, 0.1302, 0.0053, 0.0592, 0.0235, 0.0967, 0.0477], device='cuda:6'), in_proj_covar=tensor([0.0091, 0.0141, 0.0171, 0.0084, 0.0161, 0.0174, 0.0180, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-27 22:22:47,017 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.441e+02 3.888e+02 4.838e+02 6.309e+02 1.139e+03, threshold=9.677e+02, percent-clipped=4.0 2023-04-27 22:22:50,341 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23933.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:22:50,859 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-27 22:23:14,441 INFO [train.py:904] (6/8) Epoch 3, batch 3650, loss[loss=0.2672, simple_loss=0.3265, pruned_loss=0.1039, over 11391.00 frames. ], tot_loss[loss=0.2374, simple_loss=0.3071, pruned_loss=0.08382, over 3304580.57 frames. ], batch size: 246, lr: 2.11e-02, grad_scale: 8.0 2023-04-27 22:23:36,422 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-27 22:24:29,889 INFO [train.py:904] (6/8) Epoch 3, batch 3700, loss[loss=0.2366, simple_loss=0.3073, pruned_loss=0.083, over 16163.00 frames. ], tot_loss[loss=0.2387, simple_loss=0.3059, pruned_loss=0.08579, over 3308932.22 frames. ], batch size: 165, lr: 2.10e-02, grad_scale: 8.0 2023-04-27 22:25:13,795 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.162e+02 3.909e+02 4.503e+02 5.530e+02 1.035e+03, threshold=9.006e+02, percent-clipped=1.0 2023-04-27 22:25:22,683 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24037.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:25:42,963 INFO [train.py:904] (6/8) Epoch 3, batch 3750, loss[loss=0.2376, simple_loss=0.2986, pruned_loss=0.08832, over 16306.00 frames. ], tot_loss[loss=0.2412, simple_loss=0.3073, pruned_loss=0.08756, over 3281431.62 frames. ], batch size: 165, lr: 2.10e-02, grad_scale: 8.0 2023-04-27 22:25:46,760 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0587, 3.2858, 3.5155, 3.4927, 3.4753, 3.2687, 3.2641, 3.3516], device='cuda:6'), covar=tensor([0.0318, 0.0378, 0.0392, 0.0446, 0.0392, 0.0345, 0.0678, 0.0318], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0181, 0.0194, 0.0198, 0.0239, 0.0199, 0.0296, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-27 22:26:46,928 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24098.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:26:50,660 INFO [train.py:904] (6/8) Epoch 3, batch 3800, loss[loss=0.2594, simple_loss=0.3287, pruned_loss=0.09501, over 12547.00 frames. ], tot_loss[loss=0.2442, simple_loss=0.3092, pruned_loss=0.08959, over 3268188.10 frames. ], batch size: 248, lr: 2.10e-02, grad_scale: 8.0 2023-04-27 22:27:34,019 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.430e+02 3.668e+02 4.510e+02 5.659e+02 1.360e+03, threshold=9.019e+02, percent-clipped=5.0 2023-04-27 22:27:44,051 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9501, 1.5846, 2.1410, 2.7845, 2.7845, 2.7052, 1.8317, 2.7483], device='cuda:6'), covar=tensor([0.0036, 0.0198, 0.0146, 0.0084, 0.0055, 0.0073, 0.0177, 0.0045], device='cuda:6'), in_proj_covar=tensor([0.0087, 0.0124, 0.0112, 0.0105, 0.0087, 0.0070, 0.0114, 0.0067], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-27 22:27:52,888 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=24144.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 22:28:01,899 INFO [train.py:904] (6/8) Epoch 3, batch 3850, loss[loss=0.2361, simple_loss=0.3005, pruned_loss=0.08583, over 16353.00 frames. ], tot_loss[loss=0.2434, simple_loss=0.3081, pruned_loss=0.08942, over 3266403.57 frames. ], batch size: 35, lr: 2.10e-02, grad_scale: 8.0 2023-04-27 22:28:11,598 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-27 22:29:01,357 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=24192.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:29:13,623 INFO [train.py:904] (6/8) Epoch 3, batch 3900, loss[loss=0.2194, simple_loss=0.2996, pruned_loss=0.06963, over 17189.00 frames. ], tot_loss[loss=0.2431, simple_loss=0.3072, pruned_loss=0.08948, over 3267107.51 frames. ], batch size: 46, lr: 2.10e-02, grad_scale: 8.0 2023-04-27 22:29:25,707 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-27 22:29:34,169 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.88 vs. limit=5.0 2023-04-27 22:29:56,950 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.402e+02 3.854e+02 4.579e+02 5.596e+02 1.788e+03, threshold=9.157e+02, percent-clipped=5.0 2023-04-27 22:30:25,196 INFO [train.py:904] (6/8) Epoch 3, batch 3950, loss[loss=0.2577, simple_loss=0.3048, pruned_loss=0.1053, over 16874.00 frames. ], tot_loss[loss=0.2425, simple_loss=0.3061, pruned_loss=0.08939, over 3269801.63 frames. ], batch size: 116, lr: 2.09e-02, grad_scale: 8.0 2023-04-27 22:31:34,850 INFO [train.py:904] (6/8) Epoch 3, batch 4000, loss[loss=0.243, simple_loss=0.3137, pruned_loss=0.08616, over 16531.00 frames. ], tot_loss[loss=0.2418, simple_loss=0.3056, pruned_loss=0.08898, over 3264505.88 frames. ], batch size: 35, lr: 2.09e-02, grad_scale: 8.0 2023-04-27 22:32:17,081 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.388e+02 3.319e+02 4.445e+02 5.556e+02 1.324e+03, threshold=8.889e+02, percent-clipped=2.0 2023-04-27 22:32:45,198 INFO [train.py:904] (6/8) Epoch 3, batch 4050, loss[loss=0.2303, simple_loss=0.3045, pruned_loss=0.0781, over 17219.00 frames. ], tot_loss[loss=0.2392, simple_loss=0.3046, pruned_loss=0.08693, over 3253221.44 frames. ], batch size: 44, lr: 2.09e-02, grad_scale: 8.0 2023-04-27 22:33:22,594 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0814, 4.4309, 3.5645, 3.0147, 3.4787, 2.7660, 4.6373, 5.0885], device='cuda:6'), covar=tensor([0.1963, 0.0561, 0.1092, 0.0951, 0.1778, 0.1045, 0.0334, 0.0163], device='cuda:6'), in_proj_covar=tensor([0.0271, 0.0245, 0.0261, 0.0214, 0.0305, 0.0197, 0.0226, 0.0219], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 22:33:46,326 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24393.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:33:53,145 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-27 22:33:58,258 INFO [train.py:904] (6/8) Epoch 3, batch 4100, loss[loss=0.2299, simple_loss=0.3168, pruned_loss=0.07154, over 16825.00 frames. ], tot_loss[loss=0.2371, simple_loss=0.3042, pruned_loss=0.08496, over 3244138.65 frames. ], batch size: 102, lr: 2.09e-02, grad_scale: 8.0 2023-04-27 22:34:42,994 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.218e+02 2.972e+02 3.823e+02 4.924e+02 9.607e+02, threshold=7.646e+02, percent-clipped=2.0 2023-04-27 22:35:00,655 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-27 22:35:13,089 INFO [train.py:904] (6/8) Epoch 3, batch 4150, loss[loss=0.2697, simple_loss=0.3467, pruned_loss=0.09635, over 16869.00 frames. ], tot_loss[loss=0.2468, simple_loss=0.3142, pruned_loss=0.08974, over 3221870.08 frames. ], batch size: 109, lr: 2.09e-02, grad_scale: 8.0 2023-04-27 22:36:27,622 INFO [train.py:904] (6/8) Epoch 3, batch 4200, loss[loss=0.2876, simple_loss=0.3754, pruned_loss=0.09986, over 16542.00 frames. ], tot_loss[loss=0.2539, simple_loss=0.3225, pruned_loss=0.09264, over 3190857.04 frames. ], batch size: 68, lr: 2.08e-02, grad_scale: 8.0 2023-04-27 22:37:10,963 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.707e+02 3.802e+02 4.415e+02 5.397e+02 1.092e+03, threshold=8.829e+02, percent-clipped=9.0 2023-04-27 22:37:35,688 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24547.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:37:40,125 INFO [train.py:904] (6/8) Epoch 3, batch 4250, loss[loss=0.2608, simple_loss=0.3322, pruned_loss=0.09472, over 16782.00 frames. ], tot_loss[loss=0.2549, simple_loss=0.3253, pruned_loss=0.09227, over 3193731.99 frames. ], batch size: 124, lr: 2.08e-02, grad_scale: 8.0 2023-04-27 22:38:07,360 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24569.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:38:53,671 INFO [train.py:904] (6/8) Epoch 3, batch 4300, loss[loss=0.2684, simple_loss=0.3416, pruned_loss=0.09753, over 16614.00 frames. ], tot_loss[loss=0.2543, simple_loss=0.3264, pruned_loss=0.09114, over 3197242.96 frames. ], batch size: 35, lr: 2.08e-02, grad_scale: 8.0 2023-04-27 22:39:05,052 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24608.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:39:15,368 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8984, 3.9134, 3.7127, 3.8018, 3.4147, 3.8564, 3.5886, 3.5716], device='cuda:6'), covar=tensor([0.0270, 0.0136, 0.0175, 0.0122, 0.0671, 0.0149, 0.0525, 0.0274], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0117, 0.0163, 0.0137, 0.0194, 0.0146, 0.0122, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 22:39:37,622 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24630.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:39:38,216 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.269e+02 3.576e+02 4.392e+02 5.357e+02 9.860e+02, threshold=8.785e+02, percent-clipped=3.0 2023-04-27 22:40:06,098 INFO [train.py:904] (6/8) Epoch 3, batch 4350, loss[loss=0.2777, simple_loss=0.3436, pruned_loss=0.1059, over 15316.00 frames. ], tot_loss[loss=0.2585, simple_loss=0.3309, pruned_loss=0.09307, over 3185869.04 frames. ], batch size: 190, lr: 2.08e-02, grad_scale: 8.0 2023-04-27 22:40:15,316 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4594, 1.3554, 1.7994, 2.3655, 2.3324, 2.7093, 1.3952, 2.5927], device='cuda:6'), covar=tensor([0.0046, 0.0200, 0.0113, 0.0093, 0.0060, 0.0037, 0.0168, 0.0037], device='cuda:6'), in_proj_covar=tensor([0.0085, 0.0121, 0.0107, 0.0100, 0.0087, 0.0065, 0.0113, 0.0063], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-27 22:40:20,146 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9290, 3.6355, 3.6941, 1.4476, 3.9403, 3.9002, 2.8685, 2.7690], device='cuda:6'), covar=tensor([0.0991, 0.0110, 0.0142, 0.1458, 0.0043, 0.0041, 0.0363, 0.0439], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0082, 0.0077, 0.0146, 0.0073, 0.0073, 0.0114, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 22:41:09,322 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3020, 2.1430, 1.7127, 1.9807, 2.8930, 2.5358, 3.5448, 3.2182], device='cuda:6'), covar=tensor([0.0014, 0.0138, 0.0198, 0.0169, 0.0083, 0.0124, 0.0021, 0.0051], device='cuda:6'), in_proj_covar=tensor([0.0054, 0.0116, 0.0119, 0.0116, 0.0109, 0.0118, 0.0074, 0.0094], device='cuda:6'), out_proj_covar=tensor([7.8364e-05, 1.6811e-04, 1.6826e-04, 1.6741e-04, 1.6180e-04, 1.7322e-04, 1.0737e-04, 1.4221e-04], device='cuda:6') 2023-04-27 22:41:11,331 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=24693.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:41:21,564 INFO [train.py:904] (6/8) Epoch 3, batch 4400, loss[loss=0.2557, simple_loss=0.3307, pruned_loss=0.09032, over 16730.00 frames. ], tot_loss[loss=0.2596, simple_loss=0.3325, pruned_loss=0.09339, over 3198662.05 frames. ], batch size: 124, lr: 2.08e-02, grad_scale: 8.0 2023-04-27 22:41:30,426 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.99 vs. limit=5.0 2023-04-27 22:41:57,515 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3664, 4.0588, 4.1467, 1.4307, 4.5495, 4.5329, 3.0049, 3.5616], device='cuda:6'), covar=tensor([0.0842, 0.0120, 0.0180, 0.1525, 0.0041, 0.0028, 0.0297, 0.0304], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0081, 0.0079, 0.0144, 0.0073, 0.0072, 0.0113, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 22:42:05,389 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.377e+02 3.698e+02 4.447e+02 5.386e+02 9.920e+02, threshold=8.895e+02, percent-clipped=4.0 2023-04-27 22:42:21,152 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=24741.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:42:34,827 INFO [train.py:904] (6/8) Epoch 3, batch 4450, loss[loss=0.2441, simple_loss=0.333, pruned_loss=0.07762, over 16783.00 frames. ], tot_loss[loss=0.2604, simple_loss=0.3348, pruned_loss=0.09295, over 3210169.31 frames. ], batch size: 83, lr: 2.07e-02, grad_scale: 8.0 2023-04-27 22:42:52,539 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2023-04-27 22:42:55,238 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24765.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:42:57,731 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2206, 4.6605, 4.4618, 2.9599, 4.1741, 4.2945, 4.3982, 2.1539], device='cuda:6'), covar=tensor([0.0385, 0.0008, 0.0018, 0.0229, 0.0019, 0.0033, 0.0012, 0.0309], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0050, 0.0056, 0.0111, 0.0052, 0.0061, 0.0057, 0.0103], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 22:43:47,147 INFO [train.py:904] (6/8) Epoch 3, batch 4500, loss[loss=0.2434, simple_loss=0.326, pruned_loss=0.08043, over 16848.00 frames. ], tot_loss[loss=0.2595, simple_loss=0.3342, pruned_loss=0.09245, over 3213474.72 frames. ], batch size: 96, lr: 2.07e-02, grad_scale: 8.0 2023-04-27 22:44:05,724 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4184, 2.0654, 1.5531, 2.0751, 2.7927, 2.5590, 3.4068, 3.1752], device='cuda:6'), covar=tensor([0.0011, 0.0153, 0.0195, 0.0148, 0.0069, 0.0114, 0.0024, 0.0057], device='cuda:6'), in_proj_covar=tensor([0.0054, 0.0118, 0.0121, 0.0118, 0.0111, 0.0120, 0.0076, 0.0095], device='cuda:6'), out_proj_covar=tensor([7.7340e-05, 1.7019e-04, 1.7091e-04, 1.7034e-04, 1.6385e-04, 1.7536e-04, 1.0959e-04, 1.4353e-04], device='cuda:6') 2023-04-27 22:44:22,396 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2242, 4.3910, 4.2219, 2.6362, 3.9511, 4.0489, 4.1503, 2.1453], device='cuda:6'), covar=tensor([0.0359, 0.0010, 0.0020, 0.0267, 0.0025, 0.0040, 0.0014, 0.0305], device='cuda:6'), in_proj_covar=tensor([0.0111, 0.0049, 0.0056, 0.0109, 0.0051, 0.0059, 0.0056, 0.0102], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 22:44:22,438 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24826.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:44:29,087 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.999e+02 3.148e+02 3.848e+02 4.477e+02 7.288e+02, threshold=7.696e+02, percent-clipped=0.0 2023-04-27 22:44:52,872 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24848.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:44:56,540 INFO [train.py:904] (6/8) Epoch 3, batch 4550, loss[loss=0.268, simple_loss=0.3434, pruned_loss=0.09629, over 15440.00 frames. ], tot_loss[loss=0.2587, simple_loss=0.3335, pruned_loss=0.09196, over 3233279.98 frames. ], batch size: 191, lr: 2.07e-02, grad_scale: 8.0 2023-04-27 22:45:45,548 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-27 22:46:07,316 INFO [train.py:904] (6/8) Epoch 3, batch 4600, loss[loss=0.3517, simple_loss=0.3904, pruned_loss=0.1565, over 11702.00 frames. ], tot_loss[loss=0.2599, simple_loss=0.3347, pruned_loss=0.09256, over 3227079.88 frames. ], batch size: 247, lr: 2.07e-02, grad_scale: 8.0 2023-04-27 22:46:10,687 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24903.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:46:21,321 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24909.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:46:44,317 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24925.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:46:52,470 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.050e+02 3.022e+02 3.683e+02 4.480e+02 7.885e+02, threshold=7.365e+02, percent-clipped=1.0 2023-04-27 22:47:20,338 INFO [train.py:904] (6/8) Epoch 3, batch 4650, loss[loss=0.2258, simple_loss=0.3052, pruned_loss=0.07316, over 16738.00 frames. ], tot_loss[loss=0.2592, simple_loss=0.3337, pruned_loss=0.09233, over 3202510.52 frames. ], batch size: 39, lr: 2.07e-02, grad_scale: 8.0 2023-04-27 22:47:54,332 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0096, 3.1603, 3.1416, 1.3982, 3.3310, 3.3728, 2.7348, 2.4755], device='cuda:6'), covar=tensor([0.0843, 0.0119, 0.0171, 0.1360, 0.0090, 0.0052, 0.0314, 0.0405], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0081, 0.0080, 0.0148, 0.0075, 0.0072, 0.0113, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 22:47:57,320 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7698, 4.0251, 3.3593, 2.5776, 3.1490, 2.4082, 4.2652, 4.4730], device='cuda:6'), covar=tensor([0.2050, 0.0643, 0.1014, 0.0950, 0.1822, 0.1211, 0.0319, 0.0247], device='cuda:6'), in_proj_covar=tensor([0.0275, 0.0244, 0.0256, 0.0211, 0.0305, 0.0194, 0.0222, 0.0201], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 22:48:24,042 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24995.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:48:32,746 INFO [train.py:904] (6/8) Epoch 3, batch 4700, loss[loss=0.2349, simple_loss=0.3072, pruned_loss=0.08128, over 16810.00 frames. ], tot_loss[loss=0.2558, simple_loss=0.3304, pruned_loss=0.09056, over 3208066.51 frames. ], batch size: 39, lr: 2.06e-02, grad_scale: 4.0 2023-04-27 22:49:17,307 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.399e+02 3.338e+02 4.097e+02 4.784e+02 1.007e+03, threshold=8.194e+02, percent-clipped=3.0 2023-04-27 22:49:45,101 INFO [train.py:904] (6/8) Epoch 3, batch 4750, loss[loss=0.2371, simple_loss=0.316, pruned_loss=0.07906, over 16684.00 frames. ], tot_loss[loss=0.2514, simple_loss=0.3261, pruned_loss=0.08836, over 3213966.14 frames. ], batch size: 89, lr: 2.06e-02, grad_scale: 4.0 2023-04-27 22:49:52,728 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25056.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 22:50:36,923 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4875, 5.2553, 5.2505, 5.3547, 4.7162, 5.2228, 5.0956, 4.9441], device='cuda:6'), covar=tensor([0.0222, 0.0130, 0.0117, 0.0089, 0.0654, 0.0148, 0.0113, 0.0221], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0120, 0.0170, 0.0141, 0.0199, 0.0153, 0.0125, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 22:50:43,970 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7871, 3.2037, 3.0462, 2.1417, 2.9256, 2.8881, 2.8901, 1.6253], device='cuda:6'), covar=tensor([0.0352, 0.0023, 0.0036, 0.0224, 0.0039, 0.0094, 0.0034, 0.0334], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0051, 0.0060, 0.0111, 0.0053, 0.0064, 0.0059, 0.0106], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 22:50:58,608 INFO [train.py:904] (6/8) Epoch 3, batch 4800, loss[loss=0.2162, simple_loss=0.2943, pruned_loss=0.06903, over 16422.00 frames. ], tot_loss[loss=0.2471, simple_loss=0.3221, pruned_loss=0.08605, over 3206205.92 frames. ], batch size: 68, lr: 2.06e-02, grad_scale: 8.0 2023-04-27 22:51:25,555 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7136, 3.4119, 3.2829, 2.0608, 3.1315, 3.1851, 3.0596, 1.5746], device='cuda:6'), covar=tensor([0.0400, 0.0021, 0.0027, 0.0258, 0.0032, 0.0063, 0.0036, 0.0365], device='cuda:6'), in_proj_covar=tensor([0.0113, 0.0051, 0.0059, 0.0111, 0.0053, 0.0063, 0.0059, 0.0106], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 22:51:28,790 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25121.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:51:34,482 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2114, 2.2939, 2.1843, 3.5589, 1.8504, 3.2389, 2.0802, 2.0962], device='cuda:6'), covar=tensor([0.0399, 0.0776, 0.0482, 0.0231, 0.1767, 0.0265, 0.0983, 0.1350], device='cuda:6'), in_proj_covar=tensor([0.0254, 0.0230, 0.0193, 0.0254, 0.0305, 0.0203, 0.0218, 0.0300], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 22:51:40,001 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8804, 5.5221, 5.6315, 5.5047, 5.5019, 6.0876, 5.6375, 5.5338], device='cuda:6'), covar=tensor([0.0555, 0.0880, 0.0784, 0.1071, 0.1885, 0.0631, 0.0658, 0.1616], device='cuda:6'), in_proj_covar=tensor([0.0228, 0.0301, 0.0280, 0.0270, 0.0351, 0.0303, 0.0243, 0.0368], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 22:51:47,435 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.867e+02 2.955e+02 3.547e+02 4.621e+02 1.014e+03, threshold=7.094e+02, percent-clipped=1.0 2023-04-27 22:51:53,833 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.52 vs. limit=5.0 2023-04-27 22:52:13,127 INFO [train.py:904] (6/8) Epoch 3, batch 4850, loss[loss=0.2515, simple_loss=0.3321, pruned_loss=0.0855, over 16690.00 frames. ], tot_loss[loss=0.2469, simple_loss=0.3229, pruned_loss=0.08544, over 3206067.39 frames. ], batch size: 134, lr: 2.06e-02, grad_scale: 2.0 2023-04-27 22:53:25,028 INFO [train.py:904] (6/8) Epoch 3, batch 4900, loss[loss=0.2525, simple_loss=0.3118, pruned_loss=0.09657, over 16838.00 frames. ], tot_loss[loss=0.2451, simple_loss=0.3219, pruned_loss=0.08418, over 3201376.24 frames. ], batch size: 42, lr: 2.06e-02, grad_scale: 2.0 2023-04-27 22:53:28,522 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25203.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:53:29,470 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25204.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:53:59,453 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25225.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:54:10,579 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.158e+02 3.416e+02 4.003e+02 5.207e+02 1.200e+03, threshold=8.007e+02, percent-clipped=10.0 2023-04-27 22:54:35,307 INFO [train.py:904] (6/8) Epoch 3, batch 4950, loss[loss=0.2585, simple_loss=0.3365, pruned_loss=0.09028, over 16738.00 frames. ], tot_loss[loss=0.2458, simple_loss=0.3222, pruned_loss=0.08466, over 3201643.61 frames. ], batch size: 89, lr: 2.05e-02, grad_scale: 2.0 2023-04-27 22:54:36,191 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=25251.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:55:08,078 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=25273.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:55:48,336 INFO [train.py:904] (6/8) Epoch 3, batch 5000, loss[loss=0.2274, simple_loss=0.3161, pruned_loss=0.06936, over 16550.00 frames. ], tot_loss[loss=0.2478, simple_loss=0.3245, pruned_loss=0.08554, over 3200128.10 frames. ], batch size: 68, lr: 2.05e-02, grad_scale: 2.0 2023-04-27 22:55:48,798 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4606, 4.7932, 4.7306, 4.8249, 4.7508, 5.2826, 5.0053, 4.8323], device='cuda:6'), covar=tensor([0.0713, 0.1176, 0.0934, 0.1268, 0.1911, 0.0726, 0.0786, 0.1760], device='cuda:6'), in_proj_covar=tensor([0.0225, 0.0300, 0.0278, 0.0271, 0.0346, 0.0299, 0.0246, 0.0359], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 22:56:35,247 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.358e+02 3.469e+02 4.106e+02 5.107e+02 1.084e+03, threshold=8.211e+02, percent-clipped=3.0 2023-04-27 22:56:59,649 INFO [train.py:904] (6/8) Epoch 3, batch 5050, loss[loss=0.2379, simple_loss=0.3236, pruned_loss=0.07613, over 16738.00 frames. ], tot_loss[loss=0.247, simple_loss=0.3245, pruned_loss=0.08479, over 3202444.28 frames. ], batch size: 124, lr: 2.05e-02, grad_scale: 2.0 2023-04-27 22:56:59,957 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25351.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 22:57:13,787 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7236, 3.2100, 3.1693, 2.3723, 2.9157, 3.0363, 3.0904, 1.6817], device='cuda:6'), covar=tensor([0.0341, 0.0023, 0.0024, 0.0193, 0.0041, 0.0069, 0.0033, 0.0308], device='cuda:6'), in_proj_covar=tensor([0.0111, 0.0051, 0.0058, 0.0108, 0.0052, 0.0064, 0.0058, 0.0103], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 22:58:00,872 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1731, 3.2077, 1.6288, 3.2057, 2.3032, 3.2586, 1.6216, 2.4826], device='cuda:6'), covar=tensor([0.0049, 0.0195, 0.1325, 0.0041, 0.0662, 0.0280, 0.1314, 0.0549], device='cuda:6'), in_proj_covar=tensor([0.0085, 0.0129, 0.0169, 0.0077, 0.0156, 0.0155, 0.0175, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-27 22:58:08,622 INFO [train.py:904] (6/8) Epoch 3, batch 5100, loss[loss=0.2852, simple_loss=0.3463, pruned_loss=0.112, over 11854.00 frames. ], tot_loss[loss=0.2445, simple_loss=0.3221, pruned_loss=0.08339, over 3203880.25 frames. ], batch size: 247, lr: 2.05e-02, grad_scale: 2.0 2023-04-27 22:58:38,787 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25421.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:58:48,448 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=25428.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 22:58:57,532 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 3.320e+02 3.999e+02 5.208e+02 7.552e+02, threshold=7.999e+02, percent-clipped=0.0 2023-04-27 22:59:23,205 INFO [train.py:904] (6/8) Epoch 3, batch 5150, loss[loss=0.2828, simple_loss=0.3557, pruned_loss=0.1049, over 16720.00 frames. ], tot_loss[loss=0.2446, simple_loss=0.3228, pruned_loss=0.08323, over 3181462.00 frames. ], batch size: 124, lr: 2.05e-02, grad_scale: 2.0 2023-04-27 22:59:50,321 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=25469.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:00:08,762 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2828, 5.6549, 5.2787, 5.4547, 4.8240, 4.6890, 5.1061, 5.7637], device='cuda:6'), covar=tensor([0.0446, 0.0500, 0.0785, 0.0357, 0.0523, 0.0447, 0.0425, 0.0563], device='cuda:6'), in_proj_covar=tensor([0.0259, 0.0359, 0.0316, 0.0227, 0.0231, 0.0226, 0.0290, 0.0252], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 23:00:19,972 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25489.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 23:00:35,980 INFO [train.py:904] (6/8) Epoch 3, batch 5200, loss[loss=0.2215, simple_loss=0.2921, pruned_loss=0.07542, over 16692.00 frames. ], tot_loss[loss=0.2437, simple_loss=0.3211, pruned_loss=0.0831, over 3177444.81 frames. ], batch size: 57, lr: 2.04e-02, grad_scale: 4.0 2023-04-27 23:00:40,494 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25504.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:01:22,143 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.363e+02 3.525e+02 4.137e+02 5.123e+02 8.624e+02, threshold=8.275e+02, percent-clipped=4.0 2023-04-27 23:01:45,790 INFO [train.py:904] (6/8) Epoch 3, batch 5250, loss[loss=0.2529, simple_loss=0.33, pruned_loss=0.08793, over 15401.00 frames. ], tot_loss[loss=0.2421, simple_loss=0.3187, pruned_loss=0.08272, over 3189311.55 frames. ], batch size: 190, lr: 2.04e-02, grad_scale: 4.0 2023-04-27 23:01:47,893 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=25552.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:01:51,310 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-27 23:02:01,790 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1012, 4.1195, 4.1184, 4.0909, 4.0817, 4.6337, 4.3381, 3.9834], device='cuda:6'), covar=tensor([0.1192, 0.1138, 0.0981, 0.1383, 0.2138, 0.0863, 0.0883, 0.1977], device='cuda:6'), in_proj_covar=tensor([0.0235, 0.0313, 0.0291, 0.0278, 0.0359, 0.0309, 0.0251, 0.0371], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 23:02:56,098 INFO [train.py:904] (6/8) Epoch 3, batch 5300, loss[loss=0.209, simple_loss=0.2851, pruned_loss=0.06644, over 16396.00 frames. ], tot_loss[loss=0.2378, simple_loss=0.3142, pruned_loss=0.08065, over 3206704.64 frames. ], batch size: 146, lr: 2.04e-02, grad_scale: 4.0 2023-04-27 23:03:43,229 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.802e+02 3.089e+02 3.787e+02 4.480e+02 7.662e+02, threshold=7.574e+02, percent-clipped=0.0 2023-04-27 23:04:08,024 INFO [train.py:904] (6/8) Epoch 3, batch 5350, loss[loss=0.229, simple_loss=0.3126, pruned_loss=0.07265, over 16724.00 frames. ], tot_loss[loss=0.2365, simple_loss=0.3129, pruned_loss=0.08005, over 3203013.79 frames. ], batch size: 76, lr: 2.04e-02, grad_scale: 4.0 2023-04-27 23:04:08,388 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25651.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 23:04:35,607 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3378, 2.9708, 2.4985, 2.3479, 2.1982, 1.9885, 2.9374, 3.1868], device='cuda:6'), covar=tensor([0.1410, 0.0541, 0.0971, 0.0945, 0.1794, 0.1146, 0.0342, 0.0306], device='cuda:6'), in_proj_covar=tensor([0.0262, 0.0237, 0.0250, 0.0208, 0.0287, 0.0186, 0.0215, 0.0196], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 23:05:17,181 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=25699.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:05:19,278 INFO [train.py:904] (6/8) Epoch 3, batch 5400, loss[loss=0.245, simple_loss=0.328, pruned_loss=0.08099, over 16286.00 frames. ], tot_loss[loss=0.2392, simple_loss=0.3162, pruned_loss=0.08107, over 3201913.87 frames. ], batch size: 165, lr: 2.04e-02, grad_scale: 4.0 2023-04-27 23:05:51,230 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7080, 3.5136, 3.5794, 3.6186, 3.4600, 4.0775, 3.9103, 3.5910], device='cuda:6'), covar=tensor([0.1551, 0.1636, 0.1184, 0.1974, 0.2950, 0.1048, 0.0934, 0.2007], device='cuda:6'), in_proj_covar=tensor([0.0224, 0.0297, 0.0274, 0.0268, 0.0345, 0.0300, 0.0241, 0.0358], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 23:06:07,786 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.583e+02 3.733e+02 4.548e+02 5.512e+02 9.876e+02, threshold=9.097e+02, percent-clipped=3.0 2023-04-27 23:06:34,320 INFO [train.py:904] (6/8) Epoch 3, batch 5450, loss[loss=0.2604, simple_loss=0.3387, pruned_loss=0.09098, over 16525.00 frames. ], tot_loss[loss=0.245, simple_loss=0.3211, pruned_loss=0.08447, over 3198755.23 frames. ], batch size: 75, lr: 2.03e-02, grad_scale: 4.0 2023-04-27 23:07:24,732 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25784.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 23:07:49,275 INFO [train.py:904] (6/8) Epoch 3, batch 5500, loss[loss=0.3003, simple_loss=0.3628, pruned_loss=0.1189, over 16663.00 frames. ], tot_loss[loss=0.2595, simple_loss=0.3321, pruned_loss=0.09347, over 3161320.94 frames. ], batch size: 76, lr: 2.03e-02, grad_scale: 4.0 2023-04-27 23:08:09,479 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4631, 3.4432, 3.3734, 2.9967, 3.3608, 2.1158, 3.2150, 3.2687], device='cuda:6'), covar=tensor([0.0070, 0.0051, 0.0071, 0.0190, 0.0054, 0.1048, 0.0068, 0.0097], device='cuda:6'), in_proj_covar=tensor([0.0071, 0.0060, 0.0095, 0.0111, 0.0071, 0.0116, 0.0082, 0.0094], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 23:08:16,953 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0454, 3.8811, 3.4039, 1.6993, 2.7344, 2.2647, 3.2864, 3.8168], device='cuda:6'), covar=tensor([0.0270, 0.0389, 0.0435, 0.1588, 0.0719, 0.0873, 0.0684, 0.0470], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0113, 0.0156, 0.0146, 0.0139, 0.0132, 0.0146, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-27 23:08:39,220 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.596e+02 5.216e+02 6.145e+02 8.695e+02 2.860e+03, threshold=1.229e+03, percent-clipped=22.0 2023-04-27 23:09:06,197 INFO [train.py:904] (6/8) Epoch 3, batch 5550, loss[loss=0.4073, simple_loss=0.4403, pruned_loss=0.1872, over 15205.00 frames. ], tot_loss[loss=0.2727, simple_loss=0.3421, pruned_loss=0.1016, over 3126854.85 frames. ], batch size: 190, lr: 2.03e-02, grad_scale: 4.0 2023-04-27 23:09:18,164 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=25858.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 23:10:25,186 INFO [train.py:904] (6/8) Epoch 3, batch 5600, loss[loss=0.2888, simple_loss=0.3505, pruned_loss=0.1135, over 17031.00 frames. ], tot_loss[loss=0.2822, simple_loss=0.3492, pruned_loss=0.1076, over 3107006.80 frames. ], batch size: 55, lr: 2.03e-02, grad_scale: 8.0 2023-04-27 23:10:44,961 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0202, 1.7136, 1.5173, 1.5018, 1.8317, 1.7282, 1.6537, 1.9297], device='cuda:6'), covar=tensor([0.0016, 0.0073, 0.0099, 0.0093, 0.0056, 0.0067, 0.0041, 0.0045], device='cuda:6'), in_proj_covar=tensor([0.0053, 0.0114, 0.0117, 0.0119, 0.0111, 0.0121, 0.0074, 0.0096], device='cuda:6'), out_proj_covar=tensor([7.3577e-05, 1.6337e-04, 1.6314e-04, 1.7069e-04, 1.6206e-04, 1.7603e-04, 1.0586e-04, 1.4453e-04], device='cuda:6') 2023-04-27 23:10:56,792 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25919.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 23:11:21,465 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.427e+02 5.585e+02 6.873e+02 8.559e+02 2.132e+03, threshold=1.375e+03, percent-clipped=5.0 2023-04-27 23:11:48,513 INFO [train.py:904] (6/8) Epoch 3, batch 5650, loss[loss=0.4245, simple_loss=0.4442, pruned_loss=0.2024, over 11215.00 frames. ], tot_loss[loss=0.2918, simple_loss=0.3565, pruned_loss=0.1135, over 3086026.95 frames. ], batch size: 248, lr: 2.03e-02, grad_scale: 8.0 2023-04-27 23:12:01,038 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1813, 2.9795, 2.7692, 2.0468, 2.5863, 2.1784, 2.8041, 3.0406], device='cuda:6'), covar=tensor([0.0266, 0.0394, 0.0369, 0.1236, 0.0636, 0.0794, 0.0492, 0.0396], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0114, 0.0153, 0.0142, 0.0137, 0.0129, 0.0143, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-27 23:12:58,374 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9767, 5.2720, 4.9889, 5.0050, 4.6024, 4.2623, 4.7717, 5.4520], device='cuda:6'), covar=tensor([0.0436, 0.0547, 0.0787, 0.0366, 0.0483, 0.0715, 0.0465, 0.0489], device='cuda:6'), in_proj_covar=tensor([0.0259, 0.0363, 0.0323, 0.0231, 0.0234, 0.0236, 0.0293, 0.0251], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 23:13:02,891 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-27 23:13:10,042 INFO [train.py:904] (6/8) Epoch 3, batch 5700, loss[loss=0.2807, simple_loss=0.3654, pruned_loss=0.09802, over 16428.00 frames. ], tot_loss[loss=0.2943, simple_loss=0.3582, pruned_loss=0.1152, over 3080585.85 frames. ], batch size: 146, lr: 2.02e-02, grad_scale: 8.0 2023-04-27 23:13:27,502 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-27 23:13:39,175 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8756, 2.2569, 2.1976, 3.1075, 2.0420, 2.9141, 2.3651, 1.9500], device='cuda:6'), covar=tensor([0.0349, 0.0733, 0.0401, 0.0237, 0.1604, 0.0270, 0.0758, 0.1306], device='cuda:6'), in_proj_covar=tensor([0.0251, 0.0231, 0.0194, 0.0250, 0.0308, 0.0205, 0.0220, 0.0294], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 23:13:42,593 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-27 23:14:00,539 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.120e+02 5.064e+02 6.111e+02 7.661e+02 1.195e+03, threshold=1.222e+03, percent-clipped=0.0 2023-04-27 23:14:27,438 INFO [train.py:904] (6/8) Epoch 3, batch 5750, loss[loss=0.3108, simple_loss=0.3781, pruned_loss=0.1218, over 16759.00 frames. ], tot_loss[loss=0.2976, simple_loss=0.3611, pruned_loss=0.1171, over 3059013.81 frames. ], batch size: 39, lr: 2.02e-02, grad_scale: 8.0 2023-04-27 23:14:56,049 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0076, 1.4681, 2.0331, 2.7963, 2.6404, 2.7010, 1.6013, 2.8924], device='cuda:6'), covar=tensor([0.0038, 0.0220, 0.0142, 0.0079, 0.0066, 0.0079, 0.0193, 0.0045], device='cuda:6'), in_proj_covar=tensor([0.0085, 0.0120, 0.0107, 0.0101, 0.0092, 0.0069, 0.0115, 0.0063], device='cuda:6'), out_proj_covar=tensor([1.3183e-04, 1.8890e-04, 1.7564e-04, 1.6637e-04, 1.4537e-04, 1.0706e-04, 1.7929e-04, 9.8081e-05], device='cuda:6') 2023-04-27 23:15:22,008 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26084.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 23:15:34,445 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6083, 3.3523, 2.8727, 1.7595, 2.6234, 2.0913, 3.0589, 3.3457], device='cuda:6'), covar=tensor([0.0337, 0.0418, 0.0512, 0.1510, 0.0700, 0.0931, 0.0661, 0.0527], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0114, 0.0155, 0.0144, 0.0137, 0.0131, 0.0144, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-27 23:15:47,176 INFO [train.py:904] (6/8) Epoch 3, batch 5800, loss[loss=0.2353, simple_loss=0.3191, pruned_loss=0.07581, over 16925.00 frames. ], tot_loss[loss=0.296, simple_loss=0.3602, pruned_loss=0.1159, over 3045135.97 frames. ], batch size: 90, lr: 2.02e-02, grad_scale: 8.0 2023-04-27 23:15:48,976 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26102.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:16:36,164 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=26132.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 23:16:38,588 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.751e+02 5.049e+02 6.109e+02 8.114e+02 1.629e+03, threshold=1.222e+03, percent-clipped=2.0 2023-04-27 23:16:42,059 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-27 23:17:05,080 INFO [train.py:904] (6/8) Epoch 3, batch 5850, loss[loss=0.3157, simple_loss=0.3715, pruned_loss=0.13, over 16285.00 frames. ], tot_loss[loss=0.2914, simple_loss=0.3568, pruned_loss=0.113, over 3053141.83 frames. ], batch size: 165, lr: 2.02e-02, grad_scale: 8.0 2023-04-27 23:17:24,137 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26163.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:17:31,362 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26168.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:17:39,214 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.38 vs. limit=5.0 2023-04-27 23:18:26,965 INFO [train.py:904] (6/8) Epoch 3, batch 5900, loss[loss=0.2695, simple_loss=0.3357, pruned_loss=0.1016, over 15246.00 frames. ], tot_loss[loss=0.2888, simple_loss=0.3552, pruned_loss=0.1112, over 3064065.97 frames. ], batch size: 191, lr: 2.02e-02, grad_scale: 8.0 2023-04-27 23:18:51,905 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26214.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 23:19:15,279 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26229.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:19:22,630 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.765e+02 4.492e+02 5.249e+02 6.969e+02 1.603e+03, threshold=1.050e+03, percent-clipped=2.0 2023-04-27 23:19:47,796 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26250.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:19:48,574 INFO [train.py:904] (6/8) Epoch 3, batch 5950, loss[loss=0.2605, simple_loss=0.339, pruned_loss=0.09106, over 16920.00 frames. ], tot_loss[loss=0.2875, simple_loss=0.3559, pruned_loss=0.1095, over 3076427.64 frames. ], batch size: 96, lr: 2.02e-02, grad_scale: 8.0 2023-04-27 23:21:07,929 INFO [train.py:904] (6/8) Epoch 3, batch 6000, loss[loss=0.2425, simple_loss=0.3218, pruned_loss=0.08162, over 17036.00 frames. ], tot_loss[loss=0.2866, simple_loss=0.355, pruned_loss=0.1091, over 3084470.01 frames. ], batch size: 53, lr: 2.01e-02, grad_scale: 8.0 2023-04-27 23:21:07,929 INFO [train.py:929] (6/8) Computing validation loss 2023-04-27 23:21:18,890 INFO [train.py:938] (6/8) Epoch 3, validation: loss=0.2097, simple_loss=0.3184, pruned_loss=0.05055, over 944034.00 frames. 2023-04-27 23:21:18,891 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-27 23:21:34,621 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26311.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:22:07,342 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.939e+02 4.548e+02 5.519e+02 7.491e+02 1.813e+03, threshold=1.104e+03, percent-clipped=4.0 2023-04-27 23:22:36,230 INFO [train.py:904] (6/8) Epoch 3, batch 6050, loss[loss=0.2539, simple_loss=0.336, pruned_loss=0.08591, over 16612.00 frames. ], tot_loss[loss=0.2826, simple_loss=0.352, pruned_loss=0.1066, over 3110760.39 frames. ], batch size: 57, lr: 2.01e-02, grad_scale: 8.0 2023-04-27 23:22:48,876 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4969, 4.3865, 4.2410, 4.3183, 3.8366, 4.3096, 4.2765, 4.0412], device='cuda:6'), covar=tensor([0.0483, 0.0268, 0.0216, 0.0144, 0.0783, 0.0346, 0.0314, 0.0394], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0132, 0.0174, 0.0147, 0.0208, 0.0166, 0.0130, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 23:23:21,059 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1647, 2.3316, 2.1399, 3.4510, 1.8626, 3.1266, 2.2448, 2.1345], device='cuda:6'), covar=tensor([0.0401, 0.0865, 0.0549, 0.0288, 0.2006, 0.0307, 0.0928, 0.1265], device='cuda:6'), in_proj_covar=tensor([0.0258, 0.0235, 0.0197, 0.0256, 0.0313, 0.0209, 0.0224, 0.0298], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 23:23:51,471 INFO [train.py:904] (6/8) Epoch 3, batch 6100, loss[loss=0.2524, simple_loss=0.326, pruned_loss=0.08939, over 16807.00 frames. ], tot_loss[loss=0.2803, simple_loss=0.3507, pruned_loss=0.1049, over 3116588.59 frames. ], batch size: 124, lr: 2.01e-02, grad_scale: 8.0 2023-04-27 23:23:59,711 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6289, 4.7920, 4.9118, 4.9185, 4.8352, 5.3931, 5.0256, 4.7624], device='cuda:6'), covar=tensor([0.0731, 0.1358, 0.0952, 0.1447, 0.2194, 0.0734, 0.0965, 0.2148], device='cuda:6'), in_proj_covar=tensor([0.0229, 0.0316, 0.0294, 0.0277, 0.0366, 0.0317, 0.0256, 0.0370], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 23:24:12,789 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26414.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:24:42,456 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.827e+02 4.226e+02 5.750e+02 6.713e+02 1.587e+03, threshold=1.150e+03, percent-clipped=3.0 2023-04-27 23:25:11,826 INFO [train.py:904] (6/8) Epoch 3, batch 6150, loss[loss=0.2434, simple_loss=0.3133, pruned_loss=0.08679, over 16265.00 frames. ], tot_loss[loss=0.2781, simple_loss=0.3483, pruned_loss=0.1039, over 3119649.98 frames. ], batch size: 35, lr: 2.01e-02, grad_scale: 4.0 2023-04-27 23:25:23,340 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26458.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:25:50,251 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26475.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:26:28,208 INFO [train.py:904] (6/8) Epoch 3, batch 6200, loss[loss=0.3374, simple_loss=0.3776, pruned_loss=0.1486, over 11389.00 frames. ], tot_loss[loss=0.2781, simple_loss=0.3474, pruned_loss=0.1044, over 3108301.90 frames. ], batch size: 247, lr: 2.01e-02, grad_scale: 4.0 2023-04-27 23:26:48,635 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26514.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 23:27:03,497 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26524.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:27:18,520 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.570e+02 4.817e+02 5.994e+02 8.159e+02 2.733e+03, threshold=1.199e+03, percent-clipped=9.0 2023-04-27 23:27:41,889 INFO [train.py:904] (6/8) Epoch 3, batch 6250, loss[loss=0.2731, simple_loss=0.352, pruned_loss=0.09709, over 16504.00 frames. ], tot_loss[loss=0.2776, simple_loss=0.3472, pruned_loss=0.104, over 3115266.02 frames. ], batch size: 62, lr: 2.00e-02, grad_scale: 4.0 2023-04-27 23:27:58,187 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=26562.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 23:28:45,964 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7831, 3.9636, 3.8719, 3.9687, 2.8533, 4.0281, 3.9147, 3.5828], device='cuda:6'), covar=tensor([0.0797, 0.0412, 0.0362, 0.0279, 0.1745, 0.0362, 0.0526, 0.0474], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0131, 0.0169, 0.0142, 0.0200, 0.0162, 0.0126, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 23:28:54,857 INFO [train.py:904] (6/8) Epoch 3, batch 6300, loss[loss=0.2592, simple_loss=0.3305, pruned_loss=0.09397, over 16493.00 frames. ], tot_loss[loss=0.2767, simple_loss=0.3467, pruned_loss=0.1033, over 3120157.28 frames. ], batch size: 68, lr: 2.00e-02, grad_scale: 4.0 2023-04-27 23:29:02,921 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26606.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:29:48,250 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.782e+02 4.322e+02 5.254e+02 6.822e+02 1.499e+03, threshold=1.051e+03, percent-clipped=4.0 2023-04-27 23:30:11,907 INFO [train.py:904] (6/8) Epoch 3, batch 6350, loss[loss=0.2849, simple_loss=0.3541, pruned_loss=0.1079, over 16869.00 frames. ], tot_loss[loss=0.2801, simple_loss=0.3485, pruned_loss=0.1058, over 3109955.85 frames. ], batch size: 116, lr: 2.00e-02, grad_scale: 4.0 2023-04-27 23:31:25,029 INFO [train.py:904] (6/8) Epoch 3, batch 6400, loss[loss=0.251, simple_loss=0.3268, pruned_loss=0.08759, over 16339.00 frames. ], tot_loss[loss=0.2812, simple_loss=0.3488, pruned_loss=0.1068, over 3105638.47 frames. ], batch size: 146, lr: 2.00e-02, grad_scale: 8.0 2023-04-27 23:32:12,498 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.974e+02 5.048e+02 6.146e+02 7.949e+02 2.287e+03, threshold=1.229e+03, percent-clipped=11.0 2023-04-27 23:32:24,744 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-27 23:32:35,287 INFO [train.py:904] (6/8) Epoch 3, batch 6450, loss[loss=0.2428, simple_loss=0.3252, pruned_loss=0.08015, over 17063.00 frames. ], tot_loss[loss=0.2806, simple_loss=0.349, pruned_loss=0.1061, over 3095319.48 frames. ], batch size: 50, lr: 2.00e-02, grad_scale: 8.0 2023-04-27 23:32:47,318 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26758.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:33:05,757 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26770.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:33:53,754 INFO [train.py:904] (6/8) Epoch 3, batch 6500, loss[loss=0.2397, simple_loss=0.3241, pruned_loss=0.07766, over 16862.00 frames. ], tot_loss[loss=0.2773, simple_loss=0.3456, pruned_loss=0.1045, over 3093810.12 frames. ], batch size: 96, lr: 2.00e-02, grad_scale: 8.0 2023-04-27 23:34:01,035 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=26806.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:34:29,513 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26824.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:34:44,550 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.692e+02 4.621e+02 5.604e+02 6.997e+02 1.424e+03, threshold=1.121e+03, percent-clipped=2.0 2023-04-27 23:35:12,111 INFO [train.py:904] (6/8) Epoch 3, batch 6550, loss[loss=0.2912, simple_loss=0.3849, pruned_loss=0.09876, over 15257.00 frames. ], tot_loss[loss=0.2803, simple_loss=0.3488, pruned_loss=0.1059, over 3088586.77 frames. ], batch size: 190, lr: 1.99e-02, grad_scale: 8.0 2023-04-27 23:35:14,546 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-27 23:35:19,533 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.37 vs. limit=5.0 2023-04-27 23:35:45,522 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=26872.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:35:58,407 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7996, 2.5129, 2.7121, 1.9549, 2.4873, 2.5710, 2.4604, 1.8535], device='cuda:6'), covar=tensor([0.0248, 0.0030, 0.0041, 0.0162, 0.0030, 0.0057, 0.0028, 0.0224], device='cuda:6'), in_proj_covar=tensor([0.0110, 0.0049, 0.0056, 0.0108, 0.0051, 0.0061, 0.0057, 0.0103], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 23:36:15,210 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2040, 4.2029, 4.7445, 4.6698, 4.6939, 4.2858, 4.3134, 4.2179], device='cuda:6'), covar=tensor([0.0212, 0.0277, 0.0283, 0.0369, 0.0384, 0.0261, 0.0724, 0.0346], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0167, 0.0184, 0.0182, 0.0221, 0.0189, 0.0285, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-27 23:36:28,904 INFO [train.py:904] (6/8) Epoch 3, batch 6600, loss[loss=0.3223, simple_loss=0.3642, pruned_loss=0.1402, over 11313.00 frames. ], tot_loss[loss=0.2832, simple_loss=0.3522, pruned_loss=0.1071, over 3086065.82 frames. ], batch size: 246, lr: 1.99e-02, grad_scale: 8.0 2023-04-27 23:36:33,590 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26904.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 23:36:37,765 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26906.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:36:47,023 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4627, 3.4932, 3.9871, 3.9277, 3.9226, 3.5723, 3.6423, 3.6820], device='cuda:6'), covar=tensor([0.0250, 0.0302, 0.0275, 0.0370, 0.0379, 0.0297, 0.0835, 0.0359], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0169, 0.0188, 0.0186, 0.0224, 0.0192, 0.0290, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-27 23:36:52,610 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5225, 3.1748, 2.6382, 2.4593, 2.5101, 2.0141, 3.3153, 3.6206], device='cuda:6'), covar=tensor([0.1662, 0.0587, 0.0991, 0.0899, 0.1513, 0.1190, 0.0355, 0.0327], device='cuda:6'), in_proj_covar=tensor([0.0269, 0.0238, 0.0256, 0.0213, 0.0295, 0.0192, 0.0219, 0.0207], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 23:37:20,959 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.349e+02 4.809e+02 5.887e+02 7.395e+02 2.123e+03, threshold=1.177e+03, percent-clipped=7.0 2023-04-27 23:37:45,795 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26950.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:37:46,398 INFO [train.py:904] (6/8) Epoch 3, batch 6650, loss[loss=0.2332, simple_loss=0.3106, pruned_loss=0.07788, over 16636.00 frames. ], tot_loss[loss=0.2845, simple_loss=0.3526, pruned_loss=0.1082, over 3085508.63 frames. ], batch size: 68, lr: 1.99e-02, grad_scale: 8.0 2023-04-27 23:37:51,228 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=26954.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:38:09,012 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26965.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 23:39:03,454 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-27 23:39:03,588 INFO [train.py:904] (6/8) Epoch 3, batch 6700, loss[loss=0.2681, simple_loss=0.3381, pruned_loss=0.09905, over 16268.00 frames. ], tot_loss[loss=0.2834, simple_loss=0.351, pruned_loss=0.1079, over 3080226.62 frames. ], batch size: 165, lr: 1.99e-02, grad_scale: 8.0 2023-04-27 23:39:19,071 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27011.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:39:57,110 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.121e+02 4.872e+02 5.906e+02 7.141e+02 1.703e+03, threshold=1.181e+03, percent-clipped=1.0 2023-04-27 23:40:21,183 INFO [train.py:904] (6/8) Epoch 3, batch 6750, loss[loss=0.2759, simple_loss=0.3467, pruned_loss=0.1025, over 16752.00 frames. ], tot_loss[loss=0.2825, simple_loss=0.35, pruned_loss=0.1075, over 3099692.10 frames. ], batch size: 124, lr: 1.99e-02, grad_scale: 8.0 2023-04-27 23:40:51,241 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27070.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:41:02,242 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9288, 3.8260, 3.7613, 3.8048, 3.3387, 3.8138, 3.5490, 3.6164], device='cuda:6'), covar=tensor([0.0311, 0.0208, 0.0202, 0.0143, 0.0646, 0.0217, 0.0718, 0.0333], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0131, 0.0171, 0.0143, 0.0204, 0.0164, 0.0129, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-27 23:41:37,189 INFO [train.py:904] (6/8) Epoch 3, batch 6800, loss[loss=0.2675, simple_loss=0.3409, pruned_loss=0.09705, over 16615.00 frames. ], tot_loss[loss=0.2815, simple_loss=0.3494, pruned_loss=0.1067, over 3104485.41 frames. ], batch size: 62, lr: 1.98e-02, grad_scale: 8.0 2023-04-27 23:42:03,929 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=27118.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:42:19,846 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1495, 4.9383, 5.1727, 5.4414, 5.4597, 4.7861, 5.5179, 5.4440], device='cuda:6'), covar=tensor([0.0619, 0.0530, 0.0878, 0.0306, 0.0371, 0.0458, 0.0297, 0.0286], device='cuda:6'), in_proj_covar=tensor([0.0284, 0.0349, 0.0450, 0.0343, 0.0261, 0.0249, 0.0283, 0.0283], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 23:42:31,272 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.734e+02 4.279e+02 5.417e+02 7.296e+02 1.152e+03, threshold=1.083e+03, percent-clipped=0.0 2023-04-27 23:42:47,853 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5855, 4.1738, 4.1689, 1.7249, 4.4484, 4.4655, 3.6554, 3.6152], device='cuda:6'), covar=tensor([0.0770, 0.0051, 0.0093, 0.1338, 0.0037, 0.0027, 0.0199, 0.0281], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0079, 0.0079, 0.0143, 0.0072, 0.0070, 0.0114, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 23:42:51,698 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=27148.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:42:55,392 INFO [train.py:904] (6/8) Epoch 3, batch 6850, loss[loss=0.2645, simple_loss=0.3641, pruned_loss=0.08247, over 16727.00 frames. ], tot_loss[loss=0.2825, simple_loss=0.3508, pruned_loss=0.1071, over 3100276.05 frames. ], batch size: 89, lr: 1.98e-02, grad_scale: 8.0 2023-04-27 23:44:09,591 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5260, 4.3386, 3.6818, 1.8623, 2.9537, 2.6410, 3.9454, 4.2808], device='cuda:6'), covar=tensor([0.0203, 0.0337, 0.0453, 0.1429, 0.0613, 0.0808, 0.0450, 0.0496], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0115, 0.0156, 0.0143, 0.0135, 0.0129, 0.0143, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-27 23:44:10,168 INFO [train.py:904] (6/8) Epoch 3, batch 6900, loss[loss=0.2831, simple_loss=0.3477, pruned_loss=0.1092, over 16806.00 frames. ], tot_loss[loss=0.2833, simple_loss=0.3532, pruned_loss=0.1067, over 3109145.44 frames. ], batch size: 57, lr: 1.98e-02, grad_scale: 8.0 2023-04-27 23:44:22,746 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27209.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:44:32,926 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-27 23:45:02,575 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.150e+02 4.514e+02 5.203e+02 6.904e+02 1.170e+03, threshold=1.041e+03, percent-clipped=2.0 2023-04-27 23:45:28,456 INFO [train.py:904] (6/8) Epoch 3, batch 6950, loss[loss=0.305, simple_loss=0.3639, pruned_loss=0.123, over 16313.00 frames. ], tot_loss[loss=0.2882, simple_loss=0.3559, pruned_loss=0.1102, over 3078996.00 frames. ], batch size: 165, lr: 1.98e-02, grad_scale: 8.0 2023-04-27 23:45:42,184 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27260.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 23:46:43,369 INFO [train.py:904] (6/8) Epoch 3, batch 7000, loss[loss=0.2594, simple_loss=0.3408, pruned_loss=0.08898, over 16706.00 frames. ], tot_loss[loss=0.2873, simple_loss=0.3555, pruned_loss=0.1096, over 3072955.44 frames. ], batch size: 57, lr: 1.98e-02, grad_scale: 8.0 2023-04-27 23:46:51,663 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27306.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:47:35,868 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.986e+02 5.041e+02 6.145e+02 8.860e+02 1.565e+03, threshold=1.229e+03, percent-clipped=7.0 2023-04-27 23:47:47,033 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0622, 2.7246, 2.6108, 1.7858, 2.8206, 2.7986, 2.4376, 2.3560], device='cuda:6'), covar=tensor([0.0591, 0.0128, 0.0157, 0.0881, 0.0080, 0.0087, 0.0343, 0.0382], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0079, 0.0078, 0.0140, 0.0070, 0.0070, 0.0112, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 23:48:01,107 INFO [train.py:904] (6/8) Epoch 3, batch 7050, loss[loss=0.2748, simple_loss=0.3507, pruned_loss=0.09948, over 16746.00 frames. ], tot_loss[loss=0.2862, simple_loss=0.3557, pruned_loss=0.1083, over 3090039.93 frames. ], batch size: 134, lr: 1.98e-02, grad_scale: 8.0 2023-04-27 23:49:19,630 INFO [train.py:904] (6/8) Epoch 3, batch 7100, loss[loss=0.3121, simple_loss=0.3724, pruned_loss=0.1259, over 16903.00 frames. ], tot_loss[loss=0.2849, simple_loss=0.3541, pruned_loss=0.1078, over 3092396.50 frames. ], batch size: 116, lr: 1.97e-02, grad_scale: 8.0 2023-04-27 23:50:12,101 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.055e+02 4.788e+02 6.087e+02 7.690e+02 2.114e+03, threshold=1.217e+03, percent-clipped=3.0 2023-04-27 23:50:36,353 INFO [train.py:904] (6/8) Epoch 3, batch 7150, loss[loss=0.2563, simple_loss=0.3311, pruned_loss=0.0908, over 16859.00 frames. ], tot_loss[loss=0.2814, simple_loss=0.3505, pruned_loss=0.1061, over 3118995.15 frames. ], batch size: 96, lr: 1.97e-02, grad_scale: 8.0 2023-04-27 23:50:51,907 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8657, 3.5049, 3.5486, 2.6201, 3.4252, 3.4084, 3.6348, 1.9862], device='cuda:6'), covar=tensor([0.0344, 0.0022, 0.0027, 0.0196, 0.0029, 0.0061, 0.0018, 0.0295], device='cuda:6'), in_proj_covar=tensor([0.0110, 0.0051, 0.0057, 0.0109, 0.0052, 0.0061, 0.0058, 0.0105], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-27 23:51:51,150 INFO [train.py:904] (6/8) Epoch 3, batch 7200, loss[loss=0.273, simple_loss=0.3353, pruned_loss=0.1054, over 11748.00 frames. ], tot_loss[loss=0.2784, simple_loss=0.3481, pruned_loss=0.1043, over 3099864.81 frames. ], batch size: 248, lr: 1.97e-02, grad_scale: 8.0 2023-04-27 23:51:55,800 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27504.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:52:45,515 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.403e+02 3.669e+02 4.524e+02 6.083e+02 1.066e+03, threshold=9.047e+02, percent-clipped=1.0 2023-04-27 23:53:12,430 INFO [train.py:904] (6/8) Epoch 3, batch 7250, loss[loss=0.2353, simple_loss=0.3123, pruned_loss=0.07914, over 16761.00 frames. ], tot_loss[loss=0.2741, simple_loss=0.3448, pruned_loss=0.1017, over 3108485.52 frames. ], batch size: 124, lr: 1.97e-02, grad_scale: 8.0 2023-04-27 23:53:26,533 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27560.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 23:53:45,171 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2857, 4.1308, 4.6730, 4.6316, 4.6897, 4.2145, 4.3446, 4.1146], device='cuda:6'), covar=tensor([0.0185, 0.0255, 0.0276, 0.0353, 0.0314, 0.0235, 0.0575, 0.0304], device='cuda:6'), in_proj_covar=tensor([0.0189, 0.0173, 0.0192, 0.0188, 0.0228, 0.0196, 0.0296, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-27 23:53:47,236 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4751, 3.4212, 2.8524, 2.3043, 2.5808, 2.1586, 3.4693, 3.7870], device='cuda:6'), covar=tensor([0.1975, 0.0705, 0.1117, 0.1074, 0.1686, 0.1206, 0.0386, 0.0420], device='cuda:6'), in_proj_covar=tensor([0.0268, 0.0241, 0.0254, 0.0213, 0.0299, 0.0191, 0.0220, 0.0207], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 23:54:24,394 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2265, 3.5042, 3.7031, 1.5584, 3.9547, 3.9278, 3.0514, 2.7911], device='cuda:6'), covar=tensor([0.0852, 0.0117, 0.0135, 0.1161, 0.0050, 0.0047, 0.0289, 0.0406], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0081, 0.0078, 0.0143, 0.0071, 0.0071, 0.0113, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-27 23:54:26,522 INFO [train.py:904] (6/8) Epoch 3, batch 7300, loss[loss=0.2641, simple_loss=0.3409, pruned_loss=0.09369, over 16576.00 frames. ], tot_loss[loss=0.2737, simple_loss=0.3439, pruned_loss=0.1017, over 3094226.38 frames. ], batch size: 62, lr: 1.97e-02, grad_scale: 8.0 2023-04-27 23:54:27,661 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2513, 4.0313, 4.2061, 4.4592, 4.5224, 3.9598, 4.5931, 4.4783], device='cuda:6'), covar=tensor([0.0615, 0.0604, 0.0959, 0.0370, 0.0322, 0.0707, 0.0249, 0.0393], device='cuda:6'), in_proj_covar=tensor([0.0285, 0.0353, 0.0454, 0.0344, 0.0261, 0.0250, 0.0278, 0.0287], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-27 23:54:35,482 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27606.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:54:38,058 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=27608.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 23:55:17,367 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.622e+02 4.400e+02 5.943e+02 7.946e+02 1.369e+03, threshold=1.189e+03, percent-clipped=13.0 2023-04-27 23:55:40,973 INFO [train.py:904] (6/8) Epoch 3, batch 7350, loss[loss=0.2528, simple_loss=0.3287, pruned_loss=0.08846, over 16850.00 frames. ], tot_loss[loss=0.2745, simple_loss=0.3443, pruned_loss=0.1024, over 3068428.86 frames. ], batch size: 116, lr: 1.97e-02, grad_scale: 8.0 2023-04-27 23:55:46,792 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=27654.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:56:59,424 INFO [train.py:904] (6/8) Epoch 3, batch 7400, loss[loss=0.2763, simple_loss=0.3529, pruned_loss=0.09989, over 16376.00 frames. ], tot_loss[loss=0.2771, simple_loss=0.346, pruned_loss=0.104, over 3057825.08 frames. ], batch size: 146, lr: 1.96e-02, grad_scale: 4.0 2023-04-27 23:57:08,502 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-27 23:57:19,661 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2023-04-27 23:57:20,356 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=27714.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 23:57:55,128 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.894e+02 4.696e+02 5.728e+02 6.981e+02 1.392e+03, threshold=1.146e+03, percent-clipped=2.0 2023-04-27 23:58:18,285 INFO [train.py:904] (6/8) Epoch 3, batch 7450, loss[loss=0.252, simple_loss=0.3256, pruned_loss=0.08914, over 16823.00 frames. ], tot_loss[loss=0.2795, simple_loss=0.3477, pruned_loss=0.1056, over 3049798.05 frames. ], batch size: 42, lr: 1.96e-02, grad_scale: 4.0 2023-04-27 23:58:58,364 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27775.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 23:59:39,384 INFO [train.py:904] (6/8) Epoch 3, batch 7500, loss[loss=0.2526, simple_loss=0.3324, pruned_loss=0.08634, over 16849.00 frames. ], tot_loss[loss=0.2787, simple_loss=0.348, pruned_loss=0.1047, over 3047498.75 frames. ], batch size: 102, lr: 1.96e-02, grad_scale: 4.0 2023-04-27 23:59:44,254 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27804.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:00:33,135 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.883e+02 4.957e+02 5.924e+02 7.629e+02 1.550e+03, threshold=1.185e+03, percent-clipped=6.0 2023-04-28 00:00:41,308 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8525, 2.8325, 2.4376, 4.0029, 3.7874, 3.7996, 1.6632, 2.9946], device='cuda:6'), covar=tensor([0.1221, 0.0466, 0.1036, 0.0070, 0.0214, 0.0283, 0.1153, 0.0598], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0133, 0.0160, 0.0071, 0.0136, 0.0144, 0.0152, 0.0156], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-28 00:00:55,939 INFO [train.py:904] (6/8) Epoch 3, batch 7550, loss[loss=0.2655, simple_loss=0.3416, pruned_loss=0.09474, over 16250.00 frames. ], tot_loss[loss=0.2772, simple_loss=0.3464, pruned_loss=0.104, over 3058225.84 frames. ], batch size: 165, lr: 1.96e-02, grad_scale: 4.0 2023-04-28 00:00:58,762 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=27852.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:02:13,359 INFO [train.py:904] (6/8) Epoch 3, batch 7600, loss[loss=0.2689, simple_loss=0.3406, pruned_loss=0.09862, over 16656.00 frames. ], tot_loss[loss=0.2769, simple_loss=0.3456, pruned_loss=0.104, over 3071380.65 frames. ], batch size: 57, lr: 1.96e-02, grad_scale: 8.0 2023-04-28 00:02:37,615 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0780, 3.9861, 3.9466, 3.2875, 3.9750, 1.7507, 3.7051, 3.8054], device='cuda:6'), covar=tensor([0.0063, 0.0060, 0.0079, 0.0294, 0.0055, 0.1359, 0.0071, 0.0115], device='cuda:6'), in_proj_covar=tensor([0.0073, 0.0062, 0.0095, 0.0109, 0.0069, 0.0119, 0.0082, 0.0093], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:03:08,897 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.035e+02 4.816e+02 6.068e+02 7.688e+02 1.978e+03, threshold=1.214e+03, percent-clipped=6.0 2023-04-28 00:03:18,098 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-28 00:03:31,739 INFO [train.py:904] (6/8) Epoch 3, batch 7650, loss[loss=0.2848, simple_loss=0.3578, pruned_loss=0.1059, over 16721.00 frames. ], tot_loss[loss=0.2791, simple_loss=0.3473, pruned_loss=0.1054, over 3084774.81 frames. ], batch size: 124, lr: 1.96e-02, grad_scale: 8.0 2023-04-28 00:04:32,141 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.97 vs. limit=5.0 2023-04-28 00:04:40,451 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5914, 4.2633, 4.0023, 1.9703, 3.3545, 2.4406, 4.0826, 4.3526], device='cuda:6'), covar=tensor([0.0240, 0.0534, 0.0510, 0.1905, 0.0702, 0.1065, 0.0520, 0.0789], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0120, 0.0157, 0.0147, 0.0138, 0.0132, 0.0146, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 00:04:52,572 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.99 vs. limit=5.0 2023-04-28 00:04:52,685 INFO [train.py:904] (6/8) Epoch 3, batch 7700, loss[loss=0.3419, simple_loss=0.379, pruned_loss=0.1524, over 11413.00 frames. ], tot_loss[loss=0.2814, simple_loss=0.3486, pruned_loss=0.1071, over 3079541.16 frames. ], batch size: 248, lr: 1.95e-02, grad_scale: 8.0 2023-04-28 00:05:46,975 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.821e+02 4.680e+02 5.750e+02 6.748e+02 1.214e+03, threshold=1.150e+03, percent-clipped=1.0 2023-04-28 00:06:11,241 INFO [train.py:904] (6/8) Epoch 3, batch 7750, loss[loss=0.238, simple_loss=0.3187, pruned_loss=0.07868, over 16754.00 frames. ], tot_loss[loss=0.2813, simple_loss=0.3489, pruned_loss=0.1069, over 3086083.30 frames. ], batch size: 83, lr: 1.95e-02, grad_scale: 8.0 2023-04-28 00:06:15,695 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1375, 3.9033, 3.5525, 1.9146, 2.7544, 2.3481, 3.4878, 3.7224], device='cuda:6'), covar=tensor([0.0246, 0.0464, 0.0443, 0.1687, 0.0750, 0.0959, 0.0637, 0.0681], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0121, 0.0159, 0.0149, 0.0140, 0.0133, 0.0148, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-28 00:06:40,888 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28070.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 00:07:28,338 INFO [train.py:904] (6/8) Epoch 3, batch 7800, loss[loss=0.4006, simple_loss=0.4245, pruned_loss=0.1883, over 11343.00 frames. ], tot_loss[loss=0.2826, simple_loss=0.35, pruned_loss=0.1075, over 3090006.93 frames. ], batch size: 248, lr: 1.95e-02, grad_scale: 8.0 2023-04-28 00:08:22,842 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.157e+02 4.949e+02 5.902e+02 7.570e+02 1.555e+03, threshold=1.180e+03, percent-clipped=4.0 2023-04-28 00:08:45,059 INFO [train.py:904] (6/8) Epoch 3, batch 7850, loss[loss=0.2877, simple_loss=0.3682, pruned_loss=0.1036, over 16261.00 frames. ], tot_loss[loss=0.2825, simple_loss=0.3507, pruned_loss=0.1072, over 3085165.43 frames. ], batch size: 165, lr: 1.95e-02, grad_scale: 8.0 2023-04-28 00:10:00,834 INFO [train.py:904] (6/8) Epoch 3, batch 7900, loss[loss=0.3396, simple_loss=0.3792, pruned_loss=0.1499, over 11517.00 frames. ], tot_loss[loss=0.2807, simple_loss=0.3492, pruned_loss=0.1061, over 3098253.89 frames. ], batch size: 246, lr: 1.95e-02, grad_scale: 8.0 2023-04-28 00:10:45,272 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28229.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:10:55,797 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.692e+02 4.334e+02 4.942e+02 5.999e+02 2.073e+03, threshold=9.884e+02, percent-clipped=3.0 2023-04-28 00:11:18,498 INFO [train.py:904] (6/8) Epoch 3, batch 7950, loss[loss=0.3469, simple_loss=0.3913, pruned_loss=0.1513, over 11408.00 frames. ], tot_loss[loss=0.2811, simple_loss=0.349, pruned_loss=0.1065, over 3087538.80 frames. ], batch size: 248, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:11:24,745 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0557, 5.3379, 5.0252, 5.1297, 4.7229, 4.4955, 4.8111, 5.4094], device='cuda:6'), covar=tensor([0.0430, 0.0507, 0.0793, 0.0345, 0.0460, 0.0595, 0.0446, 0.0521], device='cuda:6'), in_proj_covar=tensor([0.0278, 0.0381, 0.0342, 0.0244, 0.0245, 0.0254, 0.0307, 0.0264], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:11:26,004 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28255.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 00:11:29,684 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2895, 4.2105, 1.8221, 4.4854, 2.6110, 4.5194, 2.0500, 3.0098], device='cuda:6'), covar=tensor([0.0051, 0.0175, 0.1582, 0.0021, 0.0752, 0.0211, 0.1433, 0.0526], device='cuda:6'), in_proj_covar=tensor([0.0089, 0.0134, 0.0173, 0.0080, 0.0161, 0.0161, 0.0180, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 00:12:14,008 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28287.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:12:18,748 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28290.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:12:33,513 INFO [train.py:904] (6/8) Epoch 3, batch 8000, loss[loss=0.3238, simple_loss=0.3673, pruned_loss=0.1401, over 11550.00 frames. ], tot_loss[loss=0.2833, simple_loss=0.3504, pruned_loss=0.1081, over 3064169.37 frames. ], batch size: 246, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:12:44,901 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4990, 4.4980, 4.9786, 4.9952, 4.9902, 4.5661, 4.5754, 4.3416], device='cuda:6'), covar=tensor([0.0202, 0.0234, 0.0240, 0.0304, 0.0302, 0.0220, 0.0640, 0.0300], device='cuda:6'), in_proj_covar=tensor([0.0191, 0.0181, 0.0195, 0.0197, 0.0232, 0.0202, 0.0301, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 00:12:56,276 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28316.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 00:13:27,063 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.455e+02 4.065e+02 5.202e+02 6.886e+02 1.574e+03, threshold=1.040e+03, percent-clipped=4.0 2023-04-28 00:13:46,465 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28348.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:13:49,305 INFO [train.py:904] (6/8) Epoch 3, batch 8050, loss[loss=0.2496, simple_loss=0.3282, pruned_loss=0.08551, over 16911.00 frames. ], tot_loss[loss=0.2818, simple_loss=0.3494, pruned_loss=0.1071, over 3074994.05 frames. ], batch size: 109, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:14:18,757 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28370.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 00:14:58,047 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5255, 2.8338, 2.9829, 1.9409, 2.9070, 2.9222, 2.9278, 1.6685], device='cuda:6'), covar=tensor([0.0402, 0.0034, 0.0042, 0.0261, 0.0040, 0.0068, 0.0028, 0.0320], device='cuda:6'), in_proj_covar=tensor([0.0112, 0.0054, 0.0057, 0.0111, 0.0053, 0.0063, 0.0058, 0.0106], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 00:15:05,404 INFO [train.py:904] (6/8) Epoch 3, batch 8100, loss[loss=0.2624, simple_loss=0.3351, pruned_loss=0.09488, over 16896.00 frames. ], tot_loss[loss=0.2796, simple_loss=0.3479, pruned_loss=0.1056, over 3088674.09 frames. ], batch size: 96, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:15:21,936 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9002, 3.9820, 2.9695, 2.5054, 3.1988, 2.4752, 4.1019, 4.4887], device='cuda:6'), covar=tensor([0.1974, 0.0563, 0.1248, 0.1158, 0.1785, 0.1173, 0.0425, 0.0280], device='cuda:6'), in_proj_covar=tensor([0.0267, 0.0237, 0.0251, 0.0215, 0.0300, 0.0192, 0.0221, 0.0210], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:15:30,655 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=28418.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 00:15:57,041 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.864e+02 4.688e+02 5.875e+02 7.578e+02 1.406e+03, threshold=1.175e+03, percent-clipped=5.0 2023-04-28 00:16:20,001 INFO [train.py:904] (6/8) Epoch 3, batch 8150, loss[loss=0.253, simple_loss=0.3247, pruned_loss=0.09063, over 16857.00 frames. ], tot_loss[loss=0.278, simple_loss=0.3461, pruned_loss=0.105, over 3092042.95 frames. ], batch size: 96, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:16:27,465 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-28 00:17:15,186 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28486.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:17:23,246 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28492.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:17:36,817 INFO [train.py:904] (6/8) Epoch 3, batch 8200, loss[loss=0.2828, simple_loss=0.3566, pruned_loss=0.1045, over 16251.00 frames. ], tot_loss[loss=0.2753, simple_loss=0.3432, pruned_loss=0.1037, over 3102221.00 frames. ], batch size: 165, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:17:46,264 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-28 00:18:21,628 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.78 vs. limit=5.0 2023-04-28 00:18:31,946 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.657e+02 4.728e+02 5.816e+02 7.305e+02 1.524e+03, threshold=1.163e+03, percent-clipped=3.0 2023-04-28 00:18:32,576 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7036, 4.9365, 4.7245, 4.7066, 4.3810, 4.2120, 4.5250, 4.9416], device='cuda:6'), covar=tensor([0.0424, 0.0566, 0.0674, 0.0365, 0.0451, 0.0658, 0.0454, 0.0621], device='cuda:6'), in_proj_covar=tensor([0.0278, 0.0383, 0.0334, 0.0249, 0.0247, 0.0254, 0.0304, 0.0263], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:18:50,489 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28547.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:18:55,851 INFO [train.py:904] (6/8) Epoch 3, batch 8250, loss[loss=0.2295, simple_loss=0.3065, pruned_loss=0.0763, over 11940.00 frames. ], tot_loss[loss=0.2727, simple_loss=0.3423, pruned_loss=0.1015, over 3085865.27 frames. ], batch size: 246, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:19:00,274 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28553.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:19:51,769 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28585.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:20:16,982 INFO [train.py:904] (6/8) Epoch 3, batch 8300, loss[loss=0.2472, simple_loss=0.3336, pruned_loss=0.08041, over 16906.00 frames. ], tot_loss[loss=0.2653, simple_loss=0.3375, pruned_loss=0.09659, over 3086520.68 frames. ], batch size: 116, lr: 1.93e-02, grad_scale: 8.0 2023-04-28 00:20:34,453 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28611.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 00:21:14,648 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.491e+02 3.724e+02 4.538e+02 5.868e+02 1.330e+03, threshold=9.076e+02, percent-clipped=2.0 2023-04-28 00:21:27,064 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28643.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:21:39,706 INFO [train.py:904] (6/8) Epoch 3, batch 8350, loss[loss=0.2509, simple_loss=0.3329, pruned_loss=0.08446, over 16852.00 frames. ], tot_loss[loss=0.2603, simple_loss=0.335, pruned_loss=0.09279, over 3082750.92 frames. ], batch size: 116, lr: 1.93e-02, grad_scale: 8.0 2023-04-28 00:21:52,525 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-28 00:23:00,530 INFO [train.py:904] (6/8) Epoch 3, batch 8400, loss[loss=0.2477, simple_loss=0.3133, pruned_loss=0.09102, over 12032.00 frames. ], tot_loss[loss=0.2543, simple_loss=0.3304, pruned_loss=0.08907, over 3076167.50 frames. ], batch size: 247, lr: 1.93e-02, grad_scale: 8.0 2023-04-28 00:23:03,311 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1024, 4.0019, 3.7038, 1.9190, 2.7467, 2.6334, 3.2693, 3.8804], device='cuda:6'), covar=tensor([0.0320, 0.0515, 0.0469, 0.1613, 0.0824, 0.0828, 0.0878, 0.0614], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0115, 0.0151, 0.0144, 0.0134, 0.0128, 0.0141, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 00:23:30,012 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-28 00:23:58,231 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.191e+02 3.936e+02 4.649e+02 5.360e+02 9.184e+02, threshold=9.299e+02, percent-clipped=1.0 2023-04-28 00:23:58,829 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8629, 3.1185, 2.9912, 2.1845, 3.0097, 3.0301, 2.9622, 1.7633], device='cuda:6'), covar=tensor([0.0316, 0.0022, 0.0035, 0.0216, 0.0028, 0.0045, 0.0028, 0.0320], device='cuda:6'), in_proj_covar=tensor([0.0112, 0.0054, 0.0057, 0.0111, 0.0052, 0.0063, 0.0058, 0.0108], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 00:24:20,129 INFO [train.py:904] (6/8) Epoch 3, batch 8450, loss[loss=0.2325, simple_loss=0.3159, pruned_loss=0.07456, over 16578.00 frames. ], tot_loss[loss=0.2502, simple_loss=0.3277, pruned_loss=0.08637, over 3071529.64 frames. ], batch size: 134, lr: 1.93e-02, grad_scale: 4.0 2023-04-28 00:24:38,296 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7678, 3.6512, 3.7360, 3.6894, 3.7228, 4.1563, 3.9324, 3.5684], device='cuda:6'), covar=tensor([0.1642, 0.1360, 0.1081, 0.1523, 0.2258, 0.1107, 0.1021, 0.2173], device='cuda:6'), in_proj_covar=tensor([0.0220, 0.0306, 0.0282, 0.0264, 0.0338, 0.0316, 0.0249, 0.0354], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:25:07,818 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7151, 4.6658, 4.5397, 3.6248, 4.4263, 1.9565, 4.2709, 4.4322], device='cuda:6'), covar=tensor([0.0066, 0.0053, 0.0073, 0.0316, 0.0068, 0.1482, 0.0089, 0.0127], device='cuda:6'), in_proj_covar=tensor([0.0073, 0.0062, 0.0097, 0.0104, 0.0071, 0.0123, 0.0084, 0.0092], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:25:41,659 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2023-04-28 00:25:42,012 INFO [train.py:904] (6/8) Epoch 3, batch 8500, loss[loss=0.2243, simple_loss=0.3054, pruned_loss=0.07163, over 16712.00 frames. ], tot_loss[loss=0.2432, simple_loss=0.3222, pruned_loss=0.08214, over 3073066.35 frames. ], batch size: 124, lr: 1.93e-02, grad_scale: 4.0 2023-04-28 00:26:40,691 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.012e+02 3.668e+02 4.634e+02 5.862e+02 2.485e+03, threshold=9.268e+02, percent-clipped=6.0 2023-04-28 00:26:50,002 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28842.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:27:01,008 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:27:05,698 INFO [train.py:904] (6/8) Epoch 3, batch 8550, loss[loss=0.2557, simple_loss=0.3383, pruned_loss=0.08656, over 16309.00 frames. ], tot_loss[loss=0.2411, simple_loss=0.3199, pruned_loss=0.08112, over 3056688.38 frames. ], batch size: 165, lr: 1.93e-02, grad_scale: 4.0 2023-04-28 00:27:25,983 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28861.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:28:09,078 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2106, 2.2512, 2.2221, 3.5566, 1.8737, 3.2181, 2.1678, 2.0316], device='cuda:6'), covar=tensor([0.0363, 0.0971, 0.0564, 0.0227, 0.1987, 0.0328, 0.1064, 0.1574], device='cuda:6'), in_proj_covar=tensor([0.0255, 0.0239, 0.0200, 0.0257, 0.0313, 0.0211, 0.0228, 0.0289], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:28:13,032 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28885.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:28:46,471 INFO [train.py:904] (6/8) Epoch 3, batch 8600, loss[loss=0.2347, simple_loss=0.3043, pruned_loss=0.08259, over 12503.00 frames. ], tot_loss[loss=0.2404, simple_loss=0.3205, pruned_loss=0.08019, over 3060514.35 frames. ], batch size: 248, lr: 1.92e-02, grad_scale: 4.0 2023-04-28 00:29:07,419 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28911.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 00:29:30,835 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0881, 3.8121, 4.1661, 4.3216, 4.4168, 3.9562, 4.4800, 4.3743], device='cuda:6'), covar=tensor([0.0763, 0.0725, 0.1022, 0.0529, 0.0456, 0.0566, 0.0322, 0.0432], device='cuda:6'), in_proj_covar=tensor([0.0277, 0.0345, 0.0438, 0.0338, 0.0257, 0.0244, 0.0281, 0.0281], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:29:30,889 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28922.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:29:47,745 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-28 00:29:50,787 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=28933.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:29:57,553 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.629e+02 3.755e+02 4.530e+02 5.746e+02 8.888e+02, threshold=9.061e+02, percent-clipped=0.0 2023-04-28 00:30:11,195 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28943.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:30:25,441 INFO [train.py:904] (6/8) Epoch 3, batch 8650, loss[loss=0.1957, simple_loss=0.2949, pruned_loss=0.04826, over 16219.00 frames. ], tot_loss[loss=0.2366, simple_loss=0.3179, pruned_loss=0.07766, over 3068408.40 frames. ], batch size: 165, lr: 1.92e-02, grad_scale: 4.0 2023-04-28 00:30:36,710 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3349, 4.5984, 4.2975, 4.4193, 4.0123, 3.9783, 4.2182, 4.5785], device='cuda:6'), covar=tensor([0.0523, 0.0644, 0.0885, 0.0339, 0.0560, 0.0892, 0.0503, 0.0629], device='cuda:6'), in_proj_covar=tensor([0.0260, 0.0367, 0.0312, 0.0238, 0.0234, 0.0241, 0.0288, 0.0254], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:30:45,826 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=28959.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 00:31:06,780 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7535, 3.3850, 2.9444, 4.6165, 4.3432, 4.3833, 1.5512, 3.3857], device='cuda:6'), covar=tensor([0.1404, 0.0384, 0.0841, 0.0051, 0.0149, 0.0209, 0.1312, 0.0485], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0134, 0.0162, 0.0071, 0.0134, 0.0143, 0.0152, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-04-28 00:31:08,586 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5207, 3.6523, 3.9859, 3.9249, 3.9516, 3.6165, 3.6837, 3.6997], device='cuda:6'), covar=tensor([0.0246, 0.0288, 0.0321, 0.0440, 0.0411, 0.0295, 0.0642, 0.0314], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0167, 0.0181, 0.0184, 0.0211, 0.0186, 0.0273, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-04-28 00:31:52,871 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=28991.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:32:12,025 INFO [train.py:904] (6/8) Epoch 3, batch 8700, loss[loss=0.2227, simple_loss=0.2994, pruned_loss=0.07297, over 17028.00 frames. ], tot_loss[loss=0.2325, simple_loss=0.3139, pruned_loss=0.07555, over 3061749.29 frames. ], batch size: 55, lr: 1.92e-02, grad_scale: 4.0 2023-04-28 00:32:15,184 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29002.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:33:18,499 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 3.379e+02 3.929e+02 4.901e+02 7.493e+02, threshold=7.858e+02, percent-clipped=0.0 2023-04-28 00:33:42,185 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-28 00:33:46,756 INFO [train.py:904] (6/8) Epoch 3, batch 8750, loss[loss=0.2283, simple_loss=0.2986, pruned_loss=0.07896, over 12061.00 frames. ], tot_loss[loss=0.2314, simple_loss=0.3132, pruned_loss=0.07482, over 3057390.53 frames. ], batch size: 248, lr: 1.92e-02, grad_scale: 4.0 2023-04-28 00:34:19,094 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29063.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:35:03,651 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 00:35:38,457 INFO [train.py:904] (6/8) Epoch 3, batch 8800, loss[loss=0.2194, simple_loss=0.3048, pruned_loss=0.06697, over 16246.00 frames. ], tot_loss[loss=0.2286, simple_loss=0.3111, pruned_loss=0.07311, over 3070975.71 frames. ], batch size: 165, lr: 1.92e-02, grad_scale: 8.0 2023-04-28 00:36:52,045 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.420e+02 3.673e+02 4.669e+02 5.868e+02 1.100e+03, threshold=9.339e+02, percent-clipped=8.0 2023-04-28 00:37:03,961 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29142.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:37:16,935 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29148.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:37:21,268 INFO [train.py:904] (6/8) Epoch 3, batch 8850, loss[loss=0.194, simple_loss=0.281, pruned_loss=0.05344, over 12509.00 frames. ], tot_loss[loss=0.2286, simple_loss=0.3126, pruned_loss=0.07233, over 3048035.17 frames. ], batch size: 249, lr: 1.92e-02, grad_scale: 8.0 2023-04-28 00:38:45,753 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=29190.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:38:57,574 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=29196.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:39:06,873 INFO [train.py:904] (6/8) Epoch 3, batch 8900, loss[loss=0.2241, simple_loss=0.3019, pruned_loss=0.07314, over 12750.00 frames. ], tot_loss[loss=0.2281, simple_loss=0.3133, pruned_loss=0.07142, over 3060481.22 frames. ], batch size: 250, lr: 1.91e-02, grad_scale: 8.0 2023-04-28 00:39:12,691 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9241, 5.4718, 5.4980, 5.2516, 5.3380, 5.8637, 5.5821, 5.2720], device='cuda:6'), covar=tensor([0.0528, 0.1080, 0.0931, 0.1453, 0.1864, 0.0682, 0.0819, 0.1674], device='cuda:6'), in_proj_covar=tensor([0.0220, 0.0302, 0.0288, 0.0267, 0.0344, 0.0315, 0.0253, 0.0347], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:39:40,986 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29217.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:39:41,213 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6149, 2.4543, 2.2258, 4.0584, 1.8756, 3.6310, 2.1721, 2.2188], device='cuda:6'), covar=tensor([0.0327, 0.0950, 0.0556, 0.0190, 0.2015, 0.0276, 0.1099, 0.1584], device='cuda:6'), in_proj_covar=tensor([0.0259, 0.0245, 0.0204, 0.0264, 0.0316, 0.0214, 0.0231, 0.0292], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:40:35,439 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29236.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:40:36,245 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.258e+02 3.766e+02 4.506e+02 5.745e+02 1.186e+03, threshold=9.012e+02, percent-clipped=1.0 2023-04-28 00:41:11,857 INFO [train.py:904] (6/8) Epoch 3, batch 8950, loss[loss=0.2411, simple_loss=0.3139, pruned_loss=0.08413, over 12696.00 frames. ], tot_loss[loss=0.2283, simple_loss=0.3131, pruned_loss=0.0718, over 3076639.42 frames. ], batch size: 248, lr: 1.91e-02, grad_scale: 8.0 2023-04-28 00:42:27,566 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2023-04-28 00:42:54,282 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29297.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:43:00,670 INFO [train.py:904] (6/8) Epoch 3, batch 9000, loss[loss=0.2178, simple_loss=0.3047, pruned_loss=0.06546, over 15367.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.309, pruned_loss=0.06965, over 3080638.43 frames. ], batch size: 190, lr: 1.91e-02, grad_scale: 8.0 2023-04-28 00:43:00,670 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 00:43:11,837 INFO [train.py:938] (6/8) Epoch 3, validation: loss=0.1886, simple_loss=0.2904, pruned_loss=0.04341, over 944034.00 frames. 2023-04-28 00:43:11,840 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-28 00:44:26,933 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.245e+02 3.595e+02 4.170e+02 5.160e+02 8.461e+02, threshold=8.339e+02, percent-clipped=0.0 2023-04-28 00:44:57,071 INFO [train.py:904] (6/8) Epoch 3, batch 9050, loss[loss=0.2304, simple_loss=0.3062, pruned_loss=0.07727, over 12856.00 frames. ], tot_loss[loss=0.2262, simple_loss=0.311, pruned_loss=0.07072, over 3092503.41 frames. ], batch size: 246, lr: 1.91e-02, grad_scale: 8.0 2023-04-28 00:45:12,879 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29358.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:46:41,627 INFO [train.py:904] (6/8) Epoch 3, batch 9100, loss[loss=0.235, simple_loss=0.3229, pruned_loss=0.07358, over 16163.00 frames. ], tot_loss[loss=0.2273, simple_loss=0.3111, pruned_loss=0.07176, over 3090334.78 frames. ], batch size: 165, lr: 1.91e-02, grad_scale: 8.0 2023-04-28 00:47:44,891 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 00:48:08,504 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.097e+02 4.371e+02 5.293e+02 6.986e+02 1.215e+03, threshold=1.059e+03, percent-clipped=12.0 2023-04-28 00:48:40,648 INFO [train.py:904] (6/8) Epoch 3, batch 9150, loss[loss=0.2088, simple_loss=0.297, pruned_loss=0.0603, over 16653.00 frames. ], tot_loss[loss=0.2264, simple_loss=0.3107, pruned_loss=0.07104, over 3072263.55 frames. ], batch size: 134, lr: 1.91e-02, grad_scale: 8.0 2023-04-28 00:49:15,046 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5474, 4.2891, 4.4632, 4.7207, 4.8456, 4.1996, 4.8991, 4.7497], device='cuda:6'), covar=tensor([0.0584, 0.0546, 0.0860, 0.0354, 0.0350, 0.0552, 0.0281, 0.0374], device='cuda:6'), in_proj_covar=tensor([0.0289, 0.0356, 0.0452, 0.0350, 0.0262, 0.0250, 0.0279, 0.0289], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:50:20,738 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9537, 3.2249, 3.4879, 3.4762, 3.4944, 3.2526, 3.2507, 3.3275], device='cuda:6'), covar=tensor([0.0316, 0.0349, 0.0447, 0.0459, 0.0392, 0.0327, 0.0779, 0.0349], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0167, 0.0176, 0.0179, 0.0209, 0.0185, 0.0272, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-04-28 00:50:24,885 INFO [train.py:904] (6/8) Epoch 3, batch 9200, loss[loss=0.2008, simple_loss=0.282, pruned_loss=0.05982, over 12561.00 frames. ], tot_loss[loss=0.2229, simple_loss=0.306, pruned_loss=0.06993, over 3064668.26 frames. ], batch size: 247, lr: 1.90e-02, grad_scale: 8.0 2023-04-28 00:50:27,522 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2353, 4.3352, 4.2823, 3.2098, 4.0427, 4.2530, 4.2890, 2.5995], device='cuda:6'), covar=tensor([0.0351, 0.0011, 0.0032, 0.0192, 0.0027, 0.0024, 0.0015, 0.0275], device='cuda:6'), in_proj_covar=tensor([0.0111, 0.0052, 0.0055, 0.0107, 0.0052, 0.0060, 0.0055, 0.0105], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 00:50:55,571 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29517.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:51:32,422 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.365e+02 4.043e+02 4.899e+02 6.737e+02 1.438e+03, threshold=9.797e+02, percent-clipped=1.0 2023-04-28 00:52:01,382 INFO [train.py:904] (6/8) Epoch 3, batch 9250, loss[loss=0.205, simple_loss=0.2797, pruned_loss=0.06508, over 12277.00 frames. ], tot_loss[loss=0.2232, simple_loss=0.3062, pruned_loss=0.07011, over 3070001.87 frames. ], batch size: 248, lr: 1.90e-02, grad_scale: 8.0 2023-04-28 00:52:02,555 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29551.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:52:08,638 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2739, 3.1787, 3.3093, 3.4775, 3.4744, 3.0811, 3.4667, 3.4764], device='cuda:6'), covar=tensor([0.0647, 0.0566, 0.0973, 0.0395, 0.0484, 0.1728, 0.0523, 0.0457], device='cuda:6'), in_proj_covar=tensor([0.0286, 0.0347, 0.0440, 0.0343, 0.0258, 0.0244, 0.0274, 0.0280], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:52:29,836 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=29565.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:53:31,915 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29592.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:53:50,312 INFO [train.py:904] (6/8) Epoch 3, batch 9300, loss[loss=0.1926, simple_loss=0.2681, pruned_loss=0.05849, over 12456.00 frames. ], tot_loss[loss=0.221, simple_loss=0.304, pruned_loss=0.06897, over 3061425.05 frames. ], batch size: 250, lr: 1.90e-02, grad_scale: 4.0 2023-04-28 00:54:15,980 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29612.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:55:10,763 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9034, 2.8362, 2.3774, 3.7982, 3.6830, 3.7701, 1.5737, 2.8852], device='cuda:6'), covar=tensor([0.1234, 0.0440, 0.1063, 0.0078, 0.0202, 0.0270, 0.1321, 0.0636], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0132, 0.0160, 0.0067, 0.0129, 0.0140, 0.0152, 0.0156], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-04-28 00:55:11,308 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.186e+02 3.534e+02 4.207e+02 4.979e+02 1.086e+03, threshold=8.414e+02, percent-clipped=1.0 2023-04-28 00:55:35,075 INFO [train.py:904] (6/8) Epoch 3, batch 9350, loss[loss=0.2218, simple_loss=0.2945, pruned_loss=0.07451, over 12493.00 frames. ], tot_loss[loss=0.2205, simple_loss=0.3038, pruned_loss=0.06863, over 3064722.38 frames. ], batch size: 248, lr: 1.90e-02, grad_scale: 4.0 2023-04-28 00:55:51,583 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29658.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:56:00,222 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29662.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:56:35,084 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.28 vs. limit=5.0 2023-04-28 00:56:40,520 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7073, 2.6258, 2.5995, 1.9140, 2.4663, 2.5215, 2.6390, 1.9078], device='cuda:6'), covar=tensor([0.0274, 0.0027, 0.0037, 0.0199, 0.0043, 0.0040, 0.0030, 0.0258], device='cuda:6'), in_proj_covar=tensor([0.0113, 0.0052, 0.0056, 0.0110, 0.0054, 0.0060, 0.0057, 0.0107], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 00:57:16,961 INFO [train.py:904] (6/8) Epoch 3, batch 9400, loss[loss=0.2059, simple_loss=0.3065, pruned_loss=0.05261, over 16713.00 frames. ], tot_loss[loss=0.2207, simple_loss=0.3038, pruned_loss=0.06882, over 3051785.81 frames. ], batch size: 83, lr: 1.90e-02, grad_scale: 4.0 2023-04-28 00:57:26,375 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4085, 3.4087, 2.8778, 2.1448, 2.4717, 2.0359, 3.5109, 3.7117], device='cuda:6'), covar=tensor([0.1941, 0.0656, 0.0934, 0.1242, 0.1569, 0.1275, 0.0318, 0.0444], device='cuda:6'), in_proj_covar=tensor([0.0259, 0.0239, 0.0251, 0.0210, 0.0229, 0.0191, 0.0213, 0.0201], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 00:57:27,970 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=29706.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:57:46,900 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9774, 3.1316, 3.5321, 3.5032, 3.5142, 3.2144, 3.3567, 3.3006], device='cuda:6'), covar=tensor([0.0255, 0.0407, 0.0367, 0.0421, 0.0388, 0.0302, 0.0572, 0.0303], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0161, 0.0167, 0.0173, 0.0200, 0.0180, 0.0256, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-04-28 00:58:02,161 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29723.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 00:58:32,711 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.403e+02 3.993e+02 5.001e+02 6.201e+02 1.324e+03, threshold=1.000e+03, percent-clipped=7.0 2023-04-28 00:58:58,179 INFO [train.py:904] (6/8) Epoch 3, batch 9450, loss[loss=0.219, simple_loss=0.3028, pruned_loss=0.06766, over 16304.00 frames. ], tot_loss[loss=0.2231, simple_loss=0.3059, pruned_loss=0.07016, over 3024194.81 frames. ], batch size: 146, lr: 1.90e-02, grad_scale: 4.0 2023-04-28 01:00:37,284 INFO [train.py:904] (6/8) Epoch 3, batch 9500, loss[loss=0.203, simple_loss=0.2975, pruned_loss=0.05424, over 16144.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.3047, pruned_loss=0.06903, over 3035239.81 frames. ], batch size: 165, lr: 1.90e-02, grad_scale: 4.0 2023-04-28 01:00:41,456 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29802.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:01:15,261 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9247, 1.1293, 1.6505, 1.6006, 1.6303, 1.7738, 1.2984, 1.7642], device='cuda:6'), covar=tensor([0.0060, 0.0166, 0.0096, 0.0115, 0.0083, 0.0061, 0.0146, 0.0045], device='cuda:6'), in_proj_covar=tensor([0.0093, 0.0127, 0.0112, 0.0103, 0.0101, 0.0073, 0.0118, 0.0065], device='cuda:6'), out_proj_covar=tensor([1.3806e-04, 1.9509e-04, 1.7565e-04, 1.6041e-04, 1.5321e-04, 1.0848e-04, 1.7855e-04, 9.7551e-05], device='cuda:6') 2023-04-28 01:01:32,310 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9344, 3.2516, 3.1995, 1.5692, 3.3588, 3.3674, 2.8997, 2.7055], device='cuda:6'), covar=tensor([0.0794, 0.0089, 0.0118, 0.1297, 0.0059, 0.0063, 0.0285, 0.0352], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0078, 0.0075, 0.0146, 0.0070, 0.0070, 0.0109, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 01:01:50,890 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.435e+02 3.682e+02 4.369e+02 5.544e+02 1.182e+03, threshold=8.738e+02, percent-clipped=2.0 2023-04-28 01:02:22,322 INFO [train.py:904] (6/8) Epoch 3, batch 9550, loss[loss=0.2438, simple_loss=0.3281, pruned_loss=0.07974, over 16935.00 frames. ], tot_loss[loss=0.2218, simple_loss=0.3051, pruned_loss=0.06931, over 3043675.90 frames. ], batch size: 109, lr: 1.89e-02, grad_scale: 4.0 2023-04-28 01:02:39,023 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7857, 1.1206, 1.5575, 1.6367, 1.7058, 1.7584, 1.2816, 1.7643], device='cuda:6'), covar=tensor([0.0066, 0.0179, 0.0092, 0.0104, 0.0079, 0.0060, 0.0156, 0.0044], device='cuda:6'), in_proj_covar=tensor([0.0092, 0.0125, 0.0110, 0.0102, 0.0099, 0.0072, 0.0117, 0.0063], device='cuda:6'), out_proj_covar=tensor([1.3741e-04, 1.9219e-04, 1.7218e-04, 1.5771e-04, 1.5019e-04, 1.0693e-04, 1.7705e-04, 9.5307e-05], device='cuda:6') 2023-04-28 01:02:49,750 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29863.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:03:49,018 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5303, 3.5328, 3.3448, 3.3598, 3.1280, 3.4394, 3.2528, 3.2374], device='cuda:6'), covar=tensor([0.0248, 0.0196, 0.0170, 0.0140, 0.0507, 0.0202, 0.0756, 0.0287], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0124, 0.0164, 0.0134, 0.0190, 0.0152, 0.0117, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 01:03:49,028 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29892.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:04:04,157 INFO [train.py:904] (6/8) Epoch 3, batch 9600, loss[loss=0.2371, simple_loss=0.3204, pruned_loss=0.0769, over 16979.00 frames. ], tot_loss[loss=0.2233, simple_loss=0.3063, pruned_loss=0.07019, over 3043075.10 frames. ], batch size: 109, lr: 1.89e-02, grad_scale: 8.0 2023-04-28 01:04:16,871 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29907.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:04:43,333 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0521, 2.5182, 2.4497, 3.3494, 3.1383, 3.2790, 1.8482, 2.8575], device='cuda:6'), covar=tensor([0.1112, 0.0410, 0.0873, 0.0084, 0.0183, 0.0373, 0.1153, 0.0585], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0134, 0.0162, 0.0068, 0.0131, 0.0144, 0.0154, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-04-28 01:05:18,578 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.324e+02 3.589e+02 4.321e+02 5.133e+02 1.002e+03, threshold=8.641e+02, percent-clipped=3.0 2023-04-28 01:05:23,702 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=29940.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:05:51,206 INFO [train.py:904] (6/8) Epoch 3, batch 9650, loss[loss=0.253, simple_loss=0.3309, pruned_loss=0.08748, over 15339.00 frames. ], tot_loss[loss=0.2251, simple_loss=0.3085, pruned_loss=0.07084, over 3030305.26 frames. ], batch size: 190, lr: 1.89e-02, grad_scale: 8.0 2023-04-28 01:07:41,652 INFO [train.py:904] (6/8) Epoch 3, batch 9700, loss[loss=0.217, simple_loss=0.3003, pruned_loss=0.06681, over 15389.00 frames. ], tot_loss[loss=0.2244, simple_loss=0.3076, pruned_loss=0.07058, over 3034770.51 frames. ], batch size: 191, lr: 1.89e-02, grad_scale: 8.0 2023-04-28 01:08:16,060 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30018.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 01:08:59,978 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.947e+02 3.446e+02 4.381e+02 5.437e+02 9.929e+02, threshold=8.763e+02, percent-clipped=3.0 2023-04-28 01:09:24,291 INFO [train.py:904] (6/8) Epoch 3, batch 9750, loss[loss=0.2, simple_loss=0.2933, pruned_loss=0.05334, over 16960.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.3066, pruned_loss=0.07077, over 3038916.22 frames. ], batch size: 109, lr: 1.89e-02, grad_scale: 8.0 2023-04-28 01:11:02,181 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0288, 3.9205, 4.1136, 4.3304, 4.4111, 3.8742, 4.4160, 4.4318], device='cuda:6'), covar=tensor([0.0664, 0.0547, 0.0915, 0.0376, 0.0386, 0.0762, 0.0359, 0.0288], device='cuda:6'), in_proj_covar=tensor([0.0280, 0.0339, 0.0436, 0.0338, 0.0257, 0.0241, 0.0272, 0.0277], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 01:11:02,931 INFO [train.py:904] (6/8) Epoch 3, batch 9800, loss[loss=0.2098, simple_loss=0.3066, pruned_loss=0.05647, over 16738.00 frames. ], tot_loss[loss=0.2228, simple_loss=0.3068, pruned_loss=0.06937, over 3058887.69 frames. ], batch size: 134, lr: 1.89e-02, grad_scale: 8.0 2023-04-28 01:11:22,432 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-28 01:11:48,348 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30125.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:12:14,773 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.212e+02 3.552e+02 4.452e+02 5.972e+02 1.192e+03, threshold=8.904e+02, percent-clipped=4.0 2023-04-28 01:12:47,470 INFO [train.py:904] (6/8) Epoch 3, batch 9850, loss[loss=0.2024, simple_loss=0.2941, pruned_loss=0.05541, over 16845.00 frames. ], tot_loss[loss=0.2224, simple_loss=0.3071, pruned_loss=0.06884, over 3052390.86 frames. ], batch size: 90, lr: 1.88e-02, grad_scale: 8.0 2023-04-28 01:13:02,424 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30158.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:13:35,553 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-28 01:14:04,069 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8923, 3.8536, 3.7705, 3.3434, 3.7332, 1.6683, 3.5214, 3.7660], device='cuda:6'), covar=tensor([0.0069, 0.0053, 0.0082, 0.0195, 0.0064, 0.1482, 0.0082, 0.0109], device='cuda:6'), in_proj_covar=tensor([0.0069, 0.0059, 0.0090, 0.0091, 0.0065, 0.0118, 0.0079, 0.0087], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 01:14:06,716 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30186.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:14:39,079 INFO [train.py:904] (6/8) Epoch 3, batch 9900, loss[loss=0.2149, simple_loss=0.2914, pruned_loss=0.06917, over 12543.00 frames. ], tot_loss[loss=0.2223, simple_loss=0.3069, pruned_loss=0.06878, over 3025329.81 frames. ], batch size: 247, lr: 1.88e-02, grad_scale: 8.0 2023-04-28 01:14:54,574 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30207.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:14:59,660 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30209.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:16:05,826 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.521e+02 3.778e+02 4.995e+02 6.247e+02 1.716e+03, threshold=9.990e+02, percent-clipped=10.0 2023-04-28 01:16:35,560 INFO [train.py:904] (6/8) Epoch 3, batch 9950, loss[loss=0.2368, simple_loss=0.326, pruned_loss=0.07378, over 16273.00 frames. ], tot_loss[loss=0.2244, simple_loss=0.31, pruned_loss=0.06939, over 3051114.06 frames. ], batch size: 166, lr: 1.88e-02, grad_scale: 8.0 2023-04-28 01:16:47,149 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=30255.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:17:23,105 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30270.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:18:21,498 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-28 01:18:37,068 INFO [train.py:904] (6/8) Epoch 3, batch 10000, loss[loss=0.2205, simple_loss=0.3126, pruned_loss=0.06422, over 16786.00 frames. ], tot_loss[loss=0.2228, simple_loss=0.308, pruned_loss=0.06876, over 3054311.42 frames. ], batch size: 124, lr: 1.88e-02, grad_scale: 8.0 2023-04-28 01:19:12,405 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30318.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 01:19:55,680 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.113e+02 3.684e+02 4.714e+02 5.759e+02 1.067e+03, threshold=9.428e+02, percent-clipped=2.0 2023-04-28 01:20:19,869 INFO [train.py:904] (6/8) Epoch 3, batch 10050, loss[loss=0.2386, simple_loss=0.3255, pruned_loss=0.07587, over 12166.00 frames. ], tot_loss[loss=0.2218, simple_loss=0.3074, pruned_loss=0.06811, over 3052670.13 frames. ], batch size: 247, lr: 1.88e-02, grad_scale: 8.0 2023-04-28 01:20:38,369 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5015, 3.3761, 3.4224, 3.0542, 3.4194, 1.8353, 3.2136, 3.2071], device='cuda:6'), covar=tensor([0.0069, 0.0060, 0.0082, 0.0156, 0.0058, 0.1356, 0.0085, 0.0106], device='cuda:6'), in_proj_covar=tensor([0.0071, 0.0062, 0.0093, 0.0093, 0.0068, 0.0122, 0.0082, 0.0091], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 01:20:38,392 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30360.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:20:40,280 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4018, 1.7008, 1.5694, 1.7174, 2.2502, 2.0126, 2.3898, 2.4979], device='cuda:6'), covar=tensor([0.0021, 0.0188, 0.0176, 0.0180, 0.0108, 0.0143, 0.0050, 0.0079], device='cuda:6'), in_proj_covar=tensor([0.0054, 0.0127, 0.0124, 0.0123, 0.0114, 0.0122, 0.0077, 0.0097], device='cuda:6'), out_proj_covar=tensor([6.7489e-05, 1.7546e-04, 1.6568e-04, 1.6698e-04, 1.5852e-04, 1.6871e-04, 1.0342e-04, 1.3318e-04], device='cuda:6') 2023-04-28 01:20:50,750 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=30366.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:21:15,904 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0651, 4.1438, 3.9276, 3.9877, 3.5707, 4.0332, 3.8315, 3.7938], device='cuda:6'), covar=tensor([0.0370, 0.0182, 0.0184, 0.0130, 0.0658, 0.0236, 0.0383, 0.0331], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0124, 0.0163, 0.0135, 0.0189, 0.0152, 0.0116, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 01:21:19,674 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-28 01:21:54,399 INFO [train.py:904] (6/8) Epoch 3, batch 10100, loss[loss=0.2123, simple_loss=0.2936, pruned_loss=0.06548, over 16184.00 frames. ], tot_loss[loss=0.2232, simple_loss=0.3084, pruned_loss=0.06897, over 3068962.81 frames. ], batch size: 165, lr: 1.88e-02, grad_scale: 8.0 2023-04-28 01:22:06,126 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8963, 3.8650, 3.8334, 3.3663, 3.7621, 1.7553, 3.6010, 3.8010], device='cuda:6'), covar=tensor([0.0076, 0.0071, 0.0087, 0.0202, 0.0075, 0.1443, 0.0094, 0.0119], device='cuda:6'), in_proj_covar=tensor([0.0068, 0.0059, 0.0091, 0.0089, 0.0066, 0.0117, 0.0080, 0.0088], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 01:22:37,260 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30421.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:23:00,421 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.870e+02 3.919e+02 4.872e+02 6.072e+02 1.469e+03, threshold=9.744e+02, percent-clipped=6.0 2023-04-28 01:23:38,627 INFO [train.py:904] (6/8) Epoch 4, batch 0, loss[loss=0.2913, simple_loss=0.3653, pruned_loss=0.1087, over 17067.00 frames. ], tot_loss[loss=0.2913, simple_loss=0.3653, pruned_loss=0.1087, over 17067.00 frames. ], batch size: 50, lr: 1.75e-02, grad_scale: 8.0 2023-04-28 01:23:38,627 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 01:23:46,515 INFO [train.py:938] (6/8) Epoch 4, validation: loss=0.188, simple_loss=0.2904, pruned_loss=0.04284, over 944034.00 frames. 2023-04-28 01:23:46,516 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-28 01:23:57,007 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8020, 4.5710, 4.1441, 2.0692, 3.2803, 2.5875, 3.9887, 4.4855], device='cuda:6'), covar=tensor([0.0195, 0.0349, 0.0454, 0.1493, 0.0606, 0.0976, 0.0636, 0.0506], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0109, 0.0152, 0.0143, 0.0133, 0.0129, 0.0136, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 01:23:58,017 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30458.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:24:12,727 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6270, 4.0845, 4.0559, 2.0738, 4.2935, 4.2727, 3.2825, 3.2921], device='cuda:6'), covar=tensor([0.0711, 0.0090, 0.0110, 0.1132, 0.0036, 0.0043, 0.0305, 0.0344], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0081, 0.0079, 0.0147, 0.0072, 0.0072, 0.0112, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 01:24:18,653 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-28 01:24:29,162 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30481.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:24:56,021 INFO [train.py:904] (6/8) Epoch 4, batch 50, loss[loss=0.2324, simple_loss=0.3049, pruned_loss=0.07998, over 15990.00 frames. ], tot_loss[loss=0.2687, simple_loss=0.3317, pruned_loss=0.1028, over 749823.65 frames. ], batch size: 35, lr: 1.75e-02, grad_scale: 1.0 2023-04-28 01:25:02,245 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=30506.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:25:49,219 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2595, 4.4559, 2.3130, 4.8434, 2.8158, 4.6878, 2.3258, 3.2904], device='cuda:6'), covar=tensor([0.0102, 0.0194, 0.1403, 0.0021, 0.0747, 0.0268, 0.1343, 0.0520], device='cuda:6'), in_proj_covar=tensor([0.0098, 0.0138, 0.0173, 0.0081, 0.0160, 0.0162, 0.0181, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 01:25:49,876 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.617e+02 4.176e+02 5.221e+02 6.159e+02 1.340e+03, threshold=1.044e+03, percent-clipped=3.0 2023-04-28 01:26:02,162 INFO [train.py:904] (6/8) Epoch 4, batch 100, loss[loss=0.2744, simple_loss=0.3181, pruned_loss=0.1154, over 16774.00 frames. ], tot_loss[loss=0.255, simple_loss=0.3227, pruned_loss=0.0936, over 1325077.06 frames. ], batch size: 83, lr: 1.75e-02, grad_scale: 1.0 2023-04-28 01:26:23,715 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30565.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:27:11,715 INFO [train.py:904] (6/8) Epoch 4, batch 150, loss[loss=0.2548, simple_loss=0.3025, pruned_loss=0.1036, over 16905.00 frames. ], tot_loss[loss=0.2531, simple_loss=0.3209, pruned_loss=0.09263, over 1773094.07 frames. ], batch size: 116, lr: 1.75e-02, grad_scale: 1.0 2023-04-28 01:28:04,888 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.524e+02 3.830e+02 4.808e+02 5.813e+02 1.621e+03, threshold=9.616e+02, percent-clipped=3.0 2023-04-28 01:28:09,242 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3876, 3.8178, 3.7688, 1.6897, 3.8783, 3.9180, 3.1227, 3.0275], device='cuda:6'), covar=tensor([0.0775, 0.0077, 0.0109, 0.1179, 0.0050, 0.0053, 0.0297, 0.0373], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0081, 0.0080, 0.0146, 0.0073, 0.0073, 0.0111, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 01:28:18,514 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3999, 1.8807, 2.2654, 3.1155, 3.0350, 3.4737, 1.7793, 3.2336], device='cuda:6'), covar=tensor([0.0048, 0.0179, 0.0143, 0.0088, 0.0068, 0.0057, 0.0183, 0.0049], device='cuda:6'), in_proj_covar=tensor([0.0095, 0.0128, 0.0113, 0.0105, 0.0102, 0.0077, 0.0118, 0.0065], device='cuda:6'), out_proj_covar=tensor([1.4124e-04, 1.9372e-04, 1.7578e-04, 1.6137e-04, 1.5294e-04, 1.1298e-04, 1.7657e-04, 9.7277e-05], device='cuda:6') 2023-04-28 01:28:19,173 INFO [train.py:904] (6/8) Epoch 4, batch 200, loss[loss=0.1982, simple_loss=0.2782, pruned_loss=0.05914, over 16978.00 frames. ], tot_loss[loss=0.2545, simple_loss=0.322, pruned_loss=0.09345, over 2103026.21 frames. ], batch size: 41, lr: 1.75e-02, grad_scale: 1.0 2023-04-28 01:28:23,995 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30655.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:29:26,978 INFO [train.py:904] (6/8) Epoch 4, batch 250, loss[loss=0.2119, simple_loss=0.293, pruned_loss=0.06546, over 17238.00 frames. ], tot_loss[loss=0.2505, simple_loss=0.3176, pruned_loss=0.09168, over 2374040.72 frames. ], batch size: 45, lr: 1.75e-02, grad_scale: 1.0 2023-04-28 01:29:48,571 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30716.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:29:48,714 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30716.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 01:29:52,394 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7668, 3.4677, 2.9682, 1.8352, 2.5305, 1.9367, 3.3775, 3.4216], device='cuda:6'), covar=tensor([0.0199, 0.0417, 0.0458, 0.1539, 0.0735, 0.1024, 0.0497, 0.0621], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0118, 0.0154, 0.0146, 0.0136, 0.0130, 0.0139, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-28 01:30:21,642 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.145e+02 3.686e+02 4.713e+02 5.844e+02 8.594e+02, threshold=9.427e+02, percent-clipped=0.0 2023-04-28 01:30:36,156 INFO [train.py:904] (6/8) Epoch 4, batch 300, loss[loss=0.2123, simple_loss=0.2915, pruned_loss=0.06654, over 16827.00 frames. ], tot_loss[loss=0.2453, simple_loss=0.313, pruned_loss=0.08875, over 2587152.95 frames. ], batch size: 42, lr: 1.75e-02, grad_scale: 1.0 2023-04-28 01:31:16,694 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30781.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:31:43,552 INFO [train.py:904] (6/8) Epoch 4, batch 350, loss[loss=0.2451, simple_loss=0.3067, pruned_loss=0.09178, over 16323.00 frames. ], tot_loss[loss=0.2397, simple_loss=0.3084, pruned_loss=0.08553, over 2745284.58 frames. ], batch size: 165, lr: 1.74e-02, grad_scale: 1.0 2023-04-28 01:31:56,507 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6838, 3.7203, 1.8159, 4.0031, 2.6099, 3.9090, 1.8875, 2.9570], device='cuda:6'), covar=tensor([0.0072, 0.0203, 0.1387, 0.0057, 0.0651, 0.0280, 0.1154, 0.0452], device='cuda:6'), in_proj_covar=tensor([0.0097, 0.0140, 0.0170, 0.0080, 0.0157, 0.0165, 0.0176, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-28 01:32:20,667 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=30829.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:32:38,673 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.414e+02 3.784e+02 4.816e+02 5.903e+02 1.455e+03, threshold=9.632e+02, percent-clipped=4.0 2023-04-28 01:32:51,066 INFO [train.py:904] (6/8) Epoch 4, batch 400, loss[loss=0.2011, simple_loss=0.2856, pruned_loss=0.05826, over 17241.00 frames. ], tot_loss[loss=0.2383, simple_loss=0.3073, pruned_loss=0.08464, over 2872026.03 frames. ], batch size: 45, lr: 1.74e-02, grad_scale: 2.0 2023-04-28 01:33:11,854 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30865.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:33:16,986 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5077, 2.1501, 1.5934, 1.8679, 2.5748, 2.5298, 2.7076, 2.7390], device='cuda:6'), covar=tensor([0.0046, 0.0153, 0.0189, 0.0187, 0.0081, 0.0118, 0.0082, 0.0077], device='cuda:6'), in_proj_covar=tensor([0.0062, 0.0131, 0.0127, 0.0129, 0.0119, 0.0129, 0.0090, 0.0106], device='cuda:6'), out_proj_covar=tensor([7.7868e-05, 1.7953e-04, 1.6869e-04, 1.7540e-04, 1.6528e-04, 1.7781e-04, 1.2089e-04, 1.4619e-04], device='cuda:6') 2023-04-28 01:33:20,203 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-28 01:34:01,538 INFO [train.py:904] (6/8) Epoch 4, batch 450, loss[loss=0.2347, simple_loss=0.3121, pruned_loss=0.07868, over 17006.00 frames. ], tot_loss[loss=0.2354, simple_loss=0.3052, pruned_loss=0.08277, over 2981518.45 frames. ], batch size: 55, lr: 1.74e-02, grad_scale: 2.0 2023-04-28 01:34:05,019 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30903.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:34:18,179 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=30913.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:34:35,381 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30925.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:34:39,012 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4153, 4.3392, 3.6725, 1.9404, 2.9923, 2.2063, 3.7330, 4.0302], device='cuda:6'), covar=tensor([0.0252, 0.0407, 0.0471, 0.1614, 0.0693, 0.1065, 0.0633, 0.0828], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0123, 0.0156, 0.0146, 0.0136, 0.0130, 0.0140, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 01:34:46,500 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30934.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:34:56,469 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.382e+02 3.455e+02 4.330e+02 5.569e+02 1.727e+03, threshold=8.660e+02, percent-clipped=4.0 2023-04-28 01:35:01,025 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0549, 3.2296, 3.1373, 1.5816, 3.2848, 3.3301, 2.6926, 2.5386], device='cuda:6'), covar=tensor([0.0793, 0.0113, 0.0181, 0.1209, 0.0075, 0.0085, 0.0401, 0.0429], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0082, 0.0081, 0.0145, 0.0074, 0.0075, 0.0113, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 01:35:09,305 INFO [train.py:904] (6/8) Epoch 4, batch 500, loss[loss=0.2104, simple_loss=0.288, pruned_loss=0.06642, over 17229.00 frames. ], tot_loss[loss=0.2338, simple_loss=0.3038, pruned_loss=0.08193, over 3054879.16 frames. ], batch size: 44, lr: 1.74e-02, grad_scale: 2.0 2023-04-28 01:35:13,955 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2023-04-28 01:35:28,309 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30964.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:35:59,017 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30986.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:36:03,719 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6372, 2.4550, 2.2716, 4.0713, 1.8930, 3.4997, 2.2063, 2.3508], device='cuda:6'), covar=tensor([0.0421, 0.1109, 0.0654, 0.0318, 0.2082, 0.0439, 0.1347, 0.1628], device='cuda:6'), in_proj_covar=tensor([0.0279, 0.0267, 0.0218, 0.0282, 0.0329, 0.0236, 0.0246, 0.0333], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 01:36:09,639 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30995.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:36:17,824 INFO [train.py:904] (6/8) Epoch 4, batch 550, loss[loss=0.2136, simple_loss=0.3012, pruned_loss=0.06301, over 17059.00 frames. ], tot_loss[loss=0.2333, simple_loss=0.3032, pruned_loss=0.08168, over 3112217.24 frames. ], batch size: 50, lr: 1.74e-02, grad_scale: 2.0 2023-04-28 01:36:33,158 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31011.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 01:36:33,348 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31011.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 01:36:39,296 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31016.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:37:13,481 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.455e+02 3.696e+02 4.486e+02 5.421e+02 1.042e+03, threshold=8.971e+02, percent-clipped=3.0 2023-04-28 01:37:28,709 INFO [train.py:904] (6/8) Epoch 4, batch 600, loss[loss=0.218, simple_loss=0.3003, pruned_loss=0.06782, over 17006.00 frames. ], tot_loss[loss=0.2335, simple_loss=0.3028, pruned_loss=0.08215, over 3160580.33 frames. ], batch size: 55, lr: 1.74e-02, grad_scale: 2.0 2023-04-28 01:37:46,829 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=31064.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:37:58,262 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31072.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 01:38:36,813 INFO [train.py:904] (6/8) Epoch 4, batch 650, loss[loss=0.2506, simple_loss=0.3116, pruned_loss=0.09479, over 16676.00 frames. ], tot_loss[loss=0.2324, simple_loss=0.302, pruned_loss=0.0814, over 3194321.78 frames. ], batch size: 89, lr: 1.74e-02, grad_scale: 2.0 2023-04-28 01:39:13,290 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-28 01:39:30,533 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.918e+02 3.618e+02 4.311e+02 5.742e+02 1.399e+03, threshold=8.622e+02, percent-clipped=6.0 2023-04-28 01:39:45,330 INFO [train.py:904] (6/8) Epoch 4, batch 700, loss[loss=0.2323, simple_loss=0.2953, pruned_loss=0.08472, over 16733.00 frames. ], tot_loss[loss=0.2308, simple_loss=0.3011, pruned_loss=0.08024, over 3229969.79 frames. ], batch size: 124, lr: 1.73e-02, grad_scale: 2.0 2023-04-28 01:40:04,660 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-04-28 01:40:53,738 INFO [train.py:904] (6/8) Epoch 4, batch 750, loss[loss=0.2421, simple_loss=0.3225, pruned_loss=0.08086, over 16648.00 frames. ], tot_loss[loss=0.2303, simple_loss=0.3015, pruned_loss=0.07956, over 3244191.93 frames. ], batch size: 62, lr: 1.73e-02, grad_scale: 2.0 2023-04-28 01:41:36,122 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9777, 4.0427, 3.3597, 2.5718, 3.0999, 2.3225, 4.3968, 4.4131], device='cuda:6'), covar=tensor([0.1850, 0.0633, 0.1091, 0.1294, 0.2041, 0.1340, 0.0310, 0.0450], device='cuda:6'), in_proj_covar=tensor([0.0270, 0.0248, 0.0260, 0.0222, 0.0290, 0.0200, 0.0221, 0.0234], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 01:41:43,306 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.04 vs. limit=5.0 2023-04-28 01:41:48,308 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.232e+02 3.368e+02 4.224e+02 5.033e+02 8.537e+02, threshold=8.448e+02, percent-clipped=0.0 2023-04-28 01:42:01,619 INFO [train.py:904] (6/8) Epoch 4, batch 800, loss[loss=0.2007, simple_loss=0.2857, pruned_loss=0.05787, over 17270.00 frames. ], tot_loss[loss=0.2294, simple_loss=0.3012, pruned_loss=0.07875, over 3259979.30 frames. ], batch size: 52, lr: 1.73e-02, grad_scale: 4.0 2023-04-28 01:42:14,289 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31259.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:42:44,446 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31281.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:42:56,253 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31290.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:43:11,602 INFO [train.py:904] (6/8) Epoch 4, batch 850, loss[loss=0.1938, simple_loss=0.2673, pruned_loss=0.06021, over 16760.00 frames. ], tot_loss[loss=0.2296, simple_loss=0.3008, pruned_loss=0.07924, over 3276710.59 frames. ], batch size: 124, lr: 1.73e-02, grad_scale: 4.0 2023-04-28 01:43:24,656 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31311.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:43:57,944 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-28 01:44:07,361 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.752e+02 3.699e+02 4.562e+02 5.706e+02 1.340e+03, threshold=9.123e+02, percent-clipped=5.0 2023-04-28 01:44:19,702 INFO [train.py:904] (6/8) Epoch 4, batch 900, loss[loss=0.2426, simple_loss=0.3076, pruned_loss=0.08881, over 16883.00 frames. ], tot_loss[loss=0.2275, simple_loss=0.2992, pruned_loss=0.0779, over 3288739.62 frames. ], batch size: 116, lr: 1.73e-02, grad_scale: 4.0 2023-04-28 01:44:32,831 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=31359.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:44:42,770 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31367.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 01:45:31,310 INFO [train.py:904] (6/8) Epoch 4, batch 950, loss[loss=0.2495, simple_loss=0.3067, pruned_loss=0.0961, over 16775.00 frames. ], tot_loss[loss=0.2266, simple_loss=0.2984, pruned_loss=0.07743, over 3291902.33 frames. ], batch size: 134, lr: 1.73e-02, grad_scale: 4.0 2023-04-28 01:45:41,797 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31409.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:45:55,548 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8542, 4.6164, 4.7698, 5.1396, 5.1945, 4.6343, 5.2535, 5.1549], device='cuda:6'), covar=tensor([0.0613, 0.0646, 0.1283, 0.0407, 0.0388, 0.0501, 0.0351, 0.0342], device='cuda:6'), in_proj_covar=tensor([0.0345, 0.0419, 0.0566, 0.0431, 0.0328, 0.0306, 0.0338, 0.0346], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 01:45:57,950 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-28 01:46:26,066 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.766e+02 3.401e+02 4.040e+02 4.698e+02 9.697e+02, threshold=8.080e+02, percent-clipped=2.0 2023-04-28 01:46:27,167 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3262, 2.1459, 1.5951, 1.8748, 2.5252, 2.4484, 2.6525, 2.7418], device='cuda:6'), covar=tensor([0.0048, 0.0153, 0.0188, 0.0188, 0.0082, 0.0118, 0.0067, 0.0093], device='cuda:6'), in_proj_covar=tensor([0.0069, 0.0135, 0.0133, 0.0132, 0.0126, 0.0135, 0.0097, 0.0114], device='cuda:6'), out_proj_covar=tensor([8.7820e-05, 1.8491e-04, 1.7500e-04, 1.7721e-04, 1.7305e-04, 1.8493e-04, 1.3042e-04, 1.5602e-04], device='cuda:6') 2023-04-28 01:46:34,622 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9590, 5.0450, 5.4848, 5.4231, 5.3521, 4.9790, 4.9961, 4.7096], device='cuda:6'), covar=tensor([0.0241, 0.0251, 0.0275, 0.0385, 0.0375, 0.0264, 0.0607, 0.0308], device='cuda:6'), in_proj_covar=tensor([0.0212, 0.0214, 0.0219, 0.0216, 0.0258, 0.0231, 0.0321, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 01:46:38,675 INFO [train.py:904] (6/8) Epoch 4, batch 1000, loss[loss=0.244, simple_loss=0.2989, pruned_loss=0.09457, over 16890.00 frames. ], tot_loss[loss=0.2259, simple_loss=0.2968, pruned_loss=0.07754, over 3305943.31 frames. ], batch size: 90, lr: 1.73e-02, grad_scale: 4.0 2023-04-28 01:46:50,275 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9264, 4.1949, 4.4721, 3.4876, 4.1979, 4.4503, 4.2744, 2.3287], device='cuda:6'), covar=tensor([0.0232, 0.0020, 0.0018, 0.0147, 0.0025, 0.0031, 0.0018, 0.0263], device='cuda:6'), in_proj_covar=tensor([0.0106, 0.0055, 0.0056, 0.0104, 0.0056, 0.0060, 0.0055, 0.0099], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 01:46:54,143 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4441, 4.2901, 4.0051, 1.9953, 3.0209, 2.3692, 3.7901, 4.1631], device='cuda:6'), covar=tensor([0.0316, 0.0481, 0.0400, 0.1584, 0.0708, 0.0962, 0.0706, 0.0792], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0127, 0.0155, 0.0146, 0.0137, 0.0130, 0.0142, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 01:47:06,234 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31470.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:47:36,273 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31491.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 01:47:39,952 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-28 01:47:48,995 INFO [train.py:904] (6/8) Epoch 4, batch 1050, loss[loss=0.2725, simple_loss=0.3202, pruned_loss=0.1125, over 16702.00 frames. ], tot_loss[loss=0.2253, simple_loss=0.297, pruned_loss=0.07683, over 3311375.47 frames. ], batch size: 124, lr: 1.72e-02, grad_scale: 4.0 2023-04-28 01:48:45,339 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.288e+02 3.437e+02 4.188e+02 5.032e+02 1.579e+03, threshold=8.377e+02, percent-clipped=2.0 2023-04-28 01:49:00,844 INFO [train.py:904] (6/8) Epoch 4, batch 1100, loss[loss=0.2294, simple_loss=0.3079, pruned_loss=0.07539, over 17018.00 frames. ], tot_loss[loss=0.2245, simple_loss=0.2963, pruned_loss=0.07641, over 3305738.79 frames. ], batch size: 55, lr: 1.72e-02, grad_scale: 4.0 2023-04-28 01:49:02,431 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31552.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 01:49:11,912 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31559.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:49:17,307 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5274, 4.3716, 4.4276, 3.7122, 4.3799, 1.9004, 4.1594, 4.3249], device='cuda:6'), covar=tensor([0.0069, 0.0058, 0.0077, 0.0291, 0.0056, 0.1345, 0.0085, 0.0108], device='cuda:6'), in_proj_covar=tensor([0.0084, 0.0076, 0.0114, 0.0123, 0.0082, 0.0130, 0.0099, 0.0112], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 01:49:24,332 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 01:49:41,928 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31581.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:49:55,718 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31590.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:49:55,908 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8966, 3.4475, 2.9807, 5.1737, 4.9265, 4.3700, 1.8628, 3.1004], device='cuda:6'), covar=tensor([0.1302, 0.0483, 0.1065, 0.0077, 0.0217, 0.0309, 0.1296, 0.0751], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0135, 0.0165, 0.0075, 0.0164, 0.0156, 0.0155, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 01:50:09,609 INFO [train.py:904] (6/8) Epoch 4, batch 1150, loss[loss=0.2339, simple_loss=0.295, pruned_loss=0.08641, over 15735.00 frames. ], tot_loss[loss=0.2242, simple_loss=0.2957, pruned_loss=0.07638, over 3299907.18 frames. ], batch size: 191, lr: 1.72e-02, grad_scale: 4.0 2023-04-28 01:50:18,997 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=31607.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:50:49,047 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=31629.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:50:52,305 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1390, 2.7289, 2.6096, 4.5971, 2.0282, 3.9939, 2.5039, 2.7088], device='cuda:6'), covar=tensor([0.0359, 0.1202, 0.0608, 0.0221, 0.2273, 0.0375, 0.1322, 0.1864], device='cuda:6'), in_proj_covar=tensor([0.0288, 0.0270, 0.0223, 0.0286, 0.0331, 0.0244, 0.0250, 0.0340], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 01:51:00,560 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=31638.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:51:04,259 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.256e+02 3.893e+02 4.649e+02 5.891e+02 1.405e+03, threshold=9.297e+02, percent-clipped=5.0 2023-04-28 01:51:19,068 INFO [train.py:904] (6/8) Epoch 4, batch 1200, loss[loss=0.2136, simple_loss=0.2872, pruned_loss=0.07003, over 16735.00 frames. ], tot_loss[loss=0.2229, simple_loss=0.2948, pruned_loss=0.07552, over 3300323.26 frames. ], batch size: 39, lr: 1.72e-02, grad_scale: 8.0 2023-04-28 01:51:31,726 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-28 01:51:40,949 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31667.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 01:52:27,378 INFO [train.py:904] (6/8) Epoch 4, batch 1250, loss[loss=0.2156, simple_loss=0.3031, pruned_loss=0.06406, over 17191.00 frames. ], tot_loss[loss=0.2236, simple_loss=0.295, pruned_loss=0.07608, over 3311450.96 frames. ], batch size: 46, lr: 1.72e-02, grad_scale: 8.0 2023-04-28 01:52:31,882 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31704.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:52:47,962 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=31715.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 01:52:53,543 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-28 01:53:22,524 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.448e+02 3.415e+02 4.132e+02 5.466e+02 7.512e+02, threshold=8.265e+02, percent-clipped=0.0 2023-04-28 01:53:35,322 INFO [train.py:904] (6/8) Epoch 4, batch 1300, loss[loss=0.2629, simple_loss=0.315, pruned_loss=0.1054, over 16780.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.2948, pruned_loss=0.07668, over 3301501.57 frames. ], batch size: 124, lr: 1.72e-02, grad_scale: 8.0 2023-04-28 01:53:55,319 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31765.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:53:55,491 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31765.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:54:08,712 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 01:54:26,616 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31788.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:54:44,573 INFO [train.py:904] (6/8) Epoch 4, batch 1350, loss[loss=0.2076, simple_loss=0.2859, pruned_loss=0.06464, over 17226.00 frames. ], tot_loss[loss=0.2221, simple_loss=0.294, pruned_loss=0.07505, over 3310476.32 frames. ], batch size: 45, lr: 1.72e-02, grad_scale: 8.0 2023-04-28 01:55:14,645 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31823.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:55:39,665 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.154e+02 3.282e+02 4.037e+02 5.139e+02 9.544e+02, threshold=8.075e+02, percent-clipped=2.0 2023-04-28 01:55:48,223 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31847.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 01:55:50,756 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31849.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:55:52,765 INFO [train.py:904] (6/8) Epoch 4, batch 1400, loss[loss=0.1738, simple_loss=0.2554, pruned_loss=0.04607, over 16839.00 frames. ], tot_loss[loss=0.2226, simple_loss=0.2942, pruned_loss=0.07548, over 3316035.56 frames. ], batch size: 42, lr: 1.72e-02, grad_scale: 8.0 2023-04-28 01:56:10,417 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2864, 5.2915, 5.0461, 4.4168, 5.1569, 2.1541, 4.7883, 5.1877], device='cuda:6'), covar=tensor([0.0051, 0.0047, 0.0070, 0.0292, 0.0049, 0.1338, 0.0076, 0.0098], device='cuda:6'), in_proj_covar=tensor([0.0087, 0.0078, 0.0116, 0.0127, 0.0085, 0.0133, 0.0103, 0.0117], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 01:56:32,727 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31878.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:56:40,456 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31884.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:57:02,739 INFO [train.py:904] (6/8) Epoch 4, batch 1450, loss[loss=0.2257, simple_loss=0.301, pruned_loss=0.07526, over 17109.00 frames. ], tot_loss[loss=0.2227, simple_loss=0.2943, pruned_loss=0.07558, over 3321561.87 frames. ], batch size: 48, lr: 1.71e-02, grad_scale: 4.0 2023-04-28 01:57:22,499 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31915.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:57:30,248 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-28 01:57:50,119 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-28 01:57:56,609 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31939.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:58:00,355 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.224e+02 3.023e+02 3.906e+02 5.046e+02 9.760e+02, threshold=7.812e+02, percent-clipped=3.0 2023-04-28 01:58:12,663 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31950.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:58:13,457 INFO [train.py:904] (6/8) Epoch 4, batch 1500, loss[loss=0.2492, simple_loss=0.2981, pruned_loss=0.1001, over 16770.00 frames. ], tot_loss[loss=0.2223, simple_loss=0.2937, pruned_loss=0.07541, over 3322639.85 frames. ], batch size: 124, lr: 1.71e-02, grad_scale: 4.0 2023-04-28 01:58:48,012 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31976.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:59:26,620 INFO [train.py:904] (6/8) Epoch 4, batch 1550, loss[loss=0.1868, simple_loss=0.2676, pruned_loss=0.05299, over 16987.00 frames. ], tot_loss[loss=0.2251, simple_loss=0.2954, pruned_loss=0.0774, over 3312259.52 frames. ], batch size: 41, lr: 1.71e-02, grad_scale: 4.0 2023-04-28 01:59:40,601 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=32011.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:00:21,778 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.322e+02 3.491e+02 4.515e+02 5.293e+02 8.780e+02, threshold=9.030e+02, percent-clipped=2.0 2023-04-28 02:00:34,386 INFO [train.py:904] (6/8) Epoch 4, batch 1600, loss[loss=0.2295, simple_loss=0.2961, pruned_loss=0.08149, over 16805.00 frames. ], tot_loss[loss=0.2261, simple_loss=0.297, pruned_loss=0.0776, over 3321810.75 frames. ], batch size: 83, lr: 1.71e-02, grad_scale: 8.0 2023-04-28 02:00:40,628 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.96 vs. limit=5.0 2023-04-28 02:00:46,080 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32060.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:00:52,775 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32065.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:01:30,840 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1882, 5.1401, 5.1196, 3.7429, 5.0565, 1.7049, 4.7901, 5.1375], device='cuda:6'), covar=tensor([0.0128, 0.0088, 0.0093, 0.0552, 0.0085, 0.1926, 0.0109, 0.0159], device='cuda:6'), in_proj_covar=tensor([0.0085, 0.0076, 0.0112, 0.0124, 0.0083, 0.0126, 0.0101, 0.0114], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 02:01:41,685 INFO [train.py:904] (6/8) Epoch 4, batch 1650, loss[loss=0.2091, simple_loss=0.2872, pruned_loss=0.06554, over 17174.00 frames. ], tot_loss[loss=0.2283, simple_loss=0.2995, pruned_loss=0.07854, over 3321774.99 frames. ], batch size: 46, lr: 1.71e-02, grad_scale: 8.0 2023-04-28 02:01:58,773 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32113.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:02:37,464 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.433e+02 3.425e+02 4.282e+02 5.685e+02 1.161e+03, threshold=8.563e+02, percent-clipped=3.0 2023-04-28 02:02:40,713 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32144.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:02:44,916 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32147.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 02:02:50,056 INFO [train.py:904] (6/8) Epoch 4, batch 1700, loss[loss=0.222, simple_loss=0.314, pruned_loss=0.06499, over 17016.00 frames. ], tot_loss[loss=0.2308, simple_loss=0.3026, pruned_loss=0.07948, over 3324123.16 frames. ], batch size: 50, lr: 1.71e-02, grad_scale: 8.0 2023-04-28 02:03:29,746 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32179.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:03:52,661 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-28 02:03:53,202 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32195.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 02:04:01,130 INFO [train.py:904] (6/8) Epoch 4, batch 1750, loss[loss=0.1798, simple_loss=0.2644, pruned_loss=0.04757, over 16856.00 frames. ], tot_loss[loss=0.2309, simple_loss=0.303, pruned_loss=0.07937, over 3324012.36 frames. ], batch size: 42, lr: 1.71e-02, grad_scale: 8.0 2023-04-28 02:04:48,291 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32234.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:04:58,762 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.044e+02 3.415e+02 4.164e+02 5.127e+02 8.958e+02, threshold=8.329e+02, percent-clipped=1.0 2023-04-28 02:05:11,464 INFO [train.py:904] (6/8) Epoch 4, batch 1800, loss[loss=0.2154, simple_loss=0.3061, pruned_loss=0.06236, over 17113.00 frames. ], tot_loss[loss=0.2315, simple_loss=0.3042, pruned_loss=0.0794, over 3326912.50 frames. ], batch size: 49, lr: 1.71e-02, grad_scale: 8.0 2023-04-28 02:05:39,602 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32271.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:06:07,454 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7087, 4.7257, 5.2539, 5.2776, 5.2864, 4.7185, 4.7948, 4.5363], device='cuda:6'), covar=tensor([0.0228, 0.0307, 0.0304, 0.0329, 0.0329, 0.0253, 0.0723, 0.0301], device='cuda:6'), in_proj_covar=tensor([0.0221, 0.0217, 0.0225, 0.0223, 0.0270, 0.0244, 0.0343, 0.0200], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 02:06:18,944 INFO [train.py:904] (6/8) Epoch 4, batch 1850, loss[loss=0.2234, simple_loss=0.3111, pruned_loss=0.06784, over 17131.00 frames. ], tot_loss[loss=0.2313, simple_loss=0.3048, pruned_loss=0.07885, over 3324683.63 frames. ], batch size: 49, lr: 1.70e-02, grad_scale: 4.0 2023-04-28 02:06:26,588 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32306.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:06:32,662 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4093, 5.3090, 5.1939, 5.0723, 4.6931, 5.2677, 5.1855, 4.8669], device='cuda:6'), covar=tensor([0.0339, 0.0176, 0.0142, 0.0149, 0.0831, 0.0175, 0.0140, 0.0357], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0161, 0.0207, 0.0174, 0.0247, 0.0194, 0.0150, 0.0210], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 02:07:11,864 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-28 02:07:18,153 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.462e+02 3.616e+02 4.272e+02 5.393e+02 1.345e+03, threshold=8.544e+02, percent-clipped=8.0 2023-04-28 02:07:29,364 INFO [train.py:904] (6/8) Epoch 4, batch 1900, loss[loss=0.277, simple_loss=0.3399, pruned_loss=0.1071, over 12022.00 frames. ], tot_loss[loss=0.2288, simple_loss=0.3028, pruned_loss=0.07743, over 3326616.68 frames. ], batch size: 248, lr: 1.70e-02, grad_scale: 4.0 2023-04-28 02:07:31,940 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5824, 4.6078, 3.9936, 2.1167, 2.9177, 2.7759, 3.7525, 4.1412], device='cuda:6'), covar=tensor([0.0317, 0.0414, 0.0469, 0.1465, 0.0773, 0.0857, 0.0808, 0.0818], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0129, 0.0154, 0.0142, 0.0135, 0.0127, 0.0141, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-28 02:07:42,144 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4214, 2.2976, 2.2657, 3.8117, 1.8474, 3.3720, 2.0939, 2.2143], device='cuda:6'), covar=tensor([0.0426, 0.1183, 0.0668, 0.0265, 0.2051, 0.0418, 0.1425, 0.1604], device='cuda:6'), in_proj_covar=tensor([0.0289, 0.0271, 0.0223, 0.0286, 0.0335, 0.0250, 0.0248, 0.0340], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 02:07:43,045 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32360.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:08:41,346 INFO [train.py:904] (6/8) Epoch 4, batch 1950, loss[loss=0.2151, simple_loss=0.299, pruned_loss=0.06554, over 17266.00 frames. ], tot_loss[loss=0.2271, simple_loss=0.3015, pruned_loss=0.0763, over 3320397.47 frames. ], batch size: 52, lr: 1.70e-02, grad_scale: 4.0 2023-04-28 02:08:50,794 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32408.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:09:16,982 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8507, 2.1552, 2.0677, 3.1628, 1.9769, 2.8559, 2.1314, 1.9746], device='cuda:6'), covar=tensor([0.0467, 0.1189, 0.0634, 0.0304, 0.1840, 0.0451, 0.1323, 0.1702], device='cuda:6'), in_proj_covar=tensor([0.0288, 0.0269, 0.0222, 0.0283, 0.0332, 0.0248, 0.0246, 0.0335], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 02:09:37,980 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.345e+02 3.206e+02 3.800e+02 4.937e+02 1.195e+03, threshold=7.601e+02, percent-clipped=2.0 2023-04-28 02:09:39,365 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32444.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:09:47,991 INFO [train.py:904] (6/8) Epoch 4, batch 2000, loss[loss=0.2089, simple_loss=0.2866, pruned_loss=0.06557, over 16856.00 frames. ], tot_loss[loss=0.2275, simple_loss=0.3015, pruned_loss=0.07677, over 3311870.36 frames. ], batch size: 42, lr: 1.70e-02, grad_scale: 8.0 2023-04-28 02:10:28,171 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32479.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:10:45,039 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32492.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:10:58,284 INFO [train.py:904] (6/8) Epoch 4, batch 2050, loss[loss=0.2383, simple_loss=0.312, pruned_loss=0.08233, over 16725.00 frames. ], tot_loss[loss=0.2281, simple_loss=0.3019, pruned_loss=0.07718, over 3314569.10 frames. ], batch size: 57, lr: 1.70e-02, grad_scale: 8.0 2023-04-28 02:11:34,833 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32527.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:11:44,300 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32534.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:11:57,876 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.075e+02 3.479e+02 4.226e+02 5.349e+02 1.056e+03, threshold=8.453e+02, percent-clipped=7.0 2023-04-28 02:12:07,692 INFO [train.py:904] (6/8) Epoch 4, batch 2100, loss[loss=0.2491, simple_loss=0.3285, pruned_loss=0.08483, over 16823.00 frames. ], tot_loss[loss=0.2305, simple_loss=0.3041, pruned_loss=0.07846, over 3321939.43 frames. ], batch size: 57, lr: 1.70e-02, grad_scale: 8.0 2023-04-28 02:12:36,520 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32571.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:12:51,051 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32582.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:13:17,653 INFO [train.py:904] (6/8) Epoch 4, batch 2150, loss[loss=0.2412, simple_loss=0.3179, pruned_loss=0.08227, over 16505.00 frames. ], tot_loss[loss=0.2338, simple_loss=0.3062, pruned_loss=0.08064, over 3313687.57 frames. ], batch size: 75, lr: 1.70e-02, grad_scale: 8.0 2023-04-28 02:13:25,227 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32606.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:13:25,353 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2881, 3.9728, 3.4505, 1.9154, 2.7846, 2.2486, 3.7108, 3.7922], device='cuda:6'), covar=tensor([0.0187, 0.0395, 0.0436, 0.1337, 0.0645, 0.0852, 0.0469, 0.0503], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0129, 0.0153, 0.0140, 0.0133, 0.0126, 0.0138, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-28 02:13:31,027 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3467, 5.6931, 5.3901, 5.5992, 4.9905, 4.7175, 5.2211, 5.8195], device='cuda:6'), covar=tensor([0.0530, 0.0611, 0.0933, 0.0322, 0.0541, 0.0655, 0.0555, 0.0652], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0447, 0.0381, 0.0280, 0.0284, 0.0283, 0.0351, 0.0309], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 02:13:42,049 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32619.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:14:15,065 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.513e+02 3.522e+02 4.278e+02 5.461e+02 9.146e+02, threshold=8.556e+02, percent-clipped=3.0 2023-04-28 02:14:26,822 INFO [train.py:904] (6/8) Epoch 4, batch 2200, loss[loss=0.2759, simple_loss=0.3293, pruned_loss=0.1113, over 15416.00 frames. ], tot_loss[loss=0.2338, simple_loss=0.3063, pruned_loss=0.08069, over 3319965.87 frames. ], batch size: 190, lr: 1.70e-02, grad_scale: 8.0 2023-04-28 02:14:30,431 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32654.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:15:34,029 INFO [train.py:904] (6/8) Epoch 4, batch 2250, loss[loss=0.2484, simple_loss=0.3295, pruned_loss=0.08368, over 16683.00 frames. ], tot_loss[loss=0.2339, simple_loss=0.3065, pruned_loss=0.08067, over 3324857.32 frames. ], batch size: 57, lr: 1.69e-02, grad_scale: 8.0 2023-04-28 02:16:09,981 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-28 02:16:32,022 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.152e+02 3.558e+02 4.554e+02 5.446e+02 8.366e+02, threshold=9.108e+02, percent-clipped=0.0 2023-04-28 02:16:35,794 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.54 vs. limit=5.0 2023-04-28 02:16:42,205 INFO [train.py:904] (6/8) Epoch 4, batch 2300, loss[loss=0.2333, simple_loss=0.2961, pruned_loss=0.08521, over 16715.00 frames. ], tot_loss[loss=0.2338, simple_loss=0.3066, pruned_loss=0.08047, over 3326925.53 frames. ], batch size: 124, lr: 1.69e-02, grad_scale: 4.0 2023-04-28 02:17:51,370 INFO [train.py:904] (6/8) Epoch 4, batch 2350, loss[loss=0.2582, simple_loss=0.335, pruned_loss=0.09065, over 16623.00 frames. ], tot_loss[loss=0.2328, simple_loss=0.3063, pruned_loss=0.07964, over 3330080.84 frames. ], batch size: 62, lr: 1.69e-02, grad_scale: 4.0 2023-04-28 02:18:04,096 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6198, 5.9496, 5.7186, 5.8356, 5.2006, 4.6939, 5.5160, 6.0552], device='cuda:6'), covar=tensor([0.0625, 0.0639, 0.0780, 0.0329, 0.0530, 0.0652, 0.0518, 0.0619], device='cuda:6'), in_proj_covar=tensor([0.0326, 0.0458, 0.0386, 0.0285, 0.0289, 0.0284, 0.0355, 0.0314], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 02:18:48,982 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.340e+02 3.303e+02 3.984e+02 5.110e+02 1.375e+03, threshold=7.969e+02, percent-clipped=3.0 2023-04-28 02:18:57,203 INFO [train.py:904] (6/8) Epoch 4, batch 2400, loss[loss=0.258, simple_loss=0.3182, pruned_loss=0.09894, over 16389.00 frames. ], tot_loss[loss=0.2336, simple_loss=0.3071, pruned_loss=0.08005, over 3328451.80 frames. ], batch size: 146, lr: 1.69e-02, grad_scale: 8.0 2023-04-28 02:20:06,099 INFO [train.py:904] (6/8) Epoch 4, batch 2450, loss[loss=0.2116, simple_loss=0.2944, pruned_loss=0.06442, over 17127.00 frames. ], tot_loss[loss=0.2335, simple_loss=0.3074, pruned_loss=0.07977, over 3326867.19 frames. ], batch size: 48, lr: 1.69e-02, grad_scale: 8.0 2023-04-28 02:21:03,695 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.564e+02 3.644e+02 4.364e+02 5.421e+02 9.049e+02, threshold=8.728e+02, percent-clipped=5.0 2023-04-28 02:21:13,293 INFO [train.py:904] (6/8) Epoch 4, batch 2500, loss[loss=0.192, simple_loss=0.2758, pruned_loss=0.05408, over 16837.00 frames. ], tot_loss[loss=0.2313, simple_loss=0.3057, pruned_loss=0.07842, over 3332594.20 frames. ], batch size: 42, lr: 1.69e-02, grad_scale: 8.0 2023-04-28 02:21:24,650 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-28 02:21:34,427 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=32966.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:22:01,862 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9060, 4.1376, 3.2690, 2.5265, 3.2374, 2.3372, 4.5448, 4.5997], device='cuda:6'), covar=tensor([0.1966, 0.0693, 0.1158, 0.1313, 0.1986, 0.1315, 0.0298, 0.0405], device='cuda:6'), in_proj_covar=tensor([0.0270, 0.0250, 0.0256, 0.0231, 0.0300, 0.0194, 0.0227, 0.0243], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 02:22:21,139 INFO [train.py:904] (6/8) Epoch 4, batch 2550, loss[loss=0.223, simple_loss=0.2931, pruned_loss=0.07643, over 15952.00 frames. ], tot_loss[loss=0.2317, simple_loss=0.3061, pruned_loss=0.07869, over 3322313.06 frames. ], batch size: 35, lr: 1.69e-02, grad_scale: 8.0 2023-04-28 02:22:29,482 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.32 vs. limit=5.0 2023-04-28 02:22:38,152 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6234, 2.4658, 2.4754, 4.2448, 2.0284, 3.5198, 2.1786, 2.3982], device='cuda:6'), covar=tensor([0.0403, 0.1127, 0.0586, 0.0205, 0.2096, 0.0440, 0.1443, 0.1533], device='cuda:6'), in_proj_covar=tensor([0.0293, 0.0273, 0.0226, 0.0288, 0.0337, 0.0256, 0.0252, 0.0340], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 02:22:57,640 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33027.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:23:05,670 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-28 02:23:20,059 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.536e+02 3.626e+02 4.489e+02 5.750e+02 1.218e+03, threshold=8.978e+02, percent-clipped=4.0 2023-04-28 02:23:30,980 INFO [train.py:904] (6/8) Epoch 4, batch 2600, loss[loss=0.2429, simple_loss=0.3161, pruned_loss=0.08484, over 16491.00 frames. ], tot_loss[loss=0.2311, simple_loss=0.3058, pruned_loss=0.0782, over 3333246.09 frames. ], batch size: 75, lr: 1.69e-02, grad_scale: 8.0 2023-04-28 02:23:55,328 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33069.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:24:39,065 INFO [train.py:904] (6/8) Epoch 4, batch 2650, loss[loss=0.2302, simple_loss=0.3013, pruned_loss=0.0795, over 16428.00 frames. ], tot_loss[loss=0.2312, simple_loss=0.3062, pruned_loss=0.07811, over 3331970.87 frames. ], batch size: 68, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:25:19,524 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33130.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:25:38,560 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.050e+02 3.285e+02 4.083e+02 4.956e+02 1.136e+03, threshold=8.166e+02, percent-clipped=3.0 2023-04-28 02:25:46,693 INFO [train.py:904] (6/8) Epoch 4, batch 2700, loss[loss=0.214, simple_loss=0.3097, pruned_loss=0.05919, over 17238.00 frames. ], tot_loss[loss=0.2311, simple_loss=0.3069, pruned_loss=0.07764, over 3337896.62 frames. ], batch size: 52, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:25:54,257 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33155.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:26:34,370 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9461, 3.9490, 4.4384, 4.4262, 4.4214, 4.0593, 4.1260, 3.9916], device='cuda:6'), covar=tensor([0.0263, 0.0351, 0.0316, 0.0410, 0.0399, 0.0312, 0.0715, 0.0426], device='cuda:6'), in_proj_covar=tensor([0.0224, 0.0221, 0.0228, 0.0231, 0.0278, 0.0238, 0.0346, 0.0208], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 02:26:55,036 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-28 02:26:56,823 INFO [train.py:904] (6/8) Epoch 4, batch 2750, loss[loss=0.1822, simple_loss=0.27, pruned_loss=0.04716, over 16991.00 frames. ], tot_loss[loss=0.2298, simple_loss=0.3058, pruned_loss=0.07694, over 3344251.62 frames. ], batch size: 41, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:27:16,300 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33216.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:27:47,473 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8395, 4.7458, 4.6981, 4.0335, 4.6848, 1.9506, 4.4281, 4.6977], device='cuda:6'), covar=tensor([0.0052, 0.0051, 0.0070, 0.0289, 0.0054, 0.1408, 0.0081, 0.0103], device='cuda:6'), in_proj_covar=tensor([0.0088, 0.0079, 0.0117, 0.0127, 0.0087, 0.0128, 0.0104, 0.0117], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 02:27:54,081 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.476e+02 3.299e+02 3.913e+02 4.982e+02 9.564e+02, threshold=7.826e+02, percent-clipped=3.0 2023-04-28 02:28:04,360 INFO [train.py:904] (6/8) Epoch 4, batch 2800, loss[loss=0.2, simple_loss=0.2818, pruned_loss=0.05908, over 16844.00 frames. ], tot_loss[loss=0.2297, simple_loss=0.3058, pruned_loss=0.07681, over 3340903.82 frames. ], batch size: 42, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:29:10,594 INFO [train.py:904] (6/8) Epoch 4, batch 2850, loss[loss=0.2221, simple_loss=0.307, pruned_loss=0.06856, over 17044.00 frames. ], tot_loss[loss=0.2289, simple_loss=0.3046, pruned_loss=0.07658, over 3328691.16 frames. ], batch size: 50, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:29:39,677 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33322.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:29:55,124 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-28 02:30:09,811 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.138e+02 3.527e+02 4.279e+02 5.315e+02 1.559e+03, threshold=8.559e+02, percent-clipped=4.0 2023-04-28 02:30:20,485 INFO [train.py:904] (6/8) Epoch 4, batch 2900, loss[loss=0.1947, simple_loss=0.2823, pruned_loss=0.05354, over 17187.00 frames. ], tot_loss[loss=0.2285, simple_loss=0.3036, pruned_loss=0.0767, over 3334596.50 frames. ], batch size: 46, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:30:33,812 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1188, 4.7574, 5.1064, 5.4509, 5.5481, 4.7363, 5.5185, 5.4365], device='cuda:6'), covar=tensor([0.0744, 0.0732, 0.1217, 0.0408, 0.0370, 0.0473, 0.0302, 0.0349], device='cuda:6'), in_proj_covar=tensor([0.0360, 0.0434, 0.0585, 0.0461, 0.0342, 0.0327, 0.0354, 0.0370], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 02:31:29,007 INFO [train.py:904] (6/8) Epoch 4, batch 2950, loss[loss=0.2995, simple_loss=0.3454, pruned_loss=0.1268, over 15448.00 frames. ], tot_loss[loss=0.2302, simple_loss=0.3038, pruned_loss=0.0783, over 3328926.17 frames. ], batch size: 190, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:31:53,419 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.22 vs. limit=5.0 2023-04-28 02:32:01,382 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33425.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:32:21,951 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-28 02:32:27,159 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.112e+02 3.646e+02 4.580e+02 6.032e+02 1.054e+03, threshold=9.160e+02, percent-clipped=6.0 2023-04-28 02:32:35,893 INFO [train.py:904] (6/8) Epoch 4, batch 3000, loss[loss=0.2148, simple_loss=0.2959, pruned_loss=0.06682, over 16834.00 frames. ], tot_loss[loss=0.2302, simple_loss=0.3038, pruned_loss=0.07828, over 3334822.05 frames. ], batch size: 83, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:32:35,894 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 02:32:45,553 INFO [train.py:938] (6/8) Epoch 4, validation: loss=0.1627, simple_loss=0.2694, pruned_loss=0.02796, over 944034.00 frames. 2023-04-28 02:32:45,554 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-28 02:32:54,964 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2023-04-28 02:33:13,576 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33472.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:33:15,661 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.69 vs. limit=5.0 2023-04-28 02:33:54,593 INFO [train.py:904] (6/8) Epoch 4, batch 3050, loss[loss=0.2399, simple_loss=0.2984, pruned_loss=0.09069, over 16817.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.3037, pruned_loss=0.07821, over 3334152.49 frames. ], batch size: 116, lr: 1.67e-02, grad_scale: 4.0 2023-04-28 02:34:07,904 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33511.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:34:20,631 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7318, 4.3890, 4.2091, 4.9512, 5.0220, 4.5851, 4.9760, 5.0759], device='cuda:6'), covar=tensor([0.0771, 0.0836, 0.2564, 0.0865, 0.0814, 0.0662, 0.0919, 0.0654], device='cuda:6'), in_proj_covar=tensor([0.0356, 0.0432, 0.0582, 0.0455, 0.0341, 0.0322, 0.0353, 0.0366], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 02:34:29,727 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6696, 4.3687, 4.3629, 1.6309, 4.5304, 4.5920, 3.2426, 3.8949], device='cuda:6'), covar=tensor([0.0755, 0.0110, 0.0150, 0.1176, 0.0068, 0.0066, 0.0296, 0.0233], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0083, 0.0083, 0.0142, 0.0073, 0.0075, 0.0113, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 02:34:35,912 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0247, 3.5026, 3.1804, 5.1543, 4.9739, 4.3936, 1.9401, 3.3763], device='cuda:6'), covar=tensor([0.1159, 0.0455, 0.0811, 0.0064, 0.0217, 0.0283, 0.1172, 0.0616], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0137, 0.0164, 0.0080, 0.0171, 0.0156, 0.0155, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 02:34:38,039 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33533.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:34:54,165 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.952e+02 3.585e+02 4.279e+02 5.215e+02 1.682e+03, threshold=8.559e+02, percent-clipped=1.0 2023-04-28 02:35:02,730 INFO [train.py:904] (6/8) Epoch 4, batch 3100, loss[loss=0.2144, simple_loss=0.2817, pruned_loss=0.07355, over 16666.00 frames. ], tot_loss[loss=0.2297, simple_loss=0.3032, pruned_loss=0.0781, over 3329606.09 frames. ], batch size: 37, lr: 1.67e-02, grad_scale: 4.0 2023-04-28 02:35:59,588 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6128, 4.5489, 4.5081, 3.8835, 4.5006, 2.0635, 4.3093, 4.5581], device='cuda:6'), covar=tensor([0.0073, 0.0063, 0.0087, 0.0335, 0.0063, 0.1319, 0.0079, 0.0115], device='cuda:6'), in_proj_covar=tensor([0.0090, 0.0080, 0.0120, 0.0129, 0.0089, 0.0129, 0.0107, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 02:36:10,677 INFO [train.py:904] (6/8) Epoch 4, batch 3150, loss[loss=0.2283, simple_loss=0.3155, pruned_loss=0.07055, over 17022.00 frames. ], tot_loss[loss=0.2279, simple_loss=0.3015, pruned_loss=0.07717, over 3322858.67 frames. ], batch size: 50, lr: 1.67e-02, grad_scale: 4.0 2023-04-28 02:36:20,314 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6419, 3.5965, 2.7594, 2.2492, 2.6847, 2.0235, 3.7085, 3.8054], device='cuda:6'), covar=tensor([0.1824, 0.0633, 0.1069, 0.1327, 0.2059, 0.1407, 0.0341, 0.0539], device='cuda:6'), in_proj_covar=tensor([0.0268, 0.0248, 0.0259, 0.0228, 0.0301, 0.0193, 0.0226, 0.0244], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 02:36:39,457 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33622.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:37:11,328 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.300e+02 3.211e+02 3.811e+02 4.570e+02 1.076e+03, threshold=7.622e+02, percent-clipped=2.0 2023-04-28 02:37:18,477 INFO [train.py:904] (6/8) Epoch 4, batch 3200, loss[loss=0.2243, simple_loss=0.2894, pruned_loss=0.07959, over 16386.00 frames. ], tot_loss[loss=0.2261, simple_loss=0.3003, pruned_loss=0.07589, over 3323526.41 frames. ], batch size: 146, lr: 1.67e-02, grad_scale: 8.0 2023-04-28 02:37:44,874 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=33670.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:37:54,609 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33677.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:38:06,157 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1628, 3.7706, 2.9584, 5.2115, 5.0709, 4.6123, 1.9066, 3.7109], device='cuda:6'), covar=tensor([0.1116, 0.0399, 0.0921, 0.0073, 0.0209, 0.0271, 0.1133, 0.0500], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0139, 0.0165, 0.0081, 0.0172, 0.0158, 0.0157, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 02:38:25,339 INFO [train.py:904] (6/8) Epoch 4, batch 3250, loss[loss=0.3198, simple_loss=0.3747, pruned_loss=0.1324, over 12148.00 frames. ], tot_loss[loss=0.2294, simple_loss=0.3027, pruned_loss=0.07806, over 3314380.12 frames. ], batch size: 246, lr: 1.67e-02, grad_scale: 8.0 2023-04-28 02:38:58,289 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33725.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:39:16,413 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33738.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 02:39:26,579 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.718e+02 3.624e+02 4.331e+02 5.289e+02 1.384e+03, threshold=8.662e+02, percent-clipped=5.0 2023-04-28 02:39:36,477 INFO [train.py:904] (6/8) Epoch 4, batch 3300, loss[loss=0.2415, simple_loss=0.3136, pruned_loss=0.08469, over 16489.00 frames. ], tot_loss[loss=0.2317, simple_loss=0.3048, pruned_loss=0.07928, over 3299703.52 frames. ], batch size: 146, lr: 1.67e-02, grad_scale: 8.0 2023-04-28 02:39:37,865 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9216, 3.8908, 4.3416, 4.3852, 4.3824, 3.9971, 4.0807, 4.0147], device='cuda:6'), covar=tensor([0.0256, 0.0464, 0.0313, 0.0361, 0.0317, 0.0284, 0.0667, 0.0371], device='cuda:6'), in_proj_covar=tensor([0.0228, 0.0225, 0.0231, 0.0234, 0.0278, 0.0243, 0.0350, 0.0207], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 02:40:07,419 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=33773.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:40:45,511 INFO [train.py:904] (6/8) Epoch 4, batch 3350, loss[loss=0.2102, simple_loss=0.2908, pruned_loss=0.0648, over 16469.00 frames. ], tot_loss[loss=0.23, simple_loss=0.3041, pruned_loss=0.07799, over 3313355.54 frames. ], batch size: 68, lr: 1.67e-02, grad_scale: 8.0 2023-04-28 02:40:49,945 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4430, 3.5535, 4.0048, 2.8291, 3.8646, 4.0812, 3.8931, 2.0294], device='cuda:6'), covar=tensor([0.0276, 0.0080, 0.0028, 0.0191, 0.0028, 0.0033, 0.0026, 0.0281], device='cuda:6'), in_proj_covar=tensor([0.0109, 0.0053, 0.0057, 0.0107, 0.0057, 0.0064, 0.0058, 0.0102], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 02:40:58,998 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33811.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:41:22,101 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33828.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:41:44,267 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.269e+02 3.302e+02 3.975e+02 4.780e+02 1.300e+03, threshold=7.950e+02, percent-clipped=1.0 2023-04-28 02:41:52,912 INFO [train.py:904] (6/8) Epoch 4, batch 3400, loss[loss=0.2132, simple_loss=0.307, pruned_loss=0.05965, over 17059.00 frames. ], tot_loss[loss=0.2284, simple_loss=0.3025, pruned_loss=0.07713, over 3316443.61 frames. ], batch size: 53, lr: 1.67e-02, grad_scale: 8.0 2023-04-28 02:42:04,794 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=33859.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:43:03,517 INFO [train.py:904] (6/8) Epoch 4, batch 3450, loss[loss=0.2148, simple_loss=0.2995, pruned_loss=0.06506, over 17042.00 frames. ], tot_loss[loss=0.2259, simple_loss=0.3004, pruned_loss=0.07571, over 3313914.84 frames. ], batch size: 50, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:43:58,418 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33939.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:44:05,158 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.980e+02 3.338e+02 4.153e+02 5.120e+02 1.104e+03, threshold=8.306e+02, percent-clipped=4.0 2023-04-28 02:44:13,344 INFO [train.py:904] (6/8) Epoch 4, batch 3500, loss[loss=0.2065, simple_loss=0.2993, pruned_loss=0.05689, over 17064.00 frames. ], tot_loss[loss=0.2245, simple_loss=0.2986, pruned_loss=0.07514, over 3324121.44 frames. ], batch size: 50, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:44:29,090 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0804, 3.1419, 3.3893, 2.3660, 3.2368, 3.4434, 3.2988, 1.8234], device='cuda:6'), covar=tensor([0.0244, 0.0048, 0.0026, 0.0187, 0.0034, 0.0032, 0.0029, 0.0251], device='cuda:6'), in_proj_covar=tensor([0.0111, 0.0054, 0.0058, 0.0109, 0.0059, 0.0065, 0.0058, 0.0104], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 02:44:59,098 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3821, 4.4098, 3.5695, 1.8301, 2.8041, 2.3248, 3.7084, 3.9176], device='cuda:6'), covar=tensor([0.0286, 0.0369, 0.0486, 0.1605, 0.0790, 0.0997, 0.0696, 0.0777], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0132, 0.0154, 0.0141, 0.0133, 0.0127, 0.0142, 0.0134], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-28 02:45:26,936 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34000.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:45:27,576 INFO [train.py:904] (6/8) Epoch 4, batch 3550, loss[loss=0.2069, simple_loss=0.2958, pruned_loss=0.05903, over 17153.00 frames. ], tot_loss[loss=0.2237, simple_loss=0.298, pruned_loss=0.07469, over 3320432.76 frames. ], batch size: 48, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:46:11,009 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34033.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 02:46:27,202 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34044.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:46:27,926 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.116e+02 3.324e+02 4.237e+02 5.025e+02 1.251e+03, threshold=8.474e+02, percent-clipped=3.0 2023-04-28 02:46:30,761 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3577, 4.1584, 4.3213, 4.5872, 4.6126, 4.1199, 4.3309, 4.5883], device='cuda:6'), covar=tensor([0.0714, 0.0562, 0.1021, 0.0453, 0.0515, 0.0723, 0.1202, 0.0427], device='cuda:6'), in_proj_covar=tensor([0.0369, 0.0442, 0.0599, 0.0471, 0.0351, 0.0335, 0.0363, 0.0381], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 02:46:35,600 INFO [train.py:904] (6/8) Epoch 4, batch 3600, loss[loss=0.2695, simple_loss=0.3195, pruned_loss=0.1098, over 11736.00 frames. ], tot_loss[loss=0.2221, simple_loss=0.296, pruned_loss=0.07414, over 3321426.51 frames. ], batch size: 248, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:47:48,226 INFO [train.py:904] (6/8) Epoch 4, batch 3650, loss[loss=0.2275, simple_loss=0.2955, pruned_loss=0.0797, over 16259.00 frames. ], tot_loss[loss=0.2217, simple_loss=0.295, pruned_loss=0.07418, over 3319760.24 frames. ], batch size: 165, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:47:54,676 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34105.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:48:29,121 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34128.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:48:46,115 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-28 02:48:53,739 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.336e+02 3.266e+02 3.802e+02 4.883e+02 8.817e+02, threshold=7.604e+02, percent-clipped=1.0 2023-04-28 02:48:54,281 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34145.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:49:02,642 INFO [train.py:904] (6/8) Epoch 4, batch 3700, loss[loss=0.2202, simple_loss=0.2809, pruned_loss=0.07977, over 16217.00 frames. ], tot_loss[loss=0.2233, simple_loss=0.2943, pruned_loss=0.07615, over 3288404.97 frames. ], batch size: 165, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:49:41,506 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=34176.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:50:17,551 INFO [train.py:904] (6/8) Epoch 4, batch 3750, loss[loss=0.2308, simple_loss=0.3052, pruned_loss=0.07821, over 17242.00 frames. ], tot_loss[loss=0.2261, simple_loss=0.2955, pruned_loss=0.07834, over 3265492.51 frames. ], batch size: 45, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:50:25,293 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34206.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:50:35,384 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4190, 3.8899, 3.9605, 2.0631, 4.0971, 4.1355, 3.1774, 2.9372], device='cuda:6'), covar=tensor([0.0766, 0.0090, 0.0117, 0.1033, 0.0054, 0.0050, 0.0296, 0.0380], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0082, 0.0081, 0.0142, 0.0073, 0.0075, 0.0114, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 02:51:21,419 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.239e+02 3.358e+02 4.060e+02 5.073e+02 1.417e+03, threshold=8.120e+02, percent-clipped=4.0 2023-04-28 02:51:29,934 INFO [train.py:904] (6/8) Epoch 4, batch 3800, loss[loss=0.1948, simple_loss=0.2688, pruned_loss=0.06043, over 16450.00 frames. ], tot_loss[loss=0.2283, simple_loss=0.2967, pruned_loss=0.0799, over 3264811.54 frames. ], batch size: 75, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:52:01,874 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5715, 2.1582, 1.5704, 2.0532, 2.6408, 2.5462, 2.8271, 2.7035], device='cuda:6'), covar=tensor([0.0055, 0.0157, 0.0216, 0.0194, 0.0090, 0.0131, 0.0077, 0.0107], device='cuda:6'), in_proj_covar=tensor([0.0076, 0.0139, 0.0136, 0.0133, 0.0129, 0.0137, 0.0105, 0.0116], device='cuda:6'), out_proj_covar=tensor([9.9689e-05, 1.8510e-04, 1.7553e-04, 1.7386e-04, 1.7323e-04, 1.8445e-04, 1.3943e-04, 1.5676e-04], device='cuda:6') 2023-04-28 02:52:36,322 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34295.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:52:37,647 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4215, 4.3016, 4.3414, 3.8771, 4.3180, 1.7851, 4.1137, 4.3377], device='cuda:6'), covar=tensor([0.0068, 0.0061, 0.0075, 0.0273, 0.0055, 0.1474, 0.0081, 0.0112], device='cuda:6'), in_proj_covar=tensor([0.0090, 0.0080, 0.0119, 0.0129, 0.0089, 0.0129, 0.0105, 0.0117], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 02:52:44,483 INFO [train.py:904] (6/8) Epoch 4, batch 3850, loss[loss=0.229, simple_loss=0.289, pruned_loss=0.08455, over 16845.00 frames. ], tot_loss[loss=0.2305, simple_loss=0.2982, pruned_loss=0.08141, over 3252319.64 frames. ], batch size: 116, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 02:52:55,849 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5190, 3.8994, 3.9302, 1.9902, 4.0851, 4.1508, 3.2859, 2.8759], device='cuda:6'), covar=tensor([0.0686, 0.0092, 0.0119, 0.1107, 0.0061, 0.0051, 0.0261, 0.0422], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0082, 0.0080, 0.0140, 0.0072, 0.0074, 0.0112, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 02:53:14,256 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-28 02:53:31,106 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34333.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:53:49,432 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.155e+02 3.189e+02 3.757e+02 4.739e+02 9.284e+02, threshold=7.515e+02, percent-clipped=3.0 2023-04-28 02:53:57,007 INFO [train.py:904] (6/8) Epoch 4, batch 3900, loss[loss=0.2221, simple_loss=0.2885, pruned_loss=0.07789, over 16795.00 frames. ], tot_loss[loss=0.23, simple_loss=0.2973, pruned_loss=0.08138, over 3254295.05 frames. ], batch size: 102, lr: 1.65e-02, grad_scale: 4.0 2023-04-28 02:54:40,914 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=34381.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:55:09,561 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34400.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:55:10,264 INFO [train.py:904] (6/8) Epoch 4, batch 3950, loss[loss=0.2392, simple_loss=0.2911, pruned_loss=0.09367, over 16913.00 frames. ], tot_loss[loss=0.2306, simple_loss=0.297, pruned_loss=0.08211, over 3263589.87 frames. ], batch size: 109, lr: 1.65e-02, grad_scale: 4.0 2023-04-28 02:56:16,758 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 3.579e+02 4.210e+02 5.438e+02 1.110e+03, threshold=8.420e+02, percent-clipped=9.0 2023-04-28 02:56:24,481 INFO [train.py:904] (6/8) Epoch 4, batch 4000, loss[loss=0.2024, simple_loss=0.2779, pruned_loss=0.06343, over 16995.00 frames. ], tot_loss[loss=0.2298, simple_loss=0.2963, pruned_loss=0.08162, over 3274768.52 frames. ], batch size: 53, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 02:57:31,978 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-28 02:57:36,428 INFO [train.py:904] (6/8) Epoch 4, batch 4050, loss[loss=0.2182, simple_loss=0.2955, pruned_loss=0.07048, over 16491.00 frames. ], tot_loss[loss=0.2256, simple_loss=0.2943, pruned_loss=0.0784, over 3281473.06 frames. ], batch size: 68, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 02:57:36,852 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34501.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:58:41,516 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.628e+02 2.624e+02 2.984e+02 3.642e+02 7.677e+02, threshold=5.968e+02, percent-clipped=0.0 2023-04-28 02:58:48,881 INFO [train.py:904] (6/8) Epoch 4, batch 4100, loss[loss=0.2606, simple_loss=0.3414, pruned_loss=0.08988, over 16411.00 frames. ], tot_loss[loss=0.224, simple_loss=0.2944, pruned_loss=0.07678, over 3279933.74 frames. ], batch size: 75, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 02:59:53,980 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34593.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:59:57,537 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34595.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:00:06,227 INFO [train.py:904] (6/8) Epoch 4, batch 4150, loss[loss=0.232, simple_loss=0.3112, pruned_loss=0.07644, over 17043.00 frames. ], tot_loss[loss=0.231, simple_loss=0.3023, pruned_loss=0.07985, over 3269992.70 frames. ], batch size: 53, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 03:00:15,250 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 03:00:17,355 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9768, 2.7827, 2.0529, 2.3694, 3.2573, 2.8608, 4.0362, 3.5217], device='cuda:6'), covar=tensor([0.0011, 0.0108, 0.0184, 0.0155, 0.0066, 0.0119, 0.0028, 0.0054], device='cuda:6'), in_proj_covar=tensor([0.0074, 0.0137, 0.0137, 0.0135, 0.0129, 0.0139, 0.0104, 0.0117], device='cuda:6'), out_proj_covar=tensor([9.6843e-05, 1.8159e-04, 1.7672e-04, 1.7658e-04, 1.7240e-04, 1.8678e-04, 1.3779e-04, 1.5764e-04], device='cuda:6') 2023-04-28 03:00:38,856 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34622.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:00:44,929 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34626.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:01:10,674 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=34643.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:01:15,082 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.947e+02 3.447e+02 4.221e+02 5.425e+02 1.080e+03, threshold=8.441e+02, percent-clipped=15.0 2023-04-28 03:01:22,995 INFO [train.py:904] (6/8) Epoch 4, batch 4200, loss[loss=0.2879, simple_loss=0.3473, pruned_loss=0.1143, over 11190.00 frames. ], tot_loss[loss=0.2391, simple_loss=0.3115, pruned_loss=0.08334, over 3221237.04 frames. ], batch size: 246, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 03:01:28,604 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34654.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:02:13,302 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34683.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 03:02:19,900 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34687.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:02:21,000 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8015, 5.3042, 5.4003, 5.2158, 5.1276, 5.7256, 5.4751, 5.2843], device='cuda:6'), covar=tensor([0.0685, 0.1058, 0.0900, 0.1549, 0.2155, 0.0721, 0.0866, 0.1837], device='cuda:6'), in_proj_covar=tensor([0.0253, 0.0352, 0.0326, 0.0301, 0.0391, 0.0361, 0.0285, 0.0405], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:02:24,817 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9533, 3.2820, 3.0759, 1.5807, 3.3856, 3.3598, 2.7514, 2.6812], device='cuda:6'), covar=tensor([0.0764, 0.0086, 0.0126, 0.1131, 0.0046, 0.0053, 0.0284, 0.0339], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0081, 0.0079, 0.0141, 0.0069, 0.0071, 0.0112, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 03:02:27,243 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7044, 3.4651, 3.3407, 2.3581, 3.2688, 3.2428, 3.2949, 1.8749], device='cuda:6'), covar=tensor([0.0325, 0.0017, 0.0027, 0.0188, 0.0031, 0.0063, 0.0022, 0.0253], device='cuda:6'), in_proj_covar=tensor([0.0108, 0.0050, 0.0054, 0.0106, 0.0057, 0.0061, 0.0056, 0.0101], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:02:39,607 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34700.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:02:40,350 INFO [train.py:904] (6/8) Epoch 4, batch 4250, loss[loss=0.2293, simple_loss=0.3102, pruned_loss=0.07424, over 16240.00 frames. ], tot_loss[loss=0.2432, simple_loss=0.3163, pruned_loss=0.08503, over 3179817.76 frames. ], batch size: 165, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 03:02:53,627 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34710.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 03:03:47,976 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.098e+02 2.926e+02 3.520e+02 4.399e+02 1.246e+03, threshold=7.039e+02, percent-clipped=2.0 2023-04-28 03:03:52,077 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=34748.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:03:56,212 INFO [train.py:904] (6/8) Epoch 4, batch 4300, loss[loss=0.2467, simple_loss=0.3346, pruned_loss=0.07942, over 16702.00 frames. ], tot_loss[loss=0.2425, simple_loss=0.3174, pruned_loss=0.08377, over 3180717.83 frames. ], batch size: 89, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:04:27,388 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34771.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 03:04:42,087 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7196, 4.4083, 4.3246, 1.7232, 4.9121, 4.9231, 3.4031, 3.3828], device='cuda:6'), covar=tensor([0.0712, 0.0088, 0.0177, 0.1240, 0.0018, 0.0015, 0.0275, 0.0380], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0081, 0.0080, 0.0139, 0.0069, 0.0071, 0.0112, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 03:04:49,130 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6810, 2.4202, 2.3041, 3.9677, 1.6586, 3.3415, 2.1230, 2.1714], device='cuda:6'), covar=tensor([0.0534, 0.1570, 0.0849, 0.0380, 0.3158, 0.0711, 0.1591, 0.2679], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0277, 0.0229, 0.0288, 0.0341, 0.0255, 0.0253, 0.0346], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 03:04:55,448 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-28 03:04:58,575 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8626, 2.7123, 2.4306, 1.6042, 2.6752, 2.7395, 2.4318, 2.1555], device='cuda:6'), covar=tensor([0.0768, 0.0137, 0.0182, 0.1011, 0.0066, 0.0082, 0.0358, 0.0499], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0081, 0.0081, 0.0139, 0.0069, 0.0071, 0.0112, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 03:05:10,251 INFO [train.py:904] (6/8) Epoch 4, batch 4350, loss[loss=0.2654, simple_loss=0.3364, pruned_loss=0.09727, over 16891.00 frames. ], tot_loss[loss=0.2453, simple_loss=0.3205, pruned_loss=0.08501, over 3172723.83 frames. ], batch size: 116, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:05:10,588 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34801.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:05:12,406 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2023-04-28 03:05:35,857 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0861, 2.0372, 1.6226, 1.8989, 2.5709, 2.2928, 3.0175, 2.8767], device='cuda:6'), covar=tensor([0.0021, 0.0148, 0.0198, 0.0175, 0.0091, 0.0143, 0.0035, 0.0069], device='cuda:6'), in_proj_covar=tensor([0.0071, 0.0137, 0.0138, 0.0135, 0.0130, 0.0140, 0.0100, 0.0116], device='cuda:6'), out_proj_covar=tensor([9.3365e-05, 1.8167e-04, 1.7768e-04, 1.7505e-04, 1.7460e-04, 1.8672e-04, 1.3208e-04, 1.5648e-04], device='cuda:6') 2023-04-28 03:06:03,041 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3621, 4.0342, 3.9134, 1.5612, 4.3823, 4.4015, 3.1305, 3.2496], device='cuda:6'), covar=tensor([0.0748, 0.0107, 0.0236, 0.1271, 0.0022, 0.0027, 0.0312, 0.0332], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0082, 0.0081, 0.0139, 0.0069, 0.0071, 0.0111, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 03:06:15,521 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.361e+02 3.334e+02 4.019e+02 4.887e+02 8.589e+02, threshold=8.038e+02, percent-clipped=2.0 2023-04-28 03:06:20,217 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=34849.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:06:22,168 INFO [train.py:904] (6/8) Epoch 4, batch 4400, loss[loss=0.2939, simple_loss=0.3484, pruned_loss=0.1197, over 11624.00 frames. ], tot_loss[loss=0.2465, simple_loss=0.3219, pruned_loss=0.0856, over 3169170.98 frames. ], batch size: 246, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:06:41,079 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 03:07:32,093 INFO [train.py:904] (6/8) Epoch 4, batch 4450, loss[loss=0.2326, simple_loss=0.32, pruned_loss=0.07261, over 16782.00 frames. ], tot_loss[loss=0.2478, simple_loss=0.3245, pruned_loss=0.08558, over 3177365.77 frames. ], batch size: 124, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:08:36,347 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.991e+02 2.921e+02 3.471e+02 4.143e+02 7.527e+02, threshold=6.942e+02, percent-clipped=0.0 2023-04-28 03:08:40,980 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34949.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:08:43,009 INFO [train.py:904] (6/8) Epoch 4, batch 4500, loss[loss=0.2504, simple_loss=0.3183, pruned_loss=0.09121, over 16606.00 frames. ], tot_loss[loss=0.2467, simple_loss=0.3239, pruned_loss=0.08477, over 3196273.57 frames. ], batch size: 57, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:08:54,146 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0706, 2.1441, 2.1768, 3.5706, 1.7834, 3.0330, 2.2407, 2.0562], device='cuda:6'), covar=tensor([0.0473, 0.1294, 0.0704, 0.0264, 0.2383, 0.0475, 0.1289, 0.1892], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0279, 0.0231, 0.0290, 0.0347, 0.0253, 0.0253, 0.0350], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 03:09:23,015 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34978.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:09:28,663 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34982.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:09:54,753 INFO [train.py:904] (6/8) Epoch 4, batch 4550, loss[loss=0.2306, simple_loss=0.3172, pruned_loss=0.07203, over 16739.00 frames. ], tot_loss[loss=0.2466, simple_loss=0.3237, pruned_loss=0.08475, over 3213943.83 frames. ], batch size: 76, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:10:48,290 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-28 03:10:55,215 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2023-04-28 03:10:57,620 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.276e+02 2.863e+02 3.417e+02 4.239e+02 9.930e+02, threshold=6.834e+02, percent-clipped=3.0 2023-04-28 03:11:04,051 INFO [train.py:904] (6/8) Epoch 4, batch 4600, loss[loss=0.2373, simple_loss=0.3157, pruned_loss=0.07948, over 16452.00 frames. ], tot_loss[loss=0.2459, simple_loss=0.3237, pruned_loss=0.08408, over 3220721.18 frames. ], batch size: 146, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:11:25,508 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35066.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 03:12:02,603 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35092.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:12:06,779 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9681, 3.9682, 4.3594, 4.3773, 4.3946, 3.9669, 4.0375, 3.8645], device='cuda:6'), covar=tensor([0.0217, 0.0241, 0.0278, 0.0341, 0.0310, 0.0262, 0.0607, 0.0375], device='cuda:6'), in_proj_covar=tensor([0.0202, 0.0197, 0.0204, 0.0207, 0.0249, 0.0215, 0.0311, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 03:12:15,705 INFO [train.py:904] (6/8) Epoch 4, batch 4650, loss[loss=0.2312, simple_loss=0.3109, pruned_loss=0.07573, over 17262.00 frames. ], tot_loss[loss=0.245, simple_loss=0.3226, pruned_loss=0.0837, over 3224058.23 frames. ], batch size: 45, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:12:16,718 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3558, 5.6910, 5.3031, 5.5219, 4.9963, 4.4065, 5.1539, 5.8214], device='cuda:6'), covar=tensor([0.0634, 0.0503, 0.0804, 0.0343, 0.0548, 0.0701, 0.0503, 0.0589], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0413, 0.0355, 0.0267, 0.0266, 0.0269, 0.0331, 0.0294], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 03:13:20,502 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.712e+02 3.466e+02 4.003e+02 6.879e+02, threshold=6.932e+02, percent-clipped=1.0 2023-04-28 03:13:28,256 INFO [train.py:904] (6/8) Epoch 4, batch 4700, loss[loss=0.229, simple_loss=0.3058, pruned_loss=0.07612, over 16788.00 frames. ], tot_loss[loss=0.2416, simple_loss=0.3191, pruned_loss=0.08204, over 3222875.48 frames. ], batch size: 39, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:13:32,475 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35153.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:14:41,628 INFO [train.py:904] (6/8) Epoch 4, batch 4750, loss[loss=0.2541, simple_loss=0.3295, pruned_loss=0.08932, over 15460.00 frames. ], tot_loss[loss=0.2382, simple_loss=0.3154, pruned_loss=0.08048, over 3221124.72 frames. ], batch size: 190, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:15:36,118 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2023-04-28 03:15:45,876 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.092e+02 2.820e+02 3.474e+02 4.196e+02 7.562e+02, threshold=6.948e+02, percent-clipped=1.0 2023-04-28 03:15:52,044 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35249.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:15:53,750 INFO [train.py:904] (6/8) Epoch 4, batch 4800, loss[loss=0.2391, simple_loss=0.3137, pruned_loss=0.08226, over 16484.00 frames. ], tot_loss[loss=0.2344, simple_loss=0.3116, pruned_loss=0.07861, over 3214028.11 frames. ], batch size: 68, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:16:13,403 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9361, 3.9851, 3.7367, 3.7915, 3.4910, 3.9234, 3.5798, 3.6756], device='cuda:6'), covar=tensor([0.0362, 0.0177, 0.0191, 0.0129, 0.0576, 0.0186, 0.0480, 0.0354], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0151, 0.0197, 0.0160, 0.0224, 0.0178, 0.0142, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:16:30,539 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6207, 3.3109, 3.1479, 2.0608, 2.9969, 3.1093, 2.9979, 1.6245], device='cuda:6'), covar=tensor([0.0375, 0.0022, 0.0031, 0.0252, 0.0049, 0.0062, 0.0043, 0.0373], device='cuda:6'), in_proj_covar=tensor([0.0110, 0.0050, 0.0055, 0.0109, 0.0057, 0.0063, 0.0058, 0.0104], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:16:33,400 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35278.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:16:38,626 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35282.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:16:59,985 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=35297.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:17:03,230 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9811, 3.1766, 3.1383, 1.5701, 3.3465, 3.3515, 2.6766, 2.5372], device='cuda:6'), covar=tensor([0.0848, 0.0105, 0.0129, 0.1204, 0.0058, 0.0067, 0.0337, 0.0456], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0084, 0.0079, 0.0142, 0.0069, 0.0074, 0.0113, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 03:17:05,714 INFO [train.py:904] (6/8) Epoch 4, batch 4850, loss[loss=0.2183, simple_loss=0.3075, pruned_loss=0.06457, over 16885.00 frames. ], tot_loss[loss=0.2356, simple_loss=0.3137, pruned_loss=0.07878, over 3187627.99 frames. ], batch size: 109, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:17:41,858 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=35326.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:17:48,531 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=35330.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:18:11,850 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35345.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 03:18:12,515 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.805e+02 3.003e+02 3.470e+02 4.055e+02 1.018e+03, threshold=6.940e+02, percent-clipped=2.0 2023-04-28 03:18:19,083 INFO [train.py:904] (6/8) Epoch 4, batch 4900, loss[loss=0.2326, simple_loss=0.3037, pruned_loss=0.08075, over 17105.00 frames. ], tot_loss[loss=0.2341, simple_loss=0.3129, pruned_loss=0.07762, over 3169593.55 frames. ], batch size: 49, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:18:42,062 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35366.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:18:54,383 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35374.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:19:14,213 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2023-04-28 03:19:32,913 INFO [train.py:904] (6/8) Epoch 4, batch 4950, loss[loss=0.2416, simple_loss=0.3364, pruned_loss=0.07337, over 16694.00 frames. ], tot_loss[loss=0.2345, simple_loss=0.3135, pruned_loss=0.07777, over 3184572.42 frames. ], batch size: 83, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:19:41,963 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35406.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:19:53,434 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=35414.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 03:20:22,807 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35435.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:20:37,429 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.040e+02 3.403e+02 3.998e+02 5.018e+02 8.259e+02, threshold=7.996e+02, percent-clipped=7.0 2023-04-28 03:20:41,763 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35448.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:20:45,623 INFO [train.py:904] (6/8) Epoch 4, batch 5000, loss[loss=0.2616, simple_loss=0.3352, pruned_loss=0.09396, over 11612.00 frames. ], tot_loss[loss=0.2358, simple_loss=0.3155, pruned_loss=0.0781, over 3177940.76 frames. ], batch size: 248, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:21:12,563 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9124, 3.8721, 3.6926, 3.7255, 3.3787, 3.8746, 3.6094, 3.6072], device='cuda:6'), covar=tensor([0.0400, 0.0277, 0.0232, 0.0163, 0.0796, 0.0265, 0.0671, 0.0383], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0158, 0.0202, 0.0166, 0.0231, 0.0188, 0.0145, 0.0206], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:21:57,694 INFO [train.py:904] (6/8) Epoch 4, batch 5050, loss[loss=0.1992, simple_loss=0.2882, pruned_loss=0.05507, over 16732.00 frames. ], tot_loss[loss=0.2345, simple_loss=0.3149, pruned_loss=0.07704, over 3209314.50 frames. ], batch size: 76, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:22:39,969 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35530.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:23:03,492 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.171e+02 3.027e+02 3.520e+02 4.379e+02 8.983e+02, threshold=7.040e+02, percent-clipped=1.0 2023-04-28 03:23:10,553 INFO [train.py:904] (6/8) Epoch 4, batch 5100, loss[loss=0.2181, simple_loss=0.3031, pruned_loss=0.06659, over 16745.00 frames. ], tot_loss[loss=0.2322, simple_loss=0.3126, pruned_loss=0.07592, over 3212557.11 frames. ], batch size: 134, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:24:00,669 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5775, 4.2673, 4.2576, 3.2671, 3.9767, 4.2066, 3.8662, 2.4498], device='cuda:6'), covar=tensor([0.0280, 0.0015, 0.0014, 0.0155, 0.0034, 0.0056, 0.0037, 0.0245], device='cuda:6'), in_proj_covar=tensor([0.0111, 0.0051, 0.0055, 0.0110, 0.0058, 0.0064, 0.0058, 0.0103], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:24:08,062 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35591.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:24:22,823 INFO [train.py:904] (6/8) Epoch 4, batch 5150, loss[loss=0.2331, simple_loss=0.3225, pruned_loss=0.07185, over 16435.00 frames. ], tot_loss[loss=0.2319, simple_loss=0.3131, pruned_loss=0.07531, over 3214035.38 frames. ], batch size: 146, lr: 1.62e-02, grad_scale: 8.0 2023-04-28 03:25:09,545 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7768, 3.8718, 3.6012, 3.6732, 3.1082, 3.7738, 3.5613, 3.4267], device='cuda:6'), covar=tensor([0.0460, 0.0263, 0.0277, 0.0202, 0.1006, 0.0297, 0.0848, 0.0429], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0159, 0.0200, 0.0164, 0.0230, 0.0188, 0.0144, 0.0205], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:25:29,065 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.054e+02 2.805e+02 3.383e+02 4.164e+02 1.002e+03, threshold=6.766e+02, percent-clipped=7.0 2023-04-28 03:25:36,102 INFO [train.py:904] (6/8) Epoch 4, batch 5200, loss[loss=0.2316, simple_loss=0.3082, pruned_loss=0.07744, over 16536.00 frames. ], tot_loss[loss=0.231, simple_loss=0.3119, pruned_loss=0.07507, over 3199459.63 frames. ], batch size: 75, lr: 1.62e-02, grad_scale: 8.0 2023-04-28 03:25:51,452 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4545, 4.2947, 4.5115, 1.8984, 4.9114, 4.8886, 3.5031, 3.6666], device='cuda:6'), covar=tensor([0.0815, 0.0084, 0.0118, 0.1324, 0.0024, 0.0020, 0.0231, 0.0313], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0084, 0.0080, 0.0142, 0.0069, 0.0073, 0.0112, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 03:25:58,396 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2656, 3.4652, 1.5004, 3.6015, 2.3034, 3.5657, 1.7685, 2.5835], device='cuda:6'), covar=tensor([0.0116, 0.0205, 0.1805, 0.0040, 0.0842, 0.0291, 0.1462, 0.0625], device='cuda:6'), in_proj_covar=tensor([0.0108, 0.0144, 0.0176, 0.0079, 0.0161, 0.0171, 0.0183, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 03:26:01,394 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0891, 3.8198, 3.6175, 2.5197, 3.4772, 3.6328, 3.4660, 1.6550], device='cuda:6'), covar=tensor([0.0318, 0.0028, 0.0049, 0.0232, 0.0053, 0.0081, 0.0072, 0.0403], device='cuda:6'), in_proj_covar=tensor([0.0110, 0.0051, 0.0055, 0.0110, 0.0058, 0.0063, 0.0058, 0.0103], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:26:46,332 INFO [train.py:904] (6/8) Epoch 4, batch 5250, loss[loss=0.2133, simple_loss=0.2932, pruned_loss=0.06668, over 16581.00 frames. ], tot_loss[loss=0.2291, simple_loss=0.3089, pruned_loss=0.07463, over 3207557.68 frames. ], batch size: 68, lr: 1.62e-02, grad_scale: 8.0 2023-04-28 03:26:47,352 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35701.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 03:27:28,651 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35730.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:27:52,066 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.847e+02 2.886e+02 3.406e+02 4.102e+02 7.546e+02, threshold=6.813e+02, percent-clipped=1.0 2023-04-28 03:27:55,802 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35748.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:28:00,137 INFO [train.py:904] (6/8) Epoch 4, batch 5300, loss[loss=0.2135, simple_loss=0.2908, pruned_loss=0.06809, over 16769.00 frames. ], tot_loss[loss=0.2261, simple_loss=0.3055, pruned_loss=0.07339, over 3202513.74 frames. ], batch size: 83, lr: 1.62e-02, grad_scale: 8.0 2023-04-28 03:28:28,774 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35771.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:28:30,117 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2023-04-28 03:29:02,750 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=35796.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:29:11,333 INFO [train.py:904] (6/8) Epoch 4, batch 5350, loss[loss=0.2611, simple_loss=0.3538, pruned_loss=0.08421, over 16711.00 frames. ], tot_loss[loss=0.2243, simple_loss=0.3038, pruned_loss=0.07241, over 3214178.50 frames. ], batch size: 124, lr: 1.62e-02, grad_scale: 4.0 2023-04-28 03:29:18,521 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-28 03:29:24,871 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35811.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:29:57,503 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35832.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:30:17,961 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.047e+02 3.081e+02 3.703e+02 4.509e+02 8.729e+02, threshold=7.406e+02, percent-clipped=5.0 2023-04-28 03:30:23,338 INFO [train.py:904] (6/8) Epoch 4, batch 5400, loss[loss=0.2928, simple_loss=0.3721, pruned_loss=0.1068, over 15308.00 frames. ], tot_loss[loss=0.2271, simple_loss=0.3069, pruned_loss=0.07362, over 3214274.07 frames. ], batch size: 190, lr: 1.62e-02, grad_scale: 4.0 2023-04-28 03:30:53,699 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35872.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:31:00,244 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3668, 4.2674, 4.1998, 3.5143, 4.2562, 1.7770, 4.0027, 4.1396], device='cuda:6'), covar=tensor([0.0050, 0.0047, 0.0057, 0.0304, 0.0048, 0.1448, 0.0076, 0.0094], device='cuda:6'), in_proj_covar=tensor([0.0080, 0.0070, 0.0105, 0.0120, 0.0080, 0.0125, 0.0094, 0.0107], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 03:31:13,915 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35886.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:31:38,572 INFO [train.py:904] (6/8) Epoch 4, batch 5450, loss[loss=0.2761, simple_loss=0.3473, pruned_loss=0.1025, over 16396.00 frames. ], tot_loss[loss=0.2321, simple_loss=0.3113, pruned_loss=0.07647, over 3213792.36 frames. ], batch size: 146, lr: 1.62e-02, grad_scale: 4.0 2023-04-28 03:32:36,455 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7871, 3.8688, 3.3957, 2.5865, 3.0047, 2.2879, 4.1417, 4.3760], device='cuda:6'), covar=tensor([0.2014, 0.0709, 0.0987, 0.1157, 0.2141, 0.1256, 0.0349, 0.0308], device='cuda:6'), in_proj_covar=tensor([0.0268, 0.0240, 0.0252, 0.0228, 0.0298, 0.0192, 0.0226, 0.0228], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 03:32:50,253 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.590e+02 3.796e+02 5.362e+02 7.541e+02 3.116e+03, threshold=1.072e+03, percent-clipped=25.0 2023-04-28 03:32:56,434 INFO [train.py:904] (6/8) Epoch 4, batch 5500, loss[loss=0.3184, simple_loss=0.3786, pruned_loss=0.1291, over 15372.00 frames. ], tot_loss[loss=0.2442, simple_loss=0.3207, pruned_loss=0.08381, over 3175414.12 frames. ], batch size: 190, lr: 1.62e-02, grad_scale: 4.0 2023-04-28 03:33:19,318 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4876, 3.4414, 3.3454, 2.8796, 3.3671, 2.1413, 3.1967, 3.0900], device='cuda:6'), covar=tensor([0.0094, 0.0067, 0.0086, 0.0218, 0.0064, 0.1150, 0.0082, 0.0121], device='cuda:6'), in_proj_covar=tensor([0.0080, 0.0071, 0.0107, 0.0120, 0.0081, 0.0127, 0.0095, 0.0107], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 03:33:22,574 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-04-28 03:34:16,678 INFO [train.py:904] (6/8) Epoch 4, batch 5550, loss[loss=0.3093, simple_loss=0.3655, pruned_loss=0.1266, over 15218.00 frames. ], tot_loss[loss=0.2561, simple_loss=0.3299, pruned_loss=0.09112, over 3168588.53 frames. ], batch size: 190, lr: 1.62e-02, grad_scale: 4.0 2023-04-28 03:34:17,790 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36001.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 03:34:37,144 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9100, 3.9256, 3.1726, 1.7909, 2.7845, 2.2613, 3.4606, 3.6497], device='cuda:6'), covar=tensor([0.0276, 0.0365, 0.0540, 0.1633, 0.0722, 0.0902, 0.0611, 0.0498], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0121, 0.0153, 0.0142, 0.0135, 0.0126, 0.0141, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-28 03:35:02,199 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36030.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:35:29,646 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.087e+02 4.484e+02 5.536e+02 7.020e+02 1.224e+03, threshold=1.107e+03, percent-clipped=3.0 2023-04-28 03:35:33,035 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=36049.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 03:35:35,559 INFO [train.py:904] (6/8) Epoch 4, batch 5600, loss[loss=0.3075, simple_loss=0.3823, pruned_loss=0.1163, over 16227.00 frames. ], tot_loss[loss=0.2657, simple_loss=0.3372, pruned_loss=0.09707, over 3135849.56 frames. ], batch size: 165, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:36:20,419 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=36078.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:36:57,479 INFO [train.py:904] (6/8) Epoch 4, batch 5650, loss[loss=0.3824, simple_loss=0.4161, pruned_loss=0.1743, over 11481.00 frames. ], tot_loss[loss=0.2751, simple_loss=0.3442, pruned_loss=0.103, over 3097589.63 frames. ], batch size: 246, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:37:41,227 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36127.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:38:03,601 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7702, 5.2277, 5.3584, 5.2628, 5.3015, 5.7862, 5.3717, 5.1642], device='cuda:6'), covar=tensor([0.0833, 0.1489, 0.1186, 0.1367, 0.1792, 0.0758, 0.1134, 0.2199], device='cuda:6'), in_proj_covar=tensor([0.0255, 0.0351, 0.0336, 0.0312, 0.0404, 0.0363, 0.0285, 0.0420], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:38:09,922 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.856e+02 5.092e+02 6.431e+02 7.893e+02 1.649e+03, threshold=1.286e+03, percent-clipped=5.0 2023-04-28 03:38:17,753 INFO [train.py:904] (6/8) Epoch 4, batch 5700, loss[loss=0.2641, simple_loss=0.3522, pruned_loss=0.08803, over 16468.00 frames. ], tot_loss[loss=0.2777, simple_loss=0.3455, pruned_loss=0.105, over 3086348.82 frames. ], batch size: 75, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:38:37,455 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6541, 2.8229, 2.4164, 4.2002, 3.8345, 3.8274, 1.6790, 2.7761], device='cuda:6'), covar=tensor([0.1335, 0.0478, 0.1125, 0.0071, 0.0237, 0.0324, 0.1226, 0.0769], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0134, 0.0160, 0.0072, 0.0149, 0.0154, 0.0152, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 03:38:42,667 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36167.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:39:05,516 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36181.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:39:13,397 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36186.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:39:36,185 INFO [train.py:904] (6/8) Epoch 4, batch 5750, loss[loss=0.2827, simple_loss=0.3467, pruned_loss=0.1094, over 16535.00 frames. ], tot_loss[loss=0.2817, simple_loss=0.3488, pruned_loss=0.1073, over 3039016.03 frames. ], batch size: 75, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:39:50,129 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2356, 1.5311, 2.1576, 2.7976, 2.7663, 3.2962, 1.5512, 3.0773], device='cuda:6'), covar=tensor([0.0043, 0.0250, 0.0162, 0.0104, 0.0093, 0.0046, 0.0236, 0.0043], device='cuda:6'), in_proj_covar=tensor([0.0098, 0.0129, 0.0116, 0.0111, 0.0113, 0.0080, 0.0129, 0.0076], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 03:40:29,639 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=36234.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:40:37,036 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8139, 4.6196, 4.6420, 4.4940, 4.0955, 4.6283, 4.6707, 4.3561], device='cuda:6'), covar=tensor([0.0496, 0.0359, 0.0229, 0.0186, 0.0925, 0.0271, 0.0231, 0.0479], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0151, 0.0191, 0.0158, 0.0219, 0.0179, 0.0141, 0.0200], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:40:42,467 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36242.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:40:49,464 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.564e+02 4.215e+02 5.340e+02 6.535e+02 1.180e+03, threshold=1.068e+03, percent-clipped=0.0 2023-04-28 03:40:55,811 INFO [train.py:904] (6/8) Epoch 4, batch 5800, loss[loss=0.2438, simple_loss=0.322, pruned_loss=0.08284, over 16773.00 frames. ], tot_loss[loss=0.2806, simple_loss=0.3487, pruned_loss=0.1063, over 3028358.30 frames. ], batch size: 124, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:40:59,856 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-28 03:41:17,638 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2023-04-28 03:42:12,358 INFO [train.py:904] (6/8) Epoch 4, batch 5850, loss[loss=0.2727, simple_loss=0.3423, pruned_loss=0.1016, over 15381.00 frames. ], tot_loss[loss=0.2766, simple_loss=0.3458, pruned_loss=0.1037, over 3038120.01 frames. ], batch size: 190, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:43:29,064 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.296e+02 4.315e+02 5.493e+02 6.918e+02 1.909e+03, threshold=1.099e+03, percent-clipped=3.0 2023-04-28 03:43:33,995 INFO [train.py:904] (6/8) Epoch 4, batch 5900, loss[loss=0.3061, simple_loss=0.3564, pruned_loss=0.1279, over 11512.00 frames. ], tot_loss[loss=0.2742, simple_loss=0.3443, pruned_loss=0.1021, over 3056976.27 frames. ], batch size: 250, lr: 1.61e-02, grad_scale: 4.0 2023-04-28 03:44:56,229 INFO [train.py:904] (6/8) Epoch 4, batch 5950, loss[loss=0.297, simple_loss=0.3502, pruned_loss=0.1219, over 11508.00 frames. ], tot_loss[loss=0.273, simple_loss=0.3446, pruned_loss=0.1007, over 3047990.00 frames. ], batch size: 246, lr: 1.61e-02, grad_scale: 4.0 2023-04-28 03:45:37,494 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36427.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:46:10,130 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.426e+02 4.337e+02 5.242e+02 6.578e+02 1.706e+03, threshold=1.048e+03, percent-clipped=3.0 2023-04-28 03:46:14,570 INFO [train.py:904] (6/8) Epoch 4, batch 6000, loss[loss=0.3126, simple_loss=0.3725, pruned_loss=0.1263, over 11767.00 frames. ], tot_loss[loss=0.2716, simple_loss=0.3432, pruned_loss=0.1, over 3051850.13 frames. ], batch size: 247, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:46:14,570 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 03:46:25,217 INFO [train.py:938] (6/8) Epoch 4, validation: loss=0.1966, simple_loss=0.3069, pruned_loss=0.04309, over 944034.00 frames. 2023-04-28 03:46:25,218 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-28 03:46:38,340 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0836, 4.0205, 3.8988, 3.9049, 3.5491, 3.9927, 3.8489, 3.7448], device='cuda:6'), covar=tensor([0.0400, 0.0267, 0.0204, 0.0170, 0.0674, 0.0260, 0.0518, 0.0428], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0154, 0.0191, 0.0159, 0.0218, 0.0181, 0.0142, 0.0200], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:46:49,464 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36467.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:47:01,423 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=36475.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:47:44,055 INFO [train.py:904] (6/8) Epoch 4, batch 6050, loss[loss=0.2347, simple_loss=0.3153, pruned_loss=0.07702, over 16254.00 frames. ], tot_loss[loss=0.2691, simple_loss=0.341, pruned_loss=0.09861, over 3063180.13 frames. ], batch size: 35, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:48:02,620 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-28 03:48:06,439 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=36515.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:48:42,916 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36537.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:48:58,310 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.180e+02 4.133e+02 5.686e+02 7.938e+02 1.625e+03, threshold=1.137e+03, percent-clipped=9.0 2023-04-28 03:49:03,036 INFO [train.py:904] (6/8) Epoch 4, batch 6100, loss[loss=0.2499, simple_loss=0.3236, pruned_loss=0.08812, over 17288.00 frames. ], tot_loss[loss=0.2651, simple_loss=0.339, pruned_loss=0.0956, over 3104471.71 frames. ], batch size: 52, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:49:19,167 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3884, 1.8719, 1.4725, 1.6317, 2.2471, 2.0502, 2.3834, 2.4373], device='cuda:6'), covar=tensor([0.0030, 0.0198, 0.0241, 0.0233, 0.0098, 0.0177, 0.0063, 0.0100], device='cuda:6'), in_proj_covar=tensor([0.0069, 0.0141, 0.0144, 0.0141, 0.0132, 0.0144, 0.0101, 0.0121], device='cuda:6'), out_proj_covar=tensor([8.9031e-05, 1.8607e-04, 1.8272e-04, 1.8144e-04, 1.7507e-04, 1.8927e-04, 1.3031e-04, 1.6032e-04], device='cuda:6') 2023-04-28 03:50:22,075 INFO [train.py:904] (6/8) Epoch 4, batch 6150, loss[loss=0.2519, simple_loss=0.334, pruned_loss=0.0849, over 16725.00 frames. ], tot_loss[loss=0.2649, simple_loss=0.3381, pruned_loss=0.09589, over 3086866.86 frames. ], batch size: 89, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:50:29,628 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36605.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:50:35,016 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36608.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:50:41,724 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36612.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:51:38,876 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.478e+02 3.823e+02 4.755e+02 6.026e+02 1.290e+03, threshold=9.509e+02, percent-clipped=1.0 2023-04-28 03:51:44,328 INFO [train.py:904] (6/8) Epoch 4, batch 6200, loss[loss=0.2636, simple_loss=0.3371, pruned_loss=0.09502, over 16888.00 frames. ], tot_loss[loss=0.2634, simple_loss=0.3362, pruned_loss=0.09533, over 3092505.52 frames. ], batch size: 116, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:51:50,619 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-28 03:52:07,298 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36666.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:52:11,541 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36669.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:52:17,404 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9911, 4.8771, 4.6723, 3.8924, 4.7223, 1.7486, 4.5357, 4.7120], device='cuda:6'), covar=tensor([0.0048, 0.0048, 0.0077, 0.0313, 0.0048, 0.1569, 0.0064, 0.0098], device='cuda:6'), in_proj_covar=tensor([0.0082, 0.0070, 0.0109, 0.0120, 0.0080, 0.0131, 0.0096, 0.0109], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 03:52:17,466 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36673.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:52:28,060 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8887, 3.4164, 3.5028, 1.3980, 3.5925, 3.6313, 2.8181, 2.6444], device='cuda:6'), covar=tensor([0.0908, 0.0106, 0.0137, 0.1339, 0.0051, 0.0052, 0.0320, 0.0450], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0083, 0.0078, 0.0140, 0.0066, 0.0072, 0.0112, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 03:52:57,791 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36700.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:52:59,110 INFO [train.py:904] (6/8) Epoch 4, batch 6250, loss[loss=0.3538, simple_loss=0.4045, pruned_loss=0.1515, over 11635.00 frames. ], tot_loss[loss=0.2654, simple_loss=0.3375, pruned_loss=0.09667, over 3080486.86 frames. ], batch size: 246, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:53:35,488 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-28 03:54:09,138 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.662e+02 4.271e+02 5.056e+02 6.233e+02 1.402e+03, threshold=1.011e+03, percent-clipped=5.0 2023-04-28 03:54:14,280 INFO [train.py:904] (6/8) Epoch 4, batch 6300, loss[loss=0.2492, simple_loss=0.3226, pruned_loss=0.08786, over 16526.00 frames. ], tot_loss[loss=0.2636, simple_loss=0.3363, pruned_loss=0.09545, over 3090729.65 frames. ], batch size: 68, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:54:31,278 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36761.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 03:54:48,582 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 03:55:03,311 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1971, 5.0336, 4.9480, 4.8634, 4.4664, 4.9589, 5.0269, 4.6396], device='cuda:6'), covar=tensor([0.0375, 0.0258, 0.0180, 0.0155, 0.0858, 0.0271, 0.0174, 0.0356], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0159, 0.0197, 0.0163, 0.0225, 0.0189, 0.0146, 0.0209], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:55:32,004 INFO [train.py:904] (6/8) Epoch 4, batch 6350, loss[loss=0.2488, simple_loss=0.3281, pruned_loss=0.08474, over 17011.00 frames. ], tot_loss[loss=0.2665, simple_loss=0.3379, pruned_loss=0.09753, over 3084383.54 frames. ], batch size: 50, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:55:36,604 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.10 vs. limit=5.0 2023-04-28 03:55:37,439 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6907, 3.3352, 3.1917, 2.1702, 2.9906, 3.0560, 3.1529, 1.6647], device='cuda:6'), covar=tensor([0.0348, 0.0022, 0.0035, 0.0229, 0.0040, 0.0063, 0.0035, 0.0307], device='cuda:6'), in_proj_covar=tensor([0.0112, 0.0049, 0.0055, 0.0110, 0.0056, 0.0063, 0.0058, 0.0102], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:55:38,773 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36805.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:56:28,015 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36837.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:56:43,685 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.626e+02 4.373e+02 5.288e+02 7.014e+02 2.020e+03, threshold=1.058e+03, percent-clipped=6.0 2023-04-28 03:56:48,093 INFO [train.py:904] (6/8) Epoch 4, batch 6400, loss[loss=0.2398, simple_loss=0.3204, pruned_loss=0.07958, over 16798.00 frames. ], tot_loss[loss=0.2661, simple_loss=0.337, pruned_loss=0.09755, over 3088351.95 frames. ], batch size: 83, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:57:10,874 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36866.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 03:57:23,261 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-04-28 03:57:40,001 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=36885.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:58:03,613 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8871, 3.9173, 1.7677, 4.1247, 2.5716, 4.1491, 2.0259, 2.9911], device='cuda:6'), covar=tensor([0.0082, 0.0239, 0.1507, 0.0030, 0.0611, 0.0335, 0.1210, 0.0468], device='cuda:6'), in_proj_covar=tensor([0.0107, 0.0145, 0.0174, 0.0075, 0.0157, 0.0175, 0.0182, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 03:58:04,368 INFO [train.py:904] (6/8) Epoch 4, batch 6450, loss[loss=0.2353, simple_loss=0.3195, pruned_loss=0.07555, over 16904.00 frames. ], tot_loss[loss=0.264, simple_loss=0.3363, pruned_loss=0.09581, over 3099972.79 frames. ], batch size: 90, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:58:20,090 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36911.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:59:18,063 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.485e+02 3.772e+02 4.534e+02 6.144e+02 1.455e+03, threshold=9.067e+02, percent-clipped=1.0 2023-04-28 03:59:23,779 INFO [train.py:904] (6/8) Epoch 4, batch 6500, loss[loss=0.2721, simple_loss=0.3366, pruned_loss=0.1038, over 16146.00 frames. ], tot_loss[loss=0.2613, simple_loss=0.3334, pruned_loss=0.09453, over 3100834.84 frames. ], batch size: 165, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:59:32,103 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4414, 4.6333, 4.7031, 4.7239, 4.6250, 5.2153, 4.8491, 4.6246], device='cuda:6'), covar=tensor([0.0980, 0.1452, 0.1271, 0.1759, 0.2394, 0.0895, 0.1039, 0.2004], device='cuda:6'), in_proj_covar=tensor([0.0249, 0.0342, 0.0333, 0.0311, 0.0399, 0.0360, 0.0278, 0.0406], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 03:59:35,179 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36958.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:59:39,317 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36961.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:59:44,135 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36964.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:59:50,176 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36968.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:59:56,657 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36972.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:00:18,366 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3225, 1.3256, 1.8061, 2.1710, 2.3660, 2.4148, 1.4023, 2.5040], device='cuda:6'), covar=tensor([0.0061, 0.0233, 0.0147, 0.0122, 0.0088, 0.0083, 0.0210, 0.0052], device='cuda:6'), in_proj_covar=tensor([0.0102, 0.0133, 0.0122, 0.0114, 0.0119, 0.0083, 0.0131, 0.0076], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 04:00:44,408 INFO [train.py:904] (6/8) Epoch 4, batch 6550, loss[loss=0.3387, simple_loss=0.389, pruned_loss=0.1442, over 11664.00 frames. ], tot_loss[loss=0.2643, simple_loss=0.3369, pruned_loss=0.09583, over 3102279.97 frames. ], batch size: 247, lr: 1.59e-02, grad_scale: 8.0 2023-04-28 04:00:50,636 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37004.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:00:53,099 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2482, 3.8112, 3.8740, 1.7058, 4.0305, 4.0862, 3.0680, 3.1749], device='cuda:6'), covar=tensor([0.0701, 0.0083, 0.0138, 0.1053, 0.0039, 0.0046, 0.0242, 0.0298], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0083, 0.0080, 0.0140, 0.0066, 0.0072, 0.0112, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 04:01:13,075 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37019.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:01:30,564 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8995, 2.1992, 2.1767, 3.1821, 2.0627, 2.8282, 2.3042, 1.9563], device='cuda:6'), covar=tensor([0.0455, 0.1365, 0.0694, 0.0296, 0.2100, 0.0603, 0.1280, 0.1835], device='cuda:6'), in_proj_covar=tensor([0.0293, 0.0280, 0.0233, 0.0292, 0.0347, 0.0255, 0.0257, 0.0344], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 04:01:56,565 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.603e+02 4.116e+02 5.104e+02 6.571e+02 1.425e+03, threshold=1.021e+03, percent-clipped=7.0 2023-04-28 04:02:01,478 INFO [train.py:904] (6/8) Epoch 4, batch 6600, loss[loss=0.2676, simple_loss=0.3454, pruned_loss=0.0949, over 16734.00 frames. ], tot_loss[loss=0.2672, simple_loss=0.3398, pruned_loss=0.09729, over 3083354.68 frames. ], batch size: 134, lr: 1.59e-02, grad_scale: 8.0 2023-04-28 04:02:09,609 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37056.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 04:02:23,394 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37065.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:03:19,954 INFO [train.py:904] (6/8) Epoch 4, batch 6650, loss[loss=0.2312, simple_loss=0.3052, pruned_loss=0.07855, over 16439.00 frames. ], tot_loss[loss=0.2695, simple_loss=0.341, pruned_loss=0.099, over 3078218.26 frames. ], batch size: 68, lr: 1.59e-02, grad_scale: 8.0 2023-04-28 04:04:18,934 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5685, 2.3617, 1.9660, 2.2238, 2.9567, 2.6483, 3.5956, 3.3047], device='cuda:6'), covar=tensor([0.0017, 0.0175, 0.0209, 0.0200, 0.0087, 0.0164, 0.0043, 0.0083], device='cuda:6'), in_proj_covar=tensor([0.0067, 0.0137, 0.0142, 0.0137, 0.0131, 0.0143, 0.0101, 0.0120], device='cuda:6'), out_proj_covar=tensor([8.6206e-05, 1.7951e-04, 1.7959e-04, 1.7453e-04, 1.7121e-04, 1.8682e-04, 1.2884e-04, 1.5761e-04], device='cuda:6') 2023-04-28 04:04:33,021 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.678e+02 4.226e+02 5.482e+02 6.973e+02 1.276e+03, threshold=1.096e+03, percent-clipped=4.0 2023-04-28 04:04:37,775 INFO [train.py:904] (6/8) Epoch 4, batch 6700, loss[loss=0.3042, simple_loss=0.3753, pruned_loss=0.1165, over 16282.00 frames. ], tot_loss[loss=0.2679, simple_loss=0.3392, pruned_loss=0.09829, over 3081333.61 frames. ], batch size: 165, lr: 1.59e-02, grad_scale: 8.0 2023-04-28 04:04:53,201 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37161.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 04:05:47,885 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9669, 3.9555, 3.8289, 3.8187, 3.5076, 3.9279, 3.6417, 3.6345], device='cuda:6'), covar=tensor([0.0366, 0.0204, 0.0181, 0.0151, 0.0667, 0.0213, 0.0620, 0.0403], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0156, 0.0194, 0.0161, 0.0222, 0.0184, 0.0146, 0.0207], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 04:05:53,495 INFO [train.py:904] (6/8) Epoch 4, batch 6750, loss[loss=0.2506, simple_loss=0.3263, pruned_loss=0.08748, over 16754.00 frames. ], tot_loss[loss=0.2671, simple_loss=0.3378, pruned_loss=0.09817, over 3088714.26 frames. ], batch size: 124, lr: 1.59e-02, grad_scale: 8.0 2023-04-28 04:06:02,154 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8033, 2.7170, 2.6257, 1.6260, 2.6875, 2.7957, 2.4535, 2.2055], device='cuda:6'), covar=tensor([0.0903, 0.0126, 0.0148, 0.1068, 0.0099, 0.0082, 0.0363, 0.0495], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0083, 0.0079, 0.0141, 0.0067, 0.0070, 0.0112, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 04:06:12,820 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.96 vs. limit=5.0 2023-04-28 04:07:06,101 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.556e+02 4.104e+02 5.227e+02 6.545e+02 1.583e+03, threshold=1.045e+03, percent-clipped=5.0 2023-04-28 04:07:10,550 INFO [train.py:904] (6/8) Epoch 4, batch 6800, loss[loss=0.2658, simple_loss=0.3344, pruned_loss=0.09863, over 17023.00 frames. ], tot_loss[loss=0.2661, simple_loss=0.337, pruned_loss=0.09763, over 3074202.37 frames. ], batch size: 55, lr: 1.59e-02, grad_scale: 8.0 2023-04-28 04:07:27,213 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37261.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:07:31,552 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37264.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:07:35,433 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37267.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:07:37,393 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37268.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:07:37,439 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37268.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:07:53,649 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4020, 3.4561, 3.2703, 3.2854, 3.0291, 3.3508, 3.1838, 3.2337], device='cuda:6'), covar=tensor([0.0423, 0.0244, 0.0195, 0.0172, 0.0540, 0.0257, 0.0763, 0.0367], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0155, 0.0192, 0.0159, 0.0221, 0.0185, 0.0145, 0.0206], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 04:08:10,890 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-28 04:08:27,077 INFO [train.py:904] (6/8) Epoch 4, batch 6850, loss[loss=0.2527, simple_loss=0.3459, pruned_loss=0.07977, over 17284.00 frames. ], tot_loss[loss=0.2668, simple_loss=0.3385, pruned_loss=0.09759, over 3085867.69 frames. ], batch size: 52, lr: 1.59e-02, grad_scale: 4.0 2023-04-28 04:08:38,764 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37309.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:08:43,642 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37312.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:08:46,868 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37314.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:08:49,186 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37316.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:09:08,854 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37329.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:09:36,159 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.986e+02 4.056e+02 4.810e+02 6.733e+02 1.178e+03, threshold=9.620e+02, percent-clipped=3.0 2023-04-28 04:09:39,830 INFO [train.py:904] (6/8) Epoch 4, batch 6900, loss[loss=0.3374, simple_loss=0.3775, pruned_loss=0.1486, over 11185.00 frames. ], tot_loss[loss=0.2677, simple_loss=0.3409, pruned_loss=0.09727, over 3092117.72 frames. ], batch size: 246, lr: 1.59e-02, grad_scale: 4.0 2023-04-28 04:09:48,105 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37356.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 04:09:53,437 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37360.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:09:59,808 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37364.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:10:16,634 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6814, 3.6010, 2.9279, 1.7519, 2.5933, 2.1228, 3.0751, 3.3703], device='cuda:6'), covar=tensor([0.0304, 0.0479, 0.0574, 0.1718, 0.0824, 0.0975, 0.0786, 0.0637], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0123, 0.0154, 0.0143, 0.0137, 0.0127, 0.0144, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 04:10:55,352 INFO [train.py:904] (6/8) Epoch 4, batch 6950, loss[loss=0.3183, simple_loss=0.3735, pruned_loss=0.1315, over 11278.00 frames. ], tot_loss[loss=0.2734, simple_loss=0.3446, pruned_loss=0.1011, over 3073211.07 frames. ], batch size: 248, lr: 1.59e-02, grad_scale: 4.0 2023-04-28 04:11:00,664 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37404.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 04:11:32,953 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37425.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:12:06,736 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-28 04:12:10,354 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.259e+02 4.534e+02 5.752e+02 7.141e+02 1.491e+03, threshold=1.150e+03, percent-clipped=11.0 2023-04-28 04:12:11,785 INFO [train.py:904] (6/8) Epoch 4, batch 7000, loss[loss=0.2581, simple_loss=0.3414, pruned_loss=0.08737, over 16567.00 frames. ], tot_loss[loss=0.2716, simple_loss=0.344, pruned_loss=0.09963, over 3072299.91 frames. ], batch size: 68, lr: 1.58e-02, grad_scale: 2.0 2023-04-28 04:12:28,099 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37461.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 04:12:50,540 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9979, 4.1095, 1.9258, 4.5135, 2.6847, 4.4713, 2.2486, 2.8852], device='cuda:6'), covar=tensor([0.0102, 0.0219, 0.1467, 0.0028, 0.0756, 0.0213, 0.1396, 0.0652], device='cuda:6'), in_proj_covar=tensor([0.0109, 0.0145, 0.0173, 0.0075, 0.0157, 0.0171, 0.0181, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 04:13:29,621 INFO [train.py:904] (6/8) Epoch 4, batch 7050, loss[loss=0.2645, simple_loss=0.341, pruned_loss=0.09397, over 16734.00 frames. ], tot_loss[loss=0.2696, simple_loss=0.3435, pruned_loss=0.09781, over 3105897.57 frames. ], batch size: 134, lr: 1.58e-02, grad_scale: 2.0 2023-04-28 04:13:38,704 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-28 04:13:43,149 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37509.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 04:13:51,462 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6178, 3.5383, 2.9455, 1.7820, 2.4611, 2.0956, 2.9410, 3.3933], device='cuda:6'), covar=tensor([0.0303, 0.0446, 0.0575, 0.1594, 0.0822, 0.0888, 0.0792, 0.0607], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0123, 0.0155, 0.0143, 0.0137, 0.0128, 0.0145, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 04:14:31,217 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.57 vs. limit=5.0 2023-04-28 04:14:45,396 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 3.107e+02 4.211e+02 5.076e+02 6.594e+02 1.566e+03, threshold=1.015e+03, percent-clipped=2.0 2023-04-28 04:14:46,710 INFO [train.py:904] (6/8) Epoch 4, batch 7100, loss[loss=0.2425, simple_loss=0.3272, pruned_loss=0.07897, over 16461.00 frames. ], tot_loss[loss=0.2673, simple_loss=0.3407, pruned_loss=0.09693, over 3112257.35 frames. ], batch size: 68, lr: 1.58e-02, grad_scale: 2.0 2023-04-28 04:14:47,530 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-28 04:15:13,003 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37567.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:16:02,652 INFO [train.py:904] (6/8) Epoch 4, batch 7150, loss[loss=0.2726, simple_loss=0.3446, pruned_loss=0.1003, over 16312.00 frames. ], tot_loss[loss=0.2663, simple_loss=0.3391, pruned_loss=0.09678, over 3103340.57 frames. ], batch size: 165, lr: 1.58e-02, grad_scale: 2.0 2023-04-28 04:16:23,201 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37614.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:16:24,834 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37615.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:16:38,278 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37624.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:16:57,213 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8482, 2.1386, 1.6054, 1.8207, 2.5790, 2.3243, 2.8910, 2.8286], device='cuda:6'), covar=tensor([0.0030, 0.0191, 0.0245, 0.0233, 0.0103, 0.0175, 0.0060, 0.0095], device='cuda:6'), in_proj_covar=tensor([0.0066, 0.0137, 0.0143, 0.0139, 0.0132, 0.0144, 0.0101, 0.0120], device='cuda:6'), out_proj_covar=tensor([8.5092e-05, 1.7826e-04, 1.8096e-04, 1.7679e-04, 1.7196e-04, 1.8764e-04, 1.2786e-04, 1.5741e-04], device='cuda:6') 2023-04-28 04:17:17,119 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.158e+02 3.757e+02 4.855e+02 6.210e+02 1.451e+03, threshold=9.710e+02, percent-clipped=1.0 2023-04-28 04:17:18,898 INFO [train.py:904] (6/8) Epoch 4, batch 7200, loss[loss=0.2142, simple_loss=0.3036, pruned_loss=0.06236, over 16902.00 frames. ], tot_loss[loss=0.2623, simple_loss=0.336, pruned_loss=0.09431, over 3107499.91 frames. ], batch size: 96, lr: 1.58e-02, grad_scale: 4.0 2023-04-28 04:17:33,225 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37660.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:17:36,207 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37662.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:18:40,009 INFO [train.py:904] (6/8) Epoch 4, batch 7250, loss[loss=0.2357, simple_loss=0.3074, pruned_loss=0.08202, over 16926.00 frames. ], tot_loss[loss=0.2601, simple_loss=0.3334, pruned_loss=0.09333, over 3090298.98 frames. ], batch size: 109, lr: 1.58e-02, grad_scale: 4.0 2023-04-28 04:18:52,049 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37708.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:19:09,517 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.42 vs. limit=5.0 2023-04-28 04:19:10,398 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37720.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:19:55,472 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.396e+02 3.869e+02 4.810e+02 5.847e+02 1.256e+03, threshold=9.620e+02, percent-clipped=3.0 2023-04-28 04:19:57,415 INFO [train.py:904] (6/8) Epoch 4, batch 7300, loss[loss=0.2235, simple_loss=0.3109, pruned_loss=0.068, over 16506.00 frames. ], tot_loss[loss=0.2596, simple_loss=0.3331, pruned_loss=0.09308, over 3091479.81 frames. ], batch size: 75, lr: 1.58e-02, grad_scale: 4.0 2023-04-28 04:21:14,972 INFO [train.py:904] (6/8) Epoch 4, batch 7350, loss[loss=0.2143, simple_loss=0.2945, pruned_loss=0.06704, over 17105.00 frames. ], tot_loss[loss=0.2587, simple_loss=0.3322, pruned_loss=0.09263, over 3088887.75 frames. ], batch size: 47, lr: 1.58e-02, grad_scale: 4.0 2023-04-28 04:22:15,778 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4022, 3.3312, 3.3236, 2.7897, 3.3699, 2.1682, 3.1150, 3.0431], device='cuda:6'), covar=tensor([0.0077, 0.0056, 0.0082, 0.0200, 0.0048, 0.1180, 0.0080, 0.0112], device='cuda:6'), in_proj_covar=tensor([0.0082, 0.0070, 0.0106, 0.0117, 0.0078, 0.0131, 0.0094, 0.0105], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 04:22:29,890 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.483e+02 3.907e+02 4.910e+02 6.467e+02 1.073e+03, threshold=9.820e+02, percent-clipped=2.0 2023-04-28 04:22:31,804 INFO [train.py:904] (6/8) Epoch 4, batch 7400, loss[loss=0.3648, simple_loss=0.3977, pruned_loss=0.166, over 11103.00 frames. ], tot_loss[loss=0.2611, simple_loss=0.3344, pruned_loss=0.09392, over 3070618.98 frames. ], batch size: 246, lr: 1.58e-02, grad_scale: 4.0 2023-04-28 04:23:49,001 INFO [train.py:904] (6/8) Epoch 4, batch 7450, loss[loss=0.2342, simple_loss=0.3304, pruned_loss=0.06902, over 16847.00 frames. ], tot_loss[loss=0.2648, simple_loss=0.3367, pruned_loss=0.0964, over 3038605.35 frames. ], batch size: 83, lr: 1.58e-02, grad_scale: 4.0 2023-04-28 04:24:26,711 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37924.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:24:36,521 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7034, 3.4713, 3.1270, 1.9417, 2.6632, 2.0635, 3.2058, 3.2846], device='cuda:6'), covar=tensor([0.0223, 0.0509, 0.0471, 0.1520, 0.0727, 0.1012, 0.0603, 0.0708], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0120, 0.0154, 0.0140, 0.0135, 0.0126, 0.0142, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 04:25:06,399 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.829e+02 4.545e+02 5.686e+02 6.937e+02 1.455e+03, threshold=1.137e+03, percent-clipped=4.0 2023-04-28 04:25:07,720 INFO [train.py:904] (6/8) Epoch 4, batch 7500, loss[loss=0.2637, simple_loss=0.3378, pruned_loss=0.09485, over 16882.00 frames. ], tot_loss[loss=0.2651, simple_loss=0.3372, pruned_loss=0.09652, over 3022917.96 frames. ], batch size: 96, lr: 1.57e-02, grad_scale: 4.0 2023-04-28 04:25:40,565 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37972.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:25:50,847 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37979.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:26:08,619 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6863, 1.4346, 1.9810, 2.2762, 2.5008, 2.8001, 1.3712, 2.7946], device='cuda:6'), covar=tensor([0.0050, 0.0234, 0.0137, 0.0112, 0.0083, 0.0064, 0.0249, 0.0034], device='cuda:6'), in_proj_covar=tensor([0.0101, 0.0131, 0.0118, 0.0111, 0.0117, 0.0083, 0.0131, 0.0073], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 04:26:27,335 INFO [train.py:904] (6/8) Epoch 4, batch 7550, loss[loss=0.2254, simple_loss=0.312, pruned_loss=0.06935, over 16802.00 frames. ], tot_loss[loss=0.265, simple_loss=0.3365, pruned_loss=0.09668, over 3013722.55 frames. ], batch size: 83, lr: 1.57e-02, grad_scale: 4.0 2023-04-28 04:26:39,890 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.03 vs. limit=5.0 2023-04-28 04:26:54,909 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=38020.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:27:24,039 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38040.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:27:38,138 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.661e+02 3.786e+02 4.899e+02 6.266e+02 1.464e+03, threshold=9.797e+02, percent-clipped=2.0 2023-04-28 04:27:40,072 INFO [train.py:904] (6/8) Epoch 4, batch 7600, loss[loss=0.2249, simple_loss=0.3065, pruned_loss=0.07166, over 16462.00 frames. ], tot_loss[loss=0.2641, simple_loss=0.3354, pruned_loss=0.09638, over 3030068.85 frames. ], batch size: 68, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:27:52,608 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6960, 3.5086, 3.6246, 3.5426, 3.6521, 4.1003, 3.8739, 3.4714], device='cuda:6'), covar=tensor([0.1415, 0.1800, 0.1612, 0.2105, 0.2480, 0.1285, 0.1218, 0.2479], device='cuda:6'), in_proj_covar=tensor([0.0253, 0.0357, 0.0339, 0.0307, 0.0408, 0.0373, 0.0284, 0.0417], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 04:28:05,542 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=38068.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:28:10,813 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3448, 1.7762, 1.4595, 1.5181, 2.0920, 1.8980, 2.2083, 2.3119], device='cuda:6'), covar=tensor([0.0031, 0.0165, 0.0204, 0.0200, 0.0099, 0.0148, 0.0060, 0.0078], device='cuda:6'), in_proj_covar=tensor([0.0067, 0.0139, 0.0143, 0.0140, 0.0130, 0.0144, 0.0103, 0.0119], device='cuda:6'), out_proj_covar=tensor([8.6071e-05, 1.7933e-04, 1.8052e-04, 1.7645e-04, 1.6870e-04, 1.8687e-04, 1.2952e-04, 1.5569e-04], device='cuda:6') 2023-04-28 04:28:38,616 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1644, 4.1238, 4.0354, 4.0167, 3.5983, 4.0847, 3.9054, 3.8007], device='cuda:6'), covar=tensor([0.0464, 0.0278, 0.0211, 0.0181, 0.0778, 0.0303, 0.0556, 0.0472], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0156, 0.0190, 0.0158, 0.0220, 0.0187, 0.0144, 0.0204], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 04:28:54,621 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-28 04:28:55,058 INFO [train.py:904] (6/8) Epoch 4, batch 7650, loss[loss=0.2787, simple_loss=0.3493, pruned_loss=0.1041, over 16719.00 frames. ], tot_loss[loss=0.266, simple_loss=0.3371, pruned_loss=0.09742, over 3061473.95 frames. ], batch size: 134, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:29:05,627 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-28 04:30:08,813 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.664e+02 4.200e+02 5.231e+02 7.341e+02 1.286e+03, threshold=1.046e+03, percent-clipped=4.0 2023-04-28 04:30:09,971 INFO [train.py:904] (6/8) Epoch 4, batch 7700, loss[loss=0.2573, simple_loss=0.3344, pruned_loss=0.09006, over 16818.00 frames. ], tot_loss[loss=0.2674, simple_loss=0.338, pruned_loss=0.09845, over 3057252.50 frames. ], batch size: 102, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:31:20,835 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 04:31:26,736 INFO [train.py:904] (6/8) Epoch 4, batch 7750, loss[loss=0.2563, simple_loss=0.3357, pruned_loss=0.0884, over 16938.00 frames. ], tot_loss[loss=0.2657, simple_loss=0.3373, pruned_loss=0.09709, over 3077682.84 frames. ], batch size: 109, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:31:33,141 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0434, 3.8510, 4.0167, 4.2992, 4.3536, 3.9068, 4.3383, 4.3421], device='cuda:6'), covar=tensor([0.0990, 0.0781, 0.1379, 0.0507, 0.0498, 0.0924, 0.0554, 0.0438], device='cuda:6'), in_proj_covar=tensor([0.0342, 0.0412, 0.0539, 0.0427, 0.0320, 0.0311, 0.0347, 0.0347], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 04:32:40,399 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.200e+02 4.651e+02 5.629e+02 7.358e+02 1.215e+03, threshold=1.126e+03, percent-clipped=5.0 2023-04-28 04:32:42,173 INFO [train.py:904] (6/8) Epoch 4, batch 7800, loss[loss=0.2886, simple_loss=0.3576, pruned_loss=0.1098, over 16851.00 frames. ], tot_loss[loss=0.2669, simple_loss=0.3385, pruned_loss=0.09768, over 3102659.09 frames. ], batch size: 116, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:33:58,994 INFO [train.py:904] (6/8) Epoch 4, batch 7850, loss[loss=0.2347, simple_loss=0.3208, pruned_loss=0.07432, over 16851.00 frames. ], tot_loss[loss=0.2662, simple_loss=0.3385, pruned_loss=0.0969, over 3101288.41 frames. ], batch size: 102, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:34:05,072 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3945, 3.4506, 2.5759, 2.2516, 2.5349, 2.2172, 3.3505, 3.7132], device='cuda:6'), covar=tensor([0.2211, 0.0627, 0.1375, 0.1415, 0.2062, 0.1304, 0.0494, 0.0501], device='cuda:6'), in_proj_covar=tensor([0.0277, 0.0245, 0.0262, 0.0232, 0.0305, 0.0196, 0.0231, 0.0234], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 04:34:05,366 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-28 04:34:31,675 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8164, 3.9520, 3.0191, 2.6647, 3.0672, 2.4914, 4.0760, 4.3485], device='cuda:6'), covar=tensor([0.2026, 0.0705, 0.1305, 0.1376, 0.1852, 0.1179, 0.0455, 0.0449], device='cuda:6'), in_proj_covar=tensor([0.0274, 0.0243, 0.0260, 0.0231, 0.0302, 0.0194, 0.0230, 0.0231], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 04:34:50,566 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=38335.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:35:12,443 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.779e+02 3.909e+02 4.709e+02 5.998e+02 1.164e+03, threshold=9.418e+02, percent-clipped=1.0 2023-04-28 04:35:14,733 INFO [train.py:904] (6/8) Epoch 4, batch 7900, loss[loss=0.3378, simple_loss=0.3824, pruned_loss=0.1466, over 12017.00 frames. ], tot_loss[loss=0.2653, simple_loss=0.3375, pruned_loss=0.09656, over 3088417.64 frames. ], batch size: 246, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:35:28,550 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=38360.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:36:15,083 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2757, 3.7431, 3.8217, 1.5148, 3.9621, 4.0053, 2.9646, 2.9637], device='cuda:6'), covar=tensor([0.0791, 0.0096, 0.0118, 0.1365, 0.0054, 0.0053, 0.0326, 0.0377], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0085, 0.0084, 0.0144, 0.0071, 0.0076, 0.0116, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 04:36:34,681 INFO [train.py:904] (6/8) Epoch 4, batch 7950, loss[loss=0.28, simple_loss=0.3491, pruned_loss=0.1054, over 15389.00 frames. ], tot_loss[loss=0.2663, simple_loss=0.3381, pruned_loss=0.09732, over 3087492.25 frames. ], batch size: 191, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:37:05,059 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38421.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:37:14,509 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6279, 4.4260, 4.6703, 4.9248, 5.0184, 4.4131, 4.9919, 4.9596], device='cuda:6'), covar=tensor([0.0810, 0.0746, 0.1093, 0.0392, 0.0371, 0.0509, 0.0371, 0.0346], device='cuda:6'), in_proj_covar=tensor([0.0344, 0.0416, 0.0536, 0.0426, 0.0319, 0.0306, 0.0347, 0.0351], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 04:37:49,210 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.619e+02 4.198e+02 4.896e+02 6.041e+02 1.381e+03, threshold=9.792e+02, percent-clipped=5.0 2023-04-28 04:37:50,953 INFO [train.py:904] (6/8) Epoch 4, batch 8000, loss[loss=0.2582, simple_loss=0.3335, pruned_loss=0.0915, over 15647.00 frames. ], tot_loss[loss=0.2661, simple_loss=0.3376, pruned_loss=0.09725, over 3080033.55 frames. ], batch size: 190, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:39:04,864 INFO [train.py:904] (6/8) Epoch 4, batch 8050, loss[loss=0.2659, simple_loss=0.3478, pruned_loss=0.09199, over 17025.00 frames. ], tot_loss[loss=0.2653, simple_loss=0.3371, pruned_loss=0.09678, over 3081583.73 frames. ], batch size: 53, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:39:10,861 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.39 vs. limit=5.0 2023-04-28 04:39:51,908 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6666, 4.4010, 4.3620, 3.0885, 4.0436, 4.3258, 4.0748, 2.1898], device='cuda:6'), covar=tensor([0.0268, 0.0013, 0.0022, 0.0183, 0.0026, 0.0043, 0.0027, 0.0269], device='cuda:6'), in_proj_covar=tensor([0.0112, 0.0049, 0.0056, 0.0110, 0.0057, 0.0065, 0.0059, 0.0103], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 04:40:21,960 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.546e+02 4.083e+02 4.841e+02 6.395e+02 1.721e+03, threshold=9.682e+02, percent-clipped=6.0 2023-04-28 04:40:23,265 INFO [train.py:904] (6/8) Epoch 4, batch 8100, loss[loss=0.2744, simple_loss=0.346, pruned_loss=0.1014, over 16430.00 frames. ], tot_loss[loss=0.2644, simple_loss=0.3369, pruned_loss=0.09598, over 3104020.07 frames. ], batch size: 146, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:41:15,414 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7542, 2.4970, 2.4904, 4.5054, 1.9736, 3.5199, 2.4954, 2.3342], device='cuda:6'), covar=tensor([0.0491, 0.1447, 0.0824, 0.0220, 0.2598, 0.0574, 0.1491, 0.2215], device='cuda:6'), in_proj_covar=tensor([0.0299, 0.0287, 0.0238, 0.0298, 0.0359, 0.0263, 0.0263, 0.0353], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 04:41:41,522 INFO [train.py:904] (6/8) Epoch 4, batch 8150, loss[loss=0.2496, simple_loss=0.3231, pruned_loss=0.08799, over 16838.00 frames. ], tot_loss[loss=0.2612, simple_loss=0.3336, pruned_loss=0.09441, over 3111213.32 frames. ], batch size: 102, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:42:09,078 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0249, 3.3356, 3.2482, 2.1798, 3.0758, 3.2939, 3.1766, 1.6967], device='cuda:6'), covar=tensor([0.0312, 0.0025, 0.0041, 0.0240, 0.0046, 0.0060, 0.0043, 0.0308], device='cuda:6'), in_proj_covar=tensor([0.0116, 0.0050, 0.0058, 0.0112, 0.0057, 0.0067, 0.0061, 0.0106], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 04:42:34,012 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=38635.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:42:57,098 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.420e+02 4.459e+02 5.761e+02 7.393e+02 1.688e+03, threshold=1.152e+03, percent-clipped=6.0 2023-04-28 04:42:59,071 INFO [train.py:904] (6/8) Epoch 4, batch 8200, loss[loss=0.2603, simple_loss=0.336, pruned_loss=0.09225, over 16716.00 frames. ], tot_loss[loss=0.2599, simple_loss=0.3316, pruned_loss=0.09414, over 3088756.53 frames. ], batch size: 83, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:43:31,901 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0825, 4.0344, 3.9269, 3.8835, 3.4761, 3.9832, 3.7956, 3.6947], device='cuda:6'), covar=tensor([0.0421, 0.0275, 0.0235, 0.0172, 0.0920, 0.0270, 0.0525, 0.0493], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0159, 0.0192, 0.0159, 0.0220, 0.0185, 0.0145, 0.0205], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 04:43:53,409 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=38683.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:44:23,316 INFO [train.py:904] (6/8) Epoch 4, batch 8250, loss[loss=0.2353, simple_loss=0.3213, pruned_loss=0.0747, over 15279.00 frames. ], tot_loss[loss=0.2587, simple_loss=0.3317, pruned_loss=0.09282, over 3065593.85 frames. ], batch size: 191, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:44:40,986 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2023-04-28 04:44:49,014 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=38716.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:45:18,456 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5080, 4.4492, 4.3387, 4.2696, 3.8865, 4.4246, 4.2377, 4.1385], device='cuda:6'), covar=tensor([0.0321, 0.0255, 0.0161, 0.0136, 0.0772, 0.0208, 0.0290, 0.0330], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0154, 0.0188, 0.0156, 0.0214, 0.0180, 0.0142, 0.0201], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 04:45:43,964 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.138e+02 3.522e+02 4.130e+02 5.467e+02 1.123e+03, threshold=8.260e+02, percent-clipped=0.0 2023-04-28 04:45:45,838 INFO [train.py:904] (6/8) Epoch 4, batch 8300, loss[loss=0.2183, simple_loss=0.3083, pruned_loss=0.06415, over 16871.00 frames. ], tot_loss[loss=0.2515, simple_loss=0.3268, pruned_loss=0.08807, over 3055555.56 frames. ], batch size: 116, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:46:30,240 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8128, 3.4039, 3.3940, 2.3933, 3.3431, 3.3497, 3.3150, 1.8530], device='cuda:6'), covar=tensor([0.0324, 0.0017, 0.0026, 0.0193, 0.0022, 0.0037, 0.0027, 0.0286], device='cuda:6'), in_proj_covar=tensor([0.0113, 0.0049, 0.0056, 0.0110, 0.0056, 0.0065, 0.0060, 0.0103], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 04:46:35,605 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1419, 3.8472, 3.9379, 2.8942, 3.7822, 3.9343, 3.7590, 2.2568], device='cuda:6'), covar=tensor([0.0317, 0.0015, 0.0022, 0.0175, 0.0019, 0.0033, 0.0024, 0.0255], device='cuda:6'), in_proj_covar=tensor([0.0113, 0.0049, 0.0056, 0.0110, 0.0056, 0.0066, 0.0060, 0.0104], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 04:47:07,344 INFO [train.py:904] (6/8) Epoch 4, batch 8350, loss[loss=0.2082, simple_loss=0.3029, pruned_loss=0.05669, over 16837.00 frames. ], tot_loss[loss=0.2476, simple_loss=0.325, pruned_loss=0.08508, over 3058371.51 frames. ], batch size: 102, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:47:17,602 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6144, 3.6211, 4.0253, 4.0355, 4.0151, 3.7180, 3.7831, 3.7417], device='cuda:6'), covar=tensor([0.0251, 0.0373, 0.0319, 0.0328, 0.0347, 0.0299, 0.0687, 0.0343], device='cuda:6'), in_proj_covar=tensor([0.0211, 0.0201, 0.0209, 0.0213, 0.0253, 0.0221, 0.0317, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 04:47:44,457 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=38823.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:48:29,399 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.235e+02 3.176e+02 3.884e+02 4.976e+02 1.172e+03, threshold=7.768e+02, percent-clipped=5.0 2023-04-28 04:48:30,622 INFO [train.py:904] (6/8) Epoch 4, batch 8400, loss[loss=0.244, simple_loss=0.3243, pruned_loss=0.08184, over 16721.00 frames. ], tot_loss[loss=0.2435, simple_loss=0.3216, pruned_loss=0.08267, over 3033935.81 frames. ], batch size: 134, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:48:41,839 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8309, 3.0002, 3.1413, 2.2640, 2.9165, 3.0524, 2.9923, 1.8807], device='cuda:6'), covar=tensor([0.0321, 0.0021, 0.0027, 0.0191, 0.0033, 0.0048, 0.0035, 0.0274], device='cuda:6'), in_proj_covar=tensor([0.0111, 0.0048, 0.0055, 0.0108, 0.0055, 0.0064, 0.0059, 0.0102], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 04:49:08,759 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0523, 2.5814, 2.3123, 3.2846, 2.9308, 3.3191, 1.9432, 2.6694], device='cuda:6'), covar=tensor([0.1157, 0.0474, 0.0964, 0.0082, 0.0217, 0.0355, 0.1080, 0.0696], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0134, 0.0161, 0.0076, 0.0152, 0.0159, 0.0154, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 04:49:26,906 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38884.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:49:53,140 INFO [train.py:904] (6/8) Epoch 4, batch 8450, loss[loss=0.2265, simple_loss=0.3083, pruned_loss=0.07236, over 15225.00 frames. ], tot_loss[loss=0.2397, simple_loss=0.3189, pruned_loss=0.08025, over 3031173.87 frames. ], batch size: 190, lr: 1.56e-02, grad_scale: 4.0 2023-04-28 04:50:26,980 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3324, 4.3092, 4.3699, 4.4769, 4.4456, 4.9484, 4.5274, 4.2765], device='cuda:6'), covar=tensor([0.0886, 0.1594, 0.1139, 0.1560, 0.2271, 0.0877, 0.1139, 0.2077], device='cuda:6'), in_proj_covar=tensor([0.0240, 0.0334, 0.0328, 0.0290, 0.0388, 0.0355, 0.0277, 0.0394], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 04:51:13,821 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.235e+02 3.132e+02 4.045e+02 5.414e+02 1.265e+03, threshold=8.090e+02, percent-clipped=8.0 2023-04-28 04:51:13,836 INFO [train.py:904] (6/8) Epoch 4, batch 8500, loss[loss=0.2339, simple_loss=0.3088, pruned_loss=0.07949, over 16716.00 frames. ], tot_loss[loss=0.2342, simple_loss=0.3143, pruned_loss=0.07702, over 3033604.49 frames. ], batch size: 124, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 04:52:39,867 INFO [train.py:904] (6/8) Epoch 4, batch 8550, loss[loss=0.2541, simple_loss=0.3425, pruned_loss=0.08284, over 16203.00 frames. ], tot_loss[loss=0.2299, simple_loss=0.3107, pruned_loss=0.07455, over 3043241.34 frames. ], batch size: 165, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 04:53:09,933 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39016.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:54:21,289 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.792e+02 3.391e+02 3.909e+02 5.367e+02 1.114e+03, threshold=7.819e+02, percent-clipped=4.0 2023-04-28 04:54:21,304 INFO [train.py:904] (6/8) Epoch 4, batch 8600, loss[loss=0.2327, simple_loss=0.3172, pruned_loss=0.07408, over 15349.00 frames. ], tot_loss[loss=0.2296, simple_loss=0.3118, pruned_loss=0.0737, over 3061243.21 frames. ], batch size: 191, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 04:54:51,178 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=39064.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:55:57,997 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8182, 1.1644, 1.5541, 1.7187, 1.8606, 1.8405, 1.4755, 1.7056], device='cuda:6'), covar=tensor([0.0074, 0.0162, 0.0090, 0.0114, 0.0084, 0.0072, 0.0152, 0.0049], device='cuda:6'), in_proj_covar=tensor([0.0102, 0.0131, 0.0117, 0.0113, 0.0118, 0.0080, 0.0128, 0.0072], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 04:55:58,609 INFO [train.py:904] (6/8) Epoch 4, batch 8650, loss[loss=0.2155, simple_loss=0.3008, pruned_loss=0.06511, over 12323.00 frames. ], tot_loss[loss=0.2258, simple_loss=0.3089, pruned_loss=0.07135, over 3054857.39 frames. ], batch size: 248, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 04:57:44,932 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.040e+02 3.067e+02 3.859e+02 4.894e+02 7.251e+02, threshold=7.718e+02, percent-clipped=0.0 2023-04-28 04:57:44,948 INFO [train.py:904] (6/8) Epoch 4, batch 8700, loss[loss=0.2022, simple_loss=0.2833, pruned_loss=0.06057, over 12524.00 frames. ], tot_loss[loss=0.2216, simple_loss=0.3051, pruned_loss=0.06904, over 3059432.15 frames. ], batch size: 250, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 04:58:12,730 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-28 04:58:36,852 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39179.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:58:50,404 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39185.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:59:20,254 INFO [train.py:904] (6/8) Epoch 4, batch 8750, loss[loss=0.218, simple_loss=0.3012, pruned_loss=0.06738, over 12618.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.3043, pruned_loss=0.06872, over 3049658.62 frames. ], batch size: 248, lr: 1.55e-02, grad_scale: 2.0 2023-04-28 05:00:20,866 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-04-28 05:00:42,690 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5124, 3.7907, 1.6192, 3.9664, 2.4590, 3.9246, 2.0258, 2.8422], device='cuda:6'), covar=tensor([0.0110, 0.0181, 0.1713, 0.0034, 0.0892, 0.0297, 0.1456, 0.0604], device='cuda:6'), in_proj_covar=tensor([0.0107, 0.0138, 0.0173, 0.0076, 0.0156, 0.0164, 0.0180, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 05:01:05,176 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39246.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:01:14,444 INFO [train.py:904] (6/8) Epoch 4, batch 8800, loss[loss=0.2198, simple_loss=0.3155, pruned_loss=0.06209, over 16684.00 frames. ], tot_loss[loss=0.2179, simple_loss=0.3016, pruned_loss=0.06707, over 3051182.37 frames. ], batch size: 62, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 05:01:15,986 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.302e+02 3.583e+02 4.376e+02 5.208e+02 1.205e+03, threshold=8.753e+02, percent-clipped=6.0 2023-04-28 05:02:23,716 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39284.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:02:58,289 INFO [train.py:904] (6/8) Epoch 4, batch 8850, loss[loss=0.2275, simple_loss=0.3204, pruned_loss=0.06725, over 16289.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.3046, pruned_loss=0.06641, over 3056215.13 frames. ], batch size: 165, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 05:04:22,060 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8190, 5.1655, 4.8871, 4.9343, 4.4087, 4.2733, 4.5590, 5.1409], device='cuda:6'), covar=tensor([0.0610, 0.0611, 0.0766, 0.0357, 0.0637, 0.0757, 0.0543, 0.0690], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0413, 0.0355, 0.0270, 0.0269, 0.0280, 0.0337, 0.0300], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 05:04:32,517 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39345.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:04:44,178 INFO [train.py:904] (6/8) Epoch 4, batch 8900, loss[loss=0.2484, simple_loss=0.332, pruned_loss=0.08246, over 15412.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.3034, pruned_loss=0.06493, over 3042709.35 frames. ], batch size: 190, lr: 1.55e-02, grad_scale: 2.0 2023-04-28 05:04:49,516 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.906e+02 3.441e+02 4.038e+02 4.851e+02 9.886e+02, threshold=8.076e+02, percent-clipped=2.0 2023-04-28 05:05:25,665 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0735, 4.1491, 3.9138, 3.8255, 3.5469, 4.0179, 3.8446, 3.7346], device='cuda:6'), covar=tensor([0.0388, 0.0183, 0.0214, 0.0159, 0.0657, 0.0249, 0.0442, 0.0376], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0149, 0.0186, 0.0151, 0.0202, 0.0178, 0.0136, 0.0196], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 05:05:31,115 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1876, 3.0756, 2.8425, 1.9816, 2.6502, 2.1135, 2.7064, 2.8264], device='cuda:6'), covar=tensor([0.0229, 0.0344, 0.0450, 0.1293, 0.0561, 0.0851, 0.0538, 0.0601], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0115, 0.0155, 0.0141, 0.0134, 0.0128, 0.0138, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-28 05:06:47,800 INFO [train.py:904] (6/8) Epoch 4, batch 8950, loss[loss=0.2263, simple_loss=0.3117, pruned_loss=0.07044, over 12592.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.3036, pruned_loss=0.06548, over 3045524.72 frames. ], batch size: 248, lr: 1.55e-02, grad_scale: 2.0 2023-04-28 05:06:51,629 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5012, 2.1660, 1.8312, 1.8464, 2.4769, 2.2660, 2.7298, 2.6110], device='cuda:6'), covar=tensor([0.0025, 0.0172, 0.0212, 0.0227, 0.0101, 0.0171, 0.0052, 0.0108], device='cuda:6'), in_proj_covar=tensor([0.0066, 0.0141, 0.0142, 0.0140, 0.0133, 0.0143, 0.0099, 0.0118], device='cuda:6'), out_proj_covar=tensor([8.2721e-05, 1.8171e-04, 1.7709e-04, 1.7616e-04, 1.7155e-04, 1.8422e-04, 1.2209e-04, 1.5158e-04], device='cuda:6') 2023-04-28 05:08:35,741 INFO [train.py:904] (6/8) Epoch 4, batch 9000, loss[loss=0.1982, simple_loss=0.2838, pruned_loss=0.05626, over 16324.00 frames. ], tot_loss[loss=0.2138, simple_loss=0.3001, pruned_loss=0.06377, over 3043439.35 frames. ], batch size: 146, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:08:35,742 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 05:08:45,782 INFO [train.py:938] (6/8) Epoch 4, validation: loss=0.1802, simple_loss=0.283, pruned_loss=0.0387, over 944034.00 frames. 2023-04-28 05:08:45,782 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-28 05:08:49,857 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.055e+02 3.044e+02 3.757e+02 4.852e+02 9.817e+02, threshold=7.514e+02, percent-clipped=1.0 2023-04-28 05:09:11,771 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39463.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:09:43,593 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39479.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:10:29,884 INFO [train.py:904] (6/8) Epoch 4, batch 9050, loss[loss=0.1984, simple_loss=0.2949, pruned_loss=0.05096, over 16633.00 frames. ], tot_loss[loss=0.2159, simple_loss=0.3019, pruned_loss=0.06493, over 3054449.70 frames. ], batch size: 89, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:10:52,481 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-28 05:11:17,816 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39524.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:11:21,928 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=39527.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:11:54,712 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39541.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:12:14,626 INFO [train.py:904] (6/8) Epoch 4, batch 9100, loss[loss=0.2263, simple_loss=0.3166, pruned_loss=0.06797, over 16258.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.3016, pruned_loss=0.06531, over 3079876.68 frames. ], batch size: 166, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:12:18,749 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.564e+02 3.376e+02 4.042e+02 5.182e+02 1.418e+03, threshold=8.084e+02, percent-clipped=5.0 2023-04-28 05:14:15,379 INFO [train.py:904] (6/8) Epoch 4, batch 9150, loss[loss=0.1815, simple_loss=0.2745, pruned_loss=0.04425, over 16920.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.3023, pruned_loss=0.06545, over 3069260.05 frames. ], batch size: 90, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:15:41,151 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39639.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 05:15:43,231 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39640.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:16:00,925 INFO [train.py:904] (6/8) Epoch 4, batch 9200, loss[loss=0.1761, simple_loss=0.2585, pruned_loss=0.04688, over 12346.00 frames. ], tot_loss[loss=0.2129, simple_loss=0.2974, pruned_loss=0.06416, over 3051566.58 frames. ], batch size: 248, lr: 1.54e-02, grad_scale: 4.0 2023-04-28 05:16:04,310 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.381e+02 3.414e+02 4.371e+02 6.281e+02 1.474e+03, threshold=8.741e+02, percent-clipped=12.0 2023-04-28 05:17:35,028 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39700.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 05:17:35,637 INFO [train.py:904] (6/8) Epoch 4, batch 9250, loss[loss=0.212, simple_loss=0.2871, pruned_loss=0.06844, over 12242.00 frames. ], tot_loss[loss=0.2127, simple_loss=0.2971, pruned_loss=0.06414, over 3050819.10 frames. ], batch size: 248, lr: 1.54e-02, grad_scale: 4.0 2023-04-28 05:18:04,927 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8899, 4.8555, 5.4580, 5.4240, 5.3494, 5.0395, 5.0074, 4.6815], device='cuda:6'), covar=tensor([0.0210, 0.0324, 0.0283, 0.0302, 0.0362, 0.0234, 0.0590, 0.0273], device='cuda:6'), in_proj_covar=tensor([0.0203, 0.0194, 0.0202, 0.0202, 0.0237, 0.0215, 0.0294, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-28 05:18:04,955 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0452, 4.7980, 5.0172, 5.2938, 5.3868, 4.6718, 5.4264, 5.3439], device='cuda:6'), covar=tensor([0.0734, 0.0593, 0.0985, 0.0400, 0.0417, 0.0440, 0.0305, 0.0340], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0387, 0.0489, 0.0396, 0.0301, 0.0291, 0.0321, 0.0323], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 05:18:21,982 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-28 05:18:47,990 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1134, 4.0363, 3.5227, 1.8684, 2.7185, 2.4298, 3.2890, 3.6930], device='cuda:6'), covar=tensor([0.0261, 0.0438, 0.0474, 0.1636, 0.0801, 0.0874, 0.0785, 0.0668], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0114, 0.0154, 0.0142, 0.0134, 0.0128, 0.0141, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-28 05:18:56,811 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7682, 3.6295, 3.7344, 3.7067, 3.8273, 4.2203, 3.9577, 3.6235], device='cuda:6'), covar=tensor([0.1591, 0.1937, 0.1342, 0.1995, 0.2853, 0.1223, 0.1218, 0.2671], device='cuda:6'), in_proj_covar=tensor([0.0236, 0.0335, 0.0326, 0.0292, 0.0378, 0.0352, 0.0276, 0.0388], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 05:19:26,129 INFO [train.py:904] (6/8) Epoch 4, batch 9300, loss[loss=0.1872, simple_loss=0.2866, pruned_loss=0.04392, over 17050.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.2951, pruned_loss=0.06277, over 3058173.05 frames. ], batch size: 50, lr: 1.54e-02, grad_scale: 4.0 2023-04-28 05:19:30,019 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.165e+02 3.034e+02 3.554e+02 4.321e+02 7.893e+02, threshold=7.107e+02, percent-clipped=0.0 2023-04-28 05:20:06,133 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39767.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:21:11,784 INFO [train.py:904] (6/8) Epoch 4, batch 9350, loss[loss=0.2128, simple_loss=0.2947, pruned_loss=0.06548, over 16250.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2951, pruned_loss=0.06238, over 3078354.67 frames. ], batch size: 165, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:21:41,298 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39815.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:21:48,590 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39819.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:22:05,644 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39828.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:22:30,661 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39841.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:22:51,022 INFO [train.py:904] (6/8) Epoch 4, batch 9400, loss[loss=0.2044, simple_loss=0.298, pruned_loss=0.05545, over 15275.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2951, pruned_loss=0.0622, over 3081784.30 frames. ], batch size: 190, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:22:57,590 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.054e+02 3.057e+02 3.854e+02 4.791e+02 1.161e+03, threshold=7.709e+02, percent-clipped=3.0 2023-04-28 05:23:42,103 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39876.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:24:09,958 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=39889.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:24:32,546 INFO [train.py:904] (6/8) Epoch 4, batch 9450, loss[loss=0.2015, simple_loss=0.2877, pruned_loss=0.05761, over 16869.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2966, pruned_loss=0.06268, over 3057804.23 frames. ], batch size: 116, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:25:54,129 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39940.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:26:13,856 INFO [train.py:904] (6/8) Epoch 4, batch 9500, loss[loss=0.2047, simple_loss=0.2949, pruned_loss=0.05721, over 16680.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2958, pruned_loss=0.06214, over 3058611.63 frames. ], batch size: 57, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:26:21,219 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.115e+02 3.608e+02 4.489e+02 5.705e+02 9.491e+02, threshold=8.979e+02, percent-clipped=6.0 2023-04-28 05:26:54,733 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4079, 1.9791, 1.6289, 1.6471, 2.4409, 2.1334, 2.5686, 2.5758], device='cuda:6'), covar=tensor([0.0023, 0.0170, 0.0226, 0.0222, 0.0100, 0.0171, 0.0061, 0.0093], device='cuda:6'), in_proj_covar=tensor([0.0063, 0.0141, 0.0142, 0.0139, 0.0134, 0.0143, 0.0098, 0.0115], device='cuda:6'), out_proj_covar=tensor([7.8302e-05, 1.7938e-04, 1.7716e-04, 1.7377e-04, 1.7238e-04, 1.8305e-04, 1.1958e-04, 1.4677e-04], device='cuda:6') 2023-04-28 05:27:31,419 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=39988.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:27:48,827 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39995.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 05:28:04,405 INFO [train.py:904] (6/8) Epoch 4, batch 9550, loss[loss=0.2415, simple_loss=0.3281, pruned_loss=0.07751, over 15464.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2963, pruned_loss=0.06282, over 3065078.58 frames. ], batch size: 191, lr: 1.53e-02, grad_scale: 2.0 2023-04-28 05:29:35,812 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4426, 2.0258, 1.6767, 1.5616, 2.3790, 2.1557, 2.5651, 2.5465], device='cuda:6'), covar=tensor([0.0023, 0.0186, 0.0223, 0.0254, 0.0106, 0.0164, 0.0071, 0.0087], device='cuda:6'), in_proj_covar=tensor([0.0064, 0.0140, 0.0141, 0.0139, 0.0133, 0.0141, 0.0098, 0.0114], device='cuda:6'), out_proj_covar=tensor([7.8446e-05, 1.7819e-04, 1.7529e-04, 1.7331e-04, 1.7078e-04, 1.8099e-04, 1.1969e-04, 1.4541e-04], device='cuda:6') 2023-04-28 05:29:46,549 INFO [train.py:904] (6/8) Epoch 4, batch 9600, loss[loss=0.2071, simple_loss=0.2972, pruned_loss=0.05847, over 16575.00 frames. ], tot_loss[loss=0.213, simple_loss=0.298, pruned_loss=0.06403, over 3053992.89 frames. ], batch size: 68, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:29:52,056 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.278e+02 3.502e+02 4.489e+02 5.494e+02 1.109e+03, threshold=8.977e+02, percent-clipped=3.0 2023-04-28 05:31:16,048 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-28 05:31:33,069 INFO [train.py:904] (6/8) Epoch 4, batch 9650, loss[loss=0.2426, simple_loss=0.3232, pruned_loss=0.08099, over 16237.00 frames. ], tot_loss[loss=0.2149, simple_loss=0.3006, pruned_loss=0.0646, over 3067793.19 frames. ], batch size: 165, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:32:18,232 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40119.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:32:25,199 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40123.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:33:05,533 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0283, 2.8006, 2.7052, 1.7539, 2.9292, 2.9616, 2.6210, 2.3838], device='cuda:6'), covar=tensor([0.0645, 0.0140, 0.0154, 0.0895, 0.0082, 0.0078, 0.0301, 0.0387], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0080, 0.0074, 0.0136, 0.0067, 0.0070, 0.0106, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 05:33:10,274 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0040, 3.7749, 4.0291, 4.2389, 4.3131, 3.8117, 4.3062, 4.2961], device='cuda:6'), covar=tensor([0.0820, 0.0686, 0.0953, 0.0422, 0.0364, 0.0937, 0.0431, 0.0383], device='cuda:6'), in_proj_covar=tensor([0.0336, 0.0401, 0.0504, 0.0414, 0.0305, 0.0298, 0.0331, 0.0337], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 05:33:21,181 INFO [train.py:904] (6/8) Epoch 4, batch 9700, loss[loss=0.1847, simple_loss=0.2796, pruned_loss=0.04492, over 16780.00 frames. ], tot_loss[loss=0.2132, simple_loss=0.299, pruned_loss=0.06368, over 3076979.14 frames. ], batch size: 83, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:33:26,548 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.999e+02 3.154e+02 3.870e+02 5.322e+02 1.510e+03, threshold=7.740e+02, percent-clipped=2.0 2023-04-28 05:33:53,099 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40167.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:34:01,353 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40171.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:34:32,602 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-28 05:34:54,793 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40195.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:35:03,567 INFO [train.py:904] (6/8) Epoch 4, batch 9750, loss[loss=0.1999, simple_loss=0.2738, pruned_loss=0.06297, over 12241.00 frames. ], tot_loss[loss=0.2128, simple_loss=0.2977, pruned_loss=0.06393, over 3067458.50 frames. ], batch size: 249, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:35:28,222 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-28 05:35:38,806 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40218.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:36:45,110 INFO [train.py:904] (6/8) Epoch 4, batch 9800, loss[loss=0.2226, simple_loss=0.3309, pruned_loss=0.0571, over 16755.00 frames. ], tot_loss[loss=0.2105, simple_loss=0.2967, pruned_loss=0.06214, over 3075467.62 frames. ], batch size: 83, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:36:51,067 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.012e+02 3.170e+02 3.844e+02 4.650e+02 7.827e+02, threshold=7.689e+02, percent-clipped=1.0 2023-04-28 05:36:56,652 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40256.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:37:25,560 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40272.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:37:37,840 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40279.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:38:18,098 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40295.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 05:38:29,681 INFO [train.py:904] (6/8) Epoch 4, batch 9850, loss[loss=0.2196, simple_loss=0.3058, pruned_loss=0.06675, over 15446.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.2976, pruned_loss=0.06197, over 3055692.05 frames. ], batch size: 191, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:39:31,332 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40330.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:39:39,202 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40333.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:40:05,939 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40343.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 05:40:17,525 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8220, 3.6118, 3.2743, 1.8250, 2.6687, 2.3076, 3.0876, 3.3276], device='cuda:6'), covar=tensor([0.0223, 0.0365, 0.0394, 0.1426, 0.0683, 0.0801, 0.0646, 0.0585], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0112, 0.0154, 0.0142, 0.0133, 0.0126, 0.0137, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-28 05:40:21,695 INFO [train.py:904] (6/8) Epoch 4, batch 9900, loss[loss=0.2138, simple_loss=0.3069, pruned_loss=0.06035, over 16922.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2986, pruned_loss=0.0619, over 3066620.43 frames. ], batch size: 116, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:40:27,910 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.084e+02 3.210e+02 3.891e+02 5.069e+02 8.589e+02, threshold=7.781e+02, percent-clipped=1.0 2023-04-28 05:41:56,842 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40391.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:42:03,007 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-04-28 05:42:18,075 INFO [train.py:904] (6/8) Epoch 4, batch 9950, loss[loss=0.186, simple_loss=0.2785, pruned_loss=0.04677, over 16587.00 frames. ], tot_loss[loss=0.2129, simple_loss=0.3008, pruned_loss=0.0625, over 3059419.68 frames. ], batch size: 62, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:43:13,866 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40423.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:44:20,283 INFO [train.py:904] (6/8) Epoch 4, batch 10000, loss[loss=0.2026, simple_loss=0.2991, pruned_loss=0.05304, over 15580.00 frames. ], tot_loss[loss=0.2105, simple_loss=0.2981, pruned_loss=0.06138, over 3071604.60 frames. ], batch size: 192, lr: 1.53e-02, grad_scale: 8.0 2023-04-28 05:44:26,639 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.009e+02 3.138e+02 3.849e+02 4.764e+02 1.083e+03, threshold=7.697e+02, percent-clipped=5.0 2023-04-28 05:44:44,800 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2431, 3.5004, 1.5851, 3.6023, 2.2807, 3.5546, 1.7624, 2.6104], device='cuda:6'), covar=tensor([0.0128, 0.0251, 0.1677, 0.0061, 0.0913, 0.0440, 0.1642, 0.0688], device='cuda:6'), in_proj_covar=tensor([0.0109, 0.0136, 0.0171, 0.0077, 0.0154, 0.0163, 0.0181, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 05:45:01,303 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40471.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:45:01,357 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40471.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:45:47,312 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40492.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:46:02,920 INFO [train.py:904] (6/8) Epoch 4, batch 10050, loss[loss=0.2208, simple_loss=0.3042, pruned_loss=0.06874, over 16933.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2979, pruned_loss=0.06104, over 3085239.73 frames. ], batch size: 109, lr: 1.53e-02, grad_scale: 8.0 2023-04-28 05:46:07,695 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2023-04-28 05:46:10,301 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3979, 4.3939, 4.4794, 4.5450, 4.5230, 5.0410, 4.7653, 4.3557], device='cuda:6'), covar=tensor([0.0803, 0.1529, 0.1476, 0.1755, 0.2680, 0.1101, 0.1218, 0.2212], device='cuda:6'), in_proj_covar=tensor([0.0232, 0.0335, 0.0328, 0.0293, 0.0383, 0.0356, 0.0270, 0.0386], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 05:46:38,833 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40519.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:47:38,802 INFO [train.py:904] (6/8) Epoch 4, batch 10100, loss[loss=0.2065, simple_loss=0.2915, pruned_loss=0.06071, over 16309.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2987, pruned_loss=0.06196, over 3084047.36 frames. ], batch size: 165, lr: 1.52e-02, grad_scale: 8.0 2023-04-28 05:47:39,369 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40551.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:47:42,199 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40553.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:47:42,814 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.191e+02 3.527e+02 4.401e+02 5.596e+02 1.013e+03, threshold=8.802e+02, percent-clipped=4.0 2023-04-28 05:48:27,647 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40574.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:48:41,088 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0118, 2.0556, 2.1987, 3.2148, 1.9734, 2.6840, 2.2071, 1.8745], device='cuda:6'), covar=tensor([0.0470, 0.1689, 0.0745, 0.0325, 0.2517, 0.0741, 0.1591, 0.2378], device='cuda:6'), in_proj_covar=tensor([0.0288, 0.0284, 0.0231, 0.0287, 0.0343, 0.0256, 0.0259, 0.0327], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 05:49:25,172 INFO [train.py:904] (6/8) Epoch 5, batch 0, loss[loss=0.3772, simple_loss=0.3949, pruned_loss=0.1798, over 16698.00 frames. ], tot_loss[loss=0.3772, simple_loss=0.3949, pruned_loss=0.1798, over 16698.00 frames. ], batch size: 134, lr: 1.42e-02, grad_scale: 8.0 2023-04-28 05:49:25,172 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 05:49:32,549 INFO [train.py:938] (6/8) Epoch 5, validation: loss=0.1789, simple_loss=0.2817, pruned_loss=0.03802, over 944034.00 frames. 2023-04-28 05:49:32,549 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-28 05:49:49,566 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4430, 3.4093, 2.6295, 2.3104, 2.5147, 2.2208, 3.3765, 3.4024], device='cuda:6'), covar=tensor([0.1711, 0.0516, 0.1013, 0.1198, 0.1511, 0.1228, 0.0365, 0.0554], device='cuda:6'), in_proj_covar=tensor([0.0259, 0.0238, 0.0253, 0.0227, 0.0234, 0.0190, 0.0220, 0.0219], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 05:50:11,883 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40628.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:50:42,807 INFO [train.py:904] (6/8) Epoch 5, batch 50, loss[loss=0.2441, simple_loss=0.3083, pruned_loss=0.08992, over 16848.00 frames. ], tot_loss[loss=0.2616, simple_loss=0.3256, pruned_loss=0.09881, over 748735.19 frames. ], batch size: 102, lr: 1.42e-02, grad_scale: 1.0 2023-04-28 05:50:43,302 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40651.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:50:49,840 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.398e+02 3.847e+02 4.852e+02 6.011e+02 1.299e+03, threshold=9.705e+02, percent-clipped=3.0 2023-04-28 05:50:59,519 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-28 05:51:21,205 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0401, 4.3280, 4.4707, 2.1431, 4.8143, 4.7972, 3.5806, 3.8209], device='cuda:6'), covar=tensor([0.0629, 0.0105, 0.0168, 0.1009, 0.0031, 0.0039, 0.0247, 0.0259], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0082, 0.0076, 0.0140, 0.0068, 0.0072, 0.0109, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 05:51:29,207 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40686.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:51:50,912 INFO [train.py:904] (6/8) Epoch 5, batch 100, loss[loss=0.2509, simple_loss=0.3116, pruned_loss=0.09503, over 16793.00 frames. ], tot_loss[loss=0.2458, simple_loss=0.315, pruned_loss=0.08833, over 1325590.02 frames. ], batch size: 124, lr: 1.42e-02, grad_scale: 1.0 2023-04-28 05:52:07,375 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40712.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:52:34,012 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4387, 3.4085, 3.2914, 2.9552, 3.3863, 1.9915, 3.2797, 2.8214], device='cuda:6'), covar=tensor([0.0073, 0.0062, 0.0089, 0.0197, 0.0061, 0.1424, 0.0082, 0.0133], device='cuda:6'), in_proj_covar=tensor([0.0084, 0.0070, 0.0111, 0.0110, 0.0082, 0.0136, 0.0097, 0.0106], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 05:52:59,301 INFO [train.py:904] (6/8) Epoch 5, batch 150, loss[loss=0.2641, simple_loss=0.3198, pruned_loss=0.1042, over 16735.00 frames. ], tot_loss[loss=0.2426, simple_loss=0.313, pruned_loss=0.08605, over 1759274.70 frames. ], batch size: 124, lr: 1.42e-02, grad_scale: 1.0 2023-04-28 05:53:08,165 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.263e+02 3.680e+02 4.214e+02 5.175e+02 1.144e+03, threshold=8.427e+02, percent-clipped=1.0 2023-04-28 05:53:46,382 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8617, 5.3827, 5.4231, 5.3157, 5.3462, 5.8823, 5.4964, 5.2455], device='cuda:6'), covar=tensor([0.0696, 0.1352, 0.1222, 0.1549, 0.2327, 0.0917, 0.1050, 0.2073], device='cuda:6'), in_proj_covar=tensor([0.0258, 0.0377, 0.0362, 0.0327, 0.0426, 0.0385, 0.0297, 0.0433], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 05:54:09,224 INFO [train.py:904] (6/8) Epoch 5, batch 200, loss[loss=0.2912, simple_loss=0.3361, pruned_loss=0.1232, over 16720.00 frames. ], tot_loss[loss=0.242, simple_loss=0.3126, pruned_loss=0.08576, over 2100195.61 frames. ], batch size: 134, lr: 1.42e-02, grad_scale: 1.0 2023-04-28 05:55:02,257 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0658, 1.5578, 2.1321, 2.8577, 2.9587, 3.1063, 1.7254, 2.9263], device='cuda:6'), covar=tensor([0.0053, 0.0209, 0.0134, 0.0110, 0.0065, 0.0064, 0.0181, 0.0052], device='cuda:6'), in_proj_covar=tensor([0.0107, 0.0137, 0.0122, 0.0120, 0.0118, 0.0085, 0.0134, 0.0074], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 05:55:13,932 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:55:17,642 INFO [train.py:904] (6/8) Epoch 5, batch 250, loss[loss=0.2673, simple_loss=0.3314, pruned_loss=0.1016, over 16985.00 frames. ], tot_loss[loss=0.2393, simple_loss=0.3095, pruned_loss=0.08458, over 2374301.51 frames. ], batch size: 109, lr: 1.41e-02, grad_scale: 1.0 2023-04-28 05:55:18,020 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40851.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:55:25,557 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.087e+02 3.707e+02 4.511e+02 5.174e+02 8.449e+02, threshold=9.022e+02, percent-clipped=1.0 2023-04-28 05:55:38,002 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40865.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:55:44,645 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-28 05:55:50,484 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40874.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:56:24,969 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40899.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:56:27,000 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40900.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:56:27,682 INFO [train.py:904] (6/8) Epoch 5, batch 300, loss[loss=0.214, simple_loss=0.2957, pruned_loss=0.06617, over 17115.00 frames. ], tot_loss[loss=0.2344, simple_loss=0.3057, pruned_loss=0.0815, over 2590482.26 frames. ], batch size: 47, lr: 1.41e-02, grad_scale: 1.0 2023-04-28 05:56:52,375 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-28 05:56:58,126 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40922.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:57:05,081 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40926.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:57:07,869 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40928.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:57:39,677 INFO [train.py:904] (6/8) Epoch 5, batch 350, loss[loss=0.2367, simple_loss=0.308, pruned_loss=0.08271, over 16858.00 frames. ], tot_loss[loss=0.2289, simple_loss=0.301, pruned_loss=0.0784, over 2761617.08 frames. ], batch size: 42, lr: 1.41e-02, grad_scale: 1.0 2023-04-28 05:57:48,102 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.380e+02 3.200e+02 3.923e+02 5.275e+02 9.514e+02, threshold=7.845e+02, percent-clipped=1.0 2023-04-28 05:57:54,078 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40961.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 05:58:14,895 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40976.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:58:28,782 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40986.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:58:44,176 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40997.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:58:48,861 INFO [train.py:904] (6/8) Epoch 5, batch 400, loss[loss=0.2518, simple_loss=0.3142, pruned_loss=0.09475, over 16752.00 frames. ], tot_loss[loss=0.2263, simple_loss=0.2988, pruned_loss=0.0769, over 2892223.48 frames. ], batch size: 124, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 05:58:50,582 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5311, 4.5010, 4.4621, 3.8810, 4.4176, 1.8659, 4.2889, 4.3924], device='cuda:6'), covar=tensor([0.0072, 0.0053, 0.0071, 0.0244, 0.0059, 0.1410, 0.0076, 0.0104], device='cuda:6'), in_proj_covar=tensor([0.0091, 0.0075, 0.0120, 0.0120, 0.0087, 0.0141, 0.0103, 0.0114], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 05:58:59,025 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41007.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:59:08,297 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9086, 2.5570, 1.9590, 2.2289, 2.8378, 2.6471, 3.1405, 3.0046], device='cuda:6'), covar=tensor([0.0043, 0.0153, 0.0215, 0.0184, 0.0101, 0.0154, 0.0084, 0.0081], device='cuda:6'), in_proj_covar=tensor([0.0074, 0.0147, 0.0149, 0.0145, 0.0141, 0.0148, 0.0112, 0.0125], device='cuda:6'), out_proj_covar=tensor([8.9614e-05, 1.8656e-04, 1.8452e-04, 1.7861e-04, 1.7918e-04, 1.8865e-04, 1.3627e-04, 1.5875e-04], device='cuda:6') 2023-04-28 05:59:36,590 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41034.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:59:44,181 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.87 vs. limit=5.0 2023-04-28 05:59:51,130 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41045.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:00:00,530 INFO [train.py:904] (6/8) Epoch 5, batch 450, loss[loss=0.2006, simple_loss=0.2921, pruned_loss=0.05452, over 17131.00 frames. ], tot_loss[loss=0.223, simple_loss=0.2962, pruned_loss=0.07493, over 2986027.64 frames. ], batch size: 49, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:00:07,274 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 06:00:09,409 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.105e+02 3.241e+02 3.884e+02 4.938e+02 1.053e+03, threshold=7.768e+02, percent-clipped=3.0 2023-04-28 06:00:11,813 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41058.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:00:19,867 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3834, 3.5181, 3.2205, 3.1498, 2.7714, 3.2742, 3.1443, 3.1520], device='cuda:6'), covar=tensor([0.0517, 0.0286, 0.0300, 0.0251, 0.0805, 0.0300, 0.1252, 0.0456], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0172, 0.0215, 0.0180, 0.0242, 0.0206, 0.0153, 0.0231], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 06:01:08,825 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7097, 1.4900, 2.0738, 2.5608, 2.7014, 2.4859, 1.5646, 2.6981], device='cuda:6'), covar=tensor([0.0060, 0.0197, 0.0127, 0.0090, 0.0072, 0.0098, 0.0172, 0.0027], device='cuda:6'), in_proj_covar=tensor([0.0109, 0.0135, 0.0120, 0.0119, 0.0120, 0.0084, 0.0134, 0.0074], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 06:01:10,604 INFO [train.py:904] (6/8) Epoch 5, batch 500, loss[loss=0.1857, simple_loss=0.2642, pruned_loss=0.05362, over 16808.00 frames. ], tot_loss[loss=0.2205, simple_loss=0.2939, pruned_loss=0.07354, over 3057654.33 frames. ], batch size: 39, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:01:12,340 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41102.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:01:16,679 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41106.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:02:15,305 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41148.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:02:19,017 INFO [train.py:904] (6/8) Epoch 5, batch 550, loss[loss=0.2306, simple_loss=0.3081, pruned_loss=0.07656, over 17077.00 frames. ], tot_loss[loss=0.221, simple_loss=0.2948, pruned_loss=0.07359, over 3115240.78 frames. ], batch size: 55, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:02:27,243 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.046e+02 3.287e+02 3.927e+02 4.639e+02 9.245e+02, threshold=7.855e+02, percent-clipped=2.0 2023-04-28 06:02:35,968 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41163.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:03:22,053 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41196.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:03:27,688 INFO [train.py:904] (6/8) Epoch 5, batch 600, loss[loss=0.2081, simple_loss=0.291, pruned_loss=0.06262, over 16667.00 frames. ], tot_loss[loss=0.2206, simple_loss=0.2937, pruned_loss=0.0737, over 3151968.26 frames. ], batch size: 62, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:03:54,863 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41221.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:04:33,941 INFO [train.py:904] (6/8) Epoch 5, batch 650, loss[loss=0.2323, simple_loss=0.2901, pruned_loss=0.08723, over 16699.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.292, pruned_loss=0.07272, over 3190667.87 frames. ], batch size: 124, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:04:41,760 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41256.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 06:04:42,445 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.330e+02 3.128e+02 3.726e+02 4.799e+02 9.270e+02, threshold=7.452e+02, percent-clipped=1.0 2023-04-28 06:04:51,735 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.84 vs. limit=5.0 2023-04-28 06:05:37,238 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4234, 2.5033, 2.0453, 2.1524, 2.9135, 2.7619, 3.4812, 3.1688], device='cuda:6'), covar=tensor([0.0025, 0.0154, 0.0198, 0.0206, 0.0107, 0.0149, 0.0075, 0.0096], device='cuda:6'), in_proj_covar=tensor([0.0076, 0.0149, 0.0149, 0.0144, 0.0143, 0.0148, 0.0116, 0.0127], device='cuda:6'), out_proj_covar=tensor([9.3127e-05, 1.8747e-04, 1.8407e-04, 1.7783e-04, 1.8224e-04, 1.8866e-04, 1.4239e-04, 1.6137e-04], device='cuda:6') 2023-04-28 06:05:39,909 INFO [train.py:904] (6/8) Epoch 5, batch 700, loss[loss=0.2388, simple_loss=0.3011, pruned_loss=0.08827, over 16429.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.291, pruned_loss=0.07133, over 3224789.94 frames. ], batch size: 146, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:05:49,062 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41307.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:06:49,378 INFO [train.py:904] (6/8) Epoch 5, batch 750, loss[loss=0.2135, simple_loss=0.2791, pruned_loss=0.07398, over 16399.00 frames. ], tot_loss[loss=0.2171, simple_loss=0.2906, pruned_loss=0.07179, over 3244820.52 frames. ], batch size: 75, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:06:52,781 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41353.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:06:55,088 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41355.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:06:57,955 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 3.021e+02 3.672e+02 4.220e+02 6.723e+02, threshold=7.344e+02, percent-clipped=0.0 2023-04-28 06:07:58,821 INFO [train.py:904] (6/8) Epoch 5, batch 800, loss[loss=0.202, simple_loss=0.2911, pruned_loss=0.05649, over 17088.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.2905, pruned_loss=0.07209, over 3255804.24 frames. ], batch size: 49, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:07:59,171 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41401.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:09:04,054 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 06:09:08,622 INFO [train.py:904] (6/8) Epoch 5, batch 850, loss[loss=0.2172, simple_loss=0.3012, pruned_loss=0.0666, over 16612.00 frames. ], tot_loss[loss=0.2165, simple_loss=0.29, pruned_loss=0.07147, over 3264121.90 frames. ], batch size: 62, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:09:16,402 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.580e+02 3.257e+02 3.799e+02 4.682e+02 9.659e+02, threshold=7.597e+02, percent-clipped=4.0 2023-04-28 06:09:17,868 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41458.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:09:33,313 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5404, 3.6397, 4.0238, 4.0039, 3.9819, 3.6821, 3.7033, 3.7207], device='cuda:6'), covar=tensor([0.0347, 0.0430, 0.0329, 0.0405, 0.0419, 0.0326, 0.0749, 0.0408], device='cuda:6'), in_proj_covar=tensor([0.0243, 0.0235, 0.0240, 0.0244, 0.0283, 0.0250, 0.0360, 0.0219], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 06:10:16,012 INFO [train.py:904] (6/8) Epoch 5, batch 900, loss[loss=0.2193, simple_loss=0.2878, pruned_loss=0.07542, over 16252.00 frames. ], tot_loss[loss=0.2159, simple_loss=0.2901, pruned_loss=0.07089, over 3282424.37 frames. ], batch size: 165, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:10:44,982 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41521.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:11:27,561 INFO [train.py:904] (6/8) Epoch 5, batch 950, loss[loss=0.2172, simple_loss=0.2884, pruned_loss=0.07304, over 15426.00 frames. ], tot_loss[loss=0.2162, simple_loss=0.2899, pruned_loss=0.07126, over 3281203.34 frames. ], batch size: 190, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:11:34,594 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41556.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 06:11:35,283 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.063e+02 2.970e+02 3.801e+02 5.319e+02 1.620e+03, threshold=7.602e+02, percent-clipped=9.0 2023-04-28 06:11:53,682 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41569.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:12:01,349 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-28 06:12:25,931 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3590, 5.2332, 5.0486, 4.5394, 5.0803, 2.1703, 4.8105, 5.1843], device='cuda:6'), covar=tensor([0.0047, 0.0041, 0.0074, 0.0252, 0.0051, 0.1404, 0.0082, 0.0087], device='cuda:6'), in_proj_covar=tensor([0.0096, 0.0079, 0.0127, 0.0130, 0.0095, 0.0143, 0.0110, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 06:12:37,216 INFO [train.py:904] (6/8) Epoch 5, batch 1000, loss[loss=0.203, simple_loss=0.2912, pruned_loss=0.05738, over 17044.00 frames. ], tot_loss[loss=0.2149, simple_loss=0.2884, pruned_loss=0.07068, over 3291320.79 frames. ], batch size: 50, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:12:41,514 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41604.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:13:38,131 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.48 vs. limit=5.0 2023-04-28 06:13:45,696 INFO [train.py:904] (6/8) Epoch 5, batch 1050, loss[loss=0.2083, simple_loss=0.2898, pruned_loss=0.06342, over 17044.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2883, pruned_loss=0.06994, over 3300656.06 frames. ], batch size: 50, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:13:49,020 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41653.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:13:50,953 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41654.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:13:54,767 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.676e+02 2.880e+02 3.783e+02 4.446e+02 1.011e+03, threshold=7.566e+02, percent-clipped=5.0 2023-04-28 06:13:59,142 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9252, 5.3731, 5.3925, 5.3881, 5.2939, 5.9712, 5.6294, 5.3170], device='cuda:6'), covar=tensor([0.0768, 0.1548, 0.1528, 0.1972, 0.3092, 0.0982, 0.1113, 0.2434], device='cuda:6'), in_proj_covar=tensor([0.0269, 0.0395, 0.0380, 0.0339, 0.0455, 0.0410, 0.0312, 0.0454], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 06:14:21,619 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9141, 4.2594, 1.9965, 4.5561, 2.7709, 4.4431, 2.1864, 3.0469], device='cuda:6'), covar=tensor([0.0106, 0.0189, 0.1419, 0.0066, 0.0771, 0.0295, 0.1333, 0.0575], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0151, 0.0173, 0.0086, 0.0158, 0.0182, 0.0183, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 06:14:51,385 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7304, 2.3456, 2.3158, 4.3282, 2.0994, 3.3092, 2.3031, 2.5205], device='cuda:6'), covar=tensor([0.0485, 0.1640, 0.0898, 0.0242, 0.2366, 0.0689, 0.1777, 0.1849], device='cuda:6'), in_proj_covar=tensor([0.0310, 0.0307, 0.0248, 0.0310, 0.0358, 0.0291, 0.0273, 0.0378], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 06:14:56,103 INFO [train.py:904] (6/8) Epoch 5, batch 1100, loss[loss=0.2214, simple_loss=0.2867, pruned_loss=0.078, over 16860.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.2877, pruned_loss=0.06968, over 3313414.09 frames. ], batch size: 116, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:14:56,431 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41701.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:14:56,498 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41701.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:15:11,444 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2342, 4.2437, 4.1195, 4.0861, 3.7505, 4.1570, 3.9857, 3.9239], device='cuda:6'), covar=tensor([0.0442, 0.0230, 0.0220, 0.0167, 0.0761, 0.0270, 0.0550, 0.0418], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0180, 0.0224, 0.0189, 0.0257, 0.0213, 0.0161, 0.0239], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 06:15:15,811 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41715.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 06:16:02,694 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41749.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:16:05,297 INFO [train.py:904] (6/8) Epoch 5, batch 1150, loss[loss=0.2187, simple_loss=0.2796, pruned_loss=0.07892, over 16816.00 frames. ], tot_loss[loss=0.213, simple_loss=0.2872, pruned_loss=0.06937, over 3320494.34 frames. ], batch size: 124, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:16:12,830 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 2.850e+02 3.603e+02 4.605e+02 7.939e+02, threshold=7.207e+02, percent-clipped=2.0 2023-04-28 06:16:15,797 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41758.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:17:02,155 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-28 06:17:14,374 INFO [train.py:904] (6/8) Epoch 5, batch 1200, loss[loss=0.1878, simple_loss=0.2672, pruned_loss=0.0542, over 16831.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2858, pruned_loss=0.06879, over 3323898.43 frames. ], batch size: 42, lr: 1.40e-02, grad_scale: 8.0 2023-04-28 06:17:21,141 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41806.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:17:27,923 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7298, 4.6388, 3.7097, 1.8620, 2.8524, 2.3292, 3.7757, 4.3025], device='cuda:6'), covar=tensor([0.0236, 0.0388, 0.0523, 0.1693, 0.0853, 0.1030, 0.0614, 0.0694], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0133, 0.0156, 0.0144, 0.0136, 0.0126, 0.0143, 0.0139], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 06:17:33,634 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-28 06:17:46,947 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7226, 4.0246, 1.8527, 4.0031, 2.6841, 4.1123, 1.8700, 3.0303], device='cuda:6'), covar=tensor([0.0135, 0.0186, 0.1559, 0.0128, 0.0686, 0.0343, 0.1470, 0.0502], device='cuda:6'), in_proj_covar=tensor([0.0117, 0.0154, 0.0175, 0.0087, 0.0162, 0.0186, 0.0187, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 06:18:23,576 INFO [train.py:904] (6/8) Epoch 5, batch 1250, loss[loss=0.2416, simple_loss=0.2952, pruned_loss=0.09402, over 16855.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2855, pruned_loss=0.06893, over 3331592.77 frames. ], batch size: 116, lr: 1.40e-02, grad_scale: 8.0 2023-04-28 06:18:31,521 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.030e+02 3.380e+02 4.131e+02 4.913e+02 1.055e+03, threshold=8.263e+02, percent-clipped=6.0 2023-04-28 06:19:30,797 INFO [train.py:904] (6/8) Epoch 5, batch 1300, loss[loss=0.2153, simple_loss=0.2812, pruned_loss=0.07471, over 16649.00 frames. ], tot_loss[loss=0.212, simple_loss=0.2859, pruned_loss=0.069, over 3328883.05 frames. ], batch size: 134, lr: 1.40e-02, grad_scale: 8.0 2023-04-28 06:19:39,597 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41906.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:20:38,658 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41948.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 06:20:42,274 INFO [train.py:904] (6/8) Epoch 5, batch 1350, loss[loss=0.1806, simple_loss=0.255, pruned_loss=0.05308, over 15948.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2854, pruned_loss=0.06831, over 3319810.94 frames. ], batch size: 35, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:20:51,205 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.161e+02 3.407e+02 4.000e+02 4.882e+02 1.065e+03, threshold=8.000e+02, percent-clipped=1.0 2023-04-28 06:21:05,575 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41967.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:21:55,630 INFO [train.py:904] (6/8) Epoch 5, batch 1400, loss[loss=0.2254, simple_loss=0.2855, pruned_loss=0.08267, over 12147.00 frames. ], tot_loss[loss=0.2121, simple_loss=0.2856, pruned_loss=0.06932, over 3305031.44 frames. ], batch size: 246, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:22:08,318 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42009.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 06:22:09,976 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42010.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 06:23:05,348 INFO [train.py:904] (6/8) Epoch 5, batch 1450, loss[loss=0.1733, simple_loss=0.261, pruned_loss=0.04282, over 17199.00 frames. ], tot_loss[loss=0.2121, simple_loss=0.2864, pruned_loss=0.06895, over 3318096.55 frames. ], batch size: 44, lr: 1.39e-02, grad_scale: 4.0 2023-04-28 06:23:05,792 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3018, 4.3180, 4.2688, 3.5346, 4.2198, 1.6412, 4.0740, 4.0457], device='cuda:6'), covar=tensor([0.0084, 0.0058, 0.0082, 0.0308, 0.0066, 0.1731, 0.0083, 0.0145], device='cuda:6'), in_proj_covar=tensor([0.0097, 0.0081, 0.0128, 0.0134, 0.0097, 0.0143, 0.0111, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 06:23:15,558 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.035e+02 3.059e+02 3.831e+02 4.678e+02 9.639e+02, threshold=7.661e+02, percent-clipped=1.0 2023-04-28 06:24:14,181 INFO [train.py:904] (6/8) Epoch 5, batch 1500, loss[loss=0.1906, simple_loss=0.2841, pruned_loss=0.04854, over 17151.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.285, pruned_loss=0.06882, over 3314442.52 frames. ], batch size: 49, lr: 1.39e-02, grad_scale: 4.0 2023-04-28 06:25:21,154 INFO [train.py:904] (6/8) Epoch 5, batch 1550, loss[loss=0.2414, simple_loss=0.3001, pruned_loss=0.09141, over 16747.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2869, pruned_loss=0.07066, over 3299374.44 frames. ], batch size: 83, lr: 1.39e-02, grad_scale: 4.0 2023-04-28 06:25:32,957 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.189e+02 3.587e+02 4.011e+02 4.553e+02 8.694e+02, threshold=8.021e+02, percent-clipped=2.0 2023-04-28 06:26:32,627 INFO [train.py:904] (6/8) Epoch 5, batch 1600, loss[loss=0.1891, simple_loss=0.2808, pruned_loss=0.04867, over 17114.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.2894, pruned_loss=0.07174, over 3312695.37 frames. ], batch size: 47, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:27:37,607 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-28 06:27:39,757 INFO [train.py:904] (6/8) Epoch 5, batch 1650, loss[loss=0.2412, simple_loss=0.2998, pruned_loss=0.09134, over 16737.00 frames. ], tot_loss[loss=0.2184, simple_loss=0.2915, pruned_loss=0.0727, over 3318250.04 frames. ], batch size: 89, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:27:49,823 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.983e+02 3.262e+02 3.932e+02 4.837e+02 9.371e+02, threshold=7.863e+02, percent-clipped=1.0 2023-04-28 06:27:56,371 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42262.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:28:49,287 INFO [train.py:904] (6/8) Epoch 5, batch 1700, loss[loss=0.2045, simple_loss=0.2886, pruned_loss=0.06025, over 17116.00 frames. ], tot_loss[loss=0.2216, simple_loss=0.2947, pruned_loss=0.07428, over 3306458.26 frames. ], batch size: 49, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:28:54,202 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42304.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 06:29:01,266 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42310.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 06:29:03,710 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2389, 4.8841, 5.0867, 5.4734, 5.5562, 4.6937, 5.5059, 5.4974], device='cuda:6'), covar=tensor([0.0712, 0.0609, 0.1220, 0.0370, 0.0347, 0.0533, 0.0305, 0.0338], device='cuda:6'), in_proj_covar=tensor([0.0403, 0.0493, 0.0635, 0.0499, 0.0376, 0.0367, 0.0391, 0.0416], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 06:29:56,498 INFO [train.py:904] (6/8) Epoch 5, batch 1750, loss[loss=0.2181, simple_loss=0.2944, pruned_loss=0.0709, over 16466.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.2946, pruned_loss=0.0736, over 3308075.84 frames. ], batch size: 68, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:30:05,726 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.019e+02 3.418e+02 4.322e+02 5.636e+02 1.741e+03, threshold=8.644e+02, percent-clipped=7.0 2023-04-28 06:30:05,997 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=42358.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:30:20,049 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5860, 4.4920, 4.4624, 3.9019, 4.4667, 1.7117, 4.2935, 4.3931], device='cuda:6'), covar=tensor([0.0071, 0.0072, 0.0088, 0.0298, 0.0062, 0.1690, 0.0093, 0.0113], device='cuda:6'), in_proj_covar=tensor([0.0098, 0.0082, 0.0129, 0.0135, 0.0097, 0.0142, 0.0111, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 06:30:35,459 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42379.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:31:06,096 INFO [train.py:904] (6/8) Epoch 5, batch 1800, loss[loss=0.1737, simple_loss=0.2628, pruned_loss=0.04227, over 16862.00 frames. ], tot_loss[loss=0.2204, simple_loss=0.2949, pruned_loss=0.07292, over 3311078.01 frames. ], batch size: 42, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:31:27,701 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-28 06:32:01,416 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42440.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:32:16,842 INFO [train.py:904] (6/8) Epoch 5, batch 1850, loss[loss=0.2019, simple_loss=0.2871, pruned_loss=0.05829, over 17142.00 frames. ], tot_loss[loss=0.2197, simple_loss=0.2947, pruned_loss=0.0723, over 3312255.58 frames. ], batch size: 49, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:32:26,219 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.852e+02 3.144e+02 3.803e+02 4.355e+02 7.438e+02, threshold=7.606e+02, percent-clipped=0.0 2023-04-28 06:32:27,941 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6808, 2.2314, 2.3593, 4.3667, 2.0130, 3.3240, 2.2072, 2.4029], device='cuda:6'), covar=tensor([0.0519, 0.1725, 0.0887, 0.0242, 0.2454, 0.0713, 0.1805, 0.1875], device='cuda:6'), in_proj_covar=tensor([0.0310, 0.0311, 0.0250, 0.0310, 0.0357, 0.0297, 0.0275, 0.0382], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 06:33:25,573 INFO [train.py:904] (6/8) Epoch 5, batch 1900, loss[loss=0.2739, simple_loss=0.3396, pruned_loss=0.1041, over 11944.00 frames. ], tot_loss[loss=0.2194, simple_loss=0.2947, pruned_loss=0.07201, over 3298801.62 frames. ], batch size: 247, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:33:32,591 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42506.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:34:14,247 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0448, 5.4615, 5.6367, 5.5303, 5.4930, 5.9816, 5.6796, 5.4707], device='cuda:6'), covar=tensor([0.0699, 0.1560, 0.1253, 0.1332, 0.2318, 0.0781, 0.0910, 0.2011], device='cuda:6'), in_proj_covar=tensor([0.0276, 0.0406, 0.0382, 0.0342, 0.0460, 0.0410, 0.0312, 0.0459], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 06:34:35,498 INFO [train.py:904] (6/8) Epoch 5, batch 1950, loss[loss=0.1735, simple_loss=0.2586, pruned_loss=0.04419, over 16741.00 frames. ], tot_loss[loss=0.2172, simple_loss=0.2932, pruned_loss=0.07063, over 3306584.88 frames. ], batch size: 39, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:34:47,175 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 3.199e+02 3.788e+02 4.472e+02 9.515e+02, threshold=7.576e+02, percent-clipped=2.0 2023-04-28 06:34:52,191 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42562.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:34:52,279 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8008, 3.0801, 2.5453, 4.4126, 4.0366, 4.0945, 1.5319, 3.0657], device='cuda:6'), covar=tensor([0.1248, 0.0489, 0.1005, 0.0077, 0.0239, 0.0356, 0.1307, 0.0616], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0141, 0.0168, 0.0086, 0.0176, 0.0172, 0.0160, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 06:34:52,605 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-04-28 06:34:59,479 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42567.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:35:48,039 INFO [train.py:904] (6/8) Epoch 5, batch 2000, loss[loss=0.2586, simple_loss=0.3098, pruned_loss=0.1037, over 16731.00 frames. ], tot_loss[loss=0.2172, simple_loss=0.2924, pruned_loss=0.07104, over 3308946.58 frames. ], batch size: 134, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:35:51,754 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42604.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 06:35:59,414 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.00 vs. limit=5.0 2023-04-28 06:36:00,881 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=42610.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:36:43,024 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6269, 2.6418, 2.0666, 2.3870, 3.0369, 2.8909, 3.8893, 3.3868], device='cuda:6'), covar=tensor([0.0025, 0.0140, 0.0188, 0.0167, 0.0088, 0.0145, 0.0055, 0.0080], device='cuda:6'), in_proj_covar=tensor([0.0085, 0.0152, 0.0152, 0.0150, 0.0148, 0.0155, 0.0128, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 06:36:55,100 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42649.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:36:57,536 INFO [train.py:904] (6/8) Epoch 5, batch 2050, loss[loss=0.2499, simple_loss=0.3254, pruned_loss=0.08716, over 15429.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.2916, pruned_loss=0.07025, over 3313506.77 frames. ], batch size: 190, lr: 1.38e-02, grad_scale: 8.0 2023-04-28 06:36:59,090 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=42652.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 06:37:06,886 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.791e+02 3.076e+02 3.576e+02 4.219e+02 7.856e+02, threshold=7.152e+02, percent-clipped=1.0 2023-04-28 06:37:18,205 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5236, 3.5548, 3.2237, 3.1231, 3.0426, 3.1482, 3.2464, 3.0501], device='cuda:6'), covar=tensor([0.0398, 0.0292, 0.0197, 0.0201, 0.0634, 0.0195, 0.0970, 0.0390], device='cuda:6'), in_proj_covar=tensor([0.0195, 0.0194, 0.0233, 0.0203, 0.0269, 0.0223, 0.0170, 0.0255], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 06:38:03,648 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-28 06:38:05,313 INFO [train.py:904] (6/8) Epoch 5, batch 2100, loss[loss=0.2708, simple_loss=0.3329, pruned_loss=0.1043, over 16832.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.2946, pruned_loss=0.07256, over 3311770.54 frames. ], batch size: 116, lr: 1.38e-02, grad_scale: 8.0 2023-04-28 06:38:18,835 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42710.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:38:52,985 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42735.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:39:14,541 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2769, 5.3210, 5.1529, 5.0604, 4.5669, 5.2025, 5.1874, 4.8257], device='cuda:6'), covar=tensor([0.0437, 0.0242, 0.0190, 0.0152, 0.0954, 0.0242, 0.0187, 0.0390], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0192, 0.0231, 0.0201, 0.0265, 0.0222, 0.0169, 0.0253], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 06:39:15,224 INFO [train.py:904] (6/8) Epoch 5, batch 2150, loss[loss=0.1936, simple_loss=0.272, pruned_loss=0.05757, over 16864.00 frames. ], tot_loss[loss=0.2191, simple_loss=0.2941, pruned_loss=0.07204, over 3309157.68 frames. ], batch size: 42, lr: 1.38e-02, grad_scale: 8.0 2023-04-28 06:39:24,110 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.183e+02 3.251e+02 3.882e+02 4.566e+02 8.930e+02, threshold=7.764e+02, percent-clipped=4.0 2023-04-28 06:40:23,566 INFO [train.py:904] (6/8) Epoch 5, batch 2200, loss[loss=0.2144, simple_loss=0.2965, pruned_loss=0.06615, over 17141.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.2953, pruned_loss=0.07261, over 3300874.50 frames. ], batch size: 48, lr: 1.38e-02, grad_scale: 8.0 2023-04-28 06:40:35,433 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8898, 4.2401, 1.8900, 4.6213, 2.7469, 4.5479, 2.2901, 3.0938], device='cuda:6'), covar=tensor([0.0128, 0.0216, 0.1490, 0.0030, 0.0727, 0.0263, 0.1188, 0.0501], device='cuda:6'), in_proj_covar=tensor([0.0116, 0.0157, 0.0175, 0.0087, 0.0160, 0.0189, 0.0184, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 06:40:44,817 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 06:40:53,068 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7141, 2.6740, 2.2594, 3.7844, 3.3915, 3.6079, 1.4156, 2.7723], device='cuda:6'), covar=tensor([0.1250, 0.0500, 0.1115, 0.0083, 0.0228, 0.0356, 0.1377, 0.0707], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0141, 0.0166, 0.0085, 0.0177, 0.0169, 0.0157, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 06:41:34,694 INFO [train.py:904] (6/8) Epoch 5, batch 2250, loss[loss=0.2373, simple_loss=0.3189, pruned_loss=0.07781, over 17264.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.2954, pruned_loss=0.07225, over 3313834.35 frames. ], batch size: 52, lr: 1.38e-02, grad_scale: 8.0 2023-04-28 06:41:43,707 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.043e+02 3.207e+02 3.862e+02 4.939e+02 1.226e+03, threshold=7.724e+02, percent-clipped=4.0 2023-04-28 06:41:46,064 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42859.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:41:49,532 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42862.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:42:14,454 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42879.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:42:32,803 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2019, 5.6088, 5.6472, 5.5205, 5.5516, 6.0483, 5.7550, 5.4941], device='cuda:6'), covar=tensor([0.0673, 0.1774, 0.1497, 0.1775, 0.2248, 0.0836, 0.1001, 0.1939], device='cuda:6'), in_proj_covar=tensor([0.0280, 0.0408, 0.0386, 0.0348, 0.0469, 0.0414, 0.0320, 0.0459], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 06:42:44,587 INFO [train.py:904] (6/8) Epoch 5, batch 2300, loss[loss=0.1832, simple_loss=0.2674, pruned_loss=0.04948, over 17082.00 frames. ], tot_loss[loss=0.221, simple_loss=0.2963, pruned_loss=0.07286, over 3314883.01 frames. ], batch size: 50, lr: 1.38e-02, grad_scale: 4.0 2023-04-28 06:42:53,803 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3309, 4.2690, 4.2561, 3.8009, 4.2440, 1.6361, 4.0563, 4.0914], device='cuda:6'), covar=tensor([0.0066, 0.0055, 0.0078, 0.0226, 0.0058, 0.1483, 0.0081, 0.0106], device='cuda:6'), in_proj_covar=tensor([0.0097, 0.0082, 0.0127, 0.0134, 0.0096, 0.0138, 0.0112, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 06:43:11,617 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42920.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:43:38,221 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42940.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:43:53,244 INFO [train.py:904] (6/8) Epoch 5, batch 2350, loss[loss=0.2228, simple_loss=0.2922, pruned_loss=0.07664, over 16796.00 frames. ], tot_loss[loss=0.222, simple_loss=0.2967, pruned_loss=0.07364, over 3324904.98 frames. ], batch size: 83, lr: 1.38e-02, grad_scale: 4.0 2023-04-28 06:44:03,299 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.280e+02 3.235e+02 3.889e+02 4.889e+02 8.777e+02, threshold=7.778e+02, percent-clipped=2.0 2023-04-28 06:44:05,982 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-28 06:45:02,225 INFO [train.py:904] (6/8) Epoch 5, batch 2400, loss[loss=0.2443, simple_loss=0.31, pruned_loss=0.0893, over 16761.00 frames. ], tot_loss[loss=0.2225, simple_loss=0.2975, pruned_loss=0.07375, over 3325315.15 frames. ], batch size: 89, lr: 1.38e-02, grad_scale: 8.0 2023-04-28 06:45:08,767 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43005.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:45:31,270 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2023-04-28 06:45:52,176 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43035.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:46:13,638 INFO [train.py:904] (6/8) Epoch 5, batch 2450, loss[loss=0.1954, simple_loss=0.2764, pruned_loss=0.05719, over 17201.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.2987, pruned_loss=0.07409, over 3312181.94 frames. ], batch size: 45, lr: 1.38e-02, grad_scale: 4.0 2023-04-28 06:46:26,010 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 3.112e+02 3.702e+02 4.495e+02 8.852e+02, threshold=7.404e+02, percent-clipped=2.0 2023-04-28 06:46:49,739 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0896, 4.2053, 4.4081, 1.8815, 4.5598, 4.6323, 3.3063, 3.7067], device='cuda:6'), covar=tensor([0.0535, 0.0105, 0.0113, 0.1088, 0.0055, 0.0056, 0.0302, 0.0280], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0085, 0.0086, 0.0140, 0.0071, 0.0079, 0.0115, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 06:46:58,054 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=43083.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:47:23,966 INFO [train.py:904] (6/8) Epoch 5, batch 2500, loss[loss=0.2117, simple_loss=0.3048, pruned_loss=0.05926, over 17110.00 frames. ], tot_loss[loss=0.2225, simple_loss=0.298, pruned_loss=0.07349, over 3315495.48 frames. ], batch size: 49, lr: 1.38e-02, grad_scale: 4.0 2023-04-28 06:48:33,485 INFO [train.py:904] (6/8) Epoch 5, batch 2550, loss[loss=0.2073, simple_loss=0.2999, pruned_loss=0.05735, over 17112.00 frames. ], tot_loss[loss=0.223, simple_loss=0.2988, pruned_loss=0.07362, over 3318295.23 frames. ], batch size: 49, lr: 1.38e-02, grad_scale: 4.0 2023-04-28 06:48:45,558 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.843e+02 3.229e+02 3.908e+02 4.674e+02 8.443e+02, threshold=7.816e+02, percent-clipped=2.0 2023-04-28 06:48:48,798 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43162.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:49:04,667 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-28 06:49:15,310 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7443, 4.8591, 4.9068, 4.8880, 4.7555, 5.4103, 5.0957, 4.7805], device='cuda:6'), covar=tensor([0.0932, 0.1698, 0.1312, 0.1603, 0.2646, 0.0977, 0.1197, 0.2210], device='cuda:6'), in_proj_covar=tensor([0.0281, 0.0398, 0.0383, 0.0341, 0.0465, 0.0413, 0.0321, 0.0457], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 06:49:43,647 INFO [train.py:904] (6/8) Epoch 5, batch 2600, loss[loss=0.2271, simple_loss=0.2993, pruned_loss=0.07744, over 16678.00 frames. ], tot_loss[loss=0.223, simple_loss=0.2989, pruned_loss=0.07362, over 3321292.14 frames. ], batch size: 134, lr: 1.38e-02, grad_scale: 4.0 2023-04-28 06:49:55,553 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=43210.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:50:01,775 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43215.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:50:31,268 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43235.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:50:53,869 INFO [train.py:904] (6/8) Epoch 5, batch 2650, loss[loss=0.1882, simple_loss=0.2809, pruned_loss=0.04775, over 17210.00 frames. ], tot_loss[loss=0.2229, simple_loss=0.2995, pruned_loss=0.07313, over 3316932.31 frames. ], batch size: 44, lr: 1.37e-02, grad_scale: 4.0 2023-04-28 06:51:05,580 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.996e+02 3.400e+02 3.860e+02 4.730e+02 9.799e+02, threshold=7.720e+02, percent-clipped=2.0 2023-04-28 06:52:02,500 INFO [train.py:904] (6/8) Epoch 5, batch 2700, loss[loss=0.2113, simple_loss=0.2877, pruned_loss=0.06746, over 17212.00 frames. ], tot_loss[loss=0.2208, simple_loss=0.2986, pruned_loss=0.07151, over 3329472.84 frames. ], batch size: 44, lr: 1.37e-02, grad_scale: 4.0 2023-04-28 06:52:07,143 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-28 06:52:09,055 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43305.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:53:12,583 INFO [train.py:904] (6/8) Epoch 5, batch 2750, loss[loss=0.2645, simple_loss=0.3173, pruned_loss=0.1058, over 12723.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.2979, pruned_loss=0.07023, over 3331117.45 frames. ], batch size: 246, lr: 1.37e-02, grad_scale: 4.0 2023-04-28 06:53:15,877 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=43353.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:53:25,426 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.808e+02 2.783e+02 3.325e+02 4.453e+02 8.515e+02, threshold=6.649e+02, percent-clipped=2.0 2023-04-28 06:53:34,979 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5738, 3.5150, 2.6877, 2.2457, 2.5770, 2.0853, 3.3680, 3.5324], device='cuda:6'), covar=tensor([0.1935, 0.0495, 0.1089, 0.1545, 0.1889, 0.1450, 0.0437, 0.0602], device='cuda:6'), in_proj_covar=tensor([0.0273, 0.0249, 0.0264, 0.0241, 0.0302, 0.0200, 0.0237, 0.0259], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 06:54:07,313 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6350, 6.0740, 5.7775, 5.8397, 5.2893, 5.0400, 5.5845, 6.1390], device='cuda:6'), covar=tensor([0.0726, 0.0571, 0.0818, 0.0384, 0.0566, 0.0535, 0.0544, 0.0631], device='cuda:6'), in_proj_covar=tensor([0.0377, 0.0510, 0.0429, 0.0327, 0.0318, 0.0326, 0.0406, 0.0358], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 06:54:22,964 INFO [train.py:904] (6/8) Epoch 5, batch 2800, loss[loss=0.2175, simple_loss=0.304, pruned_loss=0.06543, over 17059.00 frames. ], tot_loss[loss=0.2175, simple_loss=0.2969, pruned_loss=0.0691, over 3335912.21 frames. ], batch size: 55, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 06:55:33,350 INFO [train.py:904] (6/8) Epoch 5, batch 2850, loss[loss=0.2072, simple_loss=0.2856, pruned_loss=0.0644, over 15975.00 frames. ], tot_loss[loss=0.2169, simple_loss=0.2958, pruned_loss=0.06902, over 3333232.82 frames. ], batch size: 35, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 06:55:45,526 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.398e+02 3.090e+02 3.971e+02 4.824e+02 1.597e+03, threshold=7.942e+02, percent-clipped=16.0 2023-04-28 06:56:41,595 INFO [train.py:904] (6/8) Epoch 5, batch 2900, loss[loss=0.2372, simple_loss=0.2971, pruned_loss=0.08866, over 16718.00 frames. ], tot_loss[loss=0.2175, simple_loss=0.2957, pruned_loss=0.06962, over 3331924.61 frames. ], batch size: 124, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 06:57:00,337 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43514.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:57:01,545 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43515.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:57:28,531 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43535.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:57:49,458 INFO [train.py:904] (6/8) Epoch 5, batch 2950, loss[loss=0.1995, simple_loss=0.2898, pruned_loss=0.05463, over 17032.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.2948, pruned_loss=0.07074, over 3326722.58 frames. ], batch size: 50, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 06:58:02,028 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.717e+02 3.523e+02 4.392e+02 5.474e+02 1.022e+03, threshold=8.784e+02, percent-clipped=2.0 2023-04-28 06:58:05,995 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=43563.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:58:24,106 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43575.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:58:34,162 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=43583.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:58:59,870 INFO [train.py:904] (6/8) Epoch 5, batch 3000, loss[loss=0.2249, simple_loss=0.3172, pruned_loss=0.06628, over 17249.00 frames. ], tot_loss[loss=0.2201, simple_loss=0.2957, pruned_loss=0.07224, over 3317309.86 frames. ], batch size: 52, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 06:58:59,871 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 06:59:08,833 INFO [train.py:938] (6/8) Epoch 5, validation: loss=0.1564, simple_loss=0.2632, pruned_loss=0.02477, over 944034.00 frames. 2023-04-28 06:59:08,834 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-28 07:00:18,526 INFO [train.py:904] (6/8) Epoch 5, batch 3050, loss[loss=0.1922, simple_loss=0.2768, pruned_loss=0.05381, over 17220.00 frames. ], tot_loss[loss=0.2197, simple_loss=0.2955, pruned_loss=0.07192, over 3323370.94 frames. ], batch size: 45, lr: 1.37e-02, grad_scale: 4.0 2023-04-28 07:00:26,147 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-28 07:00:31,429 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.472e+02 3.368e+02 3.840e+02 5.233e+02 1.219e+03, threshold=7.679e+02, percent-clipped=3.0 2023-04-28 07:00:51,988 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3040, 5.7020, 5.3692, 5.4426, 4.9270, 4.9487, 5.1902, 5.8097], device='cuda:6'), covar=tensor([0.0647, 0.0748, 0.1018, 0.0558, 0.0801, 0.0600, 0.0671, 0.0749], device='cuda:6'), in_proj_covar=tensor([0.0376, 0.0510, 0.0436, 0.0330, 0.0320, 0.0329, 0.0411, 0.0358], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 07:01:25,947 INFO [train.py:904] (6/8) Epoch 5, batch 3100, loss[loss=0.2286, simple_loss=0.2999, pruned_loss=0.07865, over 16510.00 frames. ], tot_loss[loss=0.2193, simple_loss=0.2949, pruned_loss=0.07183, over 3329662.61 frames. ], batch size: 75, lr: 1.37e-02, grad_scale: 4.0 2023-04-28 07:01:32,839 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4355, 1.9129, 2.8150, 3.3219, 3.1795, 3.6522, 2.2906, 3.4677], device='cuda:6'), covar=tensor([0.0053, 0.0204, 0.0110, 0.0084, 0.0078, 0.0058, 0.0174, 0.0042], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0141, 0.0127, 0.0130, 0.0127, 0.0093, 0.0138, 0.0081], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 07:02:16,282 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4848, 4.4088, 4.3667, 3.8895, 4.3886, 1.6311, 4.2194, 4.2691], device='cuda:6'), covar=tensor([0.0072, 0.0065, 0.0079, 0.0277, 0.0061, 0.1667, 0.0084, 0.0113], device='cuda:6'), in_proj_covar=tensor([0.0097, 0.0086, 0.0129, 0.0136, 0.0099, 0.0139, 0.0114, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 07:02:33,549 INFO [train.py:904] (6/8) Epoch 5, batch 3150, loss[loss=0.2268, simple_loss=0.3109, pruned_loss=0.07135, over 17015.00 frames. ], tot_loss[loss=0.2193, simple_loss=0.2941, pruned_loss=0.07226, over 3324792.22 frames. ], batch size: 50, lr: 1.37e-02, grad_scale: 4.0 2023-04-28 07:02:46,878 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.292e+02 2.988e+02 3.776e+02 4.574e+02 1.068e+03, threshold=7.553e+02, percent-clipped=4.0 2023-04-28 07:02:56,597 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.54 vs. limit=5.0 2023-04-28 07:03:22,280 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43787.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:03:42,787 INFO [train.py:904] (6/8) Epoch 5, batch 3200, loss[loss=0.247, simple_loss=0.3305, pruned_loss=0.08176, over 17115.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.2939, pruned_loss=0.07188, over 3332404.26 frames. ], batch size: 48, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 07:04:48,295 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:04:52,117 INFO [train.py:904] (6/8) Epoch 5, batch 3250, loss[loss=0.2232, simple_loss=0.2929, pruned_loss=0.07673, over 16694.00 frames. ], tot_loss[loss=0.2184, simple_loss=0.294, pruned_loss=0.07139, over 3335071.44 frames. ], batch size: 89, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 07:05:05,946 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9628, 3.8869, 3.8708, 3.3521, 3.9013, 1.7741, 3.7038, 3.6226], device='cuda:6'), covar=tensor([0.0074, 0.0066, 0.0095, 0.0232, 0.0062, 0.1581, 0.0087, 0.0142], device='cuda:6'), in_proj_covar=tensor([0.0097, 0.0086, 0.0129, 0.0135, 0.0099, 0.0138, 0.0113, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 07:05:06,669 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.826e+02 2.952e+02 3.819e+02 4.745e+02 7.662e+02, threshold=7.638e+02, percent-clipped=2.0 2023-04-28 07:05:19,487 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43870.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:05:35,697 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0997, 1.8602, 2.4720, 2.9040, 2.9335, 3.2552, 2.1206, 3.3474], device='cuda:6'), covar=tensor([0.0072, 0.0217, 0.0117, 0.0134, 0.0097, 0.0078, 0.0185, 0.0056], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0143, 0.0128, 0.0132, 0.0127, 0.0094, 0.0137, 0.0083], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 07:05:44,007 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6563, 1.5337, 2.1117, 2.5599, 2.6446, 2.4793, 1.7039, 2.7114], device='cuda:6'), covar=tensor([0.0054, 0.0230, 0.0138, 0.0110, 0.0076, 0.0111, 0.0192, 0.0050], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0143, 0.0128, 0.0131, 0.0127, 0.0094, 0.0137, 0.0083], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 07:06:03,319 INFO [train.py:904] (6/8) Epoch 5, batch 3300, loss[loss=0.271, simple_loss=0.3314, pruned_loss=0.1053, over 16419.00 frames. ], tot_loss[loss=0.2193, simple_loss=0.2946, pruned_loss=0.07202, over 3333951.02 frames. ], batch size: 146, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:06:06,955 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.11 vs. limit=5.0 2023-04-28 07:06:14,819 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43909.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:07:12,748 INFO [train.py:904] (6/8) Epoch 5, batch 3350, loss[loss=0.2205, simple_loss=0.2924, pruned_loss=0.07433, over 16232.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.294, pruned_loss=0.07069, over 3333766.52 frames. ], batch size: 165, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:07:24,201 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43959.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:07:26,941 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.942e+02 2.864e+02 3.521e+02 4.539e+02 9.461e+02, threshold=7.042e+02, percent-clipped=3.0 2023-04-28 07:07:39,205 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43970.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:08:05,412 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0975, 4.1020, 4.5686, 4.5246, 4.5066, 4.1544, 4.1441, 4.0117], device='cuda:6'), covar=tensor([0.0279, 0.0433, 0.0274, 0.0348, 0.0353, 0.0293, 0.0756, 0.0538], device='cuda:6'), in_proj_covar=tensor([0.0256, 0.0250, 0.0256, 0.0255, 0.0303, 0.0267, 0.0384, 0.0226], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 07:08:22,241 INFO [train.py:904] (6/8) Epoch 5, batch 3400, loss[loss=0.2113, simple_loss=0.2941, pruned_loss=0.06427, over 17134.00 frames. ], tot_loss[loss=0.2165, simple_loss=0.2933, pruned_loss=0.06983, over 3335122.81 frames. ], batch size: 48, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:08:40,661 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-28 07:08:47,583 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44020.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:09:25,436 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5302, 2.2161, 1.5784, 1.9954, 2.6912, 2.5222, 2.7747, 2.8205], device='cuda:6'), covar=tensor([0.0049, 0.0148, 0.0199, 0.0188, 0.0073, 0.0121, 0.0076, 0.0082], device='cuda:6'), in_proj_covar=tensor([0.0085, 0.0152, 0.0152, 0.0148, 0.0150, 0.0155, 0.0132, 0.0137], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 07:09:31,564 INFO [train.py:904] (6/8) Epoch 5, batch 3450, loss[loss=0.2171, simple_loss=0.3019, pruned_loss=0.06612, over 17028.00 frames. ], tot_loss[loss=0.2149, simple_loss=0.2914, pruned_loss=0.06917, over 3335116.81 frames. ], batch size: 53, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:09:44,808 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.795e+02 3.034e+02 3.683e+02 4.468e+02 1.074e+03, threshold=7.367e+02, percent-clipped=2.0 2023-04-28 07:10:39,128 INFO [train.py:904] (6/8) Epoch 5, batch 3500, loss[loss=0.1885, simple_loss=0.2715, pruned_loss=0.05274, over 17183.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.29, pruned_loss=0.06852, over 3325995.38 frames. ], batch size: 46, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:11:37,832 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44143.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:11:49,615 INFO [train.py:904] (6/8) Epoch 5, batch 3550, loss[loss=0.2078, simple_loss=0.2813, pruned_loss=0.06713, over 16838.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2883, pruned_loss=0.06753, over 3325600.91 frames. ], batch size: 90, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:12:03,022 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.484e+02 3.064e+02 3.621e+02 4.319e+02 6.916e+02, threshold=7.242e+02, percent-clipped=0.0 2023-04-28 07:12:17,159 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44170.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:12:25,814 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2023-04-28 07:12:59,202 INFO [train.py:904] (6/8) Epoch 5, batch 3600, loss[loss=0.1912, simple_loss=0.2582, pruned_loss=0.0621, over 16821.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.2871, pruned_loss=0.06727, over 3318824.73 frames. ], batch size: 96, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:13:06,578 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-28 07:13:23,274 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=44218.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:14:09,666 INFO [train.py:904] (6/8) Epoch 5, batch 3650, loss[loss=0.2082, simple_loss=0.2688, pruned_loss=0.0738, over 16869.00 frames. ], tot_loss[loss=0.2119, simple_loss=0.2876, pruned_loss=0.0681, over 3314003.37 frames. ], batch size: 116, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:14:25,432 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.873e+02 3.015e+02 3.671e+02 4.591e+02 9.976e+02, threshold=7.342e+02, percent-clipped=3.0 2023-04-28 07:14:30,536 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44265.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:14:51,623 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6912, 2.6078, 2.4280, 3.6419, 3.2833, 3.6209, 1.6233, 2.7572], device='cuda:6'), covar=tensor([0.1330, 0.0542, 0.1012, 0.0100, 0.0226, 0.0330, 0.1289, 0.0719], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0141, 0.0165, 0.0088, 0.0184, 0.0172, 0.0158, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 07:15:22,478 INFO [train.py:904] (6/8) Epoch 5, batch 3700, loss[loss=0.2167, simple_loss=0.2748, pruned_loss=0.07931, over 16821.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2863, pruned_loss=0.06992, over 3299139.97 frames. ], batch size: 116, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:15:37,917 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.14 vs. limit=5.0 2023-04-28 07:15:45,302 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44315.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:16:36,542 INFO [train.py:904] (6/8) Epoch 5, batch 3750, loss[loss=0.2347, simple_loss=0.3109, pruned_loss=0.07929, over 15560.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.2871, pruned_loss=0.0719, over 3288230.96 frames. ], batch size: 191, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:16:52,716 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.995e+02 2.955e+02 3.644e+02 4.475e+02 7.424e+02, threshold=7.289e+02, percent-clipped=1.0 2023-04-28 07:17:43,070 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-28 07:17:49,965 INFO [train.py:904] (6/8) Epoch 5, batch 3800, loss[loss=0.2913, simple_loss=0.3523, pruned_loss=0.1152, over 12379.00 frames. ], tot_loss[loss=0.2186, simple_loss=0.2895, pruned_loss=0.0739, over 3269636.12 frames. ], batch size: 246, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:18:50,506 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44443.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:19:01,027 INFO [train.py:904] (6/8) Epoch 5, batch 3850, loss[loss=0.214, simple_loss=0.2784, pruned_loss=0.07482, over 16924.00 frames. ], tot_loss[loss=0.2197, simple_loss=0.2898, pruned_loss=0.07484, over 3272285.39 frames. ], batch size: 116, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:19:16,579 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.887e+02 2.978e+02 3.515e+02 4.161e+02 7.942e+02, threshold=7.030e+02, percent-clipped=2.0 2023-04-28 07:19:50,324 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3692, 2.1603, 1.6510, 1.9282, 2.5658, 2.3993, 2.6353, 2.7189], device='cuda:6'), covar=tensor([0.0052, 0.0166, 0.0205, 0.0202, 0.0081, 0.0140, 0.0080, 0.0093], device='cuda:6'), in_proj_covar=tensor([0.0083, 0.0151, 0.0151, 0.0148, 0.0147, 0.0154, 0.0130, 0.0139], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 07:19:57,731 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=44491.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:20:11,835 INFO [train.py:904] (6/8) Epoch 5, batch 3900, loss[loss=0.2355, simple_loss=0.3006, pruned_loss=0.08518, over 16296.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.2887, pruned_loss=0.07489, over 3278304.40 frames. ], batch size: 165, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:20:16,622 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7181, 4.0572, 4.2684, 1.9967, 4.4350, 4.5353, 3.1576, 3.1720], device='cuda:6'), covar=tensor([0.0686, 0.0119, 0.0121, 0.1148, 0.0048, 0.0030, 0.0327, 0.0393], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0086, 0.0082, 0.0139, 0.0072, 0.0078, 0.0115, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 07:20:39,194 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7648, 5.0521, 5.1756, 5.1726, 5.0443, 5.5991, 5.3287, 5.0399], device='cuda:6'), covar=tensor([0.0925, 0.1295, 0.1233, 0.1476, 0.2068, 0.0805, 0.1008, 0.1957], device='cuda:6'), in_proj_covar=tensor([0.0285, 0.0397, 0.0389, 0.0339, 0.0449, 0.0411, 0.0321, 0.0459], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 07:20:49,769 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=44527.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:21:02,882 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5816, 2.3417, 1.7020, 2.1652, 2.8937, 2.7064, 2.9575, 2.8653], device='cuda:6'), covar=tensor([0.0065, 0.0133, 0.0180, 0.0171, 0.0069, 0.0124, 0.0082, 0.0099], device='cuda:6'), in_proj_covar=tensor([0.0083, 0.0151, 0.0150, 0.0148, 0.0147, 0.0154, 0.0130, 0.0138], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 07:21:22,540 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=44550.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:21:23,251 INFO [train.py:904] (6/8) Epoch 5, batch 3950, loss[loss=0.1959, simple_loss=0.2729, pruned_loss=0.05944, over 17237.00 frames. ], tot_loss[loss=0.2193, simple_loss=0.2878, pruned_loss=0.07542, over 3280015.57 frames. ], batch size: 43, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:21:37,715 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.214e+02 3.071e+02 3.714e+02 4.451e+02 1.184e+03, threshold=7.429e+02, percent-clipped=3.0 2023-04-28 07:21:44,194 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44565.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:22:16,336 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44588.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:22:34,879 INFO [train.py:904] (6/8) Epoch 5, batch 4000, loss[loss=0.2316, simple_loss=0.2892, pruned_loss=0.08694, over 16425.00 frames. ], tot_loss[loss=0.219, simple_loss=0.2872, pruned_loss=0.07539, over 3281690.56 frames. ], batch size: 146, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:22:49,517 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44611.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:22:51,785 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=44613.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:22:54,866 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44615.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:23:40,942 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5298, 3.3272, 3.4074, 2.9310, 3.4228, 2.0226, 3.1109, 2.8455], device='cuda:6'), covar=tensor([0.0078, 0.0067, 0.0108, 0.0210, 0.0061, 0.1467, 0.0097, 0.0156], device='cuda:6'), in_proj_covar=tensor([0.0096, 0.0086, 0.0126, 0.0134, 0.0097, 0.0139, 0.0113, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 07:23:45,848 INFO [train.py:904] (6/8) Epoch 5, batch 4050, loss[loss=0.225, simple_loss=0.2898, pruned_loss=0.08012, over 12247.00 frames. ], tot_loss[loss=0.2169, simple_loss=0.2868, pruned_loss=0.07349, over 3274321.67 frames. ], batch size: 247, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:24:02,948 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.760e+02 3.326e+02 4.025e+02 7.816e+02, threshold=6.651e+02, percent-clipped=1.0 2023-04-28 07:24:04,435 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=44663.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:25:01,112 INFO [train.py:904] (6/8) Epoch 5, batch 4100, loss[loss=0.1878, simple_loss=0.2707, pruned_loss=0.05249, over 17276.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.2866, pruned_loss=0.07119, over 3283528.84 frames. ], batch size: 52, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:25:09,057 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7371, 2.9145, 2.5657, 4.3855, 3.8445, 3.9841, 1.6443, 3.0200], device='cuda:6'), covar=tensor([0.1269, 0.0572, 0.1074, 0.0090, 0.0259, 0.0291, 0.1344, 0.0706], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0141, 0.0165, 0.0087, 0.0181, 0.0171, 0.0158, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 07:26:16,728 INFO [train.py:904] (6/8) Epoch 5, batch 4150, loss[loss=0.2396, simple_loss=0.3218, pruned_loss=0.07869, over 16665.00 frames. ], tot_loss[loss=0.2232, simple_loss=0.2954, pruned_loss=0.07556, over 3240861.85 frames. ], batch size: 134, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:26:34,907 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.630e+02 2.818e+02 3.479e+02 4.379e+02 9.731e+02, threshold=6.958e+02, percent-clipped=3.0 2023-04-28 07:27:35,372 INFO [train.py:904] (6/8) Epoch 5, batch 4200, loss[loss=0.2263, simple_loss=0.3086, pruned_loss=0.07197, over 16489.00 frames. ], tot_loss[loss=0.2288, simple_loss=0.3028, pruned_loss=0.07736, over 3223920.80 frames. ], batch size: 68, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:28:40,860 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.4459, 2.9147, 2.3361, 4.5297, 3.7105, 4.0598, 1.5290, 2.9913], device='cuda:6'), covar=tensor([0.1481, 0.0565, 0.1192, 0.0060, 0.0206, 0.0288, 0.1460, 0.0720], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0135, 0.0161, 0.0082, 0.0169, 0.0166, 0.0154, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 07:28:51,232 INFO [train.py:904] (6/8) Epoch 5, batch 4250, loss[loss=0.1978, simple_loss=0.2943, pruned_loss=0.05062, over 16751.00 frames. ], tot_loss[loss=0.2302, simple_loss=0.3059, pruned_loss=0.0772, over 3201830.01 frames. ], batch size: 89, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:28:54,764 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 07:28:57,941 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5989, 2.8385, 2.2571, 4.1895, 3.4231, 3.8800, 1.4267, 2.8643], device='cuda:6'), covar=tensor([0.1282, 0.0492, 0.1126, 0.0060, 0.0224, 0.0285, 0.1376, 0.0693], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0135, 0.0161, 0.0082, 0.0168, 0.0166, 0.0154, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 07:29:07,183 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 3.072e+02 3.670e+02 4.674e+02 1.163e+03, threshold=7.340e+02, percent-clipped=6.0 2023-04-28 07:29:38,867 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44883.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:30:06,001 INFO [train.py:904] (6/8) Epoch 5, batch 4300, loss[loss=0.2428, simple_loss=0.3416, pruned_loss=0.07198, over 16798.00 frames. ], tot_loss[loss=0.2297, simple_loss=0.3075, pruned_loss=0.076, over 3209806.58 frames. ], batch size: 124, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:30:13,087 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44906.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:30:54,691 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.20 vs. limit=5.0 2023-04-28 07:31:19,635 INFO [train.py:904] (6/8) Epoch 5, batch 4350, loss[loss=0.2631, simple_loss=0.33, pruned_loss=0.09808, over 11541.00 frames. ], tot_loss[loss=0.2328, simple_loss=0.3109, pruned_loss=0.07741, over 3197835.23 frames. ], batch size: 246, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:31:36,572 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.041e+02 3.143e+02 3.858e+02 4.477e+02 9.466e+02, threshold=7.715e+02, percent-clipped=1.0 2023-04-28 07:32:35,462 INFO [train.py:904] (6/8) Epoch 5, batch 4400, loss[loss=0.2518, simple_loss=0.3267, pruned_loss=0.0885, over 16578.00 frames. ], tot_loss[loss=0.2352, simple_loss=0.3132, pruned_loss=0.07855, over 3206597.61 frames. ], batch size: 62, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:33:49,609 INFO [train.py:904] (6/8) Epoch 5, batch 4450, loss[loss=0.2442, simple_loss=0.3264, pruned_loss=0.08104, over 16644.00 frames. ], tot_loss[loss=0.236, simple_loss=0.3152, pruned_loss=0.07834, over 3224200.11 frames. ], batch size: 62, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:33:54,473 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-28 07:33:57,024 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45056.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:34:06,116 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.921e+02 2.841e+02 3.317e+02 4.350e+02 8.803e+02, threshold=6.635e+02, percent-clipped=2.0 2023-04-28 07:34:33,605 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45081.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:35:01,854 INFO [train.py:904] (6/8) Epoch 5, batch 4500, loss[loss=0.2178, simple_loss=0.3047, pruned_loss=0.06546, over 16664.00 frames. ], tot_loss[loss=0.2355, simple_loss=0.3149, pruned_loss=0.07799, over 3239219.62 frames. ], batch size: 76, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:35:26,545 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45117.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 07:36:00,926 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2965, 5.2995, 5.1453, 5.0728, 4.6669, 5.1089, 5.0855, 4.7621], device='cuda:6'), covar=tensor([0.0437, 0.0181, 0.0171, 0.0121, 0.0735, 0.0216, 0.0153, 0.0405], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0170, 0.0203, 0.0178, 0.0235, 0.0198, 0.0148, 0.0219], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 07:36:02,281 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45142.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:36:13,662 INFO [train.py:904] (6/8) Epoch 5, batch 4550, loss[loss=0.2597, simple_loss=0.3441, pruned_loss=0.08772, over 16426.00 frames. ], tot_loss[loss=0.2363, simple_loss=0.3155, pruned_loss=0.07855, over 3237621.39 frames. ], batch size: 75, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:36:30,358 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.972e+02 2.621e+02 3.061e+02 3.788e+02 6.051e+02, threshold=6.121e+02, percent-clipped=0.0 2023-04-28 07:37:00,753 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45183.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:37:27,614 INFO [train.py:904] (6/8) Epoch 5, batch 4600, loss[loss=0.2064, simple_loss=0.2915, pruned_loss=0.06063, over 16552.00 frames. ], tot_loss[loss=0.2364, simple_loss=0.3162, pruned_loss=0.07829, over 3235258.05 frames. ], batch size: 75, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:37:35,943 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1965, 4.1075, 4.5711, 4.5301, 4.5308, 4.0961, 4.1424, 3.9825], device='cuda:6'), covar=tensor([0.0205, 0.0294, 0.0253, 0.0327, 0.0283, 0.0256, 0.0783, 0.0400], device='cuda:6'), in_proj_covar=tensor([0.0230, 0.0219, 0.0224, 0.0228, 0.0272, 0.0241, 0.0347, 0.0202], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 07:37:35,971 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45206.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:38:12,281 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=45231.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:38:14,977 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5875, 2.7072, 2.2754, 4.1779, 3.2547, 3.7321, 1.5573, 2.8275], device='cuda:6'), covar=tensor([0.1378, 0.0566, 0.1148, 0.0058, 0.0188, 0.0319, 0.1365, 0.0716], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0138, 0.0163, 0.0082, 0.0170, 0.0165, 0.0157, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 07:38:29,996 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-28 07:38:40,728 INFO [train.py:904] (6/8) Epoch 5, batch 4650, loss[loss=0.2152, simple_loss=0.2936, pruned_loss=0.06841, over 16687.00 frames. ], tot_loss[loss=0.2353, simple_loss=0.3147, pruned_loss=0.07794, over 3233621.98 frames. ], batch size: 62, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:38:45,219 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=45254.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:38:57,179 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.863e+02 2.661e+02 3.124e+02 3.951e+02 6.864e+02, threshold=6.249e+02, percent-clipped=3.0 2023-04-28 07:39:20,908 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9227, 4.1679, 1.9438, 4.6920, 2.8442, 4.5467, 2.1019, 3.0510], device='cuda:6'), covar=tensor([0.0124, 0.0191, 0.1689, 0.0027, 0.0779, 0.0176, 0.1399, 0.0603], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0151, 0.0176, 0.0081, 0.0161, 0.0182, 0.0185, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 07:39:54,560 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0541, 4.9855, 4.8599, 4.7502, 4.3154, 4.8392, 4.8038, 4.6411], device='cuda:6'), covar=tensor([0.0389, 0.0301, 0.0173, 0.0138, 0.0893, 0.0353, 0.0195, 0.0384], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0173, 0.0205, 0.0179, 0.0237, 0.0202, 0.0150, 0.0218], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 07:39:55,268 INFO [train.py:904] (6/8) Epoch 5, batch 4700, loss[loss=0.2059, simple_loss=0.2864, pruned_loss=0.06271, over 16624.00 frames. ], tot_loss[loss=0.2317, simple_loss=0.3113, pruned_loss=0.07608, over 3236715.23 frames. ], batch size: 57, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:41:03,085 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45348.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 07:41:06,793 INFO [train.py:904] (6/8) Epoch 5, batch 4750, loss[loss=0.1878, simple_loss=0.2752, pruned_loss=0.05021, over 16669.00 frames. ], tot_loss[loss=0.2282, simple_loss=0.3075, pruned_loss=0.07445, over 3230033.62 frames. ], batch size: 76, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:41:22,807 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.785e+02 2.710e+02 3.330e+02 4.145e+02 7.902e+02, threshold=6.661e+02, percent-clipped=5.0 2023-04-28 07:41:30,377 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7290, 3.9502, 1.5841, 4.5706, 2.7103, 4.3815, 1.8285, 2.9104], device='cuda:6'), covar=tensor([0.0152, 0.0189, 0.1821, 0.0024, 0.0667, 0.0196, 0.1648, 0.0504], device='cuda:6'), in_proj_covar=tensor([0.0118, 0.0149, 0.0173, 0.0079, 0.0158, 0.0180, 0.0183, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 07:41:45,738 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9878, 1.4828, 2.0789, 2.8009, 2.7478, 3.1654, 1.7459, 2.9857], device='cuda:6'), covar=tensor([0.0056, 0.0237, 0.0151, 0.0112, 0.0094, 0.0050, 0.0202, 0.0043], device='cuda:6'), in_proj_covar=tensor([0.0115, 0.0141, 0.0127, 0.0125, 0.0125, 0.0089, 0.0137, 0.0081], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 07:42:18,716 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-28 07:42:20,546 INFO [train.py:904] (6/8) Epoch 5, batch 4800, loss[loss=0.2189, simple_loss=0.2948, pruned_loss=0.07146, over 16659.00 frames. ], tot_loss[loss=0.2249, simple_loss=0.3042, pruned_loss=0.07279, over 3230393.64 frames. ], batch size: 62, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:42:33,325 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45409.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 07:42:38,678 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45412.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 07:43:14,515 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45437.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:43:34,380 INFO [train.py:904] (6/8) Epoch 5, batch 4850, loss[loss=0.204, simple_loss=0.3004, pruned_loss=0.05377, over 16231.00 frames. ], tot_loss[loss=0.2256, simple_loss=0.3053, pruned_loss=0.07298, over 3215051.45 frames. ], batch size: 165, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:43:50,675 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 2.510e+02 3.182e+02 3.913e+02 7.711e+02, threshold=6.364e+02, percent-clipped=1.0 2023-04-28 07:44:47,402 INFO [train.py:904] (6/8) Epoch 5, batch 4900, loss[loss=0.2036, simple_loss=0.2938, pruned_loss=0.05666, over 16854.00 frames. ], tot_loss[loss=0.2244, simple_loss=0.3051, pruned_loss=0.07183, over 3203161.01 frames. ], batch size: 83, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:44:50,554 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.79 vs. limit=5.0 2023-04-28 07:46:01,013 INFO [train.py:904] (6/8) Epoch 5, batch 4950, loss[loss=0.2371, simple_loss=0.322, pruned_loss=0.0761, over 16732.00 frames. ], tot_loss[loss=0.225, simple_loss=0.3056, pruned_loss=0.07221, over 3188755.71 frames. ], batch size: 83, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:46:09,461 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2650, 4.6149, 4.6262, 2.5318, 3.8700, 4.5654, 4.2563, 2.2095], device='cuda:6'), covar=tensor([0.0312, 0.0010, 0.0018, 0.0255, 0.0032, 0.0025, 0.0023, 0.0285], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0053, 0.0057, 0.0113, 0.0059, 0.0065, 0.0062, 0.0105], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 07:46:15,482 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.470e+02 3.018e+02 3.573e+02 4.428e+02 9.346e+02, threshold=7.147e+02, percent-clipped=9.0 2023-04-28 07:47:13,802 INFO [train.py:904] (6/8) Epoch 5, batch 5000, loss[loss=0.2216, simple_loss=0.3051, pruned_loss=0.06902, over 16725.00 frames. ], tot_loss[loss=0.2265, simple_loss=0.3075, pruned_loss=0.07272, over 3183402.80 frames. ], batch size: 124, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:48:18,778 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-28 07:48:27,073 INFO [train.py:904] (6/8) Epoch 5, batch 5050, loss[loss=0.2267, simple_loss=0.3099, pruned_loss=0.07178, over 17213.00 frames. ], tot_loss[loss=0.226, simple_loss=0.3078, pruned_loss=0.07214, over 3196864.05 frames. ], batch size: 45, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:48:43,127 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.866e+02 2.781e+02 3.302e+02 3.985e+02 6.386e+02, threshold=6.604e+02, percent-clipped=0.0 2023-04-28 07:48:48,932 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-28 07:49:18,906 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4750, 3.3378, 2.8149, 1.8026, 2.6486, 2.0911, 2.9821, 3.1363], device='cuda:6'), covar=tensor([0.0286, 0.0468, 0.0589, 0.1530, 0.0722, 0.0859, 0.0671, 0.0639], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0128, 0.0154, 0.0140, 0.0134, 0.0126, 0.0140, 0.0137], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 07:49:39,972 INFO [train.py:904] (6/8) Epoch 5, batch 5100, loss[loss=0.2086, simple_loss=0.2968, pruned_loss=0.06023, over 16577.00 frames. ], tot_loss[loss=0.2238, simple_loss=0.3057, pruned_loss=0.07096, over 3202953.03 frames. ], batch size: 68, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:49:42,330 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45702.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:49:44,513 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45704.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 07:49:56,759 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45712.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 07:50:06,229 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9629, 3.6837, 3.6152, 1.3786, 3.8627, 3.8921, 2.8515, 2.4901], device='cuda:6'), covar=tensor([0.1065, 0.0088, 0.0121, 0.1386, 0.0049, 0.0055, 0.0366, 0.0618], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0087, 0.0079, 0.0140, 0.0069, 0.0074, 0.0115, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 07:50:33,814 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45737.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:50:48,595 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-28 07:50:53,044 INFO [train.py:904] (6/8) Epoch 5, batch 5150, loss[loss=0.2095, simple_loss=0.299, pruned_loss=0.05998, over 16829.00 frames. ], tot_loss[loss=0.2227, simple_loss=0.3056, pruned_loss=0.06988, over 3206336.21 frames. ], batch size: 116, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:51:07,065 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=45760.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:51:08,845 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.833e+02 2.626e+02 3.241e+02 3.847e+02 8.905e+02, threshold=6.482e+02, percent-clipped=5.0 2023-04-28 07:51:10,597 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45763.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:51:26,070 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45774.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:51:43,500 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=45785.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:52:03,152 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-04-28 07:52:05,635 INFO [train.py:904] (6/8) Epoch 5, batch 5200, loss[loss=0.196, simple_loss=0.2796, pruned_loss=0.05624, over 16763.00 frames. ], tot_loss[loss=0.222, simple_loss=0.3043, pruned_loss=0.06984, over 3209591.83 frames. ], batch size: 83, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:52:54,232 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45835.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:53:16,335 INFO [train.py:904] (6/8) Epoch 5, batch 5250, loss[loss=0.223, simple_loss=0.3059, pruned_loss=0.07003, over 16325.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.3019, pruned_loss=0.06935, over 3208428.06 frames. ], batch size: 165, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:53:27,269 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45859.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:53:31,050 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.833e+02 2.734e+02 3.218e+02 3.993e+02 9.159e+02, threshold=6.436e+02, percent-clipped=4.0 2023-04-28 07:53:42,952 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5115, 5.4122, 5.2644, 4.5621, 5.3573, 2.2493, 5.1865, 5.3367], device='cuda:6'), covar=tensor([0.0039, 0.0044, 0.0068, 0.0312, 0.0050, 0.1429, 0.0064, 0.0100], device='cuda:6'), in_proj_covar=tensor([0.0088, 0.0077, 0.0115, 0.0125, 0.0089, 0.0134, 0.0103, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 07:53:46,415 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0071, 4.7630, 4.9333, 5.2860, 5.4641, 4.7759, 5.4046, 5.4025], device='cuda:6'), covar=tensor([0.0880, 0.0597, 0.1075, 0.0377, 0.0313, 0.0456, 0.0280, 0.0270], device='cuda:6'), in_proj_covar=tensor([0.0378, 0.0457, 0.0586, 0.0470, 0.0351, 0.0350, 0.0366, 0.0386], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 07:54:26,347 INFO [train.py:904] (6/8) Epoch 5, batch 5300, loss[loss=0.1661, simple_loss=0.2497, pruned_loss=0.04121, over 16823.00 frames. ], tot_loss[loss=0.2169, simple_loss=0.2981, pruned_loss=0.06781, over 3220162.53 frames. ], batch size: 83, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:54:53,089 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45920.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:55:36,415 INFO [train.py:904] (6/8) Epoch 5, batch 5350, loss[loss=0.2226, simple_loss=0.3084, pruned_loss=0.06835, over 16923.00 frames. ], tot_loss[loss=0.2149, simple_loss=0.296, pruned_loss=0.06695, over 3212283.73 frames. ], batch size: 96, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 07:55:53,055 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.681e+02 2.936e+02 3.263e+02 4.063e+02 1.076e+03, threshold=6.526e+02, percent-clipped=3.0 2023-04-28 07:56:04,525 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-28 07:56:30,340 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6707, 1.2077, 1.5108, 1.6881, 1.8515, 1.9540, 1.5021, 1.8119], device='cuda:6'), covar=tensor([0.0081, 0.0180, 0.0087, 0.0130, 0.0091, 0.0049, 0.0167, 0.0039], device='cuda:6'), in_proj_covar=tensor([0.0118, 0.0142, 0.0125, 0.0123, 0.0124, 0.0087, 0.0139, 0.0080], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 07:56:52,487 INFO [train.py:904] (6/8) Epoch 5, batch 5400, loss[loss=0.2137, simple_loss=0.3051, pruned_loss=0.06109, over 16492.00 frames. ], tot_loss[loss=0.2174, simple_loss=0.2987, pruned_loss=0.06807, over 3202945.60 frames. ], batch size: 75, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 07:56:58,419 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46004.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 07:57:06,461 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3067, 4.3016, 4.1193, 4.0086, 3.7101, 4.1700, 4.0487, 3.9798], device='cuda:6'), covar=tensor([0.0394, 0.0315, 0.0215, 0.0184, 0.0867, 0.0308, 0.0377, 0.0406], device='cuda:6'), in_proj_covar=tensor([0.0182, 0.0190, 0.0218, 0.0190, 0.0247, 0.0217, 0.0155, 0.0234], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 07:57:12,250 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.12 vs. limit=5.0 2023-04-28 07:57:58,013 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-28 07:58:09,281 INFO [train.py:904] (6/8) Epoch 5, batch 5450, loss[loss=0.2992, simple_loss=0.3684, pruned_loss=0.115, over 15344.00 frames. ], tot_loss[loss=0.2224, simple_loss=0.3032, pruned_loss=0.0708, over 3187510.39 frames. ], batch size: 190, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 07:58:11,185 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=46052.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 07:58:19,248 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46058.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:58:24,986 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.677e+02 3.104e+02 3.655e+02 4.572e+02 1.003e+03, threshold=7.310e+02, percent-clipped=8.0 2023-04-28 07:59:09,161 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.78 vs. limit=5.0 2023-04-28 07:59:10,663 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 07:59:22,159 INFO [train.py:904] (6/8) Epoch 5, batch 5500, loss[loss=0.255, simple_loss=0.3302, pruned_loss=0.08992, over 15380.00 frames. ], tot_loss[loss=0.2336, simple_loss=0.3124, pruned_loss=0.0774, over 3173362.87 frames. ], batch size: 190, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:00:07,587 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46130.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:00:39,133 INFO [train.py:904] (6/8) Epoch 5, batch 5550, loss[loss=0.3965, simple_loss=0.4306, pruned_loss=0.1812, over 11091.00 frames. ], tot_loss[loss=0.2464, simple_loss=0.3216, pruned_loss=0.08554, over 3118831.83 frames. ], batch size: 248, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:00:56,874 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.500e+02 4.429e+02 5.470e+02 6.755e+02 1.488e+03, threshold=1.094e+03, percent-clipped=17.0 2023-04-28 08:01:58,727 INFO [train.py:904] (6/8) Epoch 5, batch 5600, loss[loss=0.3016, simple_loss=0.3683, pruned_loss=0.1175, over 15197.00 frames. ], tot_loss[loss=0.2553, simple_loss=0.3283, pruned_loss=0.0912, over 3100659.26 frames. ], batch size: 190, lr: 1.33e-02, grad_scale: 16.0 2023-04-28 08:02:23,964 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46215.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:02:27,521 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5632, 2.6720, 1.7538, 2.2405, 3.0353, 2.6249, 3.6120, 3.2508], device='cuda:6'), covar=tensor([0.0022, 0.0180, 0.0293, 0.0220, 0.0104, 0.0180, 0.0058, 0.0095], device='cuda:6'), in_proj_covar=tensor([0.0077, 0.0151, 0.0154, 0.0149, 0.0146, 0.0154, 0.0122, 0.0133], device='cuda:6'), out_proj_covar=tensor([9.6626e-05, 1.8487e-04, 1.8542e-04, 1.7970e-04, 1.8181e-04, 1.8949e-04, 1.4590e-04, 1.6463e-04], device='cuda:6') 2023-04-28 08:02:32,053 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7709, 3.0858, 3.0746, 1.9168, 2.7841, 3.0561, 3.0814, 1.8207], device='cuda:6'), covar=tensor([0.0361, 0.0028, 0.0045, 0.0285, 0.0071, 0.0071, 0.0036, 0.0312], device='cuda:6'), in_proj_covar=tensor([0.0117, 0.0055, 0.0060, 0.0116, 0.0060, 0.0069, 0.0064, 0.0109], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 08:03:19,339 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2212, 3.3103, 1.5757, 3.5383, 2.2637, 3.4825, 1.9230, 2.6445], device='cuda:6'), covar=tensor([0.0144, 0.0303, 0.1684, 0.0045, 0.0809, 0.0354, 0.1385, 0.0610], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0151, 0.0177, 0.0078, 0.0162, 0.0182, 0.0188, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 08:03:23,171 INFO [train.py:904] (6/8) Epoch 5, batch 5650, loss[loss=0.2421, simple_loss=0.325, pruned_loss=0.0796, over 16661.00 frames. ], tot_loss[loss=0.2649, simple_loss=0.335, pruned_loss=0.09739, over 3065968.95 frames. ], batch size: 62, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:03:34,503 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9643, 5.7799, 5.9226, 5.6836, 5.6484, 6.1661, 5.8805, 5.7169], device='cuda:6'), covar=tensor([0.0896, 0.1421, 0.1147, 0.1576, 0.2336, 0.0936, 0.1230, 0.2172], device='cuda:6'), in_proj_covar=tensor([0.0275, 0.0374, 0.0370, 0.0326, 0.0436, 0.0393, 0.0308, 0.0449], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 08:03:42,413 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.653e+02 4.676e+02 6.240e+02 7.908e+02 1.367e+03, threshold=1.248e+03, percent-clipped=1.0 2023-04-28 08:04:04,261 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46276.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:04:43,824 INFO [train.py:904] (6/8) Epoch 5, batch 5700, loss[loss=0.2374, simple_loss=0.3301, pruned_loss=0.07233, over 16682.00 frames. ], tot_loss[loss=0.2666, simple_loss=0.3362, pruned_loss=0.09848, over 3065159.97 frames. ], batch size: 89, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:05:41,869 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46337.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:06:04,707 INFO [train.py:904] (6/8) Epoch 5, batch 5750, loss[loss=0.2593, simple_loss=0.3347, pruned_loss=0.09201, over 16920.00 frames. ], tot_loss[loss=0.2709, simple_loss=0.34, pruned_loss=0.1009, over 3030880.34 frames. ], batch size: 109, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:06:16,599 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46358.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:06:23,495 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.622e+02 4.192e+02 4.948e+02 6.377e+02 1.174e+03, threshold=9.897e+02, percent-clipped=0.0 2023-04-28 08:06:53,514 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1428, 3.0038, 2.6372, 2.0320, 2.5311, 2.0823, 2.6817, 2.8009], device='cuda:6'), covar=tensor([0.0368, 0.0454, 0.0543, 0.1366, 0.0659, 0.0876, 0.0598, 0.0518], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0126, 0.0152, 0.0140, 0.0134, 0.0125, 0.0139, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 08:07:26,296 INFO [train.py:904] (6/8) Epoch 5, batch 5800, loss[loss=0.3205, simple_loss=0.3621, pruned_loss=0.1395, over 12079.00 frames. ], tot_loss[loss=0.2703, simple_loss=0.34, pruned_loss=0.1003, over 3014703.76 frames. ], batch size: 248, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:07:29,212 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9334, 3.9119, 3.8165, 3.3241, 3.8473, 1.8187, 3.6696, 3.7352], device='cuda:6'), covar=tensor([0.0073, 0.0063, 0.0096, 0.0251, 0.0063, 0.1659, 0.0085, 0.0131], device='cuda:6'), in_proj_covar=tensor([0.0089, 0.0077, 0.0117, 0.0126, 0.0089, 0.0140, 0.0104, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:07:35,756 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=46406.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:07:58,732 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-28 08:08:14,580 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46430.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:08:20,556 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-28 08:08:46,735 INFO [train.py:904] (6/8) Epoch 5, batch 5850, loss[loss=0.2204, simple_loss=0.2957, pruned_loss=0.07262, over 17117.00 frames. ], tot_loss[loss=0.2657, simple_loss=0.3367, pruned_loss=0.09731, over 3025553.15 frames. ], batch size: 48, lr: 1.33e-02, grad_scale: 4.0 2023-04-28 08:09:06,683 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.458e+02 3.799e+02 4.852e+02 6.197e+02 1.255e+03, threshold=9.704e+02, percent-clipped=4.0 2023-04-28 08:09:29,289 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=46478.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:10:09,066 INFO [train.py:904] (6/8) Epoch 5, batch 5900, loss[loss=0.2429, simple_loss=0.3212, pruned_loss=0.08228, over 16721.00 frames. ], tot_loss[loss=0.2639, simple_loss=0.3357, pruned_loss=0.09603, over 3055957.01 frames. ], batch size: 83, lr: 1.33e-02, grad_scale: 4.0 2023-04-28 08:10:36,202 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46515.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:10:53,851 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9309, 3.8907, 4.3411, 4.2730, 4.2960, 3.8720, 3.9446, 3.8740], device='cuda:6'), covar=tensor([0.0269, 0.0408, 0.0323, 0.0438, 0.0392, 0.0348, 0.0914, 0.0405], device='cuda:6'), in_proj_covar=tensor([0.0231, 0.0221, 0.0228, 0.0229, 0.0277, 0.0241, 0.0345, 0.0206], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 08:11:31,951 INFO [train.py:904] (6/8) Epoch 5, batch 5950, loss[loss=0.2507, simple_loss=0.3337, pruned_loss=0.08389, over 16402.00 frames. ], tot_loss[loss=0.2616, simple_loss=0.3356, pruned_loss=0.09379, over 3069278.30 frames. ], batch size: 146, lr: 1.33e-02, grad_scale: 4.0 2023-04-28 08:11:52,755 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=46563.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:11:53,494 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.225e+02 3.478e+02 4.456e+02 5.303e+02 9.511e+02, threshold=8.911e+02, percent-clipped=0.0 2023-04-28 08:12:43,367 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46595.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:12:51,974 INFO [train.py:904] (6/8) Epoch 5, batch 6000, loss[loss=0.248, simple_loss=0.3289, pruned_loss=0.08356, over 16690.00 frames. ], tot_loss[loss=0.2602, simple_loss=0.3346, pruned_loss=0.09289, over 3084118.07 frames. ], batch size: 134, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:12:51,974 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 08:13:04,052 INFO [train.py:938] (6/8) Epoch 5, validation: loss=0.1879, simple_loss=0.2992, pruned_loss=0.03826, over 944034.00 frames. 2023-04-28 08:13:04,052 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-28 08:13:05,707 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46602.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:13:51,745 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46632.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:14:27,390 INFO [train.py:904] (6/8) Epoch 5, batch 6050, loss[loss=0.3039, simple_loss=0.3539, pruned_loss=0.1269, over 11987.00 frames. ], tot_loss[loss=0.2583, simple_loss=0.333, pruned_loss=0.09179, over 3095701.92 frames. ], batch size: 248, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:14:37,547 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46656.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 08:14:40,464 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0175, 5.0046, 4.7619, 4.1080, 4.8087, 1.9706, 4.6018, 4.7219], device='cuda:6'), covar=tensor([0.0062, 0.0047, 0.0088, 0.0319, 0.0055, 0.1617, 0.0097, 0.0129], device='cuda:6'), in_proj_covar=tensor([0.0090, 0.0078, 0.0118, 0.0125, 0.0090, 0.0140, 0.0105, 0.0117], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:14:48,382 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46663.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:14:48,996 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.236e+02 4.108e+02 4.980e+02 6.205e+02 1.363e+03, threshold=9.960e+02, percent-clipped=7.0 2023-04-28 08:14:49,987 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-28 08:15:45,451 INFO [train.py:904] (6/8) Epoch 5, batch 6100, loss[loss=0.2611, simple_loss=0.3188, pruned_loss=0.1017, over 11910.00 frames. ], tot_loss[loss=0.2578, simple_loss=0.3323, pruned_loss=0.0916, over 3065996.74 frames. ], batch size: 248, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:16:45,839 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9034, 2.1910, 2.2673, 4.4441, 2.1033, 3.2492, 2.3587, 2.4846], device='cuda:6'), covar=tensor([0.0503, 0.1943, 0.1098, 0.0254, 0.2725, 0.0859, 0.1761, 0.2282], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0314, 0.0259, 0.0306, 0.0369, 0.0301, 0.0281, 0.0377], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:17:01,186 INFO [train.py:904] (6/8) Epoch 5, batch 6150, loss[loss=0.2723, simple_loss=0.3434, pruned_loss=0.1006, over 15281.00 frames. ], tot_loss[loss=0.2566, simple_loss=0.3307, pruned_loss=0.09121, over 3065657.09 frames. ], batch size: 190, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:17:22,716 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.183e+02 3.664e+02 4.778e+02 7.050e+02 1.596e+03, threshold=9.557e+02, percent-clipped=5.0 2023-04-28 08:17:28,686 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1786, 4.2357, 4.2887, 4.3184, 4.2853, 4.7510, 4.3849, 4.1527], device='cuda:6'), covar=tensor([0.1231, 0.1463, 0.1261, 0.1312, 0.2109, 0.0842, 0.1090, 0.2011], device='cuda:6'), in_proj_covar=tensor([0.0274, 0.0377, 0.0378, 0.0330, 0.0440, 0.0394, 0.0313, 0.0456], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 08:18:21,097 INFO [train.py:904] (6/8) Epoch 5, batch 6200, loss[loss=0.2615, simple_loss=0.3287, pruned_loss=0.09709, over 16218.00 frames. ], tot_loss[loss=0.2541, simple_loss=0.3283, pruned_loss=0.08997, over 3068192.48 frames. ], batch size: 165, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:18:36,417 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46810.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:18:37,750 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8968, 1.7573, 1.5083, 1.4484, 1.8763, 1.7275, 1.7454, 1.9338], device='cuda:6'), covar=tensor([0.0034, 0.0100, 0.0142, 0.0145, 0.0076, 0.0109, 0.0061, 0.0084], device='cuda:6'), in_proj_covar=tensor([0.0076, 0.0151, 0.0153, 0.0149, 0.0144, 0.0153, 0.0124, 0.0133], device='cuda:6'), out_proj_covar=tensor([9.5103e-05, 1.8487e-04, 1.8330e-04, 1.7902e-04, 1.7877e-04, 1.8810e-04, 1.4771e-04, 1.6476e-04], device='cuda:6') 2023-04-28 08:18:57,910 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1993, 1.9324, 2.0340, 3.6138, 1.8254, 2.7394, 2.1291, 1.9837], device='cuda:6'), covar=tensor([0.0618, 0.2050, 0.1110, 0.0353, 0.2871, 0.0983, 0.1821, 0.2254], device='cuda:6'), in_proj_covar=tensor([0.0310, 0.0311, 0.0257, 0.0305, 0.0367, 0.0299, 0.0280, 0.0375], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:19:02,139 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5703, 2.6382, 1.6320, 2.7671, 2.1764, 2.7327, 1.8992, 2.3625], device='cuda:6'), covar=tensor([0.0137, 0.0288, 0.1140, 0.0064, 0.0502, 0.0366, 0.1010, 0.0446], device='cuda:6'), in_proj_covar=tensor([0.0118, 0.0146, 0.0178, 0.0080, 0.0160, 0.0183, 0.0185, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 08:19:30,207 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1258, 4.2045, 4.1959, 3.0287, 3.8547, 4.0457, 3.9314, 2.2494], device='cuda:6'), covar=tensor([0.0375, 0.0020, 0.0030, 0.0201, 0.0040, 0.0081, 0.0035, 0.0300], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0052, 0.0057, 0.0113, 0.0058, 0.0068, 0.0062, 0.0106], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 08:19:37,520 INFO [train.py:904] (6/8) Epoch 5, batch 6250, loss[loss=0.2529, simple_loss=0.3399, pruned_loss=0.08293, over 16651.00 frames. ], tot_loss[loss=0.2537, simple_loss=0.3279, pruned_loss=0.08974, over 3078402.94 frames. ], batch size: 76, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:19:57,338 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.405e+02 3.628e+02 4.524e+02 5.799e+02 1.377e+03, threshold=9.049e+02, percent-clipped=5.0 2023-04-28 08:20:10,111 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46871.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:20:34,963 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-28 08:20:55,067 INFO [train.py:904] (6/8) Epoch 5, batch 6300, loss[loss=0.2416, simple_loss=0.3191, pruned_loss=0.08206, over 16873.00 frames. ], tot_loss[loss=0.2524, simple_loss=0.3275, pruned_loss=0.08868, over 3087403.37 frames. ], batch size: 83, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:21:43,897 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46932.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:22:12,093 INFO [train.py:904] (6/8) Epoch 5, batch 6350, loss[loss=0.2213, simple_loss=0.3019, pruned_loss=0.07041, over 16933.00 frames. ], tot_loss[loss=0.2551, simple_loss=0.3291, pruned_loss=0.0905, over 3098013.07 frames. ], batch size: 96, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:22:13,436 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46951.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 08:22:22,790 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46958.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:22:23,059 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7259, 3.9234, 3.1452, 2.4228, 2.8855, 2.2915, 4.0762, 4.0948], device='cuda:6'), covar=tensor([0.2124, 0.0587, 0.1183, 0.1548, 0.1977, 0.1405, 0.0406, 0.0520], device='cuda:6'), in_proj_covar=tensor([0.0277, 0.0249, 0.0267, 0.0242, 0.0291, 0.0200, 0.0239, 0.0251], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:22:31,677 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.349e+02 3.778e+02 4.701e+02 5.946e+02 1.466e+03, threshold=9.402e+02, percent-clipped=8.0 2023-04-28 08:22:55,750 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-28 08:22:57,125 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=46980.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:23:28,011 INFO [train.py:904] (6/8) Epoch 5, batch 6400, loss[loss=0.2558, simple_loss=0.3212, pruned_loss=0.09516, over 16268.00 frames. ], tot_loss[loss=0.2548, simple_loss=0.3284, pruned_loss=0.09058, over 3113865.21 frames. ], batch size: 35, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:24:34,461 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8317, 2.6849, 2.5889, 1.8812, 2.5431, 2.6330, 2.6345, 1.7547], device='cuda:6'), covar=tensor([0.0251, 0.0031, 0.0042, 0.0194, 0.0047, 0.0050, 0.0037, 0.0259], device='cuda:6'), in_proj_covar=tensor([0.0112, 0.0051, 0.0057, 0.0111, 0.0057, 0.0066, 0.0061, 0.0105], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 08:24:42,540 INFO [train.py:904] (6/8) Epoch 5, batch 6450, loss[loss=0.2172, simple_loss=0.2984, pruned_loss=0.06801, over 16270.00 frames. ], tot_loss[loss=0.2527, simple_loss=0.3271, pruned_loss=0.08916, over 3106771.69 frames. ], batch size: 165, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:25:02,352 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.185e+02 3.689e+02 4.668e+02 5.826e+02 1.210e+03, threshold=9.337e+02, percent-clipped=4.0 2023-04-28 08:25:59,689 INFO [train.py:904] (6/8) Epoch 5, batch 6500, loss[loss=0.2688, simple_loss=0.3193, pruned_loss=0.1091, over 11466.00 frames. ], tot_loss[loss=0.2502, simple_loss=0.3246, pruned_loss=0.08784, over 3110462.88 frames. ], batch size: 246, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:26:12,814 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3878, 3.7305, 3.8690, 1.5360, 4.1222, 4.1661, 2.7904, 2.9960], device='cuda:6'), covar=tensor([0.0826, 0.0142, 0.0182, 0.1420, 0.0054, 0.0062, 0.0432, 0.0432], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0089, 0.0081, 0.0140, 0.0071, 0.0077, 0.0116, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 08:26:46,696 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8560, 5.4346, 5.5503, 5.4862, 5.4753, 6.0552, 5.5998, 5.3620], device='cuda:6'), covar=tensor([0.0769, 0.1409, 0.1458, 0.1583, 0.2328, 0.0782, 0.1127, 0.2151], device='cuda:6'), in_proj_covar=tensor([0.0274, 0.0371, 0.0380, 0.0327, 0.0434, 0.0396, 0.0308, 0.0451], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 08:27:18,005 INFO [train.py:904] (6/8) Epoch 5, batch 6550, loss[loss=0.2299, simple_loss=0.3329, pruned_loss=0.06341, over 16808.00 frames. ], tot_loss[loss=0.2522, simple_loss=0.3271, pruned_loss=0.08861, over 3103717.18 frames. ], batch size: 102, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:27:37,107 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.214e+02 3.842e+02 4.732e+02 5.920e+02 1.050e+03, threshold=9.464e+02, percent-clipped=2.0 2023-04-28 08:27:41,302 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47166.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:28:33,655 INFO [train.py:904] (6/8) Epoch 5, batch 6600, loss[loss=0.2503, simple_loss=0.3254, pruned_loss=0.08757, over 16655.00 frames. ], tot_loss[loss=0.2545, simple_loss=0.3299, pruned_loss=0.08957, over 3104605.56 frames. ], batch size: 76, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:29:35,094 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47241.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:29:51,328 INFO [train.py:904] (6/8) Epoch 5, batch 6650, loss[loss=0.2503, simple_loss=0.3252, pruned_loss=0.08764, over 15335.00 frames. ], tot_loss[loss=0.258, simple_loss=0.3322, pruned_loss=0.09193, over 3094036.48 frames. ], batch size: 190, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:29:52,504 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47251.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:30:02,675 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47258.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:30:11,434 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.376e+02 3.750e+02 4.777e+02 6.524e+02 9.688e+02, threshold=9.554e+02, percent-clipped=1.0 2023-04-28 08:31:04,889 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=47299.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:31:06,905 INFO [train.py:904] (6/8) Epoch 5, batch 6700, loss[loss=0.329, simple_loss=0.3623, pruned_loss=0.1479, over 11569.00 frames. ], tot_loss[loss=0.2577, simple_loss=0.3311, pruned_loss=0.09216, over 3075055.77 frames. ], batch size: 248, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:31:09,284 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47302.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:31:16,567 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=47306.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:32:02,825 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4828, 3.4274, 3.3983, 2.8657, 3.3787, 1.9683, 3.1569, 2.8316], device='cuda:6'), covar=tensor([0.0075, 0.0058, 0.0094, 0.0219, 0.0055, 0.1555, 0.0085, 0.0120], device='cuda:6'), in_proj_covar=tensor([0.0090, 0.0078, 0.0119, 0.0124, 0.0089, 0.0141, 0.0105, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:32:25,998 INFO [train.py:904] (6/8) Epoch 5, batch 6750, loss[loss=0.2776, simple_loss=0.3593, pruned_loss=0.09792, over 15347.00 frames. ], tot_loss[loss=0.257, simple_loss=0.3299, pruned_loss=0.09198, over 3071166.50 frames. ], batch size: 190, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:32:45,841 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.669e+02 3.980e+02 4.802e+02 5.958e+02 9.394e+02, threshold=9.603e+02, percent-clipped=0.0 2023-04-28 08:33:09,254 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47379.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:33:40,654 INFO [train.py:904] (6/8) Epoch 5, batch 6800, loss[loss=0.2782, simple_loss=0.3502, pruned_loss=0.1031, over 15254.00 frames. ], tot_loss[loss=0.256, simple_loss=0.3295, pruned_loss=0.09124, over 3078181.70 frames. ], batch size: 190, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:33:51,941 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9826, 3.5271, 3.5885, 2.2704, 3.3825, 3.4589, 3.5064, 1.8832], device='cuda:6'), covar=tensor([0.0325, 0.0023, 0.0028, 0.0241, 0.0036, 0.0060, 0.0028, 0.0295], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0053, 0.0057, 0.0112, 0.0058, 0.0068, 0.0062, 0.0106], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 08:34:41,093 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47440.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:34:57,631 INFO [train.py:904] (6/8) Epoch 5, batch 6850, loss[loss=0.2852, simple_loss=0.3376, pruned_loss=0.1164, over 11635.00 frames. ], tot_loss[loss=0.257, simple_loss=0.3305, pruned_loss=0.09172, over 3077128.71 frames. ], batch size: 250, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:35:17,575 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.600e+02 3.467e+02 4.392e+02 5.504e+02 1.073e+03, threshold=8.784e+02, percent-clipped=2.0 2023-04-28 08:35:20,998 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47466.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:36:12,141 INFO [train.py:904] (6/8) Epoch 5, batch 6900, loss[loss=0.2849, simple_loss=0.3591, pruned_loss=0.1053, over 16724.00 frames. ], tot_loss[loss=0.258, simple_loss=0.3327, pruned_loss=0.09164, over 3079682.70 frames. ], batch size: 124, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:36:33,890 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=47514.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:36:42,500 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4489, 4.4609, 4.3573, 3.7372, 4.3863, 1.6701, 4.0959, 4.2332], device='cuda:6'), covar=tensor([0.0060, 0.0052, 0.0079, 0.0285, 0.0049, 0.1676, 0.0085, 0.0124], device='cuda:6'), in_proj_covar=tensor([0.0089, 0.0077, 0.0118, 0.0123, 0.0088, 0.0140, 0.0104, 0.0114], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:36:47,080 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-28 08:37:30,501 INFO [train.py:904] (6/8) Epoch 5, batch 6950, loss[loss=0.2489, simple_loss=0.327, pruned_loss=0.08542, over 16659.00 frames. ], tot_loss[loss=0.2623, simple_loss=0.3356, pruned_loss=0.09447, over 3060032.50 frames. ], batch size: 57, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:37:50,871 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.715e+02 3.906e+02 5.173e+02 7.099e+02 1.028e+03, threshold=1.035e+03, percent-clipped=6.0 2023-04-28 08:37:53,603 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-28 08:38:42,852 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47597.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:38:49,110 INFO [train.py:904] (6/8) Epoch 5, batch 7000, loss[loss=0.2655, simple_loss=0.3462, pruned_loss=0.09239, over 16972.00 frames. ], tot_loss[loss=0.2613, simple_loss=0.3357, pruned_loss=0.09342, over 3070952.72 frames. ], batch size: 41, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:40:06,954 INFO [train.py:904] (6/8) Epoch 5, batch 7050, loss[loss=0.2397, simple_loss=0.3202, pruned_loss=0.07958, over 16911.00 frames. ], tot_loss[loss=0.261, simple_loss=0.3358, pruned_loss=0.09306, over 3069995.11 frames. ], batch size: 109, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:40:26,882 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.583e+02 3.724e+02 4.498e+02 5.703e+02 1.501e+03, threshold=8.996e+02, percent-clipped=4.0 2023-04-28 08:40:30,560 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-04-28 08:40:52,053 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7579, 2.2577, 2.4526, 4.3499, 2.0320, 3.1735, 2.3634, 2.4350], device='cuda:6'), covar=tensor([0.0546, 0.1867, 0.1027, 0.0273, 0.2795, 0.0934, 0.1729, 0.2249], device='cuda:6'), in_proj_covar=tensor([0.0316, 0.0317, 0.0261, 0.0307, 0.0375, 0.0309, 0.0286, 0.0381], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:41:26,199 INFO [train.py:904] (6/8) Epoch 5, batch 7100, loss[loss=0.2794, simple_loss=0.3295, pruned_loss=0.1146, over 11172.00 frames. ], tot_loss[loss=0.26, simple_loss=0.3342, pruned_loss=0.09287, over 3050599.91 frames. ], batch size: 247, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:42:20,347 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47735.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:42:24,141 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47738.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:42:44,668 INFO [train.py:904] (6/8) Epoch 5, batch 7150, loss[loss=0.2646, simple_loss=0.3208, pruned_loss=0.1042, over 11028.00 frames. ], tot_loss[loss=0.2572, simple_loss=0.331, pruned_loss=0.09169, over 3061099.19 frames. ], batch size: 248, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:43:03,966 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.186e+02 3.709e+02 4.695e+02 5.943e+02 1.245e+03, threshold=9.389e+02, percent-clipped=5.0 2023-04-28 08:43:40,023 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47788.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:43:57,756 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47799.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:44:00,018 INFO [train.py:904] (6/8) Epoch 5, batch 7200, loss[loss=0.2629, simple_loss=0.3232, pruned_loss=0.1012, over 11601.00 frames. ], tot_loss[loss=0.2546, simple_loss=0.3292, pruned_loss=0.09006, over 3049179.16 frames. ], batch size: 248, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:44:14,044 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47809.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 08:44:17,155 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3913, 4.4116, 4.9624, 4.8741, 4.8462, 4.4860, 4.5689, 4.2151], device='cuda:6'), covar=tensor([0.0231, 0.0339, 0.0223, 0.0314, 0.0344, 0.0253, 0.0668, 0.0372], device='cuda:6'), in_proj_covar=tensor([0.0231, 0.0223, 0.0229, 0.0231, 0.0276, 0.0244, 0.0347, 0.0205], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 08:44:30,446 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0223, 2.3426, 2.3771, 3.1619, 2.5558, 3.2359, 1.7805, 2.7362], device='cuda:6'), covar=tensor([0.0959, 0.0478, 0.0903, 0.0094, 0.0166, 0.0354, 0.1111, 0.0598], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0143, 0.0167, 0.0085, 0.0180, 0.0177, 0.0160, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 08:44:53,052 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5847, 5.8443, 5.5792, 5.6492, 5.1190, 5.0349, 5.3807, 5.9310], device='cuda:6'), covar=tensor([0.0458, 0.0517, 0.0772, 0.0420, 0.0514, 0.0492, 0.0465, 0.0570], device='cuda:6'), in_proj_covar=tensor([0.0363, 0.0468, 0.0404, 0.0302, 0.0291, 0.0317, 0.0379, 0.0338], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:45:22,253 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47849.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:45:24,157 INFO [train.py:904] (6/8) Epoch 5, batch 7250, loss[loss=0.245, simple_loss=0.3168, pruned_loss=0.08659, over 16144.00 frames. ], tot_loss[loss=0.2527, simple_loss=0.3271, pruned_loss=0.08915, over 3042522.82 frames. ], batch size: 165, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:45:43,494 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.043e+02 3.368e+02 4.070e+02 5.805e+02 1.027e+03, threshold=8.141e+02, percent-clipped=1.0 2023-04-28 08:45:53,821 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47870.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 08:46:34,139 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47897.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:46:40,680 INFO [train.py:904] (6/8) Epoch 5, batch 7300, loss[loss=0.2855, simple_loss=0.3397, pruned_loss=0.1156, over 11347.00 frames. ], tot_loss[loss=0.2519, simple_loss=0.3264, pruned_loss=0.08874, over 3044577.61 frames. ], batch size: 248, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:47:16,994 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.61 vs. limit=2.0 2023-04-28 08:47:49,457 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=47945.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:47:58,197 INFO [train.py:904] (6/8) Epoch 5, batch 7350, loss[loss=0.227, simple_loss=0.3061, pruned_loss=0.07396, over 16295.00 frames. ], tot_loss[loss=0.2512, simple_loss=0.3255, pruned_loss=0.08841, over 3037170.02 frames. ], batch size: 146, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:48:17,667 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.639e+02 4.067e+02 5.121e+02 6.309e+02 1.579e+03, threshold=1.024e+03, percent-clipped=9.0 2023-04-28 08:48:26,824 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1147, 5.4667, 5.1601, 5.1790, 4.7730, 4.6494, 4.9855, 5.5244], device='cuda:6'), covar=tensor([0.0672, 0.0616, 0.0995, 0.0442, 0.0625, 0.0693, 0.0597, 0.0682], device='cuda:6'), in_proj_covar=tensor([0.0360, 0.0465, 0.0402, 0.0300, 0.0291, 0.0315, 0.0376, 0.0339], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:48:33,015 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-28 08:49:00,556 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0191, 3.4340, 3.4556, 2.3166, 3.3485, 3.4659, 3.4966, 1.8641], device='cuda:6'), covar=tensor([0.0321, 0.0032, 0.0036, 0.0251, 0.0040, 0.0065, 0.0031, 0.0303], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0054, 0.0060, 0.0118, 0.0061, 0.0071, 0.0064, 0.0111], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 08:49:18,342 INFO [train.py:904] (6/8) Epoch 5, batch 7400, loss[loss=0.21, simple_loss=0.2994, pruned_loss=0.06035, over 16836.00 frames. ], tot_loss[loss=0.253, simple_loss=0.3271, pruned_loss=0.0894, over 3036928.48 frames. ], batch size: 39, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:50:13,258 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48035.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:50:13,362 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8194, 3.3672, 3.4748, 2.2671, 3.1371, 3.3939, 3.4436, 1.9203], device='cuda:6'), covar=tensor([0.0354, 0.0026, 0.0033, 0.0242, 0.0053, 0.0066, 0.0032, 0.0287], device='cuda:6'), in_proj_covar=tensor([0.0117, 0.0054, 0.0059, 0.0116, 0.0061, 0.0070, 0.0063, 0.0109], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 08:50:38,165 INFO [train.py:904] (6/8) Epoch 5, batch 7450, loss[loss=0.2953, simple_loss=0.3377, pruned_loss=0.1265, over 11151.00 frames. ], tot_loss[loss=0.2539, simple_loss=0.3278, pruned_loss=0.08995, over 3048686.61 frames. ], batch size: 248, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:51:00,810 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.428e+02 4.070e+02 4.599e+02 6.099e+02 1.111e+03, threshold=9.198e+02, percent-clipped=2.0 2023-04-28 08:51:14,555 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48072.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:51:31,925 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=48083.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:51:49,960 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48094.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:52:00,148 INFO [train.py:904] (6/8) Epoch 5, batch 7500, loss[loss=0.3049, simple_loss=0.3543, pruned_loss=0.1277, over 11148.00 frames. ], tot_loss[loss=0.2546, simple_loss=0.3289, pruned_loss=0.09013, over 3046079.20 frames. ], batch size: 247, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 08:52:32,861 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5906, 4.5368, 4.4036, 4.3376, 3.8139, 4.4750, 4.4833, 4.1281], device='cuda:6'), covar=tensor([0.0492, 0.0284, 0.0247, 0.0180, 0.1013, 0.0296, 0.0255, 0.0483], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0181, 0.0204, 0.0177, 0.0234, 0.0208, 0.0150, 0.0229], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:52:50,990 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48133.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:53:08,579 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48144.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:53:18,028 INFO [train.py:904] (6/8) Epoch 5, batch 7550, loss[loss=0.2486, simple_loss=0.318, pruned_loss=0.08956, over 16656.00 frames. ], tot_loss[loss=0.2541, simple_loss=0.3277, pruned_loss=0.0903, over 3031775.13 frames. ], batch size: 134, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 08:53:38,522 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.261e+02 3.568e+02 4.415e+02 5.679e+02 1.356e+03, threshold=8.829e+02, percent-clipped=4.0 2023-04-28 08:53:40,712 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48165.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 08:54:34,586 INFO [train.py:904] (6/8) Epoch 5, batch 7600, loss[loss=0.264, simple_loss=0.3358, pruned_loss=0.09611, over 16420.00 frames. ], tot_loss[loss=0.2539, simple_loss=0.3268, pruned_loss=0.09048, over 3030913.59 frames. ], batch size: 146, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 08:54:46,607 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3227, 1.8539, 1.5132, 1.6385, 2.1212, 1.9506, 2.2243, 2.3066], device='cuda:6'), covar=tensor([0.0037, 0.0169, 0.0239, 0.0217, 0.0110, 0.0177, 0.0096, 0.0113], device='cuda:6'), in_proj_covar=tensor([0.0072, 0.0147, 0.0149, 0.0145, 0.0141, 0.0150, 0.0122, 0.0131], device='cuda:6'), out_proj_covar=tensor([8.8981e-05, 1.7849e-04, 1.7824e-04, 1.7332e-04, 1.7393e-04, 1.8232e-04, 1.4519e-04, 1.6098e-04], device='cuda:6') 2023-04-28 08:54:52,612 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48212.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 08:55:06,512 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3241, 3.2201, 2.7566, 2.0939, 2.3475, 2.1400, 3.3251, 3.4811], device='cuda:6'), covar=tensor([0.2263, 0.0724, 0.1201, 0.1589, 0.1689, 0.1415, 0.0430, 0.0635], device='cuda:6'), in_proj_covar=tensor([0.0281, 0.0249, 0.0269, 0.0242, 0.0291, 0.0202, 0.0241, 0.0253], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:55:51,251 INFO [train.py:904] (6/8) Epoch 5, batch 7650, loss[loss=0.2314, simple_loss=0.3137, pruned_loss=0.07457, over 16720.00 frames. ], tot_loss[loss=0.2552, simple_loss=0.328, pruned_loss=0.09122, over 3043547.54 frames. ], batch size: 89, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 08:56:00,242 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4339, 3.3904, 3.3591, 2.8628, 3.3396, 2.1113, 3.1521, 2.9333], device='cuda:6'), covar=tensor([0.0104, 0.0091, 0.0115, 0.0222, 0.0072, 0.1435, 0.0096, 0.0143], device='cuda:6'), in_proj_covar=tensor([0.0091, 0.0078, 0.0120, 0.0125, 0.0090, 0.0143, 0.0105, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:56:12,529 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.451e+02 3.891e+02 4.972e+02 6.319e+02 1.654e+03, threshold=9.944e+02, percent-clipped=8.0 2023-04-28 08:56:22,967 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4064, 1.9540, 2.0445, 3.8782, 1.9051, 2.8060, 2.1556, 2.0975], device='cuda:6'), covar=tensor([0.0585, 0.2262, 0.1159, 0.0303, 0.2994, 0.1104, 0.1933, 0.2365], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0318, 0.0261, 0.0307, 0.0374, 0.0308, 0.0286, 0.0378], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 08:56:25,981 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48273.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 08:57:08,646 INFO [train.py:904] (6/8) Epoch 5, batch 7700, loss[loss=0.2313, simple_loss=0.3101, pruned_loss=0.07626, over 16919.00 frames. ], tot_loss[loss=0.2553, simple_loss=0.328, pruned_loss=0.09129, over 3063758.79 frames. ], batch size: 109, lr: 1.30e-02, grad_scale: 4.0 2023-04-28 08:57:30,282 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-28 08:57:44,763 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48324.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:57:56,255 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-28 08:58:26,511 INFO [train.py:904] (6/8) Epoch 5, batch 7750, loss[loss=0.2943, simple_loss=0.3473, pruned_loss=0.1206, over 11522.00 frames. ], tot_loss[loss=0.2552, simple_loss=0.3281, pruned_loss=0.09109, over 3058943.95 frames. ], batch size: 248, lr: 1.30e-02, grad_scale: 4.0 2023-04-28 08:58:47,830 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.847e+02 4.062e+02 4.544e+02 6.017e+02 9.038e+02, threshold=9.088e+02, percent-clipped=0.0 2023-04-28 08:58:59,664 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-28 08:59:20,060 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48385.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:59:34,673 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48394.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:59:36,641 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0237, 2.4180, 2.3149, 3.2278, 2.5099, 3.2162, 1.8667, 2.6035], device='cuda:6'), covar=tensor([0.1152, 0.0471, 0.0974, 0.0120, 0.0244, 0.0374, 0.1181, 0.0706], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0142, 0.0168, 0.0085, 0.0181, 0.0178, 0.0159, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 08:59:39,574 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48397.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:59:45,047 INFO [train.py:904] (6/8) Epoch 5, batch 7800, loss[loss=0.2398, simple_loss=0.3166, pruned_loss=0.08153, over 15373.00 frames. ], tot_loss[loss=0.2554, simple_loss=0.3284, pruned_loss=0.09117, over 3070146.72 frames. ], batch size: 190, lr: 1.30e-02, grad_scale: 4.0 2023-04-28 08:59:48,726 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.43 vs. limit=5.0 2023-04-28 09:00:12,196 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3660, 5.6421, 5.3529, 5.4928, 4.9039, 4.7865, 5.0893, 5.7334], device='cuda:6'), covar=tensor([0.0725, 0.0673, 0.1009, 0.0484, 0.0704, 0.0588, 0.0656, 0.0632], device='cuda:6'), in_proj_covar=tensor([0.0367, 0.0477, 0.0417, 0.0310, 0.0300, 0.0323, 0.0391, 0.0347], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 09:00:21,093 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48423.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:00:28,781 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48428.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:00:28,901 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8286, 2.4444, 2.6055, 1.8687, 2.4292, 2.5909, 2.6013, 1.7741], device='cuda:6'), covar=tensor([0.0251, 0.0030, 0.0035, 0.0197, 0.0050, 0.0053, 0.0033, 0.0250], device='cuda:6'), in_proj_covar=tensor([0.0116, 0.0053, 0.0058, 0.0114, 0.0059, 0.0068, 0.0062, 0.0107], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 09:00:49,708 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=48442.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:00:52,201 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48444.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:01:02,288 INFO [train.py:904] (6/8) Epoch 5, batch 7850, loss[loss=0.2549, simple_loss=0.335, pruned_loss=0.08742, over 16916.00 frames. ], tot_loss[loss=0.2556, simple_loss=0.3295, pruned_loss=0.09087, over 3073923.77 frames. ], batch size: 116, lr: 1.30e-02, grad_scale: 4.0 2023-04-28 09:01:13,085 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48458.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:01:23,602 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.309e+02 3.753e+02 4.508e+02 5.540e+02 1.129e+03, threshold=9.017e+02, percent-clipped=6.0 2023-04-28 09:01:24,108 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48465.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 09:01:52,611 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48484.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:02:04,608 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=48492.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:02:16,621 INFO [train.py:904] (6/8) Epoch 5, batch 7900, loss[loss=0.2735, simple_loss=0.3514, pruned_loss=0.09776, over 15409.00 frames. ], tot_loss[loss=0.2545, simple_loss=0.3284, pruned_loss=0.09035, over 3072568.84 frames. ], batch size: 190, lr: 1.30e-02, grad_scale: 4.0 2023-04-28 09:02:35,278 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=48513.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 09:03:21,964 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8093, 4.6635, 4.6437, 4.4543, 4.1478, 4.6886, 4.5782, 4.2931], device='cuda:6'), covar=tensor([0.0451, 0.0309, 0.0205, 0.0184, 0.0902, 0.0326, 0.0262, 0.0485], device='cuda:6'), in_proj_covar=tensor([0.0183, 0.0187, 0.0207, 0.0183, 0.0238, 0.0211, 0.0153, 0.0236], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 09:03:36,607 INFO [train.py:904] (6/8) Epoch 5, batch 7950, loss[loss=0.2286, simple_loss=0.307, pruned_loss=0.07511, over 17070.00 frames. ], tot_loss[loss=0.2554, simple_loss=0.3288, pruned_loss=0.09098, over 3077967.46 frames. ], batch size: 53, lr: 1.30e-02, grad_scale: 4.0 2023-04-28 09:03:53,367 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3448, 3.5082, 3.2224, 3.0471, 2.9175, 3.3427, 3.1891, 3.1154], device='cuda:6'), covar=tensor([0.0630, 0.0416, 0.0264, 0.0220, 0.0726, 0.0345, 0.1178, 0.0511], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0189, 0.0209, 0.0184, 0.0240, 0.0213, 0.0154, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 09:03:53,434 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8358, 1.1509, 1.5725, 1.6639, 1.8150, 1.8423, 1.3591, 1.6125], device='cuda:6'), covar=tensor([0.0092, 0.0210, 0.0093, 0.0127, 0.0103, 0.0063, 0.0202, 0.0042], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0146, 0.0126, 0.0124, 0.0130, 0.0094, 0.0142, 0.0080], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 09:03:56,964 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.950e+02 3.717e+02 4.429e+02 5.526e+02 1.268e+03, threshold=8.857e+02, percent-clipped=1.0 2023-04-28 09:04:03,018 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48568.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 09:04:50,721 INFO [train.py:904] (6/8) Epoch 5, batch 8000, loss[loss=0.2364, simple_loss=0.3138, pruned_loss=0.07951, over 16510.00 frames. ], tot_loss[loss=0.2565, simple_loss=0.3295, pruned_loss=0.09176, over 3076034.36 frames. ], batch size: 75, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 09:05:33,358 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1903, 3.3021, 3.2611, 1.5165, 3.3058, 3.4757, 2.8558, 2.6687], device='cuda:6'), covar=tensor([0.0797, 0.0111, 0.0141, 0.1376, 0.0109, 0.0071, 0.0341, 0.0434], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0086, 0.0082, 0.0141, 0.0071, 0.0076, 0.0116, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 09:06:06,080 INFO [train.py:904] (6/8) Epoch 5, batch 8050, loss[loss=0.2549, simple_loss=0.3368, pruned_loss=0.08652, over 16401.00 frames. ], tot_loss[loss=0.2545, simple_loss=0.3282, pruned_loss=0.0904, over 3092715.43 frames. ], batch size: 146, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 09:06:26,963 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.628e+02 3.970e+02 4.648e+02 5.523e+02 1.617e+03, threshold=9.296e+02, percent-clipped=6.0 2023-04-28 09:06:49,656 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48680.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:06:58,087 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5824, 3.8715, 1.7249, 4.1307, 2.4361, 4.1098, 1.9229, 2.8270], device='cuda:6'), covar=tensor([0.0148, 0.0249, 0.1824, 0.0043, 0.0829, 0.0336, 0.1589, 0.0637], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0148, 0.0179, 0.0079, 0.0160, 0.0178, 0.0184, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 09:07:21,588 INFO [train.py:904] (6/8) Epoch 5, batch 8100, loss[loss=0.2785, simple_loss=0.3436, pruned_loss=0.1067, over 15323.00 frames. ], tot_loss[loss=0.2525, simple_loss=0.327, pruned_loss=0.08898, over 3105807.89 frames. ], batch size: 190, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 09:08:05,987 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48728.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:08:40,601 INFO [train.py:904] (6/8) Epoch 5, batch 8150, loss[loss=0.2186, simple_loss=0.2964, pruned_loss=0.07041, over 16749.00 frames. ], tot_loss[loss=0.2504, simple_loss=0.3247, pruned_loss=0.08803, over 3103234.17 frames. ], batch size: 89, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 09:08:44,670 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48753.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:09:01,945 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.147e+02 3.868e+02 4.692e+02 5.916e+02 1.218e+03, threshold=9.385e+02, percent-clipped=3.0 2023-04-28 09:09:21,416 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=48776.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:09:25,158 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48779.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:10:00,528 INFO [train.py:904] (6/8) Epoch 5, batch 8200, loss[loss=0.2541, simple_loss=0.3355, pruned_loss=0.08633, over 16253.00 frames. ], tot_loss[loss=0.2482, simple_loss=0.3222, pruned_loss=0.08709, over 3108367.58 frames. ], batch size: 165, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 09:10:32,535 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48821.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:11:01,494 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48838.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:11:22,112 INFO [train.py:904] (6/8) Epoch 5, batch 8250, loss[loss=0.2512, simple_loss=0.3373, pruned_loss=0.0825, over 16243.00 frames. ], tot_loss[loss=0.2467, simple_loss=0.3216, pruned_loss=0.0859, over 3051508.01 frames. ], batch size: 165, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:11:44,535 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.248e+02 3.760e+02 4.474e+02 5.646e+02 1.182e+03, threshold=8.947e+02, percent-clipped=2.0 2023-04-28 09:11:50,802 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48868.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 09:12:13,847 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48882.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:12:41,260 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48899.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:12:43,300 INFO [train.py:904] (6/8) Epoch 5, batch 8300, loss[loss=0.2002, simple_loss=0.2768, pruned_loss=0.06182, over 12222.00 frames. ], tot_loss[loss=0.2415, simple_loss=0.3184, pruned_loss=0.08228, over 3050314.09 frames. ], batch size: 247, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:13:08,661 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=48916.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 09:14:06,485 INFO [train.py:904] (6/8) Epoch 5, batch 8350, loss[loss=0.2102, simple_loss=0.3052, pruned_loss=0.05766, over 16806.00 frames. ], tot_loss[loss=0.2375, simple_loss=0.3164, pruned_loss=0.07926, over 3045694.40 frames. ], batch size: 102, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:14:30,518 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.953e+02 3.108e+02 3.875e+02 4.541e+02 1.583e+03, threshold=7.750e+02, percent-clipped=2.0 2023-04-28 09:14:54,743 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48980.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:14:56,699 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7872, 5.0978, 5.1353, 5.2449, 5.1091, 5.7026, 5.3101, 5.0487], device='cuda:6'), covar=tensor([0.0668, 0.1300, 0.1166, 0.1273, 0.1987, 0.0752, 0.0992, 0.1855], device='cuda:6'), in_proj_covar=tensor([0.0262, 0.0367, 0.0367, 0.0319, 0.0427, 0.0391, 0.0301, 0.0434], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 09:15:29,928 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 09:15:30,221 INFO [train.py:904] (6/8) Epoch 5, batch 8400, loss[loss=0.1977, simple_loss=0.2914, pruned_loss=0.05203, over 16761.00 frames. ], tot_loss[loss=0.2329, simple_loss=0.3127, pruned_loss=0.07651, over 3014085.28 frames. ], batch size: 102, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:15:37,392 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1367, 3.0300, 2.7798, 1.9640, 2.6594, 2.0783, 2.7865, 2.8198], device='cuda:6'), covar=tensor([0.0235, 0.0464, 0.0405, 0.1338, 0.0562, 0.0876, 0.0584, 0.0603], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0120, 0.0147, 0.0136, 0.0127, 0.0122, 0.0135, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-28 09:16:14,940 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=49028.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:16:51,001 INFO [train.py:904] (6/8) Epoch 5, batch 8450, loss[loss=0.1899, simple_loss=0.2823, pruned_loss=0.04875, over 16862.00 frames. ], tot_loss[loss=0.2297, simple_loss=0.3105, pruned_loss=0.0744, over 3030710.21 frames. ], batch size: 90, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:16:55,252 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49053.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:17:13,777 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.140e+02 2.885e+02 3.477e+02 4.145e+02 8.809e+02, threshold=6.955e+02, percent-clipped=1.0 2023-04-28 09:17:38,431 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49079.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:17:47,006 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9987, 4.2873, 2.1264, 4.4800, 2.7063, 4.3160, 2.4187, 3.0655], device='cuda:6'), covar=tensor([0.0109, 0.0134, 0.1400, 0.0027, 0.0721, 0.0265, 0.1252, 0.0560], device='cuda:6'), in_proj_covar=tensor([0.0116, 0.0147, 0.0175, 0.0076, 0.0159, 0.0174, 0.0184, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 09:18:08,426 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9322, 2.5930, 2.6265, 1.9737, 2.5996, 2.5207, 2.6114, 1.7562], device='cuda:6'), covar=tensor([0.0225, 0.0027, 0.0039, 0.0184, 0.0047, 0.0039, 0.0038, 0.0277], device='cuda:6'), in_proj_covar=tensor([0.0108, 0.0050, 0.0055, 0.0106, 0.0057, 0.0063, 0.0059, 0.0101], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 09:18:13,187 INFO [train.py:904] (6/8) Epoch 5, batch 8500, loss[loss=0.2068, simple_loss=0.295, pruned_loss=0.05934, over 16752.00 frames. ], tot_loss[loss=0.2237, simple_loss=0.3056, pruned_loss=0.07083, over 3040006.84 frames. ], batch size: 124, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:18:13,694 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=49101.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:18:57,905 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=49127.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:19:32,607 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 09:19:38,542 INFO [train.py:904] (6/8) Epoch 5, batch 8550, loss[loss=0.2307, simple_loss=0.3194, pruned_loss=0.07097, over 16780.00 frames. ], tot_loss[loss=0.2215, simple_loss=0.3035, pruned_loss=0.06975, over 3016751.51 frames. ], batch size: 124, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:19:40,875 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-28 09:20:04,107 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.964e+02 3.233e+02 3.773e+02 4.691e+02 1.040e+03, threshold=7.547e+02, percent-clipped=6.0 2023-04-28 09:20:29,011 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49177.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:21:04,531 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49194.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:21:17,317 INFO [train.py:904] (6/8) Epoch 5, batch 8600, loss[loss=0.2313, simple_loss=0.3017, pruned_loss=0.08047, over 12280.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.3033, pruned_loss=0.06825, over 3021686.74 frames. ], batch size: 247, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:21:47,226 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1288, 1.3585, 1.8075, 2.2373, 2.1847, 2.3273, 1.4146, 2.1888], device='cuda:6'), covar=tensor([0.0090, 0.0280, 0.0143, 0.0127, 0.0124, 0.0093, 0.0272, 0.0077], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0145, 0.0130, 0.0125, 0.0132, 0.0093, 0.0143, 0.0079], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 09:22:28,807 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49236.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:22:44,445 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0877, 3.3381, 3.5610, 3.5269, 3.5495, 3.2750, 3.3637, 3.3790], device='cuda:6'), covar=tensor([0.0298, 0.0432, 0.0445, 0.0519, 0.0446, 0.0380, 0.0719, 0.0360], device='cuda:6'), in_proj_covar=tensor([0.0227, 0.0219, 0.0228, 0.0223, 0.0273, 0.0239, 0.0334, 0.0200], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-28 09:22:44,720 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2023-04-28 09:22:53,007 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5464, 3.6336, 3.3857, 3.2959, 3.1660, 3.5030, 3.2045, 3.2911], device='cuda:6'), covar=tensor([0.0432, 0.0301, 0.0230, 0.0183, 0.0548, 0.0388, 0.1195, 0.0394], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0179, 0.0205, 0.0175, 0.0227, 0.0204, 0.0146, 0.0231], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 09:22:58,656 INFO [train.py:904] (6/8) Epoch 5, batch 8650, loss[loss=0.1973, simple_loss=0.2763, pruned_loss=0.05918, over 12308.00 frames. ], tot_loss[loss=0.2176, simple_loss=0.3014, pruned_loss=0.06695, over 3011453.61 frames. ], batch size: 248, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:23:33,960 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.003e+02 3.105e+02 3.609e+02 4.531e+02 9.141e+02, threshold=7.218e+02, percent-clipped=3.0 2023-04-28 09:24:39,668 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49297.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:24:47,014 INFO [train.py:904] (6/8) Epoch 5, batch 8700, loss[loss=0.2043, simple_loss=0.2904, pruned_loss=0.05914, over 15402.00 frames. ], tot_loss[loss=0.2138, simple_loss=0.2979, pruned_loss=0.06487, over 3013677.28 frames. ], batch size: 191, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:25:33,436 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2557, 3.0781, 2.8395, 2.0201, 2.5897, 2.0561, 2.8062, 2.8152], device='cuda:6'), covar=tensor([0.0289, 0.0445, 0.0418, 0.1243, 0.0573, 0.0872, 0.0614, 0.0615], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0118, 0.0147, 0.0136, 0.0129, 0.0124, 0.0134, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-28 09:26:24,152 INFO [train.py:904] (6/8) Epoch 5, batch 8750, loss[loss=0.2278, simple_loss=0.3103, pruned_loss=0.0726, over 17131.00 frames. ], tot_loss[loss=0.2129, simple_loss=0.2976, pruned_loss=0.06415, over 3013437.08 frames. ], batch size: 49, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:27:05,486 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.641e+02 2.935e+02 3.673e+02 5.022e+02 7.467e+02, threshold=7.345e+02, percent-clipped=1.0 2023-04-28 09:28:04,620 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-28 09:28:17,602 INFO [train.py:904] (6/8) Epoch 5, batch 8800, loss[loss=0.2041, simple_loss=0.2855, pruned_loss=0.0613, over 16645.00 frames. ], tot_loss[loss=0.2102, simple_loss=0.2955, pruned_loss=0.06245, over 3035590.30 frames. ], batch size: 62, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:28:18,869 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-28 09:29:02,385 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49422.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:29:20,559 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6332, 2.7651, 1.6961, 2.7988, 2.0815, 2.8068, 1.8955, 2.4222], device='cuda:6'), covar=tensor([0.0163, 0.0327, 0.1324, 0.0114, 0.0802, 0.0431, 0.1259, 0.0580], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0144, 0.0173, 0.0075, 0.0155, 0.0168, 0.0182, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-28 09:29:24,479 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49432.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 09:29:59,593 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4253, 3.6544, 3.9983, 1.6865, 4.1261, 4.2391, 3.2669, 3.0539], device='cuda:6'), covar=tensor([0.0736, 0.0148, 0.0146, 0.1161, 0.0044, 0.0041, 0.0265, 0.0356], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0085, 0.0076, 0.0137, 0.0066, 0.0073, 0.0112, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 09:30:04,510 INFO [train.py:904] (6/8) Epoch 5, batch 8850, loss[loss=0.2049, simple_loss=0.3006, pruned_loss=0.05464, over 16262.00 frames. ], tot_loss[loss=0.2105, simple_loss=0.2976, pruned_loss=0.06174, over 3022689.68 frames. ], batch size: 165, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:30:15,628 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-28 09:30:38,663 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 3.290e+02 4.056e+02 5.025e+02 1.173e+03, threshold=8.111e+02, percent-clipped=7.0 2023-04-28 09:31:02,585 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49477.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:31:16,935 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49483.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:31:37,627 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49493.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 09:31:39,809 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49494.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:31:54,235 INFO [train.py:904] (6/8) Epoch 5, batch 8900, loss[loss=0.2187, simple_loss=0.3041, pruned_loss=0.06669, over 15256.00 frames. ], tot_loss[loss=0.2095, simple_loss=0.2974, pruned_loss=0.06084, over 3032864.21 frames. ], batch size: 190, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:32:48,443 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=49525.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:33:22,560 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8401, 4.5482, 4.8270, 5.0864, 5.2611, 4.4978, 5.1815, 5.1862], device='cuda:6'), covar=tensor([0.0821, 0.0731, 0.1143, 0.0467, 0.0351, 0.0708, 0.0361, 0.0325], device='cuda:6'), in_proj_covar=tensor([0.0347, 0.0438, 0.0539, 0.0451, 0.0339, 0.0332, 0.0360, 0.0370], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 09:33:39,643 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=49542.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:33:59,795 INFO [train.py:904] (6/8) Epoch 5, batch 8950, loss[loss=0.2036, simple_loss=0.2836, pruned_loss=0.06183, over 15528.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2984, pruned_loss=0.06181, over 3042398.78 frames. ], batch size: 194, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:34:01,271 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3094, 4.0875, 4.3823, 4.6106, 4.7192, 4.1861, 4.7229, 4.6799], device='cuda:6'), covar=tensor([0.0957, 0.0800, 0.1173, 0.0459, 0.0390, 0.0710, 0.0342, 0.0369], device='cuda:6'), in_proj_covar=tensor([0.0350, 0.0440, 0.0541, 0.0453, 0.0340, 0.0333, 0.0361, 0.0371], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 09:34:35,630 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.980e+02 2.796e+02 3.504e+02 4.655e+02 7.441e+02, threshold=7.007e+02, percent-clipped=0.0 2023-04-28 09:34:50,036 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8208, 1.1979, 1.5734, 1.7468, 1.8785, 1.8636, 1.4855, 1.8091], device='cuda:6'), covar=tensor([0.0100, 0.0184, 0.0101, 0.0126, 0.0120, 0.0096, 0.0182, 0.0051], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0143, 0.0129, 0.0124, 0.0129, 0.0089, 0.0141, 0.0077], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 09:35:33,169 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49592.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:35:51,807 INFO [train.py:904] (6/8) Epoch 5, batch 9000, loss[loss=0.1886, simple_loss=0.277, pruned_loss=0.05005, over 16963.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2953, pruned_loss=0.0606, over 3018502.01 frames. ], batch size: 116, lr: 1.28e-02, grad_scale: 4.0 2023-04-28 09:35:51,807 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 09:36:02,102 INFO [train.py:938] (6/8) Epoch 5, validation: loss=0.1735, simple_loss=0.2766, pruned_loss=0.0352, over 944034.00 frames. 2023-04-28 09:36:02,103 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-28 09:36:07,830 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3653, 2.9811, 2.5256, 2.2617, 2.0758, 2.0202, 2.9356, 2.9828], device='cuda:6'), covar=tensor([0.1860, 0.0591, 0.1116, 0.1388, 0.2024, 0.1544, 0.0385, 0.0706], device='cuda:6'), in_proj_covar=tensor([0.0272, 0.0237, 0.0259, 0.0237, 0.0246, 0.0196, 0.0233, 0.0234], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 09:36:36,063 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 09:37:44,739 INFO [train.py:904] (6/8) Epoch 5, batch 9050, loss[loss=0.2431, simple_loss=0.3148, pruned_loss=0.08571, over 12955.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2975, pruned_loss=0.06199, over 3032377.80 frames. ], batch size: 248, lr: 1.28e-02, grad_scale: 4.0 2023-04-28 09:38:18,678 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.067e+02 3.141e+02 3.840e+02 5.023e+02 8.628e+02, threshold=7.679e+02, percent-clipped=5.0 2023-04-28 09:38:58,517 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4919, 3.9018, 1.8568, 4.0280, 2.4597, 3.8959, 2.0711, 2.7981], device='cuda:6'), covar=tensor([0.0162, 0.0178, 0.1530, 0.0045, 0.0748, 0.0348, 0.1415, 0.0614], device='cuda:6'), in_proj_covar=tensor([0.0116, 0.0143, 0.0175, 0.0075, 0.0154, 0.0170, 0.0185, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 09:39:29,262 INFO [train.py:904] (6/8) Epoch 5, batch 9100, loss[loss=0.2206, simple_loss=0.3121, pruned_loss=0.06455, over 16671.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2971, pruned_loss=0.0624, over 3033966.12 frames. ], batch size: 134, lr: 1.28e-02, grad_scale: 4.0 2023-04-28 09:40:13,879 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4963, 3.5271, 3.3297, 3.2248, 3.0953, 3.4793, 3.2691, 3.2566], device='cuda:6'), covar=tensor([0.0421, 0.0337, 0.0205, 0.0176, 0.0498, 0.0312, 0.0828, 0.0428], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0174, 0.0198, 0.0168, 0.0218, 0.0199, 0.0140, 0.0224], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 09:41:26,339 INFO [train.py:904] (6/8) Epoch 5, batch 9150, loss[loss=0.1887, simple_loss=0.2834, pruned_loss=0.04697, over 16897.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2968, pruned_loss=0.06129, over 3047384.03 frames. ], batch size: 90, lr: 1.28e-02, grad_scale: 4.0 2023-04-28 09:42:00,629 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49766.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:42:01,428 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.296e+02 3.040e+02 3.844e+02 4.947e+02 8.229e+02, threshold=7.688e+02, percent-clipped=4.0 2023-04-28 09:42:25,722 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49778.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:42:27,224 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9189, 1.9718, 2.4276, 4.5080, 1.9079, 2.8372, 2.4047, 2.1633], device='cuda:6'), covar=tensor([0.0429, 0.2295, 0.1102, 0.0213, 0.3159, 0.1318, 0.1860, 0.2840], device='cuda:6'), in_proj_covar=tensor([0.0304, 0.0303, 0.0255, 0.0298, 0.0360, 0.0304, 0.0284, 0.0360], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 09:42:47,437 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49788.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 09:43:09,755 INFO [train.py:904] (6/8) Epoch 5, batch 9200, loss[loss=0.2102, simple_loss=0.2935, pruned_loss=0.06348, over 16859.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2917, pruned_loss=0.05978, over 3051657.18 frames. ], batch size: 116, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:43:20,752 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-28 09:43:58,095 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49827.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:44:46,684 INFO [train.py:904] (6/8) Epoch 5, batch 9250, loss[loss=0.1938, simple_loss=0.286, pruned_loss=0.05085, over 15470.00 frames. ], tot_loss[loss=0.205, simple_loss=0.291, pruned_loss=0.05946, over 3052820.12 frames. ], batch size: 191, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:45:18,929 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.407e+02 3.305e+02 4.013e+02 4.841e+02 9.780e+02, threshold=8.027e+02, percent-clipped=1.0 2023-04-28 09:46:13,251 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-28 09:46:21,528 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49892.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:46:40,058 INFO [train.py:904] (6/8) Epoch 5, batch 9300, loss[loss=0.1744, simple_loss=0.2584, pruned_loss=0.04515, over 17063.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2887, pruned_loss=0.05808, over 3065853.44 frames. ], batch size: 53, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:47:31,870 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-28 09:47:40,983 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9305, 3.9908, 3.7948, 3.6588, 3.4541, 3.9193, 3.5953, 3.6763], device='cuda:6'), covar=tensor([0.0389, 0.0267, 0.0201, 0.0190, 0.0651, 0.0283, 0.0746, 0.0383], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0174, 0.0197, 0.0170, 0.0220, 0.0201, 0.0140, 0.0224], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 09:48:05,870 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=49940.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:48:26,418 INFO [train.py:904] (6/8) Epoch 5, batch 9350, loss[loss=0.2108, simple_loss=0.2836, pruned_loss=0.06903, over 12068.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.289, pruned_loss=0.05824, over 3068092.68 frames. ], batch size: 248, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:49:00,528 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.703e+02 3.133e+02 3.801e+02 4.449e+02 7.989e+02, threshold=7.602e+02, percent-clipped=0.0 2023-04-28 09:50:11,379 INFO [train.py:904] (6/8) Epoch 5, batch 9400, loss[loss=0.2119, simple_loss=0.3026, pruned_loss=0.06061, over 15287.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2894, pruned_loss=0.05793, over 3075726.67 frames. ], batch size: 192, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:51:51,664 INFO [train.py:904] (6/8) Epoch 5, batch 9450, loss[loss=0.1997, simple_loss=0.2885, pruned_loss=0.05544, over 15411.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2909, pruned_loss=0.05843, over 3052026.21 frames. ], batch size: 191, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:52:21,841 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 2.955e+02 3.683e+02 5.023e+02 1.227e+03, threshold=7.366e+02, percent-clipped=5.0 2023-04-28 09:52:47,323 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50078.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:53:09,012 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50088.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 09:53:21,763 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-28 09:53:32,539 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9771, 4.0224, 4.3948, 4.3706, 4.3876, 4.0451, 4.1095, 3.8892], device='cuda:6'), covar=tensor([0.0260, 0.0390, 0.0343, 0.0425, 0.0394, 0.0275, 0.0647, 0.0365], device='cuda:6'), in_proj_covar=tensor([0.0214, 0.0209, 0.0214, 0.0214, 0.0261, 0.0229, 0.0312, 0.0189], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-04-28 09:53:33,283 INFO [train.py:904] (6/8) Epoch 5, batch 9500, loss[loss=0.1945, simple_loss=0.291, pruned_loss=0.04897, over 16755.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2901, pruned_loss=0.05738, over 3078738.43 frames. ], batch size: 134, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:54:18,306 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50122.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:54:25,651 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=50126.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:54:29,349 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50128.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:54:44,420 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=50136.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 09:55:18,505 INFO [train.py:904] (6/8) Epoch 5, batch 9550, loss[loss=0.1929, simple_loss=0.2905, pruned_loss=0.04764, over 16852.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2889, pruned_loss=0.05733, over 3073698.19 frames. ], batch size: 96, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:55:53,362 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.669e+02 2.772e+02 3.459e+02 4.281e+02 6.687e+02, threshold=6.919e+02, percent-clipped=0.0 2023-04-28 09:56:35,121 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8498, 3.5598, 3.2300, 1.7908, 2.8150, 2.2721, 3.1917, 3.3087], device='cuda:6'), covar=tensor([0.0258, 0.0484, 0.0494, 0.1625, 0.0690, 0.0976, 0.0738, 0.0684], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0116, 0.0151, 0.0138, 0.0130, 0.0127, 0.0135, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 09:56:38,104 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50189.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:56:52,921 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50197.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 09:56:59,037 INFO [train.py:904] (6/8) Epoch 5, batch 9600, loss[loss=0.2306, simple_loss=0.3229, pruned_loss=0.06913, over 15395.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2912, pruned_loss=0.05886, over 3063615.19 frames. ], batch size: 191, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:58:20,080 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1230, 2.2717, 2.4321, 4.8570, 1.9888, 3.2420, 2.5680, 2.5389], device='cuda:6'), covar=tensor([0.0467, 0.2216, 0.1132, 0.0207, 0.3265, 0.1129, 0.1878, 0.2648], device='cuda:6'), in_proj_covar=tensor([0.0303, 0.0304, 0.0258, 0.0300, 0.0359, 0.0304, 0.0283, 0.0357], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 09:58:44,139 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6526, 2.7050, 1.4450, 2.7523, 2.0094, 2.7665, 1.7190, 2.2821], device='cuda:6'), covar=tensor([0.0184, 0.0361, 0.1672, 0.0110, 0.0821, 0.0475, 0.1560, 0.0665], device='cuda:6'), in_proj_covar=tensor([0.0115, 0.0144, 0.0174, 0.0074, 0.0156, 0.0169, 0.0181, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 09:58:47,432 INFO [train.py:904] (6/8) Epoch 5, batch 9650, loss[loss=0.1954, simple_loss=0.2831, pruned_loss=0.05387, over 16862.00 frames. ], tot_loss[loss=0.2067, simple_loss=0.2939, pruned_loss=0.05978, over 3058204.62 frames. ], batch size: 124, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:59:05,744 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50258.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 09:59:27,461 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.258e+02 2.991e+02 3.657e+02 4.626e+02 9.582e+02, threshold=7.315e+02, percent-clipped=7.0 2023-04-28 10:00:35,789 INFO [train.py:904] (6/8) Epoch 5, batch 9700, loss[loss=0.1941, simple_loss=0.2821, pruned_loss=0.05309, over 16663.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2929, pruned_loss=0.05974, over 3064674.47 frames. ], batch size: 57, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 10:01:21,895 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0982, 4.4460, 2.0841, 4.7486, 2.6953, 4.6718, 2.4956, 3.0462], device='cuda:6'), covar=tensor([0.0135, 0.0132, 0.1645, 0.0027, 0.0856, 0.0204, 0.1353, 0.0663], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0142, 0.0172, 0.0073, 0.0156, 0.0166, 0.0179, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 10:01:44,047 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4677, 3.5864, 1.7841, 3.7239, 2.3529, 3.6914, 1.9513, 2.7824], device='cuda:6'), covar=tensor([0.0137, 0.0203, 0.1633, 0.0055, 0.0851, 0.0404, 0.1503, 0.0630], device='cuda:6'), in_proj_covar=tensor([0.0115, 0.0144, 0.0174, 0.0074, 0.0157, 0.0168, 0.0181, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 10:02:18,556 INFO [train.py:904] (6/8) Epoch 5, batch 9750, loss[loss=0.1854, simple_loss=0.2811, pruned_loss=0.04487, over 16897.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2916, pruned_loss=0.05983, over 3049893.91 frames. ], batch size: 90, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 10:02:50,575 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.871e+02 3.120e+02 3.818e+02 4.621e+02 8.897e+02, threshold=7.636e+02, percent-clipped=1.0 2023-04-28 10:03:56,536 INFO [train.py:904] (6/8) Epoch 5, batch 9800, loss[loss=0.2133, simple_loss=0.3083, pruned_loss=0.05914, over 16667.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2906, pruned_loss=0.05837, over 3043661.85 frames. ], batch size: 134, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:04:36,920 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50422.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:04:40,420 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-28 10:04:42,225 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7161, 4.0402, 3.1790, 2.3431, 2.8342, 2.3452, 4.1687, 3.9570], device='cuda:6'), covar=tensor([0.2105, 0.0564, 0.1197, 0.1634, 0.1835, 0.1355, 0.0349, 0.0565], device='cuda:6'), in_proj_covar=tensor([0.0269, 0.0236, 0.0259, 0.0233, 0.0234, 0.0193, 0.0231, 0.0227], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 10:05:41,977 INFO [train.py:904] (6/8) Epoch 5, batch 9850, loss[loss=0.2107, simple_loss=0.2972, pruned_loss=0.06211, over 16181.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2917, pruned_loss=0.05785, over 3057810.41 frames. ], batch size: 165, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:06:14,681 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 2.952e+02 3.639e+02 4.340e+02 9.232e+02, threshold=7.278e+02, percent-clipped=1.0 2023-04-28 10:06:20,055 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0280, 2.6335, 2.2610, 3.3705, 2.9654, 3.3252, 1.6696, 2.6597], device='cuda:6'), covar=tensor([0.1052, 0.0382, 0.0969, 0.0082, 0.0196, 0.0453, 0.1203, 0.0643], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0140, 0.0166, 0.0083, 0.0150, 0.0173, 0.0161, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 10:06:21,890 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=50470.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:06:52,947 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50484.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:07:31,480 INFO [train.py:904] (6/8) Epoch 5, batch 9900, loss[loss=0.2191, simple_loss=0.3125, pruned_loss=0.06286, over 16942.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.2924, pruned_loss=0.05788, over 3043135.74 frames. ], batch size: 109, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:08:04,949 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3746, 3.3137, 3.3912, 3.5330, 3.5310, 3.2083, 3.5657, 3.5668], device='cuda:6'), covar=tensor([0.0740, 0.0615, 0.0902, 0.0498, 0.0494, 0.1624, 0.0515, 0.0465], device='cuda:6'), in_proj_covar=tensor([0.0350, 0.0439, 0.0549, 0.0456, 0.0337, 0.0337, 0.0357, 0.0377], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 10:09:17,005 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50546.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:09:27,670 INFO [train.py:904] (6/8) Epoch 5, batch 9950, loss[loss=0.2197, simple_loss=0.3071, pruned_loss=0.06618, over 15405.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2938, pruned_loss=0.05802, over 3034801.93 frames. ], batch size: 191, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:09:33,880 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50553.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 10:10:04,528 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.956e+02 3.001e+02 3.652e+02 4.906e+02 2.314e+03, threshold=7.303e+02, percent-clipped=6.0 2023-04-28 10:11:08,286 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.63 vs. limit=5.0 2023-04-28 10:11:29,333 INFO [train.py:904] (6/8) Epoch 5, batch 10000, loss[loss=0.1736, simple_loss=0.2778, pruned_loss=0.03473, over 17064.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2917, pruned_loss=0.0568, over 3068578.53 frames. ], batch size: 97, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:11:44,478 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50607.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:13:08,531 INFO [train.py:904] (6/8) Epoch 5, batch 10050, loss[loss=0.2239, simple_loss=0.323, pruned_loss=0.06235, over 16900.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2914, pruned_loss=0.05636, over 3075867.68 frames. ], batch size: 116, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:13:38,879 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.875e+02 2.547e+02 3.217e+02 3.917e+02 1.118e+03, threshold=6.434e+02, percent-clipped=1.0 2023-04-28 10:14:24,897 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8904, 3.6859, 3.3094, 1.7724, 2.7167, 2.2338, 3.1479, 3.2870], device='cuda:6'), covar=tensor([0.0305, 0.0409, 0.0443, 0.1543, 0.0716, 0.0899, 0.0764, 0.0676], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0114, 0.0151, 0.0139, 0.0129, 0.0127, 0.0134, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 10:14:38,999 INFO [train.py:904] (6/8) Epoch 5, batch 10100, loss[loss=0.2109, simple_loss=0.2893, pruned_loss=0.06628, over 16691.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2923, pruned_loss=0.05717, over 3060810.67 frames. ], batch size: 134, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:15:25,112 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1130, 3.9289, 4.1766, 4.2758, 4.4328, 3.9396, 4.4588, 4.3997], device='cuda:6'), covar=tensor([0.0906, 0.0771, 0.1125, 0.0601, 0.0380, 0.0904, 0.0368, 0.0436], device='cuda:6'), in_proj_covar=tensor([0.0348, 0.0442, 0.0552, 0.0461, 0.0338, 0.0336, 0.0358, 0.0374], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 10:16:20,268 INFO [train.py:904] (6/8) Epoch 6, batch 0, loss[loss=0.1978, simple_loss=0.272, pruned_loss=0.06175, over 16877.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.272, pruned_loss=0.06175, over 16877.00 frames. ], batch size: 42, lr: 1.19e-02, grad_scale: 8.0 2023-04-28 10:16:20,269 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 10:16:27,646 INFO [train.py:938] (6/8) Epoch 6, validation: loss=0.1727, simple_loss=0.2755, pruned_loss=0.03501, over 944034.00 frames. 2023-04-28 10:16:27,647 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-28 10:16:52,395 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.958e+02 3.546e+02 4.478e+02 5.747e+02 1.222e+03, threshold=8.956e+02, percent-clipped=19.0 2023-04-28 10:17:10,718 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50782.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:17:13,781 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50784.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:17:31,931 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-28 10:17:35,095 INFO [train.py:904] (6/8) Epoch 6, batch 50, loss[loss=0.2638, simple_loss=0.3184, pruned_loss=0.1046, over 16750.00 frames. ], tot_loss[loss=0.236, simple_loss=0.3112, pruned_loss=0.08037, over 757873.64 frames. ], batch size: 134, lr: 1.19e-02, grad_scale: 2.0 2023-04-28 10:17:39,099 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1658, 5.5845, 5.3240, 5.3781, 4.8246, 4.6339, 5.0354, 5.6937], device='cuda:6'), covar=tensor([0.0834, 0.0885, 0.1094, 0.0542, 0.0798, 0.0767, 0.0773, 0.0746], device='cuda:6'), in_proj_covar=tensor([0.0363, 0.0482, 0.0404, 0.0311, 0.0307, 0.0313, 0.0389, 0.0343], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 10:17:46,534 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50808.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:18:02,610 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.33 vs. limit=5.0 2023-04-28 10:18:20,195 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=50832.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:18:27,758 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5974, 1.4800, 2.1161, 2.3408, 2.4717, 2.3377, 1.5168, 2.5139], device='cuda:6'), covar=tensor([0.0079, 0.0220, 0.0161, 0.0125, 0.0096, 0.0108, 0.0198, 0.0041], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0143, 0.0129, 0.0124, 0.0129, 0.0091, 0.0141, 0.0078], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 10:18:34,744 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50843.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:18:47,052 INFO [train.py:904] (6/8) Epoch 6, batch 100, loss[loss=0.2488, simple_loss=0.3157, pruned_loss=0.09095, over 16295.00 frames. ], tot_loss[loss=0.2337, simple_loss=0.3079, pruned_loss=0.0798, over 1325197.37 frames. ], batch size: 165, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:18:49,623 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50853.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 10:19:11,789 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.233e+02 3.517e+02 4.205e+02 5.493e+02 1.012e+03, threshold=8.410e+02, percent-clipped=3.0 2023-04-28 10:19:12,326 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50869.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 10:19:54,850 INFO [train.py:904] (6/8) Epoch 6, batch 150, loss[loss=0.2355, simple_loss=0.3, pruned_loss=0.08547, over 16870.00 frames. ], tot_loss[loss=0.2323, simple_loss=0.3053, pruned_loss=0.07969, over 1753975.14 frames. ], batch size: 116, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:19:55,162 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=50901.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:19:56,155 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50902.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:20:57,561 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50946.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:21:05,274 INFO [train.py:904] (6/8) Epoch 6, batch 200, loss[loss=0.2573, simple_loss=0.3062, pruned_loss=0.1042, over 16794.00 frames. ], tot_loss[loss=0.2313, simple_loss=0.3042, pruned_loss=0.07915, over 2101430.02 frames. ], batch size: 83, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:21:28,624 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.959e+02 3.434e+02 3.943e+02 4.796e+02 1.132e+03, threshold=7.886e+02, percent-clipped=3.0 2023-04-28 10:21:29,174 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8909, 3.3863, 2.7446, 4.5292, 4.0638, 4.2923, 1.5104, 3.1094], device='cuda:6'), covar=tensor([0.1307, 0.0421, 0.0988, 0.0065, 0.0260, 0.0316, 0.1364, 0.0679], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0141, 0.0168, 0.0088, 0.0163, 0.0178, 0.0162, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 10:22:12,665 INFO [train.py:904] (6/8) Epoch 6, batch 250, loss[loss=0.1796, simple_loss=0.2581, pruned_loss=0.05061, over 16788.00 frames. ], tot_loss[loss=0.228, simple_loss=0.301, pruned_loss=0.07748, over 2379600.20 frames. ], batch size: 39, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:22:21,462 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51007.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:23:20,591 INFO [train.py:904] (6/8) Epoch 6, batch 300, loss[loss=0.191, simple_loss=0.278, pruned_loss=0.05201, over 17113.00 frames. ], tot_loss[loss=0.2231, simple_loss=0.2971, pruned_loss=0.07449, over 2588828.80 frames. ], batch size: 48, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:23:22,785 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8182, 1.6349, 1.3259, 1.3532, 1.9271, 1.6776, 1.6623, 1.9339], device='cuda:6'), covar=tensor([0.0049, 0.0138, 0.0215, 0.0195, 0.0101, 0.0141, 0.0092, 0.0095], device='cuda:6'), in_proj_covar=tensor([0.0081, 0.0160, 0.0161, 0.0155, 0.0153, 0.0159, 0.0135, 0.0140], device='cuda:6'), out_proj_covar=tensor([9.5677e-05, 1.9199e-04, 1.8850e-04, 1.8275e-04, 1.8552e-04, 1.9084e-04, 1.5593e-04, 1.6807e-04], device='cuda:6') 2023-04-28 10:23:37,403 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51062.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:23:45,700 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.679e+02 2.797e+02 3.691e+02 4.478e+02 8.058e+02, threshold=7.381e+02, percent-clipped=1.0 2023-04-28 10:23:50,000 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-04-28 10:24:30,407 INFO [train.py:904] (6/8) Epoch 6, batch 350, loss[loss=0.2215, simple_loss=0.2791, pruned_loss=0.08195, over 16714.00 frames. ], tot_loss[loss=0.22, simple_loss=0.2941, pruned_loss=0.07292, over 2751396.11 frames. ], batch size: 134, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:24:41,128 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7020, 4.1388, 3.2264, 2.3361, 3.0508, 2.3158, 4.4370, 4.1825], device='cuda:6'), covar=tensor([0.2424, 0.0572, 0.1229, 0.1788, 0.2165, 0.1534, 0.0383, 0.0646], device='cuda:6'), in_proj_covar=tensor([0.0281, 0.0253, 0.0273, 0.0249, 0.0271, 0.0205, 0.0245, 0.0256], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 10:24:42,709 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8266, 5.6785, 5.7398, 5.5400, 5.5558, 6.0787, 5.7308, 5.4172], device='cuda:6'), covar=tensor([0.0731, 0.1320, 0.1431, 0.1564, 0.2251, 0.0829, 0.0985, 0.2083], device='cuda:6'), in_proj_covar=tensor([0.0284, 0.0411, 0.0401, 0.0353, 0.0473, 0.0430, 0.0328, 0.0473], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 10:25:01,828 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51123.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:25:19,514 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51138.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:25:37,101 INFO [train.py:904] (6/8) Epoch 6, batch 400, loss[loss=0.2082, simple_loss=0.2959, pruned_loss=0.06031, over 17274.00 frames. ], tot_loss[loss=0.2185, simple_loss=0.2925, pruned_loss=0.07225, over 2882748.40 frames. ], batch size: 52, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:25:55,420 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51164.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:25:56,566 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1895, 5.1602, 4.9293, 4.3506, 4.8887, 1.8015, 4.7210, 4.9693], device='cuda:6'), covar=tensor([0.0047, 0.0043, 0.0087, 0.0262, 0.0064, 0.1784, 0.0082, 0.0113], device='cuda:6'), in_proj_covar=tensor([0.0096, 0.0083, 0.0131, 0.0124, 0.0097, 0.0151, 0.0111, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 10:25:58,017 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-28 10:26:01,707 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.109e+02 2.940e+02 3.610e+02 4.239e+02 7.005e+02, threshold=7.220e+02, percent-clipped=1.0 2023-04-28 10:26:05,684 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51172.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 10:26:45,054 INFO [train.py:904] (6/8) Epoch 6, batch 450, loss[loss=0.1981, simple_loss=0.2849, pruned_loss=0.05563, over 17025.00 frames. ], tot_loss[loss=0.2169, simple_loss=0.291, pruned_loss=0.07137, over 2983181.29 frames. ], batch size: 50, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:26:47,151 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51202.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:27:15,626 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5298, 3.5354, 3.2941, 3.1137, 3.0505, 3.3898, 3.1835, 3.1662], device='cuda:6'), covar=tensor([0.0459, 0.0324, 0.0207, 0.0198, 0.0623, 0.0253, 0.1165, 0.0417], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0208, 0.0227, 0.0196, 0.0260, 0.0236, 0.0161, 0.0263], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 10:27:29,537 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51233.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 10:27:52,970 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=51250.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:27:53,840 INFO [train.py:904] (6/8) Epoch 6, batch 500, loss[loss=0.2051, simple_loss=0.2863, pruned_loss=0.06197, over 16720.00 frames. ], tot_loss[loss=0.2147, simple_loss=0.2887, pruned_loss=0.07034, over 3055784.47 frames. ], batch size: 57, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:28:17,361 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.948e+02 3.090e+02 3.738e+02 4.449e+02 8.454e+02, threshold=7.475e+02, percent-clipped=2.0 2023-04-28 10:29:01,774 INFO [train.py:904] (6/8) Epoch 6, batch 550, loss[loss=0.238, simple_loss=0.2967, pruned_loss=0.08961, over 16558.00 frames. ], tot_loss[loss=0.2127, simple_loss=0.2866, pruned_loss=0.06938, over 3104347.51 frames. ], batch size: 146, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:29:03,895 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51302.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:30:12,926 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-28 10:30:13,391 INFO [train.py:904] (6/8) Epoch 6, batch 600, loss[loss=0.2235, simple_loss=0.2838, pruned_loss=0.08159, over 16704.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.2852, pruned_loss=0.06905, over 3144002.22 frames. ], batch size: 134, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:30:30,467 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2564, 5.0531, 5.0343, 4.7568, 4.5593, 5.0007, 4.9850, 4.6629], device='cuda:6'), covar=tensor([0.0399, 0.0284, 0.0194, 0.0191, 0.0918, 0.0304, 0.0235, 0.0547], device='cuda:6'), in_proj_covar=tensor([0.0197, 0.0209, 0.0229, 0.0199, 0.0262, 0.0236, 0.0162, 0.0265], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 10:30:38,609 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.911e+02 2.906e+02 3.488e+02 4.318e+02 1.008e+03, threshold=6.975e+02, percent-clipped=1.0 2023-04-28 10:30:55,691 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51382.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:31:10,948 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51392.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:31:23,783 INFO [train.py:904] (6/8) Epoch 6, batch 650, loss[loss=0.2198, simple_loss=0.3074, pruned_loss=0.06613, over 16612.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.285, pruned_loss=0.06923, over 3177864.77 frames. ], batch size: 62, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:31:47,306 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51418.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:31:55,780 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 10:32:14,599 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51438.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:32:21,247 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51443.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:32:28,219 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6102, 2.0647, 2.2790, 4.1856, 1.9964, 2.8193, 2.2182, 2.1882], device='cuda:6'), covar=tensor([0.0581, 0.2370, 0.1172, 0.0316, 0.3012, 0.1224, 0.2245, 0.2407], device='cuda:6'), in_proj_covar=tensor([0.0324, 0.0327, 0.0272, 0.0318, 0.0374, 0.0338, 0.0300, 0.0396], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 10:32:31,561 INFO [train.py:904] (6/8) Epoch 6, batch 700, loss[loss=0.3022, simple_loss=0.3498, pruned_loss=0.1273, over 15423.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2848, pruned_loss=0.06879, over 3211436.54 frames. ], batch size: 190, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:32:35,225 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51453.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:32:49,202 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51464.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 10:32:57,145 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.941e+02 3.067e+02 3.781e+02 4.816e+02 1.338e+03, threshold=7.562e+02, percent-clipped=6.0 2023-04-28 10:33:19,513 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=51486.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:33:32,453 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51495.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:33:39,717 INFO [train.py:904] (6/8) Epoch 6, batch 750, loss[loss=0.2452, simple_loss=0.3077, pruned_loss=0.09133, over 15624.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2856, pruned_loss=0.06905, over 3238278.14 frames. ], batch size: 191, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:33:56,286 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=51512.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:34:19,277 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51528.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 10:34:49,286 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9102, 1.7306, 2.3358, 2.8636, 2.6489, 3.1459, 2.0069, 3.0616], device='cuda:6'), covar=tensor([0.0094, 0.0253, 0.0172, 0.0143, 0.0130, 0.0096, 0.0214, 0.0067], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0147, 0.0133, 0.0131, 0.0133, 0.0096, 0.0146, 0.0082], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 10:34:53,034 INFO [train.py:904] (6/8) Epoch 6, batch 800, loss[loss=0.1949, simple_loss=0.2801, pruned_loss=0.05488, over 16636.00 frames. ], tot_loss[loss=0.2111, simple_loss=0.2857, pruned_loss=0.06825, over 3255317.07 frames. ], batch size: 62, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:34:59,964 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51556.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:35:19,940 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.998e+02 2.921e+02 3.386e+02 4.262e+02 8.289e+02, threshold=6.772e+02, percent-clipped=2.0 2023-04-28 10:36:01,900 INFO [train.py:904] (6/8) Epoch 6, batch 850, loss[loss=0.215, simple_loss=0.2803, pruned_loss=0.07486, over 16846.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2843, pruned_loss=0.06678, over 3277036.04 frames. ], batch size: 109, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:36:04,149 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51602.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:37:10,642 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=51650.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:37:11,527 INFO [train.py:904] (6/8) Epoch 6, batch 900, loss[loss=0.1998, simple_loss=0.2744, pruned_loss=0.06262, over 15557.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2823, pruned_loss=0.06546, over 3285821.70 frames. ], batch size: 190, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:37:39,516 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.821e+02 2.905e+02 3.526e+02 4.406e+02 8.198e+02, threshold=7.052e+02, percent-clipped=7.0 2023-04-28 10:38:22,587 INFO [train.py:904] (6/8) Epoch 6, batch 950, loss[loss=0.2261, simple_loss=0.3085, pruned_loss=0.07188, over 16756.00 frames. ], tot_loss[loss=0.2067, simple_loss=0.2826, pruned_loss=0.06543, over 3289179.91 frames. ], batch size: 57, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:38:46,306 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51718.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:39:11,273 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-28 10:39:12,962 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51738.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:39:25,723 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51747.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:39:26,806 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51748.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:39:30,790 INFO [train.py:904] (6/8) Epoch 6, batch 1000, loss[loss=0.1946, simple_loss=0.2814, pruned_loss=0.05392, over 17056.00 frames. ], tot_loss[loss=0.2063, simple_loss=0.2816, pruned_loss=0.06551, over 3299117.63 frames. ], batch size: 50, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:39:51,036 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=51766.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:39:51,404 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-28 10:39:57,333 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.119e+02 2.827e+02 3.297e+02 4.150e+02 8.828e+02, threshold=6.593e+02, percent-clipped=4.0 2023-04-28 10:40:12,051 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6787, 3.8604, 4.2683, 2.9621, 3.8399, 4.1938, 4.0613, 2.4423], device='cuda:6'), covar=tensor([0.0324, 0.0036, 0.0024, 0.0218, 0.0043, 0.0037, 0.0033, 0.0296], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0064, 0.0061, 0.0117, 0.0064, 0.0071, 0.0064, 0.0110], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 10:40:27,261 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8812, 3.8645, 4.0370, 3.0183, 3.8057, 3.9755, 3.9234, 2.1211], device='cuda:6'), covar=tensor([0.0292, 0.0051, 0.0040, 0.0229, 0.0058, 0.0060, 0.0042, 0.0381], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0064, 0.0061, 0.0118, 0.0064, 0.0072, 0.0065, 0.0110], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 10:40:39,947 INFO [train.py:904] (6/8) Epoch 6, batch 1050, loss[loss=0.178, simple_loss=0.2573, pruned_loss=0.04934, over 16363.00 frames. ], tot_loss[loss=0.206, simple_loss=0.2815, pruned_loss=0.06523, over 3291105.41 frames. ], batch size: 36, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:40:49,364 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51808.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:41:18,002 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51828.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:41:27,842 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8251, 3.4377, 2.6818, 4.5731, 4.0732, 4.3283, 1.7419, 3.3340], device='cuda:6'), covar=tensor([0.1265, 0.0426, 0.0973, 0.0091, 0.0294, 0.0305, 0.1166, 0.0548], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0143, 0.0168, 0.0093, 0.0182, 0.0185, 0.0161, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 10:41:32,900 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-28 10:41:49,452 INFO [train.py:904] (6/8) Epoch 6, batch 1100, loss[loss=0.173, simple_loss=0.2547, pruned_loss=0.04561, over 16808.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2804, pruned_loss=0.06462, over 3299473.49 frames. ], batch size: 39, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:41:49,769 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51851.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:42:16,638 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.653e+02 2.702e+02 3.374e+02 4.258e+02 9.547e+02, threshold=6.748e+02, percent-clipped=3.0 2023-04-28 10:42:25,309 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=51876.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 10:42:59,168 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-28 10:42:59,413 INFO [train.py:904] (6/8) Epoch 6, batch 1150, loss[loss=0.1638, simple_loss=0.2458, pruned_loss=0.04092, over 16844.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2797, pruned_loss=0.06395, over 3311587.34 frames. ], batch size: 39, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:44:08,004 INFO [train.py:904] (6/8) Epoch 6, batch 1200, loss[loss=0.1959, simple_loss=0.2798, pruned_loss=0.05604, over 17125.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2792, pruned_loss=0.0629, over 3317658.39 frames. ], batch size: 49, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:44:12,547 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1905, 3.0211, 3.5012, 2.3074, 3.2554, 3.4622, 3.3560, 2.1113], device='cuda:6'), covar=tensor([0.0310, 0.0124, 0.0027, 0.0207, 0.0044, 0.0043, 0.0044, 0.0252], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0064, 0.0061, 0.0116, 0.0063, 0.0070, 0.0065, 0.0108], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 10:44:33,628 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.991e+02 2.935e+02 3.395e+02 4.046e+02 1.078e+03, threshold=6.791e+02, percent-clipped=4.0 2023-04-28 10:44:51,266 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8729, 3.0530, 2.4222, 4.1981, 3.8227, 3.9610, 1.6438, 2.7845], device='cuda:6'), covar=tensor([0.1051, 0.0403, 0.0949, 0.0086, 0.0255, 0.0400, 0.1051, 0.0704], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0143, 0.0166, 0.0092, 0.0183, 0.0185, 0.0160, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 10:45:20,379 INFO [train.py:904] (6/8) Epoch 6, batch 1250, loss[loss=0.2385, simple_loss=0.292, pruned_loss=0.09254, over 16901.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2795, pruned_loss=0.06393, over 3315971.49 frames. ], batch size: 96, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:45:23,184 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 10:45:28,079 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8259, 4.0390, 1.9882, 4.3350, 2.6763, 4.3746, 2.1553, 3.0880], device='cuda:6'), covar=tensor([0.0137, 0.0227, 0.1474, 0.0091, 0.0710, 0.0323, 0.1316, 0.0571], device='cuda:6'), in_proj_covar=tensor([0.0122, 0.0161, 0.0179, 0.0086, 0.0161, 0.0192, 0.0185, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 10:46:13,396 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52038.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:46:15,752 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52040.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:46:26,712 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52048.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:46:30,467 INFO [train.py:904] (6/8) Epoch 6, batch 1300, loss[loss=0.2067, simple_loss=0.2925, pruned_loss=0.06044, over 16609.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2784, pruned_loss=0.06362, over 3313688.85 frames. ], batch size: 62, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:46:58,328 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.022e+02 2.918e+02 3.490e+02 4.202e+02 7.834e+02, threshold=6.979e+02, percent-clipped=4.0 2023-04-28 10:46:59,885 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4290, 1.4501, 1.9262, 2.3121, 2.4215, 2.2006, 1.5006, 2.3573], device='cuda:6'), covar=tensor([0.0075, 0.0230, 0.0142, 0.0117, 0.0094, 0.0131, 0.0208, 0.0047], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0145, 0.0130, 0.0129, 0.0132, 0.0096, 0.0142, 0.0081], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 10:47:20,307 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=52086.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:47:34,137 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=52096.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:47:41,500 INFO [train.py:904] (6/8) Epoch 6, batch 1350, loss[loss=0.1581, simple_loss=0.2484, pruned_loss=0.03387, over 17165.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2786, pruned_loss=0.06369, over 3315558.36 frames. ], batch size: 46, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:47:42,030 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52101.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:47:44,907 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52103.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:48:12,051 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-28 10:48:30,916 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-28 10:48:51,203 INFO [train.py:904] (6/8) Epoch 6, batch 1400, loss[loss=0.1976, simple_loss=0.2847, pruned_loss=0.05521, over 16976.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2784, pruned_loss=0.06391, over 3320440.43 frames. ], batch size: 41, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:48:51,540 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52151.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:49:19,188 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.784e+02 2.727e+02 3.349e+02 4.009e+02 8.330e+02, threshold=6.698e+02, percent-clipped=2.0 2023-04-28 10:49:23,748 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3882, 2.9430, 2.5642, 2.3651, 2.3118, 2.1481, 2.9290, 2.9274], device='cuda:6'), covar=tensor([0.1776, 0.0711, 0.1143, 0.1405, 0.1797, 0.1453, 0.0425, 0.0741], device='cuda:6'), in_proj_covar=tensor([0.0282, 0.0255, 0.0272, 0.0248, 0.0285, 0.0205, 0.0250, 0.0272], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 10:49:49,762 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52193.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:49:58,142 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=52199.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:49:58,280 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1623, 4.9095, 5.1301, 5.4009, 5.5573, 4.7809, 5.4903, 5.4926], device='cuda:6'), covar=tensor([0.0892, 0.0771, 0.1319, 0.0520, 0.0396, 0.0619, 0.0412, 0.0378], device='cuda:6'), in_proj_covar=tensor([0.0445, 0.0545, 0.0697, 0.0559, 0.0421, 0.0414, 0.0436, 0.0470], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 10:50:00,566 INFO [train.py:904] (6/8) Epoch 6, batch 1450, loss[loss=0.1808, simple_loss=0.2638, pruned_loss=0.0489, over 16968.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2778, pruned_loss=0.0632, over 3319395.32 frames. ], batch size: 41, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:51:10,831 INFO [train.py:904] (6/8) Epoch 6, batch 1500, loss[loss=0.154, simple_loss=0.2433, pruned_loss=0.03235, over 17226.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2782, pruned_loss=0.06333, over 3315446.03 frames. ], batch size: 44, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:51:15,519 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52254.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:51:38,533 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.881e+02 3.021e+02 3.666e+02 4.541e+02 9.509e+02, threshold=7.332e+02, percent-clipped=3.0 2023-04-28 10:52:11,062 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-28 10:52:18,640 INFO [train.py:904] (6/8) Epoch 6, batch 1550, loss[loss=0.233, simple_loss=0.3009, pruned_loss=0.08251, over 15530.00 frames. ], tot_loss[loss=0.205, simple_loss=0.28, pruned_loss=0.06501, over 3318565.13 frames. ], batch size: 191, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:52:34,654 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52312.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:53:01,170 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2064, 5.2092, 4.9612, 4.3067, 4.9736, 1.9186, 4.7531, 5.0192], device='cuda:6'), covar=tensor([0.0059, 0.0047, 0.0096, 0.0368, 0.0064, 0.1747, 0.0108, 0.0130], device='cuda:6'), in_proj_covar=tensor([0.0103, 0.0090, 0.0139, 0.0138, 0.0105, 0.0152, 0.0122, 0.0134], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 10:53:28,027 INFO [train.py:904] (6/8) Epoch 6, batch 1600, loss[loss=0.1968, simple_loss=0.2861, pruned_loss=0.05378, over 17107.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2819, pruned_loss=0.06543, over 3313893.90 frames. ], batch size: 48, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:53:55,822 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.894e+02 2.932e+02 3.332e+02 3.992e+02 7.476e+02, threshold=6.664e+02, percent-clipped=1.0 2023-04-28 10:53:59,252 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52373.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:54:21,510 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-28 10:54:30,508 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52396.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:54:36,390 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0623, 4.0279, 4.4397, 3.2894, 3.9653, 4.2707, 3.8561, 2.5084], device='cuda:6'), covar=tensor([0.0240, 0.0040, 0.0020, 0.0176, 0.0034, 0.0041, 0.0038, 0.0250], device='cuda:6'), in_proj_covar=tensor([0.0117, 0.0062, 0.0060, 0.0114, 0.0062, 0.0071, 0.0065, 0.0107], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 10:54:36,725 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-28 10:54:37,129 INFO [train.py:904] (6/8) Epoch 6, batch 1650, loss[loss=0.1983, simple_loss=0.2889, pruned_loss=0.0539, over 17299.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2829, pruned_loss=0.06579, over 3320687.75 frames. ], batch size: 52, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:54:40,268 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52403.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:54:51,299 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0964, 4.9720, 4.9256, 4.6785, 4.5082, 4.9075, 4.9784, 4.5738], device='cuda:6'), covar=tensor([0.0441, 0.0343, 0.0176, 0.0206, 0.0813, 0.0307, 0.0246, 0.0462], device='cuda:6'), in_proj_covar=tensor([0.0204, 0.0220, 0.0240, 0.0212, 0.0275, 0.0248, 0.0171, 0.0275], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 10:55:25,837 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8454, 4.3088, 2.3564, 4.7405, 3.0579, 4.7490, 2.4445, 3.3057], device='cuda:6'), covar=tensor([0.0173, 0.0252, 0.1379, 0.0052, 0.0692, 0.0300, 0.1347, 0.0533], device='cuda:6'), in_proj_covar=tensor([0.0122, 0.0162, 0.0178, 0.0087, 0.0160, 0.0193, 0.0186, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 10:55:45,565 INFO [train.py:904] (6/8) Epoch 6, batch 1700, loss[loss=0.2306, simple_loss=0.3136, pruned_loss=0.0738, over 16492.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2848, pruned_loss=0.06604, over 3315070.95 frames. ], batch size: 68, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:55:45,911 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=52451.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:56:14,195 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.976e+02 2.939e+02 3.544e+02 4.398e+02 9.866e+02, threshold=7.088e+02, percent-clipped=4.0 2023-04-28 10:56:16,458 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7919, 4.7756, 4.5960, 4.1041, 4.6802, 1.7987, 4.4448, 4.6012], device='cuda:6'), covar=tensor([0.0067, 0.0049, 0.0102, 0.0273, 0.0056, 0.1733, 0.0101, 0.0112], device='cuda:6'), in_proj_covar=tensor([0.0103, 0.0090, 0.0140, 0.0137, 0.0104, 0.0150, 0.0121, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 10:56:57,691 INFO [train.py:904] (6/8) Epoch 6, batch 1750, loss[loss=0.1785, simple_loss=0.2614, pruned_loss=0.04779, over 16755.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2853, pruned_loss=0.06569, over 3316702.29 frames. ], batch size: 39, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:58:05,489 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52549.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:58:07,618 INFO [train.py:904] (6/8) Epoch 6, batch 1800, loss[loss=0.1932, simple_loss=0.2732, pruned_loss=0.05665, over 16004.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2861, pruned_loss=0.06585, over 3319221.79 frames. ], batch size: 35, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:58:13,006 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2836, 3.2638, 3.8723, 2.6200, 3.4752, 3.6599, 3.5660, 2.2603], device='cuda:6'), covar=tensor([0.0321, 0.0106, 0.0028, 0.0204, 0.0046, 0.0055, 0.0044, 0.0261], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0061, 0.0059, 0.0111, 0.0061, 0.0070, 0.0064, 0.0105], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 10:58:14,087 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52555.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 10:58:36,301 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.850e+02 3.092e+02 3.773e+02 4.797e+02 9.614e+02, threshold=7.547e+02, percent-clipped=5.0 2023-04-28 10:59:00,744 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6897, 3.4043, 2.6762, 4.8514, 4.1897, 4.4121, 1.6391, 3.1595], device='cuda:6'), covar=tensor([0.1364, 0.0512, 0.1037, 0.0072, 0.0267, 0.0323, 0.1353, 0.0696], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0144, 0.0167, 0.0095, 0.0186, 0.0186, 0.0161, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 10:59:17,730 INFO [train.py:904] (6/8) Epoch 6, batch 1850, loss[loss=0.238, simple_loss=0.3064, pruned_loss=0.08481, over 16797.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2866, pruned_loss=0.06554, over 3315884.35 frames. ], batch size: 102, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 10:59:38,405 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52616.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 11:00:27,090 INFO [train.py:904] (6/8) Epoch 6, batch 1900, loss[loss=0.2007, simple_loss=0.2719, pruned_loss=0.06471, over 16842.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2849, pruned_loss=0.06454, over 3313195.74 frames. ], batch size: 116, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:00:34,441 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2560, 4.2831, 4.7682, 4.8314, 4.7796, 4.3906, 4.3864, 4.2017], device='cuda:6'), covar=tensor([0.0303, 0.0389, 0.0297, 0.0354, 0.0330, 0.0260, 0.0747, 0.0460], device='cuda:6'), in_proj_covar=tensor([0.0269, 0.0268, 0.0266, 0.0265, 0.0322, 0.0280, 0.0395, 0.0234], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 11:00:51,239 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52668.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:00:54,524 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.841e+02 2.667e+02 3.390e+02 4.181e+02 1.051e+03, threshold=6.780e+02, percent-clipped=5.0 2023-04-28 11:01:07,527 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5006, 3.3806, 3.9243, 2.6520, 3.5359, 3.6859, 3.6572, 2.3214], device='cuda:6'), covar=tensor([0.0278, 0.0119, 0.0023, 0.0212, 0.0058, 0.0050, 0.0038, 0.0267], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0061, 0.0059, 0.0112, 0.0061, 0.0070, 0.0064, 0.0106], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 11:01:29,542 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9949, 2.3197, 2.2376, 3.0699, 2.6987, 3.3328, 1.7381, 2.6459], device='cuda:6'), covar=tensor([0.0997, 0.0473, 0.0881, 0.0094, 0.0204, 0.0305, 0.1108, 0.0634], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0143, 0.0166, 0.0095, 0.0185, 0.0186, 0.0160, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 11:01:30,644 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52696.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:01:36,633 INFO [train.py:904] (6/8) Epoch 6, batch 1950, loss[loss=0.2271, simple_loss=0.3028, pruned_loss=0.07573, over 15676.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2851, pruned_loss=0.06393, over 3316224.67 frames. ], batch size: 191, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:02:17,508 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0759, 5.4752, 5.5465, 5.4385, 5.3463, 5.9942, 5.5189, 5.3249], device='cuda:6'), covar=tensor([0.0633, 0.1744, 0.1341, 0.1741, 0.2722, 0.0864, 0.1230, 0.2408], device='cuda:6'), in_proj_covar=tensor([0.0304, 0.0435, 0.0422, 0.0369, 0.0496, 0.0460, 0.0351, 0.0497], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 11:02:18,823 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0111, 1.9669, 2.3275, 2.8349, 2.6159, 3.2728, 1.7740, 3.4123], device='cuda:6'), covar=tensor([0.0099, 0.0215, 0.0157, 0.0141, 0.0140, 0.0081, 0.0229, 0.0055], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0150, 0.0134, 0.0134, 0.0139, 0.0099, 0.0144, 0.0084], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 11:02:36,906 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=52744.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:02:47,588 INFO [train.py:904] (6/8) Epoch 6, batch 2000, loss[loss=0.2058, simple_loss=0.2753, pruned_loss=0.0681, over 16475.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2851, pruned_loss=0.06398, over 3307501.78 frames. ], batch size: 68, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:02:49,214 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52752.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:03:15,694 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.007e+02 2.923e+02 3.553e+02 4.132e+02 6.210e+02, threshold=7.105e+02, percent-clipped=0.0 2023-04-28 11:03:57,170 INFO [train.py:904] (6/8) Epoch 6, batch 2050, loss[loss=0.1875, simple_loss=0.2757, pruned_loss=0.0496, over 17213.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2851, pruned_loss=0.06401, over 3308326.97 frames. ], batch size: 45, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:04:15,394 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52813.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:04:58,603 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-28 11:05:06,140 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52849.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:05:08,343 INFO [train.py:904] (6/8) Epoch 6, batch 2100, loss[loss=0.213, simple_loss=0.2994, pruned_loss=0.0633, over 17071.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2863, pruned_loss=0.06493, over 3306102.75 frames. ], batch size: 53, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:05:19,607 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 11:05:36,281 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.933e+02 3.490e+02 4.297e+02 7.985e+02, threshold=6.979e+02, percent-clipped=3.0 2023-04-28 11:06:12,953 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=52897.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:06:18,010 INFO [train.py:904] (6/8) Epoch 6, batch 2150, loss[loss=0.2325, simple_loss=0.3041, pruned_loss=0.08043, over 17033.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2871, pruned_loss=0.065, over 3318201.25 frames. ], batch size: 50, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:06:33,072 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52911.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 11:07:14,992 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5621, 4.3059, 3.9192, 2.0384, 3.0882, 2.4049, 3.8771, 3.8370], device='cuda:6'), covar=tensor([0.0219, 0.0431, 0.0394, 0.1456, 0.0649, 0.0885, 0.0525, 0.0864], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0135, 0.0152, 0.0140, 0.0130, 0.0125, 0.0139, 0.0143], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 11:07:29,308 INFO [train.py:904] (6/8) Epoch 6, batch 2200, loss[loss=0.1894, simple_loss=0.2818, pruned_loss=0.04852, over 17118.00 frames. ], tot_loss[loss=0.2105, simple_loss=0.2887, pruned_loss=0.06618, over 3314805.77 frames. ], batch size: 48, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:07:52,243 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52968.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:07:56,704 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.110e+02 3.315e+02 3.926e+02 4.640e+02 8.894e+02, threshold=7.853e+02, percent-clipped=3.0 2023-04-28 11:08:38,582 INFO [train.py:904] (6/8) Epoch 6, batch 2250, loss[loss=0.1723, simple_loss=0.2559, pruned_loss=0.04437, over 16823.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.2897, pruned_loss=0.06668, over 3311789.54 frames. ], batch size: 39, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:08:59,896 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=53016.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:09:13,351 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.78 vs. limit=5.0 2023-04-28 11:09:21,285 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6890, 2.3493, 1.6887, 2.1337, 2.8556, 2.6426, 3.0425, 3.0004], device='cuda:6'), covar=tensor([0.0062, 0.0183, 0.0249, 0.0226, 0.0091, 0.0139, 0.0090, 0.0088], device='cuda:6'), in_proj_covar=tensor([0.0094, 0.0165, 0.0164, 0.0161, 0.0163, 0.0167, 0.0153, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:09:32,751 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0774, 5.0583, 4.8625, 4.3160, 4.8943, 2.0503, 4.6672, 4.8516], device='cuda:6'), covar=tensor([0.0048, 0.0048, 0.0097, 0.0271, 0.0051, 0.1565, 0.0080, 0.0117], device='cuda:6'), in_proj_covar=tensor([0.0102, 0.0090, 0.0138, 0.0136, 0.0104, 0.0148, 0.0120, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:09:47,158 INFO [train.py:904] (6/8) Epoch 6, batch 2300, loss[loss=0.1741, simple_loss=0.2654, pruned_loss=0.04145, over 17200.00 frames. ], tot_loss[loss=0.2128, simple_loss=0.2904, pruned_loss=0.06755, over 3308659.37 frames. ], batch size: 44, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:09:51,878 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-28 11:09:54,565 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53056.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:10:15,630 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.951e+02 3.084e+02 3.786e+02 4.874e+02 1.349e+03, threshold=7.572e+02, percent-clipped=4.0 2023-04-28 11:10:57,588 INFO [train.py:904] (6/8) Epoch 6, batch 2350, loss[loss=0.1831, simple_loss=0.2621, pruned_loss=0.05204, over 17042.00 frames. ], tot_loss[loss=0.2129, simple_loss=0.2898, pruned_loss=0.06793, over 3315247.21 frames. ], batch size: 41, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:11:08,084 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53108.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:11:20,213 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53117.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:12:08,722 INFO [train.py:904] (6/8) Epoch 6, batch 2400, loss[loss=0.2122, simple_loss=0.2912, pruned_loss=0.06663, over 16738.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2913, pruned_loss=0.06847, over 3316669.68 frames. ], batch size: 89, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:12:21,076 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2300, 2.1094, 1.5778, 1.8482, 2.4818, 2.3920, 2.6245, 2.6630], device='cuda:6'), covar=tensor([0.0060, 0.0191, 0.0243, 0.0230, 0.0094, 0.0150, 0.0102, 0.0108], device='cuda:6'), in_proj_covar=tensor([0.0094, 0.0166, 0.0164, 0.0161, 0.0164, 0.0168, 0.0155, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:12:36,944 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.873e+02 2.840e+02 3.422e+02 4.151e+02 8.672e+02, threshold=6.844e+02, percent-clipped=2.0 2023-04-28 11:12:50,469 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4735, 4.2176, 3.6637, 1.8461, 3.0770, 2.5547, 3.6769, 3.7937], device='cuda:6'), covar=tensor([0.0234, 0.0432, 0.0470, 0.1587, 0.0687, 0.0832, 0.0597, 0.0732], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0135, 0.0151, 0.0139, 0.0129, 0.0124, 0.0138, 0.0142], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 11:12:51,818 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-28 11:13:19,222 INFO [train.py:904] (6/8) Epoch 6, batch 2450, loss[loss=0.2254, simple_loss=0.2955, pruned_loss=0.07763, over 16476.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.292, pruned_loss=0.0681, over 3323808.82 frames. ], batch size: 146, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:13:33,763 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53211.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 11:14:16,113 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53241.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:14:29,168 INFO [train.py:904] (6/8) Epoch 6, batch 2500, loss[loss=0.2054, simple_loss=0.2983, pruned_loss=0.05621, over 16735.00 frames. ], tot_loss[loss=0.2139, simple_loss=0.2918, pruned_loss=0.06796, over 3319106.34 frames. ], batch size: 62, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:14:35,697 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0804, 1.6455, 2.3610, 2.9290, 2.5238, 3.1639, 1.9650, 3.2467], device='cuda:6'), covar=tensor([0.0107, 0.0242, 0.0164, 0.0133, 0.0160, 0.0102, 0.0211, 0.0063], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0150, 0.0134, 0.0135, 0.0140, 0.0100, 0.0143, 0.0085], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 11:14:40,021 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=53259.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 11:14:52,965 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-28 11:14:57,074 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.925e+02 2.700e+02 3.263e+02 4.236e+02 1.034e+03, threshold=6.525e+02, percent-clipped=4.0 2023-04-28 11:15:38,466 INFO [train.py:904] (6/8) Epoch 6, batch 2550, loss[loss=0.1997, simple_loss=0.2782, pruned_loss=0.06064, over 17237.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.2915, pruned_loss=0.06773, over 3327801.96 frames. ], batch size: 43, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:15:40,617 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9369, 4.6225, 4.8718, 5.1265, 5.2743, 4.6156, 5.2697, 5.2413], device='cuda:6'), covar=tensor([0.0966, 0.0881, 0.1476, 0.0498, 0.0397, 0.0647, 0.0399, 0.0400], device='cuda:6'), in_proj_covar=tensor([0.0435, 0.0531, 0.0688, 0.0550, 0.0416, 0.0408, 0.0431, 0.0461], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:15:40,677 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53302.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:15:43,218 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53304.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:15:58,344 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53315.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:16:48,720 INFO [train.py:904] (6/8) Epoch 6, batch 2600, loss[loss=0.209, simple_loss=0.2918, pruned_loss=0.06313, over 16513.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2917, pruned_loss=0.06774, over 3326164.60 frames. ], batch size: 75, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:17:09,422 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53365.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:17:11,713 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53367.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:17:16,414 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.993e+02 2.796e+02 3.424e+02 4.100e+02 9.306e+02, threshold=6.847e+02, percent-clipped=3.0 2023-04-28 11:17:24,079 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53376.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:17:59,456 INFO [train.py:904] (6/8) Epoch 6, batch 2650, loss[loss=0.2246, simple_loss=0.3039, pruned_loss=0.0727, over 16675.00 frames. ], tot_loss[loss=0.213, simple_loss=0.292, pruned_loss=0.06699, over 3329816.96 frames. ], batch size: 134, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:18:10,392 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53408.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:18:10,542 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4720, 2.0244, 2.2008, 4.0647, 1.9951, 2.8644, 2.1783, 2.2716], device='cuda:6'), covar=tensor([0.0595, 0.2192, 0.1201, 0.0272, 0.2625, 0.1296, 0.2115, 0.2153], device='cuda:6'), in_proj_covar=tensor([0.0336, 0.0341, 0.0280, 0.0321, 0.0380, 0.0358, 0.0308, 0.0410], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:18:14,703 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53412.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:18:37,161 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53428.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 11:18:47,331 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-28 11:19:09,533 INFO [train.py:904] (6/8) Epoch 6, batch 2700, loss[loss=0.202, simple_loss=0.2845, pruned_loss=0.05975, over 16492.00 frames. ], tot_loss[loss=0.212, simple_loss=0.2916, pruned_loss=0.06617, over 3332855.90 frames. ], batch size: 68, lr: 1.16e-02, grad_scale: 4.0 2023-04-28 11:19:16,987 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=53456.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:19:38,362 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.962e+02 2.946e+02 3.581e+02 4.267e+02 8.371e+02, threshold=7.163e+02, percent-clipped=3.0 2023-04-28 11:20:19,909 INFO [train.py:904] (6/8) Epoch 6, batch 2750, loss[loss=0.2051, simple_loss=0.2781, pruned_loss=0.06608, over 16917.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2903, pruned_loss=0.06473, over 3333400.76 frames. ], batch size: 96, lr: 1.16e-02, grad_scale: 4.0 2023-04-28 11:20:43,862 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-28 11:21:31,891 INFO [train.py:904] (6/8) Epoch 6, batch 2800, loss[loss=0.1793, simple_loss=0.2602, pruned_loss=0.04919, over 16891.00 frames. ], tot_loss[loss=0.2095, simple_loss=0.2901, pruned_loss=0.06443, over 3333126.83 frames. ], batch size: 39, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:22:01,718 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.012e+02 2.830e+02 3.513e+02 4.760e+02 1.218e+03, threshold=7.026e+02, percent-clipped=5.0 2023-04-28 11:22:37,935 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53597.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:22:43,361 INFO [train.py:904] (6/8) Epoch 6, batch 2850, loss[loss=0.212, simple_loss=0.2997, pruned_loss=0.06212, over 17239.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2896, pruned_loss=0.06495, over 3335408.85 frames. ], batch size: 52, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:23:51,283 INFO [train.py:904] (6/8) Epoch 6, batch 2900, loss[loss=0.1662, simple_loss=0.2475, pruned_loss=0.04248, over 16912.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2887, pruned_loss=0.06561, over 3328720.78 frames. ], batch size: 42, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:24:04,555 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53660.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:24:19,424 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53671.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:24:20,247 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.792e+02 2.863e+02 3.412e+02 4.656e+02 8.517e+02, threshold=6.823e+02, percent-clipped=1.0 2023-04-28 11:24:59,799 INFO [train.py:904] (6/8) Epoch 6, batch 2950, loss[loss=0.1787, simple_loss=0.2645, pruned_loss=0.04641, over 17215.00 frames. ], tot_loss[loss=0.2091, simple_loss=0.2876, pruned_loss=0.06531, over 3324348.28 frames. ], batch size: 45, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:25:15,585 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53712.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:25:30,209 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53723.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 11:25:35,052 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2826, 5.2013, 5.1380, 4.8666, 4.6876, 5.1232, 5.1226, 4.7561], device='cuda:6'), covar=tensor([0.0420, 0.0297, 0.0184, 0.0194, 0.0987, 0.0256, 0.0219, 0.0564], device='cuda:6'), in_proj_covar=tensor([0.0212, 0.0230, 0.0249, 0.0221, 0.0285, 0.0251, 0.0176, 0.0280], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 11:26:08,566 INFO [train.py:904] (6/8) Epoch 6, batch 3000, loss[loss=0.211, simple_loss=0.2828, pruned_loss=0.06957, over 16774.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2857, pruned_loss=0.06445, over 3324611.88 frames. ], batch size: 83, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:26:08,566 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 11:26:17,403 INFO [train.py:938] (6/8) Epoch 6, validation: loss=0.1514, simple_loss=0.258, pruned_loss=0.02246, over 944034.00 frames. 2023-04-28 11:26:17,404 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17533MB 2023-04-28 11:26:29,471 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=53760.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:26:45,640 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-28 11:26:45,970 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.839e+02 3.365e+02 4.010e+02 8.699e+02, threshold=6.731e+02, percent-clipped=2.0 2023-04-28 11:27:25,829 INFO [train.py:904] (6/8) Epoch 6, batch 3050, loss[loss=0.2487, simple_loss=0.3105, pruned_loss=0.09347, over 16780.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2865, pruned_loss=0.06486, over 3322910.05 frames. ], batch size: 134, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:28:28,613 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1515, 5.0604, 4.9767, 4.2826, 4.9090, 1.7719, 4.7356, 4.9821], device='cuda:6'), covar=tensor([0.0059, 0.0058, 0.0099, 0.0387, 0.0072, 0.1852, 0.0098, 0.0148], device='cuda:6'), in_proj_covar=tensor([0.0104, 0.0093, 0.0144, 0.0140, 0.0107, 0.0150, 0.0124, 0.0136], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:28:35,659 INFO [train.py:904] (6/8) Epoch 6, batch 3100, loss[loss=0.1624, simple_loss=0.2431, pruned_loss=0.0409, over 16792.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2862, pruned_loss=0.06505, over 3323940.69 frames. ], batch size: 39, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:29:05,910 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.546e+02 2.841e+02 3.355e+02 4.130e+02 6.611e+02, threshold=6.710e+02, percent-clipped=0.0 2023-04-28 11:29:13,927 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7048, 5.0393, 4.7419, 4.8371, 4.4854, 4.4924, 4.5647, 5.1252], device='cuda:6'), covar=tensor([0.0845, 0.0769, 0.0956, 0.0552, 0.0688, 0.0836, 0.0710, 0.0688], device='cuda:6'), in_proj_covar=tensor([0.0423, 0.0565, 0.0469, 0.0358, 0.0344, 0.0348, 0.0453, 0.0394], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:29:40,337 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53897.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:29:45,141 INFO [train.py:904] (6/8) Epoch 6, batch 3150, loss[loss=0.1687, simple_loss=0.2519, pruned_loss=0.04275, over 17027.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2857, pruned_loss=0.06528, over 3325442.05 frames. ], batch size: 41, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:30:49,059 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=53945.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:30:56,670 INFO [train.py:904] (6/8) Epoch 6, batch 3200, loss[loss=0.189, simple_loss=0.2701, pruned_loss=0.05395, over 16780.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2844, pruned_loss=0.06479, over 3330509.59 frames. ], batch size: 42, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:31:09,494 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53960.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:31:25,153 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.71 vs. limit=5.0 2023-04-28 11:31:25,922 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53971.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:31:26,625 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.709e+02 2.936e+02 3.448e+02 4.117e+02 1.255e+03, threshold=6.896e+02, percent-clipped=5.0 2023-04-28 11:32:09,113 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4524, 4.1257, 3.6486, 2.0453, 2.9221, 2.4655, 3.6898, 3.8784], device='cuda:6'), covar=tensor([0.0241, 0.0457, 0.0531, 0.1453, 0.0717, 0.0864, 0.0574, 0.0692], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0139, 0.0156, 0.0140, 0.0131, 0.0125, 0.0141, 0.0146], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 11:32:09,686 INFO [train.py:904] (6/8) Epoch 6, batch 3250, loss[loss=0.2168, simple_loss=0.2894, pruned_loss=0.07213, over 16582.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2842, pruned_loss=0.06451, over 3337673.10 frames. ], batch size: 68, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:32:19,694 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=54008.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:32:35,419 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=54019.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:32:36,766 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6963, 3.3007, 2.6957, 4.4963, 3.9673, 4.2742, 1.6027, 3.0506], device='cuda:6'), covar=tensor([0.1332, 0.0440, 0.0924, 0.0092, 0.0273, 0.0293, 0.1283, 0.0646], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0146, 0.0168, 0.0098, 0.0197, 0.0193, 0.0163, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 11:32:40,103 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=54023.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 11:33:17,041 INFO [train.py:904] (6/8) Epoch 6, batch 3300, loss[loss=0.2156, simple_loss=0.302, pruned_loss=0.06462, over 17080.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2861, pruned_loss=0.06554, over 3330407.79 frames. ], batch size: 55, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:33:23,676 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1063, 5.0583, 4.9950, 4.3378, 4.9656, 1.7040, 4.7510, 5.0392], device='cuda:6'), covar=tensor([0.0058, 0.0051, 0.0088, 0.0310, 0.0058, 0.1854, 0.0091, 0.0112], device='cuda:6'), in_proj_covar=tensor([0.0107, 0.0096, 0.0147, 0.0143, 0.0110, 0.0154, 0.0128, 0.0141], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:33:45,194 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=54071.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:33:46,072 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.774e+02 3.091e+02 3.848e+02 4.464e+02 8.976e+02, threshold=7.696e+02, percent-clipped=3.0 2023-04-28 11:33:53,630 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7507, 1.5608, 2.1116, 2.6771, 2.7308, 2.6784, 1.6658, 2.7109], device='cuda:6'), covar=tensor([0.0074, 0.0230, 0.0182, 0.0113, 0.0098, 0.0097, 0.0217, 0.0071], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0151, 0.0136, 0.0136, 0.0140, 0.0101, 0.0144, 0.0089], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 11:33:57,832 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1513, 4.1050, 4.1119, 3.6228, 4.1252, 1.6607, 3.9504, 3.8831], device='cuda:6'), covar=tensor([0.0085, 0.0076, 0.0107, 0.0236, 0.0069, 0.1879, 0.0096, 0.0137], device='cuda:6'), in_proj_covar=tensor([0.0107, 0.0096, 0.0147, 0.0143, 0.0110, 0.0154, 0.0127, 0.0140], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:34:26,275 INFO [train.py:904] (6/8) Epoch 6, batch 3350, loss[loss=0.228, simple_loss=0.293, pruned_loss=0.08148, over 16282.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.2868, pruned_loss=0.06535, over 3334597.97 frames. ], batch size: 165, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:34:27,396 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1842, 4.1364, 4.5717, 4.5203, 4.5662, 4.1026, 4.2565, 4.0431], device='cuda:6'), covar=tensor([0.0268, 0.0424, 0.0303, 0.0391, 0.0358, 0.0327, 0.0688, 0.0526], device='cuda:6'), in_proj_covar=tensor([0.0273, 0.0270, 0.0268, 0.0266, 0.0323, 0.0283, 0.0400, 0.0239], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 11:35:01,835 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5431, 4.8770, 4.6153, 4.6791, 4.2419, 4.3098, 4.4158, 4.9465], device='cuda:6'), covar=tensor([0.0839, 0.0776, 0.0877, 0.0556, 0.0758, 0.1019, 0.0782, 0.0728], device='cuda:6'), in_proj_covar=tensor([0.0431, 0.0574, 0.0475, 0.0361, 0.0353, 0.0356, 0.0459, 0.0399], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:35:10,858 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4910, 4.1260, 4.0385, 1.9921, 3.1039, 2.4371, 3.7755, 3.8241], device='cuda:6'), covar=tensor([0.0279, 0.0537, 0.0361, 0.1497, 0.0638, 0.0901, 0.0704, 0.0983], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0137, 0.0152, 0.0138, 0.0130, 0.0123, 0.0138, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-28 11:35:26,986 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5313, 3.2555, 2.7806, 1.8577, 2.4630, 2.2248, 3.0091, 3.1063], device='cuda:6'), covar=tensor([0.0291, 0.0587, 0.0580, 0.1491, 0.0745, 0.0832, 0.0651, 0.0696], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0137, 0.0152, 0.0138, 0.0130, 0.0123, 0.0139, 0.0145], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 11:35:34,483 INFO [train.py:904] (6/8) Epoch 6, batch 3400, loss[loss=0.2021, simple_loss=0.2855, pruned_loss=0.05929, over 17125.00 frames. ], tot_loss[loss=0.209, simple_loss=0.2874, pruned_loss=0.06529, over 3336365.98 frames. ], batch size: 48, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:36:05,329 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.856e+02 2.871e+02 3.468e+02 4.191e+02 6.688e+02, threshold=6.936e+02, percent-clipped=0.0 2023-04-28 11:36:46,918 INFO [train.py:904] (6/8) Epoch 6, batch 3450, loss[loss=0.1728, simple_loss=0.2548, pruned_loss=0.04542, over 17225.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.287, pruned_loss=0.06504, over 3339680.91 frames. ], batch size: 45, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:36:50,876 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6646, 2.5617, 2.0243, 2.3611, 2.8858, 2.7698, 3.7325, 3.2492], device='cuda:6'), covar=tensor([0.0039, 0.0180, 0.0226, 0.0202, 0.0127, 0.0174, 0.0083, 0.0102], device='cuda:6'), in_proj_covar=tensor([0.0097, 0.0165, 0.0164, 0.0162, 0.0161, 0.0166, 0.0156, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:37:58,616 INFO [train.py:904] (6/8) Epoch 6, batch 3500, loss[loss=0.2152, simple_loss=0.2964, pruned_loss=0.06698, over 17128.00 frames. ], tot_loss[loss=0.2077, simple_loss=0.2853, pruned_loss=0.06506, over 3333293.43 frames. ], batch size: 49, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:38:25,859 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2840, 4.6117, 4.3273, 4.5002, 4.1024, 4.0898, 4.2405, 4.5940], device='cuda:6'), covar=tensor([0.0850, 0.0790, 0.0995, 0.0515, 0.0701, 0.1374, 0.0796, 0.0881], device='cuda:6'), in_proj_covar=tensor([0.0422, 0.0566, 0.0467, 0.0355, 0.0346, 0.0352, 0.0453, 0.0395], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:38:28,902 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.000e+02 2.700e+02 3.164e+02 3.995e+02 8.205e+02, threshold=6.329e+02, percent-clipped=2.0 2023-04-28 11:38:58,350 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-28 11:39:10,634 INFO [train.py:904] (6/8) Epoch 6, batch 3550, loss[loss=0.1917, simple_loss=0.284, pruned_loss=0.04973, over 17273.00 frames. ], tot_loss[loss=0.205, simple_loss=0.2827, pruned_loss=0.06364, over 3335853.71 frames. ], batch size: 52, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:39:35,456 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-28 11:40:21,754 INFO [train.py:904] (6/8) Epoch 6, batch 3600, loss[loss=0.2107, simple_loss=0.2678, pruned_loss=0.07682, over 16886.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2819, pruned_loss=0.06364, over 3340837.27 frames. ], batch size: 109, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:40:51,130 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.869e+02 3.018e+02 3.592e+02 4.498e+02 7.311e+02, threshold=7.184e+02, percent-clipped=7.0 2023-04-28 11:41:33,849 INFO [train.py:904] (6/8) Epoch 6, batch 3650, loss[loss=0.1751, simple_loss=0.2533, pruned_loss=0.04845, over 16787.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.281, pruned_loss=0.06405, over 3331582.07 frames. ], batch size: 39, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:41:57,945 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9874, 4.0133, 4.4332, 4.4184, 4.3874, 4.0472, 4.1400, 3.9712], device='cuda:6'), covar=tensor([0.0287, 0.0440, 0.0282, 0.0335, 0.0347, 0.0306, 0.0622, 0.0464], device='cuda:6'), in_proj_covar=tensor([0.0268, 0.0269, 0.0262, 0.0262, 0.0322, 0.0279, 0.0393, 0.0233], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 11:42:46,527 INFO [train.py:904] (6/8) Epoch 6, batch 3700, loss[loss=0.2283, simple_loss=0.3041, pruned_loss=0.07627, over 16578.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2809, pruned_loss=0.06577, over 3305714.62 frames. ], batch size: 62, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:42:58,719 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-28 11:43:13,866 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-28 11:43:17,519 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.788e+02 2.928e+02 3.594e+02 4.333e+02 7.725e+02, threshold=7.187e+02, percent-clipped=2.0 2023-04-28 11:43:59,568 INFO [train.py:904] (6/8) Epoch 6, batch 3750, loss[loss=0.2089, simple_loss=0.2735, pruned_loss=0.07214, over 16730.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2816, pruned_loss=0.06767, over 3290788.68 frames. ], batch size: 83, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:44:10,802 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3337, 4.3644, 4.3265, 3.8477, 4.2759, 1.8927, 4.1661, 4.2017], device='cuda:6'), covar=tensor([0.0072, 0.0062, 0.0096, 0.0254, 0.0067, 0.1616, 0.0095, 0.0122], device='cuda:6'), in_proj_covar=tensor([0.0103, 0.0093, 0.0142, 0.0139, 0.0108, 0.0149, 0.0124, 0.0136], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:44:52,348 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5895, 3.2572, 2.8093, 1.8603, 2.5589, 2.0622, 3.1273, 3.1340], device='cuda:6'), covar=tensor([0.0270, 0.0484, 0.0570, 0.1496, 0.0751, 0.0911, 0.0621, 0.0788], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0136, 0.0154, 0.0140, 0.0131, 0.0124, 0.0138, 0.0146], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 11:45:13,202 INFO [train.py:904] (6/8) Epoch 6, batch 3800, loss[loss=0.2369, simple_loss=0.2947, pruned_loss=0.0895, over 16876.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2831, pruned_loss=0.06934, over 3291406.30 frames. ], batch size: 109, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:45:18,646 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-28 11:45:46,020 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.146e+02 2.814e+02 3.370e+02 4.517e+02 7.386e+02, threshold=6.740e+02, percent-clipped=3.0 2023-04-28 11:46:28,642 INFO [train.py:904] (6/8) Epoch 6, batch 3850, loss[loss=0.2021, simple_loss=0.2707, pruned_loss=0.06676, over 16935.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2825, pruned_loss=0.06975, over 3294376.59 frames. ], batch size: 109, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:47:40,668 INFO [train.py:904] (6/8) Epoch 6, batch 3900, loss[loss=0.1994, simple_loss=0.271, pruned_loss=0.06389, over 16757.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2819, pruned_loss=0.07035, over 3291663.05 frames. ], batch size: 83, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:47:45,942 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4400, 1.5438, 1.9948, 2.4037, 2.5441, 2.3437, 1.6670, 2.5485], device='cuda:6'), covar=tensor([0.0083, 0.0240, 0.0161, 0.0105, 0.0087, 0.0133, 0.0203, 0.0054], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0151, 0.0137, 0.0137, 0.0141, 0.0103, 0.0145, 0.0090], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 11:47:59,704 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=54664.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 11:48:10,861 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 2.751e+02 3.451e+02 4.155e+02 1.044e+03, threshold=6.903e+02, percent-clipped=2.0 2023-04-28 11:48:53,476 INFO [train.py:904] (6/8) Epoch 6, batch 3950, loss[loss=0.2088, simple_loss=0.2783, pruned_loss=0.06971, over 16463.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2812, pruned_loss=0.07063, over 3288525.62 frames. ], batch size: 146, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:49:27,794 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=54725.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 11:50:04,728 INFO [train.py:904] (6/8) Epoch 6, batch 4000, loss[loss=0.1939, simple_loss=0.2772, pruned_loss=0.05532, over 16363.00 frames. ], tot_loss[loss=0.212, simple_loss=0.2816, pruned_loss=0.07113, over 3280100.91 frames. ], batch size: 35, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:50:17,503 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4244, 3.7133, 3.8810, 1.9423, 4.0969, 4.1430, 3.0684, 3.2723], device='cuda:6'), covar=tensor([0.0739, 0.0141, 0.0124, 0.1012, 0.0049, 0.0074, 0.0303, 0.0272], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0095, 0.0082, 0.0138, 0.0072, 0.0082, 0.0116, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 11:50:30,575 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-28 11:50:36,654 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.045e+02 2.875e+02 3.507e+02 4.223e+02 8.153e+02, threshold=7.015e+02, percent-clipped=4.0 2023-04-28 11:51:17,359 INFO [train.py:904] (6/8) Epoch 6, batch 4050, loss[loss=0.1989, simple_loss=0.2756, pruned_loss=0.06115, over 17212.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2808, pruned_loss=0.06896, over 3288436.28 frames. ], batch size: 45, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:51:42,439 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1366, 1.8395, 2.0357, 3.6016, 1.6526, 2.4203, 1.9624, 1.8873], device='cuda:6'), covar=tensor([0.0834, 0.2944, 0.1534, 0.0510, 0.3899, 0.1752, 0.2837, 0.3076], device='cuda:6'), in_proj_covar=tensor([0.0342, 0.0347, 0.0283, 0.0324, 0.0384, 0.0368, 0.0315, 0.0419], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:52:28,290 INFO [train.py:904] (6/8) Epoch 6, batch 4100, loss[loss=0.2652, simple_loss=0.3393, pruned_loss=0.09551, over 15496.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2815, pruned_loss=0.06781, over 3283135.95 frames. ], batch size: 191, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:52:45,933 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=54862.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:53:01,850 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.687e+02 2.498e+02 3.136e+02 3.951e+02 8.635e+02, threshold=6.272e+02, percent-clipped=2.0 2023-04-28 11:53:45,438 INFO [train.py:904] (6/8) Epoch 6, batch 4150, loss[loss=0.31, simple_loss=0.3589, pruned_loss=0.1305, over 11437.00 frames. ], tot_loss[loss=0.2163, simple_loss=0.2899, pruned_loss=0.0714, over 3249573.81 frames. ], batch size: 246, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:54:19,622 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=54923.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 11:55:01,101 INFO [train.py:904] (6/8) Epoch 6, batch 4200, loss[loss=0.2303, simple_loss=0.3168, pruned_loss=0.07188, over 16889.00 frames. ], tot_loss[loss=0.222, simple_loss=0.2976, pruned_loss=0.07318, over 3237322.62 frames. ], batch size: 116, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:55:30,438 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-28 11:55:33,716 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.052e+02 3.214e+02 3.756e+02 4.661e+02 1.168e+03, threshold=7.512e+02, percent-clipped=6.0 2023-04-28 11:56:15,278 INFO [train.py:904] (6/8) Epoch 6, batch 4250, loss[loss=0.2079, simple_loss=0.2972, pruned_loss=0.05928, over 17244.00 frames. ], tot_loss[loss=0.2243, simple_loss=0.3013, pruned_loss=0.07368, over 3205916.38 frames. ], batch size: 52, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:56:44,665 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55020.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 11:57:03,364 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7965, 4.8131, 4.6383, 4.0831, 4.7401, 1.8145, 4.3532, 4.5347], device='cuda:6'), covar=tensor([0.0050, 0.0040, 0.0091, 0.0234, 0.0047, 0.1628, 0.0091, 0.0112], device='cuda:6'), in_proj_covar=tensor([0.0099, 0.0088, 0.0135, 0.0133, 0.0102, 0.0145, 0.0119, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:57:29,394 INFO [train.py:904] (6/8) Epoch 6, batch 4300, loss[loss=0.2425, simple_loss=0.3223, pruned_loss=0.08138, over 16496.00 frames. ], tot_loss[loss=0.2236, simple_loss=0.302, pruned_loss=0.07262, over 3192348.77 frames. ], batch size: 68, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:58:01,800 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.697e+02 2.606e+02 3.247e+02 3.921e+02 6.703e+02, threshold=6.494e+02, percent-clipped=0.0 2023-04-28 11:58:43,132 INFO [train.py:904] (6/8) Epoch 6, batch 4350, loss[loss=0.2493, simple_loss=0.3215, pruned_loss=0.08849, over 11782.00 frames. ], tot_loss[loss=0.226, simple_loss=0.3052, pruned_loss=0.07335, over 3201332.84 frames. ], batch size: 246, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:59:34,711 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9251, 3.5646, 3.4627, 2.1629, 3.1276, 3.4340, 3.1885, 1.9124], device='cuda:6'), covar=tensor([0.0364, 0.0016, 0.0024, 0.0273, 0.0043, 0.0042, 0.0036, 0.0278], device='cuda:6'), in_proj_covar=tensor([0.0118, 0.0057, 0.0060, 0.0115, 0.0063, 0.0070, 0.0065, 0.0107], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 11:59:34,832 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1875, 1.8553, 2.0370, 3.7790, 1.7410, 2.7301, 2.1478, 1.9842], device='cuda:6'), covar=tensor([0.0676, 0.2443, 0.1388, 0.0327, 0.3213, 0.1183, 0.2018, 0.2614], device='cuda:6'), in_proj_covar=tensor([0.0336, 0.0346, 0.0281, 0.0319, 0.0386, 0.0359, 0.0310, 0.0415], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 11:59:52,184 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-28 11:59:53,084 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55148.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:59:56,594 INFO [train.py:904] (6/8) Epoch 6, batch 4400, loss[loss=0.3056, simple_loss=0.3614, pruned_loss=0.1249, over 11603.00 frames. ], tot_loss[loss=0.2285, simple_loss=0.3074, pruned_loss=0.07477, over 3181483.61 frames. ], batch size: 248, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 12:00:10,677 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-28 12:00:27,147 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.875e+02 2.770e+02 3.222e+02 4.108e+02 7.043e+02, threshold=6.445e+02, percent-clipped=2.0 2023-04-28 12:01:04,270 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8354, 4.1423, 3.8874, 3.9627, 3.5562, 3.7378, 3.8149, 4.0536], device='cuda:6'), covar=tensor([0.0821, 0.0717, 0.0849, 0.0492, 0.0634, 0.1340, 0.0655, 0.0928], device='cuda:6'), in_proj_covar=tensor([0.0388, 0.0519, 0.0435, 0.0332, 0.0317, 0.0334, 0.0422, 0.0366], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:01:06,617 INFO [train.py:904] (6/8) Epoch 6, batch 4450, loss[loss=0.2204, simple_loss=0.3004, pruned_loss=0.07018, over 16635.00 frames. ], tot_loss[loss=0.2303, simple_loss=0.3104, pruned_loss=0.07509, over 3191560.27 frames. ], batch size: 62, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 12:01:19,269 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55209.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:01:21,591 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55211.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:01:31,359 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55218.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 12:02:16,913 INFO [train.py:904] (6/8) Epoch 6, batch 4500, loss[loss=0.2311, simple_loss=0.3182, pruned_loss=0.072, over 17214.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.31, pruned_loss=0.07506, over 3176950.74 frames. ], batch size: 44, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 12:02:47,723 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55272.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:02:48,405 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.800e+02 2.322e+02 2.860e+02 3.458e+02 5.535e+02, threshold=5.719e+02, percent-clipped=0.0 2023-04-28 12:03:29,865 INFO [train.py:904] (6/8) Epoch 6, batch 4550, loss[loss=0.2718, simple_loss=0.3406, pruned_loss=0.1015, over 17270.00 frames. ], tot_loss[loss=0.231, simple_loss=0.3108, pruned_loss=0.07564, over 3193426.57 frames. ], batch size: 52, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 12:03:57,502 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55320.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 12:04:41,589 INFO [train.py:904] (6/8) Epoch 6, batch 4600, loss[loss=0.2147, simple_loss=0.2999, pruned_loss=0.06478, over 16305.00 frames. ], tot_loss[loss=0.2298, simple_loss=0.3104, pruned_loss=0.07465, over 3210476.01 frames. ], batch size: 165, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 12:05:07,045 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=55368.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 12:05:13,324 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.543e+02 2.321e+02 2.807e+02 3.689e+02 1.502e+03, threshold=5.614e+02, percent-clipped=3.0 2023-04-28 12:05:52,743 INFO [train.py:904] (6/8) Epoch 6, batch 4650, loss[loss=0.2058, simple_loss=0.2844, pruned_loss=0.0636, over 16312.00 frames. ], tot_loss[loss=0.228, simple_loss=0.3084, pruned_loss=0.07382, over 3222660.32 frames. ], batch size: 35, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 12:06:12,031 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8359, 5.3911, 5.4635, 5.3735, 5.2637, 5.9947, 5.4274, 5.2870], device='cuda:6'), covar=tensor([0.0742, 0.1177, 0.1051, 0.1505, 0.2377, 0.0814, 0.0931, 0.1997], device='cuda:6'), in_proj_covar=tensor([0.0290, 0.0396, 0.0393, 0.0347, 0.0454, 0.0417, 0.0324, 0.0466], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 12:07:03,014 INFO [train.py:904] (6/8) Epoch 6, batch 4700, loss[loss=0.2158, simple_loss=0.2949, pruned_loss=0.06829, over 16771.00 frames. ], tot_loss[loss=0.2252, simple_loss=0.3056, pruned_loss=0.0724, over 3222086.14 frames. ], batch size: 124, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:07:34,099 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.678e+02 2.601e+02 3.111e+02 3.691e+02 9.251e+02, threshold=6.221e+02, percent-clipped=2.0 2023-04-28 12:08:12,120 INFO [train.py:904] (6/8) Epoch 6, batch 4750, loss[loss=0.2121, simple_loss=0.2951, pruned_loss=0.06458, over 16828.00 frames. ], tot_loss[loss=0.222, simple_loss=0.3024, pruned_loss=0.07082, over 3212246.90 frames. ], batch size: 102, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:08:17,387 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55504.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:08:29,616 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-28 12:08:36,490 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55518.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 12:09:22,952 INFO [train.py:904] (6/8) Epoch 6, batch 4800, loss[loss=0.2243, simple_loss=0.3124, pruned_loss=0.06808, over 15373.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.2987, pruned_loss=0.06882, over 3209978.17 frames. ], batch size: 190, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:09:45,465 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=55566.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:09:46,595 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55567.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:09:53,990 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.469e+02 2.878e+02 3.568e+02 5.612e+02, threshold=5.756e+02, percent-clipped=0.0 2023-04-28 12:10:35,249 INFO [train.py:904] (6/8) Epoch 6, batch 4850, loss[loss=0.2242, simple_loss=0.3132, pruned_loss=0.06766, over 16362.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.3009, pruned_loss=0.06873, over 3192384.81 frames. ], batch size: 146, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:11:30,728 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8100, 3.7667, 4.2019, 4.1694, 4.1803, 3.8705, 3.8785, 3.8328], device='cuda:6'), covar=tensor([0.0244, 0.0474, 0.0297, 0.0367, 0.0352, 0.0258, 0.0667, 0.0371], device='cuda:6'), in_proj_covar=tensor([0.0247, 0.0243, 0.0245, 0.0243, 0.0293, 0.0261, 0.0362, 0.0215], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 12:11:48,112 INFO [train.py:904] (6/8) Epoch 6, batch 4900, loss[loss=0.1858, simple_loss=0.2797, pruned_loss=0.04599, over 16827.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2989, pruned_loss=0.06716, over 3185275.13 frames. ], batch size: 102, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:12:17,310 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2083, 1.7160, 2.5241, 3.2483, 3.0672, 3.6889, 1.7063, 3.3719], device='cuda:6'), covar=tensor([0.0088, 0.0269, 0.0163, 0.0113, 0.0110, 0.0063, 0.0275, 0.0052], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0148, 0.0130, 0.0131, 0.0134, 0.0098, 0.0143, 0.0086], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 12:12:19,092 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.308e+02 2.929e+02 3.604e+02 9.107e+02, threshold=5.857e+02, percent-clipped=2.0 2023-04-28 12:12:59,378 INFO [train.py:904] (6/8) Epoch 6, batch 4950, loss[loss=0.225, simple_loss=0.3074, pruned_loss=0.07132, over 12262.00 frames. ], tot_loss[loss=0.2167, simple_loss=0.2995, pruned_loss=0.06693, over 3189486.74 frames. ], batch size: 246, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:14:05,672 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55748.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:14:08,593 INFO [train.py:904] (6/8) Epoch 6, batch 5000, loss[loss=0.2439, simple_loss=0.3181, pruned_loss=0.08488, over 12388.00 frames. ], tot_loss[loss=0.2189, simple_loss=0.302, pruned_loss=0.06792, over 3179267.92 frames. ], batch size: 246, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:14:26,505 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2858, 3.9699, 3.4729, 1.9341, 2.9769, 2.3377, 3.5954, 3.8062], device='cuda:6'), covar=tensor([0.0221, 0.0460, 0.0522, 0.1499, 0.0704, 0.0897, 0.0557, 0.0512], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0128, 0.0152, 0.0140, 0.0131, 0.0123, 0.0138, 0.0139], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 12:14:38,593 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 2.723e+02 3.199e+02 3.755e+02 7.199e+02, threshold=6.398e+02, percent-clipped=5.0 2023-04-28 12:15:21,195 INFO [train.py:904] (6/8) Epoch 6, batch 5050, loss[loss=0.243, simple_loss=0.3276, pruned_loss=0.07922, over 16498.00 frames. ], tot_loss[loss=0.2185, simple_loss=0.3017, pruned_loss=0.06767, over 3189527.48 frames. ], batch size: 146, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:15:22,212 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-28 12:15:25,536 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55804.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:15:27,349 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6452, 3.6497, 4.0431, 3.9816, 3.9790, 3.7339, 3.7729, 3.7066], device='cuda:6'), covar=tensor([0.0269, 0.0430, 0.0312, 0.0389, 0.0401, 0.0273, 0.0640, 0.0407], device='cuda:6'), in_proj_covar=tensor([0.0252, 0.0243, 0.0248, 0.0249, 0.0298, 0.0266, 0.0366, 0.0218], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 12:15:33,302 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55809.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:16:10,618 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0734, 5.0144, 4.9907, 4.2406, 4.9294, 2.0410, 4.7060, 4.9676], device='cuda:6'), covar=tensor([0.0046, 0.0049, 0.0068, 0.0356, 0.0054, 0.1503, 0.0074, 0.0100], device='cuda:6'), in_proj_covar=tensor([0.0098, 0.0087, 0.0133, 0.0135, 0.0100, 0.0147, 0.0115, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:16:15,044 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3107, 3.4816, 1.8137, 3.7105, 2.3422, 3.5950, 1.8768, 2.5208], device='cuda:6'), covar=tensor([0.0177, 0.0254, 0.1625, 0.0052, 0.0869, 0.0349, 0.1444, 0.0716], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0157, 0.0178, 0.0083, 0.0164, 0.0189, 0.0186, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 12:16:28,808 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:16:31,556 INFO [train.py:904] (6/8) Epoch 6, batch 5100, loss[loss=0.2186, simple_loss=0.2976, pruned_loss=0.06978, over 16913.00 frames. ], tot_loss[loss=0.2163, simple_loss=0.2998, pruned_loss=0.06641, over 3198050.51 frames. ], batch size: 109, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:16:32,965 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=55852.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:16:33,369 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.51 vs. limit=5.0 2023-04-28 12:16:54,894 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55867.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:17:03,854 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8105, 3.6837, 3.7464, 4.0558, 4.1448, 3.8206, 4.1386, 4.1204], device='cuda:6'), covar=tensor([0.1021, 0.0800, 0.1697, 0.0614, 0.0561, 0.1141, 0.0506, 0.0567], device='cuda:6'), in_proj_covar=tensor([0.0411, 0.0493, 0.0642, 0.0508, 0.0385, 0.0386, 0.0395, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:17:04,477 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.683e+02 3.138e+02 3.930e+02 6.713e+02, threshold=6.277e+02, percent-clipped=2.0 2023-04-28 12:17:06,764 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7508, 2.0946, 1.6863, 1.9461, 2.5192, 2.2651, 2.8355, 2.7203], device='cuda:6'), covar=tensor([0.0050, 0.0202, 0.0260, 0.0224, 0.0104, 0.0187, 0.0073, 0.0114], device='cuda:6'), in_proj_covar=tensor([0.0088, 0.0159, 0.0160, 0.0158, 0.0154, 0.0165, 0.0142, 0.0146], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:17:09,972 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-04-28 12:17:43,209 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-28 12:17:45,240 INFO [train.py:904] (6/8) Epoch 6, batch 5150, loss[loss=0.2158, simple_loss=0.3103, pruned_loss=0.0606, over 15406.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.2997, pruned_loss=0.06561, over 3192425.97 frames. ], batch size: 190, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:17:57,067 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55909.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:18:06,166 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=55915.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:18:31,839 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-28 12:18:58,946 INFO [train.py:904] (6/8) Epoch 6, batch 5200, loss[loss=0.1877, simple_loss=0.267, pruned_loss=0.05418, over 16626.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.2974, pruned_loss=0.06478, over 3207054.39 frames. ], batch size: 62, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:19:30,989 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.992e+02 2.558e+02 3.011e+02 3.644e+02 8.071e+02, threshold=6.021e+02, percent-clipped=1.0 2023-04-28 12:20:15,819 INFO [train.py:904] (6/8) Epoch 6, batch 5250, loss[loss=0.2009, simple_loss=0.2778, pruned_loss=0.06206, over 17255.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2943, pruned_loss=0.0642, over 3217446.87 frames. ], batch size: 52, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:21:15,402 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-28 12:21:28,015 INFO [train.py:904] (6/8) Epoch 6, batch 5300, loss[loss=0.1844, simple_loss=0.2605, pruned_loss=0.05414, over 16815.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2908, pruned_loss=0.06296, over 3220422.41 frames. ], batch size: 42, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:21:59,801 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.725e+02 2.658e+02 3.030e+02 3.663e+02 6.971e+02, threshold=6.060e+02, percent-clipped=3.0 2023-04-28 12:22:04,076 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6290, 3.7316, 2.8545, 2.2032, 2.6380, 2.2506, 3.7651, 3.8395], device='cuda:6'), covar=tensor([0.2116, 0.0596, 0.1326, 0.1728, 0.1818, 0.1359, 0.0474, 0.0555], device='cuda:6'), in_proj_covar=tensor([0.0277, 0.0248, 0.0269, 0.0250, 0.0280, 0.0201, 0.0247, 0.0261], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:22:40,507 INFO [train.py:904] (6/8) Epoch 6, batch 5350, loss[loss=0.1921, simple_loss=0.2783, pruned_loss=0.05299, over 16691.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2903, pruned_loss=0.0627, over 3220636.30 frames. ], batch size: 39, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:22:45,575 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56104.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:23:44,961 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56145.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:23:52,596 INFO [train.py:904] (6/8) Epoch 6, batch 5400, loss[loss=0.2274, simple_loss=0.3117, pruned_loss=0.0715, over 16503.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.2942, pruned_loss=0.06441, over 3206896.32 frames. ], batch size: 68, lr: 1.13e-02, grad_scale: 4.0 2023-04-28 12:23:53,422 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-28 12:24:07,791 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 12:24:26,634 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.771e+02 2.610e+02 3.124e+02 3.605e+02 7.209e+02, threshold=6.248e+02, percent-clipped=3.0 2023-04-28 12:25:08,990 INFO [train.py:904] (6/8) Epoch 6, batch 5450, loss[loss=0.221, simple_loss=0.3028, pruned_loss=0.06954, over 16721.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.2978, pruned_loss=0.0665, over 3219578.04 frames. ], batch size: 89, lr: 1.13e-02, grad_scale: 4.0 2023-04-28 12:25:14,956 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56204.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:25:17,632 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56206.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:26:07,980 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 12:26:18,163 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2881, 2.8961, 2.5100, 2.2144, 2.2377, 2.1021, 2.8349, 3.0130], device='cuda:6'), covar=tensor([0.1813, 0.0687, 0.1162, 0.1438, 0.1938, 0.1486, 0.0448, 0.0706], device='cuda:6'), in_proj_covar=tensor([0.0278, 0.0247, 0.0266, 0.0249, 0.0280, 0.0201, 0.0246, 0.0258], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:26:27,336 INFO [train.py:904] (6/8) Epoch 6, batch 5500, loss[loss=0.2619, simple_loss=0.3319, pruned_loss=0.09597, over 16507.00 frames. ], tot_loss[loss=0.2255, simple_loss=0.3064, pruned_loss=0.07232, over 3192587.92 frames. ], batch size: 68, lr: 1.13e-02, grad_scale: 4.0 2023-04-28 12:26:50,232 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9218, 1.6117, 2.2736, 2.9370, 2.7791, 3.2071, 1.6868, 2.9092], device='cuda:6'), covar=tensor([0.0071, 0.0226, 0.0154, 0.0097, 0.0098, 0.0052, 0.0216, 0.0054], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0148, 0.0129, 0.0131, 0.0136, 0.0097, 0.0142, 0.0088], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 12:27:01,683 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.186e+02 3.422e+02 4.497e+02 5.690e+02 1.317e+03, threshold=8.994e+02, percent-clipped=17.0 2023-04-28 12:27:46,059 INFO [train.py:904] (6/8) Epoch 6, batch 5550, loss[loss=0.302, simple_loss=0.3679, pruned_loss=0.118, over 16177.00 frames. ], tot_loss[loss=0.237, simple_loss=0.3154, pruned_loss=0.07934, over 3152256.05 frames. ], batch size: 165, lr: 1.13e-02, grad_scale: 4.0 2023-04-28 12:28:12,348 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-28 12:28:34,886 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4404, 2.5845, 1.9134, 2.2397, 2.9549, 2.6174, 3.3860, 3.1780], device='cuda:6'), covar=tensor([0.0029, 0.0188, 0.0264, 0.0227, 0.0111, 0.0197, 0.0073, 0.0095], device='cuda:6'), in_proj_covar=tensor([0.0088, 0.0161, 0.0164, 0.0162, 0.0157, 0.0166, 0.0146, 0.0145], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:29:07,880 INFO [train.py:904] (6/8) Epoch 6, batch 5600, loss[loss=0.3351, simple_loss=0.3737, pruned_loss=0.1483, over 11279.00 frames. ], tot_loss[loss=0.2476, simple_loss=0.3225, pruned_loss=0.08638, over 3101196.23 frames. ], batch size: 248, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:29:45,251 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.502e+02 4.330e+02 5.114e+02 6.903e+02 1.749e+03, threshold=1.023e+03, percent-clipped=9.0 2023-04-28 12:30:29,968 INFO [train.py:904] (6/8) Epoch 6, batch 5650, loss[loss=0.3088, simple_loss=0.3568, pruned_loss=0.1304, over 11100.00 frames. ], tot_loss[loss=0.2566, simple_loss=0.3294, pruned_loss=0.09194, over 3073687.51 frames. ], batch size: 246, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:30:34,921 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56404.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:31:09,219 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 12:31:19,268 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-28 12:31:37,655 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6013, 3.7139, 2.8259, 2.1027, 2.7068, 2.2769, 3.9511, 3.7737], device='cuda:6'), covar=tensor([0.2300, 0.0601, 0.1304, 0.1674, 0.2069, 0.1459, 0.0380, 0.0654], device='cuda:6'), in_proj_covar=tensor([0.0279, 0.0250, 0.0268, 0.0251, 0.0284, 0.0203, 0.0248, 0.0261], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:31:48,519 INFO [train.py:904] (6/8) Epoch 6, batch 5700, loss[loss=0.3249, simple_loss=0.3659, pruned_loss=0.142, over 11762.00 frames. ], tot_loss[loss=0.2612, simple_loss=0.3324, pruned_loss=0.09502, over 3049801.27 frames. ], batch size: 248, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:31:51,663 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=56452.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:32:25,511 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.692e+02 4.150e+02 5.062e+02 6.862e+02 1.724e+03, threshold=1.012e+03, percent-clipped=2.0 2023-04-28 12:33:08,898 INFO [train.py:904] (6/8) Epoch 6, batch 5750, loss[loss=0.2211, simple_loss=0.3011, pruned_loss=0.07056, over 16858.00 frames. ], tot_loss[loss=0.2629, simple_loss=0.3342, pruned_loss=0.09577, over 3051944.06 frames. ], batch size: 116, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:33:09,384 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56501.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:33:13,906 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56504.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:33:31,698 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8827, 1.9596, 2.2873, 3.2253, 2.1063, 2.4502, 2.2666, 2.0535], device='cuda:6'), covar=tensor([0.0762, 0.2307, 0.1179, 0.0440, 0.2917, 0.1354, 0.2017, 0.2219], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0336, 0.0277, 0.0311, 0.0381, 0.0348, 0.0301, 0.0396], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:33:44,549 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56523.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 12:34:30,320 INFO [train.py:904] (6/8) Epoch 6, batch 5800, loss[loss=0.2418, simple_loss=0.3238, pruned_loss=0.07987, over 16774.00 frames. ], tot_loss[loss=0.2618, simple_loss=0.3341, pruned_loss=0.09479, over 3027262.89 frames. ], batch size: 124, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:34:32,766 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=56552.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:35:05,967 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.365e+02 3.759e+02 4.783e+02 6.283e+02 1.634e+03, threshold=9.566e+02, percent-clipped=2.0 2023-04-28 12:35:23,452 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56584.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 12:35:49,148 INFO [train.py:904] (6/8) Epoch 6, batch 5850, loss[loss=0.2352, simple_loss=0.3155, pruned_loss=0.07742, over 16176.00 frames. ], tot_loss[loss=0.2572, simple_loss=0.3309, pruned_loss=0.09181, over 3037566.40 frames. ], batch size: 35, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:36:30,298 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0494, 5.0253, 4.8760, 4.6945, 4.3829, 4.9181, 4.7844, 4.5465], device='cuda:6'), covar=tensor([0.0452, 0.0276, 0.0196, 0.0196, 0.0883, 0.0257, 0.0272, 0.0501], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0212, 0.0226, 0.0200, 0.0255, 0.0229, 0.0160, 0.0256], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:37:02,989 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9051, 1.9608, 2.2484, 3.1986, 2.0107, 2.4313, 2.2630, 2.0307], device='cuda:6'), covar=tensor([0.0590, 0.2100, 0.1066, 0.0372, 0.2768, 0.1137, 0.1682, 0.2123], device='cuda:6'), in_proj_covar=tensor([0.0328, 0.0336, 0.0279, 0.0310, 0.0383, 0.0349, 0.0301, 0.0397], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:37:11,643 INFO [train.py:904] (6/8) Epoch 6, batch 5900, loss[loss=0.2573, simple_loss=0.3304, pruned_loss=0.09206, over 15393.00 frames. ], tot_loss[loss=0.2552, simple_loss=0.3294, pruned_loss=0.09051, over 3051911.58 frames. ], batch size: 191, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:37:52,137 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.523e+02 3.504e+02 4.467e+02 5.504e+02 1.096e+03, threshold=8.934e+02, percent-clipped=2.0 2023-04-28 12:38:33,939 INFO [train.py:904] (6/8) Epoch 6, batch 5950, loss[loss=0.2766, simple_loss=0.3497, pruned_loss=0.1017, over 12024.00 frames. ], tot_loss[loss=0.2536, simple_loss=0.3302, pruned_loss=0.08853, over 3077688.91 frames. ], batch size: 248, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:39:52,035 INFO [train.py:904] (6/8) Epoch 6, batch 6000, loss[loss=0.2581, simple_loss=0.3356, pruned_loss=0.09033, over 16695.00 frames. ], tot_loss[loss=0.2529, simple_loss=0.3292, pruned_loss=0.08833, over 3073551.32 frames. ], batch size: 134, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:39:52,035 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 12:40:01,521 INFO [train.py:938] (6/8) Epoch 6, validation: loss=0.18, simple_loss=0.2922, pruned_loss=0.03386, over 944034.00 frames. 2023-04-28 12:40:01,521 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-28 12:40:07,711 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8274, 3.5562, 3.4894, 2.1884, 3.1265, 3.4621, 3.3451, 1.7567], device='cuda:6'), covar=tensor([0.0367, 0.0022, 0.0032, 0.0236, 0.0059, 0.0058, 0.0038, 0.0300], device='cuda:6'), in_proj_covar=tensor([0.0117, 0.0055, 0.0059, 0.0115, 0.0061, 0.0071, 0.0065, 0.0107], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 12:40:36,510 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.218e+02 3.558e+02 4.306e+02 5.339e+02 1.394e+03, threshold=8.611e+02, percent-clipped=5.0 2023-04-28 12:41:19,697 INFO [train.py:904] (6/8) Epoch 6, batch 6050, loss[loss=0.2474, simple_loss=0.3371, pruned_loss=0.0789, over 16940.00 frames. ], tot_loss[loss=0.2502, simple_loss=0.3267, pruned_loss=0.08686, over 3077424.81 frames. ], batch size: 109, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:41:20,605 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56801.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:41:23,932 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56803.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:41:38,530 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-28 12:42:34,242 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=56849.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:42:36,667 INFO [train.py:904] (6/8) Epoch 6, batch 6100, loss[loss=0.2025, simple_loss=0.2829, pruned_loss=0.06106, over 16456.00 frames. ], tot_loss[loss=0.2485, simple_loss=0.3258, pruned_loss=0.08559, over 3080743.98 frames. ], batch size: 75, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:42:58,733 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56864.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:43:13,026 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.259e+02 3.305e+02 4.282e+02 5.206e+02 1.267e+03, threshold=8.564e+02, percent-clipped=3.0 2023-04-28 12:43:20,803 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56879.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 12:43:56,259 INFO [train.py:904] (6/8) Epoch 6, batch 6150, loss[loss=0.228, simple_loss=0.3041, pruned_loss=0.07594, over 16849.00 frames. ], tot_loss[loss=0.2455, simple_loss=0.3229, pruned_loss=0.08405, over 3100222.58 frames. ], batch size: 116, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:44:08,575 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56908.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:44:25,980 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3001, 4.0560, 4.2562, 4.4853, 4.6416, 4.1773, 4.5392, 4.5781], device='cuda:6'), covar=tensor([0.1014, 0.0781, 0.1255, 0.0513, 0.0396, 0.0780, 0.0483, 0.0424], device='cuda:6'), in_proj_covar=tensor([0.0417, 0.0500, 0.0653, 0.0522, 0.0392, 0.0385, 0.0409, 0.0432], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:44:32,206 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3016, 1.8924, 1.5420, 1.7072, 2.1537, 2.0094, 2.2381, 2.3088], device='cuda:6'), covar=tensor([0.0042, 0.0184, 0.0228, 0.0210, 0.0101, 0.0152, 0.0092, 0.0107], device='cuda:6'), in_proj_covar=tensor([0.0087, 0.0160, 0.0163, 0.0161, 0.0156, 0.0165, 0.0147, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:45:17,263 INFO [train.py:904] (6/8) Epoch 6, batch 6200, loss[loss=0.2503, simple_loss=0.3227, pruned_loss=0.08888, over 16439.00 frames. ], tot_loss[loss=0.2435, simple_loss=0.3208, pruned_loss=0.08311, over 3121303.85 frames. ], batch size: 75, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:45:46,258 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56969.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:45:53,728 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.179e+02 3.479e+02 4.387e+02 5.634e+02 1.000e+03, threshold=8.774e+02, percent-clipped=2.0 2023-04-28 12:46:00,142 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1039, 1.8451, 2.1735, 3.5517, 1.8503, 2.5284, 2.1244, 1.9582], device='cuda:6'), covar=tensor([0.0752, 0.2529, 0.1348, 0.0410, 0.3192, 0.1376, 0.2195, 0.2493], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0339, 0.0278, 0.0315, 0.0385, 0.0349, 0.0304, 0.0399], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:46:22,762 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56993.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:46:26,580 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56995.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:46:34,334 INFO [train.py:904] (6/8) Epoch 6, batch 6250, loss[loss=0.3165, simple_loss=0.3616, pruned_loss=0.1357, over 11806.00 frames. ], tot_loss[loss=0.2416, simple_loss=0.3193, pruned_loss=0.08197, over 3128115.18 frames. ], batch size: 247, lr: 1.12e-02, grad_scale: 4.0 2023-04-28 12:46:45,435 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9032, 5.1866, 4.8948, 4.8703, 4.5445, 4.5265, 4.6256, 5.2105], device='cuda:6'), covar=tensor([0.0680, 0.0598, 0.0852, 0.0553, 0.0634, 0.0699, 0.0789, 0.0621], device='cuda:6'), in_proj_covar=tensor([0.0403, 0.0528, 0.0449, 0.0348, 0.0326, 0.0342, 0.0435, 0.0379], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:47:50,596 INFO [train.py:904] (6/8) Epoch 6, batch 6300, loss[loss=0.2471, simple_loss=0.3324, pruned_loss=0.08085, over 16734.00 frames. ], tot_loss[loss=0.2417, simple_loss=0.3196, pruned_loss=0.08193, over 3117680.39 frames. ], batch size: 83, lr: 1.12e-02, grad_scale: 4.0 2023-04-28 12:47:56,650 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57054.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:47:59,787 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57056.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:48:29,802 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.167e+02 3.660e+02 4.420e+02 5.766e+02 1.393e+03, threshold=8.841e+02, percent-clipped=5.0 2023-04-28 12:48:30,954 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9270, 2.2998, 1.6856, 2.0657, 2.7804, 2.5588, 3.0366, 3.0089], device='cuda:6'), covar=tensor([0.0050, 0.0207, 0.0308, 0.0242, 0.0123, 0.0171, 0.0109, 0.0112], device='cuda:6'), in_proj_covar=tensor([0.0086, 0.0160, 0.0163, 0.0161, 0.0156, 0.0165, 0.0148, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:49:09,383 INFO [train.py:904] (6/8) Epoch 6, batch 6350, loss[loss=0.2767, simple_loss=0.344, pruned_loss=0.1048, over 15395.00 frames. ], tot_loss[loss=0.2439, simple_loss=0.3208, pruned_loss=0.0835, over 3115007.46 frames. ], batch size: 190, lr: 1.12e-02, grad_scale: 4.0 2023-04-28 12:50:26,695 INFO [train.py:904] (6/8) Epoch 6, batch 6400, loss[loss=0.2892, simple_loss=0.3471, pruned_loss=0.1156, over 15429.00 frames. ], tot_loss[loss=0.2481, simple_loss=0.3229, pruned_loss=0.08662, over 3069951.41 frames. ], batch size: 190, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:50:38,727 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57159.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:50:42,171 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5420, 3.5325, 2.7356, 2.1587, 2.5012, 2.1452, 3.5759, 3.5602], device='cuda:6'), covar=tensor([0.2225, 0.0651, 0.1298, 0.1716, 0.1828, 0.1569, 0.0461, 0.0733], device='cuda:6'), in_proj_covar=tensor([0.0280, 0.0248, 0.0267, 0.0252, 0.0282, 0.0202, 0.0247, 0.0262], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:50:49,457 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8858, 4.1617, 3.3026, 2.3625, 3.2405, 2.4906, 4.4971, 4.2650], device='cuda:6'), covar=tensor([0.2303, 0.0642, 0.1249, 0.1691, 0.1892, 0.1437, 0.0371, 0.0644], device='cuda:6'), in_proj_covar=tensor([0.0279, 0.0248, 0.0267, 0.0252, 0.0282, 0.0201, 0.0247, 0.0261], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:51:01,324 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.642e+02 4.267e+02 5.177e+02 6.433e+02 1.516e+03, threshold=1.035e+03, percent-clipped=6.0 2023-04-28 12:51:08,480 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57179.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 12:51:19,382 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57186.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:51:36,909 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57197.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:51:42,173 INFO [train.py:904] (6/8) Epoch 6, batch 6450, loss[loss=0.2246, simple_loss=0.3124, pruned_loss=0.06838, over 16870.00 frames. ], tot_loss[loss=0.2466, simple_loss=0.3226, pruned_loss=0.08525, over 3091250.53 frames. ], batch size: 116, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:51:56,532 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7313, 3.8763, 1.8544, 4.1985, 2.6819, 4.1710, 2.1907, 2.8679], device='cuda:6'), covar=tensor([0.0137, 0.0260, 0.1793, 0.0044, 0.0751, 0.0356, 0.1435, 0.0626], device='cuda:6'), in_proj_covar=tensor([0.0123, 0.0153, 0.0176, 0.0081, 0.0161, 0.0188, 0.0187, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 12:52:26,740 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57227.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 12:52:52,972 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57244.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:52:57,841 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57247.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:53:03,828 INFO [train.py:904] (6/8) Epoch 6, batch 6500, loss[loss=0.2293, simple_loss=0.3109, pruned_loss=0.07387, over 16411.00 frames. ], tot_loss[loss=0.2428, simple_loss=0.3189, pruned_loss=0.0833, over 3101590.99 frames. ], batch size: 68, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:53:15,533 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57258.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 12:53:25,365 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57264.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:53:25,498 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4950, 3.4303, 3.4128, 2.8320, 3.3037, 2.1156, 3.1307, 2.9124], device='cuda:6'), covar=tensor([0.0090, 0.0078, 0.0104, 0.0205, 0.0073, 0.1559, 0.0099, 0.0152], device='cuda:6'), in_proj_covar=tensor([0.0098, 0.0086, 0.0130, 0.0129, 0.0098, 0.0148, 0.0113, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:53:41,776 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.959e+02 3.641e+02 4.251e+02 5.070e+02 1.044e+03, threshold=8.501e+02, percent-clipped=1.0 2023-04-28 12:54:09,072 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7901, 4.1710, 4.3855, 1.7495, 4.5744, 4.6131, 3.3705, 3.3805], device='cuda:6'), covar=tensor([0.0710, 0.0100, 0.0112, 0.1241, 0.0052, 0.0063, 0.0273, 0.0376], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0091, 0.0082, 0.0138, 0.0072, 0.0081, 0.0115, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 12:54:25,246 INFO [train.py:904] (6/8) Epoch 6, batch 6550, loss[loss=0.2298, simple_loss=0.3325, pruned_loss=0.06357, over 16895.00 frames. ], tot_loss[loss=0.2457, simple_loss=0.3224, pruned_loss=0.08452, over 3098757.19 frames. ], batch size: 90, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:54:31,839 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57305.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:55:05,083 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-28 12:55:40,128 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57349.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:55:42,178 INFO [train.py:904] (6/8) Epoch 6, batch 6600, loss[loss=0.2596, simple_loss=0.339, pruned_loss=0.09015, over 16454.00 frames. ], tot_loss[loss=0.2486, simple_loss=0.3254, pruned_loss=0.08587, over 3081290.57 frames. ], batch size: 146, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:55:43,213 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57351.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:56:18,767 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.503e+02 3.780e+02 4.683e+02 5.878e+02 1.016e+03, threshold=9.365e+02, percent-clipped=5.0 2023-04-28 12:56:56,491 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57398.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:57:00,521 INFO [train.py:904] (6/8) Epoch 6, batch 6650, loss[loss=0.2299, simple_loss=0.3046, pruned_loss=0.07755, over 16199.00 frames. ], tot_loss[loss=0.2504, simple_loss=0.326, pruned_loss=0.0874, over 3063113.18 frames. ], batch size: 35, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:58:15,912 INFO [train.py:904] (6/8) Epoch 6, batch 6700, loss[loss=0.2318, simple_loss=0.32, pruned_loss=0.07179, over 16835.00 frames. ], tot_loss[loss=0.2494, simple_loss=0.3245, pruned_loss=0.08719, over 3068788.37 frames. ], batch size: 102, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 12:58:29,525 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57459.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:58:29,697 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57459.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:58:34,646 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57462.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:58:54,092 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.152e+02 3.718e+02 4.462e+02 5.305e+02 7.989e+02, threshold=8.923e+02, percent-clipped=0.0 2023-04-28 12:58:56,517 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6554, 2.6848, 2.3993, 3.6720, 2.8183, 3.8006, 1.3695, 2.9150], device='cuda:6'), covar=tensor([0.1399, 0.0592, 0.1156, 0.0114, 0.0246, 0.0359, 0.1561, 0.0744], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0147, 0.0173, 0.0096, 0.0199, 0.0194, 0.0167, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 12:59:23,708 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9227, 4.5986, 4.8613, 5.1065, 5.2072, 4.5969, 5.1791, 5.1654], device='cuda:6'), covar=tensor([0.0930, 0.0803, 0.1146, 0.0414, 0.0487, 0.0654, 0.0384, 0.0375], device='cuda:6'), in_proj_covar=tensor([0.0404, 0.0494, 0.0636, 0.0513, 0.0386, 0.0381, 0.0400, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 12:59:24,898 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57495.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:59:34,111 INFO [train.py:904] (6/8) Epoch 6, batch 6750, loss[loss=0.2363, simple_loss=0.304, pruned_loss=0.08431, over 16569.00 frames. ], tot_loss[loss=0.2479, simple_loss=0.3227, pruned_loss=0.08653, over 3084909.24 frames. ], batch size: 35, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 12:59:43,646 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57507.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:00:08,165 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57523.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:00:36,608 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57542.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:00:42,257 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6901, 3.9709, 3.1685, 2.4443, 3.0969, 2.5271, 4.2587, 4.1307], device='cuda:6'), covar=tensor([0.2183, 0.0613, 0.1180, 0.1563, 0.1846, 0.1275, 0.0358, 0.0523], device='cuda:6'), in_proj_covar=tensor([0.0281, 0.0249, 0.0266, 0.0251, 0.0282, 0.0201, 0.0247, 0.0259], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:00:49,718 INFO [train.py:904] (6/8) Epoch 6, batch 6800, loss[loss=0.2424, simple_loss=0.3367, pruned_loss=0.07408, over 16797.00 frames. ], tot_loss[loss=0.2472, simple_loss=0.3224, pruned_loss=0.08595, over 3095812.91 frames. ], batch size: 102, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:00:52,719 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2222, 5.1960, 5.0767, 4.8503, 4.5258, 5.1369, 5.0623, 4.8084], device='cuda:6'), covar=tensor([0.0501, 0.0308, 0.0184, 0.0191, 0.0880, 0.0286, 0.0173, 0.0483], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0212, 0.0223, 0.0195, 0.0252, 0.0228, 0.0162, 0.0258], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:00:54,406 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57553.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 13:00:58,711 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57556.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:01:10,578 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57564.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:01:27,504 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.272e+02 3.624e+02 4.504e+02 5.898e+02 1.165e+03, threshold=9.008e+02, percent-clipped=3.0 2023-04-28 13:01:44,236 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-28 13:02:03,574 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6574, 4.4339, 4.7261, 4.9323, 5.0654, 4.5474, 5.0453, 5.0322], device='cuda:6'), covar=tensor([0.1102, 0.0823, 0.1205, 0.0480, 0.0417, 0.0557, 0.0416, 0.0382], device='cuda:6'), in_proj_covar=tensor([0.0406, 0.0496, 0.0639, 0.0513, 0.0389, 0.0382, 0.0401, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:02:06,899 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57600.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:02:07,769 INFO [train.py:904] (6/8) Epoch 6, batch 6850, loss[loss=0.2322, simple_loss=0.3208, pruned_loss=0.07179, over 16363.00 frames. ], tot_loss[loss=0.2492, simple_loss=0.3246, pruned_loss=0.08685, over 3082372.99 frames. ], batch size: 165, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:02:24,716 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57612.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:02:40,399 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-04-28 13:03:10,636 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-28 13:03:12,929 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0182, 3.6866, 3.1995, 1.7016, 2.8611, 2.3275, 3.4217, 3.6549], device='cuda:6'), covar=tensor([0.0252, 0.0490, 0.0633, 0.1723, 0.0752, 0.0903, 0.0670, 0.0663], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0127, 0.0153, 0.0139, 0.0133, 0.0124, 0.0138, 0.0138], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 13:03:22,360 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57649.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:03:24,925 INFO [train.py:904] (6/8) Epoch 6, batch 6900, loss[loss=0.3751, simple_loss=0.403, pruned_loss=0.1736, over 11754.00 frames. ], tot_loss[loss=0.2498, simple_loss=0.3264, pruned_loss=0.08663, over 3084423.32 frames. ], batch size: 247, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:03:25,403 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57651.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:03:35,411 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1848, 3.8757, 3.8013, 4.2808, 4.3995, 4.0639, 4.4028, 4.4285], device='cuda:6'), covar=tensor([0.0940, 0.0946, 0.2090, 0.0899, 0.0771, 0.1029, 0.0714, 0.0771], device='cuda:6'), in_proj_covar=tensor([0.0409, 0.0500, 0.0642, 0.0517, 0.0393, 0.0384, 0.0406, 0.0432], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:04:02,712 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.992e+02 3.443e+02 4.296e+02 5.667e+02 1.150e+03, threshold=8.592e+02, percent-clipped=3.0 2023-04-28 13:04:37,916 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57697.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:04:41,447 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57699.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:04:43,532 INFO [train.py:904] (6/8) Epoch 6, batch 6950, loss[loss=0.2476, simple_loss=0.3237, pruned_loss=0.08574, over 17210.00 frames. ], tot_loss[loss=0.2531, simple_loss=0.3287, pruned_loss=0.08871, over 3074024.63 frames. ], batch size: 44, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:05:34,117 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57733.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:05:42,855 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1316, 2.4115, 2.4138, 4.8538, 2.2388, 3.1615, 2.4391, 2.6615], device='cuda:6'), covar=tensor([0.0595, 0.2605, 0.1444, 0.0248, 0.3300, 0.1408, 0.2360, 0.2526], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0338, 0.0280, 0.0314, 0.0387, 0.0352, 0.0305, 0.0400], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:06:01,322 INFO [train.py:904] (6/8) Epoch 6, batch 7000, loss[loss=0.2129, simple_loss=0.3085, pruned_loss=0.05868, over 17011.00 frames. ], tot_loss[loss=0.2522, simple_loss=0.3285, pruned_loss=0.08789, over 3077267.35 frames. ], batch size: 50, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:06:06,471 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57754.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:06:38,242 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.975e+02 3.231e+02 4.685e+02 6.535e+02 1.606e+03, threshold=9.370e+02, percent-clipped=8.0 2023-04-28 13:07:07,579 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57794.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:07:17,559 INFO [train.py:904] (6/8) Epoch 6, batch 7050, loss[loss=0.2183, simple_loss=0.3004, pruned_loss=0.06806, over 16568.00 frames. ], tot_loss[loss=0.2536, simple_loss=0.3295, pruned_loss=0.08879, over 3049047.46 frames. ], batch size: 62, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:07:44,475 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57818.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:08:20,868 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57842.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:08:34,169 INFO [train.py:904] (6/8) Epoch 6, batch 7100, loss[loss=0.202, simple_loss=0.2888, pruned_loss=0.05763, over 17032.00 frames. ], tot_loss[loss=0.2517, simple_loss=0.3275, pruned_loss=0.08791, over 3059873.24 frames. ], batch size: 41, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:08:35,324 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57851.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:08:37,920 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57853.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 13:09:10,524 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.050e+02 3.825e+02 4.803e+02 6.106e+02 1.425e+03, threshold=9.606e+02, percent-clipped=4.0 2023-04-28 13:09:32,897 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57890.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:09:48,868 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57900.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:09:49,600 INFO [train.py:904] (6/8) Epoch 6, batch 7150, loss[loss=0.2269, simple_loss=0.3043, pruned_loss=0.07477, over 16417.00 frames. ], tot_loss[loss=0.2496, simple_loss=0.325, pruned_loss=0.08707, over 3063157.57 frames. ], batch size: 68, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:09:49,970 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57901.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:10:01,547 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0087, 3.1002, 1.7385, 3.2807, 2.3102, 3.2400, 1.8511, 2.5102], device='cuda:6'), covar=tensor([0.0183, 0.0338, 0.1498, 0.0092, 0.0783, 0.0483, 0.1418, 0.0653], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0155, 0.0179, 0.0085, 0.0164, 0.0190, 0.0188, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 13:10:13,015 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7881, 3.8489, 3.0748, 2.4819, 3.0655, 2.5198, 4.1932, 3.9866], device='cuda:6'), covar=tensor([0.2234, 0.0749, 0.1316, 0.1578, 0.1826, 0.1349, 0.0430, 0.0636], device='cuda:6'), in_proj_covar=tensor([0.0281, 0.0250, 0.0266, 0.0250, 0.0281, 0.0201, 0.0247, 0.0260], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:10:23,453 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2837, 3.8329, 3.7998, 2.4537, 3.3907, 3.6839, 3.6298, 2.2100], device='cuda:6'), covar=tensor([0.0327, 0.0019, 0.0023, 0.0251, 0.0054, 0.0063, 0.0032, 0.0264], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0056, 0.0059, 0.0116, 0.0063, 0.0074, 0.0065, 0.0108], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 13:11:00,229 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57948.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:11:04,887 INFO [train.py:904] (6/8) Epoch 6, batch 7200, loss[loss=0.2083, simple_loss=0.2929, pruned_loss=0.06183, over 16667.00 frames. ], tot_loss[loss=0.2452, simple_loss=0.3217, pruned_loss=0.08435, over 3060929.54 frames. ], batch size: 134, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:11:41,800 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.198e+02 3.384e+02 4.240e+02 5.535e+02 1.102e+03, threshold=8.480e+02, percent-clipped=1.0 2023-04-28 13:12:28,071 INFO [train.py:904] (6/8) Epoch 6, batch 7250, loss[loss=0.1977, simple_loss=0.2767, pruned_loss=0.05933, over 16741.00 frames. ], tot_loss[loss=0.2429, simple_loss=0.3198, pruned_loss=0.08298, over 3068796.92 frames. ], batch size: 83, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:12:42,962 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6818, 2.9567, 2.5985, 4.3029, 3.1976, 3.9752, 1.7287, 2.9202], device='cuda:6'), covar=tensor([0.1471, 0.0630, 0.1149, 0.0103, 0.0350, 0.0362, 0.1505, 0.0815], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0142, 0.0169, 0.0094, 0.0195, 0.0190, 0.0163, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 13:12:59,194 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5989, 4.9149, 4.9872, 4.8662, 4.8455, 5.3755, 4.9054, 4.7118], device='cuda:6'), covar=tensor([0.0953, 0.1379, 0.1341, 0.1477, 0.2091, 0.0909, 0.1304, 0.2284], device='cuda:6'), in_proj_covar=tensor([0.0293, 0.0397, 0.0402, 0.0350, 0.0457, 0.0433, 0.0331, 0.0472], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 13:13:04,541 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9301, 2.6811, 2.0937, 2.6170, 3.0996, 2.8216, 3.8012, 3.4685], device='cuda:6'), covar=tensor([0.0021, 0.0205, 0.0286, 0.0214, 0.0127, 0.0198, 0.0077, 0.0111], device='cuda:6'), in_proj_covar=tensor([0.0086, 0.0159, 0.0164, 0.0162, 0.0155, 0.0165, 0.0145, 0.0146], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:13:44,368 INFO [train.py:904] (6/8) Epoch 6, batch 7300, loss[loss=0.2634, simple_loss=0.3363, pruned_loss=0.09528, over 15335.00 frames. ], tot_loss[loss=0.242, simple_loss=0.319, pruned_loss=0.08249, over 3072676.15 frames. ], batch size: 190, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:13:49,485 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58054.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:13:50,000 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-28 13:14:15,965 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-28 13:14:21,941 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.373e+02 3.804e+02 4.653e+02 5.753e+02 1.061e+03, threshold=9.306e+02, percent-clipped=1.0 2023-04-28 13:14:33,116 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9366, 2.6415, 2.0788, 2.3656, 3.1790, 2.7557, 3.7902, 3.4721], device='cuda:6'), covar=tensor([0.0017, 0.0208, 0.0275, 0.0239, 0.0108, 0.0189, 0.0060, 0.0089], device='cuda:6'), in_proj_covar=tensor([0.0085, 0.0158, 0.0164, 0.0161, 0.0155, 0.0165, 0.0145, 0.0145], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:14:43,446 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58089.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:15:02,128 INFO [train.py:904] (6/8) Epoch 6, batch 7350, loss[loss=0.226, simple_loss=0.3056, pruned_loss=0.07324, over 16639.00 frames. ], tot_loss[loss=0.243, simple_loss=0.3197, pruned_loss=0.08321, over 3060866.43 frames. ], batch size: 57, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:15:03,786 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=58102.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:15:18,332 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2023-04-28 13:15:28,189 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58118.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:15:38,250 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2841, 3.5066, 3.6576, 1.7234, 3.8717, 3.8822, 3.0468, 2.7283], device='cuda:6'), covar=tensor([0.0754, 0.0127, 0.0120, 0.1115, 0.0045, 0.0066, 0.0316, 0.0424], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0089, 0.0080, 0.0137, 0.0071, 0.0079, 0.0114, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 13:16:04,976 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58141.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:16:09,250 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58144.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:16:17,123 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-28 13:16:19,318 INFO [train.py:904] (6/8) Epoch 6, batch 7400, loss[loss=0.3118, simple_loss=0.3574, pruned_loss=0.1331, over 11511.00 frames. ], tot_loss[loss=0.2441, simple_loss=0.3205, pruned_loss=0.08381, over 3054653.71 frames. ], batch size: 248, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:16:19,742 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58151.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:16:43,941 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=58166.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:16:57,162 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.315e+02 3.701e+02 4.363e+02 5.153e+02 1.099e+03, threshold=8.726e+02, percent-clipped=1.0 2023-04-28 13:17:13,550 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5518, 1.9942, 2.2143, 4.1250, 1.9549, 2.8182, 2.3387, 2.2228], device='cuda:6'), covar=tensor([0.0674, 0.2606, 0.1480, 0.0333, 0.3306, 0.1420, 0.2155, 0.2579], device='cuda:6'), in_proj_covar=tensor([0.0329, 0.0342, 0.0282, 0.0318, 0.0389, 0.0357, 0.0306, 0.0402], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:17:34,942 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=58199.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:17:37,742 INFO [train.py:904] (6/8) Epoch 6, batch 7450, loss[loss=0.2504, simple_loss=0.3271, pruned_loss=0.08684, over 16853.00 frames. ], tot_loss[loss=0.246, simple_loss=0.3217, pruned_loss=0.08513, over 3058676.84 frames. ], batch size: 116, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:17:39,872 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58202.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:17:45,724 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58205.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:18:59,284 INFO [train.py:904] (6/8) Epoch 6, batch 7500, loss[loss=0.2411, simple_loss=0.3148, pruned_loss=0.0837, over 16476.00 frames. ], tot_loss[loss=0.2466, simple_loss=0.3227, pruned_loss=0.08527, over 3042163.13 frames. ], batch size: 62, lr: 1.11e-02, grad_scale: 4.0 2023-04-28 13:19:24,827 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-28 13:19:39,441 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.280e+02 3.913e+02 4.700e+02 5.850e+02 1.352e+03, threshold=9.401e+02, percent-clipped=2.0 2023-04-28 13:20:17,030 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2815, 1.8268, 2.1567, 3.7293, 1.6563, 2.5021, 2.0255, 1.9361], device='cuda:6'), covar=tensor([0.0913, 0.3343, 0.1720, 0.0507, 0.4534, 0.1892, 0.2766, 0.3259], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0341, 0.0280, 0.0316, 0.0387, 0.0353, 0.0304, 0.0401], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:20:17,498 INFO [train.py:904] (6/8) Epoch 6, batch 7550, loss[loss=0.2256, simple_loss=0.3157, pruned_loss=0.06772, over 16823.00 frames. ], tot_loss[loss=0.2458, simple_loss=0.3216, pruned_loss=0.085, over 3047966.46 frames. ], batch size: 102, lr: 1.11e-02, grad_scale: 4.0 2023-04-28 13:21:33,637 INFO [train.py:904] (6/8) Epoch 6, batch 7600, loss[loss=0.2353, simple_loss=0.3204, pruned_loss=0.07511, over 16802.00 frames. ], tot_loss[loss=0.2451, simple_loss=0.3209, pruned_loss=0.08471, over 3070928.47 frames. ], batch size: 102, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:21:56,460 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3052, 4.2976, 4.3054, 2.8492, 3.8121, 4.2331, 3.8999, 2.4318], device='cuda:6'), covar=tensor([0.0339, 0.0013, 0.0020, 0.0225, 0.0040, 0.0051, 0.0033, 0.0264], device='cuda:6'), in_proj_covar=tensor([0.0118, 0.0055, 0.0058, 0.0115, 0.0063, 0.0074, 0.0064, 0.0108], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 13:22:14,297 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.923e+02 3.777e+02 4.488e+02 5.757e+02 1.119e+03, threshold=8.975e+02, percent-clipped=2.0 2023-04-28 13:22:14,944 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7500, 1.2928, 1.6309, 1.6264, 1.7643, 1.9689, 1.4807, 1.6657], device='cuda:6'), covar=tensor([0.0114, 0.0165, 0.0097, 0.0131, 0.0098, 0.0066, 0.0165, 0.0047], device='cuda:6'), in_proj_covar=tensor([0.0124, 0.0145, 0.0126, 0.0128, 0.0132, 0.0097, 0.0143, 0.0084], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 13:22:33,778 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58389.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:22:52,233 INFO [train.py:904] (6/8) Epoch 6, batch 7650, loss[loss=0.221, simple_loss=0.3108, pruned_loss=0.06566, over 16544.00 frames. ], tot_loss[loss=0.2464, simple_loss=0.3217, pruned_loss=0.08555, over 3064601.75 frames. ], batch size: 75, lr: 1.11e-02, grad_scale: 4.0 2023-04-28 13:23:04,937 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1910, 4.2578, 4.0595, 3.9779, 3.6992, 4.1401, 4.0136, 3.8906], device='cuda:6'), covar=tensor([0.0507, 0.0286, 0.0222, 0.0198, 0.0826, 0.0339, 0.0414, 0.0468], device='cuda:6'), in_proj_covar=tensor([0.0188, 0.0210, 0.0220, 0.0191, 0.0248, 0.0227, 0.0161, 0.0254], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:23:49,830 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=58437.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:24:02,638 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6334, 4.5961, 5.1307, 5.0525, 5.0744, 4.6966, 4.7074, 4.3638], device='cuda:6'), covar=tensor([0.0214, 0.0374, 0.0252, 0.0356, 0.0343, 0.0270, 0.0707, 0.0423], device='cuda:6'), in_proj_covar=tensor([0.0249, 0.0246, 0.0251, 0.0246, 0.0297, 0.0264, 0.0364, 0.0220], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 13:24:11,082 INFO [train.py:904] (6/8) Epoch 6, batch 7700, loss[loss=0.236, simple_loss=0.3188, pruned_loss=0.07658, over 16168.00 frames. ], tot_loss[loss=0.2475, simple_loss=0.3223, pruned_loss=0.08638, over 3055652.28 frames. ], batch size: 165, lr: 1.11e-02, grad_scale: 4.0 2023-04-28 13:24:51,994 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.098e+02 3.672e+02 4.554e+02 5.723e+02 8.873e+02, threshold=9.107e+02, percent-clipped=1.0 2023-04-28 13:25:19,489 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9977, 5.3147, 5.0409, 5.0956, 4.6668, 4.5710, 4.8222, 5.3635], device='cuda:6'), covar=tensor([0.0747, 0.0740, 0.0947, 0.0472, 0.0642, 0.0733, 0.0733, 0.0703], device='cuda:6'), in_proj_covar=tensor([0.0396, 0.0516, 0.0439, 0.0335, 0.0317, 0.0342, 0.0421, 0.0374], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:25:23,657 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58497.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:25:28,762 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58500.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:25:29,687 INFO [train.py:904] (6/8) Epoch 6, batch 7750, loss[loss=0.25, simple_loss=0.3294, pruned_loss=0.08533, over 16647.00 frames. ], tot_loss[loss=0.2473, simple_loss=0.3224, pruned_loss=0.0861, over 3051087.91 frames. ], batch size: 35, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:25:45,615 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-28 13:25:50,356 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1035, 4.0933, 4.6407, 4.5374, 4.5612, 4.2332, 4.2364, 4.0476], device='cuda:6'), covar=tensor([0.0237, 0.0422, 0.0281, 0.0380, 0.0350, 0.0284, 0.0727, 0.0410], device='cuda:6'), in_proj_covar=tensor([0.0249, 0.0247, 0.0252, 0.0248, 0.0297, 0.0263, 0.0366, 0.0220], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 13:26:46,071 INFO [train.py:904] (6/8) Epoch 6, batch 7800, loss[loss=0.2484, simple_loss=0.3253, pruned_loss=0.08569, over 17013.00 frames. ], tot_loss[loss=0.2483, simple_loss=0.3231, pruned_loss=0.08671, over 3049538.88 frames. ], batch size: 41, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:27:26,211 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.190e+02 3.955e+02 4.761e+02 5.914e+02 1.211e+03, threshold=9.523e+02, percent-clipped=2.0 2023-04-28 13:27:49,912 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-28 13:28:01,703 INFO [train.py:904] (6/8) Epoch 6, batch 7850, loss[loss=0.2422, simple_loss=0.3241, pruned_loss=0.08013, over 16395.00 frames. ], tot_loss[loss=0.2469, simple_loss=0.3232, pruned_loss=0.08535, over 3062418.93 frames. ], batch size: 68, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:29:14,704 INFO [train.py:904] (6/8) Epoch 6, batch 7900, loss[loss=0.2332, simple_loss=0.3137, pruned_loss=0.07637, over 16671.00 frames. ], tot_loss[loss=0.2465, simple_loss=0.3228, pruned_loss=0.0851, over 3068366.55 frames. ], batch size: 134, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:29:53,308 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.999e+02 3.363e+02 4.237e+02 5.568e+02 1.033e+03, threshold=8.474e+02, percent-clipped=1.0 2023-04-28 13:30:02,281 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58681.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:30:31,227 INFO [train.py:904] (6/8) Epoch 6, batch 7950, loss[loss=0.233, simple_loss=0.3167, pruned_loss=0.07463, over 16868.00 frames. ], tot_loss[loss=0.2467, simple_loss=0.3228, pruned_loss=0.08533, over 3079188.55 frames. ], batch size: 96, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:31:32,260 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7607, 4.7384, 5.3203, 5.2289, 5.3074, 4.8065, 4.8923, 4.5770], device='cuda:6'), covar=tensor([0.0220, 0.0330, 0.0258, 0.0352, 0.0326, 0.0258, 0.0718, 0.0313], device='cuda:6'), in_proj_covar=tensor([0.0252, 0.0251, 0.0254, 0.0250, 0.0301, 0.0268, 0.0371, 0.0221], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 13:31:33,530 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58742.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:31:41,805 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-28 13:31:46,443 INFO [train.py:904] (6/8) Epoch 6, batch 8000, loss[loss=0.2514, simple_loss=0.326, pruned_loss=0.08836, over 16608.00 frames. ], tot_loss[loss=0.2487, simple_loss=0.3238, pruned_loss=0.08682, over 3060356.07 frames. ], batch size: 57, lr: 1.10e-02, grad_scale: 8.0 2023-04-28 13:32:25,114 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58775.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:32:27,145 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.294e+02 3.253e+02 3.971e+02 4.697e+02 8.095e+02, threshold=7.943e+02, percent-clipped=0.0 2023-04-28 13:32:58,008 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58797.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:33:02,601 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58800.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:33:03,398 INFO [train.py:904] (6/8) Epoch 6, batch 8050, loss[loss=0.2376, simple_loss=0.3109, pruned_loss=0.0822, over 16501.00 frames. ], tot_loss[loss=0.2464, simple_loss=0.3223, pruned_loss=0.08528, over 3087931.35 frames. ], batch size: 68, lr: 1.10e-02, grad_scale: 8.0 2023-04-28 13:33:51,560 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.27 vs. limit=5.0 2023-04-28 13:33:58,072 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58836.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:34:10,391 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-04-28 13:34:13,201 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=58845.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:34:16,883 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=58848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:34:20,745 INFO [train.py:904] (6/8) Epoch 6, batch 8100, loss[loss=0.263, simple_loss=0.3505, pruned_loss=0.08781, over 16671.00 frames. ], tot_loss[loss=0.2451, simple_loss=0.3216, pruned_loss=0.08429, over 3095019.92 frames. ], batch size: 83, lr: 1.10e-02, grad_scale: 2.0 2023-04-28 13:34:46,406 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0446, 3.4347, 3.4884, 1.6720, 3.7015, 3.6885, 2.6719, 2.6794], device='cuda:6'), covar=tensor([0.0877, 0.0119, 0.0151, 0.1167, 0.0049, 0.0079, 0.0387, 0.0423], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0093, 0.0085, 0.0143, 0.0073, 0.0084, 0.0119, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 13:35:03,716 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.302e+02 3.498e+02 4.445e+02 5.357e+02 1.535e+03, threshold=8.891e+02, percent-clipped=10.0 2023-04-28 13:35:38,303 INFO [train.py:904] (6/8) Epoch 6, batch 8150, loss[loss=0.2702, simple_loss=0.3243, pruned_loss=0.1081, over 11630.00 frames. ], tot_loss[loss=0.244, simple_loss=0.3202, pruned_loss=0.08393, over 3100489.36 frames. ], batch size: 248, lr: 1.10e-02, grad_scale: 2.0 2023-04-28 13:35:46,453 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58906.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:36:54,146 INFO [train.py:904] (6/8) Epoch 6, batch 8200, loss[loss=0.2685, simple_loss=0.3298, pruned_loss=0.1036, over 11797.00 frames. ], tot_loss[loss=0.2414, simple_loss=0.3171, pruned_loss=0.08289, over 3109031.90 frames. ], batch size: 247, lr: 1.10e-02, grad_scale: 2.0 2023-04-28 13:36:57,282 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0969, 5.0540, 4.8304, 4.1952, 4.8026, 1.7914, 4.7061, 4.8573], device='cuda:6'), covar=tensor([0.0049, 0.0051, 0.0097, 0.0311, 0.0064, 0.1886, 0.0082, 0.0123], device='cuda:6'), in_proj_covar=tensor([0.0098, 0.0085, 0.0132, 0.0130, 0.0100, 0.0152, 0.0114, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:37:08,185 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0257, 2.6945, 2.6499, 1.7914, 2.8464, 2.8047, 2.3924, 2.3757], device='cuda:6'), covar=tensor([0.0692, 0.0143, 0.0164, 0.0949, 0.0099, 0.0150, 0.0402, 0.0404], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0093, 0.0084, 0.0142, 0.0073, 0.0084, 0.0119, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 13:37:21,758 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58967.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:37:40,736 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.295e+02 3.597e+02 4.719e+02 5.797e+02 1.131e+03, threshold=9.438e+02, percent-clipped=6.0 2023-04-28 13:38:17,119 INFO [train.py:904] (6/8) Epoch 6, batch 8250, loss[loss=0.2251, simple_loss=0.3119, pruned_loss=0.06913, over 16264.00 frames. ], tot_loss[loss=0.2383, simple_loss=0.3159, pruned_loss=0.08041, over 3086573.71 frames. ], batch size: 165, lr: 1.10e-02, grad_scale: 2.0 2023-04-28 13:39:16,131 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59037.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:39:22,574 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-28 13:39:37,601 INFO [train.py:904] (6/8) Epoch 6, batch 8300, loss[loss=0.2193, simple_loss=0.3069, pruned_loss=0.0658, over 16706.00 frames. ], tot_loss[loss=0.2334, simple_loss=0.3126, pruned_loss=0.07712, over 3072885.17 frames. ], batch size: 76, lr: 1.10e-02, grad_scale: 2.0 2023-04-28 13:40:22,347 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.568e+02 2.786e+02 3.488e+02 4.487e+02 8.597e+02, threshold=6.977e+02, percent-clipped=0.0 2023-04-28 13:40:59,219 INFO [train.py:904] (6/8) Epoch 6, batch 8350, loss[loss=0.2023, simple_loss=0.2993, pruned_loss=0.05269, over 16451.00 frames. ], tot_loss[loss=0.2289, simple_loss=0.3104, pruned_loss=0.07373, over 3071362.20 frames. ], batch size: 75, lr: 1.10e-02, grad_scale: 2.0 2023-04-28 13:41:49,265 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59131.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:42:21,175 INFO [train.py:904] (6/8) Epoch 6, batch 8400, loss[loss=0.2402, simple_loss=0.3293, pruned_loss=0.07551, over 16389.00 frames. ], tot_loss[loss=0.2246, simple_loss=0.3071, pruned_loss=0.07103, over 3063856.90 frames. ], batch size: 146, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:42:41,260 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5179, 3.5570, 2.8206, 2.1249, 2.4393, 2.1939, 3.6407, 3.4814], device='cuda:6'), covar=tensor([0.2234, 0.0633, 0.1191, 0.1793, 0.1807, 0.1490, 0.0343, 0.0663], device='cuda:6'), in_proj_covar=tensor([0.0275, 0.0236, 0.0260, 0.0243, 0.0263, 0.0196, 0.0239, 0.0247], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:42:42,355 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59164.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:43:05,314 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.645e+02 2.850e+02 3.501e+02 4.655e+02 7.905e+02, threshold=7.001e+02, percent-clipped=2.0 2023-04-28 13:43:40,728 INFO [train.py:904] (6/8) Epoch 6, batch 8450, loss[loss=0.2004, simple_loss=0.2948, pruned_loss=0.05296, over 16609.00 frames. ], tot_loss[loss=0.2222, simple_loss=0.3058, pruned_loss=0.06928, over 3064892.47 frames. ], batch size: 89, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:44:09,940 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6470, 4.5867, 4.4300, 3.8626, 4.4150, 1.7509, 4.2718, 4.3521], device='cuda:6'), covar=tensor([0.0060, 0.0055, 0.0104, 0.0269, 0.0065, 0.1870, 0.0086, 0.0126], device='cuda:6'), in_proj_covar=tensor([0.0095, 0.0082, 0.0129, 0.0124, 0.0097, 0.0150, 0.0111, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:44:19,854 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7441, 3.7146, 3.8059, 3.7634, 3.7875, 4.2275, 3.9582, 3.6701], device='cuda:6'), covar=tensor([0.1824, 0.1849, 0.1640, 0.2134, 0.2738, 0.1391, 0.1183, 0.2534], device='cuda:6'), in_proj_covar=tensor([0.0277, 0.0381, 0.0394, 0.0333, 0.0438, 0.0418, 0.0316, 0.0450], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:44:19,954 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59225.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:45:00,570 INFO [train.py:904] (6/8) Epoch 6, batch 8500, loss[loss=0.1844, simple_loss=0.2556, pruned_loss=0.05662, over 11826.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.3011, pruned_loss=0.06606, over 3057530.32 frames. ], batch size: 247, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:45:19,659 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59262.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:45:46,051 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.033e+02 2.693e+02 3.373e+02 4.229e+02 1.183e+03, threshold=6.745e+02, percent-clipped=1.0 2023-04-28 13:45:59,212 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5543, 3.5115, 2.7622, 2.1551, 2.4275, 2.1865, 3.7555, 3.5460], device='cuda:6'), covar=tensor([0.2284, 0.0698, 0.1350, 0.1813, 0.1869, 0.1578, 0.0422, 0.0670], device='cuda:6'), in_proj_covar=tensor([0.0273, 0.0236, 0.0258, 0.0243, 0.0260, 0.0197, 0.0236, 0.0244], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:46:05,599 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-28 13:46:25,539 INFO [train.py:904] (6/8) Epoch 6, batch 8550, loss[loss=0.1871, simple_loss=0.2754, pruned_loss=0.04941, over 17196.00 frames. ], tot_loss[loss=0.2137, simple_loss=0.2982, pruned_loss=0.06461, over 3042935.63 frames. ], batch size: 45, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:46:27,113 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-28 13:47:38,715 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59337.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:48:07,962 INFO [train.py:904] (6/8) Epoch 6, batch 8600, loss[loss=0.2341, simple_loss=0.3223, pruned_loss=0.07294, over 16213.00 frames. ], tot_loss[loss=0.2129, simple_loss=0.2982, pruned_loss=0.06384, over 3040350.12 frames. ], batch size: 165, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:49:03,381 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.797e+02 2.892e+02 3.852e+02 5.028e+02 1.593e+03, threshold=7.704e+02, percent-clipped=11.0 2023-04-28 13:49:04,185 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5688, 3.5760, 3.4126, 3.2050, 3.2222, 3.5426, 3.2372, 3.3068], device='cuda:6'), covar=tensor([0.0415, 0.0300, 0.0208, 0.0171, 0.0491, 0.0286, 0.1181, 0.0388], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0205, 0.0219, 0.0188, 0.0238, 0.0222, 0.0156, 0.0247], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:49:16,114 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=59385.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:49:46,454 INFO [train.py:904] (6/8) Epoch 6, batch 8650, loss[loss=0.1856, simple_loss=0.2869, pruned_loss=0.04213, over 16673.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2963, pruned_loss=0.06184, over 3046390.49 frames. ], batch size: 89, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:50:55,817 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59431.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:51:31,166 INFO [train.py:904] (6/8) Epoch 6, batch 8700, loss[loss=0.2181, simple_loss=0.3, pruned_loss=0.06809, over 16225.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2931, pruned_loss=0.06028, over 3061255.20 frames. ], batch size: 165, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:52:21,220 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.834e+02 2.911e+02 3.776e+02 4.464e+02 8.360e+02, threshold=7.553e+02, percent-clipped=2.0 2023-04-28 13:52:22,278 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=59479.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:53:05,570 INFO [train.py:904] (6/8) Epoch 6, batch 8750, loss[loss=0.2351, simple_loss=0.3296, pruned_loss=0.07031, over 15424.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2921, pruned_loss=0.05968, over 3035514.23 frames. ], batch size: 191, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:53:08,761 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59502.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:53:54,428 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59520.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:54:51,139 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4097, 4.3723, 4.2056, 3.9971, 3.8608, 4.2873, 4.1708, 3.9702], device='cuda:6'), covar=tensor([0.0407, 0.0320, 0.0246, 0.0215, 0.0789, 0.0319, 0.0362, 0.0514], device='cuda:6'), in_proj_covar=tensor([0.0182, 0.0203, 0.0220, 0.0189, 0.0239, 0.0220, 0.0155, 0.0247], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 13:54:51,456 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-28 13:54:55,917 INFO [train.py:904] (6/8) Epoch 6, batch 8800, loss[loss=0.2191, simple_loss=0.289, pruned_loss=0.07458, over 12594.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2907, pruned_loss=0.05843, over 3053454.00 frames. ], batch size: 248, lr: 1.10e-02, grad_scale: 8.0 2023-04-28 13:55:06,706 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-28 13:55:17,692 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59562.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:55:19,976 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59563.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:55:52,198 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.676e+02 2.649e+02 3.204e+02 3.980e+02 7.645e+02, threshold=6.408e+02, percent-clipped=1.0 2023-04-28 13:56:37,974 INFO [train.py:904] (6/8) Epoch 6, batch 8850, loss[loss=0.1971, simple_loss=0.2997, pruned_loss=0.0472, over 16635.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2928, pruned_loss=0.05741, over 3054119.10 frames. ], batch size: 57, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 13:56:56,204 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=59610.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:57:11,468 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-28 13:58:21,072 INFO [train.py:904] (6/8) Epoch 6, batch 8900, loss[loss=0.1898, simple_loss=0.2782, pruned_loss=0.05076, over 16589.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2924, pruned_loss=0.05674, over 3042222.79 frames. ], batch size: 68, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 13:59:22,389 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.981e+02 2.897e+02 3.325e+02 4.292e+02 6.901e+02, threshold=6.651e+02, percent-clipped=1.0 2023-04-28 14:00:23,135 INFO [train.py:904] (6/8) Epoch 6, batch 8950, loss[loss=0.1974, simple_loss=0.2948, pruned_loss=0.05, over 16310.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2921, pruned_loss=0.05669, over 3070345.46 frames. ], batch size: 166, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:01:47,917 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-28 14:02:12,299 INFO [train.py:904] (6/8) Epoch 6, batch 9000, loss[loss=0.1906, simple_loss=0.2765, pruned_loss=0.05235, over 16803.00 frames. ], tot_loss[loss=0.1987, simple_loss=0.2882, pruned_loss=0.05462, over 3072252.88 frames. ], batch size: 124, lr: 1.09e-02, grad_scale: 4.0 2023-04-28 14:02:12,299 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 14:02:22,190 INFO [train.py:938] (6/8) Epoch 6, validation: loss=0.1682, simple_loss=0.2716, pruned_loss=0.03235, over 944034.00 frames. 2023-04-28 14:02:22,191 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-28 14:02:48,438 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2613, 3.3272, 2.4806, 2.0231, 2.2868, 1.8878, 3.3218, 3.2717], device='cuda:6'), covar=tensor([0.2633, 0.0834, 0.1472, 0.1883, 0.1879, 0.1975, 0.0595, 0.0826], device='cuda:6'), in_proj_covar=tensor([0.0276, 0.0239, 0.0260, 0.0245, 0.0248, 0.0199, 0.0238, 0.0244], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:03:21,582 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.989e+02 2.787e+02 3.475e+02 4.351e+02 1.064e+03, threshold=6.950e+02, percent-clipped=4.0 2023-04-28 14:04:06,172 INFO [train.py:904] (6/8) Epoch 6, batch 9050, loss[loss=0.1847, simple_loss=0.2714, pruned_loss=0.04904, over 16942.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2895, pruned_loss=0.05535, over 3087113.46 frames. ], batch size: 102, lr: 1.09e-02, grad_scale: 4.0 2023-04-28 14:04:44,598 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59820.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:05:45,886 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:05:50,582 INFO [train.py:904] (6/8) Epoch 6, batch 9100, loss[loss=0.1981, simple_loss=0.2779, pruned_loss=0.05916, over 12453.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2891, pruned_loss=0.0559, over 3083851.50 frames. ], batch size: 247, lr: 1.09e-02, grad_scale: 4.0 2023-04-28 14:06:04,382 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59858.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:06:24,658 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=59868.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:06:55,730 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 2.998e+02 3.675e+02 4.654e+02 1.190e+03, threshold=7.350e+02, percent-clipped=4.0 2023-04-28 14:07:19,838 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59889.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:07:47,009 INFO [train.py:904] (6/8) Epoch 6, batch 9150, loss[loss=0.1927, simple_loss=0.2737, pruned_loss=0.05583, over 12166.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2898, pruned_loss=0.05597, over 3076459.15 frames. ], batch size: 247, lr: 1.09e-02, grad_scale: 4.0 2023-04-28 14:08:06,847 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59909.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:08:19,407 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-28 14:08:32,099 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-28 14:09:28,437 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7425, 3.9906, 1.8464, 4.2859, 2.4542, 4.1150, 2.0831, 2.8759], device='cuda:6'), covar=tensor([0.0167, 0.0189, 0.1735, 0.0053, 0.0850, 0.0337, 0.1558, 0.0614], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0145, 0.0174, 0.0080, 0.0154, 0.0177, 0.0186, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-28 14:09:30,391 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59950.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:09:31,017 INFO [train.py:904] (6/8) Epoch 6, batch 9200, loss[loss=0.2138, simple_loss=0.2983, pruned_loss=0.06465, over 15326.00 frames. ], tot_loss[loss=0.1969, simple_loss=0.2848, pruned_loss=0.05447, over 3090179.65 frames. ], batch size: 190, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:10:12,298 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6391, 3.0766, 2.4676, 4.5098, 3.4062, 4.4140, 1.5673, 2.7368], device='cuda:6'), covar=tensor([0.1494, 0.0636, 0.1212, 0.0080, 0.0207, 0.0293, 0.1512, 0.0960], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0141, 0.0165, 0.0092, 0.0166, 0.0185, 0.0163, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 14:10:22,291 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.923e+02 2.993e+02 3.687e+02 4.643e+02 6.569e+02, threshold=7.374e+02, percent-clipped=0.0 2023-04-28 14:11:09,068 INFO [train.py:904] (6/8) Epoch 6, batch 9250, loss[loss=0.1862, simple_loss=0.2757, pruned_loss=0.04837, over 16402.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2843, pruned_loss=0.05464, over 3071943.45 frames. ], batch size: 146, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:12:03,060 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5532, 3.4616, 3.4584, 3.0370, 3.3933, 2.0034, 3.2500, 2.9764], device='cuda:6'), covar=tensor([0.0078, 0.0072, 0.0101, 0.0172, 0.0060, 0.1665, 0.0091, 0.0136], device='cuda:6'), in_proj_covar=tensor([0.0093, 0.0078, 0.0124, 0.0114, 0.0094, 0.0147, 0.0108, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:12:33,985 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60040.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:12:55,394 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5847, 3.5066, 3.4752, 3.0384, 3.4085, 2.0322, 3.2677, 2.9824], device='cuda:6'), covar=tensor([0.0095, 0.0093, 0.0102, 0.0199, 0.0071, 0.1714, 0.0098, 0.0147], device='cuda:6'), in_proj_covar=tensor([0.0093, 0.0078, 0.0123, 0.0114, 0.0094, 0.0147, 0.0108, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:12:58,476 INFO [train.py:904] (6/8) Epoch 6, batch 9300, loss[loss=0.1907, simple_loss=0.2739, pruned_loss=0.05375, over 15421.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.2825, pruned_loss=0.05392, over 3050244.12 frames. ], batch size: 193, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:13:01,527 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60052.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:13:33,216 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60065.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:14:01,609 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60078.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:14:05,093 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.815e+02 2.894e+02 3.605e+02 4.509e+02 1.051e+03, threshold=7.210e+02, percent-clipped=5.0 2023-04-28 14:14:25,659 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3353, 2.9545, 2.6204, 2.1946, 2.1240, 2.1289, 2.9604, 2.9205], device='cuda:6'), covar=tensor([0.1901, 0.0659, 0.1076, 0.1615, 0.1770, 0.1561, 0.0407, 0.0819], device='cuda:6'), in_proj_covar=tensor([0.0273, 0.0239, 0.0261, 0.0245, 0.0244, 0.0200, 0.0238, 0.0244], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:14:44,853 INFO [train.py:904] (6/8) Epoch 6, batch 9350, loss[loss=0.1989, simple_loss=0.2852, pruned_loss=0.0563, over 16597.00 frames. ], tot_loss[loss=0.1962, simple_loss=0.2832, pruned_loss=0.05458, over 3040622.35 frames. ], batch size: 62, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:14:46,369 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60101.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 14:15:13,261 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60113.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 14:15:16,824 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-28 14:15:37,187 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60126.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:16:03,936 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60139.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:16:22,779 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6942, 3.9313, 1.5189, 4.1623, 2.6489, 3.9978, 1.5933, 2.7796], device='cuda:6'), covar=tensor([0.0132, 0.0174, 0.1810, 0.0050, 0.0665, 0.0313, 0.1923, 0.0606], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0145, 0.0173, 0.0082, 0.0153, 0.0176, 0.0185, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-28 14:16:25,945 INFO [train.py:904] (6/8) Epoch 6, batch 9400, loss[loss=0.1612, simple_loss=0.249, pruned_loss=0.0367, over 12374.00 frames. ], tot_loss[loss=0.1962, simple_loss=0.2841, pruned_loss=0.05421, over 3050835.86 frames. ], batch size: 248, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:16:41,557 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60158.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:17:17,606 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.45 vs. limit=5.0 2023-04-28 14:17:25,097 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.674e+02 3.031e+02 3.474e+02 4.387e+02 1.018e+03, threshold=6.948e+02, percent-clipped=2.0 2023-04-28 14:17:33,117 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4717, 3.5546, 3.3438, 3.1188, 3.1475, 3.4305, 3.2503, 3.2293], device='cuda:6'), covar=tensor([0.0495, 0.0320, 0.0205, 0.0187, 0.0576, 0.0272, 0.0938, 0.0339], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0200, 0.0216, 0.0187, 0.0237, 0.0218, 0.0152, 0.0240], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:18:05,539 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60198.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:18:09,619 INFO [train.py:904] (6/8) Epoch 6, batch 9450, loss[loss=0.2101, simple_loss=0.2883, pruned_loss=0.06592, over 12509.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2864, pruned_loss=0.05468, over 3056768.45 frames. ], batch size: 248, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:18:15,332 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60204.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:18:15,419 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3601, 4.2758, 4.8794, 4.8331, 4.8152, 4.4189, 4.4781, 4.2554], device='cuda:6'), covar=tensor([0.0224, 0.0430, 0.0304, 0.0342, 0.0347, 0.0274, 0.0748, 0.0316], device='cuda:6'), in_proj_covar=tensor([0.0235, 0.0231, 0.0241, 0.0232, 0.0276, 0.0254, 0.0341, 0.0206], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-28 14:18:19,857 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60206.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:18:40,004 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9385, 3.8232, 4.3982, 4.3468, 4.3138, 3.9823, 4.0522, 3.9483], device='cuda:6'), covar=tensor([0.0252, 0.0552, 0.0317, 0.0423, 0.0461, 0.0339, 0.0723, 0.0359], device='cuda:6'), in_proj_covar=tensor([0.0235, 0.0230, 0.0240, 0.0232, 0.0276, 0.0253, 0.0340, 0.0206], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-28 14:19:39,941 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60245.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:19:50,583 INFO [train.py:904] (6/8) Epoch 6, batch 9500, loss[loss=0.1845, simple_loss=0.274, pruned_loss=0.04745, over 16116.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2851, pruned_loss=0.05389, over 3060350.27 frames. ], batch size: 165, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:20:11,515 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60259.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:20:51,248 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.861e+02 2.773e+02 3.351e+02 4.137e+02 1.111e+03, threshold=6.703e+02, percent-clipped=2.0 2023-04-28 14:21:39,640 INFO [train.py:904] (6/8) Epoch 6, batch 9550, loss[loss=0.1868, simple_loss=0.2687, pruned_loss=0.05249, over 12335.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2843, pruned_loss=0.05377, over 3067116.41 frames. ], batch size: 247, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:22:13,986 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7272, 3.1443, 2.6729, 4.4219, 3.5921, 4.2593, 1.4928, 3.1903], device='cuda:6'), covar=tensor([0.1438, 0.0567, 0.1065, 0.0111, 0.0207, 0.0337, 0.1525, 0.0667], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0143, 0.0167, 0.0093, 0.0165, 0.0186, 0.0164, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 14:22:32,840 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7624, 5.0446, 4.8137, 4.8323, 4.5010, 4.4054, 4.6099, 5.0843], device='cuda:6'), covar=tensor([0.0769, 0.0771, 0.0824, 0.0471, 0.0628, 0.0808, 0.0673, 0.0719], device='cuda:6'), in_proj_covar=tensor([0.0388, 0.0505, 0.0415, 0.0330, 0.0310, 0.0335, 0.0415, 0.0368], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:23:15,604 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9883, 1.9528, 2.3221, 3.2778, 2.0526, 2.3695, 2.2416, 2.0016], device='cuda:6'), covar=tensor([0.0640, 0.2525, 0.1261, 0.0370, 0.3217, 0.1569, 0.2095, 0.2776], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0321, 0.0273, 0.0298, 0.0371, 0.0336, 0.0294, 0.0375], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:23:21,495 INFO [train.py:904] (6/8) Epoch 6, batch 9600, loss[loss=0.2024, simple_loss=0.2776, pruned_loss=0.06359, over 12621.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2856, pruned_loss=0.05485, over 3057816.90 frames. ], batch size: 248, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:23:35,932 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9233, 3.9825, 3.7978, 3.6086, 3.4932, 3.8733, 3.5926, 3.6068], device='cuda:6'), covar=tensor([0.0475, 0.0380, 0.0254, 0.0206, 0.0762, 0.0367, 0.0647, 0.0513], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0198, 0.0212, 0.0184, 0.0235, 0.0216, 0.0149, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:23:46,788 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0702, 3.4974, 3.4524, 1.5407, 3.5956, 3.6727, 2.8736, 2.7181], device='cuda:6'), covar=tensor([0.0881, 0.0125, 0.0180, 0.1306, 0.0087, 0.0085, 0.0358, 0.0430], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0089, 0.0078, 0.0138, 0.0067, 0.0080, 0.0113, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 14:24:17,215 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.609e+02 2.768e+02 3.406e+02 4.197e+02 1.161e+03, threshold=6.812e+02, percent-clipped=3.0 2023-04-28 14:24:57,326 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5310, 4.5950, 4.6203, 4.6347, 4.6292, 5.1693, 4.7641, 4.4929], device='cuda:6'), covar=tensor([0.0883, 0.1756, 0.1559, 0.1807, 0.2461, 0.1018, 0.1161, 0.2363], device='cuda:6'), in_proj_covar=tensor([0.0266, 0.0379, 0.0383, 0.0326, 0.0433, 0.0416, 0.0315, 0.0438], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:24:57,336 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60396.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 14:25:09,699 INFO [train.py:904] (6/8) Epoch 6, batch 9650, loss[loss=0.2043, simple_loss=0.2957, pruned_loss=0.0564, over 16982.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2878, pruned_loss=0.0556, over 3057906.95 frames. ], batch size: 116, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:25:28,131 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60408.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 14:25:57,580 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60421.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:26:22,171 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60434.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:26:33,544 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7076, 3.8826, 3.0049, 2.3830, 2.8052, 2.3937, 4.0870, 3.7344], device='cuda:6'), covar=tensor([0.2061, 0.0597, 0.1224, 0.1752, 0.1743, 0.1358, 0.0318, 0.0571], device='cuda:6'), in_proj_covar=tensor([0.0276, 0.0241, 0.0262, 0.0246, 0.0240, 0.0200, 0.0239, 0.0242], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:26:58,005 INFO [train.py:904] (6/8) Epoch 6, batch 9700, loss[loss=0.1734, simple_loss=0.2658, pruned_loss=0.04053, over 17028.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.287, pruned_loss=0.05526, over 3067266.94 frames. ], batch size: 55, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:27:59,429 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.972e+02 2.947e+02 3.794e+02 4.727e+02 9.838e+02, threshold=7.588e+02, percent-clipped=5.0 2023-04-28 14:28:17,497 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7815, 4.5295, 4.7279, 4.9780, 5.1268, 4.5936, 5.0762, 5.0016], device='cuda:6'), covar=tensor([0.0952, 0.0877, 0.1164, 0.0463, 0.0366, 0.0542, 0.0375, 0.0469], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0479, 0.0595, 0.0481, 0.0367, 0.0363, 0.0383, 0.0415], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:28:41,228 INFO [train.py:904] (6/8) Epoch 6, batch 9750, loss[loss=0.2001, simple_loss=0.2922, pruned_loss=0.05403, over 15340.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2859, pruned_loss=0.05549, over 3051237.02 frames. ], batch size: 191, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:28:47,824 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60504.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:30:01,982 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9489, 4.6995, 4.9091, 5.1304, 5.3009, 4.6823, 5.2226, 5.2202], device='cuda:6'), covar=tensor([0.0864, 0.0829, 0.1092, 0.0416, 0.0334, 0.0546, 0.0385, 0.0366], device='cuda:6'), in_proj_covar=tensor([0.0388, 0.0476, 0.0590, 0.0480, 0.0365, 0.0360, 0.0379, 0.0410], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:30:11,649 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60545.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:30:19,564 INFO [train.py:904] (6/8) Epoch 6, batch 9800, loss[loss=0.2109, simple_loss=0.3067, pruned_loss=0.05761, over 16450.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2855, pruned_loss=0.05395, over 3062870.17 frames. ], batch size: 146, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:30:22,402 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60552.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:30:27,282 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60554.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:31:14,367 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.779e+02 2.802e+02 3.388e+02 4.163e+02 9.547e+02, threshold=6.775e+02, percent-clipped=1.0 2023-04-28 14:31:48,142 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60593.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:32:03,896 INFO [train.py:904] (6/8) Epoch 6, batch 9850, loss[loss=0.2258, simple_loss=0.3089, pruned_loss=0.07136, over 16252.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2861, pruned_loss=0.05357, over 3058356.61 frames. ], batch size: 165, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:32:46,916 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-28 14:33:04,503 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7128, 3.6348, 3.8793, 3.7518, 3.8601, 4.2167, 3.9062, 3.6413], device='cuda:6'), covar=tensor([0.1993, 0.1748, 0.1392, 0.2008, 0.2265, 0.1373, 0.1198, 0.2363], device='cuda:6'), in_proj_covar=tensor([0.0265, 0.0377, 0.0383, 0.0324, 0.0433, 0.0411, 0.0315, 0.0436], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:33:56,849 INFO [train.py:904] (6/8) Epoch 6, batch 9900, loss[loss=0.1823, simple_loss=0.2818, pruned_loss=0.04143, over 16759.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2863, pruned_loss=0.05351, over 3043110.41 frames. ], batch size: 83, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:35:04,552 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 2.757e+02 3.371e+02 4.013e+02 6.843e+02, threshold=6.741e+02, percent-clipped=1.0 2023-04-28 14:35:16,998 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.63 vs. limit=5.0 2023-04-28 14:35:42,463 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60696.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:35:52,854 INFO [train.py:904] (6/8) Epoch 6, batch 9950, loss[loss=0.1687, simple_loss=0.2714, pruned_loss=0.03305, over 16719.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2881, pruned_loss=0.0538, over 3038548.80 frames. ], batch size: 83, lr: 1.08e-02, grad_scale: 8.0 2023-04-28 14:36:10,159 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60708.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:36:34,896 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60718.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:36:43,501 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60721.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:37:13,871 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60734.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:37:38,498 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60744.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:37:54,649 INFO [train.py:904] (6/8) Epoch 6, batch 10000, loss[loss=0.2123, simple_loss=0.3086, pruned_loss=0.05803, over 16352.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2861, pruned_loss=0.05301, over 3056926.17 frames. ], batch size: 146, lr: 1.08e-02, grad_scale: 8.0 2023-04-28 14:38:06,642 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60756.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:38:29,851 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60769.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:38:50,609 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60779.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:38:51,224 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 2.599e+02 3.122e+02 4.143e+02 8.746e+02, threshold=6.243e+02, percent-clipped=4.0 2023-04-28 14:38:57,417 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60782.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:39:00,261 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8696, 1.8750, 2.2206, 3.1093, 1.9805, 2.2075, 2.1131, 1.8237], device='cuda:6'), covar=tensor([0.0692, 0.3086, 0.1338, 0.0480, 0.3512, 0.1940, 0.2595, 0.3184], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0324, 0.0278, 0.0304, 0.0373, 0.0340, 0.0299, 0.0378], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:39:33,417 INFO [train.py:904] (6/8) Epoch 6, batch 10050, loss[loss=0.2011, simple_loss=0.2961, pruned_loss=0.05311, over 16933.00 frames. ], tot_loss[loss=0.1963, simple_loss=0.2864, pruned_loss=0.0531, over 3061536.83 frames. ], batch size: 109, lr: 1.08e-02, grad_scale: 8.0 2023-04-28 14:39:48,671 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6917, 3.7113, 4.0998, 4.0724, 4.0831, 3.7910, 3.8141, 3.7956], device='cuda:6'), covar=tensor([0.0252, 0.0398, 0.0339, 0.0389, 0.0315, 0.0293, 0.0635, 0.0343], device='cuda:6'), in_proj_covar=tensor([0.0244, 0.0239, 0.0245, 0.0236, 0.0278, 0.0260, 0.0347, 0.0214], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-28 14:40:19,346 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1357, 3.2605, 3.2268, 1.5854, 3.4198, 3.3957, 2.8140, 2.6908], device='cuda:6'), covar=tensor([0.0741, 0.0105, 0.0146, 0.1170, 0.0059, 0.0091, 0.0298, 0.0354], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0087, 0.0075, 0.0134, 0.0064, 0.0079, 0.0109, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 14:41:05,272 INFO [train.py:904] (6/8) Epoch 6, batch 10100, loss[loss=0.1902, simple_loss=0.2735, pruned_loss=0.05346, over 16892.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2871, pruned_loss=0.05344, over 3074412.13 frames. ], batch size: 116, lr: 1.08e-02, grad_scale: 8.0 2023-04-28 14:41:10,988 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60854.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:42:01,261 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.991e+02 2.889e+02 3.366e+02 3.931e+02 7.986e+02, threshold=6.733e+02, percent-clipped=7.0 2023-04-28 14:42:05,762 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0075, 3.1433, 1.6761, 3.3339, 2.1765, 3.2935, 1.9725, 2.5044], device='cuda:6'), covar=tensor([0.0251, 0.0320, 0.1651, 0.0097, 0.0922, 0.0438, 0.1494, 0.0682], device='cuda:6'), in_proj_covar=tensor([0.0123, 0.0148, 0.0178, 0.0084, 0.0157, 0.0179, 0.0188, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 14:42:22,731 INFO [train.py:904] (6/8) Epoch 6, batch 10150, loss[loss=0.2088, simple_loss=0.2823, pruned_loss=0.06765, over 12309.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2856, pruned_loss=0.05363, over 3045708.04 frames. ], batch size: 248, lr: 1.08e-02, grad_scale: 8.0 2023-04-28 14:42:48,314 INFO [train.py:904] (6/8) Epoch 7, batch 0, loss[loss=0.3132, simple_loss=0.3415, pruned_loss=0.1424, over 16803.00 frames. ], tot_loss[loss=0.3132, simple_loss=0.3415, pruned_loss=0.1424, over 16803.00 frames. ], batch size: 102, lr: 1.02e-02, grad_scale: 8.0 2023-04-28 14:42:48,314 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 14:42:55,779 INFO [train.py:938] (6/8) Epoch 7, validation: loss=0.1665, simple_loss=0.2702, pruned_loss=0.03141, over 944034.00 frames. 2023-04-28 14:42:55,779 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-28 14:42:56,632 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60902.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:43:38,144 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5193, 4.0159, 4.1421, 2.1084, 3.4050, 2.6573, 3.8759, 3.8711], device='cuda:6'), covar=tensor([0.0262, 0.0568, 0.0368, 0.1476, 0.0574, 0.0811, 0.0613, 0.0932], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0119, 0.0150, 0.0137, 0.0130, 0.0122, 0.0134, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-28 14:43:43,543 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6491, 2.5411, 2.1462, 2.3452, 3.0534, 2.8506, 3.4883, 3.2685], device='cuda:6'), covar=tensor([0.0028, 0.0213, 0.0267, 0.0253, 0.0128, 0.0190, 0.0103, 0.0119], device='cuda:6'), in_proj_covar=tensor([0.0083, 0.0165, 0.0164, 0.0162, 0.0159, 0.0165, 0.0142, 0.0145], device='cuda:6'), out_proj_covar=tensor([9.4271e-05, 1.9052e-04, 1.8500e-04, 1.8335e-04, 1.8436e-04, 1.8964e-04, 1.5666e-04, 1.6599e-04], device='cuda:6') 2023-04-28 14:44:05,495 INFO [train.py:904] (6/8) Epoch 7, batch 50, loss[loss=0.1909, simple_loss=0.2763, pruned_loss=0.05275, over 17215.00 frames. ], tot_loss[loss=0.2412, simple_loss=0.313, pruned_loss=0.08471, over 740896.58 frames. ], batch size: 45, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:44:29,616 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4404, 4.5417, 4.5763, 4.6175, 4.5221, 5.1398, 4.7537, 4.4812], device='cuda:6'), covar=tensor([0.1173, 0.1764, 0.1797, 0.1890, 0.3010, 0.1178, 0.1432, 0.2468], device='cuda:6'), in_proj_covar=tensor([0.0286, 0.0402, 0.0407, 0.0344, 0.0458, 0.0439, 0.0335, 0.0465], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 14:44:49,443 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.919e+02 3.058e+02 3.620e+02 4.703e+02 1.122e+03, threshold=7.241e+02, percent-clipped=6.0 2023-04-28 14:45:08,331 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9194, 4.2205, 1.9833, 4.6229, 2.6852, 4.5690, 2.2577, 3.2008], device='cuda:6'), covar=tensor([0.0187, 0.0276, 0.1819, 0.0096, 0.0931, 0.0364, 0.1531, 0.0666], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0153, 0.0182, 0.0088, 0.0159, 0.0187, 0.0191, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 14:45:15,337 INFO [train.py:904] (6/8) Epoch 7, batch 100, loss[loss=0.1789, simple_loss=0.2639, pruned_loss=0.04693, over 17183.00 frames. ], tot_loss[loss=0.2289, simple_loss=0.3023, pruned_loss=0.07774, over 1313244.68 frames. ], batch size: 44, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:46:02,533 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-28 14:46:12,685 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9667, 1.5843, 2.2583, 2.8805, 2.6902, 3.2900, 1.6032, 3.1635], device='cuda:6'), covar=tensor([0.0083, 0.0262, 0.0155, 0.0111, 0.0108, 0.0070, 0.0296, 0.0062], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0150, 0.0133, 0.0132, 0.0137, 0.0097, 0.0145, 0.0084], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 14:46:24,663 INFO [train.py:904] (6/8) Epoch 7, batch 150, loss[loss=0.1914, simple_loss=0.2845, pruned_loss=0.04911, over 17054.00 frames. ], tot_loss[loss=0.2231, simple_loss=0.2988, pruned_loss=0.07372, over 1748522.99 frames. ], batch size: 50, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:46:37,938 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61061.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:46:56,479 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=61074.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:47:07,163 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.794e+02 2.994e+02 3.585e+02 4.192e+02 7.211e+02, threshold=7.170e+02, percent-clipped=0.0 2023-04-28 14:47:11,387 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6412, 6.0005, 5.6655, 5.7923, 5.3025, 5.0126, 5.4746, 6.0860], device='cuda:6'), covar=tensor([0.0732, 0.0625, 0.0958, 0.0491, 0.0621, 0.0564, 0.0729, 0.0665], device='cuda:6'), in_proj_covar=tensor([0.0414, 0.0544, 0.0453, 0.0358, 0.0337, 0.0355, 0.0448, 0.0395], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:47:34,196 INFO [train.py:904] (6/8) Epoch 7, batch 200, loss[loss=0.2381, simple_loss=0.3036, pruned_loss=0.08625, over 16709.00 frames. ], tot_loss[loss=0.2216, simple_loss=0.2977, pruned_loss=0.07278, over 2100715.98 frames. ], batch size: 134, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:48:01,648 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61122.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:48:03,708 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-28 14:48:44,069 INFO [train.py:904] (6/8) Epoch 7, batch 250, loss[loss=0.2177, simple_loss=0.2831, pruned_loss=0.07617, over 16826.00 frames. ], tot_loss[loss=0.2195, simple_loss=0.295, pruned_loss=0.07197, over 2366169.75 frames. ], batch size: 102, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:48:58,576 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2023-04-28 14:49:24,097 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2669, 4.9344, 5.2350, 5.4343, 5.6350, 4.9427, 5.5104, 5.5408], device='cuda:6'), covar=tensor([0.1112, 0.0874, 0.1203, 0.0470, 0.0400, 0.0530, 0.0427, 0.0417], device='cuda:6'), in_proj_covar=tensor([0.0430, 0.0525, 0.0659, 0.0525, 0.0399, 0.0396, 0.0420, 0.0451], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:49:24,840 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.754e+02 3.071e+02 3.589e+02 4.451e+02 1.172e+03, threshold=7.179e+02, percent-clipped=5.0 2023-04-28 14:49:52,014 INFO [train.py:904] (6/8) Epoch 7, batch 300, loss[loss=0.2156, simple_loss=0.2734, pruned_loss=0.07886, over 16716.00 frames. ], tot_loss[loss=0.2147, simple_loss=0.2908, pruned_loss=0.06934, over 2573501.77 frames. ], batch size: 83, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:50:45,077 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1440, 5.0884, 4.8672, 4.3243, 4.8452, 1.9497, 4.7462, 5.0751], device='cuda:6'), covar=tensor([0.0060, 0.0047, 0.0110, 0.0312, 0.0076, 0.1952, 0.0094, 0.0113], device='cuda:6'), in_proj_covar=tensor([0.0100, 0.0086, 0.0135, 0.0124, 0.0101, 0.0156, 0.0118, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:50:59,330 INFO [train.py:904] (6/8) Epoch 7, batch 350, loss[loss=0.1845, simple_loss=0.273, pruned_loss=0.04801, over 17268.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2866, pruned_loss=0.06739, over 2736488.03 frames. ], batch size: 52, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:51:41,729 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.762e+02 2.825e+02 3.324e+02 4.016e+02 8.512e+02, threshold=6.649e+02, percent-clipped=2.0 2023-04-28 14:52:08,570 INFO [train.py:904] (6/8) Epoch 7, batch 400, loss[loss=0.2118, simple_loss=0.2751, pruned_loss=0.07427, over 16494.00 frames. ], tot_loss[loss=0.2102, simple_loss=0.2854, pruned_loss=0.0675, over 2865300.45 frames. ], batch size: 75, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:52:45,130 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0854, 4.7212, 4.9863, 5.2345, 5.4764, 4.7378, 5.4021, 5.3972], device='cuda:6'), covar=tensor([0.1081, 0.0959, 0.1477, 0.0595, 0.0433, 0.0683, 0.0450, 0.0412], device='cuda:6'), in_proj_covar=tensor([0.0441, 0.0538, 0.0680, 0.0536, 0.0412, 0.0405, 0.0430, 0.0465], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:53:17,203 INFO [train.py:904] (6/8) Epoch 7, batch 450, loss[loss=0.2317, simple_loss=0.3071, pruned_loss=0.07814, over 16803.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2827, pruned_loss=0.06559, over 2965591.12 frames. ], batch size: 102, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:53:48,648 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=61374.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:53:55,665 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9505, 4.9228, 4.7335, 4.5367, 4.2898, 4.8271, 4.7762, 4.4273], device='cuda:6'), covar=tensor([0.0545, 0.0375, 0.0256, 0.0229, 0.0977, 0.0383, 0.0354, 0.0591], device='cuda:6'), in_proj_covar=tensor([0.0208, 0.0233, 0.0244, 0.0216, 0.0280, 0.0250, 0.0173, 0.0281], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 14:54:00,435 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.039e+02 2.768e+02 3.236e+02 4.177e+02 7.606e+02, threshold=6.472e+02, percent-clipped=3.0 2023-04-28 14:54:10,871 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-04-28 14:54:27,747 INFO [train.py:904] (6/8) Epoch 7, batch 500, loss[loss=0.2021, simple_loss=0.28, pruned_loss=0.06214, over 16539.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2817, pruned_loss=0.06464, over 3030053.64 frames. ], batch size: 68, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:54:47,746 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=61417.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:54:54,990 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=61422.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:55:35,821 INFO [train.py:904] (6/8) Epoch 7, batch 550, loss[loss=0.2562, simple_loss=0.3135, pruned_loss=0.09945, over 16405.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2811, pruned_loss=0.06439, over 3091685.42 frames. ], batch size: 146, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:56:09,102 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6651, 4.6695, 4.6707, 4.8065, 4.6932, 5.3126, 4.9317, 4.6836], device='cuda:6'), covar=tensor([0.1097, 0.1614, 0.1677, 0.1723, 0.2559, 0.0930, 0.1202, 0.2357], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0445, 0.0442, 0.0378, 0.0507, 0.0474, 0.0360, 0.0507], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 14:56:16,803 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8809, 4.3821, 3.5081, 2.4224, 3.1188, 2.4794, 4.6070, 4.2344], device='cuda:6'), covar=tensor([0.2321, 0.0601, 0.1119, 0.1831, 0.2273, 0.1548, 0.0284, 0.0754], device='cuda:6'), in_proj_covar=tensor([0.0285, 0.0253, 0.0273, 0.0255, 0.0271, 0.0210, 0.0250, 0.0271], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:56:17,397 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.772e+02 2.888e+02 3.437e+02 4.263e+02 9.933e+02, threshold=6.875e+02, percent-clipped=6.0 2023-04-28 14:56:43,945 INFO [train.py:904] (6/8) Epoch 7, batch 600, loss[loss=0.1993, simple_loss=0.2684, pruned_loss=0.06505, over 16571.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.28, pruned_loss=0.06413, over 3147069.69 frames. ], batch size: 75, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:57:53,637 INFO [train.py:904] (6/8) Epoch 7, batch 650, loss[loss=0.2028, simple_loss=0.2774, pruned_loss=0.06413, over 16726.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2779, pruned_loss=0.06278, over 3193073.06 frames. ], batch size: 134, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:58:26,506 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0672, 5.4384, 5.1095, 5.2224, 4.7862, 4.6859, 4.8477, 5.4548], device='cuda:6'), covar=tensor([0.0770, 0.0740, 0.0959, 0.0495, 0.0679, 0.0750, 0.0766, 0.0771], device='cuda:6'), in_proj_covar=tensor([0.0435, 0.0569, 0.0474, 0.0376, 0.0353, 0.0369, 0.0472, 0.0416], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:58:35,762 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.778e+02 2.890e+02 3.388e+02 4.314e+02 1.492e+03, threshold=6.776e+02, percent-clipped=7.0 2023-04-28 14:58:39,031 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2669, 4.2421, 4.1996, 3.3981, 4.2442, 1.5721, 3.9538, 3.8465], device='cuda:6'), covar=tensor([0.0106, 0.0079, 0.0135, 0.0369, 0.0078, 0.2348, 0.0128, 0.0208], device='cuda:6'), in_proj_covar=tensor([0.0104, 0.0089, 0.0140, 0.0131, 0.0105, 0.0158, 0.0122, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 14:59:01,942 INFO [train.py:904] (6/8) Epoch 7, batch 700, loss[loss=0.1948, simple_loss=0.2757, pruned_loss=0.05695, over 15890.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2783, pruned_loss=0.06325, over 3204965.10 frames. ], batch size: 35, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 15:00:08,122 INFO [train.py:904] (6/8) Epoch 7, batch 750, loss[loss=0.2216, simple_loss=0.2977, pruned_loss=0.07282, over 16809.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2792, pruned_loss=0.06357, over 3231126.14 frames. ], batch size: 102, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 15:00:51,555 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.695e+02 2.818e+02 3.177e+02 3.782e+02 5.959e+02, threshold=6.355e+02, percent-clipped=0.0 2023-04-28 15:00:55,572 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0535, 2.1239, 2.3882, 4.7785, 2.0757, 2.8576, 2.3744, 2.3868], device='cuda:6'), covar=tensor([0.0582, 0.2795, 0.1452, 0.0247, 0.3300, 0.1627, 0.2173, 0.2990], device='cuda:6'), in_proj_covar=tensor([0.0335, 0.0346, 0.0291, 0.0322, 0.0388, 0.0376, 0.0314, 0.0415], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 15:01:16,767 INFO [train.py:904] (6/8) Epoch 7, batch 800, loss[loss=0.2035, simple_loss=0.2704, pruned_loss=0.06827, over 16898.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2787, pruned_loss=0.06324, over 3252262.27 frames. ], batch size: 90, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:01:28,259 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9161, 3.8445, 4.3103, 4.2919, 4.3036, 3.9374, 4.0239, 3.9541], device='cuda:6'), covar=tensor([0.0305, 0.0556, 0.0412, 0.0426, 0.0407, 0.0377, 0.0745, 0.0435], device='cuda:6'), in_proj_covar=tensor([0.0282, 0.0279, 0.0284, 0.0271, 0.0330, 0.0298, 0.0403, 0.0245], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 15:01:37,370 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=61717.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:01:57,093 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61732.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:02:13,271 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61743.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:02:24,039 INFO [train.py:904] (6/8) Epoch 7, batch 850, loss[loss=0.193, simple_loss=0.271, pruned_loss=0.05753, over 16463.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2773, pruned_loss=0.06204, over 3266283.43 frames. ], batch size: 75, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:02:42,591 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=61765.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:03:06,806 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.892e+02 2.876e+02 3.503e+02 4.271e+02 9.516e+02, threshold=7.005e+02, percent-clipped=8.0 2023-04-28 15:03:13,790 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.24 vs. limit=5.0 2023-04-28 15:03:19,393 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61793.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:03:33,685 INFO [train.py:904] (6/8) Epoch 7, batch 900, loss[loss=0.173, simple_loss=0.2598, pruned_loss=0.04308, over 17145.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2768, pruned_loss=0.06115, over 3282608.51 frames. ], batch size: 46, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:03:36,345 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61804.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:04:40,613 INFO [train.py:904] (6/8) Epoch 7, batch 950, loss[loss=0.1902, simple_loss=0.2645, pruned_loss=0.0579, over 16453.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.277, pruned_loss=0.06115, over 3294713.22 frames. ], batch size: 68, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:05:20,758 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.798e+02 2.637e+02 3.238e+02 3.897e+02 7.916e+02, threshold=6.477e+02, percent-clipped=2.0 2023-04-28 15:05:23,795 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7727, 3.8476, 3.0450, 2.4496, 2.7486, 2.4045, 3.8085, 3.6000], device='cuda:6'), covar=tensor([0.2291, 0.0631, 0.1305, 0.2036, 0.2184, 0.1579, 0.0581, 0.1006], device='cuda:6'), in_proj_covar=tensor([0.0286, 0.0257, 0.0272, 0.0255, 0.0277, 0.0211, 0.0251, 0.0274], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 15:05:46,847 INFO [train.py:904] (6/8) Epoch 7, batch 1000, loss[loss=0.2023, simple_loss=0.2637, pruned_loss=0.07041, over 16799.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2756, pruned_loss=0.06084, over 3305965.59 frames. ], batch size: 102, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:06:05,345 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61914.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:06:56,255 INFO [train.py:904] (6/8) Epoch 7, batch 1050, loss[loss=0.214, simple_loss=0.2829, pruned_loss=0.07253, over 16894.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2755, pruned_loss=0.05984, over 3310167.85 frames. ], batch size: 116, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:07:23,199 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3903, 3.2114, 2.5866, 2.1461, 2.3525, 2.1891, 3.2229, 3.1117], device='cuda:6'), covar=tensor([0.2237, 0.0719, 0.1327, 0.1822, 0.1996, 0.1529, 0.0465, 0.1053], device='cuda:6'), in_proj_covar=tensor([0.0286, 0.0257, 0.0273, 0.0255, 0.0279, 0.0212, 0.0252, 0.0276], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 15:07:28,117 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61975.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:07:37,640 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.626e+02 2.768e+02 3.346e+02 4.170e+02 9.065e+02, threshold=6.692e+02, percent-clipped=2.0 2023-04-28 15:08:07,086 INFO [train.py:904] (6/8) Epoch 7, batch 1100, loss[loss=0.214, simple_loss=0.282, pruned_loss=0.07294, over 16784.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2758, pruned_loss=0.05975, over 3311617.04 frames. ], batch size: 102, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:08:59,107 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4284, 4.1348, 4.0426, 2.0448, 3.3875, 2.4165, 3.8317, 3.8375], device='cuda:6'), covar=tensor([0.0256, 0.0530, 0.0385, 0.1513, 0.0633, 0.0910, 0.0532, 0.0838], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0133, 0.0153, 0.0139, 0.0132, 0.0124, 0.0137, 0.0142], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 15:09:05,205 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 15:09:15,539 INFO [train.py:904] (6/8) Epoch 7, batch 1150, loss[loss=0.2044, simple_loss=0.2893, pruned_loss=0.05972, over 16654.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.2755, pruned_loss=0.05975, over 3303417.66 frames. ], batch size: 62, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:09:57,383 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 2.657e+02 3.127e+02 3.921e+02 1.096e+03, threshold=6.253e+02, percent-clipped=5.0 2023-04-28 15:10:04,080 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62088.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:10:05,230 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8250, 3.6158, 3.7478, 3.6474, 3.6754, 4.1640, 3.8671, 3.5177], device='cuda:6'), covar=tensor([0.1984, 0.1788, 0.1588, 0.2230, 0.2862, 0.1807, 0.1232, 0.2698], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0449, 0.0450, 0.0379, 0.0515, 0.0481, 0.0358, 0.0512], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 15:10:18,775 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62099.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:10:22,593 INFO [train.py:904] (6/8) Epoch 7, batch 1200, loss[loss=0.1949, simple_loss=0.2844, pruned_loss=0.05272, over 17031.00 frames. ], tot_loss[loss=0.1962, simple_loss=0.2748, pruned_loss=0.05875, over 3312583.00 frames. ], batch size: 50, lr: 1.01e-02, grad_scale: 8.0 2023-04-28 15:10:51,388 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62122.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:11:30,497 INFO [train.py:904] (6/8) Epoch 7, batch 1250, loss[loss=0.1736, simple_loss=0.2617, pruned_loss=0.04278, over 17128.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2751, pruned_loss=0.0592, over 3317190.81 frames. ], batch size: 47, lr: 1.01e-02, grad_scale: 8.0 2023-04-28 15:11:36,775 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-28 15:12:13,504 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 2.700e+02 3.234e+02 4.045e+02 7.667e+02, threshold=6.469e+02, percent-clipped=6.0 2023-04-28 15:12:13,944 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62183.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:12:39,181 INFO [train.py:904] (6/8) Epoch 7, batch 1300, loss[loss=0.1838, simple_loss=0.2779, pruned_loss=0.04489, over 17122.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2749, pruned_loss=0.05935, over 3323313.18 frames. ], batch size: 48, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:12:53,632 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0318, 4.6115, 4.6237, 3.4103, 4.0771, 4.6006, 4.2775, 3.0272], device='cuda:6'), covar=tensor([0.0263, 0.0022, 0.0023, 0.0199, 0.0037, 0.0036, 0.0027, 0.0246], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0063, 0.0063, 0.0116, 0.0065, 0.0077, 0.0068, 0.0114], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 15:13:17,929 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6367, 2.2976, 1.7866, 2.0524, 2.8188, 2.6391, 2.9567, 2.9008], device='cuda:6'), covar=tensor([0.0088, 0.0188, 0.0255, 0.0233, 0.0108, 0.0156, 0.0132, 0.0119], device='cuda:6'), in_proj_covar=tensor([0.0103, 0.0173, 0.0172, 0.0174, 0.0170, 0.0176, 0.0166, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 15:13:49,052 INFO [train.py:904] (6/8) Epoch 7, batch 1350, loss[loss=0.1771, simple_loss=0.2698, pruned_loss=0.04222, over 17120.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2752, pruned_loss=0.05917, over 3321199.13 frames. ], batch size: 49, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:14:15,364 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62270.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:14:19,023 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62273.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:14:29,420 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-28 15:14:31,591 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.487e+02 3.145e+02 3.874e+02 9.778e+02, threshold=6.290e+02, percent-clipped=3.0 2023-04-28 15:14:40,548 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2023-04-28 15:14:56,760 INFO [train.py:904] (6/8) Epoch 7, batch 1400, loss[loss=0.2021, simple_loss=0.2716, pruned_loss=0.06628, over 16406.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2757, pruned_loss=0.05938, over 3315539.48 frames. ], batch size: 146, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:15:42,107 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62334.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:15:53,103 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8920, 4.3276, 2.2712, 4.6748, 2.9550, 4.5921, 2.1858, 3.0255], device='cuda:6'), covar=tensor([0.0208, 0.0233, 0.1437, 0.0060, 0.0739, 0.0374, 0.1502, 0.0685], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0162, 0.0179, 0.0096, 0.0159, 0.0201, 0.0188, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 15:16:06,314 INFO [train.py:904] (6/8) Epoch 7, batch 1450, loss[loss=0.2064, simple_loss=0.2772, pruned_loss=0.06783, over 16850.00 frames. ], tot_loss[loss=0.1963, simple_loss=0.2741, pruned_loss=0.05927, over 3316347.63 frames. ], batch size: 116, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:16:47,642 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.994e+02 2.891e+02 3.326e+02 4.338e+02 8.815e+02, threshold=6.653e+02, percent-clipped=8.0 2023-04-28 15:16:56,028 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62388.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:17:00,002 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62391.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:17:10,969 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62399.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:17:14,668 INFO [train.py:904] (6/8) Epoch 7, batch 1500, loss[loss=0.1908, simple_loss=0.2895, pruned_loss=0.0461, over 17144.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2745, pruned_loss=0.05925, over 3320791.19 frames. ], batch size: 49, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:18:01,110 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=62436.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:18:16,545 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=62447.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:18:23,647 INFO [train.py:904] (6/8) Epoch 7, batch 1550, loss[loss=0.2066, simple_loss=0.3014, pruned_loss=0.05591, over 17067.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2754, pruned_loss=0.06048, over 3310764.60 frames. ], batch size: 55, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:18:24,110 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62452.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:18:54,720 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1836, 5.5239, 5.5794, 5.3931, 5.4191, 5.9791, 5.6144, 5.3380], device='cuda:6'), covar=tensor([0.0672, 0.1491, 0.1443, 0.1773, 0.2403, 0.0886, 0.0962, 0.2127], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0444, 0.0447, 0.0376, 0.0501, 0.0475, 0.0354, 0.0505], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 15:18:59,836 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62478.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:19:06,018 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.614e+02 2.773e+02 3.245e+02 3.904e+02 9.637e+02, threshold=6.490e+02, percent-clipped=1.0 2023-04-28 15:19:32,093 INFO [train.py:904] (6/8) Epoch 7, batch 1600, loss[loss=0.211, simple_loss=0.3006, pruned_loss=0.06067, over 17257.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2783, pruned_loss=0.062, over 3307245.59 frames. ], batch size: 52, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:20:00,338 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1493, 4.8331, 5.0639, 5.3262, 5.5027, 4.7568, 5.4348, 5.4712], device='cuda:6'), covar=tensor([0.1312, 0.0926, 0.1520, 0.0609, 0.0466, 0.0677, 0.0497, 0.0455], device='cuda:6'), in_proj_covar=tensor([0.0473, 0.0575, 0.0727, 0.0581, 0.0442, 0.0435, 0.0464, 0.0498], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 15:20:04,832 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.76 vs. limit=5.0 2023-04-28 15:20:31,585 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62545.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:20:40,969 INFO [train.py:904] (6/8) Epoch 7, batch 1650, loss[loss=0.1661, simple_loss=0.2509, pruned_loss=0.04062, over 16999.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2785, pruned_loss=0.06214, over 3317623.82 frames. ], batch size: 41, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:21:05,426 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62570.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:21:22,701 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.048e+02 2.793e+02 3.410e+02 4.977e+02 9.439e+02, threshold=6.821e+02, percent-clipped=5.0 2023-04-28 15:21:25,421 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62585.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:21:40,505 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4088, 2.3736, 1.9805, 2.1243, 2.7433, 2.6001, 3.3537, 2.9598], device='cuda:6'), covar=tensor([0.0051, 0.0249, 0.0297, 0.0280, 0.0151, 0.0228, 0.0135, 0.0162], device='cuda:6'), in_proj_covar=tensor([0.0105, 0.0175, 0.0173, 0.0175, 0.0171, 0.0177, 0.0168, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 15:21:46,131 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-28 15:21:49,314 INFO [train.py:904] (6/8) Epoch 7, batch 1700, loss[loss=0.2974, simple_loss=0.358, pruned_loss=0.1184, over 12373.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2809, pruned_loss=0.06284, over 3309050.55 frames. ], batch size: 246, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:21:55,105 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62606.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:22:12,979 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=62618.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:22:27,921 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62629.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:22:51,318 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62646.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:22:58,544 INFO [train.py:904] (6/8) Epoch 7, batch 1750, loss[loss=0.2191, simple_loss=0.2976, pruned_loss=0.07028, over 16843.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2811, pruned_loss=0.06212, over 3315168.35 frames. ], batch size: 42, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:23:04,875 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7785, 4.1156, 3.0888, 2.4485, 2.9571, 2.4394, 4.4091, 4.0061], device='cuda:6'), covar=tensor([0.2382, 0.0690, 0.1379, 0.1872, 0.2206, 0.1541, 0.0403, 0.0744], device='cuda:6'), in_proj_covar=tensor([0.0284, 0.0256, 0.0272, 0.0256, 0.0282, 0.0209, 0.0250, 0.0278], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 15:23:04,968 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.69 vs. limit=5.0 2023-04-28 15:23:18,135 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62666.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:23:41,547 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.869e+02 2.853e+02 3.319e+02 4.250e+02 7.879e+02, threshold=6.638e+02, percent-clipped=1.0 2023-04-28 15:24:08,263 INFO [train.py:904] (6/8) Epoch 7, batch 1800, loss[loss=0.1856, simple_loss=0.2678, pruned_loss=0.05176, over 16852.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2822, pruned_loss=0.06232, over 3317985.97 frames. ], batch size: 42, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:24:42,430 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62727.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:25:10,912 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62747.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:25:17,218 INFO [train.py:904] (6/8) Epoch 7, batch 1850, loss[loss=0.1937, simple_loss=0.2677, pruned_loss=0.05984, over 16835.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2838, pruned_loss=0.06263, over 3309896.27 frames. ], batch size: 96, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:25:45,872 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62773.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:25:53,022 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62778.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:25:59,964 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.663e+02 2.724e+02 3.242e+02 3.816e+02 7.402e+02, threshold=6.484e+02, percent-clipped=2.0 2023-04-28 15:26:26,217 INFO [train.py:904] (6/8) Epoch 7, batch 1900, loss[loss=0.2152, simple_loss=0.282, pruned_loss=0.07422, over 16900.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2827, pruned_loss=0.06198, over 3296963.17 frames. ], batch size: 109, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:26:39,031 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4941, 3.5311, 3.2131, 3.0147, 3.0819, 3.2961, 3.2294, 3.1095], device='cuda:6'), covar=tensor([0.0506, 0.0363, 0.0207, 0.0208, 0.0532, 0.0270, 0.1177, 0.0403], device='cuda:6'), in_proj_covar=tensor([0.0215, 0.0251, 0.0255, 0.0229, 0.0291, 0.0258, 0.0181, 0.0293], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 15:27:00,817 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=62826.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:27:11,592 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62834.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:27:36,783 INFO [train.py:904] (6/8) Epoch 7, batch 1950, loss[loss=0.1953, simple_loss=0.2738, pruned_loss=0.05839, over 16631.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2822, pruned_loss=0.06126, over 3293915.50 frames. ], batch size: 76, lr: 9.99e-03, grad_scale: 8.0 2023-04-28 15:27:43,255 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-28 15:28:19,259 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.715e+02 3.351e+02 4.043e+02 9.306e+02, threshold=6.703e+02, percent-clipped=2.0 2023-04-28 15:28:44,390 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62901.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:28:45,196 INFO [train.py:904] (6/8) Epoch 7, batch 2000, loss[loss=0.1553, simple_loss=0.2378, pruned_loss=0.03638, over 16785.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2829, pruned_loss=0.06181, over 3296199.88 frames. ], batch size: 39, lr: 9.99e-03, grad_scale: 8.0 2023-04-28 15:29:24,390 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62929.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:29:39,603 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62941.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:29:55,689 INFO [train.py:904] (6/8) Epoch 7, batch 2050, loss[loss=0.2287, simple_loss=0.3089, pruned_loss=0.07421, over 16759.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2832, pruned_loss=0.06213, over 3302045.43 frames. ], batch size: 57, lr: 9.99e-03, grad_scale: 16.0 2023-04-28 15:30:31,333 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=62977.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:30:39,208 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.104e+02 2.968e+02 3.465e+02 4.226e+02 9.943e+02, threshold=6.931e+02, percent-clipped=3.0 2023-04-28 15:31:05,139 INFO [train.py:904] (6/8) Epoch 7, batch 2100, loss[loss=0.2088, simple_loss=0.2915, pruned_loss=0.06304, over 17071.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2849, pruned_loss=0.06367, over 3303489.22 frames. ], batch size: 55, lr: 9.98e-03, grad_scale: 8.0 2023-04-28 15:31:33,198 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63022.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:32:09,403 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63047.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:32:16,868 INFO [train.py:904] (6/8) Epoch 7, batch 2150, loss[loss=0.2217, simple_loss=0.2918, pruned_loss=0.07576, over 16427.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2851, pruned_loss=0.06423, over 3311977.13 frames. ], batch size: 146, lr: 9.98e-03, grad_scale: 8.0 2023-04-28 15:33:00,561 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.945e+02 2.751e+02 3.323e+02 4.118e+02 9.925e+02, threshold=6.647e+02, percent-clipped=2.0 2023-04-28 15:33:17,098 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=63095.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:33:25,757 INFO [train.py:904] (6/8) Epoch 7, batch 2200, loss[loss=0.2034, simple_loss=0.2718, pruned_loss=0.06747, over 16922.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2865, pruned_loss=0.06483, over 3299313.05 frames. ], batch size: 90, lr: 9.98e-03, grad_scale: 8.0 2023-04-28 15:34:03,305 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63129.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:34:33,869 INFO [train.py:904] (6/8) Epoch 7, batch 2250, loss[loss=0.2107, simple_loss=0.3002, pruned_loss=0.06057, over 17071.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2868, pruned_loss=0.06459, over 3306659.91 frames. ], batch size: 55, lr: 9.97e-03, grad_scale: 8.0 2023-04-28 15:35:18,522 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 2.937e+02 3.503e+02 4.061e+02 7.762e+02, threshold=7.006e+02, percent-clipped=5.0 2023-04-28 15:35:42,369 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63201.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:35:43,089 INFO [train.py:904] (6/8) Epoch 7, batch 2300, loss[loss=0.2136, simple_loss=0.311, pruned_loss=0.05809, over 17035.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2864, pruned_loss=0.0641, over 3318740.16 frames. ], batch size: 50, lr: 9.97e-03, grad_scale: 8.0 2023-04-28 15:36:04,383 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9839, 1.8104, 2.2886, 2.8825, 2.7097, 3.1949, 1.8744, 3.1192], device='cuda:6'), covar=tensor([0.0125, 0.0283, 0.0196, 0.0156, 0.0149, 0.0104, 0.0296, 0.0073], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0159, 0.0142, 0.0143, 0.0146, 0.0103, 0.0151, 0.0094], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 15:36:10,210 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63221.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:36:20,581 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6728, 3.3975, 2.7204, 5.0837, 4.5053, 4.6826, 1.5873, 3.5192], device='cuda:6'), covar=tensor([0.1321, 0.0516, 0.1123, 0.0111, 0.0274, 0.0356, 0.1427, 0.0571], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0147, 0.0169, 0.0104, 0.0199, 0.0201, 0.0166, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 15:36:37,801 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63241.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:36:45,661 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-28 15:36:48,218 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=63249.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:36:51,523 INFO [train.py:904] (6/8) Epoch 7, batch 2350, loss[loss=0.2028, simple_loss=0.2929, pruned_loss=0.05638, over 17284.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2868, pruned_loss=0.06442, over 3321281.92 frames. ], batch size: 52, lr: 9.96e-03, grad_scale: 4.0 2023-04-28 15:36:51,902 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.7161, 6.0827, 5.7599, 5.8315, 5.3065, 5.0318, 5.5171, 6.1731], device='cuda:6'), covar=tensor([0.0801, 0.0653, 0.1080, 0.0500, 0.0682, 0.0702, 0.0722, 0.0678], device='cuda:6'), in_proj_covar=tensor([0.0454, 0.0589, 0.0493, 0.0389, 0.0365, 0.0379, 0.0484, 0.0430], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 15:37:19,095 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1387, 5.7569, 5.8635, 5.7331, 5.6955, 6.2278, 5.8095, 5.5358], device='cuda:6'), covar=tensor([0.0714, 0.1472, 0.1563, 0.1522, 0.2292, 0.0833, 0.1010, 0.2108], device='cuda:6'), in_proj_covar=tensor([0.0323, 0.0463, 0.0459, 0.0386, 0.0527, 0.0493, 0.0371, 0.0525], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 15:37:34,021 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63282.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:37:38,006 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.053e+02 2.699e+02 3.274e+02 4.399e+02 8.377e+02, threshold=6.549e+02, percent-clipped=2.0 2023-04-28 15:37:43,610 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=63289.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:38:00,171 INFO [train.py:904] (6/8) Epoch 7, batch 2400, loss[loss=0.2055, simple_loss=0.2833, pruned_loss=0.06379, over 16746.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.2877, pruned_loss=0.06409, over 3334506.61 frames. ], batch size: 89, lr: 9.96e-03, grad_scale: 8.0 2023-04-28 15:38:27,721 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63322.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:39:02,447 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0530, 4.4823, 1.8967, 4.6648, 2.9520, 4.6980, 2.2139, 3.2114], device='cuda:6'), covar=tensor([0.0160, 0.0186, 0.1565, 0.0082, 0.0662, 0.0266, 0.1389, 0.0542], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0163, 0.0179, 0.0099, 0.0160, 0.0203, 0.0187, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 15:39:10,360 INFO [train.py:904] (6/8) Epoch 7, batch 2450, loss[loss=0.2007, simple_loss=0.2762, pruned_loss=0.0626, over 16505.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2872, pruned_loss=0.06285, over 3336707.84 frames. ], batch size: 75, lr: 9.96e-03, grad_scale: 8.0 2023-04-28 15:39:34,202 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=63370.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:39:55,096 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.710e+02 2.805e+02 3.565e+02 4.384e+02 1.263e+03, threshold=7.129e+02, percent-clipped=5.0 2023-04-28 15:40:19,245 INFO [train.py:904] (6/8) Epoch 7, batch 2500, loss[loss=0.1887, simple_loss=0.279, pruned_loss=0.04914, over 17047.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2858, pruned_loss=0.06237, over 3335769.93 frames. ], batch size: 50, lr: 9.95e-03, grad_scale: 4.0 2023-04-28 15:40:44,237 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63420.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:40:46,558 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63422.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:40:55,629 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63429.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:41:26,923 INFO [train.py:904] (6/8) Epoch 7, batch 2550, loss[loss=0.193, simple_loss=0.2654, pruned_loss=0.06028, over 16832.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2861, pruned_loss=0.06232, over 3333878.36 frames. ], batch size: 102, lr: 9.95e-03, grad_scale: 4.0 2023-04-28 15:42:02,624 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=63477.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:42:08,310 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63481.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:42:11,094 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63483.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:42:14,138 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 3.238e+02 3.870e+02 4.521e+02 8.304e+02, threshold=7.740e+02, percent-clipped=2.0 2023-04-28 15:42:31,081 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3294, 1.9248, 2.1487, 3.9083, 1.9743, 2.5772, 2.0817, 2.0964], device='cuda:6'), covar=tensor([0.0741, 0.2930, 0.1609, 0.0349, 0.2924, 0.1763, 0.2599, 0.2600], device='cuda:6'), in_proj_covar=tensor([0.0341, 0.0353, 0.0295, 0.0322, 0.0387, 0.0388, 0.0319, 0.0420], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 15:42:35,898 INFO [train.py:904] (6/8) Epoch 7, batch 2600, loss[loss=0.217, simple_loss=0.293, pruned_loss=0.07052, over 15909.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2854, pruned_loss=0.06175, over 3325814.50 frames. ], batch size: 35, lr: 9.94e-03, grad_scale: 4.0 2023-04-28 15:42:52,179 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0251, 3.8072, 4.0383, 4.2319, 4.3043, 3.8538, 4.0536, 4.3087], device='cuda:6'), covar=tensor([0.1067, 0.0962, 0.1198, 0.0628, 0.0553, 0.1350, 0.1603, 0.0489], device='cuda:6'), in_proj_covar=tensor([0.0474, 0.0581, 0.0735, 0.0582, 0.0449, 0.0442, 0.0458, 0.0506], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 15:43:42,047 INFO [train.py:904] (6/8) Epoch 7, batch 2650, loss[loss=0.1938, simple_loss=0.2933, pruned_loss=0.04714, over 17248.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2859, pruned_loss=0.06142, over 3323549.28 frames. ], batch size: 52, lr: 9.94e-03, grad_scale: 4.0 2023-04-28 15:44:06,153 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63569.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:44:11,472 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-28 15:44:15,584 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63577.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:44:27,533 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.875e+02 2.764e+02 3.348e+02 3.954e+02 9.438e+02, threshold=6.696e+02, percent-clipped=1.0 2023-04-28 15:44:27,939 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63586.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:44:48,751 INFO [train.py:904] (6/8) Epoch 7, batch 2700, loss[loss=0.2151, simple_loss=0.2891, pruned_loss=0.0706, over 15499.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.2859, pruned_loss=0.06116, over 3332414.37 frames. ], batch size: 190, lr: 9.94e-03, grad_scale: 4.0 2023-04-28 15:45:28,031 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63630.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:45:52,003 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63647.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:45:57,998 INFO [train.py:904] (6/8) Epoch 7, batch 2750, loss[loss=0.2041, simple_loss=0.287, pruned_loss=0.0606, over 16050.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2859, pruned_loss=0.06031, over 3336685.05 frames. ], batch size: 35, lr: 9.93e-03, grad_scale: 4.0 2023-04-28 15:46:45,843 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.888e+02 2.696e+02 3.173e+02 3.925e+02 8.532e+02, threshold=6.347e+02, percent-clipped=1.0 2023-04-28 15:47:08,534 INFO [train.py:904] (6/8) Epoch 7, batch 2800, loss[loss=0.218, simple_loss=0.2907, pruned_loss=0.07261, over 16235.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2845, pruned_loss=0.05978, over 3334043.65 frames. ], batch size: 165, lr: 9.93e-03, grad_scale: 8.0 2023-04-28 15:47:24,078 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8327, 1.4791, 2.2416, 2.7473, 2.5707, 3.1834, 1.8651, 3.1275], device='cuda:6'), covar=tensor([0.0137, 0.0324, 0.0187, 0.0175, 0.0153, 0.0103, 0.0285, 0.0085], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0159, 0.0142, 0.0143, 0.0148, 0.0105, 0.0150, 0.0096], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 15:48:12,843 INFO [train.py:904] (6/8) Epoch 7, batch 2850, loss[loss=0.2076, simple_loss=0.2938, pruned_loss=0.06067, over 16773.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2835, pruned_loss=0.05987, over 3333233.08 frames. ], batch size: 62, lr: 9.92e-03, grad_scale: 8.0 2023-04-28 15:48:47,982 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63776.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:48:50,232 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63778.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:49:02,036 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.714e+02 3.034e+02 3.568e+02 4.214e+02 8.361e+02, threshold=7.135e+02, percent-clipped=4.0 2023-04-28 15:49:22,552 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2270, 3.7351, 3.1702, 2.0313, 2.9005, 2.4215, 3.6367, 3.6813], device='cuda:6'), covar=tensor([0.0223, 0.0541, 0.0539, 0.1438, 0.0614, 0.0885, 0.0425, 0.0674], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0143, 0.0159, 0.0143, 0.0136, 0.0126, 0.0141, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 15:49:23,198 INFO [train.py:904] (6/8) Epoch 7, batch 2900, loss[loss=0.2198, simple_loss=0.2859, pruned_loss=0.07688, over 16807.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2834, pruned_loss=0.06097, over 3327404.72 frames. ], batch size: 102, lr: 9.92e-03, grad_scale: 8.0 2023-04-28 15:49:35,873 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63810.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:49:49,436 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.61 vs. limit=5.0 2023-04-28 15:50:22,344 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63845.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:50:32,271 INFO [train.py:904] (6/8) Epoch 7, batch 2950, loss[loss=0.2136, simple_loss=0.2812, pruned_loss=0.07295, over 16879.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2823, pruned_loss=0.06187, over 3324022.74 frames. ], batch size: 116, lr: 9.92e-03, grad_scale: 8.0 2023-04-28 15:50:58,376 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63871.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:51:05,358 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63877.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:51:17,101 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.041e+02 3.004e+02 3.573e+02 4.238e+02 9.088e+02, threshold=7.147e+02, percent-clipped=1.0 2023-04-28 15:51:38,107 INFO [train.py:904] (6/8) Epoch 7, batch 3000, loss[loss=0.1664, simple_loss=0.2456, pruned_loss=0.04364, over 17042.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2822, pruned_loss=0.06205, over 3331333.96 frames. ], batch size: 41, lr: 9.91e-03, grad_scale: 8.0 2023-04-28 15:51:38,108 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 15:51:45,769 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3527, 2.4512, 2.1836, 2.2013, 2.8525, 2.6538, 3.2968, 3.0295], device='cuda:6'), covar=tensor([0.0063, 0.0236, 0.0246, 0.0274, 0.0151, 0.0201, 0.0122, 0.0158], device='cuda:6'), in_proj_covar=tensor([0.0109, 0.0177, 0.0172, 0.0175, 0.0172, 0.0177, 0.0174, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 15:51:46,842 INFO [train.py:938] (6/8) Epoch 7, validation: loss=0.1489, simple_loss=0.2553, pruned_loss=0.02124, over 944034.00 frames. 2023-04-28 15:51:46,842 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-28 15:51:54,313 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63906.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:52:07,324 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63916.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:52:20,148 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=63925.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:52:20,162 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63925.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:52:42,754 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63942.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:52:56,533 INFO [train.py:904] (6/8) Epoch 7, batch 3050, loss[loss=0.2032, simple_loss=0.2909, pruned_loss=0.05777, over 17030.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2822, pruned_loss=0.06203, over 3320373.96 frames. ], batch size: 55, lr: 9.91e-03, grad_scale: 4.0 2023-04-28 15:53:26,926 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0463, 2.9370, 2.6808, 1.8706, 2.4714, 2.0818, 2.6617, 2.8705], device='cuda:6'), covar=tensor([0.0256, 0.0540, 0.0497, 0.1553, 0.0707, 0.0841, 0.0571, 0.0553], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0140, 0.0156, 0.0141, 0.0133, 0.0125, 0.0139, 0.0149], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 15:53:30,197 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63977.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:53:31,582 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 15:53:43,704 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.653e+02 2.789e+02 3.428e+02 3.925e+02 6.614e+02, threshold=6.856e+02, percent-clipped=0.0 2023-04-28 15:54:06,476 INFO [train.py:904] (6/8) Epoch 7, batch 3100, loss[loss=0.1891, simple_loss=0.2591, pruned_loss=0.05956, over 16757.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2813, pruned_loss=0.06219, over 3321438.53 frames. ], batch size: 102, lr: 9.91e-03, grad_scale: 4.0 2023-04-28 15:54:36,531 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64023.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:55:16,377 INFO [train.py:904] (6/8) Epoch 7, batch 3150, loss[loss=0.2071, simple_loss=0.2962, pruned_loss=0.05903, over 17134.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2811, pruned_loss=0.06174, over 3327792.68 frames. ], batch size: 49, lr: 9.90e-03, grad_scale: 4.0 2023-04-28 15:55:30,795 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-28 15:55:49,728 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64076.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:55:52,214 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64078.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:56:00,097 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64084.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:56:03,680 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.615e+02 2.613e+02 3.331e+02 4.706e+02 9.304e+02, threshold=6.662e+02, percent-clipped=4.0 2023-04-28 15:56:24,283 INFO [train.py:904] (6/8) Epoch 7, batch 3200, loss[loss=0.1778, simple_loss=0.2485, pruned_loss=0.05351, over 17006.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2796, pruned_loss=0.06121, over 3322964.84 frames. ], batch size: 41, lr: 9.90e-03, grad_scale: 8.0 2023-04-28 15:56:32,236 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.97 vs. limit=5.0 2023-04-28 15:56:46,786 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2523, 4.5526, 4.3337, 4.3491, 4.0678, 4.0632, 4.0747, 4.5810], device='cuda:6'), covar=tensor([0.0883, 0.0880, 0.1012, 0.0575, 0.0770, 0.1304, 0.0876, 0.0842], device='cuda:6'), in_proj_covar=tensor([0.0479, 0.0621, 0.0517, 0.0405, 0.0390, 0.0399, 0.0507, 0.0455], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 15:56:47,148 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-28 15:56:54,788 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64124.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:56:57,814 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64126.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:57:19,149 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-04-28 15:57:33,481 INFO [train.py:904] (6/8) Epoch 7, batch 3250, loss[loss=0.2068, simple_loss=0.2932, pruned_loss=0.06017, over 17033.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2801, pruned_loss=0.06156, over 3324499.43 frames. ], batch size: 53, lr: 9.89e-03, grad_scale: 8.0 2023-04-28 15:57:53,166 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64166.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:58:21,282 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.755e+02 2.809e+02 3.257e+02 3.904e+02 7.986e+02, threshold=6.515e+02, percent-clipped=1.0 2023-04-28 15:58:30,713 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4518, 4.1163, 3.7811, 1.9614, 2.9747, 2.4458, 3.7290, 3.8807], device='cuda:6'), covar=tensor([0.0324, 0.0598, 0.0494, 0.1710, 0.0782, 0.1014, 0.0753, 0.1012], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0140, 0.0156, 0.0140, 0.0133, 0.0125, 0.0139, 0.0149], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 15:58:42,120 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64201.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:58:43,054 INFO [train.py:904] (6/8) Epoch 7, batch 3300, loss[loss=0.1893, simple_loss=0.2848, pruned_loss=0.04687, over 17238.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2829, pruned_loss=0.06292, over 3313143.03 frames. ], batch size: 52, lr: 9.89e-03, grad_scale: 8.0 2023-04-28 15:59:08,268 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-28 15:59:16,274 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64225.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:59:18,093 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-04-28 15:59:25,473 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64232.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:59:41,161 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64242.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:59:51,969 INFO [train.py:904] (6/8) Epoch 7, batch 3350, loss[loss=0.2088, simple_loss=0.2821, pruned_loss=0.06772, over 16190.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2822, pruned_loss=0.06194, over 3320523.32 frames. ], batch size: 165, lr: 9.89e-03, grad_scale: 8.0 2023-04-28 16:00:20,548 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64272.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:00:22,691 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64273.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:00:31,235 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7515, 4.2686, 3.2961, 2.5139, 3.0512, 2.4505, 4.5994, 4.0439], device='cuda:6'), covar=tensor([0.2439, 0.0572, 0.1248, 0.1828, 0.2126, 0.1554, 0.0330, 0.0745], device='cuda:6'), in_proj_covar=tensor([0.0285, 0.0256, 0.0271, 0.0257, 0.0287, 0.0211, 0.0253, 0.0280], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:00:42,046 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.854e+02 2.871e+02 3.440e+02 4.131e+02 8.469e+02, threshold=6.880e+02, percent-clipped=3.0 2023-04-28 16:00:45,938 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64290.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:00:49,500 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64293.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:01:02,162 INFO [train.py:904] (6/8) Epoch 7, batch 3400, loss[loss=0.2071, simple_loss=0.286, pruned_loss=0.06408, over 16368.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2811, pruned_loss=0.06105, over 3321224.75 frames. ], batch size: 165, lr: 9.88e-03, grad_scale: 8.0 2023-04-28 16:02:00,680 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3780, 3.2499, 3.6730, 2.5072, 3.3024, 3.6431, 3.5641, 1.9785], device='cuda:6'), covar=tensor([0.0298, 0.0084, 0.0029, 0.0214, 0.0053, 0.0060, 0.0038, 0.0309], device='cuda:6'), in_proj_covar=tensor([0.0118, 0.0062, 0.0062, 0.0115, 0.0066, 0.0077, 0.0068, 0.0112], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 16:02:02,932 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64345.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:02:03,008 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3608, 2.0946, 1.6210, 1.8760, 2.4129, 2.1845, 2.5052, 2.5653], device='cuda:6'), covar=tensor([0.0077, 0.0193, 0.0258, 0.0235, 0.0105, 0.0190, 0.0124, 0.0128], device='cuda:6'), in_proj_covar=tensor([0.0111, 0.0177, 0.0174, 0.0175, 0.0174, 0.0178, 0.0177, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:02:10,852 INFO [train.py:904] (6/8) Epoch 7, batch 3450, loss[loss=0.177, simple_loss=0.2546, pruned_loss=0.04968, over 16712.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2804, pruned_loss=0.06071, over 3312560.19 frames. ], batch size: 39, lr: 9.88e-03, grad_scale: 8.0 2023-04-28 16:02:49,995 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64379.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:02:52,874 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4263, 4.4209, 4.5525, 4.5358, 4.4394, 5.0551, 4.6983, 4.3640], device='cuda:6'), covar=tensor([0.1585, 0.1854, 0.1566, 0.1801, 0.2711, 0.1057, 0.1290, 0.2438], device='cuda:6'), in_proj_covar=tensor([0.0320, 0.0453, 0.0453, 0.0383, 0.0516, 0.0486, 0.0366, 0.0518], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 16:03:01,785 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.744e+02 2.661e+02 3.216e+02 3.896e+02 8.084e+02, threshold=6.431e+02, percent-clipped=1.0 2023-04-28 16:03:07,009 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-28 16:03:21,643 INFO [train.py:904] (6/8) Epoch 7, batch 3500, loss[loss=0.1904, simple_loss=0.2739, pruned_loss=0.05349, over 17191.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2797, pruned_loss=0.05997, over 3318993.73 frames. ], batch size: 46, lr: 9.87e-03, grad_scale: 8.0 2023-04-28 16:03:28,834 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64406.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:04:30,074 INFO [train.py:904] (6/8) Epoch 7, batch 3550, loss[loss=0.1871, simple_loss=0.2594, pruned_loss=0.05743, over 16724.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2785, pruned_loss=0.05983, over 3318593.77 frames. ], batch size: 124, lr: 9.87e-03, grad_scale: 8.0 2023-04-28 16:04:48,283 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64465.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:04:50,574 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64466.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:05:18,414 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.738e+02 2.678e+02 3.341e+02 4.465e+02 1.003e+03, threshold=6.682e+02, percent-clipped=6.0 2023-04-28 16:05:38,576 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64501.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:05:39,310 INFO [train.py:904] (6/8) Epoch 7, batch 3600, loss[loss=0.1966, simple_loss=0.2726, pruned_loss=0.06024, over 17186.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2763, pruned_loss=0.05951, over 3316126.88 frames. ], batch size: 46, lr: 9.87e-03, grad_scale: 8.0 2023-04-28 16:05:56,510 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64514.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:06:07,121 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9058, 4.3137, 4.5606, 3.1540, 3.9624, 4.4699, 4.1630, 2.6053], device='cuda:6'), covar=tensor([0.0298, 0.0024, 0.0021, 0.0219, 0.0036, 0.0041, 0.0031, 0.0299], device='cuda:6'), in_proj_covar=tensor([0.0117, 0.0061, 0.0061, 0.0115, 0.0065, 0.0076, 0.0068, 0.0111], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 16:06:14,289 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64526.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:06:22,394 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-28 16:06:45,521 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64549.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:06:48,842 INFO [train.py:904] (6/8) Epoch 7, batch 3650, loss[loss=0.2324, simple_loss=0.2973, pruned_loss=0.08376, over 11383.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2763, pruned_loss=0.0607, over 3304256.73 frames. ], batch size: 250, lr: 9.86e-03, grad_scale: 8.0 2023-04-28 16:07:19,318 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64572.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:07:34,792 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3832, 1.9176, 2.2639, 3.9122, 1.9572, 2.5837, 2.1319, 2.1583], device='cuda:6'), covar=tensor([0.0789, 0.2794, 0.1501, 0.0356, 0.2959, 0.1715, 0.2575, 0.2457], device='cuda:6'), in_proj_covar=tensor([0.0345, 0.0357, 0.0298, 0.0328, 0.0389, 0.0393, 0.0321, 0.0423], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:07:40,346 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.014e+02 2.562e+02 3.124e+02 4.018e+02 1.198e+03, threshold=6.249e+02, percent-clipped=5.0 2023-04-28 16:07:43,316 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64588.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:07:44,605 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64589.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:08:01,998 INFO [train.py:904] (6/8) Epoch 7, batch 3700, loss[loss=0.2068, simple_loss=0.2685, pruned_loss=0.07262, over 16896.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2754, pruned_loss=0.0627, over 3279893.78 frames. ], batch size: 116, lr: 9.86e-03, grad_scale: 8.0 2023-04-28 16:08:27,963 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64620.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:09:09,995 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64648.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:09:12,297 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64650.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:09:14,069 INFO [train.py:904] (6/8) Epoch 7, batch 3750, loss[loss=0.2032, simple_loss=0.2674, pruned_loss=0.06949, over 16808.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2762, pruned_loss=0.06402, over 3270651.07 frames. ], batch size: 96, lr: 9.86e-03, grad_scale: 8.0 2023-04-28 16:09:55,076 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64679.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:10:04,723 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.745e+02 2.843e+02 3.454e+02 4.284e+02 7.405e+02, threshold=6.908e+02, percent-clipped=2.0 2023-04-28 16:10:24,968 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64701.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:10:25,831 INFO [train.py:904] (6/8) Epoch 7, batch 3800, loss[loss=0.224, simple_loss=0.2981, pruned_loss=0.07493, over 16813.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2778, pruned_loss=0.06544, over 3271669.37 frames. ], batch size: 42, lr: 9.85e-03, grad_scale: 4.0 2023-04-28 16:10:27,990 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64703.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:10:38,579 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64709.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:11:05,637 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64727.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:11:09,667 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.92 vs. limit=5.0 2023-04-28 16:11:31,468 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64745.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:11:35,248 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-28 16:11:39,107 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7345, 5.0200, 5.1309, 5.1472, 5.1404, 5.6529, 5.3066, 5.0293], device='cuda:6'), covar=tensor([0.1072, 0.1442, 0.1392, 0.1529, 0.2265, 0.0895, 0.1031, 0.2090], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0447, 0.0446, 0.0373, 0.0505, 0.0475, 0.0358, 0.0505], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 16:11:39,992 INFO [train.py:904] (6/8) Epoch 7, batch 3850, loss[loss=0.1831, simple_loss=0.2528, pruned_loss=0.05672, over 16677.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2783, pruned_loss=0.06637, over 3262478.30 frames. ], batch size: 83, lr: 9.85e-03, grad_scale: 4.0 2023-04-28 16:11:41,119 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64752.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:11:42,181 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5971, 4.6599, 4.8017, 4.8233, 4.8069, 5.3336, 4.9633, 4.7357], device='cuda:6'), covar=tensor([0.1177, 0.1522, 0.1511, 0.1687, 0.2519, 0.0947, 0.1147, 0.2169], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0448, 0.0446, 0.0373, 0.0505, 0.0475, 0.0359, 0.0505], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 16:11:58,636 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64764.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:12:33,101 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.798e+02 2.797e+02 3.034e+02 3.668e+02 1.110e+03, threshold=6.069e+02, percent-clipped=1.0 2023-04-28 16:12:52,610 INFO [train.py:904] (6/8) Epoch 7, batch 3900, loss[loss=0.1942, simple_loss=0.2768, pruned_loss=0.05576, over 17180.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2785, pruned_loss=0.06694, over 3264711.04 frames. ], batch size: 46, lr: 9.84e-03, grad_scale: 4.0 2023-04-28 16:13:00,336 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64806.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:13:01,434 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5477, 3.8397, 4.3416, 1.7345, 4.4997, 4.6011, 3.2739, 3.2988], device='cuda:6'), covar=tensor([0.0871, 0.0171, 0.0101, 0.1397, 0.0051, 0.0040, 0.0317, 0.0432], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0096, 0.0083, 0.0137, 0.0072, 0.0090, 0.0119, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 16:13:08,290 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64813.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:13:20,539 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64821.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:14:02,898 INFO [train.py:904] (6/8) Epoch 7, batch 3950, loss[loss=0.1809, simple_loss=0.2546, pruned_loss=0.05361, over 16943.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2775, pruned_loss=0.06709, over 3270598.19 frames. ], batch size: 90, lr: 9.84e-03, grad_scale: 4.0 2023-04-28 16:14:05,062 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9639, 3.9220, 3.9496, 3.2860, 3.9679, 1.7080, 3.7276, 3.5096], device='cuda:6'), covar=tensor([0.0102, 0.0072, 0.0120, 0.0294, 0.0073, 0.2094, 0.0114, 0.0187], device='cuda:6'), in_proj_covar=tensor([0.0112, 0.0099, 0.0150, 0.0146, 0.0115, 0.0158, 0.0133, 0.0143], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:14:53,035 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.973e+02 2.882e+02 3.338e+02 4.272e+02 6.667e+02, threshold=6.675e+02, percent-clipped=4.0 2023-04-28 16:14:54,203 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64888.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:15:12,235 INFO [train.py:904] (6/8) Epoch 7, batch 4000, loss[loss=0.194, simple_loss=0.2731, pruned_loss=0.05742, over 16594.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2767, pruned_loss=0.06696, over 3285946.04 frames. ], batch size: 75, lr: 9.84e-03, grad_scale: 8.0 2023-04-28 16:15:55,103 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-28 16:16:02,045 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64936.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:16:17,026 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64945.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:16:19,761 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-28 16:16:24,726 INFO [train.py:904] (6/8) Epoch 7, batch 4050, loss[loss=0.222, simple_loss=0.2923, pruned_loss=0.0759, over 16501.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2758, pruned_loss=0.06508, over 3291081.55 frames. ], batch size: 68, lr: 9.83e-03, grad_scale: 8.0 2023-04-28 16:16:59,833 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64976.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:17:16,149 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.141e+02 2.464e+02 3.078e+02 9.943e+02, threshold=4.928e+02, percent-clipped=3.0 2023-04-28 16:17:31,149 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2557, 4.1201, 4.0980, 2.4918, 3.5542, 4.0391, 3.8498, 2.3802], device='cuda:6'), covar=tensor([0.0351, 0.0015, 0.0019, 0.0248, 0.0040, 0.0048, 0.0029, 0.0255], device='cuda:6'), in_proj_covar=tensor([0.0117, 0.0059, 0.0060, 0.0114, 0.0064, 0.0073, 0.0066, 0.0109], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 16:17:35,748 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65001.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:17:36,414 INFO [train.py:904] (6/8) Epoch 7, batch 4100, loss[loss=0.2065, simple_loss=0.2896, pruned_loss=0.06169, over 16438.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2767, pruned_loss=0.06391, over 3275083.36 frames. ], batch size: 75, lr: 9.83e-03, grad_scale: 8.0 2023-04-28 16:17:39,308 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65004.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:17:47,294 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1796, 2.3019, 1.6639, 2.1230, 2.7192, 2.4129, 3.0737, 3.0102], device='cuda:6'), covar=tensor([0.0042, 0.0210, 0.0317, 0.0243, 0.0129, 0.0215, 0.0091, 0.0123], device='cuda:6'), in_proj_covar=tensor([0.0106, 0.0173, 0.0171, 0.0170, 0.0168, 0.0174, 0.0168, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:18:29,699 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65037.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 16:18:31,354 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6621, 4.9512, 5.1416, 5.0239, 5.0364, 5.5931, 5.2320, 4.9713], device='cuda:6'), covar=tensor([0.0855, 0.1540, 0.1436, 0.1452, 0.2279, 0.0867, 0.0991, 0.1853], device='cuda:6'), in_proj_covar=tensor([0.0311, 0.0446, 0.0441, 0.0370, 0.0502, 0.0471, 0.0355, 0.0501], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 16:18:37,702 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7913, 4.9866, 4.7937, 4.8597, 4.4760, 4.3293, 4.5412, 5.0727], device='cuda:6'), covar=tensor([0.0737, 0.0678, 0.0814, 0.0490, 0.0624, 0.0858, 0.0704, 0.0635], device='cuda:6'), in_proj_covar=tensor([0.0442, 0.0576, 0.0478, 0.0377, 0.0362, 0.0379, 0.0473, 0.0419], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:18:47,164 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65049.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:18:51,076 INFO [train.py:904] (6/8) Epoch 7, batch 4150, loss[loss=0.2265, simple_loss=0.3151, pruned_loss=0.06899, over 16138.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.2855, pruned_loss=0.06754, over 3236547.80 frames. ], batch size: 165, lr: 9.83e-03, grad_scale: 8.0 2023-04-28 16:18:56,815 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65055.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:19:01,080 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-28 16:19:01,897 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65059.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:19:44,054 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.893e+02 2.789e+02 3.341e+02 3.997e+02 8.596e+02, threshold=6.682e+02, percent-clipped=9.0 2023-04-28 16:20:03,731 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65101.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:20:04,500 INFO [train.py:904] (6/8) Epoch 7, batch 4200, loss[loss=0.273, simple_loss=0.3312, pruned_loss=0.1074, over 11453.00 frames. ], tot_loss[loss=0.2151, simple_loss=0.2924, pruned_loss=0.06889, over 3222587.42 frames. ], batch size: 250, lr: 9.82e-03, grad_scale: 8.0 2023-04-28 16:20:13,327 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65108.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:20:24,636 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65116.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:20:32,644 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65121.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:20:55,675 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-28 16:21:16,725 INFO [train.py:904] (6/8) Epoch 7, batch 4250, loss[loss=0.2188, simple_loss=0.3052, pruned_loss=0.06624, over 16682.00 frames. ], tot_loss[loss=0.216, simple_loss=0.2949, pruned_loss=0.06855, over 3215167.52 frames. ], batch size: 134, lr: 9.82e-03, grad_scale: 8.0 2023-04-28 16:21:31,505 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-28 16:21:42,347 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65169.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:22:08,033 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.682e+02 2.671e+02 3.307e+02 3.991e+02 9.211e+02, threshold=6.614e+02, percent-clipped=3.0 2023-04-28 16:22:29,057 INFO [train.py:904] (6/8) Epoch 7, batch 4300, loss[loss=0.2287, simple_loss=0.3144, pruned_loss=0.07148, over 16439.00 frames. ], tot_loss[loss=0.2153, simple_loss=0.2958, pruned_loss=0.0674, over 3212736.18 frames. ], batch size: 68, lr: 9.81e-03, grad_scale: 8.0 2023-04-28 16:23:31,787 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65245.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:23:41,835 INFO [train.py:904] (6/8) Epoch 7, batch 4350, loss[loss=0.2416, simple_loss=0.3247, pruned_loss=0.0793, over 16727.00 frames. ], tot_loss[loss=0.2185, simple_loss=0.2995, pruned_loss=0.0687, over 3202720.99 frames. ], batch size: 124, lr: 9.81e-03, grad_scale: 8.0 2023-04-28 16:24:30,740 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65285.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:24:33,877 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.656e+02 2.885e+02 3.415e+02 4.232e+02 9.360e+02, threshold=6.829e+02, percent-clipped=2.0 2023-04-28 16:24:41,438 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.04 vs. limit=5.0 2023-04-28 16:24:42,764 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65293.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:24:54,966 INFO [train.py:904] (6/8) Epoch 7, batch 4400, loss[loss=0.2108, simple_loss=0.2958, pruned_loss=0.06295, over 16639.00 frames. ], tot_loss[loss=0.2213, simple_loss=0.3021, pruned_loss=0.07019, over 3192319.16 frames. ], batch size: 75, lr: 9.81e-03, grad_scale: 8.0 2023-04-28 16:24:57,737 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65304.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:25:39,691 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65332.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 16:25:59,874 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65346.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:26:07,820 INFO [train.py:904] (6/8) Epoch 7, batch 4450, loss[loss=0.2184, simple_loss=0.301, pruned_loss=0.06793, over 16354.00 frames. ], tot_loss[loss=0.2237, simple_loss=0.3053, pruned_loss=0.07105, over 3209111.48 frames. ], batch size: 35, lr: 9.80e-03, grad_scale: 8.0 2023-04-28 16:26:08,159 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65352.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:26:17,503 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65359.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:26:34,052 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8555, 4.0994, 4.3670, 1.8503, 4.6780, 4.6907, 3.3042, 3.7540], device='cuda:6'), covar=tensor([0.0714, 0.0146, 0.0141, 0.1231, 0.0042, 0.0043, 0.0322, 0.0296], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0098, 0.0083, 0.0141, 0.0071, 0.0089, 0.0121, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 16:26:35,175 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7353, 3.6335, 3.8098, 3.9491, 4.0212, 3.5761, 3.9916, 4.0693], device='cuda:6'), covar=tensor([0.0949, 0.0768, 0.1041, 0.0470, 0.0386, 0.1560, 0.0480, 0.0399], device='cuda:6'), in_proj_covar=tensor([0.0437, 0.0532, 0.0665, 0.0537, 0.0406, 0.0416, 0.0416, 0.0460], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:26:58,768 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.495e+02 2.995e+02 3.500e+02 6.394e+02, threshold=5.989e+02, percent-clipped=0.0 2023-04-28 16:27:19,320 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65401.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:27:20,139 INFO [train.py:904] (6/8) Epoch 7, batch 4500, loss[loss=0.2044, simple_loss=0.2959, pruned_loss=0.05649, over 16459.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.305, pruned_loss=0.07089, over 3206742.69 frames. ], batch size: 68, lr: 9.80e-03, grad_scale: 8.0 2023-04-28 16:27:26,554 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65407.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:27:28,724 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65408.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:27:32,341 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65411.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:27:34,118 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1508, 5.1572, 4.9865, 4.7982, 4.6053, 4.9857, 4.9743, 4.7076], device='cuda:6'), covar=tensor([0.0355, 0.0136, 0.0156, 0.0152, 0.0730, 0.0191, 0.0179, 0.0414], device='cuda:6'), in_proj_covar=tensor([0.0197, 0.0229, 0.0237, 0.0212, 0.0267, 0.0235, 0.0167, 0.0267], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:28:07,581 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 16:28:27,210 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65449.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:28:27,512 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7605, 2.2897, 2.4541, 4.6796, 1.9680, 2.7271, 2.3027, 2.4149], device='cuda:6'), covar=tensor([0.0744, 0.2776, 0.1537, 0.0268, 0.3633, 0.1699, 0.2418, 0.3001], device='cuda:6'), in_proj_covar=tensor([0.0341, 0.0358, 0.0297, 0.0322, 0.0390, 0.0391, 0.0320, 0.0425], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:28:31,287 INFO [train.py:904] (6/8) Epoch 7, batch 4550, loss[loss=0.2254, simple_loss=0.3108, pruned_loss=0.06995, over 16776.00 frames. ], tot_loss[loss=0.2246, simple_loss=0.3058, pruned_loss=0.07171, over 3209542.71 frames. ], batch size: 83, lr: 9.80e-03, grad_scale: 8.0 2023-04-28 16:28:38,293 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65456.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:29:21,410 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.690e+02 2.206e+02 2.596e+02 3.051e+02 6.081e+02, threshold=5.193e+02, percent-clipped=1.0 2023-04-28 16:29:41,302 INFO [train.py:904] (6/8) Epoch 7, batch 4600, loss[loss=0.2277, simple_loss=0.3056, pruned_loss=0.07487, over 16499.00 frames. ], tot_loss[loss=0.2248, simple_loss=0.3065, pruned_loss=0.07153, over 3226765.56 frames. ], batch size: 146, lr: 9.79e-03, grad_scale: 8.0 2023-04-28 16:29:41,946 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9866, 2.6108, 2.4092, 3.0677, 2.5901, 3.2153, 1.8268, 2.7051], device='cuda:6'), covar=tensor([0.0968, 0.0414, 0.0865, 0.0097, 0.0179, 0.0360, 0.1102, 0.0609], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0145, 0.0168, 0.0102, 0.0198, 0.0192, 0.0165, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 16:29:56,855 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6612, 2.6721, 1.6863, 2.7297, 2.0949, 2.7615, 2.0217, 2.3703], device='cuda:6'), covar=tensor([0.0205, 0.0354, 0.1241, 0.0097, 0.0623, 0.0461, 0.1081, 0.0550], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0155, 0.0177, 0.0093, 0.0160, 0.0194, 0.0186, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 16:30:14,990 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4001, 1.8780, 2.5896, 3.2272, 3.2378, 3.8028, 2.0589, 3.5418], device='cuda:6'), covar=tensor([0.0084, 0.0283, 0.0164, 0.0115, 0.0112, 0.0058, 0.0272, 0.0059], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0156, 0.0138, 0.0142, 0.0148, 0.0106, 0.0153, 0.0095], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 16:30:24,210 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65532.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:30:50,580 INFO [train.py:904] (6/8) Epoch 7, batch 4650, loss[loss=0.2245, simple_loss=0.2933, pruned_loss=0.07787, over 11441.00 frames. ], tot_loss[loss=0.224, simple_loss=0.3051, pruned_loss=0.0714, over 3211562.66 frames. ], batch size: 246, lr: 9.79e-03, grad_scale: 8.0 2023-04-28 16:31:41,970 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.329e+02 2.212e+02 2.707e+02 3.141e+02 8.468e+02, threshold=5.414e+02, percent-clipped=4.0 2023-04-28 16:31:49,847 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65593.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 16:32:02,429 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8290, 5.2029, 4.8941, 5.0137, 4.6234, 4.5081, 4.6331, 5.2809], device='cuda:6'), covar=tensor([0.0993, 0.0693, 0.0976, 0.0501, 0.0697, 0.0821, 0.0807, 0.0804], device='cuda:6'), in_proj_covar=tensor([0.0428, 0.0552, 0.0469, 0.0361, 0.0349, 0.0369, 0.0458, 0.0407], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:32:03,154 INFO [train.py:904] (6/8) Epoch 7, batch 4700, loss[loss=0.1974, simple_loss=0.2868, pruned_loss=0.05401, over 16852.00 frames. ], tot_loss[loss=0.2211, simple_loss=0.3024, pruned_loss=0.06992, over 3207921.41 frames. ], batch size: 102, lr: 9.78e-03, grad_scale: 8.0 2023-04-28 16:32:45,695 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65632.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:32:56,221 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65641.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:33:12,630 INFO [train.py:904] (6/8) Epoch 7, batch 4750, loss[loss=0.178, simple_loss=0.2575, pruned_loss=0.04923, over 16642.00 frames. ], tot_loss[loss=0.2178, simple_loss=0.299, pruned_loss=0.06824, over 3193080.91 frames. ], batch size: 57, lr: 9.78e-03, grad_scale: 8.0 2023-04-28 16:33:53,641 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65680.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:34:03,333 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.519e+02 2.191e+02 2.638e+02 3.130e+02 6.213e+02, threshold=5.275e+02, percent-clipped=2.0 2023-04-28 16:34:22,789 INFO [train.py:904] (6/8) Epoch 7, batch 4800, loss[loss=0.1968, simple_loss=0.2813, pruned_loss=0.05617, over 16725.00 frames. ], tot_loss[loss=0.2129, simple_loss=0.2946, pruned_loss=0.06559, over 3206386.62 frames. ], batch size: 89, lr: 9.78e-03, grad_scale: 8.0 2023-04-28 16:34:36,983 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65711.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:35:36,916 INFO [train.py:904] (6/8) Epoch 7, batch 4850, loss[loss=0.2028, simple_loss=0.287, pruned_loss=0.05936, over 16668.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2958, pruned_loss=0.06517, over 3200825.32 frames. ], batch size: 62, lr: 9.77e-03, grad_scale: 8.0 2023-04-28 16:35:45,500 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65759.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:36:28,553 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.411e+02 2.781e+02 3.259e+02 6.496e+02, threshold=5.561e+02, percent-clipped=1.0 2023-04-28 16:36:48,038 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1725, 5.4579, 5.1139, 5.2632, 4.9230, 4.7918, 4.9121, 5.4998], device='cuda:6'), covar=tensor([0.0750, 0.0671, 0.0904, 0.0502, 0.0564, 0.0663, 0.0770, 0.0735], device='cuda:6'), in_proj_covar=tensor([0.0425, 0.0552, 0.0466, 0.0360, 0.0347, 0.0368, 0.0453, 0.0403], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:36:48,886 INFO [train.py:904] (6/8) Epoch 7, batch 4900, loss[loss=0.1889, simple_loss=0.2789, pruned_loss=0.04942, over 16806.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2956, pruned_loss=0.06439, over 3186449.00 frames. ], batch size: 83, lr: 9.77e-03, grad_scale: 8.0 2023-04-28 16:37:44,030 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65838.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 16:38:01,719 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4251, 2.0670, 2.2164, 4.1645, 1.8836, 2.6298, 2.1458, 2.3196], device='cuda:6'), covar=tensor([0.0738, 0.2628, 0.1603, 0.0322, 0.3287, 0.1594, 0.2489, 0.2405], device='cuda:6'), in_proj_covar=tensor([0.0333, 0.0348, 0.0293, 0.0317, 0.0385, 0.0381, 0.0314, 0.0414], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:38:02,371 INFO [train.py:904] (6/8) Epoch 7, batch 4950, loss[loss=0.2049, simple_loss=0.2941, pruned_loss=0.05778, over 16622.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2942, pruned_loss=0.06361, over 3195635.21 frames. ], batch size: 83, lr: 9.77e-03, grad_scale: 8.0 2023-04-28 16:38:53,492 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.703e+02 2.669e+02 3.212e+02 3.846e+02 7.226e+02, threshold=6.424e+02, percent-clipped=6.0 2023-04-28 16:38:54,453 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65888.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 16:39:08,408 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65899.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 16:39:12,681 INFO [train.py:904] (6/8) Epoch 7, batch 5000, loss[loss=0.2133, simple_loss=0.3104, pruned_loss=0.05808, over 16423.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2964, pruned_loss=0.06401, over 3209398.67 frames. ], batch size: 146, lr: 9.76e-03, grad_scale: 8.0 2023-04-28 16:40:09,141 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65941.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:40:23,702 INFO [train.py:904] (6/8) Epoch 7, batch 5050, loss[loss=0.2382, simple_loss=0.3213, pruned_loss=0.07756, over 15446.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2968, pruned_loss=0.06377, over 3204710.59 frames. ], batch size: 190, lr: 9.76e-03, grad_scale: 8.0 2023-04-28 16:40:46,956 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-28 16:41:14,576 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.911e+02 2.609e+02 2.967e+02 3.441e+02 6.265e+02, threshold=5.933e+02, percent-clipped=0.0 2023-04-28 16:41:16,169 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65989.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:41:35,930 INFO [train.py:904] (6/8) Epoch 7, batch 5100, loss[loss=0.2019, simple_loss=0.2793, pruned_loss=0.06226, over 17030.00 frames. ], tot_loss[loss=0.2101, simple_loss=0.2947, pruned_loss=0.06274, over 3217469.90 frames. ], batch size: 55, lr: 9.75e-03, grad_scale: 8.0 2023-04-28 16:42:46,401 INFO [train.py:904] (6/8) Epoch 7, batch 5150, loss[loss=0.2194, simple_loss=0.3075, pruned_loss=0.0657, over 16765.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2953, pruned_loss=0.06236, over 3213169.12 frames. ], batch size: 83, lr: 9.75e-03, grad_scale: 8.0 2023-04-28 16:43:37,177 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.675e+02 2.591e+02 3.145e+02 3.855e+02 6.325e+02, threshold=6.290e+02, percent-clipped=3.0 2023-04-28 16:43:55,945 INFO [train.py:904] (6/8) Epoch 7, batch 5200, loss[loss=0.2049, simple_loss=0.2921, pruned_loss=0.05883, over 15407.00 frames. ], tot_loss[loss=0.2091, simple_loss=0.2941, pruned_loss=0.06206, over 3215959.57 frames. ], batch size: 190, lr: 9.75e-03, grad_scale: 8.0 2023-04-28 16:44:56,218 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66144.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:45:06,327 INFO [train.py:904] (6/8) Epoch 7, batch 5250, loss[loss=0.2506, simple_loss=0.3165, pruned_loss=0.09235, over 12292.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2923, pruned_loss=0.06226, over 3215168.28 frames. ], batch size: 246, lr: 9.74e-03, grad_scale: 8.0 2023-04-28 16:45:16,805 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5107, 2.6339, 2.1037, 4.1430, 2.6282, 3.8950, 1.4377, 2.6052], device='cuda:6'), covar=tensor([0.1482, 0.0704, 0.1374, 0.0091, 0.0197, 0.0318, 0.1521, 0.0922], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0148, 0.0170, 0.0103, 0.0196, 0.0195, 0.0167, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 16:45:19,113 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66160.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:45:59,715 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.672e+02 2.502e+02 2.872e+02 3.517e+02 7.118e+02, threshold=5.743e+02, percent-clipped=1.0 2023-04-28 16:46:00,025 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66188.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:46:06,939 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66194.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 16:46:18,392 INFO [train.py:904] (6/8) Epoch 7, batch 5300, loss[loss=0.1862, simple_loss=0.2691, pruned_loss=0.05162, over 16515.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2891, pruned_loss=0.06119, over 3206889.20 frames. ], batch size: 75, lr: 9.74e-03, grad_scale: 8.0 2023-04-28 16:46:23,342 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66205.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:46:45,719 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66221.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 16:47:06,138 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=66236.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 16:47:23,450 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3267, 3.2104, 3.3569, 3.4375, 3.5193, 3.1684, 3.5348, 3.5601], device='cuda:6'), covar=tensor([0.0828, 0.0719, 0.0993, 0.0453, 0.0515, 0.2123, 0.0621, 0.0506], device='cuda:6'), in_proj_covar=tensor([0.0445, 0.0540, 0.0679, 0.0552, 0.0414, 0.0412, 0.0420, 0.0468], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:47:28,949 INFO [train.py:904] (6/8) Epoch 7, batch 5350, loss[loss=0.2175, simple_loss=0.3129, pruned_loss=0.06109, over 16765.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.2872, pruned_loss=0.06043, over 3209513.11 frames. ], batch size: 76, lr: 9.74e-03, grad_scale: 8.0 2023-04-28 16:47:36,958 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66258.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:47:43,354 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-28 16:48:20,491 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.718e+02 2.423e+02 2.781e+02 3.267e+02 5.283e+02, threshold=5.561e+02, percent-clipped=0.0 2023-04-28 16:48:40,524 INFO [train.py:904] (6/8) Epoch 7, batch 5400, loss[loss=0.234, simple_loss=0.3058, pruned_loss=0.08107, over 16253.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.29, pruned_loss=0.06122, over 3205070.65 frames. ], batch size: 35, lr: 9.73e-03, grad_scale: 8.0 2023-04-28 16:49:01,680 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5309, 4.4957, 4.4475, 3.7396, 4.4641, 1.7242, 4.2358, 4.3447], device='cuda:6'), covar=tensor([0.0068, 0.0060, 0.0079, 0.0356, 0.0057, 0.1954, 0.0081, 0.0141], device='cuda:6'), in_proj_covar=tensor([0.0106, 0.0093, 0.0142, 0.0141, 0.0109, 0.0155, 0.0125, 0.0134], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:49:05,945 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66319.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:49:55,998 INFO [train.py:904] (6/8) Epoch 7, batch 5450, loss[loss=0.2952, simple_loss=0.3506, pruned_loss=0.1199, over 11778.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2933, pruned_loss=0.06334, over 3186074.98 frames. ], batch size: 246, lr: 9.73e-03, grad_scale: 8.0 2023-04-28 16:50:00,557 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3608, 1.9416, 2.1867, 3.8775, 1.8917, 2.6497, 2.1683, 2.2427], device='cuda:6'), covar=tensor([0.0744, 0.2685, 0.1504, 0.0350, 0.3262, 0.1549, 0.2350, 0.2465], device='cuda:6'), in_proj_covar=tensor([0.0333, 0.0347, 0.0292, 0.0317, 0.0386, 0.0380, 0.0312, 0.0412], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 16:50:50,980 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.058e+02 3.043e+02 3.833e+02 5.336e+02 9.922e+02, threshold=7.666e+02, percent-clipped=19.0 2023-04-28 16:51:12,547 INFO [train.py:904] (6/8) Epoch 7, batch 5500, loss[loss=0.2732, simple_loss=0.3422, pruned_loss=0.1021, over 15345.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.3027, pruned_loss=0.06957, over 3175372.87 frames. ], batch size: 190, lr: 9.73e-03, grad_scale: 8.0 2023-04-28 16:52:29,859 INFO [train.py:904] (6/8) Epoch 7, batch 5550, loss[loss=0.2288, simple_loss=0.312, pruned_loss=0.07274, over 16994.00 frames. ], tot_loss[loss=0.2316, simple_loss=0.3109, pruned_loss=0.07619, over 3144045.55 frames. ], batch size: 55, lr: 9.72e-03, grad_scale: 8.0 2023-04-28 16:53:27,053 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.385e+02 3.933e+02 4.760e+02 6.147e+02 1.111e+03, threshold=9.520e+02, percent-clipped=8.0 2023-04-28 16:53:36,378 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66494.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:53:38,474 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9538, 2.5975, 2.6122, 1.8130, 2.7202, 2.7332, 2.4113, 2.3313], device='cuda:6'), covar=tensor([0.0744, 0.0180, 0.0170, 0.0928, 0.0101, 0.0149, 0.0409, 0.0428], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0097, 0.0082, 0.0141, 0.0072, 0.0087, 0.0121, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 16:53:47,101 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66500.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:53:49,152 INFO [train.py:904] (6/8) Epoch 7, batch 5600, loss[loss=0.2691, simple_loss=0.3374, pruned_loss=0.1004, over 16757.00 frames. ], tot_loss[loss=0.2392, simple_loss=0.3166, pruned_loss=0.08087, over 3137320.01 frames. ], batch size: 124, lr: 9.72e-03, grad_scale: 8.0 2023-04-28 16:54:11,813 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66516.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 16:54:21,731 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66521.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:54:54,214 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=66542.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:55:08,328 INFO [train.py:904] (6/8) Epoch 7, batch 5650, loss[loss=0.2463, simple_loss=0.3297, pruned_loss=0.08146, over 16878.00 frames. ], tot_loss[loss=0.2495, simple_loss=0.324, pruned_loss=0.0875, over 3073399.52 frames. ], batch size: 96, lr: 9.71e-03, grad_scale: 8.0 2023-04-28 16:55:19,133 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3862, 4.5765, 4.7077, 4.6071, 4.6142, 5.1265, 4.6714, 4.4308], device='cuda:6'), covar=tensor([0.1142, 0.1618, 0.1554, 0.1516, 0.2335, 0.0946, 0.1252, 0.2408], device='cuda:6'), in_proj_covar=tensor([0.0306, 0.0433, 0.0436, 0.0369, 0.0499, 0.0470, 0.0348, 0.0507], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 16:55:53,694 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66582.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 16:56:03,089 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.602e+02 4.256e+02 5.191e+02 6.863e+02 2.150e+03, threshold=1.038e+03, percent-clipped=9.0 2023-04-28 16:56:23,902 INFO [train.py:904] (6/8) Epoch 7, batch 5700, loss[loss=0.2361, simple_loss=0.3291, pruned_loss=0.07154, over 16699.00 frames. ], tot_loss[loss=0.2506, simple_loss=0.3247, pruned_loss=0.08824, over 3088665.32 frames. ], batch size: 89, lr: 9.71e-03, grad_scale: 8.0 2023-04-28 16:56:42,483 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66614.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:57:41,027 INFO [train.py:904] (6/8) Epoch 7, batch 5750, loss[loss=0.2616, simple_loss=0.317, pruned_loss=0.1031, over 10901.00 frames. ], tot_loss[loss=0.2535, simple_loss=0.3272, pruned_loss=0.08985, over 3051488.75 frames. ], batch size: 248, lr: 9.71e-03, grad_scale: 8.0 2023-04-28 16:58:39,454 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.231e+02 3.661e+02 4.395e+02 5.448e+02 7.543e+02, threshold=8.790e+02, percent-clipped=0.0 2023-04-28 16:59:02,126 INFO [train.py:904] (6/8) Epoch 7, batch 5800, loss[loss=0.2643, simple_loss=0.3355, pruned_loss=0.09655, over 16661.00 frames. ], tot_loss[loss=0.2525, simple_loss=0.3271, pruned_loss=0.08894, over 3035326.05 frames. ], batch size: 57, lr: 9.70e-03, grad_scale: 16.0 2023-04-28 16:59:48,391 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66731.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:00:19,932 INFO [train.py:904] (6/8) Epoch 7, batch 5850, loss[loss=0.2147, simple_loss=0.3039, pruned_loss=0.06281, over 16773.00 frames. ], tot_loss[loss=0.2479, simple_loss=0.3238, pruned_loss=0.086, over 3055115.71 frames. ], batch size: 124, lr: 9.70e-03, grad_scale: 8.0 2023-04-28 17:01:20,001 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.089e+02 3.451e+02 4.238e+02 5.428e+02 1.064e+03, threshold=8.477e+02, percent-clipped=4.0 2023-04-28 17:01:25,739 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66792.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 17:01:38,961 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66800.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:01:41,100 INFO [train.py:904] (6/8) Epoch 7, batch 5900, loss[loss=0.233, simple_loss=0.3155, pruned_loss=0.07531, over 16761.00 frames. ], tot_loss[loss=0.2473, simple_loss=0.3233, pruned_loss=0.08561, over 3062006.32 frames. ], batch size: 76, lr: 9.70e-03, grad_scale: 8.0 2023-04-28 17:01:42,840 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.96 vs. limit=5.0 2023-04-28 17:02:06,002 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0073, 3.7998, 4.0601, 4.2397, 4.3405, 3.8768, 4.3136, 4.3357], device='cuda:6'), covar=tensor([0.1212, 0.1009, 0.1344, 0.0527, 0.0474, 0.1169, 0.0512, 0.0475], device='cuda:6'), in_proj_covar=tensor([0.0443, 0.0540, 0.0674, 0.0544, 0.0418, 0.0415, 0.0422, 0.0469], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:02:08,249 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66816.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:02:45,062 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-28 17:02:56,753 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=66848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:03:02,863 INFO [train.py:904] (6/8) Epoch 7, batch 5950, loss[loss=0.2476, simple_loss=0.3273, pruned_loss=0.08399, over 15303.00 frames. ], tot_loss[loss=0.2454, simple_loss=0.323, pruned_loss=0.08388, over 3050013.29 frames. ], batch size: 190, lr: 9.69e-03, grad_scale: 8.0 2023-04-28 17:03:23,048 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=66864.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 17:03:32,657 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1283, 3.6006, 3.5659, 2.3841, 3.3109, 3.5302, 3.4657, 1.9468], device='cuda:6'), covar=tensor([0.0351, 0.0028, 0.0033, 0.0264, 0.0047, 0.0064, 0.0036, 0.0313], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0058, 0.0059, 0.0117, 0.0066, 0.0077, 0.0067, 0.0111], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 17:03:41,755 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66877.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:04:00,333 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.040e+02 3.507e+02 4.238e+02 5.487e+02 1.167e+03, threshold=8.477e+02, percent-clipped=3.0 2023-04-28 17:04:22,267 INFO [train.py:904] (6/8) Epoch 7, batch 6000, loss[loss=0.2173, simple_loss=0.2867, pruned_loss=0.07399, over 16601.00 frames. ], tot_loss[loss=0.244, simple_loss=0.3218, pruned_loss=0.08313, over 3070204.36 frames. ], batch size: 57, lr: 9.69e-03, grad_scale: 8.0 2023-04-28 17:04:22,267 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 17:04:32,878 INFO [train.py:938] (6/8) Epoch 7, validation: loss=0.1758, simple_loss=0.2891, pruned_loss=0.03127, over 944034.00 frames. 2023-04-28 17:04:32,879 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-28 17:04:51,898 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66914.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:05:18,084 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-28 17:05:53,525 INFO [train.py:904] (6/8) Epoch 7, batch 6050, loss[loss=0.2419, simple_loss=0.3182, pruned_loss=0.08278, over 15430.00 frames. ], tot_loss[loss=0.2425, simple_loss=0.3204, pruned_loss=0.08235, over 3079099.62 frames. ], batch size: 190, lr: 9.69e-03, grad_scale: 8.0 2023-04-28 17:05:54,528 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2618, 1.4268, 1.8851, 2.1591, 2.2873, 2.4141, 1.4547, 2.3808], device='cuda:6'), covar=tensor([0.0105, 0.0270, 0.0157, 0.0176, 0.0141, 0.0094, 0.0287, 0.0080], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0155, 0.0141, 0.0140, 0.0147, 0.0103, 0.0156, 0.0095], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 17:06:02,621 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66957.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:06:09,581 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=66962.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:06:50,891 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.964e+02 3.435e+02 4.153e+02 5.087e+02 8.285e+02, threshold=8.306e+02, percent-clipped=0.0 2023-04-28 17:07:12,264 INFO [train.py:904] (6/8) Epoch 7, batch 6100, loss[loss=0.2812, simple_loss=0.3388, pruned_loss=0.1118, over 11138.00 frames. ], tot_loss[loss=0.2411, simple_loss=0.3194, pruned_loss=0.08141, over 3074016.45 frames. ], batch size: 248, lr: 9.68e-03, grad_scale: 8.0 2023-04-28 17:07:39,915 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67018.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 17:07:52,757 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-28 17:08:32,436 INFO [train.py:904] (6/8) Epoch 7, batch 6150, loss[loss=0.2479, simple_loss=0.3232, pruned_loss=0.08629, over 16262.00 frames. ], tot_loss[loss=0.2399, simple_loss=0.3178, pruned_loss=0.08099, over 3075749.71 frames. ], batch size: 165, lr: 9.68e-03, grad_scale: 8.0 2023-04-28 17:09:23,228 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67083.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:09:29,304 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67087.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 17:09:31,359 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.004e+02 3.456e+02 4.175e+02 5.279e+02 9.590e+02, threshold=8.351e+02, percent-clipped=3.0 2023-04-28 17:09:51,771 INFO [train.py:904] (6/8) Epoch 7, batch 6200, loss[loss=0.2606, simple_loss=0.3364, pruned_loss=0.09247, over 16599.00 frames. ], tot_loss[loss=0.2397, simple_loss=0.3167, pruned_loss=0.0813, over 3058662.85 frames. ], batch size: 62, lr: 9.67e-03, grad_scale: 4.0 2023-04-28 17:09:53,371 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-28 17:11:02,461 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67144.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:11:14,124 INFO [train.py:904] (6/8) Epoch 7, batch 6250, loss[loss=0.2036, simple_loss=0.3016, pruned_loss=0.05282, over 16894.00 frames. ], tot_loss[loss=0.2392, simple_loss=0.3163, pruned_loss=0.081, over 3070988.53 frames. ], batch size: 96, lr: 9.67e-03, grad_scale: 4.0 2023-04-28 17:11:14,641 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1573, 3.3585, 3.5124, 1.7781, 3.6989, 3.7762, 2.7895, 2.8499], device='cuda:6'), covar=tensor([0.0907, 0.0173, 0.0179, 0.1175, 0.0060, 0.0084, 0.0382, 0.0405], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0095, 0.0081, 0.0137, 0.0070, 0.0085, 0.0118, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 17:11:52,554 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67177.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:12:11,783 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.341e+02 3.226e+02 4.153e+02 4.983e+02 8.969e+02, threshold=8.306e+02, percent-clipped=3.0 2023-04-28 17:12:30,906 INFO [train.py:904] (6/8) Epoch 7, batch 6300, loss[loss=0.2216, simple_loss=0.3072, pruned_loss=0.06799, over 16796.00 frames. ], tot_loss[loss=0.2372, simple_loss=0.3158, pruned_loss=0.07927, over 3108146.07 frames. ], batch size: 83, lr: 9.67e-03, grad_scale: 4.0 2023-04-28 17:12:49,618 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5089, 2.0891, 1.6072, 1.8033, 2.4223, 2.1189, 2.4749, 2.6178], device='cuda:6'), covar=tensor([0.0067, 0.0206, 0.0305, 0.0284, 0.0118, 0.0200, 0.0112, 0.0125], device='cuda:6'), in_proj_covar=tensor([0.0099, 0.0172, 0.0169, 0.0168, 0.0165, 0.0171, 0.0164, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:12:53,938 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7655, 5.1061, 4.8631, 4.8196, 4.6037, 4.4947, 4.5795, 5.1488], device='cuda:6'), covar=tensor([0.0717, 0.0609, 0.0758, 0.0481, 0.0581, 0.0780, 0.0724, 0.0658], device='cuda:6'), in_proj_covar=tensor([0.0438, 0.0553, 0.0475, 0.0362, 0.0347, 0.0369, 0.0461, 0.0410], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:13:08,747 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=67225.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 17:13:49,867 INFO [train.py:904] (6/8) Epoch 7, batch 6350, loss[loss=0.2455, simple_loss=0.3199, pruned_loss=0.08554, over 16178.00 frames. ], tot_loss[loss=0.2398, simple_loss=0.3173, pruned_loss=0.08117, over 3101755.85 frames. ], batch size: 165, lr: 9.66e-03, grad_scale: 4.0 2023-04-28 17:14:31,088 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-28 17:14:47,871 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.015e+02 3.646e+02 4.526e+02 5.415e+02 1.244e+03, threshold=9.052e+02, percent-clipped=4.0 2023-04-28 17:15:05,416 INFO [train.py:904] (6/8) Epoch 7, batch 6400, loss[loss=0.2307, simple_loss=0.3105, pruned_loss=0.0755, over 17104.00 frames. ], tot_loss[loss=0.2409, simple_loss=0.3177, pruned_loss=0.08202, over 3099958.29 frames. ], batch size: 49, lr: 9.66e-03, grad_scale: 8.0 2023-04-28 17:15:22,394 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67313.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:15:24,889 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9090, 3.7331, 3.9570, 4.1101, 4.2105, 3.8049, 4.1616, 4.1534], device='cuda:6'), covar=tensor([0.1147, 0.0894, 0.1247, 0.0538, 0.0511, 0.1289, 0.0568, 0.0539], device='cuda:6'), in_proj_covar=tensor([0.0439, 0.0539, 0.0671, 0.0541, 0.0414, 0.0408, 0.0421, 0.0471], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:16:20,298 INFO [train.py:904] (6/8) Epoch 7, batch 6450, loss[loss=0.2343, simple_loss=0.2953, pruned_loss=0.08669, over 11588.00 frames. ], tot_loss[loss=0.2397, simple_loss=0.3171, pruned_loss=0.0811, over 3101410.13 frames. ], batch size: 247, lr: 9.66e-03, grad_scale: 4.0 2023-04-28 17:17:16,539 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67387.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 17:17:22,057 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.155e+02 3.316e+02 4.009e+02 5.317e+02 1.482e+03, threshold=8.018e+02, percent-clipped=5.0 2023-04-28 17:17:38,435 INFO [train.py:904] (6/8) Epoch 7, batch 6500, loss[loss=0.2397, simple_loss=0.3175, pruned_loss=0.08093, over 15423.00 frames. ], tot_loss[loss=0.237, simple_loss=0.3142, pruned_loss=0.07987, over 3090606.00 frames. ], batch size: 190, lr: 9.65e-03, grad_scale: 4.0 2023-04-28 17:18:18,256 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9671, 2.6064, 2.6030, 1.8371, 2.7881, 2.7948, 2.3682, 2.3705], device='cuda:6'), covar=tensor([0.0702, 0.0159, 0.0171, 0.0870, 0.0085, 0.0157, 0.0392, 0.0367], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0094, 0.0082, 0.0138, 0.0070, 0.0086, 0.0118, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 17:18:23,282 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4523, 3.2452, 2.6555, 2.1263, 2.3012, 2.1222, 3.2092, 3.2407], device='cuda:6'), covar=tensor([0.2279, 0.0739, 0.1347, 0.1898, 0.1961, 0.1610, 0.0540, 0.0879], device='cuda:6'), in_proj_covar=tensor([0.0286, 0.0246, 0.0270, 0.0257, 0.0277, 0.0206, 0.0247, 0.0266], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:18:29,453 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=67435.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:18:36,524 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67439.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:18:57,738 INFO [train.py:904] (6/8) Epoch 7, batch 6550, loss[loss=0.2932, simple_loss=0.3536, pruned_loss=0.1164, over 11811.00 frames. ], tot_loss[loss=0.2394, simple_loss=0.3176, pruned_loss=0.08059, over 3095408.97 frames. ], batch size: 247, lr: 9.65e-03, grad_scale: 4.0 2023-04-28 17:19:56,001 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.411e+02 3.629e+02 4.533e+02 5.660e+02 1.346e+03, threshold=9.066e+02, percent-clipped=11.0 2023-04-28 17:20:13,100 INFO [train.py:904] (6/8) Epoch 7, batch 6600, loss[loss=0.2345, simple_loss=0.317, pruned_loss=0.07598, over 16658.00 frames. ], tot_loss[loss=0.2419, simple_loss=0.3205, pruned_loss=0.0817, over 3097643.90 frames. ], batch size: 89, lr: 9.65e-03, grad_scale: 4.0 2023-04-28 17:20:30,856 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67513.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:21:10,571 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.22 vs. limit=5.0 2023-04-28 17:21:29,932 INFO [train.py:904] (6/8) Epoch 7, batch 6650, loss[loss=0.3231, simple_loss=0.3677, pruned_loss=0.1392, over 11331.00 frames. ], tot_loss[loss=0.2422, simple_loss=0.3204, pruned_loss=0.08201, over 3107561.46 frames. ], batch size: 250, lr: 9.64e-03, grad_scale: 4.0 2023-04-28 17:22:02,865 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67574.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:22:27,319 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.256e+02 3.378e+02 3.968e+02 4.972e+02 1.148e+03, threshold=7.935e+02, percent-clipped=1.0 2023-04-28 17:22:43,252 INFO [train.py:904] (6/8) Epoch 7, batch 6700, loss[loss=0.225, simple_loss=0.3086, pruned_loss=0.07072, over 16452.00 frames. ], tot_loss[loss=0.2406, simple_loss=0.3181, pruned_loss=0.08151, over 3114724.60 frames. ], batch size: 146, lr: 9.64e-03, grad_scale: 4.0 2023-04-28 17:23:00,348 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67613.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:23:04,312 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-28 17:23:56,588 INFO [train.py:904] (6/8) Epoch 7, batch 6750, loss[loss=0.2322, simple_loss=0.3019, pruned_loss=0.08126, over 16310.00 frames. ], tot_loss[loss=0.2396, simple_loss=0.3169, pruned_loss=0.0812, over 3115592.02 frames. ], batch size: 35, lr: 9.64e-03, grad_scale: 4.0 2023-04-28 17:24:07,908 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-28 17:24:11,054 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=67661.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:24:37,942 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67680.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:24:45,709 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2818, 3.6093, 3.6536, 1.4526, 3.8894, 3.9415, 2.9258, 2.8499], device='cuda:6'), covar=tensor([0.0847, 0.0149, 0.0167, 0.1408, 0.0061, 0.0094, 0.0395, 0.0427], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0095, 0.0083, 0.0140, 0.0071, 0.0087, 0.0120, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 17:24:54,246 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.424e+02 3.570e+02 4.418e+02 5.557e+02 1.177e+03, threshold=8.835e+02, percent-clipped=2.0 2023-04-28 17:24:55,937 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67692.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:25:02,963 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-28 17:25:10,437 INFO [train.py:904] (6/8) Epoch 7, batch 6800, loss[loss=0.2479, simple_loss=0.3289, pruned_loss=0.08342, over 16920.00 frames. ], tot_loss[loss=0.2397, simple_loss=0.3171, pruned_loss=0.0811, over 3121408.22 frames. ], batch size: 109, lr: 9.63e-03, grad_scale: 8.0 2023-04-28 17:26:06,879 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67739.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:26:10,696 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67741.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:26:23,541 INFO [train.py:904] (6/8) Epoch 7, batch 6850, loss[loss=0.2501, simple_loss=0.335, pruned_loss=0.08266, over 16869.00 frames. ], tot_loss[loss=0.2409, simple_loss=0.3182, pruned_loss=0.08176, over 3109386.50 frames. ], batch size: 116, lr: 9.63e-03, grad_scale: 8.0 2023-04-28 17:26:26,408 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67753.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 17:26:36,942 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7707, 5.0724, 5.2488, 5.0620, 5.0062, 5.6907, 5.1535, 4.8616], device='cuda:6'), covar=tensor([0.0875, 0.1550, 0.1485, 0.1654, 0.2845, 0.0977, 0.1326, 0.2198], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0435, 0.0451, 0.0374, 0.0506, 0.0479, 0.0357, 0.0514], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 17:27:14,417 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=67787.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:27:20,977 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.331e+02 3.323e+02 3.993e+02 4.819e+02 7.017e+02, threshold=7.986e+02, percent-clipped=0.0 2023-04-28 17:27:35,614 INFO [train.py:904] (6/8) Epoch 7, batch 6900, loss[loss=0.2659, simple_loss=0.3492, pruned_loss=0.09135, over 16779.00 frames. ], tot_loss[loss=0.2423, simple_loss=0.321, pruned_loss=0.08184, over 3111141.99 frames. ], batch size: 102, lr: 9.63e-03, grad_scale: 2.0 2023-04-28 17:27:41,195 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0961, 2.9808, 2.7222, 2.0305, 2.6350, 2.1379, 2.8109, 2.9392], device='cuda:6'), covar=tensor([0.0275, 0.0454, 0.0459, 0.1420, 0.0633, 0.0883, 0.0416, 0.0518], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0130, 0.0155, 0.0140, 0.0133, 0.0125, 0.0135, 0.0140], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 17:28:20,778 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2048, 4.4817, 4.7289, 2.1923, 4.9944, 5.0193, 3.4404, 3.6498], device='cuda:6'), covar=tensor([0.0652, 0.0129, 0.0134, 0.1166, 0.0039, 0.0047, 0.0313, 0.0364], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0095, 0.0083, 0.0140, 0.0071, 0.0087, 0.0120, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 17:28:46,543 INFO [train.py:904] (6/8) Epoch 7, batch 6950, loss[loss=0.3101, simple_loss=0.3516, pruned_loss=0.1343, over 11213.00 frames. ], tot_loss[loss=0.2453, simple_loss=0.3229, pruned_loss=0.08385, over 3108793.49 frames. ], batch size: 246, lr: 9.62e-03, grad_scale: 2.0 2023-04-28 17:29:13,151 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67869.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 17:29:23,604 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-04-28 17:29:46,383 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.351e+02 3.618e+02 4.486e+02 5.724e+02 1.200e+03, threshold=8.972e+02, percent-clipped=9.0 2023-04-28 17:29:59,774 INFO [train.py:904] (6/8) Epoch 7, batch 7000, loss[loss=0.2121, simple_loss=0.3037, pruned_loss=0.06019, over 17000.00 frames. ], tot_loss[loss=0.2446, simple_loss=0.323, pruned_loss=0.0831, over 3092521.94 frames. ], batch size: 41, lr: 9.62e-03, grad_scale: 2.0 2023-04-28 17:30:33,529 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.7757, 6.1309, 5.7586, 6.0022, 5.3986, 5.0790, 5.6815, 6.2657], device='cuda:6'), covar=tensor([0.0829, 0.0659, 0.1117, 0.0483, 0.0703, 0.0587, 0.0729, 0.0659], device='cuda:6'), in_proj_covar=tensor([0.0449, 0.0560, 0.0484, 0.0371, 0.0355, 0.0381, 0.0468, 0.0417], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:31:13,283 INFO [train.py:904] (6/8) Epoch 7, batch 7050, loss[loss=0.2759, simple_loss=0.3391, pruned_loss=0.1063, over 15382.00 frames. ], tot_loss[loss=0.2439, simple_loss=0.3232, pruned_loss=0.08231, over 3095353.26 frames. ], batch size: 190, lr: 9.61e-03, grad_scale: 2.0 2023-04-28 17:31:57,441 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67981.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:32:14,208 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.912e+02 3.870e+02 4.768e+02 5.699e+02 1.240e+03, threshold=9.537e+02, percent-clipped=2.0 2023-04-28 17:32:29,898 INFO [train.py:904] (6/8) Epoch 7, batch 7100, loss[loss=0.2524, simple_loss=0.3304, pruned_loss=0.08716, over 16214.00 frames. ], tot_loss[loss=0.2426, simple_loss=0.3213, pruned_loss=0.08197, over 3082818.61 frames. ], batch size: 165, lr: 9.61e-03, grad_scale: 2.0 2023-04-28 17:33:21,414 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68036.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:33:30,270 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68042.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:33:39,220 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68048.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 17:33:44,541 INFO [train.py:904] (6/8) Epoch 7, batch 7150, loss[loss=0.2468, simple_loss=0.3232, pruned_loss=0.08518, over 16338.00 frames. ], tot_loss[loss=0.2407, simple_loss=0.319, pruned_loss=0.08124, over 3109604.66 frames. ], batch size: 165, lr: 9.61e-03, grad_scale: 2.0 2023-04-28 17:34:03,709 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-28 17:34:37,158 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68087.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 17:34:44,522 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.291e+02 3.518e+02 4.375e+02 5.869e+02 1.278e+03, threshold=8.749e+02, percent-clipped=4.0 2023-04-28 17:34:58,212 INFO [train.py:904] (6/8) Epoch 7, batch 7200, loss[loss=0.2491, simple_loss=0.3214, pruned_loss=0.08843, over 11499.00 frames. ], tot_loss[loss=0.2384, simple_loss=0.317, pruned_loss=0.07995, over 3083373.22 frames. ], batch size: 247, lr: 9.60e-03, grad_scale: 4.0 2023-04-28 17:35:20,176 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3477, 3.2348, 3.3570, 3.4567, 3.4854, 3.2314, 3.4547, 3.5299], device='cuda:6'), covar=tensor([0.0910, 0.0735, 0.0959, 0.0481, 0.0541, 0.2002, 0.0739, 0.0513], device='cuda:6'), in_proj_covar=tensor([0.0432, 0.0532, 0.0664, 0.0533, 0.0406, 0.0406, 0.0424, 0.0460], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:36:01,310 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0210, 2.3860, 1.8395, 2.0837, 2.7023, 2.4369, 2.9833, 2.9640], device='cuda:6'), covar=tensor([0.0051, 0.0211, 0.0281, 0.0258, 0.0133, 0.0207, 0.0113, 0.0120], device='cuda:6'), in_proj_covar=tensor([0.0101, 0.0173, 0.0174, 0.0172, 0.0169, 0.0177, 0.0166, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:36:11,125 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68148.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 17:36:16,039 INFO [train.py:904] (6/8) Epoch 7, batch 7250, loss[loss=0.2413, simple_loss=0.3029, pruned_loss=0.08985, over 11617.00 frames. ], tot_loss[loss=0.2354, simple_loss=0.3143, pruned_loss=0.07821, over 3091025.90 frames. ], batch size: 247, lr: 9.60e-03, grad_scale: 4.0 2023-04-28 17:36:19,965 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.17 vs. limit=5.0 2023-04-28 17:36:41,752 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68169.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:37:16,279 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.105e+02 3.214e+02 3.991e+02 5.016e+02 1.051e+03, threshold=7.982e+02, percent-clipped=2.0 2023-04-28 17:37:29,946 INFO [train.py:904] (6/8) Epoch 7, batch 7300, loss[loss=0.2235, simple_loss=0.3095, pruned_loss=0.06876, over 16643.00 frames. ], tot_loss[loss=0.2345, simple_loss=0.3131, pruned_loss=0.07792, over 3078979.94 frames. ], batch size: 134, lr: 9.60e-03, grad_scale: 4.0 2023-04-28 17:37:52,949 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=68217.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:37:59,685 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-28 17:38:44,625 INFO [train.py:904] (6/8) Epoch 7, batch 7350, loss[loss=0.2229, simple_loss=0.3047, pruned_loss=0.07052, over 16505.00 frames. ], tot_loss[loss=0.235, simple_loss=0.3138, pruned_loss=0.07804, over 3077834.24 frames. ], batch size: 35, lr: 9.59e-03, grad_scale: 4.0 2023-04-28 17:39:32,900 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9823, 2.6154, 2.6116, 1.7469, 2.7965, 2.7949, 2.3451, 2.3699], device='cuda:6'), covar=tensor([0.0662, 0.0166, 0.0173, 0.0907, 0.0083, 0.0126, 0.0393, 0.0365], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0094, 0.0082, 0.0138, 0.0069, 0.0085, 0.0119, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 17:39:48,567 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.908e+02 3.085e+02 3.772e+02 4.630e+02 1.239e+03, threshold=7.544e+02, percent-clipped=2.0 2023-04-28 17:40:02,136 INFO [train.py:904] (6/8) Epoch 7, batch 7400, loss[loss=0.3044, simple_loss=0.3544, pruned_loss=0.1272, over 11484.00 frames. ], tot_loss[loss=0.2372, simple_loss=0.3159, pruned_loss=0.07924, over 3081506.59 frames. ], batch size: 247, lr: 9.59e-03, grad_scale: 4.0 2023-04-28 17:40:53,707 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68336.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:40:54,927 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68337.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:41:14,126 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68348.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:41:19,088 INFO [train.py:904] (6/8) Epoch 7, batch 7450, loss[loss=0.2362, simple_loss=0.3205, pruned_loss=0.07593, over 16683.00 frames. ], tot_loss[loss=0.238, simple_loss=0.3164, pruned_loss=0.07978, over 3088851.60 frames. ], batch size: 134, lr: 9.59e-03, grad_scale: 4.0 2023-04-28 17:42:11,321 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=68384.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:42:13,295 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5381, 3.6199, 1.7974, 3.8522, 2.4485, 3.8805, 1.9575, 2.6405], device='cuda:6'), covar=tensor([0.0177, 0.0289, 0.1655, 0.0090, 0.0882, 0.0418, 0.1604, 0.0723], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0154, 0.0179, 0.0095, 0.0162, 0.0192, 0.0188, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 17:42:26,329 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.512e+02 3.665e+02 4.367e+02 5.646e+02 8.920e+02, threshold=8.734e+02, percent-clipped=5.0 2023-04-28 17:42:30,796 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=68396.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:42:35,735 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68399.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:42:39,840 INFO [train.py:904] (6/8) Epoch 7, batch 7500, loss[loss=0.2389, simple_loss=0.3185, pruned_loss=0.07972, over 16833.00 frames. ], tot_loss[loss=0.2378, simple_loss=0.3169, pruned_loss=0.07938, over 3084282.23 frames. ], batch size: 116, lr: 9.58e-03, grad_scale: 4.0 2023-04-28 17:43:44,358 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68443.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 17:43:52,994 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-28 17:43:57,923 INFO [train.py:904] (6/8) Epoch 7, batch 7550, loss[loss=0.2203, simple_loss=0.3033, pruned_loss=0.06863, over 16742.00 frames. ], tot_loss[loss=0.239, simple_loss=0.3169, pruned_loss=0.08054, over 3074155.17 frames. ], batch size: 83, lr: 9.58e-03, grad_scale: 4.0 2023-04-28 17:44:10,825 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68460.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:44:59,737 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.309e+02 3.638e+02 4.521e+02 5.512e+02 1.190e+03, threshold=9.043e+02, percent-clipped=2.0 2023-04-28 17:45:12,770 INFO [train.py:904] (6/8) Epoch 7, batch 7600, loss[loss=0.2117, simple_loss=0.2958, pruned_loss=0.06381, over 17194.00 frames. ], tot_loss[loss=0.238, simple_loss=0.3157, pruned_loss=0.08014, over 3083482.45 frames. ], batch size: 46, lr: 9.58e-03, grad_scale: 8.0 2023-04-28 17:46:28,266 INFO [train.py:904] (6/8) Epoch 7, batch 7650, loss[loss=0.2489, simple_loss=0.3309, pruned_loss=0.08345, over 16366.00 frames. ], tot_loss[loss=0.2405, simple_loss=0.317, pruned_loss=0.08197, over 3043634.43 frames. ], batch size: 146, lr: 9.57e-03, grad_scale: 8.0 2023-04-28 17:47:08,453 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68578.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:47:29,673 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.494e+02 3.559e+02 4.201e+02 5.687e+02 1.184e+03, threshold=8.403e+02, percent-clipped=4.0 2023-04-28 17:47:42,699 INFO [train.py:904] (6/8) Epoch 7, batch 7700, loss[loss=0.262, simple_loss=0.3357, pruned_loss=0.09414, over 16325.00 frames. ], tot_loss[loss=0.241, simple_loss=0.317, pruned_loss=0.08252, over 3041954.59 frames. ], batch size: 146, lr: 9.57e-03, grad_scale: 8.0 2023-04-28 17:48:35,293 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68637.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:48:38,391 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68639.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:48:43,249 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1924, 5.1060, 4.9972, 4.7598, 4.5256, 4.9593, 4.9454, 4.7076], device='cuda:6'), covar=tensor([0.0506, 0.0436, 0.0249, 0.0264, 0.0943, 0.0409, 0.0289, 0.0579], device='cuda:6'), in_proj_covar=tensor([0.0203, 0.0236, 0.0235, 0.0209, 0.0265, 0.0238, 0.0165, 0.0272], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:48:57,123 INFO [train.py:904] (6/8) Epoch 7, batch 7750, loss[loss=0.281, simple_loss=0.336, pruned_loss=0.1129, over 16796.00 frames. ], tot_loss[loss=0.2405, simple_loss=0.3173, pruned_loss=0.08181, over 3072864.49 frames. ], batch size: 39, lr: 9.57e-03, grad_scale: 8.0 2023-04-28 17:49:44,494 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=68685.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:49:44,588 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68685.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:49:57,425 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.246e+02 3.570e+02 4.272e+02 5.150e+02 9.085e+02, threshold=8.545e+02, percent-clipped=2.0 2023-04-28 17:50:11,059 INFO [train.py:904] (6/8) Epoch 7, batch 7800, loss[loss=0.2986, simple_loss=0.3483, pruned_loss=0.1244, over 11520.00 frames. ], tot_loss[loss=0.2419, simple_loss=0.3181, pruned_loss=0.08286, over 3063001.52 frames. ], batch size: 246, lr: 9.56e-03, grad_scale: 8.0 2023-04-28 17:50:11,606 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5752, 3.7029, 1.4458, 4.0287, 2.4512, 4.0555, 1.6949, 2.8225], device='cuda:6'), covar=tensor([0.0194, 0.0306, 0.2138, 0.0126, 0.0880, 0.0343, 0.1940, 0.0668], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0157, 0.0180, 0.0097, 0.0165, 0.0195, 0.0189, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 17:50:43,274 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0821, 1.2386, 1.6838, 1.9653, 2.0490, 2.1861, 1.5924, 2.2023], device='cuda:6'), covar=tensor([0.0112, 0.0319, 0.0170, 0.0195, 0.0181, 0.0112, 0.0280, 0.0085], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0155, 0.0136, 0.0136, 0.0147, 0.0102, 0.0154, 0.0092], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 17:51:13,401 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68743.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:51:17,559 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68746.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:51:25,160 INFO [train.py:904] (6/8) Epoch 7, batch 7850, loss[loss=0.2699, simple_loss=0.352, pruned_loss=0.09384, over 15221.00 frames. ], tot_loss[loss=0.2427, simple_loss=0.3195, pruned_loss=0.08296, over 3058417.01 frames. ], batch size: 190, lr: 9.56e-03, grad_scale: 8.0 2023-04-28 17:51:30,857 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68755.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:51:48,471 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7591, 4.8077, 5.2924, 5.2383, 5.2709, 4.8699, 4.8859, 4.5679], device='cuda:6'), covar=tensor([0.0239, 0.0332, 0.0268, 0.0379, 0.0387, 0.0256, 0.0752, 0.0356], device='cuda:6'), in_proj_covar=tensor([0.0270, 0.0270, 0.0272, 0.0264, 0.0315, 0.0292, 0.0388, 0.0239], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 17:51:57,792 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-28 17:52:24,298 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=68791.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:52:24,436 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9922, 3.0451, 3.1605, 1.6928, 3.3458, 3.3527, 2.6312, 2.5154], device='cuda:6'), covar=tensor([0.0835, 0.0173, 0.0136, 0.1030, 0.0060, 0.0111, 0.0383, 0.0461], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0093, 0.0081, 0.0137, 0.0068, 0.0085, 0.0118, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 17:52:26,860 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.225e+02 3.517e+02 4.266e+02 5.466e+02 1.247e+03, threshold=8.532e+02, percent-clipped=3.0 2023-04-28 17:52:40,546 INFO [train.py:904] (6/8) Epoch 7, batch 7900, loss[loss=0.2234, simple_loss=0.2983, pruned_loss=0.07428, over 17113.00 frames. ], tot_loss[loss=0.2412, simple_loss=0.3181, pruned_loss=0.08221, over 3060448.70 frames. ], batch size: 48, lr: 9.56e-03, grad_scale: 8.0 2023-04-28 17:53:58,005 INFO [train.py:904] (6/8) Epoch 7, batch 7950, loss[loss=0.2824, simple_loss=0.3318, pruned_loss=0.1165, over 11438.00 frames. ], tot_loss[loss=0.2416, simple_loss=0.3182, pruned_loss=0.0825, over 3068071.02 frames. ], batch size: 248, lr: 9.55e-03, grad_scale: 2.0 2023-04-28 17:54:03,792 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9657, 4.0185, 3.8126, 3.7291, 3.5333, 3.9161, 3.5857, 3.6532], device='cuda:6'), covar=tensor([0.0491, 0.0401, 0.0242, 0.0201, 0.0732, 0.0324, 0.0909, 0.0562], device='cuda:6'), in_proj_covar=tensor([0.0207, 0.0239, 0.0239, 0.0213, 0.0267, 0.0241, 0.0169, 0.0276], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:54:43,399 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2023-04-28 17:55:03,363 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.354e+02 3.365e+02 3.868e+02 4.958e+02 9.415e+02, threshold=7.735e+02, percent-clipped=1.0 2023-04-28 17:55:12,757 INFO [train.py:904] (6/8) Epoch 7, batch 8000, loss[loss=0.2345, simple_loss=0.3187, pruned_loss=0.07511, over 16723.00 frames. ], tot_loss[loss=0.2427, simple_loss=0.3189, pruned_loss=0.08327, over 3059413.83 frames. ], batch size: 124, lr: 9.55e-03, grad_scale: 4.0 2023-04-28 17:56:00,095 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68934.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:56:02,505 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8293, 4.1403, 3.9126, 3.9276, 3.6636, 3.7289, 3.7751, 4.0635], device='cuda:6'), covar=tensor([0.0999, 0.0831, 0.0995, 0.0618, 0.0719, 0.1535, 0.0885, 0.1081], device='cuda:6'), in_proj_covar=tensor([0.0447, 0.0565, 0.0482, 0.0375, 0.0355, 0.0385, 0.0471, 0.0420], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:56:11,689 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 17:56:18,117 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5614, 3.9615, 3.7054, 1.9387, 3.2055, 2.5127, 3.7909, 3.9891], device='cuda:6'), covar=tensor([0.0247, 0.0557, 0.0550, 0.1763, 0.0711, 0.0881, 0.0586, 0.0795], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0131, 0.0152, 0.0139, 0.0133, 0.0123, 0.0135, 0.0138], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 17:56:25,017 INFO [train.py:904] (6/8) Epoch 7, batch 8050, loss[loss=0.2477, simple_loss=0.3325, pruned_loss=0.08142, over 16191.00 frames. ], tot_loss[loss=0.243, simple_loss=0.319, pruned_loss=0.08344, over 3039934.26 frames. ], batch size: 165, lr: 9.54e-03, grad_scale: 4.0 2023-04-28 17:56:45,051 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8253, 3.6171, 3.8467, 3.9712, 4.0481, 3.6308, 3.9647, 4.0667], device='cuda:6'), covar=tensor([0.1026, 0.0890, 0.1041, 0.0512, 0.0508, 0.1421, 0.0631, 0.0493], device='cuda:6'), in_proj_covar=tensor([0.0441, 0.0545, 0.0676, 0.0545, 0.0413, 0.0417, 0.0438, 0.0474], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:57:30,025 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.364e+02 3.694e+02 4.392e+02 5.257e+02 9.021e+02, threshold=8.783e+02, percent-clipped=4.0 2023-04-28 17:57:41,511 INFO [train.py:904] (6/8) Epoch 7, batch 8100, loss[loss=0.2336, simple_loss=0.316, pruned_loss=0.07562, over 16929.00 frames. ], tot_loss[loss=0.2406, simple_loss=0.3174, pruned_loss=0.08188, over 3055269.31 frames. ], batch size: 109, lr: 9.54e-03, grad_scale: 4.0 2023-04-28 17:58:38,470 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3387, 4.1195, 4.3823, 4.5289, 4.6598, 4.1938, 4.6052, 4.6288], device='cuda:6'), covar=tensor([0.1244, 0.0903, 0.1208, 0.0555, 0.0486, 0.0968, 0.0524, 0.0491], device='cuda:6'), in_proj_covar=tensor([0.0444, 0.0546, 0.0678, 0.0546, 0.0415, 0.0420, 0.0439, 0.0476], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:58:40,407 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7747, 5.2209, 5.3561, 5.2681, 5.3180, 5.8044, 5.2129, 5.0378], device='cuda:6'), covar=tensor([0.0912, 0.1511, 0.1537, 0.1309, 0.2009, 0.0844, 0.1361, 0.1900], device='cuda:6'), in_proj_covar=tensor([0.0310, 0.0427, 0.0443, 0.0369, 0.0495, 0.0472, 0.0357, 0.0505], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 17:58:40,416 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69041.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:58:43,447 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69043.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:58:53,410 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2243, 4.2751, 4.0731, 3.9666, 3.7253, 4.1633, 4.0261, 3.9213], device='cuda:6'), covar=tensor([0.0555, 0.0340, 0.0246, 0.0212, 0.0842, 0.0356, 0.0420, 0.0568], device='cuda:6'), in_proj_covar=tensor([0.0207, 0.0240, 0.0238, 0.0211, 0.0266, 0.0241, 0.0169, 0.0277], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 17:58:55,812 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.64 vs. limit=5.0 2023-04-28 17:58:55,988 INFO [train.py:904] (6/8) Epoch 7, batch 8150, loss[loss=0.1999, simple_loss=0.283, pruned_loss=0.0584, over 16864.00 frames. ], tot_loss[loss=0.2374, simple_loss=0.3144, pruned_loss=0.08014, over 3069374.99 frames. ], batch size: 96, lr: 9.54e-03, grad_scale: 4.0 2023-04-28 17:59:00,799 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69055.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:59:04,140 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.08 vs. limit=5.0 2023-04-28 17:59:59,698 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.706e+02 3.772e+02 4.723e+02 6.532e+02 2.181e+03, threshold=9.446e+02, percent-clipped=10.0 2023-04-28 18:00:10,161 INFO [train.py:904] (6/8) Epoch 7, batch 8200, loss[loss=0.2384, simple_loss=0.3221, pruned_loss=0.07728, over 16850.00 frames. ], tot_loss[loss=0.2346, simple_loss=0.3117, pruned_loss=0.07876, over 3086439.50 frames. ], batch size: 116, lr: 9.53e-03, grad_scale: 4.0 2023-04-28 18:00:12,049 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69103.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:00:13,533 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69104.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:00:17,444 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-28 18:01:30,744 INFO [train.py:904] (6/8) Epoch 7, batch 8250, loss[loss=0.2324, simple_loss=0.3183, pruned_loss=0.07328, over 15427.00 frames. ], tot_loss[loss=0.232, simple_loss=0.3108, pruned_loss=0.07666, over 3067196.57 frames. ], batch size: 190, lr: 9.53e-03, grad_scale: 4.0 2023-04-28 18:01:47,282 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69162.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:01:58,231 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69168.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:02:11,996 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0304, 5.5527, 5.6313, 5.5675, 5.4813, 6.0287, 5.5715, 5.3494], device='cuda:6'), covar=tensor([0.0669, 0.1342, 0.1421, 0.1678, 0.2135, 0.0814, 0.0973, 0.2066], device='cuda:6'), in_proj_covar=tensor([0.0304, 0.0417, 0.0437, 0.0362, 0.0485, 0.0463, 0.0348, 0.0497], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 18:02:40,830 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.107e+02 2.942e+02 3.716e+02 4.471e+02 8.485e+02, threshold=7.433e+02, percent-clipped=0.0 2023-04-28 18:02:46,490 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5514, 3.6015, 3.3672, 3.2344, 3.1918, 3.4111, 3.2278, 3.2680], device='cuda:6'), covar=tensor([0.0392, 0.0311, 0.0220, 0.0194, 0.0515, 0.0273, 0.1264, 0.0435], device='cuda:6'), in_proj_covar=tensor([0.0203, 0.0238, 0.0236, 0.0210, 0.0262, 0.0236, 0.0167, 0.0274], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 18:02:52,070 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.69 vs. limit=5.0 2023-04-28 18:02:52,449 INFO [train.py:904] (6/8) Epoch 7, batch 8300, loss[loss=0.2189, simple_loss=0.3029, pruned_loss=0.06746, over 15414.00 frames. ], tot_loss[loss=0.2273, simple_loss=0.3076, pruned_loss=0.07347, over 3059182.58 frames. ], batch size: 191, lr: 9.53e-03, grad_scale: 4.0 2023-04-28 18:03:26,090 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69223.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:03:29,156 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69225.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:03:35,541 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69229.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:03:42,621 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69234.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:04:00,594 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69245.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 18:04:11,882 INFO [train.py:904] (6/8) Epoch 7, batch 8350, loss[loss=0.2143, simple_loss=0.2899, pruned_loss=0.06938, over 11876.00 frames. ], tot_loss[loss=0.2242, simple_loss=0.3063, pruned_loss=0.07106, over 3039341.61 frames. ], batch size: 248, lr: 9.52e-03, grad_scale: 4.0 2023-04-28 18:04:46,459 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69273.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:05:00,051 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69282.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:05:07,157 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69286.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:05:20,697 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 2.640e+02 3.244e+02 4.107e+02 9.415e+02, threshold=6.488e+02, percent-clipped=2.0 2023-04-28 18:05:32,118 INFO [train.py:904] (6/8) Epoch 7, batch 8400, loss[loss=0.2077, simple_loss=0.2947, pruned_loss=0.06037, over 15361.00 frames. ], tot_loss[loss=0.2194, simple_loss=0.3028, pruned_loss=0.06803, over 3050451.72 frames. ], batch size: 190, lr: 9.52e-03, grad_scale: 8.0 2023-04-28 18:05:39,331 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69306.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 18:06:02,559 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69320.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:06:23,791 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69334.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 18:06:35,230 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69341.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:06:52,097 INFO [train.py:904] (6/8) Epoch 7, batch 8450, loss[loss=0.1922, simple_loss=0.2829, pruned_loss=0.05073, over 16843.00 frames. ], tot_loss[loss=0.2159, simple_loss=0.3008, pruned_loss=0.0655, over 3087042.12 frames. ], batch size: 116, lr: 9.52e-03, grad_scale: 8.0 2023-04-28 18:07:10,163 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0171, 4.0773, 4.5125, 4.4565, 4.4320, 4.1595, 4.1510, 4.0980], device='cuda:6'), covar=tensor([0.0279, 0.0518, 0.0293, 0.0382, 0.0399, 0.0318, 0.0888, 0.0370], device='cuda:6'), in_proj_covar=tensor([0.0269, 0.0274, 0.0271, 0.0267, 0.0317, 0.0293, 0.0387, 0.0240], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 18:07:18,742 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69368.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:07:23,143 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.90 vs. limit=5.0 2023-04-28 18:07:30,306 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1355, 2.8803, 2.8900, 1.9513, 2.6544, 2.1228, 2.8364, 2.9068], device='cuda:6'), covar=tensor([0.0298, 0.0644, 0.0438, 0.1548, 0.0708, 0.0907, 0.0537, 0.0717], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0127, 0.0150, 0.0138, 0.0131, 0.0123, 0.0133, 0.0135], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-28 18:07:37,245 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-28 18:07:38,809 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69381.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:07:50,679 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69389.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:07:58,614 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69394.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:07:59,250 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.708e+02 3.167e+02 3.913e+02 6.077e+02, threshold=6.334e+02, percent-clipped=0.0 2023-04-28 18:08:06,577 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69399.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:08:10,404 INFO [train.py:904] (6/8) Epoch 7, batch 8500, loss[loss=0.179, simple_loss=0.2725, pruned_loss=0.04275, over 16328.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2962, pruned_loss=0.06229, over 3089062.12 frames. ], batch size: 146, lr: 9.51e-03, grad_scale: 8.0 2023-04-28 18:08:55,018 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69429.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:09:33,177 INFO [train.py:904] (6/8) Epoch 7, batch 8550, loss[loss=0.241, simple_loss=0.3245, pruned_loss=0.07872, over 16720.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.2947, pruned_loss=0.06223, over 3044016.42 frames. ], batch size: 124, lr: 9.51e-03, grad_scale: 8.0 2023-04-28 18:09:41,796 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69455.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:10:58,770 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.959e+02 3.006e+02 3.588e+02 4.597e+02 9.217e+02, threshold=7.177e+02, percent-clipped=6.0 2023-04-28 18:11:11,595 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-28 18:11:12,033 INFO [train.py:904] (6/8) Epoch 7, batch 8600, loss[loss=0.2183, simple_loss=0.3091, pruned_loss=0.06374, over 16210.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2946, pruned_loss=0.06136, over 3025576.45 frames. ], batch size: 165, lr: 9.51e-03, grad_scale: 8.0 2023-04-28 18:11:46,306 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69518.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:11:56,249 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69524.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:12:05,305 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.64 vs. limit=5.0 2023-04-28 18:12:17,032 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9471, 1.6984, 1.5063, 1.4944, 1.8117, 1.6326, 1.7235, 1.9298], device='cuda:6'), covar=tensor([0.0060, 0.0190, 0.0230, 0.0230, 0.0115, 0.0176, 0.0095, 0.0119], device='cuda:6'), in_proj_covar=tensor([0.0097, 0.0172, 0.0167, 0.0168, 0.0164, 0.0170, 0.0156, 0.0153], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 18:12:24,845 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-28 18:12:47,740 INFO [train.py:904] (6/8) Epoch 7, batch 8650, loss[loss=0.1884, simple_loss=0.2831, pruned_loss=0.04688, over 15233.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2921, pruned_loss=0.05926, over 3032648.91 frames. ], batch size: 191, lr: 9.50e-03, grad_scale: 8.0 2023-04-28 18:13:54,357 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69581.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:14:19,150 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.635e+02 2.551e+02 2.997e+02 3.527e+02 6.166e+02, threshold=5.994e+02, percent-clipped=0.0 2023-04-28 18:14:31,865 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69601.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 18:14:32,624 INFO [train.py:904] (6/8) Epoch 7, batch 8700, loss[loss=0.2038, simple_loss=0.2781, pruned_loss=0.06474, over 12781.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2886, pruned_loss=0.05747, over 3016362.99 frames. ], batch size: 248, lr: 9.50e-03, grad_scale: 8.0 2023-04-28 18:15:05,548 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-28 18:15:22,830 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69629.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 18:16:08,632 INFO [train.py:904] (6/8) Epoch 7, batch 8750, loss[loss=0.172, simple_loss=0.2601, pruned_loss=0.04193, over 12445.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2877, pruned_loss=0.05637, over 3016456.88 frames. ], batch size: 248, lr: 9.50e-03, grad_scale: 8.0 2023-04-28 18:17:04,395 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69674.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 18:17:07,493 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69676.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:17:16,308 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9087, 5.3519, 5.4929, 5.4352, 5.4543, 5.9232, 5.4211, 5.2105], device='cuda:6'), covar=tensor([0.0702, 0.1404, 0.1195, 0.1504, 0.1811, 0.0770, 0.1037, 0.2089], device='cuda:6'), in_proj_covar=tensor([0.0290, 0.0405, 0.0420, 0.0356, 0.0466, 0.0452, 0.0338, 0.0472], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-28 18:17:39,650 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5367, 1.9757, 1.6415, 1.8370, 2.3421, 2.1293, 2.4308, 2.4734], device='cuda:6'), covar=tensor([0.0067, 0.0242, 0.0291, 0.0258, 0.0137, 0.0198, 0.0121, 0.0139], device='cuda:6'), in_proj_covar=tensor([0.0098, 0.0172, 0.0167, 0.0168, 0.0164, 0.0169, 0.0154, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 18:17:45,985 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.541e+02 2.592e+02 3.192e+02 4.101e+02 7.533e+02, threshold=6.383e+02, percent-clipped=6.0 2023-04-28 18:17:55,725 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69699.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:18:01,164 INFO [train.py:904] (6/8) Epoch 7, batch 8800, loss[loss=0.2003, simple_loss=0.2916, pruned_loss=0.05445, over 15385.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2864, pruned_loss=0.05537, over 3026661.22 frames. ], batch size: 191, lr: 9.49e-03, grad_scale: 8.0 2023-04-28 18:18:47,923 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69724.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:19:11,466 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69735.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 18:19:26,658 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.41 vs. limit=5.0 2023-04-28 18:19:36,494 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69747.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:19:43,874 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69750.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:19:46,919 INFO [train.py:904] (6/8) Epoch 7, batch 8850, loss[loss=0.2037, simple_loss=0.3012, pruned_loss=0.05305, over 16401.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.289, pruned_loss=0.0548, over 3036036.15 frames. ], batch size: 146, lr: 9.49e-03, grad_scale: 4.0 2023-04-28 18:21:21,320 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.819e+02 2.840e+02 3.473e+02 4.436e+02 7.105e+02, threshold=6.946e+02, percent-clipped=4.0 2023-04-28 18:21:34,922 INFO [train.py:904] (6/8) Epoch 7, batch 8900, loss[loss=0.2001, simple_loss=0.2895, pruned_loss=0.05533, over 16925.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2882, pruned_loss=0.05359, over 3038980.64 frames. ], batch size: 116, lr: 9.49e-03, grad_scale: 4.0 2023-04-28 18:22:06,687 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69818.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:22:21,948 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69824.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:23:38,919 INFO [train.py:904] (6/8) Epoch 7, batch 8950, loss[loss=0.1733, simple_loss=0.2639, pruned_loss=0.04128, over 16266.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2879, pruned_loss=0.05373, over 3064168.79 frames. ], batch size: 165, lr: 9.48e-03, grad_scale: 4.0 2023-04-28 18:24:09,591 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69866.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:24:22,085 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69872.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:24:43,043 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69881.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:25:14,732 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 2.573e+02 2.911e+02 3.451e+02 7.486e+02, threshold=5.822e+02, percent-clipped=1.0 2023-04-28 18:25:27,295 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69901.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 18:25:27,971 INFO [train.py:904] (6/8) Epoch 7, batch 9000, loss[loss=0.1992, simple_loss=0.2926, pruned_loss=0.05284, over 16856.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.2853, pruned_loss=0.05255, over 3078052.13 frames. ], batch size: 124, lr: 9.48e-03, grad_scale: 4.0 2023-04-28 18:25:27,971 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 18:25:37,199 INFO [train.py:938] (6/8) Epoch 7, validation: loss=0.1647, simple_loss=0.2682, pruned_loss=0.03062, over 944034.00 frames. 2023-04-28 18:25:37,199 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-28 18:26:33,159 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69929.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:26:33,294 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69929.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 18:26:42,520 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.81 vs. limit=5.0 2023-04-28 18:27:14,855 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69949.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 18:27:18,970 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5976, 3.6642, 4.0385, 4.0464, 4.0445, 3.7479, 3.8092, 3.7865], device='cuda:6'), covar=tensor([0.0294, 0.0525, 0.0367, 0.0367, 0.0392, 0.0345, 0.0712, 0.0381], device='cuda:6'), in_proj_covar=tensor([0.0261, 0.0266, 0.0265, 0.0257, 0.0306, 0.0286, 0.0373, 0.0233], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-28 18:27:19,718 INFO [train.py:904] (6/8) Epoch 7, batch 9050, loss[loss=0.1832, simple_loss=0.269, pruned_loss=0.04868, over 15441.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2866, pruned_loss=0.05328, over 3098712.37 frames. ], batch size: 191, lr: 9.48e-03, grad_scale: 4.0 2023-04-28 18:28:08,841 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69976.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:28:10,101 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69977.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:28:43,409 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0543, 3.8977, 4.0825, 4.2393, 4.3476, 3.9775, 4.3322, 4.3285], device='cuda:6'), covar=tensor([0.1191, 0.0824, 0.1158, 0.0549, 0.0492, 0.0915, 0.0485, 0.0501], device='cuda:6'), in_proj_covar=tensor([0.0417, 0.0519, 0.0634, 0.0521, 0.0390, 0.0398, 0.0409, 0.0457], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 18:28:50,067 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.782e+02 2.800e+02 3.566e+02 4.608e+02 8.238e+02, threshold=7.132e+02, percent-clipped=9.0 2023-04-28 18:29:06,534 INFO [train.py:904] (6/8) Epoch 7, batch 9100, loss[loss=0.217, simple_loss=0.3086, pruned_loss=0.06267, over 16173.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2856, pruned_loss=0.05373, over 3078116.23 frames. ], batch size: 165, lr: 9.47e-03, grad_scale: 4.0 2023-04-28 18:29:58,315 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=70024.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:29:58,448 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70024.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:30:13,182 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70030.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 18:31:01,019 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70050.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:31:03,190 INFO [train.py:904] (6/8) Epoch 7, batch 9150, loss[loss=0.1678, simple_loss=0.2612, pruned_loss=0.03714, over 16465.00 frames. ], tot_loss[loss=0.1963, simple_loss=0.2858, pruned_loss=0.05337, over 3063154.32 frames. ], batch size: 68, lr: 9.47e-03, grad_scale: 4.0 2023-04-28 18:31:47,583 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=70072.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:32:35,334 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.681e+02 3.453e+02 4.314e+02 7.152e+02, threshold=6.905e+02, percent-clipped=1.0 2023-04-28 18:32:39,001 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=70098.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:32:45,141 INFO [train.py:904] (6/8) Epoch 7, batch 9200, loss[loss=0.2149, simple_loss=0.2964, pruned_loss=0.06671, over 16329.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2812, pruned_loss=0.05225, over 3047301.26 frames. ], batch size: 146, lr: 9.47e-03, grad_scale: 8.0 2023-04-28 18:33:01,835 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70110.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:34:22,191 INFO [train.py:904] (6/8) Epoch 7, batch 9250, loss[loss=0.1922, simple_loss=0.263, pruned_loss=0.06073, over 11982.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2806, pruned_loss=0.05228, over 3047450.62 frames. ], batch size: 246, lr: 9.46e-03, grad_scale: 8.0 2023-04-28 18:35:02,180 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=70171.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:36:01,187 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.738e+02 2.717e+02 3.344e+02 3.929e+02 8.889e+02, threshold=6.688e+02, percent-clipped=4.0 2023-04-28 18:36:06,642 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2348, 3.3244, 1.7407, 3.4849, 2.3097, 3.4364, 1.8989, 2.6551], device='cuda:6'), covar=tensor([0.0172, 0.0288, 0.1464, 0.0111, 0.0761, 0.0495, 0.1466, 0.0578], device='cuda:6'), in_proj_covar=tensor([0.0129, 0.0147, 0.0174, 0.0090, 0.0155, 0.0178, 0.0183, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-28 18:36:13,550 INFO [train.py:904] (6/8) Epoch 7, batch 9300, loss[loss=0.1923, simple_loss=0.2892, pruned_loss=0.0477, over 16194.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.279, pruned_loss=0.05138, over 3037226.64 frames. ], batch size: 165, lr: 9.46e-03, grad_scale: 8.0 2023-04-28 18:36:15,895 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6944, 4.6715, 4.4797, 4.3090, 4.1660, 4.5902, 4.4234, 4.2687], device='cuda:6'), covar=tensor([0.0403, 0.0345, 0.0218, 0.0203, 0.0718, 0.0322, 0.0308, 0.0566], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0226, 0.0228, 0.0202, 0.0248, 0.0229, 0.0158, 0.0266], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 18:37:59,027 INFO [train.py:904] (6/8) Epoch 7, batch 9350, loss[loss=0.1899, simple_loss=0.2776, pruned_loss=0.05112, over 15312.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2789, pruned_loss=0.05136, over 3044690.91 frames. ], batch size: 190, lr: 9.46e-03, grad_scale: 8.0 2023-04-28 18:38:06,690 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5245, 1.5368, 1.8673, 2.4246, 2.3300, 2.5240, 1.6031, 2.5920], device='cuda:6'), covar=tensor([0.0098, 0.0289, 0.0195, 0.0162, 0.0146, 0.0124, 0.0317, 0.0062], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0149, 0.0132, 0.0134, 0.0141, 0.0096, 0.0149, 0.0088], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 18:38:18,307 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70260.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:39:30,773 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7717, 3.4808, 3.2666, 1.7983, 2.8735, 2.2341, 3.1987, 3.3912], device='cuda:6'), covar=tensor([0.0251, 0.0494, 0.0456, 0.1631, 0.0646, 0.0852, 0.0672, 0.0768], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0123, 0.0152, 0.0138, 0.0131, 0.0123, 0.0132, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 18:39:31,294 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.755e+02 2.689e+02 3.082e+02 3.847e+02 8.568e+02, threshold=6.165e+02, percent-clipped=3.0 2023-04-28 18:39:40,128 INFO [train.py:904] (6/8) Epoch 7, batch 9400, loss[loss=0.1952, simple_loss=0.2903, pruned_loss=0.05004, over 15320.00 frames. ], tot_loss[loss=0.1906, simple_loss=0.2793, pruned_loss=0.05094, over 3052859.61 frames. ], batch size: 191, lr: 9.45e-03, grad_scale: 4.0 2023-04-28 18:39:56,056 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-04-28 18:40:18,707 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=70321.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:40:24,105 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1199, 5.7469, 5.9157, 5.6006, 5.7787, 6.2410, 5.7992, 5.4877], device='cuda:6'), covar=tensor([0.0583, 0.1323, 0.1270, 0.1598, 0.1993, 0.0845, 0.0996, 0.1892], device='cuda:6'), in_proj_covar=tensor([0.0275, 0.0391, 0.0406, 0.0342, 0.0448, 0.0433, 0.0328, 0.0457], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 18:40:36,522 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70330.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 18:41:20,567 INFO [train.py:904] (6/8) Epoch 7, batch 9450, loss[loss=0.1974, simple_loss=0.2851, pruned_loss=0.05489, over 15228.00 frames. ], tot_loss[loss=0.1923, simple_loss=0.2815, pruned_loss=0.05152, over 3062307.79 frames. ], batch size: 191, lr: 9.45e-03, grad_scale: 4.0 2023-04-28 18:41:23,633 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4735, 3.6089, 3.6281, 1.7569, 3.8786, 3.8750, 2.8448, 3.0791], device='cuda:6'), covar=tensor([0.0716, 0.0119, 0.0120, 0.1179, 0.0036, 0.0079, 0.0387, 0.0356], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0092, 0.0079, 0.0136, 0.0064, 0.0081, 0.0114, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 18:42:14,166 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=70378.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 18:42:50,911 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.795e+02 2.717e+02 3.594e+02 4.518e+02 8.204e+02, threshold=7.189e+02, percent-clipped=4.0 2023-04-28 18:43:02,096 INFO [train.py:904] (6/8) Epoch 7, batch 9500, loss[loss=0.2032, simple_loss=0.3056, pruned_loss=0.05045, over 16764.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.2813, pruned_loss=0.05107, over 3064131.05 frames. ], batch size: 83, lr: 9.45e-03, grad_scale: 4.0 2023-04-28 18:44:47,822 INFO [train.py:904] (6/8) Epoch 7, batch 9550, loss[loss=0.2166, simple_loss=0.3096, pruned_loss=0.06184, over 15619.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2814, pruned_loss=0.05117, over 3079750.84 frames. ], batch size: 191, lr: 9.44e-03, grad_scale: 4.0 2023-04-28 18:45:18,615 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70466.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:46:18,719 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.839e+02 3.500e+02 4.170e+02 7.519e+02, threshold=6.999e+02, percent-clipped=3.0 2023-04-28 18:46:26,839 INFO [train.py:904] (6/8) Epoch 7, batch 9600, loss[loss=0.2229, simple_loss=0.3132, pruned_loss=0.06635, over 16214.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2825, pruned_loss=0.05207, over 3083058.65 frames. ], batch size: 165, lr: 9.44e-03, grad_scale: 8.0 2023-04-28 18:47:05,613 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.85 vs. limit=5.0 2023-04-28 18:48:15,879 INFO [train.py:904] (6/8) Epoch 7, batch 9650, loss[loss=0.2116, simple_loss=0.3041, pruned_loss=0.05956, over 16999.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.2845, pruned_loss=0.05297, over 3067611.78 frames. ], batch size: 55, lr: 9.44e-03, grad_scale: 8.0 2023-04-28 18:49:55,421 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.918e+02 2.673e+02 3.296e+02 3.931e+02 1.187e+03, threshold=6.593e+02, percent-clipped=2.0 2023-04-28 18:50:06,090 INFO [train.py:904] (6/8) Epoch 7, batch 9700, loss[loss=0.1655, simple_loss=0.2624, pruned_loss=0.03433, over 16879.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2832, pruned_loss=0.05232, over 3075999.13 frames. ], batch size: 102, lr: 9.43e-03, grad_scale: 8.0 2023-04-28 18:50:33,466 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70616.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:51:47,796 INFO [train.py:904] (6/8) Epoch 7, batch 9750, loss[loss=0.1899, simple_loss=0.267, pruned_loss=0.05637, over 12509.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2819, pruned_loss=0.05229, over 3079427.46 frames. ], batch size: 250, lr: 9.43e-03, grad_scale: 8.0 2023-04-28 18:53:01,640 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.63 vs. limit=2.0 2023-04-28 18:53:18,481 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.561e+02 2.655e+02 3.090e+02 3.865e+02 1.050e+03, threshold=6.180e+02, percent-clipped=3.0 2023-04-28 18:53:27,256 INFO [train.py:904] (6/8) Epoch 7, batch 9800, loss[loss=0.2361, simple_loss=0.341, pruned_loss=0.06563, over 16308.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2823, pruned_loss=0.05146, over 3076506.33 frames. ], batch size: 146, lr: 9.43e-03, grad_scale: 8.0 2023-04-28 18:54:22,943 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7807, 4.1194, 4.2679, 3.1790, 3.8687, 4.0550, 4.0236, 2.5397], device='cuda:6'), covar=tensor([0.0287, 0.0015, 0.0018, 0.0191, 0.0044, 0.0046, 0.0030, 0.0287], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0056, 0.0059, 0.0117, 0.0066, 0.0073, 0.0066, 0.0113], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 18:55:12,739 INFO [train.py:904] (6/8) Epoch 7, batch 9850, loss[loss=0.1661, simple_loss=0.2668, pruned_loss=0.03265, over 16655.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2823, pruned_loss=0.05071, over 3071303.01 frames. ], batch size: 89, lr: 9.42e-03, grad_scale: 8.0 2023-04-28 18:55:43,257 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70766.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:56:38,815 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7927, 3.0918, 2.9509, 4.9620, 4.0283, 4.6322, 1.5805, 3.3711], device='cuda:6'), covar=tensor([0.1279, 0.0575, 0.0878, 0.0075, 0.0136, 0.0240, 0.1419, 0.0602], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0145, 0.0167, 0.0102, 0.0170, 0.0192, 0.0169, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 18:56:54,659 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.862e+02 2.569e+02 3.072e+02 3.785e+02 9.202e+02, threshold=6.144e+02, percent-clipped=3.0 2023-04-28 18:57:04,338 INFO [train.py:904] (6/8) Epoch 7, batch 9900, loss[loss=0.2002, simple_loss=0.2966, pruned_loss=0.05186, over 15353.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2826, pruned_loss=0.05057, over 3058699.73 frames. ], batch size: 191, lr: 9.42e-03, grad_scale: 8.0 2023-04-28 18:57:33,473 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=70814.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:57:35,120 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1620, 3.2888, 1.7640, 3.4791, 2.3230, 3.4158, 1.7733, 2.5900], device='cuda:6'), covar=tensor([0.0214, 0.0335, 0.1608, 0.0106, 0.0814, 0.0491, 0.1682, 0.0701], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0149, 0.0177, 0.0091, 0.0160, 0.0180, 0.0188, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-28 18:58:35,163 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 18:59:01,222 INFO [train.py:904] (6/8) Epoch 7, batch 9950, loss[loss=0.1987, simple_loss=0.2956, pruned_loss=0.05087, over 16265.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2852, pruned_loss=0.05116, over 3069816.41 frames. ], batch size: 165, lr: 9.42e-03, grad_scale: 8.0 2023-04-28 19:00:47,805 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 2.726e+02 3.249e+02 3.973e+02 1.098e+03, threshold=6.498e+02, percent-clipped=6.0 2023-04-28 19:01:00,566 INFO [train.py:904] (6/8) Epoch 7, batch 10000, loss[loss=0.186, simple_loss=0.2829, pruned_loss=0.04459, over 16391.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2841, pruned_loss=0.05089, over 3090372.45 frames. ], batch size: 146, lr: 9.41e-03, grad_scale: 8.0 2023-04-28 19:01:29,810 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70916.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:02:40,827 INFO [train.py:904] (6/8) Epoch 7, batch 10050, loss[loss=0.2136, simple_loss=0.3103, pruned_loss=0.05842, over 15307.00 frames. ], tot_loss[loss=0.1921, simple_loss=0.2837, pruned_loss=0.05028, over 3103755.55 frames. ], batch size: 191, lr: 9.41e-03, grad_scale: 8.0 2023-04-28 19:03:04,294 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=70964.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:03:37,406 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-28 19:03:56,295 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70992.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:04:04,309 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.660e+02 2.613e+02 3.026e+02 3.875e+02 7.173e+02, threshold=6.053e+02, percent-clipped=4.0 2023-04-28 19:04:12,143 INFO [train.py:904] (6/8) Epoch 7, batch 10100, loss[loss=0.2039, simple_loss=0.2858, pruned_loss=0.06096, over 12112.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2841, pruned_loss=0.05072, over 3090063.31 frames. ], batch size: 247, lr: 9.41e-03, grad_scale: 8.0 2023-04-28 19:04:17,683 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2040, 1.9748, 2.1395, 3.7318, 1.8856, 2.4513, 2.1572, 2.1077], device='cuda:6'), covar=tensor([0.0725, 0.2767, 0.1677, 0.0334, 0.3338, 0.1655, 0.2494, 0.2716], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0335, 0.0289, 0.0302, 0.0377, 0.0360, 0.0303, 0.0393], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:05:07,016 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71030.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:05:23,202 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9983, 3.5800, 3.5864, 2.0972, 2.9485, 2.5030, 3.3744, 3.4466], device='cuda:6'), covar=tensor([0.0349, 0.0645, 0.0426, 0.1553, 0.0681, 0.0826, 0.0781, 0.0953], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0121, 0.0151, 0.0137, 0.0129, 0.0122, 0.0132, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-28 19:05:54,946 INFO [train.py:904] (6/8) Epoch 8, batch 0, loss[loss=0.1808, simple_loss=0.258, pruned_loss=0.05181, over 16838.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.258, pruned_loss=0.05181, over 16838.00 frames. ], batch size: 42, lr: 8.86e-03, grad_scale: 8.0 2023-04-28 19:05:54,947 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 19:06:02,579 INFO [train.py:938] (6/8) Epoch 8, validation: loss=0.1627, simple_loss=0.2663, pruned_loss=0.02958, over 944034.00 frames. 2023-04-28 19:06:02,579 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-28 19:06:04,048 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71053.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:06:31,253 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.14 vs. limit=5.0 2023-04-28 19:06:35,511 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-28 19:06:38,399 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7390, 1.5038, 2.0441, 2.5477, 2.5284, 2.4501, 1.6552, 2.7704], device='cuda:6'), covar=tensor([0.0086, 0.0309, 0.0209, 0.0145, 0.0154, 0.0168, 0.0279, 0.0061], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0153, 0.0137, 0.0136, 0.0143, 0.0097, 0.0152, 0.0089], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 19:06:55,606 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71091.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:07:07,690 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.561e+02 3.480e+02 4.237e+02 5.161e+02 1.332e+03, threshold=8.475e+02, percent-clipped=15.0 2023-04-28 19:07:09,983 INFO [train.py:904] (6/8) Epoch 8, batch 50, loss[loss=0.2388, simple_loss=0.3215, pruned_loss=0.07805, over 16714.00 frames. ], tot_loss[loss=0.2292, simple_loss=0.3042, pruned_loss=0.07707, over 736282.32 frames. ], batch size: 62, lr: 8.86e-03, grad_scale: 1.0 2023-04-28 19:07:11,438 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.69 vs. limit=5.0 2023-04-28 19:08:10,541 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-28 19:08:17,890 INFO [train.py:904] (6/8) Epoch 8, batch 100, loss[loss=0.1742, simple_loss=0.2555, pruned_loss=0.04645, over 16996.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2944, pruned_loss=0.06934, over 1313339.63 frames. ], batch size: 41, lr: 8.85e-03, grad_scale: 1.0 2023-04-28 19:09:23,827 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.200e+02 2.858e+02 3.485e+02 4.000e+02 7.217e+02, threshold=6.971e+02, percent-clipped=0.0 2023-04-28 19:09:26,722 INFO [train.py:904] (6/8) Epoch 8, batch 150, loss[loss=0.2374, simple_loss=0.3033, pruned_loss=0.08569, over 16286.00 frames. ], tot_loss[loss=0.2139, simple_loss=0.2914, pruned_loss=0.06824, over 1760286.30 frames. ], batch size: 165, lr: 8.85e-03, grad_scale: 1.0 2023-04-28 19:09:55,871 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7347, 3.8200, 1.8714, 3.8953, 2.7671, 3.9296, 1.9488, 2.9747], device='cuda:6'), covar=tensor([0.0138, 0.0222, 0.1420, 0.0146, 0.0632, 0.0356, 0.1337, 0.0516], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0153, 0.0178, 0.0096, 0.0160, 0.0184, 0.0188, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-28 19:10:26,325 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9417, 1.5088, 2.1892, 2.7031, 2.5550, 3.0716, 1.8291, 3.0350], device='cuda:6'), covar=tensor([0.0110, 0.0343, 0.0216, 0.0192, 0.0175, 0.0107, 0.0326, 0.0100], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0155, 0.0138, 0.0139, 0.0144, 0.0099, 0.0152, 0.0091], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 19:10:33,447 INFO [train.py:904] (6/8) Epoch 8, batch 200, loss[loss=0.1881, simple_loss=0.2817, pruned_loss=0.04727, over 17106.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2893, pruned_loss=0.06766, over 2109989.33 frames. ], batch size: 53, lr: 8.85e-03, grad_scale: 1.0 2023-04-28 19:10:58,187 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-28 19:11:08,452 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5293, 2.6072, 1.7133, 2.5713, 2.0952, 2.7399, 1.7458, 2.3163], device='cuda:6'), covar=tensor([0.0200, 0.0327, 0.1414, 0.0138, 0.0688, 0.0406, 0.1449, 0.0562], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0155, 0.0181, 0.0098, 0.0163, 0.0188, 0.0191, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 19:11:40,016 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 2.597e+02 3.114e+02 3.652e+02 1.080e+03, threshold=6.228e+02, percent-clipped=1.0 2023-04-28 19:11:41,390 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1192, 4.1355, 4.5990, 4.5920, 4.6645, 4.3092, 4.1181, 4.1714], device='cuda:6'), covar=tensor([0.0475, 0.0742, 0.0624, 0.0615, 0.0576, 0.0480, 0.1283, 0.0584], device='cuda:6'), in_proj_covar=tensor([0.0266, 0.0276, 0.0277, 0.0262, 0.0311, 0.0294, 0.0385, 0.0240], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 19:11:42,945 INFO [train.py:904] (6/8) Epoch 8, batch 250, loss[loss=0.1911, simple_loss=0.2795, pruned_loss=0.05133, over 17146.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2873, pruned_loss=0.06729, over 2365436.68 frames. ], batch size: 47, lr: 8.84e-03, grad_scale: 1.0 2023-04-28 19:11:45,639 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2756, 4.1557, 4.0999, 3.9116, 3.7528, 4.1493, 3.9887, 3.9150], device='cuda:6'), covar=tensor([0.0493, 0.0419, 0.0282, 0.0279, 0.0901, 0.0408, 0.0476, 0.0604], device='cuda:6'), in_proj_covar=tensor([0.0208, 0.0246, 0.0245, 0.0218, 0.0274, 0.0249, 0.0166, 0.0285], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:12:47,180 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71348.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:12:52,127 INFO [train.py:904] (6/8) Epoch 8, batch 300, loss[loss=0.1928, simple_loss=0.2684, pruned_loss=0.05859, over 15914.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2843, pruned_loss=0.06629, over 2572715.81 frames. ], batch size: 35, lr: 8.84e-03, grad_scale: 1.0 2023-04-28 19:13:08,605 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3397, 4.0185, 4.1351, 1.8181, 4.4107, 4.5640, 3.1944, 3.2507], device='cuda:6'), covar=tensor([0.1078, 0.0129, 0.0196, 0.1255, 0.0070, 0.0070, 0.0341, 0.0477], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0093, 0.0081, 0.0139, 0.0067, 0.0087, 0.0117, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 19:13:16,166 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4814, 3.3499, 2.6032, 2.1561, 2.1700, 2.1987, 3.2931, 3.1693], device='cuda:6'), covar=tensor([0.2461, 0.0726, 0.1460, 0.1908, 0.2226, 0.1638, 0.0510, 0.1093], device='cuda:6'), in_proj_covar=tensor([0.0287, 0.0247, 0.0276, 0.0258, 0.0261, 0.0210, 0.0252, 0.0272], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:13:39,944 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71386.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:13:43,769 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0741, 3.8997, 3.2341, 5.2382, 4.5890, 4.8514, 1.8514, 3.5377], device='cuda:6'), covar=tensor([0.1226, 0.0415, 0.0890, 0.0113, 0.0256, 0.0311, 0.1322, 0.0631], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0149, 0.0172, 0.0108, 0.0182, 0.0201, 0.0172, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 19:13:44,034 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 19:13:58,499 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.899e+02 2.511e+02 3.030e+02 3.934e+02 1.129e+03, threshold=6.060e+02, percent-clipped=6.0 2023-04-28 19:14:01,159 INFO [train.py:904] (6/8) Epoch 8, batch 350, loss[loss=0.2179, simple_loss=0.2863, pruned_loss=0.07474, over 16237.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2809, pruned_loss=0.06352, over 2740934.25 frames. ], batch size: 165, lr: 8.84e-03, grad_scale: 1.0 2023-04-28 19:14:50,953 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7663, 4.2093, 3.1476, 2.3518, 2.8712, 2.3804, 4.5085, 3.8929], device='cuda:6'), covar=tensor([0.2426, 0.0592, 0.1369, 0.1940, 0.2266, 0.1576, 0.0352, 0.0958], device='cuda:6'), in_proj_covar=tensor([0.0285, 0.0245, 0.0272, 0.0256, 0.0260, 0.0208, 0.0250, 0.0272], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:14:56,173 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5675, 4.5429, 4.7275, 4.6837, 4.6081, 5.2322, 4.8035, 4.4703], device='cuda:6'), covar=tensor([0.1070, 0.1856, 0.1629, 0.1796, 0.2864, 0.1008, 0.1348, 0.2485], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0445, 0.0458, 0.0384, 0.0511, 0.0483, 0.0366, 0.0518], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 19:15:10,453 INFO [train.py:904] (6/8) Epoch 8, batch 400, loss[loss=0.2245, simple_loss=0.2887, pruned_loss=0.08022, over 16673.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2801, pruned_loss=0.06348, over 2861488.29 frames. ], batch size: 89, lr: 8.83e-03, grad_scale: 2.0 2023-04-28 19:15:12,770 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5905, 4.3872, 4.5562, 4.8253, 4.9427, 4.4839, 4.8814, 4.8801], device='cuda:6'), covar=tensor([0.1424, 0.0992, 0.1498, 0.0648, 0.0538, 0.0876, 0.0879, 0.0615], device='cuda:6'), in_proj_covar=tensor([0.0475, 0.0587, 0.0718, 0.0586, 0.0443, 0.0443, 0.0455, 0.0509], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:15:57,640 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7095, 4.8258, 5.3145, 5.2570, 5.2702, 4.8386, 4.8609, 4.6701], device='cuda:6'), covar=tensor([0.0290, 0.0410, 0.0365, 0.0447, 0.0463, 0.0303, 0.0837, 0.0367], device='cuda:6'), in_proj_covar=tensor([0.0280, 0.0289, 0.0289, 0.0274, 0.0327, 0.0308, 0.0405, 0.0251], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 19:16:08,463 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2348, 2.8312, 2.5063, 2.2462, 2.0875, 2.1937, 2.7796, 2.7493], device='cuda:6'), covar=tensor([0.2201, 0.0791, 0.1422, 0.1686, 0.1709, 0.1618, 0.0473, 0.0817], device='cuda:6'), in_proj_covar=tensor([0.0284, 0.0245, 0.0271, 0.0257, 0.0262, 0.0207, 0.0250, 0.0271], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:16:17,821 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.026e+02 2.889e+02 3.429e+02 4.074e+02 1.363e+03, threshold=6.859e+02, percent-clipped=4.0 2023-04-28 19:16:20,152 INFO [train.py:904] (6/8) Epoch 8, batch 450, loss[loss=0.1984, simple_loss=0.2776, pruned_loss=0.05964, over 16631.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.2792, pruned_loss=0.06225, over 2964223.48 frames. ], batch size: 62, lr: 8.83e-03, grad_scale: 2.0 2023-04-28 19:17:28,362 INFO [train.py:904] (6/8) Epoch 8, batch 500, loss[loss=0.2248, simple_loss=0.3018, pruned_loss=0.07392, over 16734.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2777, pruned_loss=0.06164, over 3049524.74 frames. ], batch size: 62, lr: 8.83e-03, grad_scale: 2.0 2023-04-28 19:17:37,419 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5958, 3.7515, 3.9104, 1.8126, 4.0554, 4.1175, 3.1520, 2.9925], device='cuda:6'), covar=tensor([0.0655, 0.0120, 0.0133, 0.1075, 0.0056, 0.0102, 0.0267, 0.0385], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0092, 0.0081, 0.0137, 0.0067, 0.0087, 0.0116, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 19:18:14,802 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4581, 3.5722, 3.0963, 2.9216, 2.9952, 3.3493, 3.2299, 3.1641], device='cuda:6'), covar=tensor([0.0521, 0.0389, 0.0241, 0.0231, 0.0558, 0.0330, 0.1228, 0.0462], device='cuda:6'), in_proj_covar=tensor([0.0216, 0.0256, 0.0255, 0.0226, 0.0284, 0.0259, 0.0174, 0.0296], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:18:33,694 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.499e+02 3.026e+02 3.865e+02 9.130e+02, threshold=6.053e+02, percent-clipped=1.0 2023-04-28 19:18:37,346 INFO [train.py:904] (6/8) Epoch 8, batch 550, loss[loss=0.1904, simple_loss=0.2824, pruned_loss=0.04917, over 17033.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2764, pruned_loss=0.06058, over 3111246.74 frames. ], batch size: 50, lr: 8.83e-03, grad_scale: 2.0 2023-04-28 19:18:51,616 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-28 19:18:56,853 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71616.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:19:41,735 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=71648.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:19:45,774 INFO [train.py:904] (6/8) Epoch 8, batch 600, loss[loss=0.2229, simple_loss=0.2843, pruned_loss=0.08069, over 16769.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2755, pruned_loss=0.06052, over 3161963.36 frames. ], batch size: 124, lr: 8.82e-03, grad_scale: 2.0 2023-04-28 19:20:19,042 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71677.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:20:34,025 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=71686.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:20:45,947 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=71696.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:20:50,937 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.658e+02 2.476e+02 3.033e+02 3.542e+02 7.789e+02, threshold=6.066e+02, percent-clipped=2.0 2023-04-28 19:20:53,309 INFO [train.py:904] (6/8) Epoch 8, batch 650, loss[loss=0.1699, simple_loss=0.2543, pruned_loss=0.04274, over 16998.00 frames. ], tot_loss[loss=0.1962, simple_loss=0.274, pruned_loss=0.05925, over 3202907.41 frames. ], batch size: 41, lr: 8.82e-03, grad_scale: 2.0 2023-04-28 19:21:37,503 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=71734.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:22:01,894 INFO [train.py:904] (6/8) Epoch 8, batch 700, loss[loss=0.1854, simple_loss=0.2568, pruned_loss=0.05702, over 16769.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.273, pruned_loss=0.05887, over 3233670.92 frames. ], batch size: 124, lr: 8.82e-03, grad_scale: 2.0 2023-04-28 19:23:05,954 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.350e+02 2.585e+02 3.182e+02 3.748e+02 9.363e+02, threshold=6.364e+02, percent-clipped=4.0 2023-04-28 19:23:08,640 INFO [train.py:904] (6/8) Epoch 8, batch 750, loss[loss=0.2387, simple_loss=0.3045, pruned_loss=0.08644, over 16808.00 frames. ], tot_loss[loss=0.194, simple_loss=0.2728, pruned_loss=0.05765, over 3262406.65 frames. ], batch size: 83, lr: 8.81e-03, grad_scale: 2.0 2023-04-28 19:24:10,628 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7430, 2.3053, 2.3843, 4.4191, 2.1850, 2.8400, 2.3338, 2.4288], device='cuda:6'), covar=tensor([0.0760, 0.2800, 0.1723, 0.0371, 0.3438, 0.1744, 0.2577, 0.3089], device='cuda:6'), in_proj_covar=tensor([0.0347, 0.0359, 0.0304, 0.0326, 0.0397, 0.0398, 0.0325, 0.0428], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:24:17,945 INFO [train.py:904] (6/8) Epoch 8, batch 800, loss[loss=0.2101, simple_loss=0.2772, pruned_loss=0.07147, over 16359.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2727, pruned_loss=0.05794, over 3272137.73 frames. ], batch size: 146, lr: 8.81e-03, grad_scale: 4.0 2023-04-28 19:24:58,189 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9554, 4.1929, 4.5032, 2.0913, 4.7109, 4.7863, 3.3401, 3.6127], device='cuda:6'), covar=tensor([0.0614, 0.0126, 0.0118, 0.0974, 0.0049, 0.0062, 0.0301, 0.0310], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0094, 0.0082, 0.0140, 0.0068, 0.0089, 0.0120, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 19:25:23,720 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.711e+02 2.663e+02 3.116e+02 3.838e+02 7.032e+02, threshold=6.231e+02, percent-clipped=1.0 2023-04-28 19:25:25,971 INFO [train.py:904] (6/8) Epoch 8, batch 850, loss[loss=0.2202, simple_loss=0.3032, pruned_loss=0.06857, over 16688.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2721, pruned_loss=0.0572, over 3295435.82 frames. ], batch size: 62, lr: 8.81e-03, grad_scale: 4.0 2023-04-28 19:25:39,375 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8253, 3.9064, 3.0802, 2.3572, 2.6631, 2.2361, 3.8786, 3.6972], device='cuda:6'), covar=tensor([0.2173, 0.0552, 0.1248, 0.2077, 0.2189, 0.1735, 0.0439, 0.0967], device='cuda:6'), in_proj_covar=tensor([0.0290, 0.0252, 0.0278, 0.0262, 0.0272, 0.0214, 0.0256, 0.0282], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:25:49,833 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71919.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:26:15,613 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-28 19:26:32,709 INFO [train.py:904] (6/8) Epoch 8, batch 900, loss[loss=0.1827, simple_loss=0.2537, pruned_loss=0.05587, over 12463.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2714, pruned_loss=0.05705, over 3285796.68 frames. ], batch size: 248, lr: 8.80e-03, grad_scale: 4.0 2023-04-28 19:26:40,625 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6717, 4.0781, 4.2686, 1.8530, 4.5311, 4.6306, 3.2941, 3.3111], device='cuda:6'), covar=tensor([0.0772, 0.0151, 0.0172, 0.1136, 0.0059, 0.0077, 0.0326, 0.0402], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0093, 0.0082, 0.0139, 0.0067, 0.0089, 0.0118, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 19:26:48,500 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-04-28 19:27:00,564 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71972.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:27:13,385 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71980.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:27:38,431 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.818e+02 2.573e+02 3.042e+02 3.857e+02 8.032e+02, threshold=6.083e+02, percent-clipped=5.0 2023-04-28 19:27:44,211 INFO [train.py:904] (6/8) Epoch 8, batch 950, loss[loss=0.1975, simple_loss=0.2662, pruned_loss=0.06438, over 16444.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2722, pruned_loss=0.05712, over 3296723.50 frames. ], batch size: 146, lr: 8.80e-03, grad_scale: 4.0 2023-04-28 19:28:46,106 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2506, 4.6217, 4.0624, 4.5777, 4.1983, 4.1910, 4.2737, 4.6787], device='cuda:6'), covar=tensor([0.1936, 0.1808, 0.2556, 0.1025, 0.1573, 0.1763, 0.1636, 0.1688], device='cuda:6'), in_proj_covar=tensor([0.0477, 0.0611, 0.0508, 0.0409, 0.0390, 0.0403, 0.0508, 0.0451], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:28:52,161 INFO [train.py:904] (6/8) Epoch 8, batch 1000, loss[loss=0.1965, simple_loss=0.263, pruned_loss=0.065, over 16902.00 frames. ], tot_loss[loss=0.1921, simple_loss=0.2704, pruned_loss=0.05689, over 3305579.19 frames. ], batch size: 96, lr: 8.80e-03, grad_scale: 4.0 2023-04-28 19:29:09,364 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5717, 4.1319, 3.8615, 2.0398, 3.2217, 2.6391, 3.9045, 3.9634], device='cuda:6'), covar=tensor([0.0312, 0.0626, 0.0490, 0.1669, 0.0729, 0.0895, 0.0725, 0.1020], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0133, 0.0153, 0.0138, 0.0132, 0.0123, 0.0134, 0.0142], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 19:29:58,311 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.793e+02 2.648e+02 3.149e+02 3.724e+02 6.885e+02, threshold=6.299e+02, percent-clipped=3.0 2023-04-28 19:30:01,638 INFO [train.py:904] (6/8) Epoch 8, batch 1050, loss[loss=0.1944, simple_loss=0.2845, pruned_loss=0.05214, over 16839.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2698, pruned_loss=0.05625, over 3309557.33 frames. ], batch size: 57, lr: 8.79e-03, grad_scale: 4.0 2023-04-28 19:30:40,194 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72129.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:30:51,054 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2023-04-28 19:31:10,578 INFO [train.py:904] (6/8) Epoch 8, batch 1100, loss[loss=0.2063, simple_loss=0.2744, pruned_loss=0.06913, over 16784.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2689, pruned_loss=0.05572, over 3312247.68 frames. ], batch size: 83, lr: 8.79e-03, grad_scale: 4.0 2023-04-28 19:32:03,586 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72190.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:32:15,691 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.462e+02 2.908e+02 3.717e+02 7.729e+02, threshold=5.817e+02, percent-clipped=1.0 2023-04-28 19:32:18,231 INFO [train.py:904] (6/8) Epoch 8, batch 1150, loss[loss=0.1963, simple_loss=0.2789, pruned_loss=0.0568, over 17113.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2693, pruned_loss=0.05512, over 3314484.76 frames. ], batch size: 48, lr: 8.79e-03, grad_scale: 4.0 2023-04-28 19:32:39,437 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-28 19:32:41,352 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8541, 1.2977, 1.6046, 1.7351, 1.8524, 1.9338, 1.5346, 1.8851], device='cuda:6'), covar=tensor([0.0127, 0.0194, 0.0115, 0.0145, 0.0121, 0.0092, 0.0216, 0.0050], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0156, 0.0142, 0.0142, 0.0149, 0.0105, 0.0156, 0.0095], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 19:33:26,642 INFO [train.py:904] (6/8) Epoch 8, batch 1200, loss[loss=0.2244, simple_loss=0.2844, pruned_loss=0.0822, over 16752.00 frames. ], tot_loss[loss=0.1897, simple_loss=0.2691, pruned_loss=0.05516, over 3311723.43 frames. ], batch size: 83, lr: 8.79e-03, grad_scale: 8.0 2023-04-28 19:33:33,440 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2198, 5.1458, 4.8954, 4.3496, 4.9714, 2.0402, 4.7150, 4.9196], device='cuda:6'), covar=tensor([0.0055, 0.0055, 0.0133, 0.0328, 0.0067, 0.1903, 0.0111, 0.0146], device='cuda:6'), in_proj_covar=tensor([0.0113, 0.0101, 0.0152, 0.0145, 0.0117, 0.0165, 0.0138, 0.0143], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:33:50,385 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7121, 3.1445, 2.9126, 1.9149, 2.5837, 2.2034, 3.1996, 3.1357], device='cuda:6'), covar=tensor([0.0280, 0.0668, 0.0595, 0.1614, 0.0813, 0.0901, 0.0610, 0.0820], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0136, 0.0155, 0.0140, 0.0135, 0.0124, 0.0136, 0.0145], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 19:33:53,176 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72272.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:33:56,990 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72275.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:34:30,638 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.919e+02 2.822e+02 3.117e+02 3.864e+02 9.153e+02, threshold=6.234e+02, percent-clipped=4.0 2023-04-28 19:34:32,975 INFO [train.py:904] (6/8) Epoch 8, batch 1250, loss[loss=0.1909, simple_loss=0.282, pruned_loss=0.04987, over 17027.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2696, pruned_loss=0.05548, over 3317401.39 frames. ], batch size: 55, lr: 8.78e-03, grad_scale: 8.0 2023-04-28 19:34:42,212 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5910, 2.4432, 1.8670, 2.2194, 2.8745, 2.6771, 2.9206, 2.9977], device='cuda:6'), covar=tensor([0.0116, 0.0229, 0.0335, 0.0308, 0.0122, 0.0190, 0.0177, 0.0131], device='cuda:6'), in_proj_covar=tensor([0.0116, 0.0184, 0.0180, 0.0180, 0.0179, 0.0187, 0.0183, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:34:58,483 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=72320.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:35:06,170 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72326.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:35:41,557 INFO [train.py:904] (6/8) Epoch 8, batch 1300, loss[loss=0.2176, simple_loss=0.2883, pruned_loss=0.07343, over 12386.00 frames. ], tot_loss[loss=0.1899, simple_loss=0.2695, pruned_loss=0.05514, over 3319041.04 frames. ], batch size: 246, lr: 8.78e-03, grad_scale: 8.0 2023-04-28 19:36:30,417 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72387.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:36:49,363 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.738e+02 2.612e+02 3.072e+02 3.829e+02 7.494e+02, threshold=6.144e+02, percent-clipped=3.0 2023-04-28 19:36:52,118 INFO [train.py:904] (6/8) Epoch 8, batch 1350, loss[loss=0.21, simple_loss=0.2791, pruned_loss=0.07044, over 16725.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2698, pruned_loss=0.05523, over 3316635.77 frames. ], batch size: 124, lr: 8.78e-03, grad_scale: 8.0 2023-04-28 19:36:52,781 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4178, 3.9798, 3.9648, 2.0509, 3.3007, 2.4511, 3.8039, 3.7478], device='cuda:6'), covar=tensor([0.0262, 0.0516, 0.0476, 0.1584, 0.0637, 0.0878, 0.0585, 0.0896], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0135, 0.0154, 0.0139, 0.0133, 0.0123, 0.0135, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 19:38:01,924 INFO [train.py:904] (6/8) Epoch 8, batch 1400, loss[loss=0.2034, simple_loss=0.2707, pruned_loss=0.06805, over 16545.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2694, pruned_loss=0.05524, over 3315820.12 frames. ], batch size: 68, lr: 8.77e-03, grad_scale: 8.0 2023-04-28 19:38:09,942 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-28 19:38:34,248 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3806, 1.4391, 1.8814, 2.2093, 2.3487, 2.2693, 1.4941, 2.4087], device='cuda:6'), covar=tensor([0.0108, 0.0291, 0.0178, 0.0162, 0.0138, 0.0137, 0.0298, 0.0072], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0157, 0.0143, 0.0143, 0.0150, 0.0105, 0.0158, 0.0097], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 19:38:47,782 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72485.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:39:07,466 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.787e+02 2.583e+02 2.954e+02 3.513e+02 6.005e+02, threshold=5.909e+02, percent-clipped=0.0 2023-04-28 19:39:11,103 INFO [train.py:904] (6/8) Epoch 8, batch 1450, loss[loss=0.2088, simple_loss=0.2681, pruned_loss=0.0747, over 16744.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2688, pruned_loss=0.05518, over 3324221.67 frames. ], batch size: 134, lr: 8.77e-03, grad_scale: 8.0 2023-04-28 19:40:20,589 INFO [train.py:904] (6/8) Epoch 8, batch 1500, loss[loss=0.1864, simple_loss=0.2564, pruned_loss=0.05815, over 11951.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2696, pruned_loss=0.05605, over 3314354.71 frames. ], batch size: 246, lr: 8.77e-03, grad_scale: 8.0 2023-04-28 19:40:27,852 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 19:40:41,595 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2566, 3.4320, 1.8234, 3.6134, 2.4934, 3.5305, 1.8677, 2.6944], device='cuda:6'), covar=tensor([0.0204, 0.0280, 0.1380, 0.0123, 0.0683, 0.0463, 0.1338, 0.0583], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0159, 0.0180, 0.0103, 0.0162, 0.0197, 0.0189, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 19:40:52,030 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72575.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:40:57,834 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-28 19:41:17,186 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5458, 4.5381, 4.5143, 3.7500, 4.4564, 1.7456, 4.2893, 4.2768], device='cuda:6'), covar=tensor([0.0084, 0.0069, 0.0115, 0.0339, 0.0086, 0.2047, 0.0125, 0.0171], device='cuda:6'), in_proj_covar=tensor([0.0115, 0.0103, 0.0155, 0.0148, 0.0119, 0.0166, 0.0141, 0.0145], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:41:25,445 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.563e+02 3.021e+02 3.528e+02 6.404e+02, threshold=6.041e+02, percent-clipped=2.0 2023-04-28 19:41:28,319 INFO [train.py:904] (6/8) Epoch 8, batch 1550, loss[loss=0.2068, simple_loss=0.2951, pruned_loss=0.05928, over 17074.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2713, pruned_loss=0.0574, over 3316292.98 frames. ], batch size: 55, lr: 8.76e-03, grad_scale: 8.0 2023-04-28 19:41:40,960 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5542, 2.2191, 2.3919, 4.3583, 2.1443, 2.8428, 2.2784, 2.4383], device='cuda:6'), covar=tensor([0.0884, 0.2840, 0.1697, 0.0332, 0.3160, 0.1689, 0.2630, 0.2449], device='cuda:6'), in_proj_covar=tensor([0.0348, 0.0364, 0.0306, 0.0327, 0.0394, 0.0405, 0.0327, 0.0432], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:41:58,136 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=72623.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:42:39,324 INFO [train.py:904] (6/8) Epoch 8, batch 1600, loss[loss=0.2157, simple_loss=0.3056, pruned_loss=0.06288, over 16542.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.2737, pruned_loss=0.05828, over 3305061.00 frames. ], batch size: 62, lr: 8.76e-03, grad_scale: 8.0 2023-04-28 19:43:20,429 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72682.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:43:44,410 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.962e+02 2.814e+02 3.483e+02 4.056e+02 7.586e+02, threshold=6.967e+02, percent-clipped=5.0 2023-04-28 19:43:47,306 INFO [train.py:904] (6/8) Epoch 8, batch 1650, loss[loss=0.1815, simple_loss=0.2653, pruned_loss=0.04881, over 15892.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2749, pruned_loss=0.05859, over 3298106.82 frames. ], batch size: 35, lr: 8.76e-03, grad_scale: 8.0 2023-04-28 19:44:58,253 INFO [train.py:904] (6/8) Epoch 8, batch 1700, loss[loss=0.1988, simple_loss=0.2981, pruned_loss=0.04977, over 17255.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2779, pruned_loss=0.05992, over 3295028.99 frames. ], batch size: 52, lr: 8.76e-03, grad_scale: 8.0 2023-04-28 19:45:44,093 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72785.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:46:05,113 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.772e+02 2.648e+02 3.116e+02 3.582e+02 6.867e+02, threshold=6.231e+02, percent-clipped=0.0 2023-04-28 19:46:07,546 INFO [train.py:904] (6/8) Epoch 8, batch 1750, loss[loss=0.1806, simple_loss=0.2584, pruned_loss=0.05136, over 16981.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2791, pruned_loss=0.0596, over 3303803.47 frames. ], batch size: 41, lr: 8.75e-03, grad_scale: 8.0 2023-04-28 19:46:12,314 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0835, 4.8982, 5.1550, 5.3821, 5.5183, 4.8854, 5.4953, 5.4641], device='cuda:6'), covar=tensor([0.1274, 0.0985, 0.1403, 0.0564, 0.0444, 0.0619, 0.0407, 0.0498], device='cuda:6'), in_proj_covar=tensor([0.0490, 0.0612, 0.0760, 0.0616, 0.0463, 0.0466, 0.0476, 0.0523], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:46:21,541 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1685, 4.1387, 4.0338, 3.8627, 3.7746, 4.1259, 3.8503, 3.8811], device='cuda:6'), covar=tensor([0.0519, 0.0386, 0.0246, 0.0221, 0.0699, 0.0343, 0.0698, 0.0530], device='cuda:6'), in_proj_covar=tensor([0.0234, 0.0275, 0.0273, 0.0245, 0.0306, 0.0277, 0.0187, 0.0316], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 19:46:50,158 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=72833.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:47:03,675 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8160, 4.8130, 5.3742, 5.3293, 5.2968, 4.9397, 4.8475, 4.5986], device='cuda:6'), covar=tensor([0.0283, 0.0410, 0.0314, 0.0390, 0.0393, 0.0304, 0.0808, 0.0414], device='cuda:6'), in_proj_covar=tensor([0.0303, 0.0311, 0.0310, 0.0295, 0.0351, 0.0329, 0.0431, 0.0266], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 19:47:16,118 INFO [train.py:904] (6/8) Epoch 8, batch 1800, loss[loss=0.1794, simple_loss=0.2713, pruned_loss=0.04375, over 17074.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2794, pruned_loss=0.05964, over 3310575.88 frames. ], batch size: 53, lr: 8.75e-03, grad_scale: 4.0 2023-04-28 19:47:27,971 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1113, 5.1406, 4.9524, 4.2667, 4.8561, 1.6546, 4.6459, 4.9190], device='cuda:6'), covar=tensor([0.0069, 0.0052, 0.0130, 0.0428, 0.0083, 0.2368, 0.0135, 0.0180], device='cuda:6'), in_proj_covar=tensor([0.0114, 0.0102, 0.0154, 0.0148, 0.0119, 0.0166, 0.0140, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:48:11,278 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9648, 2.1264, 2.3256, 4.6365, 2.0084, 2.9203, 2.3306, 2.4076], device='cuda:6'), covar=tensor([0.0679, 0.3153, 0.1753, 0.0269, 0.3502, 0.1839, 0.2615, 0.3110], device='cuda:6'), in_proj_covar=tensor([0.0346, 0.0362, 0.0305, 0.0324, 0.0392, 0.0403, 0.0325, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:48:15,265 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72894.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:48:24,987 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.805e+02 2.735e+02 3.300e+02 4.209e+02 8.740e+02, threshold=6.600e+02, percent-clipped=4.0 2023-04-28 19:48:26,869 INFO [train.py:904] (6/8) Epoch 8, batch 1850, loss[loss=0.228, simple_loss=0.3119, pruned_loss=0.0721, over 16586.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2808, pruned_loss=0.0601, over 3307862.42 frames. ], batch size: 68, lr: 8.75e-03, grad_scale: 4.0 2023-04-28 19:49:35,219 INFO [train.py:904] (6/8) Epoch 8, batch 1900, loss[loss=0.2198, simple_loss=0.2899, pruned_loss=0.07481, over 16463.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2793, pruned_loss=0.05933, over 3309098.07 frames. ], batch size: 146, lr: 8.74e-03, grad_scale: 4.0 2023-04-28 19:49:39,775 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72955.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:49:52,065 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72964.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 19:50:12,146 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5697, 4.4916, 4.4795, 3.9946, 4.4659, 1.7914, 4.3176, 4.2470], device='cuda:6'), covar=tensor([0.0074, 0.0065, 0.0103, 0.0253, 0.0065, 0.1933, 0.0097, 0.0154], device='cuda:6'), in_proj_covar=tensor([0.0116, 0.0104, 0.0157, 0.0150, 0.0121, 0.0169, 0.0142, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:50:16,710 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72982.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:50:41,095 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.639e+02 2.549e+02 2.878e+02 3.342e+02 6.835e+02, threshold=5.756e+02, percent-clipped=2.0 2023-04-28 19:50:43,026 INFO [train.py:904] (6/8) Epoch 8, batch 1950, loss[loss=0.2196, simple_loss=0.2955, pruned_loss=0.07183, over 16763.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2787, pruned_loss=0.05853, over 3304356.12 frames. ], batch size: 134, lr: 8.74e-03, grad_scale: 4.0 2023-04-28 19:51:14,211 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73025.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 19:51:21,295 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=73030.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:51:50,194 INFO [train.py:904] (6/8) Epoch 8, batch 2000, loss[loss=0.1669, simple_loss=0.2497, pruned_loss=0.04202, over 17026.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2785, pruned_loss=0.05865, over 3309407.20 frames. ], batch size: 41, lr: 8.74e-03, grad_scale: 8.0 2023-04-28 19:52:42,916 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2023-04-28 19:52:58,783 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.718e+02 3.255e+02 3.787e+02 1.366e+03, threshold=6.510e+02, percent-clipped=4.0 2023-04-28 19:53:00,015 INFO [train.py:904] (6/8) Epoch 8, batch 2050, loss[loss=0.1917, simple_loss=0.2638, pruned_loss=0.05983, over 16833.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2789, pruned_loss=0.05906, over 3312796.53 frames. ], batch size: 96, lr: 8.73e-03, grad_scale: 8.0 2023-04-28 19:53:34,982 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8674, 4.7695, 4.6759, 4.2024, 4.7603, 2.0088, 4.5649, 4.6690], device='cuda:6'), covar=tensor([0.0079, 0.0070, 0.0141, 0.0290, 0.0070, 0.1928, 0.0108, 0.0130], device='cuda:6'), in_proj_covar=tensor([0.0117, 0.0105, 0.0157, 0.0151, 0.0121, 0.0169, 0.0142, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:53:45,986 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0879, 5.0107, 5.5328, 5.5389, 5.5444, 5.1462, 5.1682, 4.9030], device='cuda:6'), covar=tensor([0.0248, 0.0335, 0.0367, 0.0451, 0.0410, 0.0270, 0.0827, 0.0359], device='cuda:6'), in_proj_covar=tensor([0.0304, 0.0310, 0.0311, 0.0296, 0.0350, 0.0330, 0.0432, 0.0264], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 19:53:54,673 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9777, 4.2503, 4.3900, 3.3876, 3.8180, 4.4114, 3.9739, 2.8099], device='cuda:6'), covar=tensor([0.0304, 0.0032, 0.0025, 0.0195, 0.0066, 0.0043, 0.0044, 0.0275], device='cuda:6'), in_proj_covar=tensor([0.0122, 0.0067, 0.0064, 0.0119, 0.0069, 0.0080, 0.0072, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 19:54:09,332 INFO [train.py:904] (6/8) Epoch 8, batch 2100, loss[loss=0.1507, simple_loss=0.2423, pruned_loss=0.02952, over 17213.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2791, pruned_loss=0.05906, over 3312944.11 frames. ], batch size: 44, lr: 8.73e-03, grad_scale: 8.0 2023-04-28 19:55:14,761 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.921e+02 2.680e+02 3.272e+02 3.798e+02 7.634e+02, threshold=6.545e+02, percent-clipped=2.0 2023-04-28 19:55:16,491 INFO [train.py:904] (6/8) Epoch 8, batch 2150, loss[loss=0.2882, simple_loss=0.3442, pruned_loss=0.1162, over 12043.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2801, pruned_loss=0.0593, over 3309214.97 frames. ], batch size: 246, lr: 8.73e-03, grad_scale: 8.0 2023-04-28 19:55:31,760 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1411, 4.5293, 3.2947, 2.5166, 3.0454, 2.4920, 4.6831, 4.0381], device='cuda:6'), covar=tensor([0.2168, 0.0586, 0.1309, 0.2043, 0.2524, 0.1685, 0.0335, 0.1023], device='cuda:6'), in_proj_covar=tensor([0.0287, 0.0255, 0.0275, 0.0262, 0.0279, 0.0213, 0.0254, 0.0286], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:56:08,735 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 19:56:21,752 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73250.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:56:23,750 INFO [train.py:904] (6/8) Epoch 8, batch 2200, loss[loss=0.2034, simple_loss=0.277, pruned_loss=0.06492, over 16787.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2808, pruned_loss=0.06003, over 3314569.85 frames. ], batch size: 124, lr: 8.73e-03, grad_scale: 8.0 2023-04-28 19:56:47,763 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-28 19:57:20,374 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73293.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:57:29,785 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.707e+02 2.689e+02 3.130e+02 3.780e+02 7.244e+02, threshold=6.259e+02, percent-clipped=1.0 2023-04-28 19:57:31,974 INFO [train.py:904] (6/8) Epoch 8, batch 2250, loss[loss=0.1769, simple_loss=0.2684, pruned_loss=0.04264, over 17140.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2811, pruned_loss=0.05968, over 3312757.60 frames. ], batch size: 47, lr: 8.72e-03, grad_scale: 8.0 2023-04-28 19:57:56,961 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73320.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 19:57:58,315 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0868, 4.7954, 5.0176, 5.3151, 5.5086, 4.7146, 5.4331, 5.3839], device='cuda:6'), covar=tensor([0.1183, 0.1003, 0.1480, 0.0589, 0.0488, 0.0737, 0.0472, 0.0496], device='cuda:6'), in_proj_covar=tensor([0.0492, 0.0615, 0.0766, 0.0616, 0.0467, 0.0471, 0.0482, 0.0527], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:58:34,985 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9994, 4.3331, 3.3917, 2.4077, 3.1117, 2.4001, 4.7084, 4.1250], device='cuda:6'), covar=tensor([0.2226, 0.0625, 0.1299, 0.2100, 0.2475, 0.1730, 0.0339, 0.0843], device='cuda:6'), in_proj_covar=tensor([0.0289, 0.0256, 0.0277, 0.0264, 0.0281, 0.0213, 0.0256, 0.0288], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 19:58:40,764 INFO [train.py:904] (6/8) Epoch 8, batch 2300, loss[loss=0.1739, simple_loss=0.2576, pruned_loss=0.04505, over 16738.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2814, pruned_loss=0.05986, over 3312647.38 frames. ], batch size: 39, lr: 8.72e-03, grad_scale: 8.0 2023-04-28 19:58:44,944 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73354.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:58:46,332 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-28 19:59:48,860 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.976e+02 2.910e+02 3.420e+02 4.127e+02 1.105e+03, threshold=6.841e+02, percent-clipped=4.0 2023-04-28 19:59:49,986 INFO [train.py:904] (6/8) Epoch 8, batch 2350, loss[loss=0.215, simple_loss=0.2814, pruned_loss=0.07425, over 16909.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.282, pruned_loss=0.06084, over 3317220.17 frames. ], batch size: 96, lr: 8.72e-03, grad_scale: 8.0 2023-04-28 20:00:58,114 INFO [train.py:904] (6/8) Epoch 8, batch 2400, loss[loss=0.1838, simple_loss=0.2636, pruned_loss=0.05199, over 17209.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2825, pruned_loss=0.06134, over 3319642.18 frames. ], batch size: 44, lr: 8.71e-03, grad_scale: 8.0 2023-04-28 20:02:04,599 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.977e+02 2.637e+02 3.145e+02 3.743e+02 9.455e+02, threshold=6.290e+02, percent-clipped=3.0 2023-04-28 20:02:05,789 INFO [train.py:904] (6/8) Epoch 8, batch 2450, loss[loss=0.2079, simple_loss=0.2989, pruned_loss=0.05847, over 17144.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2829, pruned_loss=0.06054, over 3332276.86 frames. ], batch size: 47, lr: 8.71e-03, grad_scale: 8.0 2023-04-28 20:03:01,830 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6678, 3.7394, 4.0932, 4.0620, 4.0717, 3.7699, 3.8256, 3.6972], device='cuda:6'), covar=tensor([0.0347, 0.0494, 0.0333, 0.0341, 0.0369, 0.0361, 0.0721, 0.0515], device='cuda:6'), in_proj_covar=tensor([0.0301, 0.0308, 0.0307, 0.0293, 0.0348, 0.0327, 0.0429, 0.0265], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 20:03:12,633 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73550.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:03:14,439 INFO [train.py:904] (6/8) Epoch 8, batch 2500, loss[loss=0.2303, simple_loss=0.2914, pruned_loss=0.08462, over 16856.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2825, pruned_loss=0.06043, over 3331280.95 frames. ], batch size: 109, lr: 8.71e-03, grad_scale: 8.0 2023-04-28 20:04:07,521 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73590.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:04:18,852 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=73598.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:04:22,384 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.774e+02 3.514e+02 4.177e+02 6.912e+02, threshold=7.028e+02, percent-clipped=4.0 2023-04-28 20:04:23,448 INFO [train.py:904] (6/8) Epoch 8, batch 2550, loss[loss=0.2258, simple_loss=0.3249, pruned_loss=0.06335, over 16687.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2827, pruned_loss=0.06033, over 3333848.12 frames. ], batch size: 62, lr: 8.70e-03, grad_scale: 8.0 2023-04-28 20:04:37,714 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-28 20:04:49,477 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73620.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 20:05:29,133 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73649.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:05:32,257 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73651.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:05:32,975 INFO [train.py:904] (6/8) Epoch 8, batch 2600, loss[loss=0.1851, simple_loss=0.2812, pruned_loss=0.04447, over 17104.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2826, pruned_loss=0.0597, over 3331500.78 frames. ], batch size: 49, lr: 8.70e-03, grad_scale: 8.0 2023-04-28 20:05:55,318 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=73668.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 20:06:42,999 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.639e+02 3.290e+02 4.275e+02 7.096e+02, threshold=6.579e+02, percent-clipped=1.0 2023-04-28 20:06:43,015 INFO [train.py:904] (6/8) Epoch 8, batch 2650, loss[loss=0.166, simple_loss=0.2629, pruned_loss=0.03457, over 17111.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2819, pruned_loss=0.05885, over 3330622.51 frames. ], batch size: 47, lr: 8.70e-03, grad_scale: 4.0 2023-04-28 20:07:23,267 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73732.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:07:50,956 INFO [train.py:904] (6/8) Epoch 8, batch 2700, loss[loss=0.2264, simple_loss=0.2988, pruned_loss=0.07699, over 12717.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2814, pruned_loss=0.05808, over 3336888.88 frames. ], batch size: 246, lr: 8.70e-03, grad_scale: 4.0 2023-04-28 20:08:46,283 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73793.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:08:57,112 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.547e+02 3.040e+02 3.898e+02 5.528e+02, threshold=6.080e+02, percent-clipped=0.0 2023-04-28 20:08:57,127 INFO [train.py:904] (6/8) Epoch 8, batch 2750, loss[loss=0.2193, simple_loss=0.3035, pruned_loss=0.06759, over 16496.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2813, pruned_loss=0.05766, over 3340679.79 frames. ], batch size: 68, lr: 8.69e-03, grad_scale: 4.0 2023-04-28 20:09:52,017 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8244, 4.2623, 4.2757, 1.6420, 4.6141, 4.6752, 3.3447, 3.2086], device='cuda:6'), covar=tensor([0.0783, 0.0084, 0.0191, 0.1269, 0.0057, 0.0054, 0.0291, 0.0490], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0094, 0.0083, 0.0136, 0.0069, 0.0091, 0.0118, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 20:10:05,127 INFO [train.py:904] (6/8) Epoch 8, batch 2800, loss[loss=0.1965, simple_loss=0.274, pruned_loss=0.05953, over 16746.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2808, pruned_loss=0.05725, over 3337369.01 frames. ], batch size: 124, lr: 8.69e-03, grad_scale: 8.0 2023-04-28 20:10:53,006 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-28 20:11:14,609 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.850e+02 2.765e+02 3.378e+02 4.166e+02 1.030e+03, threshold=6.755e+02, percent-clipped=2.0 2023-04-28 20:11:14,625 INFO [train.py:904] (6/8) Epoch 8, batch 2850, loss[loss=0.1883, simple_loss=0.2782, pruned_loss=0.04917, over 17041.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2804, pruned_loss=0.0571, over 3333209.10 frames. ], batch size: 50, lr: 8.69e-03, grad_scale: 8.0 2023-04-28 20:11:24,916 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4031, 5.3743, 5.1775, 4.4652, 5.2879, 2.2077, 4.9726, 5.2110], device='cuda:6'), covar=tensor([0.0060, 0.0056, 0.0139, 0.0349, 0.0063, 0.1852, 0.0107, 0.0147], device='cuda:6'), in_proj_covar=tensor([0.0117, 0.0104, 0.0155, 0.0150, 0.0122, 0.0166, 0.0143, 0.0149], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 20:12:16,940 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73946.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:12:18,697 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-28 20:12:20,790 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73949.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:12:23,990 INFO [train.py:904] (6/8) Epoch 8, batch 2900, loss[loss=0.1841, simple_loss=0.2727, pruned_loss=0.0478, over 16788.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.281, pruned_loss=0.05827, over 3322050.88 frames. ], batch size: 57, lr: 8.68e-03, grad_scale: 8.0 2023-04-28 20:12:57,720 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8310, 3.9883, 3.0117, 2.4299, 2.7069, 2.4178, 3.8723, 3.6551], device='cuda:6'), covar=tensor([0.2108, 0.0523, 0.1378, 0.2025, 0.2263, 0.1552, 0.0465, 0.0952], device='cuda:6'), in_proj_covar=tensor([0.0288, 0.0256, 0.0277, 0.0263, 0.0281, 0.0211, 0.0255, 0.0285], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 20:13:28,885 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=73997.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:13:38,231 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.885e+02 2.787e+02 3.423e+02 3.955e+02 9.086e+02, threshold=6.847e+02, percent-clipped=3.0 2023-04-28 20:13:38,246 INFO [train.py:904] (6/8) Epoch 8, batch 2950, loss[loss=0.237, simple_loss=0.3016, pruned_loss=0.08615, over 16721.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2793, pruned_loss=0.05836, over 3329586.81 frames. ], batch size: 134, lr: 8.68e-03, grad_scale: 8.0 2023-04-28 20:13:44,486 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2831, 4.4343, 4.7448, 2.2673, 5.0376, 4.9938, 3.3240, 3.9278], device='cuda:6'), covar=tensor([0.0578, 0.0096, 0.0135, 0.1006, 0.0038, 0.0097, 0.0322, 0.0291], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0095, 0.0084, 0.0137, 0.0070, 0.0094, 0.0119, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 20:14:01,324 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-28 20:14:27,658 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5666, 2.1025, 2.3438, 4.2925, 2.0679, 2.7791, 2.2085, 2.4003], device='cuda:6'), covar=tensor([0.0802, 0.3073, 0.1641, 0.0318, 0.3181, 0.1762, 0.2664, 0.2637], device='cuda:6'), in_proj_covar=tensor([0.0353, 0.0367, 0.0308, 0.0328, 0.0395, 0.0412, 0.0330, 0.0437], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 20:14:46,251 INFO [train.py:904] (6/8) Epoch 8, batch 3000, loss[loss=0.2184, simple_loss=0.3131, pruned_loss=0.06181, over 17106.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2789, pruned_loss=0.05854, over 3330382.66 frames. ], batch size: 48, lr: 8.68e-03, grad_scale: 8.0 2023-04-28 20:14:46,251 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 20:14:55,853 INFO [train.py:938] (6/8) Epoch 8, validation: loss=0.1462, simple_loss=0.2525, pruned_loss=0.01995, over 944034.00 frames. 2023-04-28 20:14:55,853 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-28 20:15:13,495 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74064.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:15:45,913 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74088.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:15:57,064 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-28 20:16:06,763 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 2.715e+02 3.274e+02 3.889e+02 8.736e+02, threshold=6.548e+02, percent-clipped=1.0 2023-04-28 20:16:06,779 INFO [train.py:904] (6/8) Epoch 8, batch 3050, loss[loss=0.2221, simple_loss=0.2915, pruned_loss=0.07635, over 16836.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2796, pruned_loss=0.05937, over 3330675.33 frames. ], batch size: 116, lr: 8.68e-03, grad_scale: 8.0 2023-04-28 20:16:30,472 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-28 20:16:38,679 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74125.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:17:13,362 INFO [train.py:904] (6/8) Epoch 8, batch 3100, loss[loss=0.1729, simple_loss=0.2683, pruned_loss=0.03879, over 17264.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2791, pruned_loss=0.05935, over 3325309.25 frames. ], batch size: 52, lr: 8.67e-03, grad_scale: 8.0 2023-04-28 20:17:17,084 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.23 vs. limit=5.0 2023-04-28 20:18:21,290 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.831e+02 2.546e+02 2.917e+02 3.424e+02 8.256e+02, threshold=5.834e+02, percent-clipped=1.0 2023-04-28 20:18:21,306 INFO [train.py:904] (6/8) Epoch 8, batch 3150, loss[loss=0.2229, simple_loss=0.2969, pruned_loss=0.07443, over 16529.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2779, pruned_loss=0.05859, over 3329210.27 frames. ], batch size: 68, lr: 8.67e-03, grad_scale: 8.0 2023-04-28 20:19:15,041 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74240.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:19:23,310 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74246.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:19:32,370 INFO [train.py:904] (6/8) Epoch 8, batch 3200, loss[loss=0.1832, simple_loss=0.2832, pruned_loss=0.04158, over 16764.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2776, pruned_loss=0.05832, over 3332019.24 frames. ], batch size: 57, lr: 8.67e-03, grad_scale: 8.0 2023-04-28 20:20:08,720 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1810, 5.1368, 4.9462, 4.6727, 4.5369, 5.0062, 5.1147, 4.5982], device='cuda:6'), covar=tensor([0.0504, 0.0401, 0.0240, 0.0248, 0.1084, 0.0366, 0.0226, 0.0679], device='cuda:6'), in_proj_covar=tensor([0.0239, 0.0284, 0.0280, 0.0253, 0.0318, 0.0284, 0.0192, 0.0324], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 20:20:31,326 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=74294.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:20:40,613 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74301.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:20:41,301 INFO [train.py:904] (6/8) Epoch 8, batch 3250, loss[loss=0.2226, simple_loss=0.2912, pruned_loss=0.07702, over 16755.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2777, pruned_loss=0.05825, over 3333004.38 frames. ], batch size: 124, lr: 8.66e-03, grad_scale: 4.0 2023-04-28 20:20:42,357 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.676e+02 2.591e+02 3.147e+02 3.802e+02 9.622e+02, threshold=6.293e+02, percent-clipped=5.0 2023-04-28 20:21:52,159 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-28 20:21:52,434 INFO [train.py:904] (6/8) Epoch 8, batch 3300, loss[loss=0.2008, simple_loss=0.2793, pruned_loss=0.06118, over 16869.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2784, pruned_loss=0.05821, over 3338631.42 frames. ], batch size: 96, lr: 8.66e-03, grad_scale: 4.0 2023-04-28 20:22:41,305 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74388.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:23:01,500 INFO [train.py:904] (6/8) Epoch 8, batch 3350, loss[loss=0.1834, simple_loss=0.2762, pruned_loss=0.04533, over 17046.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2786, pruned_loss=0.05801, over 3327578.94 frames. ], batch size: 50, lr: 8.66e-03, grad_scale: 4.0 2023-04-28 20:23:02,761 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.835e+02 2.621e+02 3.247e+02 4.157e+02 8.305e+02, threshold=6.494e+02, percent-clipped=1.0 2023-04-28 20:23:27,943 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74420.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:23:49,803 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=74436.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:24:11,179 INFO [train.py:904] (6/8) Epoch 8, batch 3400, loss[loss=0.2088, simple_loss=0.3066, pruned_loss=0.0555, over 17075.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.279, pruned_loss=0.05831, over 3329187.29 frames. ], batch size: 53, lr: 8.66e-03, grad_scale: 4.0 2023-04-28 20:25:21,711 INFO [train.py:904] (6/8) Epoch 8, batch 3450, loss[loss=0.1654, simple_loss=0.2594, pruned_loss=0.03572, over 17165.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.2766, pruned_loss=0.05681, over 3327324.61 frames. ], batch size: 46, lr: 8.65e-03, grad_scale: 4.0 2023-04-28 20:25:22,838 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.611e+02 3.016e+02 3.656e+02 6.729e+02, threshold=6.032e+02, percent-clipped=3.0 2023-04-28 20:25:32,128 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9211, 4.3236, 2.2251, 4.6176, 3.0116, 4.6437, 2.4611, 3.3127], device='cuda:6'), covar=tensor([0.0200, 0.0271, 0.1519, 0.0146, 0.0826, 0.0382, 0.1375, 0.0608], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0163, 0.0178, 0.0110, 0.0165, 0.0205, 0.0190, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 20:25:50,511 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0934, 3.9922, 4.4313, 1.9843, 4.7355, 4.6348, 3.2954, 3.8513], device='cuda:6'), covar=tensor([0.0600, 0.0168, 0.0201, 0.1010, 0.0048, 0.0113, 0.0342, 0.0277], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0094, 0.0084, 0.0134, 0.0069, 0.0092, 0.0117, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 20:26:24,799 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2149, 3.0629, 3.0691, 1.7769, 3.3415, 3.2513, 2.6954, 2.5940], device='cuda:6'), covar=tensor([0.0736, 0.0176, 0.0259, 0.1019, 0.0087, 0.0189, 0.0433, 0.0439], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0095, 0.0085, 0.0136, 0.0070, 0.0093, 0.0118, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 20:26:26,854 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-28 20:26:30,765 INFO [train.py:904] (6/8) Epoch 8, batch 3500, loss[loss=0.1908, simple_loss=0.2825, pruned_loss=0.04958, over 17249.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2758, pruned_loss=0.05695, over 3316771.57 frames. ], batch size: 52, lr: 8.65e-03, grad_scale: 4.0 2023-04-28 20:26:37,539 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1298, 3.9225, 4.0495, 4.2852, 4.3826, 3.9241, 4.0380, 4.3385], device='cuda:6'), covar=tensor([0.1061, 0.0855, 0.1265, 0.0535, 0.0476, 0.1334, 0.1812, 0.0544], device='cuda:6'), in_proj_covar=tensor([0.0511, 0.0629, 0.0788, 0.0641, 0.0483, 0.0485, 0.0498, 0.0549], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 20:27:32,520 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74595.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:27:34,460 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74596.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:27:42,864 INFO [train.py:904] (6/8) Epoch 8, batch 3550, loss[loss=0.2125, simple_loss=0.2824, pruned_loss=0.07129, over 16742.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2748, pruned_loss=0.05648, over 3320340.12 frames. ], batch size: 124, lr: 8.65e-03, grad_scale: 4.0 2023-04-28 20:27:43,959 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.810e+02 2.461e+02 3.024e+02 3.861e+02 7.667e+02, threshold=6.049e+02, percent-clipped=4.0 2023-04-28 20:28:51,895 INFO [train.py:904] (6/8) Epoch 8, batch 3600, loss[loss=0.1787, simple_loss=0.2539, pruned_loss=0.05176, over 16202.00 frames. ], tot_loss[loss=0.1925, simple_loss=0.2735, pruned_loss=0.05577, over 3319360.24 frames. ], batch size: 165, lr: 8.64e-03, grad_scale: 8.0 2023-04-28 20:28:57,000 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74656.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:29:35,220 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0510, 5.0661, 5.6335, 5.6054, 5.5676, 5.2114, 5.1861, 5.0160], device='cuda:6'), covar=tensor([0.0286, 0.0435, 0.0287, 0.0343, 0.0433, 0.0250, 0.0832, 0.0346], device='cuda:6'), in_proj_covar=tensor([0.0309, 0.0314, 0.0313, 0.0303, 0.0358, 0.0332, 0.0440, 0.0269], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 20:30:00,903 INFO [train.py:904] (6/8) Epoch 8, batch 3650, loss[loss=0.1809, simple_loss=0.2504, pruned_loss=0.05568, over 16477.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2726, pruned_loss=0.05673, over 3306524.74 frames. ], batch size: 75, lr: 8.64e-03, grad_scale: 8.0 2023-04-28 20:30:02,113 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.583e+02 2.393e+02 2.913e+02 3.939e+02 7.282e+02, threshold=5.826e+02, percent-clipped=2.0 2023-04-28 20:30:27,988 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74720.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:31:13,893 INFO [train.py:904] (6/8) Epoch 8, batch 3700, loss[loss=0.1765, simple_loss=0.251, pruned_loss=0.05106, over 16888.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2714, pruned_loss=0.05854, over 3290628.95 frames. ], batch size: 96, lr: 8.64e-03, grad_scale: 8.0 2023-04-28 20:31:38,766 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=74768.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:32:12,015 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2340, 4.3227, 4.7283, 4.6687, 4.7189, 4.3394, 4.3871, 4.1815], device='cuda:6'), covar=tensor([0.0294, 0.0502, 0.0291, 0.0415, 0.0361, 0.0306, 0.0706, 0.0464], device='cuda:6'), in_proj_covar=tensor([0.0301, 0.0306, 0.0305, 0.0295, 0.0350, 0.0324, 0.0429, 0.0262], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 20:32:29,720 INFO [train.py:904] (6/8) Epoch 8, batch 3750, loss[loss=0.2295, simple_loss=0.3066, pruned_loss=0.07623, over 16599.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2718, pruned_loss=0.06018, over 3283230.81 frames. ], batch size: 57, lr: 8.64e-03, grad_scale: 8.0 2023-04-28 20:32:30,688 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.639e+02 3.130e+02 3.703e+02 1.006e+03, threshold=6.260e+02, percent-clipped=3.0 2023-04-28 20:33:41,190 INFO [train.py:904] (6/8) Epoch 8, batch 3800, loss[loss=0.1998, simple_loss=0.2777, pruned_loss=0.06095, over 16667.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2739, pruned_loss=0.06227, over 3276608.41 frames. ], batch size: 57, lr: 8.63e-03, grad_scale: 8.0 2023-04-28 20:34:04,329 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3802, 5.3012, 5.1909, 4.3760, 5.3272, 1.8244, 5.0552, 5.0896], device='cuda:6'), covar=tensor([0.0053, 0.0048, 0.0092, 0.0331, 0.0047, 0.2162, 0.0086, 0.0140], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0108, 0.0159, 0.0153, 0.0124, 0.0168, 0.0143, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 20:34:20,820 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3138, 2.9232, 2.5080, 2.1782, 2.1769, 2.0577, 2.8281, 2.8995], device='cuda:6'), covar=tensor([0.2291, 0.0783, 0.1358, 0.1992, 0.1900, 0.1823, 0.0535, 0.0830], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0258, 0.0279, 0.0267, 0.0290, 0.0215, 0.0259, 0.0290], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 20:34:45,384 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74896.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:34:45,597 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0734, 3.4311, 3.2316, 2.0046, 2.7119, 2.3440, 3.5997, 3.4073], device='cuda:6'), covar=tensor([0.0206, 0.0572, 0.0528, 0.1485, 0.0750, 0.0885, 0.0449, 0.0734], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0141, 0.0155, 0.0139, 0.0134, 0.0124, 0.0135, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 20:34:51,947 INFO [train.py:904] (6/8) Epoch 8, batch 3850, loss[loss=0.1895, simple_loss=0.2542, pruned_loss=0.06236, over 16902.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2745, pruned_loss=0.06288, over 3271200.29 frames. ], batch size: 90, lr: 8.63e-03, grad_scale: 8.0 2023-04-28 20:34:53,140 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 2.503e+02 3.025e+02 3.649e+02 5.657e+02, threshold=6.049e+02, percent-clipped=0.0 2023-04-28 20:35:03,392 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74909.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:35:30,633 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6809, 3.8198, 4.0241, 2.0028, 4.1510, 4.1289, 3.2507, 3.0595], device='cuda:6'), covar=tensor([0.0718, 0.0138, 0.0123, 0.1116, 0.0053, 0.0094, 0.0307, 0.0387], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0098, 0.0085, 0.0139, 0.0070, 0.0093, 0.0121, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 20:35:39,054 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8375, 3.0800, 2.4413, 4.3757, 3.5882, 4.1966, 1.6139, 2.9148], device='cuda:6'), covar=tensor([0.1325, 0.0575, 0.1115, 0.0142, 0.0293, 0.0347, 0.1353, 0.0826], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0152, 0.0172, 0.0120, 0.0201, 0.0207, 0.0170, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 20:35:52,792 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=74944.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:36:01,456 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74951.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:36:02,297 INFO [train.py:904] (6/8) Epoch 8, batch 3900, loss[loss=0.1747, simple_loss=0.2542, pruned_loss=0.04757, over 16575.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2732, pruned_loss=0.06299, over 3274014.51 frames. ], batch size: 75, lr: 8.63e-03, grad_scale: 8.0 2023-04-28 20:36:20,391 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74964.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:36:28,119 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74970.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:37:08,142 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-04-28 20:37:12,341 INFO [train.py:904] (6/8) Epoch 8, batch 3950, loss[loss=0.1923, simple_loss=0.2638, pruned_loss=0.06039, over 16862.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2729, pruned_loss=0.06366, over 3280795.29 frames. ], batch size: 96, lr: 8.62e-03, grad_scale: 8.0 2023-04-28 20:37:14,091 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.920e+02 2.648e+02 3.131e+02 3.735e+02 8.073e+02, threshold=6.262e+02, percent-clipped=3.0 2023-04-28 20:37:21,788 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7285, 4.9012, 5.0022, 4.9470, 5.0100, 5.5546, 5.1077, 4.8500], device='cuda:6'), covar=tensor([0.1143, 0.1587, 0.1629, 0.1574, 0.2103, 0.0824, 0.1192, 0.1957], device='cuda:6'), in_proj_covar=tensor([0.0329, 0.0463, 0.0482, 0.0406, 0.0531, 0.0506, 0.0387, 0.0542], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 20:37:46,109 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75025.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:38:23,636 INFO [train.py:904] (6/8) Epoch 8, batch 4000, loss[loss=0.2052, simple_loss=0.273, pruned_loss=0.06865, over 16740.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2725, pruned_loss=0.06359, over 3287147.85 frames. ], batch size: 124, lr: 8.62e-03, grad_scale: 8.0 2023-04-28 20:38:45,930 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9318, 1.6360, 2.2342, 2.7403, 2.6616, 2.9609, 1.8540, 3.0103], device='cuda:6'), covar=tensor([0.0110, 0.0289, 0.0204, 0.0154, 0.0163, 0.0098, 0.0266, 0.0043], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0160, 0.0144, 0.0148, 0.0154, 0.0111, 0.0159, 0.0100], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 20:38:49,557 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9286, 3.3382, 3.3606, 2.1944, 3.0893, 3.4038, 3.2437, 1.8195], device='cuda:6'), covar=tensor([0.0381, 0.0043, 0.0028, 0.0260, 0.0056, 0.0061, 0.0045, 0.0325], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0065, 0.0064, 0.0118, 0.0069, 0.0078, 0.0070, 0.0113], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 20:39:36,987 INFO [train.py:904] (6/8) Epoch 8, batch 4050, loss[loss=0.1697, simple_loss=0.2553, pruned_loss=0.04206, over 16742.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2722, pruned_loss=0.0621, over 3289204.73 frames. ], batch size: 83, lr: 8.62e-03, grad_scale: 8.0 2023-04-28 20:39:38,164 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.447e+02 2.780e+02 3.424e+02 6.417e+02, threshold=5.561e+02, percent-clipped=2.0 2023-04-28 20:39:42,894 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-04-28 20:40:49,053 INFO [train.py:904] (6/8) Epoch 8, batch 4100, loss[loss=0.1791, simple_loss=0.2699, pruned_loss=0.04417, over 16352.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2738, pruned_loss=0.06134, over 3279858.92 frames. ], batch size: 146, lr: 8.62e-03, grad_scale: 8.0 2023-04-28 20:42:02,418 INFO [train.py:904] (6/8) Epoch 8, batch 4150, loss[loss=0.1888, simple_loss=0.2719, pruned_loss=0.05288, over 17020.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2814, pruned_loss=0.06402, over 3274285.29 frames. ], batch size: 55, lr: 8.61e-03, grad_scale: 8.0 2023-04-28 20:42:04,251 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.780e+02 2.372e+02 3.014e+02 3.829e+02 9.608e+02, threshold=6.029e+02, percent-clipped=8.0 2023-04-28 20:42:43,304 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=75228.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:43:19,560 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75251.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:43:20,232 INFO [train.py:904] (6/8) Epoch 8, batch 4200, loss[loss=0.2561, simple_loss=0.3258, pruned_loss=0.09322, over 11301.00 frames. ], tot_loss[loss=0.2111, simple_loss=0.2889, pruned_loss=0.06659, over 3220210.06 frames. ], batch size: 248, lr: 8.61e-03, grad_scale: 8.0 2023-04-28 20:43:41,423 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75265.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:43:48,069 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=75270.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:43:52,336 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3391, 3.4440, 1.7713, 3.6246, 2.5429, 3.6799, 2.0768, 2.7585], device='cuda:6'), covar=tensor([0.0209, 0.0309, 0.1694, 0.0216, 0.0729, 0.0494, 0.1479, 0.0706], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0164, 0.0184, 0.0107, 0.0166, 0.0203, 0.0191, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 20:44:16,455 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75289.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:44:30,800 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=75299.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:44:34,932 INFO [train.py:904] (6/8) Epoch 8, batch 4250, loss[loss=0.1932, simple_loss=0.2867, pruned_loss=0.04992, over 16764.00 frames. ], tot_loss[loss=0.2133, simple_loss=0.2924, pruned_loss=0.0671, over 3183935.02 frames. ], batch size: 83, lr: 8.61e-03, grad_scale: 8.0 2023-04-28 20:44:36,196 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.726e+02 2.663e+02 3.246e+02 3.785e+02 9.237e+02, threshold=6.492e+02, percent-clipped=4.0 2023-04-28 20:44:49,340 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 20:45:01,375 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75320.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:45:18,380 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75331.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 20:45:48,478 INFO [train.py:904] (6/8) Epoch 8, batch 4300, loss[loss=0.2129, simple_loss=0.299, pruned_loss=0.06343, over 16293.00 frames. ], tot_loss[loss=0.2129, simple_loss=0.2937, pruned_loss=0.06612, over 3187065.73 frames. ], batch size: 165, lr: 8.60e-03, grad_scale: 8.0 2023-04-28 20:47:02,655 INFO [train.py:904] (6/8) Epoch 8, batch 4350, loss[loss=0.2153, simple_loss=0.304, pruned_loss=0.06333, over 16782.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.2972, pruned_loss=0.06689, over 3185910.57 frames. ], batch size: 83, lr: 8.60e-03, grad_scale: 8.0 2023-04-28 20:47:03,852 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.683e+02 2.615e+02 3.065e+02 3.856e+02 8.729e+02, threshold=6.129e+02, percent-clipped=2.0 2023-04-28 20:47:21,314 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-28 20:48:17,345 INFO [train.py:904] (6/8) Epoch 8, batch 4400, loss[loss=0.1963, simple_loss=0.2799, pruned_loss=0.05641, over 16971.00 frames. ], tot_loss[loss=0.2184, simple_loss=0.2999, pruned_loss=0.06844, over 3162464.47 frames. ], batch size: 41, lr: 8.60e-03, grad_scale: 8.0 2023-04-28 20:48:38,812 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9037, 2.7114, 2.6050, 1.9333, 2.5306, 2.7326, 2.6465, 1.7682], device='cuda:6'), covar=tensor([0.0307, 0.0037, 0.0033, 0.0242, 0.0062, 0.0058, 0.0049, 0.0281], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0064, 0.0063, 0.0119, 0.0069, 0.0077, 0.0070, 0.0112], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 20:49:26,864 INFO [train.py:904] (6/8) Epoch 8, batch 4450, loss[loss=0.2418, simple_loss=0.3098, pruned_loss=0.08685, over 11503.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.3026, pruned_loss=0.06903, over 3166214.31 frames. ], batch size: 247, lr: 8.60e-03, grad_scale: 8.0 2023-04-28 20:49:28,913 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.480e+02 2.422e+02 2.913e+02 3.508e+02 6.103e+02, threshold=5.826e+02, percent-clipped=0.0 2023-04-28 20:50:38,184 INFO [train.py:904] (6/8) Epoch 8, batch 4500, loss[loss=0.2227, simple_loss=0.305, pruned_loss=0.07023, over 16698.00 frames. ], tot_loss[loss=0.2206, simple_loss=0.3024, pruned_loss=0.06937, over 3171040.06 frames. ], batch size: 76, lr: 8.59e-03, grad_scale: 8.0 2023-04-28 20:50:57,799 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75565.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:51:24,937 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75584.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:51:41,675 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.62 vs. limit=2.0 2023-04-28 20:51:51,099 INFO [train.py:904] (6/8) Epoch 8, batch 4550, loss[loss=0.2164, simple_loss=0.3079, pruned_loss=0.06245, over 17242.00 frames. ], tot_loss[loss=0.2207, simple_loss=0.3026, pruned_loss=0.06934, over 3192893.38 frames. ], batch size: 52, lr: 8.59e-03, grad_scale: 8.0 2023-04-28 20:51:52,274 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.189e+02 2.672e+02 3.111e+02 5.807e+02, threshold=5.345e+02, percent-clipped=0.0 2023-04-28 20:52:06,458 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=75613.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:52:17,125 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75620.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:52:25,163 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75626.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 20:52:52,907 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4415, 3.5292, 3.2639, 3.1309, 3.0258, 3.3793, 3.1923, 3.1383], device='cuda:6'), covar=tensor([0.0467, 0.0286, 0.0244, 0.0209, 0.0572, 0.0295, 0.1547, 0.0515], device='cuda:6'), in_proj_covar=tensor([0.0214, 0.0251, 0.0252, 0.0225, 0.0280, 0.0253, 0.0173, 0.0285], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 20:53:02,628 INFO [train.py:904] (6/8) Epoch 8, batch 4600, loss[loss=0.2115, simple_loss=0.3017, pruned_loss=0.06067, over 16573.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.3036, pruned_loss=0.06912, over 3204314.91 frames. ], batch size: 75, lr: 8.59e-03, grad_scale: 8.0 2023-04-28 20:53:25,887 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=75668.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:53:32,545 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.73 vs. limit=5.0 2023-04-28 20:54:09,595 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2023-04-28 20:54:12,064 INFO [train.py:904] (6/8) Epoch 8, batch 4650, loss[loss=0.1965, simple_loss=0.2852, pruned_loss=0.05391, over 16264.00 frames. ], tot_loss[loss=0.22, simple_loss=0.3024, pruned_loss=0.06879, over 3210911.13 frames. ], batch size: 165, lr: 8.58e-03, grad_scale: 8.0 2023-04-28 20:54:13,263 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.294e+02 2.810e+02 3.370e+02 9.861e+02, threshold=5.619e+02, percent-clipped=3.0 2023-04-28 20:55:23,490 INFO [train.py:904] (6/8) Epoch 8, batch 4700, loss[loss=0.2321, simple_loss=0.323, pruned_loss=0.07059, over 15389.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.3002, pruned_loss=0.06809, over 3188321.45 frames. ], batch size: 190, lr: 8.58e-03, grad_scale: 8.0 2023-04-28 20:55:33,704 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9565, 4.9084, 4.9475, 3.4122, 4.2028, 4.7951, 4.2086, 2.7768], device='cuda:6'), covar=tensor([0.0321, 0.0012, 0.0012, 0.0214, 0.0038, 0.0037, 0.0032, 0.0249], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0063, 0.0062, 0.0119, 0.0068, 0.0078, 0.0070, 0.0112], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 20:55:58,165 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0417, 1.8915, 2.4975, 3.1859, 3.0564, 3.6105, 1.9802, 3.4438], device='cuda:6'), covar=tensor([0.0129, 0.0320, 0.0217, 0.0157, 0.0141, 0.0085, 0.0315, 0.0065], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0160, 0.0144, 0.0146, 0.0153, 0.0107, 0.0160, 0.0100], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 20:56:16,004 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4342, 4.2049, 4.4212, 4.6332, 4.7577, 4.3118, 4.6978, 4.7709], device='cuda:6'), covar=tensor([0.1069, 0.0952, 0.1262, 0.0462, 0.0370, 0.0860, 0.0442, 0.0366], device='cuda:6'), in_proj_covar=tensor([0.0458, 0.0562, 0.0698, 0.0569, 0.0427, 0.0435, 0.0437, 0.0484], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 20:56:31,924 INFO [train.py:904] (6/8) Epoch 8, batch 4750, loss[loss=0.2066, simple_loss=0.2869, pruned_loss=0.06311, over 16266.00 frames. ], tot_loss[loss=0.2133, simple_loss=0.2953, pruned_loss=0.06564, over 3198696.89 frames. ], batch size: 165, lr: 8.58e-03, grad_scale: 8.0 2023-04-28 20:56:33,066 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.310e+02 2.114e+02 2.531e+02 3.127e+02 7.196e+02, threshold=5.061e+02, percent-clipped=1.0 2023-04-28 20:57:44,161 INFO [train.py:904] (6/8) Epoch 8, batch 4800, loss[loss=0.2454, simple_loss=0.3174, pruned_loss=0.08671, over 11695.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2913, pruned_loss=0.06352, over 3191917.71 frames. ], batch size: 246, lr: 8.58e-03, grad_scale: 8.0 2023-04-28 20:58:08,999 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-04-28 20:58:22,995 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4816, 4.1152, 4.1091, 3.0168, 3.5588, 4.0474, 3.6936, 2.2910], device='cuda:6'), covar=tensor([0.0372, 0.0019, 0.0018, 0.0210, 0.0053, 0.0065, 0.0049, 0.0304], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0063, 0.0062, 0.0119, 0.0068, 0.0078, 0.0070, 0.0112], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 20:58:23,047 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1772, 2.3672, 1.8354, 2.0925, 2.7827, 2.4364, 3.1121, 3.0642], device='cuda:6'), covar=tensor([0.0055, 0.0260, 0.0383, 0.0314, 0.0150, 0.0277, 0.0095, 0.0127], device='cuda:6'), in_proj_covar=tensor([0.0113, 0.0179, 0.0177, 0.0178, 0.0177, 0.0182, 0.0177, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 20:58:32,165 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75884.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:58:58,545 INFO [train.py:904] (6/8) Epoch 8, batch 4850, loss[loss=0.2214, simple_loss=0.3043, pruned_loss=0.06928, over 16593.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.2924, pruned_loss=0.06255, over 3201041.64 frames. ], batch size: 57, lr: 8.57e-03, grad_scale: 8.0 2023-04-28 20:59:01,503 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.327e+02 2.698e+02 3.138e+02 6.949e+02, threshold=5.395e+02, percent-clipped=1.0 2023-04-28 20:59:36,927 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75926.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 20:59:46,581 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=75932.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:00:17,705 INFO [train.py:904] (6/8) Epoch 8, batch 4900, loss[loss=0.1861, simple_loss=0.2801, pruned_loss=0.04601, over 16718.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2915, pruned_loss=0.0614, over 3187494.18 frames. ], batch size: 134, lr: 8.57e-03, grad_scale: 8.0 2023-04-28 21:00:49,946 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=75974.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:01:34,529 INFO [train.py:904] (6/8) Epoch 8, batch 4950, loss[loss=0.211, simple_loss=0.2944, pruned_loss=0.06379, over 17151.00 frames. ], tot_loss[loss=0.2058, simple_loss=0.2909, pruned_loss=0.06037, over 3192452.15 frames. ], batch size: 47, lr: 8.57e-03, grad_scale: 8.0 2023-04-28 21:01:36,817 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.340e+02 2.828e+02 3.481e+02 8.052e+02, threshold=5.656e+02, percent-clipped=2.0 2023-04-28 21:02:11,507 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-04-28 21:02:12,108 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9547, 4.8471, 4.8174, 4.1429, 4.9107, 1.6967, 4.5875, 4.7685], device='cuda:6'), covar=tensor([0.0063, 0.0064, 0.0084, 0.0347, 0.0065, 0.1946, 0.0092, 0.0125], device='cuda:6'), in_proj_covar=tensor([0.0110, 0.0098, 0.0146, 0.0142, 0.0113, 0.0159, 0.0131, 0.0136], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:02:20,678 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5046, 3.5806, 3.2703, 3.0633, 3.0765, 3.3679, 3.2551, 3.2206], device='cuda:6'), covar=tensor([0.0491, 0.0325, 0.0234, 0.0202, 0.0670, 0.0318, 0.1163, 0.0469], device='cuda:6'), in_proj_covar=tensor([0.0215, 0.0253, 0.0252, 0.0225, 0.0282, 0.0255, 0.0170, 0.0287], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:02:45,330 INFO [train.py:904] (6/8) Epoch 8, batch 5000, loss[loss=0.2058, simple_loss=0.2973, pruned_loss=0.05718, over 15567.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.292, pruned_loss=0.06047, over 3190161.01 frames. ], batch size: 191, lr: 8.56e-03, grad_scale: 8.0 2023-04-28 21:03:55,754 INFO [train.py:904] (6/8) Epoch 8, batch 5050, loss[loss=0.2198, simple_loss=0.3107, pruned_loss=0.06446, over 17019.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2929, pruned_loss=0.06042, over 3188170.36 frames. ], batch size: 55, lr: 8.56e-03, grad_scale: 8.0 2023-04-28 21:03:57,927 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.576e+02 2.969e+02 3.606e+02 8.836e+02, threshold=5.938e+02, percent-clipped=5.0 2023-04-28 21:04:04,452 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7965, 4.5522, 4.7517, 4.9293, 5.1273, 4.6369, 5.0866, 5.0708], device='cuda:6'), covar=tensor([0.1110, 0.0881, 0.1271, 0.0503, 0.0358, 0.0576, 0.0376, 0.0375], device='cuda:6'), in_proj_covar=tensor([0.0468, 0.0571, 0.0713, 0.0583, 0.0439, 0.0443, 0.0447, 0.0497], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:04:57,180 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-28 21:05:07,314 INFO [train.py:904] (6/8) Epoch 8, batch 5100, loss[loss=0.1844, simple_loss=0.2671, pruned_loss=0.05085, over 16393.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2912, pruned_loss=0.05984, over 3193527.70 frames. ], batch size: 35, lr: 8.56e-03, grad_scale: 8.0 2023-04-28 21:05:12,264 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.20 vs. limit=5.0 2023-04-28 21:06:20,870 INFO [train.py:904] (6/8) Epoch 8, batch 5150, loss[loss=0.1981, simple_loss=0.2849, pruned_loss=0.05564, over 16725.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2912, pruned_loss=0.05908, over 3196786.81 frames. ], batch size: 62, lr: 8.56e-03, grad_scale: 8.0 2023-04-28 21:06:24,096 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.361e+02 2.714e+02 3.177e+02 5.443e+02, threshold=5.429e+02, percent-clipped=0.0 2023-04-28 21:06:57,151 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5194, 4.8198, 4.5257, 4.5677, 4.2509, 4.2853, 4.3084, 4.8263], device='cuda:6'), covar=tensor([0.0811, 0.0675, 0.0932, 0.0589, 0.0701, 0.1025, 0.0842, 0.0837], device='cuda:6'), in_proj_covar=tensor([0.0462, 0.0586, 0.0495, 0.0400, 0.0373, 0.0386, 0.0489, 0.0431], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:07:13,097 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4681, 4.5311, 4.6717, 4.5797, 4.6888, 5.1948, 4.7617, 4.4491], device='cuda:6'), covar=tensor([0.1003, 0.1986, 0.1579, 0.1623, 0.2312, 0.0919, 0.1132, 0.2005], device='cuda:6'), in_proj_covar=tensor([0.0317, 0.0432, 0.0448, 0.0378, 0.0506, 0.0478, 0.0363, 0.0513], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 21:07:33,712 INFO [train.py:904] (6/8) Epoch 8, batch 5200, loss[loss=0.1986, simple_loss=0.2893, pruned_loss=0.0539, over 16692.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.29, pruned_loss=0.05865, over 3200789.65 frames. ], batch size: 134, lr: 8.55e-03, grad_scale: 8.0 2023-04-28 21:07:34,623 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-28 21:07:59,804 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76269.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 21:08:36,426 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76294.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:08:48,737 INFO [train.py:904] (6/8) Epoch 8, batch 5250, loss[loss=0.2546, simple_loss=0.3155, pruned_loss=0.09687, over 12511.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2875, pruned_loss=0.05864, over 3209001.67 frames. ], batch size: 247, lr: 8.55e-03, grad_scale: 8.0 2023-04-28 21:08:51,147 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.354e+02 2.757e+02 3.436e+02 6.643e+02, threshold=5.515e+02, percent-clipped=1.0 2023-04-28 21:09:19,327 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76323.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 21:09:30,938 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76330.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 21:10:01,830 INFO [train.py:904] (6/8) Epoch 8, batch 5300, loss[loss=0.1532, simple_loss=0.2359, pruned_loss=0.03524, over 16410.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2839, pruned_loss=0.05745, over 3208063.61 frames. ], batch size: 35, lr: 8.55e-03, grad_scale: 8.0 2023-04-28 21:10:06,497 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76355.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:10:48,054 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76384.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 21:11:13,477 INFO [train.py:904] (6/8) Epoch 8, batch 5350, loss[loss=0.2069, simple_loss=0.2941, pruned_loss=0.05987, over 16587.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2827, pruned_loss=0.0567, over 3209825.65 frames. ], batch size: 68, lr: 8.54e-03, grad_scale: 8.0 2023-04-28 21:11:15,924 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.303e+02 2.727e+02 3.252e+02 5.747e+02, threshold=5.455e+02, percent-clipped=1.0 2023-04-28 21:11:50,792 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-28 21:12:26,519 INFO [train.py:904] (6/8) Epoch 8, batch 5400, loss[loss=0.2465, simple_loss=0.3167, pruned_loss=0.08817, over 12467.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2849, pruned_loss=0.05744, over 3202483.09 frames. ], batch size: 246, lr: 8.54e-03, grad_scale: 8.0 2023-04-28 21:13:07,233 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9017, 4.1645, 3.9124, 3.9728, 3.7234, 3.8031, 3.8220, 4.1278], device='cuda:6'), covar=tensor([0.0869, 0.0824, 0.0912, 0.0630, 0.0703, 0.1285, 0.0731, 0.0950], device='cuda:6'), in_proj_covar=tensor([0.0469, 0.0592, 0.0499, 0.0406, 0.0378, 0.0387, 0.0493, 0.0434], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:13:43,648 INFO [train.py:904] (6/8) Epoch 8, batch 5450, loss[loss=0.2185, simple_loss=0.2979, pruned_loss=0.0695, over 16303.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2883, pruned_loss=0.05933, over 3201917.95 frames. ], batch size: 35, lr: 8.54e-03, grad_scale: 8.0 2023-04-28 21:13:46,706 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.643e+02 3.277e+02 3.879e+02 8.643e+02, threshold=6.553e+02, percent-clipped=9.0 2023-04-28 21:13:48,357 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-28 21:14:50,575 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3081, 1.8190, 2.2083, 3.7202, 1.6883, 2.3657, 2.0073, 1.9582], device='cuda:6'), covar=tensor([0.0922, 0.3471, 0.1890, 0.0509, 0.4344, 0.2192, 0.2922, 0.3791], device='cuda:6'), in_proj_covar=tensor([0.0343, 0.0362, 0.0304, 0.0317, 0.0393, 0.0405, 0.0326, 0.0430], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:15:01,742 INFO [train.py:904] (6/8) Epoch 8, batch 5500, loss[loss=0.2947, simple_loss=0.3548, pruned_loss=0.1174, over 11635.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2969, pruned_loss=0.06563, over 3153796.35 frames. ], batch size: 248, lr: 8.54e-03, grad_scale: 8.0 2023-04-28 21:15:29,521 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-28 21:16:22,242 INFO [train.py:904] (6/8) Epoch 8, batch 5550, loss[loss=0.3348, simple_loss=0.381, pruned_loss=0.1443, over 10751.00 frames. ], tot_loss[loss=0.2236, simple_loss=0.3051, pruned_loss=0.0711, over 3159838.93 frames. ], batch size: 249, lr: 8.53e-03, grad_scale: 8.0 2023-04-28 21:16:26,052 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.250e+02 3.650e+02 4.259e+02 5.295e+02 1.196e+03, threshold=8.517e+02, percent-clipped=11.0 2023-04-28 21:16:59,356 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.23 vs. limit=5.0 2023-04-28 21:17:00,762 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76625.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 21:17:28,519 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3400, 1.9531, 1.6384, 1.6731, 2.2177, 1.8962, 2.2483, 2.3655], device='cuda:6'), covar=tensor([0.0070, 0.0197, 0.0274, 0.0273, 0.0129, 0.0220, 0.0115, 0.0139], device='cuda:6'), in_proj_covar=tensor([0.0110, 0.0180, 0.0179, 0.0179, 0.0177, 0.0181, 0.0177, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:17:40,983 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76650.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:17:43,813 INFO [train.py:904] (6/8) Epoch 8, batch 5600, loss[loss=0.3566, simple_loss=0.3862, pruned_loss=0.1635, over 11052.00 frames. ], tot_loss[loss=0.232, simple_loss=0.311, pruned_loss=0.07647, over 3118523.16 frames. ], batch size: 246, lr: 8.53e-03, grad_scale: 8.0 2023-04-28 21:18:28,942 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76679.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 21:18:59,114 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1848, 4.0284, 4.2149, 4.4366, 4.4739, 4.0979, 4.4356, 4.4986], device='cuda:6'), covar=tensor([0.1207, 0.0871, 0.1165, 0.0455, 0.0495, 0.1009, 0.0566, 0.0436], device='cuda:6'), in_proj_covar=tensor([0.0463, 0.0564, 0.0698, 0.0571, 0.0432, 0.0434, 0.0447, 0.0496], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:19:06,822 INFO [train.py:904] (6/8) Epoch 8, batch 5650, loss[loss=0.2391, simple_loss=0.3195, pruned_loss=0.07937, over 16516.00 frames. ], tot_loss[loss=0.2392, simple_loss=0.3163, pruned_loss=0.08108, over 3079080.94 frames. ], batch size: 75, lr: 8.53e-03, grad_scale: 8.0 2023-04-28 21:19:10,210 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.391e+02 3.706e+02 4.487e+02 5.532e+02 1.181e+03, threshold=8.975e+02, percent-clipped=2.0 2023-04-28 21:19:28,354 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1723, 1.8277, 2.8332, 3.2097, 2.9433, 3.6048, 2.0007, 3.6548], device='cuda:6'), covar=tensor([0.0108, 0.0299, 0.0166, 0.0139, 0.0147, 0.0073, 0.0295, 0.0053], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0158, 0.0143, 0.0142, 0.0151, 0.0106, 0.0159, 0.0098], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 21:20:27,955 INFO [train.py:904] (6/8) Epoch 8, batch 5700, loss[loss=0.2555, simple_loss=0.315, pruned_loss=0.09798, over 11550.00 frames. ], tot_loss[loss=0.2435, simple_loss=0.3191, pruned_loss=0.08392, over 3062911.06 frames. ], batch size: 248, lr: 8.53e-03, grad_scale: 8.0 2023-04-28 21:21:12,221 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3177, 5.6362, 5.3120, 5.3959, 4.9951, 4.8883, 5.1141, 5.7612], device='cuda:6'), covar=tensor([0.0738, 0.0702, 0.0991, 0.0665, 0.0700, 0.0671, 0.0771, 0.0747], device='cuda:6'), in_proj_covar=tensor([0.0467, 0.0586, 0.0499, 0.0401, 0.0374, 0.0385, 0.0489, 0.0432], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:21:49,376 INFO [train.py:904] (6/8) Epoch 8, batch 5750, loss[loss=0.2285, simple_loss=0.3068, pruned_loss=0.0751, over 16710.00 frames. ], tot_loss[loss=0.2468, simple_loss=0.322, pruned_loss=0.08575, over 3038220.50 frames. ], batch size: 134, lr: 8.52e-03, grad_scale: 4.0 2023-04-28 21:21:54,072 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.644e+02 3.742e+02 4.943e+02 6.173e+02 1.099e+03, threshold=9.887e+02, percent-clipped=1.0 2023-04-28 21:22:52,021 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76839.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:22:52,113 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4379, 2.0756, 1.5880, 1.8282, 2.3790, 2.0524, 2.5000, 2.5456], device='cuda:6'), covar=tensor([0.0090, 0.0224, 0.0323, 0.0314, 0.0137, 0.0266, 0.0137, 0.0163], device='cuda:6'), in_proj_covar=tensor([0.0110, 0.0180, 0.0179, 0.0179, 0.0178, 0.0182, 0.0177, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:23:12,543 INFO [train.py:904] (6/8) Epoch 8, batch 5800, loss[loss=0.2463, simple_loss=0.3115, pruned_loss=0.09057, over 11912.00 frames. ], tot_loss[loss=0.2451, simple_loss=0.3215, pruned_loss=0.08434, over 3036786.92 frames. ], batch size: 248, lr: 8.52e-03, grad_scale: 4.0 2023-04-28 21:24:03,459 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9677, 4.9729, 5.4869, 5.4016, 5.4424, 5.0627, 5.0236, 4.5764], device='cuda:6'), covar=tensor([0.0227, 0.0324, 0.0289, 0.0384, 0.0426, 0.0253, 0.0774, 0.0425], device='cuda:6'), in_proj_covar=tensor([0.0285, 0.0290, 0.0290, 0.0279, 0.0330, 0.0310, 0.0412, 0.0254], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 21:24:30,495 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76900.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:24:32,999 INFO [train.py:904] (6/8) Epoch 8, batch 5850, loss[loss=0.2268, simple_loss=0.3096, pruned_loss=0.07198, over 16592.00 frames. ], tot_loss[loss=0.2413, simple_loss=0.3188, pruned_loss=0.0819, over 3064837.18 frames. ], batch size: 62, lr: 8.52e-03, grad_scale: 4.0 2023-04-28 21:24:37,987 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 3.434e+02 4.196e+02 5.124e+02 1.086e+03, threshold=8.393e+02, percent-clipped=1.0 2023-04-28 21:25:09,410 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76925.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 21:25:50,143 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6901, 2.7010, 2.2982, 4.3397, 3.2439, 4.1288, 1.3220, 3.0619], device='cuda:6'), covar=tensor([0.1280, 0.0661, 0.1211, 0.0108, 0.0273, 0.0340, 0.1545, 0.0724], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0151, 0.0174, 0.0117, 0.0200, 0.0204, 0.0172, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 21:25:53,539 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76950.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:25:56,469 INFO [train.py:904] (6/8) Epoch 8, batch 5900, loss[loss=0.227, simple_loss=0.306, pruned_loss=0.07397, over 15540.00 frames. ], tot_loss[loss=0.2398, simple_loss=0.3177, pruned_loss=0.08097, over 3068272.03 frames. ], batch size: 191, lr: 8.51e-03, grad_scale: 4.0 2023-04-28 21:26:34,319 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=76973.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 21:26:42,792 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76979.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 21:27:11,066 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=76998.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:27:17,803 INFO [train.py:904] (6/8) Epoch 8, batch 5950, loss[loss=0.2243, simple_loss=0.3153, pruned_loss=0.06666, over 16637.00 frames. ], tot_loss[loss=0.2397, simple_loss=0.3191, pruned_loss=0.08016, over 3073518.85 frames. ], batch size: 68, lr: 8.51e-03, grad_scale: 4.0 2023-04-28 21:27:21,560 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.299e+02 3.417e+02 4.127e+02 4.793e+02 1.224e+03, threshold=8.253e+02, percent-clipped=3.0 2023-04-28 21:27:55,398 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=77027.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 21:28:33,565 INFO [train.py:904] (6/8) Epoch 8, batch 6000, loss[loss=0.2192, simple_loss=0.3014, pruned_loss=0.06848, over 16920.00 frames. ], tot_loss[loss=0.2375, simple_loss=0.3173, pruned_loss=0.07889, over 3091318.75 frames. ], batch size: 109, lr: 8.51e-03, grad_scale: 8.0 2023-04-28 21:28:33,565 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 21:28:44,113 INFO [train.py:938] (6/8) Epoch 8, validation: loss=0.1707, simple_loss=0.284, pruned_loss=0.02871, over 944034.00 frames. 2023-04-28 21:28:44,114 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-28 21:29:48,594 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77095.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:30:00,317 INFO [train.py:904] (6/8) Epoch 8, batch 6050, loss[loss=0.2197, simple_loss=0.3108, pruned_loss=0.06431, over 16828.00 frames. ], tot_loss[loss=0.2363, simple_loss=0.3157, pruned_loss=0.07848, over 3094404.53 frames. ], batch size: 116, lr: 8.51e-03, grad_scale: 8.0 2023-04-28 21:30:04,223 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.264e+02 3.621e+02 4.460e+02 5.601e+02 1.783e+03, threshold=8.919e+02, percent-clipped=7.0 2023-04-28 21:30:48,222 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8553, 4.0023, 4.2893, 2.0021, 4.6311, 4.5557, 3.1565, 3.3429], device='cuda:6'), covar=tensor([0.0701, 0.0141, 0.0141, 0.1126, 0.0040, 0.0060, 0.0332, 0.0388], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0098, 0.0083, 0.0142, 0.0070, 0.0092, 0.0119, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 21:31:11,099 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-28 21:31:19,254 INFO [train.py:904] (6/8) Epoch 8, batch 6100, loss[loss=0.2093, simple_loss=0.2962, pruned_loss=0.06117, over 16840.00 frames. ], tot_loss[loss=0.234, simple_loss=0.3145, pruned_loss=0.07675, over 3099856.60 frames. ], batch size: 42, lr: 8.50e-03, grad_scale: 8.0 2023-04-28 21:31:26,124 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77156.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 21:31:39,223 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8049, 1.5985, 2.2770, 2.7645, 2.5996, 3.0394, 1.5809, 2.8816], device='cuda:6'), covar=tensor([0.0108, 0.0319, 0.0194, 0.0150, 0.0164, 0.0085, 0.0368, 0.0083], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0159, 0.0144, 0.0143, 0.0152, 0.0107, 0.0161, 0.0099], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 21:31:48,562 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0046, 5.0091, 4.7799, 4.1991, 4.8651, 1.9671, 4.6374, 4.7409], device='cuda:6'), covar=tensor([0.0054, 0.0044, 0.0105, 0.0286, 0.0059, 0.1827, 0.0080, 0.0105], device='cuda:6'), in_proj_covar=tensor([0.0110, 0.0097, 0.0145, 0.0142, 0.0114, 0.0160, 0.0130, 0.0136], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:32:26,662 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77195.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:32:37,672 INFO [train.py:904] (6/8) Epoch 8, batch 6150, loss[loss=0.2348, simple_loss=0.3081, pruned_loss=0.08075, over 15435.00 frames. ], tot_loss[loss=0.2317, simple_loss=0.3124, pruned_loss=0.07548, over 3126601.30 frames. ], batch size: 191, lr: 8.50e-03, grad_scale: 8.0 2023-04-28 21:32:42,674 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.089e+02 3.253e+02 3.802e+02 4.768e+02 1.072e+03, threshold=7.605e+02, percent-clipped=4.0 2023-04-28 21:32:52,989 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77211.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:32:56,353 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0534, 5.0488, 4.7997, 4.2007, 4.8356, 1.8989, 4.6445, 4.7127], device='cuda:6'), covar=tensor([0.0051, 0.0043, 0.0111, 0.0310, 0.0060, 0.1947, 0.0089, 0.0128], device='cuda:6'), in_proj_covar=tensor([0.0111, 0.0098, 0.0146, 0.0142, 0.0115, 0.0162, 0.0131, 0.0136], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:32:57,688 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6900, 5.2160, 5.4347, 5.1763, 5.2834, 5.8040, 5.2727, 5.0217], device='cuda:6'), covar=tensor([0.0972, 0.1485, 0.1243, 0.1508, 0.1999, 0.0754, 0.1245, 0.2104], device='cuda:6'), in_proj_covar=tensor([0.0319, 0.0435, 0.0457, 0.0385, 0.0514, 0.0481, 0.0371, 0.0522], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 21:33:00,141 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-28 21:33:10,358 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5496, 3.4854, 2.8144, 2.1431, 2.4787, 2.1178, 3.6221, 3.3986], device='cuda:6'), covar=tensor([0.2414, 0.0639, 0.1358, 0.1968, 0.2014, 0.1715, 0.0461, 0.0846], device='cuda:6'), in_proj_covar=tensor([0.0294, 0.0252, 0.0275, 0.0263, 0.0280, 0.0211, 0.0258, 0.0279], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:33:59,283 INFO [train.py:904] (6/8) Epoch 8, batch 6200, loss[loss=0.2141, simple_loss=0.2946, pruned_loss=0.06682, over 17242.00 frames. ], tot_loss[loss=0.2303, simple_loss=0.3103, pruned_loss=0.07513, over 3116282.21 frames. ], batch size: 45, lr: 8.50e-03, grad_scale: 4.0 2023-04-28 21:34:31,356 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77272.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:34:33,112 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77273.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:34:39,235 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77277.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:34:52,023 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 21:35:16,096 INFO [train.py:904] (6/8) Epoch 8, batch 6250, loss[loss=0.2149, simple_loss=0.2964, pruned_loss=0.06673, over 17133.00 frames. ], tot_loss[loss=0.2308, simple_loss=0.3106, pruned_loss=0.07555, over 3097497.66 frames. ], batch size: 48, lr: 8.50e-03, grad_scale: 4.0 2023-04-28 21:35:22,795 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.200e+02 3.064e+02 3.768e+02 4.806e+02 9.942e+02, threshold=7.536e+02, percent-clipped=2.0 2023-04-28 21:35:52,005 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0766, 2.8603, 2.7856, 2.0027, 2.5862, 2.1677, 2.8126, 2.9327], device='cuda:6'), covar=tensor([0.0260, 0.0534, 0.0488, 0.1487, 0.0673, 0.0879, 0.0472, 0.0587], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0135, 0.0156, 0.0141, 0.0136, 0.0125, 0.0136, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 21:36:07,329 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77334.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:36:11,761 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3169, 1.9284, 2.0892, 3.8927, 1.9209, 2.4471, 2.0923, 2.0735], device='cuda:6'), covar=tensor([0.0852, 0.2986, 0.1893, 0.0355, 0.3489, 0.1937, 0.2744, 0.2899], device='cuda:6'), in_proj_covar=tensor([0.0343, 0.0364, 0.0303, 0.0318, 0.0397, 0.0405, 0.0325, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:36:13,007 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77338.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:36:35,474 INFO [train.py:904] (6/8) Epoch 8, batch 6300, loss[loss=0.2172, simple_loss=0.2977, pruned_loss=0.06839, over 16923.00 frames. ], tot_loss[loss=0.2296, simple_loss=0.3103, pruned_loss=0.07444, over 3120339.87 frames. ], batch size: 116, lr: 8.49e-03, grad_scale: 4.0 2023-04-28 21:37:54,077 INFO [train.py:904] (6/8) Epoch 8, batch 6350, loss[loss=0.2151, simple_loss=0.2956, pruned_loss=0.0673, over 17044.00 frames. ], tot_loss[loss=0.2326, simple_loss=0.3123, pruned_loss=0.07644, over 3112055.59 frames. ], batch size: 53, lr: 8.49e-03, grad_scale: 4.0 2023-04-28 21:38:00,461 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.093e+02 3.385e+02 4.021e+02 5.215e+02 8.857e+02, threshold=8.042e+02, percent-clipped=2.0 2023-04-28 21:38:01,107 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2185, 5.1796, 4.9362, 4.3734, 5.0229, 1.7036, 4.8194, 4.8866], device='cuda:6'), covar=tensor([0.0050, 0.0040, 0.0095, 0.0272, 0.0052, 0.2016, 0.0075, 0.0121], device='cuda:6'), in_proj_covar=tensor([0.0109, 0.0097, 0.0144, 0.0140, 0.0114, 0.0160, 0.0130, 0.0134], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:39:10,452 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77451.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 21:39:11,215 INFO [train.py:904] (6/8) Epoch 8, batch 6400, loss[loss=0.2658, simple_loss=0.3319, pruned_loss=0.09991, over 11369.00 frames. ], tot_loss[loss=0.2349, simple_loss=0.3132, pruned_loss=0.0783, over 3080682.45 frames. ], batch size: 248, lr: 8.49e-03, grad_scale: 8.0 2023-04-28 21:40:12,254 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7646, 3.6727, 3.8536, 3.7133, 3.8011, 4.1975, 3.9120, 3.6874], device='cuda:6'), covar=tensor([0.2210, 0.1984, 0.1928, 0.2492, 0.2947, 0.1562, 0.1492, 0.2732], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0436, 0.0464, 0.0390, 0.0515, 0.0488, 0.0378, 0.0527], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 21:40:17,184 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77495.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:40:26,236 INFO [train.py:904] (6/8) Epoch 8, batch 6450, loss[loss=0.2206, simple_loss=0.3104, pruned_loss=0.06543, over 16481.00 frames. ], tot_loss[loss=0.233, simple_loss=0.3124, pruned_loss=0.07682, over 3093323.16 frames. ], batch size: 68, lr: 8.48e-03, grad_scale: 8.0 2023-04-28 21:40:33,076 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.067e+02 3.422e+02 4.688e+02 6.041e+02 9.477e+02, threshold=9.377e+02, percent-clipped=7.0 2023-04-28 21:41:31,678 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=77543.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:41:45,631 INFO [train.py:904] (6/8) Epoch 8, batch 6500, loss[loss=0.2431, simple_loss=0.324, pruned_loss=0.08105, over 16791.00 frames. ], tot_loss[loss=0.2307, simple_loss=0.3099, pruned_loss=0.07577, over 3103720.47 frames. ], batch size: 83, lr: 8.48e-03, grad_scale: 8.0 2023-04-28 21:42:09,115 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77567.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:42:32,409 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-28 21:42:47,389 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3606, 5.1006, 5.3391, 5.5054, 5.6770, 4.9843, 5.6198, 5.6307], device='cuda:6'), covar=tensor([0.1215, 0.0951, 0.1148, 0.0567, 0.0462, 0.0564, 0.0478, 0.0476], device='cuda:6'), in_proj_covar=tensor([0.0462, 0.0568, 0.0695, 0.0580, 0.0438, 0.0433, 0.0453, 0.0502], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:43:05,177 INFO [train.py:904] (6/8) Epoch 8, batch 6550, loss[loss=0.2378, simple_loss=0.3302, pruned_loss=0.07269, over 16432.00 frames. ], tot_loss[loss=0.2338, simple_loss=0.3134, pruned_loss=0.07715, over 3101449.26 frames. ], batch size: 146, lr: 8.48e-03, grad_scale: 8.0 2023-04-28 21:43:11,238 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.922e+02 3.133e+02 3.759e+02 4.463e+02 1.371e+03, threshold=7.519e+02, percent-clipped=1.0 2023-04-28 21:43:28,227 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77616.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:43:47,979 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77629.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:43:53,996 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77633.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:44:22,368 INFO [train.py:904] (6/8) Epoch 8, batch 6600, loss[loss=0.2482, simple_loss=0.3299, pruned_loss=0.08325, over 16431.00 frames. ], tot_loss[loss=0.2357, simple_loss=0.3156, pruned_loss=0.07789, over 3091339.71 frames. ], batch size: 146, lr: 8.48e-03, grad_scale: 8.0 2023-04-28 21:44:34,979 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5461, 4.5474, 5.0069, 4.9587, 4.9540, 4.5305, 4.5646, 4.3542], device='cuda:6'), covar=tensor([0.0245, 0.0400, 0.0326, 0.0342, 0.0418, 0.0367, 0.0920, 0.0428], device='cuda:6'), in_proj_covar=tensor([0.0290, 0.0296, 0.0299, 0.0286, 0.0333, 0.0317, 0.0421, 0.0256], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 21:45:00,788 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77677.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:45:00,834 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77677.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:45:22,404 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6941, 5.0193, 4.7362, 4.7436, 4.4111, 4.3928, 4.5167, 5.0746], device='cuda:6'), covar=tensor([0.0798, 0.0678, 0.0906, 0.0582, 0.0733, 0.0841, 0.0800, 0.0677], device='cuda:6'), in_proj_covar=tensor([0.0468, 0.0585, 0.0496, 0.0399, 0.0370, 0.0383, 0.0486, 0.0428], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:45:38,843 INFO [train.py:904] (6/8) Epoch 8, batch 6650, loss[loss=0.2329, simple_loss=0.3063, pruned_loss=0.07977, over 16910.00 frames. ], tot_loss[loss=0.2368, simple_loss=0.316, pruned_loss=0.07881, over 3091453.39 frames. ], batch size: 109, lr: 8.47e-03, grad_scale: 8.0 2023-04-28 21:45:45,535 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.195e+02 3.528e+02 4.149e+02 5.029e+02 1.289e+03, threshold=8.299e+02, percent-clipped=5.0 2023-04-28 21:46:35,358 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77738.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:46:53,679 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77751.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 21:46:54,547 INFO [train.py:904] (6/8) Epoch 8, batch 6700, loss[loss=0.2375, simple_loss=0.3189, pruned_loss=0.07805, over 16287.00 frames. ], tot_loss[loss=0.2364, simple_loss=0.315, pruned_loss=0.07884, over 3085909.53 frames. ], batch size: 165, lr: 8.47e-03, grad_scale: 4.0 2023-04-28 21:47:59,590 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 21:48:06,760 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=77799.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:48:10,946 INFO [train.py:904] (6/8) Epoch 8, batch 6750, loss[loss=0.2484, simple_loss=0.3228, pruned_loss=0.08704, over 16841.00 frames. ], tot_loss[loss=0.2346, simple_loss=0.3131, pruned_loss=0.07803, over 3095418.95 frames. ], batch size: 116, lr: 8.47e-03, grad_scale: 4.0 2023-04-28 21:48:17,759 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4399, 2.0900, 2.2110, 4.1739, 1.9162, 2.5733, 2.1606, 2.2815], device='cuda:6'), covar=tensor([0.0835, 0.2925, 0.1941, 0.0344, 0.3598, 0.1891, 0.2610, 0.2752], device='cuda:6'), in_proj_covar=tensor([0.0344, 0.0367, 0.0307, 0.0320, 0.0401, 0.0406, 0.0328, 0.0432], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:48:18,410 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.361e+02 3.344e+02 4.001e+02 5.088e+02 1.053e+03, threshold=8.003e+02, percent-clipped=2.0 2023-04-28 21:49:25,315 INFO [train.py:904] (6/8) Epoch 8, batch 6800, loss[loss=0.224, simple_loss=0.3068, pruned_loss=0.07065, over 16577.00 frames. ], tot_loss[loss=0.2341, simple_loss=0.313, pruned_loss=0.07758, over 3118124.32 frames. ], batch size: 68, lr: 8.47e-03, grad_scale: 8.0 2023-04-28 21:49:49,109 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77867.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:50:17,008 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4362, 5.7830, 5.4516, 5.5013, 4.9865, 4.9676, 5.3221, 5.9116], device='cuda:6'), covar=tensor([0.0957, 0.0723, 0.1045, 0.0664, 0.0833, 0.0625, 0.0847, 0.0729], device='cuda:6'), in_proj_covar=tensor([0.0470, 0.0584, 0.0495, 0.0398, 0.0369, 0.0384, 0.0483, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:50:40,224 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3757, 4.1511, 4.3911, 4.5950, 4.7333, 4.2331, 4.6247, 4.6705], device='cuda:6'), covar=tensor([0.1305, 0.0988, 0.1330, 0.0584, 0.0463, 0.0935, 0.0624, 0.0480], device='cuda:6'), in_proj_covar=tensor([0.0463, 0.0572, 0.0705, 0.0582, 0.0441, 0.0439, 0.0462, 0.0505], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:50:44,170 INFO [train.py:904] (6/8) Epoch 8, batch 6850, loss[loss=0.2164, simple_loss=0.3203, pruned_loss=0.05626, over 16659.00 frames. ], tot_loss[loss=0.2347, simple_loss=0.3141, pruned_loss=0.07766, over 3123241.19 frames. ], batch size: 62, lr: 8.46e-03, grad_scale: 4.0 2023-04-28 21:50:48,323 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1790, 4.4312, 4.6710, 1.9795, 4.9715, 4.9730, 3.5696, 3.7181], device='cuda:6'), covar=tensor([0.0577, 0.0110, 0.0091, 0.1171, 0.0030, 0.0062, 0.0235, 0.0315], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0097, 0.0084, 0.0138, 0.0068, 0.0090, 0.0118, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 21:50:53,211 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.176e+02 3.270e+02 3.895e+02 4.577e+02 9.421e+02, threshold=7.790e+02, percent-clipped=1.0 2023-04-28 21:51:03,725 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=77915.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:51:24,353 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77929.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:51:30,507 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77933.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:51:59,510 INFO [train.py:904] (6/8) Epoch 8, batch 6900, loss[loss=0.2482, simple_loss=0.3263, pruned_loss=0.08505, over 16755.00 frames. ], tot_loss[loss=0.2358, simple_loss=0.3163, pruned_loss=0.07772, over 3114222.60 frames. ], batch size: 124, lr: 8.46e-03, grad_scale: 4.0 2023-04-28 21:52:31,656 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77972.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:52:34,754 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3921, 4.4087, 4.8219, 4.7884, 4.8012, 4.4506, 4.4289, 4.2430], device='cuda:6'), covar=tensor([0.0281, 0.0434, 0.0408, 0.0440, 0.0411, 0.0337, 0.0844, 0.0442], device='cuda:6'), in_proj_covar=tensor([0.0293, 0.0301, 0.0302, 0.0290, 0.0340, 0.0320, 0.0421, 0.0261], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 21:52:39,052 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=77977.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:52:45,337 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=77981.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:53:20,813 INFO [train.py:904] (6/8) Epoch 8, batch 6950, loss[loss=0.2292, simple_loss=0.3011, pruned_loss=0.07867, over 16675.00 frames. ], tot_loss[loss=0.2394, simple_loss=0.3187, pruned_loss=0.08007, over 3092452.51 frames. ], batch size: 57, lr: 8.46e-03, grad_scale: 4.0 2023-04-28 21:53:29,769 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.004e+02 3.295e+02 4.352e+02 5.795e+02 9.816e+02, threshold=8.703e+02, percent-clipped=9.0 2023-04-28 21:54:01,541 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.84 vs. limit=5.0 2023-04-28 21:54:10,659 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78033.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:54:38,871 INFO [train.py:904] (6/8) Epoch 8, batch 7000, loss[loss=0.2017, simple_loss=0.3039, pruned_loss=0.04982, over 16846.00 frames. ], tot_loss[loss=0.2385, simple_loss=0.3187, pruned_loss=0.07915, over 3097549.92 frames. ], batch size: 102, lr: 8.45e-03, grad_scale: 4.0 2023-04-28 21:55:00,486 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4668, 3.5151, 3.2654, 3.0303, 3.0646, 3.3837, 3.2776, 3.2144], device='cuda:6'), covar=tensor([0.0525, 0.0378, 0.0221, 0.0174, 0.0546, 0.0278, 0.1091, 0.0425], device='cuda:6'), in_proj_covar=tensor([0.0214, 0.0254, 0.0249, 0.0220, 0.0275, 0.0253, 0.0172, 0.0287], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:55:53,774 INFO [train.py:904] (6/8) Epoch 8, batch 7050, loss[loss=0.288, simple_loss=0.3399, pruned_loss=0.1181, over 11539.00 frames. ], tot_loss[loss=0.2391, simple_loss=0.3195, pruned_loss=0.07931, over 3083510.41 frames. ], batch size: 247, lr: 8.45e-03, grad_scale: 4.0 2023-04-28 21:55:58,647 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6392, 2.1356, 2.2460, 4.3621, 2.0495, 2.5954, 2.2653, 2.3385], device='cuda:6'), covar=tensor([0.0737, 0.2891, 0.1858, 0.0318, 0.3467, 0.1805, 0.2577, 0.2891], device='cuda:6'), in_proj_covar=tensor([0.0343, 0.0367, 0.0307, 0.0319, 0.0402, 0.0404, 0.0327, 0.0431], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 21:56:03,900 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.852e+02 3.078e+02 3.817e+02 4.566e+02 8.440e+02, threshold=7.634e+02, percent-clipped=0.0 2023-04-28 21:56:19,330 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2092, 3.8954, 3.8647, 2.4132, 3.3680, 3.8218, 3.5721, 1.8749], device='cuda:6'), covar=tensor([0.0400, 0.0024, 0.0032, 0.0324, 0.0068, 0.0068, 0.0048, 0.0372], device='cuda:6'), in_proj_covar=tensor([0.0123, 0.0062, 0.0064, 0.0122, 0.0069, 0.0080, 0.0071, 0.0114], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 21:57:11,219 INFO [train.py:904] (6/8) Epoch 8, batch 7100, loss[loss=0.2782, simple_loss=0.3347, pruned_loss=0.1108, over 11597.00 frames. ], tot_loss[loss=0.236, simple_loss=0.3166, pruned_loss=0.07776, over 3103327.64 frames. ], batch size: 246, lr: 8.45e-03, grad_scale: 2.0 2023-04-28 21:57:52,069 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0769, 1.3478, 1.7726, 2.0119, 2.1711, 2.2205, 1.5317, 2.1455], device='cuda:6'), covar=tensor([0.0149, 0.0332, 0.0201, 0.0190, 0.0164, 0.0114, 0.0294, 0.0088], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0161, 0.0144, 0.0143, 0.0151, 0.0108, 0.0158, 0.0099], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 21:58:04,978 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9767, 2.6353, 2.6148, 1.8207, 2.7698, 2.8157, 2.4123, 2.3273], device='cuda:6'), covar=tensor([0.0732, 0.0190, 0.0203, 0.0990, 0.0098, 0.0169, 0.0406, 0.0426], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0097, 0.0085, 0.0140, 0.0068, 0.0090, 0.0119, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 21:58:26,606 INFO [train.py:904] (6/8) Epoch 8, batch 7150, loss[loss=0.2735, simple_loss=0.3262, pruned_loss=0.1104, over 11230.00 frames. ], tot_loss[loss=0.236, simple_loss=0.3154, pruned_loss=0.07833, over 3097917.89 frames. ], batch size: 248, lr: 8.45e-03, grad_scale: 2.0 2023-04-28 21:58:36,168 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.673e+02 3.732e+02 4.325e+02 5.377e+02 1.184e+03, threshold=8.651e+02, percent-clipped=7.0 2023-04-28 21:59:09,648 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78232.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:59:39,845 INFO [train.py:904] (6/8) Epoch 8, batch 7200, loss[loss=0.1971, simple_loss=0.2867, pruned_loss=0.05374, over 16900.00 frames. ], tot_loss[loss=0.2326, simple_loss=0.3127, pruned_loss=0.07629, over 3089830.89 frames. ], batch size: 109, lr: 8.44e-03, grad_scale: 4.0 2023-04-28 22:00:09,925 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78272.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:00:44,929 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78293.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 22:00:59,550 INFO [train.py:904] (6/8) Epoch 8, batch 7250, loss[loss=0.2013, simple_loss=0.2856, pruned_loss=0.05851, over 16468.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.3103, pruned_loss=0.07501, over 3084005.02 frames. ], batch size: 68, lr: 8.44e-03, grad_scale: 4.0 2023-04-28 22:01:07,688 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5560, 3.1487, 3.0476, 1.7829, 2.7301, 2.2577, 3.1649, 3.1847], device='cuda:6'), covar=tensor([0.0287, 0.0566, 0.0532, 0.1740, 0.0739, 0.0885, 0.0601, 0.0743], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0133, 0.0156, 0.0141, 0.0133, 0.0124, 0.0135, 0.0146], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 22:01:10,046 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.993e+02 2.959e+02 3.722e+02 4.557e+02 8.668e+02, threshold=7.444e+02, percent-clipped=1.0 2023-04-28 22:01:27,084 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=78320.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:01:47,723 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78333.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:01:53,894 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5830, 2.6228, 2.1414, 3.9887, 2.6411, 3.8261, 1.4504, 2.5680], device='cuda:6'), covar=tensor([0.1467, 0.0736, 0.1394, 0.0148, 0.0326, 0.0443, 0.1643, 0.1076], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0152, 0.0171, 0.0118, 0.0200, 0.0202, 0.0173, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 22:02:16,538 INFO [train.py:904] (6/8) Epoch 8, batch 7300, loss[loss=0.2324, simple_loss=0.3174, pruned_loss=0.07371, over 16838.00 frames. ], tot_loss[loss=0.2285, simple_loss=0.3088, pruned_loss=0.07411, over 3088929.06 frames. ], batch size: 116, lr: 8.44e-03, grad_scale: 4.0 2023-04-28 22:02:20,650 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8200, 1.1776, 1.6986, 1.7148, 1.8442, 1.9348, 1.4461, 1.8138], device='cuda:6'), covar=tensor([0.0136, 0.0249, 0.0136, 0.0134, 0.0134, 0.0083, 0.0246, 0.0055], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0159, 0.0142, 0.0142, 0.0149, 0.0106, 0.0157, 0.0098], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 22:03:02,793 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=78381.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:03:34,469 INFO [train.py:904] (6/8) Epoch 8, batch 7350, loss[loss=0.2617, simple_loss=0.3222, pruned_loss=0.1007, over 10949.00 frames. ], tot_loss[loss=0.2291, simple_loss=0.3093, pruned_loss=0.07451, over 3075071.93 frames. ], batch size: 248, lr: 8.44e-03, grad_scale: 4.0 2023-04-28 22:03:45,277 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.445e+02 3.549e+02 4.373e+02 5.467e+02 2.414e+03, threshold=8.746e+02, percent-clipped=10.0 2023-04-28 22:03:57,493 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6204, 4.3182, 4.3558, 2.9354, 3.7420, 4.3436, 3.9943, 2.4579], device='cuda:6'), covar=tensor([0.0336, 0.0021, 0.0021, 0.0239, 0.0051, 0.0066, 0.0034, 0.0300], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0061, 0.0063, 0.0120, 0.0068, 0.0078, 0.0069, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 22:04:55,407 INFO [train.py:904] (6/8) Epoch 8, batch 7400, loss[loss=0.2355, simple_loss=0.3198, pruned_loss=0.07559, over 16421.00 frames. ], tot_loss[loss=0.2307, simple_loss=0.3105, pruned_loss=0.07549, over 3070901.88 frames. ], batch size: 146, lr: 8.43e-03, grad_scale: 4.0 2023-04-28 22:05:08,375 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78460.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:05:49,200 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2023-04-28 22:06:13,365 INFO [train.py:904] (6/8) Epoch 8, batch 7450, loss[loss=0.1951, simple_loss=0.2806, pruned_loss=0.0548, over 16634.00 frames. ], tot_loss[loss=0.2324, simple_loss=0.312, pruned_loss=0.07642, over 3091191.15 frames. ], batch size: 62, lr: 8.43e-03, grad_scale: 4.0 2023-04-28 22:06:26,555 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.920e+02 3.338e+02 4.187e+02 5.103e+02 1.080e+03, threshold=8.375e+02, percent-clipped=2.0 2023-04-28 22:06:47,455 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78521.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:07:34,683 INFO [train.py:904] (6/8) Epoch 8, batch 7500, loss[loss=0.2055, simple_loss=0.2951, pruned_loss=0.05793, over 16815.00 frames. ], tot_loss[loss=0.2336, simple_loss=0.3131, pruned_loss=0.07705, over 3072267.98 frames. ], batch size: 102, lr: 8.43e-03, grad_scale: 4.0 2023-04-28 22:07:36,202 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 22:07:42,093 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78556.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:08:33,359 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78588.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 22:08:40,380 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1611, 3.1856, 3.3154, 1.5891, 3.5012, 3.5398, 2.7042, 2.5834], device='cuda:6'), covar=tensor([0.0738, 0.0192, 0.0158, 0.1187, 0.0054, 0.0106, 0.0404, 0.0424], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0094, 0.0081, 0.0135, 0.0065, 0.0087, 0.0116, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 22:08:55,085 INFO [train.py:904] (6/8) Epoch 8, batch 7550, loss[loss=0.1868, simple_loss=0.2795, pruned_loss=0.04702, over 16802.00 frames. ], tot_loss[loss=0.2334, simple_loss=0.3124, pruned_loss=0.07722, over 3068476.30 frames. ], batch size: 102, lr: 8.42e-03, grad_scale: 4.0 2023-04-28 22:09:05,665 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.406e+02 3.301e+02 4.162e+02 5.431e+02 1.258e+03, threshold=8.325e+02, percent-clipped=7.0 2023-04-28 22:09:20,046 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78617.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:10:12,497 INFO [train.py:904] (6/8) Epoch 8, batch 7600, loss[loss=0.2045, simple_loss=0.2921, pruned_loss=0.05841, over 16307.00 frames. ], tot_loss[loss=0.2319, simple_loss=0.3104, pruned_loss=0.07664, over 3081910.56 frames. ], batch size: 35, lr: 8.42e-03, grad_scale: 8.0 2023-04-28 22:10:49,247 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4122, 3.3775, 3.3821, 2.6360, 3.3295, 2.0551, 3.0224, 2.7203], device='cuda:6'), covar=tensor([0.0163, 0.0119, 0.0171, 0.0385, 0.0106, 0.2277, 0.0144, 0.0285], device='cuda:6'), in_proj_covar=tensor([0.0108, 0.0095, 0.0142, 0.0138, 0.0113, 0.0160, 0.0126, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:11:30,196 INFO [train.py:904] (6/8) Epoch 8, batch 7650, loss[loss=0.2642, simple_loss=0.334, pruned_loss=0.09718, over 16198.00 frames. ], tot_loss[loss=0.233, simple_loss=0.3115, pruned_loss=0.07721, over 3087825.88 frames. ], batch size: 165, lr: 8.42e-03, grad_scale: 8.0 2023-04-28 22:11:40,445 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.288e+02 3.310e+02 4.230e+02 5.150e+02 8.626e+02, threshold=8.460e+02, percent-clipped=2.0 2023-04-28 22:12:07,601 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.73 vs. limit=5.0 2023-04-28 22:12:45,888 INFO [train.py:904] (6/8) Epoch 8, batch 7700, loss[loss=0.2126, simple_loss=0.2997, pruned_loss=0.06272, over 16843.00 frames. ], tot_loss[loss=0.2333, simple_loss=0.3115, pruned_loss=0.07753, over 3092146.38 frames. ], batch size: 102, lr: 8.42e-03, grad_scale: 4.0 2023-04-28 22:13:00,511 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5879, 3.8090, 3.1258, 2.3260, 2.7247, 2.2314, 4.1308, 3.6735], device='cuda:6'), covar=tensor([0.2531, 0.0675, 0.1352, 0.1869, 0.2117, 0.1658, 0.0387, 0.0789], device='cuda:6'), in_proj_covar=tensor([0.0294, 0.0253, 0.0275, 0.0262, 0.0277, 0.0211, 0.0259, 0.0276], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:13:08,938 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.42 vs. limit=5.0 2023-04-28 22:14:03,411 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3157, 3.8472, 3.6484, 2.2053, 3.3006, 2.6291, 3.5642, 3.9483], device='cuda:6'), covar=tensor([0.0232, 0.0538, 0.0576, 0.1663, 0.0650, 0.0850, 0.0614, 0.0746], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0134, 0.0157, 0.0142, 0.0133, 0.0125, 0.0137, 0.0146], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 22:14:03,979 INFO [train.py:904] (6/8) Epoch 8, batch 7750, loss[loss=0.2138, simple_loss=0.2966, pruned_loss=0.0655, over 16649.00 frames. ], tot_loss[loss=0.2321, simple_loss=0.3111, pruned_loss=0.07656, over 3105803.41 frames. ], batch size: 62, lr: 8.41e-03, grad_scale: 2.0 2023-04-28 22:14:17,807 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.129e+02 3.510e+02 4.111e+02 5.372e+02 9.340e+02, threshold=8.221e+02, percent-clipped=3.0 2023-04-28 22:14:26,064 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78816.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:15:19,808 INFO [train.py:904] (6/8) Epoch 8, batch 7800, loss[loss=0.2003, simple_loss=0.2836, pruned_loss=0.05849, over 16538.00 frames. ], tot_loss[loss=0.2336, simple_loss=0.312, pruned_loss=0.07755, over 3095080.46 frames. ], batch size: 75, lr: 8.41e-03, grad_scale: 2.0 2023-04-28 22:15:45,196 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8896, 5.4399, 5.6116, 5.3697, 5.4028, 6.0453, 5.5491, 5.2541], device='cuda:6'), covar=tensor([0.0734, 0.1582, 0.1544, 0.1696, 0.2236, 0.0897, 0.1195, 0.2279], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0437, 0.0470, 0.0395, 0.0518, 0.0487, 0.0379, 0.0532], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 22:16:16,734 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78888.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:16:18,301 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-28 22:16:37,569 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4960, 4.4459, 4.3192, 4.1217, 3.9546, 4.3741, 4.3113, 4.0773], device='cuda:6'), covar=tensor([0.0482, 0.0403, 0.0257, 0.0227, 0.0912, 0.0372, 0.0395, 0.0558], device='cuda:6'), in_proj_covar=tensor([0.0217, 0.0257, 0.0252, 0.0223, 0.0279, 0.0261, 0.0176, 0.0292], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:16:38,223 INFO [train.py:904] (6/8) Epoch 8, batch 7850, loss[loss=0.2284, simple_loss=0.3094, pruned_loss=0.07374, over 16882.00 frames. ], tot_loss[loss=0.2342, simple_loss=0.3133, pruned_loss=0.07752, over 3093819.41 frames. ], batch size: 116, lr: 8.41e-03, grad_scale: 2.0 2023-04-28 22:16:46,573 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8168, 3.6792, 3.8708, 3.9831, 4.0556, 3.6511, 3.9738, 4.0521], device='cuda:6'), covar=tensor([0.1102, 0.0833, 0.1007, 0.0512, 0.0488, 0.1435, 0.0670, 0.0508], device='cuda:6'), in_proj_covar=tensor([0.0465, 0.0569, 0.0698, 0.0579, 0.0443, 0.0428, 0.0462, 0.0505], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:16:50,958 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.229e+02 3.386e+02 3.845e+02 4.858e+02 8.286e+02, threshold=7.691e+02, percent-clipped=1.0 2023-04-28 22:16:53,959 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78912.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:17:29,667 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=78936.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:17:54,236 INFO [train.py:904] (6/8) Epoch 8, batch 7900, loss[loss=0.2453, simple_loss=0.3241, pruned_loss=0.08326, over 15333.00 frames. ], tot_loss[loss=0.2334, simple_loss=0.3128, pruned_loss=0.07696, over 3099105.98 frames. ], batch size: 190, lr: 8.41e-03, grad_scale: 2.0 2023-04-28 22:18:15,523 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78965.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:18:36,640 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9843, 3.9173, 3.9222, 3.2886, 3.9218, 1.6175, 3.7431, 3.5711], device='cuda:6'), covar=tensor([0.0094, 0.0084, 0.0112, 0.0284, 0.0078, 0.2288, 0.0106, 0.0171], device='cuda:6'), in_proj_covar=tensor([0.0109, 0.0096, 0.0142, 0.0140, 0.0113, 0.0161, 0.0128, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:19:13,438 INFO [train.py:904] (6/8) Epoch 8, batch 7950, loss[loss=0.2116, simple_loss=0.2934, pruned_loss=0.06491, over 16857.00 frames. ], tot_loss[loss=0.2343, simple_loss=0.3131, pruned_loss=0.07778, over 3089495.76 frames. ], batch size: 42, lr: 8.40e-03, grad_scale: 2.0 2023-04-28 22:19:28,063 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.154e+02 3.667e+02 4.128e+02 4.915e+02 9.776e+02, threshold=8.255e+02, percent-clipped=2.0 2023-04-28 22:19:52,464 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79026.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:20:32,279 INFO [train.py:904] (6/8) Epoch 8, batch 8000, loss[loss=0.2362, simple_loss=0.3163, pruned_loss=0.07803, over 16867.00 frames. ], tot_loss[loss=0.2346, simple_loss=0.3132, pruned_loss=0.07795, over 3091945.16 frames. ], batch size: 116, lr: 8.40e-03, grad_scale: 4.0 2023-04-28 22:21:48,606 INFO [train.py:904] (6/8) Epoch 8, batch 8050, loss[loss=0.2265, simple_loss=0.3129, pruned_loss=0.07004, over 16418.00 frames. ], tot_loss[loss=0.2343, simple_loss=0.3128, pruned_loss=0.07792, over 3089977.90 frames. ], batch size: 146, lr: 8.40e-03, grad_scale: 4.0 2023-04-28 22:22:02,033 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.318e+02 3.198e+02 3.767e+02 4.640e+02 1.099e+03, threshold=7.534e+02, percent-clipped=1.0 2023-04-28 22:22:08,320 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4628, 3.1195, 2.9681, 1.8586, 2.6537, 2.2067, 2.9672, 3.1439], device='cuda:6'), covar=tensor([0.0326, 0.0569, 0.0576, 0.1730, 0.0785, 0.0927, 0.0689, 0.0793], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0133, 0.0155, 0.0141, 0.0133, 0.0124, 0.0136, 0.0145], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 22:22:11,170 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79116.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:22:16,838 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6093, 3.7460, 2.6125, 2.1991, 2.7077, 2.2442, 3.7605, 3.5658], device='cuda:6'), covar=tensor([0.2735, 0.0740, 0.1746, 0.2046, 0.2293, 0.1698, 0.0605, 0.0883], device='cuda:6'), in_proj_covar=tensor([0.0298, 0.0254, 0.0278, 0.0266, 0.0281, 0.0213, 0.0261, 0.0277], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:22:28,466 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-28 22:23:05,315 INFO [train.py:904] (6/8) Epoch 8, batch 8100, loss[loss=0.223, simple_loss=0.3048, pruned_loss=0.07061, over 16547.00 frames. ], tot_loss[loss=0.2324, simple_loss=0.3117, pruned_loss=0.07653, over 3099528.20 frames. ], batch size: 62, lr: 8.40e-03, grad_scale: 4.0 2023-04-28 22:23:22,826 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9997, 3.6752, 3.6741, 2.3500, 3.4521, 3.6689, 3.4556, 2.0979], device='cuda:6'), covar=tensor([0.0404, 0.0025, 0.0038, 0.0281, 0.0051, 0.0077, 0.0047, 0.0296], device='cuda:6'), in_proj_covar=tensor([0.0125, 0.0062, 0.0064, 0.0121, 0.0069, 0.0080, 0.0070, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 22:23:24,437 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=79164.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:24:22,975 INFO [train.py:904] (6/8) Epoch 8, batch 8150, loss[loss=0.1908, simple_loss=0.2704, pruned_loss=0.05565, over 17000.00 frames. ], tot_loss[loss=0.229, simple_loss=0.3084, pruned_loss=0.07484, over 3121811.76 frames. ], batch size: 41, lr: 8.39e-03, grad_scale: 4.0 2023-04-28 22:24:36,887 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.139e+02 3.692e+02 4.407e+02 5.311e+02 8.589e+02, threshold=8.814e+02, percent-clipped=3.0 2023-04-28 22:24:38,552 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79212.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:25:28,063 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7436, 1.6347, 2.1709, 2.5711, 2.4859, 2.9044, 1.6876, 2.7360], device='cuda:6'), covar=tensor([0.0103, 0.0311, 0.0195, 0.0161, 0.0168, 0.0092, 0.0332, 0.0083], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0157, 0.0141, 0.0139, 0.0148, 0.0107, 0.0156, 0.0098], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-28 22:25:42,513 INFO [train.py:904] (6/8) Epoch 8, batch 8200, loss[loss=0.2297, simple_loss=0.3122, pruned_loss=0.0736, over 15342.00 frames. ], tot_loss[loss=0.2284, simple_loss=0.307, pruned_loss=0.07496, over 3109008.42 frames. ], batch size: 191, lr: 8.39e-03, grad_scale: 4.0 2023-04-28 22:25:55,508 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=79260.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:26:36,815 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79285.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 22:26:45,176 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7577, 4.7121, 4.6320, 3.2518, 4.5898, 1.5394, 4.2703, 4.4091], device='cuda:6'), covar=tensor([0.0118, 0.0092, 0.0171, 0.0682, 0.0103, 0.2728, 0.0158, 0.0313], device='cuda:6'), in_proj_covar=tensor([0.0110, 0.0095, 0.0143, 0.0141, 0.0115, 0.0162, 0.0129, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:27:04,307 INFO [train.py:904] (6/8) Epoch 8, batch 8250, loss[loss=0.2034, simple_loss=0.2971, pruned_loss=0.05489, over 15233.00 frames. ], tot_loss[loss=0.2252, simple_loss=0.3061, pruned_loss=0.07214, over 3111284.93 frames. ], batch size: 190, lr: 8.39e-03, grad_scale: 4.0 2023-04-28 22:27:19,441 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.098e+02 3.091e+02 3.915e+02 5.485e+02 9.423e+02, threshold=7.829e+02, percent-clipped=2.0 2023-04-28 22:27:37,102 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79321.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:27:56,968 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79333.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:28:17,695 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79346.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 22:28:26,464 INFO [train.py:904] (6/8) Epoch 8, batch 8300, loss[loss=0.2111, simple_loss=0.3008, pruned_loss=0.06068, over 16175.00 frames. ], tot_loss[loss=0.2208, simple_loss=0.3029, pruned_loss=0.06931, over 3088965.95 frames. ], batch size: 165, lr: 8.39e-03, grad_scale: 4.0 2023-04-28 22:28:29,026 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8689, 4.1830, 3.9720, 4.0370, 3.6850, 3.7809, 3.8343, 4.1555], device='cuda:6'), covar=tensor([0.1066, 0.0929, 0.1023, 0.0647, 0.0778, 0.1358, 0.0854, 0.0995], device='cuda:6'), in_proj_covar=tensor([0.0463, 0.0574, 0.0491, 0.0399, 0.0360, 0.0387, 0.0482, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:28:46,975 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79364.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:29:05,004 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79375.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 22:29:36,134 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79394.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:29:48,237 INFO [train.py:904] (6/8) Epoch 8, batch 8350, loss[loss=0.201, simple_loss=0.3006, pruned_loss=0.05076, over 16782.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.3021, pruned_loss=0.06719, over 3079173.36 frames. ], batch size: 102, lr: 8.38e-03, grad_scale: 4.0 2023-04-28 22:30:02,871 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.936e+02 2.645e+02 3.311e+02 4.098e+02 6.833e+02, threshold=6.622e+02, percent-clipped=0.0 2023-04-28 22:30:26,321 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79425.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:30:43,791 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79436.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 22:30:47,777 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0190, 2.3756, 2.3611, 3.0543, 2.1322, 3.3617, 1.6507, 2.8492], device='cuda:6'), covar=tensor([0.1173, 0.0528, 0.0864, 0.0114, 0.0106, 0.0347, 0.1355, 0.0586], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0152, 0.0172, 0.0117, 0.0198, 0.0200, 0.0172, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 22:31:09,026 INFO [train.py:904] (6/8) Epoch 8, batch 8400, loss[loss=0.1885, simple_loss=0.2811, pruned_loss=0.04789, over 16848.00 frames. ], tot_loss[loss=0.215, simple_loss=0.2993, pruned_loss=0.06529, over 3063081.98 frames. ], batch size: 116, lr: 8.38e-03, grad_scale: 8.0 2023-04-28 22:32:11,719 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-28 22:32:20,182 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9555, 5.3973, 5.5466, 5.3758, 5.4512, 5.9213, 5.4458, 5.2418], device='cuda:6'), covar=tensor([0.0758, 0.1378, 0.1221, 0.1866, 0.2233, 0.0816, 0.1298, 0.2147], device='cuda:6'), in_proj_covar=tensor([0.0310, 0.0421, 0.0452, 0.0379, 0.0496, 0.0474, 0.0368, 0.0503], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 22:32:27,061 INFO [train.py:904] (6/8) Epoch 8, batch 8450, loss[loss=0.1905, simple_loss=0.2818, pruned_loss=0.04965, over 16271.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.297, pruned_loss=0.06333, over 3060622.28 frames. ], batch size: 165, lr: 8.38e-03, grad_scale: 8.0 2023-04-28 22:32:42,128 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.576e+02 3.232e+02 4.028e+02 7.324e+02, threshold=6.464e+02, percent-clipped=2.0 2023-04-28 22:33:13,715 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9654, 2.3133, 1.9422, 2.0292, 2.6815, 2.3826, 2.9307, 2.9037], device='cuda:6'), covar=tensor([0.0067, 0.0264, 0.0330, 0.0313, 0.0179, 0.0230, 0.0123, 0.0134], device='cuda:6'), in_proj_covar=tensor([0.0106, 0.0179, 0.0177, 0.0176, 0.0174, 0.0178, 0.0171, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:33:47,271 INFO [train.py:904] (6/8) Epoch 8, batch 8500, loss[loss=0.1731, simple_loss=0.2494, pruned_loss=0.04841, over 11563.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2925, pruned_loss=0.06026, over 3053451.29 frames. ], batch size: 247, lr: 8.37e-03, grad_scale: 8.0 2023-04-28 22:34:40,491 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79585.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:35:09,510 INFO [train.py:904] (6/8) Epoch 8, batch 8550, loss[loss=0.1938, simple_loss=0.2872, pruned_loss=0.05022, over 16899.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2898, pruned_loss=0.05876, over 3047920.69 frames. ], batch size: 96, lr: 8.37e-03, grad_scale: 8.0 2023-04-28 22:35:26,482 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.911e+02 2.652e+02 3.316e+02 4.199e+02 1.038e+03, threshold=6.632e+02, percent-clipped=3.0 2023-04-28 22:35:29,797 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5566, 3.2944, 2.7714, 2.1015, 2.3135, 2.1595, 3.4285, 3.1494], device='cuda:6'), covar=tensor([0.2114, 0.0693, 0.1412, 0.2144, 0.2173, 0.1662, 0.0395, 0.0849], device='cuda:6'), in_proj_covar=tensor([0.0285, 0.0245, 0.0268, 0.0256, 0.0263, 0.0206, 0.0249, 0.0264], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:35:47,547 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79621.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:36:28,566 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79641.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 22:36:37,273 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79646.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:36:48,481 INFO [train.py:904] (6/8) Epoch 8, batch 8600, loss[loss=0.1956, simple_loss=0.2748, pruned_loss=0.05823, over 12567.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2906, pruned_loss=0.05808, over 3055352.10 frames. ], batch size: 247, lr: 8.37e-03, grad_scale: 8.0 2023-04-28 22:36:59,468 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79657.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:37:24,935 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=79669.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:37:33,105 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0690, 5.0393, 5.5125, 5.4851, 5.4828, 5.1878, 5.1170, 4.8643], device='cuda:6'), covar=tensor([0.0213, 0.0469, 0.0336, 0.0364, 0.0397, 0.0245, 0.0717, 0.0292], device='cuda:6'), in_proj_covar=tensor([0.0276, 0.0281, 0.0281, 0.0275, 0.0319, 0.0296, 0.0391, 0.0243], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-28 22:37:54,267 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.86 vs. limit=5.0 2023-04-28 22:38:02,630 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79689.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:38:26,215 INFO [train.py:904] (6/8) Epoch 8, batch 8650, loss[loss=0.1622, simple_loss=0.2655, pruned_loss=0.02945, over 16893.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2878, pruned_loss=0.05582, over 3051793.36 frames. ], batch size: 102, lr: 8.37e-03, grad_scale: 4.0 2023-04-28 22:38:50,263 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.014e+02 2.698e+02 3.186e+02 3.919e+02 1.176e+03, threshold=6.372e+02, percent-clipped=4.0 2023-04-28 22:39:04,746 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79718.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:39:09,483 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79720.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:39:32,960 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79731.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 22:39:51,454 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-04-28 22:40:12,066 INFO [train.py:904] (6/8) Epoch 8, batch 8700, loss[loss=0.1906, simple_loss=0.2813, pruned_loss=0.04992, over 15240.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2846, pruned_loss=0.05427, over 3051899.68 frames. ], batch size: 190, lr: 8.36e-03, grad_scale: 4.0 2023-04-28 22:40:33,027 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79762.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:41:50,475 INFO [train.py:904] (6/8) Epoch 8, batch 8750, loss[loss=0.1851, simple_loss=0.2791, pruned_loss=0.04556, over 16707.00 frames. ], tot_loss[loss=0.1963, simple_loss=0.2848, pruned_loss=0.05387, over 3061792.89 frames. ], batch size: 76, lr: 8.36e-03, grad_scale: 4.0 2023-04-28 22:42:15,364 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.669e+02 3.166e+02 4.113e+02 7.290e+02, threshold=6.332e+02, percent-clipped=2.0 2023-04-28 22:42:44,077 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79823.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:43:44,531 INFO [train.py:904] (6/8) Epoch 8, batch 8800, loss[loss=0.1838, simple_loss=0.2769, pruned_loss=0.04538, over 16299.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2827, pruned_loss=0.05257, over 3071506.14 frames. ], batch size: 146, lr: 8.36e-03, grad_scale: 8.0 2023-04-28 22:45:31,356 INFO [train.py:904] (6/8) Epoch 8, batch 8850, loss[loss=0.1939, simple_loss=0.2993, pruned_loss=0.04424, over 16850.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2851, pruned_loss=0.05178, over 3075319.19 frames. ], batch size: 124, lr: 8.36e-03, grad_scale: 8.0 2023-04-28 22:45:51,485 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.661e+02 3.236e+02 3.873e+02 8.211e+02, threshold=6.471e+02, percent-clipped=3.0 2023-04-28 22:46:58,213 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79941.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:46:58,238 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79941.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 22:47:20,936 INFO [train.py:904] (6/8) Epoch 8, batch 8900, loss[loss=0.18, simple_loss=0.2779, pruned_loss=0.04105, over 16866.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2847, pruned_loss=0.05081, over 3075045.09 frames. ], batch size: 90, lr: 8.35e-03, grad_scale: 8.0 2023-04-28 22:48:55,755 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=79989.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 22:48:55,783 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79989.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:49:29,483 INFO [train.py:904] (6/8) Epoch 8, batch 8950, loss[loss=0.2013, simple_loss=0.2788, pruned_loss=0.06194, over 12361.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2848, pruned_loss=0.05139, over 3081384.23 frames. ], batch size: 248, lr: 8.35e-03, grad_scale: 8.0 2023-04-28 22:49:50,485 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 2.495e+02 3.051e+02 3.831e+02 8.360e+02, threshold=6.103e+02, percent-clipped=2.0 2023-04-28 22:49:54,996 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80013.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:50:08,396 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80020.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:50:13,965 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2427, 4.3742, 4.1313, 3.9379, 3.6274, 4.2271, 4.0538, 3.8741], device='cuda:6'), covar=tensor([0.0599, 0.0334, 0.0324, 0.0284, 0.0965, 0.0376, 0.0545, 0.0629], device='cuda:6'), in_proj_covar=tensor([0.0204, 0.0244, 0.0242, 0.0214, 0.0259, 0.0246, 0.0166, 0.0276], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:50:32,382 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80031.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 22:50:46,302 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80037.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:50:59,150 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-28 22:51:17,195 INFO [train.py:904] (6/8) Epoch 8, batch 9000, loss[loss=0.1739, simple_loss=0.2577, pruned_loss=0.04509, over 16605.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.281, pruned_loss=0.04934, over 3105233.18 frames. ], batch size: 68, lr: 8.35e-03, grad_scale: 8.0 2023-04-28 22:51:17,195 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 22:51:27,533 INFO [train.py:938] (6/8) Epoch 8, validation: loss=0.1608, simple_loss=0.265, pruned_loss=0.02828, over 944034.00 frames. 2023-04-28 22:51:27,534 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-28 22:52:04,337 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80068.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:52:26,508 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80079.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 22:52:44,479 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.42 vs. limit=5.0 2023-04-28 22:53:14,309 INFO [train.py:904] (6/8) Epoch 8, batch 9050, loss[loss=0.184, simple_loss=0.2662, pruned_loss=0.05088, over 16752.00 frames. ], tot_loss[loss=0.191, simple_loss=0.2821, pruned_loss=0.05, over 3101155.53 frames. ], batch size: 124, lr: 8.35e-03, grad_scale: 8.0 2023-04-28 22:53:34,489 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7996, 3.7439, 3.9225, 3.7065, 3.8693, 4.2396, 3.9984, 3.6859], device='cuda:6'), covar=tensor([0.1761, 0.1896, 0.1458, 0.2312, 0.2332, 0.1310, 0.1280, 0.2324], device='cuda:6'), in_proj_covar=tensor([0.0299, 0.0411, 0.0438, 0.0370, 0.0481, 0.0458, 0.0358, 0.0483], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:53:35,355 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.717e+02 2.777e+02 3.475e+02 4.225e+02 7.695e+02, threshold=6.949e+02, percent-clipped=5.0 2023-04-28 22:53:48,954 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80118.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:54:59,312 INFO [train.py:904] (6/8) Epoch 8, batch 9100, loss[loss=0.1911, simple_loss=0.2901, pruned_loss=0.04607, over 16190.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2824, pruned_loss=0.05085, over 3107596.29 frames. ], batch size: 165, lr: 8.34e-03, grad_scale: 8.0 2023-04-28 22:56:28,802 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5032, 3.7275, 2.9953, 2.1555, 2.5348, 2.2151, 3.9158, 3.4116], device='cuda:6'), covar=tensor([0.2525, 0.0612, 0.1272, 0.1986, 0.2062, 0.1722, 0.0353, 0.0801], device='cuda:6'), in_proj_covar=tensor([0.0281, 0.0241, 0.0266, 0.0254, 0.0248, 0.0203, 0.0246, 0.0258], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 22:56:59,465 INFO [train.py:904] (6/8) Epoch 8, batch 9150, loss[loss=0.1723, simple_loss=0.2689, pruned_loss=0.03779, over 15361.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2819, pruned_loss=0.05025, over 3081811.79 frames. ], batch size: 191, lr: 8.34e-03, grad_scale: 8.0 2023-04-28 22:57:20,184 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.002e+02 2.760e+02 3.111e+02 3.901e+02 6.426e+02, threshold=6.222e+02, percent-clipped=0.0 2023-04-28 22:57:59,342 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-28 22:58:26,888 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80241.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:58:44,954 INFO [train.py:904] (6/8) Epoch 8, batch 9200, loss[loss=0.1618, simple_loss=0.2468, pruned_loss=0.03845, over 12240.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2776, pruned_loss=0.04897, over 3085433.36 frames. ], batch size: 248, lr: 8.34e-03, grad_scale: 8.0 2023-04-28 22:59:43,160 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80283.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:59:55,958 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80289.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:00:22,118 INFO [train.py:904] (6/8) Epoch 8, batch 9250, loss[loss=0.1497, simple_loss=0.233, pruned_loss=0.03319, over 12332.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2778, pruned_loss=0.0492, over 3085355.86 frames. ], batch size: 248, lr: 8.34e-03, grad_scale: 4.0 2023-04-28 23:00:42,874 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.717e+02 2.839e+02 3.478e+02 4.215e+02 8.620e+02, threshold=6.955e+02, percent-clipped=3.0 2023-04-28 23:00:44,620 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80313.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:00:57,700 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-28 23:01:56,438 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80344.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:02:12,076 INFO [train.py:904] (6/8) Epoch 8, batch 9300, loss[loss=0.1645, simple_loss=0.2595, pruned_loss=0.03472, over 16741.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2763, pruned_loss=0.04835, over 3084704.44 frames. ], batch size: 89, lr: 8.33e-03, grad_scale: 4.0 2023-04-28 23:02:23,319 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5098, 1.9791, 1.6049, 1.7185, 2.3311, 2.0489, 2.3427, 2.5043], device='cuda:6'), covar=tensor([0.0066, 0.0304, 0.0346, 0.0340, 0.0150, 0.0259, 0.0124, 0.0139], device='cuda:6'), in_proj_covar=tensor([0.0105, 0.0182, 0.0178, 0.0180, 0.0175, 0.0180, 0.0169, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:02:32,766 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80361.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:02:49,481 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4255, 2.9860, 2.5992, 2.2006, 2.1045, 2.1441, 2.9070, 2.8917], device='cuda:6'), covar=tensor([0.2182, 0.0814, 0.1296, 0.1984, 0.2047, 0.1662, 0.0474, 0.1055], device='cuda:6'), in_proj_covar=tensor([0.0281, 0.0243, 0.0266, 0.0255, 0.0248, 0.0204, 0.0248, 0.0259], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:03:25,907 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-28 23:03:34,054 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-28 23:03:42,327 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80393.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:03:57,480 INFO [train.py:904] (6/8) Epoch 8, batch 9350, loss[loss=0.1829, simple_loss=0.275, pruned_loss=0.04538, over 16720.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.276, pruned_loss=0.0483, over 3087692.85 frames. ], batch size: 83, lr: 8.33e-03, grad_scale: 4.0 2023-04-28 23:04:22,274 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.742e+02 2.575e+02 3.007e+02 3.554e+02 5.975e+02, threshold=6.013e+02, percent-clipped=0.0 2023-04-28 23:04:30,117 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9862, 3.0907, 3.0501, 2.0905, 2.9182, 3.0877, 2.9972, 1.7188], device='cuda:6'), covar=tensor([0.0354, 0.0034, 0.0034, 0.0301, 0.0067, 0.0065, 0.0056, 0.0394], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0060, 0.0062, 0.0118, 0.0068, 0.0077, 0.0068, 0.0113], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 23:04:34,124 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80418.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:05:40,898 INFO [train.py:904] (6/8) Epoch 8, batch 9400, loss[loss=0.1905, simple_loss=0.2686, pruned_loss=0.05622, over 12482.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2765, pruned_loss=0.04827, over 3083890.58 frames. ], batch size: 248, lr: 8.33e-03, grad_scale: 4.0 2023-04-28 23:05:46,325 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80454.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:06:05,202 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80463.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:06:10,060 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80466.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:07:07,267 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80494.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:07:19,864 INFO [train.py:904] (6/8) Epoch 8, batch 9450, loss[loss=0.1807, simple_loss=0.27, pruned_loss=0.04566, over 16907.00 frames. ], tot_loss[loss=0.188, simple_loss=0.278, pruned_loss=0.04905, over 3063424.00 frames. ], batch size: 109, lr: 8.33e-03, grad_scale: 4.0 2023-04-28 23:07:38,811 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.685e+02 2.597e+02 3.133e+02 4.086e+02 1.022e+03, threshold=6.266e+02, percent-clipped=6.0 2023-04-28 23:07:53,873 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-28 23:08:04,844 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80524.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:08:58,862 INFO [train.py:904] (6/8) Epoch 8, batch 9500, loss[loss=0.1932, simple_loss=0.2813, pruned_loss=0.05251, over 15459.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2773, pruned_loss=0.04851, over 3066169.64 frames. ], batch size: 191, lr: 8.32e-03, grad_scale: 4.0 2023-04-28 23:09:08,419 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80555.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:10:31,296 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6499, 4.4664, 4.5785, 4.8519, 4.9597, 4.5556, 5.0172, 5.0054], device='cuda:6'), covar=tensor([0.1246, 0.1061, 0.1598, 0.0720, 0.0694, 0.0586, 0.0509, 0.0632], device='cuda:6'), in_proj_covar=tensor([0.0438, 0.0538, 0.0657, 0.0554, 0.0414, 0.0411, 0.0434, 0.0477], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:10:46,592 INFO [train.py:904] (6/8) Epoch 8, batch 9550, loss[loss=0.184, simple_loss=0.2671, pruned_loss=0.05052, over 12583.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2776, pruned_loss=0.04908, over 3053783.89 frames. ], batch size: 247, lr: 8.32e-03, grad_scale: 4.0 2023-04-28 23:11:10,119 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.584e+02 3.110e+02 3.705e+02 8.746e+02, threshold=6.220e+02, percent-clipped=3.0 2023-04-28 23:11:55,714 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-28 23:12:04,483 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80639.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:12:21,302 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9762, 3.2565, 3.2717, 2.1174, 3.1275, 3.2224, 3.2239, 1.7210], device='cuda:6'), covar=tensor([0.0372, 0.0030, 0.0037, 0.0309, 0.0073, 0.0068, 0.0045, 0.0389], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0059, 0.0061, 0.0115, 0.0067, 0.0075, 0.0066, 0.0111], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 23:12:27,095 INFO [train.py:904] (6/8) Epoch 8, batch 9600, loss[loss=0.2373, simple_loss=0.3209, pruned_loss=0.07684, over 15479.00 frames. ], tot_loss[loss=0.1899, simple_loss=0.2796, pruned_loss=0.05011, over 3048434.18 frames. ], batch size: 191, lr: 8.32e-03, grad_scale: 8.0 2023-04-28 23:14:15,055 INFO [train.py:904] (6/8) Epoch 8, batch 9650, loss[loss=0.1866, simple_loss=0.2975, pruned_loss=0.03785, over 16898.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.2817, pruned_loss=0.05061, over 3038580.97 frames. ], batch size: 116, lr: 8.31e-03, grad_scale: 8.0 2023-04-28 23:14:33,448 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6278, 3.6532, 4.0954, 4.0443, 4.0398, 3.7345, 3.7616, 3.7799], device='cuda:6'), covar=tensor([0.0342, 0.0518, 0.0359, 0.0387, 0.0449, 0.0395, 0.0785, 0.0394], device='cuda:6'), in_proj_covar=tensor([0.0271, 0.0272, 0.0275, 0.0271, 0.0310, 0.0293, 0.0378, 0.0239], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-28 23:14:42,537 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-28 23:14:42,864 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.581e+02 3.070e+02 3.886e+02 7.619e+02, threshold=6.139e+02, percent-clipped=2.0 2023-04-28 23:15:30,100 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80736.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:15:59,669 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80749.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:16:03,264 INFO [train.py:904] (6/8) Epoch 8, batch 9700, loss[loss=0.1886, simple_loss=0.2733, pruned_loss=0.05195, over 12592.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.28, pruned_loss=0.0503, over 3032065.22 frames. ], batch size: 247, lr: 8.31e-03, grad_scale: 8.0 2023-04-28 23:16:46,594 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80773.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:17:38,603 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80797.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:17:46,348 INFO [train.py:904] (6/8) Epoch 8, batch 9750, loss[loss=0.176, simple_loss=0.276, pruned_loss=0.03804, over 15401.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2787, pruned_loss=0.0506, over 3012442.84 frames. ], batch size: 190, lr: 8.31e-03, grad_scale: 8.0 2023-04-28 23:18:08,278 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.771e+02 2.800e+02 3.435e+02 4.029e+02 7.858e+02, threshold=6.871e+02, percent-clipped=2.0 2023-04-28 23:18:21,090 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80819.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:18:55,741 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80834.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:18:59,940 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-28 23:19:24,253 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80850.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:19:26,320 INFO [train.py:904] (6/8) Epoch 8, batch 9800, loss[loss=0.196, simple_loss=0.2751, pruned_loss=0.05842, over 12340.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2788, pruned_loss=0.04945, over 3028244.14 frames. ], batch size: 248, lr: 8.31e-03, grad_scale: 8.0 2023-04-28 23:19:28,116 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.64 vs. limit=2.0 2023-04-28 23:21:11,973 INFO [train.py:904] (6/8) Epoch 8, batch 9850, loss[loss=0.1941, simple_loss=0.2862, pruned_loss=0.05098, over 16219.00 frames. ], tot_loss[loss=0.189, simple_loss=0.28, pruned_loss=0.04899, over 3039346.24 frames. ], batch size: 165, lr: 8.30e-03, grad_scale: 8.0 2023-04-28 23:21:33,316 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.396e+02 3.091e+02 3.765e+02 1.204e+03, threshold=6.182e+02, percent-clipped=3.0 2023-04-28 23:21:37,041 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1852, 3.0682, 3.1743, 1.7317, 3.3640, 3.4119, 2.7401, 2.6221], device='cuda:6'), covar=tensor([0.0782, 0.0196, 0.0166, 0.1235, 0.0064, 0.0092, 0.0398, 0.0440], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0094, 0.0078, 0.0138, 0.0064, 0.0085, 0.0114, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 23:22:37,459 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80939.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:23:04,047 INFO [train.py:904] (6/8) Epoch 8, batch 9900, loss[loss=0.1993, simple_loss=0.2904, pruned_loss=0.05405, over 15237.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2799, pruned_loss=0.04889, over 3023082.75 frames. ], batch size: 191, lr: 8.30e-03, grad_scale: 8.0 2023-04-28 23:23:12,137 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5360, 3.5214, 3.4654, 3.0960, 3.4631, 2.0136, 3.2806, 3.0987], device='cuda:6'), covar=tensor([0.0106, 0.0105, 0.0138, 0.0186, 0.0093, 0.1854, 0.0109, 0.0167], device='cuda:6'), in_proj_covar=tensor([0.0106, 0.0095, 0.0137, 0.0127, 0.0110, 0.0163, 0.0123, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-04-28 23:23:59,696 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5513, 4.8577, 4.6053, 4.6364, 4.3138, 4.2945, 4.4450, 4.8627], device='cuda:6'), covar=tensor([0.0898, 0.0821, 0.0816, 0.0551, 0.0729, 0.1059, 0.0778, 0.0814], device='cuda:6'), in_proj_covar=tensor([0.0454, 0.0577, 0.0473, 0.0394, 0.0356, 0.0384, 0.0475, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:24:29,639 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80987.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:24:57,518 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4926, 4.8363, 4.6112, 4.6117, 4.2994, 4.2807, 4.3055, 4.8435], device='cuda:6'), covar=tensor([0.0995, 0.0844, 0.0844, 0.0513, 0.0754, 0.1103, 0.0901, 0.0898], device='cuda:6'), in_proj_covar=tensor([0.0452, 0.0575, 0.0472, 0.0392, 0.0354, 0.0382, 0.0474, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:25:03,326 INFO [train.py:904] (6/8) Epoch 8, batch 9950, loss[loss=0.1821, simple_loss=0.2796, pruned_loss=0.0423, over 16881.00 frames. ], tot_loss[loss=0.1899, simple_loss=0.2817, pruned_loss=0.04901, over 3028288.41 frames. ], batch size: 116, lr: 8.30e-03, grad_scale: 8.0 2023-04-28 23:25:29,444 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.549e+02 2.486e+02 3.063e+02 3.674e+02 7.431e+02, threshold=6.127e+02, percent-clipped=1.0 2023-04-28 23:27:02,293 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81049.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:27:07,185 INFO [train.py:904] (6/8) Epoch 8, batch 10000, loss[loss=0.1885, simple_loss=0.282, pruned_loss=0.04751, over 15399.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2799, pruned_loss=0.04858, over 3028959.25 frames. ], batch size: 191, lr: 8.30e-03, grad_scale: 8.0 2023-04-28 23:28:04,996 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-28 23:28:30,841 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81092.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:28:40,810 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=81097.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:28:47,984 INFO [train.py:904] (6/8) Epoch 8, batch 10050, loss[loss=0.1664, simple_loss=0.2653, pruned_loss=0.0338, over 16874.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.2801, pruned_loss=0.04833, over 3048460.63 frames. ], batch size: 96, lr: 8.29e-03, grad_scale: 8.0 2023-04-28 23:29:08,273 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.518e+02 2.989e+02 3.596e+02 8.754e+02, threshold=5.978e+02, percent-clipped=2.0 2023-04-28 23:29:15,166 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81116.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:29:21,119 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81119.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:29:35,119 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0373, 2.6724, 2.6990, 1.8283, 2.8934, 2.9338, 2.5005, 2.3894], device='cuda:6'), covar=tensor([0.0714, 0.0202, 0.0181, 0.1050, 0.0081, 0.0123, 0.0413, 0.0399], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0092, 0.0076, 0.0135, 0.0063, 0.0083, 0.0112, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-28 23:29:40,302 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81129.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:29:44,782 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-28 23:30:10,822 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7336, 2.9509, 2.5885, 4.4607, 3.2393, 4.2758, 1.3510, 3.3231], device='cuda:6'), covar=tensor([0.1227, 0.0559, 0.0973, 0.0099, 0.0134, 0.0285, 0.1401, 0.0544], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0148, 0.0169, 0.0111, 0.0172, 0.0194, 0.0168, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 23:30:18,745 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81150.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:30:21,529 INFO [train.py:904] (6/8) Epoch 8, batch 10100, loss[loss=0.1769, simple_loss=0.2677, pruned_loss=0.04302, over 15433.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.2802, pruned_loss=0.04827, over 3064299.69 frames. ], batch size: 191, lr: 8.29e-03, grad_scale: 8.0 2023-04-28 23:30:43,058 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4309, 1.9552, 1.6696, 1.6505, 2.2818, 1.9106, 2.2293, 2.3846], device='cuda:6'), covar=tensor([0.0057, 0.0261, 0.0323, 0.0313, 0.0161, 0.0240, 0.0106, 0.0142], device='cuda:6'), in_proj_covar=tensor([0.0105, 0.0183, 0.0178, 0.0179, 0.0176, 0.0181, 0.0168, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:30:51,849 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=81167.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:31:14,301 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81177.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:31:16,861 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3488, 3.8995, 3.9874, 2.8149, 3.5809, 3.8771, 3.6222, 2.1303], device='cuda:6'), covar=tensor([0.0360, 0.0024, 0.0026, 0.0224, 0.0052, 0.0056, 0.0040, 0.0349], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0060, 0.0062, 0.0117, 0.0067, 0.0075, 0.0067, 0.0113], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 23:31:37,453 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=81198.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:32:08,878 INFO [train.py:904] (6/8) Epoch 9, batch 0, loss[loss=0.2778, simple_loss=0.3454, pruned_loss=0.1051, over 15634.00 frames. ], tot_loss[loss=0.2778, simple_loss=0.3454, pruned_loss=0.1051, over 15634.00 frames. ], batch size: 191, lr: 7.85e-03, grad_scale: 8.0 2023-04-28 23:32:08,878 INFO [train.py:929] (6/8) Computing validation loss 2023-04-28 23:32:16,263 INFO [train.py:938] (6/8) Epoch 9, validation: loss=0.1602, simple_loss=0.2637, pruned_loss=0.02837, over 944034.00 frames. 2023-04-28 23:32:16,263 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-28 23:32:36,784 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.828e+02 3.571e+02 4.681e+02 1.164e+03, threshold=7.142e+02, percent-clipped=9.0 2023-04-28 23:33:21,311 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6409, 2.8012, 2.4315, 2.6336, 3.0597, 2.9949, 3.6735, 3.3575], device='cuda:6'), covar=tensor([0.0064, 0.0243, 0.0289, 0.0255, 0.0165, 0.0208, 0.0117, 0.0146], device='cuda:6'), in_proj_covar=tensor([0.0107, 0.0184, 0.0180, 0.0179, 0.0178, 0.0182, 0.0170, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:33:25,079 INFO [train.py:904] (6/8) Epoch 9, batch 50, loss[loss=0.2244, simple_loss=0.2939, pruned_loss=0.07745, over 16366.00 frames. ], tot_loss[loss=0.2121, simple_loss=0.2898, pruned_loss=0.06724, over 751110.23 frames. ], batch size: 145, lr: 7.85e-03, grad_scale: 1.0 2023-04-28 23:34:31,186 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81299.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:34:34,783 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81301.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:34:35,496 INFO [train.py:904] (6/8) Epoch 9, batch 100, loss[loss=0.1744, simple_loss=0.2574, pruned_loss=0.04571, over 17246.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.2861, pruned_loss=0.06489, over 1325254.24 frames. ], batch size: 45, lr: 7.84e-03, grad_scale: 1.0 2023-04-28 23:34:54,535 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.676e+02 3.088e+02 4.301e+02 1.054e+03, threshold=6.177e+02, percent-clipped=2.0 2023-04-28 23:35:42,946 INFO [train.py:904] (6/8) Epoch 9, batch 150, loss[loss=0.1989, simple_loss=0.2759, pruned_loss=0.06099, over 16518.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2864, pruned_loss=0.06461, over 1759529.22 frames. ], batch size: 75, lr: 7.84e-03, grad_scale: 1.0 2023-04-28 23:35:55,902 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81360.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:35:58,281 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81362.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:36:40,035 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81392.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:36:54,427 INFO [train.py:904] (6/8) Epoch 9, batch 200, loss[loss=0.2564, simple_loss=0.3167, pruned_loss=0.09809, over 16546.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2879, pruned_loss=0.0657, over 2105600.26 frames. ], batch size: 75, lr: 7.84e-03, grad_scale: 1.0 2023-04-28 23:37:13,043 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.798e+02 2.694e+02 3.180e+02 3.951e+02 1.154e+03, threshold=6.361e+02, percent-clipped=2.0 2023-04-28 23:37:24,047 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-28 23:37:27,365 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.72 vs. limit=5.0 2023-04-28 23:37:31,959 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81429.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:37:46,218 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=81440.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:38:01,970 INFO [train.py:904] (6/8) Epoch 9, batch 250, loss[loss=0.2006, simple_loss=0.2793, pruned_loss=0.06099, over 15411.00 frames. ], tot_loss[loss=0.2082, simple_loss=0.2855, pruned_loss=0.06544, over 2372499.93 frames. ], batch size: 190, lr: 7.84e-03, grad_scale: 1.0 2023-04-28 23:38:17,616 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-28 23:38:29,698 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81472.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:38:36,874 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=81477.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:39:10,399 INFO [train.py:904] (6/8) Epoch 9, batch 300, loss[loss=0.1815, simple_loss=0.2662, pruned_loss=0.04837, over 17149.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2815, pruned_loss=0.06261, over 2586956.76 frames. ], batch size: 46, lr: 7.83e-03, grad_scale: 1.0 2023-04-28 23:39:29,675 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.655e+02 2.475e+02 2.919e+02 3.755e+02 7.155e+02, threshold=5.837e+02, percent-clipped=3.0 2023-04-28 23:40:17,834 INFO [train.py:904] (6/8) Epoch 9, batch 350, loss[loss=0.1751, simple_loss=0.2523, pruned_loss=0.04896, over 16352.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2787, pruned_loss=0.06099, over 2756698.26 frames. ], batch size: 36, lr: 7.83e-03, grad_scale: 1.0 2023-04-28 23:41:07,943 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-28 23:41:08,694 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7930, 4.9577, 4.6769, 4.4707, 3.7470, 4.8977, 4.9837, 4.3190], device='cuda:6'), covar=tensor([0.0903, 0.0436, 0.0441, 0.0364, 0.1964, 0.0412, 0.0314, 0.0775], device='cuda:6'), in_proj_covar=tensor([0.0226, 0.0264, 0.0260, 0.0233, 0.0286, 0.0267, 0.0179, 0.0302], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:41:23,824 INFO [train.py:904] (6/8) Epoch 9, batch 400, loss[loss=0.2046, simple_loss=0.2718, pruned_loss=0.06872, over 16796.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2777, pruned_loss=0.06052, over 2872863.55 frames. ], batch size: 83, lr: 7.83e-03, grad_scale: 2.0 2023-04-28 23:41:38,570 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8090, 4.5887, 4.8343, 5.0123, 5.1965, 4.6554, 5.1912, 5.1582], device='cuda:6'), covar=tensor([0.1443, 0.1057, 0.1541, 0.0625, 0.0477, 0.0783, 0.0527, 0.0515], device='cuda:6'), in_proj_covar=tensor([0.0495, 0.0605, 0.0747, 0.0615, 0.0460, 0.0454, 0.0486, 0.0532], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:41:43,584 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.573e+02 2.386e+02 2.843e+02 3.463e+02 6.249e+02, threshold=5.687e+02, percent-clipped=1.0 2023-04-28 23:42:27,876 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5331, 4.5500, 4.7452, 3.5668, 4.0488, 4.6053, 4.0720, 3.1327], device='cuda:6'), covar=tensor([0.0221, 0.0035, 0.0021, 0.0202, 0.0054, 0.0053, 0.0036, 0.0253], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0065, 0.0065, 0.0120, 0.0069, 0.0078, 0.0069, 0.0114], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-28 23:42:33,320 INFO [train.py:904] (6/8) Epoch 9, batch 450, loss[loss=0.2123, simple_loss=0.2782, pruned_loss=0.07317, over 16887.00 frames. ], tot_loss[loss=0.1969, simple_loss=0.2751, pruned_loss=0.05932, over 2972443.69 frames. ], batch size: 116, lr: 7.83e-03, grad_scale: 2.0 2023-04-28 23:42:37,318 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81655.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:42:37,439 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4473, 4.3820, 4.2834, 4.1154, 3.9519, 4.3384, 4.2195, 4.0682], device='cuda:6'), covar=tensor([0.0573, 0.0467, 0.0252, 0.0226, 0.0888, 0.0433, 0.0431, 0.0541], device='cuda:6'), in_proj_covar=tensor([0.0228, 0.0267, 0.0263, 0.0236, 0.0289, 0.0271, 0.0181, 0.0304], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:42:40,765 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81657.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:42:58,904 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81670.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:43:09,350 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7574, 2.8218, 2.6388, 4.6858, 3.7828, 4.2486, 1.5217, 3.1069], device='cuda:6'), covar=tensor([0.1414, 0.0736, 0.1156, 0.0147, 0.0332, 0.0415, 0.1590, 0.0761], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0153, 0.0175, 0.0122, 0.0189, 0.0207, 0.0174, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-28 23:43:40,761 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-28 23:43:42,338 INFO [train.py:904] (6/8) Epoch 9, batch 500, loss[loss=0.1853, simple_loss=0.2736, pruned_loss=0.04849, over 17058.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2733, pruned_loss=0.05787, over 3061078.18 frames. ], batch size: 53, lr: 7.82e-03, grad_scale: 2.0 2023-04-28 23:44:01,900 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.446e+02 2.865e+02 3.583e+02 8.926e+02, threshold=5.729e+02, percent-clipped=4.0 2023-04-28 23:44:22,636 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81731.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 23:44:35,744 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8134, 2.1884, 2.1786, 4.6559, 2.0321, 2.8402, 2.2467, 2.4094], device='cuda:6'), covar=tensor([0.0772, 0.3109, 0.2071, 0.0301, 0.3711, 0.1837, 0.2736, 0.3250], device='cuda:6'), in_proj_covar=tensor([0.0346, 0.0370, 0.0311, 0.0323, 0.0399, 0.0407, 0.0330, 0.0433], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:44:39,564 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6309, 4.9766, 4.7332, 4.7500, 4.4040, 4.3981, 4.4749, 5.0231], device='cuda:6'), covar=tensor([0.1083, 0.0900, 0.1065, 0.0604, 0.0858, 0.1068, 0.0956, 0.0834], device='cuda:6'), in_proj_covar=tensor([0.0497, 0.0640, 0.0533, 0.0436, 0.0393, 0.0414, 0.0528, 0.0471], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:44:49,401 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.69 vs. limit=5.0 2023-04-28 23:44:50,828 INFO [train.py:904] (6/8) Epoch 9, batch 550, loss[loss=0.2024, simple_loss=0.2847, pruned_loss=0.06005, over 16618.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.2726, pruned_loss=0.05731, over 3123810.55 frames. ], batch size: 62, lr: 7.82e-03, grad_scale: 2.0 2023-04-28 23:45:19,903 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81772.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:46:02,608 INFO [train.py:904] (6/8) Epoch 9, batch 600, loss[loss=0.1881, simple_loss=0.2485, pruned_loss=0.0639, over 16554.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2718, pruned_loss=0.05731, over 3150150.46 frames. ], batch size: 75, lr: 7.82e-03, grad_scale: 2.0 2023-04-28 23:46:06,174 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-28 23:46:21,575 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.753e+02 2.670e+02 3.296e+02 4.052e+02 8.860e+02, threshold=6.592e+02, percent-clipped=6.0 2023-04-28 23:46:26,574 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=81820.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:46:28,902 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-28 23:46:40,849 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-28 23:47:09,564 INFO [train.py:904] (6/8) Epoch 9, batch 650, loss[loss=0.177, simple_loss=0.2489, pruned_loss=0.05252, over 16707.00 frames. ], tot_loss[loss=0.1923, simple_loss=0.2708, pruned_loss=0.05687, over 3189555.35 frames. ], batch size: 89, lr: 7.82e-03, grad_scale: 2.0 2023-04-28 23:48:17,847 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-28 23:48:18,161 INFO [train.py:904] (6/8) Epoch 9, batch 700, loss[loss=0.1709, simple_loss=0.2526, pruned_loss=0.04462, over 16828.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2702, pruned_loss=0.05681, over 3222289.02 frames. ], batch size: 42, lr: 7.81e-03, grad_scale: 2.0 2023-04-28 23:48:37,201 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.675e+02 2.553e+02 3.037e+02 4.738e+02 2.852e+03, threshold=6.073e+02, percent-clipped=12.0 2023-04-28 23:49:20,036 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1848, 5.5068, 5.2500, 5.2664, 4.8974, 4.8247, 4.9256, 5.5797], device='cuda:6'), covar=tensor([0.0871, 0.0803, 0.0989, 0.0576, 0.0676, 0.0778, 0.0833, 0.0751], device='cuda:6'), in_proj_covar=tensor([0.0496, 0.0639, 0.0533, 0.0436, 0.0394, 0.0411, 0.0528, 0.0469], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:49:25,153 INFO [train.py:904] (6/8) Epoch 9, batch 750, loss[loss=0.1887, simple_loss=0.2791, pruned_loss=0.04918, over 17175.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.27, pruned_loss=0.05622, over 3245843.61 frames. ], batch size: 46, lr: 7.81e-03, grad_scale: 2.0 2023-04-28 23:49:28,078 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5657, 3.5716, 2.6553, 2.2124, 2.4610, 2.1249, 3.5200, 3.3889], device='cuda:6'), covar=tensor([0.2088, 0.0623, 0.1377, 0.1912, 0.2048, 0.1702, 0.0474, 0.0986], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0253, 0.0278, 0.0266, 0.0270, 0.0214, 0.0257, 0.0281], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:49:29,106 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81955.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:49:31,501 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81957.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:50:05,358 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6925, 4.8315, 5.2844, 5.2428, 5.2244, 4.8784, 4.8216, 4.5423], device='cuda:6'), covar=tensor([0.0286, 0.0400, 0.0315, 0.0354, 0.0368, 0.0298, 0.0844, 0.0403], device='cuda:6'), in_proj_covar=tensor([0.0307, 0.0315, 0.0314, 0.0304, 0.0353, 0.0334, 0.0431, 0.0270], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 23:50:14,435 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81987.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:50:37,832 INFO [train.py:904] (6/8) Epoch 9, batch 800, loss[loss=0.1859, simple_loss=0.2746, pruned_loss=0.04864, over 17164.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2694, pruned_loss=0.05612, over 3262980.63 frames. ], batch size: 46, lr: 7.81e-03, grad_scale: 4.0 2023-04-28 23:50:39,973 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=82003.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:50:42,266 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=82005.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:50:49,961 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1073, 5.5381, 5.7288, 5.4186, 5.5489, 6.0981, 5.7368, 5.4486], device='cuda:6'), covar=tensor([0.0708, 0.1665, 0.1712, 0.2034, 0.2665, 0.0975, 0.1056, 0.2106], device='cuda:6'), in_proj_covar=tensor([0.0331, 0.0465, 0.0497, 0.0413, 0.0550, 0.0522, 0.0395, 0.0547], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-28 23:50:56,930 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.806e+02 2.619e+02 2.968e+02 3.371e+02 5.495e+02, threshold=5.935e+02, percent-clipped=0.0 2023-04-28 23:51:10,864 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82026.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 23:51:40,959 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82048.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:51:45,036 INFO [train.py:904] (6/8) Epoch 9, batch 850, loss[loss=0.1873, simple_loss=0.2767, pruned_loss=0.04898, over 17060.00 frames. ], tot_loss[loss=0.1906, simple_loss=0.2699, pruned_loss=0.05567, over 3277022.76 frames. ], batch size: 50, lr: 7.81e-03, grad_scale: 4.0 2023-04-28 23:51:49,569 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82055.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:52:24,224 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8450, 2.1137, 2.1701, 4.6548, 2.0374, 2.7452, 2.2472, 2.4424], device='cuda:6'), covar=tensor([0.0750, 0.3222, 0.2141, 0.0289, 0.3692, 0.2068, 0.2825, 0.3192], device='cuda:6'), in_proj_covar=tensor([0.0351, 0.0373, 0.0315, 0.0325, 0.0402, 0.0414, 0.0333, 0.0439], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:52:54,644 INFO [train.py:904] (6/8) Epoch 9, batch 900, loss[loss=0.1798, simple_loss=0.2673, pruned_loss=0.0462, over 17047.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2685, pruned_loss=0.05437, over 3293272.60 frames. ], batch size: 53, lr: 7.80e-03, grad_scale: 4.0 2023-04-28 23:53:13,855 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.661e+02 2.604e+02 3.157e+02 3.589e+02 5.638e+02, threshold=6.315e+02, percent-clipped=0.0 2023-04-28 23:53:14,988 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82116.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:53:17,312 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0280, 2.1921, 2.4390, 4.8119, 2.1885, 2.8520, 2.3725, 2.5320], device='cuda:6'), covar=tensor([0.0718, 0.3208, 0.1963, 0.0270, 0.3590, 0.2109, 0.2808, 0.3126], device='cuda:6'), in_proj_covar=tensor([0.0351, 0.0372, 0.0315, 0.0326, 0.0401, 0.0415, 0.0332, 0.0439], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:53:38,581 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9255, 5.0147, 5.5441, 5.4969, 5.4742, 5.1259, 5.0730, 4.8519], device='cuda:6'), covar=tensor([0.0276, 0.0342, 0.0308, 0.0380, 0.0374, 0.0282, 0.0864, 0.0351], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0318, 0.0318, 0.0311, 0.0362, 0.0339, 0.0437, 0.0272], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-28 23:53:52,810 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8005, 3.9652, 2.2416, 4.4592, 2.7336, 4.4221, 2.2076, 3.1274], device='cuda:6'), covar=tensor([0.0209, 0.0340, 0.1477, 0.0163, 0.0837, 0.0385, 0.1484, 0.0620], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0162, 0.0185, 0.0113, 0.0165, 0.0199, 0.0192, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 23:54:03,806 INFO [train.py:904] (6/8) Epoch 9, batch 950, loss[loss=0.1798, simple_loss=0.2728, pruned_loss=0.04333, over 17139.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.269, pruned_loss=0.05465, over 3304993.50 frames. ], batch size: 48, lr: 7.80e-03, grad_scale: 4.0 2023-04-28 23:54:24,205 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6807, 3.8600, 2.1327, 4.0280, 2.7549, 3.9829, 2.0899, 3.0022], device='cuda:6'), covar=tensor([0.0194, 0.0302, 0.1496, 0.0208, 0.0784, 0.0470, 0.1402, 0.0592], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0163, 0.0186, 0.0114, 0.0165, 0.0200, 0.0193, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 23:55:11,097 INFO [train.py:904] (6/8) Epoch 9, batch 1000, loss[loss=0.2002, simple_loss=0.2685, pruned_loss=0.06594, over 12100.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2677, pruned_loss=0.05411, over 3302091.28 frames. ], batch size: 247, lr: 7.80e-03, grad_scale: 4.0 2023-04-28 23:55:31,950 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 2.342e+02 2.920e+02 3.464e+02 5.943e+02, threshold=5.839e+02, percent-clipped=0.0 2023-04-28 23:55:37,150 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1253, 2.0123, 2.2633, 3.4992, 1.9949, 2.3238, 2.1376, 2.1848], device='cuda:6'), covar=tensor([0.0943, 0.2862, 0.1894, 0.0536, 0.3358, 0.2062, 0.2893, 0.2730], device='cuda:6'), in_proj_covar=tensor([0.0353, 0.0374, 0.0316, 0.0327, 0.0401, 0.0416, 0.0333, 0.0439], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:56:20,454 INFO [train.py:904] (6/8) Epoch 9, batch 1050, loss[loss=0.1671, simple_loss=0.2415, pruned_loss=0.04633, over 16795.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2666, pruned_loss=0.05397, over 3291991.42 frames. ], batch size: 39, lr: 7.80e-03, grad_scale: 4.0 2023-04-28 23:57:13,007 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-28 23:57:28,537 INFO [train.py:904] (6/8) Epoch 9, batch 1100, loss[loss=0.1733, simple_loss=0.252, pruned_loss=0.04728, over 15916.00 frames. ], tot_loss[loss=0.187, simple_loss=0.2665, pruned_loss=0.05379, over 3290629.90 frames. ], batch size: 35, lr: 7.80e-03, grad_scale: 4.0 2023-04-28 23:57:39,205 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3152, 4.3038, 4.2312, 3.7776, 4.2490, 1.8169, 4.0338, 3.9610], device='cuda:6'), covar=tensor([0.0088, 0.0072, 0.0119, 0.0274, 0.0074, 0.1997, 0.0110, 0.0148], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0108, 0.0156, 0.0150, 0.0125, 0.0174, 0.0142, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-28 23:57:47,216 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.669e+02 2.514e+02 3.056e+02 3.616e+02 1.290e+03, threshold=6.113e+02, percent-clipped=7.0 2023-04-28 23:58:01,825 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82326.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:58:23,392 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4487, 4.0006, 3.8956, 2.1091, 3.1624, 2.5584, 3.9778, 3.9868], device='cuda:6'), covar=tensor([0.0260, 0.0581, 0.0466, 0.1563, 0.0730, 0.0868, 0.0571, 0.0867], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0136, 0.0155, 0.0141, 0.0134, 0.0124, 0.0134, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 23:58:24,280 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82343.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:58:28,003 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82346.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:58:35,310 INFO [train.py:904] (6/8) Epoch 9, batch 1150, loss[loss=0.1719, simple_loss=0.2511, pruned_loss=0.04632, over 16797.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2662, pruned_loss=0.05311, over 3290022.83 frames. ], batch size: 39, lr: 7.79e-03, grad_scale: 4.0 2023-04-28 23:58:54,946 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6647, 3.7915, 2.0717, 3.9456, 2.7662, 3.9095, 2.2270, 2.8692], device='cuda:6'), covar=tensor([0.0185, 0.0295, 0.1509, 0.0186, 0.0696, 0.0560, 0.1360, 0.0621], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0163, 0.0185, 0.0114, 0.0165, 0.0202, 0.0193, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-28 23:59:04,895 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=82374.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:59:44,518 INFO [train.py:904] (6/8) Epoch 9, batch 1200, loss[loss=0.1708, simple_loss=0.2487, pruned_loss=0.04651, over 16808.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2654, pruned_loss=0.05276, over 3298143.00 frames. ], batch size: 42, lr: 7.79e-03, grad_scale: 8.0 2023-04-28 23:59:50,700 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82407.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:59:55,272 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82411.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:00:02,690 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.435e+02 2.375e+02 3.047e+02 3.937e+02 1.504e+03, threshold=6.095e+02, percent-clipped=2.0 2023-04-29 00:00:50,233 INFO [train.py:904] (6/8) Epoch 9, batch 1250, loss[loss=0.1865, simple_loss=0.2594, pruned_loss=0.0568, over 16671.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2654, pruned_loss=0.0534, over 3305896.96 frames. ], batch size: 62, lr: 7.79e-03, grad_scale: 4.0 2023-04-29 00:00:57,267 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-29 00:01:47,069 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82493.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 00:01:58,487 INFO [train.py:904] (6/8) Epoch 9, batch 1300, loss[loss=0.1849, simple_loss=0.2555, pruned_loss=0.05714, over 12203.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.266, pruned_loss=0.05421, over 3302988.98 frames. ], batch size: 247, lr: 7.79e-03, grad_scale: 4.0 2023-04-29 00:02:17,334 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4542, 4.4151, 4.3481, 3.9338, 4.3525, 1.6483, 4.1202, 4.1223], device='cuda:6'), covar=tensor([0.0086, 0.0071, 0.0124, 0.0262, 0.0079, 0.2186, 0.0117, 0.0155], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0109, 0.0158, 0.0152, 0.0127, 0.0174, 0.0143, 0.0149], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:02:18,040 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.645e+02 2.404e+02 2.965e+02 3.987e+02 6.881e+02, threshold=5.930e+02, percent-clipped=4.0 2023-04-29 00:03:05,225 INFO [train.py:904] (6/8) Epoch 9, batch 1350, loss[loss=0.1897, simple_loss=0.2569, pruned_loss=0.0612, over 16685.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2663, pruned_loss=0.05339, over 3313662.72 frames. ], batch size: 89, lr: 7.78e-03, grad_scale: 4.0 2023-04-29 00:03:07,954 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82554.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 00:03:29,880 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82569.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:04:10,508 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-29 00:04:12,575 INFO [train.py:904] (6/8) Epoch 9, batch 1400, loss[loss=0.1886, simple_loss=0.264, pruned_loss=0.05658, over 16488.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2668, pruned_loss=0.0539, over 3319927.76 frames. ], batch size: 75, lr: 7.78e-03, grad_scale: 4.0 2023-04-29 00:04:21,807 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1409, 5.7156, 5.9464, 5.6945, 5.7361, 6.2237, 5.7647, 5.4217], device='cuda:6'), covar=tensor([0.0746, 0.1627, 0.1495, 0.1704, 0.2434, 0.0872, 0.1375, 0.2444], device='cuda:6'), in_proj_covar=tensor([0.0332, 0.0466, 0.0497, 0.0412, 0.0548, 0.0521, 0.0396, 0.0550], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 00:04:31,977 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-04-29 00:04:33,568 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.521e+02 2.260e+02 2.970e+02 4.048e+02 1.598e+03, threshold=5.940e+02, percent-clipped=4.0 2023-04-29 00:04:53,117 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82630.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:05:09,543 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82643.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:05:22,656 INFO [train.py:904] (6/8) Epoch 9, batch 1450, loss[loss=0.1543, simple_loss=0.2373, pruned_loss=0.03565, over 16816.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2659, pruned_loss=0.0537, over 3317695.29 frames. ], batch size: 39, lr: 7.78e-03, grad_scale: 4.0 2023-04-29 00:05:38,607 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9102, 4.0847, 1.8593, 4.5411, 2.7373, 4.5021, 2.0766, 3.0082], device='cuda:6'), covar=tensor([0.0193, 0.0249, 0.1705, 0.0148, 0.0822, 0.0311, 0.1509, 0.0660], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0162, 0.0183, 0.0116, 0.0164, 0.0202, 0.0193, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 00:06:05,922 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9655, 1.7460, 2.4366, 2.9066, 2.8465, 2.8841, 1.9565, 2.9893], device='cuda:6'), covar=tensor([0.0108, 0.0303, 0.0179, 0.0155, 0.0148, 0.0131, 0.0296, 0.0092], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0164, 0.0148, 0.0150, 0.0155, 0.0112, 0.0162, 0.0103], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 00:06:16,062 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=82691.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:06:29,712 INFO [train.py:904] (6/8) Epoch 9, batch 1500, loss[loss=0.2006, simple_loss=0.2739, pruned_loss=0.06366, over 16402.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2662, pruned_loss=0.05422, over 3301348.02 frames. ], batch size: 68, lr: 7.78e-03, grad_scale: 4.0 2023-04-29 00:06:29,979 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82702.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:06:42,806 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82711.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:06:51,612 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.540e+02 3.048e+02 3.562e+02 8.559e+02, threshold=6.096e+02, percent-clipped=4.0 2023-04-29 00:07:31,451 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0890, 4.1370, 4.5436, 4.4948, 4.5536, 4.1805, 4.2613, 4.0951], device='cuda:6'), covar=tensor([0.0352, 0.0600, 0.0389, 0.0446, 0.0400, 0.0398, 0.0754, 0.0508], device='cuda:6'), in_proj_covar=tensor([0.0317, 0.0323, 0.0327, 0.0311, 0.0368, 0.0343, 0.0444, 0.0274], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 00:07:34,644 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2578, 1.9236, 2.2124, 3.7876, 2.0213, 2.5002, 2.1073, 2.0955], device='cuda:6'), covar=tensor([0.0861, 0.2967, 0.1872, 0.0415, 0.3225, 0.1823, 0.2759, 0.2682], device='cuda:6'), in_proj_covar=tensor([0.0354, 0.0375, 0.0317, 0.0327, 0.0401, 0.0421, 0.0335, 0.0442], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:07:39,171 INFO [train.py:904] (6/8) Epoch 9, batch 1550, loss[loss=0.2533, simple_loss=0.3322, pruned_loss=0.08722, over 12092.00 frames. ], tot_loss[loss=0.1895, simple_loss=0.2681, pruned_loss=0.05542, over 3297735.63 frames. ], batch size: 246, lr: 7.77e-03, grad_scale: 4.0 2023-04-29 00:07:49,786 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=82759.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:07:51,056 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4227, 4.2241, 4.4379, 4.6118, 4.7184, 4.2855, 4.4783, 4.6575], device='cuda:6'), covar=tensor([0.1113, 0.0959, 0.1191, 0.0532, 0.0487, 0.0975, 0.1942, 0.0569], device='cuda:6'), in_proj_covar=tensor([0.0526, 0.0640, 0.0807, 0.0658, 0.0492, 0.0486, 0.0512, 0.0570], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:07:52,839 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-29 00:07:55,213 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-29 00:08:48,744 INFO [train.py:904] (6/8) Epoch 9, batch 1600, loss[loss=0.1688, simple_loss=0.2695, pruned_loss=0.03406, over 17100.00 frames. ], tot_loss[loss=0.1916, simple_loss=0.2702, pruned_loss=0.05645, over 3310834.49 frames. ], batch size: 47, lr: 7.77e-03, grad_scale: 8.0 2023-04-29 00:09:03,056 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9746, 3.9242, 4.2436, 1.9032, 4.5470, 4.4857, 3.2651, 3.5099], device='cuda:6'), covar=tensor([0.0620, 0.0158, 0.0165, 0.1160, 0.0045, 0.0128, 0.0348, 0.0353], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0098, 0.0086, 0.0141, 0.0070, 0.0097, 0.0119, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 00:09:09,713 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.618e+02 2.591e+02 3.276e+02 4.037e+02 8.145e+02, threshold=6.551e+02, percent-clipped=5.0 2023-04-29 00:09:11,446 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8977, 4.0701, 2.2001, 4.5551, 2.7169, 4.5227, 2.1866, 3.1652], device='cuda:6'), covar=tensor([0.0210, 0.0288, 0.1570, 0.0140, 0.0872, 0.0365, 0.1602, 0.0665], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0164, 0.0186, 0.0118, 0.0166, 0.0205, 0.0196, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 00:09:31,283 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-29 00:09:52,304 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82849.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 00:09:56,031 INFO [train.py:904] (6/8) Epoch 9, batch 1650, loss[loss=0.17, simple_loss=0.2647, pruned_loss=0.03768, over 17113.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2719, pruned_loss=0.05685, over 3315232.06 frames. ], batch size: 48, lr: 7.77e-03, grad_scale: 8.0 2023-04-29 00:10:36,939 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0983, 5.6273, 5.8237, 5.5753, 5.6708, 6.1898, 5.7938, 5.4577], device='cuda:6'), covar=tensor([0.0769, 0.1505, 0.1571, 0.1778, 0.2542, 0.0858, 0.1213, 0.2130], device='cuda:6'), in_proj_covar=tensor([0.0341, 0.0474, 0.0504, 0.0420, 0.0558, 0.0532, 0.0402, 0.0560], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 00:11:03,715 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-29 00:11:05,715 INFO [train.py:904] (6/8) Epoch 9, batch 1700, loss[loss=0.1958, simple_loss=0.2692, pruned_loss=0.06115, over 16881.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.2744, pruned_loss=0.05797, over 3306022.30 frames. ], batch size: 109, lr: 7.77e-03, grad_scale: 8.0 2023-04-29 00:11:24,536 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.504e+02 2.790e+02 3.377e+02 4.170e+02 9.157e+02, threshold=6.754e+02, percent-clipped=3.0 2023-04-29 00:11:32,129 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4902, 4.0751, 4.2070, 2.7879, 3.7978, 4.2192, 3.8952, 2.3858], device='cuda:6'), covar=tensor([0.0353, 0.0057, 0.0031, 0.0259, 0.0052, 0.0056, 0.0044, 0.0299], device='cuda:6'), in_proj_covar=tensor([0.0123, 0.0067, 0.0065, 0.0121, 0.0070, 0.0080, 0.0071, 0.0114], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 00:11:37,172 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82925.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:11:40,675 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0447, 5.0185, 4.7691, 4.2438, 4.8741, 1.8683, 4.6224, 4.7950], device='cuda:6'), covar=tensor([0.0070, 0.0059, 0.0143, 0.0361, 0.0081, 0.2241, 0.0118, 0.0147], device='cuda:6'), in_proj_covar=tensor([0.0124, 0.0111, 0.0161, 0.0155, 0.0130, 0.0176, 0.0145, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:11:40,923 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 00:12:06,726 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0871, 2.3025, 1.6671, 2.0629, 2.7817, 2.5413, 3.2319, 2.9581], device='cuda:6'), covar=tensor([0.0153, 0.0355, 0.0461, 0.0409, 0.0207, 0.0286, 0.0197, 0.0204], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0197, 0.0190, 0.0193, 0.0192, 0.0196, 0.0199, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:12:13,189 INFO [train.py:904] (6/8) Epoch 9, batch 1750, loss[loss=0.2117, simple_loss=0.2877, pruned_loss=0.06788, over 16708.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2751, pruned_loss=0.05815, over 3311042.50 frames. ], batch size: 134, lr: 7.76e-03, grad_scale: 8.0 2023-04-29 00:12:28,824 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82963.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:12:32,295 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2023-04-29 00:12:33,798 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5231, 2.1827, 2.3213, 4.3504, 2.1755, 2.7827, 2.2966, 2.4667], device='cuda:6'), covar=tensor([0.0874, 0.3064, 0.1849, 0.0350, 0.3293, 0.1958, 0.2755, 0.2638], device='cuda:6'), in_proj_covar=tensor([0.0350, 0.0373, 0.0313, 0.0323, 0.0399, 0.0420, 0.0334, 0.0440], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:13:15,137 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7620, 3.4345, 2.9814, 5.1855, 4.4107, 4.8584, 1.5984, 3.5529], device='cuda:6'), covar=tensor([0.1332, 0.0535, 0.0932, 0.0118, 0.0279, 0.0314, 0.1391, 0.0591], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0152, 0.0174, 0.0126, 0.0196, 0.0210, 0.0171, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 00:13:19,612 INFO [train.py:904] (6/8) Epoch 9, batch 1800, loss[loss=0.1856, simple_loss=0.2799, pruned_loss=0.04563, over 17045.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.2758, pruned_loss=0.05793, over 3316729.33 frames. ], batch size: 53, lr: 7.76e-03, grad_scale: 8.0 2023-04-29 00:13:20,534 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83002.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:13:39,506 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5489, 5.3719, 5.3631, 5.0331, 4.9405, 5.3597, 5.4346, 5.0211], device='cuda:6'), covar=tensor([0.0439, 0.0374, 0.0229, 0.0217, 0.0924, 0.0349, 0.0191, 0.0556], device='cuda:6'), in_proj_covar=tensor([0.0242, 0.0290, 0.0285, 0.0255, 0.0308, 0.0291, 0.0196, 0.0326], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 00:13:40,258 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.771e+02 2.500e+02 2.970e+02 3.668e+02 6.624e+02, threshold=5.940e+02, percent-clipped=0.0 2023-04-29 00:13:50,874 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83024.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:14:06,212 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-29 00:14:26,663 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=83050.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:14:28,691 INFO [train.py:904] (6/8) Epoch 9, batch 1850, loss[loss=0.2065, simple_loss=0.2964, pruned_loss=0.05831, over 16679.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.2764, pruned_loss=0.05764, over 3318515.56 frames. ], batch size: 57, lr: 7.76e-03, grad_scale: 8.0 2023-04-29 00:15:37,693 INFO [train.py:904] (6/8) Epoch 9, batch 1900, loss[loss=0.2037, simple_loss=0.2752, pruned_loss=0.06604, over 16901.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2748, pruned_loss=0.05638, over 3326611.19 frames. ], batch size: 109, lr: 7.76e-03, grad_scale: 8.0 2023-04-29 00:15:59,152 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.589e+02 2.476e+02 2.876e+02 3.479e+02 6.930e+02, threshold=5.751e+02, percent-clipped=2.0 2023-04-29 00:16:36,171 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.4660, 3.0522, 2.4890, 4.8473, 3.9905, 4.4252, 1.3450, 3.0061], device='cuda:6'), covar=tensor([0.1591, 0.0652, 0.1311, 0.0188, 0.0303, 0.0428, 0.1777, 0.0874], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0152, 0.0175, 0.0126, 0.0195, 0.0210, 0.0171, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 00:16:42,418 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83149.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 00:16:46,825 INFO [train.py:904] (6/8) Epoch 9, batch 1950, loss[loss=0.2229, simple_loss=0.2957, pruned_loss=0.07508, over 16797.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2751, pruned_loss=0.05627, over 3325348.99 frames. ], batch size: 116, lr: 7.76e-03, grad_scale: 8.0 2023-04-29 00:17:04,155 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-29 00:17:48,339 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83196.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:17:49,339 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=83197.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 00:17:55,081 INFO [train.py:904] (6/8) Epoch 9, batch 2000, loss[loss=0.2702, simple_loss=0.3448, pruned_loss=0.09784, over 15355.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2754, pruned_loss=0.05657, over 3316154.04 frames. ], batch size: 190, lr: 7.75e-03, grad_scale: 8.0 2023-04-29 00:18:05,280 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7663, 3.9865, 3.1001, 2.2595, 2.7358, 2.3256, 3.9799, 3.7079], device='cuda:6'), covar=tensor([0.2145, 0.0535, 0.1208, 0.2165, 0.2367, 0.1662, 0.0437, 0.0988], device='cuda:6'), in_proj_covar=tensor([0.0294, 0.0256, 0.0281, 0.0269, 0.0280, 0.0217, 0.0263, 0.0289], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:18:17,511 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.705e+02 2.480e+02 2.857e+02 3.443e+02 6.900e+02, threshold=5.715e+02, percent-clipped=1.0 2023-04-29 00:18:26,837 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83225.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:19:04,142 INFO [train.py:904] (6/8) Epoch 9, batch 2050, loss[loss=0.1601, simple_loss=0.2512, pruned_loss=0.03451, over 17209.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2757, pruned_loss=0.0567, over 3316241.21 frames. ], batch size: 46, lr: 7.75e-03, grad_scale: 8.0 2023-04-29 00:19:11,370 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7318, 5.0790, 4.8173, 4.8741, 4.5785, 4.5259, 4.5690, 5.1251], device='cuda:6'), covar=tensor([0.0965, 0.0897, 0.0999, 0.0598, 0.0821, 0.1045, 0.0966, 0.0867], device='cuda:6'), in_proj_covar=tensor([0.0516, 0.0656, 0.0547, 0.0448, 0.0407, 0.0421, 0.0539, 0.0485], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:19:12,763 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83257.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:19:16,228 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7251, 4.2612, 3.2479, 2.2201, 2.8555, 2.3203, 4.5016, 3.9302], device='cuda:6'), covar=tensor([0.2602, 0.0636, 0.1379, 0.2340, 0.2377, 0.1784, 0.0362, 0.0935], device='cuda:6'), in_proj_covar=tensor([0.0294, 0.0256, 0.0280, 0.0269, 0.0281, 0.0217, 0.0263, 0.0289], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:19:33,982 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=83273.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:19:48,945 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83284.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:20:13,196 INFO [train.py:904] (6/8) Epoch 9, batch 2100, loss[loss=0.1803, simple_loss=0.2732, pruned_loss=0.04373, over 17255.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2761, pruned_loss=0.05633, over 3318355.01 frames. ], batch size: 52, lr: 7.75e-03, grad_scale: 8.0 2023-04-29 00:20:35,023 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.834e+02 2.645e+02 3.321e+02 4.089e+02 1.048e+03, threshold=6.642e+02, percent-clipped=7.0 2023-04-29 00:20:37,110 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83319.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:21:11,819 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83345.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:21:20,735 INFO [train.py:904] (6/8) Epoch 9, batch 2150, loss[loss=0.2026, simple_loss=0.2716, pruned_loss=0.06687, over 16756.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2761, pruned_loss=0.05631, over 3324445.87 frames. ], batch size: 83, lr: 7.75e-03, grad_scale: 8.0 2023-04-29 00:22:26,667 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0528, 2.5579, 2.6241, 1.8655, 2.7963, 2.7797, 2.3775, 2.3985], device='cuda:6'), covar=tensor([0.0655, 0.0194, 0.0219, 0.0911, 0.0086, 0.0171, 0.0435, 0.0409], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0098, 0.0087, 0.0142, 0.0070, 0.0098, 0.0121, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 00:22:31,367 INFO [train.py:904] (6/8) Epoch 9, batch 2200, loss[loss=0.2663, simple_loss=0.3215, pruned_loss=0.1056, over 12066.00 frames. ], tot_loss[loss=0.1953, simple_loss=0.2765, pruned_loss=0.05703, over 3313460.19 frames. ], batch size: 247, lr: 7.74e-03, grad_scale: 8.0 2023-04-29 00:22:54,056 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.989e+02 2.797e+02 3.398e+02 4.283e+02 9.169e+02, threshold=6.797e+02, percent-clipped=3.0 2023-04-29 00:23:21,323 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4351, 5.3006, 5.3112, 4.8830, 4.8418, 5.3268, 5.2362, 4.9056], device='cuda:6'), covar=tensor([0.0537, 0.0398, 0.0210, 0.0226, 0.0974, 0.0361, 0.0240, 0.0639], device='cuda:6'), in_proj_covar=tensor([0.0245, 0.0294, 0.0284, 0.0258, 0.0312, 0.0294, 0.0196, 0.0329], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 00:23:41,045 INFO [train.py:904] (6/8) Epoch 9, batch 2250, loss[loss=0.1999, simple_loss=0.276, pruned_loss=0.06188, over 16754.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.2769, pruned_loss=0.05734, over 3319091.34 frames. ], batch size: 83, lr: 7.74e-03, grad_scale: 8.0 2023-04-29 00:24:44,134 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8496, 4.1446, 3.9352, 4.0179, 3.6095, 3.7877, 3.8632, 4.1004], device='cuda:6'), covar=tensor([0.1014, 0.0909, 0.1039, 0.0622, 0.0793, 0.1495, 0.0758, 0.0996], device='cuda:6'), in_proj_covar=tensor([0.0506, 0.0649, 0.0538, 0.0444, 0.0400, 0.0411, 0.0534, 0.0480], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:24:49,198 INFO [train.py:904] (6/8) Epoch 9, batch 2300, loss[loss=0.1746, simple_loss=0.2532, pruned_loss=0.04796, over 15871.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2778, pruned_loss=0.05808, over 3324034.82 frames. ], batch size: 35, lr: 7.74e-03, grad_scale: 8.0 2023-04-29 00:25:11,715 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 00:25:12,015 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.674e+02 3.137e+02 4.026e+02 1.366e+03, threshold=6.274e+02, percent-clipped=5.0 2023-04-29 00:25:51,394 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0062, 4.3276, 3.2961, 2.3747, 2.9362, 2.3993, 4.6152, 4.0560], device='cuda:6'), covar=tensor([0.2136, 0.0538, 0.1307, 0.2125, 0.2384, 0.1667, 0.0366, 0.0817], device='cuda:6'), in_proj_covar=tensor([0.0293, 0.0256, 0.0280, 0.0270, 0.0283, 0.0216, 0.0263, 0.0289], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:25:59,039 INFO [train.py:904] (6/8) Epoch 9, batch 2350, loss[loss=0.2055, simple_loss=0.2892, pruned_loss=0.06094, over 16511.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2779, pruned_loss=0.05824, over 3316142.59 frames. ], batch size: 68, lr: 7.74e-03, grad_scale: 8.0 2023-04-29 00:25:59,325 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83552.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:26:34,416 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.75 vs. limit=5.0 2023-04-29 00:26:37,500 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0032, 4.1687, 2.4278, 4.6814, 2.9410, 4.6109, 2.3759, 3.3366], device='cuda:6'), covar=tensor([0.0189, 0.0253, 0.1337, 0.0233, 0.0777, 0.0373, 0.1427, 0.0587], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0164, 0.0182, 0.0117, 0.0164, 0.0204, 0.0191, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 00:27:06,421 INFO [train.py:904] (6/8) Epoch 9, batch 2400, loss[loss=0.1678, simple_loss=0.261, pruned_loss=0.03729, over 17134.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2782, pruned_loss=0.05827, over 3325507.53 frames. ], batch size: 46, lr: 7.73e-03, grad_scale: 8.0 2023-04-29 00:27:29,672 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.715e+02 2.500e+02 2.959e+02 3.579e+02 6.970e+02, threshold=5.919e+02, percent-clipped=2.0 2023-04-29 00:27:31,344 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83619.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:27:59,236 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83640.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:28:14,955 INFO [train.py:904] (6/8) Epoch 9, batch 2450, loss[loss=0.2475, simple_loss=0.3174, pruned_loss=0.08885, over 15540.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2788, pruned_loss=0.05791, over 3316689.82 frames. ], batch size: 190, lr: 7.73e-03, grad_scale: 8.0 2023-04-29 00:28:26,680 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4238, 4.3540, 4.8142, 4.7791, 4.8398, 4.4899, 4.5204, 4.2889], device='cuda:6'), covar=tensor([0.0287, 0.0484, 0.0347, 0.0434, 0.0409, 0.0311, 0.0747, 0.0419], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0319, 0.0322, 0.0307, 0.0364, 0.0341, 0.0442, 0.0270], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 00:28:35,101 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=83667.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:29:22,511 INFO [train.py:904] (6/8) Epoch 9, batch 2500, loss[loss=0.2299, simple_loss=0.3009, pruned_loss=0.07948, over 16799.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.278, pruned_loss=0.05716, over 3325394.41 frames. ], batch size: 124, lr: 7.73e-03, grad_scale: 8.0 2023-04-29 00:29:44,587 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.720e+02 2.505e+02 2.999e+02 3.639e+02 7.354e+02, threshold=5.999e+02, percent-clipped=3.0 2023-04-29 00:30:28,782 INFO [train.py:904] (6/8) Epoch 9, batch 2550, loss[loss=0.148, simple_loss=0.2366, pruned_loss=0.02966, over 16833.00 frames. ], tot_loss[loss=0.1963, simple_loss=0.2779, pruned_loss=0.05737, over 3321613.75 frames. ], batch size: 42, lr: 7.73e-03, grad_scale: 8.0 2023-04-29 00:30:34,684 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8165, 2.4459, 1.8314, 2.1401, 2.8689, 2.6208, 3.0304, 2.9888], device='cuda:6'), covar=tensor([0.0094, 0.0218, 0.0299, 0.0275, 0.0119, 0.0203, 0.0113, 0.0123], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0192, 0.0187, 0.0192, 0.0190, 0.0194, 0.0198, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:30:41,786 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83760.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:31:13,047 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-29 00:31:38,615 INFO [train.py:904] (6/8) Epoch 9, batch 2600, loss[loss=0.1987, simple_loss=0.2729, pruned_loss=0.06228, over 16531.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2782, pruned_loss=0.05755, over 3326595.02 frames. ], batch size: 75, lr: 7.73e-03, grad_scale: 8.0 2023-04-29 00:31:59,379 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.494e+02 3.064e+02 3.833e+02 6.863e+02, threshold=6.129e+02, percent-clipped=2.0 2023-04-29 00:32:04,237 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83821.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:32:45,547 INFO [train.py:904] (6/8) Epoch 9, batch 2650, loss[loss=0.1945, simple_loss=0.2822, pruned_loss=0.05336, over 16767.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2788, pruned_loss=0.05707, over 3335255.51 frames. ], batch size: 102, lr: 7.72e-03, grad_scale: 8.0 2023-04-29 00:32:45,851 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83852.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:33:24,533 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6791, 2.5427, 2.0663, 2.2693, 2.9349, 2.6701, 3.4999, 3.2040], device='cuda:6'), covar=tensor([0.0058, 0.0282, 0.0333, 0.0304, 0.0178, 0.0248, 0.0141, 0.0159], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0194, 0.0189, 0.0192, 0.0191, 0.0195, 0.0200, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:33:51,931 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=83900.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:33:53,984 INFO [train.py:904] (6/8) Epoch 9, batch 2700, loss[loss=0.18, simple_loss=0.2696, pruned_loss=0.04518, over 17137.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.2777, pruned_loss=0.05585, over 3342312.14 frames. ], batch size: 48, lr: 7.72e-03, grad_scale: 8.0 2023-04-29 00:34:17,453 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.501e+02 2.856e+02 3.411e+02 7.585e+02, threshold=5.711e+02, percent-clipped=1.0 2023-04-29 00:34:47,158 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83940.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:34:58,151 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83949.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:35:02,459 INFO [train.py:904] (6/8) Epoch 9, batch 2750, loss[loss=0.2013, simple_loss=0.278, pruned_loss=0.06227, over 16849.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.2781, pruned_loss=0.05562, over 3340822.26 frames. ], batch size: 90, lr: 7.72e-03, grad_scale: 8.0 2023-04-29 00:35:46,157 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8140, 2.1150, 2.2343, 4.7055, 2.0100, 2.7631, 2.3310, 2.4385], device='cuda:6'), covar=tensor([0.0771, 0.3284, 0.2017, 0.0292, 0.3617, 0.2143, 0.2773, 0.3140], device='cuda:6'), in_proj_covar=tensor([0.0355, 0.0376, 0.0316, 0.0325, 0.0400, 0.0427, 0.0335, 0.0445], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:35:48,389 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83985.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:35:51,691 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=83988.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:36:15,269 INFO [train.py:904] (6/8) Epoch 9, batch 2800, loss[loss=0.1934, simple_loss=0.2764, pruned_loss=0.05519, over 17007.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2778, pruned_loss=0.05547, over 3337977.50 frames. ], batch size: 55, lr: 7.72e-03, grad_scale: 8.0 2023-04-29 00:36:25,601 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84010.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 00:36:36,295 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.816e+02 2.751e+02 3.290e+02 3.928e+02 7.223e+02, threshold=6.581e+02, percent-clipped=1.0 2023-04-29 00:37:16,070 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84046.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:37:23,280 INFO [train.py:904] (6/8) Epoch 9, batch 2850, loss[loss=0.1728, simple_loss=0.2531, pruned_loss=0.04625, over 16819.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2769, pruned_loss=0.05528, over 3337743.40 frames. ], batch size: 42, lr: 7.71e-03, grad_scale: 8.0 2023-04-29 00:37:36,005 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9065, 4.1676, 2.3400, 4.6350, 3.0291, 4.5240, 2.4439, 3.1744], device='cuda:6'), covar=tensor([0.0204, 0.0260, 0.1341, 0.0126, 0.0647, 0.0420, 0.1264, 0.0621], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0166, 0.0186, 0.0121, 0.0165, 0.0209, 0.0192, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 00:38:25,942 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84097.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:38:32,592 INFO [train.py:904] (6/8) Epoch 9, batch 2900, loss[loss=0.1935, simple_loss=0.2821, pruned_loss=0.05243, over 16757.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2754, pruned_loss=0.05552, over 3329883.10 frames. ], batch size: 57, lr: 7.71e-03, grad_scale: 8.0 2023-04-29 00:38:52,445 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84116.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:38:54,442 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.515e+02 3.057e+02 3.611e+02 6.139e+02, threshold=6.114e+02, percent-clipped=0.0 2023-04-29 00:39:30,628 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 00:39:43,177 INFO [train.py:904] (6/8) Epoch 9, batch 2950, loss[loss=0.2008, simple_loss=0.2851, pruned_loss=0.0582, over 17109.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.275, pruned_loss=0.05608, over 3323676.04 frames. ], batch size: 49, lr: 7.71e-03, grad_scale: 8.0 2023-04-29 00:39:51,225 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84158.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:40:21,143 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-04-29 00:40:48,059 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84199.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:40:50,439 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1128, 2.5384, 2.6996, 4.8989, 2.4312, 3.1682, 2.6674, 2.9302], device='cuda:6'), covar=tensor([0.0690, 0.2799, 0.1691, 0.0298, 0.3031, 0.1733, 0.2375, 0.2606], device='cuda:6'), in_proj_covar=tensor([0.0357, 0.0378, 0.0316, 0.0326, 0.0400, 0.0428, 0.0337, 0.0447], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:40:52,809 INFO [train.py:904] (6/8) Epoch 9, batch 3000, loss[loss=0.1816, simple_loss=0.2563, pruned_loss=0.05346, over 16841.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2753, pruned_loss=0.05661, over 3330005.27 frames. ], batch size: 102, lr: 7.71e-03, grad_scale: 8.0 2023-04-29 00:40:52,809 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 00:41:02,059 INFO [train.py:938] (6/8) Epoch 9, validation: loss=0.1444, simple_loss=0.2507, pruned_loss=0.019, over 944034.00 frames. 2023-04-29 00:41:02,060 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 00:41:23,148 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.086e+02 2.665e+02 3.010e+02 4.048e+02 8.692e+02, threshold=6.019e+02, percent-clipped=1.0 2023-04-29 00:41:47,923 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84236.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:42:09,946 INFO [train.py:904] (6/8) Epoch 9, batch 3050, loss[loss=0.1961, simple_loss=0.2684, pruned_loss=0.06192, over 16763.00 frames. ], tot_loss[loss=0.1946, simple_loss=0.2754, pruned_loss=0.05691, over 3321919.53 frames. ], batch size: 96, lr: 7.70e-03, grad_scale: 8.0 2023-04-29 00:42:20,940 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84260.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:42:50,452 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6466, 5.9931, 5.7030, 5.7715, 5.3325, 5.0454, 5.4078, 6.1381], device='cuda:6'), covar=tensor([0.1002, 0.0797, 0.0910, 0.0535, 0.0811, 0.0644, 0.0808, 0.0699], device='cuda:6'), in_proj_covar=tensor([0.0524, 0.0664, 0.0555, 0.0452, 0.0410, 0.0421, 0.0547, 0.0491], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:43:11,786 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84297.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:43:17,163 INFO [train.py:904] (6/8) Epoch 9, batch 3100, loss[loss=0.2237, simple_loss=0.3117, pruned_loss=0.06785, over 16672.00 frames. ], tot_loss[loss=0.1946, simple_loss=0.2752, pruned_loss=0.057, over 3317885.52 frames. ], batch size: 62, lr: 7.70e-03, grad_scale: 8.0 2023-04-29 00:43:22,114 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84305.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 00:43:39,988 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.846e+02 2.519e+02 3.109e+02 3.888e+02 1.112e+03, threshold=6.217e+02, percent-clipped=5.0 2023-04-29 00:43:42,078 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-29 00:44:06,167 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84336.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:44:12,264 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84341.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:44:28,370 INFO [train.py:904] (6/8) Epoch 9, batch 3150, loss[loss=0.1807, simple_loss=0.2592, pruned_loss=0.05109, over 16990.00 frames. ], tot_loss[loss=0.194, simple_loss=0.274, pruned_loss=0.05704, over 3319938.77 frames. ], batch size: 41, lr: 7.70e-03, grad_scale: 8.0 2023-04-29 00:45:30,293 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84397.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:45:36,485 INFO [train.py:904] (6/8) Epoch 9, batch 3200, loss[loss=0.2114, simple_loss=0.2749, pruned_loss=0.07393, over 16733.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2733, pruned_loss=0.05664, over 3318116.48 frames. ], batch size: 134, lr: 7.70e-03, grad_scale: 8.0 2023-04-29 00:45:50,043 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7552, 1.5861, 2.3299, 2.6184, 2.6326, 2.6685, 1.7158, 2.7702], device='cuda:6'), covar=tensor([0.0090, 0.0299, 0.0181, 0.0159, 0.0142, 0.0132, 0.0294, 0.0081], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0164, 0.0151, 0.0154, 0.0157, 0.0115, 0.0162, 0.0107], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 00:45:56,356 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84416.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:45:59,092 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.715e+02 2.601e+02 3.159e+02 4.189e+02 1.012e+03, threshold=6.318e+02, percent-clipped=3.0 2023-04-29 00:46:32,986 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2023-04-29 00:46:45,891 INFO [train.py:904] (6/8) Epoch 9, batch 3250, loss[loss=0.2225, simple_loss=0.301, pruned_loss=0.072, over 16495.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2743, pruned_loss=0.05722, over 3312515.55 frames. ], batch size: 68, lr: 7.70e-03, grad_scale: 8.0 2023-04-29 00:46:47,462 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84453.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:47:03,267 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=84464.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:47:26,511 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1477, 5.0659, 4.9407, 4.6191, 4.5410, 4.9907, 5.0079, 4.6612], device='cuda:6'), covar=tensor([0.0489, 0.0397, 0.0251, 0.0248, 0.1008, 0.0365, 0.0277, 0.0606], device='cuda:6'), in_proj_covar=tensor([0.0247, 0.0301, 0.0290, 0.0263, 0.0316, 0.0300, 0.0199, 0.0335], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 00:47:55,748 INFO [train.py:904] (6/8) Epoch 9, batch 3300, loss[loss=0.2321, simple_loss=0.296, pruned_loss=0.08415, over 16842.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2763, pruned_loss=0.05834, over 3310873.70 frames. ], batch size: 116, lr: 7.69e-03, grad_scale: 8.0 2023-04-29 00:48:18,092 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 2.660e+02 3.024e+02 3.740e+02 8.260e+02, threshold=6.049e+02, percent-clipped=1.0 2023-04-29 00:48:27,966 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84525.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:49:04,553 INFO [train.py:904] (6/8) Epoch 9, batch 3350, loss[loss=0.1694, simple_loss=0.268, pruned_loss=0.03535, over 17253.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.277, pruned_loss=0.05809, over 3305477.68 frames. ], batch size: 52, lr: 7.69e-03, grad_scale: 8.0 2023-04-29 00:49:10,236 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84555.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:49:53,249 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84586.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:50:01,514 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84592.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:50:16,103 INFO [train.py:904] (6/8) Epoch 9, batch 3400, loss[loss=0.1736, simple_loss=0.2591, pruned_loss=0.04399, over 17154.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.2764, pruned_loss=0.05724, over 3300821.05 frames. ], batch size: 46, lr: 7.69e-03, grad_scale: 8.0 2023-04-29 00:50:21,094 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84605.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:50:38,674 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.593e+02 2.352e+02 2.993e+02 3.807e+02 8.102e+02, threshold=5.985e+02, percent-clipped=2.0 2023-04-29 00:50:59,410 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84633.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:51:10,730 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84641.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:51:14,530 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 00:51:21,528 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2413, 4.6827, 3.6049, 2.7397, 3.3874, 2.7182, 4.9029, 4.2545], device='cuda:6'), covar=tensor([0.1999, 0.0467, 0.1236, 0.1846, 0.2102, 0.1474, 0.0296, 0.0786], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0255, 0.0277, 0.0267, 0.0280, 0.0215, 0.0262, 0.0290], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:51:24,588 INFO [train.py:904] (6/8) Epoch 9, batch 3450, loss[loss=0.193, simple_loss=0.2859, pruned_loss=0.05008, over 16756.00 frames. ], tot_loss[loss=0.194, simple_loss=0.2746, pruned_loss=0.05664, over 3304014.06 frames. ], batch size: 57, lr: 7.69e-03, grad_scale: 8.0 2023-04-29 00:51:26,679 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=84653.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:52:16,067 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=84689.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:52:20,950 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84692.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:52:24,235 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84694.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:52:32,589 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0123, 4.9574, 4.7979, 4.4869, 4.4403, 4.9194, 4.8752, 4.5941], device='cuda:6'), covar=tensor([0.0499, 0.0395, 0.0300, 0.0270, 0.1051, 0.0330, 0.0288, 0.0601], device='cuda:6'), in_proj_covar=tensor([0.0250, 0.0306, 0.0296, 0.0267, 0.0322, 0.0305, 0.0202, 0.0342], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 00:52:35,101 INFO [train.py:904] (6/8) Epoch 9, batch 3500, loss[loss=0.1681, simple_loss=0.2654, pruned_loss=0.03544, over 17270.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2731, pruned_loss=0.05613, over 3306666.54 frames. ], batch size: 52, lr: 7.68e-03, grad_scale: 8.0 2023-04-29 00:52:36,746 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9131, 3.3713, 3.0584, 1.8572, 2.7398, 2.2210, 3.4874, 3.3715], device='cuda:6'), covar=tensor([0.0240, 0.0652, 0.0634, 0.1705, 0.0786, 0.0911, 0.0538, 0.0718], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0145, 0.0157, 0.0142, 0.0136, 0.0125, 0.0137, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 00:52:56,975 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.766e+02 2.287e+02 2.792e+02 3.603e+02 8.093e+02, threshold=5.583e+02, percent-clipped=2.0 2023-04-29 00:53:14,144 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-29 00:53:44,898 INFO [train.py:904] (6/8) Epoch 9, batch 3550, loss[loss=0.2394, simple_loss=0.3012, pruned_loss=0.08874, over 11979.00 frames. ], tot_loss[loss=0.1921, simple_loss=0.2725, pruned_loss=0.05581, over 3298277.86 frames. ], batch size: 248, lr: 7.68e-03, grad_scale: 8.0 2023-04-29 00:53:47,047 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84753.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:54:31,103 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3576, 5.0795, 5.3114, 5.4991, 5.6655, 4.9469, 5.6099, 5.5891], device='cuda:6'), covar=tensor([0.1233, 0.0978, 0.1392, 0.0601, 0.0477, 0.0624, 0.0373, 0.0448], device='cuda:6'), in_proj_covar=tensor([0.0536, 0.0656, 0.0824, 0.0673, 0.0506, 0.0508, 0.0522, 0.0582], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 00:54:53,033 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84800.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:54:54,908 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=84801.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:54:55,836 INFO [train.py:904] (6/8) Epoch 9, batch 3600, loss[loss=0.1625, simple_loss=0.2477, pruned_loss=0.0386, over 16986.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2707, pruned_loss=0.05484, over 3303836.27 frames. ], batch size: 41, lr: 7.68e-03, grad_scale: 8.0 2023-04-29 00:55:17,924 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.706e+02 2.544e+02 2.999e+02 3.772e+02 8.043e+02, threshold=5.998e+02, percent-clipped=5.0 2023-04-29 00:56:07,140 INFO [train.py:904] (6/8) Epoch 9, batch 3650, loss[loss=0.1821, simple_loss=0.2467, pruned_loss=0.05872, over 16487.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.2699, pruned_loss=0.05553, over 3295873.29 frames. ], batch size: 75, lr: 7.68e-03, grad_scale: 8.0 2023-04-29 00:56:12,843 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84855.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:56:21,103 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84861.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:56:50,579 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84881.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:57:07,345 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84892.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:57:14,839 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.64 vs. limit=5.0 2023-04-29 00:57:21,439 INFO [train.py:904] (6/8) Epoch 9, batch 3700, loss[loss=0.1741, simple_loss=0.2652, pruned_loss=0.04147, over 16846.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2688, pruned_loss=0.05683, over 3274545.60 frames. ], batch size: 42, lr: 7.68e-03, grad_scale: 8.0 2023-04-29 00:57:23,529 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=84903.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:57:46,406 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.577e+02 2.579e+02 2.947e+02 3.773e+02 6.988e+02, threshold=5.894e+02, percent-clipped=1.0 2023-04-29 00:58:19,041 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=84940.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:58:24,181 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5068, 3.5417, 3.6581, 2.0277, 3.8389, 3.8507, 3.0688, 2.7770], device='cuda:6'), covar=tensor([0.0681, 0.0150, 0.0133, 0.0979, 0.0057, 0.0106, 0.0340, 0.0384], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0098, 0.0087, 0.0140, 0.0071, 0.0101, 0.0122, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 00:58:34,977 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84951.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:58:35,654 INFO [train.py:904] (6/8) Epoch 9, batch 3750, loss[loss=0.1862, simple_loss=0.2549, pruned_loss=0.0587, over 16799.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2692, pruned_loss=0.05803, over 3277831.06 frames. ], batch size: 90, lr: 7.67e-03, grad_scale: 8.0 2023-04-29 00:59:28,398 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84989.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:59:31,091 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9386, 4.2945, 1.9174, 4.6390, 2.7639, 4.7863, 2.2223, 2.9271], device='cuda:6'), covar=tensor([0.0178, 0.0216, 0.1981, 0.0059, 0.0792, 0.0187, 0.1555, 0.0753], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0165, 0.0184, 0.0121, 0.0165, 0.0208, 0.0193, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 00:59:34,077 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84992.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:59:47,558 INFO [train.py:904] (6/8) Epoch 9, batch 3800, loss[loss=0.1813, simple_loss=0.2584, pruned_loss=0.05213, over 16757.00 frames. ], tot_loss[loss=0.1942, simple_loss=0.2698, pruned_loss=0.05931, over 3270495.59 frames. ], batch size: 89, lr: 7.67e-03, grad_scale: 8.0 2023-04-29 01:00:04,169 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85012.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:00:11,015 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.825e+02 2.585e+02 2.931e+02 3.484e+02 5.921e+02, threshold=5.862e+02, percent-clipped=1.0 2023-04-29 01:00:43,934 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85040.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:01:00,353 INFO [train.py:904] (6/8) Epoch 9, batch 3850, loss[loss=0.1881, simple_loss=0.2622, pruned_loss=0.05699, over 16567.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2697, pruned_loss=0.05961, over 3280875.13 frames. ], batch size: 35, lr: 7.67e-03, grad_scale: 8.0 2023-04-29 01:01:10,435 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7051, 3.7798, 4.0366, 2.0107, 4.2418, 4.2720, 3.2925, 3.1445], device='cuda:6'), covar=tensor([0.0718, 0.0179, 0.0142, 0.1215, 0.0057, 0.0092, 0.0305, 0.0413], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0097, 0.0086, 0.0140, 0.0070, 0.0101, 0.0121, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 01:01:34,392 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1200, 3.2101, 1.8638, 3.2720, 2.4369, 3.3147, 1.9851, 2.6059], device='cuda:6'), covar=tensor([0.0215, 0.0367, 0.1534, 0.0191, 0.0735, 0.0456, 0.1340, 0.0598], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0165, 0.0184, 0.0120, 0.0165, 0.0207, 0.0192, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 01:02:13,000 INFO [train.py:904] (6/8) Epoch 9, batch 3900, loss[loss=0.198, simple_loss=0.2632, pruned_loss=0.06635, over 16568.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2711, pruned_loss=0.06093, over 3250497.82 frames. ], batch size: 76, lr: 7.67e-03, grad_scale: 8.0 2023-04-29 01:02:21,759 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85107.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:02:32,604 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85115.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:02:36,232 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.667e+02 2.478e+02 2.844e+02 3.462e+02 6.399e+02, threshold=5.687e+02, percent-clipped=3.0 2023-04-29 01:03:16,824 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.78 vs. limit=5.0 2023-04-29 01:03:24,358 INFO [train.py:904] (6/8) Epoch 9, batch 3950, loss[loss=0.188, simple_loss=0.2583, pruned_loss=0.05878, over 16827.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2701, pruned_loss=0.06095, over 3262692.08 frames. ], batch size: 83, lr: 7.66e-03, grad_scale: 8.0 2023-04-29 01:03:27,580 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3171, 1.9776, 2.6217, 3.1903, 3.0010, 3.6356, 1.8845, 3.4640], device='cuda:6'), covar=tensor([0.0104, 0.0300, 0.0178, 0.0156, 0.0157, 0.0068, 0.0327, 0.0066], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0164, 0.0151, 0.0153, 0.0158, 0.0116, 0.0162, 0.0106], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 01:03:32,223 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85156.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:03:50,058 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85168.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 01:04:00,697 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85176.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:04:07,222 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85181.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:04:37,100 INFO [train.py:904] (6/8) Epoch 9, batch 4000, loss[loss=0.2118, simple_loss=0.283, pruned_loss=0.07027, over 17204.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.27, pruned_loss=0.06139, over 3262828.91 frames. ], batch size: 44, lr: 7.66e-03, grad_scale: 16.0 2023-04-29 01:05:00,771 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.397e+02 2.490e+02 3.071e+02 3.606e+02 5.108e+02, threshold=6.141e+02, percent-clipped=0.0 2023-04-29 01:05:17,721 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85229.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:05:51,192 INFO [train.py:904] (6/8) Epoch 9, batch 4050, loss[loss=0.2307, simple_loss=0.2993, pruned_loss=0.08109, over 12103.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2698, pruned_loss=0.05994, over 3254051.98 frames. ], batch size: 246, lr: 7.66e-03, grad_scale: 16.0 2023-04-29 01:06:10,192 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2101, 3.5555, 3.4633, 1.9190, 3.0237, 2.2320, 3.6830, 3.6606], device='cuda:6'), covar=tensor([0.0222, 0.0540, 0.0510, 0.1683, 0.0681, 0.0867, 0.0478, 0.0720], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0145, 0.0157, 0.0142, 0.0136, 0.0126, 0.0137, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 01:06:45,551 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85289.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:07:04,621 INFO [train.py:904] (6/8) Epoch 9, batch 4100, loss[loss=0.198, simple_loss=0.2774, pruned_loss=0.05924, over 17067.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2711, pruned_loss=0.05899, over 3255989.40 frames. ], batch size: 55, lr: 7.66e-03, grad_scale: 16.0 2023-04-29 01:07:10,058 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6905, 2.7238, 2.0438, 2.4146, 3.0773, 2.7921, 3.4697, 3.2952], device='cuda:6'), covar=tensor([0.0036, 0.0226, 0.0332, 0.0288, 0.0136, 0.0213, 0.0103, 0.0141], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0191, 0.0188, 0.0187, 0.0187, 0.0190, 0.0195, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 01:07:12,414 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85307.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:07:28,042 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.051e+02 2.369e+02 2.790e+02 6.834e+02, threshold=4.737e+02, percent-clipped=1.0 2023-04-29 01:07:34,834 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-29 01:07:57,982 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85337.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:07:59,350 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2352, 1.5334, 1.9090, 2.1767, 2.2665, 2.5341, 1.6115, 2.3428], device='cuda:6'), covar=tensor([0.0148, 0.0301, 0.0180, 0.0193, 0.0169, 0.0111, 0.0277, 0.0078], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0165, 0.0152, 0.0153, 0.0158, 0.0117, 0.0164, 0.0106], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 01:08:20,246 INFO [train.py:904] (6/8) Epoch 9, batch 4150, loss[loss=0.2058, simple_loss=0.2961, pruned_loss=0.05775, over 16223.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2788, pruned_loss=0.06208, over 3221148.60 frames. ], batch size: 165, lr: 7.66e-03, grad_scale: 16.0 2023-04-29 01:08:38,537 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0792, 3.9773, 4.1295, 4.2703, 4.3701, 3.9633, 4.3300, 4.3877], device='cuda:6'), covar=tensor([0.1196, 0.0797, 0.1125, 0.0516, 0.0411, 0.1222, 0.0505, 0.0453], device='cuda:6'), in_proj_covar=tensor([0.0505, 0.0612, 0.0764, 0.0624, 0.0473, 0.0478, 0.0490, 0.0546], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 01:08:57,312 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8843, 2.1858, 1.6866, 1.9214, 2.5500, 2.2760, 2.8173, 2.8371], device='cuda:6'), covar=tensor([0.0076, 0.0290, 0.0373, 0.0316, 0.0159, 0.0267, 0.0135, 0.0157], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0192, 0.0189, 0.0188, 0.0189, 0.0192, 0.0196, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 01:09:37,262 INFO [train.py:904] (6/8) Epoch 9, batch 4200, loss[loss=0.2185, simple_loss=0.3078, pruned_loss=0.06462, over 16507.00 frames. ], tot_loss[loss=0.2075, simple_loss=0.2863, pruned_loss=0.06436, over 3184666.55 frames. ], batch size: 75, lr: 7.65e-03, grad_scale: 16.0 2023-04-29 01:10:02,491 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.568e+02 2.639e+02 3.175e+02 3.973e+02 9.081e+02, threshold=6.349e+02, percent-clipped=14.0 2023-04-29 01:10:17,790 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85428.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:10:26,795 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0104, 3.2181, 2.9453, 5.2482, 4.1026, 4.5691, 1.5956, 3.4252], device='cuda:6'), covar=tensor([0.1100, 0.0573, 0.0971, 0.0091, 0.0273, 0.0353, 0.1452, 0.0676], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0154, 0.0175, 0.0128, 0.0202, 0.0208, 0.0175, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 01:10:38,315 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0239, 2.4015, 2.2823, 2.9079, 2.2753, 3.2899, 1.6642, 2.7337], device='cuda:6'), covar=tensor([0.0897, 0.0514, 0.0843, 0.0117, 0.0145, 0.0372, 0.1237, 0.0570], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0154, 0.0175, 0.0128, 0.0202, 0.0208, 0.0175, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 01:10:52,278 INFO [train.py:904] (6/8) Epoch 9, batch 4250, loss[loss=0.1988, simple_loss=0.2879, pruned_loss=0.05484, over 16928.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.29, pruned_loss=0.06466, over 3166505.34 frames. ], batch size: 109, lr: 7.65e-03, grad_scale: 16.0 2023-04-29 01:10:59,126 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85456.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:11:08,336 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85463.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 01:11:20,275 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85471.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:11:39,022 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85483.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:11:48,364 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85489.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:12:07,380 INFO [train.py:904] (6/8) Epoch 9, batch 4300, loss[loss=0.2218, simple_loss=0.3086, pruned_loss=0.06752, over 16499.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2913, pruned_loss=0.06366, over 3173168.07 frames. ], batch size: 146, lr: 7.65e-03, grad_scale: 16.0 2023-04-29 01:12:11,407 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85504.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:12:30,788 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.243e+02 2.412e+02 2.949e+02 3.511e+02 9.601e+02, threshold=5.898e+02, percent-clipped=2.0 2023-04-29 01:12:49,917 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85530.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:13:10,355 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85544.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:13:21,362 INFO [train.py:904] (6/8) Epoch 9, batch 4350, loss[loss=0.2073, simple_loss=0.2989, pruned_loss=0.05782, over 16622.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2943, pruned_loss=0.06408, over 3198795.96 frames. ], batch size: 62, lr: 7.65e-03, grad_scale: 16.0 2023-04-29 01:13:24,352 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1322, 3.9475, 3.9233, 2.2629, 3.4637, 3.8152, 3.5151, 2.0534], device='cuda:6'), covar=tensor([0.0386, 0.0019, 0.0022, 0.0308, 0.0053, 0.0046, 0.0044, 0.0318], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0065, 0.0065, 0.0119, 0.0071, 0.0081, 0.0071, 0.0112], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 01:13:29,626 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9853, 3.8559, 3.7839, 2.0986, 3.3976, 3.7265, 3.4677, 2.0585], device='cuda:6'), covar=tensor([0.0413, 0.0019, 0.0026, 0.0337, 0.0055, 0.0049, 0.0042, 0.0309], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0065, 0.0065, 0.0119, 0.0071, 0.0081, 0.0071, 0.0112], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 01:14:19,060 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85591.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:14:34,737 INFO [train.py:904] (6/8) Epoch 9, batch 4400, loss[loss=0.2225, simple_loss=0.3095, pruned_loss=0.06777, over 16706.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.2965, pruned_loss=0.06529, over 3212222.45 frames. ], batch size: 62, lr: 7.64e-03, grad_scale: 16.0 2023-04-29 01:14:41,572 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85607.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:14:56,974 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.792e+02 2.712e+02 3.095e+02 3.599e+02 6.298e+02, threshold=6.190e+02, percent-clipped=2.0 2023-04-29 01:15:06,820 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85625.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:15:27,622 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8950, 4.8789, 5.3332, 5.2582, 5.3298, 4.8410, 4.9101, 4.4025], device='cuda:6'), covar=tensor([0.0216, 0.0298, 0.0245, 0.0334, 0.0303, 0.0293, 0.0657, 0.0403], device='cuda:6'), in_proj_covar=tensor([0.0297, 0.0300, 0.0307, 0.0296, 0.0349, 0.0324, 0.0424, 0.0259], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 01:15:46,022 INFO [train.py:904] (6/8) Epoch 9, batch 4450, loss[loss=0.2166, simple_loss=0.2956, pruned_loss=0.06879, over 16806.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.2993, pruned_loss=0.06578, over 3211677.21 frames. ], batch size: 39, lr: 7.64e-03, grad_scale: 16.0 2023-04-29 01:15:50,458 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85655.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:16:34,289 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85686.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 01:16:56,347 INFO [train.py:904] (6/8) Epoch 9, batch 4500, loss[loss=0.2306, simple_loss=0.3062, pruned_loss=0.07754, over 15368.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.2995, pruned_loss=0.06636, over 3206735.38 frames. ], batch size: 191, lr: 7.64e-03, grad_scale: 16.0 2023-04-29 01:17:10,864 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-29 01:17:12,498 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1135, 3.5913, 3.4346, 1.9277, 2.9036, 2.3523, 3.2614, 3.5817], device='cuda:6'), covar=tensor([0.0344, 0.0655, 0.0557, 0.1787, 0.0802, 0.0893, 0.0856, 0.0852], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0141, 0.0155, 0.0140, 0.0133, 0.0123, 0.0135, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 01:17:20,314 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 2.131e+02 2.570e+02 3.015e+02 5.229e+02, threshold=5.140e+02, percent-clipped=0.0 2023-04-29 01:17:33,536 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6530, 3.8996, 4.1650, 1.7877, 4.7173, 4.6964, 3.2597, 3.4074], device='cuda:6'), covar=tensor([0.0777, 0.0181, 0.0153, 0.1164, 0.0026, 0.0041, 0.0300, 0.0382], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0095, 0.0084, 0.0136, 0.0067, 0.0095, 0.0117, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-29 01:18:07,069 INFO [train.py:904] (6/8) Epoch 9, batch 4550, loss[loss=0.2378, simple_loss=0.3144, pruned_loss=0.08057, over 15170.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.2997, pruned_loss=0.06656, over 3217805.98 frames. ], batch size: 190, lr: 7.64e-03, grad_scale: 16.0 2023-04-29 01:18:23,718 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85763.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:18:34,917 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85771.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:18:54,469 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85784.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:19:20,632 INFO [train.py:904] (6/8) Epoch 9, batch 4600, loss[loss=0.2018, simple_loss=0.2872, pruned_loss=0.05818, over 17123.00 frames. ], tot_loss[loss=0.2167, simple_loss=0.3001, pruned_loss=0.06669, over 3212451.20 frames. ], batch size: 47, lr: 7.64e-03, grad_scale: 16.0 2023-04-29 01:19:32,652 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85811.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:19:42,422 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.222e+02 2.575e+02 3.036e+02 5.036e+02, threshold=5.150e+02, percent-clipped=0.0 2023-04-29 01:19:44,978 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85819.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:19:52,087 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.25 vs. limit=5.0 2023-04-29 01:20:12,995 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85839.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:20:17,321 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6921, 4.0669, 4.1939, 2.1936, 3.3622, 2.8625, 3.8499, 4.0854], device='cuda:6'), covar=tensor([0.0208, 0.0558, 0.0411, 0.1598, 0.0679, 0.0769, 0.0565, 0.0762], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0140, 0.0155, 0.0140, 0.0133, 0.0123, 0.0134, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 01:20:30,859 INFO [train.py:904] (6/8) Epoch 9, batch 4650, loss[loss=0.2528, simple_loss=0.3177, pruned_loss=0.09397, over 11847.00 frames. ], tot_loss[loss=0.2151, simple_loss=0.2982, pruned_loss=0.06598, over 3224271.10 frames. ], batch size: 248, lr: 7.63e-03, grad_scale: 16.0 2023-04-29 01:20:43,221 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-29 01:21:04,810 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85875.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:21:20,428 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85886.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:21:23,418 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1375, 4.9340, 5.2081, 5.4326, 5.5517, 4.9035, 5.5264, 5.5578], device='cuda:6'), covar=tensor([0.1176, 0.0864, 0.0994, 0.0375, 0.0358, 0.0487, 0.0324, 0.0361], device='cuda:6'), in_proj_covar=tensor([0.0482, 0.0592, 0.0736, 0.0599, 0.0457, 0.0461, 0.0471, 0.0526], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 01:21:32,732 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-29 01:21:42,719 INFO [train.py:904] (6/8) Epoch 9, batch 4700, loss[loss=0.2288, simple_loss=0.3032, pruned_loss=0.07717, over 11806.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2952, pruned_loss=0.06471, over 3214209.79 frames. ], batch size: 247, lr: 7.63e-03, grad_scale: 16.0 2023-04-29 01:22:06,316 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.169e+02 2.505e+02 2.903e+02 6.033e+02, threshold=5.010e+02, percent-clipped=1.0 2023-04-29 01:22:25,267 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85930.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 01:22:33,494 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85936.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:22:42,583 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9118, 4.1609, 3.9500, 4.0390, 3.6942, 3.7325, 3.8241, 4.1010], device='cuda:6'), covar=tensor([0.0868, 0.0813, 0.0867, 0.0562, 0.0688, 0.1440, 0.0767, 0.0977], device='cuda:6'), in_proj_covar=tensor([0.0486, 0.0608, 0.0510, 0.0416, 0.0381, 0.0395, 0.0507, 0.0460], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 01:22:55,504 INFO [train.py:904] (6/8) Epoch 9, batch 4750, loss[loss=0.2091, simple_loss=0.2868, pruned_loss=0.06575, over 11696.00 frames. ], tot_loss[loss=0.2091, simple_loss=0.2917, pruned_loss=0.06328, over 3201129.40 frames. ], batch size: 248, lr: 7.63e-03, grad_scale: 16.0 2023-04-29 01:23:04,348 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85957.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:23:39,897 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85981.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 01:23:49,277 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-29 01:23:50,182 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6658, 3.1397, 3.0529, 1.7820, 2.6376, 2.1200, 3.2151, 3.1751], device='cuda:6'), covar=tensor([0.0265, 0.0675, 0.0603, 0.1736, 0.0807, 0.0887, 0.0617, 0.0747], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0142, 0.0157, 0.0142, 0.0135, 0.0125, 0.0136, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 01:23:55,257 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85991.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 01:24:13,771 INFO [train.py:904] (6/8) Epoch 9, batch 4800, loss[loss=0.1946, simple_loss=0.2731, pruned_loss=0.05802, over 16986.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2882, pruned_loss=0.06123, over 3195132.01 frames. ], batch size: 55, lr: 7.63e-03, grad_scale: 16.0 2023-04-29 01:24:14,285 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86002.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:24:37,221 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.079e+02 2.389e+02 2.964e+02 6.421e+02, threshold=4.777e+02, percent-clipped=3.0 2023-04-29 01:24:38,381 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86018.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:25:20,100 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6950, 2.9499, 2.5261, 4.4875, 3.2631, 4.1525, 1.3744, 3.0487], device='cuda:6'), covar=tensor([0.1286, 0.0561, 0.1050, 0.0094, 0.0214, 0.0321, 0.1519, 0.0704], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0151, 0.0173, 0.0125, 0.0197, 0.0203, 0.0172, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 01:25:28,114 INFO [train.py:904] (6/8) Epoch 9, batch 4850, loss[loss=0.1984, simple_loss=0.276, pruned_loss=0.06036, over 16639.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.289, pruned_loss=0.06074, over 3175975.87 frames. ], batch size: 57, lr: 7.62e-03, grad_scale: 16.0 2023-04-29 01:25:29,151 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86052.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:25:36,414 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1094, 1.4650, 1.7897, 2.0717, 2.2533, 2.3403, 1.6008, 2.1676], device='cuda:6'), covar=tensor([0.0132, 0.0296, 0.0185, 0.0188, 0.0146, 0.0107, 0.0278, 0.0069], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0163, 0.0148, 0.0152, 0.0156, 0.0114, 0.0162, 0.0104], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 01:25:45,118 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86063.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:25:54,712 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86069.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:26:16,502 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86084.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:26:42,269 INFO [train.py:904] (6/8) Epoch 9, batch 4900, loss[loss=0.1756, simple_loss=0.2722, pruned_loss=0.0395, over 16857.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2881, pruned_loss=0.0594, over 3181347.93 frames. ], batch size: 102, lr: 7.62e-03, grad_scale: 16.0 2023-04-29 01:26:59,805 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86113.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:27:05,788 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.491e+02 2.278e+02 2.647e+02 3.027e+02 5.480e+02, threshold=5.294e+02, percent-clipped=1.0 2023-04-29 01:27:23,822 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86130.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:27:26,527 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86132.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:27:37,259 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86139.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:27:55,672 INFO [train.py:904] (6/8) Epoch 9, batch 4950, loss[loss=0.1934, simple_loss=0.2926, pruned_loss=0.0471, over 16864.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2875, pruned_loss=0.05847, over 3199526.75 frames. ], batch size: 96, lr: 7.62e-03, grad_scale: 16.0 2023-04-29 01:28:45,914 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86186.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:28:47,118 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86187.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:29:08,681 INFO [train.py:904] (6/8) Epoch 9, batch 5000, loss[loss=0.2041, simple_loss=0.292, pruned_loss=0.05812, over 16697.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2898, pruned_loss=0.05894, over 3204678.67 frames. ], batch size: 124, lr: 7.62e-03, grad_scale: 16.0 2023-04-29 01:29:32,025 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.770e+02 2.365e+02 2.931e+02 3.414e+02 6.733e+02, threshold=5.862e+02, percent-clipped=2.0 2023-04-29 01:29:50,440 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86231.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:29:54,585 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86234.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:30:11,777 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86246.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:30:19,720 INFO [train.py:904] (6/8) Epoch 9, batch 5050, loss[loss=0.2099, simple_loss=0.3047, pruned_loss=0.05759, over 16863.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2903, pruned_loss=0.05906, over 3214611.80 frames. ], batch size: 116, lr: 7.62e-03, grad_scale: 16.0 2023-04-29 01:30:36,926 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 01:31:01,225 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86281.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:31:07,983 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86286.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 01:31:29,531 INFO [train.py:904] (6/8) Epoch 9, batch 5100, loss[loss=0.175, simple_loss=0.2597, pruned_loss=0.04513, over 17070.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2885, pruned_loss=0.05831, over 3212192.34 frames. ], batch size: 49, lr: 7.61e-03, grad_scale: 16.0 2023-04-29 01:31:37,598 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86307.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:31:43,063 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-29 01:31:45,611 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86313.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:31:52,763 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.623e+02 2.422e+02 2.762e+02 3.223e+02 5.276e+02, threshold=5.525e+02, percent-clipped=0.0 2023-04-29 01:31:55,814 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 01:32:08,055 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:32:41,540 INFO [train.py:904] (6/8) Epoch 9, batch 5150, loss[loss=0.2034, simple_loss=0.2867, pruned_loss=0.06002, over 12161.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2884, pruned_loss=0.05742, over 3196558.25 frames. ], batch size: 248, lr: 7.61e-03, grad_scale: 16.0 2023-04-29 01:32:50,286 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86358.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:33:16,068 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2407, 1.9842, 1.7397, 1.8370, 2.2865, 2.0685, 2.1840, 2.4774], device='cuda:6'), covar=tensor([0.0086, 0.0278, 0.0334, 0.0325, 0.0142, 0.0243, 0.0111, 0.0166], device='cuda:6'), in_proj_covar=tensor([0.0125, 0.0190, 0.0187, 0.0184, 0.0186, 0.0191, 0.0188, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 01:33:54,429 INFO [train.py:904] (6/8) Epoch 9, batch 5200, loss[loss=0.2142, simple_loss=0.2945, pruned_loss=0.06702, over 16380.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2869, pruned_loss=0.05706, over 3196416.54 frames. ], batch size: 68, lr: 7.61e-03, grad_scale: 16.0 2023-04-29 01:34:02,830 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86408.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:34:02,973 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8502, 4.7521, 4.6130, 4.4313, 4.2320, 4.7092, 4.6809, 4.3763], device='cuda:6'), covar=tensor([0.0494, 0.0344, 0.0279, 0.0216, 0.0914, 0.0369, 0.0278, 0.0514], device='cuda:6'), in_proj_covar=tensor([0.0224, 0.0274, 0.0268, 0.0241, 0.0290, 0.0276, 0.0182, 0.0308], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 01:34:17,298 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.533e+02 2.298e+02 2.670e+02 3.115e+02 5.719e+02, threshold=5.340e+02, percent-clipped=1.0 2023-04-29 01:34:27,688 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86425.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:34:50,074 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86440.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:34:53,424 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3588, 1.9643, 2.1645, 3.8461, 1.8591, 2.4076, 2.0526, 2.1438], device='cuda:6'), covar=tensor([0.1026, 0.3475, 0.2013, 0.0543, 0.4240, 0.2491, 0.2917, 0.3155], device='cuda:6'), in_proj_covar=tensor([0.0349, 0.0374, 0.0312, 0.0317, 0.0399, 0.0424, 0.0332, 0.0438], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 01:34:57,838 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86445.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:35:06,327 INFO [train.py:904] (6/8) Epoch 9, batch 5250, loss[loss=0.2087, simple_loss=0.3011, pruned_loss=0.05819, over 16415.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2847, pruned_loss=0.05664, over 3205432.63 frames. ], batch size: 146, lr: 7.61e-03, grad_scale: 16.0 2023-04-29 01:35:51,367 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 01:36:17,581 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86501.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:36:19,005 INFO [train.py:904] (6/8) Epoch 9, batch 5300, loss[loss=0.1885, simple_loss=0.2662, pruned_loss=0.05541, over 16680.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2807, pruned_loss=0.05512, over 3216468.32 frames. ], batch size: 134, lr: 7.60e-03, grad_scale: 16.0 2023-04-29 01:36:25,577 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86506.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 01:36:42,024 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.448e+02 2.767e+02 3.292e+02 5.111e+02, threshold=5.534e+02, percent-clipped=0.0 2023-04-29 01:37:01,831 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86531.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:37:19,706 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9464, 4.2629, 3.4053, 2.6297, 3.3337, 2.8032, 4.7895, 4.1843], device='cuda:6'), covar=tensor([0.2433, 0.0664, 0.1318, 0.1769, 0.1943, 0.1335, 0.0381, 0.0705], device='cuda:6'), in_proj_covar=tensor([0.0293, 0.0252, 0.0276, 0.0269, 0.0276, 0.0213, 0.0262, 0.0278], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 01:37:32,902 INFO [train.py:904] (6/8) Epoch 9, batch 5350, loss[loss=0.1831, simple_loss=0.2797, pruned_loss=0.04323, over 15444.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2794, pruned_loss=0.05435, over 3226184.13 frames. ], batch size: 190, lr: 7.60e-03, grad_scale: 16.0 2023-04-29 01:38:11,779 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86579.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:38:13,855 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86580.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:38:23,799 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86586.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 01:38:45,869 INFO [train.py:904] (6/8) Epoch 9, batch 5400, loss[loss=0.2067, simple_loss=0.2948, pruned_loss=0.05924, over 16755.00 frames. ], tot_loss[loss=0.1953, simple_loss=0.2814, pruned_loss=0.0546, over 3241071.58 frames. ], batch size: 83, lr: 7.60e-03, grad_scale: 16.0 2023-04-29 01:38:46,249 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86602.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:39:01,716 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86613.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:39:09,034 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.152e+02 2.423e+02 2.907e+02 4.710e+02, threshold=4.847e+02, percent-clipped=0.0 2023-04-29 01:39:18,893 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-29 01:39:29,365 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7035, 2.8994, 2.6672, 4.5390, 3.2595, 4.1318, 1.5459, 3.0601], device='cuda:6'), covar=tensor([0.1222, 0.0595, 0.0983, 0.0101, 0.0232, 0.0337, 0.1347, 0.0719], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0153, 0.0175, 0.0126, 0.0198, 0.0204, 0.0173, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 01:39:32,145 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86634.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 01:39:41,849 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86641.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:39:58,142 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7240, 2.9146, 2.5993, 4.5880, 3.3932, 4.1151, 1.4451, 3.0062], device='cuda:6'), covar=tensor([0.1221, 0.0604, 0.1043, 0.0093, 0.0326, 0.0344, 0.1452, 0.0766], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0150, 0.0173, 0.0124, 0.0196, 0.0201, 0.0171, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 01:40:00,485 INFO [train.py:904] (6/8) Epoch 9, batch 5450, loss[loss=0.2523, simple_loss=0.3261, pruned_loss=0.08925, over 16442.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2849, pruned_loss=0.05649, over 3236831.29 frames. ], batch size: 146, lr: 7.60e-03, grad_scale: 16.0 2023-04-29 01:40:10,935 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86658.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:40:15,978 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86661.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:41:18,955 INFO [train.py:904] (6/8) Epoch 9, batch 5500, loss[loss=0.248, simple_loss=0.3211, pruned_loss=0.08747, over 16833.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2942, pruned_loss=0.06283, over 3202176.75 frames. ], batch size: 42, lr: 7.60e-03, grad_scale: 16.0 2023-04-29 01:41:24,418 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86706.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:41:27,693 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86708.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:41:43,592 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.927e+02 3.483e+02 4.212e+02 5.366e+02 1.166e+03, threshold=8.424e+02, percent-clipped=36.0 2023-04-29 01:41:54,531 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86725.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:42:36,877 INFO [train.py:904] (6/8) Epoch 9, batch 5550, loss[loss=0.245, simple_loss=0.3202, pruned_loss=0.08491, over 16937.00 frames. ], tot_loss[loss=0.2202, simple_loss=0.3022, pruned_loss=0.06908, over 3156460.70 frames. ], batch size: 109, lr: 7.59e-03, grad_scale: 16.0 2023-04-29 01:42:43,143 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86756.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:43:09,698 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86773.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:43:45,366 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86796.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:43:54,001 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86801.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 01:43:54,665 INFO [train.py:904] (6/8) Epoch 9, batch 5600, loss[loss=0.3059, simple_loss=0.3569, pruned_loss=0.1275, over 11059.00 frames. ], tot_loss[loss=0.2312, simple_loss=0.3096, pruned_loss=0.0764, over 3071029.54 frames. ], batch size: 246, lr: 7.59e-03, grad_scale: 8.0 2023-04-29 01:44:07,427 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7254, 1.7060, 1.5692, 1.5917, 1.9017, 1.6619, 1.6780, 1.9150], device='cuda:6'), covar=tensor([0.0111, 0.0145, 0.0241, 0.0202, 0.0119, 0.0155, 0.0127, 0.0116], device='cuda:6'), in_proj_covar=tensor([0.0125, 0.0189, 0.0187, 0.0186, 0.0187, 0.0190, 0.0188, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 01:44:14,640 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86814.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:44:21,510 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.583e+02 3.990e+02 4.962e+02 6.279e+02 1.585e+03, threshold=9.923e+02, percent-clipped=6.0 2023-04-29 01:45:17,292 INFO [train.py:904] (6/8) Epoch 9, batch 5650, loss[loss=0.198, simple_loss=0.2859, pruned_loss=0.05501, over 17114.00 frames. ], tot_loss[loss=0.2365, simple_loss=0.3138, pruned_loss=0.07958, over 3067649.94 frames. ], batch size: 49, lr: 7.59e-03, grad_scale: 4.0 2023-04-29 01:45:53,451 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86875.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:45:58,541 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86878.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:46:33,109 INFO [train.py:904] (6/8) Epoch 9, batch 5700, loss[loss=0.2197, simple_loss=0.312, pruned_loss=0.06376, over 16341.00 frames. ], tot_loss[loss=0.2382, simple_loss=0.3151, pruned_loss=0.08068, over 3053203.05 frames. ], batch size: 146, lr: 7.59e-03, grad_scale: 4.0 2023-04-29 01:46:33,474 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86902.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:46:37,244 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8011, 2.7347, 2.6107, 1.9293, 2.5258, 2.7079, 2.5588, 1.8194], device='cuda:6'), covar=tensor([0.0323, 0.0039, 0.0055, 0.0258, 0.0074, 0.0079, 0.0057, 0.0288], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0064, 0.0064, 0.0120, 0.0071, 0.0081, 0.0070, 0.0113], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 01:46:59,880 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.388e+02 3.612e+02 4.435e+02 5.746e+02 9.391e+02, threshold=8.870e+02, percent-clipped=0.0 2023-04-29 01:47:25,188 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86936.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:47:31,021 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86939.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:47:40,057 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2096, 3.2087, 3.0392, 5.2276, 4.2271, 4.6312, 1.8333, 3.5584], device='cuda:6'), covar=tensor([0.1193, 0.0645, 0.1056, 0.0122, 0.0308, 0.0340, 0.1455, 0.0674], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0152, 0.0176, 0.0126, 0.0199, 0.0205, 0.0174, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 01:47:48,290 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86950.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:47:50,854 INFO [train.py:904] (6/8) Epoch 9, batch 5750, loss[loss=0.2398, simple_loss=0.3231, pruned_loss=0.07827, over 16946.00 frames. ], tot_loss[loss=0.2421, simple_loss=0.3187, pruned_loss=0.08271, over 3052024.13 frames. ], batch size: 109, lr: 7.58e-03, grad_scale: 4.0 2023-04-29 01:49:10,127 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8232, 2.6322, 2.5938, 1.9135, 2.5348, 2.6517, 2.5453, 1.8064], device='cuda:6'), covar=tensor([0.0327, 0.0057, 0.0061, 0.0272, 0.0085, 0.0085, 0.0071, 0.0315], device='cuda:6'), in_proj_covar=tensor([0.0122, 0.0065, 0.0065, 0.0121, 0.0072, 0.0082, 0.0070, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 01:49:12,515 INFO [train.py:904] (6/8) Epoch 9, batch 5800, loss[loss=0.2219, simple_loss=0.3073, pruned_loss=0.06831, over 15420.00 frames. ], tot_loss[loss=0.2402, simple_loss=0.3182, pruned_loss=0.08104, over 3050120.11 frames. ], batch size: 190, lr: 7.58e-03, grad_scale: 4.0 2023-04-29 01:49:13,806 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87002.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 01:49:40,719 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.964e+02 3.181e+02 3.841e+02 4.855e+02 7.782e+02, threshold=7.681e+02, percent-clipped=0.0 2023-04-29 01:50:30,446 INFO [train.py:904] (6/8) Epoch 9, batch 5850, loss[loss=0.2303, simple_loss=0.3107, pruned_loss=0.07491, over 16459.00 frames. ], tot_loss[loss=0.2372, simple_loss=0.3157, pruned_loss=0.07932, over 3050051.67 frames. ], batch size: 146, lr: 7.58e-03, grad_scale: 4.0 2023-04-29 01:50:47,497 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87063.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 01:51:26,628 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0026, 1.6930, 2.4238, 2.8824, 2.7311, 3.1769, 1.9419, 3.1205], device='cuda:6'), covar=tensor([0.0115, 0.0349, 0.0201, 0.0155, 0.0164, 0.0087, 0.0330, 0.0077], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0161, 0.0144, 0.0147, 0.0154, 0.0111, 0.0162, 0.0102], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 01:51:42,104 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87096.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:51:50,071 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87101.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 01:51:50,871 INFO [train.py:904] (6/8) Epoch 9, batch 5900, loss[loss=0.207, simple_loss=0.2902, pruned_loss=0.06188, over 15414.00 frames. ], tot_loss[loss=0.237, simple_loss=0.3154, pruned_loss=0.07925, over 3061231.60 frames. ], batch size: 191, lr: 7.58e-03, grad_scale: 4.0 2023-04-29 01:52:22,918 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.063e+02 3.002e+02 3.675e+02 4.500e+02 8.851e+02, threshold=7.350e+02, percent-clipped=1.0 2023-04-29 01:53:01,034 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87144.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:53:09,080 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87149.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:53:13,003 INFO [train.py:904] (6/8) Epoch 9, batch 5950, loss[loss=0.2356, simple_loss=0.3164, pruned_loss=0.07737, over 16683.00 frames. ], tot_loss[loss=0.2354, simple_loss=0.3155, pruned_loss=0.07762, over 3074405.29 frames. ], batch size: 134, lr: 7.58e-03, grad_scale: 4.0 2023-04-29 01:53:42,083 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87170.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:53:48,775 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9506, 2.7962, 2.6152, 1.9820, 2.5984, 2.0877, 2.7511, 2.8839], device='cuda:6'), covar=tensor([0.0262, 0.0530, 0.0554, 0.1472, 0.0651, 0.0899, 0.0508, 0.0648], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0139, 0.0158, 0.0142, 0.0134, 0.0125, 0.0136, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 01:54:02,320 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-29 01:54:33,096 INFO [train.py:904] (6/8) Epoch 9, batch 6000, loss[loss=0.1954, simple_loss=0.2798, pruned_loss=0.05549, over 16590.00 frames. ], tot_loss[loss=0.2331, simple_loss=0.3137, pruned_loss=0.07628, over 3090411.93 frames. ], batch size: 75, lr: 7.57e-03, grad_scale: 8.0 2023-04-29 01:54:33,096 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 01:54:44,302 INFO [train.py:938] (6/8) Epoch 9, validation: loss=0.1674, simple_loss=0.2809, pruned_loss=0.02692, over 944034.00 frames. 2023-04-29 01:54:44,302 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 01:55:11,799 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.951e+02 3.247e+02 3.795e+02 4.870e+02 1.523e+03, threshold=7.589e+02, percent-clipped=3.0 2023-04-29 01:55:28,586 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87230.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:55:34,789 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87234.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:55:37,576 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87236.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:55:40,315 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6334, 3.7763, 3.9753, 1.8724, 4.2583, 4.3242, 3.1617, 3.2257], device='cuda:6'), covar=tensor([0.0750, 0.0177, 0.0169, 0.1244, 0.0053, 0.0083, 0.0354, 0.0403], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0097, 0.0084, 0.0137, 0.0066, 0.0094, 0.0118, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-29 01:56:03,542 INFO [train.py:904] (6/8) Epoch 9, batch 6050, loss[loss=0.2228, simple_loss=0.3136, pruned_loss=0.06605, over 16903.00 frames. ], tot_loss[loss=0.2312, simple_loss=0.3119, pruned_loss=0.07519, over 3108221.66 frames. ], batch size: 96, lr: 7.57e-03, grad_scale: 8.0 2023-04-29 01:56:56,177 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87284.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:57:06,976 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87291.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:57:23,335 INFO [train.py:904] (6/8) Epoch 9, batch 6100, loss[loss=0.2103, simple_loss=0.2978, pruned_loss=0.06136, over 16171.00 frames. ], tot_loss[loss=0.229, simple_loss=0.3107, pruned_loss=0.07369, over 3110005.01 frames. ], batch size: 165, lr: 7.57e-03, grad_scale: 8.0 2023-04-29 01:57:52,501 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.809e+02 2.938e+02 3.700e+02 4.608e+02 9.243e+02, threshold=7.400e+02, percent-clipped=2.0 2023-04-29 01:58:42,520 INFO [train.py:904] (6/8) Epoch 9, batch 6150, loss[loss=0.2202, simple_loss=0.3081, pruned_loss=0.06609, over 16708.00 frames. ], tot_loss[loss=0.227, simple_loss=0.3086, pruned_loss=0.07267, over 3127516.45 frames. ], batch size: 124, lr: 7.57e-03, grad_scale: 8.0 2023-04-29 01:58:52,865 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87358.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 01:58:56,074 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3166, 3.3449, 3.6325, 1.7259, 3.8368, 3.8734, 2.9220, 2.7513], device='cuda:6'), covar=tensor([0.0721, 0.0179, 0.0127, 0.1168, 0.0042, 0.0091, 0.0354, 0.0453], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0098, 0.0085, 0.0139, 0.0067, 0.0096, 0.0120, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 01:58:59,203 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87362.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:00:00,627 INFO [train.py:904] (6/8) Epoch 9, batch 6200, loss[loss=0.2156, simple_loss=0.2967, pruned_loss=0.06721, over 16402.00 frames. ], tot_loss[loss=0.2266, simple_loss=0.3074, pruned_loss=0.07291, over 3115057.73 frames. ], batch size: 75, lr: 7.57e-03, grad_scale: 8.0 2023-04-29 02:00:28,303 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.203e+02 3.670e+02 4.311e+02 5.768e+02 9.493e+02, threshold=8.622e+02, percent-clipped=7.0 2023-04-29 02:00:34,890 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87423.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:01:18,823 INFO [train.py:904] (6/8) Epoch 9, batch 6250, loss[loss=0.1917, simple_loss=0.2865, pruned_loss=0.04841, over 16830.00 frames. ], tot_loss[loss=0.2268, simple_loss=0.3075, pruned_loss=0.07304, over 3108185.58 frames. ], batch size: 102, lr: 7.56e-03, grad_scale: 4.0 2023-04-29 02:01:32,693 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87461.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:01:38,844 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7909, 1.6986, 1.5296, 1.4978, 1.8768, 1.6298, 1.7315, 1.9687], device='cuda:6'), covar=tensor([0.0091, 0.0177, 0.0254, 0.0230, 0.0118, 0.0180, 0.0112, 0.0132], device='cuda:6'), in_proj_covar=tensor([0.0123, 0.0188, 0.0186, 0.0185, 0.0186, 0.0189, 0.0188, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:01:46,084 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87470.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:02:33,518 INFO [train.py:904] (6/8) Epoch 9, batch 6300, loss[loss=0.2267, simple_loss=0.3154, pruned_loss=0.06893, over 16539.00 frames. ], tot_loss[loss=0.2254, simple_loss=0.307, pruned_loss=0.07189, over 3117797.59 frames. ], batch size: 75, lr: 7.56e-03, grad_scale: 4.0 2023-04-29 02:02:38,939 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87505.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:02:54,506 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0987, 3.2964, 3.5151, 3.4912, 3.4853, 3.2809, 3.3304, 3.4059], device='cuda:6'), covar=tensor([0.0394, 0.0618, 0.0475, 0.0502, 0.0509, 0.0500, 0.0873, 0.0467], device='cuda:6'), in_proj_covar=tensor([0.0309, 0.0308, 0.0320, 0.0306, 0.0361, 0.0329, 0.0441, 0.0269], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 02:02:59,938 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87518.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:03:04,115 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.747e+02 3.209e+02 3.966e+02 4.812e+02 1.231e+03, threshold=7.932e+02, percent-clipped=2.0 2023-04-29 02:03:06,833 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87522.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:03:24,767 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87534.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:03:50,418 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7050, 4.9692, 5.1444, 4.9889, 5.0183, 5.5576, 5.0286, 4.8305], device='cuda:6'), covar=tensor([0.1012, 0.1720, 0.1918, 0.1823, 0.2306, 0.0988, 0.1527, 0.2304], device='cuda:6'), in_proj_covar=tensor([0.0328, 0.0453, 0.0485, 0.0401, 0.0526, 0.0507, 0.0391, 0.0539], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 02:03:51,224 INFO [train.py:904] (6/8) Epoch 9, batch 6350, loss[loss=0.2034, simple_loss=0.29, pruned_loss=0.05834, over 16915.00 frames. ], tot_loss[loss=0.2284, simple_loss=0.3085, pruned_loss=0.0741, over 3093345.11 frames. ], batch size: 90, lr: 7.56e-03, grad_scale: 4.0 2023-04-29 02:04:13,056 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87566.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 02:04:37,493 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87582.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:04:40,682 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87584.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:04:44,423 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87586.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:04:50,113 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87589.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:05:08,848 INFO [train.py:904] (6/8) Epoch 9, batch 6400, loss[loss=0.2091, simple_loss=0.2862, pruned_loss=0.06601, over 17127.00 frames. ], tot_loss[loss=0.2278, simple_loss=0.3076, pruned_loss=0.074, over 3109617.55 frames. ], batch size: 49, lr: 7.56e-03, grad_scale: 8.0 2023-04-29 02:05:14,056 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87605.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:05:37,573 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.591e+02 3.429e+02 4.257e+02 5.158e+02 9.236e+02, threshold=8.515e+02, percent-clipped=3.0 2023-04-29 02:05:51,898 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7975, 5.0985, 4.8306, 4.8880, 4.6091, 4.4551, 4.5020, 5.1620], device='cuda:6'), covar=tensor([0.0906, 0.0802, 0.0971, 0.0639, 0.0697, 0.0948, 0.0913, 0.0787], device='cuda:6'), in_proj_covar=tensor([0.0497, 0.0626, 0.0525, 0.0426, 0.0388, 0.0405, 0.0523, 0.0466], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:06:02,396 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6780, 3.3603, 3.1435, 1.8710, 2.7111, 2.0984, 3.3081, 3.4811], device='cuda:6'), covar=tensor([0.0245, 0.0520, 0.0494, 0.1803, 0.0807, 0.0959, 0.0512, 0.0558], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0138, 0.0157, 0.0141, 0.0134, 0.0124, 0.0135, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 02:06:05,300 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0043, 4.9289, 4.8087, 4.1848, 4.8504, 1.7391, 4.5982, 4.6743], device='cuda:6'), covar=tensor([0.0076, 0.0080, 0.0124, 0.0333, 0.0079, 0.2272, 0.0118, 0.0156], device='cuda:6'), in_proj_covar=tensor([0.0116, 0.0104, 0.0152, 0.0145, 0.0121, 0.0166, 0.0136, 0.0142], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:06:15,000 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87645.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 02:06:22,230 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87650.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:06:24,601 INFO [train.py:904] (6/8) Epoch 9, batch 6450, loss[loss=0.2374, simple_loss=0.3226, pruned_loss=0.07606, over 16675.00 frames. ], tot_loss[loss=0.2277, simple_loss=0.3078, pruned_loss=0.07383, over 3098520.43 frames. ], batch size: 76, lr: 7.55e-03, grad_scale: 4.0 2023-04-29 02:06:27,961 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-29 02:06:34,935 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87658.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 02:06:38,950 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8962, 4.1692, 3.9782, 4.0002, 3.6857, 3.7752, 3.8677, 4.1445], device='cuda:6'), covar=tensor([0.0994, 0.1070, 0.0990, 0.0715, 0.0791, 0.1498, 0.0860, 0.1013], device='cuda:6'), in_proj_covar=tensor([0.0494, 0.0622, 0.0522, 0.0425, 0.0387, 0.0402, 0.0519, 0.0463], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:06:46,460 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87666.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 02:07:41,886 INFO [train.py:904] (6/8) Epoch 9, batch 6500, loss[loss=0.2358, simple_loss=0.301, pruned_loss=0.08537, over 11416.00 frames. ], tot_loss[loss=0.2255, simple_loss=0.3056, pruned_loss=0.07275, over 3099407.74 frames. ], batch size: 248, lr: 7.55e-03, grad_scale: 2.0 2023-04-29 02:07:44,339 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5182, 4.4383, 4.3713, 3.7964, 4.4042, 1.5907, 4.1743, 4.2332], device='cuda:6'), covar=tensor([0.0082, 0.0078, 0.0141, 0.0334, 0.0085, 0.2355, 0.0111, 0.0158], device='cuda:6'), in_proj_covar=tensor([0.0116, 0.0104, 0.0151, 0.0144, 0.0120, 0.0165, 0.0135, 0.0141], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:07:48,426 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87706.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 02:07:59,893 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-29 02:08:06,567 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87718.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:08:13,251 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.348e+02 3.236e+02 4.021e+02 5.179e+02 1.078e+03, threshold=8.043e+02, percent-clipped=2.0 2023-04-29 02:09:00,843 INFO [train.py:904] (6/8) Epoch 9, batch 6550, loss[loss=0.235, simple_loss=0.3306, pruned_loss=0.06965, over 16294.00 frames. ], tot_loss[loss=0.228, simple_loss=0.3087, pruned_loss=0.07362, over 3106970.48 frames. ], batch size: 165, lr: 7.55e-03, grad_scale: 2.0 2023-04-29 02:10:18,518 INFO [train.py:904] (6/8) Epoch 9, batch 6600, loss[loss=0.2121, simple_loss=0.2989, pruned_loss=0.06262, over 16514.00 frames. ], tot_loss[loss=0.2312, simple_loss=0.3116, pruned_loss=0.07535, over 3083440.27 frames. ], batch size: 75, lr: 7.55e-03, grad_scale: 2.0 2023-04-29 02:10:36,708 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-29 02:10:42,561 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87817.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:10:51,645 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.077e+02 3.184e+02 4.033e+02 5.145e+02 1.254e+03, threshold=8.065e+02, percent-clipped=5.0 2023-04-29 02:11:33,654 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-29 02:11:36,976 INFO [train.py:904] (6/8) Epoch 9, batch 6650, loss[loss=0.2183, simple_loss=0.3008, pruned_loss=0.06787, over 16759.00 frames. ], tot_loss[loss=0.2322, simple_loss=0.3121, pruned_loss=0.07616, over 3074114.70 frames. ], batch size: 83, lr: 7.55e-03, grad_scale: 2.0 2023-04-29 02:11:51,502 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87861.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 02:12:29,314 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87886.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:12:53,898 INFO [train.py:904] (6/8) Epoch 9, batch 6700, loss[loss=0.2365, simple_loss=0.314, pruned_loss=0.07953, over 16733.00 frames. ], tot_loss[loss=0.231, simple_loss=0.3103, pruned_loss=0.07585, over 3076017.70 frames. ], batch size: 124, lr: 7.54e-03, grad_scale: 2.0 2023-04-29 02:13:20,547 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 02:13:26,700 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.105e+02 3.383e+02 4.194e+02 5.361e+02 9.838e+02, threshold=8.388e+02, percent-clipped=4.0 2023-04-29 02:13:44,326 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87934.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:13:52,491 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87940.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 02:14:00,295 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87945.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:14:10,407 INFO [train.py:904] (6/8) Epoch 9, batch 6750, loss[loss=0.2578, simple_loss=0.3284, pruned_loss=0.09361, over 16455.00 frames. ], tot_loss[loss=0.2305, simple_loss=0.309, pruned_loss=0.07602, over 3062467.69 frames. ], batch size: 146, lr: 7.54e-03, grad_scale: 2.0 2023-04-29 02:14:15,195 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6540, 4.5832, 4.4402, 4.2488, 4.0674, 4.5017, 4.3413, 4.1671], device='cuda:6'), covar=tensor([0.0397, 0.0309, 0.0257, 0.0236, 0.0943, 0.0306, 0.0395, 0.0718], device='cuda:6'), in_proj_covar=tensor([0.0225, 0.0272, 0.0263, 0.0238, 0.0286, 0.0276, 0.0180, 0.0309], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:14:24,882 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87961.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 02:14:50,161 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8998, 3.7999, 3.9687, 4.1363, 4.1945, 3.7870, 4.1243, 4.2208], device='cuda:6'), covar=tensor([0.1378, 0.0993, 0.1297, 0.0579, 0.0571, 0.1519, 0.0700, 0.0551], device='cuda:6'), in_proj_covar=tensor([0.0486, 0.0601, 0.0745, 0.0616, 0.0471, 0.0467, 0.0491, 0.0542], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:15:29,294 INFO [train.py:904] (6/8) Epoch 9, batch 6800, loss[loss=0.2042, simple_loss=0.2982, pruned_loss=0.05514, over 16771.00 frames. ], tot_loss[loss=0.2292, simple_loss=0.3082, pruned_loss=0.07513, over 3077095.71 frames. ], batch size: 102, lr: 7.54e-03, grad_scale: 4.0 2023-04-29 02:15:54,172 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88018.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:16:02,234 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.936e+02 3.136e+02 3.707e+02 4.804e+02 8.151e+02, threshold=7.414e+02, percent-clipped=0.0 2023-04-29 02:16:19,644 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4553, 4.4167, 4.2758, 4.1262, 3.9324, 4.3272, 4.1688, 4.0264], device='cuda:6'), covar=tensor([0.0427, 0.0346, 0.0270, 0.0210, 0.0901, 0.0360, 0.0457, 0.0607], device='cuda:6'), in_proj_covar=tensor([0.0224, 0.0270, 0.0261, 0.0237, 0.0284, 0.0275, 0.0180, 0.0305], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:16:45,306 INFO [train.py:904] (6/8) Epoch 9, batch 6850, loss[loss=0.257, simple_loss=0.3228, pruned_loss=0.09558, over 11627.00 frames. ], tot_loss[loss=0.2314, simple_loss=0.3103, pruned_loss=0.07624, over 3075766.17 frames. ], batch size: 247, lr: 7.54e-03, grad_scale: 4.0 2023-04-29 02:17:00,369 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-29 02:17:06,663 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88066.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:17:47,991 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 02:17:59,052 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=88100.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:18:00,999 INFO [train.py:904] (6/8) Epoch 9, batch 6900, loss[loss=0.2549, simple_loss=0.3261, pruned_loss=0.09183, over 16902.00 frames. ], tot_loss[loss=0.2324, simple_loss=0.3131, pruned_loss=0.07584, over 3096937.02 frames. ], batch size: 109, lr: 7.54e-03, grad_scale: 4.0 2023-04-29 02:18:24,688 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88117.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:18:33,102 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.937e+02 3.211e+02 3.844e+02 4.463e+02 9.504e+02, threshold=7.687e+02, percent-clipped=4.0 2023-04-29 02:19:18,074 INFO [train.py:904] (6/8) Epoch 9, batch 6950, loss[loss=0.2192, simple_loss=0.3073, pruned_loss=0.06559, over 16727.00 frames. ], tot_loss[loss=0.2362, simple_loss=0.3154, pruned_loss=0.07848, over 3061084.11 frames. ], batch size: 89, lr: 7.53e-03, grad_scale: 4.0 2023-04-29 02:19:31,338 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-29 02:19:32,162 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88161.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 02:19:32,305 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=88161.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:19:37,235 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4045, 3.1329, 3.0135, 1.8702, 2.6509, 2.2067, 3.0806, 3.1940], device='cuda:6'), covar=tensor([0.0276, 0.0586, 0.0565, 0.1742, 0.0796, 0.0924, 0.0633, 0.0708], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0138, 0.0158, 0.0142, 0.0134, 0.0125, 0.0136, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 02:19:38,181 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88165.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:20:07,255 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 02:20:31,302 INFO [train.py:904] (6/8) Epoch 9, batch 7000, loss[loss=0.2625, simple_loss=0.322, pruned_loss=0.1015, over 11667.00 frames. ], tot_loss[loss=0.235, simple_loss=0.3151, pruned_loss=0.0775, over 3052835.03 frames. ], batch size: 247, lr: 7.53e-03, grad_scale: 4.0 2023-04-29 02:20:43,255 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88209.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:21:01,197 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0467, 1.6797, 2.3928, 3.0663, 2.6890, 3.2863, 1.8708, 3.2255], device='cuda:6'), covar=tensor([0.0116, 0.0357, 0.0229, 0.0156, 0.0207, 0.0110, 0.0386, 0.0084], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0162, 0.0144, 0.0146, 0.0155, 0.0113, 0.0163, 0.0103], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 02:21:03,406 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.948e+02 3.320e+02 4.209e+02 4.909e+02 8.638e+02, threshold=8.417e+02, percent-clipped=2.0 2023-04-29 02:21:25,673 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.59 vs. limit=5.0 2023-04-29 02:21:29,057 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88240.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:21:36,647 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88245.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:21:47,202 INFO [train.py:904] (6/8) Epoch 9, batch 7050, loss[loss=0.1943, simple_loss=0.2815, pruned_loss=0.05352, over 16359.00 frames. ], tot_loss[loss=0.2339, simple_loss=0.3152, pruned_loss=0.07631, over 3079693.90 frames. ], batch size: 35, lr: 7.53e-03, grad_scale: 4.0 2023-04-29 02:22:00,409 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88261.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 02:22:41,149 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88288.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:22:44,380 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1662, 3.2443, 1.7331, 3.5036, 2.3444, 3.4455, 1.9006, 2.5361], device='cuda:6'), covar=tensor([0.0233, 0.0337, 0.1677, 0.0107, 0.0792, 0.0487, 0.1499, 0.0779], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0160, 0.0184, 0.0111, 0.0164, 0.0200, 0.0192, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 02:22:45,508 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=88291.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 02:22:47,891 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88293.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:23:02,082 INFO [train.py:904] (6/8) Epoch 9, batch 7100, loss[loss=0.2677, simple_loss=0.319, pruned_loss=0.1082, over 11831.00 frames. ], tot_loss[loss=0.2321, simple_loss=0.3132, pruned_loss=0.07551, over 3099199.36 frames. ], batch size: 248, lr: 7.53e-03, grad_scale: 4.0 2023-04-29 02:23:12,647 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88309.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:23:33,099 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.153e+02 3.325e+02 4.085e+02 5.101e+02 9.859e+02, threshold=8.169e+02, percent-clipped=1.0 2023-04-29 02:24:13,913 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2023-04-29 02:24:15,977 INFO [train.py:904] (6/8) Epoch 9, batch 7150, loss[loss=0.2541, simple_loss=0.31, pruned_loss=0.0991, over 11517.00 frames. ], tot_loss[loss=0.2307, simple_loss=0.3111, pruned_loss=0.07511, over 3087148.95 frames. ], batch size: 247, lr: 7.52e-03, grad_scale: 4.0 2023-04-29 02:24:17,099 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=88352.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 02:25:28,098 INFO [train.py:904] (6/8) Epoch 9, batch 7200, loss[loss=0.2238, simple_loss=0.2915, pruned_loss=0.0781, over 11572.00 frames. ], tot_loss[loss=0.2289, simple_loss=0.3093, pruned_loss=0.07424, over 3046148.87 frames. ], batch size: 246, lr: 7.52e-03, grad_scale: 8.0 2023-04-29 02:26:00,148 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.048e+02 3.005e+02 3.439e+02 4.228e+02 8.504e+02, threshold=6.879e+02, percent-clipped=1.0 2023-04-29 02:26:47,224 INFO [train.py:904] (6/8) Epoch 9, batch 7250, loss[loss=0.188, simple_loss=0.2602, pruned_loss=0.0579, over 17074.00 frames. ], tot_loss[loss=0.2261, simple_loss=0.3067, pruned_loss=0.07278, over 3056753.62 frames. ], batch size: 53, lr: 7.52e-03, grad_scale: 8.0 2023-04-29 02:26:53,321 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=88456.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:28:00,985 INFO [train.py:904] (6/8) Epoch 9, batch 7300, loss[loss=0.22, simple_loss=0.3085, pruned_loss=0.06575, over 16566.00 frames. ], tot_loss[loss=0.2259, simple_loss=0.3063, pruned_loss=0.0728, over 3061385.13 frames. ], batch size: 62, lr: 7.52e-03, grad_scale: 8.0 2023-04-29 02:28:19,691 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9237, 5.5630, 5.7855, 5.5664, 5.6600, 6.1775, 5.5790, 5.3887], device='cuda:6'), covar=tensor([0.0813, 0.1680, 0.1460, 0.1521, 0.2037, 0.0765, 0.1395, 0.2354], device='cuda:6'), in_proj_covar=tensor([0.0333, 0.0453, 0.0486, 0.0403, 0.0530, 0.0514, 0.0393, 0.0547], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 02:28:33,453 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 3.255e+02 4.092e+02 5.788e+02 1.345e+03, threshold=8.184e+02, percent-clipped=12.0 2023-04-29 02:28:36,380 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6444, 3.8521, 3.0926, 2.4283, 2.9996, 2.4654, 4.2641, 3.6145], device='cuda:6'), covar=tensor([0.2603, 0.0734, 0.1446, 0.1780, 0.2066, 0.1661, 0.0381, 0.0917], device='cuda:6'), in_proj_covar=tensor([0.0296, 0.0255, 0.0280, 0.0269, 0.0280, 0.0214, 0.0263, 0.0281], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:29:14,111 INFO [train.py:904] (6/8) Epoch 9, batch 7350, loss[loss=0.2126, simple_loss=0.2933, pruned_loss=0.06596, over 16487.00 frames. ], tot_loss[loss=0.2269, simple_loss=0.3071, pruned_loss=0.07336, over 3056099.42 frames. ], batch size: 68, lr: 7.52e-03, grad_scale: 4.0 2023-04-29 02:30:27,377 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9772, 3.4605, 3.2298, 1.8031, 2.8057, 2.2711, 3.4818, 3.5269], device='cuda:6'), covar=tensor([0.0249, 0.0625, 0.0648, 0.2004, 0.0860, 0.0976, 0.0682, 0.0885], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0140, 0.0158, 0.0143, 0.0135, 0.0127, 0.0138, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 02:30:27,925 INFO [train.py:904] (6/8) Epoch 9, batch 7400, loss[loss=0.2353, simple_loss=0.3191, pruned_loss=0.0758, over 16876.00 frames. ], tot_loss[loss=0.229, simple_loss=0.3085, pruned_loss=0.07472, over 3039376.03 frames. ], batch size: 102, lr: 7.51e-03, grad_scale: 4.0 2023-04-29 02:30:41,952 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-29 02:30:56,956 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8388, 4.1236, 3.8754, 3.9412, 3.6489, 3.7680, 3.8086, 4.1201], device='cuda:6'), covar=tensor([0.1009, 0.0884, 0.1074, 0.0692, 0.0776, 0.1352, 0.0864, 0.0927], device='cuda:6'), in_proj_covar=tensor([0.0489, 0.0617, 0.0520, 0.0418, 0.0381, 0.0400, 0.0511, 0.0465], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:31:01,751 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.901e+02 3.181e+02 3.584e+02 4.494e+02 7.659e+02, threshold=7.169e+02, percent-clipped=0.0 2023-04-29 02:31:38,444 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=88647.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 02:31:44,621 INFO [train.py:904] (6/8) Epoch 9, batch 7450, loss[loss=0.2166, simple_loss=0.3095, pruned_loss=0.0619, over 16713.00 frames. ], tot_loss[loss=0.2305, simple_loss=0.3097, pruned_loss=0.07568, over 3043211.70 frames. ], batch size: 76, lr: 7.51e-03, grad_scale: 4.0 2023-04-29 02:32:51,451 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-29 02:33:03,522 INFO [train.py:904] (6/8) Epoch 9, batch 7500, loss[loss=0.1966, simple_loss=0.2786, pruned_loss=0.05732, over 16945.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.3103, pruned_loss=0.07492, over 3039151.27 frames. ], batch size: 109, lr: 7.51e-03, grad_scale: 4.0 2023-04-29 02:33:36,928 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.091e+02 3.551e+02 4.142e+02 5.326e+02 1.060e+03, threshold=8.283e+02, percent-clipped=5.0 2023-04-29 02:34:09,644 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6414, 2.1460, 1.8412, 2.0811, 2.5393, 2.2672, 2.7236, 2.8201], device='cuda:6'), covar=tensor([0.0082, 0.0270, 0.0337, 0.0287, 0.0155, 0.0254, 0.0124, 0.0152], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0187, 0.0185, 0.0184, 0.0185, 0.0187, 0.0186, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:34:18,158 INFO [train.py:904] (6/8) Epoch 9, batch 7550, loss[loss=0.2167, simple_loss=0.2988, pruned_loss=0.06734, over 16756.00 frames. ], tot_loss[loss=0.2295, simple_loss=0.309, pruned_loss=0.07493, over 3044700.69 frames. ], batch size: 83, lr: 7.51e-03, grad_scale: 4.0 2023-04-29 02:34:23,600 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88756.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:34:45,545 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 02:35:25,811 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6383, 4.0927, 3.7034, 2.0830, 3.2709, 3.0167, 3.7771, 4.1825], device='cuda:6'), covar=tensor([0.0244, 0.0523, 0.0605, 0.1718, 0.0695, 0.0738, 0.0601, 0.0821], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0140, 0.0157, 0.0142, 0.0134, 0.0126, 0.0137, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 02:35:33,596 INFO [train.py:904] (6/8) Epoch 9, batch 7600, loss[loss=0.258, simple_loss=0.3191, pruned_loss=0.09847, over 11194.00 frames. ], tot_loss[loss=0.2288, simple_loss=0.3081, pruned_loss=0.07473, over 3038056.82 frames. ], batch size: 246, lr: 7.51e-03, grad_scale: 8.0 2023-04-29 02:35:37,865 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88804.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:36:05,554 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.836e+02 3.313e+02 3.789e+02 4.639e+02 1.052e+03, threshold=7.577e+02, percent-clipped=2.0 2023-04-29 02:36:29,534 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7459, 3.6858, 3.8368, 3.6917, 3.7291, 4.1907, 3.9019, 3.6095], device='cuda:6'), covar=tensor([0.1984, 0.2227, 0.2235, 0.2335, 0.3017, 0.1581, 0.1631, 0.2734], device='cuda:6'), in_proj_covar=tensor([0.0332, 0.0449, 0.0483, 0.0400, 0.0524, 0.0513, 0.0390, 0.0540], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 02:36:45,916 INFO [train.py:904] (6/8) Epoch 9, batch 7650, loss[loss=0.2626, simple_loss=0.323, pruned_loss=0.1011, over 11400.00 frames. ], tot_loss[loss=0.2292, simple_loss=0.3085, pruned_loss=0.07496, over 3054353.48 frames. ], batch size: 246, lr: 7.50e-03, grad_scale: 8.0 2023-04-29 02:37:15,865 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1660, 1.8911, 1.5676, 1.5955, 2.1969, 1.9023, 2.1860, 2.3792], device='cuda:6'), covar=tensor([0.0099, 0.0258, 0.0373, 0.0339, 0.0154, 0.0244, 0.0147, 0.0151], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0188, 0.0186, 0.0183, 0.0185, 0.0187, 0.0186, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:38:01,886 INFO [train.py:904] (6/8) Epoch 9, batch 7700, loss[loss=0.2999, simple_loss=0.3399, pruned_loss=0.13, over 11646.00 frames. ], tot_loss[loss=0.2309, simple_loss=0.3095, pruned_loss=0.07618, over 3058743.87 frames. ], batch size: 248, lr: 7.50e-03, grad_scale: 8.0 2023-04-29 02:38:04,162 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 02:38:34,565 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.958e+02 3.502e+02 4.245e+02 5.386e+02 9.706e+02, threshold=8.489e+02, percent-clipped=2.0 2023-04-29 02:39:08,840 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88947.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 02:39:16,827 INFO [train.py:904] (6/8) Epoch 9, batch 7750, loss[loss=0.2517, simple_loss=0.3076, pruned_loss=0.09792, over 11469.00 frames. ], tot_loss[loss=0.2305, simple_loss=0.3095, pruned_loss=0.0757, over 3074024.10 frames. ], batch size: 247, lr: 7.50e-03, grad_scale: 8.0 2023-04-29 02:39:58,798 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8781, 3.4738, 2.8129, 1.6456, 2.5556, 2.0978, 3.1297, 3.5249], device='cuda:6'), covar=tensor([0.0255, 0.0481, 0.0663, 0.1924, 0.0897, 0.1003, 0.0690, 0.0649], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0139, 0.0157, 0.0142, 0.0134, 0.0125, 0.0137, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 02:40:20,785 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88995.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 02:40:29,342 INFO [train.py:904] (6/8) Epoch 9, batch 7800, loss[loss=0.197, simple_loss=0.2891, pruned_loss=0.05246, over 16764.00 frames. ], tot_loss[loss=0.2306, simple_loss=0.3098, pruned_loss=0.07571, over 3093922.56 frames. ], batch size: 102, lr: 7.50e-03, grad_scale: 8.0 2023-04-29 02:41:02,672 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.440e+02 3.396e+02 4.306e+02 5.337e+02 1.534e+03, threshold=8.611e+02, percent-clipped=4.0 2023-04-29 02:41:07,718 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89027.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:41:42,368 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7994, 5.2593, 5.5213, 5.2189, 5.3166, 5.8499, 5.3168, 5.0649], device='cuda:6'), covar=tensor([0.0934, 0.1645, 0.1714, 0.1601, 0.1932, 0.0892, 0.1446, 0.2540], device='cuda:6'), in_proj_covar=tensor([0.0332, 0.0451, 0.0485, 0.0402, 0.0524, 0.0515, 0.0391, 0.0541], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 02:41:44,457 INFO [train.py:904] (6/8) Epoch 9, batch 7850, loss[loss=0.2381, simple_loss=0.3195, pruned_loss=0.07833, over 16193.00 frames. ], tot_loss[loss=0.23, simple_loss=0.31, pruned_loss=0.07498, over 3088613.03 frames. ], batch size: 165, lr: 7.49e-03, grad_scale: 8.0 2023-04-29 02:42:39,060 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89088.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:42:49,409 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7107, 1.2663, 1.6089, 1.6370, 1.8020, 1.8540, 1.4968, 1.6321], device='cuda:6'), covar=tensor([0.0167, 0.0244, 0.0130, 0.0184, 0.0159, 0.0104, 0.0251, 0.0065], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0161, 0.0144, 0.0146, 0.0156, 0.0114, 0.0163, 0.0103], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 02:42:57,227 INFO [train.py:904] (6/8) Epoch 9, batch 7900, loss[loss=0.2565, simple_loss=0.3474, pruned_loss=0.08285, over 16219.00 frames. ], tot_loss[loss=0.2292, simple_loss=0.3092, pruned_loss=0.07463, over 3069681.14 frames. ], batch size: 165, lr: 7.49e-03, grad_scale: 8.0 2023-04-29 02:43:18,472 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0462, 2.2600, 1.7035, 2.0248, 2.6985, 2.4300, 2.9663, 2.9660], device='cuda:6'), covar=tensor([0.0067, 0.0277, 0.0382, 0.0357, 0.0150, 0.0259, 0.0137, 0.0154], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0188, 0.0187, 0.0186, 0.0188, 0.0189, 0.0189, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:43:28,442 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.998e+02 3.084e+02 3.809e+02 4.466e+02 8.463e+02, threshold=7.617e+02, percent-clipped=0.0 2023-04-29 02:44:12,923 INFO [train.py:904] (6/8) Epoch 9, batch 7950, loss[loss=0.2563, simple_loss=0.3334, pruned_loss=0.0896, over 16326.00 frames. ], tot_loss[loss=0.2293, simple_loss=0.3093, pruned_loss=0.07467, over 3079613.71 frames. ], batch size: 146, lr: 7.49e-03, grad_scale: 8.0 2023-04-29 02:44:23,072 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2023-04-29 02:44:39,153 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89169.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:45:26,762 INFO [train.py:904] (6/8) Epoch 9, batch 8000, loss[loss=0.2195, simple_loss=0.3009, pruned_loss=0.06906, over 16841.00 frames. ], tot_loss[loss=0.2308, simple_loss=0.3101, pruned_loss=0.07572, over 3076165.33 frames. ], batch size: 116, lr: 7.49e-03, grad_scale: 8.0 2023-04-29 02:45:59,520 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.399e+02 3.668e+02 4.137e+02 4.605e+02 8.803e+02, threshold=8.275e+02, percent-clipped=3.0 2023-04-29 02:46:09,133 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89230.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:46:16,413 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1541, 5.1309, 4.9519, 4.7055, 4.4733, 5.0211, 4.9459, 4.6591], device='cuda:6'), covar=tensor([0.0576, 0.0422, 0.0259, 0.0232, 0.1080, 0.0439, 0.0224, 0.0627], device='cuda:6'), in_proj_covar=tensor([0.0225, 0.0271, 0.0259, 0.0240, 0.0280, 0.0272, 0.0179, 0.0304], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:46:40,072 INFO [train.py:904] (6/8) Epoch 9, batch 8050, loss[loss=0.2288, simple_loss=0.3157, pruned_loss=0.07094, over 16201.00 frames. ], tot_loss[loss=0.2315, simple_loss=0.3107, pruned_loss=0.07614, over 3069652.54 frames. ], batch size: 165, lr: 7.49e-03, grad_scale: 4.0 2023-04-29 02:47:18,126 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-29 02:47:55,802 INFO [train.py:904] (6/8) Epoch 9, batch 8100, loss[loss=0.242, simple_loss=0.3064, pruned_loss=0.08876, over 11575.00 frames. ], tot_loss[loss=0.2302, simple_loss=0.3097, pruned_loss=0.07536, over 3070692.82 frames. ], batch size: 246, lr: 7.48e-03, grad_scale: 4.0 2023-04-29 02:48:29,433 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.308e+02 2.948e+02 3.621e+02 4.670e+02 8.254e+02, threshold=7.243e+02, percent-clipped=0.0 2023-04-29 02:48:40,996 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4820, 3.5364, 3.2131, 3.1166, 3.0977, 3.3929, 3.2743, 3.2533], device='cuda:6'), covar=tensor([0.0538, 0.0475, 0.0240, 0.0204, 0.0600, 0.0373, 0.0994, 0.0474], device='cuda:6'), in_proj_covar=tensor([0.0229, 0.0276, 0.0263, 0.0244, 0.0285, 0.0276, 0.0182, 0.0308], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:49:10,970 INFO [train.py:904] (6/8) Epoch 9, batch 8150, loss[loss=0.1913, simple_loss=0.2687, pruned_loss=0.05692, over 17229.00 frames. ], tot_loss[loss=0.2274, simple_loss=0.3068, pruned_loss=0.07398, over 3079065.43 frames. ], batch size: 52, lr: 7.48e-03, grad_scale: 4.0 2023-04-29 02:49:34,620 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0255, 2.5807, 2.6329, 1.9441, 2.7895, 2.8396, 2.4524, 2.3730], device='cuda:6'), covar=tensor([0.0718, 0.0160, 0.0179, 0.0883, 0.0078, 0.0172, 0.0393, 0.0429], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0099, 0.0087, 0.0142, 0.0068, 0.0096, 0.0120, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 02:49:45,625 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89374.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:49:59,922 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89383.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:50:01,716 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6461, 4.9123, 4.6758, 4.6377, 4.3321, 4.3996, 4.4057, 4.9832], device='cuda:6'), covar=tensor([0.0836, 0.0798, 0.0963, 0.0716, 0.0789, 0.0965, 0.0922, 0.0868], device='cuda:6'), in_proj_covar=tensor([0.0497, 0.0630, 0.0530, 0.0429, 0.0391, 0.0411, 0.0526, 0.0474], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:50:16,417 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-04-29 02:50:28,004 INFO [train.py:904] (6/8) Epoch 9, batch 8200, loss[loss=0.2103, simple_loss=0.2978, pruned_loss=0.06142, over 16731.00 frames. ], tot_loss[loss=0.2252, simple_loss=0.3042, pruned_loss=0.07307, over 3091884.02 frames. ], batch size: 83, lr: 7.48e-03, grad_scale: 4.0 2023-04-29 02:51:05,359 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89425.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:51:06,153 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.992e+02 3.313e+02 4.005e+02 4.576e+02 8.683e+02, threshold=8.011e+02, percent-clipped=3.0 2023-04-29 02:51:16,480 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2526, 1.4272, 1.8423, 2.1399, 2.2571, 2.4323, 1.6497, 2.2612], device='cuda:6'), covar=tensor([0.0121, 0.0337, 0.0213, 0.0204, 0.0185, 0.0125, 0.0314, 0.0110], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0161, 0.0144, 0.0147, 0.0157, 0.0114, 0.0163, 0.0103], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 02:51:20,805 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89435.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:51:27,529 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89439.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 02:51:47,430 INFO [train.py:904] (6/8) Epoch 9, batch 8250, loss[loss=0.215, simple_loss=0.3129, pruned_loss=0.05852, over 16909.00 frames. ], tot_loss[loss=0.223, simple_loss=0.3042, pruned_loss=0.07088, over 3092295.31 frames. ], batch size: 116, lr: 7.48e-03, grad_scale: 2.0 2023-04-29 02:52:02,741 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6645, 2.6658, 1.8210, 2.7974, 2.1501, 2.8082, 2.0018, 2.4563], device='cuda:6'), covar=tensor([0.0215, 0.0311, 0.1128, 0.0172, 0.0655, 0.0460, 0.1123, 0.0507], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0160, 0.0186, 0.0112, 0.0165, 0.0200, 0.0194, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 02:52:34,065 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8369, 2.5380, 2.6038, 1.9268, 2.4486, 2.5023, 2.6024, 1.8029], device='cuda:6'), covar=tensor([0.0345, 0.0048, 0.0046, 0.0263, 0.0083, 0.0070, 0.0071, 0.0342], device='cuda:6'), in_proj_covar=tensor([0.0124, 0.0063, 0.0065, 0.0123, 0.0072, 0.0083, 0.0072, 0.0117], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 02:52:43,710 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89486.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:53:06,320 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89500.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 02:53:08,808 INFO [train.py:904] (6/8) Epoch 9, batch 8300, loss[loss=0.2116, simple_loss=0.3151, pruned_loss=0.0541, over 16731.00 frames. ], tot_loss[loss=0.2178, simple_loss=0.3009, pruned_loss=0.06731, over 3090010.44 frames. ], batch size: 124, lr: 7.48e-03, grad_scale: 2.0 2023-04-29 02:53:21,216 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89509.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:53:47,800 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89525.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:53:48,527 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.763e+02 2.745e+02 3.164e+02 3.682e+02 6.265e+02, threshold=6.329e+02, percent-clipped=0.0 2023-04-29 02:54:31,537 INFO [train.py:904] (6/8) Epoch 9, batch 8350, loss[loss=0.2328, simple_loss=0.3079, pruned_loss=0.07884, over 12329.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.2993, pruned_loss=0.0655, over 3054517.33 frames. ], batch size: 248, lr: 7.47e-03, grad_scale: 2.0 2023-04-29 02:55:01,180 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89570.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:55:08,104 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2688, 3.8734, 3.7786, 2.0648, 3.1342, 2.5497, 3.7291, 3.7687], device='cuda:6'), covar=tensor([0.0232, 0.0520, 0.0428, 0.1719, 0.0645, 0.0838, 0.0556, 0.0740], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0135, 0.0153, 0.0139, 0.0130, 0.0123, 0.0133, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 02:55:45,378 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5671, 3.6237, 3.2576, 3.0961, 3.1578, 3.4780, 3.2656, 3.3134], device='cuda:6'), covar=tensor([0.0498, 0.0488, 0.0247, 0.0223, 0.0566, 0.0380, 0.1195, 0.0496], device='cuda:6'), in_proj_covar=tensor([0.0220, 0.0267, 0.0256, 0.0236, 0.0277, 0.0267, 0.0177, 0.0300], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:55:52,200 INFO [train.py:904] (6/8) Epoch 9, batch 8400, loss[loss=0.209, simple_loss=0.2934, pruned_loss=0.06226, over 16438.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2967, pruned_loss=0.06312, over 3061072.78 frames. ], batch size: 146, lr: 7.47e-03, grad_scale: 4.0 2023-04-29 02:56:07,035 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3927, 3.5262, 1.8301, 3.8534, 2.3263, 3.8244, 2.0272, 2.8797], device='cuda:6'), covar=tensor([0.0219, 0.0349, 0.1689, 0.0106, 0.0913, 0.0394, 0.1623, 0.0606], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0157, 0.0183, 0.0111, 0.0163, 0.0196, 0.0191, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 02:56:31,315 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.203e+02 2.904e+02 3.369e+02 3.930e+02 8.032e+02, threshold=6.737e+02, percent-clipped=2.0 2023-04-29 02:57:13,775 INFO [train.py:904] (6/8) Epoch 9, batch 8450, loss[loss=0.1853, simple_loss=0.2673, pruned_loss=0.0516, over 12659.00 frames. ], tot_loss[loss=0.2094, simple_loss=0.2953, pruned_loss=0.06168, over 3069346.06 frames. ], batch size: 247, lr: 7.47e-03, grad_scale: 4.0 2023-04-29 02:57:25,962 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2162, 3.4002, 3.6170, 3.5984, 3.6095, 3.4239, 3.4464, 3.4836], device='cuda:6'), covar=tensor([0.0335, 0.0604, 0.0417, 0.0459, 0.0466, 0.0377, 0.0703, 0.0390], device='cuda:6'), in_proj_covar=tensor([0.0298, 0.0300, 0.0305, 0.0293, 0.0343, 0.0319, 0.0422, 0.0259], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-29 02:58:00,146 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89680.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:58:04,750 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=89683.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:58:34,963 INFO [train.py:904] (6/8) Epoch 9, batch 8500, loss[loss=0.1728, simple_loss=0.2662, pruned_loss=0.03972, over 15242.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2905, pruned_loss=0.05861, over 3042617.60 frames. ], batch size: 191, lr: 7.47e-03, grad_scale: 4.0 2023-04-29 02:59:11,550 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5645, 5.9162, 5.6889, 5.6935, 5.2973, 5.1848, 5.4776, 6.0040], device='cuda:6'), covar=tensor([0.1075, 0.0828, 0.0857, 0.0606, 0.0707, 0.0639, 0.0777, 0.0820], device='cuda:6'), in_proj_covar=tensor([0.0474, 0.0602, 0.0502, 0.0410, 0.0371, 0.0394, 0.0502, 0.0457], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 02:59:14,097 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 2.444e+02 3.113e+02 3.731e+02 7.658e+02, threshold=6.225e+02, percent-clipped=1.0 2023-04-29 02:59:20,978 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89730.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:59:22,205 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=89731.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:59:39,468 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89741.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:59:58,592 INFO [train.py:904] (6/8) Epoch 9, batch 8550, loss[loss=0.2099, simple_loss=0.2872, pruned_loss=0.0663, over 12197.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2882, pruned_loss=0.05739, over 3033164.56 frames. ], batch size: 246, lr: 7.47e-03, grad_scale: 4.0 2023-04-29 03:00:55,440 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89781.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:01:24,770 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89795.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 03:01:37,953 INFO [train.py:904] (6/8) Epoch 9, batch 8600, loss[loss=0.1712, simple_loss=0.2702, pruned_loss=0.03612, over 16680.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2885, pruned_loss=0.05657, over 3030067.23 frames. ], batch size: 89, lr: 7.46e-03, grad_scale: 4.0 2023-04-29 03:02:25,766 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=89825.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:02:26,423 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.775e+02 2.695e+02 3.313e+02 4.176e+02 8.025e+02, threshold=6.625e+02, percent-clipped=2.0 2023-04-29 03:03:15,780 INFO [train.py:904] (6/8) Epoch 9, batch 8650, loss[loss=0.1766, simple_loss=0.265, pruned_loss=0.04408, over 12387.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2858, pruned_loss=0.05443, over 3032649.98 frames. ], batch size: 248, lr: 7.46e-03, grad_scale: 4.0 2023-04-29 03:03:47,723 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89865.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:03:47,837 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89865.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:04:05,356 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=89873.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:05:02,347 INFO [train.py:904] (6/8) Epoch 9, batch 8700, loss[loss=0.1982, simple_loss=0.2775, pruned_loss=0.05949, over 16760.00 frames. ], tot_loss[loss=0.1953, simple_loss=0.2833, pruned_loss=0.05371, over 3028561.30 frames. ], batch size: 83, lr: 7.46e-03, grad_scale: 4.0 2023-04-29 03:05:34,175 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9471, 2.3470, 2.2224, 3.0019, 2.0554, 3.2793, 1.6213, 2.7497], device='cuda:6'), covar=tensor([0.1205, 0.0558, 0.1074, 0.0142, 0.0095, 0.0418, 0.1386, 0.0668], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0150, 0.0174, 0.0124, 0.0190, 0.0202, 0.0172, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 03:05:45,059 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.721e+02 2.547e+02 3.087e+02 3.551e+02 5.785e+02, threshold=6.175e+02, percent-clipped=0.0 2023-04-29 03:05:45,891 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4485, 1.8970, 1.6861, 1.5838, 2.2345, 1.8650, 2.2153, 2.2637], device='cuda:6'), covar=tensor([0.0091, 0.0263, 0.0334, 0.0321, 0.0155, 0.0267, 0.0120, 0.0187], device='cuda:6'), in_proj_covar=tensor([0.0120, 0.0186, 0.0184, 0.0184, 0.0183, 0.0185, 0.0180, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:05:45,895 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89926.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 03:06:36,263 INFO [train.py:904] (6/8) Epoch 9, batch 8750, loss[loss=0.1925, simple_loss=0.2898, pruned_loss=0.04759, over 16887.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2825, pruned_loss=0.05265, over 3045764.49 frames. ], batch size: 116, lr: 7.46e-03, grad_scale: 4.0 2023-04-29 03:08:32,637 INFO [train.py:904] (6/8) Epoch 9, batch 8800, loss[loss=0.2068, simple_loss=0.2908, pruned_loss=0.06134, over 15370.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.2809, pruned_loss=0.0513, over 3065978.97 frames. ], batch size: 191, lr: 7.46e-03, grad_scale: 8.0 2023-04-29 03:09:21,968 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.546e+02 3.019e+02 3.613e+02 7.330e+02, threshold=6.037e+02, percent-clipped=3.0 2023-04-29 03:09:31,860 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90030.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:09:44,772 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90036.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:10:17,856 INFO [train.py:904] (6/8) Epoch 9, batch 8850, loss[loss=0.1859, simple_loss=0.2879, pruned_loss=0.04193, over 16351.00 frames. ], tot_loss[loss=0.1916, simple_loss=0.2827, pruned_loss=0.05022, over 3057775.88 frames. ], batch size: 166, lr: 7.45e-03, grad_scale: 8.0 2023-04-29 03:10:32,668 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6269, 4.5965, 4.3743, 4.0185, 4.4309, 1.5164, 4.1911, 4.2805], device='cuda:6'), covar=tensor([0.0061, 0.0055, 0.0120, 0.0224, 0.0067, 0.2200, 0.0098, 0.0153], device='cuda:6'), in_proj_covar=tensor([0.0115, 0.0101, 0.0147, 0.0139, 0.0118, 0.0167, 0.0133, 0.0137], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:11:12,094 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90078.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:11:18,646 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90081.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:11:50,076 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90095.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 03:11:54,111 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6630, 3.9220, 2.8028, 2.3809, 2.7607, 2.2687, 4.0911, 3.6587], device='cuda:6'), covar=tensor([0.2702, 0.0869, 0.1751, 0.2194, 0.2341, 0.1884, 0.0558, 0.0904], device='cuda:6'), in_proj_covar=tensor([0.0287, 0.0242, 0.0270, 0.0259, 0.0255, 0.0206, 0.0251, 0.0265], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:12:02,856 INFO [train.py:904] (6/8) Epoch 9, batch 8900, loss[loss=0.1841, simple_loss=0.2803, pruned_loss=0.04389, over 16676.00 frames. ], tot_loss[loss=0.1913, simple_loss=0.2832, pruned_loss=0.04969, over 3070814.75 frames. ], batch size: 89, lr: 7.45e-03, grad_scale: 4.0 2023-04-29 03:12:57,513 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.859e+02 2.641e+02 3.095e+02 3.699e+02 6.742e+02, threshold=6.190e+02, percent-clipped=4.0 2023-04-29 03:13:07,421 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90129.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:13:10,555 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90130.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:13:46,999 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-29 03:13:48,683 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90143.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 03:14:08,705 INFO [train.py:904] (6/8) Epoch 9, batch 8950, loss[loss=0.1624, simple_loss=0.2585, pruned_loss=0.03319, over 16736.00 frames. ], tot_loss[loss=0.1913, simple_loss=0.2829, pruned_loss=0.04986, over 3076200.47 frames. ], batch size: 83, lr: 7.45e-03, grad_scale: 4.0 2023-04-29 03:14:12,481 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 03:14:39,012 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90165.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:15:35,023 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90191.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:15:50,842 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-29 03:15:57,359 INFO [train.py:904] (6/8) Epoch 9, batch 9000, loss[loss=0.1969, simple_loss=0.278, pruned_loss=0.05787, over 12209.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2801, pruned_loss=0.04876, over 3070478.23 frames. ], batch size: 247, lr: 7.45e-03, grad_scale: 4.0 2023-04-29 03:15:57,359 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 03:16:07,533 INFO [train.py:938] (6/8) Epoch 9, validation: loss=0.1581, simple_loss=0.2623, pruned_loss=0.02697, over 944034.00 frames. 2023-04-29 03:16:07,533 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 03:16:31,273 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90213.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:16:40,095 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3050, 4.1191, 4.3880, 4.5200, 4.6545, 4.1475, 4.6701, 4.6294], device='cuda:6'), covar=tensor([0.1510, 0.1140, 0.1328, 0.0670, 0.0532, 0.1065, 0.0476, 0.0622], device='cuda:6'), in_proj_covar=tensor([0.0458, 0.0566, 0.0687, 0.0573, 0.0440, 0.0441, 0.0460, 0.0516], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:16:48,419 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90221.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 03:16:58,704 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.701e+02 2.410e+02 2.881e+02 3.616e+02 7.746e+02, threshold=5.761e+02, percent-clipped=2.0 2023-04-29 03:17:09,177 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90232.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:17:44,078 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2036, 3.3728, 3.6561, 3.6407, 3.6244, 3.4005, 3.4616, 3.4815], device='cuda:6'), covar=tensor([0.0320, 0.0575, 0.0399, 0.0407, 0.0433, 0.0445, 0.0649, 0.0353], device='cuda:6'), in_proj_covar=tensor([0.0294, 0.0293, 0.0300, 0.0292, 0.0337, 0.0315, 0.0411, 0.0254], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-29 03:17:51,878 INFO [train.py:904] (6/8) Epoch 9, batch 9050, loss[loss=0.1784, simple_loss=0.2637, pruned_loss=0.04654, over 17002.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.281, pruned_loss=0.04929, over 3065539.58 frames. ], batch size: 41, lr: 7.45e-03, grad_scale: 4.0 2023-04-29 03:19:19,922 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90293.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:19:24,860 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4089, 3.3323, 3.4324, 3.5400, 3.5816, 3.2538, 3.5583, 3.6082], device='cuda:6'), covar=tensor([0.1081, 0.0894, 0.1048, 0.0576, 0.0560, 0.2268, 0.0776, 0.0679], device='cuda:6'), in_proj_covar=tensor([0.0464, 0.0574, 0.0695, 0.0578, 0.0444, 0.0443, 0.0462, 0.0521], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:19:36,841 INFO [train.py:904] (6/8) Epoch 9, batch 9100, loss[loss=0.1937, simple_loss=0.2719, pruned_loss=0.05779, over 12401.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2806, pruned_loss=0.04984, over 3061546.53 frames. ], batch size: 248, lr: 7.44e-03, grad_scale: 4.0 2023-04-29 03:19:42,828 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4124, 3.3240, 3.4294, 3.5476, 3.5774, 3.2323, 3.5455, 3.6029], device='cuda:6'), covar=tensor([0.1033, 0.0861, 0.1048, 0.0541, 0.0568, 0.2541, 0.0843, 0.0635], device='cuda:6'), in_proj_covar=tensor([0.0464, 0.0574, 0.0695, 0.0578, 0.0444, 0.0444, 0.0462, 0.0521], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:20:34,112 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.733e+02 2.617e+02 3.274e+02 4.462e+02 7.495e+02, threshold=6.548e+02, percent-clipped=8.0 2023-04-29 03:20:43,652 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2328, 3.3540, 3.5958, 3.5985, 3.5891, 3.3502, 3.4014, 3.4922], device='cuda:6'), covar=tensor([0.0336, 0.0543, 0.0415, 0.0407, 0.0398, 0.0427, 0.0737, 0.0373], device='cuda:6'), in_proj_covar=tensor([0.0297, 0.0294, 0.0303, 0.0293, 0.0340, 0.0318, 0.0414, 0.0257], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-29 03:20:58,616 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90336.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:21:15,847 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2256, 4.0864, 4.2806, 2.0047, 4.5148, 4.5596, 3.3197, 3.4261], device='cuda:6'), covar=tensor([0.0509, 0.0153, 0.0153, 0.1165, 0.0041, 0.0054, 0.0314, 0.0314], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0096, 0.0082, 0.0138, 0.0065, 0.0091, 0.0117, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-29 03:21:34,608 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90351.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:21:35,356 INFO [train.py:904] (6/8) Epoch 9, batch 9150, loss[loss=0.1897, simple_loss=0.2733, pruned_loss=0.05305, over 11825.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2804, pruned_loss=0.04913, over 3051399.17 frames. ], batch size: 248, lr: 7.44e-03, grad_scale: 4.0 2023-04-29 03:22:35,148 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4288, 3.8640, 3.9185, 2.6824, 3.5570, 3.8034, 3.6269, 1.9677], device='cuda:6'), covar=tensor([0.0368, 0.0029, 0.0036, 0.0286, 0.0072, 0.0081, 0.0058, 0.0451], device='cuda:6'), in_proj_covar=tensor([0.0123, 0.0063, 0.0064, 0.0120, 0.0071, 0.0080, 0.0071, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 03:22:45,425 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90384.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:22:59,693 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.93 vs. limit=5.0 2023-04-29 03:23:22,748 INFO [train.py:904] (6/8) Epoch 9, batch 9200, loss[loss=0.1694, simple_loss=0.2515, pruned_loss=0.04363, over 11942.00 frames. ], tot_loss[loss=0.1858, simple_loss=0.2758, pruned_loss=0.04793, over 3061708.84 frames. ], batch size: 247, lr: 7.44e-03, grad_scale: 8.0 2023-04-29 03:23:40,240 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2888, 2.3324, 1.9765, 2.1827, 2.7770, 2.5170, 3.2242, 2.9606], device='cuda:6'), covar=tensor([0.0073, 0.0302, 0.0363, 0.0342, 0.0198, 0.0257, 0.0116, 0.0175], device='cuda:6'), in_proj_covar=tensor([0.0119, 0.0188, 0.0186, 0.0185, 0.0185, 0.0188, 0.0179, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:23:42,450 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90412.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:24:07,624 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.818e+02 2.608e+02 2.952e+02 3.728e+02 7.219e+02, threshold=5.904e+02, percent-clipped=3.0 2023-04-29 03:24:58,585 INFO [train.py:904] (6/8) Epoch 9, batch 9250, loss[loss=0.1782, simple_loss=0.271, pruned_loss=0.04271, over 16436.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2753, pruned_loss=0.04781, over 3055860.27 frames. ], batch size: 147, lr: 7.44e-03, grad_scale: 8.0 2023-04-29 03:25:14,816 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90459.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:26:08,299 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-29 03:26:13,012 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90486.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:26:38,475 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90496.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:26:49,728 INFO [train.py:904] (6/8) Epoch 9, batch 9300, loss[loss=0.1623, simple_loss=0.2549, pruned_loss=0.0349, over 16583.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2736, pruned_loss=0.04701, over 3054448.89 frames. ], batch size: 57, lr: 7.43e-03, grad_scale: 8.0 2023-04-29 03:27:33,264 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4310, 3.3785, 3.4767, 3.5761, 3.6229, 3.2878, 3.5740, 3.6421], device='cuda:6'), covar=tensor([0.1095, 0.0820, 0.0981, 0.0570, 0.0534, 0.2171, 0.0845, 0.0597], device='cuda:6'), in_proj_covar=tensor([0.0456, 0.0561, 0.0679, 0.0568, 0.0438, 0.0436, 0.0455, 0.0511], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:27:35,130 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90520.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:27:36,965 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90521.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 03:27:46,913 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.664e+02 2.420e+02 2.895e+02 3.675e+02 7.033e+02, threshold=5.790e+02, percent-clipped=1.0 2023-04-29 03:28:36,083 INFO [train.py:904] (6/8) Epoch 9, batch 9350, loss[loss=0.195, simple_loss=0.2746, pruned_loss=0.05766, over 12134.00 frames. ], tot_loss[loss=0.1841, simple_loss=0.2736, pruned_loss=0.04729, over 3036073.65 frames. ], batch size: 248, lr: 7.43e-03, grad_scale: 8.0 2023-04-29 03:28:48,568 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90557.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:29:14,437 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90569.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:29:42,506 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-04-29 03:29:50,775 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90588.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:30:18,677 INFO [train.py:904] (6/8) Epoch 9, batch 9400, loss[loss=0.1704, simple_loss=0.2775, pruned_loss=0.03163, over 16856.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2737, pruned_loss=0.04704, over 3038239.57 frames. ], batch size: 96, lr: 7.43e-03, grad_scale: 8.0 2023-04-29 03:31:09,564 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 2.474e+02 2.959e+02 3.714e+02 8.908e+02, threshold=5.918e+02, percent-clipped=5.0 2023-04-29 03:32:00,895 INFO [train.py:904] (6/8) Epoch 9, batch 9450, loss[loss=0.1963, simple_loss=0.2897, pruned_loss=0.05148, over 16185.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2754, pruned_loss=0.04727, over 3037490.77 frames. ], batch size: 165, lr: 7.43e-03, grad_scale: 8.0 2023-04-29 03:32:21,309 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7565, 3.4985, 3.3459, 1.9160, 2.7586, 2.4011, 3.0892, 3.3982], device='cuda:6'), covar=tensor([0.0405, 0.0644, 0.0474, 0.1716, 0.0781, 0.0826, 0.0927, 0.0869], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0129, 0.0150, 0.0139, 0.0129, 0.0121, 0.0131, 0.0140], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 03:33:43,967 INFO [train.py:904] (6/8) Epoch 9, batch 9500, loss[loss=0.1714, simple_loss=0.2669, pruned_loss=0.03797, over 15281.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2741, pruned_loss=0.04683, over 3034160.94 frames. ], batch size: 191, lr: 7.43e-03, grad_scale: 8.0 2023-04-29 03:33:58,715 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90707.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:34:05,053 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.31 vs. limit=5.0 2023-04-29 03:34:07,639 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0599, 3.4770, 3.5480, 2.3308, 3.1642, 3.4126, 3.3441, 2.0487], device='cuda:6'), covar=tensor([0.0399, 0.0027, 0.0035, 0.0269, 0.0074, 0.0075, 0.0049, 0.0337], device='cuda:6'), in_proj_covar=tensor([0.0123, 0.0062, 0.0064, 0.0121, 0.0072, 0.0081, 0.0071, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 03:34:35,314 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.646e+02 2.298e+02 2.954e+02 3.660e+02 6.291e+02, threshold=5.908e+02, percent-clipped=3.0 2023-04-29 03:35:30,037 INFO [train.py:904] (6/8) Epoch 9, batch 9550, loss[loss=0.2071, simple_loss=0.3004, pruned_loss=0.05687, over 16658.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2752, pruned_loss=0.04728, over 3060823.40 frames. ], batch size: 134, lr: 7.42e-03, grad_scale: 8.0 2023-04-29 03:35:49,807 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0225, 4.0258, 3.8735, 3.3857, 3.9162, 1.6906, 3.7225, 3.5668], device='cuda:6'), covar=tensor([0.0084, 0.0074, 0.0136, 0.0252, 0.0084, 0.2327, 0.0108, 0.0199], device='cuda:6'), in_proj_covar=tensor([0.0111, 0.0099, 0.0144, 0.0133, 0.0115, 0.0165, 0.0130, 0.0135], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-04-29 03:36:43,091 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90786.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:36:57,656 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8439, 3.8277, 4.2427, 4.2378, 4.2204, 3.8956, 3.9729, 3.8806], device='cuda:6'), covar=tensor([0.0311, 0.0526, 0.0439, 0.0435, 0.0420, 0.0388, 0.0851, 0.0436], device='cuda:6'), in_proj_covar=tensor([0.0293, 0.0290, 0.0297, 0.0288, 0.0335, 0.0313, 0.0405, 0.0253], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-29 03:37:11,179 INFO [train.py:904] (6/8) Epoch 9, batch 9600, loss[loss=0.2387, simple_loss=0.3273, pruned_loss=0.07504, over 15439.00 frames. ], tot_loss[loss=0.1859, simple_loss=0.2761, pruned_loss=0.04779, over 3055622.86 frames. ], batch size: 191, lr: 7.42e-03, grad_scale: 8.0 2023-04-29 03:37:37,096 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90815.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:37:59,119 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 2.542e+02 3.036e+02 4.023e+02 8.440e+02, threshold=6.073e+02, percent-clipped=4.0 2023-04-29 03:38:08,227 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 03:38:13,850 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-04-29 03:38:17,844 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90834.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:38:55,254 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90849.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:39:01,498 INFO [train.py:904] (6/8) Epoch 9, batch 9650, loss[loss=0.2033, simple_loss=0.2952, pruned_loss=0.05573, over 16394.00 frames. ], tot_loss[loss=0.1869, simple_loss=0.2778, pruned_loss=0.04803, over 3056363.09 frames. ], batch size: 146, lr: 7.42e-03, grad_scale: 8.0 2023-04-29 03:39:03,589 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90852.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:39:55,210 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-29 03:40:19,165 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90888.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:40:48,115 INFO [train.py:904] (6/8) Epoch 9, batch 9700, loss[loss=0.1693, simple_loss=0.2639, pruned_loss=0.03732, over 16739.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2768, pruned_loss=0.04776, over 3051109.37 frames. ], batch size: 76, lr: 7.42e-03, grad_scale: 8.0 2023-04-29 03:41:04,648 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90910.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:41:40,521 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.339e+02 3.062e+02 3.716e+02 7.920e+02, threshold=6.123e+02, percent-clipped=1.0 2023-04-29 03:42:00,639 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90936.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:42:31,486 INFO [train.py:904] (6/8) Epoch 9, batch 9750, loss[loss=0.1893, simple_loss=0.2839, pruned_loss=0.04736, over 16211.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2752, pruned_loss=0.04763, over 3069735.77 frames. ], batch size: 165, lr: 7.42e-03, grad_scale: 8.0 2023-04-29 03:44:10,866 INFO [train.py:904] (6/8) Epoch 9, batch 9800, loss[loss=0.1569, simple_loss=0.2414, pruned_loss=0.03613, over 12296.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2746, pruned_loss=0.04638, over 3071398.27 frames. ], batch size: 250, lr: 7.41e-03, grad_scale: 8.0 2023-04-29 03:44:21,823 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91007.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:44:57,913 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.670e+02 2.313e+02 2.751e+02 3.431e+02 5.847e+02, threshold=5.502e+02, percent-clipped=0.0 2023-04-29 03:45:57,974 INFO [train.py:904] (6/8) Epoch 9, batch 9850, loss[loss=0.1752, simple_loss=0.2715, pruned_loss=0.03944, over 16057.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2757, pruned_loss=0.04581, over 3071479.37 frames. ], batch size: 35, lr: 7.41e-03, grad_scale: 8.0 2023-04-29 03:46:05,839 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=91055.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:46:14,351 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0543, 5.0058, 4.8726, 4.5692, 4.5478, 4.9524, 4.8836, 4.5137], device='cuda:6'), covar=tensor([0.0480, 0.0413, 0.0223, 0.0237, 0.0836, 0.0359, 0.0243, 0.0597], device='cuda:6'), in_proj_covar=tensor([0.0211, 0.0254, 0.0247, 0.0227, 0.0265, 0.0255, 0.0169, 0.0286], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-04-29 03:47:48,699 INFO [train.py:904] (6/8) Epoch 9, batch 9900, loss[loss=0.1806, simple_loss=0.2642, pruned_loss=0.04847, over 12727.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.2766, pruned_loss=0.04611, over 3067826.63 frames. ], batch size: 248, lr: 7.41e-03, grad_scale: 8.0 2023-04-29 03:48:20,569 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91115.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:48:47,428 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1745, 2.1752, 1.8602, 1.9267, 2.6226, 2.2333, 2.9525, 2.8584], device='cuda:6'), covar=tensor([0.0072, 0.0350, 0.0439, 0.0371, 0.0226, 0.0308, 0.0154, 0.0195], device='cuda:6'), in_proj_covar=tensor([0.0117, 0.0189, 0.0183, 0.0183, 0.0184, 0.0186, 0.0177, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:48:47,932 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 2.434e+02 3.232e+02 4.008e+02 9.044e+02, threshold=6.464e+02, percent-clipped=5.0 2023-04-29 03:49:47,990 INFO [train.py:904] (6/8) Epoch 9, batch 9950, loss[loss=0.1949, simple_loss=0.2807, pruned_loss=0.05455, over 12457.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2791, pruned_loss=0.04666, over 3073194.17 frames. ], batch size: 248, lr: 7.41e-03, grad_scale: 8.0 2023-04-29 03:49:49,254 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91152.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:50:07,661 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3771, 3.3440, 2.7159, 2.1312, 2.2744, 2.2638, 3.3917, 3.1868], device='cuda:6'), covar=tensor([0.2598, 0.0569, 0.1500, 0.2264, 0.1840, 0.1545, 0.0429, 0.0794], device='cuda:6'), in_proj_covar=tensor([0.0285, 0.0241, 0.0266, 0.0258, 0.0243, 0.0205, 0.0250, 0.0261], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:50:14,537 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=91163.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:50:30,352 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 03:51:44,349 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=91200.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:51:47,243 INFO [train.py:904] (6/8) Epoch 9, batch 10000, loss[loss=0.1856, simple_loss=0.2725, pruned_loss=0.04937, over 12860.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.278, pruned_loss=0.0467, over 3079421.17 frames. ], batch size: 248, lr: 7.41e-03, grad_scale: 8.0 2023-04-29 03:51:47,882 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3403, 1.9614, 1.6681, 1.7537, 2.2421, 1.9339, 2.1380, 2.3539], device='cuda:6'), covar=tensor([0.0074, 0.0244, 0.0318, 0.0316, 0.0138, 0.0244, 0.0129, 0.0142], device='cuda:6'), in_proj_covar=tensor([0.0118, 0.0190, 0.0184, 0.0185, 0.0184, 0.0188, 0.0178, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:51:53,872 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91205.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:52:35,744 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.574e+02 2.985e+02 3.622e+02 7.238e+02, threshold=5.969e+02, percent-clipped=2.0 2023-04-29 03:53:27,231 INFO [train.py:904] (6/8) Epoch 9, batch 10050, loss[loss=0.1674, simple_loss=0.2581, pruned_loss=0.03836, over 12443.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2776, pruned_loss=0.04635, over 3059662.40 frames. ], batch size: 248, lr: 7.40e-03, grad_scale: 8.0 2023-04-29 03:53:30,498 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 03:54:30,781 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91285.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:55:01,275 INFO [train.py:904] (6/8) Epoch 9, batch 10100, loss[loss=0.1743, simple_loss=0.2612, pruned_loss=0.04367, over 16351.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2778, pruned_loss=0.04648, over 3075307.51 frames. ], batch size: 146, lr: 7.40e-03, grad_scale: 8.0 2023-04-29 03:55:08,676 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-29 03:55:18,426 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9065, 3.3870, 3.5187, 1.9466, 2.7772, 2.3426, 3.2918, 3.4278], device='cuda:6'), covar=tensor([0.0316, 0.0697, 0.0442, 0.1759, 0.0821, 0.0925, 0.0772, 0.0920], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0129, 0.0154, 0.0140, 0.0131, 0.0123, 0.0131, 0.0140], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 03:55:25,695 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-04-29 03:55:52,059 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.558e+02 3.237e+02 3.853e+02 8.628e+02, threshold=6.474e+02, percent-clipped=2.0 2023-04-29 03:56:14,323 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91346.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:56:45,261 INFO [train.py:904] (6/8) Epoch 10, batch 0, loss[loss=0.1947, simple_loss=0.2757, pruned_loss=0.05687, over 16982.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.2757, pruned_loss=0.05687, over 16982.00 frames. ], batch size: 41, lr: 7.04e-03, grad_scale: 8.0 2023-04-29 03:56:45,262 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 03:56:50,410 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5321, 4.8992, 4.6539, 4.7740, 4.4789, 4.6148, 4.3769, 4.9082], device='cuda:6'), covar=tensor([0.0887, 0.0735, 0.0747, 0.0549, 0.0769, 0.0378, 0.0939, 0.0717], device='cuda:6'), in_proj_covar=tensor([0.0467, 0.0596, 0.0485, 0.0407, 0.0373, 0.0388, 0.0496, 0.0443], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:56:52,894 INFO [train.py:938] (6/8) Epoch 10, validation: loss=0.158, simple_loss=0.2614, pruned_loss=0.02732, over 944034.00 frames. 2023-04-29 03:56:52,895 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 03:57:32,661 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8935, 1.7336, 2.1871, 2.7470, 2.7074, 2.6945, 1.9111, 3.0049], device='cuda:6'), covar=tensor([0.0107, 0.0318, 0.0216, 0.0152, 0.0150, 0.0151, 0.0336, 0.0065], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0164, 0.0149, 0.0148, 0.0158, 0.0114, 0.0166, 0.0101], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 03:57:52,215 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0129, 2.2410, 2.3276, 4.8033, 2.1154, 2.7957, 2.3532, 2.4883], device='cuda:6'), covar=tensor([0.0648, 0.3034, 0.1975, 0.0257, 0.3537, 0.1962, 0.2676, 0.2921], device='cuda:6'), in_proj_covar=tensor([0.0335, 0.0361, 0.0307, 0.0304, 0.0390, 0.0401, 0.0325, 0.0418], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:58:02,501 INFO [train.py:904] (6/8) Epoch 10, batch 50, loss[loss=0.2357, simple_loss=0.3113, pruned_loss=0.08003, over 15497.00 frames. ], tot_loss[loss=0.2196, simple_loss=0.2961, pruned_loss=0.0716, over 749493.73 frames. ], batch size: 190, lr: 7.04e-03, grad_scale: 2.0 2023-04-29 03:58:07,526 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-29 03:58:09,123 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9675, 2.1777, 2.3261, 4.7293, 2.0826, 2.6224, 2.3167, 2.4164], device='cuda:6'), covar=tensor([0.0654, 0.3126, 0.2035, 0.0293, 0.3817, 0.2223, 0.2669, 0.3325], device='cuda:6'), in_proj_covar=tensor([0.0337, 0.0362, 0.0308, 0.0305, 0.0392, 0.0403, 0.0326, 0.0419], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 03:58:17,021 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9715, 4.9000, 5.4233, 5.4506, 5.4393, 5.0977, 4.9847, 4.8020], device='cuda:6'), covar=tensor([0.0259, 0.0413, 0.0411, 0.0406, 0.0384, 0.0289, 0.0937, 0.0391], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0293, 0.0297, 0.0283, 0.0331, 0.0313, 0.0404, 0.0252], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-29 03:58:39,941 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.984e+02 3.603e+02 4.569e+02 8.591e+02, threshold=7.207e+02, percent-clipped=1.0 2023-04-29 03:59:08,801 INFO [train.py:904] (6/8) Epoch 10, batch 100, loss[loss=0.1675, simple_loss=0.2547, pruned_loss=0.04015, over 17233.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.29, pruned_loss=0.06617, over 1315009.12 frames. ], batch size: 44, lr: 7.03e-03, grad_scale: 1.0 2023-04-29 03:59:18,103 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8382, 1.2601, 1.6242, 1.6223, 1.7815, 1.8375, 1.5284, 1.8539], device='cuda:6'), covar=tensor([0.0163, 0.0263, 0.0152, 0.0188, 0.0166, 0.0156, 0.0282, 0.0079], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0166, 0.0150, 0.0150, 0.0160, 0.0115, 0.0167, 0.0102], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 04:00:16,885 INFO [train.py:904] (6/8) Epoch 10, batch 150, loss[loss=0.1705, simple_loss=0.2515, pruned_loss=0.04474, over 16220.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2851, pruned_loss=0.06351, over 1759549.61 frames. ], batch size: 36, lr: 7.03e-03, grad_scale: 1.0 2023-04-29 04:00:22,942 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91505.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:00:56,394 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.669e+02 2.683e+02 3.349e+02 4.064e+02 6.042e+02, threshold=6.698e+02, percent-clipped=0.0 2023-04-29 04:00:59,634 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5781, 4.8932, 4.6658, 4.6289, 4.3233, 4.4090, 4.4332, 4.9825], device='cuda:6'), covar=tensor([0.1042, 0.0897, 0.1141, 0.0689, 0.0863, 0.1006, 0.0919, 0.0879], device='cuda:6'), in_proj_covar=tensor([0.0501, 0.0634, 0.0520, 0.0432, 0.0396, 0.0410, 0.0529, 0.0468], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:01:12,390 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91542.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 04:01:26,562 INFO [train.py:904] (6/8) Epoch 10, batch 200, loss[loss=0.2432, simple_loss=0.2953, pruned_loss=0.09554, over 16913.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2843, pruned_loss=0.0627, over 2102624.79 frames. ], batch size: 109, lr: 7.03e-03, grad_scale: 1.0 2023-04-29 04:01:28,065 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=91553.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:01:32,461 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8523, 4.7860, 5.3306, 5.3016, 5.3406, 4.9753, 4.9476, 4.6767], device='cuda:6'), covar=tensor([0.0244, 0.0396, 0.0325, 0.0384, 0.0338, 0.0276, 0.0750, 0.0358], device='cuda:6'), in_proj_covar=tensor([0.0306, 0.0309, 0.0311, 0.0297, 0.0347, 0.0331, 0.0425, 0.0264], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-29 04:02:34,674 INFO [train.py:904] (6/8) Epoch 10, batch 250, loss[loss=0.2022, simple_loss=0.2777, pruned_loss=0.06337, over 12259.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2821, pruned_loss=0.06153, over 2371190.81 frames. ], batch size: 247, lr: 7.03e-03, grad_scale: 1.0 2023-04-29 04:02:36,468 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91603.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 04:02:37,967 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-29 04:02:58,316 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6979, 2.1241, 2.2228, 4.5183, 2.0671, 2.6345, 2.3043, 2.3264], device='cuda:6'), covar=tensor([0.0799, 0.3332, 0.2192, 0.0330, 0.3754, 0.2288, 0.2826, 0.3280], device='cuda:6'), in_proj_covar=tensor([0.0350, 0.0375, 0.0318, 0.0319, 0.0405, 0.0420, 0.0338, 0.0438], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:03:07,004 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0189, 4.9736, 5.5278, 5.5486, 5.5235, 5.1334, 5.1411, 4.8813], device='cuda:6'), covar=tensor([0.0308, 0.0450, 0.0348, 0.0338, 0.0394, 0.0276, 0.0837, 0.0339], device='cuda:6'), in_proj_covar=tensor([0.0310, 0.0314, 0.0314, 0.0302, 0.0354, 0.0335, 0.0433, 0.0268], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 04:03:11,342 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.913e+02 2.504e+02 3.085e+02 3.691e+02 6.376e+02, threshold=6.169e+02, percent-clipped=0.0 2023-04-29 04:03:28,097 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91641.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:03:42,030 INFO [train.py:904] (6/8) Epoch 10, batch 300, loss[loss=0.1855, simple_loss=0.259, pruned_loss=0.05597, over 16209.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2782, pruned_loss=0.05969, over 2584911.73 frames. ], batch size: 165, lr: 7.03e-03, grad_scale: 1.0 2023-04-29 04:04:01,941 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6272, 3.6992, 2.8693, 2.3393, 2.6096, 2.2792, 3.7827, 3.4962], device='cuda:6'), covar=tensor([0.2431, 0.0606, 0.1450, 0.2251, 0.2102, 0.1726, 0.0501, 0.1191], device='cuda:6'), in_proj_covar=tensor([0.0293, 0.0250, 0.0276, 0.0267, 0.0259, 0.0213, 0.0260, 0.0281], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:04:39,250 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1904, 1.3548, 1.8506, 2.1077, 2.2095, 2.1996, 1.5997, 2.2079], device='cuda:6'), covar=tensor([0.0175, 0.0358, 0.0207, 0.0184, 0.0181, 0.0187, 0.0320, 0.0087], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0167, 0.0151, 0.0152, 0.0161, 0.0118, 0.0167, 0.0104], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 04:04:40,905 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7240, 3.7454, 4.0699, 4.0571, 4.0596, 3.7591, 3.8362, 3.8152], device='cuda:6'), covar=tensor([0.0367, 0.0552, 0.0385, 0.0380, 0.0417, 0.0407, 0.0775, 0.0456], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0315, 0.0317, 0.0303, 0.0357, 0.0337, 0.0435, 0.0271], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 04:04:51,283 INFO [train.py:904] (6/8) Epoch 10, batch 350, loss[loss=0.2055, simple_loss=0.2903, pruned_loss=0.06039, over 16766.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2758, pruned_loss=0.05817, over 2742267.41 frames. ], batch size: 62, lr: 7.02e-03, grad_scale: 1.0 2023-04-29 04:05:02,685 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-29 04:05:04,688 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4308, 3.7459, 3.7807, 2.0539, 3.0889, 2.5175, 3.8101, 3.7955], device='cuda:6'), covar=tensor([0.0293, 0.0698, 0.0520, 0.1667, 0.0705, 0.0919, 0.0566, 0.1049], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0135, 0.0155, 0.0140, 0.0132, 0.0124, 0.0133, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 04:05:25,463 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0206, 5.2490, 4.9664, 4.6585, 4.0870, 5.1115, 5.2262, 4.5943], device='cuda:6'), covar=tensor([0.0940, 0.0439, 0.0433, 0.0376, 0.2028, 0.0458, 0.0235, 0.0698], device='cuda:6'), in_proj_covar=tensor([0.0236, 0.0286, 0.0273, 0.0251, 0.0295, 0.0287, 0.0187, 0.0319], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:05:28,619 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.553e+02 2.384e+02 2.940e+02 3.585e+02 5.710e+02, threshold=5.881e+02, percent-clipped=0.0 2023-04-29 04:05:59,423 INFO [train.py:904] (6/8) Epoch 10, batch 400, loss[loss=0.1599, simple_loss=0.2399, pruned_loss=0.04001, over 16784.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2736, pruned_loss=0.05709, over 2877566.55 frames. ], batch size: 39, lr: 7.02e-03, grad_scale: 2.0 2023-04-29 04:07:11,414 INFO [train.py:904] (6/8) Epoch 10, batch 450, loss[loss=0.2173, simple_loss=0.2833, pruned_loss=0.0757, over 16714.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.2714, pruned_loss=0.05598, over 2973512.40 frames. ], batch size: 134, lr: 7.02e-03, grad_scale: 2.0 2023-04-29 04:07:46,477 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6375, 3.7008, 2.9026, 2.1921, 2.4190, 2.2601, 3.7176, 3.3233], device='cuda:6'), covar=tensor([0.2307, 0.0544, 0.1348, 0.2289, 0.2280, 0.1613, 0.0474, 0.1069], device='cuda:6'), in_proj_covar=tensor([0.0297, 0.0253, 0.0279, 0.0271, 0.0265, 0.0216, 0.0264, 0.0286], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:07:50,758 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.774e+02 2.266e+02 2.904e+02 3.599e+02 6.333e+02, threshold=5.808e+02, percent-clipped=1.0 2023-04-29 04:07:57,036 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1181, 2.0165, 2.4325, 2.9925, 2.9978, 3.4760, 2.2166, 3.3697], device='cuda:6'), covar=tensor([0.0145, 0.0323, 0.0217, 0.0194, 0.0158, 0.0120, 0.0303, 0.0102], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0169, 0.0153, 0.0154, 0.0163, 0.0120, 0.0169, 0.0106], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 04:07:58,131 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0995, 4.7927, 4.9950, 5.2917, 5.4707, 4.7432, 5.4474, 5.4076], device='cuda:6'), covar=tensor([0.1249, 0.1067, 0.1450, 0.0598, 0.0435, 0.0721, 0.0437, 0.0475], device='cuda:6'), in_proj_covar=tensor([0.0515, 0.0628, 0.0765, 0.0633, 0.0483, 0.0482, 0.0506, 0.0567], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:08:20,248 INFO [train.py:904] (6/8) Epoch 10, batch 500, loss[loss=0.2074, simple_loss=0.2741, pruned_loss=0.07032, over 16800.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.2695, pruned_loss=0.05448, over 3050447.57 frames. ], batch size: 124, lr: 7.02e-03, grad_scale: 2.0 2023-04-29 04:08:31,956 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1405, 3.3105, 3.1926, 2.0403, 2.6905, 2.3805, 3.5151, 3.4519], device='cuda:6'), covar=tensor([0.0189, 0.0649, 0.0525, 0.1530, 0.0739, 0.0869, 0.0441, 0.0811], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0136, 0.0155, 0.0140, 0.0133, 0.0124, 0.0134, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 04:09:15,093 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91891.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:09:23,881 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91898.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 04:09:29,464 INFO [train.py:904] (6/8) Epoch 10, batch 550, loss[loss=0.1625, simple_loss=0.2534, pruned_loss=0.03577, over 17201.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2687, pruned_loss=0.05359, over 3106266.91 frames. ], batch size: 44, lr: 7.02e-03, grad_scale: 2.0 2023-04-29 04:10:07,828 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 2.517e+02 3.166e+02 3.975e+02 1.037e+03, threshold=6.333e+02, percent-clipped=5.0 2023-04-29 04:10:22,735 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91941.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:10:37,528 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-29 04:10:38,342 INFO [train.py:904] (6/8) Epoch 10, batch 600, loss[loss=0.1565, simple_loss=0.2446, pruned_loss=0.03419, over 17236.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2682, pruned_loss=0.05465, over 3155719.21 frames. ], batch size: 45, lr: 7.01e-03, grad_scale: 2.0 2023-04-29 04:10:38,871 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0474, 4.0153, 2.2994, 4.5129, 2.8436, 4.4909, 2.3631, 3.2961], device='cuda:6'), covar=tensor([0.0169, 0.0304, 0.1453, 0.0185, 0.0803, 0.0411, 0.1394, 0.0544], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0163, 0.0187, 0.0121, 0.0167, 0.0202, 0.0193, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 04:10:38,893 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91952.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:10:42,821 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91955.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:10:55,412 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7975, 2.1324, 2.2224, 4.6123, 2.0178, 2.5976, 2.3158, 2.3213], device='cuda:6'), covar=tensor([0.0816, 0.3478, 0.2207, 0.0312, 0.4078, 0.2302, 0.2895, 0.3598], device='cuda:6'), in_proj_covar=tensor([0.0352, 0.0375, 0.0320, 0.0319, 0.0405, 0.0424, 0.0338, 0.0442], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:11:19,557 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91981.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 04:11:24,844 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6933, 4.2518, 4.3946, 3.2091, 3.8052, 4.2599, 3.9092, 2.4705], device='cuda:6'), covar=tensor([0.0370, 0.0043, 0.0026, 0.0244, 0.0074, 0.0063, 0.0061, 0.0346], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0068, 0.0066, 0.0124, 0.0075, 0.0083, 0.0075, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 04:11:30,063 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=91989.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:11:51,293 INFO [train.py:904] (6/8) Epoch 10, batch 650, loss[loss=0.1786, simple_loss=0.2516, pruned_loss=0.05279, over 16782.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2669, pruned_loss=0.05373, over 3188797.78 frames. ], batch size: 83, lr: 7.01e-03, grad_scale: 2.0 2023-04-29 04:12:12,140 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92016.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:12:25,134 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8901, 3.8528, 2.1474, 4.1285, 2.9720, 4.0812, 2.3095, 3.1464], device='cuda:6'), covar=tensor([0.0163, 0.0295, 0.1499, 0.0207, 0.0615, 0.0537, 0.1285, 0.0520], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0164, 0.0188, 0.0124, 0.0167, 0.0203, 0.0194, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 04:12:31,804 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.739e+02 2.417e+02 3.017e+02 3.569e+02 8.138e+02, threshold=6.033e+02, percent-clipped=1.0 2023-04-29 04:12:49,537 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92042.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 04:13:01,359 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.31 vs. limit=5.0 2023-04-29 04:13:03,077 INFO [train.py:904] (6/8) Epoch 10, batch 700, loss[loss=0.2007, simple_loss=0.2722, pruned_loss=0.06459, over 16454.00 frames. ], tot_loss[loss=0.1859, simple_loss=0.2661, pruned_loss=0.0528, over 3219461.53 frames. ], batch size: 146, lr: 7.01e-03, grad_scale: 2.0 2023-04-29 04:13:30,492 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2236, 4.4872, 4.8173, 4.8113, 4.8135, 4.4984, 4.1261, 4.2379], device='cuda:6'), covar=tensor([0.0642, 0.0847, 0.0622, 0.0647, 0.0605, 0.0593, 0.1484, 0.0653], device='cuda:6'), in_proj_covar=tensor([0.0324, 0.0329, 0.0331, 0.0317, 0.0369, 0.0348, 0.0452, 0.0281], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 04:14:09,234 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8131, 3.0268, 2.8843, 4.8538, 4.0488, 4.6013, 1.5539, 3.2778], device='cuda:6'), covar=tensor([0.1246, 0.0607, 0.1007, 0.0143, 0.0212, 0.0333, 0.1438, 0.0661], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0152, 0.0177, 0.0132, 0.0190, 0.0208, 0.0176, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 04:14:12,201 INFO [train.py:904] (6/8) Epoch 10, batch 750, loss[loss=0.1926, simple_loss=0.2827, pruned_loss=0.05124, over 17017.00 frames. ], tot_loss[loss=0.1876, simple_loss=0.2673, pruned_loss=0.05395, over 3232782.08 frames. ], batch size: 55, lr: 7.01e-03, grad_scale: 2.0 2023-04-29 04:14:34,821 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-29 04:14:52,014 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.724e+02 2.681e+02 3.065e+02 4.087e+02 6.685e+02, threshold=6.130e+02, percent-clipped=5.0 2023-04-29 04:15:22,687 INFO [train.py:904] (6/8) Epoch 10, batch 800, loss[loss=0.1838, simple_loss=0.258, pruned_loss=0.0548, over 16737.00 frames. ], tot_loss[loss=0.1874, simple_loss=0.2673, pruned_loss=0.0537, over 3255095.42 frames. ], batch size: 124, lr: 7.01e-03, grad_scale: 4.0 2023-04-29 04:16:27,559 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92198.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 04:16:32,818 INFO [train.py:904] (6/8) Epoch 10, batch 850, loss[loss=0.2026, simple_loss=0.2696, pruned_loss=0.06776, over 16877.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2669, pruned_loss=0.05321, over 3260413.01 frames. ], batch size: 116, lr: 7.00e-03, grad_scale: 4.0 2023-04-29 04:16:33,627 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 04:17:03,053 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.93 vs. limit=5.0 2023-04-29 04:17:10,135 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.453e+02 2.911e+02 3.766e+02 9.676e+02, threshold=5.821e+02, percent-clipped=2.0 2023-04-29 04:17:33,334 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=92246.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 04:17:34,346 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92247.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:17:41,160 INFO [train.py:904] (6/8) Epoch 10, batch 900, loss[loss=0.1663, simple_loss=0.2523, pruned_loss=0.04015, over 16500.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2663, pruned_loss=0.05292, over 3275116.99 frames. ], batch size: 75, lr: 7.00e-03, grad_scale: 4.0 2023-04-29 04:18:50,837 INFO [train.py:904] (6/8) Epoch 10, batch 950, loss[loss=0.1637, simple_loss=0.2535, pruned_loss=0.03692, over 17241.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2654, pruned_loss=0.05218, over 3288400.12 frames. ], batch size: 44, lr: 7.00e-03, grad_scale: 4.0 2023-04-29 04:19:03,284 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1630, 4.3186, 4.6482, 2.5265, 4.8484, 4.9152, 3.5211, 3.8555], device='cuda:6'), covar=tensor([0.0660, 0.0159, 0.0140, 0.0992, 0.0054, 0.0090, 0.0328, 0.0306], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0098, 0.0085, 0.0141, 0.0069, 0.0099, 0.0120, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 04:19:04,204 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92311.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:19:29,761 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.546e+02 2.335e+02 2.688e+02 3.318e+02 6.424e+02, threshold=5.375e+02, percent-clipped=3.0 2023-04-29 04:19:39,509 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92337.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 04:19:58,791 INFO [train.py:904] (6/8) Epoch 10, batch 1000, loss[loss=0.193, simple_loss=0.2617, pruned_loss=0.06221, over 16365.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2643, pruned_loss=0.05208, over 3300770.62 frames. ], batch size: 145, lr: 7.00e-03, grad_scale: 4.0 2023-04-29 04:21:09,129 INFO [train.py:904] (6/8) Epoch 10, batch 1050, loss[loss=0.1977, simple_loss=0.2679, pruned_loss=0.06372, over 16854.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2645, pruned_loss=0.05285, over 3301229.88 frames. ], batch size: 102, lr: 7.00e-03, grad_scale: 4.0 2023-04-29 04:21:11,953 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5735, 5.9731, 5.7214, 5.8195, 5.3595, 5.2341, 5.4302, 6.1010], device='cuda:6'), covar=tensor([0.1101, 0.0924, 0.0987, 0.0612, 0.0811, 0.0587, 0.0928, 0.0815], device='cuda:6'), in_proj_covar=tensor([0.0526, 0.0669, 0.0551, 0.0459, 0.0418, 0.0426, 0.0563, 0.0505], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:21:42,264 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7190, 4.8190, 4.9038, 4.8343, 4.8273, 5.3696, 4.9262, 4.6200], device='cuda:6'), covar=tensor([0.1215, 0.1638, 0.1779, 0.1815, 0.2444, 0.1021, 0.1306, 0.2232], device='cuda:6'), in_proj_covar=tensor([0.0343, 0.0494, 0.0519, 0.0423, 0.0566, 0.0539, 0.0420, 0.0569], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 04:21:48,359 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.641e+02 2.612e+02 3.041e+02 3.664e+02 7.677e+02, threshold=6.083e+02, percent-clipped=4.0 2023-04-29 04:22:18,523 INFO [train.py:904] (6/8) Epoch 10, batch 1100, loss[loss=0.2044, simple_loss=0.2797, pruned_loss=0.06457, over 16736.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2638, pruned_loss=0.05276, over 3299453.97 frames. ], batch size: 134, lr: 7.00e-03, grad_scale: 4.0 2023-04-29 04:23:28,210 INFO [train.py:904] (6/8) Epoch 10, batch 1150, loss[loss=0.1548, simple_loss=0.2381, pruned_loss=0.03572, over 16989.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2631, pruned_loss=0.05203, over 3300749.01 frames. ], batch size: 41, lr: 6.99e-03, grad_scale: 4.0 2023-04-29 04:24:08,394 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.186e+02 2.590e+02 3.516e+02 6.121e+02, threshold=5.179e+02, percent-clipped=1.0 2023-04-29 04:24:32,806 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92547.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:24:38,818 INFO [train.py:904] (6/8) Epoch 10, batch 1200, loss[loss=0.1687, simple_loss=0.2628, pruned_loss=0.03735, over 16651.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2632, pruned_loss=0.05197, over 3306419.60 frames. ], batch size: 57, lr: 6.99e-03, grad_scale: 8.0 2023-04-29 04:25:06,515 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7984, 4.8475, 5.4238, 5.3667, 5.3467, 4.9880, 4.8996, 4.6444], device='cuda:6'), covar=tensor([0.0287, 0.0451, 0.0338, 0.0425, 0.0366, 0.0317, 0.0804, 0.0385], device='cuda:6'), in_proj_covar=tensor([0.0335, 0.0339, 0.0338, 0.0326, 0.0380, 0.0356, 0.0463, 0.0286], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 04:25:11,133 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9031, 4.3752, 4.5307, 3.2726, 3.8814, 4.2925, 4.0524, 2.8255], device='cuda:6'), covar=tensor([0.0326, 0.0043, 0.0026, 0.0240, 0.0069, 0.0064, 0.0052, 0.0308], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0070, 0.0068, 0.0125, 0.0077, 0.0084, 0.0075, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 04:25:39,101 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=92595.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:25:47,778 INFO [train.py:904] (6/8) Epoch 10, batch 1250, loss[loss=0.1978, simple_loss=0.281, pruned_loss=0.05732, over 16696.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2634, pruned_loss=0.05254, over 3307165.48 frames. ], batch size: 62, lr: 6.99e-03, grad_scale: 8.0 2023-04-29 04:26:01,774 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92611.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:26:21,289 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92625.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:26:27,270 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.554e+02 2.910e+02 3.549e+02 5.849e+02, threshold=5.821e+02, percent-clipped=4.0 2023-04-29 04:26:37,467 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92637.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 04:26:58,797 INFO [train.py:904] (6/8) Epoch 10, batch 1300, loss[loss=0.1887, simple_loss=0.2601, pruned_loss=0.05864, over 16511.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2633, pruned_loss=0.05217, over 3309680.05 frames. ], batch size: 75, lr: 6.99e-03, grad_scale: 8.0 2023-04-29 04:27:07,466 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=92659.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:27:44,065 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=92685.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 04:27:45,934 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92686.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:28:07,566 INFO [train.py:904] (6/8) Epoch 10, batch 1350, loss[loss=0.1953, simple_loss=0.2826, pruned_loss=0.05406, over 17112.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2643, pruned_loss=0.05209, over 3318242.29 frames. ], batch size: 53, lr: 6.99e-03, grad_scale: 8.0 2023-04-29 04:28:34,018 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92720.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:28:47,042 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.580e+02 3.033e+02 3.471e+02 5.594e+02, threshold=6.066e+02, percent-clipped=0.0 2023-04-29 04:29:18,974 INFO [train.py:904] (6/8) Epoch 10, batch 1400, loss[loss=0.1784, simple_loss=0.255, pruned_loss=0.05094, over 15514.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2635, pruned_loss=0.0515, over 3302791.11 frames. ], batch size: 190, lr: 6.98e-03, grad_scale: 8.0 2023-04-29 04:30:00,651 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92781.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:30:06,549 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8108, 2.2015, 2.2409, 4.5532, 2.0421, 2.7670, 2.3078, 2.4650], device='cuda:6'), covar=tensor([0.0813, 0.3201, 0.2240, 0.0343, 0.3750, 0.2137, 0.2873, 0.3284], device='cuda:6'), in_proj_covar=tensor([0.0358, 0.0384, 0.0325, 0.0323, 0.0408, 0.0435, 0.0345, 0.0453], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:30:28,851 INFO [train.py:904] (6/8) Epoch 10, batch 1450, loss[loss=0.1803, simple_loss=0.2612, pruned_loss=0.0497, over 15459.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2626, pruned_loss=0.05098, over 3308011.31 frames. ], batch size: 190, lr: 6.98e-03, grad_scale: 8.0 2023-04-29 04:30:34,694 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4324, 3.4898, 2.0350, 3.7090, 2.5702, 3.5879, 2.1224, 2.7530], device='cuda:6'), covar=tensor([0.0217, 0.0411, 0.1387, 0.0197, 0.0785, 0.0744, 0.1258, 0.0622], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0168, 0.0190, 0.0126, 0.0168, 0.0207, 0.0196, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 04:31:07,980 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.451e+02 2.441e+02 3.023e+02 3.567e+02 6.383e+02, threshold=6.046e+02, percent-clipped=1.0 2023-04-29 04:31:09,670 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92831.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:31:38,221 INFO [train.py:904] (6/8) Epoch 10, batch 1500, loss[loss=0.2143, simple_loss=0.2848, pruned_loss=0.07191, over 16904.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2635, pruned_loss=0.05221, over 3317159.96 frames. ], batch size: 116, lr: 6.98e-03, grad_scale: 8.0 2023-04-29 04:31:42,969 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-29 04:32:34,152 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92892.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:32:45,590 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92899.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:32:48,539 INFO [train.py:904] (6/8) Epoch 10, batch 1550, loss[loss=0.2039, simple_loss=0.2785, pruned_loss=0.06468, over 16308.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2647, pruned_loss=0.05256, over 3324285.71 frames. ], batch size: 165, lr: 6.98e-03, grad_scale: 8.0 2023-04-29 04:33:26,143 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.809e+02 2.944e+02 3.490e+02 4.097e+02 8.318e+02, threshold=6.980e+02, percent-clipped=5.0 2023-04-29 04:33:28,357 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92931.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:33:56,442 INFO [train.py:904] (6/8) Epoch 10, batch 1600, loss[loss=0.2038, simple_loss=0.2711, pruned_loss=0.0682, over 16863.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2659, pruned_loss=0.05323, over 3329721.18 frames. ], batch size: 109, lr: 6.98e-03, grad_scale: 8.0 2023-04-29 04:34:08,408 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92960.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:34:17,378 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2218, 3.4377, 3.7966, 2.6788, 3.3885, 3.7890, 3.6791, 2.3299], device='cuda:6'), covar=tensor([0.0391, 0.0157, 0.0039, 0.0261, 0.0088, 0.0073, 0.0055, 0.0316], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0070, 0.0068, 0.0124, 0.0076, 0.0084, 0.0074, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 04:34:29,046 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92975.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:34:29,155 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2184, 1.9746, 2.2696, 3.8344, 2.0280, 2.4341, 2.1054, 2.2184], device='cuda:6'), covar=tensor([0.0960, 0.3262, 0.2000, 0.0405, 0.3364, 0.2096, 0.2948, 0.2758], device='cuda:6'), in_proj_covar=tensor([0.0358, 0.0383, 0.0325, 0.0323, 0.0407, 0.0435, 0.0344, 0.0451], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:34:37,947 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92981.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:34:52,588 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92992.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:35:06,231 INFO [train.py:904] (6/8) Epoch 10, batch 1650, loss[loss=0.1747, simple_loss=0.2656, pruned_loss=0.04188, over 17129.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2678, pruned_loss=0.0539, over 3323577.83 frames. ], batch size: 48, lr: 6.97e-03, grad_scale: 8.0 2023-04-29 04:35:18,213 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93010.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:35:21,410 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 04:35:45,203 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.722e+02 2.314e+02 2.824e+02 3.373e+02 5.658e+02, threshold=5.648e+02, percent-clipped=0.0 2023-04-29 04:35:53,083 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93036.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:35:53,121 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93036.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:35:55,522 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-04-29 04:36:15,632 INFO [train.py:904] (6/8) Epoch 10, batch 1700, loss[loss=0.1902, simple_loss=0.2634, pruned_loss=0.05846, over 16754.00 frames. ], tot_loss[loss=0.1895, simple_loss=0.2704, pruned_loss=0.05437, over 3324365.45 frames. ], batch size: 124, lr: 6.97e-03, grad_scale: 8.0 2023-04-29 04:36:42,621 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93071.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:36:48,264 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93076.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:37:17,823 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93097.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 04:37:23,690 INFO [train.py:904] (6/8) Epoch 10, batch 1750, loss[loss=0.1993, simple_loss=0.2956, pruned_loss=0.05156, over 17101.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.271, pruned_loss=0.0539, over 3329610.80 frames. ], batch size: 49, lr: 6.97e-03, grad_scale: 8.0 2023-04-29 04:37:42,817 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93115.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:38:01,414 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 2.469e+02 2.889e+02 3.646e+02 7.131e+02, threshold=5.778e+02, percent-clipped=4.0 2023-04-29 04:38:32,561 INFO [train.py:904] (6/8) Epoch 10, batch 1800, loss[loss=0.2296, simple_loss=0.2959, pruned_loss=0.0817, over 16885.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2715, pruned_loss=0.05387, over 3329836.20 frames. ], batch size: 116, lr: 6.97e-03, grad_scale: 8.0 2023-04-29 04:38:34,707 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-29 04:39:06,726 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93176.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:39:10,970 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3678, 3.4099, 2.0267, 3.5706, 2.5861, 3.5491, 1.9570, 2.7157], device='cuda:6'), covar=tensor([0.0199, 0.0379, 0.1392, 0.0208, 0.0752, 0.0674, 0.1387, 0.0603], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0167, 0.0186, 0.0126, 0.0166, 0.0206, 0.0193, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 04:39:20,478 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93187.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:39:33,098 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-04-29 04:39:42,538 INFO [train.py:904] (6/8) Epoch 10, batch 1850, loss[loss=0.1705, simple_loss=0.2559, pruned_loss=0.04256, over 17234.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2711, pruned_loss=0.05337, over 3338562.26 frames. ], batch size: 44, lr: 6.97e-03, grad_scale: 8.0 2023-04-29 04:39:56,114 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0346, 4.1404, 4.4441, 2.0955, 4.7579, 4.7090, 3.2405, 3.5361], device='cuda:6'), covar=tensor([0.0649, 0.0164, 0.0160, 0.1095, 0.0041, 0.0113, 0.0353, 0.0352], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0099, 0.0087, 0.0141, 0.0070, 0.0101, 0.0122, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 04:40:21,084 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.739e+02 2.500e+02 2.909e+02 3.446e+02 8.007e+02, threshold=5.817e+02, percent-clipped=2.0 2023-04-29 04:40:39,261 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4632, 3.4447, 3.3568, 2.7848, 3.3626, 2.0361, 3.0740, 2.8040], device='cuda:6'), covar=tensor([0.0115, 0.0099, 0.0147, 0.0200, 0.0078, 0.1978, 0.0122, 0.0195], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0116, 0.0168, 0.0158, 0.0136, 0.0181, 0.0154, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:40:52,059 INFO [train.py:904] (6/8) Epoch 10, batch 1900, loss[loss=0.1669, simple_loss=0.257, pruned_loss=0.03843, over 17122.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.2702, pruned_loss=0.0526, over 3336878.21 frames. ], batch size: 47, lr: 6.97e-03, grad_scale: 8.0 2023-04-29 04:40:56,800 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93255.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:41:33,395 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93281.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:41:41,653 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93287.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:41:48,712 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8215, 5.2029, 5.3312, 5.1848, 5.1501, 5.7713, 5.3919, 5.1453], device='cuda:6'), covar=tensor([0.1127, 0.1751, 0.1846, 0.2284, 0.3026, 0.1197, 0.1281, 0.2389], device='cuda:6'), in_proj_covar=tensor([0.0345, 0.0495, 0.0519, 0.0429, 0.0570, 0.0548, 0.0420, 0.0572], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 04:42:02,723 INFO [train.py:904] (6/8) Epoch 10, batch 1950, loss[loss=0.1646, simple_loss=0.2544, pruned_loss=0.03745, over 17222.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2706, pruned_loss=0.05269, over 3329760.94 frames. ], batch size: 45, lr: 6.96e-03, grad_scale: 8.0 2023-04-29 04:42:22,834 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3014, 5.0511, 5.1986, 5.4720, 5.6613, 4.9281, 5.5930, 5.6061], device='cuda:6'), covar=tensor([0.1128, 0.0842, 0.1536, 0.0573, 0.0463, 0.0653, 0.0405, 0.0452], device='cuda:6'), in_proj_covar=tensor([0.0542, 0.0661, 0.0817, 0.0675, 0.0506, 0.0513, 0.0530, 0.0601], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:42:40,591 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:42:41,417 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.729e+02 2.458e+02 3.015e+02 3.561e+02 8.313e+02, threshold=6.031e+02, percent-clipped=4.0 2023-04-29 04:42:43,593 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93331.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:43:12,558 INFO [train.py:904] (6/8) Epoch 10, batch 2000, loss[loss=0.1686, simple_loss=0.256, pruned_loss=0.04056, over 15894.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2706, pruned_loss=0.05293, over 3311688.87 frames. ], batch size: 35, lr: 6.96e-03, grad_scale: 8.0 2023-04-29 04:43:31,502 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93366.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:43:45,295 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93376.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:44:08,434 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93392.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 04:44:21,491 INFO [train.py:904] (6/8) Epoch 10, batch 2050, loss[loss=0.2147, simple_loss=0.304, pruned_loss=0.06265, over 16706.00 frames. ], tot_loss[loss=0.1887, simple_loss=0.2707, pruned_loss=0.05339, over 3317119.47 frames. ], batch size: 57, lr: 6.96e-03, grad_scale: 4.0 2023-04-29 04:44:23,788 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-29 04:44:51,427 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93424.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:45:00,378 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7981, 3.1637, 3.0129, 5.0405, 4.2057, 4.6841, 1.7317, 3.2450], device='cuda:6'), covar=tensor([0.1366, 0.0614, 0.0975, 0.0110, 0.0234, 0.0339, 0.1384, 0.0750], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0157, 0.0179, 0.0138, 0.0198, 0.0213, 0.0178, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 04:45:00,992 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.855e+02 2.531e+02 2.855e+02 3.295e+02 5.942e+02, threshold=5.709e+02, percent-clipped=0.0 2023-04-29 04:45:29,945 INFO [train.py:904] (6/8) Epoch 10, batch 2100, loss[loss=0.1808, simple_loss=0.2727, pruned_loss=0.0444, over 17125.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2715, pruned_loss=0.05404, over 3303526.31 frames. ], batch size: 48, lr: 6.96e-03, grad_scale: 4.0 2023-04-29 04:45:30,357 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9067, 2.9613, 2.4308, 2.6605, 3.2002, 2.9423, 3.7048, 3.4859], device='cuda:6'), covar=tensor([0.0058, 0.0213, 0.0297, 0.0261, 0.0163, 0.0229, 0.0178, 0.0136], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0201, 0.0197, 0.0196, 0.0199, 0.0200, 0.0209, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:45:56,820 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93471.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:46:09,074 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3559, 3.3056, 3.3387, 3.4760, 3.5249, 3.2196, 3.4033, 3.5532], device='cuda:6'), covar=tensor([0.1054, 0.0779, 0.1068, 0.0637, 0.0624, 0.2312, 0.1318, 0.0701], device='cuda:6'), in_proj_covar=tensor([0.0538, 0.0654, 0.0806, 0.0672, 0.0502, 0.0510, 0.0528, 0.0597], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:46:18,669 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93487.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:46:20,187 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 04:46:40,125 INFO [train.py:904] (6/8) Epoch 10, batch 2150, loss[loss=0.1966, simple_loss=0.2692, pruned_loss=0.06195, over 16734.00 frames. ], tot_loss[loss=0.1906, simple_loss=0.2725, pruned_loss=0.05441, over 3296709.81 frames. ], batch size: 124, lr: 6.96e-03, grad_scale: 4.0 2023-04-29 04:47:14,130 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5917, 4.9058, 4.6673, 4.7250, 4.4493, 4.3579, 4.4109, 4.9496], device='cuda:6'), covar=tensor([0.0916, 0.0803, 0.0948, 0.0595, 0.0758, 0.1043, 0.0941, 0.0849], device='cuda:6'), in_proj_covar=tensor([0.0530, 0.0676, 0.0556, 0.0461, 0.0424, 0.0430, 0.0562, 0.0511], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:47:18,311 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.855e+02 2.473e+02 3.061e+02 3.473e+02 5.653e+02, threshold=6.122e+02, percent-clipped=0.0 2023-04-29 04:47:24,632 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93535.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:47:46,702 INFO [train.py:904] (6/8) Epoch 10, batch 2200, loss[loss=0.202, simple_loss=0.2968, pruned_loss=0.05362, over 17072.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.2732, pruned_loss=0.05432, over 3302916.79 frames. ], batch size: 53, lr: 6.95e-03, grad_scale: 4.0 2023-04-29 04:47:52,024 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93555.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:48:00,120 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-29 04:48:35,415 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93587.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:48:50,629 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2534, 2.1446, 1.6924, 2.0010, 2.5599, 2.3975, 2.6240, 2.7092], device='cuda:6'), covar=tensor([0.0130, 0.0251, 0.0345, 0.0309, 0.0139, 0.0224, 0.0133, 0.0175], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0202, 0.0198, 0.0199, 0.0200, 0.0202, 0.0211, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:48:54,025 INFO [train.py:904] (6/8) Epoch 10, batch 2250, loss[loss=0.1899, simple_loss=0.2678, pruned_loss=0.05603, over 16794.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.2736, pruned_loss=0.05457, over 3309170.47 frames. ], batch size: 102, lr: 6.95e-03, grad_scale: 4.0 2023-04-29 04:48:55,439 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93603.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:49:06,521 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-29 04:49:06,584 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-29 04:49:10,916 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7580, 2.3948, 1.8774, 2.2322, 2.8847, 2.7107, 2.9981, 2.9783], device='cuda:6'), covar=tensor([0.0107, 0.0251, 0.0357, 0.0295, 0.0130, 0.0189, 0.0148, 0.0149], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0202, 0.0198, 0.0199, 0.0200, 0.0201, 0.0210, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 04:49:33,829 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.688e+02 2.560e+02 3.200e+02 3.958e+02 7.230e+02, threshold=6.400e+02, percent-clipped=2.0 2023-04-29 04:49:34,793 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93631.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:49:39,248 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93635.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:50:00,624 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-29 04:50:04,001 INFO [train.py:904] (6/8) Epoch 10, batch 2300, loss[loss=0.1935, simple_loss=0.2895, pruned_loss=0.04869, over 17120.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2753, pruned_loss=0.05555, over 3303553.04 frames. ], batch size: 53, lr: 6.95e-03, grad_scale: 4.0 2023-04-29 04:50:22,491 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93666.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:50:39,723 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93679.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:50:49,512 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8026, 3.0010, 2.5695, 4.7227, 3.9215, 4.4633, 1.5978, 3.3610], device='cuda:6'), covar=tensor([0.1259, 0.0621, 0.1132, 0.0216, 0.0269, 0.0349, 0.1383, 0.0639], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0156, 0.0178, 0.0137, 0.0198, 0.0212, 0.0177, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 04:50:57,944 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93692.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:51:04,576 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 04:51:09,769 INFO [train.py:904] (6/8) Epoch 10, batch 2350, loss[loss=0.2206, simple_loss=0.3005, pruned_loss=0.07031, over 16917.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2759, pruned_loss=0.05608, over 3298077.87 frames. ], batch size: 96, lr: 6.95e-03, grad_scale: 4.0 2023-04-29 04:51:27,737 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93714.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:51:49,910 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.363e+02 2.805e+02 3.305e+02 9.718e+02, threshold=5.610e+02, percent-clipped=1.0 2023-04-29 04:52:03,176 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93740.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:52:17,496 INFO [train.py:904] (6/8) Epoch 10, batch 2400, loss[loss=0.1863, simple_loss=0.2681, pruned_loss=0.05226, over 16761.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2759, pruned_loss=0.0557, over 3301876.48 frames. ], batch size: 83, lr: 6.95e-03, grad_scale: 8.0 2023-04-29 04:52:41,472 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93770.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:52:43,109 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93771.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:52:49,833 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2203, 3.2400, 3.4131, 2.3255, 3.0789, 3.4976, 3.3049, 1.9658], device='cuda:6'), covar=tensor([0.0370, 0.0078, 0.0039, 0.0281, 0.0093, 0.0060, 0.0059, 0.0361], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0070, 0.0069, 0.0125, 0.0077, 0.0085, 0.0076, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 04:53:17,227 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2023-04-29 04:53:26,813 INFO [train.py:904] (6/8) Epoch 10, batch 2450, loss[loss=0.1663, simple_loss=0.2556, pruned_loss=0.03847, over 17243.00 frames. ], tot_loss[loss=0.1935, simple_loss=0.2763, pruned_loss=0.05536, over 3290887.43 frames. ], batch size: 45, lr: 6.94e-03, grad_scale: 8.0 2023-04-29 04:53:49,967 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93819.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:54:05,073 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.668e+02 2.505e+02 3.029e+02 3.825e+02 7.236e+02, threshold=6.058e+02, percent-clipped=4.0 2023-04-29 04:54:05,487 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93831.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:54:16,421 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1409, 4.2214, 4.5232, 2.1086, 4.8218, 4.8631, 3.3394, 3.5657], device='cuda:6'), covar=tensor([0.0620, 0.0163, 0.0172, 0.1086, 0.0068, 0.0096, 0.0344, 0.0354], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0098, 0.0087, 0.0138, 0.0070, 0.0100, 0.0121, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 04:54:34,536 INFO [train.py:904] (6/8) Epoch 10, batch 2500, loss[loss=0.2043, simple_loss=0.2836, pruned_loss=0.06252, over 16716.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2759, pruned_loss=0.05494, over 3296722.07 frames. ], batch size: 134, lr: 6.94e-03, grad_scale: 8.0 2023-04-29 04:55:17,058 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-29 04:55:43,655 INFO [train.py:904] (6/8) Epoch 10, batch 2550, loss[loss=0.2203, simple_loss=0.2894, pruned_loss=0.0756, over 16663.00 frames. ], tot_loss[loss=0.192, simple_loss=0.275, pruned_loss=0.05453, over 3307520.23 frames. ], batch size: 134, lr: 6.94e-03, grad_scale: 8.0 2023-04-29 04:56:23,964 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.543e+02 2.408e+02 2.927e+02 3.582e+02 7.180e+02, threshold=5.854e+02, percent-clipped=2.0 2023-04-29 04:56:52,858 INFO [train.py:904] (6/8) Epoch 10, batch 2600, loss[loss=0.1807, simple_loss=0.2757, pruned_loss=0.04279, over 17108.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.2746, pruned_loss=0.05358, over 3317725.63 frames. ], batch size: 55, lr: 6.94e-03, grad_scale: 8.0 2023-04-29 04:57:32,378 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93981.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:58:03,849 INFO [train.py:904] (6/8) Epoch 10, batch 2650, loss[loss=0.1838, simple_loss=0.2794, pruned_loss=0.04406, over 17045.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2745, pruned_loss=0.05311, over 3318021.96 frames. ], batch size: 50, lr: 6.94e-03, grad_scale: 8.0 2023-04-29 04:58:23,134 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-29 04:58:43,799 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.633e+02 2.229e+02 2.744e+02 3.271e+02 8.724e+02, threshold=5.488e+02, percent-clipped=1.0 2023-04-29 04:59:00,264 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94042.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:59:10,880 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6977, 4.5565, 4.5234, 3.0842, 3.8043, 4.4244, 4.0648, 2.7729], device='cuda:6'), covar=tensor([0.0374, 0.0033, 0.0026, 0.0273, 0.0081, 0.0065, 0.0045, 0.0316], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0070, 0.0069, 0.0124, 0.0078, 0.0085, 0.0076, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 04:59:13,550 INFO [train.py:904] (6/8) Epoch 10, batch 2700, loss[loss=0.1985, simple_loss=0.2835, pruned_loss=0.05678, over 16134.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2748, pruned_loss=0.05284, over 3311210.93 frames. ], batch size: 35, lr: 6.94e-03, grad_scale: 8.0 2023-04-29 04:59:30,797 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 04:59:32,952 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5267, 1.4570, 2.0167, 2.3693, 2.4527, 2.3172, 1.4794, 2.5993], device='cuda:6'), covar=tensor([0.0130, 0.0368, 0.0231, 0.0224, 0.0176, 0.0209, 0.0371, 0.0082], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0171, 0.0155, 0.0160, 0.0167, 0.0122, 0.0170, 0.0112], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 04:59:35,098 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-29 05:00:23,305 INFO [train.py:904] (6/8) Epoch 10, batch 2750, loss[loss=0.1965, simple_loss=0.2838, pruned_loss=0.05457, over 17248.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2752, pruned_loss=0.05269, over 3311290.31 frames. ], batch size: 52, lr: 6.93e-03, grad_scale: 8.0 2023-04-29 05:00:55,374 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94126.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:01:01,090 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.410e+02 2.958e+02 3.419e+02 5.641e+02, threshold=5.917e+02, percent-clipped=1.0 2023-04-29 05:01:29,685 INFO [train.py:904] (6/8) Epoch 10, batch 2800, loss[loss=0.187, simple_loss=0.2781, pruned_loss=0.04794, over 17260.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.2753, pruned_loss=0.05309, over 3312256.32 frames. ], batch size: 52, lr: 6.93e-03, grad_scale: 8.0 2023-04-29 05:01:50,001 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-29 05:01:59,483 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0822, 3.4693, 3.2501, 2.0899, 2.8325, 2.3883, 3.6269, 3.5659], device='cuda:6'), covar=tensor([0.0241, 0.0668, 0.0539, 0.1476, 0.0691, 0.0871, 0.0457, 0.0669], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0144, 0.0157, 0.0143, 0.0135, 0.0125, 0.0136, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 05:02:32,106 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-29 05:02:39,398 INFO [train.py:904] (6/8) Epoch 10, batch 2850, loss[loss=0.2853, simple_loss=0.3389, pruned_loss=0.1158, over 11847.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.2747, pruned_loss=0.05312, over 3310701.48 frames. ], batch size: 247, lr: 6.93e-03, grad_scale: 8.0 2023-04-29 05:03:09,796 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=94223.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:03:20,115 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.429e+02 2.851e+02 3.333e+02 6.061e+02, threshold=5.703e+02, percent-clipped=1.0 2023-04-29 05:03:49,048 INFO [train.py:904] (6/8) Epoch 10, batch 2900, loss[loss=0.165, simple_loss=0.2587, pruned_loss=0.03563, over 16997.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2729, pruned_loss=0.05295, over 3314871.62 frames. ], batch size: 41, lr: 6.93e-03, grad_scale: 8.0 2023-04-29 05:04:23,035 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0055, 4.7426, 4.9871, 5.2120, 5.4163, 4.7458, 5.3524, 5.3342], device='cuda:6'), covar=tensor([0.1373, 0.1053, 0.1680, 0.0621, 0.0481, 0.0814, 0.0516, 0.0551], device='cuda:6'), in_proj_covar=tensor([0.0547, 0.0667, 0.0825, 0.0683, 0.0508, 0.0527, 0.0540, 0.0603], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:04:33,813 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94284.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 05:04:36,259 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7454, 2.2305, 2.3463, 4.4275, 2.1756, 2.8377, 2.3431, 2.4523], device='cuda:6'), covar=tensor([0.0809, 0.3199, 0.2039, 0.0343, 0.3373, 0.2152, 0.2764, 0.3089], device='cuda:6'), in_proj_covar=tensor([0.0363, 0.0387, 0.0325, 0.0324, 0.0407, 0.0440, 0.0346, 0.0456], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:04:58,055 INFO [train.py:904] (6/8) Epoch 10, batch 2950, loss[loss=0.1862, simple_loss=0.2571, pruned_loss=0.05764, over 16774.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2728, pruned_loss=0.05378, over 3310044.86 frames. ], batch size: 102, lr: 6.93e-03, grad_scale: 8.0 2023-04-29 05:05:39,543 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.726e+02 2.713e+02 3.214e+02 3.938e+02 7.856e+02, threshold=6.427e+02, percent-clipped=3.0 2023-04-29 05:05:49,283 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94337.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:06:03,147 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=94348.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:06:08,009 INFO [train.py:904] (6/8) Epoch 10, batch 3000, loss[loss=0.1926, simple_loss=0.2687, pruned_loss=0.05824, over 15951.00 frames. ], tot_loss[loss=0.1923, simple_loss=0.2743, pruned_loss=0.05516, over 3300337.73 frames. ], batch size: 35, lr: 6.92e-03, grad_scale: 8.0 2023-04-29 05:06:08,009 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 05:06:17,138 INFO [train.py:938] (6/8) Epoch 10, validation: loss=0.1426, simple_loss=0.2488, pruned_loss=0.01818, over 944034.00 frames. 2023-04-29 05:06:17,139 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 05:07:26,695 INFO [train.py:904] (6/8) Epoch 10, batch 3050, loss[loss=0.1772, simple_loss=0.2712, pruned_loss=0.04159, over 17121.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2747, pruned_loss=0.05562, over 3289315.83 frames. ], batch size: 47, lr: 6.92e-03, grad_scale: 8.0 2023-04-29 05:07:36,849 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94409.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:07:57,725 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94426.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:08:05,349 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.835e+02 2.667e+02 3.252e+02 4.083e+02 7.974e+02, threshold=6.505e+02, percent-clipped=3.0 2023-04-29 05:08:12,066 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6951, 3.0412, 2.4024, 4.4937, 3.6796, 4.2660, 1.6573, 2.9266], device='cuda:6'), covar=tensor([0.1334, 0.0591, 0.1130, 0.0155, 0.0297, 0.0364, 0.1360, 0.0822], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0157, 0.0179, 0.0139, 0.0200, 0.0214, 0.0176, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 05:08:33,247 INFO [train.py:904] (6/8) Epoch 10, batch 3100, loss[loss=0.1789, simple_loss=0.2693, pruned_loss=0.04428, over 16644.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2732, pruned_loss=0.05557, over 3292699.60 frames. ], batch size: 62, lr: 6.92e-03, grad_scale: 8.0 2023-04-29 05:09:04,483 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=94474.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:09:13,152 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5062, 4.5064, 4.4390, 3.9440, 4.4653, 1.7210, 4.2112, 4.1972], device='cuda:6'), covar=tensor([0.0092, 0.0068, 0.0126, 0.0280, 0.0075, 0.2209, 0.0124, 0.0170], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0118, 0.0166, 0.0160, 0.0137, 0.0178, 0.0155, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:09:43,502 INFO [train.py:904] (6/8) Epoch 10, batch 3150, loss[loss=0.1664, simple_loss=0.2629, pruned_loss=0.03497, over 17110.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.272, pruned_loss=0.05489, over 3300620.58 frames. ], batch size: 48, lr: 6.92e-03, grad_scale: 8.0 2023-04-29 05:10:02,982 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8594, 4.3137, 3.2428, 2.2998, 2.9695, 2.5298, 4.6587, 4.0435], device='cuda:6'), covar=tensor([0.2474, 0.0598, 0.1417, 0.2172, 0.2351, 0.1681, 0.0343, 0.0925], device='cuda:6'), in_proj_covar=tensor([0.0299, 0.0256, 0.0279, 0.0273, 0.0281, 0.0219, 0.0266, 0.0298], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:10:15,386 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7985, 4.4048, 4.5202, 4.9851, 5.1921, 4.5587, 5.2381, 5.1416], device='cuda:6'), covar=tensor([0.1413, 0.1426, 0.2468, 0.0865, 0.0731, 0.0876, 0.0729, 0.0753], device='cuda:6'), in_proj_covar=tensor([0.0544, 0.0670, 0.0828, 0.0678, 0.0508, 0.0526, 0.0536, 0.0598], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:10:23,683 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.666e+02 2.494e+02 3.047e+02 3.554e+02 8.509e+02, threshold=6.093e+02, percent-clipped=1.0 2023-04-29 05:10:52,169 INFO [train.py:904] (6/8) Epoch 10, batch 3200, loss[loss=0.2024, simple_loss=0.2809, pruned_loss=0.06198, over 12505.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2707, pruned_loss=0.05407, over 3295476.09 frames. ], batch size: 247, lr: 6.92e-03, grad_scale: 8.0 2023-04-29 05:11:32,211 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94579.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 05:12:04,565 INFO [train.py:904] (6/8) Epoch 10, batch 3250, loss[loss=0.215, simple_loss=0.2862, pruned_loss=0.07192, over 16711.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2716, pruned_loss=0.05386, over 3299044.30 frames. ], batch size: 134, lr: 6.92e-03, grad_scale: 8.0 2023-04-29 05:12:44,900 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.625e+02 2.346e+02 2.940e+02 3.507e+02 9.203e+02, threshold=5.881e+02, percent-clipped=1.0 2023-04-29 05:12:51,175 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 05:12:53,017 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94637.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:13:14,618 INFO [train.py:904] (6/8) Epoch 10, batch 3300, loss[loss=0.221, simple_loss=0.2984, pruned_loss=0.07183, over 16436.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.2734, pruned_loss=0.05503, over 3307003.31 frames. ], batch size: 146, lr: 6.91e-03, grad_scale: 8.0 2023-04-29 05:14:02,320 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=94685.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:14:24,570 INFO [train.py:904] (6/8) Epoch 10, batch 3350, loss[loss=0.1759, simple_loss=0.264, pruned_loss=0.0439, over 16862.00 frames. ], tot_loss[loss=0.1915, simple_loss=0.2737, pruned_loss=0.05467, over 3313052.24 frames. ], batch size: 42, lr: 6.91e-03, grad_scale: 8.0 2023-04-29 05:14:28,541 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94704.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:15:05,006 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.683e+02 2.488e+02 2.932e+02 3.912e+02 8.438e+02, threshold=5.863e+02, percent-clipped=4.0 2023-04-29 05:15:13,456 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9173, 4.9714, 4.7765, 4.2234, 4.8075, 1.9722, 4.5643, 4.7211], device='cuda:6'), covar=tensor([0.0079, 0.0063, 0.0147, 0.0353, 0.0087, 0.2130, 0.0125, 0.0172], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0121, 0.0170, 0.0163, 0.0139, 0.0181, 0.0158, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:15:28,946 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3945, 4.4229, 4.8252, 4.8392, 4.8647, 4.5154, 4.5515, 4.3324], device='cuda:6'), covar=tensor([0.0347, 0.0495, 0.0355, 0.0342, 0.0390, 0.0310, 0.0760, 0.0490], device='cuda:6'), in_proj_covar=tensor([0.0345, 0.0351, 0.0358, 0.0337, 0.0397, 0.0373, 0.0481, 0.0298], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 05:15:35,790 INFO [train.py:904] (6/8) Epoch 10, batch 3400, loss[loss=0.1982, simple_loss=0.2685, pruned_loss=0.06402, over 16848.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.2729, pruned_loss=0.05404, over 3312830.85 frames. ], batch size: 109, lr: 6.91e-03, grad_scale: 8.0 2023-04-29 05:16:35,074 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7144, 3.8705, 2.9743, 2.1575, 2.5732, 2.1760, 3.9424, 3.4872], device='cuda:6'), covar=tensor([0.2401, 0.0557, 0.1332, 0.2438, 0.2463, 0.1807, 0.0463, 0.1169], device='cuda:6'), in_proj_covar=tensor([0.0300, 0.0257, 0.0278, 0.0274, 0.0281, 0.0218, 0.0267, 0.0298], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:16:44,906 INFO [train.py:904] (6/8) Epoch 10, batch 3450, loss[loss=0.1701, simple_loss=0.2452, pruned_loss=0.04753, over 15863.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2719, pruned_loss=0.05415, over 3305624.91 frames. ], batch size: 35, lr: 6.91e-03, grad_scale: 8.0 2023-04-29 05:16:45,558 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 05:16:57,161 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.80 vs. limit=2.0 2023-04-29 05:17:00,811 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8055, 4.0070, 3.1011, 2.3979, 2.8469, 2.3824, 4.2605, 3.7519], device='cuda:6'), covar=tensor([0.2409, 0.0612, 0.1393, 0.2206, 0.2346, 0.1743, 0.0410, 0.1012], device='cuda:6'), in_proj_covar=tensor([0.0300, 0.0257, 0.0279, 0.0274, 0.0281, 0.0218, 0.0267, 0.0299], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:17:13,795 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.80 vs. limit=5.0 2023-04-29 05:17:26,304 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.690e+02 2.564e+02 2.957e+02 3.532e+02 6.981e+02, threshold=5.915e+02, percent-clipped=3.0 2023-04-29 05:17:53,299 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7543, 4.8527, 5.2416, 5.2207, 5.2203, 4.9294, 4.9008, 4.6185], device='cuda:6'), covar=tensor([0.0286, 0.0424, 0.0383, 0.0430, 0.0406, 0.0303, 0.0744, 0.0456], device='cuda:6'), in_proj_covar=tensor([0.0344, 0.0350, 0.0356, 0.0337, 0.0399, 0.0371, 0.0478, 0.0298], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 05:17:56,546 INFO [train.py:904] (6/8) Epoch 10, batch 3500, loss[loss=0.1605, simple_loss=0.248, pruned_loss=0.03649, over 17248.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2706, pruned_loss=0.05353, over 3303278.11 frames. ], batch size: 45, lr: 6.91e-03, grad_scale: 8.0 2023-04-29 05:18:27,048 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 05:18:35,833 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94879.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:19:06,954 INFO [train.py:904] (6/8) Epoch 10, batch 3550, loss[loss=0.1923, simple_loss=0.2631, pruned_loss=0.06072, over 16460.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.2699, pruned_loss=0.05344, over 3307786.03 frames. ], batch size: 75, lr: 6.90e-03, grad_scale: 8.0 2023-04-29 05:19:22,833 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1088, 4.4399, 4.4585, 3.5270, 3.8418, 4.3578, 4.0437, 2.5413], device='cuda:6'), covar=tensor([0.0301, 0.0026, 0.0024, 0.0212, 0.0078, 0.0058, 0.0047, 0.0363], device='cuda:6'), in_proj_covar=tensor([0.0125, 0.0069, 0.0069, 0.0122, 0.0078, 0.0084, 0.0075, 0.0117], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 05:19:42,075 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=94927.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:19:47,556 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.652e+02 2.230e+02 2.675e+02 3.251e+02 5.912e+02, threshold=5.350e+02, percent-clipped=0.0 2023-04-29 05:20:17,530 INFO [train.py:904] (6/8) Epoch 10, batch 3600, loss[loss=0.1596, simple_loss=0.25, pruned_loss=0.03457, over 17202.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2694, pruned_loss=0.0532, over 3301965.09 frames. ], batch size: 44, lr: 6.90e-03, grad_scale: 8.0 2023-04-29 05:20:45,809 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9832, 3.3724, 3.1382, 1.9805, 2.7543, 2.4509, 3.3006, 3.4675], device='cuda:6'), covar=tensor([0.0319, 0.0728, 0.0561, 0.1580, 0.0780, 0.0809, 0.0681, 0.0821], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0147, 0.0157, 0.0144, 0.0136, 0.0125, 0.0138, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 05:21:28,702 INFO [train.py:904] (6/8) Epoch 10, batch 3650, loss[loss=0.143, simple_loss=0.2141, pruned_loss=0.03597, over 16758.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2684, pruned_loss=0.05307, over 3292021.09 frames. ], batch size: 83, lr: 6.90e-03, grad_scale: 8.0 2023-04-29 05:21:33,024 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95004.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:22:10,213 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.931e+02 2.260e+02 2.711e+02 3.366e+02 9.321e+02, threshold=5.423e+02, percent-clipped=5.0 2023-04-29 05:22:42,298 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2737, 5.2422, 5.0850, 4.7438, 4.6780, 5.1817, 5.0810, 4.8092], device='cuda:6'), covar=tensor([0.0477, 0.0372, 0.0246, 0.0274, 0.0994, 0.0309, 0.0261, 0.0580], device='cuda:6'), in_proj_covar=tensor([0.0252, 0.0314, 0.0298, 0.0277, 0.0325, 0.0313, 0.0205, 0.0349], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 05:22:43,027 INFO [train.py:904] (6/8) Epoch 10, batch 3700, loss[loss=0.1868, simple_loss=0.2773, pruned_loss=0.04816, over 17258.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2671, pruned_loss=0.05449, over 3276854.28 frames. ], batch size: 52, lr: 6.90e-03, grad_scale: 8.0 2023-04-29 05:22:43,413 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=95052.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:23:56,093 INFO [train.py:904] (6/8) Epoch 10, batch 3750, loss[loss=0.1941, simple_loss=0.2571, pruned_loss=0.06558, over 16681.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2671, pruned_loss=0.05601, over 3270962.84 frames. ], batch size: 89, lr: 6.90e-03, grad_scale: 8.0 2023-04-29 05:24:00,855 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95105.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 05:24:25,470 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9433, 2.3993, 2.3034, 2.7291, 2.4227, 3.2412, 1.7710, 2.6340], device='cuda:6'), covar=tensor([0.1047, 0.0571, 0.0952, 0.0120, 0.0169, 0.0318, 0.1180, 0.0691], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0159, 0.0180, 0.0143, 0.0205, 0.0214, 0.0178, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 05:24:38,278 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.833e+02 2.465e+02 2.786e+02 3.363e+02 5.307e+02, threshold=5.572e+02, percent-clipped=0.0 2023-04-29 05:24:44,263 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2552, 2.7114, 2.1496, 2.3867, 3.0577, 2.7346, 3.2791, 3.2411], device='cuda:6'), covar=tensor([0.0089, 0.0204, 0.0308, 0.0288, 0.0122, 0.0241, 0.0119, 0.0160], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0200, 0.0197, 0.0197, 0.0200, 0.0203, 0.0211, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:25:07,899 INFO [train.py:904] (6/8) Epoch 10, batch 3800, loss[loss=0.1989, simple_loss=0.2748, pruned_loss=0.0615, over 16197.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2688, pruned_loss=0.05741, over 3269499.01 frames. ], batch size: 165, lr: 6.90e-03, grad_scale: 8.0 2023-04-29 05:25:28,625 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95166.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 05:25:52,510 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4442, 4.6587, 4.9632, 4.9375, 4.9602, 4.6981, 4.5366, 4.4306], device='cuda:6'), covar=tensor([0.0443, 0.0623, 0.0392, 0.0431, 0.0553, 0.0445, 0.1158, 0.0566], device='cuda:6'), in_proj_covar=tensor([0.0334, 0.0341, 0.0344, 0.0326, 0.0385, 0.0361, 0.0466, 0.0285], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 05:26:20,912 INFO [train.py:904] (6/8) Epoch 10, batch 3850, loss[loss=0.201, simple_loss=0.2769, pruned_loss=0.06252, over 15647.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2685, pruned_loss=0.058, over 3267596.59 frames. ], batch size: 190, lr: 6.89e-03, grad_scale: 8.0 2023-04-29 05:27:00,983 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.608e+02 2.452e+02 2.920e+02 3.414e+02 5.310e+02, threshold=5.839e+02, percent-clipped=0.0 2023-04-29 05:27:31,950 INFO [train.py:904] (6/8) Epoch 10, batch 3900, loss[loss=0.1955, simple_loss=0.2668, pruned_loss=0.06213, over 16776.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2678, pruned_loss=0.05828, over 3282403.39 frames. ], batch size: 102, lr: 6.89e-03, grad_scale: 8.0 2023-04-29 05:27:43,915 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95260.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:27:56,160 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95268.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:28:07,352 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5028, 4.0526, 4.2639, 2.6962, 3.8559, 4.2596, 3.9729, 2.6253], device='cuda:6'), covar=tensor([0.0383, 0.0090, 0.0026, 0.0284, 0.0045, 0.0052, 0.0039, 0.0263], device='cuda:6'), in_proj_covar=tensor([0.0124, 0.0067, 0.0067, 0.0121, 0.0075, 0.0083, 0.0073, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 05:28:45,286 INFO [train.py:904] (6/8) Epoch 10, batch 3950, loss[loss=0.2016, simple_loss=0.2691, pruned_loss=0.06703, over 16904.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2672, pruned_loss=0.05834, over 3282173.96 frames. ], batch size: 116, lr: 6.89e-03, grad_scale: 8.0 2023-04-29 05:28:50,444 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-29 05:29:13,005 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95321.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:29:23,111 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:29:25,658 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.393e+02 2.876e+02 3.494e+02 7.568e+02, threshold=5.751e+02, percent-clipped=4.0 2023-04-29 05:29:38,101 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95339.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:29:56,313 INFO [train.py:904] (6/8) Epoch 10, batch 4000, loss[loss=0.2121, simple_loss=0.2829, pruned_loss=0.07069, over 16759.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2667, pruned_loss=0.05857, over 3287527.28 frames. ], batch size: 124, lr: 6.89e-03, grad_scale: 8.0 2023-04-29 05:31:05,676 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95400.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:31:07,607 INFO [train.py:904] (6/8) Epoch 10, batch 4050, loss[loss=0.1932, simple_loss=0.2755, pruned_loss=0.05541, over 16400.00 frames. ], tot_loss[loss=0.1906, simple_loss=0.267, pruned_loss=0.05717, over 3286985.02 frames. ], batch size: 146, lr: 6.89e-03, grad_scale: 16.0 2023-04-29 05:31:49,145 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 1.964e+02 2.335e+02 2.702e+02 4.319e+02, threshold=4.671e+02, percent-clipped=0.0 2023-04-29 05:32:20,029 INFO [train.py:904] (6/8) Epoch 10, batch 4100, loss[loss=0.1981, simple_loss=0.284, pruned_loss=0.05607, over 16733.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.268, pruned_loss=0.05635, over 3272099.96 frames. ], batch size: 124, lr: 6.88e-03, grad_scale: 16.0 2023-04-29 05:32:34,840 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95461.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:33:00,749 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6093, 3.5922, 2.7602, 2.1975, 2.7294, 2.3102, 3.9047, 3.5063], device='cuda:6'), covar=tensor([0.2702, 0.0918, 0.1706, 0.2084, 0.2181, 0.1707, 0.0528, 0.0975], device='cuda:6'), in_proj_covar=tensor([0.0301, 0.0259, 0.0284, 0.0278, 0.0290, 0.0220, 0.0271, 0.0302], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 05:33:33,930 INFO [train.py:904] (6/8) Epoch 10, batch 4150, loss[loss=0.1953, simple_loss=0.2793, pruned_loss=0.05569, over 17028.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2762, pruned_loss=0.05955, over 3232329.65 frames. ], batch size: 55, lr: 6.88e-03, grad_scale: 16.0 2023-04-29 05:34:08,559 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2785, 2.4927, 2.0072, 2.2765, 2.7739, 2.5524, 3.0645, 3.0494], device='cuda:6'), covar=tensor([0.0064, 0.0275, 0.0393, 0.0342, 0.0177, 0.0255, 0.0135, 0.0155], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0197, 0.0194, 0.0193, 0.0198, 0.0199, 0.0205, 0.0188], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:34:17,108 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.022e+02 2.709e+02 3.150e+02 3.936e+02 7.135e+02, threshold=6.300e+02, percent-clipped=10.0 2023-04-29 05:34:21,720 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4185, 2.1365, 2.0946, 4.0589, 2.0193, 2.6432, 2.2963, 2.3585], device='cuda:6'), covar=tensor([0.0889, 0.3162, 0.2086, 0.0390, 0.3531, 0.2022, 0.2652, 0.2918], device='cuda:6'), in_proj_covar=tensor([0.0362, 0.0390, 0.0324, 0.0324, 0.0405, 0.0446, 0.0350, 0.0460], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:34:44,260 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2023-04-29 05:34:49,629 INFO [train.py:904] (6/8) Epoch 10, batch 4200, loss[loss=0.2282, simple_loss=0.3146, pruned_loss=0.07088, over 16460.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2839, pruned_loss=0.06195, over 3198697.41 frames. ], batch size: 75, lr: 6.88e-03, grad_scale: 16.0 2023-04-29 05:35:10,370 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 05:35:30,764 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5095, 2.0123, 2.1080, 4.1546, 1.9205, 2.6464, 2.0932, 2.2255], device='cuda:6'), covar=tensor([0.0801, 0.3631, 0.2229, 0.0337, 0.3859, 0.1964, 0.3052, 0.3167], device='cuda:6'), in_proj_covar=tensor([0.0360, 0.0389, 0.0323, 0.0323, 0.0406, 0.0444, 0.0349, 0.0459], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:36:04,061 INFO [train.py:904] (6/8) Epoch 10, batch 4250, loss[loss=0.1949, simple_loss=0.2966, pruned_loss=0.04662, over 16505.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2878, pruned_loss=0.06201, over 3198179.35 frames. ], batch size: 68, lr: 6.88e-03, grad_scale: 8.0 2023-04-29 05:36:24,753 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95616.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:36:26,120 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95617.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:36:38,762 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95624.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:36:49,130 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.325e+02 2.809e+02 3.314e+02 5.860e+02, threshold=5.619e+02, percent-clipped=0.0 2023-04-29 05:37:19,469 INFO [train.py:904] (6/8) Epoch 10, batch 4300, loss[loss=0.2162, simple_loss=0.3082, pruned_loss=0.06209, over 17010.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2884, pruned_loss=0.06084, over 3189050.39 frames. ], batch size: 55, lr: 6.88e-03, grad_scale: 8.0 2023-04-29 05:37:26,774 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3111, 4.0627, 4.0227, 2.4802, 3.6543, 3.8684, 3.7138, 2.2134], device='cuda:6'), covar=tensor([0.0384, 0.0020, 0.0023, 0.0301, 0.0054, 0.0057, 0.0045, 0.0311], device='cuda:6'), in_proj_covar=tensor([0.0125, 0.0066, 0.0067, 0.0121, 0.0076, 0.0083, 0.0073, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 05:37:59,704 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95678.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:38:24,124 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95695.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:38:28,796 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5447, 2.1759, 2.1667, 4.0804, 1.8354, 2.5166, 2.2550, 2.2284], device='cuda:6'), covar=tensor([0.0975, 0.3374, 0.2203, 0.0520, 0.4373, 0.2409, 0.2894, 0.3744], device='cuda:6'), in_proj_covar=tensor([0.0362, 0.0390, 0.0323, 0.0323, 0.0409, 0.0444, 0.0350, 0.0459], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:38:33,663 INFO [train.py:904] (6/8) Epoch 10, batch 4350, loss[loss=0.1997, simple_loss=0.2906, pruned_loss=0.05437, over 17105.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2915, pruned_loss=0.06144, over 3216676.75 frames. ], batch size: 47, lr: 6.88e-03, grad_scale: 8.0 2023-04-29 05:39:18,685 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.762e+02 2.523e+02 3.025e+02 3.679e+02 8.417e+02, threshold=6.050e+02, percent-clipped=3.0 2023-04-29 05:39:49,467 INFO [train.py:904] (6/8) Epoch 10, batch 4400, loss[loss=0.2339, simple_loss=0.3046, pruned_loss=0.08159, over 11781.00 frames. ], tot_loss[loss=0.2095, simple_loss=0.2937, pruned_loss=0.06268, over 3193086.46 frames. ], batch size: 246, lr: 6.87e-03, grad_scale: 8.0 2023-04-29 05:39:55,843 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4050, 3.3765, 3.7026, 1.6398, 3.9194, 4.0246, 2.9047, 2.9090], device='cuda:6'), covar=tensor([0.0755, 0.0221, 0.0194, 0.1209, 0.0049, 0.0061, 0.0410, 0.0403], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0101, 0.0087, 0.0139, 0.0070, 0.0099, 0.0121, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 05:40:02,693 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95761.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:41:01,592 INFO [train.py:904] (6/8) Epoch 10, batch 4450, loss[loss=0.2495, simple_loss=0.3153, pruned_loss=0.09183, over 11998.00 frames. ], tot_loss[loss=0.2119, simple_loss=0.2969, pruned_loss=0.06347, over 3205506.38 frames. ], batch size: 246, lr: 6.87e-03, grad_scale: 8.0 2023-04-29 05:41:13,573 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=95809.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:41:17,830 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95812.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:41:39,041 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.54 vs. limit=2.0 2023-04-29 05:41:46,152 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.254e+02 2.672e+02 3.297e+02 5.015e+02, threshold=5.344e+02, percent-clipped=0.0 2023-04-29 05:42:08,789 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95847.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:42:15,000 INFO [train.py:904] (6/8) Epoch 10, batch 4500, loss[loss=0.2072, simple_loss=0.297, pruned_loss=0.05872, over 17216.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2972, pruned_loss=0.06383, over 3211021.45 frames. ], batch size: 44, lr: 6.87e-03, grad_scale: 8.0 2023-04-29 05:42:46,026 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95873.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:43:04,874 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8854, 3.2774, 3.1905, 1.7705, 2.6328, 2.0552, 3.3233, 3.3668], device='cuda:6'), covar=tensor([0.0259, 0.0669, 0.0585, 0.1973, 0.0884, 0.1045, 0.0641, 0.0866], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0143, 0.0155, 0.0141, 0.0133, 0.0123, 0.0134, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 05:43:27,210 INFO [train.py:904] (6/8) Epoch 10, batch 4550, loss[loss=0.2223, simple_loss=0.3054, pruned_loss=0.06964, over 16404.00 frames. ], tot_loss[loss=0.2132, simple_loss=0.2977, pruned_loss=0.06429, over 3227695.20 frames. ], batch size: 35, lr: 6.87e-03, grad_scale: 8.0 2023-04-29 05:43:35,617 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95908.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 05:43:47,890 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95916.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:43:59,522 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95924.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:44:10,342 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.676e+02 2.125e+02 2.355e+02 2.832e+02 4.769e+02, threshold=4.710e+02, percent-clipped=0.0 2023-04-29 05:44:39,213 INFO [train.py:904] (6/8) Epoch 10, batch 4600, loss[loss=0.1931, simple_loss=0.2854, pruned_loss=0.0504, over 17233.00 frames. ], tot_loss[loss=0.2132, simple_loss=0.298, pruned_loss=0.06416, over 3220737.58 frames. ], batch size: 45, lr: 6.87e-03, grad_scale: 8.0 2023-04-29 05:44:55,071 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9393, 4.9840, 4.7579, 4.4915, 4.4405, 4.8441, 4.6565, 4.4795], device='cuda:6'), covar=tensor([0.0403, 0.0202, 0.0172, 0.0200, 0.0723, 0.0238, 0.0304, 0.0516], device='cuda:6'), in_proj_covar=tensor([0.0231, 0.0287, 0.0276, 0.0252, 0.0298, 0.0286, 0.0189, 0.0321], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:44:57,859 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=95964.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:45:10,196 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=95972.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:45:12,083 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95973.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:45:20,949 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95980.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:45:36,661 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2023-04-29 05:45:43,371 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95995.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:45:55,175 INFO [train.py:904] (6/8) Epoch 10, batch 4650, loss[loss=0.2016, simple_loss=0.2875, pruned_loss=0.05788, over 16801.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2968, pruned_loss=0.06398, over 3224108.97 frames. ], batch size: 116, lr: 6.87e-03, grad_scale: 8.0 2023-04-29 05:46:40,627 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.329e+02 2.110e+02 2.403e+02 2.925e+02 6.538e+02, threshold=4.805e+02, percent-clipped=1.0 2023-04-29 05:46:55,269 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96041.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:46:58,681 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96043.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:47:10,296 INFO [train.py:904] (6/8) Epoch 10, batch 4700, loss[loss=0.2018, simple_loss=0.2853, pruned_loss=0.05914, over 16701.00 frames. ], tot_loss[loss=0.2095, simple_loss=0.2937, pruned_loss=0.06268, over 3210658.17 frames. ], batch size: 83, lr: 6.86e-03, grad_scale: 8.0 2023-04-29 05:47:52,272 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96080.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:48:23,998 INFO [train.py:904] (6/8) Epoch 10, batch 4750, loss[loss=0.2062, simple_loss=0.2973, pruned_loss=0.05753, over 15381.00 frames. ], tot_loss[loss=0.206, simple_loss=0.2897, pruned_loss=0.06115, over 3200962.27 frames. ], batch size: 190, lr: 6.86e-03, grad_scale: 8.0 2023-04-29 05:49:08,951 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.162e+02 2.549e+02 3.399e+02 6.347e+02, threshold=5.097e+02, percent-clipped=5.0 2023-04-29 05:49:23,236 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96141.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:49:24,388 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6812, 4.5014, 4.7067, 4.9214, 5.0617, 4.5368, 5.0306, 5.0901], device='cuda:6'), covar=tensor([0.1396, 0.1034, 0.1494, 0.0544, 0.0414, 0.0743, 0.0467, 0.0422], device='cuda:6'), in_proj_covar=tensor([0.0495, 0.0614, 0.0751, 0.0623, 0.0465, 0.0480, 0.0488, 0.0551], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:49:38,025 INFO [train.py:904] (6/8) Epoch 10, batch 4800, loss[loss=0.1923, simple_loss=0.2869, pruned_loss=0.04887, over 16826.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2863, pruned_loss=0.05941, over 3198122.23 frames. ], batch size: 96, lr: 6.86e-03, grad_scale: 8.0 2023-04-29 05:49:58,829 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96165.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:50:03,011 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96168.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 05:50:54,658 INFO [train.py:904] (6/8) Epoch 10, batch 4850, loss[loss=0.2039, simple_loss=0.2953, pruned_loss=0.05624, over 16757.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2883, pruned_loss=0.05942, over 3182257.83 frames. ], batch size: 124, lr: 6.86e-03, grad_scale: 8.0 2023-04-29 05:50:56,282 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96203.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 05:51:32,219 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96226.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:51:40,157 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.214e+02 2.693e+02 3.150e+02 7.967e+02, threshold=5.386e+02, percent-clipped=5.0 2023-04-29 05:52:10,086 INFO [train.py:904] (6/8) Epoch 10, batch 4900, loss[loss=0.2021, simple_loss=0.295, pruned_loss=0.05465, over 15313.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2876, pruned_loss=0.05829, over 3181366.66 frames. ], batch size: 190, lr: 6.86e-03, grad_scale: 8.0 2023-04-29 05:52:42,998 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96273.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:52:44,372 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.99 vs. limit=5.0 2023-04-29 05:53:24,548 INFO [train.py:904] (6/8) Epoch 10, batch 4950, loss[loss=0.1955, simple_loss=0.2894, pruned_loss=0.05083, over 16812.00 frames. ], tot_loss[loss=0.2013, simple_loss=0.2873, pruned_loss=0.05764, over 3192656.27 frames. ], batch size: 83, lr: 6.85e-03, grad_scale: 8.0 2023-04-29 05:53:52,125 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96321.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:53:56,345 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96324.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:53:56,448 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4007, 2.0718, 2.0279, 4.0271, 1.9047, 2.5426, 2.1304, 2.3431], device='cuda:6'), covar=tensor([0.0835, 0.2871, 0.2052, 0.0361, 0.3382, 0.1850, 0.2718, 0.2404], device='cuda:6'), in_proj_covar=tensor([0.0354, 0.0382, 0.0317, 0.0316, 0.0402, 0.0435, 0.0344, 0.0447], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:54:05,870 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.230e+02 2.596e+02 3.100e+02 4.863e+02, threshold=5.193e+02, percent-clipped=0.0 2023-04-29 05:54:12,439 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96336.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:54:33,261 INFO [train.py:904] (6/8) Epoch 10, batch 5000, loss[loss=0.2142, simple_loss=0.3089, pruned_loss=0.05975, over 16841.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2887, pruned_loss=0.05794, over 3203711.71 frames. ], batch size: 116, lr: 6.85e-03, grad_scale: 8.0 2023-04-29 05:55:21,038 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96385.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:55:28,059 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96390.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:55:38,879 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9680, 3.9785, 1.7631, 4.7324, 2.7723, 4.5428, 2.3558, 3.0280], device='cuda:6'), covar=tensor([0.0166, 0.0286, 0.1979, 0.0048, 0.0917, 0.0279, 0.1460, 0.0719], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0163, 0.0186, 0.0120, 0.0166, 0.0200, 0.0191, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 05:55:44,796 INFO [train.py:904] (6/8) Epoch 10, batch 5050, loss[loss=0.1934, simple_loss=0.2826, pruned_loss=0.05214, over 16725.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2891, pruned_loss=0.0577, over 3200300.17 frames. ], batch size: 124, lr: 6.85e-03, grad_scale: 8.0 2023-04-29 05:56:27,889 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.480e+02 2.386e+02 2.799e+02 3.376e+02 6.474e+02, threshold=5.598e+02, percent-clipped=3.0 2023-04-29 05:56:33,452 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96436.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:56:55,115 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96451.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:56:55,806 INFO [train.py:904] (6/8) Epoch 10, batch 5100, loss[loss=0.1819, simple_loss=0.2705, pruned_loss=0.04664, over 16691.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2876, pruned_loss=0.05695, over 3192784.01 frames. ], batch size: 134, lr: 6.85e-03, grad_scale: 8.0 2023-04-29 05:57:20,342 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96468.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 05:58:03,017 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3811, 4.2631, 4.2803, 3.5643, 4.2882, 1.4832, 3.9833, 4.0854], device='cuda:6'), covar=tensor([0.0078, 0.0073, 0.0105, 0.0366, 0.0068, 0.2449, 0.0122, 0.0168], device='cuda:6'), in_proj_covar=tensor([0.0122, 0.0111, 0.0156, 0.0153, 0.0127, 0.0171, 0.0145, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:58:08,611 INFO [train.py:904] (6/8) Epoch 10, batch 5150, loss[loss=0.2155, simple_loss=0.2993, pruned_loss=0.06587, over 11695.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2874, pruned_loss=0.05581, over 3192573.79 frames. ], batch size: 248, lr: 6.85e-03, grad_scale: 8.0 2023-04-29 05:58:11,190 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96503.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:58:28,688 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2262, 2.6944, 2.7574, 4.8215, 2.5903, 3.0707, 2.7747, 3.0397], device='cuda:6'), covar=tensor([0.0663, 0.2612, 0.1749, 0.0287, 0.2996, 0.1928, 0.2411, 0.2234], device='cuda:6'), in_proj_covar=tensor([0.0355, 0.0384, 0.0319, 0.0318, 0.0404, 0.0437, 0.0344, 0.0448], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 05:58:30,253 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96516.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 05:58:37,400 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96521.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:58:52,021 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.088e+02 2.479e+02 2.937e+02 7.130e+02, threshold=4.958e+02, percent-clipped=1.0 2023-04-29 05:59:21,946 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96551.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:59:22,743 INFO [train.py:904] (6/8) Epoch 10, batch 5200, loss[loss=0.1652, simple_loss=0.2503, pruned_loss=0.04008, over 16519.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2856, pruned_loss=0.05548, over 3192047.38 frames. ], batch size: 68, lr: 6.85e-03, grad_scale: 8.0 2023-04-29 05:59:28,683 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8460, 3.3816, 3.2110, 1.9722, 2.8026, 2.3055, 3.3905, 3.4652], device='cuda:6'), covar=tensor([0.0210, 0.0576, 0.0585, 0.1635, 0.0773, 0.0845, 0.0557, 0.0613], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0143, 0.0157, 0.0142, 0.0135, 0.0124, 0.0135, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 05:59:43,071 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-29 06:00:35,342 INFO [train.py:904] (6/8) Epoch 10, batch 5250, loss[loss=0.1738, simple_loss=0.2663, pruned_loss=0.04068, over 16921.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2837, pruned_loss=0.05527, over 3196382.52 frames. ], batch size: 96, lr: 6.84e-03, grad_scale: 4.0 2023-04-29 06:01:21,025 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.301e+02 2.656e+02 3.144e+02 5.435e+02, threshold=5.311e+02, percent-clipped=2.0 2023-04-29 06:01:26,378 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96636.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:01:48,983 INFO [train.py:904] (6/8) Epoch 10, batch 5300, loss[loss=0.1715, simple_loss=0.2536, pruned_loss=0.04471, over 16702.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.279, pruned_loss=0.05374, over 3199473.68 frames. ], batch size: 83, lr: 6.84e-03, grad_scale: 4.0 2023-04-29 06:02:04,547 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4120, 2.0825, 2.2581, 3.9306, 1.9970, 2.5700, 2.1948, 2.3261], device='cuda:6'), covar=tensor([0.0901, 0.2922, 0.2059, 0.0403, 0.3516, 0.1973, 0.2836, 0.2867], device='cuda:6'), in_proj_covar=tensor([0.0354, 0.0380, 0.0317, 0.0318, 0.0401, 0.0434, 0.0342, 0.0445], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:02:30,226 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96680.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:02:36,966 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96684.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:03:03,535 INFO [train.py:904] (6/8) Epoch 10, batch 5350, loss[loss=0.2123, simple_loss=0.2947, pruned_loss=0.06492, over 17037.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2777, pruned_loss=0.05311, over 3200596.37 frames. ], batch size: 53, lr: 6.84e-03, grad_scale: 4.0 2023-04-29 06:03:26,751 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3530, 4.0164, 3.1482, 1.7794, 2.7325, 2.3547, 3.5961, 4.0339], device='cuda:6'), covar=tensor([0.0238, 0.0510, 0.0827, 0.2154, 0.1085, 0.0996, 0.0623, 0.0630], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0142, 0.0156, 0.0141, 0.0134, 0.0123, 0.0135, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 06:03:48,673 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.305e+02 2.858e+02 3.473e+02 6.088e+02, threshold=5.716e+02, percent-clipped=3.0 2023-04-29 06:03:51,959 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.75 vs. limit=5.0 2023-04-29 06:03:53,479 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96736.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:04:08,473 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96746.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:04:08,717 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9209, 2.0184, 2.2560, 3.1615, 2.1064, 2.2818, 2.1976, 2.1396], device='cuda:6'), covar=tensor([0.0904, 0.2694, 0.1771, 0.0513, 0.3322, 0.1845, 0.2614, 0.2638], device='cuda:6'), in_proj_covar=tensor([0.0355, 0.0381, 0.0318, 0.0318, 0.0402, 0.0436, 0.0343, 0.0446], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:04:15,627 INFO [train.py:904] (6/8) Epoch 10, batch 5400, loss[loss=0.1954, simple_loss=0.2871, pruned_loss=0.05178, over 15335.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.28, pruned_loss=0.0536, over 3198625.74 frames. ], batch size: 190, lr: 6.84e-03, grad_scale: 4.0 2023-04-29 06:04:18,200 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-04-29 06:04:52,064 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96776.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:05:03,164 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96784.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:05:12,475 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.58 vs. limit=5.0 2023-04-29 06:05:13,334 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9427, 2.7669, 2.2740, 2.6115, 3.1073, 2.7880, 3.6863, 3.4157], device='cuda:6'), covar=tensor([0.0034, 0.0228, 0.0310, 0.0245, 0.0162, 0.0238, 0.0100, 0.0134], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0192, 0.0189, 0.0188, 0.0191, 0.0193, 0.0193, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:05:31,412 INFO [train.py:904] (6/8) Epoch 10, batch 5450, loss[loss=0.2178, simple_loss=0.2985, pruned_loss=0.06853, over 16642.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2823, pruned_loss=0.05482, over 3192288.70 frames. ], batch size: 57, lr: 6.84e-03, grad_scale: 4.0 2023-04-29 06:06:02,299 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96821.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:06:20,032 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.545e+02 3.377e+02 4.195e+02 1.255e+03, threshold=6.754e+02, percent-clipped=10.0 2023-04-29 06:06:27,774 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96837.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:06:49,563 INFO [train.py:904] (6/8) Epoch 10, batch 5500, loss[loss=0.2075, simple_loss=0.2959, pruned_loss=0.05956, over 17095.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2907, pruned_loss=0.06013, over 3175346.16 frames. ], batch size: 47, lr: 6.83e-03, grad_scale: 4.0 2023-04-29 06:06:53,928 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8596, 3.0839, 3.0787, 2.0922, 2.8605, 3.1422, 3.0410, 1.8373], device='cuda:6'), covar=tensor([0.0397, 0.0043, 0.0040, 0.0300, 0.0083, 0.0082, 0.0065, 0.0356], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0065, 0.0067, 0.0122, 0.0076, 0.0085, 0.0074, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 06:07:00,970 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-29 06:07:17,927 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96869.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:07:45,610 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-04-29 06:08:08,495 INFO [train.py:904] (6/8) Epoch 10, batch 5550, loss[loss=0.2243, simple_loss=0.3125, pruned_loss=0.06799, over 17148.00 frames. ], tot_loss[loss=0.2162, simple_loss=0.2995, pruned_loss=0.06643, over 3159598.60 frames. ], batch size: 46, lr: 6.83e-03, grad_scale: 4.0 2023-04-29 06:08:25,922 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.05 vs. limit=5.0 2023-04-29 06:09:01,353 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.332e+02 3.697e+02 4.404e+02 5.267e+02 9.227e+02, threshold=8.809e+02, percent-clipped=8.0 2023-04-29 06:09:28,015 INFO [train.py:904] (6/8) Epoch 10, batch 5600, loss[loss=0.2138, simple_loss=0.2968, pruned_loss=0.06541, over 17128.00 frames. ], tot_loss[loss=0.2244, simple_loss=0.3053, pruned_loss=0.07181, over 3101067.43 frames. ], batch size: 47, lr: 6.83e-03, grad_scale: 2.0 2023-04-29 06:09:32,680 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-29 06:10:15,934 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96980.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:10:51,186 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97000.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:10:53,954 INFO [train.py:904] (6/8) Epoch 10, batch 5650, loss[loss=0.1944, simple_loss=0.2693, pruned_loss=0.05981, over 16339.00 frames. ], tot_loss[loss=0.2328, simple_loss=0.3114, pruned_loss=0.07708, over 3066053.72 frames. ], batch size: 35, lr: 6.83e-03, grad_scale: 2.0 2023-04-29 06:11:34,646 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=97028.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:11:41,716 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2023-04-29 06:11:43,933 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.521e+02 3.909e+02 4.849e+02 5.778e+02 1.289e+03, threshold=9.698e+02, percent-clipped=3.0 2023-04-29 06:12:02,711 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97046.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:12:11,111 INFO [train.py:904] (6/8) Epoch 10, batch 5700, loss[loss=0.2941, simple_loss=0.3476, pruned_loss=0.1203, over 11907.00 frames. ], tot_loss[loss=0.2348, simple_loss=0.3131, pruned_loss=0.07822, over 3070942.90 frames. ], batch size: 248, lr: 6.83e-03, grad_scale: 2.0 2023-04-29 06:12:25,397 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97061.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:12:53,520 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7457, 1.2833, 1.6295, 1.5734, 1.7867, 1.8828, 1.4681, 1.7512], device='cuda:6'), covar=tensor([0.0160, 0.0226, 0.0129, 0.0182, 0.0155, 0.0101, 0.0242, 0.0074], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0165, 0.0150, 0.0154, 0.0161, 0.0117, 0.0167, 0.0109], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 06:13:16,963 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=97094.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:13:29,501 INFO [train.py:904] (6/8) Epoch 10, batch 5750, loss[loss=0.2439, simple_loss=0.3269, pruned_loss=0.08042, over 16892.00 frames. ], tot_loss[loss=0.2393, simple_loss=0.3167, pruned_loss=0.08094, over 3039639.82 frames. ], batch size: 116, lr: 6.83e-03, grad_scale: 2.0 2023-04-29 06:14:12,274 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8878, 2.0210, 2.3486, 3.1597, 2.1682, 2.3390, 2.2359, 2.1582], device='cuda:6'), covar=tensor([0.0933, 0.3004, 0.1655, 0.0476, 0.3151, 0.1864, 0.2600, 0.2622], device='cuda:6'), in_proj_covar=tensor([0.0356, 0.0382, 0.0319, 0.0318, 0.0403, 0.0435, 0.0344, 0.0448], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:14:17,785 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97132.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:14:22,108 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.407e+02 3.370e+02 4.173e+02 4.986e+02 1.197e+03, threshold=8.346e+02, percent-clipped=2.0 2023-04-29 06:14:49,726 INFO [train.py:904] (6/8) Epoch 10, batch 5800, loss[loss=0.1999, simple_loss=0.2871, pruned_loss=0.05635, over 16610.00 frames. ], tot_loss[loss=0.2377, simple_loss=0.3162, pruned_loss=0.07966, over 3049056.23 frames. ], batch size: 57, lr: 6.82e-03, grad_scale: 2.0 2023-04-29 06:16:07,864 INFO [train.py:904] (6/8) Epoch 10, batch 5850, loss[loss=0.2242, simple_loss=0.3076, pruned_loss=0.07041, over 16756.00 frames. ], tot_loss[loss=0.2333, simple_loss=0.3126, pruned_loss=0.07694, over 3053724.77 frames. ], batch size: 124, lr: 6.82e-03, grad_scale: 2.0 2023-04-29 06:17:00,834 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.062e+02 2.965e+02 3.707e+02 4.622e+02 9.015e+02, threshold=7.415e+02, percent-clipped=1.0 2023-04-29 06:17:28,941 INFO [train.py:904] (6/8) Epoch 10, batch 5900, loss[loss=0.2111, simple_loss=0.2986, pruned_loss=0.06177, over 16368.00 frames. ], tot_loss[loss=0.2325, simple_loss=0.3119, pruned_loss=0.07657, over 3052295.50 frames. ], batch size: 146, lr: 6.82e-03, grad_scale: 2.0 2023-04-29 06:17:45,176 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97260.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:17:59,514 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7559, 3.6805, 3.8821, 3.6759, 3.8167, 4.1709, 3.8744, 3.7090], device='cuda:6'), covar=tensor([0.2155, 0.2010, 0.1723, 0.2361, 0.2762, 0.1819, 0.1400, 0.2374], device='cuda:6'), in_proj_covar=tensor([0.0344, 0.0475, 0.0501, 0.0413, 0.0550, 0.0537, 0.0410, 0.0565], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 06:18:49,548 INFO [train.py:904] (6/8) Epoch 10, batch 5950, loss[loss=0.2271, simple_loss=0.3083, pruned_loss=0.07295, over 15367.00 frames. ], tot_loss[loss=0.2314, simple_loss=0.3123, pruned_loss=0.07525, over 3068407.27 frames. ], batch size: 191, lr: 6.82e-03, grad_scale: 2.0 2023-04-29 06:19:17,385 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7947, 3.8833, 3.1531, 2.3155, 2.8561, 2.4353, 4.0243, 3.8038], device='cuda:6'), covar=tensor([0.2449, 0.0682, 0.1434, 0.2108, 0.2098, 0.1591, 0.0437, 0.0846], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0257, 0.0280, 0.0277, 0.0281, 0.0217, 0.0266, 0.0290], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:19:20,406 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97321.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:19:27,428 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9320, 2.2708, 1.8821, 1.9488, 2.7539, 2.3280, 2.8489, 2.8630], device='cuda:6'), covar=tensor([0.0084, 0.0253, 0.0364, 0.0355, 0.0146, 0.0284, 0.0139, 0.0166], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0192, 0.0191, 0.0190, 0.0191, 0.0194, 0.0196, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:19:36,475 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2204, 3.7970, 3.7625, 2.3496, 3.4360, 3.7475, 3.5554, 1.8716], device='cuda:6'), covar=tensor([0.0467, 0.0028, 0.0034, 0.0363, 0.0076, 0.0096, 0.0056, 0.0411], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0066, 0.0067, 0.0124, 0.0076, 0.0086, 0.0074, 0.0117], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 06:19:41,580 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.885e+02 2.897e+02 3.264e+02 4.239e+02 7.603e+02, threshold=6.529e+02, percent-clipped=1.0 2023-04-29 06:19:52,876 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97342.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:20:01,475 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97347.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:20:09,122 INFO [train.py:904] (6/8) Epoch 10, batch 6000, loss[loss=0.2095, simple_loss=0.2968, pruned_loss=0.06103, over 16929.00 frames. ], tot_loss[loss=0.2295, simple_loss=0.3109, pruned_loss=0.07402, over 3067925.78 frames. ], batch size: 109, lr: 6.82e-03, grad_scale: 4.0 2023-04-29 06:20:09,122 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 06:20:23,728 INFO [train.py:938] (6/8) Epoch 10, validation: loss=0.165, simple_loss=0.2783, pruned_loss=0.02583, over 944034.00 frames. 2023-04-29 06:20:23,729 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 06:20:30,455 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97356.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:20:58,023 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97374.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:21:42,480 INFO [train.py:904] (6/8) Epoch 10, batch 6050, loss[loss=0.22, simple_loss=0.3064, pruned_loss=0.06684, over 16878.00 frames. ], tot_loss[loss=0.227, simple_loss=0.3087, pruned_loss=0.07272, over 3096532.79 frames. ], batch size: 109, lr: 6.82e-03, grad_scale: 2.0 2023-04-29 06:21:45,451 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97403.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 06:21:52,791 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97408.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:22:30,438 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97432.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:22:35,027 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97435.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:22:35,672 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.881e+02 3.012e+02 3.614e+02 4.338e+02 6.849e+02, threshold=7.229e+02, percent-clipped=1.0 2023-04-29 06:23:02,109 INFO [train.py:904] (6/8) Epoch 10, batch 6100, loss[loss=0.1942, simple_loss=0.2854, pruned_loss=0.0515, over 17272.00 frames. ], tot_loss[loss=0.2251, simple_loss=0.3081, pruned_loss=0.07108, over 3112396.49 frames. ], batch size: 52, lr: 6.81e-03, grad_scale: 2.0 2023-04-29 06:23:49,434 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=97480.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:24:13,564 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8964, 4.9177, 4.7390, 4.5651, 4.1738, 4.7920, 4.8421, 4.4018], device='cuda:6'), covar=tensor([0.0629, 0.0413, 0.0330, 0.0280, 0.1252, 0.0485, 0.0310, 0.0670], device='cuda:6'), in_proj_covar=tensor([0.0233, 0.0291, 0.0277, 0.0253, 0.0298, 0.0289, 0.0187, 0.0319], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:24:23,702 INFO [train.py:904] (6/8) Epoch 10, batch 6150, loss[loss=0.1949, simple_loss=0.2779, pruned_loss=0.05595, over 16616.00 frames. ], tot_loss[loss=0.2232, simple_loss=0.3057, pruned_loss=0.07037, over 3117732.86 frames. ], batch size: 62, lr: 6.81e-03, grad_scale: 2.0 2023-04-29 06:25:17,520 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.161e+02 3.229e+02 3.939e+02 5.020e+02 8.476e+02, threshold=7.879e+02, percent-clipped=2.0 2023-04-29 06:25:21,152 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2620, 4.2909, 4.7427, 4.7280, 4.6545, 4.3886, 4.3413, 4.1966], device='cuda:6'), covar=tensor([0.0312, 0.0471, 0.0302, 0.0343, 0.0453, 0.0335, 0.0889, 0.0433], device='cuda:6'), in_proj_covar=tensor([0.0319, 0.0326, 0.0327, 0.0314, 0.0375, 0.0349, 0.0450, 0.0280], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 06:25:41,180 INFO [train.py:904] (6/8) Epoch 10, batch 6200, loss[loss=0.2064, simple_loss=0.2874, pruned_loss=0.06271, over 16696.00 frames. ], tot_loss[loss=0.2225, simple_loss=0.304, pruned_loss=0.07052, over 3102641.17 frames. ], batch size: 134, lr: 6.81e-03, grad_scale: 2.0 2023-04-29 06:25:46,664 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97555.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:26:57,943 INFO [train.py:904] (6/8) Epoch 10, batch 6250, loss[loss=0.2214, simple_loss=0.3169, pruned_loss=0.06292, over 16838.00 frames. ], tot_loss[loss=0.2228, simple_loss=0.3044, pruned_loss=0.07062, over 3119275.14 frames. ], batch size: 83, lr: 6.81e-03, grad_scale: 2.0 2023-04-29 06:26:59,643 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4920, 3.5540, 3.2519, 3.0384, 3.1433, 3.4177, 3.2949, 3.2434], device='cuda:6'), covar=tensor([0.0537, 0.0431, 0.0219, 0.0214, 0.0533, 0.0372, 0.1053, 0.0413], device='cuda:6'), in_proj_covar=tensor([0.0237, 0.0296, 0.0279, 0.0257, 0.0302, 0.0292, 0.0190, 0.0324], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:27:18,727 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97616.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:27:18,891 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97616.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:27:47,812 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.754e+02 3.287e+02 3.846e+02 4.993e+02 1.247e+03, threshold=7.692e+02, percent-clipped=7.0 2023-04-29 06:28:11,795 INFO [train.py:904] (6/8) Epoch 10, batch 6300, loss[loss=0.1962, simple_loss=0.2902, pruned_loss=0.05108, over 17233.00 frames. ], tot_loss[loss=0.2223, simple_loss=0.3042, pruned_loss=0.07024, over 3097810.10 frames. ], batch size: 45, lr: 6.81e-03, grad_scale: 2.0 2023-04-29 06:28:18,590 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97656.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:28:48,994 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97674.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 06:29:24,504 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97698.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 06:29:30,238 INFO [train.py:904] (6/8) Epoch 10, batch 6350, loss[loss=0.2292, simple_loss=0.3056, pruned_loss=0.07644, over 16753.00 frames. ], tot_loss[loss=0.2248, simple_loss=0.3053, pruned_loss=0.07211, over 3082296.23 frames. ], batch size: 89, lr: 6.81e-03, grad_scale: 2.0 2023-04-29 06:29:31,950 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97703.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:29:33,675 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=97704.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:30:13,649 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97730.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:30:22,265 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97735.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 06:30:22,844 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 3.422e+02 4.156e+02 5.083e+02 8.318e+02, threshold=8.312e+02, percent-clipped=1.0 2023-04-29 06:30:46,248 INFO [train.py:904] (6/8) Epoch 10, batch 6400, loss[loss=0.2325, simple_loss=0.3125, pruned_loss=0.07627, over 16788.00 frames. ], tot_loss[loss=0.2244, simple_loss=0.3048, pruned_loss=0.07195, over 3095814.98 frames. ], batch size: 83, lr: 6.80e-03, grad_scale: 4.0 2023-04-29 06:30:51,805 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-29 06:31:46,818 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9545, 4.9046, 4.8486, 4.5449, 4.4384, 4.8644, 4.7974, 4.4866], device='cuda:6'), covar=tensor([0.0699, 0.0616, 0.0274, 0.0308, 0.0963, 0.0438, 0.0367, 0.0825], device='cuda:6'), in_proj_covar=tensor([0.0237, 0.0296, 0.0277, 0.0255, 0.0300, 0.0289, 0.0189, 0.0323], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:32:00,788 INFO [train.py:904] (6/8) Epoch 10, batch 6450, loss[loss=0.2135, simple_loss=0.2966, pruned_loss=0.06518, over 16261.00 frames. ], tot_loss[loss=0.2226, simple_loss=0.3039, pruned_loss=0.07065, over 3096224.39 frames. ], batch size: 165, lr: 6.80e-03, grad_scale: 4.0 2023-04-29 06:32:57,253 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.110e+02 2.867e+02 3.528e+02 4.342e+02 1.041e+03, threshold=7.056e+02, percent-clipped=1.0 2023-04-29 06:33:21,803 INFO [train.py:904] (6/8) Epoch 10, batch 6500, loss[loss=0.2109, simple_loss=0.2881, pruned_loss=0.06684, over 16581.00 frames. ], tot_loss[loss=0.2205, simple_loss=0.3018, pruned_loss=0.0696, over 3099909.71 frames. ], batch size: 62, lr: 6.80e-03, grad_scale: 4.0 2023-04-29 06:34:17,502 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97888.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:34:33,144 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8694, 4.1886, 4.4253, 4.4011, 4.3562, 4.0869, 3.7577, 3.9942], device='cuda:6'), covar=tensor([0.0622, 0.0743, 0.0665, 0.0655, 0.0742, 0.0646, 0.1403, 0.0588], device='cuda:6'), in_proj_covar=tensor([0.0314, 0.0321, 0.0321, 0.0309, 0.0370, 0.0342, 0.0442, 0.0276], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 06:34:41,158 INFO [train.py:904] (6/8) Epoch 10, batch 6550, loss[loss=0.2303, simple_loss=0.3299, pruned_loss=0.0654, over 16502.00 frames. ], tot_loss[loss=0.2242, simple_loss=0.3052, pruned_loss=0.07156, over 3089857.20 frames. ], batch size: 68, lr: 6.80e-03, grad_scale: 4.0 2023-04-29 06:34:56,262 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97911.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:35:04,983 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97916.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:35:35,377 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.197e+02 3.236e+02 3.952e+02 5.112e+02 9.277e+02, threshold=7.905e+02, percent-clipped=2.0 2023-04-29 06:35:37,247 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1489, 3.2012, 1.6465, 3.4351, 2.3361, 3.4907, 1.8385, 2.5155], device='cuda:6'), covar=tensor([0.0230, 0.0366, 0.1720, 0.0163, 0.0837, 0.0511, 0.1584, 0.0702], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0160, 0.0184, 0.0118, 0.0163, 0.0199, 0.0188, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 06:35:41,130 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9428, 2.3637, 1.7412, 1.9913, 2.8206, 2.3770, 2.9521, 2.9589], device='cuda:6'), covar=tensor([0.0095, 0.0259, 0.0406, 0.0361, 0.0149, 0.0270, 0.0152, 0.0165], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0193, 0.0192, 0.0192, 0.0191, 0.0197, 0.0198, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:35:56,894 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97949.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:36:00,714 INFO [train.py:904] (6/8) Epoch 10, batch 6600, loss[loss=0.2615, simple_loss=0.3404, pruned_loss=0.0913, over 16761.00 frames. ], tot_loss[loss=0.2251, simple_loss=0.3071, pruned_loss=0.07159, over 3102906.13 frames. ], batch size: 124, lr: 6.80e-03, grad_scale: 4.0 2023-04-29 06:36:19,452 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=97964.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:36:22,426 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-29 06:37:14,842 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97998.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:37:22,482 INFO [train.py:904] (6/8) Epoch 10, batch 6650, loss[loss=0.2229, simple_loss=0.3021, pruned_loss=0.07189, over 16832.00 frames. ], tot_loss[loss=0.226, simple_loss=0.3073, pruned_loss=0.07232, over 3100553.25 frames. ], batch size: 116, lr: 6.79e-03, grad_scale: 4.0 2023-04-29 06:37:24,752 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98003.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:37:35,209 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0378, 4.0108, 3.9191, 3.2858, 3.9623, 1.7072, 3.7516, 3.6045], device='cuda:6'), covar=tensor([0.0086, 0.0074, 0.0123, 0.0322, 0.0078, 0.2417, 0.0113, 0.0197], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0109, 0.0156, 0.0151, 0.0127, 0.0171, 0.0144, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:38:05,465 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98030.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 06:38:05,499 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98030.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:38:14,887 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.646e+02 3.281e+02 3.856e+02 4.909e+02 8.809e+02, threshold=7.713e+02, percent-clipped=1.0 2023-04-29 06:38:31,003 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98046.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:38:34,513 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-04-29 06:38:39,003 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98051.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:38:39,792 INFO [train.py:904] (6/8) Epoch 10, batch 6700, loss[loss=0.2277, simple_loss=0.3076, pruned_loss=0.07391, over 16961.00 frames. ], tot_loss[loss=0.2254, simple_loss=0.3063, pruned_loss=0.07223, over 3091646.93 frames. ], batch size: 109, lr: 6.79e-03, grad_scale: 4.0 2023-04-29 06:39:20,940 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98078.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:39:23,128 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.06 vs. limit=5.0 2023-04-29 06:39:57,931 INFO [train.py:904] (6/8) Epoch 10, batch 6750, loss[loss=0.2223, simple_loss=0.3039, pruned_loss=0.07036, over 16504.00 frames. ], tot_loss[loss=0.225, simple_loss=0.3055, pruned_loss=0.07224, over 3093970.64 frames. ], batch size: 146, lr: 6.79e-03, grad_scale: 4.0 2023-04-29 06:40:04,474 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98106.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:40:28,069 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6429, 6.0662, 5.6721, 5.8322, 5.2394, 5.1691, 5.4378, 6.1133], device='cuda:6'), covar=tensor([0.0998, 0.0686, 0.1034, 0.0655, 0.0747, 0.0616, 0.1024, 0.0799], device='cuda:6'), in_proj_covar=tensor([0.0510, 0.0638, 0.0534, 0.0445, 0.0399, 0.0415, 0.0535, 0.0484], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:40:42,024 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 06:40:49,799 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.169e+02 3.584e+02 4.478e+02 5.468e+02 7.454e+02, threshold=8.956e+02, percent-clipped=0.0 2023-04-29 06:41:14,647 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.27 vs. limit=5.0 2023-04-29 06:41:15,032 INFO [train.py:904] (6/8) Epoch 10, batch 6800, loss[loss=0.1945, simple_loss=0.2785, pruned_loss=0.0553, over 16520.00 frames. ], tot_loss[loss=0.2249, simple_loss=0.3055, pruned_loss=0.07213, over 3108528.54 frames. ], batch size: 68, lr: 6.79e-03, grad_scale: 8.0 2023-04-29 06:41:39,154 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98167.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:42:33,969 INFO [train.py:904] (6/8) Epoch 10, batch 6850, loss[loss=0.2089, simple_loss=0.3118, pruned_loss=0.053, over 16689.00 frames. ], tot_loss[loss=0.2256, simple_loss=0.3068, pruned_loss=0.0722, over 3131360.93 frames. ], batch size: 57, lr: 6.79e-03, grad_scale: 8.0 2023-04-29 06:42:47,665 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98211.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:43:24,607 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.168e+02 2.932e+02 3.586e+02 4.673e+02 8.284e+02, threshold=7.173e+02, percent-clipped=0.0 2023-04-29 06:43:37,308 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98244.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:43:49,047 INFO [train.py:904] (6/8) Epoch 10, batch 6900, loss[loss=0.2385, simple_loss=0.3184, pruned_loss=0.07934, over 16903.00 frames. ], tot_loss[loss=0.2259, simple_loss=0.3091, pruned_loss=0.07132, over 3154449.54 frames. ], batch size: 109, lr: 6.79e-03, grad_scale: 8.0 2023-04-29 06:44:01,761 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98259.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:44:39,512 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4915, 3.5989, 2.8418, 2.1715, 2.3971, 2.2848, 3.7083, 3.3381], device='cuda:6'), covar=tensor([0.2743, 0.0638, 0.1481, 0.2257, 0.2377, 0.1737, 0.0448, 0.1014], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0255, 0.0280, 0.0275, 0.0280, 0.0216, 0.0264, 0.0288], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:45:09,159 INFO [train.py:904] (6/8) Epoch 10, batch 6950, loss[loss=0.1854, simple_loss=0.2783, pruned_loss=0.04624, over 16836.00 frames. ], tot_loss[loss=0.2297, simple_loss=0.3114, pruned_loss=0.07396, over 3109231.79 frames. ], batch size: 102, lr: 6.78e-03, grad_scale: 8.0 2023-04-29 06:45:54,268 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98330.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 06:46:01,763 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 3.204e+02 3.949e+02 4.845e+02 8.099e+02, threshold=7.898e+02, percent-clipped=3.0 2023-04-29 06:46:06,177 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9256, 4.0179, 2.0496, 4.7001, 2.7687, 4.5711, 2.3744, 2.8786], device='cuda:6'), covar=tensor([0.0200, 0.0319, 0.1857, 0.0115, 0.0927, 0.0408, 0.1486, 0.0783], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0160, 0.0185, 0.0117, 0.0164, 0.0201, 0.0189, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 06:46:27,403 INFO [train.py:904] (6/8) Epoch 10, batch 7000, loss[loss=0.218, simple_loss=0.3091, pruned_loss=0.06345, over 17047.00 frames. ], tot_loss[loss=0.2285, simple_loss=0.311, pruned_loss=0.07297, over 3111536.82 frames. ], batch size: 53, lr: 6.78e-03, grad_scale: 8.0 2023-04-29 06:46:52,163 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8264, 5.3658, 5.6197, 5.2695, 5.3379, 5.9766, 5.4191, 5.1499], device='cuda:6'), covar=tensor([0.0886, 0.1562, 0.1693, 0.1881, 0.2396, 0.0825, 0.1299, 0.2244], device='cuda:6'), in_proj_covar=tensor([0.0341, 0.0476, 0.0511, 0.0412, 0.0546, 0.0538, 0.0412, 0.0565], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 06:47:08,044 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98378.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 06:47:43,137 INFO [train.py:904] (6/8) Epoch 10, batch 7050, loss[loss=0.304, simple_loss=0.3495, pruned_loss=0.1293, over 11714.00 frames. ], tot_loss[loss=0.2296, simple_loss=0.3122, pruned_loss=0.07347, over 3099227.70 frames. ], batch size: 248, lr: 6.78e-03, grad_scale: 8.0 2023-04-29 06:47:57,212 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98410.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:48:34,409 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.770e+02 3.125e+02 3.922e+02 4.883e+02 1.063e+03, threshold=7.844e+02, percent-clipped=4.0 2023-04-29 06:48:59,553 INFO [train.py:904] (6/8) Epoch 10, batch 7100, loss[loss=0.2056, simple_loss=0.2907, pruned_loss=0.06025, over 16845.00 frames. ], tot_loss[loss=0.2282, simple_loss=0.3107, pruned_loss=0.07289, over 3096358.07 frames. ], batch size: 102, lr: 6.78e-03, grad_scale: 8.0 2023-04-29 06:49:14,093 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98462.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:49:27,983 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98471.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:50:07,928 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3444, 4.3951, 4.1869, 3.9652, 3.8789, 4.2916, 4.1096, 3.9867], device='cuda:6'), covar=tensor([0.0576, 0.0394, 0.0247, 0.0252, 0.0871, 0.0380, 0.0556, 0.0590], device='cuda:6'), in_proj_covar=tensor([0.0229, 0.0288, 0.0268, 0.0247, 0.0292, 0.0282, 0.0185, 0.0312], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:50:12,813 INFO [train.py:904] (6/8) Epoch 10, batch 7150, loss[loss=0.2411, simple_loss=0.3275, pruned_loss=0.07738, over 16480.00 frames. ], tot_loss[loss=0.2263, simple_loss=0.3083, pruned_loss=0.07218, over 3127323.09 frames. ], batch size: 68, lr: 6.78e-03, grad_scale: 8.0 2023-04-29 06:50:25,465 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9478, 1.8163, 2.1099, 3.3469, 1.8407, 2.1917, 1.9787, 1.9482], device='cuda:6'), covar=tensor([0.1147, 0.3809, 0.2260, 0.0614, 0.4495, 0.2499, 0.3271, 0.3845], device='cuda:6'), in_proj_covar=tensor([0.0357, 0.0384, 0.0320, 0.0316, 0.0409, 0.0433, 0.0346, 0.0449], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:50:51,736 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6809, 3.7721, 3.1372, 2.2445, 2.6899, 2.4060, 4.1005, 3.4939], device='cuda:6'), covar=tensor([0.2604, 0.0739, 0.1390, 0.2151, 0.2214, 0.1661, 0.0451, 0.0966], device='cuda:6'), in_proj_covar=tensor([0.0301, 0.0255, 0.0279, 0.0274, 0.0278, 0.0216, 0.0263, 0.0286], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:50:59,851 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 06:51:03,784 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.197e+02 3.117e+02 3.720e+02 4.761e+02 8.896e+02, threshold=7.439e+02, percent-clipped=3.0 2023-04-29 06:51:15,709 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98543.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:51:16,774 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98544.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:51:29,368 INFO [train.py:904] (6/8) Epoch 10, batch 7200, loss[loss=0.204, simple_loss=0.2975, pruned_loss=0.05524, over 16272.00 frames. ], tot_loss[loss=0.2238, simple_loss=0.3063, pruned_loss=0.0706, over 3127671.45 frames. ], batch size: 165, lr: 6.78e-03, grad_scale: 8.0 2023-04-29 06:51:43,055 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9332, 4.1551, 3.9454, 4.0121, 3.6640, 3.8094, 3.8528, 4.1260], device='cuda:6'), covar=tensor([0.0882, 0.0770, 0.0892, 0.0609, 0.0665, 0.1361, 0.0831, 0.0841], device='cuda:6'), in_proj_covar=tensor([0.0509, 0.0637, 0.0530, 0.0442, 0.0399, 0.0419, 0.0533, 0.0482], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:51:55,365 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7875, 3.8311, 3.1080, 2.3320, 2.9153, 2.4616, 4.1953, 3.5791], device='cuda:6'), covar=tensor([0.2362, 0.0742, 0.1416, 0.1968, 0.2040, 0.1571, 0.0399, 0.0922], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0256, 0.0281, 0.0275, 0.0280, 0.0217, 0.0264, 0.0287], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:52:32,648 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98592.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:52:49,206 INFO [train.py:904] (6/8) Epoch 10, batch 7250, loss[loss=0.259, simple_loss=0.3188, pruned_loss=0.09965, over 11076.00 frames. ], tot_loss[loss=0.2212, simple_loss=0.3038, pruned_loss=0.06928, over 3111817.87 frames. ], batch size: 248, lr: 6.77e-03, grad_scale: 2.0 2023-04-29 06:52:53,607 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98604.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:53:36,656 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0831, 2.3323, 2.2653, 2.7497, 2.0457, 3.2429, 1.7890, 2.5566], device='cuda:6'), covar=tensor([0.0970, 0.0508, 0.0886, 0.0128, 0.0118, 0.0356, 0.1140, 0.0638], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0156, 0.0179, 0.0135, 0.0201, 0.0205, 0.0178, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 06:53:45,148 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.674e+02 3.052e+02 3.513e+02 4.472e+02 1.147e+03, threshold=7.026e+02, percent-clipped=3.0 2023-04-29 06:54:05,881 INFO [train.py:904] (6/8) Epoch 10, batch 7300, loss[loss=0.2628, simple_loss=0.3185, pruned_loss=0.1035, over 11592.00 frames. ], tot_loss[loss=0.2208, simple_loss=0.3032, pruned_loss=0.06926, over 3098145.96 frames. ], batch size: 250, lr: 6.77e-03, grad_scale: 2.0 2023-04-29 06:55:23,864 INFO [train.py:904] (6/8) Epoch 10, batch 7350, loss[loss=0.2023, simple_loss=0.2797, pruned_loss=0.06245, over 16543.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.3032, pruned_loss=0.0698, over 3091131.46 frames. ], batch size: 75, lr: 6.77e-03, grad_scale: 2.0 2023-04-29 06:56:19,126 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.271e+02 3.294e+02 3.832e+02 4.622e+02 6.263e+02, threshold=7.664e+02, percent-clipped=0.0 2023-04-29 06:56:25,902 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4616, 3.4382, 3.3578, 2.7733, 3.3114, 2.0835, 3.0857, 2.7078], device='cuda:6'), covar=tensor([0.0104, 0.0087, 0.0139, 0.0195, 0.0078, 0.1927, 0.0110, 0.0179], device='cuda:6'), in_proj_covar=tensor([0.0121, 0.0107, 0.0154, 0.0149, 0.0125, 0.0172, 0.0141, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 06:56:41,002 INFO [train.py:904] (6/8) Epoch 10, batch 7400, loss[loss=0.2202, simple_loss=0.3006, pruned_loss=0.06987, over 16934.00 frames. ], tot_loss[loss=0.2231, simple_loss=0.3044, pruned_loss=0.07089, over 3066885.99 frames. ], batch size: 109, lr: 6.77e-03, grad_scale: 2.0 2023-04-29 06:56:52,010 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2023-04-29 06:56:57,712 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98762.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:57:03,070 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98766.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:57:59,093 INFO [train.py:904] (6/8) Epoch 10, batch 7450, loss[loss=0.2069, simple_loss=0.3021, pruned_loss=0.05583, over 16797.00 frames. ], tot_loss[loss=0.2245, simple_loss=0.3055, pruned_loss=0.07178, over 3079822.02 frames. ], batch size: 83, lr: 6.77e-03, grad_scale: 2.0 2023-04-29 06:58:14,250 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98810.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:58:57,288 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.965e+02 3.303e+02 3.942e+02 4.934e+02 1.157e+03, threshold=7.883e+02, percent-clipped=3.0 2023-04-29 06:59:19,898 INFO [train.py:904] (6/8) Epoch 10, batch 7500, loss[loss=0.2364, simple_loss=0.3135, pruned_loss=0.07964, over 16211.00 frames. ], tot_loss[loss=0.224, simple_loss=0.3054, pruned_loss=0.07131, over 3073101.51 frames. ], batch size: 165, lr: 6.77e-03, grad_scale: 2.0 2023-04-29 06:59:27,924 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8100, 1.7823, 1.6024, 1.4853, 1.8930, 1.6423, 1.7236, 1.9376], device='cuda:6'), covar=tensor([0.0097, 0.0192, 0.0269, 0.0243, 0.0131, 0.0189, 0.0132, 0.0131], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0193, 0.0191, 0.0190, 0.0191, 0.0195, 0.0195, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:00:20,566 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98890.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:00:27,946 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98895.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:00:34,238 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98899.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:00:38,286 INFO [train.py:904] (6/8) Epoch 10, batch 7550, loss[loss=0.2052, simple_loss=0.297, pruned_loss=0.05673, over 16741.00 frames. ], tot_loss[loss=0.2242, simple_loss=0.3051, pruned_loss=0.07164, over 3077365.80 frames. ], batch size: 89, lr: 6.76e-03, grad_scale: 2.0 2023-04-29 07:00:48,654 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2023-04-29 07:01:04,981 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8457, 4.7427, 4.7379, 3.2938, 4.1392, 4.6640, 4.2265, 2.6431], device='cuda:6'), covar=tensor([0.0404, 0.0029, 0.0028, 0.0294, 0.0061, 0.0091, 0.0060, 0.0346], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0066, 0.0068, 0.0126, 0.0076, 0.0087, 0.0074, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 07:01:30,293 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5408, 4.7831, 4.5559, 4.5329, 4.3152, 4.2809, 4.3068, 4.8778], device='cuda:6'), covar=tensor([0.0895, 0.0818, 0.1020, 0.0712, 0.0737, 0.1048, 0.1013, 0.0804], device='cuda:6'), in_proj_covar=tensor([0.0506, 0.0633, 0.0528, 0.0439, 0.0395, 0.0418, 0.0529, 0.0481], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:01:32,281 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.425e+02 3.280e+02 3.927e+02 5.001e+02 7.830e+02, threshold=7.854e+02, percent-clipped=0.0 2023-04-29 07:01:40,199 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.50 vs. limit=5.0 2023-04-29 07:01:53,299 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98951.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:01:53,973 INFO [train.py:904] (6/8) Epoch 10, batch 7600, loss[loss=0.2318, simple_loss=0.309, pruned_loss=0.07731, over 16276.00 frames. ], tot_loss[loss=0.2236, simple_loss=0.3043, pruned_loss=0.07151, over 3089692.35 frames. ], batch size: 165, lr: 6.76e-03, grad_scale: 4.0 2023-04-29 07:02:01,581 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98956.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:02:24,368 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98971.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:03:11,983 INFO [train.py:904] (6/8) Epoch 10, batch 7650, loss[loss=0.2098, simple_loss=0.2892, pruned_loss=0.06515, over 16703.00 frames. ], tot_loss[loss=0.2244, simple_loss=0.3046, pruned_loss=0.0721, over 3092111.37 frames. ], batch size: 134, lr: 6.76e-03, grad_scale: 4.0 2023-04-29 07:03:59,265 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99032.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:04:08,571 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.981e+02 3.374e+02 4.235e+02 5.349e+02 9.496e+02, threshold=8.471e+02, percent-clipped=6.0 2023-04-29 07:04:28,259 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99051.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:04:29,002 INFO [train.py:904] (6/8) Epoch 10, batch 7700, loss[loss=0.2785, simple_loss=0.3307, pruned_loss=0.1132, over 11388.00 frames. ], tot_loss[loss=0.2248, simple_loss=0.3043, pruned_loss=0.07261, over 3082217.51 frames. ], batch size: 247, lr: 6.76e-03, grad_scale: 4.0 2023-04-29 07:04:50,169 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99066.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:04:51,633 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0533, 3.4495, 3.4787, 1.8784, 2.9463, 2.3352, 3.4830, 3.5931], device='cuda:6'), covar=tensor([0.0249, 0.0652, 0.0583, 0.1891, 0.0802, 0.0967, 0.0652, 0.0826], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0143, 0.0157, 0.0143, 0.0135, 0.0126, 0.0137, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 07:05:08,166 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1969, 5.2145, 4.9858, 4.2542, 5.0603, 1.7358, 4.7412, 4.8334], device='cuda:6'), covar=tensor([0.0064, 0.0053, 0.0123, 0.0348, 0.0065, 0.2387, 0.0114, 0.0147], device='cuda:6'), in_proj_covar=tensor([0.0122, 0.0107, 0.0155, 0.0150, 0.0126, 0.0173, 0.0142, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:05:43,026 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8149, 4.1171, 3.8705, 3.9553, 3.6102, 3.7866, 3.8013, 4.0832], device='cuda:6'), covar=tensor([0.1061, 0.0906, 0.0999, 0.0707, 0.0807, 0.1511, 0.0940, 0.1038], device='cuda:6'), in_proj_covar=tensor([0.0513, 0.0639, 0.0536, 0.0444, 0.0400, 0.0424, 0.0538, 0.0487], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:05:43,695 INFO [train.py:904] (6/8) Epoch 10, batch 7750, loss[loss=0.2808, simple_loss=0.3387, pruned_loss=0.1114, over 11286.00 frames. ], tot_loss[loss=0.2253, simple_loss=0.3049, pruned_loss=0.07287, over 3086355.13 frames. ], batch size: 247, lr: 6.76e-03, grad_scale: 4.0 2023-04-29 07:06:00,269 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99112.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:06:02,527 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99114.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:06:38,056 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.181e+02 3.321e+02 3.885e+02 5.264e+02 1.269e+03, threshold=7.770e+02, percent-clipped=1.0 2023-04-29 07:06:59,326 INFO [train.py:904] (6/8) Epoch 10, batch 7800, loss[loss=0.2855, simple_loss=0.339, pruned_loss=0.116, over 11661.00 frames. ], tot_loss[loss=0.227, simple_loss=0.3062, pruned_loss=0.07391, over 3077391.33 frames. ], batch size: 246, lr: 6.76e-03, grad_scale: 4.0 2023-04-29 07:07:16,275 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8292, 4.7870, 5.3366, 5.3198, 5.2503, 4.9293, 4.8779, 4.5729], device='cuda:6'), covar=tensor([0.0293, 0.0488, 0.0311, 0.0327, 0.0401, 0.0315, 0.0936, 0.0425], device='cuda:6'), in_proj_covar=tensor([0.0315, 0.0323, 0.0325, 0.0310, 0.0374, 0.0344, 0.0444, 0.0281], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 07:07:27,852 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6225, 2.6289, 1.8078, 2.6639, 2.1188, 2.7679, 1.9915, 2.3396], device='cuda:6'), covar=tensor([0.0240, 0.0348, 0.1190, 0.0162, 0.0644, 0.0485, 0.1179, 0.0584], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0162, 0.0186, 0.0119, 0.0164, 0.0203, 0.0191, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 07:08:12,875 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99199.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:08:16,573 INFO [train.py:904] (6/8) Epoch 10, batch 7850, loss[loss=0.1952, simple_loss=0.2921, pruned_loss=0.04913, over 16841.00 frames. ], tot_loss[loss=0.2276, simple_loss=0.3072, pruned_loss=0.07397, over 3060338.41 frames. ], batch size: 96, lr: 6.75e-03, grad_scale: 4.0 2023-04-29 07:08:57,421 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99229.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:09:00,449 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0529, 3.0102, 3.1240, 1.7470, 3.3282, 3.3925, 2.5398, 2.6280], device='cuda:6'), covar=tensor([0.0770, 0.0192, 0.0173, 0.1054, 0.0058, 0.0121, 0.0436, 0.0378], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0099, 0.0085, 0.0137, 0.0068, 0.0097, 0.0121, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 07:09:10,220 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.079e+02 2.953e+02 3.751e+02 4.670e+02 9.934e+02, threshold=7.502e+02, percent-clipped=3.0 2023-04-29 07:09:22,396 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99246.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:09:24,273 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99247.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:09:29,700 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99251.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:09:30,435 INFO [train.py:904] (6/8) Epoch 10, batch 7900, loss[loss=0.2326, simple_loss=0.3193, pruned_loss=0.07294, over 16749.00 frames. ], tot_loss[loss=0.2262, simple_loss=0.3066, pruned_loss=0.07292, over 3091358.86 frames. ], batch size: 124, lr: 6.75e-03, grad_scale: 4.0 2023-04-29 07:09:55,555 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99268.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:10:31,556 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99290.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:10:49,021 INFO [train.py:904] (6/8) Epoch 10, batch 7950, loss[loss=0.2283, simple_loss=0.3074, pruned_loss=0.07461, over 16739.00 frames. ], tot_loss[loss=0.2264, simple_loss=0.3067, pruned_loss=0.07302, over 3094078.28 frames. ], batch size: 134, lr: 6.75e-03, grad_scale: 4.0 2023-04-29 07:11:04,139 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5044, 3.3927, 2.7211, 2.1038, 2.3162, 2.0689, 3.6445, 3.2395], device='cuda:6'), covar=tensor([0.2822, 0.0756, 0.1643, 0.2508, 0.2605, 0.1964, 0.0544, 0.1178], device='cuda:6'), in_proj_covar=tensor([0.0306, 0.0256, 0.0282, 0.0277, 0.0283, 0.0219, 0.0266, 0.0290], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:11:22,932 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99324.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:11:27,245 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99327.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:11:29,270 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6021, 3.6206, 1.7657, 4.0492, 2.5609, 4.0670, 2.1505, 2.7707], device='cuda:6'), covar=tensor([0.0215, 0.0373, 0.1876, 0.0138, 0.0868, 0.0505, 0.1508, 0.0742], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0162, 0.0188, 0.0120, 0.0165, 0.0204, 0.0193, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 07:11:30,467 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:11:37,334 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.31 vs. limit=5.0 2023-04-29 07:11:43,783 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.817e+02 3.138e+02 3.909e+02 4.608e+02 7.916e+02, threshold=7.818e+02, percent-clipped=1.0 2023-04-29 07:12:03,713 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3265, 2.8238, 2.6422, 2.1578, 2.2481, 2.1513, 2.8586, 2.8177], device='cuda:6'), covar=tensor([0.2129, 0.0789, 0.1308, 0.1948, 0.2066, 0.1793, 0.0536, 0.0964], device='cuda:6'), in_proj_covar=tensor([0.0304, 0.0255, 0.0281, 0.0276, 0.0281, 0.0218, 0.0265, 0.0289], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:12:06,029 INFO [train.py:904] (6/8) Epoch 10, batch 8000, loss[loss=0.2169, simple_loss=0.2979, pruned_loss=0.06794, over 17045.00 frames. ], tot_loss[loss=0.2274, simple_loss=0.3075, pruned_loss=0.07367, over 3088594.95 frames. ], batch size: 55, lr: 6.75e-03, grad_scale: 8.0 2023-04-29 07:12:11,585 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 07:12:52,497 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0345, 5.3219, 5.0500, 5.0810, 4.7274, 4.6259, 4.7626, 5.4269], device='cuda:6'), covar=tensor([0.0898, 0.0738, 0.0925, 0.0645, 0.0754, 0.0842, 0.0961, 0.0763], device='cuda:6'), in_proj_covar=tensor([0.0508, 0.0635, 0.0534, 0.0443, 0.0397, 0.0421, 0.0534, 0.0484], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:12:57,304 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99385.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:13:21,634 INFO [train.py:904] (6/8) Epoch 10, batch 8050, loss[loss=0.2258, simple_loss=0.3054, pruned_loss=0.07306, over 16213.00 frames. ], tot_loss[loss=0.2275, simple_loss=0.3075, pruned_loss=0.07369, over 3087230.94 frames. ], batch size: 165, lr: 6.75e-03, grad_scale: 8.0 2023-04-29 07:13:29,919 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99407.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:13:30,092 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5381, 3.0431, 3.0952, 1.7725, 2.6905, 2.0657, 3.0486, 3.2043], device='cuda:6'), covar=tensor([0.0300, 0.0674, 0.0548, 0.1918, 0.0833, 0.0995, 0.0767, 0.0821], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0142, 0.0157, 0.0142, 0.0135, 0.0125, 0.0136, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 07:13:45,670 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5379, 3.6379, 2.8036, 2.1207, 2.5930, 2.2884, 3.9156, 3.5817], device='cuda:6'), covar=tensor([0.2814, 0.0672, 0.1619, 0.2254, 0.2297, 0.1738, 0.0493, 0.0903], device='cuda:6'), in_proj_covar=tensor([0.0305, 0.0256, 0.0283, 0.0277, 0.0283, 0.0219, 0.0266, 0.0291], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:13:50,429 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99420.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:14:18,153 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.046e+02 3.114e+02 3.682e+02 4.495e+02 9.204e+02, threshold=7.364e+02, percent-clipped=1.0 2023-04-29 07:14:39,583 INFO [train.py:904] (6/8) Epoch 10, batch 8100, loss[loss=0.232, simple_loss=0.304, pruned_loss=0.08006, over 11624.00 frames. ], tot_loss[loss=0.226, simple_loss=0.3066, pruned_loss=0.07271, over 3083652.51 frames. ], batch size: 247, lr: 6.75e-03, grad_scale: 8.0 2023-04-29 07:15:00,294 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2853, 4.3697, 4.7731, 4.7365, 4.7259, 4.3955, 4.4137, 4.2066], device='cuda:6'), covar=tensor([0.0296, 0.0460, 0.0296, 0.0358, 0.0382, 0.0353, 0.0817, 0.0463], device='cuda:6'), in_proj_covar=tensor([0.0315, 0.0322, 0.0324, 0.0311, 0.0374, 0.0343, 0.0444, 0.0281], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 07:15:03,926 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99468.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 07:15:18,195 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-04-29 07:15:23,093 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99481.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:15:54,082 INFO [train.py:904] (6/8) Epoch 10, batch 8150, loss[loss=0.1876, simple_loss=0.2746, pruned_loss=0.05025, over 16900.00 frames. ], tot_loss[loss=0.2237, simple_loss=0.304, pruned_loss=0.0717, over 3083020.22 frames. ], batch size: 96, lr: 6.74e-03, grad_scale: 8.0 2023-04-29 07:16:11,894 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9567, 2.7413, 2.7714, 2.0034, 2.5819, 2.1629, 2.7864, 2.9401], device='cuda:6'), covar=tensor([0.0286, 0.0733, 0.0515, 0.1615, 0.0717, 0.0867, 0.0532, 0.0654], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0143, 0.0158, 0.0143, 0.0136, 0.0126, 0.0137, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 07:16:35,901 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99529.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 07:16:49,692 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.193e+02 2.958e+02 3.724e+02 4.411e+02 7.985e+02, threshold=7.447e+02, percent-clipped=3.0 2023-04-29 07:17:01,451 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99546.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:17:08,937 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99551.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:17:09,684 INFO [train.py:904] (6/8) Epoch 10, batch 8200, loss[loss=0.2787, simple_loss=0.3249, pruned_loss=0.1162, over 11246.00 frames. ], tot_loss[loss=0.2212, simple_loss=0.3011, pruned_loss=0.07062, over 3084679.83 frames. ], batch size: 247, lr: 6.74e-03, grad_scale: 4.0 2023-04-29 07:17:16,375 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0421, 2.5393, 2.5370, 1.8350, 2.7444, 2.8210, 2.3542, 2.3982], device='cuda:6'), covar=tensor([0.0652, 0.0194, 0.0187, 0.0917, 0.0092, 0.0187, 0.0420, 0.0394], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0099, 0.0084, 0.0137, 0.0068, 0.0097, 0.0119, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 07:17:36,106 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99568.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:18:04,754 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99585.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:18:11,857 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99589.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:18:20,174 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99594.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:18:28,354 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99599.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:18:32,975 INFO [train.py:904] (6/8) Epoch 10, batch 8250, loss[loss=0.209, simple_loss=0.2966, pruned_loss=0.06069, over 15242.00 frames. ], tot_loss[loss=0.2194, simple_loss=0.3006, pruned_loss=0.06916, over 3055601.14 frames. ], batch size: 190, lr: 6.74e-03, grad_scale: 4.0 2023-04-29 07:18:43,773 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0426, 2.7301, 2.8688, 2.0547, 2.6638, 2.1744, 2.7342, 2.8968], device='cuda:6'), covar=tensor([0.0351, 0.0797, 0.0440, 0.1624, 0.0693, 0.0929, 0.0607, 0.0692], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0139, 0.0155, 0.0140, 0.0134, 0.0123, 0.0135, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 07:19:03,947 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6668, 4.6061, 4.4633, 3.8795, 4.4590, 1.7634, 4.2934, 4.3940], device='cuda:6'), covar=tensor([0.0071, 0.0070, 0.0128, 0.0350, 0.0093, 0.2295, 0.0111, 0.0161], device='cuda:6'), in_proj_covar=tensor([0.0122, 0.0107, 0.0155, 0.0149, 0.0126, 0.0173, 0.0141, 0.0143], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:19:10,831 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99624.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:19:16,606 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99627.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:19:19,907 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99629.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:19:36,138 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.952e+02 2.844e+02 3.459e+02 4.111e+02 7.773e+02, threshold=6.919e+02, percent-clipped=2.0 2023-04-29 07:19:55,548 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99650.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:19:57,217 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9499, 4.2173, 4.0275, 4.0746, 3.7070, 3.8317, 3.8765, 4.2148], device='cuda:6'), covar=tensor([0.0862, 0.0915, 0.0953, 0.0660, 0.0773, 0.1326, 0.0853, 0.0932], device='cuda:6'), in_proj_covar=tensor([0.0507, 0.0635, 0.0532, 0.0441, 0.0397, 0.0420, 0.0530, 0.0481], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:19:57,954 INFO [train.py:904] (6/8) Epoch 10, batch 8300, loss[loss=0.1894, simple_loss=0.2778, pruned_loss=0.05047, over 12243.00 frames. ], tot_loss[loss=0.2144, simple_loss=0.2976, pruned_loss=0.06555, over 3061557.82 frames. ], batch size: 248, lr: 6.74e-03, grad_scale: 4.0 2023-04-29 07:20:27,064 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 07:20:37,234 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99675.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:20:44,715 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99680.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:20:44,806 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99680.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:21:21,641 INFO [train.py:904] (6/8) Epoch 10, batch 8350, loss[loss=0.2299, simple_loss=0.3016, pruned_loss=0.07914, over 12043.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.2965, pruned_loss=0.06321, over 3061850.84 frames. ], batch size: 247, lr: 6.74e-03, grad_scale: 4.0 2023-04-29 07:21:30,548 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99707.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:22:21,849 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.480e+02 2.332e+02 2.884e+02 3.614e+02 8.033e+02, threshold=5.769e+02, percent-clipped=2.0 2023-04-29 07:22:26,802 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99741.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:22:43,906 INFO [train.py:904] (6/8) Epoch 10, batch 8400, loss[loss=0.1839, simple_loss=0.2776, pruned_loss=0.04508, over 16256.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2939, pruned_loss=0.06148, over 3032147.40 frames. ], batch size: 165, lr: 6.74e-03, grad_scale: 8.0 2023-04-29 07:22:49,990 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99755.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:22:57,449 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0522, 2.2815, 2.2657, 4.7701, 2.1552, 2.8542, 2.3353, 2.5559], device='cuda:6'), covar=tensor([0.0638, 0.3356, 0.2262, 0.0239, 0.3911, 0.1969, 0.3010, 0.2952], device='cuda:6'), in_proj_covar=tensor([0.0343, 0.0376, 0.0314, 0.0309, 0.0400, 0.0422, 0.0336, 0.0437], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:23:24,137 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7900, 3.6590, 3.8628, 3.6880, 3.8066, 4.2449, 3.9591, 3.6808], device='cuda:6'), covar=tensor([0.1973, 0.2388, 0.2093, 0.2532, 0.2839, 0.1778, 0.1402, 0.2655], device='cuda:6'), in_proj_covar=tensor([0.0326, 0.0456, 0.0492, 0.0398, 0.0524, 0.0521, 0.0399, 0.0543], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 07:23:25,434 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99776.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:23:32,281 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-29 07:24:06,248 INFO [train.py:904] (6/8) Epoch 10, batch 8450, loss[loss=0.1837, simple_loss=0.2743, pruned_loss=0.04658, over 16861.00 frames. ], tot_loss[loss=0.205, simple_loss=0.2912, pruned_loss=0.05941, over 3012563.23 frames. ], batch size: 116, lr: 6.73e-03, grad_scale: 8.0 2023-04-29 07:24:42,764 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99824.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 07:24:55,141 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-29 07:25:06,071 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.564e+02 2.551e+02 3.005e+02 3.693e+02 6.066e+02, threshold=6.011e+02, percent-clipped=2.0 2023-04-29 07:25:25,949 INFO [train.py:904] (6/8) Epoch 10, batch 8500, loss[loss=0.168, simple_loss=0.2494, pruned_loss=0.0433, over 11788.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2872, pruned_loss=0.05686, over 3023180.34 frames. ], batch size: 246, lr: 6.73e-03, grad_scale: 8.0 2023-04-29 07:26:21,523 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99885.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:26:50,575 INFO [train.py:904] (6/8) Epoch 10, batch 8550, loss[loss=0.2118, simple_loss=0.2998, pruned_loss=0.06193, over 15404.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2847, pruned_loss=0.05526, over 3034399.23 frames. ], batch size: 190, lr: 6.73e-03, grad_scale: 8.0 2023-04-29 07:27:34,875 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99924.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:27:34,969 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99924.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:27:50,547 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99933.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:28:02,240 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.540e+02 3.329e+02 3.854e+02 7.309e+02, threshold=6.657e+02, percent-clipped=4.0 2023-04-29 07:28:16,656 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99945.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:28:29,552 INFO [train.py:904] (6/8) Epoch 10, batch 8600, loss[loss=0.1864, simple_loss=0.28, pruned_loss=0.04644, over 16917.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2852, pruned_loss=0.05386, over 3062801.25 frames. ], batch size: 96, lr: 6.73e-03, grad_scale: 8.0 2023-04-29 07:29:10,993 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99972.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:29:26,477 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99980.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:30:11,015 INFO [train.py:904] (6/8) Epoch 10, batch 8650, loss[loss=0.1683, simple_loss=0.2723, pruned_loss=0.03218, over 16645.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2831, pruned_loss=0.05231, over 3052294.80 frames. ], batch size: 89, lr: 6.73e-03, grad_scale: 8.0 2023-04-29 07:30:24,826 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100007.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 07:31:10,591 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100028.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:31:21,328 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-29 07:31:27,926 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100036.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:31:31,866 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.482e+02 2.900e+02 4.130e+02 1.018e+03, threshold=5.800e+02, percent-clipped=4.0 2023-04-29 07:31:54,114 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-29 07:31:56,446 INFO [train.py:904] (6/8) Epoch 10, batch 8700, loss[loss=0.1752, simple_loss=0.2589, pruned_loss=0.04576, over 12411.00 frames. ], tot_loss[loss=0.191, simple_loss=0.2799, pruned_loss=0.05103, over 3036128.65 frames. ], batch size: 246, lr: 6.73e-03, grad_scale: 8.0 2023-04-29 07:32:28,555 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100068.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 07:32:39,668 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 07:32:43,293 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100076.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:32:52,580 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9377, 2.0046, 2.2092, 3.2259, 2.0902, 2.2230, 2.1794, 2.0582], device='cuda:6'), covar=tensor([0.0850, 0.3265, 0.1998, 0.0495, 0.3793, 0.2252, 0.2856, 0.3349], device='cuda:6'), in_proj_covar=tensor([0.0338, 0.0372, 0.0309, 0.0304, 0.0395, 0.0414, 0.0332, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:33:35,355 INFO [train.py:904] (6/8) Epoch 10, batch 8750, loss[loss=0.1937, simple_loss=0.2885, pruned_loss=0.0495, over 15386.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2797, pruned_loss=0.05061, over 3029135.28 frames. ], batch size: 191, lr: 6.72e-03, grad_scale: 8.0 2023-04-29 07:34:32,195 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100124.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:34:32,326 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100124.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 07:35:02,703 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.530e+02 3.072e+02 4.161e+02 6.852e+02, threshold=6.144e+02, percent-clipped=8.0 2023-04-29 07:35:29,749 INFO [train.py:904] (6/8) Epoch 10, batch 8800, loss[loss=0.1721, simple_loss=0.2662, pruned_loss=0.03898, over 16678.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.2778, pruned_loss=0.04942, over 3033622.63 frames. ], batch size: 62, lr: 6.72e-03, grad_scale: 8.0 2023-04-29 07:36:02,079 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3075, 3.4447, 3.7244, 1.7383, 3.9060, 3.9958, 2.8959, 2.8594], device='cuda:6'), covar=tensor([0.0810, 0.0186, 0.0183, 0.1191, 0.0062, 0.0077, 0.0396, 0.0419], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0096, 0.0081, 0.0133, 0.0065, 0.0092, 0.0115, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-29 07:36:12,282 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100172.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 07:37:15,875 INFO [train.py:904] (6/8) Epoch 10, batch 8850, loss[loss=0.1836, simple_loss=0.2654, pruned_loss=0.05085, over 12393.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.2798, pruned_loss=0.04847, over 3039172.07 frames. ], batch size: 250, lr: 6.72e-03, grad_scale: 8.0 2023-04-29 07:38:04,281 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100224.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:38:34,830 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100238.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:38:36,934 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 2.435e+02 3.067e+02 3.598e+02 6.365e+02, threshold=6.134e+02, percent-clipped=1.0 2023-04-29 07:38:48,322 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100245.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:39:01,668 INFO [train.py:904] (6/8) Epoch 10, batch 8900, loss[loss=0.2059, simple_loss=0.2944, pruned_loss=0.05875, over 16155.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2806, pruned_loss=0.04791, over 3054330.14 frames. ], batch size: 165, lr: 6.72e-03, grad_scale: 4.0 2023-04-29 07:39:43,458 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100272.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:40:45,781 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100293.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:40:54,149 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.01 vs. limit=5.0 2023-04-29 07:41:01,116 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100299.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:41:05,909 INFO [train.py:904] (6/8) Epoch 10, batch 8950, loss[loss=0.1666, simple_loss=0.2641, pruned_loss=0.03456, over 16698.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2799, pruned_loss=0.04824, over 3030245.29 frames. ], batch size: 83, lr: 6.72e-03, grad_scale: 4.0 2023-04-29 07:41:22,816 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2447, 5.9541, 6.1708, 5.8609, 5.9277, 6.3775, 6.0240, 5.7514], device='cuda:6'), covar=tensor([0.0694, 0.1543, 0.1307, 0.1811, 0.2256, 0.0895, 0.1198, 0.1953], device='cuda:6'), in_proj_covar=tensor([0.0316, 0.0446, 0.0476, 0.0387, 0.0509, 0.0508, 0.0391, 0.0521], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:41:31,185 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9460, 5.2380, 5.0280, 5.0631, 4.7283, 4.6816, 4.6710, 5.3276], device='cuda:6'), covar=tensor([0.0958, 0.0846, 0.1027, 0.0627, 0.0773, 0.0816, 0.0970, 0.0785], device='cuda:6'), in_proj_covar=tensor([0.0486, 0.0614, 0.0506, 0.0425, 0.0384, 0.0407, 0.0515, 0.0469], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:42:10,222 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6704, 4.2479, 4.2326, 2.8961, 3.7037, 4.1817, 3.9239, 2.3363], device='cuda:6'), covar=tensor([0.0367, 0.0019, 0.0023, 0.0274, 0.0053, 0.0058, 0.0046, 0.0349], device='cuda:6'), in_proj_covar=tensor([0.0123, 0.0062, 0.0066, 0.0121, 0.0075, 0.0083, 0.0073, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 07:42:22,238 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100336.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:42:29,137 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.659e+02 2.569e+02 2.987e+02 3.833e+02 6.749e+02, threshold=5.974e+02, percent-clipped=2.0 2023-04-29 07:42:55,809 INFO [train.py:904] (6/8) Epoch 10, batch 9000, loss[loss=0.1699, simple_loss=0.2605, pruned_loss=0.03967, over 16300.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2766, pruned_loss=0.04673, over 3040839.72 frames. ], batch size: 146, lr: 6.71e-03, grad_scale: 4.0 2023-04-29 07:42:55,809 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 07:43:05,469 INFO [train.py:938] (6/8) Epoch 10, validation: loss=0.1565, simple_loss=0.2604, pruned_loss=0.02634, over 944034.00 frames. 2023-04-29 07:43:05,470 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 07:43:30,275 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100363.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 07:44:12,990 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100384.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:44:50,482 INFO [train.py:904] (6/8) Epoch 10, batch 9050, loss[loss=0.19, simple_loss=0.2768, pruned_loss=0.05161, over 12812.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2777, pruned_loss=0.04743, over 3042872.72 frames. ], batch size: 246, lr: 6.71e-03, grad_scale: 4.0 2023-04-29 07:44:52,450 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100402.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:45:08,597 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 07:46:08,024 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0007, 1.8123, 1.5841, 1.4890, 1.9391, 1.6050, 1.7779, 2.0246], device='cuda:6'), covar=tensor([0.0093, 0.0223, 0.0299, 0.0269, 0.0158, 0.0183, 0.0125, 0.0149], device='cuda:6'), in_proj_covar=tensor([0.0129, 0.0192, 0.0191, 0.0188, 0.0190, 0.0193, 0.0188, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:46:08,563 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.813e+02 2.533e+02 2.923e+02 3.957e+02 1.423e+03, threshold=5.846e+02, percent-clipped=5.0 2023-04-29 07:46:25,438 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0263, 3.2176, 2.9178, 4.8686, 3.7245, 4.5580, 1.6112, 3.5143], device='cuda:6'), covar=tensor([0.1181, 0.0547, 0.0943, 0.0103, 0.0149, 0.0278, 0.1377, 0.0561], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0156, 0.0179, 0.0133, 0.0189, 0.0204, 0.0179, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 07:46:37,387 INFO [train.py:904] (6/8) Epoch 10, batch 9100, loss[loss=0.1967, simple_loss=0.294, pruned_loss=0.0497, over 16178.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2773, pruned_loss=0.04758, over 3048207.25 frames. ], batch size: 165, lr: 6.71e-03, grad_scale: 4.0 2023-04-29 07:46:58,874 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100463.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 07:47:19,445 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6877, 3.6976, 4.0549, 1.8750, 4.2120, 4.3059, 3.1556, 3.0830], device='cuda:6'), covar=tensor([0.0640, 0.0173, 0.0133, 0.1116, 0.0047, 0.0068, 0.0317, 0.0381], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0096, 0.0081, 0.0134, 0.0065, 0.0093, 0.0115, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-29 07:48:34,506 INFO [train.py:904] (6/8) Epoch 10, batch 9150, loss[loss=0.1815, simple_loss=0.2678, pruned_loss=0.04762, over 16583.00 frames. ], tot_loss[loss=0.187, simple_loss=0.2787, pruned_loss=0.04758, over 3064273.74 frames. ], batch size: 148, lr: 6.71e-03, grad_scale: 4.0 2023-04-29 07:49:54,593 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 2.281e+02 2.663e+02 3.521e+02 5.779e+02, threshold=5.326e+02, percent-clipped=0.0 2023-04-29 07:50:13,980 INFO [train.py:904] (6/8) Epoch 10, batch 9200, loss[loss=0.183, simple_loss=0.2738, pruned_loss=0.04615, over 16832.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2743, pruned_loss=0.04637, over 3082456.58 frames. ], batch size: 102, lr: 6.71e-03, grad_scale: 8.0 2023-04-29 07:50:15,538 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7758, 4.9662, 5.1284, 5.0887, 4.9978, 5.5731, 5.0548, 4.8205], device='cuda:6'), covar=tensor([0.0854, 0.1507, 0.1531, 0.1724, 0.2438, 0.0846, 0.1343, 0.2238], device='cuda:6'), in_proj_covar=tensor([0.0319, 0.0445, 0.0478, 0.0389, 0.0509, 0.0508, 0.0391, 0.0527], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 07:51:35,119 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100594.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:51:50,957 INFO [train.py:904] (6/8) Epoch 10, batch 9250, loss[loss=0.1902, simple_loss=0.2753, pruned_loss=0.05252, over 12250.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.274, pruned_loss=0.04641, over 3081906.70 frames. ], batch size: 248, lr: 6.71e-03, grad_scale: 8.0 2023-04-29 07:53:14,008 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.494e+02 2.424e+02 2.858e+02 3.550e+02 5.723e+02, threshold=5.716e+02, percent-clipped=3.0 2023-04-29 07:53:42,535 INFO [train.py:904] (6/8) Epoch 10, batch 9300, loss[loss=0.1791, simple_loss=0.27, pruned_loss=0.04413, over 16406.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.272, pruned_loss=0.04587, over 3061723.16 frames. ], batch size: 146, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 07:54:09,402 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100663.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 07:55:29,730 INFO [train.py:904] (6/8) Epoch 10, batch 9350, loss[loss=0.1785, simple_loss=0.2649, pruned_loss=0.04604, over 15466.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.272, pruned_loss=0.04588, over 3065566.25 frames. ], batch size: 191, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 07:55:50,307 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100711.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 07:55:53,066 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.29 vs. limit=5.0 2023-04-29 07:56:29,088 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100730.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 07:56:48,376 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.720e+02 2.395e+02 2.806e+02 3.199e+02 7.223e+02, threshold=5.612e+02, percent-clipped=2.0 2023-04-29 07:56:51,837 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1531, 4.2193, 4.0360, 3.7530, 3.6810, 4.1044, 3.8307, 3.8681], device='cuda:6'), covar=tensor([0.0567, 0.0469, 0.0277, 0.0300, 0.0825, 0.0512, 0.0663, 0.0593], device='cuda:6'), in_proj_covar=tensor([0.0222, 0.0276, 0.0260, 0.0240, 0.0280, 0.0277, 0.0182, 0.0303], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 07:57:00,261 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-29 07:57:12,447 INFO [train.py:904] (6/8) Epoch 10, batch 9400, loss[loss=0.1713, simple_loss=0.2516, pruned_loss=0.04554, over 12458.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2721, pruned_loss=0.04573, over 3055206.94 frames. ], batch size: 248, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 07:57:25,480 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100758.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 07:58:33,077 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100791.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 07:58:52,903 INFO [train.py:904] (6/8) Epoch 10, batch 9450, loss[loss=0.1673, simple_loss=0.2547, pruned_loss=0.03996, over 12059.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2735, pruned_loss=0.04605, over 3049513.59 frames. ], batch size: 248, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 07:59:23,822 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5613, 2.5774, 2.3368, 3.7518, 2.4301, 3.8444, 1.3164, 2.8711], device='cuda:6'), covar=tensor([0.1712, 0.0719, 0.1264, 0.0167, 0.0154, 0.0347, 0.1927, 0.0760], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0153, 0.0176, 0.0131, 0.0183, 0.0200, 0.0177, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 07:59:51,636 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.56 vs. limit=2.0 2023-04-29 08:00:10,871 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.411e+02 3.092e+02 3.645e+02 7.379e+02, threshold=6.183e+02, percent-clipped=4.0 2023-04-29 08:00:34,622 INFO [train.py:904] (6/8) Epoch 10, batch 9500, loss[loss=0.1824, simple_loss=0.271, pruned_loss=0.04689, over 16626.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2726, pruned_loss=0.04551, over 3049403.59 frames. ], batch size: 134, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 08:01:30,548 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8117, 1.3075, 1.6164, 1.6634, 1.7543, 1.8447, 1.5527, 1.7741], device='cuda:6'), covar=tensor([0.0176, 0.0262, 0.0145, 0.0182, 0.0178, 0.0141, 0.0258, 0.0085], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0165, 0.0148, 0.0150, 0.0162, 0.0114, 0.0166, 0.0105], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 08:02:00,977 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9899, 4.2323, 4.0061, 4.0783, 3.7124, 3.8727, 3.8268, 4.2091], device='cuda:6'), covar=tensor([0.0853, 0.0942, 0.0988, 0.0653, 0.0728, 0.1360, 0.0906, 0.0938], device='cuda:6'), in_proj_covar=tensor([0.0488, 0.0612, 0.0504, 0.0426, 0.0387, 0.0405, 0.0514, 0.0470], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:02:03,490 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100894.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:02:20,731 INFO [train.py:904] (6/8) Epoch 10, batch 9550, loss[loss=0.1816, simple_loss=0.2774, pruned_loss=0.04292, over 15287.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2726, pruned_loss=0.0457, over 3052035.62 frames. ], batch size: 192, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 08:03:16,171 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0090, 3.9247, 4.2280, 2.0713, 4.4461, 4.5175, 3.2701, 3.5731], device='cuda:6'), covar=tensor([0.0589, 0.0199, 0.0199, 0.1147, 0.0044, 0.0087, 0.0337, 0.0320], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0095, 0.0081, 0.0134, 0.0065, 0.0093, 0.0115, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:6') 2023-04-29 08:03:40,250 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.820e+02 2.447e+02 2.728e+02 3.779e+02 7.191e+02, threshold=5.455e+02, percent-clipped=3.0 2023-04-29 08:03:45,444 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100942.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:04:00,789 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-29 08:04:02,634 INFO [train.py:904] (6/8) Epoch 10, batch 9600, loss[loss=0.1739, simple_loss=0.2604, pruned_loss=0.0437, over 12230.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2743, pruned_loss=0.04644, over 3060575.40 frames. ], batch size: 248, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 08:04:42,372 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7967, 3.6954, 3.8658, 3.9657, 4.0578, 3.6252, 4.0397, 4.0649], device='cuda:6'), covar=tensor([0.1431, 0.0972, 0.1263, 0.0653, 0.0552, 0.1504, 0.0588, 0.0543], device='cuda:6'), in_proj_covar=tensor([0.0470, 0.0591, 0.0711, 0.0608, 0.0459, 0.0461, 0.0476, 0.0536], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:05:48,870 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0848, 3.5587, 3.5815, 2.3441, 3.3019, 3.6001, 3.4431, 1.9402], device='cuda:6'), covar=tensor([0.0442, 0.0029, 0.0032, 0.0323, 0.0069, 0.0057, 0.0064, 0.0403], device='cuda:6'), in_proj_covar=tensor([0.0123, 0.0063, 0.0066, 0.0121, 0.0074, 0.0083, 0.0072, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 08:05:52,878 INFO [train.py:904] (6/8) Epoch 10, batch 9650, loss[loss=0.1816, simple_loss=0.2759, pruned_loss=0.04363, over 16226.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.2762, pruned_loss=0.0463, over 3060842.86 frames. ], batch size: 165, lr: 6.69e-03, grad_scale: 8.0 2023-04-29 08:07:15,510 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.746e+02 2.568e+02 3.331e+02 4.260e+02 1.013e+03, threshold=6.663e+02, percent-clipped=7.0 2023-04-29 08:07:41,466 INFO [train.py:904] (6/8) Epoch 10, batch 9700, loss[loss=0.1891, simple_loss=0.2801, pruned_loss=0.04909, over 16762.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2749, pruned_loss=0.04594, over 3057097.25 frames. ], batch size: 124, lr: 6.69e-03, grad_scale: 8.0 2023-04-29 08:07:52,539 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101058.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:08:22,266 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.34 vs. limit=5.0 2023-04-29 08:08:40,749 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3459, 1.9283, 2.1455, 3.8218, 1.8851, 2.1332, 2.0292, 2.0172], device='cuda:6'), covar=tensor([0.0904, 0.3656, 0.2242, 0.0443, 0.4528, 0.2608, 0.3346, 0.3801], device='cuda:6'), in_proj_covar=tensor([0.0341, 0.0368, 0.0312, 0.0304, 0.0398, 0.0411, 0.0333, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:08:54,360 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=101086.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 08:09:25,195 INFO [train.py:904] (6/8) Epoch 10, batch 9750, loss[loss=0.1969, simple_loss=0.287, pruned_loss=0.0534, over 16857.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2739, pruned_loss=0.04644, over 3050862.98 frames. ], batch size: 124, lr: 6.69e-03, grad_scale: 8.0 2023-04-29 08:09:32,958 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=101106.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:09:35,569 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101107.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:10:09,708 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7536, 2.3992, 2.3070, 3.4576, 2.2188, 3.6957, 1.3899, 2.8653], device='cuda:6'), covar=tensor([0.1302, 0.0705, 0.1122, 0.0105, 0.0114, 0.0350, 0.1537, 0.0691], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0153, 0.0176, 0.0130, 0.0181, 0.0199, 0.0178, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 08:10:12,926 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0819, 2.7941, 2.6599, 1.8701, 2.4641, 2.0592, 2.5462, 2.8544], device='cuda:6'), covar=tensor([0.0394, 0.0671, 0.0578, 0.1758, 0.0829, 0.0979, 0.0921, 0.0757], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0132, 0.0154, 0.0141, 0.0133, 0.0123, 0.0133, 0.0142], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 08:10:14,918 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-29 08:10:45,034 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.458e+02 3.043e+02 3.868e+02 6.520e+02, threshold=6.087e+02, percent-clipped=0.0 2023-04-29 08:11:05,041 INFO [train.py:904] (6/8) Epoch 10, batch 9800, loss[loss=0.19, simple_loss=0.292, pruned_loss=0.04402, over 16697.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2745, pruned_loss=0.04542, over 3075250.71 frames. ], batch size: 134, lr: 6.69e-03, grad_scale: 4.0 2023-04-29 08:11:37,106 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101168.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:12:46,555 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-29 08:12:49,186 INFO [train.py:904] (6/8) Epoch 10, batch 9850, loss[loss=0.1923, simple_loss=0.295, pruned_loss=0.04476, over 16265.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2757, pruned_loss=0.04509, over 3093618.20 frames. ], batch size: 165, lr: 6.69e-03, grad_scale: 4.0 2023-04-29 08:12:57,878 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8723, 1.7069, 2.2052, 2.8476, 2.6198, 3.0152, 2.0441, 2.9607], device='cuda:6'), covar=tensor([0.0116, 0.0350, 0.0242, 0.0164, 0.0201, 0.0138, 0.0316, 0.0088], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0166, 0.0147, 0.0150, 0.0161, 0.0114, 0.0166, 0.0105], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 08:14:17,926 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.688e+02 2.461e+02 2.965e+02 3.704e+02 9.099e+02, threshold=5.931e+02, percent-clipped=1.0 2023-04-29 08:14:25,235 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8877, 3.0241, 2.6525, 4.7558, 3.6119, 4.3730, 1.5493, 3.4458], device='cuda:6'), covar=tensor([0.1218, 0.0599, 0.1046, 0.0091, 0.0146, 0.0244, 0.1435, 0.0523], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0151, 0.0174, 0.0128, 0.0179, 0.0197, 0.0175, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 08:14:41,617 INFO [train.py:904] (6/8) Epoch 10, batch 9900, loss[loss=0.1882, simple_loss=0.2863, pruned_loss=0.04503, over 16365.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2763, pruned_loss=0.04527, over 3080690.28 frames. ], batch size: 146, lr: 6.69e-03, grad_scale: 4.0 2023-04-29 08:15:23,841 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7165, 4.0055, 3.2686, 2.2481, 2.8175, 2.6235, 4.3427, 3.6369], device='cuda:6'), covar=tensor([0.2418, 0.0505, 0.1182, 0.2112, 0.2062, 0.1525, 0.0331, 0.0878], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0245, 0.0273, 0.0265, 0.0255, 0.0212, 0.0254, 0.0275], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:16:40,042 INFO [train.py:904] (6/8) Epoch 10, batch 9950, loss[loss=0.1677, simple_loss=0.2653, pruned_loss=0.03503, over 16589.00 frames. ], tot_loss[loss=0.1848, simple_loss=0.2783, pruned_loss=0.04564, over 3090246.03 frames. ], batch size: 62, lr: 6.68e-03, grad_scale: 4.0 2023-04-29 08:16:57,566 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 08:18:10,185 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5900, 4.3834, 4.6071, 4.7707, 4.9410, 4.4483, 4.9329, 4.9053], device='cuda:6'), covar=tensor([0.1363, 0.0954, 0.1288, 0.0573, 0.0407, 0.0716, 0.0373, 0.0514], device='cuda:6'), in_proj_covar=tensor([0.0467, 0.0586, 0.0703, 0.0600, 0.0456, 0.0453, 0.0470, 0.0535], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:18:13,665 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.784e+02 2.620e+02 2.929e+02 3.358e+02 7.407e+02, threshold=5.858e+02, percent-clipped=2.0 2023-04-29 08:18:17,749 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2598, 3.2466, 1.7419, 3.5002, 2.3820, 3.4919, 2.0116, 2.6724], device='cuda:6'), covar=tensor([0.0217, 0.0287, 0.1630, 0.0173, 0.0831, 0.0496, 0.1467, 0.0659], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0154, 0.0182, 0.0116, 0.0161, 0.0192, 0.0191, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-29 08:18:26,790 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-29 08:18:41,623 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4600, 3.2547, 2.7300, 2.0845, 2.2085, 2.1623, 3.4540, 3.0885], device='cuda:6'), covar=tensor([0.2414, 0.0575, 0.1331, 0.2244, 0.2041, 0.1676, 0.0412, 0.0907], device='cuda:6'), in_proj_covar=tensor([0.0289, 0.0243, 0.0271, 0.0263, 0.0252, 0.0209, 0.0253, 0.0272], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:18:42,216 INFO [train.py:904] (6/8) Epoch 10, batch 10000, loss[loss=0.1763, simple_loss=0.2801, pruned_loss=0.03627, over 15566.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2767, pruned_loss=0.04513, over 3104769.64 frames. ], batch size: 194, lr: 6.68e-03, grad_scale: 8.0 2023-04-29 08:19:54,015 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101386.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 08:20:01,096 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 08:20:18,159 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6747, 3.5346, 3.5357, 3.8429, 3.8718, 3.5730, 3.9317, 3.9385], device='cuda:6'), covar=tensor([0.1355, 0.1153, 0.1985, 0.0917, 0.0980, 0.1767, 0.0853, 0.0871], device='cuda:6'), in_proj_covar=tensor([0.0468, 0.0587, 0.0706, 0.0601, 0.0457, 0.0454, 0.0470, 0.0533], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:20:23,593 INFO [train.py:904] (6/8) Epoch 10, batch 10050, loss[loss=0.1742, simple_loss=0.2712, pruned_loss=0.03862, over 16739.00 frames. ], tot_loss[loss=0.1831, simple_loss=0.2764, pruned_loss=0.04487, over 3112554.82 frames. ], batch size: 83, lr: 6.68e-03, grad_scale: 8.0 2023-04-29 08:21:04,543 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0023, 4.0463, 3.8827, 3.6735, 3.6222, 3.9791, 3.7069, 3.7637], device='cuda:6'), covar=tensor([0.0591, 0.0650, 0.0261, 0.0255, 0.0677, 0.0600, 0.0819, 0.0636], device='cuda:6'), in_proj_covar=tensor([0.0220, 0.0273, 0.0258, 0.0237, 0.0275, 0.0270, 0.0178, 0.0299], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-04-29 08:21:25,329 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=101434.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 08:21:36,392 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.448e+02 2.853e+02 3.607e+02 8.482e+02, threshold=5.706e+02, percent-clipped=4.0 2023-04-29 08:21:56,132 INFO [train.py:904] (6/8) Epoch 10, batch 10100, loss[loss=0.1702, simple_loss=0.2512, pruned_loss=0.0446, over 12819.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2767, pruned_loss=0.04535, over 3101777.31 frames. ], batch size: 250, lr: 6.68e-03, grad_scale: 8.0 2023-04-29 08:22:16,003 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=101463.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:23:38,425 INFO [train.py:904] (6/8) Epoch 11, batch 0, loss[loss=0.3082, simple_loss=0.3413, pruned_loss=0.1376, over 16877.00 frames. ], tot_loss[loss=0.3082, simple_loss=0.3413, pruned_loss=0.1376, over 16877.00 frames. ], batch size: 116, lr: 6.37e-03, grad_scale: 8.0 2023-04-29 08:23:38,425 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 08:23:45,830 INFO [train.py:938] (6/8) Epoch 11, validation: loss=0.1557, simple_loss=0.2595, pruned_loss=0.02591, over 944034.00 frames. 2023-04-29 08:23:45,831 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 08:24:32,832 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6255, 4.5239, 5.0004, 5.0055, 5.0375, 4.6896, 4.6889, 4.3653], device='cuda:6'), covar=tensor([0.0271, 0.0705, 0.0441, 0.0438, 0.0441, 0.0371, 0.0877, 0.0485], device='cuda:6'), in_proj_covar=tensor([0.0293, 0.0302, 0.0306, 0.0290, 0.0347, 0.0322, 0.0408, 0.0263], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-29 08:24:43,128 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.573e+02 2.991e+02 3.932e+02 8.479e+02, threshold=5.981e+02, percent-clipped=4.0 2023-04-29 08:24:43,679 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7985, 3.7159, 2.9653, 2.3693, 2.5837, 2.3690, 3.8256, 3.4646], device='cuda:6'), covar=tensor([0.2223, 0.0609, 0.1366, 0.2220, 0.2074, 0.1587, 0.0479, 0.1140], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0244, 0.0274, 0.0264, 0.0254, 0.0211, 0.0255, 0.0276], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:24:54,773 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.20 vs. limit=5.0 2023-04-29 08:24:55,132 INFO [train.py:904] (6/8) Epoch 11, batch 50, loss[loss=0.2009, simple_loss=0.2709, pruned_loss=0.06543, over 16748.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.29, pruned_loss=0.06726, over 753803.13 frames. ], batch size: 124, lr: 6.37e-03, grad_scale: 1.0 2023-04-29 08:25:29,032 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8880, 2.6995, 2.5837, 1.9111, 2.5484, 2.6739, 2.6032, 1.8024], device='cuda:6'), covar=tensor([0.0338, 0.0060, 0.0049, 0.0299, 0.0098, 0.0085, 0.0076, 0.0339], device='cuda:6'), in_proj_covar=tensor([0.0124, 0.0065, 0.0067, 0.0123, 0.0076, 0.0084, 0.0074, 0.0117], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 08:26:05,600 INFO [train.py:904] (6/8) Epoch 11, batch 100, loss[loss=0.207, simple_loss=0.2866, pruned_loss=0.06369, over 16475.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2828, pruned_loss=0.062, over 1324217.34 frames. ], batch size: 146, lr: 6.37e-03, grad_scale: 1.0 2023-04-29 08:26:15,584 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.64 vs. limit=5.0 2023-04-29 08:27:03,337 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.860e+02 2.523e+02 3.127e+02 3.796e+02 9.987e+02, threshold=6.254e+02, percent-clipped=2.0 2023-04-29 08:27:12,713 INFO [train.py:904] (6/8) Epoch 11, batch 150, loss[loss=0.2359, simple_loss=0.2942, pruned_loss=0.08875, over 16905.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2781, pruned_loss=0.05915, over 1768326.26 frames. ], batch size: 109, lr: 6.37e-03, grad_scale: 1.0 2023-04-29 08:27:59,431 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101684.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:28:18,087 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9063, 5.6118, 5.7480, 5.5541, 5.5661, 6.1508, 5.7430, 5.5135], device='cuda:6'), covar=tensor([0.0869, 0.1638, 0.1787, 0.1820, 0.2800, 0.0844, 0.1225, 0.2155], device='cuda:6'), in_proj_covar=tensor([0.0331, 0.0469, 0.0504, 0.0404, 0.0538, 0.0534, 0.0402, 0.0547], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 08:28:23,285 INFO [train.py:904] (6/8) Epoch 11, batch 200, loss[loss=0.1965, simple_loss=0.2707, pruned_loss=0.06116, over 16841.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2796, pruned_loss=0.05974, over 2099557.09 frames. ], batch size: 96, lr: 6.37e-03, grad_scale: 1.0 2023-04-29 08:28:37,299 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 08:29:21,756 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.429e+02 2.923e+02 3.406e+02 5.474e+02, threshold=5.846e+02, percent-clipped=1.0 2023-04-29 08:29:23,289 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101745.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:29:31,801 INFO [train.py:904] (6/8) Epoch 11, batch 250, loss[loss=0.2579, simple_loss=0.3231, pruned_loss=0.09632, over 12333.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2789, pruned_loss=0.05953, over 2358660.43 frames. ], batch size: 248, lr: 6.37e-03, grad_scale: 1.0 2023-04-29 08:29:35,665 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3395, 3.4812, 2.0305, 3.5710, 2.5862, 3.6279, 2.1217, 2.7890], device='cuda:6'), covar=tensor([0.0261, 0.0333, 0.1527, 0.0281, 0.0759, 0.0614, 0.1371, 0.0625], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0161, 0.0188, 0.0125, 0.0167, 0.0202, 0.0197, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 08:29:46,227 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101763.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:30:37,963 INFO [train.py:904] (6/8) Epoch 11, batch 300, loss[loss=0.1689, simple_loss=0.2535, pruned_loss=0.0422, over 16814.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2755, pruned_loss=0.05797, over 2574864.41 frames. ], batch size: 39, lr: 6.37e-03, grad_scale: 1.0 2023-04-29 08:30:51,063 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=101811.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:30:58,130 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8211, 4.5043, 4.8458, 5.0612, 5.2138, 4.5809, 5.2091, 5.2010], device='cuda:6'), covar=tensor([0.1633, 0.1263, 0.1785, 0.0749, 0.0547, 0.0847, 0.0567, 0.0589], device='cuda:6'), in_proj_covar=tensor([0.0506, 0.0633, 0.0766, 0.0649, 0.0490, 0.0489, 0.0508, 0.0571], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:31:35,542 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.476e+02 3.003e+02 3.668e+02 8.355e+02, threshold=6.006e+02, percent-clipped=2.0 2023-04-29 08:31:47,602 INFO [train.py:904] (6/8) Epoch 11, batch 350, loss[loss=0.2162, simple_loss=0.2972, pruned_loss=0.06762, over 12142.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.273, pruned_loss=0.05614, over 2735224.41 frames. ], batch size: 246, lr: 6.36e-03, grad_scale: 1.0 2023-04-29 08:32:31,451 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0873, 4.7559, 5.0887, 5.3515, 5.5229, 4.8852, 5.4758, 5.4759], device='cuda:6'), covar=tensor([0.1506, 0.1233, 0.1616, 0.0659, 0.0481, 0.0637, 0.0456, 0.0517], device='cuda:6'), in_proj_covar=tensor([0.0511, 0.0640, 0.0778, 0.0656, 0.0497, 0.0493, 0.0513, 0.0577], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:32:56,331 INFO [train.py:904] (6/8) Epoch 11, batch 400, loss[loss=0.1798, simple_loss=0.2718, pruned_loss=0.04391, over 17106.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2709, pruned_loss=0.05496, over 2865921.74 frames. ], batch size: 49, lr: 6.36e-03, grad_scale: 2.0 2023-04-29 08:33:22,128 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101920.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 08:33:54,606 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.298e+02 2.716e+02 3.213e+02 6.833e+02, threshold=5.433e+02, percent-clipped=1.0 2023-04-29 08:34:06,159 INFO [train.py:904] (6/8) Epoch 11, batch 450, loss[loss=0.2039, simple_loss=0.2706, pruned_loss=0.06854, over 16757.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.27, pruned_loss=0.05417, over 2963577.06 frames. ], batch size: 124, lr: 6.36e-03, grad_scale: 2.0 2023-04-29 08:34:43,731 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5699, 2.5838, 1.7632, 2.7130, 2.1002, 2.7816, 2.0387, 2.2905], device='cuda:6'), covar=tensor([0.0245, 0.0339, 0.1324, 0.0224, 0.0689, 0.0445, 0.1217, 0.0582], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0163, 0.0188, 0.0127, 0.0168, 0.0204, 0.0197, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 08:34:47,034 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101981.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 08:34:54,576 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101987.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:35:19,214 INFO [train.py:904] (6/8) Epoch 11, batch 500, loss[loss=0.1702, simple_loss=0.2642, pruned_loss=0.03811, over 17278.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2679, pruned_loss=0.05322, over 3043814.63 frames. ], batch size: 52, lr: 6.36e-03, grad_scale: 2.0 2023-04-29 08:35:32,788 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-29 08:36:13,471 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102040.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:36:19,094 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.707e+02 2.413e+02 2.842e+02 3.807e+02 1.018e+03, threshold=5.684e+02, percent-clipped=5.0 2023-04-29 08:36:24,828 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102048.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:36:29,128 INFO [train.py:904] (6/8) Epoch 11, batch 550, loss[loss=0.1781, simple_loss=0.2648, pruned_loss=0.04572, over 17191.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2665, pruned_loss=0.05238, over 3099096.11 frames. ], batch size: 46, lr: 6.36e-03, grad_scale: 2.0 2023-04-29 08:37:07,469 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0770, 3.1245, 1.7655, 3.2258, 2.3945, 3.3171, 2.0107, 2.5770], device='cuda:6'), covar=tensor([0.0257, 0.0402, 0.1551, 0.0283, 0.0789, 0.0533, 0.1351, 0.0657], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0164, 0.0188, 0.0128, 0.0168, 0.0206, 0.0198, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 08:37:13,799 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8915, 5.2963, 5.5070, 5.2739, 5.2388, 5.8655, 5.3963, 5.1851], device='cuda:6'), covar=tensor([0.0924, 0.1663, 0.1838, 0.1840, 0.2912, 0.1000, 0.1340, 0.2422], device='cuda:6'), in_proj_covar=tensor([0.0345, 0.0488, 0.0524, 0.0422, 0.0565, 0.0554, 0.0415, 0.0572], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 08:37:27,136 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7681, 4.0922, 4.2288, 3.2091, 3.6877, 4.1531, 3.9034, 2.4903], device='cuda:6'), covar=tensor([0.0345, 0.0052, 0.0037, 0.0236, 0.0078, 0.0072, 0.0059, 0.0339], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0069, 0.0070, 0.0126, 0.0078, 0.0087, 0.0077, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 08:37:40,169 INFO [train.py:904] (6/8) Epoch 11, batch 600, loss[loss=0.1884, simple_loss=0.2511, pruned_loss=0.06291, over 16829.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2658, pruned_loss=0.0528, over 3146703.67 frames. ], batch size: 96, lr: 6.36e-03, grad_scale: 2.0 2023-04-29 08:38:15,481 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1358, 2.4350, 2.6232, 4.9161, 2.5173, 2.8738, 2.6620, 2.5918], device='cuda:6'), covar=tensor([0.0722, 0.3096, 0.2099, 0.0298, 0.3498, 0.2139, 0.2705, 0.3382], device='cuda:6'), in_proj_covar=tensor([0.0358, 0.0386, 0.0326, 0.0320, 0.0410, 0.0437, 0.0349, 0.0453], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:38:19,598 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8529, 3.8308, 2.9660, 2.3336, 2.7081, 2.4367, 4.0277, 3.5714], device='cuda:6'), covar=tensor([0.2253, 0.0729, 0.1417, 0.2247, 0.2377, 0.1686, 0.0449, 0.1190], device='cuda:6'), in_proj_covar=tensor([0.0305, 0.0256, 0.0284, 0.0276, 0.0276, 0.0222, 0.0268, 0.0293], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:38:38,906 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.665e+02 3.250e+02 4.151e+02 1.142e+03, threshold=6.501e+02, percent-clipped=8.0 2023-04-29 08:38:49,721 INFO [train.py:904] (6/8) Epoch 11, batch 650, loss[loss=0.1695, simple_loss=0.2619, pruned_loss=0.03857, over 17061.00 frames. ], tot_loss[loss=0.1848, simple_loss=0.2649, pruned_loss=0.05236, over 3186685.38 frames. ], batch size: 50, lr: 6.35e-03, grad_scale: 2.0 2023-04-29 08:39:13,984 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-29 08:39:58,899 INFO [train.py:904] (6/8) Epoch 11, batch 700, loss[loss=0.1718, simple_loss=0.2622, pruned_loss=0.04069, over 17053.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2641, pruned_loss=0.05151, over 3210738.73 frames. ], batch size: 53, lr: 6.35e-03, grad_scale: 2.0 2023-04-29 08:40:16,836 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8161, 5.1934, 4.8512, 4.9839, 4.6966, 4.7474, 4.6700, 5.2485], device='cuda:6'), covar=tensor([0.1241, 0.0888, 0.1232, 0.0742, 0.0874, 0.0884, 0.1201, 0.0896], device='cuda:6'), in_proj_covar=tensor([0.0545, 0.0678, 0.0562, 0.0475, 0.0430, 0.0443, 0.0573, 0.0519], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:40:47,669 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4157, 3.4672, 1.9262, 3.6170, 2.5503, 3.6536, 2.1174, 2.7929], device='cuda:6'), covar=tensor([0.0213, 0.0340, 0.1411, 0.0261, 0.0784, 0.0557, 0.1319, 0.0599], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0164, 0.0187, 0.0129, 0.0168, 0.0206, 0.0196, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 08:40:57,196 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 2.406e+02 3.042e+02 3.688e+02 8.109e+02, threshold=6.084e+02, percent-clipped=2.0 2023-04-29 08:41:09,232 INFO [train.py:904] (6/8) Epoch 11, batch 750, loss[loss=0.1845, simple_loss=0.2739, pruned_loss=0.04761, over 17020.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2646, pruned_loss=0.05194, over 3229986.83 frames. ], batch size: 50, lr: 6.35e-03, grad_scale: 2.0 2023-04-29 08:41:42,361 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102276.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 08:42:01,209 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-29 08:42:10,651 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-29 08:42:18,049 INFO [train.py:904] (6/8) Epoch 11, batch 800, loss[loss=0.1988, simple_loss=0.2724, pruned_loss=0.06266, over 16267.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2642, pruned_loss=0.05155, over 3258975.84 frames. ], batch size: 165, lr: 6.35e-03, grad_scale: 4.0 2023-04-29 08:43:11,896 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102340.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:43:14,310 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4421, 5.2963, 5.2305, 4.7707, 4.7574, 5.2069, 5.2804, 4.8501], device='cuda:6'), covar=tensor([0.0498, 0.0381, 0.0246, 0.0269, 0.1132, 0.0376, 0.0239, 0.0740], device='cuda:6'), in_proj_covar=tensor([0.0254, 0.0315, 0.0294, 0.0275, 0.0318, 0.0313, 0.0202, 0.0345], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:43:15,310 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102343.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:43:16,112 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.524e+02 3.019e+02 3.517e+02 6.392e+02, threshold=6.037e+02, percent-clipped=1.0 2023-04-29 08:43:27,548 INFO [train.py:904] (6/8) Epoch 11, batch 850, loss[loss=0.1723, simple_loss=0.2518, pruned_loss=0.0464, over 17020.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2638, pruned_loss=0.05086, over 3274481.70 frames. ], batch size: 41, lr: 6.35e-03, grad_scale: 4.0 2023-04-29 08:44:17,574 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=102388.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:44:37,481 INFO [train.py:904] (6/8) Epoch 11, batch 900, loss[loss=0.1829, simple_loss=0.2741, pruned_loss=0.04589, over 17291.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2629, pruned_loss=0.05023, over 3290220.59 frames. ], batch size: 52, lr: 6.35e-03, grad_scale: 4.0 2023-04-29 08:44:39,211 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 08:45:35,216 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 2.314e+02 2.859e+02 3.467e+02 6.346e+02, threshold=5.718e+02, percent-clipped=1.0 2023-04-29 08:45:35,550 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4864, 5.8760, 5.2570, 5.8574, 5.3965, 4.9466, 5.4281, 5.9951], device='cuda:6'), covar=tensor([0.2172, 0.1862, 0.2873, 0.1197, 0.1557, 0.1306, 0.2205, 0.1740], device='cuda:6'), in_proj_covar=tensor([0.0547, 0.0686, 0.0569, 0.0478, 0.0432, 0.0446, 0.0576, 0.0523], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:45:45,392 INFO [train.py:904] (6/8) Epoch 11, batch 950, loss[loss=0.1519, simple_loss=0.2343, pruned_loss=0.03474, over 16624.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2632, pruned_loss=0.05044, over 3292829.49 frames. ], batch size: 76, lr: 6.34e-03, grad_scale: 4.0 2023-04-29 08:46:07,394 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.69 vs. limit=5.0 2023-04-29 08:46:54,282 INFO [train.py:904] (6/8) Epoch 11, batch 1000, loss[loss=0.178, simple_loss=0.2498, pruned_loss=0.05305, over 16696.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2618, pruned_loss=0.05028, over 3308089.08 frames. ], batch size: 83, lr: 6.34e-03, grad_scale: 4.0 2023-04-29 08:47:52,028 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.460e+02 2.841e+02 3.268e+02 6.263e+02, threshold=5.683e+02, percent-clipped=1.0 2023-04-29 08:48:02,251 INFO [train.py:904] (6/8) Epoch 11, batch 1050, loss[loss=0.1802, simple_loss=0.262, pruned_loss=0.04917, over 15928.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2619, pruned_loss=0.0507, over 3298860.07 frames. ], batch size: 35, lr: 6.34e-03, grad_scale: 4.0 2023-04-29 08:48:08,157 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-29 08:48:36,376 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102576.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 08:49:12,617 INFO [train.py:904] (6/8) Epoch 11, batch 1100, loss[loss=0.1886, simple_loss=0.2786, pruned_loss=0.0493, over 17067.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2612, pruned_loss=0.04999, over 3301608.15 frames. ], batch size: 53, lr: 6.34e-03, grad_scale: 4.0 2023-04-29 08:49:43,773 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=102624.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 08:49:52,399 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 08:50:07,637 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102643.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:50:09,057 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.199e+02 2.657e+02 3.494e+02 5.570e+02, threshold=5.313e+02, percent-clipped=0.0 2023-04-29 08:50:20,435 INFO [train.py:904] (6/8) Epoch 11, batch 1150, loss[loss=0.1677, simple_loss=0.2478, pruned_loss=0.04379, over 16482.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2616, pruned_loss=0.04954, over 3304067.07 frames. ], batch size: 75, lr: 6.34e-03, grad_scale: 4.0 2023-04-29 08:51:13,750 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-04-29 08:51:14,265 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=102691.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:51:27,908 INFO [train.py:904] (6/8) Epoch 11, batch 1200, loss[loss=0.1731, simple_loss=0.2671, pruned_loss=0.03954, over 17041.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2607, pruned_loss=0.04906, over 3297408.81 frames. ], batch size: 50, lr: 6.34e-03, grad_scale: 8.0 2023-04-29 08:51:30,790 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7263, 3.8519, 2.9480, 2.2982, 2.6157, 2.4123, 3.9270, 3.5071], device='cuda:6'), covar=tensor([0.2385, 0.0584, 0.1385, 0.2297, 0.2344, 0.1699, 0.0495, 0.1108], device='cuda:6'), in_proj_covar=tensor([0.0305, 0.0257, 0.0283, 0.0275, 0.0278, 0.0222, 0.0268, 0.0296], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:51:50,189 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0588, 4.5919, 3.5813, 2.4054, 3.0680, 2.7369, 4.9160, 3.9570], device='cuda:6'), covar=tensor([0.2368, 0.0507, 0.1289, 0.2209, 0.2451, 0.1656, 0.0285, 0.1046], device='cuda:6'), in_proj_covar=tensor([0.0304, 0.0257, 0.0282, 0.0275, 0.0278, 0.0222, 0.0268, 0.0296], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:52:27,631 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.474e+02 2.248e+02 2.693e+02 3.149e+02 5.178e+02, threshold=5.386e+02, percent-clipped=0.0 2023-04-29 08:52:39,081 INFO [train.py:904] (6/8) Epoch 11, batch 1250, loss[loss=0.1708, simple_loss=0.2613, pruned_loss=0.04015, over 17119.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2607, pruned_loss=0.04928, over 3306081.28 frames. ], batch size: 47, lr: 6.34e-03, grad_scale: 8.0 2023-04-29 08:52:41,209 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3796, 5.2637, 5.1643, 4.7864, 4.5547, 5.2291, 5.3000, 4.7776], device='cuda:6'), covar=tensor([0.0635, 0.0491, 0.0324, 0.0302, 0.1263, 0.0468, 0.0189, 0.0805], device='cuda:6'), in_proj_covar=tensor([0.0257, 0.0320, 0.0300, 0.0278, 0.0323, 0.0319, 0.0205, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 08:53:49,412 INFO [train.py:904] (6/8) Epoch 11, batch 1300, loss[loss=0.1928, simple_loss=0.2783, pruned_loss=0.05368, over 17048.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2604, pruned_loss=0.04859, over 3297994.07 frames. ], batch size: 55, lr: 6.33e-03, grad_scale: 8.0 2023-04-29 08:53:55,121 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.26 vs. limit=5.0 2023-04-29 08:54:46,485 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.885e+02 2.481e+02 2.988e+02 3.715e+02 8.832e+02, threshold=5.975e+02, percent-clipped=5.0 2023-04-29 08:54:51,811 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7706, 2.8362, 2.3834, 2.6978, 3.1583, 2.9962, 3.6546, 3.4191], device='cuda:6'), covar=tensor([0.0084, 0.0273, 0.0377, 0.0331, 0.0179, 0.0248, 0.0188, 0.0180], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0206, 0.0202, 0.0202, 0.0206, 0.0205, 0.0211, 0.0196], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 08:54:58,026 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=102851.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:54:58,763 INFO [train.py:904] (6/8) Epoch 11, batch 1350, loss[loss=0.1855, simple_loss=0.2585, pruned_loss=0.05627, over 16710.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2614, pruned_loss=0.04897, over 3314135.48 frames. ], batch size: 89, lr: 6.33e-03, grad_scale: 8.0 2023-04-29 08:55:45,174 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=102887.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:56:05,624 INFO [train.py:904] (6/8) Epoch 11, batch 1400, loss[loss=0.1845, simple_loss=0.2536, pruned_loss=0.0577, over 16428.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2611, pruned_loss=0.04936, over 3319843.93 frames. ], batch size: 146, lr: 6.33e-03, grad_scale: 8.0 2023-04-29 08:56:20,040 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102912.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:56:22,883 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.23 vs. limit=5.0 2023-04-29 08:56:39,727 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 08:57:02,665 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1994, 1.4963, 1.8720, 2.1517, 2.2276, 2.2422, 1.6765, 2.1927], device='cuda:6'), covar=tensor([0.0172, 0.0358, 0.0181, 0.0228, 0.0194, 0.0184, 0.0336, 0.0105], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0173, 0.0157, 0.0158, 0.0168, 0.0123, 0.0171, 0.0114], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 08:57:05,104 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.762e+02 2.362e+02 2.982e+02 3.517e+02 5.508e+02, threshold=5.963e+02, percent-clipped=1.0 2023-04-29 08:57:10,255 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102948.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:57:15,710 INFO [train.py:904] (6/8) Epoch 11, batch 1450, loss[loss=0.184, simple_loss=0.2516, pruned_loss=0.05822, over 16732.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2603, pruned_loss=0.04911, over 3313683.75 frames. ], batch size: 124, lr: 6.33e-03, grad_scale: 8.0 2023-04-29 08:58:25,002 INFO [train.py:904] (6/8) Epoch 11, batch 1500, loss[loss=0.1984, simple_loss=0.2807, pruned_loss=0.05802, over 17141.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2612, pruned_loss=0.04975, over 3314961.38 frames. ], batch size: 46, lr: 6.33e-03, grad_scale: 8.0 2023-04-29 08:58:59,926 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7637, 4.5825, 4.6131, 4.3947, 4.2739, 4.6530, 4.5214, 4.3978], device='cuda:6'), covar=tensor([0.0620, 0.0606, 0.0285, 0.0247, 0.0871, 0.0434, 0.0391, 0.0589], device='cuda:6'), in_proj_covar=tensor([0.0259, 0.0323, 0.0302, 0.0281, 0.0324, 0.0320, 0.0208, 0.0353], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 08:59:24,565 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.863e+02 2.416e+02 2.850e+02 3.321e+02 6.158e+02, threshold=5.699e+02, percent-clipped=1.0 2023-04-29 08:59:31,257 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1134, 3.2037, 3.5240, 2.3325, 3.1431, 3.5292, 3.1975, 2.0116], device='cuda:6'), covar=tensor([0.0388, 0.0097, 0.0037, 0.0287, 0.0079, 0.0059, 0.0076, 0.0337], device='cuda:6'), in_proj_covar=tensor([0.0125, 0.0069, 0.0070, 0.0124, 0.0077, 0.0088, 0.0077, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 08:59:35,064 INFO [train.py:904] (6/8) Epoch 11, batch 1550, loss[loss=0.224, simple_loss=0.2943, pruned_loss=0.07684, over 12541.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2627, pruned_loss=0.05101, over 3304148.47 frames. ], batch size: 246, lr: 6.33e-03, grad_scale: 8.0 2023-04-29 09:00:39,950 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103098.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:00:45,000 INFO [train.py:904] (6/8) Epoch 11, batch 1600, loss[loss=0.1901, simple_loss=0.274, pruned_loss=0.05304, over 16572.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2646, pruned_loss=0.05161, over 3317455.70 frames. ], batch size: 68, lr: 6.32e-03, grad_scale: 8.0 2023-04-29 09:00:47,906 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8983, 2.9128, 2.6640, 4.7097, 3.9115, 4.2872, 1.6505, 3.1159], device='cuda:6'), covar=tensor([0.1258, 0.0607, 0.1040, 0.0136, 0.0232, 0.0385, 0.1350, 0.0722], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0159, 0.0180, 0.0143, 0.0196, 0.0210, 0.0180, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 09:01:24,228 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-29 09:01:25,969 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103132.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:01:42,625 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.757e+02 2.417e+02 3.013e+02 3.629e+02 9.117e+02, threshold=6.027e+02, percent-clipped=4.0 2023-04-29 09:01:53,521 INFO [train.py:904] (6/8) Epoch 11, batch 1650, loss[loss=0.1894, simple_loss=0.2812, pruned_loss=0.04874, over 17048.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2657, pruned_loss=0.05208, over 3320463.95 frames. ], batch size: 53, lr: 6.32e-03, grad_scale: 8.0 2023-04-29 09:01:56,796 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5617, 4.7443, 4.9010, 4.7739, 4.7035, 5.3724, 4.9715, 4.5884], device='cuda:6'), covar=tensor([0.1347, 0.1731, 0.1965, 0.1957, 0.2767, 0.1005, 0.1433, 0.2580], device='cuda:6'), in_proj_covar=tensor([0.0350, 0.0496, 0.0538, 0.0428, 0.0568, 0.0562, 0.0422, 0.0576], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 09:01:57,025 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7850, 3.8299, 2.9570, 2.2718, 2.5857, 2.3154, 3.9430, 3.4433], device='cuda:6'), covar=tensor([0.2169, 0.0487, 0.1329, 0.2324, 0.2361, 0.1697, 0.0417, 0.1117], device='cuda:6'), in_proj_covar=tensor([0.0303, 0.0255, 0.0281, 0.0273, 0.0278, 0.0221, 0.0266, 0.0296], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:02:03,370 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103159.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:02:50,108 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103193.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:03:02,605 INFO [train.py:904] (6/8) Epoch 11, batch 1700, loss[loss=0.1889, simple_loss=0.2715, pruned_loss=0.05313, over 16028.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2679, pruned_loss=0.0526, over 3318551.38 frames. ], batch size: 35, lr: 6.32e-03, grad_scale: 8.0 2023-04-29 09:03:10,714 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103207.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:04:01,647 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103243.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:04:02,559 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.865e+02 2.478e+02 3.058e+02 3.640e+02 7.869e+02, threshold=6.116e+02, percent-clipped=2.0 2023-04-29 09:04:13,825 INFO [train.py:904] (6/8) Epoch 11, batch 1750, loss[loss=0.1691, simple_loss=0.2582, pruned_loss=0.04, over 17109.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2681, pruned_loss=0.05218, over 3322754.57 frames. ], batch size: 47, lr: 6.32e-03, grad_scale: 4.0 2023-04-29 09:05:22,483 INFO [train.py:904] (6/8) Epoch 11, batch 1800, loss[loss=0.1683, simple_loss=0.261, pruned_loss=0.03778, over 17269.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2688, pruned_loss=0.05231, over 3317599.27 frames. ], batch size: 52, lr: 6.32e-03, grad_scale: 4.0 2023-04-29 09:06:21,365 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.500e+02 2.964e+02 3.680e+02 1.127e+03, threshold=5.929e+02, percent-clipped=5.0 2023-04-29 09:06:31,956 INFO [train.py:904] (6/8) Epoch 11, batch 1850, loss[loss=0.1939, simple_loss=0.2884, pruned_loss=0.04972, over 16723.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2702, pruned_loss=0.05306, over 3317410.46 frames. ], batch size: 57, lr: 6.32e-03, grad_scale: 4.0 2023-04-29 09:06:51,479 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103366.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 09:07:39,139 INFO [train.py:904] (6/8) Epoch 11, batch 1900, loss[loss=0.1635, simple_loss=0.2483, pruned_loss=0.03931, over 17237.00 frames. ], tot_loss[loss=0.1869, simple_loss=0.2692, pruned_loss=0.05228, over 3314524.72 frames. ], batch size: 43, lr: 6.32e-03, grad_scale: 4.0 2023-04-29 09:07:46,966 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8996, 4.0935, 3.2085, 2.3040, 2.7675, 2.4793, 4.3112, 3.7432], device='cuda:6'), covar=tensor([0.2308, 0.0581, 0.1352, 0.2141, 0.2435, 0.1706, 0.0405, 0.1075], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0254, 0.0280, 0.0274, 0.0279, 0.0220, 0.0266, 0.0296], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:08:16,852 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103427.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 09:08:40,926 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.233e+02 2.712e+02 3.149e+02 7.257e+02, threshold=5.425e+02, percent-clipped=1.0 2023-04-29 09:08:51,894 INFO [train.py:904] (6/8) Epoch 11, batch 1950, loss[loss=0.2156, simple_loss=0.288, pruned_loss=0.07154, over 16507.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2692, pruned_loss=0.05191, over 3314142.91 frames. ], batch size: 146, lr: 6.31e-03, grad_scale: 4.0 2023-04-29 09:08:55,100 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103454.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:09:41,434 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103488.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:10:00,154 INFO [train.py:904] (6/8) Epoch 11, batch 2000, loss[loss=0.1778, simple_loss=0.2698, pruned_loss=0.04289, over 17117.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2688, pruned_loss=0.05131, over 3314692.19 frames. ], batch size: 48, lr: 6.31e-03, grad_scale: 8.0 2023-04-29 09:10:07,798 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103507.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:10:36,002 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0873, 4.8258, 5.1349, 5.3789, 5.5735, 4.8550, 5.5066, 5.4893], device='cuda:6'), covar=tensor([0.1641, 0.1188, 0.1691, 0.0622, 0.0479, 0.0742, 0.0401, 0.0476], device='cuda:6'), in_proj_covar=tensor([0.0548, 0.0682, 0.0840, 0.0697, 0.0526, 0.0526, 0.0541, 0.0616], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:10:42,121 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-29 09:10:58,929 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103543.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:11:00,877 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.679e+02 2.374e+02 2.796e+02 3.607e+02 7.066e+02, threshold=5.592e+02, percent-clipped=4.0 2023-04-29 09:11:11,322 INFO [train.py:904] (6/8) Epoch 11, batch 2050, loss[loss=0.1736, simple_loss=0.2638, pruned_loss=0.04173, over 17035.00 frames. ], tot_loss[loss=0.1858, simple_loss=0.2686, pruned_loss=0.05154, over 3318168.60 frames. ], batch size: 50, lr: 6.31e-03, grad_scale: 4.0 2023-04-29 09:11:16,490 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=103555.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:11:20,406 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.40 vs. limit=5.0 2023-04-29 09:11:30,050 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0362, 4.7371, 5.0351, 5.3168, 5.4765, 4.8207, 5.4402, 5.4127], device='cuda:6'), covar=tensor([0.1502, 0.1104, 0.1594, 0.0570, 0.0475, 0.0710, 0.0402, 0.0456], device='cuda:6'), in_proj_covar=tensor([0.0547, 0.0682, 0.0840, 0.0696, 0.0526, 0.0525, 0.0540, 0.0615], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:12:05,794 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=103591.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:12:14,836 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7902, 4.0585, 4.2522, 2.9483, 3.7782, 4.2389, 3.8982, 2.5688], device='cuda:6'), covar=tensor([0.0380, 0.0068, 0.0032, 0.0283, 0.0066, 0.0068, 0.0056, 0.0322], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0071, 0.0071, 0.0126, 0.0079, 0.0089, 0.0077, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 09:12:21,596 INFO [train.py:904] (6/8) Epoch 11, batch 2100, loss[loss=0.1813, simple_loss=0.2564, pruned_loss=0.05303, over 16850.00 frames. ], tot_loss[loss=0.1858, simple_loss=0.2687, pruned_loss=0.05141, over 3318749.50 frames. ], batch size: 96, lr: 6.31e-03, grad_scale: 4.0 2023-04-29 09:12:27,002 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-04-29 09:13:22,846 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.732e+02 2.751e+02 3.128e+02 3.657e+02 9.004e+02, threshold=6.256e+02, percent-clipped=1.0 2023-04-29 09:13:31,355 INFO [train.py:904] (6/8) Epoch 11, batch 2150, loss[loss=0.1492, simple_loss=0.2392, pruned_loss=0.02964, over 16826.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2696, pruned_loss=0.05229, over 3327201.60 frames. ], batch size: 42, lr: 6.31e-03, grad_scale: 4.0 2023-04-29 09:14:06,180 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6271, 6.0191, 5.7218, 5.8469, 5.3695, 5.2482, 5.4211, 6.1187], device='cuda:6'), covar=tensor([0.1238, 0.0822, 0.1018, 0.0670, 0.0807, 0.0647, 0.0995, 0.0888], device='cuda:6'), in_proj_covar=tensor([0.0546, 0.0688, 0.0570, 0.0482, 0.0430, 0.0444, 0.0577, 0.0524], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:14:42,109 INFO [train.py:904] (6/8) Epoch 11, batch 2200, loss[loss=0.1946, simple_loss=0.2721, pruned_loss=0.05855, over 16788.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2705, pruned_loss=0.05252, over 3327666.24 frames. ], batch size: 102, lr: 6.31e-03, grad_scale: 4.0 2023-04-29 09:15:11,431 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103722.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 09:15:35,711 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103740.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:15:44,092 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.506e+02 2.339e+02 2.759e+02 3.363e+02 7.654e+02, threshold=5.518e+02, percent-clipped=1.0 2023-04-29 09:15:51,099 INFO [train.py:904] (6/8) Epoch 11, batch 2250, loss[loss=0.1925, simple_loss=0.2963, pruned_loss=0.04433, over 17024.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2715, pruned_loss=0.05287, over 3325666.91 frames. ], batch size: 50, lr: 6.31e-03, grad_scale: 4.0 2023-04-29 09:15:54,427 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103754.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:16:12,461 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103765.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:16:44,274 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103788.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:17:01,654 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103801.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:17:02,366 INFO [train.py:904] (6/8) Epoch 11, batch 2300, loss[loss=0.1682, simple_loss=0.2635, pruned_loss=0.03643, over 17166.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2715, pruned_loss=0.05275, over 3325899.61 frames. ], batch size: 46, lr: 6.30e-03, grad_scale: 4.0 2023-04-29 09:17:02,656 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=103802.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:17:35,821 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103826.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:17:48,570 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=103836.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:18:02,531 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.826e+02 2.457e+02 2.837e+02 3.610e+02 7.865e+02, threshold=5.674e+02, percent-clipped=5.0 2023-04-29 09:18:11,652 INFO [train.py:904] (6/8) Epoch 11, batch 2350, loss[loss=0.1823, simple_loss=0.2537, pruned_loss=0.05548, over 16843.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2718, pruned_loss=0.05353, over 3300527.07 frames. ], batch size: 102, lr: 6.30e-03, grad_scale: 4.0 2023-04-29 09:18:45,290 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103876.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:19:19,669 INFO [train.py:904] (6/8) Epoch 11, batch 2400, loss[loss=0.2131, simple_loss=0.2873, pruned_loss=0.0695, over 16346.00 frames. ], tot_loss[loss=0.191, simple_loss=0.274, pruned_loss=0.05405, over 3292860.63 frames. ], batch size: 145, lr: 6.30e-03, grad_scale: 8.0 2023-04-29 09:20:09,843 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103937.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 09:20:20,789 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.318e+02 2.704e+02 3.589e+02 9.063e+02, threshold=5.409e+02, percent-clipped=3.0 2023-04-29 09:20:28,632 INFO [train.py:904] (6/8) Epoch 11, batch 2450, loss[loss=0.2063, simple_loss=0.274, pruned_loss=0.06931, over 16784.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2748, pruned_loss=0.05451, over 3302268.51 frames. ], batch size: 102, lr: 6.30e-03, grad_scale: 8.0 2023-04-29 09:20:33,869 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8890, 4.3402, 4.4748, 3.3322, 3.7172, 4.3560, 4.0633, 2.6886], device='cuda:6'), covar=tensor([0.0351, 0.0048, 0.0028, 0.0225, 0.0078, 0.0059, 0.0051, 0.0333], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0070, 0.0070, 0.0124, 0.0078, 0.0089, 0.0077, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 09:21:22,327 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103990.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:21:32,660 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103997.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:21:42,813 INFO [train.py:904] (6/8) Epoch 11, batch 2500, loss[loss=0.177, simple_loss=0.2615, pruned_loss=0.04624, over 16245.00 frames. ], tot_loss[loss=0.1915, simple_loss=0.2747, pruned_loss=0.05417, over 3301290.93 frames. ], batch size: 36, lr: 6.30e-03, grad_scale: 8.0 2023-04-29 09:21:49,045 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6813, 6.0010, 5.7735, 5.7954, 5.4342, 5.2386, 5.4006, 6.1393], device='cuda:6'), covar=tensor([0.1061, 0.0840, 0.0785, 0.0722, 0.0778, 0.0616, 0.0969, 0.0735], device='cuda:6'), in_proj_covar=tensor([0.0548, 0.0685, 0.0568, 0.0480, 0.0429, 0.0441, 0.0575, 0.0521], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:22:11,614 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104022.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 09:22:43,744 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.442e+02 2.893e+02 3.416e+02 6.492e+02, threshold=5.787e+02, percent-clipped=3.0 2023-04-29 09:22:47,579 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=104049.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:22:49,937 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104051.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:22:50,610 INFO [train.py:904] (6/8) Epoch 11, batch 2550, loss[loss=0.2059, simple_loss=0.282, pruned_loss=0.06486, over 16479.00 frames. ], tot_loss[loss=0.191, simple_loss=0.2746, pruned_loss=0.05373, over 3298174.47 frames. ], batch size: 146, lr: 6.30e-03, grad_scale: 8.0 2023-04-29 09:23:00,521 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104058.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:23:13,763 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6870, 3.8284, 2.1155, 4.0796, 2.8117, 3.9941, 2.2608, 3.0330], device='cuda:6'), covar=tensor([0.0192, 0.0271, 0.1379, 0.0200, 0.0660, 0.0597, 0.1325, 0.0554], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0168, 0.0187, 0.0134, 0.0167, 0.0212, 0.0194, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 09:23:15,997 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104070.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 09:23:17,450 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 09:23:51,819 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104096.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:23:59,572 INFO [train.py:904] (6/8) Epoch 11, batch 2600, loss[loss=0.1806, simple_loss=0.2682, pruned_loss=0.04646, over 17153.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2739, pruned_loss=0.05301, over 3299417.44 frames. ], batch size: 46, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:24:11,997 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104110.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:24:16,939 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6327, 2.5416, 1.9738, 2.2079, 2.9225, 2.6402, 3.4774, 3.2674], device='cuda:6'), covar=tensor([0.0099, 0.0374, 0.0471, 0.0415, 0.0226, 0.0330, 0.0180, 0.0195], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0208, 0.0201, 0.0202, 0.0206, 0.0207, 0.0215, 0.0197], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:24:26,714 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104121.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:25:01,154 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.655e+02 3.049e+02 3.595e+02 6.269e+02, threshold=6.098e+02, percent-clipped=2.0 2023-04-29 09:25:10,380 INFO [train.py:904] (6/8) Epoch 11, batch 2650, loss[loss=0.2097, simple_loss=0.2997, pruned_loss=0.05985, over 16555.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2746, pruned_loss=0.0528, over 3308998.46 frames. ], batch size: 62, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:26:18,846 INFO [train.py:904] (6/8) Epoch 11, batch 2700, loss[loss=0.1958, simple_loss=0.2718, pruned_loss=0.05993, over 12559.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2736, pruned_loss=0.0518, over 3300967.07 frames. ], batch size: 246, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:27:00,691 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104232.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 09:27:13,598 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0966, 5.0489, 5.5261, 5.5152, 5.5520, 5.1168, 5.1124, 4.8203], device='cuda:6'), covar=tensor([0.0301, 0.0480, 0.0379, 0.0422, 0.0465, 0.0340, 0.0889, 0.0428], device='cuda:6'), in_proj_covar=tensor([0.0335, 0.0347, 0.0350, 0.0328, 0.0396, 0.0366, 0.0466, 0.0293], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 09:27:19,005 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.677e+02 2.334e+02 2.747e+02 3.753e+02 9.915e+02, threshold=5.495e+02, percent-clipped=4.0 2023-04-29 09:27:27,257 INFO [train.py:904] (6/8) Epoch 11, batch 2750, loss[loss=0.1946, simple_loss=0.2939, pruned_loss=0.04765, over 16750.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2734, pruned_loss=0.05085, over 3314951.51 frames. ], batch size: 62, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:28:10,900 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 09:28:32,583 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5223, 3.9777, 3.9724, 2.2953, 3.3083, 2.5391, 4.0327, 4.0307], device='cuda:6'), covar=tensor([0.0229, 0.0818, 0.0443, 0.1557, 0.0674, 0.0917, 0.0560, 0.0939], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0147, 0.0158, 0.0142, 0.0136, 0.0124, 0.0136, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 09:28:36,570 INFO [train.py:904] (6/8) Epoch 11, batch 2800, loss[loss=0.1538, simple_loss=0.2428, pruned_loss=0.03241, over 16823.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2731, pruned_loss=0.05078, over 3319507.37 frames. ], batch size: 39, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:29:24,721 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-29 09:29:37,339 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.621e+02 2.406e+02 2.849e+02 3.526e+02 7.229e+02, threshold=5.698e+02, percent-clipped=5.0 2023-04-29 09:29:37,666 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104346.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:29:41,391 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0501, 2.5241, 2.5972, 4.9021, 2.4203, 3.0185, 2.5955, 2.7351], device='cuda:6'), covar=tensor([0.0790, 0.3216, 0.2155, 0.0313, 0.3610, 0.2110, 0.2853, 0.3087], device='cuda:6'), in_proj_covar=tensor([0.0367, 0.0392, 0.0331, 0.0325, 0.0412, 0.0451, 0.0355, 0.0465], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:29:44,355 INFO [train.py:904] (6/8) Epoch 11, batch 2850, loss[loss=0.1828, simple_loss=0.2615, pruned_loss=0.0521, over 15854.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2721, pruned_loss=0.05046, over 3322129.94 frames. ], batch size: 35, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:29:45,779 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104353.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:30:08,929 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-29 09:30:18,905 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6758, 2.7736, 2.4438, 4.1637, 3.5002, 4.1568, 1.5146, 2.9653], device='cuda:6'), covar=tensor([0.1288, 0.0574, 0.1030, 0.0136, 0.0144, 0.0341, 0.1334, 0.0690], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0157, 0.0180, 0.0146, 0.0198, 0.0209, 0.0177, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 09:30:44,795 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104396.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:30:52,870 INFO [train.py:904] (6/8) Epoch 11, batch 2900, loss[loss=0.2051, simple_loss=0.2702, pruned_loss=0.07001, over 16502.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2714, pruned_loss=0.05137, over 3321371.08 frames. ], batch size: 146, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:30:57,319 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104405.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:31:00,707 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 09:31:01,683 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8477, 2.9824, 2.7391, 4.3861, 3.7061, 4.2944, 1.5843, 3.1677], device='cuda:6'), covar=tensor([0.1253, 0.0578, 0.0924, 0.0142, 0.0226, 0.0348, 0.1330, 0.0671], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0158, 0.0181, 0.0147, 0.0199, 0.0211, 0.0178, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 09:31:20,097 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104421.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:31:53,053 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104444.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:31:55,136 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.712e+02 2.403e+02 2.933e+02 3.429e+02 6.056e+02, threshold=5.866e+02, percent-clipped=1.0 2023-04-29 09:32:04,097 INFO [train.py:904] (6/8) Epoch 11, batch 2950, loss[loss=0.1857, simple_loss=0.2732, pruned_loss=0.04909, over 17049.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2705, pruned_loss=0.05162, over 3328954.21 frames. ], batch size: 53, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:32:18,906 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-29 09:32:28,530 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104469.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:32:29,881 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4498, 3.2971, 2.5970, 2.1681, 2.3368, 2.0646, 3.3320, 3.1936], device='cuda:6'), covar=tensor([0.2335, 0.0692, 0.1513, 0.2072, 0.2403, 0.1940, 0.0505, 0.1110], device='cuda:6'), in_proj_covar=tensor([0.0305, 0.0258, 0.0282, 0.0276, 0.0284, 0.0222, 0.0269, 0.0301], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:33:12,380 INFO [train.py:904] (6/8) Epoch 11, batch 3000, loss[loss=0.206, simple_loss=0.2766, pruned_loss=0.06767, over 16380.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.27, pruned_loss=0.05181, over 3333951.34 frames. ], batch size: 146, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:33:12,381 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 09:33:22,063 INFO [train.py:938] (6/8) Epoch 11, validation: loss=0.1413, simple_loss=0.2475, pruned_loss=0.01754, over 944034.00 frames. 2023-04-29 09:33:22,063 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 09:34:04,295 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104532.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 09:34:20,848 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.544e+02 3.030e+02 3.628e+02 6.112e+02, threshold=6.060e+02, percent-clipped=1.0 2023-04-29 09:34:30,327 INFO [train.py:904] (6/8) Epoch 11, batch 3050, loss[loss=0.1852, simple_loss=0.2799, pruned_loss=0.04525, over 17230.00 frames. ], tot_loss[loss=0.1876, simple_loss=0.2705, pruned_loss=0.05232, over 3340039.15 frames. ], batch size: 52, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:35:07,351 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104580.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:35:37,109 INFO [train.py:904] (6/8) Epoch 11, batch 3100, loss[loss=0.2053, simple_loss=0.2819, pruned_loss=0.06434, over 16506.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2702, pruned_loss=0.05211, over 3347497.92 frames. ], batch size: 68, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:35:52,501 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8533, 3.9859, 2.0539, 4.6081, 2.8207, 4.4922, 2.0366, 3.0852], device='cuda:6'), covar=tensor([0.0190, 0.0248, 0.1548, 0.0143, 0.0695, 0.0333, 0.1658, 0.0587], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0169, 0.0188, 0.0136, 0.0171, 0.0215, 0.0196, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 09:36:03,311 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6988, 3.1208, 2.7833, 4.9794, 4.2226, 4.5200, 1.4516, 3.2773], device='cuda:6'), covar=tensor([0.1344, 0.0617, 0.1089, 0.0149, 0.0236, 0.0341, 0.1543, 0.0691], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0157, 0.0178, 0.0146, 0.0198, 0.0209, 0.0176, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 09:36:34,729 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0290, 1.8974, 2.4663, 2.9155, 2.7022, 3.4260, 2.0618, 3.2795], device='cuda:6'), covar=tensor([0.0165, 0.0349, 0.0247, 0.0208, 0.0249, 0.0116, 0.0363, 0.0119], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0173, 0.0158, 0.0160, 0.0170, 0.0125, 0.0170, 0.0117], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 09:36:39,246 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.673e+02 2.465e+02 3.037e+02 3.411e+02 8.178e+02, threshold=6.075e+02, percent-clipped=4.0 2023-04-29 09:36:39,572 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104646.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:36:47,518 INFO [train.py:904] (6/8) Epoch 11, batch 3150, loss[loss=0.1907, simple_loss=0.272, pruned_loss=0.05469, over 16507.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.27, pruned_loss=0.05205, over 3344297.64 frames. ], batch size: 68, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:36:49,984 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104653.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:37:46,968 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104694.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:37:56,591 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104701.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:37:57,453 INFO [train.py:904] (6/8) Epoch 11, batch 3200, loss[loss=0.1827, simple_loss=0.2641, pruned_loss=0.05064, over 16810.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.269, pruned_loss=0.05171, over 3336892.83 frames. ], batch size: 39, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:38:01,271 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104705.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:38:57,880 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.739e+02 2.472e+02 2.965e+02 3.891e+02 6.936e+02, threshold=5.930e+02, percent-clipped=3.0 2023-04-29 09:39:06,607 INFO [train.py:904] (6/8) Epoch 11, batch 3250, loss[loss=0.1614, simple_loss=0.2487, pruned_loss=0.03707, over 17160.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2696, pruned_loss=0.05203, over 3337683.32 frames. ], batch size: 46, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:39:08,055 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104753.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:40:15,729 INFO [train.py:904] (6/8) Epoch 11, batch 3300, loss[loss=0.1938, simple_loss=0.287, pruned_loss=0.0503, over 16726.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.2712, pruned_loss=0.05268, over 3327824.73 frames. ], batch size: 62, lr: 6.27e-03, grad_scale: 8.0 2023-04-29 09:40:56,945 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 09:41:16,318 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.335e+02 2.858e+02 3.500e+02 6.085e+02, threshold=5.716e+02, percent-clipped=1.0 2023-04-29 09:41:24,648 INFO [train.py:904] (6/8) Epoch 11, batch 3350, loss[loss=0.1761, simple_loss=0.2703, pruned_loss=0.041, over 17140.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2716, pruned_loss=0.05301, over 3322855.38 frames. ], batch size: 48, lr: 6.27e-03, grad_scale: 8.0 2023-04-29 09:41:26,224 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8517, 1.2728, 1.6815, 1.7292, 1.8068, 2.0111, 1.5567, 1.8408], device='cuda:6'), covar=tensor([0.0181, 0.0294, 0.0157, 0.0210, 0.0179, 0.0124, 0.0290, 0.0086], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0173, 0.0158, 0.0161, 0.0170, 0.0126, 0.0171, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 09:41:58,120 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3512, 2.1033, 2.2507, 4.1284, 2.0933, 2.6187, 2.2186, 2.3886], device='cuda:6'), covar=tensor([0.1013, 0.3174, 0.2106, 0.0420, 0.3367, 0.2105, 0.2979, 0.2644], device='cuda:6'), in_proj_covar=tensor([0.0368, 0.0393, 0.0331, 0.0325, 0.0413, 0.0453, 0.0356, 0.0464], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:42:33,960 INFO [train.py:904] (6/8) Epoch 11, batch 3400, loss[loss=0.1819, simple_loss=0.2702, pruned_loss=0.04684, over 17129.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2708, pruned_loss=0.05259, over 3318130.09 frames. ], batch size: 47, lr: 6.27e-03, grad_scale: 8.0 2023-04-29 09:43:23,017 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2957, 3.2569, 3.5757, 2.3585, 3.1994, 3.6146, 3.2849, 2.1165], device='cuda:6'), covar=tensor([0.0370, 0.0099, 0.0042, 0.0306, 0.0075, 0.0067, 0.0072, 0.0347], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0071, 0.0071, 0.0126, 0.0079, 0.0091, 0.0077, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 09:43:33,847 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.318e+02 2.811e+02 3.324e+02 7.497e+02, threshold=5.622e+02, percent-clipped=1.0 2023-04-29 09:43:41,667 INFO [train.py:904] (6/8) Epoch 11, batch 3450, loss[loss=0.1538, simple_loss=0.2352, pruned_loss=0.03623, over 17026.00 frames. ], tot_loss[loss=0.1869, simple_loss=0.2696, pruned_loss=0.05209, over 3322714.78 frames. ], batch size: 41, lr: 6.27e-03, grad_scale: 8.0 2023-04-29 09:44:08,682 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9078, 3.3401, 2.6965, 5.0555, 4.3522, 4.4657, 1.4479, 3.0374], device='cuda:6'), covar=tensor([0.1146, 0.0520, 0.0983, 0.0132, 0.0201, 0.0400, 0.1349, 0.0722], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0158, 0.0178, 0.0148, 0.0200, 0.0211, 0.0177, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 09:44:13,591 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0510, 4.4991, 4.3656, 3.2370, 3.7619, 4.4094, 4.0040, 2.6628], device='cuda:6'), covar=tensor([0.0355, 0.0041, 0.0047, 0.0283, 0.0081, 0.0079, 0.0063, 0.0371], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0071, 0.0071, 0.0125, 0.0079, 0.0090, 0.0077, 0.0117], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 09:44:33,230 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2477, 2.0132, 2.2536, 3.8294, 2.1115, 2.3819, 2.1420, 2.2147], device='cuda:6'), covar=tensor([0.1018, 0.3400, 0.2142, 0.0444, 0.3310, 0.2251, 0.2968, 0.2925], device='cuda:6'), in_proj_covar=tensor([0.0370, 0.0395, 0.0333, 0.0328, 0.0416, 0.0456, 0.0359, 0.0467], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:44:52,761 INFO [train.py:904] (6/8) Epoch 11, batch 3500, loss[loss=0.1584, simple_loss=0.249, pruned_loss=0.03388, over 17201.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2678, pruned_loss=0.05119, over 3325403.66 frames. ], batch size: 46, lr: 6.27e-03, grad_scale: 8.0 2023-04-29 09:45:47,632 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-29 09:45:55,145 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.324e+02 2.297e+02 2.659e+02 3.238e+02 6.959e+02, threshold=5.317e+02, percent-clipped=2.0 2023-04-29 09:46:03,258 INFO [train.py:904] (6/8) Epoch 11, batch 3550, loss[loss=0.156, simple_loss=0.2411, pruned_loss=0.03549, over 17216.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2662, pruned_loss=0.05022, over 3329210.85 frames. ], batch size: 45, lr: 6.27e-03, grad_scale: 4.0 2023-04-29 09:46:48,335 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6435, 2.2776, 2.3957, 4.3654, 2.2455, 2.6850, 2.3634, 2.5073], device='cuda:6'), covar=tensor([0.0915, 0.3135, 0.2111, 0.0351, 0.3404, 0.2073, 0.2897, 0.2819], device='cuda:6'), in_proj_covar=tensor([0.0366, 0.0390, 0.0330, 0.0323, 0.0411, 0.0452, 0.0354, 0.0463], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:47:12,588 INFO [train.py:904] (6/8) Epoch 11, batch 3600, loss[loss=0.2156, simple_loss=0.2872, pruned_loss=0.07201, over 11352.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.2649, pruned_loss=0.05002, over 3310773.24 frames. ], batch size: 248, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:47:58,168 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 09:48:17,988 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.381e+02 2.999e+02 3.559e+02 5.389e+02, threshold=5.997e+02, percent-clipped=2.0 2023-04-29 09:48:24,363 INFO [train.py:904] (6/8) Epoch 11, batch 3650, loss[loss=0.1657, simple_loss=0.2454, pruned_loss=0.04296, over 15406.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2636, pruned_loss=0.05055, over 3304480.75 frames. ], batch size: 191, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:48:53,110 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 09:49:07,486 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1042, 4.2470, 2.3351, 4.7564, 3.2518, 4.7329, 2.5443, 3.3296], device='cuda:6'), covar=tensor([0.0175, 0.0252, 0.1403, 0.0245, 0.0630, 0.0351, 0.1287, 0.0580], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0168, 0.0187, 0.0136, 0.0167, 0.0213, 0.0194, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 09:49:37,392 INFO [train.py:904] (6/8) Epoch 11, batch 3700, loss[loss=0.2173, simple_loss=0.2873, pruned_loss=0.07368, over 11416.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2623, pruned_loss=0.05169, over 3285236.03 frames. ], batch size: 248, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:49:44,128 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-29 09:50:41,009 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.627e+02 2.380e+02 2.864e+02 3.354e+02 5.031e+02, threshold=5.728e+02, percent-clipped=0.0 2023-04-29 09:50:48,711 INFO [train.py:904] (6/8) Epoch 11, batch 3750, loss[loss=0.1834, simple_loss=0.257, pruned_loss=0.05484, over 16409.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2638, pruned_loss=0.05375, over 3279393.83 frames. ], batch size: 68, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:51:57,075 INFO [train.py:904] (6/8) Epoch 11, batch 3800, loss[loss=0.1936, simple_loss=0.2616, pruned_loss=0.06281, over 16934.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2649, pruned_loss=0.0555, over 3283004.65 frames. ], batch size: 109, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:53:02,335 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.313e+02 2.608e+02 3.308e+02 8.159e+02, threshold=5.217e+02, percent-clipped=3.0 2023-04-29 09:53:06,407 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7155, 3.9249, 2.8343, 2.2294, 2.7862, 2.2606, 3.9815, 3.5358], device='cuda:6'), covar=tensor([0.2403, 0.0566, 0.1532, 0.2255, 0.2294, 0.1788, 0.0487, 0.0972], device='cuda:6'), in_proj_covar=tensor([0.0304, 0.0258, 0.0284, 0.0278, 0.0290, 0.0223, 0.0269, 0.0301], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:53:08,977 INFO [train.py:904] (6/8) Epoch 11, batch 3850, loss[loss=0.1753, simple_loss=0.254, pruned_loss=0.04836, over 16506.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2647, pruned_loss=0.05582, over 3284100.77 frames. ], batch size: 68, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:54:00,683 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105388.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:54:06,602 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.17 vs. limit=5.0 2023-04-29 09:54:19,876 INFO [train.py:904] (6/8) Epoch 11, batch 3900, loss[loss=0.1811, simple_loss=0.2541, pruned_loss=0.05404, over 16389.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2642, pruned_loss=0.05603, over 3283475.81 frames. ], batch size: 146, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:54:35,486 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3072, 4.2629, 4.2073, 4.0109, 3.9339, 4.2942, 4.0103, 4.0299], device='cuda:6'), covar=tensor([0.0629, 0.0558, 0.0311, 0.0284, 0.0845, 0.0401, 0.0660, 0.0552], device='cuda:6'), in_proj_covar=tensor([0.0260, 0.0327, 0.0306, 0.0284, 0.0330, 0.0326, 0.0209, 0.0355], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 09:55:14,946 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6179, 2.3708, 1.8708, 2.1903, 2.8100, 2.6092, 2.8931, 2.8491], device='cuda:6'), covar=tensor([0.0145, 0.0245, 0.0356, 0.0316, 0.0147, 0.0209, 0.0174, 0.0197], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0205, 0.0199, 0.0200, 0.0205, 0.0202, 0.0213, 0.0197], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 09:55:25,282 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.279e+02 2.712e+02 3.395e+02 5.874e+02, threshold=5.424e+02, percent-clipped=4.0 2023-04-29 09:55:28,972 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105449.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 09:55:31,848 INFO [train.py:904] (6/8) Epoch 11, batch 3950, loss[loss=0.1876, simple_loss=0.2592, pruned_loss=0.05794, over 16722.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2644, pruned_loss=0.05663, over 3290031.79 frames. ], batch size: 83, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 09:56:21,694 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105486.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:56:27,456 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-29 09:56:44,039 INFO [train.py:904] (6/8) Epoch 11, batch 4000, loss[loss=0.1764, simple_loss=0.2576, pruned_loss=0.04764, over 16785.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2641, pruned_loss=0.05646, over 3294677.43 frames. ], batch size: 83, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 09:56:44,478 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105502.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:57:48,231 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.126e+02 2.510e+02 3.113e+02 5.391e+02, threshold=5.020e+02, percent-clipped=0.0 2023-04-29 09:57:48,844 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105547.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:57:55,346 INFO [train.py:904] (6/8) Epoch 11, batch 4050, loss[loss=0.1845, simple_loss=0.2651, pruned_loss=0.05192, over 16893.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.264, pruned_loss=0.05531, over 3293529.30 frames. ], batch size: 109, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 09:58:07,909 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-29 09:58:10,147 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-29 09:58:11,880 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105563.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:59:08,731 INFO [train.py:904] (6/8) Epoch 11, batch 4100, loss[loss=0.1834, simple_loss=0.2691, pruned_loss=0.04883, over 16547.00 frames. ], tot_loss[loss=0.187, simple_loss=0.2653, pruned_loss=0.05435, over 3289387.11 frames. ], batch size: 68, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 09:59:18,633 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0727, 4.1757, 4.6437, 2.3431, 4.9658, 5.0472, 3.3748, 3.7136], device='cuda:6'), covar=tensor([0.0674, 0.0204, 0.0131, 0.1027, 0.0031, 0.0037, 0.0382, 0.0361], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0102, 0.0091, 0.0141, 0.0072, 0.0106, 0.0123, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 09:59:39,399 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-29 09:59:58,463 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 10:00:05,662 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7447, 3.8207, 2.1931, 4.5487, 2.8870, 4.4286, 2.5225, 3.0316], device='cuda:6'), covar=tensor([0.0233, 0.0319, 0.1495, 0.0097, 0.0696, 0.0336, 0.1197, 0.0632], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0167, 0.0188, 0.0133, 0.0167, 0.0211, 0.0195, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 10:00:17,327 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-29 10:00:18,929 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.211e+02 2.197e+02 2.575e+02 3.230e+02 7.189e+02, threshold=5.150e+02, percent-clipped=4.0 2023-04-29 10:00:26,723 INFO [train.py:904] (6/8) Epoch 11, batch 4150, loss[loss=0.2399, simple_loss=0.3246, pruned_loss=0.07757, over 16851.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.2739, pruned_loss=0.05774, over 3242087.90 frames. ], batch size: 116, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 10:01:05,142 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3204, 4.3507, 4.7310, 4.7074, 4.7291, 4.3829, 4.3674, 4.1440], device='cuda:6'), covar=tensor([0.0296, 0.0436, 0.0346, 0.0370, 0.0354, 0.0355, 0.0891, 0.0508], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0335, 0.0336, 0.0316, 0.0380, 0.0352, 0.0453, 0.0283], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 10:01:44,701 INFO [train.py:904] (6/8) Epoch 11, batch 4200, loss[loss=0.2617, simple_loss=0.3326, pruned_loss=0.09541, over 11433.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.281, pruned_loss=0.0597, over 3203508.21 frames. ], batch size: 247, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 10:02:30,037 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1382, 1.4547, 1.9515, 2.0717, 2.2240, 2.3668, 1.6199, 2.2332], device='cuda:6'), covar=tensor([0.0149, 0.0361, 0.0189, 0.0234, 0.0196, 0.0129, 0.0338, 0.0085], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0172, 0.0156, 0.0160, 0.0168, 0.0123, 0.0168, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 10:02:37,922 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-29 10:02:47,073 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0278, 2.6393, 2.6726, 1.8415, 2.8404, 2.8920, 2.4572, 2.4151], device='cuda:6'), covar=tensor([0.0696, 0.0201, 0.0199, 0.0951, 0.0099, 0.0179, 0.0415, 0.0413], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0102, 0.0091, 0.0141, 0.0071, 0.0105, 0.0123, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 10:02:49,761 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105744.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 10:02:53,622 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.308e+02 2.787e+02 3.444e+02 5.705e+02, threshold=5.573e+02, percent-clipped=4.0 2023-04-29 10:02:59,807 INFO [train.py:904] (6/8) Epoch 11, batch 4250, loss[loss=0.1983, simple_loss=0.2878, pruned_loss=0.05437, over 16305.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2846, pruned_loss=0.05985, over 3185942.07 frames. ], batch size: 165, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 10:03:30,056 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9449, 1.7661, 2.5288, 2.9525, 2.8155, 3.3861, 1.9528, 3.1638], device='cuda:6'), covar=tensor([0.0126, 0.0343, 0.0202, 0.0169, 0.0164, 0.0088, 0.0339, 0.0077], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0170, 0.0155, 0.0159, 0.0166, 0.0122, 0.0167, 0.0114], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 10:03:59,292 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105792.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:04:08,462 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105799.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:04:12,691 INFO [train.py:904] (6/8) Epoch 11, batch 4300, loss[loss=0.1828, simple_loss=0.2823, pruned_loss=0.04166, over 16868.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2853, pruned_loss=0.05874, over 3188751.14 frames. ], batch size: 96, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:05:11,664 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105842.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:05:17,304 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.245e+02 2.727e+02 3.248e+02 5.954e+02, threshold=5.453e+02, percent-clipped=2.0 2023-04-29 10:05:24,847 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2487, 1.9354, 1.4979, 1.8029, 2.2329, 1.9703, 2.2616, 2.4669], device='cuda:6'), covar=tensor([0.0108, 0.0268, 0.0410, 0.0311, 0.0174, 0.0260, 0.0142, 0.0156], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0201, 0.0197, 0.0196, 0.0201, 0.0200, 0.0206, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:05:25,482 INFO [train.py:904] (6/8) Epoch 11, batch 4350, loss[loss=0.2108, simple_loss=0.2967, pruned_loss=0.06239, over 16286.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2883, pruned_loss=0.0598, over 3170712.55 frames. ], batch size: 165, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:05:27,932 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105853.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:05:34,709 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105858.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:05:37,399 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105860.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:06:33,265 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6417, 3.9022, 4.0814, 2.4011, 3.1703, 2.4851, 4.0117, 3.9072], device='cuda:6'), covar=tensor([0.0209, 0.0616, 0.0501, 0.1638, 0.0809, 0.0877, 0.0554, 0.0899], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0145, 0.0156, 0.0144, 0.0135, 0.0124, 0.0137, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 10:06:38,330 INFO [train.py:904] (6/8) Epoch 11, batch 4400, loss[loss=0.2116, simple_loss=0.295, pruned_loss=0.06407, over 16736.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.29, pruned_loss=0.0605, over 3188511.76 frames. ], batch size: 134, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:07:01,453 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-04-29 10:07:40,705 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.666e+02 2.548e+02 2.924e+02 3.370e+02 5.572e+02, threshold=5.847e+02, percent-clipped=1.0 2023-04-29 10:07:48,892 INFO [train.py:904] (6/8) Epoch 11, batch 4450, loss[loss=0.2007, simple_loss=0.2967, pruned_loss=0.05233, over 16737.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2939, pruned_loss=0.0617, over 3202223.75 frames. ], batch size: 89, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:08:35,390 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0088, 5.0436, 4.8178, 4.5080, 4.5092, 4.9409, 4.8092, 4.5202], device='cuda:6'), covar=tensor([0.0428, 0.0212, 0.0203, 0.0219, 0.0723, 0.0251, 0.0228, 0.0509], device='cuda:6'), in_proj_covar=tensor([0.0239, 0.0301, 0.0285, 0.0264, 0.0304, 0.0299, 0.0194, 0.0329], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:09:04,902 INFO [train.py:904] (6/8) Epoch 11, batch 4500, loss[loss=0.2077, simple_loss=0.2946, pruned_loss=0.06045, over 17231.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2945, pruned_loss=0.06192, over 3214694.50 frames. ], batch size: 44, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:10:05,651 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106044.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:10:09,391 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 1.945e+02 2.292e+02 2.586e+02 4.524e+02, threshold=4.584e+02, percent-clipped=0.0 2023-04-29 10:10:17,384 INFO [train.py:904] (6/8) Epoch 11, batch 4550, loss[loss=0.2051, simple_loss=0.2851, pruned_loss=0.06254, over 16301.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.2952, pruned_loss=0.06269, over 3226613.44 frames. ], batch size: 35, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:10:18,319 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.04 vs. limit=5.0 2023-04-29 10:10:53,571 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9336, 5.1796, 4.9764, 4.9995, 4.6859, 4.6193, 4.6647, 5.2992], device='cuda:6'), covar=tensor([0.1110, 0.0719, 0.0864, 0.0659, 0.0711, 0.0814, 0.0854, 0.0744], device='cuda:6'), in_proj_covar=tensor([0.0529, 0.0658, 0.0544, 0.0459, 0.0419, 0.0427, 0.0548, 0.0504], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:10:53,635 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106077.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:11:15,421 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106092.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:11:29,147 INFO [train.py:904] (6/8) Epoch 11, batch 4600, loss[loss=0.2098, simple_loss=0.2938, pruned_loss=0.06287, over 17009.00 frames. ], tot_loss[loss=0.2105, simple_loss=0.2959, pruned_loss=0.06259, over 3232225.97 frames. ], batch size: 53, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:11:35,486 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8226, 1.2969, 1.6027, 1.7121, 1.7926, 1.9497, 1.4834, 1.7675], device='cuda:6'), covar=tensor([0.0161, 0.0263, 0.0142, 0.0181, 0.0164, 0.0101, 0.0289, 0.0078], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0171, 0.0156, 0.0160, 0.0167, 0.0123, 0.0169, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 10:12:00,397 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7603, 2.4450, 2.5355, 4.6789, 2.4338, 2.7841, 2.5344, 2.6211], device='cuda:6'), covar=tensor([0.0826, 0.2860, 0.2028, 0.0296, 0.3382, 0.1980, 0.2659, 0.2891], device='cuda:6'), in_proj_covar=tensor([0.0362, 0.0390, 0.0326, 0.0319, 0.0412, 0.0452, 0.0354, 0.0460], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:12:14,180 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3473, 2.8775, 2.5797, 2.2161, 2.1031, 2.0546, 2.8193, 2.7221], device='cuda:6'), covar=tensor([0.2299, 0.0670, 0.1476, 0.2057, 0.2161, 0.2004, 0.0502, 0.0932], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0254, 0.0282, 0.0277, 0.0285, 0.0221, 0.0266, 0.0296], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:12:22,911 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106138.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:12:28,020 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106142.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:12:35,552 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 1.926e+02 2.251e+02 2.695e+02 4.536e+02, threshold=4.502e+02, percent-clipped=0.0 2023-04-29 10:12:37,697 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106148.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:12:41,942 INFO [train.py:904] (6/8) Epoch 11, batch 4650, loss[loss=0.1816, simple_loss=0.2708, pruned_loss=0.04615, over 16874.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2942, pruned_loss=0.06216, over 3235418.56 frames. ], batch size: 96, lr: 6.23e-03, grad_scale: 8.0 2023-04-29 10:12:47,354 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106155.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:12:51,839 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106158.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:13:37,558 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106190.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:13:55,735 INFO [train.py:904] (6/8) Epoch 11, batch 4700, loss[loss=0.2077, simple_loss=0.2874, pruned_loss=0.06397, over 17054.00 frames. ], tot_loss[loss=0.2067, simple_loss=0.2911, pruned_loss=0.06114, over 3209763.10 frames. ], batch size: 53, lr: 6.23e-03, grad_scale: 8.0 2023-04-29 10:14:01,902 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106206.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:15:01,675 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.393e+02 2.142e+02 2.451e+02 2.986e+02 5.737e+02, threshold=4.902e+02, percent-clipped=3.0 2023-04-29 10:15:09,055 INFO [train.py:904] (6/8) Epoch 11, batch 4750, loss[loss=0.1879, simple_loss=0.2778, pruned_loss=0.04903, over 15338.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2873, pruned_loss=0.05928, over 3212981.84 frames. ], batch size: 191, lr: 6.23e-03, grad_scale: 8.0 2023-04-29 10:15:32,876 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106268.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:15:52,654 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8829, 4.8729, 4.8499, 4.0506, 4.8174, 1.7902, 4.5270, 4.7037], device='cuda:6'), covar=tensor([0.0095, 0.0078, 0.0101, 0.0527, 0.0078, 0.2231, 0.0126, 0.0162], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0116, 0.0164, 0.0158, 0.0134, 0.0177, 0.0151, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:16:22,052 INFO [train.py:904] (6/8) Epoch 11, batch 4800, loss[loss=0.1884, simple_loss=0.2779, pruned_loss=0.04947, over 16781.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2843, pruned_loss=0.05762, over 3200798.82 frames. ], batch size: 83, lr: 6.23e-03, grad_scale: 8.0 2023-04-29 10:17:02,209 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:17:12,260 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106336.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:17:27,376 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.167e+02 2.529e+02 2.965e+02 6.623e+02, threshold=5.058e+02, percent-clipped=3.0 2023-04-29 10:17:35,183 INFO [train.py:904] (6/8) Epoch 11, batch 4850, loss[loss=0.2031, simple_loss=0.296, pruned_loss=0.05507, over 16908.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2846, pruned_loss=0.05646, over 3199839.54 frames. ], batch size: 109, lr: 6.23e-03, grad_scale: 8.0 2023-04-29 10:18:38,647 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106397.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:18:45,468 INFO [train.py:904] (6/8) Epoch 11, batch 4900, loss[loss=0.1776, simple_loss=0.2657, pruned_loss=0.04477, over 16905.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2837, pruned_loss=0.05491, over 3195873.67 frames. ], batch size: 109, lr: 6.23e-03, grad_scale: 8.0 2023-04-29 10:19:29,953 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106433.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:19:50,195 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 2.157e+02 2.578e+02 2.904e+02 4.407e+02, threshold=5.156e+02, percent-clipped=0.0 2023-04-29 10:19:52,406 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106448.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:19:56,921 INFO [train.py:904] (6/8) Epoch 11, batch 4950, loss[loss=0.2189, simple_loss=0.3068, pruned_loss=0.06545, over 16851.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2837, pruned_loss=0.05475, over 3200980.96 frames. ], batch size: 116, lr: 6.22e-03, grad_scale: 8.0 2023-04-29 10:20:01,902 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106454.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:20:03,087 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106455.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:20:56,856 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5943, 2.1527, 2.2054, 4.0896, 1.9236, 2.6153, 2.1955, 2.3401], device='cuda:6'), covar=tensor([0.1080, 0.3699, 0.2409, 0.0553, 0.4414, 0.2544, 0.3165, 0.3561], device='cuda:6'), in_proj_covar=tensor([0.0363, 0.0392, 0.0327, 0.0319, 0.0411, 0.0451, 0.0355, 0.0460], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:21:00,657 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106496.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:21:08,412 INFO [train.py:904] (6/8) Epoch 11, batch 5000, loss[loss=0.1812, simple_loss=0.272, pruned_loss=0.04524, over 17159.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2854, pruned_loss=0.05484, over 3205336.86 frames. ], batch size: 46, lr: 6.22e-03, grad_scale: 8.0 2023-04-29 10:21:10,910 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106503.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:21:28,735 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106515.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:21:54,455 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1477, 3.3124, 3.5353, 1.5132, 3.5210, 3.7579, 2.7953, 2.6415], device='cuda:6'), covar=tensor([0.0851, 0.0188, 0.0141, 0.1357, 0.0127, 0.0081, 0.0437, 0.0505], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0101, 0.0090, 0.0140, 0.0070, 0.0102, 0.0122, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 10:22:04,200 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5022, 2.5768, 2.2465, 3.9867, 2.8135, 3.8712, 1.2625, 2.8159], device='cuda:6'), covar=tensor([0.1401, 0.0737, 0.1297, 0.0146, 0.0272, 0.0362, 0.1691, 0.0825], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0158, 0.0181, 0.0145, 0.0199, 0.0208, 0.0180, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 10:22:14,239 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.886e+02 2.415e+02 2.751e+02 3.160e+02 5.367e+02, threshold=5.502e+02, percent-clipped=2.0 2023-04-29 10:22:21,440 INFO [train.py:904] (6/8) Epoch 11, batch 5050, loss[loss=0.1974, simple_loss=0.2832, pruned_loss=0.05582, over 16759.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.2857, pruned_loss=0.05463, over 3219466.99 frames. ], batch size: 134, lr: 6.22e-03, grad_scale: 8.0 2023-04-29 10:23:32,363 INFO [train.py:904] (6/8) Epoch 11, batch 5100, loss[loss=0.187, simple_loss=0.271, pruned_loss=0.05153, over 16658.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2836, pruned_loss=0.05374, over 3230725.97 frames. ], batch size: 134, lr: 6.22e-03, grad_scale: 8.0 2023-04-29 10:23:57,222 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0512, 3.6778, 3.5837, 2.4662, 3.1499, 3.5584, 3.3296, 1.9302], device='cuda:6'), covar=tensor([0.0459, 0.0026, 0.0029, 0.0315, 0.0071, 0.0087, 0.0071, 0.0399], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0070, 0.0070, 0.0127, 0.0079, 0.0091, 0.0078, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 10:24:03,664 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106624.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:24:05,086 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-29 10:24:38,776 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.440e+02 2.185e+02 2.500e+02 2.940e+02 7.724e+02, threshold=5.000e+02, percent-clipped=1.0 2023-04-29 10:24:44,364 INFO [train.py:904] (6/8) Epoch 11, batch 5150, loss[loss=0.2116, simple_loss=0.3102, pruned_loss=0.05649, over 16249.00 frames. ], tot_loss[loss=0.195, simple_loss=0.2836, pruned_loss=0.05315, over 3219490.46 frames. ], batch size: 165, lr: 6.22e-03, grad_scale: 4.0 2023-04-29 10:25:30,529 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8618, 3.9672, 4.2480, 4.2208, 4.2104, 3.9384, 3.9164, 3.8895], device='cuda:6'), covar=tensor([0.0301, 0.0509, 0.0349, 0.0423, 0.0414, 0.0369, 0.0808, 0.0470], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0328, 0.0331, 0.0316, 0.0375, 0.0349, 0.0448, 0.0280], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 10:25:35,670 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4059, 4.3944, 4.3244, 3.6666, 4.3058, 1.6189, 4.0633, 4.1461], device='cuda:6'), covar=tensor([0.0084, 0.0071, 0.0125, 0.0369, 0.0079, 0.2387, 0.0122, 0.0173], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0117, 0.0165, 0.0160, 0.0136, 0.0179, 0.0152, 0.0153], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:25:43,747 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106692.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:25:50,472 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1714, 4.0303, 4.2296, 4.4112, 4.5287, 4.1194, 4.4791, 4.5344], device='cuda:6'), covar=tensor([0.1333, 0.0919, 0.1382, 0.0548, 0.0482, 0.1033, 0.0539, 0.0493], device='cuda:6'), in_proj_covar=tensor([0.0507, 0.0640, 0.0782, 0.0655, 0.0496, 0.0503, 0.0506, 0.0581], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:25:56,050 INFO [train.py:904] (6/8) Epoch 11, batch 5200, loss[loss=0.1799, simple_loss=0.2654, pruned_loss=0.04725, over 16883.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.283, pruned_loss=0.05332, over 3217446.45 frames. ], batch size: 96, lr: 6.22e-03, grad_scale: 8.0 2023-04-29 10:26:08,471 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7987, 3.2133, 3.2081, 1.8423, 2.8171, 2.2146, 3.2779, 3.3624], device='cuda:6'), covar=tensor([0.0215, 0.0667, 0.0596, 0.1801, 0.0804, 0.0927, 0.0614, 0.0723], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0144, 0.0158, 0.0143, 0.0136, 0.0124, 0.0137, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 10:26:42,858 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106733.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:26:51,596 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-29 10:27:02,574 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9127, 2.0270, 2.3250, 3.2342, 2.0862, 2.2520, 2.1858, 2.1557], device='cuda:6'), covar=tensor([0.1096, 0.2955, 0.1899, 0.0548, 0.3523, 0.2161, 0.2798, 0.2743], device='cuda:6'), in_proj_covar=tensor([0.0362, 0.0390, 0.0325, 0.0319, 0.0408, 0.0448, 0.0353, 0.0457], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:27:04,334 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.660e+02 2.187e+02 2.709e+02 3.054e+02 5.980e+02, threshold=5.418e+02, percent-clipped=1.0 2023-04-29 10:27:11,356 INFO [train.py:904] (6/8) Epoch 11, batch 5250, loss[loss=0.1991, simple_loss=0.2817, pruned_loss=0.05829, over 16863.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2805, pruned_loss=0.05304, over 3217257.38 frames. ], batch size: 116, lr: 6.22e-03, grad_scale: 8.0 2023-04-29 10:27:52,595 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106781.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:28:22,266 INFO [train.py:904] (6/8) Epoch 11, batch 5300, loss[loss=0.1673, simple_loss=0.2552, pruned_loss=0.03964, over 16222.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2771, pruned_loss=0.05184, over 3216601.00 frames. ], batch size: 165, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:28:32,939 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106810.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:29:27,199 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 2.160e+02 2.565e+02 2.956e+02 5.223e+02, threshold=5.130e+02, percent-clipped=0.0 2023-04-29 10:29:33,906 INFO [train.py:904] (6/8) Epoch 11, batch 5350, loss[loss=0.2262, simple_loss=0.3068, pruned_loss=0.07287, over 16567.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.275, pruned_loss=0.0508, over 3222871.80 frames. ], batch size: 75, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:30:11,414 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5118, 4.8020, 4.6094, 4.5979, 4.3029, 4.2681, 4.2726, 4.8543], device='cuda:6'), covar=tensor([0.1089, 0.0833, 0.0900, 0.0651, 0.0754, 0.1136, 0.0945, 0.0823], device='cuda:6'), in_proj_covar=tensor([0.0534, 0.0660, 0.0548, 0.0455, 0.0419, 0.0428, 0.0550, 0.0508], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:30:45,867 INFO [train.py:904] (6/8) Epoch 11, batch 5400, loss[loss=0.2157, simple_loss=0.303, pruned_loss=0.06422, over 16784.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2782, pruned_loss=0.05173, over 3231837.34 frames. ], batch size: 83, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:31:18,236 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106924.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:31:54,541 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.177e+02 2.658e+02 3.216e+02 5.881e+02, threshold=5.316e+02, percent-clipped=3.0 2023-04-29 10:32:02,032 INFO [train.py:904] (6/8) Epoch 11, batch 5450, loss[loss=0.1965, simple_loss=0.2882, pruned_loss=0.0524, over 16878.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2814, pruned_loss=0.05347, over 3205635.95 frames. ], batch size: 102, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:32:11,894 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8351, 1.2750, 1.6315, 1.7018, 1.7844, 1.9047, 1.5087, 1.8284], device='cuda:6'), covar=tensor([0.0172, 0.0281, 0.0140, 0.0199, 0.0166, 0.0112, 0.0262, 0.0074], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0173, 0.0155, 0.0162, 0.0168, 0.0124, 0.0170, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 10:32:34,433 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106972.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:33:03,987 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106992.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:33:18,600 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4651, 3.4400, 3.4408, 2.8846, 3.3393, 2.1454, 3.1829, 2.8810], device='cuda:6'), covar=tensor([0.0120, 0.0109, 0.0142, 0.0210, 0.0087, 0.1821, 0.0108, 0.0189], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0116, 0.0164, 0.0159, 0.0135, 0.0177, 0.0150, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:33:19,270 INFO [train.py:904] (6/8) Epoch 11, batch 5500, loss[loss=0.3122, simple_loss=0.3697, pruned_loss=0.1273, over 11312.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2889, pruned_loss=0.05853, over 3172622.27 frames. ], batch size: 248, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:34:18,908 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=107040.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:34:31,623 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.975e+02 3.119e+02 3.959e+02 4.735e+02 9.206e+02, threshold=7.917e+02, percent-clipped=14.0 2023-04-29 10:34:37,901 INFO [train.py:904] (6/8) Epoch 11, batch 5550, loss[loss=0.2569, simple_loss=0.3319, pruned_loss=0.09099, over 15309.00 frames. ], tot_loss[loss=0.2133, simple_loss=0.2974, pruned_loss=0.06459, over 3167715.92 frames. ], batch size: 191, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:35:57,900 INFO [train.py:904] (6/8) Epoch 11, batch 5600, loss[loss=0.1941, simple_loss=0.2826, pruned_loss=0.0528, over 16819.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.303, pruned_loss=0.06939, over 3133112.87 frames. ], batch size: 42, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:36:12,133 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107110.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:37:02,620 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107140.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 10:37:17,303 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.615e+02 3.503e+02 4.339e+02 5.380e+02 1.040e+03, threshold=8.679e+02, percent-clipped=3.0 2023-04-29 10:37:21,722 INFO [train.py:904] (6/8) Epoch 11, batch 5650, loss[loss=0.2108, simple_loss=0.29, pruned_loss=0.06581, over 16446.00 frames. ], tot_loss[loss=0.2281, simple_loss=0.3082, pruned_loss=0.07397, over 3093235.00 frames. ], batch size: 68, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:37:22,224 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107152.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:37:32,839 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=107158.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:38:29,844 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8465, 3.8973, 2.3691, 4.4472, 2.9004, 4.4311, 2.3690, 3.0764], device='cuda:6'), covar=tensor([0.0178, 0.0307, 0.1339, 0.0114, 0.0726, 0.0326, 0.1387, 0.0598], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0163, 0.0186, 0.0125, 0.0166, 0.0203, 0.0192, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 10:38:42,807 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107201.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 10:38:43,536 INFO [train.py:904] (6/8) Epoch 11, batch 5700, loss[loss=0.2491, simple_loss=0.3353, pruned_loss=0.0815, over 16466.00 frames. ], tot_loss[loss=0.2292, simple_loss=0.3089, pruned_loss=0.0747, over 3099416.80 frames. ], batch size: 146, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:38:47,298 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3545, 3.1705, 2.5828, 2.0759, 2.3357, 2.1240, 3.2951, 3.0715], device='cuda:6'), covar=tensor([0.2560, 0.0739, 0.1551, 0.2148, 0.2070, 0.1815, 0.0476, 0.1059], device='cuda:6'), in_proj_covar=tensor([0.0301, 0.0255, 0.0281, 0.0275, 0.0280, 0.0219, 0.0265, 0.0291], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:39:02,888 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107213.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 10:39:35,117 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 10:39:59,318 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.558e+02 3.709e+02 4.582e+02 6.080e+02 1.249e+03, threshold=9.164e+02, percent-clipped=1.0 2023-04-29 10:40:04,427 INFO [train.py:904] (6/8) Epoch 11, batch 5750, loss[loss=0.2385, simple_loss=0.3135, pruned_loss=0.08172, over 15271.00 frames. ], tot_loss[loss=0.2332, simple_loss=0.3124, pruned_loss=0.07698, over 3072420.23 frames. ], batch size: 190, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:40:35,629 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8651, 3.8766, 2.1954, 4.3824, 2.8850, 4.3353, 2.2910, 3.1472], device='cuda:6'), covar=tensor([0.0186, 0.0332, 0.1546, 0.0156, 0.0730, 0.0424, 0.1451, 0.0592], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0164, 0.0187, 0.0126, 0.0168, 0.0205, 0.0195, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 10:41:25,722 INFO [train.py:904] (6/8) Epoch 11, batch 5800, loss[loss=0.2264, simple_loss=0.3078, pruned_loss=0.07247, over 15181.00 frames. ], tot_loss[loss=0.2317, simple_loss=0.3119, pruned_loss=0.07571, over 3066370.95 frames. ], batch size: 191, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:41:59,214 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-29 10:42:39,024 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.001e+02 2.956e+02 3.560e+02 4.660e+02 9.709e+02, threshold=7.120e+02, percent-clipped=1.0 2023-04-29 10:42:43,689 INFO [train.py:904] (6/8) Epoch 11, batch 5850, loss[loss=0.2128, simple_loss=0.3034, pruned_loss=0.06103, over 16237.00 frames. ], tot_loss[loss=0.228, simple_loss=0.3091, pruned_loss=0.07341, over 3090472.67 frames. ], batch size: 165, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:42:52,937 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3275, 4.6408, 4.4103, 4.4164, 4.1332, 4.1090, 4.1881, 4.6906], device='cuda:6'), covar=tensor([0.1032, 0.0785, 0.0930, 0.0723, 0.0730, 0.1433, 0.0941, 0.0809], device='cuda:6'), in_proj_covar=tensor([0.0539, 0.0664, 0.0552, 0.0457, 0.0421, 0.0432, 0.0553, 0.0507], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:43:20,270 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-04-29 10:44:05,219 INFO [train.py:904] (6/8) Epoch 11, batch 5900, loss[loss=0.2199, simple_loss=0.3125, pruned_loss=0.06361, over 16866.00 frames. ], tot_loss[loss=0.2263, simple_loss=0.3079, pruned_loss=0.07235, over 3109396.96 frames. ], batch size: 96, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:45:21,997 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.026e+02 2.914e+02 3.759e+02 4.619e+02 1.179e+03, threshold=7.519e+02, percent-clipped=2.0 2023-04-29 10:45:26,032 INFO [train.py:904] (6/8) Epoch 11, batch 5950, loss[loss=0.2209, simple_loss=0.3101, pruned_loss=0.06586, over 16401.00 frames. ], tot_loss[loss=0.2255, simple_loss=0.3087, pruned_loss=0.07114, over 3108661.21 frames. ], batch size: 146, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:45:39,037 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0872, 3.0343, 3.1552, 1.6598, 3.2752, 3.3113, 2.6539, 2.5672], device='cuda:6'), covar=tensor([0.0734, 0.0188, 0.0148, 0.1127, 0.0073, 0.0142, 0.0356, 0.0447], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0098, 0.0087, 0.0137, 0.0069, 0.0101, 0.0118, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 10:46:40,585 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107496.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 10:46:48,947 INFO [train.py:904] (6/8) Epoch 11, batch 6000, loss[loss=0.2072, simple_loss=0.2923, pruned_loss=0.06104, over 16479.00 frames. ], tot_loss[loss=0.2245, simple_loss=0.3075, pruned_loss=0.07072, over 3100936.65 frames. ], batch size: 68, lr: 6.19e-03, grad_scale: 8.0 2023-04-29 10:46:48,947 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 10:46:59,888 INFO [train.py:938] (6/8) Epoch 11, validation: loss=0.163, simple_loss=0.2761, pruned_loss=0.02492, over 944034.00 frames. 2023-04-29 10:46:59,888 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 10:47:06,062 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8409, 4.0451, 3.2644, 2.4023, 3.0270, 2.5234, 4.4350, 3.9537], device='cuda:6'), covar=tensor([0.2332, 0.0640, 0.1340, 0.2000, 0.2144, 0.1574, 0.0361, 0.0815], device='cuda:6'), in_proj_covar=tensor([0.0301, 0.0256, 0.0281, 0.0276, 0.0281, 0.0219, 0.0266, 0.0291], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:47:10,249 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107508.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 10:47:13,525 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107510.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:47:13,698 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8801, 2.0449, 2.3326, 3.1533, 2.1851, 2.3207, 2.2554, 2.1681], device='cuda:6'), covar=tensor([0.0931, 0.2833, 0.1809, 0.0531, 0.3335, 0.1971, 0.2525, 0.2669], device='cuda:6'), in_proj_covar=tensor([0.0358, 0.0385, 0.0323, 0.0315, 0.0406, 0.0443, 0.0350, 0.0452], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:48:07,062 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 10:48:12,719 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.272e+02 3.327e+02 3.927e+02 4.902e+02 9.680e+02, threshold=7.854e+02, percent-clipped=6.0 2023-04-29 10:48:18,417 INFO [train.py:904] (6/8) Epoch 11, batch 6050, loss[loss=0.2157, simple_loss=0.3032, pruned_loss=0.06406, over 16914.00 frames. ], tot_loss[loss=0.2237, simple_loss=0.3063, pruned_loss=0.07059, over 3096958.35 frames. ], batch size: 109, lr: 6.19e-03, grad_scale: 8.0 2023-04-29 10:48:48,368 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107571.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:49:28,899 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-29 10:49:34,932 INFO [train.py:904] (6/8) Epoch 11, batch 6100, loss[loss=0.2623, simple_loss=0.3493, pruned_loss=0.08762, over 15478.00 frames. ], tot_loss[loss=0.2224, simple_loss=0.3056, pruned_loss=0.0696, over 3106319.17 frames. ], batch size: 190, lr: 6.19e-03, grad_scale: 8.0 2023-04-29 10:50:51,420 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.904e+02 2.980e+02 3.646e+02 4.413e+02 7.935e+02, threshold=7.292e+02, percent-clipped=3.0 2023-04-29 10:50:56,527 INFO [train.py:904] (6/8) Epoch 11, batch 6150, loss[loss=0.2077, simple_loss=0.2981, pruned_loss=0.05861, over 16673.00 frames. ], tot_loss[loss=0.2208, simple_loss=0.3036, pruned_loss=0.06904, over 3103575.20 frames. ], batch size: 89, lr: 6.19e-03, grad_scale: 8.0 2023-04-29 10:51:07,619 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107659.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:51:11,019 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3586, 4.4347, 4.7884, 4.7867, 4.7698, 4.4765, 4.4494, 4.2485], device='cuda:6'), covar=tensor([0.0288, 0.0423, 0.0329, 0.0366, 0.0470, 0.0351, 0.0896, 0.0475], device='cuda:6'), in_proj_covar=tensor([0.0326, 0.0341, 0.0343, 0.0325, 0.0388, 0.0361, 0.0463, 0.0294], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 10:51:36,920 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5614, 3.1254, 2.9880, 1.8799, 2.6886, 2.1989, 3.0940, 3.2752], device='cuda:6'), covar=tensor([0.0278, 0.0609, 0.0626, 0.1772, 0.0782, 0.0896, 0.0624, 0.0774], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0142, 0.0156, 0.0143, 0.0135, 0.0124, 0.0136, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 10:52:14,168 INFO [train.py:904] (6/8) Epoch 11, batch 6200, loss[loss=0.242, simple_loss=0.304, pruned_loss=0.09004, over 11459.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.302, pruned_loss=0.06888, over 3099413.11 frames. ], batch size: 248, lr: 6.19e-03, grad_scale: 4.0 2023-04-29 10:52:18,405 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0847, 3.4559, 3.4400, 1.8832, 2.9334, 2.4001, 3.4706, 3.6410], device='cuda:6'), covar=tensor([0.0253, 0.0640, 0.0549, 0.1925, 0.0813, 0.0879, 0.0608, 0.0810], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0142, 0.0156, 0.0143, 0.0135, 0.0124, 0.0136, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 10:52:20,579 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.51 vs. limit=5.0 2023-04-29 10:52:42,657 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107720.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:52:44,200 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8097, 2.4756, 2.4880, 3.3812, 2.3713, 3.6231, 1.4498, 2.7867], device='cuda:6'), covar=tensor([0.1272, 0.0665, 0.1046, 0.0147, 0.0199, 0.0373, 0.1551, 0.0737], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0158, 0.0180, 0.0143, 0.0198, 0.0207, 0.0180, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 10:53:07,927 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107737.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:53:27,428 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.232e+02 3.321e+02 3.959e+02 5.175e+02 1.024e+03, threshold=7.919e+02, percent-clipped=5.0 2023-04-29 10:53:29,994 INFO [train.py:904] (6/8) Epoch 11, batch 6250, loss[loss=0.2276, simple_loss=0.3198, pruned_loss=0.06772, over 16638.00 frames. ], tot_loss[loss=0.2189, simple_loss=0.3012, pruned_loss=0.06832, over 3100529.72 frames. ], batch size: 62, lr: 6.19e-03, grad_scale: 4.0 2023-04-29 10:54:36,355 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107796.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 10:54:37,578 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6198, 5.6000, 5.4186, 5.1226, 4.9830, 5.4500, 5.4099, 5.1279], device='cuda:6'), covar=tensor([0.0536, 0.0360, 0.0226, 0.0219, 0.1026, 0.0369, 0.0244, 0.0602], device='cuda:6'), in_proj_covar=tensor([0.0237, 0.0303, 0.0279, 0.0258, 0.0301, 0.0298, 0.0193, 0.0326], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:54:38,936 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107798.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 10:54:45,102 INFO [train.py:904] (6/8) Epoch 11, batch 6300, loss[loss=0.2129, simple_loss=0.2979, pruned_loss=0.0639, over 16887.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.3008, pruned_loss=0.06768, over 3109416.52 frames. ], batch size: 116, lr: 6.19e-03, grad_scale: 4.0 2023-04-29 10:54:54,778 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107808.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:55:52,955 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=107844.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 10:56:00,514 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.985e+02 3.066e+02 3.848e+02 4.977e+02 9.936e+02, threshold=7.696e+02, percent-clipped=3.0 2023-04-29 10:56:03,040 INFO [train.py:904] (6/8) Epoch 11, batch 6350, loss[loss=0.2176, simple_loss=0.299, pruned_loss=0.06815, over 16868.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.3022, pruned_loss=0.06914, over 3108508.94 frames. ], batch size: 96, lr: 6.18e-03, grad_scale: 4.0 2023-04-29 10:56:10,569 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=107856.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:56:25,828 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107866.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:57:20,953 INFO [train.py:904] (6/8) Epoch 11, batch 6400, loss[loss=0.2035, simple_loss=0.2934, pruned_loss=0.05682, over 16879.00 frames. ], tot_loss[loss=0.2217, simple_loss=0.303, pruned_loss=0.07018, over 3102460.30 frames. ], batch size: 102, lr: 6.18e-03, grad_scale: 8.0 2023-04-29 10:57:26,450 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-29 10:57:32,744 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3061, 4.0281, 3.9696, 4.4807, 4.5802, 4.2157, 4.4272, 4.5837], device='cuda:6'), covar=tensor([0.1520, 0.1245, 0.2469, 0.0993, 0.0864, 0.1292, 0.1175, 0.0983], device='cuda:6'), in_proj_covar=tensor([0.0510, 0.0636, 0.0767, 0.0645, 0.0495, 0.0498, 0.0513, 0.0577], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:58:23,910 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.73 vs. limit=5.0 2023-04-29 10:58:35,864 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.893e+02 3.293e+02 4.168e+02 5.156e+02 1.369e+03, threshold=8.335e+02, percent-clipped=6.0 2023-04-29 10:58:35,879 INFO [train.py:904] (6/8) Epoch 11, batch 6450, loss[loss=0.2485, simple_loss=0.3143, pruned_loss=0.09136, over 11710.00 frames. ], tot_loss[loss=0.2202, simple_loss=0.3022, pruned_loss=0.06913, over 3099828.41 frames. ], batch size: 247, lr: 6.18e-03, grad_scale: 2.0 2023-04-29 10:58:52,464 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8714, 1.7287, 1.5427, 1.5412, 1.8830, 1.5504, 1.6799, 1.9229], device='cuda:6'), covar=tensor([0.0131, 0.0219, 0.0326, 0.0285, 0.0159, 0.0214, 0.0196, 0.0188], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0198, 0.0196, 0.0196, 0.0198, 0.0200, 0.0203, 0.0188], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 10:59:54,943 INFO [train.py:904] (6/8) Epoch 11, batch 6500, loss[loss=0.2214, simple_loss=0.3103, pruned_loss=0.06628, over 16737.00 frames. ], tot_loss[loss=0.2186, simple_loss=0.3001, pruned_loss=0.06859, over 3092496.46 frames. ], batch size: 124, lr: 6.18e-03, grad_scale: 2.0 2023-04-29 11:00:14,833 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108015.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:00:48,964 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9750, 2.9345, 2.8768, 4.6654, 3.3797, 4.3441, 1.5708, 3.2210], device='cuda:6'), covar=tensor([0.1221, 0.0630, 0.0968, 0.0104, 0.0289, 0.0301, 0.1461, 0.0679], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0158, 0.0182, 0.0143, 0.0199, 0.0208, 0.0180, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 11:01:12,920 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.008e+02 3.175e+02 3.633e+02 4.590e+02 1.124e+03, threshold=7.266e+02, percent-clipped=3.0 2023-04-29 11:01:12,942 INFO [train.py:904] (6/8) Epoch 11, batch 6550, loss[loss=0.2197, simple_loss=0.3169, pruned_loss=0.06123, over 16137.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.3032, pruned_loss=0.06986, over 3081859.35 frames. ], batch size: 165, lr: 6.18e-03, grad_scale: 2.0 2023-04-29 11:01:26,309 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3235, 1.9584, 2.7893, 3.2900, 3.0172, 3.7659, 2.2215, 3.7306], device='cuda:6'), covar=tensor([0.0132, 0.0372, 0.0217, 0.0180, 0.0211, 0.0111, 0.0353, 0.0077], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0171, 0.0150, 0.0158, 0.0168, 0.0124, 0.0168, 0.0113], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 11:02:13,650 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108093.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 11:02:25,859 INFO [train.py:904] (6/8) Epoch 11, batch 6600, loss[loss=0.2527, simple_loss=0.3277, pruned_loss=0.0889, over 15254.00 frames. ], tot_loss[loss=0.2233, simple_loss=0.3054, pruned_loss=0.07054, over 3070624.60 frames. ], batch size: 190, lr: 6.18e-03, grad_scale: 2.0 2023-04-29 11:02:41,082 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3957, 4.3872, 4.8174, 4.7708, 4.7823, 4.4445, 4.4724, 4.2884], device='cuda:6'), covar=tensor([0.0281, 0.0448, 0.0378, 0.0405, 0.0427, 0.0361, 0.0857, 0.0460], device='cuda:6'), in_proj_covar=tensor([0.0325, 0.0341, 0.0343, 0.0322, 0.0389, 0.0360, 0.0461, 0.0291], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 11:03:41,606 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.097e+02 3.251e+02 4.038e+02 4.757e+02 1.371e+03, threshold=8.077e+02, percent-clipped=5.0 2023-04-29 11:03:41,621 INFO [train.py:904] (6/8) Epoch 11, batch 6650, loss[loss=0.2447, simple_loss=0.3213, pruned_loss=0.08408, over 15562.00 frames. ], tot_loss[loss=0.2258, simple_loss=0.3065, pruned_loss=0.07251, over 3056979.20 frames. ], batch size: 190, lr: 6.18e-03, grad_scale: 2.0 2023-04-29 11:03:51,303 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-29 11:04:03,300 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108166.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:04:56,954 INFO [train.py:904] (6/8) Epoch 11, batch 6700, loss[loss=0.2294, simple_loss=0.3093, pruned_loss=0.07474, over 16693.00 frames. ], tot_loss[loss=0.2252, simple_loss=0.3054, pruned_loss=0.07246, over 3042539.92 frames. ], batch size: 62, lr: 6.17e-03, grad_scale: 2.0 2023-04-29 11:05:15,640 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=108214.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:05:28,599 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2271, 3.3212, 3.5782, 3.5578, 3.5743, 3.3754, 3.3844, 3.4340], device='cuda:6'), covar=tensor([0.0370, 0.0655, 0.0451, 0.0446, 0.0490, 0.0454, 0.0779, 0.0475], device='cuda:6'), in_proj_covar=tensor([0.0326, 0.0341, 0.0346, 0.0323, 0.0392, 0.0360, 0.0462, 0.0291], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 11:06:13,523 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 3.397e+02 4.413e+02 5.907e+02 2.679e+03, threshold=8.826e+02, percent-clipped=7.0 2023-04-29 11:06:13,538 INFO [train.py:904] (6/8) Epoch 11, batch 6750, loss[loss=0.2166, simple_loss=0.294, pruned_loss=0.06962, over 16410.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.304, pruned_loss=0.07209, over 3049608.67 frames. ], batch size: 35, lr: 6.17e-03, grad_scale: 2.0 2023-04-29 11:07:28,508 INFO [train.py:904] (6/8) Epoch 11, batch 6800, loss[loss=0.2347, simple_loss=0.3212, pruned_loss=0.07407, over 16291.00 frames. ], tot_loss[loss=0.2228, simple_loss=0.3036, pruned_loss=0.07104, over 3080543.59 frames. ], batch size: 165, lr: 6.17e-03, grad_scale: 4.0 2023-04-29 11:07:49,241 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108315.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:08:05,617 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108325.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:08:21,334 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9121, 5.1586, 4.9060, 4.8654, 4.6484, 4.5222, 4.6281, 5.2276], device='cuda:6'), covar=tensor([0.0978, 0.0736, 0.0889, 0.0751, 0.0774, 0.0927, 0.0986, 0.0835], device='cuda:6'), in_proj_covar=tensor([0.0541, 0.0660, 0.0551, 0.0457, 0.0419, 0.0432, 0.0554, 0.0509], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:08:21,455 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5102, 3.4632, 3.6671, 1.6554, 3.8841, 3.9221, 2.8624, 2.7940], device='cuda:6'), covar=tensor([0.0691, 0.0194, 0.0163, 0.1243, 0.0052, 0.0107, 0.0391, 0.0454], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0099, 0.0088, 0.0138, 0.0069, 0.0102, 0.0120, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 11:08:45,532 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.125e+02 2.977e+02 3.381e+02 4.053e+02 7.016e+02, threshold=6.762e+02, percent-clipped=0.0 2023-04-29 11:08:45,554 INFO [train.py:904] (6/8) Epoch 11, batch 6850, loss[loss=0.2009, simple_loss=0.3064, pruned_loss=0.04769, over 17109.00 frames. ], tot_loss[loss=0.2237, simple_loss=0.3046, pruned_loss=0.07135, over 3084536.97 frames. ], batch size: 49, lr: 6.17e-03, grad_scale: 4.0 2023-04-29 11:09:01,710 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=108363.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:09:35,655 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108386.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:09:46,468 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108393.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:09:59,714 INFO [train.py:904] (6/8) Epoch 11, batch 6900, loss[loss=0.2105, simple_loss=0.299, pruned_loss=0.06099, over 16673.00 frames. ], tot_loss[loss=0.2235, simple_loss=0.3065, pruned_loss=0.07028, over 3111262.61 frames. ], batch size: 57, lr: 6.17e-03, grad_scale: 2.0 2023-04-29 11:10:00,990 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3181, 1.8123, 2.5499, 3.0798, 2.9491, 3.6708, 1.9729, 3.6904], device='cuda:6'), covar=tensor([0.0125, 0.0386, 0.0239, 0.0172, 0.0196, 0.0091, 0.0388, 0.0071], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0172, 0.0151, 0.0159, 0.0168, 0.0124, 0.0170, 0.0113], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 11:10:45,838 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108432.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:10:59,697 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=108441.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:11:15,789 INFO [train.py:904] (6/8) Epoch 11, batch 6950, loss[loss=0.2112, simple_loss=0.2992, pruned_loss=0.06165, over 16426.00 frames. ], tot_loss[loss=0.226, simple_loss=0.3082, pruned_loss=0.07188, over 3107041.66 frames. ], batch size: 68, lr: 6.17e-03, grad_scale: 2.0 2023-04-29 11:11:17,883 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.069e+02 2.933e+02 3.744e+02 4.621e+02 9.342e+02, threshold=7.489e+02, percent-clipped=9.0 2023-04-29 11:12:13,366 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2616, 5.2173, 5.0398, 4.6939, 4.6161, 5.0886, 5.1059, 4.7695], device='cuda:6'), covar=tensor([0.0621, 0.0695, 0.0303, 0.0328, 0.0982, 0.0537, 0.0298, 0.0921], device='cuda:6'), in_proj_covar=tensor([0.0232, 0.0300, 0.0275, 0.0253, 0.0294, 0.0293, 0.0191, 0.0321], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:12:20,909 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108493.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 11:12:33,547 INFO [train.py:904] (6/8) Epoch 11, batch 7000, loss[loss=0.2094, simple_loss=0.3067, pruned_loss=0.05604, over 16882.00 frames. ], tot_loss[loss=0.2261, simple_loss=0.3088, pruned_loss=0.07169, over 3094695.31 frames. ], batch size: 102, lr: 6.17e-03, grad_scale: 2.0 2023-04-29 11:13:52,263 INFO [train.py:904] (6/8) Epoch 11, batch 7050, loss[loss=0.2324, simple_loss=0.3142, pruned_loss=0.07528, over 16286.00 frames. ], tot_loss[loss=0.2257, simple_loss=0.3092, pruned_loss=0.0711, over 3098050.35 frames. ], batch size: 165, lr: 6.16e-03, grad_scale: 2.0 2023-04-29 11:13:53,481 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.056e+02 3.044e+02 3.927e+02 4.822e+02 1.171e+03, threshold=7.854e+02, percent-clipped=4.0 2023-04-29 11:14:33,465 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108577.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:14:45,336 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8080, 1.9837, 2.3967, 3.0751, 2.1367, 2.2218, 2.1969, 2.0891], device='cuda:6'), covar=tensor([0.0913, 0.3026, 0.1799, 0.0547, 0.3427, 0.2038, 0.2723, 0.2982], device='cuda:6'), in_proj_covar=tensor([0.0359, 0.0387, 0.0325, 0.0316, 0.0411, 0.0445, 0.0352, 0.0454], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:15:11,201 INFO [train.py:904] (6/8) Epoch 11, batch 7100, loss[loss=0.2677, simple_loss=0.3298, pruned_loss=0.1028, over 11735.00 frames. ], tot_loss[loss=0.225, simple_loss=0.3076, pruned_loss=0.07119, over 3093922.33 frames. ], batch size: 248, lr: 6.16e-03, grad_scale: 2.0 2023-04-29 11:15:50,300 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 11:16:07,787 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108638.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:16:27,132 INFO [train.py:904] (6/8) Epoch 11, batch 7150, loss[loss=0.2178, simple_loss=0.2969, pruned_loss=0.0693, over 16899.00 frames. ], tot_loss[loss=0.2232, simple_loss=0.3053, pruned_loss=0.07058, over 3099165.98 frames. ], batch size: 116, lr: 6.16e-03, grad_scale: 2.0 2023-04-29 11:16:28,930 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.138e+02 3.054e+02 3.516e+02 4.486e+02 8.068e+02, threshold=7.031e+02, percent-clipped=2.0 2023-04-29 11:16:51,012 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8684, 2.6889, 2.7162, 1.9067, 2.5272, 2.6871, 2.6419, 1.8510], device='cuda:6'), covar=tensor([0.0352, 0.0048, 0.0049, 0.0302, 0.0094, 0.0089, 0.0070, 0.0335], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0068, 0.0068, 0.0125, 0.0077, 0.0090, 0.0077, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 11:17:10,535 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108681.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:17:41,823 INFO [train.py:904] (6/8) Epoch 11, batch 7200, loss[loss=0.1989, simple_loss=0.2889, pruned_loss=0.05441, over 15406.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.303, pruned_loss=0.06885, over 3092078.47 frames. ], batch size: 190, lr: 6.16e-03, grad_scale: 4.0 2023-04-29 11:18:20,366 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 11:19:02,056 INFO [train.py:904] (6/8) Epoch 11, batch 7250, loss[loss=0.2064, simple_loss=0.2864, pruned_loss=0.06319, over 16889.00 frames. ], tot_loss[loss=0.2185, simple_loss=0.301, pruned_loss=0.06802, over 3073748.25 frames. ], batch size: 116, lr: 6.16e-03, grad_scale: 4.0 2023-04-29 11:19:03,145 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.903e+02 3.497e+02 4.380e+02 7.660e+02, threshold=6.994e+02, percent-clipped=1.0 2023-04-29 11:19:55,826 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108788.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 11:20:00,992 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108791.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:20:16,089 INFO [train.py:904] (6/8) Epoch 11, batch 7300, loss[loss=0.2089, simple_loss=0.2968, pruned_loss=0.06051, over 16786.00 frames. ], tot_loss[loss=0.2165, simple_loss=0.2994, pruned_loss=0.06678, over 3098746.10 frames. ], batch size: 102, lr: 6.16e-03, grad_scale: 4.0 2023-04-29 11:20:28,640 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6238, 2.5809, 1.7607, 2.7101, 2.1194, 2.7929, 2.0530, 2.3749], device='cuda:6'), covar=tensor([0.0234, 0.0331, 0.1183, 0.0150, 0.0652, 0.0382, 0.1078, 0.0514], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0160, 0.0184, 0.0123, 0.0165, 0.0201, 0.0192, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-29 11:21:34,169 INFO [train.py:904] (6/8) Epoch 11, batch 7350, loss[loss=0.2064, simple_loss=0.2955, pruned_loss=0.05868, over 16860.00 frames. ], tot_loss[loss=0.2191, simple_loss=0.3011, pruned_loss=0.06854, over 3062633.84 frames. ], batch size: 83, lr: 6.16e-03, grad_scale: 4.0 2023-04-29 11:21:34,690 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108852.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:21:35,285 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.782e+02 3.104e+02 3.844e+02 4.749e+02 1.066e+03, threshold=7.688e+02, percent-clipped=6.0 2023-04-29 11:22:54,323 INFO [train.py:904] (6/8) Epoch 11, batch 7400, loss[loss=0.2104, simple_loss=0.2972, pruned_loss=0.06183, over 16221.00 frames. ], tot_loss[loss=0.2195, simple_loss=0.3018, pruned_loss=0.06858, over 3085535.51 frames. ], batch size: 165, lr: 6.15e-03, grad_scale: 4.0 2023-04-29 11:23:28,579 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1178, 3.5816, 3.7300, 2.1753, 3.4179, 3.6739, 3.4864, 2.0698], device='cuda:6'), covar=tensor([0.0475, 0.0043, 0.0036, 0.0368, 0.0074, 0.0093, 0.0071, 0.0372], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0068, 0.0068, 0.0127, 0.0078, 0.0090, 0.0077, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 11:23:42,501 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108933.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:24:11,388 INFO [train.py:904] (6/8) Epoch 11, batch 7450, loss[loss=0.2348, simple_loss=0.3238, pruned_loss=0.07286, over 16809.00 frames. ], tot_loss[loss=0.2215, simple_loss=0.3034, pruned_loss=0.06985, over 3075472.59 frames. ], batch size: 124, lr: 6.15e-03, grad_scale: 4.0 2023-04-29 11:24:13,637 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.939e+02 3.048e+02 3.705e+02 5.111e+02 8.676e+02, threshold=7.410e+02, percent-clipped=2.0 2023-04-29 11:24:14,829 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0592, 3.4386, 3.5839, 2.1708, 3.3301, 3.5251, 3.4145, 1.8587], device='cuda:6'), covar=tensor([0.0458, 0.0048, 0.0039, 0.0362, 0.0071, 0.0090, 0.0057, 0.0414], device='cuda:6'), in_proj_covar=tensor([0.0129, 0.0068, 0.0069, 0.0127, 0.0078, 0.0090, 0.0077, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 11:24:59,228 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108981.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:25:03,261 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6132, 1.7051, 2.0901, 2.4868, 2.5386, 2.7972, 1.6791, 2.7821], device='cuda:6'), covar=tensor([0.0147, 0.0362, 0.0251, 0.0204, 0.0204, 0.0134, 0.0389, 0.0070], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0168, 0.0148, 0.0156, 0.0164, 0.0122, 0.0167, 0.0111], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 11:25:31,627 INFO [train.py:904] (6/8) Epoch 11, batch 7500, loss[loss=0.2646, simple_loss=0.3292, pruned_loss=0.1, over 11595.00 frames. ], tot_loss[loss=0.222, simple_loss=0.304, pruned_loss=0.06997, over 3058719.96 frames. ], batch size: 246, lr: 6.15e-03, grad_scale: 4.0 2023-04-29 11:26:16,354 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109029.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:26:28,526 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109037.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 11:26:51,084 INFO [train.py:904] (6/8) Epoch 11, batch 7550, loss[loss=0.1906, simple_loss=0.2692, pruned_loss=0.05596, over 16263.00 frames. ], tot_loss[loss=0.2213, simple_loss=0.303, pruned_loss=0.0698, over 3058429.39 frames. ], batch size: 35, lr: 6.15e-03, grad_scale: 4.0 2023-04-29 11:26:52,322 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.931e+02 3.123e+02 3.909e+02 4.676e+02 1.371e+03, threshold=7.817e+02, percent-clipped=2.0 2023-04-29 11:27:38,853 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0039, 4.3627, 3.4030, 2.3850, 3.0663, 2.6405, 4.6884, 3.8958], device='cuda:6'), covar=tensor([0.2403, 0.0584, 0.1444, 0.2177, 0.2569, 0.1671, 0.0374, 0.1040], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0255, 0.0282, 0.0275, 0.0280, 0.0221, 0.0265, 0.0292], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:27:46,166 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109088.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:28:02,014 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109098.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 11:28:06,702 INFO [train.py:904] (6/8) Epoch 11, batch 7600, loss[loss=0.2224, simple_loss=0.3002, pruned_loss=0.07233, over 15259.00 frames. ], tot_loss[loss=0.221, simple_loss=0.3023, pruned_loss=0.0699, over 3064981.48 frames. ], batch size: 190, lr: 6.15e-03, grad_scale: 8.0 2023-04-29 11:28:24,355 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-29 11:28:56,880 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109136.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:29:12,281 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109147.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:29:20,692 INFO [train.py:904] (6/8) Epoch 11, batch 7650, loss[loss=0.2114, simple_loss=0.2958, pruned_loss=0.06347, over 17181.00 frames. ], tot_loss[loss=0.2225, simple_loss=0.3034, pruned_loss=0.07081, over 3073912.70 frames. ], batch size: 45, lr: 6.15e-03, grad_scale: 4.0 2023-04-29 11:29:23,631 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.975e+02 2.787e+02 3.621e+02 4.390e+02 9.045e+02, threshold=7.242e+02, percent-clipped=3.0 2023-04-29 11:29:29,897 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7669, 5.0904, 5.2727, 5.0350, 5.0642, 5.6598, 5.0733, 4.8656], device='cuda:6'), covar=tensor([0.0932, 0.1914, 0.1781, 0.1757, 0.2419, 0.0995, 0.1629, 0.2606], device='cuda:6'), in_proj_covar=tensor([0.0350, 0.0485, 0.0528, 0.0418, 0.0558, 0.0553, 0.0420, 0.0573], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 11:30:12,206 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-04-29 11:30:25,787 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109194.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:30:36,526 INFO [train.py:904] (6/8) Epoch 11, batch 7700, loss[loss=0.2216, simple_loss=0.3065, pruned_loss=0.06839, over 16318.00 frames. ], tot_loss[loss=0.2221, simple_loss=0.303, pruned_loss=0.07061, over 3074319.40 frames. ], batch size: 165, lr: 6.15e-03, grad_scale: 4.0 2023-04-29 11:31:02,679 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 11:31:25,256 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109233.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:31:53,773 INFO [train.py:904] (6/8) Epoch 11, batch 7750, loss[loss=0.2204, simple_loss=0.3035, pruned_loss=0.06863, over 16568.00 frames. ], tot_loss[loss=0.2223, simple_loss=0.3031, pruned_loss=0.07075, over 3068996.86 frames. ], batch size: 62, lr: 6.14e-03, grad_scale: 4.0 2023-04-29 11:31:56,716 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.029e+02 3.061e+02 3.488e+02 4.422e+02 1.299e+03, threshold=6.977e+02, percent-clipped=6.0 2023-04-29 11:31:59,766 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109255.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:32:36,034 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8947, 5.4031, 5.5978, 5.2949, 5.3588, 5.9256, 5.4056, 5.1226], device='cuda:6'), covar=tensor([0.0905, 0.1637, 0.1809, 0.1837, 0.2195, 0.0886, 0.1517, 0.2631], device='cuda:6'), in_proj_covar=tensor([0.0350, 0.0484, 0.0524, 0.0417, 0.0553, 0.0552, 0.0417, 0.0571], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 11:32:39,669 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109281.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:32:58,639 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109294.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:33:10,715 INFO [train.py:904] (6/8) Epoch 11, batch 7800, loss[loss=0.217, simple_loss=0.3049, pruned_loss=0.0645, over 16530.00 frames. ], tot_loss[loss=0.2227, simple_loss=0.3033, pruned_loss=0.07108, over 3069774.78 frames. ], batch size: 68, lr: 6.14e-03, grad_scale: 4.0 2023-04-29 11:33:35,299 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109318.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:34:19,343 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5529, 3.5715, 1.9416, 3.9736, 2.5598, 3.9465, 2.0862, 2.8021], device='cuda:6'), covar=tensor([0.0209, 0.0327, 0.1686, 0.0148, 0.0825, 0.0549, 0.1652, 0.0730], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0164, 0.0188, 0.0125, 0.0167, 0.0204, 0.0195, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 11:34:25,574 INFO [train.py:904] (6/8) Epoch 11, batch 7850, loss[loss=0.2211, simple_loss=0.3094, pruned_loss=0.06641, over 16475.00 frames. ], tot_loss[loss=0.2233, simple_loss=0.3047, pruned_loss=0.07092, over 3084434.24 frames. ], batch size: 68, lr: 6.14e-03, grad_scale: 2.0 2023-04-29 11:34:30,495 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.254e+02 2.996e+02 3.807e+02 4.830e+02 8.310e+02, threshold=7.614e+02, percent-clipped=7.0 2023-04-29 11:34:31,699 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109355.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:35:03,635 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 11:35:07,156 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109379.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 11:35:27,274 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109393.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 11:35:41,205 INFO [train.py:904] (6/8) Epoch 11, batch 7900, loss[loss=0.2417, simple_loss=0.3078, pruned_loss=0.08783, over 11595.00 frames. ], tot_loss[loss=0.2229, simple_loss=0.304, pruned_loss=0.07085, over 3061838.78 frames. ], batch size: 248, lr: 6.14e-03, grad_scale: 2.0 2023-04-29 11:36:52,331 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109447.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:37:00,098 INFO [train.py:904] (6/8) Epoch 11, batch 7950, loss[loss=0.3063, simple_loss=0.3509, pruned_loss=0.1308, over 11520.00 frames. ], tot_loss[loss=0.2233, simple_loss=0.3041, pruned_loss=0.07122, over 3061132.20 frames. ], batch size: 248, lr: 6.14e-03, grad_scale: 2.0 2023-04-29 11:37:04,710 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.170e+02 3.029e+02 3.447e+02 4.404e+02 6.785e+02, threshold=6.894e+02, percent-clipped=0.0 2023-04-29 11:37:32,714 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5508, 2.5771, 2.2423, 3.8266, 2.2663, 3.8599, 1.3801, 2.5896], device='cuda:6'), covar=tensor([0.1648, 0.0864, 0.1448, 0.0276, 0.0366, 0.0453, 0.1971, 0.1040], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0159, 0.0182, 0.0145, 0.0201, 0.0209, 0.0181, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 11:37:35,725 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 11:38:04,155 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109495.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:38:14,151 INFO [train.py:904] (6/8) Epoch 11, batch 8000, loss[loss=0.2612, simple_loss=0.3252, pruned_loss=0.0986, over 11370.00 frames. ], tot_loss[loss=0.2243, simple_loss=0.3049, pruned_loss=0.07185, over 3060358.99 frames. ], batch size: 248, lr: 6.14e-03, grad_scale: 4.0 2023-04-29 11:38:47,876 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3713, 3.1059, 2.5966, 2.0440, 2.2386, 2.1442, 3.2835, 3.0484], device='cuda:6'), covar=tensor([0.2817, 0.0997, 0.1758, 0.2513, 0.2448, 0.1991, 0.0575, 0.1204], device='cuda:6'), in_proj_covar=tensor([0.0309, 0.0260, 0.0286, 0.0279, 0.0285, 0.0225, 0.0268, 0.0296], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:39:19,025 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4134, 4.4844, 4.2702, 4.0109, 3.9371, 4.3806, 4.1899, 4.0675], device='cuda:6'), covar=tensor([0.0627, 0.0448, 0.0296, 0.0303, 0.0917, 0.0451, 0.0501, 0.0641], device='cuda:6'), in_proj_covar=tensor([0.0235, 0.0304, 0.0277, 0.0256, 0.0295, 0.0296, 0.0191, 0.0322], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:39:24,981 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109550.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:39:27,473 INFO [train.py:904] (6/8) Epoch 11, batch 8050, loss[loss=0.2373, simple_loss=0.3195, pruned_loss=0.0776, over 16333.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.305, pruned_loss=0.07155, over 3070692.29 frames. ], batch size: 165, lr: 6.14e-03, grad_scale: 4.0 2023-04-29 11:39:31,003 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.160e+02 2.984e+02 3.790e+02 4.579e+02 1.063e+03, threshold=7.580e+02, percent-clipped=3.0 2023-04-29 11:39:36,514 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2003, 3.1975, 1.7896, 3.4470, 2.3546, 3.4934, 1.9383, 2.6770], device='cuda:6'), covar=tensor([0.0222, 0.0365, 0.1642, 0.0180, 0.0871, 0.0583, 0.1542, 0.0659], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0164, 0.0188, 0.0125, 0.0167, 0.0204, 0.0196, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 11:39:50,572 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9729, 2.2547, 2.2943, 2.7660, 2.0464, 3.1947, 1.7157, 2.6866], device='cuda:6'), covar=tensor([0.1227, 0.0559, 0.0972, 0.0171, 0.0142, 0.0351, 0.1463, 0.0676], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0161, 0.0183, 0.0146, 0.0203, 0.0210, 0.0183, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 11:40:06,474 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109579.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:40:40,895 INFO [train.py:904] (6/8) Epoch 11, batch 8100, loss[loss=0.2061, simple_loss=0.2909, pruned_loss=0.0607, over 16715.00 frames. ], tot_loss[loss=0.222, simple_loss=0.3034, pruned_loss=0.07026, over 3090754.51 frames. ], batch size: 134, lr: 6.13e-03, grad_scale: 4.0 2023-04-29 11:41:36,153 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-29 11:41:39,239 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109640.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:41:50,621 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-29 11:41:54,526 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109650.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:41:56,612 INFO [train.py:904] (6/8) Epoch 11, batch 8150, loss[loss=0.208, simple_loss=0.279, pruned_loss=0.06855, over 11542.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.3005, pruned_loss=0.0689, over 3089603.24 frames. ], batch size: 247, lr: 6.13e-03, grad_scale: 4.0 2023-04-29 11:42:01,358 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.997e+02 3.110e+02 3.865e+02 4.805e+02 7.636e+02, threshold=7.730e+02, percent-clipped=1.0 2023-04-29 11:42:05,481 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5308, 3.5167, 2.7393, 2.1068, 2.4376, 2.1829, 3.5620, 3.2313], device='cuda:6'), covar=tensor([0.2589, 0.0728, 0.1656, 0.2336, 0.2336, 0.1946, 0.0572, 0.1181], device='cuda:6'), in_proj_covar=tensor([0.0307, 0.0258, 0.0286, 0.0277, 0.0284, 0.0224, 0.0268, 0.0295], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:42:06,955 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4259, 4.5158, 4.7198, 4.5901, 4.6322, 5.1361, 4.6657, 4.4261], device='cuda:6'), covar=tensor([0.1258, 0.1796, 0.1855, 0.1762, 0.2245, 0.0955, 0.1436, 0.2426], device='cuda:6'), in_proj_covar=tensor([0.0350, 0.0484, 0.0526, 0.0420, 0.0550, 0.0552, 0.0415, 0.0568], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 11:42:08,225 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109659.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:42:30,927 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109674.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 11:42:59,249 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109693.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 11:43:04,363 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9655, 4.9856, 4.8239, 4.5181, 4.4778, 4.8881, 4.7356, 4.5007], device='cuda:6'), covar=tensor([0.0598, 0.0421, 0.0238, 0.0267, 0.0919, 0.0387, 0.0357, 0.0717], device='cuda:6'), in_proj_covar=tensor([0.0237, 0.0307, 0.0279, 0.0258, 0.0298, 0.0299, 0.0194, 0.0326], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:43:12,477 INFO [train.py:904] (6/8) Epoch 11, batch 8200, loss[loss=0.2065, simple_loss=0.2883, pruned_loss=0.06237, over 16611.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.2988, pruned_loss=0.06876, over 3078339.38 frames. ], batch size: 62, lr: 6.13e-03, grad_scale: 4.0 2023-04-29 11:43:42,608 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109720.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:44:16,311 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109741.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 11:44:33,547 INFO [train.py:904] (6/8) Epoch 11, batch 8250, loss[loss=0.2251, simple_loss=0.3147, pruned_loss=0.06774, over 15301.00 frames. ], tot_loss[loss=0.2151, simple_loss=0.2981, pruned_loss=0.06602, over 3082978.95 frames. ], batch size: 190, lr: 6.13e-03, grad_scale: 4.0 2023-04-29 11:44:38,007 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.186e+02 3.150e+02 3.966e+02 4.988e+02 1.179e+03, threshold=7.932e+02, percent-clipped=8.0 2023-04-29 11:44:46,601 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-29 11:45:14,724 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9432, 2.0233, 2.3152, 3.1960, 2.1327, 2.2702, 2.2182, 2.1002], device='cuda:6'), covar=tensor([0.0908, 0.3395, 0.2064, 0.0577, 0.3954, 0.2355, 0.3004, 0.3357], device='cuda:6'), in_proj_covar=tensor([0.0353, 0.0385, 0.0322, 0.0313, 0.0407, 0.0438, 0.0348, 0.0450], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:45:52,506 INFO [train.py:904] (6/8) Epoch 11, batch 8300, loss[loss=0.1927, simple_loss=0.2857, pruned_loss=0.04989, over 16759.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.2953, pruned_loss=0.06319, over 3068988.28 frames. ], batch size: 124, lr: 6.13e-03, grad_scale: 4.0 2023-04-29 11:47:09,188 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109850.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:47:11,569 INFO [train.py:904] (6/8) Epoch 11, batch 8350, loss[loss=0.1979, simple_loss=0.2952, pruned_loss=0.05027, over 16293.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.294, pruned_loss=0.0608, over 3068384.71 frames. ], batch size: 165, lr: 6.13e-03, grad_scale: 4.0 2023-04-29 11:47:16,940 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 2.495e+02 2.844e+02 3.376e+02 6.294e+02, threshold=5.687e+02, percent-clipped=0.0 2023-04-29 11:47:25,338 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-29 11:48:25,294 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109898.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:48:30,268 INFO [train.py:904] (6/8) Epoch 11, batch 8400, loss[loss=0.1832, simple_loss=0.2643, pruned_loss=0.05104, over 11917.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.291, pruned_loss=0.05864, over 3045240.85 frames. ], batch size: 248, lr: 6.13e-03, grad_scale: 8.0 2023-04-29 11:49:21,861 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109935.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:49:40,944 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9528, 1.7613, 1.5707, 1.3619, 1.7750, 1.5170, 1.7090, 1.9416], device='cuda:6'), covar=tensor([0.0120, 0.0225, 0.0323, 0.0304, 0.0163, 0.0223, 0.0150, 0.0171], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0197, 0.0193, 0.0193, 0.0195, 0.0196, 0.0197, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:49:45,332 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109950.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:49:47,548 INFO [train.py:904] (6/8) Epoch 11, batch 8450, loss[loss=0.1709, simple_loss=0.2636, pruned_loss=0.03913, over 16432.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2891, pruned_loss=0.05653, over 3053542.29 frames. ], batch size: 68, lr: 6.13e-03, grad_scale: 8.0 2023-04-29 11:49:52,354 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.195e+02 2.664e+02 3.370e+02 8.341e+02, threshold=5.327e+02, percent-clipped=3.0 2023-04-29 11:50:23,639 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109974.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:50:41,936 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1848, 1.9557, 2.1545, 3.5752, 2.0000, 2.3432, 2.1466, 2.1424], device='cuda:6'), covar=tensor([0.0878, 0.3777, 0.2428, 0.0448, 0.4210, 0.2477, 0.3243, 0.3571], device='cuda:6'), in_proj_covar=tensor([0.0349, 0.0382, 0.0320, 0.0308, 0.0402, 0.0433, 0.0344, 0.0445], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:51:00,422 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109998.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:51:09,669 INFO [train.py:904] (6/8) Epoch 11, batch 8500, loss[loss=0.1813, simple_loss=0.2737, pruned_loss=0.04446, over 16654.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2851, pruned_loss=0.0538, over 3066145.54 frames. ], batch size: 89, lr: 6.12e-03, grad_scale: 8.0 2023-04-29 11:51:12,666 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2535, 3.3468, 3.6104, 3.5770, 3.5896, 3.4260, 3.4371, 3.4885], device='cuda:6'), covar=tensor([0.0342, 0.0691, 0.0425, 0.0441, 0.0495, 0.0457, 0.0854, 0.0487], device='cuda:6'), in_proj_covar=tensor([0.0309, 0.0323, 0.0324, 0.0304, 0.0372, 0.0340, 0.0441, 0.0279], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-29 11:51:31,271 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110015.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:51:34,058 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110017.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:51:41,429 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110022.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:52:12,121 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9361, 2.0433, 2.3746, 3.1912, 2.1497, 2.2866, 2.2327, 2.1438], device='cuda:6'), covar=tensor([0.0888, 0.3406, 0.1910, 0.0516, 0.3833, 0.2323, 0.2822, 0.3269], device='cuda:6'), in_proj_covar=tensor([0.0349, 0.0381, 0.0320, 0.0308, 0.0403, 0.0432, 0.0343, 0.0444], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:52:17,165 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110044.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:52:31,121 INFO [train.py:904] (6/8) Epoch 11, batch 8550, loss[loss=0.185, simple_loss=0.2658, pruned_loss=0.05215, over 12125.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2822, pruned_loss=0.05265, over 3026166.73 frames. ], batch size: 248, lr: 6.12e-03, grad_scale: 8.0 2023-04-29 11:52:37,011 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.639e+02 2.340e+02 2.855e+02 3.457e+02 5.505e+02, threshold=5.709e+02, percent-clipped=2.0 2023-04-29 11:53:19,618 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110078.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:54:07,122 INFO [train.py:904] (6/8) Epoch 11, batch 8600, loss[loss=0.185, simple_loss=0.2815, pruned_loss=0.04423, over 16166.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2823, pruned_loss=0.05165, over 3037115.81 frames. ], batch size: 165, lr: 6.12e-03, grad_scale: 8.0 2023-04-29 11:54:14,581 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110105.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:54:23,193 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110110.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:54:35,014 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9323, 4.2195, 4.0089, 4.0472, 3.6958, 3.7859, 3.8795, 4.1919], device='cuda:6'), covar=tensor([0.0991, 0.0969, 0.1016, 0.0682, 0.0867, 0.1583, 0.0966, 0.0968], device='cuda:6'), in_proj_covar=tensor([0.0519, 0.0637, 0.0531, 0.0445, 0.0400, 0.0422, 0.0534, 0.0490], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:54:35,127 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110115.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:54:47,946 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8865, 5.1875, 5.0010, 4.9431, 4.6220, 4.6299, 4.6229, 5.2815], device='cuda:6'), covar=tensor([0.0925, 0.0785, 0.0951, 0.0607, 0.0758, 0.0857, 0.1001, 0.0689], device='cuda:6'), in_proj_covar=tensor([0.0520, 0.0638, 0.0532, 0.0445, 0.0400, 0.0423, 0.0535, 0.0490], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:55:43,443 INFO [train.py:904] (6/8) Epoch 11, batch 8650, loss[loss=0.1822, simple_loss=0.2645, pruned_loss=0.04998, over 12148.00 frames. ], tot_loss[loss=0.191, simple_loss=0.2807, pruned_loss=0.05063, over 3030858.47 frames. ], batch size: 246, lr: 6.12e-03, grad_scale: 4.0 2023-04-29 11:55:53,867 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.622e+02 2.527e+02 3.196e+02 4.321e+02 7.577e+02, threshold=6.393e+02, percent-clipped=5.0 2023-04-29 11:56:16,580 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6815, 4.0228, 3.5742, 3.9608, 3.4966, 3.6327, 3.5607, 4.0034], device='cuda:6'), covar=tensor([0.2521, 0.1943, 0.3419, 0.1266, 0.2115, 0.3069, 0.2622, 0.2084], device='cuda:6'), in_proj_covar=tensor([0.0521, 0.0642, 0.0533, 0.0447, 0.0401, 0.0424, 0.0537, 0.0492], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 11:56:28,754 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110171.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:56:40,327 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110176.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:57:21,112 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0537, 3.2287, 3.1518, 2.2258, 2.9205, 3.2264, 3.1010, 2.0488], device='cuda:6'), covar=tensor([0.0397, 0.0034, 0.0036, 0.0295, 0.0077, 0.0052, 0.0053, 0.0344], device='cuda:6'), in_proj_covar=tensor([0.0125, 0.0066, 0.0067, 0.0122, 0.0076, 0.0086, 0.0075, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 11:57:30,807 INFO [train.py:904] (6/8) Epoch 11, batch 8700, loss[loss=0.1767, simple_loss=0.2726, pruned_loss=0.04039, over 16747.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.2775, pruned_loss=0.04892, over 3038874.44 frames. ], batch size: 83, lr: 6.12e-03, grad_scale: 4.0 2023-04-29 11:57:58,306 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5578, 2.7349, 2.4232, 3.7818, 2.4525, 3.8645, 1.3760, 2.8424], device='cuda:6'), covar=tensor([0.1710, 0.0702, 0.1239, 0.0170, 0.0152, 0.0408, 0.1912, 0.0781], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0156, 0.0178, 0.0140, 0.0193, 0.0205, 0.0179, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 11:58:34,011 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110235.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:59:05,968 INFO [train.py:904] (6/8) Epoch 11, batch 8750, loss[loss=0.2007, simple_loss=0.2908, pruned_loss=0.05529, over 16477.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.2778, pruned_loss=0.04876, over 3040122.82 frames. ], batch size: 35, lr: 6.12e-03, grad_scale: 4.0 2023-04-29 11:59:15,715 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.697e+02 2.315e+02 2.719e+02 3.353e+02 7.427e+02, threshold=5.437e+02, percent-clipped=1.0 2023-04-29 12:00:22,511 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110283.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:00:22,687 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110283.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:01:00,290 INFO [train.py:904] (6/8) Epoch 11, batch 8800, loss[loss=0.1898, simple_loss=0.271, pruned_loss=0.05427, over 12296.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2761, pruned_loss=0.04762, over 3038530.01 frames. ], batch size: 248, lr: 6.12e-03, grad_scale: 8.0 2023-04-29 12:01:03,783 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110303.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:01:27,805 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110315.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:01:34,228 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110318.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:02:29,399 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110344.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:02:45,831 INFO [train.py:904] (6/8) Epoch 11, batch 8850, loss[loss=0.181, simple_loss=0.2684, pruned_loss=0.04683, over 12446.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2792, pruned_loss=0.04726, over 3046917.92 frames. ], batch size: 246, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:02:52,480 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.350e+02 2.510e+02 2.922e+02 3.907e+02 6.547e+02, threshold=5.844e+02, percent-clipped=7.0 2023-04-29 12:03:09,621 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110363.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:03:11,346 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110364.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:03:29,842 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110373.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:03:42,460 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110379.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:04:27,923 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110400.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:04:30,860 INFO [train.py:904] (6/8) Epoch 11, batch 8900, loss[loss=0.1765, simple_loss=0.2646, pruned_loss=0.04424, over 12672.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2788, pruned_loss=0.04616, over 3037484.97 frames. ], batch size: 250, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:05:18,768 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6194, 3.2655, 3.0578, 1.8755, 2.7072, 2.2264, 3.1950, 3.3263], device='cuda:6'), covar=tensor([0.0305, 0.0598, 0.0666, 0.1910, 0.0829, 0.0943, 0.0720, 0.0759], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0137, 0.0154, 0.0140, 0.0134, 0.0123, 0.0133, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 12:06:21,730 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-04-29 12:06:36,589 INFO [train.py:904] (6/8) Epoch 11, batch 8950, loss[loss=0.159, simple_loss=0.2521, pruned_loss=0.03293, over 15243.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.2792, pruned_loss=0.0468, over 3067987.29 frames. ], batch size: 191, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:06:45,507 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.580e+02 2.155e+02 2.674e+02 3.353e+02 7.252e+02, threshold=5.349e+02, percent-clipped=1.0 2023-04-29 12:07:08,189 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110466.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:07:08,759 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-04-29 12:07:18,604 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110471.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:08:16,802 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2588, 3.3724, 3.5128, 1.5960, 3.7285, 3.8271, 2.8838, 2.7227], device='cuda:6'), covar=tensor([0.0792, 0.0205, 0.0199, 0.1296, 0.0053, 0.0094, 0.0365, 0.0474], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0097, 0.0084, 0.0136, 0.0066, 0.0097, 0.0117, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 12:08:26,529 INFO [train.py:904] (6/8) Epoch 11, batch 9000, loss[loss=0.1565, simple_loss=0.2486, pruned_loss=0.03216, over 17216.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2755, pruned_loss=0.04512, over 3067720.81 frames. ], batch size: 44, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:08:26,529 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 12:08:36,934 INFO [train.py:938] (6/8) Epoch 11, validation: loss=0.1545, simple_loss=0.2586, pruned_loss=0.02523, over 944034.00 frames. 2023-04-29 12:08:36,935 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 12:08:45,769 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2213, 4.2966, 4.4511, 4.2905, 4.3256, 4.8015, 4.4283, 4.1165], device='cuda:6'), covar=tensor([0.1362, 0.1788, 0.1737, 0.1759, 0.2377, 0.1003, 0.1412, 0.2455], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0462, 0.0506, 0.0397, 0.0525, 0.0534, 0.0402, 0.0544], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 12:10:21,905 INFO [train.py:904] (6/8) Epoch 11, batch 9050, loss[loss=0.1928, simple_loss=0.2745, pruned_loss=0.05556, over 16280.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.2767, pruned_loss=0.04606, over 3072423.08 frames. ], batch size: 165, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:10:28,904 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.635e+02 2.375e+02 3.006e+02 3.857e+02 1.104e+03, threshold=6.012e+02, percent-clipped=5.0 2023-04-29 12:12:06,401 INFO [train.py:904] (6/8) Epoch 11, batch 9100, loss[loss=0.1805, simple_loss=0.2809, pruned_loss=0.04002, over 15365.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.2763, pruned_loss=0.04624, over 3082556.46 frames. ], batch size: 191, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:12:12,289 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5210, 3.5083, 3.7312, 1.8648, 3.9153, 3.9551, 2.9936, 2.9047], device='cuda:6'), covar=tensor([0.0689, 0.0194, 0.0153, 0.1125, 0.0041, 0.0097, 0.0338, 0.0420], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0096, 0.0083, 0.0135, 0.0065, 0.0097, 0.0116, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 12:12:25,215 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110611.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:13:11,196 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7789, 4.5949, 4.7694, 4.9378, 5.1126, 4.4751, 5.1030, 5.1138], device='cuda:6'), covar=tensor([0.1457, 0.1019, 0.1445, 0.0617, 0.0453, 0.0806, 0.0388, 0.0510], device='cuda:6'), in_proj_covar=tensor([0.0493, 0.0617, 0.0731, 0.0627, 0.0480, 0.0484, 0.0494, 0.0563], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:13:35,048 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-29 12:13:36,692 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110639.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:14:06,379 INFO [train.py:904] (6/8) Epoch 11, batch 9150, loss[loss=0.191, simple_loss=0.2733, pruned_loss=0.05434, over 11965.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.276, pruned_loss=0.0458, over 3050432.21 frames. ], batch size: 247, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:14:15,980 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.599e+02 2.395e+02 2.852e+02 3.560e+02 7.952e+02, threshold=5.704e+02, percent-clipped=1.0 2023-04-29 12:14:23,430 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110659.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:14:52,236 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110672.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:14:54,527 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110673.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:14:56,542 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110674.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:15:03,028 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 12:15:46,088 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110700.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:15:48,928 INFO [train.py:904] (6/8) Epoch 11, batch 9200, loss[loss=0.173, simple_loss=0.2661, pruned_loss=0.03994, over 15436.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.272, pruned_loss=0.04476, over 3065802.10 frames. ], batch size: 191, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:16:24,210 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110721.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:16:52,954 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110736.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:17:18,208 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110748.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:17:24,356 INFO [train.py:904] (6/8) Epoch 11, batch 9250, loss[loss=0.1828, simple_loss=0.257, pruned_loss=0.05429, over 12123.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.2716, pruned_loss=0.04499, over 3038908.05 frames. ], batch size: 249, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:17:32,955 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.609e+02 2.583e+02 3.099e+02 3.621e+02 8.759e+02, threshold=6.198e+02, percent-clipped=4.0 2023-04-29 12:17:53,826 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110766.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:18:05,244 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110771.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:18:44,461 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1102, 1.3853, 1.8167, 2.0686, 2.1353, 2.2044, 1.7557, 2.2346], device='cuda:6'), covar=tensor([0.0179, 0.0344, 0.0213, 0.0240, 0.0229, 0.0159, 0.0325, 0.0112], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0166, 0.0147, 0.0152, 0.0162, 0.0118, 0.0166, 0.0109], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 12:19:06,553 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110797.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 12:19:16,916 INFO [train.py:904] (6/8) Epoch 11, batch 9300, loss[loss=0.1579, simple_loss=0.2414, pruned_loss=0.03718, over 12052.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2695, pruned_loss=0.04416, over 3028741.84 frames. ], batch size: 246, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:19:45,794 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110814.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:19:59,850 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110819.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:21:01,169 INFO [train.py:904] (6/8) Epoch 11, batch 9350, loss[loss=0.2194, simple_loss=0.305, pruned_loss=0.06694, over 15177.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2692, pruned_loss=0.04394, over 3050830.46 frames. ], batch size: 190, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:21:10,063 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.415e+02 2.201e+02 2.572e+02 3.084e+02 6.751e+02, threshold=5.144e+02, percent-clipped=1.0 2023-04-29 12:21:11,501 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-29 12:21:36,193 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110868.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:21:44,931 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8370, 3.8021, 3.9656, 3.7421, 3.8936, 4.3389, 4.0092, 3.7027], device='cuda:6'), covar=tensor([0.2109, 0.2012, 0.1854, 0.2117, 0.2598, 0.1298, 0.1456, 0.2518], device='cuda:6'), in_proj_covar=tensor([0.0328, 0.0463, 0.0508, 0.0401, 0.0528, 0.0540, 0.0405, 0.0541], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 12:22:31,176 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6480, 4.9533, 4.7678, 4.7429, 4.4422, 4.3640, 4.4749, 5.0291], device='cuda:6'), covar=tensor([0.1083, 0.0953, 0.1045, 0.0657, 0.0842, 0.1031, 0.0925, 0.0829], device='cuda:6'), in_proj_covar=tensor([0.0507, 0.0629, 0.0517, 0.0437, 0.0394, 0.0413, 0.0526, 0.0480], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:22:41,698 INFO [train.py:904] (6/8) Epoch 11, batch 9400, loss[loss=0.1681, simple_loss=0.2587, pruned_loss=0.03874, over 12479.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2688, pruned_loss=0.04359, over 3047101.23 frames. ], batch size: 248, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:23:11,240 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-29 12:23:36,574 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110929.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:23:57,883 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110939.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:24:21,277 INFO [train.py:904] (6/8) Epoch 11, batch 9450, loss[loss=0.17, simple_loss=0.2598, pruned_loss=0.04007, over 16881.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2714, pruned_loss=0.0441, over 3044179.18 frames. ], batch size: 116, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:24:27,346 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.422e+02 3.077e+02 3.855e+02 7.743e+02, threshold=6.155e+02, percent-clipped=6.0 2023-04-29 12:24:33,087 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110958.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:24:34,543 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110959.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:24:50,469 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110967.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:25:06,693 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110974.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:25:18,508 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6784, 4.9340, 4.7511, 4.7361, 4.4311, 4.3763, 4.3726, 5.0077], device='cuda:6'), covar=tensor([0.0951, 0.0852, 0.0862, 0.0583, 0.0703, 0.0969, 0.0903, 0.0791], device='cuda:6'), in_proj_covar=tensor([0.0514, 0.0638, 0.0523, 0.0442, 0.0400, 0.0418, 0.0531, 0.0485], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:25:33,576 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110987.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:26:00,931 INFO [train.py:904] (6/8) Epoch 11, batch 9500, loss[loss=0.1704, simple_loss=0.2684, pruned_loss=0.03619, over 16502.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2702, pruned_loss=0.04333, over 3054767.03 frames. ], batch size: 68, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:26:15,566 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111007.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:26:38,128 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111019.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:26:44,355 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111022.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:27:27,039 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1938, 4.2427, 4.6312, 4.6149, 4.6317, 4.3282, 4.3483, 4.1698], device='cuda:6'), covar=tensor([0.0289, 0.0662, 0.0398, 0.0392, 0.0453, 0.0374, 0.0832, 0.0400], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0313, 0.0316, 0.0295, 0.0357, 0.0333, 0.0424, 0.0268], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-29 12:27:46,818 INFO [train.py:904] (6/8) Epoch 11, batch 9550, loss[loss=0.1903, simple_loss=0.2766, pruned_loss=0.05203, over 12299.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2703, pruned_loss=0.04379, over 3050149.41 frames. ], batch size: 246, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:27:55,298 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.451e+02 2.456e+02 2.878e+02 3.283e+02 5.727e+02, threshold=5.755e+02, percent-clipped=0.0 2023-04-29 12:28:26,865 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4873, 3.7312, 2.0775, 4.0025, 2.5628, 3.9351, 2.1758, 2.8590], device='cuda:6'), covar=tensor([0.0212, 0.0265, 0.1550, 0.0169, 0.0836, 0.0473, 0.1492, 0.0683], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0154, 0.0181, 0.0120, 0.0161, 0.0192, 0.0191, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-29 12:29:10,425 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111092.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 12:29:26,768 INFO [train.py:904] (6/8) Epoch 11, batch 9600, loss[loss=0.1888, simple_loss=0.2693, pruned_loss=0.05412, over 12414.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.271, pruned_loss=0.04429, over 3040991.89 frames. ], batch size: 248, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:30:14,523 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7015, 2.0070, 1.6333, 1.7752, 2.4012, 2.0928, 2.4705, 2.5983], device='cuda:6'), covar=tensor([0.0102, 0.0358, 0.0454, 0.0409, 0.0236, 0.0316, 0.0143, 0.0197], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0201, 0.0195, 0.0195, 0.0198, 0.0198, 0.0195, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:31:14,962 INFO [train.py:904] (6/8) Epoch 11, batch 9650, loss[loss=0.1723, simple_loss=0.2664, pruned_loss=0.03913, over 16563.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2737, pruned_loss=0.04497, over 3028623.77 frames. ], batch size: 62, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:31:24,135 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.791e+02 2.367e+02 2.756e+02 3.328e+02 5.495e+02, threshold=5.512e+02, percent-clipped=0.0 2023-04-29 12:31:33,801 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9663, 4.9202, 4.8090, 4.4603, 4.4715, 4.8408, 4.8026, 4.5110], device='cuda:6'), covar=tensor([0.0583, 0.0552, 0.0271, 0.0281, 0.0990, 0.0379, 0.0298, 0.0651], device='cuda:6'), in_proj_covar=tensor([0.0227, 0.0293, 0.0268, 0.0250, 0.0286, 0.0289, 0.0187, 0.0311], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:33:03,901 INFO [train.py:904] (6/8) Epoch 11, batch 9700, loss[loss=0.1833, simple_loss=0.2803, pruned_loss=0.04313, over 15373.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2731, pruned_loss=0.04508, over 3026399.73 frames. ], batch size: 191, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:33:13,932 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4432, 2.0639, 2.0316, 3.9912, 2.0343, 2.4915, 2.1641, 2.2271], device='cuda:6'), covar=tensor([0.0849, 0.3359, 0.2571, 0.0358, 0.4033, 0.2210, 0.3082, 0.3292], device='cuda:6'), in_proj_covar=tensor([0.0349, 0.0375, 0.0322, 0.0306, 0.0401, 0.0424, 0.0341, 0.0438], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:33:19,416 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6036, 4.5190, 5.0186, 4.9663, 5.0158, 4.7230, 4.7091, 4.4709], device='cuda:6'), covar=tensor([0.0264, 0.0663, 0.0429, 0.0461, 0.0432, 0.0346, 0.0792, 0.0385], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0313, 0.0317, 0.0297, 0.0357, 0.0334, 0.0424, 0.0270], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-29 12:33:41,299 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9832, 2.8170, 2.9353, 2.0223, 2.7426, 2.1354, 2.7121, 2.8665], device='cuda:6'), covar=tensor([0.0316, 0.0606, 0.0389, 0.1591, 0.0608, 0.0882, 0.0620, 0.0707], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0134, 0.0153, 0.0140, 0.0133, 0.0122, 0.0133, 0.0146], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 12:33:49,084 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111224.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:33:52,461 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111225.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:34:16,438 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8779, 1.2510, 1.5998, 1.7708, 1.8671, 1.9382, 1.5318, 1.8306], device='cuda:6'), covar=tensor([0.0174, 0.0322, 0.0152, 0.0186, 0.0177, 0.0142, 0.0338, 0.0078], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0165, 0.0147, 0.0151, 0.0160, 0.0118, 0.0164, 0.0107], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 12:34:46,302 INFO [train.py:904] (6/8) Epoch 11, batch 9750, loss[loss=0.1783, simple_loss=0.2758, pruned_loss=0.04038, over 16454.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.2715, pruned_loss=0.0451, over 3031870.80 frames. ], batch size: 68, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:34:53,729 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.704e+02 2.278e+02 2.798e+02 3.645e+02 8.933e+02, threshold=5.595e+02, percent-clipped=9.0 2023-04-29 12:35:17,242 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111267.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:35:17,361 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4517, 3.3984, 3.5838, 1.6881, 3.7693, 3.8015, 3.0217, 2.8667], device='cuda:6'), covar=tensor([0.0703, 0.0197, 0.0189, 0.1205, 0.0050, 0.0123, 0.0331, 0.0393], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0097, 0.0083, 0.0136, 0.0066, 0.0098, 0.0117, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 12:35:58,526 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111286.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:36:25,548 INFO [train.py:904] (6/8) Epoch 11, batch 9800, loss[loss=0.1875, simple_loss=0.2822, pruned_loss=0.04641, over 16780.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.272, pruned_loss=0.04412, over 3049307.66 frames. ], batch size: 124, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:36:49,119 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111314.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:36:51,181 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111315.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:37:01,088 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2420, 2.0049, 2.1247, 3.7598, 2.0304, 2.3336, 2.1302, 2.1741], device='cuda:6'), covar=tensor([0.0862, 0.3410, 0.2294, 0.0427, 0.3739, 0.2347, 0.3168, 0.3230], device='cuda:6'), in_proj_covar=tensor([0.0348, 0.0375, 0.0320, 0.0306, 0.0401, 0.0424, 0.0341, 0.0438], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:38:11,446 INFO [train.py:904] (6/8) Epoch 11, batch 9850, loss[loss=0.171, simple_loss=0.2651, pruned_loss=0.03844, over 12744.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2728, pruned_loss=0.04359, over 3051251.57 frames. ], batch size: 250, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:38:20,193 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.271e+02 2.860e+02 3.428e+02 8.615e+02, threshold=5.720e+02, percent-clipped=1.0 2023-04-29 12:39:42,439 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111392.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 12:40:02,790 INFO [train.py:904] (6/8) Epoch 11, batch 9900, loss[loss=0.1742, simple_loss=0.277, pruned_loss=0.03573, over 15480.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2732, pruned_loss=0.04354, over 3042980.50 frames. ], batch size: 191, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:40:41,856 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2297, 4.2515, 4.4316, 4.2335, 4.2912, 4.7946, 4.4731, 4.2204], device='cuda:6'), covar=tensor([0.1439, 0.1901, 0.1638, 0.1996, 0.2603, 0.1117, 0.1311, 0.2442], device='cuda:6'), in_proj_covar=tensor([0.0320, 0.0452, 0.0494, 0.0388, 0.0514, 0.0527, 0.0391, 0.0527], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:40:55,087 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0928, 3.2159, 1.8334, 3.5101, 2.3280, 3.4408, 2.0132, 2.6241], device='cuda:6'), covar=tensor([0.0279, 0.0374, 0.1694, 0.0163, 0.0951, 0.0519, 0.1644, 0.0775], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0156, 0.0184, 0.0121, 0.0163, 0.0195, 0.0194, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-29 12:41:34,336 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111440.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:42:01,112 INFO [train.py:904] (6/8) Epoch 11, batch 9950, loss[loss=0.1561, simple_loss=0.2558, pruned_loss=0.02823, over 16772.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2752, pruned_loss=0.04356, over 3057018.63 frames. ], batch size: 76, lr: 6.08e-03, grad_scale: 8.0 2023-04-29 12:42:11,464 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.497e+02 2.371e+02 2.780e+02 3.480e+02 6.168e+02, threshold=5.560e+02, percent-clipped=1.0 2023-04-29 12:42:41,092 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111468.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:42:57,891 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8104, 3.7617, 3.9234, 3.7596, 3.8799, 4.2938, 3.9811, 3.6983], device='cuda:6'), covar=tensor([0.2125, 0.2421, 0.2162, 0.2416, 0.3031, 0.1817, 0.1460, 0.2734], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0454, 0.0498, 0.0392, 0.0518, 0.0531, 0.0395, 0.0530], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:44:02,030 INFO [train.py:904] (6/8) Epoch 11, batch 10000, loss[loss=0.1833, simple_loss=0.2818, pruned_loss=0.04241, over 16702.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2739, pruned_loss=0.04313, over 3080665.64 frames. ], batch size: 134, lr: 6.08e-03, grad_scale: 8.0 2023-04-29 12:44:45,110 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111524.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:44:55,683 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111529.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:45:42,134 INFO [train.py:904] (6/8) Epoch 11, batch 10050, loss[loss=0.1709, simple_loss=0.2649, pruned_loss=0.03842, over 12077.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.274, pruned_loss=0.04318, over 3070571.74 frames. ], batch size: 250, lr: 6.08e-03, grad_scale: 8.0 2023-04-29 12:45:45,752 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9045, 2.3128, 2.3209, 2.9652, 2.1320, 3.3229, 1.5995, 2.7890], device='cuda:6'), covar=tensor([0.1200, 0.0582, 0.0985, 0.0106, 0.0097, 0.0390, 0.1436, 0.0608], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0155, 0.0177, 0.0137, 0.0182, 0.0203, 0.0180, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 12:45:50,236 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.518e+02 2.160e+02 2.693e+02 3.599e+02 6.171e+02, threshold=5.385e+02, percent-clipped=1.0 2023-04-29 12:46:22,430 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111572.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:46:38,822 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111581.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:47:00,534 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6180, 4.5924, 4.3679, 3.9948, 4.4323, 1.6523, 4.2047, 4.2482], device='cuda:6'), covar=tensor([0.0092, 0.0120, 0.0188, 0.0265, 0.0144, 0.2368, 0.0135, 0.0170], device='cuda:6'), in_proj_covar=tensor([0.0124, 0.0110, 0.0157, 0.0141, 0.0128, 0.0176, 0.0144, 0.0142], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:47:14,825 INFO [train.py:904] (6/8) Epoch 11, batch 10100, loss[loss=0.174, simple_loss=0.27, pruned_loss=0.039, over 16673.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2744, pruned_loss=0.04377, over 3072239.73 frames. ], batch size: 134, lr: 6.08e-03, grad_scale: 8.0 2023-04-29 12:47:35,955 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111614.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:48:58,023 INFO [train.py:904] (6/8) Epoch 12, batch 0, loss[loss=0.2538, simple_loss=0.3337, pruned_loss=0.08692, over 16666.00 frames. ], tot_loss[loss=0.2538, simple_loss=0.3337, pruned_loss=0.08692, over 16666.00 frames. ], batch size: 62, lr: 5.82e-03, grad_scale: 8.0 2023-04-29 12:48:58,023 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 12:49:05,311 INFO [train.py:938] (6/8) Epoch 12, validation: loss=0.1543, simple_loss=0.2577, pruned_loss=0.0254, over 944034.00 frames. 2023-04-29 12:49:05,312 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 12:49:12,536 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.743e+02 2.560e+02 3.107e+02 3.982e+02 7.820e+02, threshold=6.214e+02, percent-clipped=3.0 2023-04-29 12:49:21,578 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111662.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:49:22,805 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9540, 5.3298, 5.4759, 5.3091, 5.2871, 5.8431, 5.4113, 5.1510], device='cuda:6'), covar=tensor([0.0963, 0.1743, 0.2140, 0.2006, 0.2353, 0.1032, 0.1475, 0.2445], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0454, 0.0497, 0.0394, 0.0519, 0.0532, 0.0400, 0.0527], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:50:01,813 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4393, 3.4404, 3.8018, 2.5565, 3.5286, 3.7621, 3.5833, 1.8500], device='cuda:6'), covar=tensor([0.0475, 0.0268, 0.0059, 0.0356, 0.0109, 0.0114, 0.0119, 0.0566], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0066, 0.0067, 0.0124, 0.0076, 0.0085, 0.0075, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 12:50:15,964 INFO [train.py:904] (6/8) Epoch 12, batch 50, loss[loss=0.1725, simple_loss=0.2517, pruned_loss=0.04666, over 16838.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.2891, pruned_loss=0.06507, over 744594.86 frames. ], batch size: 96, lr: 5.82e-03, grad_scale: 2.0 2023-04-29 12:51:25,695 INFO [train.py:904] (6/8) Epoch 12, batch 100, loss[loss=0.2122, simple_loss=0.3023, pruned_loss=0.06106, over 16745.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.2827, pruned_loss=0.06053, over 1318080.18 frames. ], batch size: 57, lr: 5.82e-03, grad_scale: 2.0 2023-04-29 12:51:34,342 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 2.444e+02 2.855e+02 3.643e+02 7.519e+02, threshold=5.710e+02, percent-clipped=2.0 2023-04-29 12:52:18,947 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111791.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 12:52:31,943 INFO [train.py:904] (6/8) Epoch 12, batch 150, loss[loss=0.1826, simple_loss=0.2515, pruned_loss=0.05683, over 16743.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2796, pruned_loss=0.05825, over 1762626.65 frames. ], batch size: 124, lr: 5.82e-03, grad_scale: 2.0 2023-04-29 12:53:03,083 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111824.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:53:40,719 INFO [train.py:904] (6/8) Epoch 12, batch 200, loss[loss=0.2456, simple_loss=0.3066, pruned_loss=0.09232, over 16885.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2776, pruned_loss=0.05691, over 2112603.03 frames. ], batch size: 96, lr: 5.82e-03, grad_scale: 2.0 2023-04-29 12:53:41,219 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111852.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 12:53:50,234 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.787e+02 2.566e+02 3.072e+02 3.744e+02 9.632e+02, threshold=6.144e+02, percent-clipped=5.0 2023-04-29 12:53:55,987 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.7276, 6.1324, 5.8154, 5.9439, 5.4288, 5.3488, 5.5971, 6.2318], device='cuda:6'), covar=tensor([0.1060, 0.0868, 0.1192, 0.0637, 0.0806, 0.0577, 0.0871, 0.0831], device='cuda:6'), in_proj_covar=tensor([0.0536, 0.0665, 0.0544, 0.0463, 0.0418, 0.0432, 0.0556, 0.0510], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:54:21,428 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111881.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:54:49,916 INFO [train.py:904] (6/8) Epoch 12, batch 250, loss[loss=0.216, simple_loss=0.2969, pruned_loss=0.06754, over 16502.00 frames. ], tot_loss[loss=0.1925, simple_loss=0.2748, pruned_loss=0.05512, over 2377898.67 frames. ], batch size: 146, lr: 5.82e-03, grad_scale: 2.0 2023-04-29 12:55:27,221 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111929.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:55:55,943 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4679, 3.5484, 2.1811, 3.7318, 2.7841, 3.6649, 2.2523, 2.8299], device='cuda:6'), covar=tensor([0.0241, 0.0377, 0.1379, 0.0279, 0.0673, 0.0662, 0.1231, 0.0600], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0164, 0.0190, 0.0129, 0.0169, 0.0204, 0.0199, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 12:55:57,752 INFO [train.py:904] (6/8) Epoch 12, batch 300, loss[loss=0.189, simple_loss=0.2678, pruned_loss=0.05511, over 16476.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.2715, pruned_loss=0.05342, over 2579253.12 frames. ], batch size: 146, lr: 5.82e-03, grad_scale: 1.0 2023-04-29 12:56:09,471 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.668e+02 2.376e+02 2.773e+02 3.192e+02 6.381e+02, threshold=5.545e+02, percent-clipped=1.0 2023-04-29 12:56:18,585 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-29 12:57:10,668 INFO [train.py:904] (6/8) Epoch 12, batch 350, loss[loss=0.1657, simple_loss=0.2443, pruned_loss=0.04354, over 17015.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2687, pruned_loss=0.05223, over 2750296.24 frames. ], batch size: 41, lr: 5.81e-03, grad_scale: 1.0 2023-04-29 12:57:27,984 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-29 12:57:54,758 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9401, 3.8272, 4.0510, 4.1860, 4.2467, 3.8017, 4.0569, 4.2450], device='cuda:6'), covar=tensor([0.1423, 0.1004, 0.1137, 0.0575, 0.0530, 0.1621, 0.1977, 0.0631], device='cuda:6'), in_proj_covar=tensor([0.0531, 0.0660, 0.0795, 0.0668, 0.0509, 0.0515, 0.0532, 0.0602], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 12:58:17,741 INFO [train.py:904] (6/8) Epoch 12, batch 400, loss[loss=0.1557, simple_loss=0.2438, pruned_loss=0.03384, over 17241.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2676, pruned_loss=0.0517, over 2872513.18 frames. ], batch size: 44, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 12:58:27,669 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.698e+02 2.508e+02 2.981e+02 3.635e+02 6.653e+02, threshold=5.962e+02, percent-clipped=1.0 2023-04-29 12:58:36,995 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112066.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:59:26,048 INFO [train.py:904] (6/8) Epoch 12, batch 450, loss[loss=0.1837, simple_loss=0.2766, pruned_loss=0.04535, over 17136.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2656, pruned_loss=0.0512, over 2969974.08 frames. ], batch size: 49, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 12:59:56,560 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112124.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:00:00,694 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112127.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:00:27,067 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112147.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 13:00:31,884 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7765, 3.9687, 4.3452, 1.8935, 4.5002, 4.4742, 3.0979, 3.4285], device='cuda:6'), covar=tensor([0.0744, 0.0185, 0.0127, 0.1155, 0.0060, 0.0130, 0.0413, 0.0356], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0099, 0.0086, 0.0139, 0.0069, 0.0103, 0.0120, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 13:00:33,802 INFO [train.py:904] (6/8) Epoch 12, batch 500, loss[loss=0.1451, simple_loss=0.223, pruned_loss=0.03355, over 16808.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2638, pruned_loss=0.04986, over 3051977.24 frames. ], batch size: 39, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 13:00:45,214 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.270e+02 2.759e+02 3.532e+02 6.724e+02, threshold=5.519e+02, percent-clipped=2.0 2023-04-29 13:01:02,760 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=112172.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:01:19,712 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-29 13:01:30,007 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-29 13:01:32,823 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1645, 5.1540, 5.6871, 5.6262, 5.6687, 5.3291, 5.2408, 5.1082], device='cuda:6'), covar=tensor([0.0305, 0.0483, 0.0370, 0.0478, 0.0378, 0.0305, 0.0898, 0.0359], device='cuda:6'), in_proj_covar=tensor([0.0333, 0.0346, 0.0348, 0.0328, 0.0389, 0.0367, 0.0468, 0.0298], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 13:01:44,744 INFO [train.py:904] (6/8) Epoch 12, batch 550, loss[loss=0.1876, simple_loss=0.2703, pruned_loss=0.05246, over 15932.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.2632, pruned_loss=0.04984, over 3096533.24 frames. ], batch size: 35, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 13:02:55,017 INFO [train.py:904] (6/8) Epoch 12, batch 600, loss[loss=0.1874, simple_loss=0.2521, pruned_loss=0.0614, over 16804.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2634, pruned_loss=0.05035, over 3140842.41 frames. ], batch size: 83, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 13:03:06,852 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.658e+02 2.365e+02 2.773e+02 3.420e+02 1.272e+03, threshold=5.547e+02, percent-clipped=1.0 2023-04-29 13:04:02,404 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-29 13:04:04,841 INFO [train.py:904] (6/8) Epoch 12, batch 650, loss[loss=0.1718, simple_loss=0.2453, pruned_loss=0.0492, over 16304.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.262, pruned_loss=0.05011, over 3178175.24 frames. ], batch size: 165, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 13:04:12,460 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2129, 4.0385, 4.4632, 2.0339, 4.6763, 4.6920, 3.3223, 3.5757], device='cuda:6'), covar=tensor([0.0550, 0.0166, 0.0141, 0.1117, 0.0044, 0.0095, 0.0347, 0.0315], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0098, 0.0086, 0.0139, 0.0069, 0.0104, 0.0120, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 13:05:14,232 INFO [train.py:904] (6/8) Epoch 12, batch 700, loss[loss=0.1649, simple_loss=0.2403, pruned_loss=0.04476, over 12432.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2619, pruned_loss=0.04948, over 3204588.02 frames. ], batch size: 246, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 13:05:26,018 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.678e+02 2.509e+02 2.916e+02 3.539e+02 5.225e+02, threshold=5.832e+02, percent-clipped=0.0 2023-04-29 13:05:30,165 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0410, 4.6982, 4.6498, 5.2138, 5.4228, 4.8261, 5.3059, 5.3038], device='cuda:6'), covar=tensor([0.1493, 0.1286, 0.2733, 0.0861, 0.0720, 0.0925, 0.0832, 0.1028], device='cuda:6'), in_proj_covar=tensor([0.0544, 0.0677, 0.0820, 0.0683, 0.0521, 0.0527, 0.0542, 0.0619], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:06:07,768 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112389.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:06:24,418 INFO [train.py:904] (6/8) Epoch 12, batch 750, loss[loss=0.1822, simple_loss=0.2626, pruned_loss=0.05096, over 16492.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.262, pruned_loss=0.0493, over 3229264.84 frames. ], batch size: 75, lr: 5.80e-03, grad_scale: 2.0 2023-04-29 13:06:52,550 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112422.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:07:18,368 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3848, 5.7576, 5.5003, 5.5872, 5.2014, 5.0740, 5.1940, 5.8571], device='cuda:6'), covar=tensor([0.1290, 0.0956, 0.1035, 0.0674, 0.0872, 0.0734, 0.1014, 0.0971], device='cuda:6'), in_proj_covar=tensor([0.0565, 0.0701, 0.0572, 0.0488, 0.0440, 0.0450, 0.0586, 0.0537], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:07:20,867 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 13:07:27,213 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9378, 5.1811, 5.3981, 5.2308, 5.1583, 5.8216, 5.3522, 5.1006], device='cuda:6'), covar=tensor([0.1046, 0.1915, 0.2271, 0.2260, 0.3313, 0.1168, 0.1555, 0.2477], device='cuda:6'), in_proj_covar=tensor([0.0355, 0.0508, 0.0557, 0.0441, 0.0588, 0.0584, 0.0437, 0.0588], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 13:07:29,145 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112447.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 13:07:32,733 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112450.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:07:34,635 INFO [train.py:904] (6/8) Epoch 12, batch 800, loss[loss=0.1736, simple_loss=0.266, pruned_loss=0.04065, over 17029.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2622, pruned_loss=0.049, over 3251826.87 frames. ], batch size: 50, lr: 5.80e-03, grad_scale: 4.0 2023-04-29 13:07:45,050 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.661e+02 2.488e+02 3.052e+02 3.720e+02 1.064e+03, threshold=6.105e+02, percent-clipped=2.0 2023-04-29 13:08:33,906 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=112495.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 13:08:42,891 INFO [train.py:904] (6/8) Epoch 12, batch 850, loss[loss=0.1554, simple_loss=0.2396, pruned_loss=0.03563, over 16958.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2614, pruned_loss=0.04914, over 3262097.24 frames. ], batch size: 41, lr: 5.80e-03, grad_scale: 4.0 2023-04-29 13:09:51,999 INFO [train.py:904] (6/8) Epoch 12, batch 900, loss[loss=0.1816, simple_loss=0.2632, pruned_loss=0.04994, over 16562.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.261, pruned_loss=0.0492, over 3276477.23 frames. ], batch size: 68, lr: 5.80e-03, grad_scale: 4.0 2023-04-29 13:09:53,399 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-29 13:10:02,361 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.529e+02 2.271e+02 2.835e+02 3.426e+02 5.348e+02, threshold=5.671e+02, percent-clipped=0.0 2023-04-29 13:10:05,739 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112562.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:10:26,136 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112577.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:10:59,465 INFO [train.py:904] (6/8) Epoch 12, batch 950, loss[loss=0.1729, simple_loss=0.2589, pruned_loss=0.04345, over 17213.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2606, pruned_loss=0.04881, over 3287177.44 frames. ], batch size: 45, lr: 5.80e-03, grad_scale: 4.0 2023-04-29 13:11:01,690 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3710, 5.2206, 5.1601, 4.7232, 4.7643, 5.2478, 5.2278, 4.8232], device='cuda:6'), covar=tensor([0.0520, 0.0342, 0.0285, 0.0304, 0.1113, 0.0321, 0.0270, 0.0733], device='cuda:6'), in_proj_covar=tensor([0.0259, 0.0336, 0.0307, 0.0286, 0.0330, 0.0331, 0.0211, 0.0357], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:11:21,361 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112617.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:11:28,351 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112623.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:11:50,176 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112638.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 13:12:03,609 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5793, 4.9283, 4.6822, 4.6809, 4.4037, 4.3730, 4.4160, 4.9477], device='cuda:6'), covar=tensor([0.1058, 0.0874, 0.0968, 0.0681, 0.0857, 0.1195, 0.0945, 0.0908], device='cuda:6'), in_proj_covar=tensor([0.0568, 0.0706, 0.0575, 0.0490, 0.0441, 0.0454, 0.0590, 0.0540], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:12:07,851 INFO [train.py:904] (6/8) Epoch 12, batch 1000, loss[loss=0.1673, simple_loss=0.2422, pruned_loss=0.04622, over 16314.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2593, pruned_loss=0.04834, over 3305110.23 frames. ], batch size: 165, lr: 5.80e-03, grad_scale: 4.0 2023-04-29 13:12:18,367 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.238e+02 2.624e+02 3.056e+02 7.407e+02, threshold=5.248e+02, percent-clipped=1.0 2023-04-29 13:12:43,259 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112678.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 13:13:15,635 INFO [train.py:904] (6/8) Epoch 12, batch 1050, loss[loss=0.1652, simple_loss=0.2638, pruned_loss=0.03326, over 17270.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2596, pruned_loss=0.04829, over 3312157.19 frames. ], batch size: 52, lr: 5.80e-03, grad_scale: 4.0 2023-04-29 13:13:42,902 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112722.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:14:14,519 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112745.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:14:23,337 INFO [train.py:904] (6/8) Epoch 12, batch 1100, loss[loss=0.1803, simple_loss=0.274, pruned_loss=0.04328, over 17046.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2593, pruned_loss=0.04828, over 3308134.34 frames. ], batch size: 50, lr: 5.79e-03, grad_scale: 4.0 2023-04-29 13:14:34,069 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.374e+02 2.893e+02 3.790e+02 9.509e+02, threshold=5.785e+02, percent-clipped=6.0 2023-04-29 13:14:48,741 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=112770.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:15:15,204 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 13:15:33,615 INFO [train.py:904] (6/8) Epoch 12, batch 1150, loss[loss=0.1453, simple_loss=0.2323, pruned_loss=0.02912, over 17234.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2592, pruned_loss=0.04797, over 3295544.68 frames. ], batch size: 44, lr: 5.79e-03, grad_scale: 4.0 2023-04-29 13:16:42,796 INFO [train.py:904] (6/8) Epoch 12, batch 1200, loss[loss=0.1668, simple_loss=0.248, pruned_loss=0.04276, over 17221.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2584, pruned_loss=0.0474, over 3300814.65 frames. ], batch size: 45, lr: 5.79e-03, grad_scale: 8.0 2023-04-29 13:16:44,333 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6676, 2.9913, 3.0000, 2.0097, 2.6985, 2.2538, 3.0854, 3.1732], device='cuda:6'), covar=tensor([0.0383, 0.0865, 0.0561, 0.1696, 0.0876, 0.0918, 0.0750, 0.0977], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0144, 0.0158, 0.0143, 0.0137, 0.0124, 0.0137, 0.0156], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 13:16:52,675 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.635e+02 2.211e+02 2.743e+02 3.250e+02 7.821e+02, threshold=5.486e+02, percent-clipped=2.0 2023-04-29 13:17:49,331 INFO [train.py:904] (6/8) Epoch 12, batch 1250, loss[loss=0.1599, simple_loss=0.2532, pruned_loss=0.03325, over 17235.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2595, pruned_loss=0.04814, over 3308483.09 frames. ], batch size: 45, lr: 5.79e-03, grad_scale: 8.0 2023-04-29 13:18:12,292 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112918.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:18:16,598 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112921.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:18:32,350 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112933.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 13:18:57,920 INFO [train.py:904] (6/8) Epoch 12, batch 1300, loss[loss=0.1875, simple_loss=0.2689, pruned_loss=0.05306, over 16696.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2588, pruned_loss=0.04766, over 3317036.36 frames. ], batch size: 124, lr: 5.79e-03, grad_scale: 8.0 2023-04-29 13:19:09,586 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.880e+02 2.412e+02 2.831e+02 3.329e+02 6.613e+02, threshold=5.661e+02, percent-clipped=2.0 2023-04-29 13:19:27,111 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112973.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 13:19:41,097 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112982.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:20:04,773 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 13:20:08,548 INFO [train.py:904] (6/8) Epoch 12, batch 1350, loss[loss=0.1612, simple_loss=0.2524, pruned_loss=0.03498, over 17236.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2584, pruned_loss=0.04759, over 3322429.10 frames. ], batch size: 45, lr: 5.79e-03, grad_scale: 8.0 2023-04-29 13:20:15,875 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4818, 3.5512, 1.9233, 3.7024, 2.7104, 3.6994, 1.9037, 2.7456], device='cuda:6'), covar=tensor([0.0212, 0.0282, 0.1580, 0.0275, 0.0640, 0.0435, 0.1715, 0.0633], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0168, 0.0191, 0.0137, 0.0169, 0.0211, 0.0200, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 13:21:07,998 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113045.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:21:16,368 INFO [train.py:904] (6/8) Epoch 12, batch 1400, loss[loss=0.1768, simple_loss=0.2631, pruned_loss=0.04521, over 17072.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2589, pruned_loss=0.04744, over 3334086.36 frames. ], batch size: 55, lr: 5.79e-03, grad_scale: 8.0 2023-04-29 13:21:26,438 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.826e+02 2.440e+02 3.018e+02 3.818e+02 8.239e+02, threshold=6.035e+02, percent-clipped=8.0 2023-04-29 13:22:13,293 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=113093.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:22:24,189 INFO [train.py:904] (6/8) Epoch 12, batch 1450, loss[loss=0.151, simple_loss=0.235, pruned_loss=0.03348, over 16971.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2574, pruned_loss=0.04735, over 3325336.64 frames. ], batch size: 41, lr: 5.79e-03, grad_scale: 8.0 2023-04-29 13:23:25,160 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9192, 4.5398, 3.3843, 2.3777, 2.9457, 2.5343, 4.7766, 3.8863], device='cuda:6'), covar=tensor([0.2700, 0.0588, 0.1525, 0.2488, 0.2811, 0.1953, 0.0386, 0.1241], device='cuda:6'), in_proj_covar=tensor([0.0304, 0.0260, 0.0284, 0.0279, 0.0279, 0.0225, 0.0268, 0.0301], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:23:35,102 INFO [train.py:904] (6/8) Epoch 12, batch 1500, loss[loss=0.1841, simple_loss=0.2543, pruned_loss=0.05697, over 16929.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2574, pruned_loss=0.04781, over 3327590.54 frames. ], batch size: 109, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:23:45,775 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.499e+02 2.892e+02 3.478e+02 9.992e+02, threshold=5.783e+02, percent-clipped=3.0 2023-04-29 13:23:50,288 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4702, 4.4005, 4.4443, 3.3730, 4.4038, 1.6050, 4.1150, 4.0580], device='cuda:6'), covar=tensor([0.0192, 0.0138, 0.0179, 0.0611, 0.0133, 0.3104, 0.0219, 0.0282], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0121, 0.0171, 0.0158, 0.0141, 0.0185, 0.0159, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:24:43,434 INFO [train.py:904] (6/8) Epoch 12, batch 1550, loss[loss=0.2276, simple_loss=0.2791, pruned_loss=0.08806, over 16800.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2582, pruned_loss=0.0483, over 3322799.41 frames. ], batch size: 102, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:25:06,570 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113218.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:25:26,597 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-29 13:25:27,375 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113233.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:25:29,502 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113234.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:25:54,129 INFO [train.py:904] (6/8) Epoch 12, batch 1600, loss[loss=0.1669, simple_loss=0.2521, pruned_loss=0.04087, over 16991.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2607, pruned_loss=0.04944, over 3328381.40 frames. ], batch size: 41, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:26:04,701 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.814e+02 2.404e+02 2.815e+02 3.461e+02 5.184e+02, threshold=5.631e+02, percent-clipped=0.0 2023-04-29 13:26:12,640 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=113266.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:26:23,366 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113273.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 13:26:26,967 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3175, 5.3046, 5.1043, 4.5196, 5.1052, 2.1763, 4.8696, 5.1178], device='cuda:6'), covar=tensor([0.0062, 0.0048, 0.0139, 0.0321, 0.0079, 0.2029, 0.0115, 0.0140], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0121, 0.0170, 0.0159, 0.0141, 0.0184, 0.0158, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:26:28,588 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113277.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:26:33,986 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=113281.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:26:53,667 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113295.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:27:02,727 INFO [train.py:904] (6/8) Epoch 12, batch 1650, loss[loss=0.2237, simple_loss=0.2998, pruned_loss=0.07378, over 12421.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2627, pruned_loss=0.04984, over 3320841.01 frames. ], batch size: 247, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:27:29,760 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=113321.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:27:52,758 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3552, 2.1312, 2.2850, 4.0026, 2.1110, 2.5659, 2.2050, 2.3735], device='cuda:6'), covar=tensor([0.1042, 0.3320, 0.2259, 0.0452, 0.3392, 0.2180, 0.3139, 0.2645], device='cuda:6'), in_proj_covar=tensor([0.0367, 0.0396, 0.0335, 0.0323, 0.0412, 0.0453, 0.0359, 0.0464], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:28:12,390 INFO [train.py:904] (6/8) Epoch 12, batch 1700, loss[loss=0.1657, simple_loss=0.2534, pruned_loss=0.03893, over 17187.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.2641, pruned_loss=0.05041, over 3324040.89 frames. ], batch size: 46, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:28:23,615 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 2.537e+02 3.070e+02 3.736e+02 6.116e+02, threshold=6.140e+02, percent-clipped=1.0 2023-04-29 13:28:45,443 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.38 vs. limit=5.0 2023-04-29 13:29:22,277 INFO [train.py:904] (6/8) Epoch 12, batch 1750, loss[loss=0.1726, simple_loss=0.2573, pruned_loss=0.04391, over 15924.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2652, pruned_loss=0.0506, over 3324390.78 frames. ], batch size: 35, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:30:32,313 INFO [train.py:904] (6/8) Epoch 12, batch 1800, loss[loss=0.1976, simple_loss=0.273, pruned_loss=0.06115, over 16763.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2666, pruned_loss=0.05059, over 3326491.34 frames. ], batch size: 124, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:30:43,442 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.362e+02 2.904e+02 3.594e+02 5.616e+02, threshold=5.809e+02, percent-clipped=0.0 2023-04-29 13:31:03,069 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8812, 4.8378, 4.6944, 4.2266, 4.7394, 1.8513, 4.4911, 4.5168], device='cuda:6'), covar=tensor([0.0095, 0.0089, 0.0150, 0.0336, 0.0100, 0.2478, 0.0136, 0.0197], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0124, 0.0173, 0.0162, 0.0144, 0.0188, 0.0161, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:31:42,916 INFO [train.py:904] (6/8) Epoch 12, batch 1850, loss[loss=0.1634, simple_loss=0.2545, pruned_loss=0.0362, over 17232.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2674, pruned_loss=0.05058, over 3313824.63 frames. ], batch size: 45, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:32:53,253 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-29 13:32:53,584 INFO [train.py:904] (6/8) Epoch 12, batch 1900, loss[loss=0.1639, simple_loss=0.2453, pruned_loss=0.04127, over 16796.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2663, pruned_loss=0.04979, over 3319696.64 frames. ], batch size: 102, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:33:04,793 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 2.197e+02 2.521e+02 3.135e+02 9.394e+02, threshold=5.041e+02, percent-clipped=1.0 2023-04-29 13:33:29,758 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113577.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:33:48,969 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113590.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:34:05,625 INFO [train.py:904] (6/8) Epoch 12, batch 1950, loss[loss=0.163, simple_loss=0.2513, pruned_loss=0.03731, over 17225.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2659, pruned_loss=0.04911, over 3315344.84 frames. ], batch size: 45, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:34:27,379 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113617.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:34:37,525 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7544, 2.6458, 2.2178, 2.4786, 2.9341, 2.7729, 3.4070, 3.2727], device='cuda:6'), covar=tensor([0.0075, 0.0284, 0.0374, 0.0334, 0.0202, 0.0283, 0.0206, 0.0169], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0213, 0.0206, 0.0205, 0.0214, 0.0212, 0.0221, 0.0203], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:34:39,273 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=113625.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:35:08,099 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113646.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:35:16,103 INFO [train.py:904] (6/8) Epoch 12, batch 2000, loss[loss=0.1708, simple_loss=0.263, pruned_loss=0.03933, over 17209.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.2653, pruned_loss=0.04879, over 3325663.43 frames. ], batch size: 46, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:35:27,899 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.257e+02 2.755e+02 3.569e+02 6.259e+02, threshold=5.509e+02, percent-clipped=3.0 2023-04-29 13:35:52,986 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113678.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:36:25,541 INFO [train.py:904] (6/8) Epoch 12, batch 2050, loss[loss=0.1632, simple_loss=0.2525, pruned_loss=0.03694, over 17174.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2662, pruned_loss=0.04972, over 3319792.76 frames. ], batch size: 46, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:36:31,452 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0326, 4.5789, 3.2987, 2.3175, 3.0146, 2.4423, 4.8945, 3.9156], device='cuda:6'), covar=tensor([0.2363, 0.0511, 0.1510, 0.2370, 0.2652, 0.1989, 0.0296, 0.1064], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0259, 0.0286, 0.0281, 0.0281, 0.0225, 0.0269, 0.0302], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:36:32,490 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113707.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:37:33,719 INFO [train.py:904] (6/8) Epoch 12, batch 2100, loss[loss=0.1855, simple_loss=0.2783, pruned_loss=0.04639, over 17293.00 frames. ], tot_loss[loss=0.1841, simple_loss=0.2675, pruned_loss=0.0504, over 3327979.30 frames. ], batch size: 52, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:37:45,422 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.512e+02 2.602e+02 3.022e+02 3.669e+02 5.748e+02, threshold=6.044e+02, percent-clipped=1.0 2023-04-29 13:38:30,001 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0280, 4.9191, 4.9015, 4.5714, 4.5326, 4.9433, 4.8509, 4.5897], device='cuda:6'), covar=tensor([0.0612, 0.0498, 0.0276, 0.0271, 0.0978, 0.0378, 0.0375, 0.0707], device='cuda:6'), in_proj_covar=tensor([0.0268, 0.0348, 0.0316, 0.0294, 0.0337, 0.0340, 0.0215, 0.0369], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 13:38:39,696 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9752, 1.9589, 2.4276, 2.9533, 2.8904, 3.0973, 2.0629, 3.1445], device='cuda:6'), covar=tensor([0.0145, 0.0337, 0.0230, 0.0172, 0.0186, 0.0154, 0.0343, 0.0091], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0177, 0.0158, 0.0163, 0.0173, 0.0130, 0.0176, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 13:38:44,790 INFO [train.py:904] (6/8) Epoch 12, batch 2150, loss[loss=0.1904, simple_loss=0.2649, pruned_loss=0.05793, over 16417.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2682, pruned_loss=0.05122, over 3319975.37 frames. ], batch size: 146, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:39:54,111 INFO [train.py:904] (6/8) Epoch 12, batch 2200, loss[loss=0.1986, simple_loss=0.2721, pruned_loss=0.06254, over 16735.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.268, pruned_loss=0.05086, over 3319998.94 frames. ], batch size: 83, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:40:02,371 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8612, 4.0812, 4.2350, 3.0930, 3.6259, 4.0589, 3.7926, 2.6510], device='cuda:6'), covar=tensor([0.0355, 0.0079, 0.0034, 0.0270, 0.0080, 0.0088, 0.0070, 0.0309], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0072, 0.0070, 0.0125, 0.0080, 0.0090, 0.0079, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 13:40:05,118 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.389e+02 2.779e+02 3.462e+02 6.586e+02, threshold=5.558e+02, percent-clipped=1.0 2023-04-29 13:40:05,632 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1789, 4.4624, 4.4824, 3.3785, 3.7834, 4.3049, 3.9333, 2.7148], device='cuda:6'), covar=tensor([0.0331, 0.0059, 0.0033, 0.0243, 0.0089, 0.0072, 0.0069, 0.0332], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0072, 0.0071, 0.0125, 0.0080, 0.0090, 0.0079, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 13:40:48,888 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113890.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:41:03,694 INFO [train.py:904] (6/8) Epoch 12, batch 2250, loss[loss=0.1811, simple_loss=0.261, pruned_loss=0.05061, over 16879.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2687, pruned_loss=0.05098, over 3329093.67 frames. ], batch size: 90, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:41:07,969 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-04-29 13:41:54,643 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=113938.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:42:14,190 INFO [train.py:904] (6/8) Epoch 12, batch 2300, loss[loss=0.1736, simple_loss=0.2557, pruned_loss=0.04577, over 16765.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2688, pruned_loss=0.05114, over 3322689.49 frames. ], batch size: 83, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:42:24,217 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.413e+02 2.819e+02 3.398e+02 7.599e+02, threshold=5.637e+02, percent-clipped=2.0 2023-04-29 13:42:43,128 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113973.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:43:25,573 INFO [train.py:904] (6/8) Epoch 12, batch 2350, loss[loss=0.1661, simple_loss=0.2428, pruned_loss=0.04472, over 16761.00 frames. ], tot_loss[loss=0.1869, simple_loss=0.2695, pruned_loss=0.0521, over 3313611.72 frames. ], batch size: 39, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:43:25,843 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=114002.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:44:35,987 INFO [train.py:904] (6/8) Epoch 12, batch 2400, loss[loss=0.2256, simple_loss=0.3028, pruned_loss=0.07415, over 15525.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2706, pruned_loss=0.05226, over 3311256.70 frames. ], batch size: 190, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:44:48,485 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.062e+02 2.572e+02 3.152e+02 4.032e+02 9.919e+02, threshold=6.305e+02, percent-clipped=6.0 2023-04-29 13:45:41,863 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1460, 1.4536, 1.7762, 1.9767, 2.2421, 2.1971, 1.5504, 2.2129], device='cuda:6'), covar=tensor([0.0200, 0.0398, 0.0226, 0.0265, 0.0197, 0.0201, 0.0367, 0.0105], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0177, 0.0158, 0.0165, 0.0172, 0.0130, 0.0175, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 13:45:49,044 INFO [train.py:904] (6/8) Epoch 12, batch 2450, loss[loss=0.1846, simple_loss=0.2785, pruned_loss=0.04529, over 16583.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2713, pruned_loss=0.05231, over 3300273.34 frames. ], batch size: 62, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:45:53,173 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.94 vs. limit=5.0 2023-04-29 13:46:23,436 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114127.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:46:57,384 INFO [train.py:904] (6/8) Epoch 12, batch 2500, loss[loss=0.222, simple_loss=0.3016, pruned_loss=0.07117, over 16265.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2709, pruned_loss=0.0517, over 3314802.64 frames. ], batch size: 165, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:47:09,683 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.290e+02 2.668e+02 3.158e+02 5.614e+02, threshold=5.335e+02, percent-clipped=0.0 2023-04-29 13:47:21,350 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-29 13:47:42,736 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 13:47:48,662 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114188.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:47:59,698 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8515, 4.1884, 3.0414, 2.2681, 2.8578, 2.5215, 4.5082, 3.8364], device='cuda:6'), covar=tensor([0.2432, 0.0572, 0.1515, 0.2294, 0.2641, 0.1698, 0.0362, 0.1061], device='cuda:6'), in_proj_covar=tensor([0.0305, 0.0261, 0.0287, 0.0284, 0.0286, 0.0227, 0.0273, 0.0305], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 13:48:06,982 INFO [train.py:904] (6/8) Epoch 12, batch 2550, loss[loss=0.1738, simple_loss=0.2687, pruned_loss=0.0394, over 17052.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2706, pruned_loss=0.05132, over 3327928.63 frames. ], batch size: 50, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:48:58,781 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1220, 5.1458, 4.9666, 4.6471, 4.4342, 5.1232, 5.1317, 4.6124], device='cuda:6'), covar=tensor([0.0657, 0.0513, 0.0390, 0.0368, 0.1217, 0.0448, 0.0279, 0.0775], device='cuda:6'), in_proj_covar=tensor([0.0266, 0.0347, 0.0315, 0.0292, 0.0333, 0.0339, 0.0213, 0.0366], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 13:49:15,390 INFO [train.py:904] (6/8) Epoch 12, batch 2600, loss[loss=0.1849, simple_loss=0.2819, pruned_loss=0.04398, over 16545.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.271, pruned_loss=0.05089, over 3328450.28 frames. ], batch size: 68, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:49:25,935 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.778e+02 2.446e+02 2.880e+02 3.491e+02 5.288e+02, threshold=5.760e+02, percent-clipped=0.0 2023-04-29 13:49:45,801 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114273.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:50:24,130 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 13:50:24,314 INFO [train.py:904] (6/8) Epoch 12, batch 2650, loss[loss=0.1882, simple_loss=0.2882, pruned_loss=0.04411, over 17010.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2721, pruned_loss=0.05115, over 3325682.91 frames. ], batch size: 50, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:50:25,243 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114302.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:50:41,667 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8029, 2.6785, 2.3274, 2.6777, 2.9804, 2.8547, 3.5413, 3.2754], device='cuda:6'), covar=tensor([0.0081, 0.0314, 0.0382, 0.0329, 0.0207, 0.0272, 0.0190, 0.0187], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0214, 0.0205, 0.0206, 0.0213, 0.0213, 0.0222, 0.0205], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:50:45,275 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9002, 1.7225, 2.3062, 2.8287, 2.6399, 3.3237, 2.1526, 3.3019], device='cuda:6'), covar=tensor([0.0180, 0.0416, 0.0260, 0.0219, 0.0238, 0.0130, 0.0346, 0.0089], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0177, 0.0157, 0.0164, 0.0172, 0.0130, 0.0174, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 13:50:51,495 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=114321.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:50:52,018 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.20 vs. limit=5.0 2023-04-29 13:51:32,414 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=114350.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:51:35,149 INFO [train.py:904] (6/8) Epoch 12, batch 2700, loss[loss=0.1862, simple_loss=0.2804, pruned_loss=0.04598, over 17137.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2718, pruned_loss=0.05039, over 3321770.46 frames. ], batch size: 48, lr: 5.75e-03, grad_scale: 16.0 2023-04-29 13:51:45,485 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.425e+02 2.890e+02 3.523e+02 1.000e+03, threshold=5.781e+02, percent-clipped=5.0 2023-04-29 13:52:22,113 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3300, 2.4121, 1.8704, 2.2444, 2.8667, 2.6146, 3.1781, 3.1398], device='cuda:6'), covar=tensor([0.0131, 0.0373, 0.0478, 0.0395, 0.0220, 0.0287, 0.0193, 0.0199], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0215, 0.0206, 0.0206, 0.0214, 0.0213, 0.0223, 0.0206], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:52:44,855 INFO [train.py:904] (6/8) Epoch 12, batch 2750, loss[loss=0.1639, simple_loss=0.2491, pruned_loss=0.03931, over 16834.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2715, pruned_loss=0.04979, over 3324150.00 frames. ], batch size: 42, lr: 5.75e-03, grad_scale: 16.0 2023-04-29 13:53:23,570 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9540, 4.0933, 2.5574, 4.7193, 3.0681, 4.6804, 2.7257, 3.3060], device='cuda:6'), covar=tensor([0.0208, 0.0297, 0.1351, 0.0191, 0.0761, 0.0374, 0.1303, 0.0643], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0170, 0.0190, 0.0141, 0.0171, 0.0215, 0.0199, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 13:53:54,634 INFO [train.py:904] (6/8) Epoch 12, batch 2800, loss[loss=0.183, simple_loss=0.266, pruned_loss=0.04999, over 16713.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2704, pruned_loss=0.05014, over 3319803.27 frames. ], batch size: 89, lr: 5.75e-03, grad_scale: 16.0 2023-04-29 13:54:06,072 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.709e+02 2.386e+02 2.760e+02 3.486e+02 6.287e+02, threshold=5.520e+02, percent-clipped=1.0 2023-04-29 13:54:38,708 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=114483.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:55:04,045 INFO [train.py:904] (6/8) Epoch 12, batch 2850, loss[loss=0.1898, simple_loss=0.2597, pruned_loss=0.0599, over 16684.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2695, pruned_loss=0.05023, over 3328045.62 frames. ], batch size: 134, lr: 5.75e-03, grad_scale: 16.0 2023-04-29 13:55:27,402 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2807, 3.3665, 3.5685, 2.4532, 3.2186, 3.6484, 3.4349, 2.1218], device='cuda:6'), covar=tensor([0.0413, 0.0095, 0.0052, 0.0311, 0.0095, 0.0084, 0.0066, 0.0352], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0072, 0.0071, 0.0126, 0.0082, 0.0091, 0.0081, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 13:56:13,203 INFO [train.py:904] (6/8) Epoch 12, batch 2900, loss[loss=0.1663, simple_loss=0.2483, pruned_loss=0.04221, over 16989.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2686, pruned_loss=0.05079, over 3319604.78 frames. ], batch size: 41, lr: 5.75e-03, grad_scale: 8.0 2023-04-29 13:56:24,536 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.629e+02 2.459e+02 2.961e+02 3.423e+02 5.764e+02, threshold=5.923e+02, percent-clipped=1.0 2023-04-29 13:57:20,462 INFO [train.py:904] (6/8) Epoch 12, batch 2950, loss[loss=0.1617, simple_loss=0.2393, pruned_loss=0.04206, over 16800.00 frames. ], tot_loss[loss=0.186, simple_loss=0.2686, pruned_loss=0.05171, over 3318060.26 frames. ], batch size: 39, lr: 5.75e-03, grad_scale: 8.0 2023-04-29 13:57:26,309 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8699, 4.0706, 2.3051, 4.5586, 2.9790, 4.4949, 2.4672, 3.1755], device='cuda:6'), covar=tensor([0.0217, 0.0308, 0.1442, 0.0208, 0.0765, 0.0447, 0.1372, 0.0632], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0170, 0.0191, 0.0140, 0.0170, 0.0215, 0.0199, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 13:57:30,529 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.7832, 6.1610, 5.8251, 5.9903, 5.4978, 5.3106, 5.6545, 6.2929], device='cuda:6'), covar=tensor([0.1063, 0.0751, 0.1041, 0.0636, 0.0698, 0.0609, 0.0977, 0.0700], device='cuda:6'), in_proj_covar=tensor([0.0578, 0.0716, 0.0591, 0.0503, 0.0452, 0.0458, 0.0596, 0.0552], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:57:44,945 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 13:57:52,193 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0332, 5.3920, 5.0986, 5.1527, 4.8740, 4.7352, 4.8458, 5.4697], device='cuda:6'), covar=tensor([0.0905, 0.0746, 0.0870, 0.0668, 0.0686, 0.0878, 0.0925, 0.0703], device='cuda:6'), in_proj_covar=tensor([0.0578, 0.0715, 0.0590, 0.0502, 0.0451, 0.0457, 0.0594, 0.0551], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 13:58:27,869 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7192, 3.9417, 3.9136, 2.8738, 3.6211, 3.9410, 3.7221, 2.2574], device='cuda:6'), covar=tensor([0.0410, 0.0083, 0.0076, 0.0331, 0.0103, 0.0121, 0.0088, 0.0461], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0072, 0.0071, 0.0125, 0.0081, 0.0090, 0.0080, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 13:58:28,574 INFO [train.py:904] (6/8) Epoch 12, batch 3000, loss[loss=0.1796, simple_loss=0.2698, pruned_loss=0.04467, over 16625.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2678, pruned_loss=0.05156, over 3328320.98 frames. ], batch size: 62, lr: 5.75e-03, grad_scale: 8.0 2023-04-29 13:58:28,575 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 13:58:38,478 INFO [train.py:938] (6/8) Epoch 12, validation: loss=0.14, simple_loss=0.2459, pruned_loss=0.01708, over 944034.00 frames. 2023-04-29 13:58:38,479 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 13:58:50,228 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.323e+02 2.752e+02 3.480e+02 1.028e+03, threshold=5.504e+02, percent-clipped=2.0 2023-04-29 13:58:58,934 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-29 13:59:15,900 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 13:59:48,674 INFO [train.py:904] (6/8) Epoch 12, batch 3050, loss[loss=0.1743, simple_loss=0.2711, pruned_loss=0.03874, over 17071.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2683, pruned_loss=0.051, over 3332223.37 frames. ], batch size: 53, lr: 5.75e-03, grad_scale: 8.0 2023-04-29 14:00:08,966 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114716.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:00:13,501 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8407, 2.5840, 2.6477, 1.9326, 2.5935, 2.7997, 2.6669, 1.8779], device='cuda:6'), covar=tensor([0.0352, 0.0093, 0.0054, 0.0295, 0.0090, 0.0085, 0.0075, 0.0308], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0071, 0.0070, 0.0124, 0.0080, 0.0089, 0.0079, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 14:00:21,279 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 14:00:38,313 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 14:00:56,624 INFO [train.py:904] (6/8) Epoch 12, batch 3100, loss[loss=0.2122, simple_loss=0.287, pruned_loss=0.06871, over 12264.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2676, pruned_loss=0.05132, over 3324022.75 frames. ], batch size: 247, lr: 5.74e-03, grad_scale: 8.0 2023-04-29 14:01:04,355 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0166, 1.8653, 2.3127, 2.7993, 2.7921, 2.8174, 1.7398, 3.0688], device='cuda:6'), covar=tensor([0.0105, 0.0374, 0.0259, 0.0191, 0.0182, 0.0184, 0.0398, 0.0079], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0177, 0.0159, 0.0165, 0.0175, 0.0131, 0.0175, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 14:01:10,357 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.272e+02 2.804e+02 3.413e+02 6.523e+02, threshold=5.608e+02, percent-clipped=1.0 2023-04-29 14:01:32,699 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114777.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:01:40,383 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114783.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:01:56,971 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-04-29 14:02:05,274 INFO [train.py:904] (6/8) Epoch 12, batch 3150, loss[loss=0.2225, simple_loss=0.2855, pruned_loss=0.07977, over 11979.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2669, pruned_loss=0.05061, over 3316797.61 frames. ], batch size: 247, lr: 5.74e-03, grad_scale: 8.0 2023-04-29 14:02:45,206 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=114831.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:03:14,011 INFO [train.py:904] (6/8) Epoch 12, batch 3200, loss[loss=0.1743, simple_loss=0.2534, pruned_loss=0.04758, over 16708.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2667, pruned_loss=0.05059, over 3309655.65 frames. ], batch size: 134, lr: 5.74e-03, grad_scale: 8.0 2023-04-29 14:03:26,053 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.649e+02 2.483e+02 2.978e+02 3.563e+02 6.234e+02, threshold=5.956e+02, percent-clipped=4.0 2023-04-29 14:03:40,859 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114871.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:03:43,228 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 14:04:22,408 INFO [train.py:904] (6/8) Epoch 12, batch 3250, loss[loss=0.1864, simple_loss=0.2756, pruned_loss=0.04864, over 17130.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2674, pruned_loss=0.05092, over 3315918.70 frames. ], batch size: 48, lr: 5.74e-03, grad_scale: 4.0 2023-04-29 14:04:56,225 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-29 14:05:05,299 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114932.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:05:32,344 INFO [train.py:904] (6/8) Epoch 12, batch 3300, loss[loss=0.1993, simple_loss=0.2715, pruned_loss=0.06361, over 16864.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2682, pruned_loss=0.05079, over 3321128.11 frames. ], batch size: 116, lr: 5.74e-03, grad_scale: 4.0 2023-04-29 14:05:45,363 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.497e+02 2.351e+02 3.077e+02 3.881e+02 7.792e+02, threshold=6.153e+02, percent-clipped=7.0 2023-04-29 14:06:01,215 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4577, 4.2551, 4.6513, 2.1460, 4.8504, 4.7742, 3.3705, 3.9150], device='cuda:6'), covar=tensor([0.0517, 0.0165, 0.0168, 0.1058, 0.0051, 0.0140, 0.0359, 0.0271], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0101, 0.0089, 0.0138, 0.0070, 0.0109, 0.0122, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 14:06:10,279 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.90 vs. limit=5.0 2023-04-29 14:06:22,036 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2912, 3.7344, 3.8249, 1.9515, 3.0423, 2.5531, 3.8239, 3.7562], device='cuda:6'), covar=tensor([0.0293, 0.0721, 0.0501, 0.1780, 0.0758, 0.0861, 0.0593, 0.0974], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0149, 0.0159, 0.0145, 0.0137, 0.0125, 0.0138, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 14:06:41,224 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1700, 5.7358, 5.9253, 5.6396, 5.6381, 6.2509, 5.7675, 5.4858], device='cuda:6'), covar=tensor([0.0856, 0.1662, 0.1587, 0.2058, 0.2648, 0.0904, 0.1273, 0.2199], device='cuda:6'), in_proj_covar=tensor([0.0361, 0.0516, 0.0558, 0.0441, 0.0599, 0.0583, 0.0443, 0.0593], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 14:06:42,024 INFO [train.py:904] (6/8) Epoch 12, batch 3350, loss[loss=0.1811, simple_loss=0.2554, pruned_loss=0.0534, over 16803.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2691, pruned_loss=0.05096, over 3321671.18 frames. ], batch size: 42, lr: 5.74e-03, grad_scale: 4.0 2023-04-29 14:07:50,787 INFO [train.py:904] (6/8) Epoch 12, batch 3400, loss[loss=0.1895, simple_loss=0.2792, pruned_loss=0.04991, over 17125.00 frames. ], tot_loss[loss=0.1848, simple_loss=0.2684, pruned_loss=0.05055, over 3315727.36 frames. ], batch size: 49, lr: 5.74e-03, grad_scale: 4.0 2023-04-29 14:08:04,045 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.344e+02 2.703e+02 3.237e+02 1.034e+03, threshold=5.406e+02, percent-clipped=1.0 2023-04-29 14:08:08,707 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7515, 4.0904, 3.2111, 2.3745, 2.8082, 2.4816, 4.1803, 3.6441], device='cuda:6'), covar=tensor([0.2596, 0.0644, 0.1478, 0.2297, 0.2428, 0.1807, 0.0473, 0.1237], device='cuda:6'), in_proj_covar=tensor([0.0303, 0.0261, 0.0285, 0.0282, 0.0285, 0.0225, 0.0271, 0.0305], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 14:08:18,295 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115072.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:08:37,791 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115086.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:08:56,420 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8556, 1.9235, 2.3909, 2.8073, 2.7323, 3.2691, 2.1409, 3.2842], device='cuda:6'), covar=tensor([0.0182, 0.0379, 0.0234, 0.0249, 0.0238, 0.0134, 0.0355, 0.0107], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0178, 0.0160, 0.0165, 0.0177, 0.0132, 0.0176, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:09:00,025 INFO [train.py:904] (6/8) Epoch 12, batch 3450, loss[loss=0.1419, simple_loss=0.2303, pruned_loss=0.02674, over 17031.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2665, pruned_loss=0.04979, over 3308857.67 frames. ], batch size: 41, lr: 5.74e-03, grad_scale: 4.0 2023-04-29 14:10:02,281 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115147.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:10:08,329 INFO [train.py:904] (6/8) Epoch 12, batch 3500, loss[loss=0.1938, simple_loss=0.267, pruned_loss=0.06027, over 16446.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.2656, pruned_loss=0.04977, over 3312991.83 frames. ], batch size: 146, lr: 5.73e-03, grad_scale: 4.0 2023-04-29 14:10:23,310 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.509e+02 2.367e+02 2.772e+02 3.300e+02 1.193e+03, threshold=5.543e+02, percent-clipped=2.0 2023-04-29 14:11:19,452 INFO [train.py:904] (6/8) Epoch 12, batch 3550, loss[loss=0.1681, simple_loss=0.2494, pruned_loss=0.04343, over 16833.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2649, pruned_loss=0.04969, over 3308064.02 frames. ], batch size: 96, lr: 5.73e-03, grad_scale: 4.0 2023-04-29 14:11:53,680 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115227.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:12:25,875 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-29 14:12:28,659 INFO [train.py:904] (6/8) Epoch 12, batch 3600, loss[loss=0.1497, simple_loss=0.2228, pruned_loss=0.0383, over 16504.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.263, pruned_loss=0.04885, over 3305952.35 frames. ], batch size: 75, lr: 5.73e-03, grad_scale: 8.0 2023-04-29 14:12:43,848 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.756e+02 2.259e+02 2.631e+02 3.384e+02 1.021e+03, threshold=5.262e+02, percent-clipped=2.0 2023-04-29 14:12:50,587 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0642, 5.5202, 5.7277, 5.4024, 5.4745, 6.0835, 5.6215, 5.2825], device='cuda:6'), covar=tensor([0.0795, 0.1823, 0.1673, 0.1883, 0.2540, 0.0831, 0.1333, 0.2202], device='cuda:6'), in_proj_covar=tensor([0.0362, 0.0516, 0.0559, 0.0438, 0.0595, 0.0581, 0.0442, 0.0594], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 14:13:17,632 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6343, 4.7125, 4.9187, 4.7660, 4.7118, 5.3514, 4.9887, 4.6811], device='cuda:6'), covar=tensor([0.1456, 0.1827, 0.2005, 0.1945, 0.2635, 0.1037, 0.1392, 0.2295], device='cuda:6'), in_proj_covar=tensor([0.0362, 0.0515, 0.0559, 0.0437, 0.0594, 0.0580, 0.0441, 0.0593], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 14:13:40,320 INFO [train.py:904] (6/8) Epoch 12, batch 3650, loss[loss=0.165, simple_loss=0.2324, pruned_loss=0.04878, over 16742.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2614, pruned_loss=0.04904, over 3296001.68 frames. ], batch size: 83, lr: 5.73e-03, grad_scale: 8.0 2023-04-29 14:13:57,224 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0341, 3.0357, 1.8011, 3.1923, 2.3934, 3.2691, 2.0236, 2.4646], device='cuda:6'), covar=tensor([0.0266, 0.0434, 0.1547, 0.0285, 0.0783, 0.0583, 0.1296, 0.0708], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0168, 0.0189, 0.0141, 0.0169, 0.0214, 0.0199, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 14:14:55,142 INFO [train.py:904] (6/8) Epoch 12, batch 3700, loss[loss=0.1753, simple_loss=0.2451, pruned_loss=0.05275, over 16899.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2603, pruned_loss=0.05087, over 3284964.81 frames. ], batch size: 96, lr: 5.73e-03, grad_scale: 8.0 2023-04-29 14:14:55,648 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4581, 3.5591, 3.1624, 2.9390, 3.0870, 3.3596, 3.2260, 3.1457], device='cuda:6'), covar=tensor([0.0591, 0.0517, 0.0278, 0.0286, 0.0552, 0.0373, 0.1362, 0.0498], device='cuda:6'), in_proj_covar=tensor([0.0266, 0.0348, 0.0318, 0.0294, 0.0337, 0.0341, 0.0214, 0.0369], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 14:15:09,329 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.748e+02 2.309e+02 2.720e+02 3.193e+02 6.265e+02, threshold=5.440e+02, percent-clipped=2.0 2023-04-29 14:15:23,493 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115370.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:15:24,758 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115371.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:15:25,900 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115372.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:15:46,482 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8242, 1.8532, 2.3138, 2.7762, 2.7817, 2.7368, 1.7994, 2.9740], device='cuda:6'), covar=tensor([0.0162, 0.0389, 0.0255, 0.0234, 0.0202, 0.0187, 0.0427, 0.0087], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0177, 0.0160, 0.0165, 0.0177, 0.0132, 0.0177, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:16:10,074 INFO [train.py:904] (6/8) Epoch 12, batch 3750, loss[loss=0.1882, simple_loss=0.2597, pruned_loss=0.05835, over 16872.00 frames. ], tot_loss[loss=0.1831, simple_loss=0.2612, pruned_loss=0.05246, over 3273817.79 frames. ], batch size: 96, lr: 5.73e-03, grad_scale: 8.0 2023-04-29 14:16:36,737 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=115420.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:16:54,182 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115431.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:16:55,927 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115432.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:17:00,637 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5463, 3.6976, 2.0999, 3.9694, 2.8047, 3.8860, 2.0825, 2.8682], device='cuda:6'), covar=tensor([0.0210, 0.0366, 0.1463, 0.0194, 0.0716, 0.0682, 0.1461, 0.0667], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0169, 0.0191, 0.0141, 0.0170, 0.0215, 0.0199, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 14:17:09,497 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115442.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:17:23,444 INFO [train.py:904] (6/8) Epoch 12, batch 3800, loss[loss=0.1752, simple_loss=0.2474, pruned_loss=0.05154, over 16341.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2622, pruned_loss=0.05351, over 3274443.76 frames. ], batch size: 165, lr: 5.73e-03, grad_scale: 8.0 2023-04-29 14:17:26,859 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4347, 3.7432, 4.0335, 1.7587, 4.0843, 4.1275, 3.1304, 2.9521], device='cuda:6'), covar=tensor([0.0917, 0.0149, 0.0128, 0.1287, 0.0070, 0.0111, 0.0354, 0.0512], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0101, 0.0089, 0.0138, 0.0071, 0.0109, 0.0122, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 14:17:38,966 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 2.466e+02 2.817e+02 3.424e+02 6.065e+02, threshold=5.633e+02, percent-clipped=2.0 2023-04-29 14:18:37,587 INFO [train.py:904] (6/8) Epoch 12, batch 3850, loss[loss=0.1997, simple_loss=0.2664, pruned_loss=0.0665, over 16764.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2629, pruned_loss=0.05417, over 3273968.60 frames. ], batch size: 124, lr: 5.73e-03, grad_scale: 8.0 2023-04-29 14:19:16,966 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115527.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:19:27,432 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6552, 1.7806, 2.2875, 2.5396, 2.6990, 2.5458, 1.7886, 2.6997], device='cuda:6'), covar=tensor([0.0127, 0.0355, 0.0217, 0.0211, 0.0168, 0.0200, 0.0342, 0.0091], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0177, 0.0159, 0.0163, 0.0175, 0.0131, 0.0175, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:19:52,698 INFO [train.py:904] (6/8) Epoch 12, batch 3900, loss[loss=0.1963, simple_loss=0.2682, pruned_loss=0.06218, over 16751.00 frames. ], tot_loss[loss=0.1859, simple_loss=0.2626, pruned_loss=0.05457, over 3277325.12 frames. ], batch size: 134, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:19:59,911 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6685, 2.8878, 2.6900, 4.8693, 3.8534, 4.3144, 1.6341, 3.4579], device='cuda:6'), covar=tensor([0.1309, 0.0735, 0.1150, 0.0121, 0.0372, 0.0334, 0.1494, 0.0736], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0162, 0.0183, 0.0155, 0.0202, 0.0214, 0.0183, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 14:20:01,736 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4393, 3.3395, 2.6467, 2.1147, 2.2481, 2.1586, 3.3116, 3.0682], device='cuda:6'), covar=tensor([0.2360, 0.0640, 0.1475, 0.2364, 0.2282, 0.1819, 0.0532, 0.1171], device='cuda:6'), in_proj_covar=tensor([0.0299, 0.0256, 0.0283, 0.0279, 0.0285, 0.0222, 0.0267, 0.0302], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:20:07,961 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.661e+02 2.153e+02 2.652e+02 3.211e+02 6.333e+02, threshold=5.304e+02, percent-clipped=2.0 2023-04-29 14:20:29,518 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=115575.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:21:08,905 INFO [train.py:904] (6/8) Epoch 12, batch 3950, loss[loss=0.1946, simple_loss=0.2611, pruned_loss=0.06407, over 16467.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.263, pruned_loss=0.0553, over 3270810.60 frames. ], batch size: 75, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:21:16,573 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1964, 3.4641, 3.3166, 2.1480, 2.9625, 2.4641, 3.5859, 3.7681], device='cuda:6'), covar=tensor([0.0278, 0.0759, 0.0576, 0.1569, 0.0724, 0.0863, 0.0490, 0.0638], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0149, 0.0159, 0.0144, 0.0137, 0.0124, 0.0138, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 14:22:00,889 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-29 14:22:21,470 INFO [train.py:904] (6/8) Epoch 12, batch 4000, loss[loss=0.1737, simple_loss=0.2525, pruned_loss=0.04743, over 16505.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2625, pruned_loss=0.05506, over 3283969.41 frames. ], batch size: 75, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:22:34,737 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 2.376e+02 2.718e+02 3.245e+02 5.190e+02, threshold=5.435e+02, percent-clipped=0.0 2023-04-29 14:23:35,774 INFO [train.py:904] (6/8) Epoch 12, batch 4050, loss[loss=0.1916, simple_loss=0.2722, pruned_loss=0.05548, over 16175.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2629, pruned_loss=0.054, over 3277242.27 frames. ], batch size: 35, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:24:12,118 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115726.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:24:13,875 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115727.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:24:30,907 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0031, 4.1160, 3.8306, 3.6084, 3.6033, 3.9995, 3.6593, 3.7487], device='cuda:6'), covar=tensor([0.0560, 0.0529, 0.0279, 0.0276, 0.0738, 0.0437, 0.0991, 0.0515], device='cuda:6'), in_proj_covar=tensor([0.0262, 0.0340, 0.0310, 0.0287, 0.0331, 0.0332, 0.0209, 0.0360], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:24:36,718 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115742.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:24:50,500 INFO [train.py:904] (6/8) Epoch 12, batch 4100, loss[loss=0.2064, simple_loss=0.2885, pruned_loss=0.06213, over 15369.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2637, pruned_loss=0.05317, over 3270892.77 frames. ], batch size: 190, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:24:57,886 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7471, 4.7728, 4.9823, 4.8653, 4.8121, 5.4370, 4.9777, 4.7149], device='cuda:6'), covar=tensor([0.0979, 0.1618, 0.1567, 0.1645, 0.2334, 0.0824, 0.1165, 0.2116], device='cuda:6'), in_proj_covar=tensor([0.0357, 0.0507, 0.0548, 0.0433, 0.0583, 0.0572, 0.0438, 0.0587], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 14:24:58,071 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5857, 1.8895, 2.1867, 2.5488, 2.6961, 2.8688, 1.8426, 2.7205], device='cuda:6'), covar=tensor([0.0167, 0.0344, 0.0244, 0.0208, 0.0205, 0.0117, 0.0373, 0.0093], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0177, 0.0160, 0.0164, 0.0176, 0.0132, 0.0176, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:25:05,531 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.445e+02 1.888e+02 2.192e+02 2.609e+02 4.553e+02, threshold=4.384e+02, percent-clipped=0.0 2023-04-29 14:25:25,641 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-29 14:25:26,930 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7779, 5.0510, 4.8272, 4.8540, 4.5798, 4.4543, 4.5577, 5.1205], device='cuda:6'), covar=tensor([0.0945, 0.0747, 0.0850, 0.0605, 0.0671, 0.0936, 0.0812, 0.0737], device='cuda:6'), in_proj_covar=tensor([0.0575, 0.0716, 0.0589, 0.0506, 0.0451, 0.0462, 0.0593, 0.0555], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:25:48,338 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=115790.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:26:05,810 INFO [train.py:904] (6/8) Epoch 12, batch 4150, loss[loss=0.2265, simple_loss=0.3098, pruned_loss=0.07157, over 15275.00 frames. ], tot_loss[loss=0.1925, simple_loss=0.2718, pruned_loss=0.05657, over 3229842.80 frames. ], batch size: 191, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:27:23,037 INFO [train.py:904] (6/8) Epoch 12, batch 4200, loss[loss=0.2493, simple_loss=0.3172, pruned_loss=0.0907, over 11179.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2791, pruned_loss=0.05842, over 3210618.49 frames. ], batch size: 247, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:27:37,142 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.546e+02 2.889e+02 3.538e+02 7.743e+02, threshold=5.778e+02, percent-clipped=11.0 2023-04-29 14:28:22,481 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115892.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 14:28:34,017 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6340, 2.6360, 2.3465, 3.4293, 2.3201, 3.6429, 1.4763, 2.8634], device='cuda:6'), covar=tensor([0.1325, 0.0646, 0.1158, 0.0152, 0.0164, 0.0410, 0.1545, 0.0741], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0161, 0.0182, 0.0154, 0.0201, 0.0212, 0.0183, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 14:28:36,618 INFO [train.py:904] (6/8) Epoch 12, batch 4250, loss[loss=0.2247, simple_loss=0.2936, pruned_loss=0.07788, over 12303.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2829, pruned_loss=0.05895, over 3186821.35 frames. ], batch size: 248, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:29:15,823 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-29 14:29:26,865 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4662, 4.3045, 4.5273, 4.6610, 4.8181, 4.3546, 4.7238, 4.8408], device='cuda:6'), covar=tensor([0.1344, 0.1024, 0.1361, 0.0596, 0.0468, 0.0913, 0.0645, 0.0511], device='cuda:6'), in_proj_covar=tensor([0.0539, 0.0672, 0.0803, 0.0684, 0.0513, 0.0532, 0.0534, 0.0617], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:29:49,145 INFO [train.py:904] (6/8) Epoch 12, batch 4300, loss[loss=0.2213, simple_loss=0.3011, pruned_loss=0.07071, over 12064.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2839, pruned_loss=0.05789, over 3195054.65 frames. ], batch size: 248, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:29:50,874 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115953.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 14:30:04,139 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7696, 3.8319, 2.1662, 4.4880, 2.7719, 4.3585, 2.3304, 2.8929], device='cuda:6'), covar=tensor([0.0209, 0.0281, 0.1580, 0.0106, 0.0818, 0.0358, 0.1421, 0.0712], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0165, 0.0188, 0.0134, 0.0168, 0.0208, 0.0194, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 14:30:04,743 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 2.520e+02 2.885e+02 3.320e+02 6.529e+02, threshold=5.769e+02, percent-clipped=1.0 2023-04-29 14:30:45,367 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0575, 5.3829, 5.6241, 5.3914, 5.4288, 5.9723, 5.5248, 5.1988], device='cuda:6'), covar=tensor([0.0769, 0.1578, 0.1785, 0.1680, 0.2105, 0.0856, 0.1277, 0.2338], device='cuda:6'), in_proj_covar=tensor([0.0351, 0.0495, 0.0536, 0.0423, 0.0569, 0.0562, 0.0429, 0.0576], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 14:31:07,410 INFO [train.py:904] (6/8) Epoch 12, batch 4350, loss[loss=0.2157, simple_loss=0.2993, pruned_loss=0.06607, over 16692.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2873, pruned_loss=0.05904, over 3225313.59 frames. ], batch size: 134, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:31:27,537 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116014.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:31:46,087 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116026.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:31:47,878 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116027.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:32:22,085 INFO [train.py:904] (6/8) Epoch 12, batch 4400, loss[loss=0.2105, simple_loss=0.2981, pruned_loss=0.06148, over 16292.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2893, pruned_loss=0.06003, over 3210362.64 frames. ], batch size: 165, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:32:37,555 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.512e+02 3.045e+02 3.645e+02 6.621e+02, threshold=6.090e+02, percent-clipped=4.0 2023-04-29 14:32:55,644 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=116074.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:32:57,362 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=116075.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:32:57,565 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116075.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 14:32:58,648 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7717, 5.0955, 4.8686, 4.8799, 4.6464, 4.5167, 4.5230, 5.1677], device='cuda:6'), covar=tensor([0.0927, 0.0731, 0.0843, 0.0694, 0.0644, 0.0933, 0.0903, 0.0776], device='cuda:6'), in_proj_covar=tensor([0.0556, 0.0690, 0.0568, 0.0487, 0.0435, 0.0447, 0.0573, 0.0536], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:33:35,439 INFO [train.py:904] (6/8) Epoch 12, batch 4450, loss[loss=0.2056, simple_loss=0.2839, pruned_loss=0.06366, over 12145.00 frames. ], tot_loss[loss=0.2077, simple_loss=0.2926, pruned_loss=0.06145, over 3190136.76 frames. ], batch size: 248, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:33:40,160 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116105.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:33:57,023 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116116.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:34:04,077 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.03 vs. limit=5.0 2023-04-29 14:34:32,551 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4029, 3.4957, 1.8312, 3.9884, 2.6464, 3.9101, 2.0493, 2.7337], device='cuda:6'), covar=tensor([0.0231, 0.0335, 0.1823, 0.0088, 0.0751, 0.0384, 0.1545, 0.0720], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0163, 0.0187, 0.0132, 0.0165, 0.0205, 0.0192, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-29 14:34:49,327 INFO [train.py:904] (6/8) Epoch 12, batch 4500, loss[loss=0.2051, simple_loss=0.2977, pruned_loss=0.05626, over 17241.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2928, pruned_loss=0.06118, over 3212795.27 frames. ], batch size: 52, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:35:03,473 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 2.115e+02 2.423e+02 2.863e+02 5.089e+02, threshold=4.846e+02, percent-clipped=0.0 2023-04-29 14:35:10,907 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116166.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:35:26,237 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116177.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:35:32,570 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 14:36:02,110 INFO [train.py:904] (6/8) Epoch 12, batch 4550, loss[loss=0.1903, simple_loss=0.2793, pruned_loss=0.05067, over 17043.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.2929, pruned_loss=0.0615, over 3226173.18 frames. ], batch size: 50, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:37:04,522 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8765, 2.1871, 2.3523, 2.5850, 2.2565, 3.0914, 1.6879, 2.7158], device='cuda:6'), covar=tensor([0.1374, 0.0687, 0.1163, 0.0165, 0.0227, 0.0395, 0.1558, 0.0723], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0160, 0.0181, 0.0151, 0.0200, 0.0209, 0.0182, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 14:37:08,868 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116248.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 14:37:14,079 INFO [train.py:904] (6/8) Epoch 12, batch 4600, loss[loss=0.1867, simple_loss=0.2864, pruned_loss=0.04351, over 16706.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2934, pruned_loss=0.06133, over 3233440.21 frames. ], batch size: 89, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:37:29,432 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.395e+02 1.985e+02 2.258e+02 2.658e+02 3.690e+02, threshold=4.517e+02, percent-clipped=0.0 2023-04-29 14:38:26,068 INFO [train.py:904] (6/8) Epoch 12, batch 4650, loss[loss=0.1693, simple_loss=0.2601, pruned_loss=0.03922, over 16604.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.293, pruned_loss=0.06184, over 3222417.84 frames. ], batch size: 62, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:38:56,378 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116323.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:39:01,269 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8039, 2.5926, 2.0662, 2.3668, 3.0118, 2.6989, 3.4584, 3.2903], device='cuda:6'), covar=tensor([0.0049, 0.0294, 0.0420, 0.0351, 0.0193, 0.0303, 0.0138, 0.0178], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0207, 0.0203, 0.0201, 0.0207, 0.0207, 0.0213, 0.0201], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:39:38,311 INFO [train.py:904] (6/8) Epoch 12, batch 4700, loss[loss=0.1976, simple_loss=0.283, pruned_loss=0.05614, over 16752.00 frames. ], tot_loss[loss=0.206, simple_loss=0.2905, pruned_loss=0.06076, over 3207508.61 frames. ], batch size: 124, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:39:53,946 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.398e+02 2.029e+02 2.445e+02 3.086e+02 5.514e+02, threshold=4.890e+02, percent-clipped=4.0 2023-04-29 14:40:05,782 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116370.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 14:40:27,541 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116384.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:40:54,107 INFO [train.py:904] (6/8) Epoch 12, batch 4750, loss[loss=0.1643, simple_loss=0.2442, pruned_loss=0.04226, over 16597.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2857, pruned_loss=0.05835, over 3208557.16 frames. ], batch size: 57, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:41:42,637 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7386, 2.5255, 1.8387, 2.2886, 2.9753, 2.7216, 3.3414, 3.2682], device='cuda:6'), covar=tensor([0.0052, 0.0336, 0.0535, 0.0404, 0.0203, 0.0309, 0.0143, 0.0199], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0204, 0.0201, 0.0199, 0.0205, 0.0204, 0.0210, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:41:59,008 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116446.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:42:07,120 INFO [train.py:904] (6/8) Epoch 12, batch 4800, loss[loss=0.2136, simple_loss=0.3005, pruned_loss=0.06332, over 15297.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.2825, pruned_loss=0.05656, over 3201526.02 frames. ], batch size: 190, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:42:22,210 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116461.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:42:23,017 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 1.880e+02 2.223e+02 2.728e+02 4.441e+02, threshold=4.446e+02, percent-clipped=0.0 2023-04-29 14:42:39,675 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116472.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:42:56,749 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.26 vs. limit=5.0 2023-04-29 14:43:23,414 INFO [train.py:904] (6/8) Epoch 12, batch 4850, loss[loss=0.2271, simple_loss=0.3032, pruned_loss=0.07547, over 12409.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2831, pruned_loss=0.05575, over 3194566.70 frames. ], batch size: 247, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:43:28,333 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116505.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:43:31,433 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116507.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 14:43:39,748 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-29 14:44:05,928 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5499, 5.4846, 5.4174, 4.8016, 5.4040, 2.1258, 5.2208, 5.3875], device='cuda:6'), covar=tensor([0.0049, 0.0042, 0.0094, 0.0379, 0.0058, 0.2007, 0.0081, 0.0104], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0118, 0.0163, 0.0155, 0.0136, 0.0178, 0.0154, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:44:06,196 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-04-29 14:44:31,815 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116548.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 14:44:38,141 INFO [train.py:904] (6/8) Epoch 12, batch 4900, loss[loss=0.188, simple_loss=0.2754, pruned_loss=0.05031, over 16736.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2825, pruned_loss=0.05434, over 3191662.47 frames. ], batch size: 124, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:44:52,631 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.574e+02 2.192e+02 2.607e+02 3.082e+02 6.652e+02, threshold=5.215e+02, percent-clipped=3.0 2023-04-29 14:44:59,316 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116566.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:45:42,830 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=116596.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 14:45:52,018 INFO [train.py:904] (6/8) Epoch 12, batch 4950, loss[loss=0.1822, simple_loss=0.2651, pruned_loss=0.04966, over 16883.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2819, pruned_loss=0.05383, over 3197515.43 frames. ], batch size: 42, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:46:16,320 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116618.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:47:04,268 INFO [train.py:904] (6/8) Epoch 12, batch 5000, loss[loss=0.1878, simple_loss=0.2757, pruned_loss=0.04993, over 16479.00 frames. ], tot_loss[loss=0.1953, simple_loss=0.2831, pruned_loss=0.05373, over 3201842.90 frames. ], batch size: 68, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:47:17,030 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.254e+02 2.645e+02 3.532e+02 7.072e+02, threshold=5.290e+02, percent-clipped=1.0 2023-04-29 14:47:26,241 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6124, 2.6287, 1.6407, 2.7833, 2.0900, 2.8089, 2.0115, 2.4142], device='cuda:6'), covar=tensor([0.0259, 0.0378, 0.1427, 0.0156, 0.0735, 0.0426, 0.1233, 0.0595], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0161, 0.0186, 0.0129, 0.0163, 0.0202, 0.0192, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-29 14:47:30,407 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116670.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 14:47:43,260 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116679.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:47:43,371 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116679.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:48:15,687 INFO [train.py:904] (6/8) Epoch 12, batch 5050, loss[loss=0.1978, simple_loss=0.2856, pruned_loss=0.05499, over 16629.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2841, pruned_loss=0.05382, over 3199143.59 frames. ], batch size: 134, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:48:18,365 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116704.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:48:18,691 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-29 14:48:38,190 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=116718.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:49:24,608 INFO [train.py:904] (6/8) Epoch 12, batch 5100, loss[loss=0.1684, simple_loss=0.2595, pruned_loss=0.03867, over 16458.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2828, pruned_loss=0.05339, over 3201349.89 frames. ], batch size: 75, lr: 5.69e-03, grad_scale: 8.0 2023-04-29 14:49:37,470 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116761.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:49:38,225 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 2.103e+02 2.626e+02 3.146e+02 5.078e+02, threshold=5.251e+02, percent-clipped=1.0 2023-04-29 14:49:43,724 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116765.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:49:52,543 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116772.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:50:35,772 INFO [train.py:904] (6/8) Epoch 12, batch 5150, loss[loss=0.1939, simple_loss=0.2951, pruned_loss=0.04632, over 15381.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2825, pruned_loss=0.05244, over 3203602.68 frames. ], batch size: 191, lr: 5.69e-03, grad_scale: 8.0 2023-04-29 14:50:36,704 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116802.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 14:50:36,855 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116802.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:50:47,861 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=116809.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:51:03,103 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=116820.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:51:15,167 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1622, 2.7844, 2.8300, 2.0064, 2.6533, 2.1027, 2.8371, 2.9825], device='cuda:6'), covar=tensor([0.0261, 0.0648, 0.0532, 0.1557, 0.0715, 0.0890, 0.0553, 0.0643], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0148, 0.0160, 0.0145, 0.0137, 0.0125, 0.0138, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 14:51:44,587 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-29 14:51:47,911 INFO [train.py:904] (6/8) Epoch 12, batch 5200, loss[loss=0.2006, simple_loss=0.2872, pruned_loss=0.05706, over 16640.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2806, pruned_loss=0.05192, over 3203521.04 frames. ], batch size: 62, lr: 5.69e-03, grad_scale: 8.0 2023-04-29 14:51:55,172 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-29 14:52:00,767 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116861.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:52:01,643 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.630e+02 2.168e+02 2.438e+02 3.019e+02 6.927e+02, threshold=4.876e+02, percent-clipped=1.0 2023-04-29 14:52:03,386 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116863.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:52:58,581 INFO [train.py:904] (6/8) Epoch 12, batch 5250, loss[loss=0.2144, simple_loss=0.305, pruned_loss=0.06193, over 16303.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2791, pruned_loss=0.05226, over 3190494.89 frames. ], batch size: 165, lr: 5.69e-03, grad_scale: 16.0 2023-04-29 14:54:11,415 INFO [train.py:904] (6/8) Epoch 12, batch 5300, loss[loss=0.1725, simple_loss=0.253, pruned_loss=0.04597, over 16761.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.276, pruned_loss=0.05108, over 3195102.01 frames. ], batch size: 62, lr: 5.69e-03, grad_scale: 8.0 2023-04-29 14:54:27,261 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.433e+02 2.208e+02 2.588e+02 3.045e+02 5.450e+02, threshold=5.175e+02, percent-clipped=1.0 2023-04-29 14:54:42,592 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116974.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:54:48,700 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6768, 4.6931, 4.5043, 4.2037, 4.1313, 4.5865, 4.4214, 4.2429], device='cuda:6'), covar=tensor([0.0516, 0.0434, 0.0259, 0.0251, 0.0966, 0.0403, 0.0372, 0.0615], device='cuda:6'), in_proj_covar=tensor([0.0249, 0.0330, 0.0299, 0.0277, 0.0318, 0.0322, 0.0203, 0.0347], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:54:49,797 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116979.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:54:55,663 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1895, 4.0580, 3.9916, 2.4944, 3.5633, 3.9811, 3.6696, 1.9889], device='cuda:6'), covar=tensor([0.0481, 0.0025, 0.0029, 0.0349, 0.0069, 0.0080, 0.0062, 0.0430], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0068, 0.0069, 0.0124, 0.0080, 0.0089, 0.0079, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 14:55:21,983 INFO [train.py:904] (6/8) Epoch 12, batch 5350, loss[loss=0.1748, simple_loss=0.2679, pruned_loss=0.04078, over 16811.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2732, pruned_loss=0.04952, over 3194976.69 frames. ], batch size: 83, lr: 5.69e-03, grad_scale: 4.0 2023-04-29 14:55:58,442 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117027.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:56:02,313 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0611, 5.0271, 4.8862, 4.5235, 4.4527, 4.8694, 4.9356, 4.5995], device='cuda:6'), covar=tensor([0.0593, 0.0592, 0.0301, 0.0275, 0.1171, 0.0501, 0.0257, 0.0636], device='cuda:6'), in_proj_covar=tensor([0.0250, 0.0330, 0.0300, 0.0277, 0.0319, 0.0323, 0.0203, 0.0348], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:56:24,830 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-29 14:56:31,809 INFO [train.py:904] (6/8) Epoch 12, batch 5400, loss[loss=0.1868, simple_loss=0.2747, pruned_loss=0.04949, over 16719.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2756, pruned_loss=0.05023, over 3208977.93 frames. ], batch size: 83, lr: 5.69e-03, grad_scale: 4.0 2023-04-29 14:56:35,844 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0672, 5.0601, 4.8815, 4.5747, 4.5130, 4.8835, 4.9328, 4.5929], device='cuda:6'), covar=tensor([0.0557, 0.0439, 0.0272, 0.0251, 0.1067, 0.0444, 0.0260, 0.0625], device='cuda:6'), in_proj_covar=tensor([0.0249, 0.0329, 0.0299, 0.0276, 0.0318, 0.0322, 0.0203, 0.0347], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 14:56:43,949 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117060.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:56:49,066 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.392e+02 2.253e+02 2.451e+02 2.837e+02 5.241e+02, threshold=4.902e+02, percent-clipped=1.0 2023-04-29 14:57:46,200 INFO [train.py:904] (6/8) Epoch 12, batch 5450, loss[loss=0.2209, simple_loss=0.3052, pruned_loss=0.06828, over 16764.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.2785, pruned_loss=0.05166, over 3207734.56 frames. ], batch size: 83, lr: 5.69e-03, grad_scale: 4.0 2023-04-29 14:57:46,776 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117102.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:58:57,655 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117150.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:59:00,350 INFO [train.py:904] (6/8) Epoch 12, batch 5500, loss[loss=0.2025, simple_loss=0.29, pruned_loss=0.05746, over 16843.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2866, pruned_loss=0.0573, over 3152581.14 frames. ], batch size: 102, lr: 5.69e-03, grad_scale: 4.0 2023-04-29 14:59:09,312 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117158.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:59:13,805 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117161.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:59:18,158 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 2.773e+02 3.440e+02 4.409e+02 8.971e+02, threshold=6.880e+02, percent-clipped=17.0 2023-04-29 15:00:18,009 INFO [train.py:904] (6/8) Epoch 12, batch 5550, loss[loss=0.2476, simple_loss=0.3257, pruned_loss=0.08478, over 16081.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2941, pruned_loss=0.06299, over 3129525.61 frames. ], batch size: 165, lr: 5.68e-03, grad_scale: 4.0 2023-04-29 15:00:30,360 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117209.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:00:42,276 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-29 15:00:49,883 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=117221.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:00:50,185 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-29 15:01:39,143 INFO [train.py:904] (6/8) Epoch 12, batch 5600, loss[loss=0.3608, simple_loss=0.4032, pruned_loss=0.1592, over 11208.00 frames. ], tot_loss[loss=0.2184, simple_loss=0.3001, pruned_loss=0.0683, over 3078962.82 frames. ], batch size: 246, lr: 5.68e-03, grad_scale: 8.0 2023-04-29 15:01:51,446 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-29 15:01:58,900 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.404e+02 3.697e+02 4.299e+02 5.006e+02 8.998e+02, threshold=8.599e+02, percent-clipped=4.0 2023-04-29 15:02:00,078 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3813, 5.3930, 5.0725, 4.5307, 5.2302, 1.9328, 5.0392, 5.0123], device='cuda:6'), covar=tensor([0.0060, 0.0041, 0.0126, 0.0310, 0.0062, 0.2177, 0.0078, 0.0144], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0118, 0.0163, 0.0156, 0.0135, 0.0178, 0.0152, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:02:16,725 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117274.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:02:30,429 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=117282.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:03:02,066 INFO [train.py:904] (6/8) Epoch 12, batch 5650, loss[loss=0.316, simple_loss=0.3634, pruned_loss=0.1343, over 11215.00 frames. ], tot_loss[loss=0.2252, simple_loss=0.3053, pruned_loss=0.07256, over 3060716.57 frames. ], batch size: 248, lr: 5.68e-03, grad_scale: 4.0 2023-04-29 15:03:32,785 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4546, 3.4460, 3.4023, 2.7864, 3.2891, 2.1168, 3.1948, 2.8184], device='cuda:6'), covar=tensor([0.0119, 0.0099, 0.0145, 0.0204, 0.0077, 0.1903, 0.0104, 0.0166], device='cuda:6'), in_proj_covar=tensor([0.0129, 0.0118, 0.0162, 0.0155, 0.0134, 0.0176, 0.0151, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:03:33,949 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117322.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:03:44,980 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=117329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:04:10,033 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.69 vs. limit=5.0 2023-04-29 15:04:18,926 INFO [train.py:904] (6/8) Epoch 12, batch 5700, loss[loss=0.2572, simple_loss=0.314, pruned_loss=0.1003, over 11216.00 frames. ], tot_loss[loss=0.2267, simple_loss=0.3062, pruned_loss=0.0736, over 3066382.73 frames. ], batch size: 248, lr: 5.68e-03, grad_scale: 2.0 2023-04-29 15:04:22,452 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1605, 1.9612, 2.1607, 3.7994, 1.9541, 2.4256, 2.1139, 2.1916], device='cuda:6'), covar=tensor([0.1090, 0.3296, 0.2398, 0.0473, 0.3958, 0.2175, 0.3116, 0.3311], device='cuda:6'), in_proj_covar=tensor([0.0366, 0.0396, 0.0331, 0.0316, 0.0412, 0.0455, 0.0360, 0.0461], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:04:32,710 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117360.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:04:41,587 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.520e+02 3.684e+02 4.316e+02 5.371e+02 1.144e+03, threshold=8.631e+02, percent-clipped=1.0 2023-04-29 15:05:21,447 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7025, 1.2985, 1.6304, 1.5257, 1.7012, 1.8656, 1.5262, 1.6920], device='cuda:6'), covar=tensor([0.0199, 0.0241, 0.0133, 0.0180, 0.0171, 0.0109, 0.0247, 0.0080], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0169, 0.0153, 0.0157, 0.0166, 0.0124, 0.0170, 0.0116], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 15:05:21,456 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=117390.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:05:39,177 INFO [train.py:904] (6/8) Epoch 12, batch 5750, loss[loss=0.2491, simple_loss=0.3156, pruned_loss=0.09134, over 11178.00 frames. ], tot_loss[loss=0.2291, simple_loss=0.3085, pruned_loss=0.07485, over 3046333.17 frames. ], batch size: 248, lr: 5.68e-03, grad_scale: 2.0 2023-04-29 15:05:48,655 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117408.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:07:00,135 INFO [train.py:904] (6/8) Epoch 12, batch 5800, loss[loss=0.2002, simple_loss=0.2905, pruned_loss=0.05497, over 16415.00 frames. ], tot_loss[loss=0.2262, simple_loss=0.3075, pruned_loss=0.0724, over 3077740.42 frames. ], batch size: 146, lr: 5.68e-03, grad_scale: 2.0 2023-04-29 15:07:09,902 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117458.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:07:21,329 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.077e+02 3.389e+02 3.943e+02 4.772e+02 8.236e+02, threshold=7.885e+02, percent-clipped=0.0 2023-04-29 15:07:54,296 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3889, 4.5556, 4.7167, 4.5316, 4.5868, 5.1183, 4.6441, 4.3808], device='cuda:6'), covar=tensor([0.1496, 0.1751, 0.1995, 0.1994, 0.2476, 0.1006, 0.1506, 0.2624], device='cuda:6'), in_proj_covar=tensor([0.0351, 0.0487, 0.0531, 0.0420, 0.0572, 0.0556, 0.0423, 0.0576], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 15:08:16,564 INFO [train.py:904] (6/8) Epoch 12, batch 5850, loss[loss=0.215, simple_loss=0.2891, pruned_loss=0.07047, over 11671.00 frames. ], tot_loss[loss=0.2237, simple_loss=0.3054, pruned_loss=0.07101, over 3067894.34 frames. ], batch size: 248, lr: 5.68e-03, grad_scale: 2.0 2023-04-29 15:08:24,234 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117506.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:08:50,632 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6905, 3.5136, 3.5051, 3.8911, 3.8608, 3.5614, 3.8672, 3.9464], device='cuda:6'), covar=tensor([0.1415, 0.1358, 0.2153, 0.0827, 0.1007, 0.2456, 0.1061, 0.0945], device='cuda:6'), in_proj_covar=tensor([0.0521, 0.0653, 0.0780, 0.0662, 0.0496, 0.0515, 0.0521, 0.0598], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:09:36,851 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-29 15:09:37,155 INFO [train.py:904] (6/8) Epoch 12, batch 5900, loss[loss=0.2388, simple_loss=0.3007, pruned_loss=0.08849, over 11480.00 frames. ], tot_loss[loss=0.2243, simple_loss=0.3055, pruned_loss=0.07152, over 3049270.86 frames. ], batch size: 248, lr: 5.68e-03, grad_scale: 2.0 2023-04-29 15:10:01,560 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.797e+02 2.838e+02 3.606e+02 4.204e+02 7.799e+02, threshold=7.213e+02, percent-clipped=0.0 2023-04-29 15:10:18,955 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117577.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:10:55,830 INFO [train.py:904] (6/8) Epoch 12, batch 5950, loss[loss=0.2085, simple_loss=0.2965, pruned_loss=0.06029, over 16366.00 frames. ], tot_loss[loss=0.2235, simple_loss=0.306, pruned_loss=0.07052, over 3042523.59 frames. ], batch size: 146, lr: 5.67e-03, grad_scale: 2.0 2023-04-29 15:11:09,706 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2023-04-29 15:12:14,103 INFO [train.py:904] (6/8) Epoch 12, batch 6000, loss[loss=0.1802, simple_loss=0.2671, pruned_loss=0.04667, over 16700.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.3044, pruned_loss=0.06917, over 3054682.03 frames. ], batch size: 62, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:12:14,103 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 15:12:25,320 INFO [train.py:938] (6/8) Epoch 12, validation: loss=0.161, simple_loss=0.2739, pruned_loss=0.02405, over 944034.00 frames. 2023-04-29 15:12:25,320 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 15:12:27,584 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4673, 3.4660, 3.3955, 2.7176, 3.2959, 2.0743, 3.1238, 2.8116], device='cuda:6'), covar=tensor([0.0136, 0.0098, 0.0144, 0.0212, 0.0091, 0.1937, 0.0123, 0.0184], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0118, 0.0162, 0.0156, 0.0135, 0.0177, 0.0151, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:12:27,685 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0469, 2.8017, 2.8286, 2.0670, 2.5875, 2.2688, 2.6824, 2.9912], device='cuda:6'), covar=tensor([0.0311, 0.0684, 0.0456, 0.1551, 0.0732, 0.0739, 0.0593, 0.0625], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0148, 0.0161, 0.0145, 0.0138, 0.0125, 0.0140, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 15:12:46,504 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.036e+02 2.944e+02 3.518e+02 4.227e+02 7.444e+02, threshold=7.035e+02, percent-clipped=1.0 2023-04-29 15:13:15,815 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-29 15:13:18,109 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117685.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:13:46,105 INFO [train.py:904] (6/8) Epoch 12, batch 6050, loss[loss=0.2008, simple_loss=0.3069, pruned_loss=0.0474, over 16912.00 frames. ], tot_loss[loss=0.2198, simple_loss=0.3032, pruned_loss=0.06819, over 3069394.55 frames. ], batch size: 96, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:15:02,175 INFO [train.py:904] (6/8) Epoch 12, batch 6100, loss[loss=0.2196, simple_loss=0.3056, pruned_loss=0.06682, over 15205.00 frames. ], tot_loss[loss=0.2189, simple_loss=0.303, pruned_loss=0.06737, over 3081195.76 frames. ], batch size: 191, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:15:24,818 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.907e+02 2.974e+02 3.801e+02 4.781e+02 1.516e+03, threshold=7.602e+02, percent-clipped=11.0 2023-04-29 15:16:19,684 INFO [train.py:904] (6/8) Epoch 12, batch 6150, loss[loss=0.2459, simple_loss=0.3144, pruned_loss=0.08865, over 11488.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.3005, pruned_loss=0.06633, over 3103286.06 frames. ], batch size: 248, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:16:40,912 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5257, 3.9107, 3.7478, 2.0425, 3.3570, 2.3839, 3.8008, 4.0775], device='cuda:6'), covar=tensor([0.0237, 0.0629, 0.0559, 0.1850, 0.0700, 0.0884, 0.0700, 0.0885], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0146, 0.0159, 0.0143, 0.0137, 0.0124, 0.0138, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 15:16:43,530 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-29 15:17:32,201 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 15:17:38,949 INFO [train.py:904] (6/8) Epoch 12, batch 6200, loss[loss=0.2379, simple_loss=0.3152, pruned_loss=0.08037, over 11839.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.2989, pruned_loss=0.06574, over 3105120.94 frames. ], batch size: 247, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:17:49,325 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6763, 2.4632, 2.0982, 3.9793, 2.5461, 3.8370, 1.4389, 2.5494], device='cuda:6'), covar=tensor([0.1369, 0.0839, 0.1476, 0.0164, 0.0261, 0.0432, 0.1661, 0.0988], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0161, 0.0181, 0.0150, 0.0198, 0.0207, 0.0182, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 15:17:57,636 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-29 15:18:00,669 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.030e+02 3.055e+02 3.626e+02 4.280e+02 7.203e+02, threshold=7.253e+02, percent-clipped=0.0 2023-04-29 15:18:17,999 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117877.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:18:52,878 INFO [train.py:904] (6/8) Epoch 12, batch 6250, loss[loss=0.2129, simple_loss=0.3053, pruned_loss=0.06031, over 16694.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.2994, pruned_loss=0.06568, over 3115264.53 frames. ], batch size: 89, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:19:03,039 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4833, 4.6501, 4.3729, 4.1554, 3.7973, 4.5089, 4.3134, 4.1162], device='cuda:6'), covar=tensor([0.0938, 0.0922, 0.0412, 0.0459, 0.1250, 0.0650, 0.0668, 0.1004], device='cuda:6'), in_proj_covar=tensor([0.0242, 0.0323, 0.0288, 0.0269, 0.0308, 0.0311, 0.0198, 0.0338], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:19:28,072 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117925.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:20:06,465 INFO [train.py:904] (6/8) Epoch 12, batch 6300, loss[loss=0.1719, simple_loss=0.2647, pruned_loss=0.03952, over 16516.00 frames. ], tot_loss[loss=0.2137, simple_loss=0.2984, pruned_loss=0.06444, over 3136568.56 frames. ], batch size: 75, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:20:28,834 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.927e+02 3.047e+02 3.596e+02 4.345e+02 1.152e+03, threshold=7.193e+02, percent-clipped=4.0 2023-04-29 15:20:57,493 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117985.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:21:25,210 INFO [train.py:904] (6/8) Epoch 12, batch 6350, loss[loss=0.2078, simple_loss=0.2904, pruned_loss=0.06258, over 16699.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.2994, pruned_loss=0.06581, over 3134061.15 frames. ], batch size: 83, lr: 5.66e-03, grad_scale: 4.0 2023-04-29 15:22:11,813 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=118033.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:22:21,811 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118040.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:22:37,545 INFO [train.py:904] (6/8) Epoch 12, batch 6400, loss[loss=0.3036, simple_loss=0.3553, pruned_loss=0.126, over 10960.00 frames. ], tot_loss[loss=0.218, simple_loss=0.3005, pruned_loss=0.06771, over 3107164.08 frames. ], batch size: 247, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:22:57,664 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.235e+02 3.125e+02 4.036e+02 4.702e+02 8.359e+02, threshold=8.072e+02, percent-clipped=7.0 2023-04-29 15:23:50,429 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118101.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:23:50,964 INFO [train.py:904] (6/8) Epoch 12, batch 6450, loss[loss=0.2035, simple_loss=0.2849, pruned_loss=0.06102, over 16790.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.3004, pruned_loss=0.06658, over 3111096.42 frames. ], batch size: 83, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:24:20,402 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.15 vs. limit=5.0 2023-04-29 15:25:08,004 INFO [train.py:904] (6/8) Epoch 12, batch 6500, loss[loss=0.2032, simple_loss=0.2933, pruned_loss=0.05657, over 16789.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.2991, pruned_loss=0.06687, over 3088406.92 frames. ], batch size: 39, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:25:29,347 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.059e+02 2.874e+02 3.539e+02 4.309e+02 7.479e+02, threshold=7.078e+02, percent-clipped=0.0 2023-04-29 15:25:32,656 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118168.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:25:44,056 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2023-04-29 15:25:46,027 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-29 15:26:28,542 INFO [train.py:904] (6/8) Epoch 12, batch 6550, loss[loss=0.2779, simple_loss=0.3373, pruned_loss=0.1093, over 11783.00 frames. ], tot_loss[loss=0.2185, simple_loss=0.3014, pruned_loss=0.06782, over 3082097.94 frames. ], batch size: 247, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:26:35,356 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6974, 4.9310, 5.1430, 4.9183, 5.0230, 5.5714, 5.0746, 4.8341], device='cuda:6'), covar=tensor([0.1002, 0.1641, 0.2425, 0.1896, 0.2247, 0.0872, 0.1446, 0.2068], device='cuda:6'), in_proj_covar=tensor([0.0358, 0.0496, 0.0545, 0.0430, 0.0580, 0.0568, 0.0432, 0.0585], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 15:27:10,819 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118229.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 15:27:42,508 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7297, 5.0169, 5.2041, 5.0043, 5.0268, 5.6040, 5.1156, 4.9022], device='cuda:6'), covar=tensor([0.1033, 0.1779, 0.2162, 0.1882, 0.2538, 0.0970, 0.1508, 0.2271], device='cuda:6'), in_proj_covar=tensor([0.0358, 0.0496, 0.0546, 0.0431, 0.0583, 0.0570, 0.0432, 0.0586], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 15:27:44,389 INFO [train.py:904] (6/8) Epoch 12, batch 6600, loss[loss=0.216, simple_loss=0.3, pruned_loss=0.06598, over 16881.00 frames. ], tot_loss[loss=0.2202, simple_loss=0.3038, pruned_loss=0.0683, over 3095490.99 frames. ], batch size: 116, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:27:57,199 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8938, 2.5215, 2.5063, 4.6841, 2.4796, 2.9792, 2.6464, 2.7580], device='cuda:6'), covar=tensor([0.0850, 0.2902, 0.2108, 0.0304, 0.3257, 0.1947, 0.2517, 0.2739], device='cuda:6'), in_proj_covar=tensor([0.0364, 0.0392, 0.0329, 0.0314, 0.0410, 0.0452, 0.0358, 0.0458], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:28:05,473 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.686e+02 3.186e+02 3.899e+02 4.864e+02 8.341e+02, threshold=7.798e+02, percent-clipped=4.0 2023-04-29 15:28:10,192 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118269.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:28:42,135 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118289.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:29:01,033 INFO [train.py:904] (6/8) Epoch 12, batch 6650, loss[loss=0.2073, simple_loss=0.2912, pruned_loss=0.06169, over 15483.00 frames. ], tot_loss[loss=0.2212, simple_loss=0.3039, pruned_loss=0.06925, over 3092433.12 frames. ], batch size: 191, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:29:44,391 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:29:45,601 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118330.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:29:51,830 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9605, 2.6024, 2.6234, 4.8181, 2.4609, 3.1044, 2.7298, 2.8632], device='cuda:6'), covar=tensor([0.0817, 0.3131, 0.2098, 0.0285, 0.3444, 0.1918, 0.2708, 0.2600], device='cuda:6'), in_proj_covar=tensor([0.0363, 0.0392, 0.0329, 0.0315, 0.0410, 0.0452, 0.0359, 0.0458], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:30:16,335 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118350.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:30:18,843 INFO [train.py:904] (6/8) Epoch 12, batch 6700, loss[loss=0.2054, simple_loss=0.2913, pruned_loss=0.05968, over 16654.00 frames. ], tot_loss[loss=0.2198, simple_loss=0.3021, pruned_loss=0.06871, over 3120243.13 frames. ], batch size: 89, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:30:22,659 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.24 vs. limit=5.0 2023-04-29 15:30:39,918 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.931e+02 3.120e+02 3.560e+02 4.149e+02 6.786e+02, threshold=7.119e+02, percent-clipped=0.0 2023-04-29 15:31:17,597 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118390.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:31:27,356 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118396.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:31:35,206 INFO [train.py:904] (6/8) Epoch 12, batch 6750, loss[loss=0.22, simple_loss=0.2957, pruned_loss=0.07212, over 16679.00 frames. ], tot_loss[loss=0.2193, simple_loss=0.3007, pruned_loss=0.06893, over 3103070.17 frames. ], batch size: 134, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:31:57,883 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5179, 4.2665, 4.5352, 4.7127, 4.8572, 4.3660, 4.8643, 4.8415], device='cuda:6'), covar=tensor([0.1526, 0.1251, 0.1702, 0.0667, 0.0541, 0.1042, 0.0518, 0.0597], device='cuda:6'), in_proj_covar=tensor([0.0528, 0.0664, 0.0796, 0.0672, 0.0507, 0.0522, 0.0535, 0.0608], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:32:11,575 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 15:32:49,915 INFO [train.py:904] (6/8) Epoch 12, batch 6800, loss[loss=0.2451, simple_loss=0.3141, pruned_loss=0.08805, over 11794.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.3001, pruned_loss=0.06815, over 3114820.30 frames. ], batch size: 246, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:33:11,648 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.917e+02 3.071e+02 3.777e+02 4.759e+02 7.416e+02, threshold=7.554e+02, percent-clipped=1.0 2023-04-29 15:34:04,798 INFO [train.py:904] (6/8) Epoch 12, batch 6850, loss[loss=0.238, simple_loss=0.3305, pruned_loss=0.07275, over 16902.00 frames. ], tot_loss[loss=0.219, simple_loss=0.3016, pruned_loss=0.06825, over 3134143.86 frames. ], batch size: 116, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:34:32,503 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1150, 5.6388, 5.8453, 5.5308, 5.4710, 6.1856, 5.6126, 5.3929], device='cuda:6'), covar=tensor([0.0760, 0.1671, 0.1977, 0.1948, 0.2598, 0.0868, 0.1521, 0.2442], device='cuda:6'), in_proj_covar=tensor([0.0357, 0.0492, 0.0541, 0.0426, 0.0577, 0.0566, 0.0428, 0.0581], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 15:34:36,246 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118524.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 15:35:15,878 INFO [train.py:904] (6/8) Epoch 12, batch 6900, loss[loss=0.2885, simple_loss=0.3443, pruned_loss=0.1163, over 11549.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.3041, pruned_loss=0.06785, over 3138348.08 frames. ], batch size: 247, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:35:36,844 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.921e+02 3.007e+02 3.634e+02 4.757e+02 1.105e+03, threshold=7.268e+02, percent-clipped=1.0 2023-04-29 15:36:30,544 INFO [train.py:904] (6/8) Epoch 12, batch 6950, loss[loss=0.198, simple_loss=0.2874, pruned_loss=0.05431, over 16844.00 frames. ], tot_loss[loss=0.2238, simple_loss=0.3066, pruned_loss=0.07052, over 3105157.97 frames. ], batch size: 102, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:36:56,709 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0174, 1.9346, 2.1720, 3.4846, 1.9715, 2.2430, 2.1138, 2.0341], device='cuda:6'), covar=tensor([0.1025, 0.3374, 0.2313, 0.0488, 0.4108, 0.2374, 0.3075, 0.3295], device='cuda:6'), in_proj_covar=tensor([0.0361, 0.0390, 0.0328, 0.0313, 0.0409, 0.0449, 0.0358, 0.0457], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:37:04,639 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118625.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:37:15,361 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118631.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:37:23,079 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7861, 2.6791, 2.1284, 2.4501, 3.1294, 2.7608, 3.4430, 3.3565], device='cuda:6'), covar=tensor([0.0073, 0.0334, 0.0471, 0.0378, 0.0188, 0.0315, 0.0195, 0.0189], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0203, 0.0200, 0.0199, 0.0205, 0.0202, 0.0208, 0.0196], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:37:35,515 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118645.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:37:44,792 INFO [train.py:904] (6/8) Epoch 12, batch 7000, loss[loss=0.2154, simple_loss=0.3122, pruned_loss=0.05931, over 17019.00 frames. ], tot_loss[loss=0.222, simple_loss=0.3059, pruned_loss=0.0691, over 3114656.87 frames. ], batch size: 55, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:38:05,443 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 2.982e+02 3.795e+02 4.664e+02 8.323e+02, threshold=7.591e+02, percent-clipped=2.0 2023-04-29 15:38:33,852 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118685.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:38:44,689 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118692.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:38:50,375 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118696.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:38:59,535 INFO [train.py:904] (6/8) Epoch 12, batch 7050, loss[loss=0.202, simple_loss=0.2973, pruned_loss=0.05333, over 17192.00 frames. ], tot_loss[loss=0.2227, simple_loss=0.3072, pruned_loss=0.06909, over 3122179.74 frames. ], batch size: 46, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:39:16,806 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8663, 3.2499, 3.1445, 2.0815, 2.9924, 3.2034, 3.0152, 1.7633], device='cuda:6'), covar=tensor([0.0489, 0.0034, 0.0043, 0.0358, 0.0075, 0.0080, 0.0071, 0.0402], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0068, 0.0071, 0.0126, 0.0080, 0.0092, 0.0080, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 15:40:01,790 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=118744.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:40:14,607 INFO [train.py:904] (6/8) Epoch 12, batch 7100, loss[loss=0.227, simple_loss=0.3112, pruned_loss=0.07142, over 16389.00 frames. ], tot_loss[loss=0.2217, simple_loss=0.3054, pruned_loss=0.06899, over 3097752.32 frames. ], batch size: 146, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:40:36,861 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.757e+02 3.120e+02 3.747e+02 4.649e+02 1.761e+03, threshold=7.495e+02, percent-clipped=5.0 2023-04-29 15:41:29,296 INFO [train.py:904] (6/8) Epoch 12, batch 7150, loss[loss=0.2729, simple_loss=0.3387, pruned_loss=0.1035, over 11143.00 frames. ], tot_loss[loss=0.2207, simple_loss=0.3036, pruned_loss=0.06889, over 3094444.27 frames. ], batch size: 247, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:42:01,983 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118824.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:42:16,963 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-04-29 15:42:29,978 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-04-29 15:42:41,519 INFO [train.py:904] (6/8) Epoch 12, batch 7200, loss[loss=0.2054, simple_loss=0.2922, pruned_loss=0.05928, over 15361.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.3014, pruned_loss=0.06746, over 3068862.03 frames. ], batch size: 190, lr: 5.64e-03, grad_scale: 8.0 2023-04-29 15:43:03,912 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.822e+02 2.566e+02 3.185e+02 3.996e+02 6.189e+02, threshold=6.369e+02, percent-clipped=0.0 2023-04-29 15:43:12,503 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=118872.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:43:42,436 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5276, 4.3909, 4.5670, 4.7453, 4.9144, 4.4315, 4.8859, 4.8833], device='cuda:6'), covar=tensor([0.1531, 0.1168, 0.1445, 0.0612, 0.0460, 0.0830, 0.0471, 0.0530], device='cuda:6'), in_proj_covar=tensor([0.0518, 0.0649, 0.0779, 0.0661, 0.0497, 0.0511, 0.0526, 0.0595], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:44:00,069 INFO [train.py:904] (6/8) Epoch 12, batch 7250, loss[loss=0.2021, simple_loss=0.2852, pruned_loss=0.05951, over 15320.00 frames. ], tot_loss[loss=0.2153, simple_loss=0.2989, pruned_loss=0.06585, over 3062243.63 frames. ], batch size: 190, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:44:35,588 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118925.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:45:04,527 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118945.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:45:12,490 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1116, 5.4042, 5.1653, 5.1771, 4.8089, 4.7728, 4.8632, 5.4511], device='cuda:6'), covar=tensor([0.0984, 0.0748, 0.0922, 0.0676, 0.0785, 0.0802, 0.0963, 0.0879], device='cuda:6'), in_proj_covar=tensor([0.0551, 0.0686, 0.0568, 0.0484, 0.0439, 0.0451, 0.0571, 0.0532], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:45:15,112 INFO [train.py:904] (6/8) Epoch 12, batch 7300, loss[loss=0.2503, simple_loss=0.316, pruned_loss=0.09228, over 11573.00 frames. ], tot_loss[loss=0.215, simple_loss=0.2985, pruned_loss=0.06569, over 3079681.17 frames. ], batch size: 248, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:45:36,388 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.980e+02 3.034e+02 3.599e+02 4.380e+02 7.583e+02, threshold=7.199e+02, percent-clipped=5.0 2023-04-29 15:45:45,784 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=118973.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:46:03,523 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118985.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:46:06,454 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118987.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:46:15,624 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=118993.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:46:28,511 INFO [train.py:904] (6/8) Epoch 12, batch 7350, loss[loss=0.2101, simple_loss=0.295, pruned_loss=0.06258, over 16810.00 frames. ], tot_loss[loss=0.2162, simple_loss=0.299, pruned_loss=0.0667, over 3050047.96 frames. ], batch size: 42, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:46:57,943 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0378, 4.0580, 4.4970, 4.4811, 4.4716, 4.1725, 4.2224, 4.0901], device='cuda:6'), covar=tensor([0.0334, 0.0562, 0.0371, 0.0376, 0.0496, 0.0404, 0.0852, 0.0511], device='cuda:6'), in_proj_covar=tensor([0.0336, 0.0351, 0.0353, 0.0332, 0.0398, 0.0373, 0.0469, 0.0300], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 15:47:14,443 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=119033.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:47:22,334 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8560, 2.1129, 2.4211, 3.1869, 2.2192, 2.3389, 2.3149, 2.2042], device='cuda:6'), covar=tensor([0.1000, 0.2885, 0.1836, 0.0555, 0.3431, 0.1848, 0.2610, 0.2971], device='cuda:6'), in_proj_covar=tensor([0.0364, 0.0394, 0.0329, 0.0316, 0.0414, 0.0454, 0.0360, 0.0459], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:47:42,951 INFO [train.py:904] (6/8) Epoch 12, batch 7400, loss[loss=0.2748, simple_loss=0.3318, pruned_loss=0.1089, over 11351.00 frames. ], tot_loss[loss=0.2169, simple_loss=0.2998, pruned_loss=0.06706, over 3076458.67 frames. ], batch size: 247, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:47:59,576 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8911, 5.1274, 4.8840, 4.9226, 4.6544, 4.5857, 4.6841, 5.1933], device='cuda:6'), covar=tensor([0.0872, 0.0815, 0.0972, 0.0743, 0.0756, 0.0884, 0.0915, 0.0790], device='cuda:6'), in_proj_covar=tensor([0.0549, 0.0686, 0.0567, 0.0484, 0.0438, 0.0453, 0.0571, 0.0533], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:48:06,306 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 3.304e+02 4.015e+02 4.796e+02 1.422e+03, threshold=8.030e+02, percent-clipped=7.0 2023-04-29 15:48:10,157 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-29 15:48:19,123 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119075.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:48:43,293 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6671, 2.4723, 2.3409, 3.5594, 2.6095, 3.7328, 1.4153, 2.7802], device='cuda:6'), covar=tensor([0.1292, 0.0665, 0.1147, 0.0144, 0.0204, 0.0383, 0.1544, 0.0718], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0162, 0.0182, 0.0150, 0.0199, 0.0208, 0.0183, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 15:48:47,046 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119093.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:49:01,209 INFO [train.py:904] (6/8) Epoch 12, batch 7450, loss[loss=0.2359, simple_loss=0.3157, pruned_loss=0.07809, over 16634.00 frames. ], tot_loss[loss=0.219, simple_loss=0.3012, pruned_loss=0.06844, over 3074000.45 frames. ], batch size: 62, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:49:55,978 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119136.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:50:20,477 INFO [train.py:904] (6/8) Epoch 12, batch 7500, loss[loss=0.2007, simple_loss=0.2894, pruned_loss=0.05597, over 16850.00 frames. ], tot_loss[loss=0.2191, simple_loss=0.3017, pruned_loss=0.06829, over 3063748.70 frames. ], batch size: 102, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:50:24,099 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119154.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 15:50:42,266 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.862e+02 2.952e+02 3.435e+02 4.438e+02 7.679e+02, threshold=6.870e+02, percent-clipped=0.0 2023-04-29 15:51:23,819 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.89 vs. limit=5.0 2023-04-29 15:51:27,804 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4693, 2.6163, 2.0508, 2.4400, 2.9761, 2.6231, 3.2659, 3.2339], device='cuda:6'), covar=tensor([0.0066, 0.0281, 0.0399, 0.0332, 0.0183, 0.0291, 0.0128, 0.0168], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0202, 0.0197, 0.0197, 0.0202, 0.0200, 0.0204, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:51:35,616 INFO [train.py:904] (6/8) Epoch 12, batch 7550, loss[loss=0.2549, simple_loss=0.3088, pruned_loss=0.1006, over 11236.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.3014, pruned_loss=0.06917, over 3035833.50 frames. ], batch size: 247, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:52:50,119 INFO [train.py:904] (6/8) Epoch 12, batch 7600, loss[loss=0.2392, simple_loss=0.3084, pruned_loss=0.08503, over 11301.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.3005, pruned_loss=0.06892, over 3041388.47 frames. ], batch size: 246, lr: 5.64e-03, grad_scale: 8.0 2023-04-29 15:53:04,742 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-29 15:53:09,824 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1859, 2.3549, 1.8625, 2.1181, 2.7389, 2.4079, 2.9341, 2.9904], device='cuda:6'), covar=tensor([0.0084, 0.0360, 0.0476, 0.0402, 0.0216, 0.0335, 0.0231, 0.0212], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0205, 0.0201, 0.0200, 0.0206, 0.0203, 0.0209, 0.0195], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:53:12,404 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.101e+02 2.984e+02 3.563e+02 4.291e+02 1.017e+03, threshold=7.126e+02, percent-clipped=2.0 2023-04-29 15:53:30,514 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5028, 4.6009, 4.8015, 4.6534, 4.6831, 5.1682, 4.7421, 4.4853], device='cuda:6'), covar=tensor([0.1182, 0.1791, 0.2006, 0.1766, 0.2287, 0.0981, 0.1449, 0.2344], device='cuda:6'), in_proj_covar=tensor([0.0359, 0.0500, 0.0546, 0.0434, 0.0582, 0.0569, 0.0431, 0.0587], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 15:53:43,843 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119287.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:54:05,863 INFO [train.py:904] (6/8) Epoch 12, batch 7650, loss[loss=0.2186, simple_loss=0.3038, pruned_loss=0.06669, over 16774.00 frames. ], tot_loss[loss=0.2193, simple_loss=0.3006, pruned_loss=0.06901, over 3060269.32 frames. ], batch size: 124, lr: 5.63e-03, grad_scale: 8.0 2023-04-29 15:54:35,645 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3426, 2.0935, 2.2409, 4.0070, 2.0480, 2.5063, 2.2083, 2.2502], device='cuda:6'), covar=tensor([0.1017, 0.3114, 0.2408, 0.0426, 0.3897, 0.2294, 0.3090, 0.3223], device='cuda:6'), in_proj_covar=tensor([0.0363, 0.0394, 0.0330, 0.0317, 0.0413, 0.0452, 0.0359, 0.0458], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 15:54:55,654 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=119335.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:55:20,164 INFO [train.py:904] (6/8) Epoch 12, batch 7700, loss[loss=0.2074, simple_loss=0.2931, pruned_loss=0.06082, over 16481.00 frames. ], tot_loss[loss=0.2198, simple_loss=0.301, pruned_loss=0.06933, over 3056157.21 frames. ], batch size: 68, lr: 5.63e-03, grad_scale: 8.0 2023-04-29 15:55:42,613 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.234e+02 3.396e+02 4.418e+02 5.495e+02 1.012e+03, threshold=8.835e+02, percent-clipped=5.0 2023-04-29 15:56:19,972 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-29 15:56:36,117 INFO [train.py:904] (6/8) Epoch 12, batch 7750, loss[loss=0.1953, simple_loss=0.2852, pruned_loss=0.05266, over 16718.00 frames. ], tot_loss[loss=0.2195, simple_loss=0.301, pruned_loss=0.06898, over 3068590.87 frames. ], batch size: 83, lr: 5.63e-03, grad_scale: 8.0 2023-04-29 15:56:53,014 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1092, 3.1750, 1.7631, 3.4678, 2.3655, 3.4819, 2.0161, 2.5045], device='cuda:6'), covar=tensor([0.0267, 0.0372, 0.1771, 0.0169, 0.0793, 0.0573, 0.1452, 0.0783], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0164, 0.0190, 0.0130, 0.0168, 0.0204, 0.0196, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 15:57:10,054 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.74 vs. limit=5.0 2023-04-29 15:57:18,794 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119431.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:57:45,520 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119449.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 15:57:48,706 INFO [train.py:904] (6/8) Epoch 12, batch 7800, loss[loss=0.2066, simple_loss=0.2953, pruned_loss=0.059, over 16463.00 frames. ], tot_loss[loss=0.2202, simple_loss=0.3015, pruned_loss=0.06944, over 3067795.12 frames. ], batch size: 68, lr: 5.63e-03, grad_scale: 8.0 2023-04-29 15:58:11,186 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.951e+02 3.141e+02 3.879e+02 4.510e+02 7.221e+02, threshold=7.757e+02, percent-clipped=0.0 2023-04-29 15:59:02,090 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-29 15:59:04,882 INFO [train.py:904] (6/8) Epoch 12, batch 7850, loss[loss=0.2304, simple_loss=0.2974, pruned_loss=0.0817, over 11578.00 frames. ], tot_loss[loss=0.2196, simple_loss=0.3019, pruned_loss=0.06869, over 3083810.02 frames. ], batch size: 247, lr: 5.63e-03, grad_scale: 4.0 2023-04-29 15:59:19,846 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2902, 3.4477, 1.8228, 3.7091, 2.5491, 3.7046, 2.0758, 2.6256], device='cuda:6'), covar=tensor([0.0252, 0.0338, 0.1769, 0.0171, 0.0815, 0.0525, 0.1506, 0.0775], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0163, 0.0190, 0.0130, 0.0168, 0.0205, 0.0197, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 15:59:24,360 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119514.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:00:21,536 INFO [train.py:904] (6/8) Epoch 12, batch 7900, loss[loss=0.2022, simple_loss=0.2923, pruned_loss=0.05608, over 16782.00 frames. ], tot_loss[loss=0.217, simple_loss=0.2999, pruned_loss=0.06704, over 3105493.94 frames. ], batch size: 83, lr: 5.63e-03, grad_scale: 4.0 2023-04-29 16:00:45,732 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.963e+02 2.838e+02 3.537e+02 4.302e+02 9.445e+02, threshold=7.074e+02, percent-clipped=1.0 2023-04-29 16:00:56,313 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119575.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:01:38,577 INFO [train.py:904] (6/8) Epoch 12, batch 7950, loss[loss=0.2209, simple_loss=0.304, pruned_loss=0.06887, over 16103.00 frames. ], tot_loss[loss=0.2184, simple_loss=0.3007, pruned_loss=0.06807, over 3096499.63 frames. ], batch size: 165, lr: 5.63e-03, grad_scale: 4.0 2023-04-29 16:02:53,353 INFO [train.py:904] (6/8) Epoch 12, batch 8000, loss[loss=0.2157, simple_loss=0.3079, pruned_loss=0.06177, over 17130.00 frames. ], tot_loss[loss=0.219, simple_loss=0.3014, pruned_loss=0.06831, over 3106856.58 frames. ], batch size: 40, lr: 5.63e-03, grad_scale: 8.0 2023-04-29 16:03:01,147 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 16:03:12,751 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119665.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:03:17,120 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.726e+02 2.898e+02 3.526e+02 5.291e+02 1.062e+03, threshold=7.052e+02, percent-clipped=9.0 2023-04-29 16:04:04,462 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2934, 3.3301, 3.5811, 1.7284, 3.7417, 3.7402, 2.8083, 2.7723], device='cuda:6'), covar=tensor([0.0762, 0.0179, 0.0145, 0.1059, 0.0052, 0.0137, 0.0367, 0.0405], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0099, 0.0086, 0.0136, 0.0068, 0.0104, 0.0119, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 16:04:07,955 INFO [train.py:904] (6/8) Epoch 12, batch 8050, loss[loss=0.2391, simple_loss=0.3064, pruned_loss=0.08587, over 11593.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.3006, pruned_loss=0.06778, over 3104356.82 frames. ], batch size: 247, lr: 5.62e-03, grad_scale: 4.0 2023-04-29 16:04:42,908 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119726.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:04:49,833 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119731.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:05:17,337 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119749.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 16:05:21,187 INFO [train.py:904] (6/8) Epoch 12, batch 8100, loss[loss=0.204, simple_loss=0.288, pruned_loss=0.05994, over 16486.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.3003, pruned_loss=0.06718, over 3114824.50 frames. ], batch size: 68, lr: 5.62e-03, grad_scale: 4.0 2023-04-29 16:05:26,544 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3957, 2.6174, 2.0066, 2.3863, 2.9474, 2.7361, 3.1521, 3.2079], device='cuda:6'), covar=tensor([0.0082, 0.0288, 0.0431, 0.0291, 0.0194, 0.0232, 0.0184, 0.0165], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0204, 0.0200, 0.0200, 0.0205, 0.0202, 0.0208, 0.0194], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:05:28,343 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9114, 2.7378, 2.7326, 2.0412, 2.6391, 2.1365, 2.7581, 2.9419], device='cuda:6'), covar=tensor([0.0257, 0.0736, 0.0507, 0.1700, 0.0746, 0.0871, 0.0508, 0.0630], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0144, 0.0158, 0.0142, 0.0135, 0.0124, 0.0136, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 16:05:47,745 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.274e+02 3.055e+02 3.528e+02 4.290e+02 7.598e+02, threshold=7.056e+02, percent-clipped=1.0 2023-04-29 16:06:01,238 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=119779.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:06:22,802 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2018, 1.9854, 1.6268, 1.7691, 2.2023, 1.9598, 2.0518, 2.3373], device='cuda:6'), covar=tensor([0.0146, 0.0290, 0.0392, 0.0331, 0.0176, 0.0269, 0.0153, 0.0202], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0203, 0.0198, 0.0198, 0.0204, 0.0201, 0.0206, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:06:28,073 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=119797.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:06:34,772 INFO [train.py:904] (6/8) Epoch 12, batch 8150, loss[loss=0.2213, simple_loss=0.288, pruned_loss=0.07728, over 11726.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.2983, pruned_loss=0.06694, over 3090730.01 frames. ], batch size: 247, lr: 5.62e-03, grad_scale: 2.0 2023-04-29 16:07:50,608 INFO [train.py:904] (6/8) Epoch 12, batch 8200, loss[loss=0.2106, simple_loss=0.2975, pruned_loss=0.06183, over 16683.00 frames. ], tot_loss[loss=0.2133, simple_loss=0.2954, pruned_loss=0.06556, over 3101532.09 frames. ], batch size: 134, lr: 5.62e-03, grad_scale: 2.0 2023-04-29 16:08:12,188 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5677, 3.9309, 3.9232, 2.0496, 3.2404, 2.5592, 3.7800, 3.9200], device='cuda:6'), covar=tensor([0.0234, 0.0602, 0.0509, 0.1976, 0.0705, 0.0918, 0.0635, 0.0829], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0145, 0.0160, 0.0144, 0.0137, 0.0125, 0.0138, 0.0156], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 16:08:18,226 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.831e+02 3.028e+02 3.797e+02 4.737e+02 9.915e+02, threshold=7.593e+02, percent-clipped=4.0 2023-04-29 16:08:19,314 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119870.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:09:09,079 INFO [train.py:904] (6/8) Epoch 12, batch 8250, loss[loss=0.1775, simple_loss=0.278, pruned_loss=0.03849, over 16824.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2944, pruned_loss=0.0635, over 3074493.10 frames. ], batch size: 102, lr: 5.62e-03, grad_scale: 2.0 2023-04-29 16:09:57,723 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.90 vs. limit=5.0 2023-04-29 16:10:00,072 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1503, 3.2555, 1.9759, 3.5011, 2.4526, 3.5362, 2.0039, 2.6300], device='cuda:6'), covar=tensor([0.0228, 0.0290, 0.1339, 0.0158, 0.0726, 0.0372, 0.1423, 0.0608], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0160, 0.0188, 0.0129, 0.0165, 0.0202, 0.0195, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-29 16:10:28,023 INFO [train.py:904] (6/8) Epoch 12, batch 8300, loss[loss=0.2007, simple_loss=0.2896, pruned_loss=0.05588, over 16843.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2922, pruned_loss=0.06073, over 3086187.17 frames. ], batch size: 116, lr: 5.62e-03, grad_scale: 2.0 2023-04-29 16:10:57,566 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.672e+02 2.352e+02 2.901e+02 3.420e+02 5.863e+02, threshold=5.801e+02, percent-clipped=0.0 2023-04-29 16:11:52,964 INFO [train.py:904] (6/8) Epoch 12, batch 8350, loss[loss=0.1923, simple_loss=0.2918, pruned_loss=0.04639, over 16879.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2918, pruned_loss=0.05877, over 3086522.37 frames. ], batch size: 96, lr: 5.62e-03, grad_scale: 2.0 2023-04-29 16:12:24,538 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=120021.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:13:14,333 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-29 16:13:14,449 INFO [train.py:904] (6/8) Epoch 12, batch 8400, loss[loss=0.1828, simple_loss=0.2722, pruned_loss=0.04675, over 15344.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2883, pruned_loss=0.05622, over 3081517.92 frames. ], batch size: 190, lr: 5.62e-03, grad_scale: 4.0 2023-04-29 16:13:42,956 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.720e+02 2.479e+02 2.965e+02 3.428e+02 6.852e+02, threshold=5.930e+02, percent-clipped=2.0 2023-04-29 16:13:54,730 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.37 vs. limit=5.0 2023-04-29 16:14:31,500 INFO [train.py:904] (6/8) Epoch 12, batch 8450, loss[loss=0.1927, simple_loss=0.274, pruned_loss=0.05565, over 12680.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2863, pruned_loss=0.05445, over 3080427.25 frames. ], batch size: 246, lr: 5.62e-03, grad_scale: 4.0 2023-04-29 16:15:03,176 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.17 vs. limit=5.0 2023-04-29 16:15:07,167 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0236, 2.2750, 1.8897, 1.9609, 2.6789, 2.3285, 2.7818, 2.9008], device='cuda:6'), covar=tensor([0.0121, 0.0319, 0.0406, 0.0417, 0.0220, 0.0309, 0.0176, 0.0196], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0200, 0.0195, 0.0195, 0.0201, 0.0198, 0.0201, 0.0189], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:15:19,925 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-29 16:15:23,482 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-29 16:15:42,523 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2023-04-29 16:15:47,447 INFO [train.py:904] (6/8) Epoch 12, batch 8500, loss[loss=0.1868, simple_loss=0.271, pruned_loss=0.0513, over 16479.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2824, pruned_loss=0.05171, over 3079733.31 frames. ], batch size: 68, lr: 5.61e-03, grad_scale: 4.0 2023-04-29 16:16:15,429 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.656e+02 2.373e+02 2.767e+02 3.491e+02 7.213e+02, threshold=5.534e+02, percent-clipped=3.0 2023-04-29 16:16:15,939 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=120170.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:16:57,997 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4492, 4.5628, 4.7197, 4.5534, 4.5540, 5.1328, 4.7391, 4.4252], device='cuda:6'), covar=tensor([0.1246, 0.1889, 0.2178, 0.2284, 0.2618, 0.1117, 0.1537, 0.2792], device='cuda:6'), in_proj_covar=tensor([0.0344, 0.0474, 0.0524, 0.0413, 0.0555, 0.0551, 0.0418, 0.0561], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 16:17:07,726 INFO [train.py:904] (6/8) Epoch 12, batch 8550, loss[loss=0.2177, simple_loss=0.3088, pruned_loss=0.06327, over 16853.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2804, pruned_loss=0.05098, over 3060777.45 frames. ], batch size: 116, lr: 5.61e-03, grad_scale: 4.0 2023-04-29 16:17:37,325 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=120218.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:18:12,568 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2950, 3.3873, 3.6316, 3.6396, 3.6325, 3.4520, 3.4957, 3.4576], device='cuda:6'), covar=tensor([0.0360, 0.0686, 0.0469, 0.0408, 0.0462, 0.0444, 0.0714, 0.0525], device='cuda:6'), in_proj_covar=tensor([0.0333, 0.0352, 0.0346, 0.0330, 0.0398, 0.0372, 0.0465, 0.0300], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 16:18:28,168 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-29 16:18:41,889 INFO [train.py:904] (6/8) Epoch 12, batch 8600, loss[loss=0.1927, simple_loss=0.2912, pruned_loss=0.0471, over 15360.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2806, pruned_loss=0.04982, over 3058538.80 frames. ], batch size: 191, lr: 5.61e-03, grad_scale: 4.0 2023-04-29 16:18:55,745 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9038, 3.7832, 3.9902, 4.0858, 4.2177, 3.8048, 4.1762, 4.2296], device='cuda:6'), covar=tensor([0.1491, 0.1069, 0.1396, 0.0722, 0.0555, 0.1345, 0.0658, 0.0633], device='cuda:6'), in_proj_covar=tensor([0.0509, 0.0639, 0.0765, 0.0653, 0.0493, 0.0503, 0.0524, 0.0594], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:19:02,482 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3884, 4.6900, 4.4707, 4.4545, 4.1573, 4.1616, 4.2315, 4.7001], device='cuda:6'), covar=tensor([0.0982, 0.0875, 0.0973, 0.0762, 0.0782, 0.1327, 0.1022, 0.0918], device='cuda:6'), in_proj_covar=tensor([0.0532, 0.0655, 0.0544, 0.0467, 0.0418, 0.0436, 0.0550, 0.0511], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:19:19,442 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.485e+02 2.557e+02 3.073e+02 4.030e+02 6.956e+02, threshold=6.147e+02, percent-clipped=5.0 2023-04-29 16:20:19,200 INFO [train.py:904] (6/8) Epoch 12, batch 8650, loss[loss=0.1628, simple_loss=0.2675, pruned_loss=0.02899, over 16830.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2788, pruned_loss=0.04844, over 3060675.58 frames. ], batch size: 102, lr: 5.61e-03, grad_scale: 4.0 2023-04-29 16:20:21,364 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.13 vs. limit=5.0 2023-04-29 16:21:01,240 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=120321.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:22:02,145 INFO [train.py:904] (6/8) Epoch 12, batch 8700, loss[loss=0.1774, simple_loss=0.268, pruned_loss=0.0434, over 16847.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2756, pruned_loss=0.04711, over 3050040.35 frames. ], batch size: 124, lr: 5.61e-03, grad_scale: 4.0 2023-04-29 16:22:14,774 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-29 16:22:18,320 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0287, 4.8282, 5.0123, 5.2264, 5.3767, 4.7617, 5.3599, 5.3543], device='cuda:6'), covar=tensor([0.1544, 0.1046, 0.1547, 0.0602, 0.0507, 0.0745, 0.0510, 0.0590], device='cuda:6'), in_proj_covar=tensor([0.0507, 0.0637, 0.0765, 0.0649, 0.0490, 0.0502, 0.0520, 0.0592], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:22:18,788 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-29 16:22:33,633 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=120369.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:22:34,421 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.439e+02 2.303e+02 2.810e+02 3.260e+02 5.191e+02, threshold=5.620e+02, percent-clipped=0.0 2023-04-29 16:23:24,351 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7361, 3.0864, 2.8644, 4.4283, 3.1631, 4.2628, 1.6792, 3.2718], device='cuda:6'), covar=tensor([0.1336, 0.0550, 0.0939, 0.0134, 0.0186, 0.0295, 0.1419, 0.0576], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0159, 0.0180, 0.0146, 0.0192, 0.0205, 0.0182, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 16:23:36,530 INFO [train.py:904] (6/8) Epoch 12, batch 8750, loss[loss=0.2007, simple_loss=0.2956, pruned_loss=0.05293, over 16893.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2756, pruned_loss=0.04681, over 3047670.59 frames. ], batch size: 116, lr: 5.61e-03, grad_scale: 4.0 2023-04-29 16:23:58,073 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0676, 4.0605, 3.9202, 3.4161, 3.9644, 1.7002, 3.7745, 3.5663], device='cuda:6'), covar=tensor([0.0070, 0.0067, 0.0122, 0.0221, 0.0074, 0.2320, 0.0098, 0.0173], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0115, 0.0158, 0.0149, 0.0132, 0.0177, 0.0148, 0.0145], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:24:29,027 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1762, 4.3023, 4.0587, 3.8299, 3.7850, 4.1924, 3.9045, 3.9047], device='cuda:6'), covar=tensor([0.0549, 0.0363, 0.0256, 0.0267, 0.0787, 0.0341, 0.0656, 0.0563], device='cuda:6'), in_proj_covar=tensor([0.0236, 0.0315, 0.0278, 0.0258, 0.0294, 0.0298, 0.0195, 0.0324], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:25:17,388 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4944, 3.4655, 2.7429, 1.9881, 2.2447, 2.1768, 3.5640, 3.1604], device='cuda:6'), covar=tensor([0.2728, 0.0683, 0.1602, 0.2547, 0.2422, 0.1926, 0.0446, 0.1131], device='cuda:6'), in_proj_covar=tensor([0.0295, 0.0246, 0.0276, 0.0271, 0.0265, 0.0216, 0.0258, 0.0284], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:25:30,724 INFO [train.py:904] (6/8) Epoch 12, batch 8800, loss[loss=0.1811, simple_loss=0.2669, pruned_loss=0.04768, over 12824.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2745, pruned_loss=0.04605, over 3027977.92 frames. ], batch size: 248, lr: 5.61e-03, grad_scale: 8.0 2023-04-29 16:26:08,426 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.720e+02 2.538e+02 3.135e+02 3.629e+02 6.620e+02, threshold=6.271e+02, percent-clipped=2.0 2023-04-29 16:26:39,616 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-29 16:27:17,117 INFO [train.py:904] (6/8) Epoch 12, batch 8850, loss[loss=0.1912, simple_loss=0.3093, pruned_loss=0.03653, over 16945.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2775, pruned_loss=0.04542, over 3044793.12 frames. ], batch size: 96, lr: 5.61e-03, grad_scale: 8.0 2023-04-29 16:27:57,543 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1677, 2.9922, 3.1501, 1.5794, 3.2953, 3.4181, 2.6502, 2.6151], device='cuda:6'), covar=tensor([0.0781, 0.0236, 0.0154, 0.1218, 0.0077, 0.0119, 0.0442, 0.0441], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0096, 0.0083, 0.0132, 0.0065, 0.0100, 0.0116, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 16:27:57,927 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-29 16:29:04,467 INFO [train.py:904] (6/8) Epoch 12, batch 8900, loss[loss=0.192, simple_loss=0.2896, pruned_loss=0.04721, over 16890.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2773, pruned_loss=0.04431, over 3050721.55 frames. ], batch size: 116, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:29:37,104 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2375, 4.0155, 4.1258, 4.4190, 4.5631, 4.1608, 4.6050, 4.5468], device='cuda:6'), covar=tensor([0.1678, 0.1373, 0.2038, 0.0902, 0.0718, 0.1132, 0.0700, 0.0838], device='cuda:6'), in_proj_covar=tensor([0.0501, 0.0632, 0.0751, 0.0642, 0.0485, 0.0496, 0.0510, 0.0584], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:29:39,307 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.788e+02 2.488e+02 2.995e+02 3.537e+02 1.098e+03, threshold=5.991e+02, percent-clipped=1.0 2023-04-29 16:29:49,647 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0851, 4.1026, 4.4875, 4.4643, 4.4589, 4.1679, 4.1849, 4.1311], device='cuda:6'), covar=tensor([0.0290, 0.0592, 0.0386, 0.0435, 0.0460, 0.0380, 0.0864, 0.0386], device='cuda:6'), in_proj_covar=tensor([0.0321, 0.0336, 0.0334, 0.0318, 0.0382, 0.0358, 0.0445, 0.0288], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 16:31:10,991 INFO [train.py:904] (6/8) Epoch 12, batch 8950, loss[loss=0.1697, simple_loss=0.2619, pruned_loss=0.03875, over 16387.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.277, pruned_loss=0.04476, over 3064933.24 frames. ], batch size: 146, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:32:42,186 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6355, 4.7473, 4.9539, 4.7644, 4.7695, 5.3313, 4.8735, 4.5769], device='cuda:6'), covar=tensor([0.0966, 0.1804, 0.1837, 0.1946, 0.2500, 0.0958, 0.1440, 0.2582], device='cuda:6'), in_proj_covar=tensor([0.0337, 0.0469, 0.0513, 0.0403, 0.0546, 0.0542, 0.0410, 0.0546], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 16:33:00,364 INFO [train.py:904] (6/8) Epoch 12, batch 9000, loss[loss=0.1625, simple_loss=0.2534, pruned_loss=0.03574, over 16801.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2738, pruned_loss=0.04333, over 3076690.95 frames. ], batch size: 124, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:33:00,365 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 16:33:10,348 INFO [train.py:938] (6/8) Epoch 12, validation: loss=0.1532, simple_loss=0.2571, pruned_loss=0.02465, over 944034.00 frames. 2023-04-29 16:33:10,349 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 16:33:30,279 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-29 16:33:49,108 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.498e+02 2.337e+02 2.622e+02 3.241e+02 7.245e+02, threshold=5.244e+02, percent-clipped=4.0 2023-04-29 16:34:01,254 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=120676.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:34:54,273 INFO [train.py:904] (6/8) Epoch 12, batch 9050, loss[loss=0.1887, simple_loss=0.279, pruned_loss=0.04923, over 12728.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2748, pruned_loss=0.04394, over 3087374.97 frames. ], batch size: 246, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:35:21,860 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3650, 4.4871, 4.6419, 4.4652, 4.5131, 5.0075, 4.6085, 4.3192], device='cuda:6'), covar=tensor([0.1308, 0.1717, 0.1558, 0.2085, 0.2336, 0.0978, 0.1330, 0.2521], device='cuda:6'), in_proj_covar=tensor([0.0339, 0.0471, 0.0514, 0.0404, 0.0545, 0.0542, 0.0409, 0.0547], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 16:35:25,375 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6720, 2.7503, 1.7343, 2.8857, 2.1069, 2.8507, 2.0745, 2.4397], device='cuda:6'), covar=tensor([0.0249, 0.0346, 0.1329, 0.0238, 0.0655, 0.0489, 0.1172, 0.0542], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0157, 0.0183, 0.0125, 0.0164, 0.0196, 0.0192, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-29 16:36:04,415 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=120737.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:36:15,902 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-29 16:36:39,407 INFO [train.py:904] (6/8) Epoch 12, batch 9100, loss[loss=0.1742, simple_loss=0.2682, pruned_loss=0.04008, over 17166.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2746, pruned_loss=0.04435, over 3096002.29 frames. ], batch size: 46, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:37:15,444 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 2.513e+02 2.910e+02 3.503e+02 5.095e+02, threshold=5.819e+02, percent-clipped=0.0 2023-04-29 16:38:36,941 INFO [train.py:904] (6/8) Epoch 12, batch 9150, loss[loss=0.1709, simple_loss=0.2685, pruned_loss=0.03665, over 16824.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.2745, pruned_loss=0.04423, over 3077458.63 frames. ], batch size: 83, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:39:21,629 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7721, 3.7240, 3.8941, 3.9668, 4.0726, 3.6123, 4.0670, 4.0941], device='cuda:6'), covar=tensor([0.1236, 0.1047, 0.1089, 0.0634, 0.0502, 0.1693, 0.0568, 0.0678], device='cuda:6'), in_proj_covar=tensor([0.0498, 0.0629, 0.0749, 0.0639, 0.0482, 0.0494, 0.0510, 0.0581], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:40:21,513 INFO [train.py:904] (6/8) Epoch 12, batch 9200, loss[loss=0.1395, simple_loss=0.2262, pruned_loss=0.02645, over 11769.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2704, pruned_loss=0.04337, over 3064369.22 frames. ], batch size: 248, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:40:55,537 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.751e+02 2.400e+02 2.881e+02 3.588e+02 6.775e+02, threshold=5.761e+02, percent-clipped=4.0 2023-04-29 16:41:45,812 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0464, 3.0932, 1.8459, 3.3336, 2.2746, 3.2908, 2.1004, 2.6254], device='cuda:6'), covar=tensor([0.0270, 0.0372, 0.1526, 0.0192, 0.0795, 0.0603, 0.1418, 0.0638], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0156, 0.0181, 0.0125, 0.0162, 0.0194, 0.0192, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-29 16:42:00,503 INFO [train.py:904] (6/8) Epoch 12, batch 9250, loss[loss=0.1798, simple_loss=0.2687, pruned_loss=0.04545, over 16672.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.27, pruned_loss=0.04308, over 3078357.97 frames. ], batch size: 134, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:42:40,520 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-29 16:43:49,106 INFO [train.py:904] (6/8) Epoch 12, batch 9300, loss[loss=0.1482, simple_loss=0.2412, pruned_loss=0.02755, over 17138.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2682, pruned_loss=0.04245, over 3070637.93 frames. ], batch size: 40, lr: 5.60e-03, grad_scale: 4.0 2023-04-29 16:44:31,798 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.549e+02 3.003e+02 3.626e+02 5.994e+02, threshold=6.005e+02, percent-clipped=1.0 2023-04-29 16:45:32,147 INFO [train.py:904] (6/8) Epoch 12, batch 9350, loss[loss=0.2085, simple_loss=0.2966, pruned_loss=0.06025, over 16709.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2684, pruned_loss=0.04285, over 3076081.03 frames. ], batch size: 134, lr: 5.59e-03, grad_scale: 4.0 2023-04-29 16:45:35,605 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121003.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:46:33,082 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121032.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:47:13,259 INFO [train.py:904] (6/8) Epoch 12, batch 9400, loss[loss=0.159, simple_loss=0.2418, pruned_loss=0.03812, over 12439.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.268, pruned_loss=0.04274, over 3045496.44 frames. ], batch size: 247, lr: 5.59e-03, grad_scale: 4.0 2023-04-29 16:47:39,261 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121064.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:47:50,521 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.435e+02 2.897e+02 3.662e+02 7.614e+02, threshold=5.793e+02, percent-clipped=2.0 2023-04-29 16:48:35,434 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 16:48:55,032 INFO [train.py:904] (6/8) Epoch 12, batch 9450, loss[loss=0.196, simple_loss=0.2834, pruned_loss=0.05432, over 16928.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2695, pruned_loss=0.04299, over 3049504.04 frames. ], batch size: 116, lr: 5.59e-03, grad_scale: 4.0 2023-04-29 16:49:13,478 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.04 vs. limit=5.0 2023-04-29 16:50:34,751 INFO [train.py:904] (6/8) Epoch 12, batch 9500, loss[loss=0.1648, simple_loss=0.2607, pruned_loss=0.03444, over 12532.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2687, pruned_loss=0.04234, over 3063653.08 frames. ], batch size: 246, lr: 5.59e-03, grad_scale: 4.0 2023-04-29 16:51:01,902 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121164.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:51:13,620 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.296e+02 2.757e+02 3.587e+02 6.751e+02, threshold=5.515e+02, percent-clipped=2.0 2023-04-29 16:52:19,996 INFO [train.py:904] (6/8) Epoch 12, batch 9550, loss[loss=0.1666, simple_loss=0.2612, pruned_loss=0.03603, over 12688.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.268, pruned_loss=0.04211, over 3065654.32 frames. ], batch size: 248, lr: 5.59e-03, grad_scale: 4.0 2023-04-29 16:52:49,374 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0301, 1.4544, 1.8137, 2.1044, 2.1826, 2.2163, 1.6653, 2.2292], device='cuda:6'), covar=tensor([0.0216, 0.0397, 0.0242, 0.0226, 0.0259, 0.0178, 0.0379, 0.0108], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0170, 0.0155, 0.0156, 0.0167, 0.0120, 0.0168, 0.0113], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 16:53:08,748 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121225.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:53:13,022 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9794, 2.1032, 2.4057, 3.2336, 2.1634, 2.2559, 2.2745, 2.1598], device='cuda:6'), covar=tensor([0.0947, 0.3098, 0.2016, 0.0559, 0.3818, 0.2294, 0.2902, 0.3493], device='cuda:6'), in_proj_covar=tensor([0.0353, 0.0381, 0.0325, 0.0306, 0.0404, 0.0434, 0.0351, 0.0445], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:53:51,818 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4258, 4.4177, 4.2625, 3.8259, 4.2736, 1.6570, 4.0383, 4.0561], device='cuda:6'), covar=tensor([0.0079, 0.0074, 0.0152, 0.0232, 0.0089, 0.2415, 0.0123, 0.0200], device='cuda:6'), in_proj_covar=tensor([0.0126, 0.0114, 0.0156, 0.0144, 0.0131, 0.0177, 0.0146, 0.0143], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 16:53:59,925 INFO [train.py:904] (6/8) Epoch 12, batch 9600, loss[loss=0.1836, simple_loss=0.2749, pruned_loss=0.04612, over 16711.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2697, pruned_loss=0.04327, over 3056907.03 frames. ], batch size: 76, lr: 5.59e-03, grad_scale: 8.0 2023-04-29 16:54:35,188 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.521e+02 2.441e+02 3.037e+02 3.482e+02 7.490e+02, threshold=6.075e+02, percent-clipped=2.0 2023-04-29 16:55:12,853 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 16:55:32,764 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8615, 3.3972, 3.2925, 1.8749, 2.9258, 2.1620, 3.4467, 3.4272], device='cuda:6'), covar=tensor([0.0267, 0.0737, 0.0587, 0.2013, 0.0778, 0.1101, 0.0672, 0.0914], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0137, 0.0155, 0.0141, 0.0133, 0.0122, 0.0132, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 16:55:45,648 INFO [train.py:904] (6/8) Epoch 12, batch 9650, loss[loss=0.1854, simple_loss=0.2693, pruned_loss=0.05073, over 12034.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2718, pruned_loss=0.04347, over 3061307.21 frames. ], batch size: 247, lr: 5.59e-03, grad_scale: 8.0 2023-04-29 16:56:10,090 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121311.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 16:56:34,820 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-29 16:56:52,038 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121332.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:57:30,748 INFO [train.py:904] (6/8) Epoch 12, batch 9700, loss[loss=0.1782, simple_loss=0.2729, pruned_loss=0.04173, over 15348.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2715, pruned_loss=0.04348, over 3067139.25 frames. ], batch size: 191, lr: 5.59e-03, grad_scale: 8.0 2023-04-29 16:57:44,329 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121359.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:58:07,288 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.472e+02 3.175e+02 3.928e+02 9.920e+02, threshold=6.351e+02, percent-clipped=4.0 2023-04-29 16:58:11,377 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121372.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 16:58:19,888 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5467, 3.1959, 2.9645, 1.8336, 2.6208, 2.0602, 3.0518, 3.2135], device='cuda:6'), covar=tensor([0.0262, 0.0610, 0.0573, 0.1990, 0.0964, 0.1126, 0.0728, 0.0710], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0136, 0.0154, 0.0140, 0.0133, 0.0121, 0.0132, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 16:58:31,090 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=121380.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:59:14,333 INFO [train.py:904] (6/8) Epoch 12, batch 9750, loss[loss=0.1793, simple_loss=0.274, pruned_loss=0.04225, over 16830.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2703, pruned_loss=0.04357, over 3084654.74 frames. ], batch size: 124, lr: 5.59e-03, grad_scale: 8.0 2023-04-29 16:59:23,928 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121407.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:00:53,747 INFO [train.py:904] (6/8) Epoch 12, batch 9800, loss[loss=0.1638, simple_loss=0.26, pruned_loss=0.03379, over 16780.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2708, pruned_loss=0.04256, over 3097910.44 frames. ], batch size: 39, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:01:23,685 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-29 17:01:24,701 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121468.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:01:29,885 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.684e+02 2.372e+02 2.708e+02 3.035e+02 5.819e+02, threshold=5.415e+02, percent-clipped=0.0 2023-04-29 17:02:38,486 INFO [train.py:904] (6/8) Epoch 12, batch 9850, loss[loss=0.1795, simple_loss=0.2746, pruned_loss=0.04222, over 16335.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2715, pruned_loss=0.0421, over 3084318.54 frames. ], batch size: 146, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:03:17,658 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121520.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:03:33,224 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2794, 2.0273, 2.2180, 3.9084, 2.0316, 2.4847, 2.2378, 2.1844], device='cuda:6'), covar=tensor([0.0947, 0.3427, 0.2324, 0.0376, 0.3889, 0.2194, 0.3053, 0.3449], device='cuda:6'), in_proj_covar=tensor([0.0355, 0.0383, 0.0326, 0.0307, 0.0406, 0.0436, 0.0352, 0.0446], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:04:03,414 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5271, 1.9850, 1.6252, 1.7449, 2.3147, 1.9822, 2.2336, 2.4636], device='cuda:6'), covar=tensor([0.0108, 0.0362, 0.0459, 0.0452, 0.0215, 0.0329, 0.0163, 0.0208], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0204, 0.0198, 0.0197, 0.0203, 0.0201, 0.0198, 0.0188], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:04:30,193 INFO [train.py:904] (6/8) Epoch 12, batch 9900, loss[loss=0.1812, simple_loss=0.2807, pruned_loss=0.0409, over 16179.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2712, pruned_loss=0.04202, over 3063426.01 frames. ], batch size: 165, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:05:12,986 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.385e+02 2.886e+02 3.421e+02 6.283e+02, threshold=5.771e+02, percent-clipped=3.0 2023-04-29 17:06:26,640 INFO [train.py:904] (6/8) Epoch 12, batch 9950, loss[loss=0.1936, simple_loss=0.3011, pruned_loss=0.04308, over 16218.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2731, pruned_loss=0.04218, over 3063037.79 frames. ], batch size: 165, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:07:14,260 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6496, 1.6090, 2.0264, 2.5737, 2.4708, 2.7550, 2.0082, 2.7652], device='cuda:6'), covar=tensor([0.0129, 0.0399, 0.0259, 0.0213, 0.0230, 0.0146, 0.0315, 0.0116], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0171, 0.0157, 0.0159, 0.0169, 0.0123, 0.0171, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:6') 2023-04-29 17:08:27,934 INFO [train.py:904] (6/8) Epoch 12, batch 10000, loss[loss=0.1843, simple_loss=0.2792, pruned_loss=0.04474, over 16975.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2713, pruned_loss=0.04157, over 3088380.12 frames. ], batch size: 109, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:08:44,218 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121659.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:08:59,913 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121667.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 17:09:06,353 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 2.322e+02 2.719e+02 3.378e+02 7.215e+02, threshold=5.438e+02, percent-clipped=2.0 2023-04-29 17:10:10,826 INFO [train.py:904] (6/8) Epoch 12, batch 10050, loss[loss=0.178, simple_loss=0.2722, pruned_loss=0.04189, over 16485.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2714, pruned_loss=0.04163, over 3085254.44 frames. ], batch size: 68, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:10:21,424 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=121707.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:11:29,249 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1303, 5.4430, 5.2375, 5.1947, 4.9180, 4.8375, 4.7901, 5.5013], device='cuda:6'), covar=tensor([0.0988, 0.0764, 0.0814, 0.0631, 0.0702, 0.0725, 0.0961, 0.0791], device='cuda:6'), in_proj_covar=tensor([0.0516, 0.0650, 0.0529, 0.0457, 0.0414, 0.0426, 0.0538, 0.0498], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:11:42,157 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5985, 3.2161, 3.2466, 1.9747, 2.8760, 2.2258, 3.1457, 3.2879], device='cuda:6'), covar=tensor([0.0277, 0.0654, 0.0493, 0.1760, 0.0735, 0.0911, 0.0712, 0.0839], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0137, 0.0155, 0.0141, 0.0134, 0.0122, 0.0131, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 17:11:46,089 INFO [train.py:904] (6/8) Epoch 12, batch 10100, loss[loss=0.1696, simple_loss=0.2611, pruned_loss=0.03902, over 16734.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.2719, pruned_loss=0.04174, over 3098832.05 frames. ], batch size: 76, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:12:06,122 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121763.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:12:23,724 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.628e+02 2.328e+02 2.737e+02 3.349e+02 5.090e+02, threshold=5.474e+02, percent-clipped=0.0 2023-04-29 17:13:30,551 INFO [train.py:904] (6/8) Epoch 13, batch 0, loss[loss=0.2487, simple_loss=0.3175, pruned_loss=0.08994, over 16679.00 frames. ], tot_loss[loss=0.2487, simple_loss=0.3175, pruned_loss=0.08994, over 16679.00 frames. ], batch size: 134, lr: 5.36e-03, grad_scale: 8.0 2023-04-29 17:13:30,552 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 17:13:38,110 INFO [train.py:938] (6/8) Epoch 13, validation: loss=0.1523, simple_loss=0.2559, pruned_loss=0.02431, over 944034.00 frames. 2023-04-29 17:13:38,110 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 17:14:04,131 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121820.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:14:30,467 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5187, 4.9251, 4.2973, 4.7477, 4.4280, 4.2786, 4.5021, 4.9269], device='cuda:6'), covar=tensor([0.2305, 0.1853, 0.3037, 0.1422, 0.1761, 0.1922, 0.2142, 0.1972], device='cuda:6'), in_proj_covar=tensor([0.0523, 0.0658, 0.0536, 0.0463, 0.0420, 0.0431, 0.0547, 0.0504], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:14:49,498 INFO [train.py:904] (6/8) Epoch 13, batch 50, loss[loss=0.1908, simple_loss=0.2766, pruned_loss=0.05249, over 17143.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2826, pruned_loss=0.05764, over 744559.01 frames. ], batch size: 48, lr: 5.36e-03, grad_scale: 2.0 2023-04-29 17:15:01,503 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5812, 2.9860, 2.6107, 4.9457, 4.0114, 4.3561, 1.4412, 3.0831], device='cuda:6'), covar=tensor([0.1492, 0.0713, 0.1233, 0.0154, 0.0215, 0.0391, 0.1674, 0.0770], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0156, 0.0179, 0.0143, 0.0184, 0.0202, 0.0180, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 17:15:11,441 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=121868.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:15:18,817 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.864e+02 2.677e+02 3.187e+02 4.069e+02 9.250e+02, threshold=6.374e+02, percent-clipped=6.0 2023-04-29 17:15:58,269 INFO [train.py:904] (6/8) Epoch 13, batch 100, loss[loss=0.2519, simple_loss=0.3192, pruned_loss=0.09227, over 12269.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2773, pruned_loss=0.05509, over 1316019.63 frames. ], batch size: 246, lr: 5.36e-03, grad_scale: 2.0 2023-04-29 17:16:10,835 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7781, 3.9709, 4.2451, 3.1846, 3.7033, 4.1759, 3.9207, 2.5039], device='cuda:6'), covar=tensor([0.0374, 0.0094, 0.0035, 0.0252, 0.0071, 0.0063, 0.0052, 0.0360], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0070, 0.0070, 0.0127, 0.0080, 0.0088, 0.0079, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 17:16:27,103 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-29 17:16:30,913 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9792, 3.4102, 2.9502, 5.1484, 4.3112, 4.5257, 1.7538, 3.5450], device='cuda:6'), covar=tensor([0.1255, 0.0598, 0.1048, 0.0153, 0.0247, 0.0386, 0.1455, 0.0587], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0157, 0.0179, 0.0144, 0.0185, 0.0203, 0.0181, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 17:17:04,490 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121950.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:17:07,072 INFO [train.py:904] (6/8) Epoch 13, batch 150, loss[loss=0.2319, simple_loss=0.309, pruned_loss=0.07746, over 12049.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2764, pruned_loss=0.05499, over 1762767.18 frames. ], batch size: 246, lr: 5.35e-03, grad_scale: 2.0 2023-04-29 17:17:27,837 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121967.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 17:17:35,629 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.742e+02 2.538e+02 3.283e+02 4.116e+02 1.264e+03, threshold=6.566e+02, percent-clipped=3.0 2023-04-29 17:17:48,793 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9929, 4.7249, 5.0093, 5.2239, 5.4147, 4.7410, 5.4004, 5.3699], device='cuda:6'), covar=tensor([0.1573, 0.1242, 0.1578, 0.0636, 0.0448, 0.0757, 0.0437, 0.0512], device='cuda:6'), in_proj_covar=tensor([0.0516, 0.0645, 0.0777, 0.0655, 0.0495, 0.0507, 0.0521, 0.0597], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:18:00,052 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.7020, 6.1107, 5.8105, 5.8422, 5.5010, 5.4045, 5.5242, 6.2379], device='cuda:6'), covar=tensor([0.1164, 0.0824, 0.1347, 0.0728, 0.0785, 0.0632, 0.0987, 0.0757], device='cuda:6'), in_proj_covar=tensor([0.0544, 0.0684, 0.0558, 0.0482, 0.0437, 0.0446, 0.0570, 0.0522], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:18:18,195 INFO [train.py:904] (6/8) Epoch 13, batch 200, loss[loss=0.1853, simple_loss=0.2828, pruned_loss=0.04397, over 16738.00 frames. ], tot_loss[loss=0.1924, simple_loss=0.2761, pruned_loss=0.0543, over 2111576.97 frames. ], batch size: 57, lr: 5.35e-03, grad_scale: 2.0 2023-04-29 17:18:29,420 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9425, 1.9425, 2.4011, 2.9425, 2.7663, 3.2853, 2.1944, 3.2719], device='cuda:6'), covar=tensor([0.0182, 0.0393, 0.0258, 0.0237, 0.0255, 0.0157, 0.0373, 0.0110], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0174, 0.0160, 0.0163, 0.0174, 0.0126, 0.0175, 0.0118], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 17:18:30,503 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122011.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:18:35,254 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=122015.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 17:18:59,772 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7914, 4.5131, 4.7966, 4.9831, 5.1619, 4.5550, 5.1674, 5.1252], device='cuda:6'), covar=tensor([0.1326, 0.1258, 0.1518, 0.0647, 0.0460, 0.0908, 0.0513, 0.0565], device='cuda:6'), in_proj_covar=tensor([0.0517, 0.0648, 0.0778, 0.0656, 0.0497, 0.0507, 0.0523, 0.0597], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:19:26,236 INFO [train.py:904] (6/8) Epoch 13, batch 250, loss[loss=0.1499, simple_loss=0.2342, pruned_loss=0.03278, over 16723.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2738, pruned_loss=0.05431, over 2379799.47 frames. ], batch size: 37, lr: 5.35e-03, grad_scale: 2.0 2023-04-29 17:19:41,435 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122063.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:19:41,562 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2871, 3.7306, 3.9152, 2.1948, 3.0839, 2.4646, 3.7921, 3.8137], device='cuda:6'), covar=tensor([0.0270, 0.0789, 0.0450, 0.1694, 0.0748, 0.0920, 0.0598, 0.0980], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0140, 0.0157, 0.0142, 0.0136, 0.0124, 0.0133, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 17:19:54,203 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.354e+02 2.915e+02 3.517e+02 1.210e+03, threshold=5.831e+02, percent-clipped=2.0 2023-04-29 17:20:25,601 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9128, 3.9427, 4.2560, 4.2245, 4.2671, 3.9604, 4.0278, 3.9651], device='cuda:6'), covar=tensor([0.0386, 0.0629, 0.0450, 0.0501, 0.0478, 0.0465, 0.0768, 0.0496], device='cuda:6'), in_proj_covar=tensor([0.0332, 0.0348, 0.0348, 0.0332, 0.0397, 0.0373, 0.0462, 0.0300], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 17:20:34,740 INFO [train.py:904] (6/8) Epoch 13, batch 300, loss[loss=0.1844, simple_loss=0.2602, pruned_loss=0.05435, over 16923.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2704, pruned_loss=0.0527, over 2584794.14 frames. ], batch size: 90, lr: 5.35e-03, grad_scale: 2.0 2023-04-29 17:20:48,063 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=122111.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:20:55,825 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122117.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:21:38,519 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9167, 1.9106, 2.3659, 2.8720, 2.7254, 3.1722, 2.0001, 2.9830], device='cuda:6'), covar=tensor([0.0170, 0.0352, 0.0248, 0.0221, 0.0219, 0.0134, 0.0372, 0.0140], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0175, 0.0161, 0.0164, 0.0174, 0.0127, 0.0176, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 17:21:42,719 INFO [train.py:904] (6/8) Epoch 13, batch 350, loss[loss=0.1744, simple_loss=0.2444, pruned_loss=0.05221, over 16706.00 frames. ], tot_loss[loss=0.1845, simple_loss=0.2673, pruned_loss=0.0509, over 2752608.37 frames. ], batch size: 124, lr: 5.35e-03, grad_scale: 2.0 2023-04-29 17:22:13,957 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.554e+02 2.369e+02 2.759e+02 3.305e+02 7.145e+02, threshold=5.518e+02, percent-clipped=2.0 2023-04-29 17:22:20,605 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122178.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 17:22:44,402 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122195.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:22:53,137 INFO [train.py:904] (6/8) Epoch 13, batch 400, loss[loss=0.1668, simple_loss=0.242, pruned_loss=0.04577, over 15999.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2671, pruned_loss=0.05154, over 2877163.97 frames. ], batch size: 35, lr: 5.35e-03, grad_scale: 4.0 2023-04-29 17:24:03,759 INFO [train.py:904] (6/8) Epoch 13, batch 450, loss[loss=0.1596, simple_loss=0.2566, pruned_loss=0.03131, over 17061.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2646, pruned_loss=0.05035, over 2981432.35 frames. ], batch size: 50, lr: 5.35e-03, grad_scale: 4.0 2023-04-29 17:24:10,407 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122256.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:24:29,388 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3948, 3.1940, 3.5961, 2.6019, 3.2851, 3.6351, 3.4425, 2.1479], device='cuda:6'), covar=tensor([0.0409, 0.0112, 0.0042, 0.0292, 0.0090, 0.0080, 0.0063, 0.0390], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0072, 0.0071, 0.0128, 0.0081, 0.0091, 0.0081, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 17:24:32,943 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.250e+02 2.134e+02 2.674e+02 3.325e+02 9.122e+02, threshold=5.349e+02, percent-clipped=2.0 2023-04-29 17:25:04,495 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7922, 3.0101, 2.6270, 4.7284, 3.8973, 4.2838, 1.6395, 3.1230], device='cuda:6'), covar=tensor([0.1316, 0.0668, 0.1166, 0.0172, 0.0308, 0.0419, 0.1533, 0.0763], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0157, 0.0179, 0.0147, 0.0189, 0.0204, 0.0181, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 17:25:13,886 INFO [train.py:904] (6/8) Epoch 13, batch 500, loss[loss=0.1885, simple_loss=0.2687, pruned_loss=0.05414, over 16823.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2626, pruned_loss=0.0487, over 3061540.34 frames. ], batch size: 102, lr: 5.35e-03, grad_scale: 4.0 2023-04-29 17:25:19,401 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122306.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:25:32,654 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5734, 2.1880, 2.3781, 4.3336, 2.2452, 2.6715, 2.3385, 2.3870], device='cuda:6'), covar=tensor([0.1021, 0.3505, 0.2538, 0.0459, 0.3798, 0.2310, 0.3155, 0.3271], device='cuda:6'), in_proj_covar=tensor([0.0367, 0.0397, 0.0336, 0.0321, 0.0414, 0.0454, 0.0364, 0.0464], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:25:36,920 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122319.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:25:48,859 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9429, 4.0241, 4.3766, 4.3635, 4.4016, 4.0933, 4.1206, 4.0486], device='cuda:6'), covar=tensor([0.0373, 0.0610, 0.0436, 0.0426, 0.0406, 0.0430, 0.0743, 0.0491], device='cuda:6'), in_proj_covar=tensor([0.0343, 0.0360, 0.0361, 0.0342, 0.0408, 0.0384, 0.0478, 0.0310], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 17:25:58,489 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 17:26:21,438 INFO [train.py:904] (6/8) Epoch 13, batch 550, loss[loss=0.176, simple_loss=0.2708, pruned_loss=0.04058, over 17126.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2622, pruned_loss=0.04857, over 3107695.56 frames. ], batch size: 49, lr: 5.35e-03, grad_scale: 4.0 2023-04-29 17:26:50,249 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.266e+02 2.671e+02 3.065e+02 5.763e+02, threshold=5.342e+02, percent-clipped=1.0 2023-04-29 17:27:00,851 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122380.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:27:30,037 INFO [train.py:904] (6/8) Epoch 13, batch 600, loss[loss=0.1618, simple_loss=0.2557, pruned_loss=0.03397, over 16681.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2617, pruned_loss=0.04838, over 3149117.19 frames. ], batch size: 57, lr: 5.35e-03, grad_scale: 4.0 2023-04-29 17:28:37,794 INFO [train.py:904] (6/8) Epoch 13, batch 650, loss[loss=0.1695, simple_loss=0.2506, pruned_loss=0.0442, over 16818.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2612, pruned_loss=0.04895, over 3196303.38 frames. ], batch size: 102, lr: 5.34e-03, grad_scale: 4.0 2023-04-29 17:29:01,155 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122468.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:29:07,722 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.494e+02 2.897e+02 3.411e+02 6.772e+02, threshold=5.794e+02, percent-clipped=6.0 2023-04-29 17:29:08,017 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122473.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 17:29:47,551 INFO [train.py:904] (6/8) Epoch 13, batch 700, loss[loss=0.2207, simple_loss=0.2822, pruned_loss=0.07953, over 16871.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2605, pruned_loss=0.04865, over 3209518.75 frames. ], batch size: 109, lr: 5.34e-03, grad_scale: 4.0 2023-04-29 17:30:26,027 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122529.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:30:41,736 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8578, 5.1702, 4.9396, 4.9424, 4.6355, 4.6348, 4.6440, 5.2648], device='cuda:6'), covar=tensor([0.1097, 0.0903, 0.1085, 0.0706, 0.0828, 0.1030, 0.1032, 0.0942], device='cuda:6'), in_proj_covar=tensor([0.0565, 0.0711, 0.0579, 0.0503, 0.0452, 0.0464, 0.0593, 0.0543], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:30:56,849 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122551.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:30:57,588 INFO [train.py:904] (6/8) Epoch 13, batch 750, loss[loss=0.1818, simple_loss=0.2633, pruned_loss=0.05013, over 15562.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2607, pruned_loss=0.04839, over 3221807.04 frames. ], batch size: 190, lr: 5.34e-03, grad_scale: 4.0 2023-04-29 17:31:27,851 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.324e+02 2.801e+02 3.357e+02 5.795e+02, threshold=5.603e+02, percent-clipped=1.0 2023-04-29 17:31:37,399 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0580, 5.0605, 4.8053, 4.3803, 4.8488, 1.7206, 4.6411, 4.7376], device='cuda:6'), covar=tensor([0.0071, 0.0059, 0.0154, 0.0331, 0.0083, 0.2485, 0.0118, 0.0173], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0124, 0.0169, 0.0156, 0.0141, 0.0186, 0.0157, 0.0156], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:32:09,554 INFO [train.py:904] (6/8) Epoch 13, batch 800, loss[loss=0.1493, simple_loss=0.2374, pruned_loss=0.03062, over 17233.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2598, pruned_loss=0.04771, over 3241668.95 frames. ], batch size: 43, lr: 5.34e-03, grad_scale: 8.0 2023-04-29 17:32:15,153 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122606.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:33:15,727 INFO [train.py:904] (6/8) Epoch 13, batch 850, loss[loss=0.1659, simple_loss=0.2498, pruned_loss=0.04102, over 17202.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2598, pruned_loss=0.04814, over 3260864.37 frames. ], batch size: 44, lr: 5.34e-03, grad_scale: 8.0 2023-04-29 17:33:17,185 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0933, 5.1913, 5.6508, 5.6458, 5.6489, 5.2911, 5.2283, 5.0582], device='cuda:6'), covar=tensor([0.0345, 0.0536, 0.0401, 0.0436, 0.0558, 0.0357, 0.0866, 0.0391], device='cuda:6'), in_proj_covar=tensor([0.0354, 0.0375, 0.0370, 0.0352, 0.0420, 0.0397, 0.0493, 0.0320], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 17:33:18,970 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=122654.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:33:44,224 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.660e+02 2.256e+02 2.755e+02 3.448e+02 5.006e+02, threshold=5.510e+02, percent-clipped=0.0 2023-04-29 17:33:47,010 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122675.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:34:23,500 INFO [train.py:904] (6/8) Epoch 13, batch 900, loss[loss=0.1778, simple_loss=0.2488, pruned_loss=0.05342, over 16804.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2588, pruned_loss=0.04736, over 3283144.22 frames. ], batch size: 83, lr: 5.34e-03, grad_scale: 8.0 2023-04-29 17:35:33,155 INFO [train.py:904] (6/8) Epoch 13, batch 950, loss[loss=0.1924, simple_loss=0.2699, pruned_loss=0.05748, over 16850.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2587, pruned_loss=0.04761, over 3292496.34 frames. ], batch size: 96, lr: 5.34e-03, grad_scale: 8.0 2023-04-29 17:35:39,114 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8979, 5.0088, 5.4966, 5.4754, 5.4679, 5.1013, 5.0329, 4.8698], device='cuda:6'), covar=tensor([0.0339, 0.0470, 0.0364, 0.0423, 0.0455, 0.0350, 0.0967, 0.0407], device='cuda:6'), in_proj_covar=tensor([0.0355, 0.0375, 0.0371, 0.0355, 0.0422, 0.0399, 0.0495, 0.0321], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 17:36:02,666 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 2.301e+02 2.627e+02 3.023e+02 6.278e+02, threshold=5.253e+02, percent-clipped=3.0 2023-04-29 17:36:03,706 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122773.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 17:36:43,312 INFO [train.py:904] (6/8) Epoch 13, batch 1000, loss[loss=0.1954, simple_loss=0.2634, pruned_loss=0.06368, over 16530.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2579, pruned_loss=0.04788, over 3307390.62 frames. ], batch size: 68, lr: 5.34e-03, grad_scale: 8.0 2023-04-29 17:37:07,531 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1006, 3.3037, 3.2466, 2.1073, 2.9297, 2.3286, 3.5788, 3.5068], device='cuda:6'), covar=tensor([0.0214, 0.0735, 0.0559, 0.1630, 0.0760, 0.0896, 0.0486, 0.0801], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0146, 0.0159, 0.0145, 0.0139, 0.0125, 0.0137, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 17:37:10,555 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=122821.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:37:14,623 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122824.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:37:52,027 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122851.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:37:52,802 INFO [train.py:904] (6/8) Epoch 13, batch 1050, loss[loss=0.1753, simple_loss=0.2675, pruned_loss=0.04157, over 17047.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.258, pruned_loss=0.04816, over 3302443.94 frames. ], batch size: 55, lr: 5.34e-03, grad_scale: 8.0 2023-04-29 17:38:20,483 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122872.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:38:21,176 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 2.183e+02 2.662e+02 3.059e+02 5.432e+02, threshold=5.323e+02, percent-clipped=2.0 2023-04-29 17:38:33,787 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3660, 3.2866, 3.4527, 2.5829, 3.3198, 3.5979, 3.4195, 1.8123], device='cuda:6'), covar=tensor([0.0433, 0.0128, 0.0069, 0.0307, 0.0109, 0.0106, 0.0087, 0.0526], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0072, 0.0072, 0.0129, 0.0084, 0.0092, 0.0082, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 17:38:58,265 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=122899.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:39:01,466 INFO [train.py:904] (6/8) Epoch 13, batch 1100, loss[loss=0.1623, simple_loss=0.2491, pruned_loss=0.03773, over 16836.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2577, pruned_loss=0.04762, over 3311735.98 frames. ], batch size: 42, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:39:45,776 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122933.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:40:11,764 INFO [train.py:904] (6/8) Epoch 13, batch 1150, loss[loss=0.1719, simple_loss=0.2493, pruned_loss=0.04718, over 16271.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2575, pruned_loss=0.04688, over 3316673.85 frames. ], batch size: 165, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:40:39,511 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.197e+02 2.560e+02 3.227e+02 9.975e+02, threshold=5.120e+02, percent-clipped=3.0 2023-04-29 17:40:43,397 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122975.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:41:20,386 INFO [train.py:904] (6/8) Epoch 13, batch 1200, loss[loss=0.1794, simple_loss=0.2708, pruned_loss=0.04403, over 16629.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2568, pruned_loss=0.0463, over 3314773.59 frames. ], batch size: 57, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:41:50,194 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=123023.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:42:21,853 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-29 17:42:30,139 INFO [train.py:904] (6/8) Epoch 13, batch 1250, loss[loss=0.1566, simple_loss=0.2508, pruned_loss=0.03126, over 17253.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.256, pruned_loss=0.04637, over 3309924.07 frames. ], batch size: 52, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:42:59,815 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.819e+02 2.369e+02 2.845e+02 3.353e+02 6.021e+02, threshold=5.690e+02, percent-clipped=1.0 2023-04-29 17:43:40,469 INFO [train.py:904] (6/8) Epoch 13, batch 1300, loss[loss=0.1545, simple_loss=0.2411, pruned_loss=0.03391, over 17205.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2556, pruned_loss=0.04644, over 3308610.17 frames. ], batch size: 44, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:44:12,376 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=123124.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:44:49,681 INFO [train.py:904] (6/8) Epoch 13, batch 1350, loss[loss=0.1831, simple_loss=0.2748, pruned_loss=0.04571, over 16728.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2563, pruned_loss=0.04631, over 3315632.58 frames. ], batch size: 57, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:45:11,201 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-29 17:45:17,070 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=123172.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:45:17,976 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.686e+02 2.284e+02 2.773e+02 3.220e+02 7.351e+02, threshold=5.547e+02, percent-clipped=2.0 2023-04-29 17:45:56,645 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7123, 1.7577, 1.5617, 1.5536, 1.8101, 1.6086, 1.6297, 1.8869], device='cuda:6'), covar=tensor([0.0160, 0.0233, 0.0315, 0.0300, 0.0180, 0.0222, 0.0173, 0.0170], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0217, 0.0209, 0.0208, 0.0217, 0.0214, 0.0222, 0.0204], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:45:58,577 INFO [train.py:904] (6/8) Epoch 13, batch 1400, loss[loss=0.1605, simple_loss=0.2398, pruned_loss=0.04059, over 16221.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2568, pruned_loss=0.04597, over 3320121.69 frames. ], batch size: 165, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:46:35,865 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=123228.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:47:09,270 INFO [train.py:904] (6/8) Epoch 13, batch 1450, loss[loss=0.1842, simple_loss=0.2504, pruned_loss=0.05902, over 16896.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2553, pruned_loss=0.04559, over 3316590.95 frames. ], batch size: 96, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:47:27,706 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-29 17:47:38,907 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.758e+02 2.301e+02 2.596e+02 3.249e+02 6.793e+02, threshold=5.192e+02, percent-clipped=2.0 2023-04-29 17:48:19,764 INFO [train.py:904] (6/8) Epoch 13, batch 1500, loss[loss=0.1957, simple_loss=0.258, pruned_loss=0.06669, over 16869.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2553, pruned_loss=0.04562, over 3321460.80 frames. ], batch size: 109, lr: 5.33e-03, grad_scale: 4.0 2023-04-29 17:48:36,400 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-29 17:48:37,192 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=123314.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:49:20,517 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9237, 2.9436, 2.5499, 2.7568, 3.2059, 3.0872, 3.7761, 3.4540], device='cuda:6'), covar=tensor([0.0080, 0.0297, 0.0358, 0.0336, 0.0216, 0.0234, 0.0133, 0.0190], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0217, 0.0208, 0.0208, 0.0217, 0.0215, 0.0222, 0.0206], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:49:30,720 INFO [train.py:904] (6/8) Epoch 13, batch 1550, loss[loss=0.1774, simple_loss=0.2468, pruned_loss=0.05405, over 16865.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2568, pruned_loss=0.04711, over 3324137.02 frames. ], batch size: 96, lr: 5.32e-03, grad_scale: 4.0 2023-04-29 17:49:31,181 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6807, 2.5083, 2.2307, 3.4451, 2.5385, 3.5708, 1.4600, 2.7037], device='cuda:6'), covar=tensor([0.1388, 0.0675, 0.1206, 0.0215, 0.0174, 0.0423, 0.1546, 0.0774], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0160, 0.0182, 0.0153, 0.0194, 0.0209, 0.0182, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 17:50:00,254 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.779e+02 2.426e+02 2.890e+02 3.629e+02 7.930e+02, threshold=5.780e+02, percent-clipped=5.0 2023-04-29 17:50:01,838 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=123375.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:50:11,985 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0704, 4.5323, 4.6302, 3.5143, 3.8513, 4.4949, 4.1060, 2.6945], device='cuda:6'), covar=tensor([0.0359, 0.0038, 0.0027, 0.0225, 0.0085, 0.0064, 0.0058, 0.0356], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0072, 0.0072, 0.0129, 0.0084, 0.0093, 0.0083, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 17:50:27,802 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9631, 5.1926, 5.4514, 5.2544, 5.2262, 5.8301, 5.3319, 5.0938], device='cuda:6'), covar=tensor([0.1006, 0.1832, 0.1803, 0.1795, 0.2756, 0.0927, 0.1304, 0.2131], device='cuda:6'), in_proj_covar=tensor([0.0370, 0.0526, 0.0574, 0.0455, 0.0616, 0.0597, 0.0454, 0.0606], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 17:50:39,392 INFO [train.py:904] (6/8) Epoch 13, batch 1600, loss[loss=0.1627, simple_loss=0.2506, pruned_loss=0.03744, over 17186.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2593, pruned_loss=0.04839, over 3319894.84 frames. ], batch size: 46, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:50:52,968 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6075, 3.6883, 4.0571, 2.0602, 4.1706, 4.1282, 3.1459, 3.1943], device='cuda:6'), covar=tensor([0.0732, 0.0202, 0.0141, 0.1082, 0.0057, 0.0132, 0.0370, 0.0372], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0101, 0.0090, 0.0138, 0.0069, 0.0110, 0.0121, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 17:50:59,442 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2023-04-29 17:51:15,094 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=123428.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:51:47,303 INFO [train.py:904] (6/8) Epoch 13, batch 1650, loss[loss=0.1813, simple_loss=0.2605, pruned_loss=0.05102, over 16646.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2611, pruned_loss=0.04885, over 3320602.00 frames. ], batch size: 89, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:52:11,314 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5148, 2.4214, 1.9434, 2.1610, 2.7541, 2.5645, 2.7980, 2.7758], device='cuda:6'), covar=tensor([0.0178, 0.0291, 0.0381, 0.0362, 0.0181, 0.0248, 0.0170, 0.0247], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0217, 0.0209, 0.0208, 0.0218, 0.0216, 0.0223, 0.0207], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:52:18,026 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.667e+02 2.454e+02 2.938e+02 3.518e+02 6.758e+02, threshold=5.875e+02, percent-clipped=2.0 2023-04-29 17:52:38,901 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=123489.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:52:46,471 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 17:52:55,811 INFO [train.py:904] (6/8) Epoch 13, batch 1700, loss[loss=0.1858, simple_loss=0.267, pruned_loss=0.0523, over 16707.00 frames. ], tot_loss[loss=0.181, simple_loss=0.263, pruned_loss=0.04951, over 3322127.44 frames. ], batch size: 89, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:53:12,553 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8860, 4.3005, 4.4135, 3.2596, 3.6225, 4.3139, 3.9313, 2.3792], device='cuda:6'), covar=tensor([0.0392, 0.0055, 0.0032, 0.0267, 0.0100, 0.0068, 0.0062, 0.0416], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0073, 0.0072, 0.0129, 0.0085, 0.0093, 0.0084, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 17:53:32,016 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=123528.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:53:44,496 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5930, 4.9228, 4.6433, 4.7066, 4.4427, 4.3497, 4.3810, 4.9402], device='cuda:6'), covar=tensor([0.1118, 0.0860, 0.1041, 0.0749, 0.0842, 0.1245, 0.1104, 0.0968], device='cuda:6'), in_proj_covar=tensor([0.0585, 0.0737, 0.0595, 0.0524, 0.0467, 0.0477, 0.0613, 0.0565], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:54:04,796 INFO [train.py:904] (6/8) Epoch 13, batch 1750, loss[loss=0.2434, simple_loss=0.3188, pruned_loss=0.08395, over 11845.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.265, pruned_loss=0.05033, over 3301069.90 frames. ], batch size: 248, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:54:16,151 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8909, 4.4411, 3.3743, 2.3743, 2.8853, 2.6442, 4.7222, 3.8246], device='cuda:6'), covar=tensor([0.2591, 0.0522, 0.1492, 0.2446, 0.2528, 0.1789, 0.0356, 0.1089], device='cuda:6'), in_proj_covar=tensor([0.0306, 0.0262, 0.0289, 0.0283, 0.0280, 0.0228, 0.0272, 0.0304], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:54:34,129 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.793e+02 2.358e+02 2.760e+02 3.262e+02 5.842e+02, threshold=5.520e+02, percent-clipped=0.0 2023-04-29 17:54:37,989 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=123576.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:55:14,400 INFO [train.py:904] (6/8) Epoch 13, batch 1800, loss[loss=0.1892, simple_loss=0.2808, pruned_loss=0.04874, over 17081.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.2659, pruned_loss=0.04956, over 3302441.98 frames. ], batch size: 53, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:55:16,059 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1306, 5.0403, 4.8825, 4.3305, 4.9382, 1.8063, 4.6587, 4.8416], device='cuda:6'), covar=tensor([0.0080, 0.0079, 0.0166, 0.0357, 0.0090, 0.2630, 0.0134, 0.0181], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0131, 0.0176, 0.0165, 0.0149, 0.0191, 0.0165, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:56:23,364 INFO [train.py:904] (6/8) Epoch 13, batch 1850, loss[loss=0.1944, simple_loss=0.2723, pruned_loss=0.05826, over 16248.00 frames. ], tot_loss[loss=0.1841, simple_loss=0.2674, pruned_loss=0.05045, over 3296974.34 frames. ], batch size: 165, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:56:37,122 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.58 vs. limit=5.0 2023-04-29 17:56:40,471 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1973, 1.9846, 2.1574, 3.7872, 2.0834, 2.4100, 2.0937, 2.1472], device='cuda:6'), covar=tensor([0.1149, 0.3525, 0.2437, 0.0545, 0.3510, 0.2237, 0.3277, 0.3105], device='cuda:6'), in_proj_covar=tensor([0.0370, 0.0402, 0.0339, 0.0326, 0.0417, 0.0464, 0.0368, 0.0471], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 17:56:47,435 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=123670.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:56:52,511 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.694e+02 2.459e+02 2.923e+02 3.563e+02 7.345e+02, threshold=5.846e+02, percent-clipped=2.0 2023-04-29 17:57:33,302 INFO [train.py:904] (6/8) Epoch 13, batch 1900, loss[loss=0.1737, simple_loss=0.2512, pruned_loss=0.04805, over 16881.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2664, pruned_loss=0.05021, over 3293701.85 frames. ], batch size: 116, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:58:39,328 INFO [train.py:904] (6/8) Epoch 13, batch 1950, loss[loss=0.1931, simple_loss=0.2856, pruned_loss=0.05032, over 17122.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2667, pruned_loss=0.04984, over 3301473.30 frames. ], batch size: 48, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:59:09,610 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-29 17:59:09,938 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 2.247e+02 2.704e+02 3.281e+02 7.395e+02, threshold=5.408e+02, percent-clipped=2.0 2023-04-29 17:59:25,259 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=123784.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:59:48,915 INFO [train.py:904] (6/8) Epoch 13, batch 2000, loss[loss=0.2331, simple_loss=0.2963, pruned_loss=0.08498, over 16682.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2658, pruned_loss=0.04936, over 3309472.52 frames. ], batch size: 134, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:00:31,882 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-29 18:00:36,029 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-29 18:00:59,626 INFO [train.py:904] (6/8) Epoch 13, batch 2050, loss[loss=0.1833, simple_loss=0.2771, pruned_loss=0.04476, over 17028.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2658, pruned_loss=0.04992, over 3320789.93 frames. ], batch size: 50, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:01:28,834 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=123872.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:01:30,782 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.367e+02 2.767e+02 3.495e+02 6.657e+02, threshold=5.534e+02, percent-clipped=3.0 2023-04-29 18:02:09,985 INFO [train.py:904] (6/8) Epoch 13, batch 2100, loss[loss=0.2064, simple_loss=0.2861, pruned_loss=0.06339, over 16479.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2674, pruned_loss=0.0512, over 3311693.44 frames. ], batch size: 75, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:02:33,557 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-29 18:02:54,402 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2292, 4.1700, 4.1070, 3.8835, 3.8837, 4.1839, 3.9001, 3.9718], device='cuda:6'), covar=tensor([0.0672, 0.0718, 0.0261, 0.0242, 0.0726, 0.0499, 0.0795, 0.0541], device='cuda:6'), in_proj_covar=tensor([0.0271, 0.0364, 0.0322, 0.0301, 0.0344, 0.0346, 0.0218, 0.0379], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 18:02:54,438 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=123933.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:03:20,328 INFO [train.py:904] (6/8) Epoch 13, batch 2150, loss[loss=0.1752, simple_loss=0.2511, pruned_loss=0.04962, over 15999.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2681, pruned_loss=0.05165, over 3310667.74 frames. ], batch size: 35, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:03:44,943 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=123970.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:03:50,024 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.498e+02 2.337e+02 2.985e+02 3.427e+02 6.976e+02, threshold=5.971e+02, percent-clipped=3.0 2023-04-29 18:04:30,744 INFO [train.py:904] (6/8) Epoch 13, batch 2200, loss[loss=0.1843, simple_loss=0.2874, pruned_loss=0.04058, over 17254.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2675, pruned_loss=0.05089, over 3311704.45 frames. ], batch size: 52, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:04:53,398 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=124018.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:05:03,351 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6457, 3.6325, 2.8132, 2.1383, 2.4053, 2.2328, 3.7092, 3.2380], device='cuda:6'), covar=tensor([0.2422, 0.0658, 0.1543, 0.2525, 0.2479, 0.1894, 0.0502, 0.1302], device='cuda:6'), in_proj_covar=tensor([0.0306, 0.0262, 0.0288, 0.0283, 0.0284, 0.0228, 0.0271, 0.0308], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 18:05:19,553 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124036.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:05:40,212 INFO [train.py:904] (6/8) Epoch 13, batch 2250, loss[loss=0.1902, simple_loss=0.2617, pruned_loss=0.05941, over 16602.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2687, pruned_loss=0.05109, over 3311618.79 frames. ], batch size: 75, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:05:53,686 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124062.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:06:09,552 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.722e+02 2.445e+02 2.929e+02 3.478e+02 7.502e+02, threshold=5.857e+02, percent-clipped=2.0 2023-04-29 18:06:22,677 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124084.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:06:42,896 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124097.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:06:48,018 INFO [train.py:904] (6/8) Epoch 13, batch 2300, loss[loss=0.2142, simple_loss=0.2863, pruned_loss=0.07109, over 15742.00 frames. ], tot_loss[loss=0.186, simple_loss=0.2694, pruned_loss=0.05132, over 3303655.82 frames. ], batch size: 191, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:07:16,676 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6595, 4.5018, 4.7089, 4.8756, 5.0232, 4.4491, 4.9934, 5.0167], device='cuda:6'), covar=tensor([0.1481, 0.1110, 0.1348, 0.0631, 0.0518, 0.0989, 0.0760, 0.0553], device='cuda:6'), in_proj_covar=tensor([0.0573, 0.0715, 0.0862, 0.0723, 0.0541, 0.0570, 0.0576, 0.0661], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:07:17,846 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124123.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:07:22,862 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-29 18:07:29,438 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=124132.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:07:56,749 INFO [train.py:904] (6/8) Epoch 13, batch 2350, loss[loss=0.1753, simple_loss=0.2726, pruned_loss=0.03898, over 16748.00 frames. ], tot_loss[loss=0.1858, simple_loss=0.2693, pruned_loss=0.05115, over 3308357.10 frames. ], batch size: 62, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:08:26,419 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.342e+02 2.726e+02 3.219e+02 8.801e+02, threshold=5.452e+02, percent-clipped=1.0 2023-04-29 18:08:33,136 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-29 18:09:06,167 INFO [train.py:904] (6/8) Epoch 13, batch 2400, loss[loss=0.1944, simple_loss=0.2781, pruned_loss=0.05534, over 15965.00 frames. ], tot_loss[loss=0.187, simple_loss=0.2707, pruned_loss=0.05162, over 3305716.76 frames. ], batch size: 35, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:09:08,443 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2389, 4.5338, 4.5088, 3.3735, 3.9090, 4.2971, 3.9392, 2.2042], device='cuda:6'), covar=tensor([0.0375, 0.0083, 0.0044, 0.0316, 0.0115, 0.0117, 0.0116, 0.0528], device='cuda:6'), in_proj_covar=tensor([0.0129, 0.0072, 0.0071, 0.0127, 0.0083, 0.0092, 0.0082, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 18:09:42,815 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124228.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:09:53,200 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-29 18:10:04,380 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8781, 3.4161, 2.6683, 5.1270, 4.2200, 4.6084, 1.6726, 3.4123], device='cuda:6'), covar=tensor([0.1282, 0.0569, 0.1161, 0.0133, 0.0248, 0.0328, 0.1503, 0.0651], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0161, 0.0184, 0.0157, 0.0197, 0.0210, 0.0183, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 18:10:15,680 INFO [train.py:904] (6/8) Epoch 13, batch 2450, loss[loss=0.1812, simple_loss=0.2625, pruned_loss=0.04994, over 16476.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2709, pruned_loss=0.05114, over 3311158.71 frames. ], batch size: 146, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:10:43,399 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124272.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:10:46,034 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.400e+02 2.819e+02 3.354e+02 7.582e+02, threshold=5.638e+02, percent-clipped=2.0 2023-04-29 18:11:11,583 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-29 18:11:24,236 INFO [train.py:904] (6/8) Epoch 13, batch 2500, loss[loss=0.171, simple_loss=0.2689, pruned_loss=0.03656, over 17060.00 frames. ], tot_loss[loss=0.1859, simple_loss=0.2702, pruned_loss=0.05079, over 3308042.44 frames. ], batch size: 53, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:11:49,572 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-29 18:12:09,445 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124333.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:12:31,500 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124348.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 18:12:37,097 INFO [train.py:904] (6/8) Epoch 13, batch 2550, loss[loss=0.2131, simple_loss=0.2869, pruned_loss=0.06963, over 16853.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.271, pruned_loss=0.05128, over 3295670.80 frames. ], batch size: 116, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:13:06,416 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.275e+02 2.738e+02 3.268e+02 6.282e+02, threshold=5.476e+02, percent-clipped=1.0 2023-04-29 18:13:27,582 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.26 vs. limit=5.0 2023-04-29 18:13:31,385 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124392.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:13:44,301 INFO [train.py:904] (6/8) Epoch 13, batch 2600, loss[loss=0.2186, simple_loss=0.2898, pruned_loss=0.07364, over 16724.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.271, pruned_loss=0.05104, over 3287370.47 frames. ], batch size: 134, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:13:51,956 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9325, 4.4778, 3.2943, 2.3255, 2.8940, 2.6025, 4.7790, 3.8928], device='cuda:6'), covar=tensor([0.2536, 0.0578, 0.1510, 0.2597, 0.2679, 0.1887, 0.0345, 0.1124], device='cuda:6'), in_proj_covar=tensor([0.0307, 0.0262, 0.0288, 0.0285, 0.0284, 0.0229, 0.0272, 0.0308], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 18:13:54,841 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124409.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 18:14:06,941 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124418.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:14:26,610 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124432.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:14:49,344 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2087, 5.1984, 5.6564, 5.6282, 5.6925, 5.2872, 5.2466, 4.9578], device='cuda:6'), covar=tensor([0.0267, 0.0459, 0.0316, 0.0386, 0.0427, 0.0313, 0.0790, 0.0377], device='cuda:6'), in_proj_covar=tensor([0.0363, 0.0382, 0.0382, 0.0361, 0.0432, 0.0406, 0.0504, 0.0324], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 18:14:54,849 INFO [train.py:904] (6/8) Epoch 13, batch 2650, loss[loss=0.155, simple_loss=0.2499, pruned_loss=0.03001, over 17175.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2704, pruned_loss=0.05014, over 3292846.54 frames. ], batch size: 46, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:15:26,000 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.623e+02 2.199e+02 2.542e+02 3.057e+02 5.216e+02, threshold=5.084e+02, percent-clipped=0.0 2023-04-29 18:15:52,818 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124493.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:16:04,563 INFO [train.py:904] (6/8) Epoch 13, batch 2700, loss[loss=0.1847, simple_loss=0.278, pruned_loss=0.04574, over 16726.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2703, pruned_loss=0.04945, over 3301761.57 frames. ], batch size: 57, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:16:40,461 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124528.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:16:48,446 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1659, 4.0160, 4.2316, 4.3657, 4.4440, 3.9718, 4.2146, 4.4366], device='cuda:6'), covar=tensor([0.1514, 0.1052, 0.1361, 0.0651, 0.0588, 0.1438, 0.2015, 0.0696], device='cuda:6'), in_proj_covar=tensor([0.0579, 0.0722, 0.0871, 0.0729, 0.0548, 0.0578, 0.0581, 0.0672], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:16:56,180 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4684, 3.7707, 4.1379, 2.3034, 3.2503, 2.5991, 4.0200, 3.9754], device='cuda:6'), covar=tensor([0.0242, 0.0753, 0.0422, 0.1679, 0.0724, 0.0848, 0.0564, 0.0865], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0152, 0.0162, 0.0147, 0.0138, 0.0126, 0.0139, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 18:17:13,323 INFO [train.py:904] (6/8) Epoch 13, batch 2750, loss[loss=0.163, simple_loss=0.2643, pruned_loss=0.03087, over 17131.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2699, pruned_loss=0.04876, over 3305756.36 frames. ], batch size: 49, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:17:44,014 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.177e+02 2.567e+02 3.156e+02 4.821e+02, threshold=5.134e+02, percent-clipped=0.0 2023-04-29 18:17:46,649 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=124576.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:18:05,331 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0933, 2.0292, 2.2500, 3.5576, 2.1233, 2.3151, 2.2048, 2.1594], device='cuda:6'), covar=tensor([0.1172, 0.3315, 0.2328, 0.0586, 0.3651, 0.2476, 0.3029, 0.3396], device='cuda:6'), in_proj_covar=tensor([0.0374, 0.0405, 0.0339, 0.0325, 0.0417, 0.0468, 0.0369, 0.0476], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:18:22,959 INFO [train.py:904] (6/8) Epoch 13, batch 2800, loss[loss=0.153, simple_loss=0.2471, pruned_loss=0.02946, over 17169.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2698, pruned_loss=0.04839, over 3309481.23 frames. ], batch size: 46, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:18:24,631 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0067, 4.2026, 2.2715, 4.7257, 2.9701, 4.5995, 2.3752, 3.3799], device='cuda:6'), covar=tensor([0.0215, 0.0256, 0.1558, 0.0144, 0.0752, 0.0431, 0.1437, 0.0623], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0169, 0.0190, 0.0145, 0.0171, 0.0215, 0.0199, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 18:18:59,116 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124628.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:19:09,486 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-29 18:19:12,816 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124638.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:19:31,048 INFO [train.py:904] (6/8) Epoch 13, batch 2850, loss[loss=0.1771, simple_loss=0.2679, pruned_loss=0.04312, over 17127.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.2685, pruned_loss=0.04839, over 3307820.15 frames. ], batch size: 48, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:19:46,073 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124663.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:20:00,128 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.347e+02 2.816e+02 3.672e+02 8.891e+02, threshold=5.632e+02, percent-clipped=4.0 2023-04-29 18:20:24,532 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124692.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:20:26,439 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8115, 4.7955, 5.2337, 5.2052, 5.2524, 4.9446, 4.8901, 4.7061], device='cuda:6'), covar=tensor([0.0295, 0.0496, 0.0399, 0.0465, 0.0545, 0.0358, 0.0943, 0.0439], device='cuda:6'), in_proj_covar=tensor([0.0365, 0.0384, 0.0384, 0.0363, 0.0433, 0.0408, 0.0507, 0.0324], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 18:20:30,137 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1233, 3.4631, 3.5156, 5.1688, 4.5445, 4.7652, 1.9135, 3.9856], device='cuda:6'), covar=tensor([0.1251, 0.0596, 0.0830, 0.0167, 0.0256, 0.0295, 0.1448, 0.0517], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0163, 0.0186, 0.0160, 0.0201, 0.0213, 0.0186, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 18:20:34,550 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124699.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:20:37,525 INFO [train.py:904] (6/8) Epoch 13, batch 2900, loss[loss=0.1647, simple_loss=0.2508, pruned_loss=0.0393, over 16997.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2673, pruned_loss=0.04847, over 3309201.04 frames. ], batch size: 41, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:20:40,123 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124704.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 18:20:47,902 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8079, 4.6844, 4.6739, 4.3594, 4.3047, 4.7245, 4.6256, 4.4082], device='cuda:6'), covar=tensor([0.0594, 0.0727, 0.0283, 0.0313, 0.0993, 0.0506, 0.0404, 0.0721], device='cuda:6'), in_proj_covar=tensor([0.0271, 0.0367, 0.0324, 0.0304, 0.0347, 0.0349, 0.0220, 0.0380], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 18:20:58,999 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124718.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:21:07,474 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124724.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:21:29,889 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=124740.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:21:37,284 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0350, 4.0568, 4.4184, 4.4293, 4.4571, 4.1336, 4.1716, 4.0615], device='cuda:6'), covar=tensor([0.0342, 0.0625, 0.0412, 0.0462, 0.0466, 0.0450, 0.0871, 0.0583], device='cuda:6'), in_proj_covar=tensor([0.0362, 0.0381, 0.0380, 0.0361, 0.0429, 0.0405, 0.0503, 0.0322], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 18:21:46,232 INFO [train.py:904] (6/8) Epoch 13, batch 2950, loss[loss=0.1764, simple_loss=0.265, pruned_loss=0.04388, over 17094.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.267, pruned_loss=0.04934, over 3306944.37 frames. ], batch size: 47, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:22:05,844 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=124766.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:22:16,910 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.794e+02 2.554e+02 2.996e+02 3.501e+02 7.058e+02, threshold=5.993e+02, percent-clipped=3.0 2023-04-29 18:22:36,237 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124788.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:22:48,678 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3406, 5.2525, 5.0913, 4.5017, 5.1143, 2.1961, 4.8645, 5.1117], device='cuda:6'), covar=tensor([0.0066, 0.0065, 0.0156, 0.0392, 0.0087, 0.2238, 0.0121, 0.0149], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0133, 0.0180, 0.0170, 0.0151, 0.0191, 0.0169, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:22:53,497 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.99 vs. limit=5.0 2023-04-29 18:22:55,199 INFO [train.py:904] (6/8) Epoch 13, batch 3000, loss[loss=0.15, simple_loss=0.2372, pruned_loss=0.03143, over 16843.00 frames. ], tot_loss[loss=0.1841, simple_loss=0.2676, pruned_loss=0.0503, over 3295589.00 frames. ], batch size: 42, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:22:55,199 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 18:23:04,002 INFO [train.py:938] (6/8) Epoch 13, validation: loss=0.1391, simple_loss=0.2452, pruned_loss=0.01648, over 944034.00 frames. 2023-04-29 18:23:04,002 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 18:23:04,731 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-29 18:23:29,538 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124820.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:23:44,339 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2922, 3.7831, 3.9770, 2.1240, 3.0912, 2.4319, 3.8687, 3.8629], device='cuda:6'), covar=tensor([0.0287, 0.0791, 0.0464, 0.1813, 0.0750, 0.0936, 0.0615, 0.0937], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0150, 0.0159, 0.0145, 0.0136, 0.0125, 0.0138, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 18:23:59,071 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6068, 3.6370, 2.7852, 2.1040, 2.3862, 2.2031, 3.6750, 3.2073], device='cuda:6'), covar=tensor([0.2454, 0.0628, 0.1504, 0.2636, 0.2397, 0.1888, 0.0480, 0.1356], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0257, 0.0284, 0.0281, 0.0280, 0.0226, 0.0269, 0.0305], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:24:14,167 INFO [train.py:904] (6/8) Epoch 13, batch 3050, loss[loss=0.172, simple_loss=0.2665, pruned_loss=0.03878, over 17041.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2673, pruned_loss=0.05007, over 3299130.36 frames. ], batch size: 50, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:24:16,494 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6188, 3.0373, 2.7098, 1.8118, 2.4318, 1.9177, 3.1248, 3.3025], device='cuda:6'), covar=tensor([0.0279, 0.0797, 0.0689, 0.2183, 0.1064, 0.1139, 0.0703, 0.0841], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0151, 0.0160, 0.0146, 0.0137, 0.0126, 0.0138, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 18:24:44,774 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.296e+02 3.000e+02 3.565e+02 8.414e+02, threshold=5.999e+02, percent-clipped=2.0 2023-04-29 18:24:55,443 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124881.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:25:06,009 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7571, 3.7760, 4.2017, 1.9623, 4.3135, 4.3698, 3.1660, 3.2915], device='cuda:6'), covar=tensor([0.0735, 0.0225, 0.0186, 0.1208, 0.0074, 0.0147, 0.0375, 0.0401], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0102, 0.0090, 0.0138, 0.0071, 0.0112, 0.0122, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 18:25:25,282 INFO [train.py:904] (6/8) Epoch 13, batch 3100, loss[loss=0.2189, simple_loss=0.2882, pruned_loss=0.07476, over 11965.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2672, pruned_loss=0.05037, over 3296291.96 frames. ], batch size: 246, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:26:01,315 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124928.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:26:03,207 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1722, 2.1159, 2.6169, 3.0402, 2.8364, 3.5081, 2.3097, 3.3558], device='cuda:6'), covar=tensor([0.0192, 0.0389, 0.0255, 0.0263, 0.0241, 0.0140, 0.0341, 0.0145], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0181, 0.0165, 0.0172, 0.0179, 0.0133, 0.0179, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:26:34,831 INFO [train.py:904] (6/8) Epoch 13, batch 3150, loss[loss=0.2045, simple_loss=0.2754, pruned_loss=0.06675, over 16526.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2662, pruned_loss=0.05014, over 3304451.75 frames. ], batch size: 146, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:27:05,835 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 2.205e+02 2.626e+02 3.111e+02 5.476e+02, threshold=5.252e+02, percent-clipped=0.0 2023-04-29 18:27:09,147 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=124976.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:27:34,184 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124994.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:27:45,736 INFO [train.py:904] (6/8) Epoch 13, batch 3200, loss[loss=0.1678, simple_loss=0.263, pruned_loss=0.03628, over 16532.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2649, pruned_loss=0.04918, over 3307920.88 frames. ], batch size: 75, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:27:49,108 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125004.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 18:28:10,524 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125019.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:28:28,510 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.96 vs. limit=5.0 2023-04-29 18:28:56,247 INFO [train.py:904] (6/8) Epoch 13, batch 3250, loss[loss=0.1836, simple_loss=0.2774, pruned_loss=0.04489, over 16759.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.2647, pruned_loss=0.04915, over 3311487.27 frames. ], batch size: 62, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:28:56,585 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=125052.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 18:29:17,297 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5306, 3.5328, 3.9749, 1.9208, 3.9996, 4.0534, 3.1643, 3.0158], device='cuda:6'), covar=tensor([0.0762, 0.0231, 0.0127, 0.1210, 0.0075, 0.0157, 0.0334, 0.0429], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0102, 0.0089, 0.0138, 0.0071, 0.0112, 0.0122, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 18:29:27,127 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.694e+02 2.287e+02 2.914e+02 3.406e+02 7.385e+02, threshold=5.827e+02, percent-clipped=6.0 2023-04-29 18:29:35,970 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-04-29 18:29:46,788 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125088.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:29:50,292 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8141, 4.9267, 5.3452, 5.3888, 5.3198, 4.9846, 4.9773, 4.7154], device='cuda:6'), covar=tensor([0.0277, 0.0459, 0.0383, 0.0368, 0.0469, 0.0384, 0.0782, 0.0426], device='cuda:6'), in_proj_covar=tensor([0.0368, 0.0387, 0.0386, 0.0365, 0.0434, 0.0408, 0.0512, 0.0327], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 18:30:05,563 INFO [train.py:904] (6/8) Epoch 13, batch 3300, loss[loss=0.2014, simple_loss=0.2887, pruned_loss=0.05706, over 16464.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2659, pruned_loss=0.04915, over 3318451.77 frames. ], batch size: 68, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:30:42,372 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125128.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:30:53,994 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=125136.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:31:15,061 INFO [train.py:904] (6/8) Epoch 13, batch 3350, loss[loss=0.1919, simple_loss=0.273, pruned_loss=0.05545, over 16682.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2662, pruned_loss=0.04922, over 3318752.35 frames. ], batch size: 134, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:31:45,937 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.375e+02 2.737e+02 3.421e+02 8.798e+02, threshold=5.473e+02, percent-clipped=2.0 2023-04-29 18:31:49,061 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125176.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:32:06,956 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125189.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:32:24,428 INFO [train.py:904] (6/8) Epoch 13, batch 3400, loss[loss=0.1972, simple_loss=0.2772, pruned_loss=0.0586, over 16483.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.2665, pruned_loss=0.04937, over 3316636.98 frames. ], batch size: 75, lr: 5.29e-03, grad_scale: 4.0 2023-04-29 18:33:26,496 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9441, 3.9298, 4.4738, 2.0668, 4.5862, 4.7274, 3.2376, 3.7869], device='cuda:6'), covar=tensor([0.0685, 0.0208, 0.0179, 0.1110, 0.0072, 0.0105, 0.0371, 0.0291], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0102, 0.0089, 0.0137, 0.0070, 0.0111, 0.0121, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 18:33:27,562 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1683, 4.8899, 5.1549, 5.3879, 5.5547, 4.7291, 5.5088, 5.5256], device='cuda:6'), covar=tensor([0.1486, 0.1181, 0.1532, 0.0601, 0.0482, 0.0807, 0.0421, 0.0546], device='cuda:6'), in_proj_covar=tensor([0.0586, 0.0738, 0.0891, 0.0744, 0.0559, 0.0586, 0.0592, 0.0683], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:33:35,396 INFO [train.py:904] (6/8) Epoch 13, batch 3450, loss[loss=0.1805, simple_loss=0.257, pruned_loss=0.05204, over 16510.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2643, pruned_loss=0.04791, over 3329004.72 frames. ], batch size: 75, lr: 5.28e-03, grad_scale: 4.0 2023-04-29 18:33:44,654 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1650, 5.2390, 5.6600, 5.6253, 5.6744, 5.3280, 5.2750, 5.0560], device='cuda:6'), covar=tensor([0.0291, 0.0532, 0.0368, 0.0460, 0.0421, 0.0306, 0.0864, 0.0441], device='cuda:6'), in_proj_covar=tensor([0.0365, 0.0385, 0.0384, 0.0363, 0.0432, 0.0407, 0.0508, 0.0327], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 18:34:07,279 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.393e+02 2.670e+02 3.290e+02 7.203e+02, threshold=5.341e+02, percent-clipped=1.0 2023-04-29 18:34:35,536 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125294.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:34:46,233 INFO [train.py:904] (6/8) Epoch 13, batch 3500, loss[loss=0.1679, simple_loss=0.2478, pruned_loss=0.04403, over 16886.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2634, pruned_loss=0.04794, over 3333987.84 frames. ], batch size: 90, lr: 5.28e-03, grad_scale: 4.0 2023-04-29 18:35:09,869 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125319.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:35:21,534 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8671, 2.8179, 2.4694, 2.7147, 3.1180, 2.9633, 3.6558, 3.4238], device='cuda:6'), covar=tensor([0.0092, 0.0301, 0.0364, 0.0328, 0.0210, 0.0295, 0.0176, 0.0199], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0218, 0.0209, 0.0210, 0.0219, 0.0216, 0.0229, 0.0208], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:35:31,139 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2575, 4.0632, 4.3085, 4.4365, 4.5253, 4.1120, 4.3230, 4.5262], device='cuda:6'), covar=tensor([0.1316, 0.1105, 0.1223, 0.0653, 0.0551, 0.1169, 0.1682, 0.0750], device='cuda:6'), in_proj_covar=tensor([0.0582, 0.0733, 0.0886, 0.0742, 0.0556, 0.0583, 0.0588, 0.0679], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:35:43,533 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=125342.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:35:57,848 INFO [train.py:904] (6/8) Epoch 13, batch 3550, loss[loss=0.1617, simple_loss=0.2547, pruned_loss=0.0344, over 17136.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2624, pruned_loss=0.04725, over 3335782.59 frames. ], batch size: 48, lr: 5.28e-03, grad_scale: 4.0 2023-04-29 18:36:03,791 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125356.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:36:03,874 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4958, 1.7143, 2.1543, 2.3348, 2.4542, 2.3708, 1.6833, 2.6183], device='cuda:6'), covar=tensor([0.0152, 0.0378, 0.0218, 0.0215, 0.0216, 0.0226, 0.0380, 0.0122], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0180, 0.0164, 0.0170, 0.0179, 0.0134, 0.0178, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:36:19,301 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=125367.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:36:30,595 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.719e+02 2.251e+02 2.708e+02 3.279e+02 8.250e+02, threshold=5.415e+02, percent-clipped=4.0 2023-04-29 18:36:32,836 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125376.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:36:57,639 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0666, 5.6464, 5.8167, 5.6036, 5.6651, 6.1916, 5.6926, 5.4391], device='cuda:6'), covar=tensor([0.0833, 0.1831, 0.1951, 0.1972, 0.2560, 0.0810, 0.1443, 0.2215], device='cuda:6'), in_proj_covar=tensor([0.0373, 0.0530, 0.0573, 0.0458, 0.0620, 0.0597, 0.0455, 0.0609], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 18:37:08,328 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7881, 4.2553, 3.2775, 2.2744, 2.8881, 2.5367, 4.5927, 3.7484], device='cuda:6'), covar=tensor([0.2620, 0.0523, 0.1389, 0.2535, 0.2461, 0.1730, 0.0306, 0.0994], device='cuda:6'), in_proj_covar=tensor([0.0306, 0.0261, 0.0288, 0.0287, 0.0286, 0.0230, 0.0275, 0.0309], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 18:37:08,975 INFO [train.py:904] (6/8) Epoch 13, batch 3600, loss[loss=0.1838, simple_loss=0.2783, pruned_loss=0.04464, over 17044.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2624, pruned_loss=0.04766, over 3318099.78 frames. ], batch size: 55, lr: 5.28e-03, grad_scale: 8.0 2023-04-29 18:37:31,320 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125417.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 18:37:35,983 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125421.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:37:39,546 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2023-04-29 18:37:55,618 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8773, 4.7907, 4.7548, 4.4836, 4.4273, 4.8528, 4.6183, 4.5698], device='cuda:6'), covar=tensor([0.0612, 0.0669, 0.0317, 0.0310, 0.0952, 0.0420, 0.0461, 0.0687], device='cuda:6'), in_proj_covar=tensor([0.0273, 0.0368, 0.0323, 0.0305, 0.0349, 0.0351, 0.0220, 0.0381], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 18:37:59,838 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125437.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:38:21,931 INFO [train.py:904] (6/8) Epoch 13, batch 3650, loss[loss=0.1684, simple_loss=0.2617, pruned_loss=0.03749, over 17247.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2604, pruned_loss=0.04784, over 3326910.78 frames. ], batch size: 52, lr: 5.28e-03, grad_scale: 8.0 2023-04-29 18:38:57,388 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.272e+02 2.828e+02 3.424e+02 6.489e+02, threshold=5.656e+02, percent-clipped=1.0 2023-04-29 18:38:59,662 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125476.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:39:07,736 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125482.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:39:11,353 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125484.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:39:36,233 INFO [train.py:904] (6/8) Epoch 13, batch 3700, loss[loss=0.1861, simple_loss=0.2654, pruned_loss=0.05343, over 15567.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2598, pruned_loss=0.04956, over 3281282.50 frames. ], batch size: 191, lr: 5.28e-03, grad_scale: 8.0 2023-04-29 18:40:09,746 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6628, 5.0278, 4.7787, 4.7968, 4.4726, 4.4471, 4.4704, 5.0543], device='cuda:6'), covar=tensor([0.1025, 0.0742, 0.0968, 0.0649, 0.0756, 0.1129, 0.0979, 0.0875], device='cuda:6'), in_proj_covar=tensor([0.0577, 0.0728, 0.0589, 0.0517, 0.0461, 0.0465, 0.0605, 0.0561], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:40:10,905 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=125524.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:40:32,047 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7795, 1.8796, 2.3564, 2.6595, 2.7392, 2.5329, 1.8065, 2.9144], device='cuda:6'), covar=tensor([0.0129, 0.0364, 0.0237, 0.0197, 0.0204, 0.0213, 0.0394, 0.0086], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0178, 0.0163, 0.0170, 0.0178, 0.0133, 0.0177, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:40:51,172 INFO [train.py:904] (6/8) Epoch 13, batch 3750, loss[loss=0.162, simple_loss=0.2423, pruned_loss=0.04084, over 16781.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2603, pruned_loss=0.05087, over 3279256.42 frames. ], batch size: 89, lr: 5.28e-03, grad_scale: 8.0 2023-04-29 18:41:24,165 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.222e+02 2.735e+02 3.211e+02 5.083e+02, threshold=5.471e+02, percent-clipped=0.0 2023-04-29 18:41:27,347 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1779, 3.4511, 3.7581, 2.5849, 3.3422, 3.7415, 3.4744, 2.1370], device='cuda:6'), covar=tensor([0.0432, 0.0113, 0.0035, 0.0265, 0.0081, 0.0072, 0.0063, 0.0356], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0073, 0.0071, 0.0126, 0.0083, 0.0093, 0.0083, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 18:42:05,166 INFO [train.py:904] (6/8) Epoch 13, batch 3800, loss[loss=0.1838, simple_loss=0.2573, pruned_loss=0.0551, over 16454.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2621, pruned_loss=0.05217, over 3270891.64 frames. ], batch size: 75, lr: 5.28e-03, grad_scale: 8.0 2023-04-29 18:42:07,512 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125603.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:42:11,267 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0340, 2.7331, 2.6911, 1.9715, 2.5710, 2.1359, 2.7241, 2.9320], device='cuda:6'), covar=tensor([0.0305, 0.0724, 0.0525, 0.1736, 0.0795, 0.0862, 0.0647, 0.0666], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0152, 0.0160, 0.0146, 0.0138, 0.0126, 0.0138, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 18:43:18,384 INFO [train.py:904] (6/8) Epoch 13, batch 3850, loss[loss=0.1909, simple_loss=0.2606, pruned_loss=0.06065, over 16371.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2624, pruned_loss=0.05266, over 3266894.86 frames. ], batch size: 68, lr: 5.28e-03, grad_scale: 8.0 2023-04-29 18:43:35,233 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125664.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:43:39,949 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5511, 4.6578, 4.9293, 4.9165, 4.9801, 4.6862, 4.5344, 4.4233], device='cuda:6'), covar=tensor([0.0448, 0.0739, 0.0434, 0.0574, 0.0572, 0.0482, 0.1130, 0.0595], device='cuda:6'), in_proj_covar=tensor([0.0361, 0.0380, 0.0376, 0.0360, 0.0427, 0.0401, 0.0500, 0.0321], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 18:43:50,561 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.444e+02 2.805e+02 3.444e+02 9.574e+02, threshold=5.610e+02, percent-clipped=2.0 2023-04-29 18:44:29,762 INFO [train.py:904] (6/8) Epoch 13, batch 3900, loss[loss=0.1772, simple_loss=0.2622, pruned_loss=0.04604, over 16786.00 frames. ], tot_loss[loss=0.1845, simple_loss=0.2621, pruned_loss=0.05344, over 3262461.78 frames. ], batch size: 39, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:44:45,081 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125712.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 18:44:49,760 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4681, 2.2691, 2.3310, 4.3688, 2.1833, 2.6735, 2.3490, 2.4841], device='cuda:6'), covar=tensor([0.1078, 0.3275, 0.2286, 0.0401, 0.3620, 0.2259, 0.3194, 0.2786], device='cuda:6'), in_proj_covar=tensor([0.0373, 0.0408, 0.0340, 0.0326, 0.0419, 0.0472, 0.0372, 0.0478], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:44:58,107 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-29 18:45:03,902 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125725.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:45:14,792 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125732.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:45:42,347 INFO [train.py:904] (6/8) Epoch 13, batch 3950, loss[loss=0.1714, simple_loss=0.2485, pruned_loss=0.04713, over 16492.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2623, pruned_loss=0.05446, over 3269888.78 frames. ], batch size: 68, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:46:16,759 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.736e+02 2.289e+02 2.629e+02 3.210e+02 7.394e+02, threshold=5.259e+02, percent-clipped=3.0 2023-04-29 18:46:17,260 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4335, 3.5952, 3.9449, 1.7469, 3.9405, 4.1443, 3.1315, 3.0435], device='cuda:6'), covar=tensor([0.0830, 0.0230, 0.0170, 0.1341, 0.0108, 0.0164, 0.0388, 0.0433], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0102, 0.0089, 0.0138, 0.0070, 0.0111, 0.0121, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 18:46:20,058 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125777.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:46:29,352 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125784.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:46:32,795 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125786.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:46:44,350 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-04-29 18:46:53,469 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0283, 5.6338, 5.8471, 5.5168, 5.6207, 6.1882, 5.6495, 5.3654], device='cuda:6'), covar=tensor([0.0775, 0.1441, 0.1623, 0.1851, 0.2269, 0.0836, 0.1275, 0.2024], device='cuda:6'), in_proj_covar=tensor([0.0373, 0.0528, 0.0572, 0.0456, 0.0614, 0.0595, 0.0453, 0.0607], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 18:46:55,491 INFO [train.py:904] (6/8) Epoch 13, batch 4000, loss[loss=0.1929, simple_loss=0.2695, pruned_loss=0.05819, over 16930.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2618, pruned_loss=0.05445, over 3277206.01 frames. ], batch size: 116, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:47:38,001 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=125832.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:48:05,369 INFO [train.py:904] (6/8) Epoch 13, batch 4050, loss[loss=0.177, simple_loss=0.2611, pruned_loss=0.04651, over 16486.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2621, pruned_loss=0.0533, over 3281237.90 frames. ], batch size: 146, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:48:23,594 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.33 vs. limit=5.0 2023-04-29 18:48:36,495 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 1.993e+02 2.274e+02 2.780e+02 5.444e+02, threshold=4.549e+02, percent-clipped=3.0 2023-04-29 18:48:57,831 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5450, 4.9325, 5.2722, 5.1718, 5.1846, 4.8250, 4.4280, 4.5494], device='cuda:6'), covar=tensor([0.0565, 0.0559, 0.0443, 0.0651, 0.0647, 0.0491, 0.1356, 0.0493], device='cuda:6'), in_proj_covar=tensor([0.0356, 0.0373, 0.0371, 0.0356, 0.0421, 0.0395, 0.0492, 0.0316], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 18:49:15,116 INFO [train.py:904] (6/8) Epoch 13, batch 4100, loss[loss=0.1779, simple_loss=0.2714, pruned_loss=0.0422, over 16717.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2631, pruned_loss=0.05213, over 3283582.61 frames. ], batch size: 83, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:50:04,849 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 18:50:19,792 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5136, 3.3713, 2.6526, 2.1143, 2.3733, 2.2122, 3.5142, 3.2406], device='cuda:6'), covar=tensor([0.2567, 0.0772, 0.1573, 0.2381, 0.2277, 0.1819, 0.0554, 0.1058], device='cuda:6'), in_proj_covar=tensor([0.0304, 0.0260, 0.0290, 0.0287, 0.0288, 0.0229, 0.0275, 0.0307], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:50:30,154 INFO [train.py:904] (6/8) Epoch 13, batch 4150, loss[loss=0.2227, simple_loss=0.3028, pruned_loss=0.0713, over 17020.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.2707, pruned_loss=0.05537, over 3225188.56 frames. ], batch size: 50, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:50:42,582 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125959.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:51:05,863 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.529e+02 2.394e+02 2.865e+02 3.650e+02 6.775e+02, threshold=5.730e+02, percent-clipped=10.0 2023-04-29 18:51:40,580 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125998.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:51:48,696 INFO [train.py:904] (6/8) Epoch 13, batch 4200, loss[loss=0.2145, simple_loss=0.3033, pruned_loss=0.06286, over 16904.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2776, pruned_loss=0.05711, over 3197345.04 frames. ], batch size: 109, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:52:04,601 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126012.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 18:52:34,267 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126032.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:53:02,528 INFO [train.py:904] (6/8) Epoch 13, batch 4250, loss[loss=0.1975, simple_loss=0.293, pruned_loss=0.05105, over 16694.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2814, pruned_loss=0.05718, over 3180798.84 frames. ], batch size: 76, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:53:13,107 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126059.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:53:14,163 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=126060.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:53:26,952 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9099, 1.9530, 2.3525, 3.1821, 2.1169, 2.1671, 2.2189, 2.0513], device='cuda:6'), covar=tensor([0.1043, 0.3408, 0.1930, 0.0557, 0.3740, 0.2390, 0.2869, 0.3529], device='cuda:6'), in_proj_covar=tensor([0.0371, 0.0407, 0.0337, 0.0322, 0.0417, 0.0470, 0.0370, 0.0475], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:53:36,046 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.365e+02 2.942e+02 3.644e+02 5.722e+02, threshold=5.884e+02, percent-clipped=0.0 2023-04-29 18:53:39,781 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126077.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:53:43,876 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=126080.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:53:45,096 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126081.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:54:06,378 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-04-29 18:54:16,795 INFO [train.py:904] (6/8) Epoch 13, batch 4300, loss[loss=0.1848, simple_loss=0.2789, pruned_loss=0.04538, over 16484.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2823, pruned_loss=0.056, over 3180675.28 frames. ], batch size: 62, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:54:21,459 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7403, 3.7506, 3.9196, 3.7181, 3.8418, 4.2530, 3.9523, 3.6120], device='cuda:6'), covar=tensor([0.1963, 0.1917, 0.1984, 0.2228, 0.2343, 0.1501, 0.1370, 0.2506], device='cuda:6'), in_proj_covar=tensor([0.0363, 0.0511, 0.0553, 0.0438, 0.0591, 0.0579, 0.0441, 0.0592], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 18:54:51,666 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=126125.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:55:07,390 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-04-29 18:55:30,684 INFO [train.py:904] (6/8) Epoch 13, batch 4350, loss[loss=0.1893, simple_loss=0.2875, pruned_loss=0.04556, over 16892.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2858, pruned_loss=0.05743, over 3184867.06 frames. ], batch size: 96, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:56:06,023 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.441e+02 2.808e+02 3.338e+02 6.392e+02, threshold=5.616e+02, percent-clipped=2.0 2023-04-29 18:56:29,005 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-29 18:56:46,666 INFO [train.py:904] (6/8) Epoch 13, batch 4400, loss[loss=0.1994, simple_loss=0.2917, pruned_loss=0.05358, over 16734.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2874, pruned_loss=0.05805, over 3188908.38 frames. ], batch size: 83, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 18:57:59,272 INFO [train.py:904] (6/8) Epoch 13, batch 4450, loss[loss=0.2149, simple_loss=0.3072, pruned_loss=0.06128, over 16778.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2908, pruned_loss=0.05936, over 3191954.36 frames. ], batch size: 62, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 18:58:10,093 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126259.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:58:20,784 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9234, 4.9568, 4.7723, 4.4944, 4.4849, 4.8885, 4.7038, 4.5291], device='cuda:6'), covar=tensor([0.0465, 0.0245, 0.0183, 0.0192, 0.0749, 0.0216, 0.0315, 0.0494], device='cuda:6'), in_proj_covar=tensor([0.0255, 0.0343, 0.0302, 0.0283, 0.0325, 0.0325, 0.0207, 0.0353], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 18:58:33,037 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.490e+02 2.129e+02 2.519e+02 3.084e+02 5.887e+02, threshold=5.038e+02, percent-clipped=2.0 2023-04-29 18:59:14,450 INFO [train.py:904] (6/8) Epoch 13, batch 4500, loss[loss=0.1957, simple_loss=0.2842, pruned_loss=0.05364, over 16637.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2908, pruned_loss=0.05949, over 3201597.29 frames. ], batch size: 35, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 18:59:21,426 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=126307.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:00:25,574 INFO [train.py:904] (6/8) Epoch 13, batch 4550, loss[loss=0.2062, simple_loss=0.2912, pruned_loss=0.06066, over 17222.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2914, pruned_loss=0.05997, over 3228497.84 frames. ], batch size: 45, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 19:00:28,863 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126354.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:00:59,194 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.430e+02 2.007e+02 2.258e+02 2.674e+02 4.772e+02, threshold=4.515e+02, percent-clipped=0.0 2023-04-29 19:01:06,193 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=126380.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:01:07,332 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126381.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:01:14,081 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6260, 2.6317, 1.7643, 2.7902, 2.0563, 2.7906, 2.0362, 2.3112], device='cuda:6'), covar=tensor([0.0244, 0.0326, 0.1264, 0.0151, 0.0652, 0.0393, 0.1114, 0.0596], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0165, 0.0188, 0.0138, 0.0166, 0.0209, 0.0195, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 19:01:35,460 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3619, 5.3372, 5.0216, 4.1371, 5.2753, 1.5600, 4.9943, 4.7065], device='cuda:6'), covar=tensor([0.0047, 0.0042, 0.0119, 0.0410, 0.0055, 0.2970, 0.0083, 0.0223], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0126, 0.0171, 0.0161, 0.0144, 0.0182, 0.0160, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:01:36,159 INFO [train.py:904] (6/8) Epoch 13, batch 4600, loss[loss=0.2034, simple_loss=0.2885, pruned_loss=0.05914, over 17189.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2922, pruned_loss=0.06006, over 3222105.86 frames. ], batch size: 46, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 19:02:00,573 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.4643, 2.6819, 2.5340, 3.8586, 3.0884, 3.9348, 1.5183, 2.9165], device='cuda:6'), covar=tensor([0.1396, 0.0707, 0.1147, 0.0145, 0.0287, 0.0327, 0.1586, 0.0724], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0161, 0.0183, 0.0157, 0.0198, 0.0206, 0.0182, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 19:02:22,725 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=126429.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:02:23,032 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4243, 2.4088, 2.2713, 4.4785, 2.2085, 2.8490, 2.4080, 2.5024], device='cuda:6'), covar=tensor([0.1026, 0.2879, 0.2296, 0.0333, 0.3650, 0.1971, 0.2795, 0.2977], device='cuda:6'), in_proj_covar=tensor([0.0372, 0.0405, 0.0336, 0.0319, 0.0419, 0.0470, 0.0369, 0.0473], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:02:27,565 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-04-29 19:02:39,280 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126441.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:02:53,211 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=126450.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:02:55,164 INFO [train.py:904] (6/8) Epoch 13, batch 4650, loss[loss=0.1992, simple_loss=0.283, pruned_loss=0.05767, over 15484.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2912, pruned_loss=0.0601, over 3200128.64 frames. ], batch size: 191, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 19:03:27,524 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 1.885e+02 2.230e+02 2.796e+02 5.506e+02, threshold=4.460e+02, percent-clipped=0.0 2023-04-29 19:04:07,651 INFO [train.py:904] (6/8) Epoch 13, batch 4700, loss[loss=0.1891, simple_loss=0.2709, pruned_loss=0.05366, over 16633.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2886, pruned_loss=0.0591, over 3200623.75 frames. ], batch size: 62, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 19:04:21,586 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126511.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:05:13,297 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-29 19:05:21,228 INFO [train.py:904] (6/8) Epoch 13, batch 4750, loss[loss=0.172, simple_loss=0.2622, pruned_loss=0.04092, over 16404.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2841, pruned_loss=0.05702, over 3209715.93 frames. ], batch size: 68, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 19:05:53,856 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 1.907e+02 2.465e+02 2.943e+02 4.218e+02, threshold=4.930e+02, percent-clipped=1.0 2023-04-29 19:06:32,270 INFO [train.py:904] (6/8) Epoch 13, batch 4800, loss[loss=0.1903, simple_loss=0.2742, pruned_loss=0.05319, over 16649.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2809, pruned_loss=0.05547, over 3196827.90 frames. ], batch size: 62, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 19:07:39,308 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2315, 3.4141, 3.6264, 3.6030, 3.5945, 3.4192, 3.4366, 3.4360], device='cuda:6'), covar=tensor([0.0346, 0.0574, 0.0387, 0.0421, 0.0451, 0.0467, 0.0776, 0.0477], device='cuda:6'), in_proj_covar=tensor([0.0341, 0.0357, 0.0356, 0.0347, 0.0408, 0.0380, 0.0476, 0.0306], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 19:07:46,138 INFO [train.py:904] (6/8) Epoch 13, batch 4850, loss[loss=0.1897, simple_loss=0.2879, pruned_loss=0.04576, over 16764.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2817, pruned_loss=0.05476, over 3186457.10 frames. ], batch size: 124, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:07:50,043 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126654.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:08:19,805 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 1.948e+02 2.386e+02 2.716e+02 6.913e+02, threshold=4.771e+02, percent-clipped=1.0 2023-04-29 19:08:59,235 INFO [train.py:904] (6/8) Epoch 13, batch 4900, loss[loss=0.2063, simple_loss=0.2969, pruned_loss=0.05788, over 16731.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2811, pruned_loss=0.05357, over 3156915.92 frames. ], batch size: 134, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:08:59,639 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=126702.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:09:42,599 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4507, 4.4460, 4.8906, 4.8286, 4.8084, 4.4929, 4.4385, 4.2721], device='cuda:6'), covar=tensor([0.0285, 0.0499, 0.0291, 0.0418, 0.0437, 0.0319, 0.0951, 0.0466], device='cuda:6'), in_proj_covar=tensor([0.0342, 0.0358, 0.0356, 0.0347, 0.0407, 0.0380, 0.0476, 0.0306], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 19:09:49,771 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126736.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:10:12,983 INFO [train.py:904] (6/8) Epoch 13, batch 4950, loss[loss=0.257, simple_loss=0.3262, pruned_loss=0.09388, over 12249.00 frames. ], tot_loss[loss=0.1935, simple_loss=0.281, pruned_loss=0.053, over 3152974.56 frames. ], batch size: 247, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:10:45,730 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.702e+02 2.239e+02 2.710e+02 3.273e+02 4.681e+02, threshold=5.421e+02, percent-clipped=0.0 2023-04-29 19:11:26,260 INFO [train.py:904] (6/8) Epoch 13, batch 5000, loss[loss=0.2194, simple_loss=0.3111, pruned_loss=0.06383, over 15313.00 frames. ], tot_loss[loss=0.1942, simple_loss=0.2826, pruned_loss=0.05286, over 3166443.49 frames. ], batch size: 190, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:11:32,780 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126806.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:11:47,360 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=126816.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:12:11,458 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.64 vs. limit=2.0 2023-04-29 19:12:12,260 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3278, 3.3255, 3.9250, 1.7051, 4.0022, 4.0726, 2.9199, 2.8502], device='cuda:6'), covar=tensor([0.0832, 0.0233, 0.0117, 0.1255, 0.0046, 0.0081, 0.0373, 0.0463], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0101, 0.0087, 0.0135, 0.0069, 0.0107, 0.0120, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 19:12:38,197 INFO [train.py:904] (6/8) Epoch 13, batch 5050, loss[loss=0.1864, simple_loss=0.2731, pruned_loss=0.04985, over 16446.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2827, pruned_loss=0.0525, over 3183731.63 frames. ], batch size: 35, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:13:11,123 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.161e+02 2.561e+02 3.247e+02 4.827e+02, threshold=5.121e+02, percent-clipped=0.0 2023-04-29 19:13:14,623 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126877.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:13:48,011 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-29 19:13:51,462 INFO [train.py:904] (6/8) Epoch 13, batch 5100, loss[loss=0.1743, simple_loss=0.2628, pruned_loss=0.0429, over 16759.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2812, pruned_loss=0.05223, over 3196377.40 frames. ], batch size: 124, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:14:10,625 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5578, 3.6272, 3.3701, 3.0629, 3.1830, 3.4813, 3.2787, 3.3173], device='cuda:6'), covar=tensor([0.0536, 0.0459, 0.0275, 0.0248, 0.0593, 0.0413, 0.1412, 0.0483], device='cuda:6'), in_proj_covar=tensor([0.0251, 0.0343, 0.0302, 0.0282, 0.0324, 0.0327, 0.0206, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:15:04,900 INFO [train.py:904] (6/8) Epoch 13, batch 5150, loss[loss=0.1864, simple_loss=0.2857, pruned_loss=0.04354, over 15488.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2813, pruned_loss=0.05198, over 3179652.25 frames. ], batch size: 191, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:15:15,016 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4407, 4.1790, 4.4331, 4.6395, 4.8106, 4.3818, 4.7422, 4.7759], device='cuda:6'), covar=tensor([0.1424, 0.1373, 0.1624, 0.0719, 0.0455, 0.0872, 0.0569, 0.0650], device='cuda:6'), in_proj_covar=tensor([0.0541, 0.0678, 0.0810, 0.0681, 0.0514, 0.0535, 0.0547, 0.0624], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:15:37,787 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.071e+02 2.483e+02 2.841e+02 7.629e+02, threshold=4.965e+02, percent-clipped=2.0 2023-04-29 19:16:17,738 INFO [train.py:904] (6/8) Epoch 13, batch 5200, loss[loss=0.1838, simple_loss=0.2597, pruned_loss=0.05395, over 16145.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2803, pruned_loss=0.05163, over 3176380.82 frames. ], batch size: 35, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:17:08,390 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127036.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:17:31,282 INFO [train.py:904] (6/8) Epoch 13, batch 5250, loss[loss=0.1644, simple_loss=0.2459, pruned_loss=0.04148, over 17043.00 frames. ], tot_loss[loss=0.1906, simple_loss=0.2785, pruned_loss=0.05139, over 3176894.76 frames. ], batch size: 55, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:17:49,395 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-29 19:17:50,426 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.32 vs. limit=5.0 2023-04-29 19:18:04,188 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.619e+02 2.082e+02 2.366e+02 2.761e+02 5.233e+02, threshold=4.733e+02, percent-clipped=1.0 2023-04-29 19:18:18,851 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=127084.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:18:44,213 INFO [train.py:904] (6/8) Epoch 13, batch 5300, loss[loss=0.1529, simple_loss=0.2373, pruned_loss=0.03425, over 17231.00 frames. ], tot_loss[loss=0.187, simple_loss=0.2742, pruned_loss=0.04991, over 3190406.49 frames. ], batch size: 45, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:18:50,712 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127106.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:19:29,155 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1522, 2.0490, 2.6173, 3.1005, 2.9893, 3.6348, 2.0063, 3.5355], device='cuda:6'), covar=tensor([0.0148, 0.0385, 0.0246, 0.0233, 0.0213, 0.0104, 0.0422, 0.0081], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0176, 0.0160, 0.0165, 0.0174, 0.0130, 0.0175, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 19:19:58,315 INFO [train.py:904] (6/8) Epoch 13, batch 5350, loss[loss=0.2005, simple_loss=0.29, pruned_loss=0.05554, over 16429.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2724, pruned_loss=0.04912, over 3205975.14 frames. ], batch size: 68, lr: 5.24e-03, grad_scale: 4.0 2023-04-29 19:20:01,847 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=127154.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:20:27,667 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127172.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:20:32,661 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.449e+02 2.172e+02 2.538e+02 2.960e+02 5.660e+02, threshold=5.076e+02, percent-clipped=1.0 2023-04-29 19:21:10,975 INFO [train.py:904] (6/8) Epoch 13, batch 5400, loss[loss=0.2136, simple_loss=0.292, pruned_loss=0.06758, over 12211.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.2742, pruned_loss=0.0493, over 3219370.24 frames. ], batch size: 247, lr: 5.24e-03, grad_scale: 4.0 2023-04-29 19:22:20,962 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127249.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:22:26,302 INFO [train.py:904] (6/8) Epoch 13, batch 5450, loss[loss=0.2877, simple_loss=0.3503, pruned_loss=0.1126, over 15353.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2781, pruned_loss=0.05173, over 3180596.22 frames. ], batch size: 191, lr: 5.24e-03, grad_scale: 4.0 2023-04-29 19:23:02,325 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.258e+02 2.675e+02 3.738e+02 1.100e+03, threshold=5.349e+02, percent-clipped=10.0 2023-04-29 19:23:43,874 INFO [train.py:904] (6/8) Epoch 13, batch 5500, loss[loss=0.2251, simple_loss=0.3084, pruned_loss=0.07093, over 16742.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2855, pruned_loss=0.0566, over 3152595.52 frames. ], batch size: 124, lr: 5.24e-03, grad_scale: 4.0 2023-04-29 19:23:57,137 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127310.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:25:00,716 INFO [train.py:904] (6/8) Epoch 13, batch 5550, loss[loss=0.1996, simple_loss=0.2957, pruned_loss=0.05179, over 16846.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.293, pruned_loss=0.06189, over 3135086.83 frames. ], batch size: 96, lr: 5.24e-03, grad_scale: 4.0 2023-04-29 19:25:38,573 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.215e+02 3.416e+02 4.230e+02 5.078e+02 9.045e+02, threshold=8.460e+02, percent-clipped=18.0 2023-04-29 19:26:20,875 INFO [train.py:904] (6/8) Epoch 13, batch 5600, loss[loss=0.2036, simple_loss=0.2911, pruned_loss=0.05805, over 16853.00 frames. ], tot_loss[loss=0.2158, simple_loss=0.2985, pruned_loss=0.06654, over 3106698.32 frames. ], batch size: 90, lr: 5.24e-03, grad_scale: 8.0 2023-04-29 19:26:30,755 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3697, 4.6482, 4.3857, 4.4152, 4.1523, 4.1094, 4.2245, 4.6903], device='cuda:6'), covar=tensor([0.1032, 0.0848, 0.1031, 0.0741, 0.0755, 0.1348, 0.0972, 0.0882], device='cuda:6'), in_proj_covar=tensor([0.0557, 0.0696, 0.0566, 0.0493, 0.0439, 0.0445, 0.0578, 0.0535], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:27:02,310 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127427.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 19:27:41,559 INFO [train.py:904] (6/8) Epoch 13, batch 5650, loss[loss=0.2274, simple_loss=0.3075, pruned_loss=0.07364, over 16687.00 frames. ], tot_loss[loss=0.2227, simple_loss=0.3039, pruned_loss=0.07073, over 3084506.75 frames. ], batch size: 134, lr: 5.24e-03, grad_scale: 8.0 2023-04-29 19:28:12,821 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127472.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:28:17,739 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.064e+02 3.489e+02 4.637e+02 5.886e+02 1.352e+03, threshold=9.273e+02, percent-clipped=4.0 2023-04-29 19:28:36,972 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127488.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 19:28:58,354 INFO [train.py:904] (6/8) Epoch 13, batch 5700, loss[loss=0.2007, simple_loss=0.3031, pruned_loss=0.04921, over 16876.00 frames. ], tot_loss[loss=0.2256, simple_loss=0.3059, pruned_loss=0.07261, over 3076028.99 frames. ], batch size: 90, lr: 5.24e-03, grad_scale: 8.0 2023-04-29 19:29:14,398 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.43 vs. limit=5.0 2023-04-29 19:29:27,678 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=127520.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:29:42,134 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127529.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:30:18,040 INFO [train.py:904] (6/8) Epoch 13, batch 5750, loss[loss=0.2247, simple_loss=0.3087, pruned_loss=0.07033, over 16884.00 frames. ], tot_loss[loss=0.2294, simple_loss=0.3092, pruned_loss=0.07481, over 3035534.43 frames. ], batch size: 109, lr: 5.24e-03, grad_scale: 8.0 2023-04-29 19:30:20,882 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-04-29 19:30:21,774 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.52 vs. limit=2.0 2023-04-29 19:30:56,348 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.152e+02 3.042e+02 3.668e+02 4.519e+02 8.025e+02, threshold=7.337e+02, percent-clipped=0.0 2023-04-29 19:31:21,805 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127590.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:31:39,675 INFO [train.py:904] (6/8) Epoch 13, batch 5800, loss[loss=0.2087, simple_loss=0.2982, pruned_loss=0.0596, over 15367.00 frames. ], tot_loss[loss=0.2272, simple_loss=0.3084, pruned_loss=0.07297, over 3058769.75 frames. ], batch size: 191, lr: 5.24e-03, grad_scale: 4.0 2023-04-29 19:31:44,371 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127605.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:32:21,162 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7424, 1.2834, 1.6458, 1.6005, 1.7774, 1.8857, 1.5482, 1.7687], device='cuda:6'), covar=tensor([0.0189, 0.0324, 0.0162, 0.0238, 0.0196, 0.0144, 0.0308, 0.0098], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0176, 0.0160, 0.0163, 0.0174, 0.0130, 0.0175, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 19:32:46,361 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8385, 3.8843, 4.3483, 1.9819, 4.4114, 4.5799, 3.2399, 3.3415], device='cuda:6'), covar=tensor([0.0728, 0.0208, 0.0148, 0.1235, 0.0070, 0.0093, 0.0343, 0.0416], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0101, 0.0087, 0.0136, 0.0068, 0.0108, 0.0120, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 19:32:57,618 INFO [train.py:904] (6/8) Epoch 13, batch 5850, loss[loss=0.21, simple_loss=0.2957, pruned_loss=0.06211, over 16492.00 frames. ], tot_loss[loss=0.2247, simple_loss=0.3062, pruned_loss=0.07159, over 3047778.76 frames. ], batch size: 68, lr: 5.23e-03, grad_scale: 4.0 2023-04-29 19:33:15,171 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127662.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:33:36,796 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.121e+02 3.118e+02 3.715e+02 4.396e+02 1.301e+03, threshold=7.431e+02, percent-clipped=3.0 2023-04-29 19:34:19,530 INFO [train.py:904] (6/8) Epoch 13, batch 5900, loss[loss=0.1862, simple_loss=0.2775, pruned_loss=0.04748, over 16859.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.3051, pruned_loss=0.07081, over 3062706.23 frames. ], batch size: 96, lr: 5.23e-03, grad_scale: 4.0 2023-04-29 19:34:57,882 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127723.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:35:31,072 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5800, 2.5206, 2.4581, 3.8590, 2.8219, 3.9111, 1.3464, 2.7681], device='cuda:6'), covar=tensor([0.1317, 0.0721, 0.1120, 0.0150, 0.0226, 0.0347, 0.1567, 0.0769], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0163, 0.0183, 0.0157, 0.0201, 0.0208, 0.0183, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 19:35:39,925 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6566, 4.9430, 4.7001, 4.6798, 4.4615, 4.4189, 4.4082, 5.0152], device='cuda:6'), covar=tensor([0.1010, 0.0813, 0.1011, 0.0753, 0.0777, 0.1091, 0.1014, 0.0861], device='cuda:6'), in_proj_covar=tensor([0.0568, 0.0705, 0.0579, 0.0503, 0.0445, 0.0454, 0.0590, 0.0545], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:35:40,343 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-29 19:35:42,724 INFO [train.py:904] (6/8) Epoch 13, batch 5950, loss[loss=0.2193, simple_loss=0.31, pruned_loss=0.06432, over 17009.00 frames. ], tot_loss[loss=0.2223, simple_loss=0.3055, pruned_loss=0.06953, over 3059274.91 frames. ], batch size: 41, lr: 5.23e-03, grad_scale: 4.0 2023-04-29 19:36:06,760 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-29 19:36:21,828 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.141e+02 2.934e+02 3.631e+02 4.257e+02 8.531e+02, threshold=7.262e+02, percent-clipped=1.0 2023-04-29 19:36:26,821 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127779.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:36:33,431 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127783.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 19:36:52,376 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3215, 4.0582, 4.0116, 2.7331, 3.5140, 3.9667, 3.5576, 2.3161], device='cuda:6'), covar=tensor([0.0455, 0.0028, 0.0033, 0.0310, 0.0090, 0.0091, 0.0072, 0.0369], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0070, 0.0072, 0.0128, 0.0084, 0.0093, 0.0083, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 19:37:04,075 INFO [train.py:904] (6/8) Epoch 13, batch 6000, loss[loss=0.2112, simple_loss=0.2953, pruned_loss=0.0636, over 16535.00 frames. ], tot_loss[loss=0.2213, simple_loss=0.3047, pruned_loss=0.06897, over 3059059.49 frames. ], batch size: 75, lr: 5.23e-03, grad_scale: 8.0 2023-04-29 19:37:04,075 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 19:37:14,214 INFO [train.py:938] (6/8) Epoch 13, validation: loss=0.1599, simple_loss=0.2726, pruned_loss=0.02359, over 944034.00 frames. 2023-04-29 19:37:14,215 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 19:37:56,140 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-29 19:38:13,886 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127840.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:38:32,720 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127850.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:38:35,158 INFO [train.py:904] (6/8) Epoch 13, batch 6050, loss[loss=0.2057, simple_loss=0.3006, pruned_loss=0.05534, over 16733.00 frames. ], tot_loss[loss=0.2196, simple_loss=0.3028, pruned_loss=0.06821, over 3077906.47 frames. ], batch size: 89, lr: 5.23e-03, grad_scale: 8.0 2023-04-29 19:38:52,679 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9657, 4.0624, 3.8656, 3.6003, 3.5850, 3.9496, 3.6885, 3.7050], device='cuda:6'), covar=tensor([0.0736, 0.0739, 0.0310, 0.0306, 0.0780, 0.0499, 0.0946, 0.0725], device='cuda:6'), in_proj_covar=tensor([0.0253, 0.0342, 0.0300, 0.0279, 0.0320, 0.0325, 0.0203, 0.0350], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:39:07,461 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127872.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:39:09,012 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-04-29 19:39:14,937 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 2.979e+02 3.574e+02 4.345e+02 1.015e+03, threshold=7.148e+02, percent-clipped=2.0 2023-04-29 19:39:28,467 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127885.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:39:40,377 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2085, 4.2449, 4.3970, 4.2644, 4.2879, 4.7842, 4.3997, 4.1288], device='cuda:6'), covar=tensor([0.1679, 0.2049, 0.2187, 0.2046, 0.2755, 0.1150, 0.1488, 0.2614], device='cuda:6'), in_proj_covar=tensor([0.0367, 0.0513, 0.0558, 0.0436, 0.0594, 0.0580, 0.0444, 0.0591], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 19:39:55,360 INFO [train.py:904] (6/8) Epoch 13, batch 6100, loss[loss=0.2089, simple_loss=0.2994, pruned_loss=0.05924, over 16346.00 frames. ], tot_loss[loss=0.2178, simple_loss=0.3023, pruned_loss=0.0667, over 3102110.69 frames. ], batch size: 165, lr: 5.23e-03, grad_scale: 8.0 2023-04-29 19:40:01,568 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127905.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:40:12,636 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127911.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:40:47,042 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127933.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:41:15,198 INFO [train.py:904] (6/8) Epoch 13, batch 6150, loss[loss=0.2259, simple_loss=0.3107, pruned_loss=0.07048, over 15328.00 frames. ], tot_loss[loss=0.2159, simple_loss=0.3001, pruned_loss=0.06586, over 3113280.36 frames. ], batch size: 190, lr: 5.23e-03, grad_scale: 8.0 2023-04-29 19:41:17,426 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=127953.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:41:25,630 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2442, 2.1665, 2.6402, 3.1403, 3.0598, 3.7228, 2.0194, 3.6553], device='cuda:6'), covar=tensor([0.0144, 0.0365, 0.0272, 0.0203, 0.0195, 0.0091, 0.0425, 0.0079], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0176, 0.0160, 0.0163, 0.0173, 0.0130, 0.0176, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 19:41:48,816 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4114, 3.5735, 3.6241, 2.0855, 3.1258, 2.4385, 3.9552, 3.7980], device='cuda:6'), covar=tensor([0.0223, 0.0731, 0.0580, 0.1874, 0.0733, 0.0903, 0.0528, 0.0802], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0149, 0.0159, 0.0145, 0.0139, 0.0126, 0.0139, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 19:41:56,814 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.069e+02 3.037e+02 3.583e+02 4.334e+02 1.010e+03, threshold=7.167e+02, percent-clipped=2.0 2023-04-29 19:42:39,405 INFO [train.py:904] (6/8) Epoch 13, batch 6200, loss[loss=0.1988, simple_loss=0.2902, pruned_loss=0.05377, over 17115.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2981, pruned_loss=0.06519, over 3115227.96 frames. ], batch size: 49, lr: 5.23e-03, grad_scale: 4.0 2023-04-29 19:43:05,152 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128018.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:43:44,214 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2787, 2.0663, 2.2101, 3.9979, 1.9528, 2.4247, 2.1488, 2.1956], device='cuda:6'), covar=tensor([0.1086, 0.3461, 0.2433, 0.0436, 0.4224, 0.2597, 0.3456, 0.3392], device='cuda:6'), in_proj_covar=tensor([0.0366, 0.0399, 0.0332, 0.0314, 0.0414, 0.0461, 0.0366, 0.0465], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:43:57,666 INFO [train.py:904] (6/8) Epoch 13, batch 6250, loss[loss=0.1914, simple_loss=0.2858, pruned_loss=0.04851, over 16809.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2976, pruned_loss=0.06545, over 3098299.30 frames. ], batch size: 102, lr: 5.23e-03, grad_scale: 4.0 2023-04-29 19:44:37,187 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.989e+02 2.817e+02 3.684e+02 4.282e+02 1.153e+03, threshold=7.367e+02, percent-clipped=4.0 2023-04-29 19:44:45,114 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128083.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 19:45:13,035 INFO [train.py:904] (6/8) Epoch 13, batch 6300, loss[loss=0.2267, simple_loss=0.3124, pruned_loss=0.07048, over 16797.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2974, pruned_loss=0.06494, over 3099980.77 frames. ], batch size: 39, lr: 5.23e-03, grad_scale: 4.0 2023-04-29 19:45:51,358 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.06 vs. limit=5.0 2023-04-29 19:45:53,713 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5498, 4.8053, 4.6240, 4.5634, 4.3185, 4.3170, 4.3613, 4.8730], device='cuda:6'), covar=tensor([0.1014, 0.0878, 0.0914, 0.0793, 0.0782, 0.1161, 0.1010, 0.0915], device='cuda:6'), in_proj_covar=tensor([0.0562, 0.0696, 0.0571, 0.0500, 0.0439, 0.0449, 0.0585, 0.0539], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:46:02,725 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=128131.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 19:46:08,201 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128135.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:46:14,628 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9925, 2.4778, 2.6014, 1.9337, 2.6907, 2.7843, 2.3891, 2.3259], device='cuda:6'), covar=tensor([0.0735, 0.0210, 0.0194, 0.0885, 0.0088, 0.0199, 0.0411, 0.0452], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0101, 0.0088, 0.0138, 0.0069, 0.0109, 0.0121, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 19:46:33,879 INFO [train.py:904] (6/8) Epoch 13, batch 6350, loss[loss=0.2394, simple_loss=0.3187, pruned_loss=0.08005, over 16544.00 frames. ], tot_loss[loss=0.2158, simple_loss=0.2985, pruned_loss=0.06655, over 3076366.52 frames. ], batch size: 68, lr: 5.22e-03, grad_scale: 4.0 2023-04-29 19:47:00,130 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9571, 3.2373, 3.1980, 2.1191, 2.9757, 3.1930, 3.0276, 1.9791], device='cuda:6'), covar=tensor([0.0451, 0.0041, 0.0050, 0.0344, 0.0082, 0.0104, 0.0075, 0.0370], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0069, 0.0072, 0.0126, 0.0083, 0.0093, 0.0082, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 19:47:13,675 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.951e+02 3.177e+02 3.833e+02 4.854e+02 7.993e+02, threshold=7.666e+02, percent-clipped=3.0 2023-04-29 19:47:16,136 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2440, 1.8807, 2.7162, 3.0709, 3.0622, 3.5439, 1.8542, 3.4634], device='cuda:6'), covar=tensor([0.0124, 0.0421, 0.0225, 0.0169, 0.0164, 0.0093, 0.0498, 0.0069], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0176, 0.0161, 0.0162, 0.0174, 0.0131, 0.0176, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 19:47:25,349 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128185.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:47:49,923 INFO [train.py:904] (6/8) Epoch 13, batch 6400, loss[loss=0.2596, simple_loss=0.3227, pruned_loss=0.09824, over 11175.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.299, pruned_loss=0.06734, over 3093364.69 frames. ], batch size: 247, lr: 5.22e-03, grad_scale: 8.0 2023-04-29 19:47:56,272 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128206.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:47:59,991 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.93 vs. limit=5.0 2023-04-29 19:48:11,446 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5743, 2.3594, 2.2704, 3.6981, 2.2088, 3.7634, 1.4330, 2.6099], device='cuda:6'), covar=tensor([0.1608, 0.0954, 0.1430, 0.0200, 0.0251, 0.0455, 0.1927, 0.1008], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0162, 0.0181, 0.0155, 0.0200, 0.0207, 0.0183, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 19:48:30,055 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128228.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:48:35,523 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7108, 3.9615, 3.0441, 2.2083, 2.7651, 2.4079, 4.1481, 3.5811], device='cuda:6'), covar=tensor([0.2611, 0.0660, 0.1541, 0.2635, 0.2439, 0.1860, 0.0473, 0.1143], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0260, 0.0288, 0.0286, 0.0283, 0.0228, 0.0273, 0.0306], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:48:36,411 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=128233.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:48:47,735 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9505, 2.7010, 2.7111, 2.0055, 2.6711, 2.1403, 2.8220, 2.8946], device='cuda:6'), covar=tensor([0.0242, 0.0774, 0.0525, 0.1700, 0.0734, 0.0956, 0.0513, 0.0573], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0149, 0.0160, 0.0145, 0.0139, 0.0126, 0.0139, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 19:49:02,069 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4819, 3.5164, 2.0139, 3.9874, 2.6271, 3.9058, 2.1327, 2.7169], device='cuda:6'), covar=tensor([0.0231, 0.0348, 0.1614, 0.0125, 0.0746, 0.0525, 0.1505, 0.0752], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0167, 0.0190, 0.0137, 0.0167, 0.0207, 0.0197, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 19:49:04,023 INFO [train.py:904] (6/8) Epoch 13, batch 6450, loss[loss=0.2548, simple_loss=0.3199, pruned_loss=0.09483, over 11555.00 frames. ], tot_loss[loss=0.2151, simple_loss=0.2981, pruned_loss=0.066, over 3110185.00 frames. ], batch size: 247, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:49:48,882 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.913e+02 2.912e+02 3.353e+02 4.052e+02 7.402e+02, threshold=6.705e+02, percent-clipped=0.0 2023-04-29 19:49:52,059 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2023-04-29 19:49:59,249 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6090, 1.6534, 2.1306, 2.4345, 2.4953, 2.8615, 1.6761, 2.7526], device='cuda:6'), covar=tensor([0.0162, 0.0412, 0.0262, 0.0234, 0.0228, 0.0139, 0.0442, 0.0104], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0176, 0.0160, 0.0162, 0.0174, 0.0131, 0.0177, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 19:50:21,894 INFO [train.py:904] (6/8) Epoch 13, batch 6500, loss[loss=0.2602, simple_loss=0.3157, pruned_loss=0.1023, over 11594.00 frames. ], tot_loss[loss=0.2129, simple_loss=0.2958, pruned_loss=0.06496, over 3113958.37 frames. ], batch size: 247, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:50:27,156 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9835, 3.1627, 3.1552, 2.1117, 2.9257, 3.1221, 3.0177, 1.9431], device='cuda:6'), covar=tensor([0.0463, 0.0046, 0.0052, 0.0345, 0.0095, 0.0106, 0.0080, 0.0392], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0070, 0.0072, 0.0127, 0.0083, 0.0094, 0.0083, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 19:50:42,978 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0431, 2.4933, 2.6030, 1.9212, 2.7331, 2.7910, 2.3860, 2.3947], device='cuda:6'), covar=tensor([0.0675, 0.0209, 0.0206, 0.0889, 0.0103, 0.0233, 0.0415, 0.0408], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0101, 0.0088, 0.0138, 0.0069, 0.0109, 0.0121, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 19:50:45,385 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128318.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:50:48,256 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128320.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:51:03,482 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-04-29 19:51:39,453 INFO [train.py:904] (6/8) Epoch 13, batch 6550, loss[loss=0.2089, simple_loss=0.3068, pruned_loss=0.05555, over 16897.00 frames. ], tot_loss[loss=0.2146, simple_loss=0.2983, pruned_loss=0.06545, over 3109750.22 frames. ], batch size: 116, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:52:01,658 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=128366.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:52:22,244 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9512, 5.2164, 4.9730, 4.9697, 4.7220, 4.6811, 4.6327, 5.3049], device='cuda:6'), covar=tensor([0.0999, 0.0778, 0.0980, 0.0728, 0.0813, 0.0836, 0.1028, 0.0873], device='cuda:6'), in_proj_covar=tensor([0.0564, 0.0697, 0.0572, 0.0501, 0.0439, 0.0448, 0.0585, 0.0538], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:52:22,963 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.909e+02 2.847e+02 3.449e+02 4.250e+02 8.173e+02, threshold=6.898e+02, percent-clipped=4.0 2023-04-29 19:52:25,029 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128381.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:52:36,641 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2511, 4.3440, 4.1612, 3.9491, 3.8429, 4.2589, 4.0029, 4.0058], device='cuda:6'), covar=tensor([0.0640, 0.0525, 0.0260, 0.0243, 0.0832, 0.0408, 0.0606, 0.0607], device='cuda:6'), in_proj_covar=tensor([0.0248, 0.0340, 0.0297, 0.0277, 0.0317, 0.0321, 0.0202, 0.0347], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:52:56,141 INFO [train.py:904] (6/8) Epoch 13, batch 6600, loss[loss=0.2223, simple_loss=0.3, pruned_loss=0.07224, over 16649.00 frames. ], tot_loss[loss=0.2165, simple_loss=0.3008, pruned_loss=0.06612, over 3115055.63 frames. ], batch size: 57, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:53:14,133 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128413.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:53:31,211 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-29 19:53:48,832 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128435.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:54:14,487 INFO [train.py:904] (6/8) Epoch 13, batch 6650, loss[loss=0.2006, simple_loss=0.2811, pruned_loss=0.06001, over 17072.00 frames. ], tot_loss[loss=0.2184, simple_loss=0.3019, pruned_loss=0.06751, over 3104213.38 frames. ], batch size: 53, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:54:47,989 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128474.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:54:56,948 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.056e+02 3.058e+02 3.713e+02 4.648e+02 8.477e+02, threshold=7.425e+02, percent-clipped=4.0 2023-04-29 19:55:02,267 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=128483.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:55:30,391 INFO [train.py:904] (6/8) Epoch 13, batch 6700, loss[loss=0.2652, simple_loss=0.3267, pruned_loss=0.1019, over 11363.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.2995, pruned_loss=0.0666, over 3116950.10 frames. ], batch size: 246, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:55:36,449 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128506.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:55:52,245 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8536, 2.7633, 2.7632, 2.0594, 2.6227, 2.1665, 2.6928, 2.9094], device='cuda:6'), covar=tensor([0.0277, 0.0610, 0.0496, 0.1663, 0.0763, 0.0861, 0.0587, 0.0643], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0149, 0.0160, 0.0145, 0.0139, 0.0126, 0.0138, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 19:56:10,095 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128528.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:56:46,042 INFO [train.py:904] (6/8) Epoch 13, batch 6750, loss[loss=0.1846, simple_loss=0.2779, pruned_loss=0.04567, over 16890.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.2999, pruned_loss=0.06775, over 3103277.27 frames. ], batch size: 96, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:56:49,450 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=128554.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:57:07,080 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3686, 3.2757, 3.3658, 3.4775, 3.5003, 3.2802, 3.4797, 3.5305], device='cuda:6'), covar=tensor([0.1200, 0.0904, 0.1031, 0.0574, 0.0661, 0.2141, 0.0986, 0.0788], device='cuda:6'), in_proj_covar=tensor([0.0548, 0.0681, 0.0813, 0.0694, 0.0529, 0.0534, 0.0559, 0.0640], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:57:22,972 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=128576.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:57:28,178 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.231e+02 3.101e+02 3.923e+02 4.752e+02 1.055e+03, threshold=7.847e+02, percent-clipped=2.0 2023-04-29 19:58:01,812 INFO [train.py:904] (6/8) Epoch 13, batch 6800, loss[loss=0.2534, simple_loss=0.317, pruned_loss=0.09489, over 11415.00 frames. ], tot_loss[loss=0.2169, simple_loss=0.2992, pruned_loss=0.06732, over 3086348.15 frames. ], batch size: 248, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 19:58:29,213 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7815, 1.7894, 1.5121, 1.5209, 1.9182, 1.5616, 1.7364, 1.8901], device='cuda:6'), covar=tensor([0.0133, 0.0194, 0.0296, 0.0256, 0.0152, 0.0219, 0.0139, 0.0152], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0210, 0.0204, 0.0205, 0.0208, 0.0208, 0.0214, 0.0200], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:58:52,203 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0147, 5.0253, 4.8290, 4.1538, 4.8778, 2.0251, 4.6520, 4.6461], device='cuda:6'), covar=tensor([0.0063, 0.0055, 0.0130, 0.0324, 0.0065, 0.2150, 0.0102, 0.0158], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0125, 0.0171, 0.0161, 0.0143, 0.0185, 0.0159, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 19:59:19,101 INFO [train.py:904] (6/8) Epoch 13, batch 6850, loss[loss=0.2271, simple_loss=0.3201, pruned_loss=0.06702, over 16889.00 frames. ], tot_loss[loss=0.2191, simple_loss=0.3008, pruned_loss=0.06868, over 3058654.87 frames. ], batch size: 116, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 19:59:55,715 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128676.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:00:00,829 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.804e+02 2.799e+02 3.352e+02 4.100e+02 8.801e+02, threshold=6.704e+02, percent-clipped=1.0 2023-04-29 20:00:34,560 INFO [train.py:904] (6/8) Epoch 13, batch 6900, loss[loss=0.2463, simple_loss=0.3244, pruned_loss=0.08413, over 15466.00 frames. ], tot_loss[loss=0.2198, simple_loss=0.3028, pruned_loss=0.06839, over 3069466.45 frames. ], batch size: 191, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 20:01:02,034 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-29 20:01:52,386 INFO [train.py:904] (6/8) Epoch 13, batch 6950, loss[loss=0.1924, simple_loss=0.2832, pruned_loss=0.05079, over 16834.00 frames. ], tot_loss[loss=0.2205, simple_loss=0.3038, pruned_loss=0.06861, over 3096947.43 frames. ], batch size: 102, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 20:02:20,015 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128769.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:02:36,165 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.242e+02 3.339e+02 3.931e+02 4.735e+02 7.907e+02, threshold=7.862e+02, percent-clipped=3.0 2023-04-29 20:02:38,374 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128781.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:03:08,750 INFO [train.py:904] (6/8) Epoch 13, batch 7000, loss[loss=0.2389, simple_loss=0.3039, pruned_loss=0.08693, over 11343.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.304, pruned_loss=0.06789, over 3104122.14 frames. ], batch size: 247, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 20:04:01,201 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2454, 2.0439, 2.0957, 3.9556, 2.0385, 2.4802, 2.1505, 2.2265], device='cuda:6'), covar=tensor([0.1070, 0.3493, 0.2644, 0.0441, 0.4036, 0.2397, 0.3302, 0.3232], device='cuda:6'), in_proj_covar=tensor([0.0365, 0.0400, 0.0333, 0.0313, 0.0415, 0.0460, 0.0367, 0.0467], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 20:04:07,245 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128842.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:04:20,571 INFO [train.py:904] (6/8) Epoch 13, batch 7050, loss[loss=0.2495, simple_loss=0.3156, pruned_loss=0.09175, over 11474.00 frames. ], tot_loss[loss=0.2194, simple_loss=0.3041, pruned_loss=0.06735, over 3100660.58 frames. ], batch size: 246, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 20:04:45,993 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8572, 3.9267, 2.3328, 4.7145, 2.9974, 4.5766, 2.3353, 3.0360], device='cuda:6'), covar=tensor([0.0256, 0.0351, 0.1588, 0.0120, 0.0747, 0.0413, 0.1476, 0.0714], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0165, 0.0188, 0.0136, 0.0167, 0.0205, 0.0194, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 20:05:02,585 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.996e+02 2.845e+02 3.584e+02 4.333e+02 9.135e+02, threshold=7.167e+02, percent-clipped=3.0 2023-04-29 20:05:19,847 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128890.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 20:05:33,230 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3061, 3.3834, 2.0117, 3.7160, 2.5463, 3.6839, 1.9939, 2.6084], device='cuda:6'), covar=tensor([0.0241, 0.0365, 0.1530, 0.0141, 0.0754, 0.0545, 0.1527, 0.0770], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0165, 0.0188, 0.0136, 0.0167, 0.0206, 0.0194, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 20:05:38,297 INFO [train.py:904] (6/8) Epoch 13, batch 7100, loss[loss=0.2118, simple_loss=0.3025, pruned_loss=0.06055, over 16704.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.3027, pruned_loss=0.06745, over 3086724.32 frames. ], batch size: 134, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 20:06:56,862 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128951.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 20:06:57,456 INFO [train.py:904] (6/8) Epoch 13, batch 7150, loss[loss=0.2761, simple_loss=0.3378, pruned_loss=0.1071, over 11784.00 frames. ], tot_loss[loss=0.2174, simple_loss=0.3008, pruned_loss=0.06696, over 3098141.11 frames. ], batch size: 246, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 20:07:34,009 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128976.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:07:39,435 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.000e+02 3.228e+02 3.806e+02 5.170e+02 8.621e+02, threshold=7.612e+02, percent-clipped=3.0 2023-04-29 20:08:12,085 INFO [train.py:904] (6/8) Epoch 13, batch 7200, loss[loss=0.1929, simple_loss=0.2883, pruned_loss=0.04879, over 16922.00 frames. ], tot_loss[loss=0.214, simple_loss=0.2983, pruned_loss=0.06484, over 3109579.18 frames. ], batch size: 90, lr: 5.21e-03, grad_scale: 8.0 2023-04-29 20:08:45,129 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=129024.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:09:29,932 INFO [train.py:904] (6/8) Epoch 13, batch 7250, loss[loss=0.1917, simple_loss=0.274, pruned_loss=0.05474, over 16955.00 frames. ], tot_loss[loss=0.2125, simple_loss=0.2965, pruned_loss=0.06428, over 3110529.44 frames. ], batch size: 109, lr: 5.21e-03, grad_scale: 8.0 2023-04-29 20:09:56,916 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=129069.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:10:12,180 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.987e+02 2.885e+02 3.444e+02 4.352e+02 7.294e+02, threshold=6.888e+02, percent-clipped=0.0 2023-04-29 20:10:45,713 INFO [train.py:904] (6/8) Epoch 13, batch 7300, loss[loss=0.2324, simple_loss=0.3118, pruned_loss=0.07652, over 15308.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.296, pruned_loss=0.0642, over 3113918.53 frames. ], batch size: 190, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:11:09,712 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=129117.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:11:29,796 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5740, 2.5163, 1.8685, 2.6933, 2.0769, 2.6723, 2.0116, 2.2273], device='cuda:6'), covar=tensor([0.0246, 0.0289, 0.1092, 0.0157, 0.0556, 0.0343, 0.1046, 0.0582], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0165, 0.0190, 0.0136, 0.0169, 0.0207, 0.0197, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 20:11:40,200 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129137.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:11:52,940 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.25 vs. limit=5.0 2023-04-29 20:12:02,354 INFO [train.py:904] (6/8) Epoch 13, batch 7350, loss[loss=0.1945, simple_loss=0.2825, pruned_loss=0.05325, over 16237.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2973, pruned_loss=0.065, over 3106389.70 frames. ], batch size: 35, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:12:28,644 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 20:12:46,221 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.932e+02 2.924e+02 3.417e+02 4.096e+02 1.600e+03, threshold=6.834e+02, percent-clipped=4.0 2023-04-29 20:13:21,034 INFO [train.py:904] (6/8) Epoch 13, batch 7400, loss[loss=0.2174, simple_loss=0.3041, pruned_loss=0.06535, over 16789.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.2983, pruned_loss=0.06536, over 3099144.46 frames. ], batch size: 39, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:14:15,551 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1065, 5.0688, 4.8521, 3.9278, 4.9116, 1.7907, 4.6645, 4.6879], device='cuda:6'), covar=tensor([0.0097, 0.0078, 0.0172, 0.0466, 0.0112, 0.2618, 0.0132, 0.0211], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0122, 0.0167, 0.0159, 0.0140, 0.0182, 0.0157, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 20:14:32,917 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129246.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 20:14:41,688 INFO [train.py:904] (6/8) Epoch 13, batch 7450, loss[loss=0.2219, simple_loss=0.3113, pruned_loss=0.06625, over 16875.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2996, pruned_loss=0.06681, over 3106287.17 frames. ], batch size: 116, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:15:30,917 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.914e+02 3.121e+02 3.842e+02 4.432e+02 7.351e+02, threshold=7.685e+02, percent-clipped=1.0 2023-04-29 20:16:03,370 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-29 20:16:05,625 INFO [train.py:904] (6/8) Epoch 13, batch 7500, loss[loss=0.2114, simple_loss=0.284, pruned_loss=0.06935, over 11166.00 frames. ], tot_loss[loss=0.2165, simple_loss=0.3, pruned_loss=0.06647, over 3096020.81 frames. ], batch size: 246, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:16:19,391 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-29 20:17:24,560 INFO [train.py:904] (6/8) Epoch 13, batch 7550, loss[loss=0.1964, simple_loss=0.283, pruned_loss=0.05487, over 16672.00 frames. ], tot_loss[loss=0.2162, simple_loss=0.299, pruned_loss=0.06673, over 3075813.95 frames. ], batch size: 134, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:17:43,654 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4418, 5.7776, 5.4823, 5.5149, 5.1973, 5.0615, 5.1855, 5.8467], device='cuda:6'), covar=tensor([0.0983, 0.0727, 0.1051, 0.0735, 0.0740, 0.0706, 0.1020, 0.0758], device='cuda:6'), in_proj_covar=tensor([0.0568, 0.0702, 0.0576, 0.0500, 0.0442, 0.0455, 0.0587, 0.0538], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 20:17:58,698 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2023-04-29 20:18:07,709 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.866e+02 2.820e+02 3.717e+02 4.924e+02 9.310e+02, threshold=7.434e+02, percent-clipped=3.0 2023-04-29 20:18:41,433 INFO [train.py:904] (6/8) Epoch 13, batch 7600, loss[loss=0.2183, simple_loss=0.307, pruned_loss=0.06482, over 16721.00 frames. ], tot_loss[loss=0.2163, simple_loss=0.2985, pruned_loss=0.06705, over 3078936.79 frames. ], batch size: 89, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:18:59,702 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=129413.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 20:19:37,565 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=129437.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:19:41,486 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2023-04-29 20:19:44,180 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.39 vs. limit=5.0 2023-04-29 20:20:00,017 INFO [train.py:904] (6/8) Epoch 13, batch 7650, loss[loss=0.201, simple_loss=0.2872, pruned_loss=0.0574, over 16472.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.2993, pruned_loss=0.0677, over 3074308.24 frames. ], batch size: 68, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:20:12,424 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-29 20:20:36,400 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=129474.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 20:20:44,159 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.211e+02 3.428e+02 4.160e+02 4.831e+02 1.059e+03, threshold=8.320e+02, percent-clipped=3.0 2023-04-29 20:20:53,102 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=129485.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:21:06,964 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0705, 5.3788, 5.1628, 5.1260, 4.8440, 4.7867, 4.8282, 5.4716], device='cuda:6'), covar=tensor([0.1035, 0.0801, 0.0976, 0.0776, 0.0790, 0.0763, 0.1053, 0.0834], device='cuda:6'), in_proj_covar=tensor([0.0568, 0.0702, 0.0576, 0.0499, 0.0441, 0.0455, 0.0587, 0.0536], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 20:21:18,576 INFO [train.py:904] (6/8) Epoch 13, batch 7700, loss[loss=0.2113, simple_loss=0.2939, pruned_loss=0.06437, over 16399.00 frames. ], tot_loss[loss=0.2174, simple_loss=0.2992, pruned_loss=0.06784, over 3071894.42 frames. ], batch size: 146, lr: 5.20e-03, grad_scale: 4.0 2023-04-29 20:21:54,410 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6407, 1.7632, 2.3586, 2.7783, 2.6404, 3.1108, 1.8927, 3.0376], device='cuda:6'), covar=tensor([0.0208, 0.0417, 0.0250, 0.0220, 0.0231, 0.0134, 0.0394, 0.0110], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0174, 0.0158, 0.0162, 0.0173, 0.0128, 0.0175, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 20:22:26,972 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=129546.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 20:22:35,875 INFO [train.py:904] (6/8) Epoch 13, batch 7750, loss[loss=0.1917, simple_loss=0.2792, pruned_loss=0.05209, over 17047.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.2987, pruned_loss=0.06674, over 3094671.80 frames. ], batch size: 55, lr: 5.20e-03, grad_scale: 4.0 2023-04-29 20:23:20,357 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.834e+02 3.197e+02 3.787e+02 4.587e+02 8.661e+02, threshold=7.574e+02, percent-clipped=1.0 2023-04-29 20:23:40,229 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=129594.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 20:23:52,176 INFO [train.py:904] (6/8) Epoch 13, batch 7800, loss[loss=0.1881, simple_loss=0.2855, pruned_loss=0.04539, over 16481.00 frames. ], tot_loss[loss=0.2172, simple_loss=0.2997, pruned_loss=0.06733, over 3101177.31 frames. ], batch size: 75, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:24:09,311 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=129612.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:24:48,744 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6409, 4.8300, 4.9981, 4.8196, 4.8038, 5.3877, 4.8923, 4.6687], device='cuda:6'), covar=tensor([0.1029, 0.1835, 0.2147, 0.1950, 0.2490, 0.0957, 0.1474, 0.2308], device='cuda:6'), in_proj_covar=tensor([0.0366, 0.0519, 0.0567, 0.0441, 0.0597, 0.0584, 0.0450, 0.0593], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 20:25:09,849 INFO [train.py:904] (6/8) Epoch 13, batch 7850, loss[loss=0.1973, simple_loss=0.2898, pruned_loss=0.05237, over 16192.00 frames. ], tot_loss[loss=0.2178, simple_loss=0.3009, pruned_loss=0.06731, over 3111843.90 frames. ], batch size: 165, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:25:26,674 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7325, 4.8567, 5.1016, 4.9144, 4.9320, 5.5023, 4.9855, 4.7584], device='cuda:6'), covar=tensor([0.0995, 0.1959, 0.2074, 0.1835, 0.2449, 0.0943, 0.1567, 0.2365], device='cuda:6'), in_proj_covar=tensor([0.0366, 0.0518, 0.0566, 0.0441, 0.0597, 0.0584, 0.0450, 0.0593], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 20:25:40,333 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=129673.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:25:52,267 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 3.020e+02 3.492e+02 4.287e+02 1.158e+03, threshold=6.983e+02, percent-clipped=4.0 2023-04-29 20:26:02,617 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.8351, 6.1883, 5.9143, 6.0168, 5.5760, 5.4456, 5.6537, 6.2937], device='cuda:6'), covar=tensor([0.0966, 0.0710, 0.0789, 0.0689, 0.0743, 0.0537, 0.0953, 0.0785], device='cuda:6'), in_proj_covar=tensor([0.0570, 0.0703, 0.0580, 0.0502, 0.0442, 0.0456, 0.0588, 0.0538], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 20:26:22,707 INFO [train.py:904] (6/8) Epoch 13, batch 7900, loss[loss=0.2177, simple_loss=0.3035, pruned_loss=0.06593, over 15323.00 frames. ], tot_loss[loss=0.2163, simple_loss=0.2997, pruned_loss=0.06642, over 3113794.38 frames. ], batch size: 190, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:27:36,766 INFO [train.py:904] (6/8) Epoch 13, batch 7950, loss[loss=0.2706, simple_loss=0.3252, pruned_loss=0.108, over 11460.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2999, pruned_loss=0.06667, over 3120126.24 frames. ], batch size: 247, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:28:02,252 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129769.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 20:28:18,206 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.009e+02 2.840e+02 3.431e+02 4.053e+02 8.396e+02, threshold=6.863e+02, percent-clipped=2.0 2023-04-29 20:28:33,917 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=129791.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:28:49,440 INFO [train.py:904] (6/8) Epoch 13, batch 8000, loss[loss=0.2035, simple_loss=0.2881, pruned_loss=0.05948, over 17163.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.3002, pruned_loss=0.06758, over 3110531.37 frames. ], batch size: 40, lr: 5.19e-03, grad_scale: 8.0 2023-04-29 20:30:02,333 INFO [train.py:904] (6/8) Epoch 13, batch 8050, loss[loss=0.2213, simple_loss=0.3016, pruned_loss=0.07048, over 15514.00 frames. ], tot_loss[loss=0.2179, simple_loss=0.3006, pruned_loss=0.06757, over 3108699.38 frames. ], batch size: 190, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:30:02,895 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=129852.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:30:25,135 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-29 20:30:37,838 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-29 20:30:45,831 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.353e+02 3.065e+02 3.748e+02 4.862e+02 1.232e+03, threshold=7.497e+02, percent-clipped=5.0 2023-04-29 20:30:53,733 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7124, 3.6337, 4.1196, 2.0384, 4.3293, 4.3401, 2.9142, 3.1646], device='cuda:6'), covar=tensor([0.0702, 0.0212, 0.0168, 0.1014, 0.0045, 0.0110, 0.0429, 0.0396], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0101, 0.0089, 0.0138, 0.0069, 0.0109, 0.0121, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 20:31:15,197 INFO [train.py:904] (6/8) Epoch 13, batch 8100, loss[loss=0.2236, simple_loss=0.3071, pruned_loss=0.07006, over 15265.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.2997, pruned_loss=0.06702, over 3102669.14 frames. ], batch size: 191, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:32:29,503 INFO [train.py:904] (6/8) Epoch 13, batch 8150, loss[loss=0.189, simple_loss=0.2701, pruned_loss=0.05389, over 16638.00 frames. ], tot_loss[loss=0.2143, simple_loss=0.2972, pruned_loss=0.0657, over 3111769.90 frames. ], batch size: 134, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:32:53,000 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129968.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:33:14,447 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.827e+02 2.878e+02 3.437e+02 4.165e+02 9.532e+02, threshold=6.873e+02, percent-clipped=2.0 2023-04-29 20:33:46,561 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130000.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:33:48,621 INFO [train.py:904] (6/8) Epoch 13, batch 8200, loss[loss=0.2348, simple_loss=0.3152, pruned_loss=0.07719, over 15417.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2954, pruned_loss=0.0654, over 3112640.53 frames. ], batch size: 191, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:33:49,153 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130002.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:34:58,964 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5410, 3.4347, 3.4692, 2.6972, 3.4030, 1.8925, 3.1431, 2.9238], device='cuda:6'), covar=tensor([0.0132, 0.0131, 0.0163, 0.0328, 0.0118, 0.2546, 0.0151, 0.0286], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0123, 0.0169, 0.0159, 0.0141, 0.0184, 0.0157, 0.0153], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 20:35:09,258 INFO [train.py:904] (6/8) Epoch 13, batch 8250, loss[loss=0.2023, simple_loss=0.2933, pruned_loss=0.05568, over 16142.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.2942, pruned_loss=0.0632, over 3084903.34 frames. ], batch size: 165, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:35:24,107 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130061.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:35:27,855 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130063.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 20:35:37,694 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130069.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 20:35:57,079 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.490e+02 2.705e+02 3.349e+02 4.037e+02 8.257e+02, threshold=6.697e+02, percent-clipped=3.0 2023-04-29 20:36:29,983 INFO [train.py:904] (6/8) Epoch 13, batch 8300, loss[loss=0.1851, simple_loss=0.2802, pruned_loss=0.04498, over 16221.00 frames. ], tot_loss[loss=0.206, simple_loss=0.2916, pruned_loss=0.06015, over 3094535.48 frames. ], batch size: 165, lr: 5.18e-03, grad_scale: 4.0 2023-04-29 20:36:34,750 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-04-29 20:36:55,890 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=130117.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 20:37:43,810 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130147.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:37:51,484 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0336, 2.1159, 2.4075, 3.2310, 2.1935, 2.4183, 2.3107, 2.2183], device='cuda:6'), covar=tensor([0.0878, 0.3456, 0.2073, 0.0570, 0.4123, 0.2218, 0.3148, 0.3243], device='cuda:6'), in_proj_covar=tensor([0.0362, 0.0398, 0.0332, 0.0312, 0.0411, 0.0454, 0.0361, 0.0462], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 20:37:51,986 INFO [train.py:904] (6/8) Epoch 13, batch 8350, loss[loss=0.2228, simple_loss=0.3, pruned_loss=0.07279, over 11724.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2902, pruned_loss=0.05817, over 3063275.75 frames. ], batch size: 246, lr: 5.18e-03, grad_scale: 4.0 2023-04-29 20:38:39,963 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.345e+02 2.793e+02 3.218e+02 5.548e+02, threshold=5.587e+02, percent-clipped=0.0 2023-04-29 20:39:12,413 INFO [train.py:904] (6/8) Epoch 13, batch 8400, loss[loss=0.1674, simple_loss=0.2662, pruned_loss=0.03429, over 16722.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2871, pruned_loss=0.05564, over 3070425.85 frames. ], batch size: 89, lr: 5.18e-03, grad_scale: 8.0 2023-04-29 20:40:29,224 INFO [train.py:904] (6/8) Epoch 13, batch 8450, loss[loss=0.1777, simple_loss=0.2748, pruned_loss=0.0403, over 16706.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2846, pruned_loss=0.05378, over 3062128.39 frames. ], batch size: 124, lr: 5.18e-03, grad_scale: 8.0 2023-04-29 20:40:32,063 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7131, 3.9059, 3.0558, 2.1587, 2.5089, 2.3568, 4.1682, 3.3915], device='cuda:6'), covar=tensor([0.2602, 0.0628, 0.1549, 0.2740, 0.2790, 0.1910, 0.0355, 0.1215], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0252, 0.0283, 0.0280, 0.0274, 0.0225, 0.0267, 0.0297], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 20:40:40,911 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8556, 5.1003, 5.3101, 5.1224, 5.1848, 5.7215, 5.1670, 4.9225], device='cuda:6'), covar=tensor([0.0849, 0.2015, 0.1506, 0.1881, 0.2463, 0.0842, 0.1521, 0.2219], device='cuda:6'), in_proj_covar=tensor([0.0354, 0.0499, 0.0545, 0.0428, 0.0576, 0.0570, 0.0438, 0.0571], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 20:40:56,022 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130268.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:41:17,432 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.448e+02 2.422e+02 2.811e+02 3.625e+02 1.309e+03, threshold=5.622e+02, percent-clipped=6.0 2023-04-29 20:41:27,029 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1118, 3.1472, 3.1493, 2.2367, 2.9379, 3.1886, 3.0853, 1.9217], device='cuda:6'), covar=tensor([0.0389, 0.0046, 0.0043, 0.0299, 0.0085, 0.0069, 0.0065, 0.0399], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0068, 0.0070, 0.0125, 0.0082, 0.0091, 0.0080, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 20:41:49,397 INFO [train.py:904] (6/8) Epoch 13, batch 8500, loss[loss=0.161, simple_loss=0.2522, pruned_loss=0.03489, over 15129.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.2803, pruned_loss=0.05076, over 3067617.61 frames. ], batch size: 190, lr: 5.18e-03, grad_scale: 8.0 2023-04-29 20:41:59,795 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3518, 4.3265, 4.7637, 4.7541, 4.7299, 4.4547, 4.3954, 4.3402], device='cuda:6'), covar=tensor([0.0337, 0.0689, 0.0393, 0.0392, 0.0550, 0.0397, 0.1194, 0.0504], device='cuda:6'), in_proj_covar=tensor([0.0344, 0.0364, 0.0363, 0.0349, 0.0408, 0.0387, 0.0478, 0.0309], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 20:42:13,189 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=130316.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:42:45,837 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2696, 3.6065, 3.5781, 2.4262, 3.2854, 3.6055, 3.4191, 2.0234], device='cuda:6'), covar=tensor([0.0424, 0.0032, 0.0037, 0.0316, 0.0069, 0.0062, 0.0057, 0.0425], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0068, 0.0070, 0.0125, 0.0082, 0.0090, 0.0080, 0.0119], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 20:42:47,815 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4687, 3.1558, 3.3496, 1.8479, 3.6168, 3.6898, 2.8995, 2.8879], device='cuda:6'), covar=tensor([0.0670, 0.0238, 0.0211, 0.1126, 0.0052, 0.0118, 0.0367, 0.0378], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0097, 0.0085, 0.0133, 0.0066, 0.0104, 0.0116, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 20:43:10,838 INFO [train.py:904] (6/8) Epoch 13, batch 8550, loss[loss=0.1938, simple_loss=0.2743, pruned_loss=0.05668, over 16406.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2784, pruned_loss=0.05019, over 3031322.43 frames. ], batch size: 68, lr: 5.18e-03, grad_scale: 8.0 2023-04-29 20:43:19,866 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130356.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:43:23,660 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130358.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 20:43:29,884 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-29 20:44:07,712 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.521e+02 2.204e+02 2.928e+02 3.716e+02 7.366e+02, threshold=5.857e+02, percent-clipped=6.0 2023-04-29 20:44:12,840 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130384.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:44:50,594 INFO [train.py:904] (6/8) Epoch 13, batch 8600, loss[loss=0.1901, simple_loss=0.285, pruned_loss=0.04764, over 16186.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2793, pruned_loss=0.04923, over 3040269.43 frames. ], batch size: 165, lr: 5.18e-03, grad_scale: 4.0 2023-04-29 20:45:51,734 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130432.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:46:04,143 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130438.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:46:16,912 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130445.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:46:20,636 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130447.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:46:30,003 INFO [train.py:904] (6/8) Epoch 13, batch 8650, loss[loss=0.1974, simple_loss=0.2953, pruned_loss=0.04974, over 15388.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.2776, pruned_loss=0.04763, over 3056769.77 frames. ], batch size: 191, lr: 5.18e-03, grad_scale: 4.0 2023-04-29 20:47:40,991 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.301e+02 2.678e+02 3.280e+02 8.282e+02, threshold=5.356e+02, percent-clipped=3.0 2023-04-29 20:47:46,598 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8717, 2.7989, 2.6388, 1.9365, 2.5465, 2.7788, 2.6947, 1.8422], device='cuda:6'), covar=tensor([0.0384, 0.0050, 0.0044, 0.0324, 0.0089, 0.0071, 0.0073, 0.0403], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0068, 0.0070, 0.0126, 0.0082, 0.0090, 0.0080, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 20:48:01,242 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130493.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:48:05,351 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=130495.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:48:12,936 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130499.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:48:17,492 INFO [train.py:904] (6/8) Epoch 13, batch 8700, loss[loss=0.1627, simple_loss=0.263, pruned_loss=0.03122, over 16862.00 frames. ], tot_loss[loss=0.1841, simple_loss=0.2749, pruned_loss=0.04662, over 3036159.73 frames. ], batch size: 102, lr: 5.18e-03, grad_scale: 4.0 2023-04-29 20:49:02,921 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9496, 1.9923, 2.2801, 3.1981, 2.1203, 2.2524, 2.2149, 2.1265], device='cuda:6'), covar=tensor([0.1048, 0.3794, 0.2443, 0.0619, 0.4286, 0.2670, 0.3300, 0.3724], device='cuda:6'), in_proj_covar=tensor([0.0359, 0.0396, 0.0330, 0.0308, 0.0408, 0.0450, 0.0358, 0.0457], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 20:49:41,396 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0088, 1.7788, 1.6043, 1.5294, 1.9179, 1.5499, 1.7220, 1.9235], device='cuda:6'), covar=tensor([0.0125, 0.0233, 0.0304, 0.0283, 0.0173, 0.0226, 0.0137, 0.0179], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0203, 0.0197, 0.0197, 0.0201, 0.0201, 0.0201, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 20:49:47,100 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130548.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:49:54,307 INFO [train.py:904] (6/8) Epoch 13, batch 8750, loss[loss=0.193, simple_loss=0.2884, pruned_loss=0.04876, over 16279.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.274, pruned_loss=0.04592, over 3034523.13 frames. ], batch size: 166, lr: 5.18e-03, grad_scale: 4.0 2023-04-29 20:50:30,310 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130565.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:51:07,892 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.672e+02 2.544e+02 3.252e+02 4.020e+02 9.168e+02, threshold=6.504e+02, percent-clipped=9.0 2023-04-29 20:51:48,202 INFO [train.py:904] (6/8) Epoch 13, batch 8800, loss[loss=0.1858, simple_loss=0.2708, pruned_loss=0.05044, over 12487.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2724, pruned_loss=0.04475, over 3053116.43 frames. ], batch size: 248, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 20:52:02,478 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130609.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 20:52:39,868 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130626.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:53:32,459 INFO [train.py:904] (6/8) Epoch 13, batch 8850, loss[loss=0.1569, simple_loss=0.2467, pruned_loss=0.03354, over 12387.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2745, pruned_loss=0.04379, over 3052799.07 frames. ], batch size: 247, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 20:53:41,550 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130656.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:53:45,834 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130658.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 20:54:37,528 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.288e+02 2.852e+02 3.513e+02 7.607e+02, threshold=5.705e+02, percent-clipped=2.0 2023-04-29 20:55:17,113 INFO [train.py:904] (6/8) Epoch 13, batch 8900, loss[loss=0.1731, simple_loss=0.2686, pruned_loss=0.03876, over 16416.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.2754, pruned_loss=0.04374, over 3055940.09 frames. ], batch size: 75, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 20:55:22,771 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=130704.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:55:26,794 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=130706.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:55:29,343 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0977, 2.0722, 2.2011, 3.5928, 2.0085, 2.4376, 2.1688, 2.1853], device='cuda:6'), covar=tensor([0.1129, 0.3275, 0.2476, 0.0480, 0.4112, 0.2176, 0.3364, 0.3237], device='cuda:6'), in_proj_covar=tensor([0.0361, 0.0396, 0.0330, 0.0309, 0.0411, 0.0451, 0.0361, 0.0460], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 20:55:51,105 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1470, 2.6088, 2.6673, 1.9166, 2.8738, 2.9461, 2.5544, 2.4792], device='cuda:6'), covar=tensor([0.0587, 0.0205, 0.0174, 0.0887, 0.0069, 0.0137, 0.0342, 0.0381], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0096, 0.0084, 0.0132, 0.0065, 0.0103, 0.0115, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 20:56:54,227 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130740.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:57:21,721 INFO [train.py:904] (6/8) Epoch 13, batch 8950, loss[loss=0.1987, simple_loss=0.2857, pruned_loss=0.05586, over 16940.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.2745, pruned_loss=0.04349, over 3075400.88 frames. ], batch size: 109, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 20:57:38,164 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2302, 3.6718, 3.6786, 2.5958, 3.4154, 3.7486, 3.5694, 1.9848], device='cuda:6'), covar=tensor([0.0456, 0.0026, 0.0030, 0.0293, 0.0063, 0.0048, 0.0049, 0.0431], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0067, 0.0069, 0.0125, 0.0082, 0.0089, 0.0079, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 20:58:29,463 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.201e+02 2.684e+02 3.167e+02 7.959e+02, threshold=5.368e+02, percent-clipped=1.0 2023-04-29 20:58:41,743 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130788.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:58:54,888 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130794.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:58:55,307 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.40 vs. limit=5.0 2023-04-29 20:59:11,358 INFO [train.py:904] (6/8) Epoch 13, batch 9000, loss[loss=0.1508, simple_loss=0.238, pruned_loss=0.03185, over 16649.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2712, pruned_loss=0.04244, over 3068750.30 frames. ], batch size: 57, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 20:59:11,359 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 20:59:22,055 INFO [train.py:938] (6/8) Epoch 13, validation: loss=0.1517, simple_loss=0.2561, pruned_loss=0.02371, over 944034.00 frames. 2023-04-29 20:59:22,056 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 20:59:25,400 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1894, 2.9802, 3.1057, 1.7973, 3.2889, 3.3837, 2.7386, 2.5843], device='cuda:6'), covar=tensor([0.0736, 0.0215, 0.0157, 0.1071, 0.0066, 0.0117, 0.0391, 0.0441], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0097, 0.0084, 0.0134, 0.0066, 0.0103, 0.0116, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 21:00:53,826 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9707, 4.2822, 4.1252, 4.1178, 3.7221, 3.8432, 3.8885, 4.2473], device='cuda:6'), covar=tensor([0.1002, 0.0815, 0.0851, 0.0717, 0.0737, 0.1602, 0.0914, 0.1000], device='cuda:6'), in_proj_covar=tensor([0.0556, 0.0690, 0.0564, 0.0494, 0.0433, 0.0447, 0.0577, 0.0525], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 21:01:06,010 INFO [train.py:904] (6/8) Epoch 13, batch 9050, loss[loss=0.2023, simple_loss=0.2902, pruned_loss=0.05719, over 12306.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2727, pruned_loss=0.04336, over 3063998.55 frames. ], batch size: 247, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 21:02:07,080 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.631e+02 2.403e+02 2.798e+02 3.395e+02 5.131e+02, threshold=5.596e+02, percent-clipped=0.0 2023-04-29 21:02:52,476 INFO [train.py:904] (6/8) Epoch 13, batch 9100, loss[loss=0.1717, simple_loss=0.2739, pruned_loss=0.03478, over 15400.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2723, pruned_loss=0.04378, over 3060108.08 frames. ], batch size: 191, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 21:02:58,110 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130904.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 21:03:33,403 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130921.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:04:15,615 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-29 21:04:49,577 INFO [train.py:904] (6/8) Epoch 13, batch 9150, loss[loss=0.1748, simple_loss=0.2606, pruned_loss=0.04445, over 12265.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2724, pruned_loss=0.0434, over 3050967.47 frames. ], batch size: 250, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 21:05:52,920 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.726e+02 2.558e+02 2.960e+02 3.606e+02 6.781e+02, threshold=5.920e+02, percent-clipped=4.0 2023-04-29 21:06:31,915 INFO [train.py:904] (6/8) Epoch 13, batch 9200, loss[loss=0.1629, simple_loss=0.2541, pruned_loss=0.03584, over 16656.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2682, pruned_loss=0.04251, over 3064519.77 frames. ], batch size: 57, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 21:07:14,735 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0656, 1.4965, 1.8268, 2.0383, 2.1346, 2.1935, 1.7425, 2.2294], device='cuda:6'), covar=tensor([0.0174, 0.0423, 0.0237, 0.0269, 0.0248, 0.0181, 0.0371, 0.0114], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0170, 0.0153, 0.0155, 0.0167, 0.0124, 0.0170, 0.0115], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 21:07:43,532 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131040.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:07:56,328 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1862, 4.2842, 4.0600, 3.8100, 3.7838, 4.2135, 3.8632, 3.9068], device='cuda:6'), covar=tensor([0.0512, 0.0463, 0.0275, 0.0235, 0.0704, 0.0404, 0.0746, 0.0549], device='cuda:6'), in_proj_covar=tensor([0.0238, 0.0326, 0.0288, 0.0266, 0.0298, 0.0309, 0.0196, 0.0333], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 21:08:07,050 INFO [train.py:904] (6/8) Epoch 13, batch 9250, loss[loss=0.1749, simple_loss=0.2677, pruned_loss=0.04106, over 16669.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2683, pruned_loss=0.04262, over 3081081.56 frames. ], batch size: 62, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 21:09:12,849 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.311e+02 2.375e+02 2.887e+02 3.471e+02 9.563e+02, threshold=5.774e+02, percent-clipped=3.0 2023-04-29 21:09:24,456 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=131088.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:09:24,517 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131088.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:09:28,855 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1915, 4.2703, 4.4613, 4.2992, 4.3748, 4.8103, 4.4400, 4.1122], device='cuda:6'), covar=tensor([0.1525, 0.2018, 0.2056, 0.1997, 0.2481, 0.1064, 0.1299, 0.2343], device='cuda:6'), in_proj_covar=tensor([0.0347, 0.0490, 0.0537, 0.0418, 0.0563, 0.0564, 0.0429, 0.0561], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 21:09:40,185 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131094.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:09:49,112 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 21:09:55,043 INFO [train.py:904] (6/8) Epoch 13, batch 9300, loss[loss=0.1793, simple_loss=0.2668, pruned_loss=0.04588, over 15482.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2663, pruned_loss=0.04176, over 3063418.78 frames. ], batch size: 191, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 21:11:11,207 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=131136.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:11:24,001 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=131142.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:11:40,549 INFO [train.py:904] (6/8) Epoch 13, batch 9350, loss[loss=0.1874, simple_loss=0.278, pruned_loss=0.04839, over 12296.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2666, pruned_loss=0.04182, over 3067038.67 frames. ], batch size: 248, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:12:25,626 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131174.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:12:41,633 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.118e+02 2.389e+02 2.842e+02 4.349e+02, threshold=4.777e+02, percent-clipped=0.0 2023-04-29 21:13:20,691 INFO [train.py:904] (6/8) Epoch 13, batch 9400, loss[loss=0.1624, simple_loss=0.2502, pruned_loss=0.03725, over 12045.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2653, pruned_loss=0.04161, over 3029378.77 frames. ], batch size: 246, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:13:25,713 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131204.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 21:13:59,476 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131221.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:14:17,583 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-29 21:14:27,970 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131235.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 21:14:38,442 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4033, 4.6909, 4.4880, 4.5222, 4.1382, 4.1350, 4.2260, 4.7082], device='cuda:6'), covar=tensor([0.0893, 0.0854, 0.0978, 0.0646, 0.0743, 0.1333, 0.0926, 0.0850], device='cuda:6'), in_proj_covar=tensor([0.0541, 0.0676, 0.0546, 0.0479, 0.0423, 0.0435, 0.0561, 0.0513], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 21:14:59,709 INFO [train.py:904] (6/8) Epoch 13, batch 9450, loss[loss=0.1587, simple_loss=0.2543, pruned_loss=0.03154, over 16460.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2671, pruned_loss=0.04197, over 3024449.34 frames. ], batch size: 68, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:15:00,865 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=131252.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:15:12,879 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-29 21:15:34,987 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=131269.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:16:03,424 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.757e+02 2.351e+02 2.704e+02 3.479e+02 5.453e+02, threshold=5.408e+02, percent-clipped=2.0 2023-04-29 21:16:40,123 INFO [train.py:904] (6/8) Epoch 13, batch 9500, loss[loss=0.1786, simple_loss=0.268, pruned_loss=0.04461, over 16669.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2673, pruned_loss=0.04196, over 3032557.18 frames. ], batch size: 134, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:17:02,502 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0815, 3.4872, 3.6377, 2.0735, 3.0684, 2.2900, 3.5900, 3.3636], device='cuda:6'), covar=tensor([0.0228, 0.0613, 0.0421, 0.1698, 0.0615, 0.0888, 0.0578, 0.0910], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0138, 0.0153, 0.0140, 0.0132, 0.0122, 0.0131, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 21:18:25,254 INFO [train.py:904] (6/8) Epoch 13, batch 9550, loss[loss=0.1793, simple_loss=0.2664, pruned_loss=0.04605, over 12451.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2679, pruned_loss=0.04249, over 3040673.68 frames. ], batch size: 248, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:18:50,542 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-04-29 21:19:29,579 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.636e+02 2.317e+02 2.711e+02 3.253e+02 6.482e+02, threshold=5.422e+02, percent-clipped=4.0 2023-04-29 21:20:03,674 INFO [train.py:904] (6/8) Epoch 13, batch 9600, loss[loss=0.1846, simple_loss=0.2693, pruned_loss=0.05, over 12357.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.269, pruned_loss=0.04315, over 3033350.63 frames. ], batch size: 250, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:20:17,924 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131409.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:20:30,358 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131416.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:20:40,939 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-29 21:21:49,833 INFO [train.py:904] (6/8) Epoch 13, batch 9650, loss[loss=0.1949, simple_loss=0.2892, pruned_loss=0.0503, over 15340.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2711, pruned_loss=0.04374, over 3038861.15 frames. ], batch size: 191, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:22:34,304 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131470.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:22:48,226 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131477.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:22:58,034 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.434e+02 2.934e+02 3.737e+02 7.668e+02, threshold=5.867e+02, percent-clipped=4.0 2023-04-29 21:23:18,233 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131492.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:23:38,319 INFO [train.py:904] (6/8) Epoch 13, batch 9700, loss[loss=0.156, simple_loss=0.2421, pruned_loss=0.03499, over 12323.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2702, pruned_loss=0.0434, over 3045212.53 frames. ], batch size: 248, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:24:36,251 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131530.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 21:25:20,243 INFO [train.py:904] (6/8) Epoch 13, batch 9750, loss[loss=0.1877, simple_loss=0.2765, pruned_loss=0.04945, over 16892.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.269, pruned_loss=0.04336, over 3039557.76 frames. ], batch size: 109, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:25:22,335 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131553.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:26:22,844 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 2.333e+02 2.960e+02 3.674e+02 6.318e+02, threshold=5.921e+02, percent-clipped=2.0 2023-04-29 21:26:31,003 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0431, 3.0667, 1.8400, 3.3100, 2.2952, 3.2891, 2.0677, 2.5634], device='cuda:6'), covar=tensor([0.0253, 0.0338, 0.1455, 0.0190, 0.0857, 0.0482, 0.1405, 0.0697], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0160, 0.0186, 0.0130, 0.0165, 0.0197, 0.0194, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-29 21:26:57,813 INFO [train.py:904] (6/8) Epoch 13, batch 9800, loss[loss=0.1684, simple_loss=0.2791, pruned_loss=0.02884, over 16800.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2695, pruned_loss=0.04224, over 3060676.81 frames. ], batch size: 83, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:27:30,279 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2737, 4.3366, 4.7427, 4.7098, 4.7069, 4.4267, 4.4263, 4.3131], device='cuda:6'), covar=tensor([0.0319, 0.0494, 0.0367, 0.0389, 0.0411, 0.0373, 0.0795, 0.0389], device='cuda:6'), in_proj_covar=tensor([0.0332, 0.0345, 0.0345, 0.0336, 0.0391, 0.0373, 0.0453, 0.0296], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 21:28:12,796 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-29 21:28:39,800 INFO [train.py:904] (6/8) Epoch 13, batch 9850, loss[loss=0.1667, simple_loss=0.2652, pruned_loss=0.03411, over 15487.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2707, pruned_loss=0.04206, over 3073058.22 frames. ], batch size: 191, lr: 5.15e-03, grad_scale: 4.0 2023-04-29 21:29:46,601 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.743e+02 2.305e+02 2.925e+02 3.439e+02 6.228e+02, threshold=5.850e+02, percent-clipped=2.0 2023-04-29 21:30:32,565 INFO [train.py:904] (6/8) Epoch 13, batch 9900, loss[loss=0.189, simple_loss=0.2913, pruned_loss=0.04339, over 16226.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.271, pruned_loss=0.04176, over 3062419.92 frames. ], batch size: 165, lr: 5.15e-03, grad_scale: 4.0 2023-04-29 21:32:30,668 INFO [train.py:904] (6/8) Epoch 13, batch 9950, loss[loss=0.1625, simple_loss=0.2612, pruned_loss=0.03194, over 16849.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2733, pruned_loss=0.04224, over 3055587.27 frames. ], batch size: 90, lr: 5.15e-03, grad_scale: 4.0 2023-04-29 21:33:02,012 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131765.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:33:04,620 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9569, 4.2307, 4.0563, 4.0796, 3.7642, 3.8543, 3.8443, 4.1977], device='cuda:6'), covar=tensor([0.1132, 0.0928, 0.0924, 0.0680, 0.0812, 0.1508, 0.0994, 0.1058], device='cuda:6'), in_proj_covar=tensor([0.0548, 0.0689, 0.0552, 0.0487, 0.0431, 0.0442, 0.0570, 0.0525], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 21:33:20,260 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131772.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:33:47,200 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.347e+02 2.831e+02 3.741e+02 5.862e+02, threshold=5.662e+02, percent-clipped=1.0 2023-04-29 21:33:56,005 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8811, 3.8547, 4.3046, 4.2523, 4.2531, 3.9784, 3.9913, 4.0014], device='cuda:6'), covar=tensor([0.0320, 0.0723, 0.0411, 0.0492, 0.0448, 0.0462, 0.0819, 0.0407], device='cuda:6'), in_proj_covar=tensor([0.0330, 0.0344, 0.0342, 0.0334, 0.0390, 0.0370, 0.0450, 0.0296], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 21:34:16,894 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6867, 2.6786, 2.2888, 4.2514, 2.8753, 4.0810, 1.3195, 2.9676], device='cuda:6'), covar=tensor([0.1362, 0.0708, 0.1249, 0.0122, 0.0176, 0.0370, 0.1636, 0.0706], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0160, 0.0180, 0.0150, 0.0188, 0.0202, 0.0186, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 21:34:31,163 INFO [train.py:904] (6/8) Epoch 13, batch 10000, loss[loss=0.1724, simple_loss=0.2717, pruned_loss=0.03656, over 16724.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2715, pruned_loss=0.04157, over 3063269.98 frames. ], batch size: 134, lr: 5.15e-03, grad_scale: 8.0 2023-04-29 21:35:27,702 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131830.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:36:06,341 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131848.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:36:11,632 INFO [train.py:904] (6/8) Epoch 13, batch 10050, loss[loss=0.1714, simple_loss=0.2645, pruned_loss=0.03915, over 16623.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2714, pruned_loss=0.0414, over 3068166.85 frames. ], batch size: 62, lr: 5.15e-03, grad_scale: 8.0 2023-04-29 21:36:23,340 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131857.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:37:04,853 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=131878.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:37:14,088 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.561e+02 2.136e+02 2.458e+02 2.977e+02 5.348e+02, threshold=4.916e+02, percent-clipped=0.0 2023-04-29 21:37:45,829 INFO [train.py:904] (6/8) Epoch 13, batch 10100, loss[loss=0.1795, simple_loss=0.2711, pruned_loss=0.04394, over 16382.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2718, pruned_loss=0.04143, over 3083201.12 frames. ], batch size: 146, lr: 5.15e-03, grad_scale: 8.0 2023-04-29 21:38:16,836 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131918.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:39:29,405 INFO [train.py:904] (6/8) Epoch 14, batch 0, loss[loss=0.1624, simple_loss=0.2449, pruned_loss=0.03996, over 17009.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2449, pruned_loss=0.03996, over 17009.00 frames. ], batch size: 41, lr: 4.96e-03, grad_scale: 8.0 2023-04-29 21:39:29,405 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 21:39:36,899 INFO [train.py:938] (6/8) Epoch 14, validation: loss=0.1515, simple_loss=0.2551, pruned_loss=0.024, over 944034.00 frames. 2023-04-29 21:39:36,899 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 21:40:03,497 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6881, 4.1098, 4.2556, 3.0131, 3.6668, 4.2451, 3.8964, 2.4072], device='cuda:6'), covar=tensor([0.0418, 0.0046, 0.0026, 0.0294, 0.0090, 0.0065, 0.0067, 0.0404], device='cuda:6'), in_proj_covar=tensor([0.0129, 0.0069, 0.0070, 0.0127, 0.0082, 0.0089, 0.0080, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 21:40:04,651 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4141, 3.2266, 3.4509, 1.6844, 3.5577, 3.5863, 2.8434, 2.6152], device='cuda:6'), covar=tensor([0.0771, 0.0212, 0.0159, 0.1233, 0.0079, 0.0153, 0.0422, 0.0448], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0097, 0.0083, 0.0132, 0.0065, 0.0103, 0.0116, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 21:40:22,371 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.657e+02 2.510e+02 3.083e+02 4.151e+02 9.991e+02, threshold=6.165e+02, percent-clipped=13.0 2023-04-29 21:40:46,681 INFO [train.py:904] (6/8) Epoch 14, batch 50, loss[loss=0.2363, simple_loss=0.3137, pruned_loss=0.07941, over 11927.00 frames. ], tot_loss[loss=0.1946, simple_loss=0.2783, pruned_loss=0.05545, over 752612.56 frames. ], batch size: 246, lr: 4.96e-03, grad_scale: 1.0 2023-04-29 21:41:27,244 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132031.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:41:48,917 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5968, 3.6959, 3.0617, 5.3539, 4.4308, 4.6339, 2.2294, 3.4835], device='cuda:6'), covar=tensor([0.0964, 0.0545, 0.0924, 0.0173, 0.0268, 0.0395, 0.1236, 0.0648], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0162, 0.0183, 0.0154, 0.0190, 0.0205, 0.0188, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 21:41:55,078 INFO [train.py:904] (6/8) Epoch 14, batch 100, loss[loss=0.2646, simple_loss=0.326, pruned_loss=0.1016, over 12002.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.2742, pruned_loss=0.05395, over 1313861.98 frames. ], batch size: 246, lr: 4.96e-03, grad_scale: 1.0 2023-04-29 21:41:58,548 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132054.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:42:14,076 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132065.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:42:24,502 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132072.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:42:31,569 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9697, 5.4138, 5.1515, 5.0936, 4.8745, 4.8000, 4.8696, 5.4798], device='cuda:6'), covar=tensor([0.1297, 0.1023, 0.1138, 0.0950, 0.0940, 0.1007, 0.1099, 0.1011], device='cuda:6'), in_proj_covar=tensor([0.0568, 0.0709, 0.0570, 0.0500, 0.0446, 0.0453, 0.0589, 0.0539], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 21:42:44,827 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.577e+02 2.362e+02 2.775e+02 3.378e+02 5.633e+02, threshold=5.550e+02, percent-clipped=0.0 2023-04-29 21:42:45,514 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-29 21:42:51,115 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132092.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:43:04,249 INFO [train.py:904] (6/8) Epoch 14, batch 150, loss[loss=0.1877, simple_loss=0.2768, pruned_loss=0.04931, over 16707.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2737, pruned_loss=0.05442, over 1751916.01 frames. ], batch size: 62, lr: 4.96e-03, grad_scale: 1.0 2023-04-29 21:43:21,434 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=132113.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:43:24,028 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132115.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 21:43:30,575 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=132120.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:43:42,397 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-29 21:44:09,335 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132148.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:44:15,117 INFO [train.py:904] (6/8) Epoch 14, batch 200, loss[loss=0.2026, simple_loss=0.2679, pruned_loss=0.06862, over 16740.00 frames. ], tot_loss[loss=0.1887, simple_loss=0.272, pruned_loss=0.0527, over 2108531.49 frames. ], batch size: 83, lr: 4.96e-03, grad_scale: 1.0 2023-04-29 21:45:02,143 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.381e+02 2.781e+02 3.514e+02 5.258e+02, threshold=5.562e+02, percent-clipped=0.0 2023-04-29 21:45:15,515 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=132196.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:45:23,095 INFO [train.py:904] (6/8) Epoch 14, batch 250, loss[loss=0.1472, simple_loss=0.2288, pruned_loss=0.03276, over 17039.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2708, pruned_loss=0.05257, over 2377685.63 frames. ], batch size: 41, lr: 4.96e-03, grad_scale: 1.0 2023-04-29 21:45:38,859 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=132213.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:46:33,333 INFO [train.py:904] (6/8) Epoch 14, batch 300, loss[loss=0.1836, simple_loss=0.2816, pruned_loss=0.04279, over 16680.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.268, pruned_loss=0.0516, over 2582336.20 frames. ], batch size: 57, lr: 4.96e-03, grad_scale: 1.0 2023-04-29 21:46:33,773 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1657, 4.8042, 5.1434, 5.3625, 5.5747, 4.8513, 5.5545, 5.5245], device='cuda:6'), covar=tensor([0.1532, 0.1274, 0.1740, 0.0736, 0.0451, 0.0797, 0.0415, 0.0565], device='cuda:6'), in_proj_covar=tensor([0.0552, 0.0677, 0.0813, 0.0693, 0.0523, 0.0531, 0.0550, 0.0640], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 21:47:22,675 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.346e+02 2.777e+02 3.394e+02 6.651e+02, threshold=5.554e+02, percent-clipped=2.0 2023-04-29 21:47:43,453 INFO [train.py:904] (6/8) Epoch 14, batch 350, loss[loss=0.1905, simple_loss=0.2636, pruned_loss=0.05864, over 16685.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.265, pruned_loss=0.04984, over 2753988.08 frames. ], batch size: 134, lr: 4.95e-03, grad_scale: 1.0 2023-04-29 21:48:30,604 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.45 vs. limit=5.0 2023-04-29 21:48:51,145 INFO [train.py:904] (6/8) Epoch 14, batch 400, loss[loss=0.1619, simple_loss=0.2531, pruned_loss=0.03534, over 17129.00 frames. ], tot_loss[loss=0.1809, simple_loss=0.2636, pruned_loss=0.0491, over 2881898.80 frames. ], batch size: 48, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:49:38,425 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.217e+02 2.653e+02 3.223e+02 5.439e+02, threshold=5.306e+02, percent-clipped=0.0 2023-04-29 21:49:39,469 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=132387.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:49:57,688 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2656, 3.5595, 3.8229, 2.2504, 3.0833, 2.4142, 3.7065, 3.6921], device='cuda:6'), covar=tensor([0.0263, 0.0756, 0.0440, 0.1762, 0.0783, 0.0927, 0.0588, 0.0996], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0143, 0.0156, 0.0143, 0.0135, 0.0123, 0.0134, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 21:50:00,120 INFO [train.py:904] (6/8) Epoch 14, batch 450, loss[loss=0.1623, simple_loss=0.2606, pruned_loss=0.03198, over 17140.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.262, pruned_loss=0.04882, over 2967351.27 frames. ], batch size: 48, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:50:12,064 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=132410.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 21:51:11,604 INFO [train.py:904] (6/8) Epoch 14, batch 500, loss[loss=0.1768, simple_loss=0.2502, pruned_loss=0.05166, over 16678.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2611, pruned_loss=0.04823, over 3037611.03 frames. ], batch size: 89, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:51:15,510 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3905, 4.4164, 4.8162, 4.8208, 4.8511, 4.5091, 4.5075, 4.3308], device='cuda:6'), covar=tensor([0.0373, 0.0637, 0.0465, 0.0422, 0.0480, 0.0427, 0.0804, 0.0581], device='cuda:6'), in_proj_covar=tensor([0.0358, 0.0374, 0.0374, 0.0358, 0.0417, 0.0402, 0.0492, 0.0319], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 21:51:49,444 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5852, 4.5561, 4.4902, 4.0343, 4.5018, 1.7636, 4.2711, 4.1557], device='cuda:6'), covar=tensor([0.0098, 0.0081, 0.0144, 0.0255, 0.0075, 0.2375, 0.0113, 0.0172], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0127, 0.0170, 0.0157, 0.0142, 0.0189, 0.0161, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 21:51:58,581 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.766e+02 2.377e+02 2.811e+02 3.534e+02 1.282e+03, threshold=5.622e+02, percent-clipped=6.0 2023-04-29 21:52:16,264 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8470, 3.7073, 3.8982, 4.0062, 4.0446, 3.6537, 3.9240, 4.0627], device='cuda:6'), covar=tensor([0.1305, 0.0964, 0.1021, 0.0580, 0.0538, 0.1963, 0.1546, 0.0646], device='cuda:6'), in_proj_covar=tensor([0.0575, 0.0703, 0.0847, 0.0722, 0.0547, 0.0552, 0.0571, 0.0665], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 21:52:19,295 INFO [train.py:904] (6/8) Epoch 14, batch 550, loss[loss=0.211, simple_loss=0.2808, pruned_loss=0.0706, over 16841.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.261, pruned_loss=0.04877, over 3100804.84 frames. ], batch size: 109, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:52:32,211 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3815, 4.3238, 4.3412, 3.6158, 4.3100, 1.6965, 4.1042, 3.9089], device='cuda:6'), covar=tensor([0.0131, 0.0119, 0.0170, 0.0410, 0.0106, 0.2808, 0.0144, 0.0254], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0128, 0.0172, 0.0159, 0.0144, 0.0191, 0.0162, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 21:52:34,332 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132513.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:53:01,165 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8788, 4.9168, 5.3933, 5.3753, 5.3791, 5.0428, 4.9792, 4.7095], device='cuda:6'), covar=tensor([0.0359, 0.0491, 0.0465, 0.0446, 0.0429, 0.0348, 0.0860, 0.0443], device='cuda:6'), in_proj_covar=tensor([0.0360, 0.0375, 0.0375, 0.0359, 0.0418, 0.0403, 0.0493, 0.0320], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 21:53:28,147 INFO [train.py:904] (6/8) Epoch 14, batch 600, loss[loss=0.1564, simple_loss=0.2509, pruned_loss=0.0309, over 17137.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.261, pruned_loss=0.04841, over 3161302.50 frames. ], batch size: 48, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:53:39,910 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.05 vs. limit=5.0 2023-04-29 21:53:41,854 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=132561.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:54:00,100 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-29 21:54:17,820 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 2.200e+02 2.653e+02 3.173e+02 6.384e+02, threshold=5.306e+02, percent-clipped=1.0 2023-04-29 21:54:39,140 INFO [train.py:904] (6/8) Epoch 14, batch 650, loss[loss=0.1796, simple_loss=0.252, pruned_loss=0.05355, over 16308.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.26, pruned_loss=0.04775, over 3188398.15 frames. ], batch size: 165, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:55:49,508 INFO [train.py:904] (6/8) Epoch 14, batch 700, loss[loss=0.166, simple_loss=0.2469, pruned_loss=0.04251, over 16824.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2592, pruned_loss=0.04705, over 3219200.76 frames. ], batch size: 102, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:55:53,088 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1299, 4.0755, 4.5123, 2.0910, 4.6872, 4.7280, 3.3841, 3.6237], device='cuda:6'), covar=tensor([0.0649, 0.0198, 0.0183, 0.1198, 0.0048, 0.0126, 0.0379, 0.0373], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0103, 0.0088, 0.0138, 0.0069, 0.0112, 0.0122, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 21:56:01,342 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132660.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:56:38,037 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.642e+02 2.258e+02 2.690e+02 3.168e+02 6.172e+02, threshold=5.381e+02, percent-clipped=1.0 2023-04-29 21:56:39,032 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132687.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:56:58,534 INFO [train.py:904] (6/8) Epoch 14, batch 750, loss[loss=0.1895, simple_loss=0.2651, pruned_loss=0.057, over 16938.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2597, pruned_loss=0.04773, over 3231516.53 frames. ], batch size: 96, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:57:09,169 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132710.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 21:57:17,141 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4181, 2.3262, 1.8426, 2.1230, 2.6934, 2.4836, 2.6877, 2.7301], device='cuda:6'), covar=tensor([0.0181, 0.0311, 0.0418, 0.0378, 0.0184, 0.0274, 0.0203, 0.0251], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0214, 0.0208, 0.0207, 0.0213, 0.0214, 0.0218, 0.0205], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 21:57:24,394 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132721.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:57:43,962 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=132735.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:57:44,168 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1681, 2.1017, 2.7425, 3.0827, 2.9348, 3.3729, 2.3264, 3.4348], device='cuda:6'), covar=tensor([0.0167, 0.0380, 0.0212, 0.0240, 0.0206, 0.0137, 0.0364, 0.0120], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0177, 0.0161, 0.0166, 0.0174, 0.0131, 0.0179, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 21:57:59,788 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132746.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:58:07,420 INFO [train.py:904] (6/8) Epoch 14, batch 800, loss[loss=0.1435, simple_loss=0.2352, pruned_loss=0.02586, over 17196.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2592, pruned_loss=0.04734, over 3245503.97 frames. ], batch size: 44, lr: 4.95e-03, grad_scale: 4.0 2023-04-29 21:58:16,064 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=132758.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:58:52,003 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7841, 3.0912, 2.8677, 4.9802, 4.0209, 4.4314, 1.7205, 3.3065], device='cuda:6'), covar=tensor([0.1353, 0.0657, 0.1068, 0.0160, 0.0291, 0.0353, 0.1505, 0.0710], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0161, 0.0180, 0.0157, 0.0193, 0.0205, 0.0185, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-29 21:58:54,908 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.382e+02 2.716e+02 3.142e+02 6.467e+02, threshold=5.432e+02, percent-clipped=2.0 2023-04-29 21:59:14,437 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2023-04-29 21:59:15,846 INFO [train.py:904] (6/8) Epoch 14, batch 850, loss[loss=0.1691, simple_loss=0.2479, pruned_loss=0.04519, over 16822.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2585, pruned_loss=0.04685, over 3265947.16 frames. ], batch size: 96, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 21:59:23,547 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132807.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:59:29,484 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1443, 5.6383, 5.7789, 5.4593, 5.5703, 6.1528, 5.6859, 5.3965], device='cuda:6'), covar=tensor([0.0859, 0.1972, 0.2405, 0.2032, 0.2999, 0.0984, 0.1392, 0.2451], device='cuda:6'), in_proj_covar=tensor([0.0370, 0.0525, 0.0578, 0.0445, 0.0606, 0.0602, 0.0457, 0.0599], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 22:00:24,478 INFO [train.py:904] (6/8) Epoch 14, batch 900, loss[loss=0.2078, simple_loss=0.2753, pruned_loss=0.0702, over 16675.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2574, pruned_loss=0.04658, over 3271953.17 frames. ], batch size: 134, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 22:01:02,118 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6050, 4.7122, 4.8354, 4.6868, 4.6821, 5.2953, 4.8406, 4.4533], device='cuda:6'), covar=tensor([0.1377, 0.1883, 0.2207, 0.2020, 0.2705, 0.0998, 0.1451, 0.2787], device='cuda:6'), in_proj_covar=tensor([0.0368, 0.0522, 0.0573, 0.0442, 0.0604, 0.0597, 0.0455, 0.0596], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 22:01:14,368 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.740e+02 2.299e+02 2.623e+02 3.116e+02 6.405e+02, threshold=5.245e+02, percent-clipped=2.0 2023-04-29 22:01:34,126 INFO [train.py:904] (6/8) Epoch 14, batch 950, loss[loss=0.144, simple_loss=0.2288, pruned_loss=0.02957, over 16855.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2576, pruned_loss=0.0467, over 3281032.98 frames. ], batch size: 42, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 22:01:41,141 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2023-04-29 22:02:41,756 INFO [train.py:904] (6/8) Epoch 14, batch 1000, loss[loss=0.1432, simple_loss=0.2299, pruned_loss=0.02825, over 16942.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2561, pruned_loss=0.04619, over 3290294.69 frames. ], batch size: 41, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 22:02:59,136 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-29 22:03:25,053 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8334, 3.9004, 2.4346, 4.4841, 2.9889, 4.5039, 2.5010, 3.2561], device='cuda:6'), covar=tensor([0.0252, 0.0365, 0.1456, 0.0256, 0.0813, 0.0405, 0.1432, 0.0670], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0168, 0.0190, 0.0144, 0.0169, 0.0209, 0.0199, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:03:29,371 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.118e+02 2.524e+02 3.195e+02 6.689e+02, threshold=5.047e+02, percent-clipped=1.0 2023-04-29 22:03:49,998 INFO [train.py:904] (6/8) Epoch 14, batch 1050, loss[loss=0.1721, simple_loss=0.2652, pruned_loss=0.03951, over 16664.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2565, pruned_loss=0.04597, over 3286685.35 frames. ], batch size: 57, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 22:03:52,448 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2224, 3.4133, 3.4200, 2.0883, 3.0001, 2.4678, 3.6807, 3.6705], device='cuda:6'), covar=tensor([0.0202, 0.0738, 0.0566, 0.1720, 0.0748, 0.0888, 0.0476, 0.0799], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0146, 0.0157, 0.0144, 0.0137, 0.0124, 0.0135, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:04:10,798 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133016.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:04:10,980 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5553, 3.6030, 2.1646, 3.8800, 2.7503, 3.8430, 2.3063, 2.9288], device='cuda:6'), covar=tensor([0.0217, 0.0399, 0.1395, 0.0233, 0.0690, 0.0627, 0.1236, 0.0584], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0168, 0.0190, 0.0143, 0.0169, 0.0209, 0.0198, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:04:30,706 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1495, 4.4215, 4.7028, 4.7358, 4.7275, 4.4210, 4.0914, 4.2083], device='cuda:6'), covar=tensor([0.0619, 0.0857, 0.0634, 0.0668, 0.0760, 0.0651, 0.1461, 0.0810], device='cuda:6'), in_proj_covar=tensor([0.0367, 0.0386, 0.0387, 0.0366, 0.0428, 0.0412, 0.0503, 0.0328], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 22:04:59,790 INFO [train.py:904] (6/8) Epoch 14, batch 1100, loss[loss=0.183, simple_loss=0.256, pruned_loss=0.05504, over 16868.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2549, pruned_loss=0.04553, over 3287109.09 frames. ], batch size: 116, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 22:05:34,766 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4926, 2.5516, 2.1419, 2.3730, 2.8744, 2.7096, 3.2236, 3.1653], device='cuda:6'), covar=tensor([0.0126, 0.0372, 0.0432, 0.0360, 0.0257, 0.0319, 0.0221, 0.0212], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0217, 0.0210, 0.0209, 0.0216, 0.0218, 0.0222, 0.0209], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:05:47,408 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.639e+02 2.252e+02 2.587e+02 3.047e+02 6.829e+02, threshold=5.174e+02, percent-clipped=2.0 2023-04-29 22:06:08,349 INFO [train.py:904] (6/8) Epoch 14, batch 1150, loss[loss=0.2031, simple_loss=0.2956, pruned_loss=0.05524, over 17038.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2548, pruned_loss=0.04481, over 3306428.61 frames. ], batch size: 55, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 22:06:08,688 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133102.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:06:47,404 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1239, 3.3359, 2.7431, 5.1819, 4.3680, 4.4127, 1.9293, 3.2137], device='cuda:6'), covar=tensor([0.1081, 0.0544, 0.1092, 0.0126, 0.0220, 0.0448, 0.1207, 0.0730], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0163, 0.0182, 0.0160, 0.0196, 0.0208, 0.0186, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:07:16,496 INFO [train.py:904] (6/8) Epoch 14, batch 1200, loss[loss=0.2085, simple_loss=0.2799, pruned_loss=0.06853, over 16331.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2551, pruned_loss=0.04439, over 3309266.38 frames. ], batch size: 165, lr: 4.94e-03, grad_scale: 8.0 2023-04-29 22:08:05,978 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.506e+02 2.966e+02 3.654e+02 5.614e+02, threshold=5.932e+02, percent-clipped=4.0 2023-04-29 22:08:26,235 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-29 22:08:28,179 INFO [train.py:904] (6/8) Epoch 14, batch 1250, loss[loss=0.1552, simple_loss=0.2438, pruned_loss=0.03334, over 17252.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2559, pruned_loss=0.04535, over 3306675.90 frames. ], batch size: 45, lr: 4.94e-03, grad_scale: 8.0 2023-04-29 22:08:32,403 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 22:08:33,099 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5107, 5.8988, 5.6802, 5.7252, 5.3389, 5.2984, 5.3239, 5.9959], device='cuda:6'), covar=tensor([0.1209, 0.0950, 0.1022, 0.0741, 0.0794, 0.0623, 0.1081, 0.1027], device='cuda:6'), in_proj_covar=tensor([0.0599, 0.0750, 0.0609, 0.0534, 0.0475, 0.0479, 0.0627, 0.0573], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:09:37,985 INFO [train.py:904] (6/8) Epoch 14, batch 1300, loss[loss=0.1801, simple_loss=0.2673, pruned_loss=0.04644, over 17212.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2553, pruned_loss=0.04445, over 3316271.98 frames. ], batch size: 45, lr: 4.94e-03, grad_scale: 8.0 2023-04-29 22:10:27,127 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.451e+02 2.160e+02 2.631e+02 3.090e+02 7.401e+02, threshold=5.263e+02, percent-clipped=1.0 2023-04-29 22:10:30,485 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0676, 4.1813, 2.5291, 4.7822, 3.2341, 4.7539, 2.8554, 3.5240], device='cuda:6'), covar=tensor([0.0240, 0.0326, 0.1468, 0.0197, 0.0730, 0.0451, 0.1384, 0.0613], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0171, 0.0192, 0.0147, 0.0171, 0.0213, 0.0201, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:10:38,924 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-29 22:10:48,345 INFO [train.py:904] (6/8) Epoch 14, batch 1350, loss[loss=0.1757, simple_loss=0.2514, pruned_loss=0.04999, over 16676.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2559, pruned_loss=0.04452, over 3320866.57 frames. ], batch size: 134, lr: 4.94e-03, grad_scale: 8.0 2023-04-29 22:11:06,820 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=133316.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:11:22,515 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9808, 2.6904, 2.7644, 2.1408, 2.6406, 2.1991, 2.7627, 2.8472], device='cuda:6'), covar=tensor([0.0302, 0.0743, 0.0546, 0.1700, 0.0776, 0.0847, 0.0595, 0.0780], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0148, 0.0158, 0.0145, 0.0137, 0.0125, 0.0137, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:11:56,451 INFO [train.py:904] (6/8) Epoch 14, batch 1400, loss[loss=0.1746, simple_loss=0.2459, pruned_loss=0.05172, over 16887.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2558, pruned_loss=0.04438, over 3322290.19 frames. ], batch size: 116, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:12:13,482 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=133364.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:12:26,521 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133374.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 22:12:30,140 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133376.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:12:44,805 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.235e+02 2.580e+02 3.131e+02 5.456e+02, threshold=5.161e+02, percent-clipped=1.0 2023-04-29 22:13:06,248 INFO [train.py:904] (6/8) Epoch 14, batch 1450, loss[loss=0.1707, simple_loss=0.2428, pruned_loss=0.04927, over 11737.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2552, pruned_loss=0.04457, over 3320844.78 frames. ], batch size: 246, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:13:06,599 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=133402.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:13:52,498 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133435.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 22:13:55,258 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133437.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:14:12,935 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=133450.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:14:14,808 INFO [train.py:904] (6/8) Epoch 14, batch 1500, loss[loss=0.176, simple_loss=0.2478, pruned_loss=0.05217, over 12110.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2551, pruned_loss=0.04489, over 3314626.92 frames. ], batch size: 246, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:14:46,583 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133474.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:15:03,756 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.705e+02 2.267e+02 2.734e+02 3.162e+02 6.055e+02, threshold=5.467e+02, percent-clipped=3.0 2023-04-29 22:15:23,303 INFO [train.py:904] (6/8) Epoch 14, batch 1550, loss[loss=0.1948, simple_loss=0.2787, pruned_loss=0.05546, over 15476.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2563, pruned_loss=0.04603, over 3312339.76 frames. ], batch size: 191, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:16:10,853 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133535.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:16:33,557 INFO [train.py:904] (6/8) Epoch 14, batch 1600, loss[loss=0.2004, simple_loss=0.2664, pruned_loss=0.06717, over 16903.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2588, pruned_loss=0.0477, over 3311697.51 frames. ], batch size: 116, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:17:00,950 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-29 22:17:02,693 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 22:17:21,877 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.622e+02 2.312e+02 2.830e+02 3.166e+02 6.146e+02, threshold=5.659e+02, percent-clipped=1.0 2023-04-29 22:17:43,027 INFO [train.py:904] (6/8) Epoch 14, batch 1650, loss[loss=0.1687, simple_loss=0.2686, pruned_loss=0.03443, over 17113.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2603, pruned_loss=0.04806, over 3309108.01 frames. ], batch size: 48, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:18:52,510 INFO [train.py:904] (6/8) Epoch 14, batch 1700, loss[loss=0.1633, simple_loss=0.2547, pruned_loss=0.03594, over 17129.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2616, pruned_loss=0.04824, over 3319999.39 frames. ], batch size: 48, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:18:53,038 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133652.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:19:42,412 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.666e+02 2.547e+02 3.320e+02 4.199e+02 1.315e+03, threshold=6.641e+02, percent-clipped=10.0 2023-04-29 22:19:47,979 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0943, 5.0379, 5.5437, 5.5252, 5.5564, 5.1477, 5.1703, 4.8521], device='cuda:6'), covar=tensor([0.0279, 0.0453, 0.0397, 0.0389, 0.0437, 0.0345, 0.0888, 0.0421], device='cuda:6'), in_proj_covar=tensor([0.0373, 0.0390, 0.0391, 0.0373, 0.0431, 0.0417, 0.0510, 0.0331], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 22:19:59,647 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-29 22:20:03,031 INFO [train.py:904] (6/8) Epoch 14, batch 1750, loss[loss=0.1836, simple_loss=0.2613, pruned_loss=0.05292, over 16742.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2629, pruned_loss=0.04854, over 3318840.16 frames. ], batch size: 83, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:20:06,996 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.30 vs. limit=5.0 2023-04-29 22:20:10,954 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8247, 3.6581, 4.1290, 2.1370, 4.3043, 4.2697, 3.1390, 3.2632], device='cuda:6'), covar=tensor([0.0694, 0.0199, 0.0177, 0.1032, 0.0070, 0.0195, 0.0377, 0.0396], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0102, 0.0088, 0.0135, 0.0070, 0.0112, 0.0120, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 22:20:19,396 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133713.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:20:42,464 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133730.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 22:20:45,408 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133732.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:20:47,332 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-29 22:21:12,608 INFO [train.py:904] (6/8) Epoch 14, batch 1800, loss[loss=0.1869, simple_loss=0.2727, pruned_loss=0.05055, over 16613.00 frames. ], tot_loss[loss=0.1809, simple_loss=0.2641, pruned_loss=0.04888, over 3305037.58 frames. ], batch size: 89, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:22:00,849 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.275e+02 2.807e+02 3.369e+02 7.697e+02, threshold=5.615e+02, percent-clipped=1.0 2023-04-29 22:22:22,387 INFO [train.py:904] (6/8) Epoch 14, batch 1850, loss[loss=0.2176, simple_loss=0.3013, pruned_loss=0.06696, over 12043.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.265, pruned_loss=0.04903, over 3302332.32 frames. ], batch size: 249, lr: 4.93e-03, grad_scale: 4.0 2023-04-29 22:23:02,983 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133830.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:23:32,964 INFO [train.py:904] (6/8) Epoch 14, batch 1900, loss[loss=0.1986, simple_loss=0.267, pruned_loss=0.06506, over 16862.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2647, pruned_loss=0.0479, over 3305695.98 frames. ], batch size: 109, lr: 4.93e-03, grad_scale: 4.0 2023-04-29 22:24:23,478 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 2.126e+02 2.452e+02 2.920e+02 9.757e+02, threshold=4.904e+02, percent-clipped=2.0 2023-04-29 22:24:42,160 INFO [train.py:904] (6/8) Epoch 14, batch 1950, loss[loss=0.1898, simple_loss=0.2833, pruned_loss=0.04813, over 12277.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2649, pruned_loss=0.04737, over 3306509.81 frames. ], batch size: 247, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:25:52,906 INFO [train.py:904] (6/8) Epoch 14, batch 2000, loss[loss=0.2001, simple_loss=0.2902, pruned_loss=0.05502, over 16675.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2643, pruned_loss=0.04738, over 3305643.55 frames. ], batch size: 62, lr: 4.92e-03, grad_scale: 8.0 2023-04-29 22:25:54,942 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 22:26:08,649 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.51 vs. limit=2.0 2023-04-29 22:26:09,885 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-04-29 22:26:22,651 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-04-29 22:26:43,618 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.687e+02 2.429e+02 2.777e+02 3.533e+02 6.577e+02, threshold=5.554e+02, percent-clipped=5.0 2023-04-29 22:26:57,116 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133997.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:27:06,903 INFO [train.py:904] (6/8) Epoch 14, batch 2050, loss[loss=0.1708, simple_loss=0.2637, pruned_loss=0.03896, over 17111.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2642, pruned_loss=0.04765, over 3299705.03 frames. ], batch size: 49, lr: 4.92e-03, grad_scale: 8.0 2023-04-29 22:27:15,546 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134008.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:27:46,369 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134030.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 22:27:49,259 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134032.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:28:16,827 INFO [train.py:904] (6/8) Epoch 14, batch 2100, loss[loss=0.1886, simple_loss=0.2755, pruned_loss=0.05087, over 17105.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2649, pruned_loss=0.04797, over 3309904.20 frames. ], batch size: 47, lr: 4.92e-03, grad_scale: 8.0 2023-04-29 22:28:25,775 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134058.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:28:54,067 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=134078.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 22:28:56,310 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=134080.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:29:09,011 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.284e+02 2.721e+02 3.276e+02 7.722e+02, threshold=5.442e+02, percent-clipped=1.0 2023-04-29 22:29:26,680 INFO [train.py:904] (6/8) Epoch 14, batch 2150, loss[loss=0.2098, simple_loss=0.2868, pruned_loss=0.0664, over 15557.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2664, pruned_loss=0.04908, over 3305072.27 frames. ], batch size: 191, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:29:32,014 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6404, 2.2228, 2.2898, 4.5352, 2.1555, 2.6684, 2.3006, 2.4651], device='cuda:6'), covar=tensor([0.1075, 0.3685, 0.2649, 0.0400, 0.4102, 0.2543, 0.3365, 0.3458], device='cuda:6'), in_proj_covar=tensor([0.0379, 0.0411, 0.0344, 0.0328, 0.0420, 0.0474, 0.0377, 0.0482], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:30:06,183 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134130.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:30:17,610 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7808, 4.0717, 3.0711, 2.3234, 2.7478, 2.4889, 4.2633, 3.6012], device='cuda:6'), covar=tensor([0.2670, 0.0618, 0.1572, 0.2599, 0.2602, 0.1806, 0.0453, 0.1269], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0261, 0.0288, 0.0288, 0.0281, 0.0231, 0.0275, 0.0311], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 22:30:35,960 INFO [train.py:904] (6/8) Epoch 14, batch 2200, loss[loss=0.1531, simple_loss=0.2468, pruned_loss=0.02967, over 17172.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.2663, pruned_loss=0.04941, over 3294090.03 frames. ], batch size: 46, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:30:40,824 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 22:30:55,625 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8966, 5.1197, 5.2836, 5.0975, 5.1395, 5.7648, 5.2929, 5.0548], device='cuda:6'), covar=tensor([0.1132, 0.2000, 0.2158, 0.1937, 0.2827, 0.1057, 0.1389, 0.2353], device='cuda:6'), in_proj_covar=tensor([0.0380, 0.0539, 0.0590, 0.0455, 0.0618, 0.0612, 0.0466, 0.0610], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 22:31:12,754 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=134178.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:31:25,395 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8305, 3.1418, 2.7232, 4.9650, 3.8662, 4.3941, 1.8756, 3.0826], device='cuda:6'), covar=tensor([0.1469, 0.0771, 0.1231, 0.0178, 0.0287, 0.0433, 0.1586, 0.0836], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0164, 0.0182, 0.0161, 0.0199, 0.0210, 0.0186, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:31:27,737 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.617e+02 2.496e+02 2.998e+02 3.803e+02 8.330e+02, threshold=5.996e+02, percent-clipped=6.0 2023-04-29 22:31:45,982 INFO [train.py:904] (6/8) Epoch 14, batch 2250, loss[loss=0.187, simple_loss=0.2774, pruned_loss=0.04823, over 16503.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2668, pruned_loss=0.04959, over 3304490.15 frames. ], batch size: 68, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:32:34,269 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0681, 5.0125, 4.9367, 4.5665, 4.5886, 4.9822, 4.9085, 4.6183], device='cuda:6'), covar=tensor([0.0587, 0.0540, 0.0250, 0.0277, 0.0903, 0.0411, 0.0418, 0.0625], device='cuda:6'), in_proj_covar=tensor([0.0274, 0.0375, 0.0332, 0.0311, 0.0347, 0.0360, 0.0225, 0.0386], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 22:32:56,451 INFO [train.py:904] (6/8) Epoch 14, batch 2300, loss[loss=0.1812, simple_loss=0.2628, pruned_loss=0.04982, over 16522.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2668, pruned_loss=0.04882, over 3312521.05 frames. ], batch size: 75, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:33:48,101 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.438e+02 2.937e+02 3.292e+02 7.337e+02, threshold=5.875e+02, percent-clipped=1.0 2023-04-29 22:34:06,356 INFO [train.py:904] (6/8) Epoch 14, batch 2350, loss[loss=0.1996, simple_loss=0.2673, pruned_loss=0.06596, over 16951.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2664, pruned_loss=0.04902, over 3319079.00 frames. ], batch size: 116, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:34:09,703 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=134304.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:34:15,134 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134308.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:34:16,991 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=134309.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:34:47,529 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8622, 2.6078, 2.0041, 2.3809, 2.9535, 2.7686, 3.1394, 3.0630], device='cuda:6'), covar=tensor([0.0169, 0.0272, 0.0416, 0.0354, 0.0184, 0.0239, 0.0172, 0.0206], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0223, 0.0216, 0.0214, 0.0222, 0.0223, 0.0230, 0.0216], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:35:17,012 INFO [train.py:904] (6/8) Epoch 14, batch 2400, loss[loss=0.2217, simple_loss=0.298, pruned_loss=0.07271, over 11950.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2658, pruned_loss=0.04826, over 3321869.66 frames. ], batch size: 246, lr: 4.92e-03, grad_scale: 8.0 2023-04-29 22:35:19,207 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134353.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:35:23,357 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=134356.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:35:35,628 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134365.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:35:42,550 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134370.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:36:08,842 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.471e+02 2.303e+02 2.657e+02 3.180e+02 6.909e+02, threshold=5.314e+02, percent-clipped=1.0 2023-04-29 22:36:25,718 INFO [train.py:904] (6/8) Epoch 14, batch 2450, loss[loss=0.16, simple_loss=0.2464, pruned_loss=0.03684, over 17172.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.2664, pruned_loss=0.04828, over 3322044.33 frames. ], batch size: 46, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:37:33,358 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5167, 3.8486, 4.1234, 2.2495, 3.2171, 2.6222, 4.0391, 3.9335], device='cuda:6'), covar=tensor([0.0267, 0.0739, 0.0417, 0.1714, 0.0723, 0.0894, 0.0551, 0.0988], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0153, 0.0162, 0.0148, 0.0140, 0.0127, 0.0139, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:37:35,054 INFO [train.py:904] (6/8) Epoch 14, batch 2500, loss[loss=0.2096, simple_loss=0.2892, pruned_loss=0.06505, over 16483.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2664, pruned_loss=0.048, over 3321466.00 frames. ], batch size: 146, lr: 4.91e-03, grad_scale: 4.0 2023-04-29 22:38:28,208 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.356e+02 2.745e+02 3.374e+02 4.761e+02, threshold=5.489e+02, percent-clipped=0.0 2023-04-29 22:38:45,442 INFO [train.py:904] (6/8) Epoch 14, batch 2550, loss[loss=0.1816, simple_loss=0.2769, pruned_loss=0.04317, over 16738.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2668, pruned_loss=0.04819, over 3315757.32 frames. ], batch size: 57, lr: 4.91e-03, grad_scale: 4.0 2023-04-29 22:39:12,347 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8586, 2.8270, 2.4975, 4.2521, 3.5728, 4.1570, 1.4967, 2.9562], device='cuda:6'), covar=tensor([0.1276, 0.0623, 0.1028, 0.0163, 0.0161, 0.0352, 0.1421, 0.0718], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0163, 0.0182, 0.0163, 0.0199, 0.0212, 0.0186, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:39:46,304 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8799, 4.2789, 3.0970, 2.1419, 2.7558, 2.5152, 4.5625, 3.6402], device='cuda:6'), covar=tensor([0.2476, 0.0593, 0.1586, 0.2700, 0.2819, 0.1762, 0.0365, 0.1160], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0262, 0.0290, 0.0289, 0.0284, 0.0231, 0.0276, 0.0313], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-29 22:39:54,961 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-29 22:39:55,149 INFO [train.py:904] (6/8) Epoch 14, batch 2600, loss[loss=0.2131, simple_loss=0.2799, pruned_loss=0.07312, over 16862.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2662, pruned_loss=0.04761, over 3324699.60 frames. ], batch size: 109, lr: 4.91e-03, grad_scale: 4.0 2023-04-29 22:40:05,105 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.49 vs. limit=5.0 2023-04-29 22:40:46,904 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.364e+02 2.744e+02 3.247e+02 7.768e+02, threshold=5.488e+02, percent-clipped=4.0 2023-04-29 22:41:03,694 INFO [train.py:904] (6/8) Epoch 14, batch 2650, loss[loss=0.1848, simple_loss=0.2742, pruned_loss=0.04771, over 16780.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2668, pruned_loss=0.04766, over 3329209.38 frames. ], batch size: 83, lr: 4.91e-03, grad_scale: 4.0 2023-04-29 22:41:22,826 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0623, 3.0725, 1.7799, 3.2152, 2.3622, 3.2468, 2.0150, 2.5779], device='cuda:6'), covar=tensor([0.0269, 0.0384, 0.1639, 0.0350, 0.0792, 0.0628, 0.1371, 0.0683], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0173, 0.0192, 0.0149, 0.0171, 0.0215, 0.0201, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:42:06,656 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8461, 3.9709, 2.3111, 4.6779, 3.1588, 4.6110, 2.4891, 3.2586], device='cuda:6'), covar=tensor([0.0252, 0.0334, 0.1615, 0.0216, 0.0690, 0.0432, 0.1381, 0.0666], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0173, 0.0192, 0.0149, 0.0171, 0.0215, 0.0201, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:42:06,891 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-04-29 22:42:12,177 INFO [train.py:904] (6/8) Epoch 14, batch 2700, loss[loss=0.1935, simple_loss=0.2756, pruned_loss=0.05573, over 16787.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2669, pruned_loss=0.04687, over 3331351.92 frames. ], batch size: 102, lr: 4.91e-03, grad_scale: 4.0 2023-04-29 22:42:13,622 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134653.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:42:23,650 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134660.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:42:31,296 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134665.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:43:04,242 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.644e+02 2.219e+02 2.559e+02 3.067e+02 4.567e+02, threshold=5.119e+02, percent-clipped=0.0 2023-04-29 22:43:20,433 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=134701.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:43:21,288 INFO [train.py:904] (6/8) Epoch 14, batch 2750, loss[loss=0.1997, simple_loss=0.2928, pruned_loss=0.05333, over 16639.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2671, pruned_loss=0.0464, over 3332460.74 frames. ], batch size: 62, lr: 4.91e-03, grad_scale: 4.0 2023-04-29 22:44:29,009 INFO [train.py:904] (6/8) Epoch 14, batch 2800, loss[loss=0.1919, simple_loss=0.2852, pruned_loss=0.04928, over 17047.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.2678, pruned_loss=0.04684, over 3322117.44 frames. ], batch size: 55, lr: 4.91e-03, grad_scale: 8.0 2023-04-29 22:45:20,154 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.581e+02 2.461e+02 2.848e+02 3.331e+02 6.185e+02, threshold=5.697e+02, percent-clipped=2.0 2023-04-29 22:45:37,175 INFO [train.py:904] (6/8) Epoch 14, batch 2850, loss[loss=0.1937, simple_loss=0.2692, pruned_loss=0.05914, over 16294.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2663, pruned_loss=0.04632, over 3327685.19 frames. ], batch size: 165, lr: 4.91e-03, grad_scale: 8.0 2023-04-29 22:45:52,357 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0129, 2.5937, 2.6874, 4.7925, 2.5369, 3.0628, 2.6681, 2.8407], device='cuda:6'), covar=tensor([0.0917, 0.2998, 0.2313, 0.0368, 0.3500, 0.2063, 0.2833, 0.2818], device='cuda:6'), in_proj_covar=tensor([0.0378, 0.0411, 0.0344, 0.0327, 0.0418, 0.0475, 0.0377, 0.0483], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:46:18,994 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-29 22:46:28,871 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8771, 3.9912, 2.4144, 4.5984, 2.9642, 4.4934, 2.5830, 3.2351], device='cuda:6'), covar=tensor([0.0242, 0.0344, 0.1449, 0.0207, 0.0769, 0.0487, 0.1351, 0.0647], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0172, 0.0191, 0.0149, 0.0170, 0.0215, 0.0200, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:46:38,413 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5413, 4.5304, 4.5051, 4.0056, 4.4884, 1.7799, 4.2265, 4.2247], device='cuda:6'), covar=tensor([0.0105, 0.0086, 0.0157, 0.0328, 0.0099, 0.2333, 0.0137, 0.0157], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0136, 0.0183, 0.0171, 0.0154, 0.0195, 0.0174, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:46:45,299 INFO [train.py:904] (6/8) Epoch 14, batch 2900, loss[loss=0.2156, simple_loss=0.2774, pruned_loss=0.07692, over 15511.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2663, pruned_loss=0.04723, over 3324433.43 frames. ], batch size: 190, lr: 4.91e-03, grad_scale: 8.0 2023-04-29 22:46:57,828 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9171, 2.4147, 2.6791, 4.7972, 2.4368, 2.8819, 2.5279, 2.6942], device='cuda:6'), covar=tensor([0.0954, 0.3405, 0.2365, 0.0336, 0.3705, 0.2322, 0.3149, 0.3345], device='cuda:6'), in_proj_covar=tensor([0.0380, 0.0414, 0.0346, 0.0328, 0.0420, 0.0477, 0.0378, 0.0485], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:46:58,850 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=134862.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:47:21,035 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9403, 3.8009, 4.0307, 1.9926, 4.1636, 4.2243, 3.1991, 3.2236], device='cuda:6'), covar=tensor([0.0638, 0.0164, 0.0160, 0.1222, 0.0075, 0.0172, 0.0378, 0.0421], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0102, 0.0089, 0.0137, 0.0071, 0.0114, 0.0121, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 22:47:36,338 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.583e+02 2.493e+02 2.912e+02 3.442e+02 5.810e+02, threshold=5.824e+02, percent-clipped=2.0 2023-04-29 22:47:54,145 INFO [train.py:904] (6/8) Epoch 14, batch 2950, loss[loss=0.1923, simple_loss=0.2682, pruned_loss=0.05821, over 16224.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2655, pruned_loss=0.04764, over 3323098.74 frames. ], batch size: 165, lr: 4.91e-03, grad_scale: 8.0 2023-04-29 22:48:24,087 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134923.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:49:02,685 INFO [train.py:904] (6/8) Epoch 14, batch 3000, loss[loss=0.1699, simple_loss=0.2644, pruned_loss=0.03772, over 17118.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2658, pruned_loss=0.04775, over 3330552.81 frames. ], batch size: 49, lr: 4.91e-03, grad_scale: 8.0 2023-04-29 22:49:02,685 INFO [train.py:929] (6/8) Computing validation loss 2023-04-29 22:49:12,422 INFO [train.py:938] (6/8) Epoch 14, validation: loss=0.1382, simple_loss=0.244, pruned_loss=0.01621, over 944034.00 frames. 2023-04-29 22:49:12,422 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-29 22:49:24,949 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134960.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:49:31,468 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134965.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:50:02,076 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-29 22:50:06,980 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.405e+02 2.394e+02 2.911e+02 3.370e+02 7.406e+02, threshold=5.821e+02, percent-clipped=1.0 2023-04-29 22:50:07,430 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5203, 2.3413, 1.6012, 1.9716, 2.7259, 2.5287, 2.8659, 2.7884], device='cuda:6'), covar=tensor([0.0172, 0.0390, 0.0620, 0.0481, 0.0237, 0.0349, 0.0247, 0.0278], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0223, 0.0216, 0.0215, 0.0223, 0.0223, 0.0233, 0.0217], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:50:10,202 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6470, 2.5818, 2.4129, 3.8067, 3.0349, 3.8805, 1.4875, 2.7074], device='cuda:6'), covar=tensor([0.1433, 0.0700, 0.1120, 0.0198, 0.0192, 0.0397, 0.1505, 0.0869], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0163, 0.0182, 0.0163, 0.0199, 0.0211, 0.0186, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:50:24,227 INFO [train.py:904] (6/8) Epoch 14, batch 3050, loss[loss=0.1882, simple_loss=0.2784, pruned_loss=0.04894, over 16731.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2657, pruned_loss=0.04811, over 3324373.42 frames. ], batch size: 57, lr: 4.90e-03, grad_scale: 8.0 2023-04-29 22:50:33,253 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=135008.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:50:36,568 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-04-29 22:50:40,114 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=135013.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:50:40,181 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9033, 4.1917, 3.9804, 4.0381, 3.7136, 3.8150, 3.7993, 4.1641], device='cuda:6'), covar=tensor([0.1084, 0.0899, 0.1029, 0.0745, 0.0864, 0.1526, 0.0923, 0.1025], device='cuda:6'), in_proj_covar=tensor([0.0601, 0.0749, 0.0612, 0.0538, 0.0478, 0.0477, 0.0625, 0.0579], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:51:32,217 INFO [train.py:904] (6/8) Epoch 14, batch 3100, loss[loss=0.1803, simple_loss=0.257, pruned_loss=0.05179, over 16772.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2647, pruned_loss=0.04793, over 3323180.26 frames. ], batch size: 83, lr: 4.90e-03, grad_scale: 8.0 2023-04-29 22:51:47,483 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-29 22:52:00,080 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8546, 4.8396, 4.7439, 4.4795, 4.4221, 4.8297, 4.5783, 4.5245], device='cuda:6'), covar=tensor([0.0617, 0.0554, 0.0266, 0.0265, 0.0857, 0.0398, 0.0494, 0.0609], device='cuda:6'), in_proj_covar=tensor([0.0269, 0.0370, 0.0327, 0.0308, 0.0344, 0.0355, 0.0223, 0.0381], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:52:05,861 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7958, 4.8541, 5.3408, 5.3353, 5.3256, 4.9528, 4.8935, 4.6417], device='cuda:6'), covar=tensor([0.0330, 0.0545, 0.0394, 0.0420, 0.0506, 0.0404, 0.1012, 0.0476], device='cuda:6'), in_proj_covar=tensor([0.0378, 0.0395, 0.0397, 0.0376, 0.0439, 0.0420, 0.0515, 0.0336], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 22:52:24,973 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.254e+02 2.683e+02 3.212e+02 6.924e+02, threshold=5.366e+02, percent-clipped=1.0 2023-04-29 22:52:40,973 INFO [train.py:904] (6/8) Epoch 14, batch 3150, loss[loss=0.206, simple_loss=0.2765, pruned_loss=0.06775, over 16829.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.264, pruned_loss=0.04745, over 3331859.79 frames. ], batch size: 83, lr: 4.90e-03, grad_scale: 8.0 2023-04-29 22:52:43,902 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9829, 3.1703, 3.2937, 2.1106, 2.7556, 2.3944, 3.4609, 3.4435], device='cuda:6'), covar=tensor([0.0223, 0.0816, 0.0493, 0.1732, 0.0771, 0.0889, 0.0547, 0.0904], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0154, 0.0162, 0.0148, 0.0140, 0.0127, 0.0141, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:53:21,964 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3231, 2.1699, 2.4192, 4.1225, 2.1614, 2.5870, 2.3096, 2.4194], device='cuda:6'), covar=tensor([0.1241, 0.3485, 0.2395, 0.0472, 0.3599, 0.2342, 0.3233, 0.2965], device='cuda:6'), in_proj_covar=tensor([0.0381, 0.0415, 0.0347, 0.0329, 0.0421, 0.0480, 0.0379, 0.0487], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:53:50,749 INFO [train.py:904] (6/8) Epoch 14, batch 3200, loss[loss=0.1431, simple_loss=0.2293, pruned_loss=0.02849, over 17240.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2625, pruned_loss=0.04712, over 3323472.03 frames. ], batch size: 44, lr: 4.90e-03, grad_scale: 8.0 2023-04-29 22:54:17,336 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135171.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:54:41,149 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.733e+02 2.470e+02 2.854e+02 3.612e+02 7.577e+02, threshold=5.709e+02, percent-clipped=5.0 2023-04-29 22:54:59,014 INFO [train.py:904] (6/8) Epoch 14, batch 3250, loss[loss=0.1751, simple_loss=0.2495, pruned_loss=0.05037, over 16250.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2632, pruned_loss=0.04714, over 3331343.26 frames. ], batch size: 165, lr: 4.90e-03, grad_scale: 8.0 2023-04-29 22:55:03,101 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6251, 1.6564, 2.1660, 2.5400, 2.5112, 2.5892, 1.6928, 2.7608], device='cuda:6'), covar=tensor([0.0167, 0.0409, 0.0276, 0.0239, 0.0221, 0.0201, 0.0402, 0.0108], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0181, 0.0167, 0.0173, 0.0180, 0.0136, 0.0182, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:55:22,151 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=135218.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:55:22,416 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8312, 3.3082, 2.8485, 5.1269, 4.2042, 4.5549, 1.7144, 3.4803], device='cuda:6'), covar=tensor([0.1340, 0.0656, 0.1078, 0.0183, 0.0235, 0.0368, 0.1484, 0.0650], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0164, 0.0183, 0.0164, 0.0202, 0.0213, 0.0187, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 22:55:41,817 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=135232.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:56:07,846 INFO [train.py:904] (6/8) Epoch 14, batch 3300, loss[loss=0.1939, simple_loss=0.2879, pruned_loss=0.04993, over 17076.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2639, pruned_loss=0.0472, over 3329532.23 frames. ], batch size: 53, lr: 4.90e-03, grad_scale: 4.0 2023-04-29 22:56:55,783 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6109, 2.3713, 1.8996, 2.1408, 2.7352, 2.5329, 2.7906, 2.8953], device='cuda:6'), covar=tensor([0.0184, 0.0338, 0.0440, 0.0381, 0.0189, 0.0264, 0.0196, 0.0219], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0222, 0.0216, 0.0215, 0.0223, 0.0223, 0.0233, 0.0217], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:57:01,615 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 2.198e+02 2.720e+02 3.178e+02 6.678e+02, threshold=5.440e+02, percent-clipped=1.0 2023-04-29 22:57:16,993 INFO [train.py:904] (6/8) Epoch 14, batch 3350, loss[loss=0.1838, simple_loss=0.2752, pruned_loss=0.04622, over 16715.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2636, pruned_loss=0.04647, over 3335279.43 frames. ], batch size: 62, lr: 4.90e-03, grad_scale: 4.0 2023-04-29 22:57:20,382 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3639, 2.1227, 2.3029, 4.0922, 2.1526, 2.5628, 2.2540, 2.3683], device='cuda:6'), covar=tensor([0.1068, 0.3462, 0.2458, 0.0447, 0.3560, 0.2274, 0.3552, 0.2749], device='cuda:6'), in_proj_covar=tensor([0.0378, 0.0412, 0.0345, 0.0328, 0.0420, 0.0477, 0.0377, 0.0484], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:57:49,335 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0970, 4.7742, 5.0583, 5.2860, 5.4769, 4.8039, 5.4407, 5.4170], device='cuda:6'), covar=tensor([0.1661, 0.1222, 0.1818, 0.0716, 0.0549, 0.0772, 0.0515, 0.0586], device='cuda:6'), in_proj_covar=tensor([0.0599, 0.0743, 0.0896, 0.0764, 0.0573, 0.0587, 0.0601, 0.0702], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 22:58:24,247 INFO [train.py:904] (6/8) Epoch 14, batch 3400, loss[loss=0.1849, simple_loss=0.2667, pruned_loss=0.05153, over 15626.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2625, pruned_loss=0.04597, over 3324916.75 frames. ], batch size: 190, lr: 4.90e-03, grad_scale: 4.0 2023-04-29 22:58:30,481 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-04-29 22:59:16,984 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.205e+02 2.551e+02 3.042e+02 5.815e+02, threshold=5.103e+02, percent-clipped=2.0 2023-04-29 22:59:18,038 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0392, 5.0226, 5.5050, 5.4892, 5.5134, 5.1573, 5.0758, 4.8052], device='cuda:6'), covar=tensor([0.0288, 0.0624, 0.0376, 0.0432, 0.0456, 0.0369, 0.0942, 0.0479], device='cuda:6'), in_proj_covar=tensor([0.0376, 0.0398, 0.0394, 0.0374, 0.0440, 0.0419, 0.0516, 0.0336], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 22:59:32,657 INFO [train.py:904] (6/8) Epoch 14, batch 3450, loss[loss=0.2176, simple_loss=0.2851, pruned_loss=0.07508, over 11510.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2618, pruned_loss=0.046, over 3316608.12 frames. ], batch size: 246, lr: 4.90e-03, grad_scale: 4.0 2023-04-29 23:00:09,389 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-29 23:00:41,081 INFO [train.py:904] (6/8) Epoch 14, batch 3500, loss[loss=0.1981, simple_loss=0.2698, pruned_loss=0.06317, over 15588.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2607, pruned_loss=0.0457, over 3315298.09 frames. ], batch size: 191, lr: 4.90e-03, grad_scale: 4.0 2023-04-29 23:01:37,087 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.212e+02 2.743e+02 3.344e+02 6.080e+02, threshold=5.486e+02, percent-clipped=2.0 2023-04-29 23:01:51,743 INFO [train.py:904] (6/8) Epoch 14, batch 3550, loss[loss=0.1712, simple_loss=0.2638, pruned_loss=0.03926, over 17145.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2602, pruned_loss=0.0456, over 3312460.52 frames. ], batch size: 47, lr: 4.90e-03, grad_scale: 4.0 2023-04-29 23:02:15,211 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=135518.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:02:27,912 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=135527.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:03:02,250 INFO [train.py:904] (6/8) Epoch 14, batch 3600, loss[loss=0.1923, simple_loss=0.2687, pruned_loss=0.05794, over 11688.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.2591, pruned_loss=0.04528, over 3303764.52 frames. ], batch size: 247, lr: 4.89e-03, grad_scale: 8.0 2023-04-29 23:03:03,880 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7731, 4.7359, 4.9502, 4.8396, 4.8184, 5.4268, 4.9423, 4.5679], device='cuda:6'), covar=tensor([0.1209, 0.2001, 0.2101, 0.2167, 0.2740, 0.1036, 0.1546, 0.2625], device='cuda:6'), in_proj_covar=tensor([0.0379, 0.0541, 0.0592, 0.0460, 0.0624, 0.0619, 0.0469, 0.0613], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 23:03:22,563 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=135566.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:03:58,638 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.573e+02 2.279e+02 2.683e+02 3.136e+02 7.106e+02, threshold=5.366e+02, percent-clipped=1.0 2023-04-29 23:04:14,251 INFO [train.py:904] (6/8) Epoch 14, batch 3650, loss[loss=0.1866, simple_loss=0.2542, pruned_loss=0.05946, over 16415.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2585, pruned_loss=0.04596, over 3283228.78 frames. ], batch size: 68, lr: 4.89e-03, grad_scale: 8.0 2023-04-29 23:04:39,728 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7444, 3.5960, 3.9417, 2.0064, 4.0674, 4.0396, 3.3266, 2.9611], device='cuda:6'), covar=tensor([0.0673, 0.0207, 0.0140, 0.1107, 0.0061, 0.0130, 0.0297, 0.0443], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0103, 0.0089, 0.0136, 0.0072, 0.0115, 0.0121, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 23:04:44,549 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 23:04:57,419 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135630.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:05:02,274 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5722, 4.5499, 4.7245, 4.5874, 4.6301, 5.1364, 4.6816, 4.3981], device='cuda:6'), covar=tensor([0.1341, 0.1851, 0.1827, 0.2047, 0.2261, 0.0963, 0.1361, 0.2184], device='cuda:6'), in_proj_covar=tensor([0.0379, 0.0541, 0.0589, 0.0457, 0.0623, 0.0618, 0.0470, 0.0612], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 23:05:31,030 INFO [train.py:904] (6/8) Epoch 14, batch 3700, loss[loss=0.1916, simple_loss=0.2591, pruned_loss=0.0621, over 16716.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2575, pruned_loss=0.04779, over 3275644.13 frames. ], batch size: 83, lr: 4.89e-03, grad_scale: 4.0 2023-04-29 23:06:30,663 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=135691.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:06:32,147 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.568e+02 2.199e+02 2.631e+02 3.126e+02 8.071e+02, threshold=5.262e+02, percent-clipped=3.0 2023-04-29 23:06:46,314 INFO [train.py:904] (6/8) Epoch 14, batch 3750, loss[loss=0.19, simple_loss=0.2684, pruned_loss=0.05576, over 16432.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2579, pruned_loss=0.0494, over 3282994.84 frames. ], batch size: 68, lr: 4.89e-03, grad_scale: 4.0 2023-04-29 23:06:48,725 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2078, 4.1940, 4.1575, 3.6524, 4.1523, 1.6816, 3.9663, 3.6725], device='cuda:6'), covar=tensor([0.0108, 0.0090, 0.0141, 0.0241, 0.0076, 0.2448, 0.0108, 0.0175], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0135, 0.0180, 0.0168, 0.0153, 0.0192, 0.0171, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:08:01,576 INFO [train.py:904] (6/8) Epoch 14, batch 3800, loss[loss=0.2118, simple_loss=0.2751, pruned_loss=0.07419, over 16859.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2597, pruned_loss=0.05082, over 3270837.01 frames. ], batch size: 116, lr: 4.89e-03, grad_scale: 4.0 2023-04-29 23:09:00,856 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.772e+02 2.313e+02 2.548e+02 3.011e+02 5.523e+02, threshold=5.096e+02, percent-clipped=2.0 2023-04-29 23:09:16,026 INFO [train.py:904] (6/8) Epoch 14, batch 3850, loss[loss=0.1816, simple_loss=0.2529, pruned_loss=0.05515, over 16789.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.26, pruned_loss=0.05168, over 3280362.08 frames. ], batch size: 83, lr: 4.89e-03, grad_scale: 4.0 2023-04-29 23:09:25,909 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.64 vs. limit=5.0 2023-04-29 23:09:53,722 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=135827.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:10:30,025 INFO [train.py:904] (6/8) Epoch 14, batch 3900, loss[loss=0.1724, simple_loss=0.2579, pruned_loss=0.04343, over 16606.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2592, pruned_loss=0.05163, over 3280898.50 frames. ], batch size: 57, lr: 4.89e-03, grad_scale: 4.0 2023-04-29 23:11:05,206 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=135875.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:11:08,995 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7529, 2.8317, 2.0985, 2.5361, 3.1680, 2.7450, 3.4405, 3.2895], device='cuda:6'), covar=tensor([0.0053, 0.0308, 0.0466, 0.0367, 0.0193, 0.0315, 0.0143, 0.0200], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0216, 0.0211, 0.0211, 0.0218, 0.0218, 0.0227, 0.0211], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:11:30,218 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.696e+02 2.144e+02 2.448e+02 3.272e+02 6.091e+02, threshold=4.896e+02, percent-clipped=2.0 2023-04-29 23:11:45,241 INFO [train.py:904] (6/8) Epoch 14, batch 3950, loss[loss=0.1964, simple_loss=0.2645, pruned_loss=0.06413, over 16874.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.2599, pruned_loss=0.05263, over 3270613.42 frames. ], batch size: 116, lr: 4.89e-03, grad_scale: 4.0 2023-04-29 23:11:58,599 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3549, 2.7360, 2.1780, 2.4561, 3.1121, 2.7886, 3.1919, 3.1973], device='cuda:6'), covar=tensor([0.0114, 0.0270, 0.0391, 0.0354, 0.0156, 0.0268, 0.0181, 0.0190], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0215, 0.0210, 0.0210, 0.0216, 0.0217, 0.0226, 0.0210], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:12:58,424 INFO [train.py:904] (6/8) Epoch 14, batch 4000, loss[loss=0.1997, simple_loss=0.2846, pruned_loss=0.05734, over 16767.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2603, pruned_loss=0.05319, over 3268112.54 frames. ], batch size: 83, lr: 4.89e-03, grad_scale: 8.0 2023-04-29 23:13:41,817 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135982.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:13:47,604 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=135986.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:13:55,964 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.249e+02 2.823e+02 3.453e+02 5.778e+02, threshold=5.647e+02, percent-clipped=7.0 2023-04-29 23:13:58,904 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135994.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:14:13,537 INFO [train.py:904] (6/8) Epoch 14, batch 4050, loss[loss=0.149, simple_loss=0.2404, pruned_loss=0.02881, over 16826.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2604, pruned_loss=0.05197, over 3266407.53 frames. ], batch size: 102, lr: 4.89e-03, grad_scale: 8.0 2023-04-29 23:15:12,468 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136043.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:15:24,937 INFO [train.py:904] (6/8) Epoch 14, batch 4100, loss[loss=0.1925, simple_loss=0.2725, pruned_loss=0.05629, over 12569.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.2618, pruned_loss=0.05162, over 3261287.32 frames. ], batch size: 247, lr: 4.89e-03, grad_scale: 8.0 2023-04-29 23:15:30,481 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136055.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:16:08,578 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5743, 2.6212, 1.8466, 2.7346, 2.1969, 2.7541, 2.1014, 2.3446], device='cuda:6'), covar=tensor([0.0212, 0.0275, 0.1076, 0.0182, 0.0511, 0.0400, 0.0978, 0.0483], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0170, 0.0188, 0.0146, 0.0169, 0.0213, 0.0198, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 23:16:24,201 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.443e+02 1.983e+02 2.350e+02 2.941e+02 4.738e+02, threshold=4.699e+02, percent-clipped=0.0 2023-04-29 23:16:39,722 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0953, 5.4195, 5.1486, 5.2463, 4.9313, 4.7915, 4.8772, 5.4966], device='cuda:6'), covar=tensor([0.1207, 0.0762, 0.0961, 0.0666, 0.0743, 0.0698, 0.0931, 0.0833], device='cuda:6'), in_proj_covar=tensor([0.0596, 0.0738, 0.0602, 0.0534, 0.0474, 0.0474, 0.0619, 0.0572], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:16:40,531 INFO [train.py:904] (6/8) Epoch 14, batch 4150, loss[loss=0.2093, simple_loss=0.3043, pruned_loss=0.05718, over 16707.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2692, pruned_loss=0.05426, over 3234436.62 frames. ], batch size: 89, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:17:41,232 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136142.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:17:55,765 INFO [train.py:904] (6/8) Epoch 14, batch 4200, loss[loss=0.2267, simple_loss=0.3227, pruned_loss=0.0653, over 16387.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2763, pruned_loss=0.05563, over 3216543.44 frames. ], batch size: 146, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:18:54,126 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.13 vs. limit=5.0 2023-04-29 23:18:55,501 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 2.476e+02 2.870e+02 3.638e+02 7.412e+02, threshold=5.739e+02, percent-clipped=7.0 2023-04-29 23:19:03,445 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-29 23:19:10,103 INFO [train.py:904] (6/8) Epoch 14, batch 4250, loss[loss=0.169, simple_loss=0.2659, pruned_loss=0.03607, over 16678.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2794, pruned_loss=0.05579, over 3200865.78 frames. ], batch size: 62, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:19:12,652 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136203.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:19:16,600 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2497, 4.3583, 4.6063, 4.6296, 4.6379, 4.3038, 4.2533, 4.1799], device='cuda:6'), covar=tensor([0.0294, 0.0475, 0.0444, 0.0409, 0.0418, 0.0420, 0.1144, 0.0545], device='cuda:6'), in_proj_covar=tensor([0.0357, 0.0377, 0.0374, 0.0356, 0.0422, 0.0399, 0.0491, 0.0318], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 23:19:24,609 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.71 vs. limit=5.0 2023-04-29 23:19:27,049 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9564, 4.9314, 4.7045, 4.0797, 4.8322, 2.0195, 4.5948, 4.5202], device='cuda:6'), covar=tensor([0.0071, 0.0061, 0.0139, 0.0270, 0.0074, 0.2187, 0.0092, 0.0157], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0133, 0.0179, 0.0167, 0.0151, 0.0191, 0.0170, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:20:23,644 INFO [train.py:904] (6/8) Epoch 14, batch 4300, loss[loss=0.2161, simple_loss=0.2995, pruned_loss=0.06639, over 15403.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.2808, pruned_loss=0.05473, over 3204320.98 frames. ], batch size: 190, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:20:31,565 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9613, 3.4476, 3.3255, 2.0778, 3.1100, 3.3891, 3.1546, 1.8986], device='cuda:6'), covar=tensor([0.0478, 0.0034, 0.0047, 0.0398, 0.0084, 0.0079, 0.0075, 0.0399], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0073, 0.0074, 0.0130, 0.0086, 0.0095, 0.0084, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 23:21:14,672 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136286.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:21:21,637 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 23:21:23,663 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.749e+02 2.283e+02 2.653e+02 3.073e+02 4.214e+02, threshold=5.307e+02, percent-clipped=0.0 2023-04-29 23:21:38,131 INFO [train.py:904] (6/8) Epoch 14, batch 4350, loss[loss=0.1966, simple_loss=0.2814, pruned_loss=0.05586, over 17027.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2836, pruned_loss=0.05557, over 3197747.86 frames. ], batch size: 50, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:22:00,722 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0222, 5.0162, 4.8933, 4.5988, 4.5566, 4.9629, 4.7441, 4.6303], device='cuda:6'), covar=tensor([0.0480, 0.0246, 0.0214, 0.0225, 0.0829, 0.0261, 0.0330, 0.0520], device='cuda:6'), in_proj_covar=tensor([0.0257, 0.0355, 0.0312, 0.0291, 0.0328, 0.0339, 0.0212, 0.0363], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:22:27,142 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=136334.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:22:29,192 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136335.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:22:33,160 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136338.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:22:51,138 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136350.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:22:53,926 INFO [train.py:904] (6/8) Epoch 14, batch 4400, loss[loss=0.1951, simple_loss=0.281, pruned_loss=0.05464, over 16657.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2854, pruned_loss=0.0565, over 3200471.26 frames. ], batch size: 57, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:23:03,535 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7492, 1.7056, 1.4508, 1.4079, 1.8993, 1.5648, 1.6765, 1.9283], device='cuda:6'), covar=tensor([0.0138, 0.0245, 0.0388, 0.0305, 0.0161, 0.0232, 0.0162, 0.0161], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0214, 0.0210, 0.0208, 0.0216, 0.0216, 0.0222, 0.0208], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:23:51,148 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.242e+02 2.686e+02 3.350e+02 5.764e+02, threshold=5.372e+02, percent-clipped=1.0 2023-04-29 23:23:58,027 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136396.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:23:59,440 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3878, 2.3465, 2.8734, 3.3214, 3.0641, 3.8848, 2.3781, 3.5630], device='cuda:6'), covar=tensor([0.0136, 0.0343, 0.0206, 0.0171, 0.0196, 0.0081, 0.0375, 0.0110], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0179, 0.0163, 0.0172, 0.0179, 0.0135, 0.0180, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:24:06,498 INFO [train.py:904] (6/8) Epoch 14, batch 4450, loss[loss=0.2159, simple_loss=0.2929, pruned_loss=0.06943, over 12184.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2896, pruned_loss=0.05859, over 3189553.99 frames. ], batch size: 247, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:25:18,608 INFO [train.py:904] (6/8) Epoch 14, batch 4500, loss[loss=0.1961, simple_loss=0.2828, pruned_loss=0.05467, over 16825.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2896, pruned_loss=0.05856, over 3218215.34 frames. ], batch size: 83, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:25:20,047 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-04-29 23:26:18,140 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.032e+02 2.340e+02 2.822e+02 5.483e+02, threshold=4.680e+02, percent-clipped=1.0 2023-04-29 23:26:27,079 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136498.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:26:31,831 INFO [train.py:904] (6/8) Epoch 14, batch 4550, loss[loss=0.2017, simple_loss=0.2889, pruned_loss=0.05727, over 17005.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2906, pruned_loss=0.05923, over 3234024.64 frames. ], batch size: 55, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:27:39,533 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5989, 3.6716, 4.1152, 2.0508, 4.2866, 4.3385, 3.1196, 3.1454], device='cuda:6'), covar=tensor([0.0789, 0.0228, 0.0133, 0.1133, 0.0038, 0.0094, 0.0383, 0.0446], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0105, 0.0089, 0.0139, 0.0072, 0.0114, 0.0123, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 23:27:45,694 INFO [train.py:904] (6/8) Epoch 14, batch 4600, loss[loss=0.187, simple_loss=0.2766, pruned_loss=0.04874, over 16907.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2915, pruned_loss=0.05949, over 3237686.73 frames. ], batch size: 96, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:28:42,904 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.386e+02 2.145e+02 2.405e+02 2.956e+02 4.611e+02, threshold=4.810e+02, percent-clipped=0.0 2023-04-29 23:28:57,100 INFO [train.py:904] (6/8) Epoch 14, batch 4650, loss[loss=0.1858, simple_loss=0.2731, pruned_loss=0.04928, over 16699.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2902, pruned_loss=0.05937, over 3232023.38 frames. ], batch size: 76, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:29:33,796 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4374, 4.4622, 4.7792, 4.7622, 4.7922, 4.4595, 4.4466, 4.2562], device='cuda:6'), covar=tensor([0.0269, 0.0381, 0.0283, 0.0358, 0.0401, 0.0331, 0.0878, 0.0491], device='cuda:6'), in_proj_covar=tensor([0.0352, 0.0370, 0.0369, 0.0353, 0.0417, 0.0395, 0.0486, 0.0316], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 23:29:40,967 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-29 23:29:52,211 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136638.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:29:58,825 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136643.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:30:08,513 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136650.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:30:10,583 INFO [train.py:904] (6/8) Epoch 14, batch 4700, loss[loss=0.1851, simple_loss=0.271, pruned_loss=0.04964, over 16607.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2874, pruned_loss=0.05803, over 3237139.99 frames. ], batch size: 62, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:31:01,338 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=136686.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:31:09,102 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136691.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:31:09,908 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 2.115e+02 2.410e+02 3.015e+02 5.394e+02, threshold=4.821e+02, percent-clipped=2.0 2023-04-29 23:31:18,636 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=136698.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:31:20,321 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-29 23:31:23,574 INFO [train.py:904] (6/8) Epoch 14, batch 4750, loss[loss=0.1838, simple_loss=0.2717, pruned_loss=0.04794, over 16291.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2839, pruned_loss=0.05646, over 3212915.21 frames. ], batch size: 165, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:31:27,818 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136704.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:32:00,100 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0619, 5.1030, 4.9579, 4.5738, 4.5025, 5.0159, 4.8801, 4.7087], device='cuda:6'), covar=tensor([0.0532, 0.0507, 0.0241, 0.0256, 0.1045, 0.0404, 0.0305, 0.0566], device='cuda:6'), in_proj_covar=tensor([0.0254, 0.0352, 0.0308, 0.0289, 0.0326, 0.0335, 0.0211, 0.0360], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:32:36,916 INFO [train.py:904] (6/8) Epoch 14, batch 4800, loss[loss=0.2128, simple_loss=0.2996, pruned_loss=0.06296, over 15276.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2799, pruned_loss=0.05445, over 3221510.80 frames. ], batch size: 190, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:33:36,432 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.299e+02 1.970e+02 2.266e+02 2.773e+02 4.879e+02, threshold=4.532e+02, percent-clipped=1.0 2023-04-29 23:33:46,356 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136798.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:33:52,185 INFO [train.py:904] (6/8) Epoch 14, batch 4850, loss[loss=0.1528, simple_loss=0.2433, pruned_loss=0.03113, over 16622.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2806, pruned_loss=0.05402, over 3197657.59 frames. ], batch size: 57, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:33:59,122 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136806.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:34:19,909 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6363, 3.0388, 3.1634, 1.8460, 2.7992, 2.2402, 3.1915, 3.1574], device='cuda:6'), covar=tensor([0.0247, 0.0666, 0.0540, 0.1823, 0.0753, 0.0934, 0.0584, 0.0717], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0153, 0.0161, 0.0147, 0.0139, 0.0127, 0.0139, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-29 23:34:33,676 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 23:34:47,580 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3403, 3.3025, 3.6993, 1.7628, 3.8409, 3.9135, 2.9249, 2.7740], device='cuda:6'), covar=tensor([0.0855, 0.0253, 0.0148, 0.1289, 0.0054, 0.0100, 0.0398, 0.0493], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0104, 0.0089, 0.0138, 0.0072, 0.0114, 0.0123, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 23:34:56,904 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=136846.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:35:05,044 INFO [train.py:904] (6/8) Epoch 14, batch 4900, loss[loss=0.1898, simple_loss=0.2732, pruned_loss=0.05325, over 16862.00 frames. ], tot_loss[loss=0.193, simple_loss=0.28, pruned_loss=0.05297, over 3173307.66 frames. ], batch size: 116, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:35:28,824 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136867.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:35:49,278 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-29 23:35:55,922 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-29 23:36:02,817 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.076e+02 2.406e+02 2.771e+02 5.710e+02, threshold=4.811e+02, percent-clipped=1.0 2023-04-29 23:36:17,693 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0159, 5.0556, 5.4418, 5.3906, 5.4231, 5.0445, 4.9973, 4.6894], device='cuda:6'), covar=tensor([0.0252, 0.0454, 0.0264, 0.0393, 0.0460, 0.0274, 0.0914, 0.0429], device='cuda:6'), in_proj_covar=tensor([0.0351, 0.0372, 0.0370, 0.0353, 0.0416, 0.0393, 0.0484, 0.0316], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 23:36:18,495 INFO [train.py:904] (6/8) Epoch 14, batch 4950, loss[loss=0.2018, simple_loss=0.2941, pruned_loss=0.05478, over 17118.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.2792, pruned_loss=0.0518, over 3180018.44 frames. ], batch size: 48, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:37:30,153 INFO [train.py:904] (6/8) Epoch 14, batch 5000, loss[loss=0.1773, simple_loss=0.2701, pruned_loss=0.04223, over 16627.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2813, pruned_loss=0.05208, over 3195595.58 frames. ], batch size: 57, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:38:23,006 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7983, 5.1019, 5.3048, 5.0520, 5.0347, 5.6940, 5.1352, 4.8643], device='cuda:6'), covar=tensor([0.0931, 0.1624, 0.1517, 0.1753, 0.2498, 0.0775, 0.1295, 0.2289], device='cuda:6'), in_proj_covar=tensor([0.0370, 0.0516, 0.0557, 0.0438, 0.0591, 0.0586, 0.0449, 0.0588], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 23:38:24,317 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136991.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:38:25,064 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.688e+02 2.271e+02 2.523e+02 3.075e+02 5.093e+02, threshold=5.046e+02, percent-clipped=1.0 2023-04-29 23:38:36,219 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136999.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:38:39,688 INFO [train.py:904] (6/8) Epoch 14, batch 5050, loss[loss=0.1654, simple_loss=0.257, pruned_loss=0.03686, over 16470.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2811, pruned_loss=0.05132, over 3213421.92 frames. ], batch size: 68, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:38:42,835 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-04-29 23:39:31,294 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=137039.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:39:49,977 INFO [train.py:904] (6/8) Epoch 14, batch 5100, loss[loss=0.1801, simple_loss=0.2663, pruned_loss=0.04696, over 16767.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2789, pruned_loss=0.05061, over 3214991.52 frames. ], batch size: 134, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:39:53,509 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4425, 2.3188, 2.3819, 4.1876, 2.1092, 2.5984, 2.3989, 2.4822], device='cuda:6'), covar=tensor([0.1138, 0.3501, 0.2471, 0.0458, 0.3983, 0.2589, 0.3228, 0.3293], device='cuda:6'), in_proj_covar=tensor([0.0374, 0.0415, 0.0343, 0.0323, 0.0419, 0.0477, 0.0376, 0.0483], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:40:45,950 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.206e+02 2.593e+02 3.193e+02 6.541e+02, threshold=5.187e+02, percent-clipped=2.0 2023-04-29 23:41:00,262 INFO [train.py:904] (6/8) Epoch 14, batch 5150, loss[loss=0.1864, simple_loss=0.2788, pruned_loss=0.04704, over 17038.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2792, pruned_loss=0.05022, over 3215308.93 frames. ], batch size: 50, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:42:08,569 INFO [train.py:904] (6/8) Epoch 14, batch 5200, loss[loss=0.1959, simple_loss=0.2747, pruned_loss=0.05855, over 16668.00 frames. ], tot_loss[loss=0.1887, simple_loss=0.2781, pruned_loss=0.04969, over 3230801.51 frames. ], batch size: 62, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:42:22,359 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=137162.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:42:46,159 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8177, 5.1179, 4.7872, 4.8833, 4.5662, 4.6148, 4.5060, 5.1484], device='cuda:6'), covar=tensor([0.1059, 0.0775, 0.1054, 0.0758, 0.0884, 0.0820, 0.1065, 0.0935], device='cuda:6'), in_proj_covar=tensor([0.0576, 0.0712, 0.0586, 0.0516, 0.0455, 0.0459, 0.0599, 0.0554], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:43:03,327 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.223e+02 2.609e+02 3.158e+02 5.537e+02, threshold=5.218e+02, percent-clipped=1.0 2023-04-29 23:43:18,147 INFO [train.py:904] (6/8) Epoch 14, batch 5250, loss[loss=0.2022, simple_loss=0.2955, pruned_loss=0.05446, over 16281.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.2771, pruned_loss=0.04984, over 3207518.00 frames. ], batch size: 165, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:44:18,462 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3830, 2.9991, 2.6027, 2.1916, 2.1993, 2.2127, 3.0098, 2.9132], device='cuda:6'), covar=tensor([0.2354, 0.0798, 0.1582, 0.2311, 0.2024, 0.1884, 0.0616, 0.1111], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0259, 0.0289, 0.0288, 0.0282, 0.0229, 0.0276, 0.0308], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:44:26,542 INFO [train.py:904] (6/8) Epoch 14, batch 5300, loss[loss=0.1812, simple_loss=0.2646, pruned_loss=0.04894, over 16597.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2735, pruned_loss=0.04858, over 3201887.84 frames. ], batch size: 62, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:44:41,428 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6315, 4.9500, 4.6630, 4.7293, 4.4325, 4.4291, 4.3314, 5.0037], device='cuda:6'), covar=tensor([0.1178, 0.0771, 0.0959, 0.0714, 0.0753, 0.1071, 0.1056, 0.0808], device='cuda:6'), in_proj_covar=tensor([0.0581, 0.0718, 0.0592, 0.0520, 0.0459, 0.0462, 0.0605, 0.0558], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:45:23,602 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.002e+02 2.371e+02 2.932e+02 5.680e+02, threshold=4.741e+02, percent-clipped=1.0 2023-04-29 23:45:28,215 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4592, 3.9215, 4.1053, 1.7796, 4.3467, 4.5212, 3.3174, 2.9014], device='cuda:6'), covar=tensor([0.1203, 0.0167, 0.0145, 0.1580, 0.0058, 0.0068, 0.0354, 0.0659], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0104, 0.0089, 0.0138, 0.0071, 0.0112, 0.0123, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 23:45:28,448 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.89 vs. limit=5.0 2023-04-29 23:45:33,721 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=137299.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:45:36,785 INFO [train.py:904] (6/8) Epoch 14, batch 5350, loss[loss=0.1953, simple_loss=0.2724, pruned_loss=0.05909, over 12129.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2722, pruned_loss=0.04852, over 3173819.24 frames. ], batch size: 248, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:46:32,109 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3375, 4.1396, 4.3438, 4.5312, 4.6760, 4.2736, 4.6171, 4.6736], device='cuda:6'), covar=tensor([0.1389, 0.1110, 0.1392, 0.0574, 0.0418, 0.0840, 0.0509, 0.0504], device='cuda:6'), in_proj_covar=tensor([0.0564, 0.0696, 0.0833, 0.0711, 0.0537, 0.0551, 0.0553, 0.0656], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:46:40,458 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=137347.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:46:46,692 INFO [train.py:904] (6/8) Epoch 14, batch 5400, loss[loss=0.2053, simple_loss=0.2929, pruned_loss=0.05888, over 16327.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2743, pruned_loss=0.04894, over 3190262.75 frames. ], batch size: 165, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:47:46,868 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.624e+02 2.156e+02 2.492e+02 3.031e+02 4.895e+02, threshold=4.984e+02, percent-clipped=1.0 2023-04-29 23:47:49,965 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3695, 4.6312, 4.9868, 4.8869, 4.9104, 4.5374, 4.2118, 4.3256], device='cuda:6'), covar=tensor([0.0575, 0.0654, 0.0522, 0.0684, 0.0819, 0.0632, 0.1767, 0.0631], device='cuda:6'), in_proj_covar=tensor([0.0359, 0.0380, 0.0379, 0.0362, 0.0426, 0.0403, 0.0500, 0.0323], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-29 23:48:02,464 INFO [train.py:904] (6/8) Epoch 14, batch 5450, loss[loss=0.2192, simple_loss=0.306, pruned_loss=0.06623, over 16948.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2775, pruned_loss=0.05086, over 3177224.21 frames. ], batch size: 109, lr: 4.86e-03, grad_scale: 4.0 2023-04-29 23:49:08,153 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7976, 2.8526, 2.4864, 4.6973, 3.5386, 4.2255, 1.5700, 3.0595], device='cuda:6'), covar=tensor([0.1231, 0.0707, 0.1195, 0.0153, 0.0331, 0.0328, 0.1495, 0.0779], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0164, 0.0186, 0.0162, 0.0202, 0.0210, 0.0189, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 23:49:19,957 INFO [train.py:904] (6/8) Epoch 14, batch 5500, loss[loss=0.2184, simple_loss=0.3064, pruned_loss=0.06521, over 16754.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2855, pruned_loss=0.05579, over 3160894.31 frames. ], batch size: 89, lr: 4.86e-03, grad_scale: 4.0 2023-04-29 23:49:35,275 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=137462.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:50:23,833 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.042e+02 3.226e+02 3.760e+02 4.566e+02 8.180e+02, threshold=7.520e+02, percent-clipped=15.0 2023-04-29 23:50:36,998 INFO [train.py:904] (6/8) Epoch 14, batch 5550, loss[loss=0.2226, simple_loss=0.3086, pruned_loss=0.06828, over 16415.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2934, pruned_loss=0.06172, over 3126922.24 frames. ], batch size: 146, lr: 4.86e-03, grad_scale: 4.0 2023-04-29 23:50:51,277 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=137510.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:51:25,084 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5493, 5.5505, 5.3005, 4.7386, 5.4338, 2.2741, 5.1764, 5.2176], device='cuda:6'), covar=tensor([0.0067, 0.0054, 0.0128, 0.0332, 0.0065, 0.2211, 0.0085, 0.0145], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0131, 0.0177, 0.0167, 0.0149, 0.0190, 0.0166, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:51:55,242 INFO [train.py:904] (6/8) Epoch 14, batch 5600, loss[loss=0.2808, simple_loss=0.3325, pruned_loss=0.1145, over 11222.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.2979, pruned_loss=0.06554, over 3110034.48 frames. ], batch size: 248, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:53:03,057 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.471e+02 3.513e+02 4.095e+02 5.435e+02 1.324e+03, threshold=8.190e+02, percent-clipped=8.0 2023-04-29 23:53:17,473 INFO [train.py:904] (6/8) Epoch 14, batch 5650, loss[loss=0.2696, simple_loss=0.3335, pruned_loss=0.1029, over 11506.00 frames. ], tot_loss[loss=0.2219, simple_loss=0.3036, pruned_loss=0.07016, over 3076640.06 frames. ], batch size: 249, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:54:11,637 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1781, 1.7703, 2.6011, 3.0894, 2.9612, 3.5452, 1.7499, 3.3619], device='cuda:6'), covar=tensor([0.0131, 0.0466, 0.0237, 0.0187, 0.0183, 0.0101, 0.0570, 0.0096], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0175, 0.0161, 0.0167, 0.0175, 0.0132, 0.0177, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:54:38,721 INFO [train.py:904] (6/8) Epoch 14, batch 5700, loss[loss=0.2104, simple_loss=0.2993, pruned_loss=0.06072, over 16743.00 frames. ], tot_loss[loss=0.2227, simple_loss=0.304, pruned_loss=0.07067, over 3084964.60 frames. ], batch size: 83, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:55:29,698 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1102, 3.0129, 3.2511, 1.6258, 3.4320, 3.5045, 2.7383, 2.5576], device='cuda:6'), covar=tensor([0.0865, 0.0274, 0.0258, 0.1247, 0.0075, 0.0155, 0.0445, 0.0477], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0105, 0.0090, 0.0139, 0.0072, 0.0114, 0.0124, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-29 23:55:44,544 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3036, 5.2843, 4.9830, 4.3502, 5.1628, 1.6328, 4.9456, 4.8787], device='cuda:6'), covar=tensor([0.0067, 0.0069, 0.0157, 0.0365, 0.0070, 0.2736, 0.0106, 0.0176], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0130, 0.0175, 0.0164, 0.0147, 0.0188, 0.0164, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:55:45,155 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.087e+02 3.198e+02 3.940e+02 4.824e+02 8.425e+02, threshold=7.880e+02, percent-clipped=1.0 2023-04-29 23:55:59,931 INFO [train.py:904] (6/8) Epoch 14, batch 5750, loss[loss=0.2697, simple_loss=0.3297, pruned_loss=0.1049, over 11409.00 frames. ], tot_loss[loss=0.2261, simple_loss=0.3067, pruned_loss=0.07277, over 3019309.79 frames. ], batch size: 248, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:56:27,337 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7865, 3.6807, 3.9142, 3.7039, 3.8563, 4.1931, 3.9121, 3.6918], device='cuda:6'), covar=tensor([0.2135, 0.2258, 0.2214, 0.2440, 0.2645, 0.1573, 0.1515, 0.2474], device='cuda:6'), in_proj_covar=tensor([0.0370, 0.0519, 0.0564, 0.0438, 0.0593, 0.0586, 0.0447, 0.0597], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-29 23:57:21,149 INFO [train.py:904] (6/8) Epoch 14, batch 5800, loss[loss=0.23, simple_loss=0.3135, pruned_loss=0.0733, over 15469.00 frames. ], tot_loss[loss=0.2242, simple_loss=0.306, pruned_loss=0.07115, over 3036089.80 frames. ], batch size: 191, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:58:11,010 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9625, 2.4120, 2.2807, 2.8341, 2.0438, 3.2360, 1.6983, 2.6938], device='cuda:6'), covar=tensor([0.1183, 0.0558, 0.1069, 0.0157, 0.0135, 0.0378, 0.1404, 0.0657], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0162, 0.0185, 0.0160, 0.0201, 0.0208, 0.0187, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-29 23:58:26,266 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.930e+02 3.076e+02 3.658e+02 4.339e+02 6.692e+02, threshold=7.316e+02, percent-clipped=0.0 2023-04-29 23:58:41,173 INFO [train.py:904] (6/8) Epoch 14, batch 5850, loss[loss=0.2384, simple_loss=0.3085, pruned_loss=0.08414, over 11451.00 frames. ], tot_loss[loss=0.221, simple_loss=0.3033, pruned_loss=0.06941, over 3035872.36 frames. ], batch size: 248, lr: 4.85e-03, grad_scale: 8.0 2023-04-29 23:58:43,035 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9228, 2.0776, 2.4082, 3.1762, 2.1368, 2.2721, 2.3083, 2.1821], device='cuda:6'), covar=tensor([0.1038, 0.2955, 0.1971, 0.0577, 0.3556, 0.2233, 0.2674, 0.3086], device='cuda:6'), in_proj_covar=tensor([0.0371, 0.0410, 0.0340, 0.0318, 0.0416, 0.0470, 0.0373, 0.0477], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-29 23:59:11,877 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9809, 1.9987, 2.5243, 2.9220, 2.8370, 3.4204, 2.1075, 3.2043], device='cuda:6'), covar=tensor([0.0159, 0.0366, 0.0250, 0.0216, 0.0207, 0.0118, 0.0375, 0.0117], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0173, 0.0158, 0.0165, 0.0173, 0.0131, 0.0175, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-29 23:59:56,072 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.32 vs. limit=5.0 2023-04-30 00:00:03,510 INFO [train.py:904] (6/8) Epoch 14, batch 5900, loss[loss=0.1911, simple_loss=0.2839, pruned_loss=0.04914, over 16359.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.3029, pruned_loss=0.06881, over 3049447.29 frames. ], batch size: 35, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:01:10,261 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=137892.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:01:10,915 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.882e+02 2.749e+02 3.330e+02 4.231e+02 8.888e+02, threshold=6.660e+02, percent-clipped=2.0 2023-04-30 00:01:15,132 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=137895.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:01:24,299 INFO [train.py:904] (6/8) Epoch 14, batch 5950, loss[loss=0.2181, simple_loss=0.3035, pruned_loss=0.06638, over 16987.00 frames. ], tot_loss[loss=0.2195, simple_loss=0.3039, pruned_loss=0.06753, over 3062864.95 frames. ], batch size: 53, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:01:59,499 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-30 00:02:40,351 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6884, 1.6931, 1.3956, 1.4186, 1.7549, 1.3973, 1.5832, 1.8215], device='cuda:6'), covar=tensor([0.0165, 0.0299, 0.0434, 0.0357, 0.0208, 0.0289, 0.0170, 0.0221], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0210, 0.0206, 0.0206, 0.0211, 0.0213, 0.0217, 0.0206], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 00:02:44,564 INFO [train.py:904] (6/8) Epoch 14, batch 6000, loss[loss=0.2094, simple_loss=0.2894, pruned_loss=0.06469, over 16201.00 frames. ], tot_loss[loss=0.2197, simple_loss=0.3035, pruned_loss=0.06792, over 3055788.68 frames. ], batch size: 35, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:02:44,564 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 00:02:51,512 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8472, 3.9612, 2.4767, 4.3889, 2.9739, 4.3725, 2.7494, 3.1515], device='cuda:6'), covar=tensor([0.0224, 0.0279, 0.1413, 0.0200, 0.0717, 0.0398, 0.1243, 0.0663], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0168, 0.0188, 0.0138, 0.0166, 0.0207, 0.0196, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 00:02:55,351 INFO [train.py:938] (6/8) Epoch 14, validation: loss=0.1574, simple_loss=0.2706, pruned_loss=0.0221, over 944034.00 frames. 2023-04-30 00:02:55,351 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-30 00:02:57,517 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=137953.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:03:01,589 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=137956.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:03:59,086 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.967e+02 2.895e+02 3.490e+02 4.190e+02 7.002e+02, threshold=6.980e+02, percent-clipped=3.0 2023-04-30 00:04:18,770 INFO [train.py:904] (6/8) Epoch 14, batch 6050, loss[loss=0.1868, simple_loss=0.2809, pruned_loss=0.04635, over 16729.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.3018, pruned_loss=0.06685, over 3075871.36 frames. ], batch size: 68, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:05:40,154 INFO [train.py:904] (6/8) Epoch 14, batch 6100, loss[loss=0.2469, simple_loss=0.3109, pruned_loss=0.09148, over 11678.00 frames. ], tot_loss[loss=0.2167, simple_loss=0.3018, pruned_loss=0.06585, over 3093449.09 frames. ], batch size: 248, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:06:45,098 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.655e+02 2.919e+02 3.731e+02 4.841e+02 8.594e+02, threshold=7.462e+02, percent-clipped=5.0 2023-04-30 00:06:45,960 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-30 00:06:45,979 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-30 00:06:59,049 INFO [train.py:904] (6/8) Epoch 14, batch 6150, loss[loss=0.1771, simple_loss=0.2662, pruned_loss=0.04397, over 17271.00 frames. ], tot_loss[loss=0.2144, simple_loss=0.299, pruned_loss=0.06484, over 3104203.97 frames. ], batch size: 52, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:07:43,945 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-04-30 00:07:57,996 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-30 00:08:16,180 INFO [train.py:904] (6/8) Epoch 14, batch 6200, loss[loss=0.202, simple_loss=0.2853, pruned_loss=0.05931, over 16432.00 frames. ], tot_loss[loss=0.213, simple_loss=0.2973, pruned_loss=0.06433, over 3114243.41 frames. ], batch size: 146, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:08:29,386 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7966, 2.2788, 2.4781, 1.7034, 2.5324, 2.7331, 2.3839, 2.1868], device='cuda:6'), covar=tensor([0.0964, 0.0226, 0.0206, 0.1120, 0.0106, 0.0203, 0.0465, 0.0561], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0104, 0.0089, 0.0138, 0.0072, 0.0113, 0.0123, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 00:09:11,198 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3311, 3.5350, 3.1540, 2.9692, 2.8949, 3.4047, 3.2072, 3.1232], device='cuda:6'), covar=tensor([0.0820, 0.0557, 0.0384, 0.0336, 0.0956, 0.0489, 0.1986, 0.0574], device='cuda:6'), in_proj_covar=tensor([0.0252, 0.0346, 0.0300, 0.0281, 0.0319, 0.0329, 0.0204, 0.0355], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 00:09:17,890 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.837e+02 2.947e+02 3.424e+02 4.172e+02 1.390e+03, threshold=6.848e+02, percent-clipped=3.0 2023-04-30 00:09:31,817 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0968, 5.1648, 4.8885, 4.6158, 4.4639, 5.0234, 4.9629, 4.6128], device='cuda:6'), covar=tensor([0.0953, 0.0829, 0.0370, 0.0396, 0.1271, 0.0698, 0.0479, 0.1070], device='cuda:6'), in_proj_covar=tensor([0.0252, 0.0346, 0.0300, 0.0280, 0.0320, 0.0329, 0.0204, 0.0355], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 00:09:32,464 INFO [train.py:904] (6/8) Epoch 14, batch 6250, loss[loss=0.1988, simple_loss=0.2884, pruned_loss=0.05457, over 16985.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.297, pruned_loss=0.06393, over 3115817.82 frames. ], batch size: 55, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:10:42,323 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138248.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:10:46,688 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138251.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:10:47,496 INFO [train.py:904] (6/8) Epoch 14, batch 6300, loss[loss=0.1778, simple_loss=0.2783, pruned_loss=0.03863, over 16852.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2961, pruned_loss=0.06281, over 3126935.20 frames. ], batch size: 96, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:11:17,745 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138270.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:11:22,325 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8341, 5.1079, 4.8344, 4.8986, 4.5992, 4.5563, 4.5683, 5.1727], device='cuda:6'), covar=tensor([0.1200, 0.0871, 0.1120, 0.0734, 0.0869, 0.0996, 0.1011, 0.0900], device='cuda:6'), in_proj_covar=tensor([0.0588, 0.0719, 0.0595, 0.0526, 0.0459, 0.0468, 0.0606, 0.0555], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 00:11:52,435 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.960e+02 2.966e+02 3.499e+02 4.315e+02 9.401e+02, threshold=6.998e+02, percent-clipped=7.0 2023-04-30 00:12:05,896 INFO [train.py:904] (6/8) Epoch 14, batch 6350, loss[loss=0.275, simple_loss=0.3297, pruned_loss=0.1102, over 11630.00 frames. ], tot_loss[loss=0.2137, simple_loss=0.2978, pruned_loss=0.0648, over 3108638.38 frames. ], batch size: 247, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:12:19,751 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3597, 5.3480, 5.2039, 4.8542, 4.7932, 5.2713, 5.2104, 4.9294], device='cuda:6'), covar=tensor([0.0600, 0.0417, 0.0250, 0.0253, 0.1033, 0.0400, 0.0251, 0.0601], device='cuda:6'), in_proj_covar=tensor([0.0252, 0.0348, 0.0301, 0.0281, 0.0319, 0.0330, 0.0205, 0.0356], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 00:12:37,574 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138322.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:12:50,127 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138331.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:13:22,149 INFO [train.py:904] (6/8) Epoch 14, batch 6400, loss[loss=0.2109, simple_loss=0.2913, pruned_loss=0.06526, over 15491.00 frames. ], tot_loss[loss=0.2151, simple_loss=0.2985, pruned_loss=0.06589, over 3102944.52 frames. ], batch size: 191, lr: 4.84e-03, grad_scale: 8.0 2023-04-30 00:13:53,277 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5574, 3.6207, 4.0330, 1.8325, 4.1244, 4.2401, 3.2201, 2.9296], device='cuda:6'), covar=tensor([0.0795, 0.0210, 0.0140, 0.1262, 0.0060, 0.0120, 0.0302, 0.0454], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0104, 0.0089, 0.0138, 0.0072, 0.0113, 0.0123, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 00:14:09,305 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138383.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:14:23,620 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.151e+02 3.125e+02 3.740e+02 4.766e+02 1.292e+03, threshold=7.481e+02, percent-clipped=6.0 2023-04-30 00:14:37,172 INFO [train.py:904] (6/8) Epoch 14, batch 6450, loss[loss=0.1976, simple_loss=0.2857, pruned_loss=0.05474, over 16190.00 frames. ], tot_loss[loss=0.215, simple_loss=0.2987, pruned_loss=0.06564, over 3090435.64 frames. ], batch size: 165, lr: 4.84e-03, grad_scale: 8.0 2023-04-30 00:15:30,338 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1658, 5.1852, 5.5922, 5.5794, 5.6225, 5.2350, 5.1712, 4.8159], device='cuda:6'), covar=tensor([0.0269, 0.0401, 0.0380, 0.0378, 0.0470, 0.0335, 0.0975, 0.0479], device='cuda:6'), in_proj_covar=tensor([0.0359, 0.0382, 0.0380, 0.0362, 0.0427, 0.0405, 0.0500, 0.0322], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 00:15:54,739 INFO [train.py:904] (6/8) Epoch 14, batch 6500, loss[loss=0.2115, simple_loss=0.295, pruned_loss=0.06395, over 16812.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2964, pruned_loss=0.06492, over 3111791.48 frames. ], batch size: 83, lr: 4.84e-03, grad_scale: 4.0 2023-04-30 00:16:59,451 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.094e+02 2.847e+02 3.325e+02 4.301e+02 7.980e+02, threshold=6.650e+02, percent-clipped=1.0 2023-04-30 00:17:12,094 INFO [train.py:904] (6/8) Epoch 14, batch 6550, loss[loss=0.2014, simple_loss=0.3027, pruned_loss=0.05003, over 16530.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.2991, pruned_loss=0.06584, over 3110325.52 frames. ], batch size: 75, lr: 4.84e-03, grad_scale: 4.0 2023-04-30 00:18:19,012 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0280, 1.8538, 2.4017, 2.8654, 2.8425, 3.3030, 1.9430, 3.2481], device='cuda:6'), covar=tensor([0.0144, 0.0377, 0.0270, 0.0207, 0.0204, 0.0134, 0.0413, 0.0093], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0174, 0.0160, 0.0165, 0.0173, 0.0132, 0.0177, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-30 00:18:22,242 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138548.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:18:25,708 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138551.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:18:26,459 INFO [train.py:904] (6/8) Epoch 14, batch 6600, loss[loss=0.2212, simple_loss=0.3126, pruned_loss=0.06483, over 16440.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.3023, pruned_loss=0.06697, over 3106113.12 frames. ], batch size: 146, lr: 4.84e-03, grad_scale: 4.0 2023-04-30 00:19:29,570 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.028e+02 2.801e+02 3.547e+02 4.436e+02 1.538e+03, threshold=7.095e+02, percent-clipped=5.0 2023-04-30 00:19:33,986 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=138596.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:19:38,206 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=138599.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:19:42,709 INFO [train.py:904] (6/8) Epoch 14, batch 6650, loss[loss=0.2067, simple_loss=0.2894, pruned_loss=0.06203, over 16869.00 frames. ], tot_loss[loss=0.2179, simple_loss=0.3015, pruned_loss=0.06714, over 3112320.08 frames. ], batch size: 116, lr: 4.84e-03, grad_scale: 4.0 2023-04-30 00:19:55,647 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1312, 2.0498, 2.1285, 3.7675, 1.9808, 2.4376, 2.1698, 2.2128], device='cuda:6'), covar=tensor([0.1220, 0.3353, 0.2723, 0.0508, 0.4042, 0.2346, 0.3204, 0.3390], device='cuda:6'), in_proj_covar=tensor([0.0370, 0.0408, 0.0340, 0.0317, 0.0417, 0.0470, 0.0374, 0.0476], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 00:20:19,712 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138626.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:20:58,166 INFO [train.py:904] (6/8) Epoch 14, batch 6700, loss[loss=0.2304, simple_loss=0.309, pruned_loss=0.07588, over 15357.00 frames. ], tot_loss[loss=0.2174, simple_loss=0.3, pruned_loss=0.06741, over 3102336.22 frames. ], batch size: 191, lr: 4.84e-03, grad_scale: 4.0 2023-04-30 00:21:27,989 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7109, 3.7612, 2.8839, 2.1578, 2.5683, 2.2880, 4.0059, 3.4514], device='cuda:6'), covar=tensor([0.2599, 0.0695, 0.1629, 0.2450, 0.2499, 0.1964, 0.0435, 0.1104], device='cuda:6'), in_proj_covar=tensor([0.0309, 0.0261, 0.0289, 0.0290, 0.0285, 0.0231, 0.0277, 0.0309], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 00:21:40,537 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138678.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:21:44,957 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138681.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:21:52,981 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138686.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:22:03,782 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.231e+02 3.354e+02 4.045e+02 5.286e+02 1.404e+03, threshold=8.090e+02, percent-clipped=8.0 2023-04-30 00:22:09,169 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138697.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:22:15,574 INFO [train.py:904] (6/8) Epoch 14, batch 6750, loss[loss=0.1911, simple_loss=0.28, pruned_loss=0.0511, over 16831.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.2993, pruned_loss=0.06764, over 3091824.08 frames. ], batch size: 116, lr: 4.84e-03, grad_scale: 4.0 2023-04-30 00:23:14,950 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138742.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:23:23,034 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138747.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:23:29,043 INFO [train.py:904] (6/8) Epoch 14, batch 6800, loss[loss=0.2372, simple_loss=0.305, pruned_loss=0.08465, over 11320.00 frames. ], tot_loss[loss=0.218, simple_loss=0.3001, pruned_loss=0.06792, over 3079345.51 frames. ], batch size: 247, lr: 4.84e-03, grad_scale: 8.0 2023-04-30 00:23:38,536 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138758.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:24:18,254 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1597, 3.3547, 3.5341, 3.4912, 3.5051, 3.3419, 3.2070, 3.3998], device='cuda:6'), covar=tensor([0.0595, 0.0782, 0.0575, 0.0671, 0.0774, 0.0702, 0.1404, 0.0662], device='cuda:6'), in_proj_covar=tensor([0.0352, 0.0374, 0.0373, 0.0355, 0.0420, 0.0398, 0.0489, 0.0316], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 00:24:34,694 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 2.812e+02 3.433e+02 4.059e+02 1.030e+03, threshold=6.866e+02, percent-clipped=2.0 2023-04-30 00:24:46,910 INFO [train.py:904] (6/8) Epoch 14, batch 6850, loss[loss=0.2102, simple_loss=0.3153, pruned_loss=0.05262, over 16864.00 frames. ], tot_loss[loss=0.2179, simple_loss=0.3003, pruned_loss=0.06773, over 3076416.60 frames. ], batch size: 109, lr: 4.84e-03, grad_scale: 8.0 2023-04-30 00:26:02,477 INFO [train.py:904] (6/8) Epoch 14, batch 6900, loss[loss=0.2379, simple_loss=0.3328, pruned_loss=0.07151, over 16747.00 frames. ], tot_loss[loss=0.218, simple_loss=0.3024, pruned_loss=0.06676, over 3102017.68 frames. ], batch size: 83, lr: 4.84e-03, grad_scale: 8.0 2023-04-30 00:26:18,575 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-30 00:27:01,689 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4315, 3.2477, 2.6467, 2.1271, 2.2813, 2.1750, 3.3078, 3.0644], device='cuda:6'), covar=tensor([0.2595, 0.0749, 0.1584, 0.2387, 0.2278, 0.1934, 0.0490, 0.1196], device='cuda:6'), in_proj_covar=tensor([0.0310, 0.0261, 0.0291, 0.0292, 0.0286, 0.0231, 0.0278, 0.0309], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 00:27:10,346 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.725e+02 2.815e+02 3.844e+02 4.824e+02 1.140e+03, threshold=7.689e+02, percent-clipped=2.0 2023-04-30 00:27:22,848 INFO [train.py:904] (6/8) Epoch 14, batch 6950, loss[loss=0.2112, simple_loss=0.2935, pruned_loss=0.06443, over 16646.00 frames. ], tot_loss[loss=0.2196, simple_loss=0.3032, pruned_loss=0.068, over 3099439.67 frames. ], batch size: 62, lr: 4.84e-03, grad_scale: 8.0 2023-04-30 00:27:33,830 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138909.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:27:47,811 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-30 00:28:00,028 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138926.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:28:38,216 INFO [train.py:904] (6/8) Epoch 14, batch 7000, loss[loss=0.2074, simple_loss=0.3062, pruned_loss=0.05426, over 16678.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.303, pruned_loss=0.06725, over 3089686.02 frames. ], batch size: 62, lr: 4.83e-03, grad_scale: 8.0 2023-04-30 00:29:06,862 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138970.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 00:29:12,849 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=138974.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:29:18,588 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138978.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:29:42,505 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.027e+02 2.958e+02 3.558e+02 4.270e+02 7.816e+02, threshold=7.117e+02, percent-clipped=1.0 2023-04-30 00:29:55,148 INFO [train.py:904] (6/8) Epoch 14, batch 7050, loss[loss=0.2686, simple_loss=0.3328, pruned_loss=0.1022, over 11455.00 frames. ], tot_loss[loss=0.2201, simple_loss=0.3045, pruned_loss=0.06783, over 3073497.06 frames. ], batch size: 247, lr: 4.83e-03, grad_scale: 8.0 2023-04-30 00:30:33,409 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139026.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:30:50,681 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139037.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:30:58,310 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139042.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:31:13,408 INFO [train.py:904] (6/8) Epoch 14, batch 7100, loss[loss=0.1976, simple_loss=0.2864, pruned_loss=0.0544, over 16559.00 frames. ], tot_loss[loss=0.2189, simple_loss=0.303, pruned_loss=0.06741, over 3075718.17 frames. ], batch size: 75, lr: 4.83e-03, grad_scale: 8.0 2023-04-30 00:31:15,009 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139053.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:32:18,348 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.051e+02 2.828e+02 3.507e+02 4.047e+02 5.820e+02, threshold=7.014e+02, percent-clipped=0.0 2023-04-30 00:32:31,527 INFO [train.py:904] (6/8) Epoch 14, batch 7150, loss[loss=0.1825, simple_loss=0.2763, pruned_loss=0.04436, over 16307.00 frames. ], tot_loss[loss=0.2176, simple_loss=0.301, pruned_loss=0.06715, over 3056430.07 frames. ], batch size: 35, lr: 4.83e-03, grad_scale: 8.0 2023-04-30 00:33:10,445 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0136, 2.4548, 2.6348, 1.8295, 2.6786, 2.7827, 2.4190, 2.3557], device='cuda:6'), covar=tensor([0.0769, 0.0222, 0.0224, 0.0954, 0.0104, 0.0209, 0.0422, 0.0440], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0103, 0.0090, 0.0138, 0.0072, 0.0112, 0.0123, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 00:33:47,875 INFO [train.py:904] (6/8) Epoch 14, batch 7200, loss[loss=0.1865, simple_loss=0.2721, pruned_loss=0.05048, over 17242.00 frames. ], tot_loss[loss=0.215, simple_loss=0.299, pruned_loss=0.06548, over 3055262.40 frames. ], batch size: 45, lr: 4.83e-03, grad_scale: 8.0 2023-04-30 00:34:23,033 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-30 00:34:44,293 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7637, 1.6629, 2.1147, 2.5852, 2.5978, 3.0075, 1.6645, 2.8829], device='cuda:6'), covar=tensor([0.0171, 0.0447, 0.0300, 0.0276, 0.0250, 0.0148, 0.0522, 0.0118], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0174, 0.0159, 0.0165, 0.0173, 0.0130, 0.0177, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-30 00:34:55,352 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.878e+02 2.722e+02 3.151e+02 3.906e+02 1.100e+03, threshold=6.302e+02, percent-clipped=3.0 2023-04-30 00:35:07,113 INFO [train.py:904] (6/8) Epoch 14, batch 7250, loss[loss=0.1945, simple_loss=0.2771, pruned_loss=0.056, over 16411.00 frames. ], tot_loss[loss=0.2126, simple_loss=0.2967, pruned_loss=0.06418, over 3065644.26 frames. ], batch size: 146, lr: 4.83e-03, grad_scale: 4.0 2023-04-30 00:35:59,658 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4325, 2.9343, 2.6930, 2.2608, 2.3149, 2.2641, 2.8698, 2.8975], device='cuda:6'), covar=tensor([0.2245, 0.0779, 0.1387, 0.2181, 0.2018, 0.1837, 0.0444, 0.1072], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0263, 0.0291, 0.0293, 0.0287, 0.0233, 0.0278, 0.0311], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 00:36:01,873 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139238.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:36:22,186 INFO [train.py:904] (6/8) Epoch 14, batch 7300, loss[loss=0.2625, simple_loss=0.322, pruned_loss=0.1015, over 11847.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2961, pruned_loss=0.06416, over 3070987.66 frames. ], batch size: 248, lr: 4.83e-03, grad_scale: 4.0 2023-04-30 00:36:34,576 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0018, 2.3151, 2.2499, 2.8949, 2.1618, 3.1574, 1.7638, 2.6639], device='cuda:6'), covar=tensor([0.1144, 0.0628, 0.1103, 0.0163, 0.0148, 0.0347, 0.1349, 0.0691], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0165, 0.0186, 0.0163, 0.0203, 0.0210, 0.0189, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 00:36:43,285 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139265.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 00:37:00,555 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139276.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:37:20,441 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139289.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:37:29,626 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.078e+02 2.984e+02 3.521e+02 4.495e+02 8.579e+02, threshold=7.042e+02, percent-clipped=5.0 2023-04-30 00:37:37,678 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139299.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:37:40,533 INFO [train.py:904] (6/8) Epoch 14, batch 7350, loss[loss=0.2059, simple_loss=0.2934, pruned_loss=0.05915, over 16816.00 frames. ], tot_loss[loss=0.2134, simple_loss=0.2972, pruned_loss=0.06478, over 3065758.99 frames. ], batch size: 102, lr: 4.83e-03, grad_scale: 4.0 2023-04-30 00:38:36,861 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139337.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:38:37,004 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139337.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 00:38:45,007 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139342.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:38:57,268 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139350.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:38:59,361 INFO [train.py:904] (6/8) Epoch 14, batch 7400, loss[loss=0.1927, simple_loss=0.2851, pruned_loss=0.05021, over 16656.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.2977, pruned_loss=0.06462, over 3083668.88 frames. ], batch size: 57, lr: 4.83e-03, grad_scale: 4.0 2023-04-30 00:39:01,719 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139353.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:39:52,989 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139385.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:40:00,776 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139390.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:40:08,330 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.005e+02 2.794e+02 3.378e+02 4.039e+02 6.978e+02, threshold=6.757e+02, percent-clipped=0.0 2023-04-30 00:40:18,654 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139401.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:40:19,477 INFO [train.py:904] (6/8) Epoch 14, batch 7450, loss[loss=0.2001, simple_loss=0.2861, pruned_loss=0.05706, over 16597.00 frames. ], tot_loss[loss=0.214, simple_loss=0.2983, pruned_loss=0.06483, over 3101294.49 frames. ], batch size: 62, lr: 4.83e-03, grad_scale: 4.0 2023-04-30 00:40:49,148 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3573, 3.2993, 3.4098, 3.5017, 3.5430, 3.2563, 3.4791, 3.5800], device='cuda:6'), covar=tensor([0.1240, 0.0961, 0.1019, 0.0563, 0.0673, 0.2028, 0.1079, 0.0760], device='cuda:6'), in_proj_covar=tensor([0.0553, 0.0686, 0.0820, 0.0700, 0.0532, 0.0549, 0.0556, 0.0653], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 00:41:43,386 INFO [train.py:904] (6/8) Epoch 14, batch 7500, loss[loss=0.206, simple_loss=0.2886, pruned_loss=0.06169, over 17207.00 frames. ], tot_loss[loss=0.2133, simple_loss=0.2981, pruned_loss=0.06423, over 3084424.70 frames. ], batch size: 45, lr: 4.83e-03, grad_scale: 4.0 2023-04-30 00:42:53,285 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.056e+02 3.028e+02 3.537e+02 4.351e+02 9.459e+02, threshold=7.073e+02, percent-clipped=3.0 2023-04-30 00:43:02,943 INFO [train.py:904] (6/8) Epoch 14, batch 7550, loss[loss=0.1933, simple_loss=0.2813, pruned_loss=0.05264, over 16712.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.2974, pruned_loss=0.06482, over 3076797.75 frames. ], batch size: 134, lr: 4.82e-03, grad_scale: 4.0 2023-04-30 00:44:17,127 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9218, 4.1583, 3.9824, 3.9980, 3.6962, 3.8153, 3.8432, 4.1395], device='cuda:6'), covar=tensor([0.0978, 0.0827, 0.0984, 0.0791, 0.0727, 0.1414, 0.0858, 0.0950], device='cuda:6'), in_proj_covar=tensor([0.0583, 0.0716, 0.0592, 0.0517, 0.0452, 0.0467, 0.0600, 0.0552], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 00:44:19,040 INFO [train.py:904] (6/8) Epoch 14, batch 7600, loss[loss=0.2357, simple_loss=0.3103, pruned_loss=0.08057, over 16722.00 frames. ], tot_loss[loss=0.2126, simple_loss=0.296, pruned_loss=0.06461, over 3077913.21 frames. ], batch size: 134, lr: 4.82e-03, grad_scale: 8.0 2023-04-30 00:44:20,373 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0669, 1.8145, 2.4402, 2.8819, 2.7833, 3.3799, 1.8799, 3.2441], device='cuda:6'), covar=tensor([0.0154, 0.0459, 0.0263, 0.0238, 0.0242, 0.0128, 0.0490, 0.0112], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0175, 0.0159, 0.0165, 0.0173, 0.0131, 0.0177, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-30 00:44:39,736 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139565.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 00:44:49,041 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139571.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:45:06,324 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139582.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:45:14,448 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2834, 3.4038, 3.5961, 3.5825, 3.5745, 3.3958, 3.4362, 3.4596], device='cuda:6'), covar=tensor([0.0377, 0.0634, 0.0422, 0.0440, 0.0527, 0.0512, 0.0781, 0.0508], device='cuda:6'), in_proj_covar=tensor([0.0350, 0.0366, 0.0368, 0.0352, 0.0414, 0.0393, 0.0480, 0.0312], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 00:45:23,470 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139594.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:45:24,244 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.790e+02 2.745e+02 3.435e+02 4.599e+02 7.932e+02, threshold=6.869e+02, percent-clipped=2.0 2023-04-30 00:45:34,163 INFO [train.py:904] (6/8) Epoch 14, batch 7650, loss[loss=0.2041, simple_loss=0.2915, pruned_loss=0.05836, over 16716.00 frames. ], tot_loss[loss=0.2144, simple_loss=0.2971, pruned_loss=0.0658, over 3058701.14 frames. ], batch size: 89, lr: 4.82e-03, grad_scale: 8.0 2023-04-30 00:45:50,717 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139613.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:45:52,757 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7599, 1.7332, 2.2376, 2.6672, 2.6671, 3.0287, 1.8141, 2.9261], device='cuda:6'), covar=tensor([0.0174, 0.0428, 0.0273, 0.0242, 0.0223, 0.0138, 0.0436, 0.0114], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0174, 0.0158, 0.0164, 0.0172, 0.0130, 0.0176, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-30 00:46:19,767 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139632.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 00:46:19,816 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139632.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:46:37,206 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139643.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:46:39,507 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139645.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:46:49,448 INFO [train.py:904] (6/8) Epoch 14, batch 7700, loss[loss=0.2561, simple_loss=0.3229, pruned_loss=0.09463, over 11400.00 frames. ], tot_loss[loss=0.2146, simple_loss=0.297, pruned_loss=0.06603, over 3077265.88 frames. ], batch size: 246, lr: 4.82e-03, grad_scale: 8.0 2023-04-30 00:47:57,375 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.201e+02 3.391e+02 3.997e+02 4.581e+02 8.153e+02, threshold=7.994e+02, percent-clipped=6.0 2023-04-30 00:48:06,798 INFO [train.py:904] (6/8) Epoch 14, batch 7750, loss[loss=0.1974, simple_loss=0.2819, pruned_loss=0.05643, over 16648.00 frames. ], tot_loss[loss=0.216, simple_loss=0.2982, pruned_loss=0.0669, over 3061713.24 frames. ], batch size: 62, lr: 4.82e-03, grad_scale: 4.0 2023-04-30 00:48:33,549 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9846, 4.9484, 4.7863, 3.3624, 4.7858, 1.6041, 4.4392, 4.4494], device='cuda:6'), covar=tensor([0.0164, 0.0125, 0.0204, 0.0805, 0.0157, 0.3304, 0.0208, 0.0361], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0127, 0.0173, 0.0161, 0.0146, 0.0187, 0.0161, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 00:48:42,034 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.62 vs. limit=5.0 2023-04-30 00:49:24,607 INFO [train.py:904] (6/8) Epoch 14, batch 7800, loss[loss=0.1986, simple_loss=0.296, pruned_loss=0.05059, over 17198.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.3, pruned_loss=0.06825, over 3051639.24 frames. ], batch size: 44, lr: 4.82e-03, grad_scale: 4.0 2023-04-30 00:49:51,619 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139769.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:50:22,534 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139789.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:50:32,559 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.773e+02 2.870e+02 3.646e+02 4.608e+02 1.262e+03, threshold=7.291e+02, percent-clipped=3.0 2023-04-30 00:50:41,153 INFO [train.py:904] (6/8) Epoch 14, batch 7850, loss[loss=0.2215, simple_loss=0.3039, pruned_loss=0.06958, over 16675.00 frames. ], tot_loss[loss=0.2179, simple_loss=0.3006, pruned_loss=0.06758, over 3065418.76 frames. ], batch size: 134, lr: 4.82e-03, grad_scale: 4.0 2023-04-30 00:50:46,046 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139804.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:50:48,971 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139806.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:51:24,810 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139830.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:51:53,637 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139850.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:51:53,670 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139850.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:51:56,192 INFO [train.py:904] (6/8) Epoch 14, batch 7900, loss[loss=0.2218, simple_loss=0.3089, pruned_loss=0.0673, over 16356.00 frames. ], tot_loss[loss=0.2162, simple_loss=0.2991, pruned_loss=0.06665, over 3078214.94 frames. ], batch size: 146, lr: 4.82e-03, grad_scale: 4.0 2023-04-30 00:52:15,829 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139865.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:52:18,398 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139867.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:52:26,745 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5223, 3.4796, 3.8358, 1.8978, 3.9285, 4.0383, 2.9024, 3.0237], device='cuda:6'), covar=tensor([0.0771, 0.0205, 0.0159, 0.1175, 0.0064, 0.0119, 0.0425, 0.0394], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0104, 0.0090, 0.0139, 0.0072, 0.0112, 0.0124, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 00:53:01,515 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139894.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:53:02,108 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-30 00:53:03,656 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.073e+02 2.924e+02 3.560e+02 4.104e+02 6.335e+02, threshold=7.120e+02, percent-clipped=0.0 2023-04-30 00:53:12,912 INFO [train.py:904] (6/8) Epoch 14, batch 7950, loss[loss=0.2165, simple_loss=0.2925, pruned_loss=0.07021, over 16454.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2996, pruned_loss=0.06685, over 3096034.61 frames. ], batch size: 68, lr: 4.82e-03, grad_scale: 4.0 2023-04-30 00:53:28,046 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139911.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:53:51,347 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139927.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:53:59,014 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139932.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 00:54:08,414 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139938.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:54:13,648 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139942.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:54:18,429 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139945.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:54:28,354 INFO [train.py:904] (6/8) Epoch 14, batch 8000, loss[loss=0.1966, simple_loss=0.285, pruned_loss=0.05405, over 16711.00 frames. ], tot_loss[loss=0.2167, simple_loss=0.2997, pruned_loss=0.06682, over 3111418.33 frames. ], batch size: 134, lr: 4.82e-03, grad_scale: 8.0 2023-04-30 00:55:11,103 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139980.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:55:30,359 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139993.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:55:34,105 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.098e+02 3.021e+02 3.514e+02 4.508e+02 9.326e+02, threshold=7.028e+02, percent-clipped=1.0 2023-04-30 00:55:46,024 INFO [train.py:904] (6/8) Epoch 14, batch 8050, loss[loss=0.2357, simple_loss=0.3082, pruned_loss=0.08157, over 11507.00 frames. ], tot_loss[loss=0.2162, simple_loss=0.2993, pruned_loss=0.06652, over 3106714.81 frames. ], batch size: 246, lr: 4.82e-03, grad_scale: 8.0 2023-04-30 00:56:59,041 INFO [train.py:904] (6/8) Epoch 14, batch 8100, loss[loss=0.2434, simple_loss=0.3082, pruned_loss=0.0893, over 11654.00 frames. ], tot_loss[loss=0.2153, simple_loss=0.2987, pruned_loss=0.066, over 3099260.07 frames. ], batch size: 246, lr: 4.82e-03, grad_scale: 8.0 2023-04-30 00:57:49,624 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-30 00:58:06,570 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.001e+02 3.052e+02 3.743e+02 5.058e+02 1.059e+03, threshold=7.487e+02, percent-clipped=6.0 2023-04-30 00:58:14,610 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1177, 2.9150, 3.1885, 1.7196, 3.2721, 3.3555, 2.6596, 2.5267], device='cuda:6'), covar=tensor([0.0837, 0.0250, 0.0169, 0.1283, 0.0080, 0.0192, 0.0432, 0.0521], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0103, 0.0090, 0.0138, 0.0072, 0.0112, 0.0123, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 00:58:17,058 INFO [train.py:904] (6/8) Epoch 14, batch 8150, loss[loss=0.2017, simple_loss=0.2853, pruned_loss=0.05903, over 16879.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2957, pruned_loss=0.06455, over 3123687.08 frames. ], batch size: 116, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 00:58:18,944 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140103.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:58:29,454 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140110.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:58:52,679 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140125.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:59:03,613 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 00:59:09,465 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9524, 2.3619, 2.2756, 2.7971, 2.1136, 3.2365, 1.7550, 2.6671], device='cuda:6'), covar=tensor([0.1171, 0.0605, 0.1114, 0.0160, 0.0149, 0.0416, 0.1418, 0.0716], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0166, 0.0188, 0.0164, 0.0206, 0.0213, 0.0190, 0.0188], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 00:59:23,736 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140145.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:59:34,241 INFO [train.py:904] (6/8) Epoch 14, batch 8200, loss[loss=0.2381, simple_loss=0.3044, pruned_loss=0.08591, over 11429.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.2938, pruned_loss=0.06457, over 3114009.61 frames. ], batch size: 247, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 00:59:47,438 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140160.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:59:50,569 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140162.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:59:53,815 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140164.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:00:05,632 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140171.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:00:21,225 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 01:00:44,842 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.704e+02 2.531e+02 3.267e+02 4.112e+02 9.606e+02, threshold=6.535e+02, percent-clipped=4.0 2023-04-30 01:00:55,629 INFO [train.py:904] (6/8) Epoch 14, batch 8250, loss[loss=0.1871, simple_loss=0.2729, pruned_loss=0.05066, over 12435.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.293, pruned_loss=0.06266, over 3074944.75 frames. ], batch size: 246, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:01:02,322 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140206.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:01:37,189 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140227.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:01:55,206 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140238.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:02:16,398 INFO [train.py:904] (6/8) Epoch 14, batch 8300, loss[loss=0.1845, simple_loss=0.2739, pruned_loss=0.04759, over 16371.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.29, pruned_loss=0.05894, over 3086919.11 frames. ], batch size: 35, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:02:29,768 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140259.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:02:55,099 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140275.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:03:12,851 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140286.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:03:27,882 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.670e+02 2.340e+02 2.728e+02 3.506e+02 7.480e+02, threshold=5.455e+02, percent-clipped=2.0 2023-04-30 01:03:36,676 INFO [train.py:904] (6/8) Epoch 14, batch 8350, loss[loss=0.1874, simple_loss=0.2811, pruned_loss=0.04683, over 15320.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2891, pruned_loss=0.05679, over 3088072.45 frames. ], batch size: 191, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:03:43,842 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2023-04-30 01:04:07,938 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140320.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:04:34,149 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-30 01:04:57,945 INFO [train.py:904] (6/8) Epoch 14, batch 8400, loss[loss=0.1824, simple_loss=0.2756, pruned_loss=0.04458, over 16681.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2868, pruned_loss=0.05499, over 3081237.06 frames. ], batch size: 134, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:05:26,572 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-30 01:05:54,817 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140387.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:06:08,145 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.599e+02 2.349e+02 2.839e+02 3.572e+02 6.227e+02, threshold=5.678e+02, percent-clipped=3.0 2023-04-30 01:06:18,701 INFO [train.py:904] (6/8) Epoch 14, batch 8450, loss[loss=0.1882, simple_loss=0.2658, pruned_loss=0.05529, over 12467.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2851, pruned_loss=0.05341, over 3081011.36 frames. ], batch size: 248, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:06:48,786 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9803, 4.2356, 4.1095, 4.1165, 3.7508, 3.8204, 3.8525, 4.2372], device='cuda:6'), covar=tensor([0.1026, 0.0882, 0.0848, 0.0737, 0.0784, 0.1680, 0.0945, 0.0977], device='cuda:6'), in_proj_covar=tensor([0.0570, 0.0699, 0.0576, 0.0504, 0.0442, 0.0456, 0.0586, 0.0533], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:06:55,838 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140425.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:07:27,243 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140445.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:07:32,313 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140448.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:07:37,964 INFO [train.py:904] (6/8) Epoch 14, batch 8500, loss[loss=0.1961, simple_loss=0.2912, pruned_loss=0.05053, over 16712.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.2811, pruned_loss=0.0509, over 3063341.10 frames. ], batch size: 124, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:07:50,069 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140459.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:07:51,411 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140460.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:07:54,679 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140462.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:08:01,434 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140466.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:08:13,347 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140473.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:08:32,527 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3236, 1.9330, 2.1420, 3.8874, 1.9170, 2.2442, 2.1205, 2.1261], device='cuda:6'), covar=tensor([0.1226, 0.4414, 0.2931, 0.0623, 0.5281, 0.3045, 0.3923, 0.4170], device='cuda:6'), in_proj_covar=tensor([0.0364, 0.0403, 0.0336, 0.0311, 0.0412, 0.0461, 0.0367, 0.0469], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:08:38,693 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140489.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:08:46,171 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140493.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:08:51,328 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.625e+02 2.295e+02 2.729e+02 3.409e+02 8.118e+02, threshold=5.459e+02, percent-clipped=4.0 2023-04-30 01:09:01,486 INFO [train.py:904] (6/8) Epoch 14, batch 8550, loss[loss=0.1901, simple_loss=0.2856, pruned_loss=0.04733, over 15217.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2782, pruned_loss=0.04983, over 3031138.66 frames. ], batch size: 190, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:09:11,900 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140506.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:09:15,776 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140508.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:09:17,343 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140509.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:09:19,130 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140510.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:10:39,254 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140550.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:10:41,521 INFO [train.py:904] (6/8) Epoch 14, batch 8600, loss[loss=0.1871, simple_loss=0.2654, pruned_loss=0.05438, over 12297.00 frames. ], tot_loss[loss=0.1874, simple_loss=0.2776, pruned_loss=0.04862, over 3004741.21 frames. ], batch size: 246, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:10:46,550 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140554.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:10:56,211 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6047, 3.6663, 3.4388, 3.2325, 3.2579, 3.5771, 3.3197, 3.4041], device='cuda:6'), covar=tensor([0.0570, 0.0583, 0.0258, 0.0223, 0.0538, 0.0465, 0.1184, 0.0559], device='cuda:6'), in_proj_covar=tensor([0.0248, 0.0340, 0.0297, 0.0274, 0.0308, 0.0323, 0.0205, 0.0350], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:11:19,591 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140570.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:12:02,895 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6593, 2.6351, 2.5579, 4.4062, 2.8167, 4.1307, 1.6074, 2.9222], device='cuda:6'), covar=tensor([0.1529, 0.0884, 0.1217, 0.0170, 0.0253, 0.0407, 0.1712, 0.0823], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0161, 0.0182, 0.0158, 0.0197, 0.0208, 0.0186, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-30 01:12:03,112 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-30 01:12:06,768 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.419e+02 2.341e+02 2.953e+02 3.525e+02 9.042e+02, threshold=5.906e+02, percent-clipped=3.0 2023-04-30 01:12:18,463 INFO [train.py:904] (6/8) Epoch 14, batch 8650, loss[loss=0.1773, simple_loss=0.2836, pruned_loss=0.03551, over 16313.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.2756, pruned_loss=0.04663, over 3016532.60 frames. ], batch size: 146, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:12:52,263 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140615.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:13:46,497 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1529, 1.9188, 2.0516, 3.6205, 1.9243, 2.2256, 2.0912, 2.0559], device='cuda:6'), covar=tensor([0.1090, 0.4041, 0.2784, 0.0521, 0.4426, 0.2750, 0.3518, 0.3686], device='cuda:6'), in_proj_covar=tensor([0.0362, 0.0402, 0.0335, 0.0310, 0.0411, 0.0459, 0.0365, 0.0466], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:14:05,319 INFO [train.py:904] (6/8) Epoch 14, batch 8700, loss[loss=0.1875, simple_loss=0.2811, pruned_loss=0.04694, over 15466.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2729, pruned_loss=0.04565, over 3007279.09 frames. ], batch size: 191, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:15:24,089 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.380e+02 2.814e+02 3.680e+02 6.881e+02, threshold=5.627e+02, percent-clipped=2.0 2023-04-30 01:15:38,741 INFO [train.py:904] (6/8) Epoch 14, batch 8750, loss[loss=0.1762, simple_loss=0.2724, pruned_loss=0.03995, over 16767.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2725, pruned_loss=0.04505, over 3004100.75 frames. ], batch size: 124, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:16:31,520 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140723.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:17:10,062 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-30 01:17:11,078 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140743.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:17:28,573 INFO [train.py:904] (6/8) Epoch 14, batch 8800, loss[loss=0.1819, simple_loss=0.27, pruned_loss=0.04689, over 16934.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2706, pruned_loss=0.0439, over 3012275.49 frames. ], batch size: 109, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:17:37,426 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9767, 1.8085, 1.6158, 1.5401, 1.9405, 1.5754, 1.6149, 1.9308], device='cuda:6'), covar=tensor([0.0118, 0.0256, 0.0341, 0.0302, 0.0183, 0.0251, 0.0151, 0.0181], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0208, 0.0202, 0.0204, 0.0207, 0.0208, 0.0208, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:17:43,084 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140759.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:17:57,090 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140766.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:18:35,771 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140784.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:19:00,060 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.211e+02 2.894e+02 3.689e+02 7.406e+02, threshold=5.789e+02, percent-clipped=2.0 2023-04-30 01:19:04,938 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1434, 5.4315, 5.2495, 5.2689, 4.9441, 4.8422, 4.8565, 5.5056], device='cuda:6'), covar=tensor([0.1020, 0.0768, 0.0801, 0.0580, 0.0719, 0.0726, 0.1032, 0.0872], device='cuda:6'), in_proj_covar=tensor([0.0564, 0.0697, 0.0569, 0.0502, 0.0441, 0.0453, 0.0584, 0.0530], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:19:12,351 INFO [train.py:904] (6/8) Epoch 14, batch 8850, loss[loss=0.1867, simple_loss=0.2935, pruned_loss=0.03993, over 16595.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2733, pruned_loss=0.04338, over 3016370.40 frames. ], batch size: 62, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:19:23,526 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140807.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:19:27,178 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140809.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:19:37,859 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140814.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:20:44,014 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140845.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:20:58,217 INFO [train.py:904] (6/8) Epoch 14, batch 8900, loss[loss=0.1766, simple_loss=0.2752, pruned_loss=0.03901, over 16702.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2735, pruned_loss=0.04273, over 3023817.04 frames. ], batch size: 89, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:21:25,580 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140865.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:21:34,682 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140870.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:22:46,903 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.640e+02 2.342e+02 2.775e+02 3.236e+02 6.448e+02, threshold=5.549e+02, percent-clipped=1.0 2023-04-30 01:23:01,545 INFO [train.py:904] (6/8) Epoch 14, batch 8950, loss[loss=0.172, simple_loss=0.2581, pruned_loss=0.04291, over 12578.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2739, pruned_loss=0.0432, over 3032110.31 frames. ], batch size: 250, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:23:29,088 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140915.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:23:32,814 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140917.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:24:00,003 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8088, 3.8626, 3.9307, 3.7898, 3.8928, 4.3071, 4.0331, 3.7487], device='cuda:6'), covar=tensor([0.2132, 0.2131, 0.2025, 0.2372, 0.2531, 0.1482, 0.1350, 0.2377], device='cuda:6'), in_proj_covar=tensor([0.0354, 0.0496, 0.0545, 0.0422, 0.0563, 0.0576, 0.0435, 0.0570], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 01:24:38,377 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140947.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:24:48,129 INFO [train.py:904] (6/8) Epoch 14, batch 9000, loss[loss=0.1733, simple_loss=0.2603, pruned_loss=0.04312, over 11883.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2705, pruned_loss=0.0418, over 3028869.10 frames. ], batch size: 248, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:24:48,129 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 01:24:58,094 INFO [train.py:938] (6/8) Epoch 14, validation: loss=0.1514, simple_loss=0.2553, pruned_loss=0.0237, over 944034.00 frames. 2023-04-30 01:24:58,094 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-30 01:25:21,317 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140963.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:25:52,427 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140978.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:26:29,788 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.243e+02 2.199e+02 2.701e+02 3.272e+02 7.252e+02, threshold=5.402e+02, percent-clipped=4.0 2023-04-30 01:26:39,922 INFO [train.py:904] (6/8) Epoch 14, batch 9050, loss[loss=0.1732, simple_loss=0.2607, pruned_loss=0.04282, over 16811.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2716, pruned_loss=0.04249, over 3026846.97 frames. ], batch size: 124, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:26:55,112 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141008.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:27:13,659 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.92 vs. limit=5.0 2023-04-30 01:27:35,531 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4939, 4.6362, 4.7827, 4.5685, 4.6689, 5.1599, 4.7339, 4.3755], device='cuda:6'), covar=tensor([0.1158, 0.1756, 0.1872, 0.2033, 0.2236, 0.0927, 0.1454, 0.2733], device='cuda:6'), in_proj_covar=tensor([0.0353, 0.0496, 0.0545, 0.0421, 0.0564, 0.0576, 0.0435, 0.0572], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 01:28:08,147 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141043.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:28:25,839 INFO [train.py:904] (6/8) Epoch 14, batch 9100, loss[loss=0.1891, simple_loss=0.284, pruned_loss=0.04712, over 16698.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2717, pruned_loss=0.04309, over 3046259.76 frames. ], batch size: 134, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:29:30,794 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141079.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:29:58,973 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141091.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:30:13,176 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.729e+02 2.366e+02 2.876e+02 3.628e+02 8.857e+02, threshold=5.752e+02, percent-clipped=2.0 2023-04-30 01:30:22,841 INFO [train.py:904] (6/8) Epoch 14, batch 9150, loss[loss=0.1665, simple_loss=0.2587, pruned_loss=0.03714, over 16940.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2719, pruned_loss=0.04303, over 3035967.83 frames. ], batch size: 109, lr: 4.80e-03, grad_scale: 4.0 2023-04-30 01:31:37,794 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4761, 3.0036, 2.7247, 2.2120, 2.2048, 2.2531, 2.9448, 2.8538], device='cuda:6'), covar=tensor([0.2316, 0.0716, 0.1415, 0.2358, 0.2296, 0.1848, 0.0434, 0.1072], device='cuda:6'), in_proj_covar=tensor([0.0301, 0.0252, 0.0283, 0.0282, 0.0269, 0.0228, 0.0265, 0.0296], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:31:51,274 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141143.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:31:54,384 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141145.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:32:04,830 INFO [train.py:904] (6/8) Epoch 14, batch 9200, loss[loss=0.1828, simple_loss=0.2677, pruned_loss=0.04893, over 16653.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2676, pruned_loss=0.04217, over 3028003.43 frames. ], batch size: 62, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:32:15,033 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5871, 3.6440, 3.3906, 3.1898, 3.2457, 3.5430, 3.3072, 3.3574], device='cuda:6'), covar=tensor([0.0480, 0.0450, 0.0245, 0.0212, 0.0477, 0.0353, 0.1202, 0.0435], device='cuda:6'), in_proj_covar=tensor([0.0244, 0.0332, 0.0293, 0.0271, 0.0302, 0.0317, 0.0201, 0.0342], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:32:29,656 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141165.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:32:29,818 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141165.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:32:54,845 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6557, 2.7169, 1.8606, 2.8477, 2.0774, 2.7935, 2.0939, 2.3933], device='cuda:6'), covar=tensor([0.0248, 0.0292, 0.1181, 0.0266, 0.0648, 0.0554, 0.1165, 0.0590], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0161, 0.0184, 0.0133, 0.0163, 0.0199, 0.0192, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 01:33:23,826 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141193.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:33:31,526 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.509e+02 2.979e+02 4.006e+02 1.027e+03, threshold=5.958e+02, percent-clipped=7.0 2023-04-30 01:33:37,874 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5716, 2.0694, 1.7832, 1.8404, 2.3509, 2.0613, 2.1378, 2.4923], device='cuda:6'), covar=tensor([0.0127, 0.0350, 0.0444, 0.0426, 0.0224, 0.0329, 0.0160, 0.0218], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0210, 0.0204, 0.0206, 0.0209, 0.0210, 0.0208, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:33:40,990 INFO [train.py:904] (6/8) Epoch 14, batch 9250, loss[loss=0.1485, simple_loss=0.2486, pruned_loss=0.02426, over 16889.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2676, pruned_loss=0.04228, over 3030874.20 frames. ], batch size: 90, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:33:46,145 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141204.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:34:03,406 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141213.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:34:38,137 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4040, 2.1090, 2.1319, 4.1281, 2.1296, 2.5991, 2.2445, 2.2684], device='cuda:6'), covar=tensor([0.0975, 0.3577, 0.2678, 0.0372, 0.3971, 0.2203, 0.3242, 0.3417], device='cuda:6'), in_proj_covar=tensor([0.0363, 0.0402, 0.0337, 0.0311, 0.0413, 0.0457, 0.0367, 0.0467], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:34:51,879 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6549, 4.7402, 4.5091, 4.1924, 4.1931, 4.6152, 4.4140, 4.3307], device='cuda:6'), covar=tensor([0.0547, 0.0494, 0.0290, 0.0273, 0.0823, 0.0520, 0.0440, 0.0614], device='cuda:6'), in_proj_covar=tensor([0.0245, 0.0333, 0.0294, 0.0272, 0.0303, 0.0318, 0.0202, 0.0343], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:35:30,127 INFO [train.py:904] (6/8) Epoch 14, batch 9300, loss[loss=0.1513, simple_loss=0.2539, pruned_loss=0.0243, over 16736.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2649, pruned_loss=0.04128, over 3016771.52 frames. ], batch size: 83, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:36:20,430 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141273.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:37:05,690 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.463e+02 2.032e+02 2.564e+02 3.502e+02 9.246e+02, threshold=5.129e+02, percent-clipped=2.0 2023-04-30 01:37:14,522 INFO [train.py:904] (6/8) Epoch 14, batch 9350, loss[loss=0.1779, simple_loss=0.2763, pruned_loss=0.03968, over 16811.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2653, pruned_loss=0.04141, over 3038366.97 frames. ], batch size: 83, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:37:17,072 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141303.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:38:09,289 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141329.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:38:54,830 INFO [train.py:904] (6/8) Epoch 14, batch 9400, loss[loss=0.1861, simple_loss=0.2842, pruned_loss=0.04399, over 16199.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2656, pruned_loss=0.04103, over 3049061.05 frames. ], batch size: 165, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:39:49,508 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141379.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:40:12,156 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141390.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:40:25,322 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.208e+02 2.590e+02 3.171e+02 6.862e+02, threshold=5.180e+02, percent-clipped=4.0 2023-04-30 01:40:33,425 INFO [train.py:904] (6/8) Epoch 14, batch 9450, loss[loss=0.1819, simple_loss=0.2705, pruned_loss=0.0467, over 12809.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2667, pruned_loss=0.04094, over 3035485.67 frames. ], batch size: 246, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:40:50,679 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7319, 1.3303, 1.7284, 1.6991, 1.8432, 1.8755, 1.6031, 1.8021], device='cuda:6'), covar=tensor([0.0264, 0.0371, 0.0199, 0.0263, 0.0247, 0.0192, 0.0343, 0.0117], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0176, 0.0158, 0.0162, 0.0173, 0.0129, 0.0176, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-30 01:41:00,346 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0612, 3.0814, 1.8416, 3.3113, 2.1996, 3.2680, 2.0971, 2.6157], device='cuda:6'), covar=tensor([0.0263, 0.0393, 0.1486, 0.0232, 0.0823, 0.0595, 0.1332, 0.0646], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0160, 0.0182, 0.0132, 0.0162, 0.0197, 0.0190, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 01:41:03,171 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141417.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 01:41:24,366 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141427.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:41:54,096 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141441.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:42:14,326 INFO [train.py:904] (6/8) Epoch 14, batch 9500, loss[loss=0.1532, simple_loss=0.2384, pruned_loss=0.03405, over 12733.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2654, pruned_loss=0.04026, over 3034580.73 frames. ], batch size: 248, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:42:43,608 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141465.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:42:57,214 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-30 01:43:08,455 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141478.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 01:43:10,445 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7505, 2.3271, 2.2936, 3.6217, 2.2225, 3.7360, 1.3200, 2.9472], device='cuda:6'), covar=tensor([0.1328, 0.0771, 0.1193, 0.0129, 0.0105, 0.0344, 0.1681, 0.0672], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0162, 0.0182, 0.0157, 0.0191, 0.0207, 0.0187, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 01:43:19,288 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2547, 3.7215, 3.6595, 2.6279, 3.3328, 3.7302, 3.5293, 2.2547], device='cuda:6'), covar=tensor([0.0448, 0.0030, 0.0040, 0.0296, 0.0075, 0.0066, 0.0058, 0.0387], device='cuda:6'), in_proj_covar=tensor([0.0127, 0.0069, 0.0071, 0.0126, 0.0083, 0.0092, 0.0081, 0.0120], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 01:43:47,731 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.502e+02 2.299e+02 2.699e+02 3.093e+02 4.944e+02, threshold=5.398e+02, percent-clipped=0.0 2023-04-30 01:43:55,177 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141499.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:43:59,693 INFO [train.py:904] (6/8) Epoch 14, batch 9550, loss[loss=0.1747, simple_loss=0.2733, pruned_loss=0.03808, over 16799.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2656, pruned_loss=0.0405, over 3047038.55 frames. ], batch size: 124, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:44:01,090 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141502.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:44:24,361 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141513.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:44:49,567 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4094, 2.1160, 2.1275, 4.0263, 2.1224, 2.5278, 2.2123, 2.2402], device='cuda:6'), covar=tensor([0.0932, 0.3422, 0.2636, 0.0381, 0.3817, 0.2310, 0.3297, 0.3455], device='cuda:6'), in_proj_covar=tensor([0.0358, 0.0396, 0.0333, 0.0308, 0.0408, 0.0450, 0.0362, 0.0460], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:45:41,071 INFO [train.py:904] (6/8) Epoch 14, batch 9600, loss[loss=0.1954, simple_loss=0.2988, pruned_loss=0.04595, over 16880.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.268, pruned_loss=0.04169, over 3055016.31 frames. ], batch size: 116, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:45:41,579 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0773, 3.3748, 3.5536, 3.5216, 3.5539, 3.4039, 3.2093, 3.4691], device='cuda:6'), covar=tensor([0.0653, 0.0833, 0.0716, 0.0880, 0.0813, 0.0769, 0.1232, 0.0604], device='cuda:6'), in_proj_covar=tensor([0.0332, 0.0349, 0.0350, 0.0334, 0.0394, 0.0372, 0.0449, 0.0297], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:6') 2023-04-30 01:45:54,156 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0053, 3.3957, 3.3059, 2.1797, 3.0213, 3.3892, 3.2593, 1.8326], device='cuda:6'), covar=tensor([0.0500, 0.0035, 0.0043, 0.0379, 0.0094, 0.0069, 0.0063, 0.0474], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0069, 0.0071, 0.0127, 0.0083, 0.0092, 0.0081, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 01:46:19,360 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0916, 2.7704, 2.8621, 2.0640, 2.6179, 2.1404, 2.6537, 2.9158], device='cuda:6'), covar=tensor([0.0299, 0.0717, 0.0555, 0.1690, 0.0801, 0.0929, 0.0621, 0.0725], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0143, 0.0156, 0.0143, 0.0135, 0.0123, 0.0134, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 01:46:22,384 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141573.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:47:19,263 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.635e+02 2.296e+02 2.650e+02 3.286e+02 9.159e+02, threshold=5.299e+02, percent-clipped=3.0 2023-04-30 01:47:30,277 INFO [train.py:904] (6/8) Epoch 14, batch 9650, loss[loss=0.1794, simple_loss=0.2716, pruned_loss=0.04362, over 16892.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.27, pruned_loss=0.04228, over 3051296.43 frames. ], batch size: 116, lr: 4.79e-03, grad_scale: 4.0 2023-04-30 01:47:34,876 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141603.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:47:54,402 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1429, 5.4952, 5.2364, 5.2668, 4.9494, 4.9411, 4.8552, 5.5644], device='cuda:6'), covar=tensor([0.1165, 0.0822, 0.1105, 0.0697, 0.0824, 0.0740, 0.1198, 0.0824], device='cuda:6'), in_proj_covar=tensor([0.0562, 0.0691, 0.0567, 0.0500, 0.0440, 0.0451, 0.0582, 0.0528], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:48:16,938 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141621.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:48:24,233 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-30 01:49:17,044 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141651.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:49:17,846 INFO [train.py:904] (6/8) Epoch 14, batch 9700, loss[loss=0.1942, simple_loss=0.2825, pruned_loss=0.05301, over 16102.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.269, pruned_loss=0.04166, over 3080665.27 frames. ], batch size: 165, lr: 4.79e-03, grad_scale: 4.0 2023-04-30 01:50:12,760 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1446, 3.4217, 3.3034, 2.3351, 3.1497, 3.4443, 3.3326, 1.7231], device='cuda:6'), covar=tensor([0.0502, 0.0051, 0.0067, 0.0370, 0.0106, 0.0095, 0.0077, 0.0606], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0069, 0.0071, 0.0127, 0.0083, 0.0092, 0.0081, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 01:50:27,991 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141685.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:50:38,035 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9807, 3.8271, 4.0165, 4.1732, 4.2587, 3.8644, 4.2354, 4.2745], device='cuda:6'), covar=tensor([0.1537, 0.1141, 0.1305, 0.0634, 0.0555, 0.1358, 0.0579, 0.0680], device='cuda:6'), in_proj_covar=tensor([0.0531, 0.0657, 0.0775, 0.0673, 0.0507, 0.0525, 0.0529, 0.0629], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:50:45,547 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0091, 4.0861, 3.8953, 3.6406, 3.6504, 4.0222, 3.6942, 3.7971], device='cuda:6'), covar=tensor([0.0554, 0.0514, 0.0299, 0.0251, 0.0712, 0.0385, 0.0943, 0.0538], device='cuda:6'), in_proj_covar=tensor([0.0243, 0.0329, 0.0290, 0.0270, 0.0299, 0.0315, 0.0200, 0.0339], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-04-30 01:50:53,324 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.703e+02 2.310e+02 2.782e+02 3.506e+02 1.114e+03, threshold=5.564e+02, percent-clipped=5.0 2023-04-30 01:50:58,408 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-30 01:51:00,293 INFO [train.py:904] (6/8) Epoch 14, batch 9750, loss[loss=0.1766, simple_loss=0.2725, pruned_loss=0.04035, over 15414.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2678, pruned_loss=0.04177, over 3086415.16 frames. ], batch size: 191, lr: 4.79e-03, grad_scale: 4.0 2023-04-30 01:51:36,730 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.45 vs. limit=5.0 2023-04-30 01:52:38,197 INFO [train.py:904] (6/8) Epoch 14, batch 9800, loss[loss=0.194, simple_loss=0.2999, pruned_loss=0.044, over 15515.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2673, pruned_loss=0.041, over 3076293.43 frames. ], batch size: 191, lr: 4.79e-03, grad_scale: 4.0 2023-04-30 01:53:18,671 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141773.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 01:53:39,969 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6221, 3.6324, 2.1263, 4.1395, 2.5510, 4.0328, 2.3107, 2.8210], device='cuda:6'), covar=tensor([0.0265, 0.0349, 0.1572, 0.0164, 0.0937, 0.0487, 0.1448, 0.0760], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0159, 0.0183, 0.0131, 0.0162, 0.0195, 0.0191, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 01:54:13,228 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141797.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:54:13,940 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.610e+02 2.347e+02 2.793e+02 3.311e+02 9.229e+02, threshold=5.586e+02, percent-clipped=3.0 2023-04-30 01:54:17,233 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141799.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:54:21,457 INFO [train.py:904] (6/8) Epoch 14, batch 9850, loss[loss=0.1761, simple_loss=0.2701, pruned_loss=0.04108, over 16243.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2685, pruned_loss=0.04076, over 3082310.56 frames. ], batch size: 165, lr: 4.79e-03, grad_scale: 4.0 2023-04-30 01:55:34,303 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.89 vs. limit=5.0 2023-04-30 01:55:58,339 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1532, 5.6509, 5.7917, 5.5860, 5.6263, 6.1368, 5.6480, 5.3778], device='cuda:6'), covar=tensor([0.0717, 0.1465, 0.1646, 0.1763, 0.1892, 0.0825, 0.1239, 0.2201], device='cuda:6'), in_proj_covar=tensor([0.0347, 0.0490, 0.0542, 0.0416, 0.0558, 0.0569, 0.0427, 0.0562], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 01:56:05,075 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141847.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:56:14,788 INFO [train.py:904] (6/8) Epoch 14, batch 9900, loss[loss=0.1719, simple_loss=0.2735, pruned_loss=0.03517, over 16253.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2691, pruned_loss=0.04057, over 3081249.97 frames. ], batch size: 166, lr: 4.78e-03, grad_scale: 4.0 2023-04-30 01:57:46,977 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3780, 5.7404, 5.4446, 5.5108, 5.1780, 5.1248, 5.0411, 5.8202], device='cuda:6'), covar=tensor([0.1079, 0.0845, 0.1135, 0.0674, 0.0827, 0.0635, 0.1155, 0.0850], device='cuda:6'), in_proj_covar=tensor([0.0558, 0.0687, 0.0565, 0.0498, 0.0438, 0.0450, 0.0578, 0.0524], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 01:58:03,748 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.629e+02 2.142e+02 2.825e+02 3.311e+02 8.255e+02, threshold=5.650e+02, percent-clipped=2.0 2023-04-30 01:58:13,944 INFO [train.py:904] (6/8) Epoch 14, batch 9950, loss[loss=0.1983, simple_loss=0.3039, pruned_loss=0.04634, over 16261.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2714, pruned_loss=0.04104, over 3070816.21 frames. ], batch size: 165, lr: 4.78e-03, grad_scale: 4.0 2023-04-30 01:58:25,219 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9136, 3.8155, 4.3984, 2.1048, 4.5426, 4.5309, 3.3362, 3.4971], device='cuda:6'), covar=tensor([0.0627, 0.0209, 0.0109, 0.1135, 0.0030, 0.0082, 0.0315, 0.0350], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0098, 0.0083, 0.0132, 0.0067, 0.0105, 0.0118, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 01:58:48,453 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141916.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:59:21,085 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141928.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:00:06,718 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 02:00:16,381 INFO [train.py:904] (6/8) Epoch 14, batch 10000, loss[loss=0.1769, simple_loss=0.2586, pruned_loss=0.04761, over 12983.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2699, pruned_loss=0.0407, over 3065884.86 frames. ], batch size: 248, lr: 4.78e-03, grad_scale: 8.0 2023-04-30 02:01:07,146 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141977.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:01:26,024 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141985.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:01:34,657 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141989.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:01:50,846 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.061e+02 2.551e+02 3.100e+02 8.241e+02, threshold=5.102e+02, percent-clipped=1.0 2023-04-30 02:02:01,156 INFO [train.py:904] (6/8) Epoch 14, batch 10050, loss[loss=0.1781, simple_loss=0.2732, pruned_loss=0.04156, over 16471.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2696, pruned_loss=0.04022, over 3084308.56 frames. ], batch size: 68, lr: 4.78e-03, grad_scale: 8.0 2023-04-30 02:03:01,577 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=142033.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:03:16,231 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5792, 2.3103, 2.3089, 4.4107, 2.3144, 2.7157, 2.4228, 2.4946], device='cuda:6'), covar=tensor([0.0959, 0.3437, 0.2612, 0.0319, 0.3725, 0.2242, 0.3061, 0.3119], device='cuda:6'), in_proj_covar=tensor([0.0359, 0.0395, 0.0333, 0.0306, 0.0406, 0.0448, 0.0360, 0.0457], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:03:25,940 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5328, 3.5207, 2.8006, 2.0917, 2.2338, 2.2583, 3.7040, 3.2436], device='cuda:6'), covar=tensor([0.2714, 0.0690, 0.1621, 0.2621, 0.2463, 0.1902, 0.0441, 0.1099], device='cuda:6'), in_proj_covar=tensor([0.0299, 0.0251, 0.0282, 0.0279, 0.0262, 0.0225, 0.0262, 0.0292], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:03:27,351 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1302, 2.1026, 2.2576, 3.5785, 2.1069, 2.3601, 2.2237, 2.2044], device='cuda:6'), covar=tensor([0.1008, 0.3369, 0.2510, 0.0452, 0.3818, 0.2372, 0.3036, 0.3415], device='cuda:6'), in_proj_covar=tensor([0.0358, 0.0395, 0.0333, 0.0306, 0.0406, 0.0448, 0.0360, 0.0456], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:03:34,640 INFO [train.py:904] (6/8) Epoch 14, batch 10100, loss[loss=0.1736, simple_loss=0.2703, pruned_loss=0.03852, over 16536.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.27, pruned_loss=0.04045, over 3098798.40 frames. ], batch size: 75, lr: 4.78e-03, grad_scale: 8.0 2023-04-30 02:04:17,836 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1210, 4.1441, 4.5316, 4.5186, 4.5171, 4.2466, 4.2294, 4.1659], device='cuda:6'), covar=tensor([0.0334, 0.0673, 0.0406, 0.0402, 0.0447, 0.0458, 0.0855, 0.0507], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0345, 0.0348, 0.0329, 0.0389, 0.0369, 0.0444, 0.0295], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:6') 2023-04-30 02:04:19,921 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142073.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 02:04:34,887 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1873, 4.2166, 4.0653, 3.7878, 3.8075, 4.1575, 3.8880, 3.8790], device='cuda:6'), covar=tensor([0.0498, 0.0549, 0.0278, 0.0265, 0.0670, 0.0499, 0.0609, 0.0529], device='cuda:6'), in_proj_covar=tensor([0.0237, 0.0322, 0.0283, 0.0264, 0.0292, 0.0307, 0.0194, 0.0330], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-04-30 02:04:49,779 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142097.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:04:50,457 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 2.492e+02 2.995e+02 3.648e+02 7.218e+02, threshold=5.990e+02, percent-clipped=8.0 2023-04-30 02:05:19,914 INFO [train.py:904] (6/8) Epoch 15, batch 0, loss[loss=0.2877, simple_loss=0.3321, pruned_loss=0.1217, over 16864.00 frames. ], tot_loss[loss=0.2877, simple_loss=0.3321, pruned_loss=0.1217, over 16864.00 frames. ], batch size: 109, lr: 4.62e-03, grad_scale: 8.0 2023-04-30 02:05:19,914 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 02:05:27,351 INFO [train.py:938] (6/8) Epoch 15, validation: loss=0.1501, simple_loss=0.2536, pruned_loss=0.02333, over 944034.00 frames. 2023-04-30 02:05:27,352 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-30 02:05:53,982 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=142121.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 02:06:27,844 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=142145.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:06:37,637 INFO [train.py:904] (6/8) Epoch 15, batch 50, loss[loss=0.1836, simple_loss=0.2717, pruned_loss=0.04778, over 17049.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2804, pruned_loss=0.05873, over 743091.79 frames. ], batch size: 53, lr: 4.62e-03, grad_scale: 2.0 2023-04-30 02:07:24,268 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-04-30 02:07:44,550 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.745e+02 2.683e+02 3.202e+02 3.967e+02 8.381e+02, threshold=6.403e+02, percent-clipped=5.0 2023-04-30 02:07:47,914 INFO [train.py:904] (6/8) Epoch 15, batch 100, loss[loss=0.1729, simple_loss=0.2751, pruned_loss=0.03528, over 17117.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2735, pruned_loss=0.05445, over 1324799.80 frames. ], batch size: 49, lr: 4.62e-03, grad_scale: 2.0 2023-04-30 02:08:33,269 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-30 02:08:56,686 INFO [train.py:904] (6/8) Epoch 15, batch 150, loss[loss=0.1814, simple_loss=0.2718, pruned_loss=0.04547, over 16733.00 frames. ], tot_loss[loss=0.191, simple_loss=0.2737, pruned_loss=0.05415, over 1763203.09 frames. ], batch size: 57, lr: 4.62e-03, grad_scale: 2.0 2023-04-30 02:09:25,349 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=142272.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:09:42,934 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=142284.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:10:04,594 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.354e+02 2.662e+02 3.206e+02 6.929e+02, threshold=5.325e+02, percent-clipped=1.0 2023-04-30 02:10:07,770 INFO [train.py:904] (6/8) Epoch 15, batch 200, loss[loss=0.1942, simple_loss=0.2679, pruned_loss=0.06028, over 16827.00 frames. ], tot_loss[loss=0.1931, simple_loss=0.2756, pruned_loss=0.05525, over 2094864.85 frames. ], batch size: 102, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:11:16,845 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-30 02:11:17,114 INFO [train.py:904] (6/8) Epoch 15, batch 250, loss[loss=0.1443, simple_loss=0.2292, pruned_loss=0.0297, over 16976.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.2715, pruned_loss=0.05343, over 2372770.94 frames. ], batch size: 41, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:12:22,521 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.243e+02 2.346e+02 2.797e+02 3.590e+02 6.679e+02, threshold=5.594e+02, percent-clipped=6.0 2023-04-30 02:12:25,436 INFO [train.py:904] (6/8) Epoch 15, batch 300, loss[loss=0.1781, simple_loss=0.2498, pruned_loss=0.05316, over 16758.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.269, pruned_loss=0.05182, over 2588339.96 frames. ], batch size: 102, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:12:54,822 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 02:13:36,228 INFO [train.py:904] (6/8) Epoch 15, batch 350, loss[loss=0.2241, simple_loss=0.2871, pruned_loss=0.08051, over 16875.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2667, pruned_loss=0.05037, over 2752002.78 frames. ], batch size: 116, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:14:42,900 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.497e+02 2.253e+02 2.722e+02 3.272e+02 5.602e+02, threshold=5.443e+02, percent-clipped=1.0 2023-04-30 02:14:45,198 INFO [train.py:904] (6/8) Epoch 15, batch 400, loss[loss=0.2087, simple_loss=0.2795, pruned_loss=0.069, over 16746.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2653, pruned_loss=0.04955, over 2875095.87 frames. ], batch size: 134, lr: 4.61e-03, grad_scale: 4.0 2023-04-30 02:15:17,785 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.45 vs. limit=5.0 2023-04-30 02:15:31,898 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1871, 5.8339, 5.9213, 5.6715, 5.7102, 6.3013, 5.7902, 5.5114], device='cuda:6'), covar=tensor([0.0838, 0.1800, 0.1997, 0.2111, 0.2648, 0.0936, 0.1572, 0.2522], device='cuda:6'), in_proj_covar=tensor([0.0374, 0.0528, 0.0580, 0.0448, 0.0604, 0.0606, 0.0460, 0.0599], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 02:15:54,199 INFO [train.py:904] (6/8) Epoch 15, batch 450, loss[loss=0.1769, simple_loss=0.2678, pruned_loss=0.04305, over 17054.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2621, pruned_loss=0.04837, over 2975937.73 frames. ], batch size: 55, lr: 4.61e-03, grad_scale: 4.0 2023-04-30 02:16:23,521 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142572.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:16:24,749 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=142573.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:16:41,565 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142584.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:16:42,926 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2269, 3.6633, 3.9097, 2.1411, 3.0665, 2.6290, 3.6594, 3.7341], device='cuda:6'), covar=tensor([0.0312, 0.0838, 0.0484, 0.1937, 0.0843, 0.0890, 0.0657, 0.1042], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0149, 0.0160, 0.0147, 0.0138, 0.0126, 0.0138, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 02:17:03,317 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.083e+02 2.627e+02 3.131e+02 6.454e+02, threshold=5.253e+02, percent-clipped=1.0 2023-04-30 02:17:05,222 INFO [train.py:904] (6/8) Epoch 15, batch 500, loss[loss=0.1783, simple_loss=0.2517, pruned_loss=0.05243, over 16823.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2606, pruned_loss=0.04801, over 3045826.32 frames. ], batch size: 124, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:17:28,780 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=142620.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:17:28,963 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3452, 5.3716, 5.1852, 4.6292, 5.1540, 2.2115, 4.9879, 5.1385], device='cuda:6'), covar=tensor([0.0071, 0.0061, 0.0153, 0.0324, 0.0083, 0.2176, 0.0113, 0.0161], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0131, 0.0176, 0.0160, 0.0149, 0.0192, 0.0164, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:17:46,434 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=142632.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:17:49,291 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=142634.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:18:13,799 INFO [train.py:904] (6/8) Epoch 15, batch 550, loss[loss=0.1605, simple_loss=0.2431, pruned_loss=0.0389, over 15830.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2585, pruned_loss=0.04658, over 3106655.36 frames. ], batch size: 35, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:19:22,092 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.274e+02 2.667e+02 3.434e+02 1.165e+03, threshold=5.335e+02, percent-clipped=9.0 2023-04-30 02:19:23,200 INFO [train.py:904] (6/8) Epoch 15, batch 600, loss[loss=0.2085, simple_loss=0.2707, pruned_loss=0.07316, over 16816.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2592, pruned_loss=0.04714, over 3152462.74 frames. ], batch size: 124, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:19:53,077 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=142724.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:20:32,859 INFO [train.py:904] (6/8) Epoch 15, batch 650, loss[loss=0.1871, simple_loss=0.2748, pruned_loss=0.04969, over 16697.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2583, pruned_loss=0.0469, over 3184983.55 frames. ], batch size: 57, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:21:17,860 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=142785.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:21:40,890 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.547e+02 2.460e+02 3.000e+02 3.809e+02 2.810e+03, threshold=6.001e+02, percent-clipped=15.0 2023-04-30 02:21:42,123 INFO [train.py:904] (6/8) Epoch 15, batch 700, loss[loss=0.1546, simple_loss=0.2483, pruned_loss=0.03047, over 17111.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2577, pruned_loss=0.04616, over 3224748.68 frames. ], batch size: 49, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:22:43,314 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-04-30 02:22:50,283 INFO [train.py:904] (6/8) Epoch 15, batch 750, loss[loss=0.1792, simple_loss=0.2566, pruned_loss=0.05087, over 16732.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2571, pruned_loss=0.04571, over 3249222.69 frames. ], batch size: 89, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:22:50,748 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3627, 4.2021, 4.3946, 4.5752, 4.6912, 4.2355, 4.4518, 4.6876], device='cuda:6'), covar=tensor([0.1652, 0.1217, 0.1312, 0.0663, 0.0600, 0.1202, 0.2290, 0.0591], device='cuda:6'), in_proj_covar=tensor([0.0589, 0.0726, 0.0863, 0.0742, 0.0557, 0.0577, 0.0589, 0.0689], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:23:28,778 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 02:23:57,713 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.460e+02 2.171e+02 2.529e+02 2.994e+02 6.163e+02, threshold=5.058e+02, percent-clipped=1.0 2023-04-30 02:23:59,533 INFO [train.py:904] (6/8) Epoch 15, batch 800, loss[loss=0.1758, simple_loss=0.2616, pruned_loss=0.045, over 16424.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2564, pruned_loss=0.04543, over 3261345.33 frames. ], batch size: 68, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:24:36,438 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=142929.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:25:08,640 INFO [train.py:904] (6/8) Epoch 15, batch 850, loss[loss=0.1965, simple_loss=0.2695, pruned_loss=0.06177, over 16870.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2562, pruned_loss=0.04487, over 3274625.18 frames. ], batch size: 116, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:25:22,753 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-30 02:25:44,594 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-30 02:26:15,121 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.443e+02 2.170e+02 2.499e+02 2.974e+02 6.190e+02, threshold=4.998e+02, percent-clipped=2.0 2023-04-30 02:26:16,301 INFO [train.py:904] (6/8) Epoch 15, batch 900, loss[loss=0.1398, simple_loss=0.2218, pruned_loss=0.02889, over 16943.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2554, pruned_loss=0.04439, over 3284542.38 frames. ], batch size: 41, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:27:24,767 INFO [train.py:904] (6/8) Epoch 15, batch 950, loss[loss=0.1722, simple_loss=0.2631, pruned_loss=0.04064, over 16674.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2559, pruned_loss=0.04448, over 3289888.59 frames. ], batch size: 62, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:27:44,690 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-30 02:28:02,443 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=143080.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:28:30,229 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.522e+02 2.127e+02 2.534e+02 2.926e+02 6.813e+02, threshold=5.068e+02, percent-clipped=2.0 2023-04-30 02:28:31,477 INFO [train.py:904] (6/8) Epoch 15, batch 1000, loss[loss=0.1605, simple_loss=0.2558, pruned_loss=0.03266, over 17105.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2552, pruned_loss=0.04445, over 3288702.00 frames. ], batch size: 49, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:29:04,676 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8076, 3.7283, 3.8852, 3.6871, 3.8081, 4.2703, 3.9058, 3.5639], device='cuda:6'), covar=tensor([0.1912, 0.2074, 0.2143, 0.2532, 0.2878, 0.1867, 0.1616, 0.2874], device='cuda:6'), in_proj_covar=tensor([0.0385, 0.0545, 0.0603, 0.0465, 0.0624, 0.0627, 0.0477, 0.0618], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 02:29:41,403 INFO [train.py:904] (6/8) Epoch 15, batch 1050, loss[loss=0.1741, simple_loss=0.2474, pruned_loss=0.0504, over 16731.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2547, pruned_loss=0.04451, over 3293114.44 frames. ], batch size: 89, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:29:46,073 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-30 02:29:55,164 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8577, 2.8722, 2.5541, 4.5041, 3.6715, 4.2547, 1.7281, 3.0885], device='cuda:6'), covar=tensor([0.1297, 0.0700, 0.1149, 0.0214, 0.0262, 0.0459, 0.1483, 0.0786], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0165, 0.0187, 0.0167, 0.0197, 0.0214, 0.0191, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 02:30:46,963 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.748e+02 2.203e+02 2.568e+02 3.063e+02 1.237e+03, threshold=5.135e+02, percent-clipped=3.0 2023-04-30 02:30:49,005 INFO [train.py:904] (6/8) Epoch 15, batch 1100, loss[loss=0.1928, simple_loss=0.2779, pruned_loss=0.05379, over 17041.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2545, pruned_loss=0.04447, over 3294466.38 frames. ], batch size: 53, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:31:26,695 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=143229.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:31:51,229 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-30 02:31:58,384 INFO [train.py:904] (6/8) Epoch 15, batch 1150, loss[loss=0.1695, simple_loss=0.2477, pruned_loss=0.04561, over 11924.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2548, pruned_loss=0.04435, over 3291531.38 frames. ], batch size: 246, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:32:34,546 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=143277.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:33:07,199 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.631e+02 2.162e+02 2.475e+02 3.140e+02 4.983e+02, threshold=4.950e+02, percent-clipped=0.0 2023-04-30 02:33:08,293 INFO [train.py:904] (6/8) Epoch 15, batch 1200, loss[loss=0.1544, simple_loss=0.2425, pruned_loss=0.03318, over 16856.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2539, pruned_loss=0.04383, over 3296743.17 frames. ], batch size: 42, lr: 4.60e-03, grad_scale: 8.0 2023-04-30 02:33:44,546 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8094, 4.2789, 3.1329, 2.2108, 2.7593, 2.5728, 4.5495, 3.6552], device='cuda:6'), covar=tensor([0.2741, 0.0638, 0.1633, 0.2653, 0.2660, 0.1880, 0.0400, 0.1290], device='cuda:6'), in_proj_covar=tensor([0.0310, 0.0260, 0.0291, 0.0289, 0.0279, 0.0235, 0.0275, 0.0311], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 02:34:16,291 INFO [train.py:904] (6/8) Epoch 15, batch 1250, loss[loss=0.1876, simple_loss=0.2527, pruned_loss=0.06126, over 16900.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2542, pruned_loss=0.04427, over 3297128.23 frames. ], batch size: 109, lr: 4.60e-03, grad_scale: 8.0 2023-04-30 02:34:30,478 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5947, 4.4075, 4.6071, 4.7997, 4.9223, 4.4333, 4.7722, 4.9194], device='cuda:6'), covar=tensor([0.1470, 0.1207, 0.1398, 0.0661, 0.0589, 0.1116, 0.1480, 0.0644], device='cuda:6'), in_proj_covar=tensor([0.0594, 0.0730, 0.0874, 0.0746, 0.0559, 0.0580, 0.0591, 0.0696], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:34:57,384 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=143380.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:35:25,451 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=143400.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 02:35:26,163 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.199e+02 2.560e+02 2.965e+02 4.851e+02, threshold=5.120e+02, percent-clipped=0.0 2023-04-30 02:35:27,818 INFO [train.py:904] (6/8) Epoch 15, batch 1300, loss[loss=0.1847, simple_loss=0.2575, pruned_loss=0.05601, over 16774.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2539, pruned_loss=0.04437, over 3303289.49 frames. ], batch size: 83, lr: 4.60e-03, grad_scale: 8.0 2023-04-30 02:36:03,858 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=143428.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:36:13,522 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-30 02:36:37,208 INFO [train.py:904] (6/8) Epoch 15, batch 1350, loss[loss=0.1726, simple_loss=0.2702, pruned_loss=0.03754, over 17277.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2539, pruned_loss=0.04419, over 3293758.62 frames. ], batch size: 52, lr: 4.60e-03, grad_scale: 8.0 2023-04-30 02:36:50,716 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=143461.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 02:36:59,394 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0372, 2.9888, 2.6814, 4.7878, 3.8363, 4.3834, 1.7341, 3.2188], device='cuda:6'), covar=tensor([0.1215, 0.0693, 0.1132, 0.0184, 0.0321, 0.0413, 0.1466, 0.0716], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0165, 0.0186, 0.0167, 0.0197, 0.0213, 0.0190, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 02:37:06,585 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=143473.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 02:37:45,742 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.631e+02 2.266e+02 2.683e+02 3.217e+02 5.542e+02, threshold=5.366e+02, percent-clipped=1.0 2023-04-30 02:37:47,548 INFO [train.py:904] (6/8) Epoch 15, batch 1400, loss[loss=0.1875, simple_loss=0.2574, pruned_loss=0.05878, over 16696.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2539, pruned_loss=0.04406, over 3297187.55 frames. ], batch size: 134, lr: 4.59e-03, grad_scale: 8.0 2023-04-30 02:38:32,159 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=143534.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 02:38:55,992 INFO [train.py:904] (6/8) Epoch 15, batch 1450, loss[loss=0.1718, simple_loss=0.242, pruned_loss=0.05074, over 16852.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2529, pruned_loss=0.04393, over 3310429.21 frames. ], batch size: 116, lr: 4.59e-03, grad_scale: 8.0 2023-04-30 02:40:05,552 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.183e+02 2.517e+02 3.188e+02 7.484e+02, threshold=5.034e+02, percent-clipped=1.0 2023-04-30 02:40:06,726 INFO [train.py:904] (6/8) Epoch 15, batch 1500, loss[loss=0.1757, simple_loss=0.2724, pruned_loss=0.0395, over 17118.00 frames. ], tot_loss[loss=0.171, simple_loss=0.253, pruned_loss=0.04446, over 3311306.63 frames. ], batch size: 49, lr: 4.59e-03, grad_scale: 8.0 2023-04-30 02:40:18,181 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.45 vs. limit=5.0 2023-04-30 02:40:19,427 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4089, 2.2390, 2.2992, 4.3688, 2.2080, 2.6787, 2.3683, 2.5015], device='cuda:6'), covar=tensor([0.1118, 0.3485, 0.2709, 0.0428, 0.4005, 0.2406, 0.3265, 0.3329], device='cuda:6'), in_proj_covar=tensor([0.0381, 0.0417, 0.0349, 0.0328, 0.0425, 0.0481, 0.0382, 0.0489], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:40:41,154 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.59 vs. limit=5.0 2023-04-30 02:41:02,444 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2433, 4.2546, 4.6485, 2.3327, 4.8451, 4.8762, 3.3402, 3.9957], device='cuda:6'), covar=tensor([0.0605, 0.0174, 0.0148, 0.1057, 0.0052, 0.0113, 0.0406, 0.0274], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0104, 0.0091, 0.0140, 0.0073, 0.0117, 0.0126, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 02:41:14,547 INFO [train.py:904] (6/8) Epoch 15, batch 1550, loss[loss=0.202, simple_loss=0.2759, pruned_loss=0.06403, over 16396.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2542, pruned_loss=0.04492, over 3323367.88 frames. ], batch size: 146, lr: 4.59e-03, grad_scale: 8.0 2023-04-30 02:42:22,872 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.352e+02 2.733e+02 3.255e+02 5.583e+02, threshold=5.465e+02, percent-clipped=1.0 2023-04-30 02:42:24,070 INFO [train.py:904] (6/8) Epoch 15, batch 1600, loss[loss=0.1615, simple_loss=0.2452, pruned_loss=0.03887, over 16818.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2559, pruned_loss=0.04478, over 3327566.02 frames. ], batch size: 39, lr: 4.59e-03, grad_scale: 8.0 2023-04-30 02:42:34,176 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-04-30 02:43:31,068 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6128, 2.2475, 2.3046, 4.4213, 2.2983, 2.6538, 2.3836, 2.4403], device='cuda:6'), covar=tensor([0.1047, 0.3517, 0.2705, 0.0436, 0.3953, 0.2503, 0.3229, 0.3613], device='cuda:6'), in_proj_covar=tensor([0.0379, 0.0415, 0.0348, 0.0327, 0.0423, 0.0480, 0.0381, 0.0487], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:43:35,419 INFO [train.py:904] (6/8) Epoch 15, batch 1650, loss[loss=0.1783, simple_loss=0.2779, pruned_loss=0.03937, over 17190.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2587, pruned_loss=0.04584, over 3315763.59 frames. ], batch size: 46, lr: 4.59e-03, grad_scale: 8.0 2023-04-30 02:43:40,915 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=143756.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 02:43:45,486 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6559, 3.7318, 3.9864, 2.7622, 3.5522, 4.0487, 3.7353, 2.1928], device='cuda:6'), covar=tensor([0.0406, 0.0222, 0.0045, 0.0322, 0.0097, 0.0083, 0.0072, 0.0443], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0075, 0.0074, 0.0130, 0.0088, 0.0098, 0.0086, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 02:44:21,332 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2784, 5.2857, 5.0317, 4.5279, 5.1057, 2.1170, 4.8496, 5.0576], device='cuda:6'), covar=tensor([0.0074, 0.0069, 0.0168, 0.0370, 0.0086, 0.2291, 0.0134, 0.0172], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0137, 0.0184, 0.0169, 0.0156, 0.0197, 0.0172, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:44:23,104 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0626, 4.8087, 5.0956, 5.2881, 5.4794, 4.7259, 5.3919, 5.4518], device='cuda:6'), covar=tensor([0.1689, 0.1490, 0.1847, 0.0795, 0.0637, 0.0909, 0.0588, 0.0608], device='cuda:6'), in_proj_covar=tensor([0.0606, 0.0747, 0.0895, 0.0768, 0.0572, 0.0595, 0.0602, 0.0711], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:44:46,124 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.543e+02 2.448e+02 2.826e+02 3.842e+02 9.668e+02, threshold=5.653e+02, percent-clipped=8.0 2023-04-30 02:44:46,139 INFO [train.py:904] (6/8) Epoch 15, batch 1700, loss[loss=0.2086, simple_loss=0.2831, pruned_loss=0.06704, over 16239.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2595, pruned_loss=0.04588, over 3319798.60 frames. ], batch size: 165, lr: 4.59e-03, grad_scale: 4.0 2023-04-30 02:45:22,463 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=143829.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 02:45:45,778 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 02:45:54,335 INFO [train.py:904] (6/8) Epoch 15, batch 1750, loss[loss=0.1869, simple_loss=0.2735, pruned_loss=0.05011, over 16738.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2609, pruned_loss=0.04626, over 3324201.65 frames. ], batch size: 62, lr: 4.59e-03, grad_scale: 4.0 2023-04-30 02:46:00,491 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-30 02:46:45,497 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8344, 3.1483, 2.8747, 5.0636, 4.2195, 4.6037, 1.8258, 3.1691], device='cuda:6'), covar=tensor([0.1311, 0.0694, 0.1079, 0.0221, 0.0276, 0.0366, 0.1455, 0.0786], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0163, 0.0183, 0.0166, 0.0196, 0.0211, 0.0187, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 02:46:52,044 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6080, 3.6018, 3.8917, 2.1613, 3.9759, 3.9315, 3.1651, 3.0655], device='cuda:6'), covar=tensor([0.0691, 0.0179, 0.0136, 0.0959, 0.0074, 0.0177, 0.0326, 0.0381], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0106, 0.0092, 0.0142, 0.0074, 0.0119, 0.0127, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 02:47:05,601 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.250e+02 2.638e+02 3.028e+02 6.204e+02, threshold=5.275e+02, percent-clipped=1.0 2023-04-30 02:47:05,616 INFO [train.py:904] (6/8) Epoch 15, batch 1800, loss[loss=0.2201, simple_loss=0.3006, pruned_loss=0.06979, over 15586.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2628, pruned_loss=0.04706, over 3309945.40 frames. ], batch size: 191, lr: 4.59e-03, grad_scale: 4.0 2023-04-30 02:47:23,566 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 02:48:15,536 INFO [train.py:904] (6/8) Epoch 15, batch 1850, loss[loss=0.1736, simple_loss=0.2628, pruned_loss=0.04218, over 16486.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2626, pruned_loss=0.04668, over 3321019.62 frames. ], batch size: 75, lr: 4.59e-03, grad_scale: 2.0 2023-04-30 02:48:34,918 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4474, 5.2909, 5.1805, 4.6581, 4.7833, 5.2275, 5.1794, 4.8624], device='cuda:6'), covar=tensor([0.0487, 0.0461, 0.0291, 0.0315, 0.1112, 0.0458, 0.0269, 0.0711], device='cuda:6'), in_proj_covar=tensor([0.0275, 0.0378, 0.0328, 0.0309, 0.0343, 0.0358, 0.0224, 0.0388], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:49:04,440 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6708, 2.9512, 2.4377, 4.6889, 3.5720, 4.3745, 1.5571, 2.8244], device='cuda:6'), covar=tensor([0.1694, 0.0865, 0.1425, 0.0234, 0.0355, 0.0391, 0.1933, 0.0973], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0164, 0.0185, 0.0167, 0.0198, 0.0212, 0.0188, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 02:49:30,761 INFO [train.py:904] (6/8) Epoch 15, batch 1900, loss[loss=0.1812, simple_loss=0.2583, pruned_loss=0.05201, over 16811.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2613, pruned_loss=0.0461, over 3325857.66 frames. ], batch size: 96, lr: 4.59e-03, grad_scale: 2.0 2023-04-30 02:49:31,842 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 2.211e+02 2.636e+02 2.995e+02 6.158e+02, threshold=5.272e+02, percent-clipped=2.0 2023-04-30 02:50:39,887 INFO [train.py:904] (6/8) Epoch 15, batch 1950, loss[loss=0.1591, simple_loss=0.2474, pruned_loss=0.03543, over 17152.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2609, pruned_loss=0.04546, over 3333466.43 frames. ], batch size: 47, lr: 4.59e-03, grad_scale: 2.0 2023-04-30 02:50:46,707 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=144056.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 02:51:44,723 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6208, 4.9792, 4.7402, 4.7798, 4.4606, 4.4148, 4.4383, 5.0761], device='cuda:6'), covar=tensor([0.1235, 0.0904, 0.1079, 0.0790, 0.0926, 0.1077, 0.1217, 0.0875], device='cuda:6'), in_proj_covar=tensor([0.0618, 0.0768, 0.0632, 0.0551, 0.0486, 0.0492, 0.0642, 0.0582], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:51:49,541 INFO [train.py:904] (6/8) Epoch 15, batch 2000, loss[loss=0.1908, simple_loss=0.2756, pruned_loss=0.05294, over 16469.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.26, pruned_loss=0.04472, over 3330920.66 frames. ], batch size: 146, lr: 4.59e-03, grad_scale: 4.0 2023-04-30 02:51:51,363 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.196e+02 2.584e+02 2.966e+02 4.475e+02, threshold=5.169e+02, percent-clipped=0.0 2023-04-30 02:51:52,651 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=144104.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 02:52:27,340 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=144129.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 02:52:58,028 INFO [train.py:904] (6/8) Epoch 15, batch 2050, loss[loss=0.1529, simple_loss=0.236, pruned_loss=0.03491, over 16778.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2609, pruned_loss=0.04539, over 3333152.19 frames. ], batch size: 39, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 02:53:32,909 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=144177.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 02:53:54,744 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2346, 3.4905, 3.6130, 1.7237, 3.7806, 3.8305, 3.0409, 2.7172], device='cuda:6'), covar=tensor([0.1068, 0.0185, 0.0237, 0.1316, 0.0106, 0.0220, 0.0410, 0.0584], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0106, 0.0092, 0.0141, 0.0074, 0.0118, 0.0126, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 02:54:07,933 INFO [train.py:904] (6/8) Epoch 15, batch 2100, loss[loss=0.1628, simple_loss=0.2516, pruned_loss=0.037, over 17231.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.2626, pruned_loss=0.04637, over 3322035.88 frames. ], batch size: 46, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 02:54:08,980 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.706e+02 2.513e+02 2.931e+02 3.819e+02 1.829e+03, threshold=5.862e+02, percent-clipped=10.0 2023-04-30 02:54:44,697 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7270, 3.8010, 2.3694, 4.1990, 2.7714, 4.1461, 2.3747, 2.9893], device='cuda:6'), covar=tensor([0.0239, 0.0304, 0.1432, 0.0241, 0.0778, 0.0539, 0.1346, 0.0704], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0171, 0.0192, 0.0150, 0.0171, 0.0213, 0.0200, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 02:55:17,937 INFO [train.py:904] (6/8) Epoch 15, batch 2150, loss[loss=0.1861, simple_loss=0.2618, pruned_loss=0.0552, over 16506.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2642, pruned_loss=0.047, over 3311138.74 frames. ], batch size: 146, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 02:55:27,738 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2713, 3.6831, 3.7149, 2.1009, 3.0778, 2.5601, 3.6824, 3.8442], device='cuda:6'), covar=tensor([0.0296, 0.0752, 0.0505, 0.1837, 0.0771, 0.0896, 0.0641, 0.0886], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0154, 0.0162, 0.0148, 0.0140, 0.0126, 0.0139, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 02:56:25,351 INFO [train.py:904] (6/8) Epoch 15, batch 2200, loss[loss=0.1818, simple_loss=0.2542, pruned_loss=0.0547, over 16822.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2646, pruned_loss=0.0475, over 3311006.80 frames. ], batch size: 83, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 02:56:27,073 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.322e+02 2.712e+02 3.377e+02 6.214e+02, threshold=5.423e+02, percent-clipped=1.0 2023-04-30 02:57:05,566 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5236, 3.8201, 4.1562, 2.3256, 3.2659, 2.6804, 4.0267, 3.9835], device='cuda:6'), covar=tensor([0.0260, 0.0791, 0.0406, 0.1698, 0.0706, 0.0827, 0.0545, 0.0950], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0154, 0.0162, 0.0147, 0.0140, 0.0126, 0.0139, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 02:57:36,224 INFO [train.py:904] (6/8) Epoch 15, batch 2250, loss[loss=0.2451, simple_loss=0.3191, pruned_loss=0.08558, over 11722.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2652, pruned_loss=0.04837, over 3304826.41 frames. ], batch size: 246, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 02:58:46,690 INFO [train.py:904] (6/8) Epoch 15, batch 2300, loss[loss=0.2041, simple_loss=0.2927, pruned_loss=0.05773, over 16815.00 frames. ], tot_loss[loss=0.1809, simple_loss=0.265, pruned_loss=0.04838, over 3308517.93 frames. ], batch size: 90, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 02:58:47,877 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.350e+02 2.907e+02 3.506e+02 6.150e+02, threshold=5.814e+02, percent-clipped=3.0 2023-04-30 02:58:59,063 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=144411.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:59:45,613 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4938, 5.8268, 5.6179, 5.6843, 5.2244, 5.2079, 5.2288, 5.9594], device='cuda:6'), covar=tensor([0.1412, 0.1101, 0.1112, 0.0761, 0.1078, 0.0798, 0.1234, 0.1059], device='cuda:6'), in_proj_covar=tensor([0.0623, 0.0776, 0.0632, 0.0553, 0.0488, 0.0494, 0.0643, 0.0587], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 02:59:53,214 INFO [train.py:904] (6/8) Epoch 15, batch 2350, loss[loss=0.2096, simple_loss=0.2895, pruned_loss=0.06485, over 16062.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2655, pruned_loss=0.04848, over 3310861.55 frames. ], batch size: 164, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 03:00:20,666 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=144472.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:01:02,761 INFO [train.py:904] (6/8) Epoch 15, batch 2400, loss[loss=0.159, simple_loss=0.247, pruned_loss=0.03548, over 16744.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2667, pruned_loss=0.04865, over 3305855.80 frames. ], batch size: 39, lr: 4.58e-03, grad_scale: 8.0 2023-04-30 03:01:04,723 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.656e+02 2.391e+02 2.805e+02 3.317e+02 7.772e+02, threshold=5.609e+02, percent-clipped=1.0 2023-04-30 03:02:10,324 INFO [train.py:904] (6/8) Epoch 15, batch 2450, loss[loss=0.1504, simple_loss=0.2436, pruned_loss=0.02864, over 17210.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.2666, pruned_loss=0.04822, over 3312021.61 frames. ], batch size: 44, lr: 4.58e-03, grad_scale: 8.0 2023-04-30 03:02:26,658 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9068, 3.1197, 3.1180, 1.9673, 2.6497, 2.3247, 3.3023, 3.3690], device='cuda:6'), covar=tensor([0.0258, 0.0919, 0.0633, 0.1841, 0.0938, 0.0950, 0.0611, 0.0857], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0156, 0.0164, 0.0148, 0.0141, 0.0127, 0.0141, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 03:03:17,678 INFO [train.py:904] (6/8) Epoch 15, batch 2500, loss[loss=0.2243, simple_loss=0.3, pruned_loss=0.07436, over 15536.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2664, pruned_loss=0.04792, over 3313047.03 frames. ], batch size: 191, lr: 4.58e-03, grad_scale: 8.0 2023-04-30 03:03:18,674 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.258e+02 2.678e+02 3.424e+02 5.626e+02, threshold=5.355e+02, percent-clipped=1.0 2023-04-30 03:04:26,856 INFO [train.py:904] (6/8) Epoch 15, batch 2550, loss[loss=0.1771, simple_loss=0.2495, pruned_loss=0.05236, over 16736.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2672, pruned_loss=0.0481, over 3319989.93 frames. ], batch size: 89, lr: 4.58e-03, grad_scale: 8.0 2023-04-30 03:05:06,343 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-30 03:05:09,766 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-30 03:05:12,612 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5732, 3.8156, 4.1593, 2.4380, 3.3221, 2.7205, 4.0856, 4.1193], device='cuda:6'), covar=tensor([0.0260, 0.0837, 0.0439, 0.1679, 0.0757, 0.0882, 0.0529, 0.0946], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0155, 0.0163, 0.0147, 0.0140, 0.0126, 0.0140, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 03:05:34,904 INFO [train.py:904] (6/8) Epoch 15, batch 2600, loss[loss=0.1727, simple_loss=0.2596, pruned_loss=0.04292, over 16401.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2672, pruned_loss=0.04815, over 3301051.49 frames. ], batch size: 146, lr: 4.58e-03, grad_scale: 8.0 2023-04-30 03:05:36,053 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.335e+02 2.561e+02 3.164e+02 7.288e+02, threshold=5.122e+02, percent-clipped=2.0 2023-04-30 03:06:08,914 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2964, 3.6549, 3.8063, 2.1794, 3.0096, 2.4942, 3.7691, 3.8555], device='cuda:6'), covar=tensor([0.0308, 0.0865, 0.0558, 0.1789, 0.0865, 0.0945, 0.0611, 0.0938], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0155, 0.0163, 0.0147, 0.0140, 0.0126, 0.0140, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 03:06:43,553 INFO [train.py:904] (6/8) Epoch 15, batch 2650, loss[loss=0.21, simple_loss=0.2858, pruned_loss=0.06708, over 16891.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2671, pruned_loss=0.04741, over 3302851.14 frames. ], batch size: 109, lr: 4.58e-03, grad_scale: 8.0 2023-04-30 03:06:54,380 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0034, 4.2554, 3.2311, 2.3017, 2.9012, 2.7189, 4.6557, 3.7449], device='cuda:6'), covar=tensor([0.2605, 0.0673, 0.1728, 0.2775, 0.2761, 0.1782, 0.0405, 0.1350], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0266, 0.0295, 0.0293, 0.0287, 0.0239, 0.0279, 0.0319], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 03:07:05,886 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=144767.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:07:14,213 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-04-30 03:07:53,565 INFO [train.py:904] (6/8) Epoch 15, batch 2700, loss[loss=0.1744, simple_loss=0.2738, pruned_loss=0.03754, over 16748.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2667, pruned_loss=0.04649, over 3309270.40 frames. ], batch size: 62, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:07:54,729 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.168e+02 2.529e+02 3.023e+02 4.642e+02, threshold=5.059e+02, percent-clipped=0.0 2023-04-30 03:08:19,319 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1928, 5.8887, 6.0203, 5.6744, 5.7139, 6.3259, 5.7617, 5.4796], device='cuda:6'), covar=tensor([0.1085, 0.1539, 0.2054, 0.2071, 0.2645, 0.1038, 0.1515, 0.2321], device='cuda:6'), in_proj_covar=tensor([0.0392, 0.0558, 0.0612, 0.0473, 0.0636, 0.0640, 0.0485, 0.0624], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 03:08:53,777 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-04-30 03:09:02,446 INFO [train.py:904] (6/8) Epoch 15, batch 2750, loss[loss=0.158, simple_loss=0.2458, pruned_loss=0.0351, over 17204.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2666, pruned_loss=0.04604, over 3316484.91 frames. ], batch size: 45, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:10:11,018 INFO [train.py:904] (6/8) Epoch 15, batch 2800, loss[loss=0.1899, simple_loss=0.285, pruned_loss=0.04737, over 17263.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2664, pruned_loss=0.04576, over 3314396.31 frames. ], batch size: 52, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:10:12,144 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.181e+02 2.488e+02 3.014e+02 5.995e+02, threshold=4.976e+02, percent-clipped=2.0 2023-04-30 03:10:35,335 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1360, 5.1154, 4.9067, 4.1352, 4.9540, 1.7566, 4.7436, 4.7954], device='cuda:6'), covar=tensor([0.0098, 0.0088, 0.0195, 0.0472, 0.0118, 0.2853, 0.0140, 0.0221], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0139, 0.0188, 0.0173, 0.0158, 0.0198, 0.0175, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:10:47,665 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8701, 1.8990, 2.3495, 2.8583, 2.6186, 3.2510, 2.0644, 3.1461], device='cuda:6'), covar=tensor([0.0221, 0.0445, 0.0339, 0.0252, 0.0291, 0.0157, 0.0454, 0.0133], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0184, 0.0170, 0.0173, 0.0183, 0.0139, 0.0184, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:11:21,052 INFO [train.py:904] (6/8) Epoch 15, batch 2850, loss[loss=0.1638, simple_loss=0.2417, pruned_loss=0.04298, over 16819.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2662, pruned_loss=0.04565, over 3316653.23 frames. ], batch size: 102, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:11:28,531 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-30 03:12:22,175 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=144996.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:12:31,830 INFO [train.py:904] (6/8) Epoch 15, batch 2900, loss[loss=0.189, simple_loss=0.2559, pruned_loss=0.06109, over 15384.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2645, pruned_loss=0.04564, over 3320027.79 frames. ], batch size: 191, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:12:33,008 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.445e+02 2.844e+02 3.300e+02 6.709e+02, threshold=5.687e+02, percent-clipped=6.0 2023-04-30 03:12:49,248 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8105, 3.1050, 2.9369, 5.0124, 4.2106, 4.5753, 1.7570, 3.4190], device='cuda:6'), covar=tensor([0.1340, 0.0663, 0.1074, 0.0192, 0.0260, 0.0389, 0.1454, 0.0658], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0165, 0.0185, 0.0169, 0.0199, 0.0212, 0.0188, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 03:13:06,920 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145027.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:13:40,918 INFO [train.py:904] (6/8) Epoch 15, batch 2950, loss[loss=0.1769, simple_loss=0.2675, pruned_loss=0.0431, over 17091.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2632, pruned_loss=0.04557, over 3333151.61 frames. ], batch size: 49, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:13:47,697 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145057.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 03:14:01,042 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145066.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:14:02,155 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145067.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:14:31,923 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145088.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:14:49,683 INFO [train.py:904] (6/8) Epoch 15, batch 3000, loss[loss=0.1354, simple_loss=0.2193, pruned_loss=0.02578, over 16981.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.264, pruned_loss=0.04658, over 3337075.44 frames. ], batch size: 41, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:14:49,683 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 03:14:58,803 INFO [train.py:938] (6/8) Epoch 15, validation: loss=0.138, simple_loss=0.2438, pruned_loss=0.01616, over 944034.00 frames. 2023-04-30 03:14:58,804 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17572MB 2023-04-30 03:15:00,799 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.691e+02 2.402e+02 2.841e+02 3.286e+02 6.614e+02, threshold=5.681e+02, percent-clipped=1.0 2023-04-30 03:15:17,421 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=145115.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:15:34,612 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145127.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:15:37,295 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-30 03:16:07,543 INFO [train.py:904] (6/8) Epoch 15, batch 3050, loss[loss=0.191, simple_loss=0.2665, pruned_loss=0.05771, over 16805.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2637, pruned_loss=0.04656, over 3336452.01 frames. ], batch size: 124, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:17:18,167 INFO [train.py:904] (6/8) Epoch 15, batch 3100, loss[loss=0.2092, simple_loss=0.2898, pruned_loss=0.06424, over 16739.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2635, pruned_loss=0.04706, over 3333243.52 frames. ], batch size: 83, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:17:19,338 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.571e+02 2.445e+02 2.806e+02 3.389e+02 5.168e+02, threshold=5.611e+02, percent-clipped=0.0 2023-04-30 03:17:27,604 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5971, 4.9453, 4.6824, 4.7025, 4.4807, 4.4170, 4.3748, 4.9797], device='cuda:6'), covar=tensor([0.1095, 0.0858, 0.0965, 0.0795, 0.0792, 0.1181, 0.1183, 0.0843], device='cuda:6'), in_proj_covar=tensor([0.0629, 0.0779, 0.0640, 0.0559, 0.0495, 0.0497, 0.0649, 0.0593], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:17:31,425 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-30 03:18:16,647 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2394, 4.0461, 4.3271, 4.4585, 4.5553, 4.0979, 4.3125, 4.5349], device='cuda:6'), covar=tensor([0.1399, 0.1159, 0.1185, 0.0570, 0.0520, 0.1186, 0.1871, 0.0598], device='cuda:6'), in_proj_covar=tensor([0.0611, 0.0758, 0.0904, 0.0774, 0.0579, 0.0603, 0.0608, 0.0714], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:18:28,438 INFO [train.py:904] (6/8) Epoch 15, batch 3150, loss[loss=0.1981, simple_loss=0.2724, pruned_loss=0.06189, over 16721.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2632, pruned_loss=0.04712, over 3334177.54 frames. ], batch size: 124, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:18:44,288 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0807, 4.2288, 2.7405, 4.8787, 3.4209, 4.8396, 3.0674, 3.4781], device='cuda:6'), covar=tensor([0.0248, 0.0309, 0.1374, 0.0272, 0.0664, 0.0382, 0.1178, 0.0593], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0172, 0.0191, 0.0151, 0.0169, 0.0216, 0.0199, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 03:19:12,064 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1055, 4.7530, 5.0638, 5.2647, 5.5063, 4.7590, 5.4265, 5.4380], device='cuda:6'), covar=tensor([0.1591, 0.1384, 0.1903, 0.0800, 0.0517, 0.0871, 0.0517, 0.0555], device='cuda:6'), in_proj_covar=tensor([0.0615, 0.0762, 0.0911, 0.0779, 0.0583, 0.0608, 0.0613, 0.0719], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:19:37,246 INFO [train.py:904] (6/8) Epoch 15, batch 3200, loss[loss=0.1483, simple_loss=0.2378, pruned_loss=0.02936, over 16783.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2614, pruned_loss=0.04612, over 3332614.84 frames. ], batch size: 39, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:19:38,470 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.650e+02 2.241e+02 2.735e+02 3.234e+02 5.514e+02, threshold=5.469e+02, percent-clipped=0.0 2023-04-30 03:19:40,666 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4994, 2.5226, 2.2115, 2.4185, 2.9103, 2.6400, 3.2375, 3.1408], device='cuda:6'), covar=tensor([0.0114, 0.0356, 0.0423, 0.0362, 0.0229, 0.0320, 0.0240, 0.0222], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0222, 0.0211, 0.0213, 0.0222, 0.0222, 0.0229, 0.0216], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:20:46,507 INFO [train.py:904] (6/8) Epoch 15, batch 3250, loss[loss=0.211, simple_loss=0.2856, pruned_loss=0.06821, over 16695.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2615, pruned_loss=0.04634, over 3332483.74 frames. ], batch size: 134, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:20:46,772 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145352.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 03:21:00,102 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4274, 3.7091, 3.8774, 2.1304, 3.0280, 2.4466, 3.9447, 3.9130], device='cuda:6'), covar=tensor([0.0333, 0.0916, 0.0555, 0.2025, 0.0900, 0.1056, 0.0654, 0.1059], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0156, 0.0163, 0.0148, 0.0140, 0.0127, 0.0141, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 03:21:13,304 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145371.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:21:30,558 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145383.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:21:57,380 INFO [train.py:904] (6/8) Epoch 15, batch 3300, loss[loss=0.2024, simple_loss=0.2841, pruned_loss=0.06035, over 16908.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2617, pruned_loss=0.04633, over 3330356.12 frames. ], batch size: 109, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:21:58,627 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.241e+02 2.723e+02 3.266e+02 5.157e+02, threshold=5.447e+02, percent-clipped=0.0 2023-04-30 03:22:16,914 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-30 03:22:25,239 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145422.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:22:38,936 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145432.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:22:56,257 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-30 03:23:06,148 INFO [train.py:904] (6/8) Epoch 15, batch 3350, loss[loss=0.2014, simple_loss=0.2755, pruned_loss=0.06364, over 16740.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2634, pruned_loss=0.04726, over 3326600.99 frames. ], batch size: 134, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:24:17,515 INFO [train.py:904] (6/8) Epoch 15, batch 3400, loss[loss=0.1584, simple_loss=0.252, pruned_loss=0.03239, over 16636.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.264, pruned_loss=0.0471, over 3331727.96 frames. ], batch size: 62, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:24:18,607 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.624e+02 2.153e+02 2.635e+02 3.209e+02 5.771e+02, threshold=5.270e+02, percent-clipped=2.0 2023-04-30 03:24:53,586 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0828, 3.1936, 3.2414, 2.2233, 3.0830, 3.4283, 3.2280, 2.0046], device='cuda:6'), covar=tensor([0.0471, 0.0095, 0.0059, 0.0355, 0.0082, 0.0083, 0.0081, 0.0376], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0076, 0.0075, 0.0131, 0.0088, 0.0100, 0.0086, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 03:25:28,517 INFO [train.py:904] (6/8) Epoch 15, batch 3450, loss[loss=0.1937, simple_loss=0.273, pruned_loss=0.05723, over 16504.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2628, pruned_loss=0.04684, over 3312912.09 frames. ], batch size: 146, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:26:27,190 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145594.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:26:38,102 INFO [train.py:904] (6/8) Epoch 15, batch 3500, loss[loss=0.147, simple_loss=0.2325, pruned_loss=0.03069, over 15892.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.262, pruned_loss=0.04654, over 3312081.38 frames. ], batch size: 35, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:26:39,230 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.386e+02 2.268e+02 2.638e+02 3.199e+02 5.613e+02, threshold=5.276e+02, percent-clipped=1.0 2023-04-30 03:26:42,690 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2023-04-30 03:27:09,395 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9986, 4.1218, 4.3807, 4.3734, 4.3963, 4.1416, 4.1650, 4.0740], device='cuda:6'), covar=tensor([0.0346, 0.0537, 0.0374, 0.0407, 0.0489, 0.0423, 0.0773, 0.0558], device='cuda:6'), in_proj_covar=tensor([0.0381, 0.0408, 0.0403, 0.0386, 0.0455, 0.0426, 0.0524, 0.0340], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 03:27:35,951 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145643.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:27:47,267 INFO [train.py:904] (6/8) Epoch 15, batch 3550, loss[loss=0.2036, simple_loss=0.2839, pruned_loss=0.06165, over 16770.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.2608, pruned_loss=0.0459, over 3309391.10 frames. ], batch size: 134, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:27:48,252 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145652.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 03:27:53,473 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145655.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 03:28:27,643 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6635, 4.2451, 4.1601, 3.1930, 3.5954, 4.2354, 3.9373, 2.5196], device='cuda:6'), covar=tensor([0.0442, 0.0051, 0.0043, 0.0281, 0.0100, 0.0084, 0.0064, 0.0365], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0075, 0.0074, 0.0129, 0.0088, 0.0099, 0.0085, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 03:28:32,351 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145683.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:28:51,476 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9697, 2.0592, 2.4102, 2.9405, 2.6858, 3.3998, 2.2019, 3.2801], device='cuda:6'), covar=tensor([0.0196, 0.0411, 0.0305, 0.0269, 0.0287, 0.0162, 0.0408, 0.0151], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0187, 0.0171, 0.0176, 0.0185, 0.0142, 0.0187, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:28:54,823 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=145700.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:28:57,583 INFO [train.py:904] (6/8) Epoch 15, batch 3600, loss[loss=0.1542, simple_loss=0.2371, pruned_loss=0.03567, over 16995.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2596, pruned_loss=0.04485, over 3317043.89 frames. ], batch size: 41, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:28:58,725 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.177e+02 2.263e+02 2.507e+02 3.007e+02 5.256e+02, threshold=5.015e+02, percent-clipped=0.0 2023-04-30 03:29:00,407 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145704.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:29:26,973 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145722.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:29:34,304 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145727.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:29:39,859 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=145731.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:29:50,772 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.88 vs. limit=5.0 2023-04-30 03:30:10,701 INFO [train.py:904] (6/8) Epoch 15, batch 3650, loss[loss=0.181, simple_loss=0.2675, pruned_loss=0.04724, over 16559.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2576, pruned_loss=0.04496, over 3311230.63 frames. ], batch size: 62, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:30:38,222 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=145770.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:31:24,084 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 03:31:24,497 INFO [train.py:904] (6/8) Epoch 15, batch 3700, loss[loss=0.1828, simple_loss=0.2535, pruned_loss=0.0561, over 16392.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.2563, pruned_loss=0.04668, over 3300972.90 frames. ], batch size: 75, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:31:26,279 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.174e+02 2.701e+02 3.170e+02 5.249e+02, threshold=5.403e+02, percent-clipped=2.0 2023-04-30 03:31:53,387 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145821.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 03:31:55,917 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3706, 3.4748, 3.4887, 2.1415, 2.9931, 2.5362, 3.7460, 3.8327], device='cuda:6'), covar=tensor([0.0210, 0.0755, 0.0568, 0.1694, 0.0810, 0.0888, 0.0470, 0.0728], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0155, 0.0160, 0.0146, 0.0138, 0.0125, 0.0139, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 03:32:38,908 INFO [train.py:904] (6/8) Epoch 15, batch 3750, loss[loss=0.1728, simple_loss=0.2439, pruned_loss=0.05085, over 16913.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2572, pruned_loss=0.0484, over 3273520.01 frames. ], batch size: 116, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:32:50,841 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-30 03:33:23,529 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145882.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 03:33:51,792 INFO [train.py:904] (6/8) Epoch 15, batch 3800, loss[loss=0.1546, simple_loss=0.2353, pruned_loss=0.037, over 16801.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2579, pruned_loss=0.04948, over 3283195.92 frames. ], batch size: 102, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:33:53,680 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.221e+02 2.643e+02 3.115e+02 5.182e+02, threshold=5.286e+02, percent-clipped=0.0 2023-04-30 03:34:26,692 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4907, 3.8082, 3.9472, 2.8785, 3.6355, 4.0417, 3.7227, 2.0957], device='cuda:6'), covar=tensor([0.0444, 0.0135, 0.0044, 0.0306, 0.0077, 0.0085, 0.0077, 0.0436], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0075, 0.0074, 0.0129, 0.0088, 0.0099, 0.0086, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 03:35:01,658 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145950.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 03:35:04,377 INFO [train.py:904] (6/8) Epoch 15, batch 3850, loss[loss=0.1718, simple_loss=0.26, pruned_loss=0.04174, over 17125.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2583, pruned_loss=0.0505, over 3282920.08 frames. ], batch size: 48, lr: 4.56e-03, grad_scale: 16.0 2023-04-30 03:35:09,088 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2283, 4.1982, 4.1150, 3.9092, 3.8866, 4.1986, 3.8988, 3.9745], device='cuda:6'), covar=tensor([0.0621, 0.0612, 0.0274, 0.0261, 0.0700, 0.0439, 0.0792, 0.0559], device='cuda:6'), in_proj_covar=tensor([0.0278, 0.0382, 0.0328, 0.0311, 0.0344, 0.0357, 0.0221, 0.0389], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:36:13,420 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145999.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:36:21,005 INFO [train.py:904] (6/8) Epoch 15, batch 3900, loss[loss=0.1741, simple_loss=0.2582, pruned_loss=0.04496, over 16625.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2582, pruned_loss=0.05108, over 3265404.52 frames. ], batch size: 62, lr: 4.56e-03, grad_scale: 16.0 2023-04-30 03:36:22,203 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.721e+02 2.255e+02 2.647e+02 3.185e+02 6.041e+02, threshold=5.295e+02, percent-clipped=2.0 2023-04-30 03:36:25,908 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6024, 2.4076, 1.9400, 2.1196, 2.7504, 2.4800, 2.7534, 2.8558], device='cuda:6'), covar=tensor([0.0179, 0.0312, 0.0416, 0.0418, 0.0193, 0.0276, 0.0181, 0.0226], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0220, 0.0212, 0.0214, 0.0221, 0.0220, 0.0230, 0.0215], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:36:58,025 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146027.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:37:32,699 INFO [train.py:904] (6/8) Epoch 15, batch 3950, loss[loss=0.182, simple_loss=0.2497, pruned_loss=0.05715, over 16830.00 frames. ], tot_loss[loss=0.181, simple_loss=0.258, pruned_loss=0.05194, over 3273113.52 frames. ], batch size: 102, lr: 4.55e-03, grad_scale: 16.0 2023-04-30 03:38:06,970 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=146075.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:38:33,672 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2729, 3.2160, 3.4885, 1.8339, 3.5842, 3.6298, 2.9718, 2.6569], device='cuda:6'), covar=tensor([0.0777, 0.0210, 0.0162, 0.1133, 0.0092, 0.0166, 0.0380, 0.0482], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0106, 0.0091, 0.0139, 0.0074, 0.0118, 0.0124, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 03:38:46,132 INFO [train.py:904] (6/8) Epoch 15, batch 4000, loss[loss=0.1882, simple_loss=0.2687, pruned_loss=0.05381, over 16906.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.2584, pruned_loss=0.05224, over 3276351.68 frames. ], batch size: 116, lr: 4.55e-03, grad_scale: 16.0 2023-04-30 03:38:47,409 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 2.290e+02 2.701e+02 3.084e+02 7.730e+02, threshold=5.403e+02, percent-clipped=2.0 2023-04-30 03:38:54,860 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3246, 3.8137, 3.9236, 2.7811, 3.5419, 4.0431, 3.6064, 2.0671], device='cuda:6'), covar=tensor([0.0454, 0.0089, 0.0038, 0.0294, 0.0084, 0.0080, 0.0080, 0.0424], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0074, 0.0074, 0.0129, 0.0088, 0.0098, 0.0086, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 03:39:56,181 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146149.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:39:59,984 INFO [train.py:904] (6/8) Epoch 15, batch 4050, loss[loss=0.1867, simple_loss=0.2755, pruned_loss=0.04896, over 17170.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.259, pruned_loss=0.05088, over 3282935.62 frames. ], batch size: 46, lr: 4.55e-03, grad_scale: 16.0 2023-04-30 03:40:36,974 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146177.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 03:40:58,502 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3663, 5.3837, 5.2135, 4.8727, 4.8462, 5.2821, 5.1248, 5.0157], device='cuda:6'), covar=tensor([0.0525, 0.0319, 0.0225, 0.0239, 0.0894, 0.0337, 0.0261, 0.0519], device='cuda:6'), in_proj_covar=tensor([0.0278, 0.0379, 0.0327, 0.0310, 0.0344, 0.0357, 0.0220, 0.0388], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:41:13,945 INFO [train.py:904] (6/8) Epoch 15, batch 4100, loss[loss=0.2247, simple_loss=0.3052, pruned_loss=0.07208, over 11863.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.261, pruned_loss=0.05076, over 3259421.12 frames. ], batch size: 246, lr: 4.55e-03, grad_scale: 16.0 2023-04-30 03:41:15,750 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 1.975e+02 2.401e+02 2.875e+02 5.931e+02, threshold=4.803e+02, percent-clipped=1.0 2023-04-30 03:41:19,669 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-30 03:41:26,699 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146210.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:42:30,111 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146250.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:42:32,939 INFO [train.py:904] (6/8) Epoch 15, batch 4150, loss[loss=0.2286, simple_loss=0.3111, pruned_loss=0.07302, over 15409.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.268, pruned_loss=0.05281, over 3262901.91 frames. ], batch size: 191, lr: 4.55e-03, grad_scale: 8.0 2023-04-30 03:43:45,581 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=146298.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:43:46,707 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146299.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:43:50,403 INFO [train.py:904] (6/8) Epoch 15, batch 4200, loss[loss=0.247, simple_loss=0.3208, pruned_loss=0.08656, over 11414.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2751, pruned_loss=0.05468, over 3233023.32 frames. ], batch size: 247, lr: 4.55e-03, grad_scale: 8.0 2023-04-30 03:43:53,463 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.342e+02 2.800e+02 3.448e+02 4.997e+02, threshold=5.600e+02, percent-clipped=3.0 2023-04-30 03:44:21,165 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9139, 4.0094, 4.2507, 4.2679, 4.2710, 4.0309, 3.9627, 3.9689], device='cuda:6'), covar=tensor([0.0294, 0.0497, 0.0424, 0.0376, 0.0379, 0.0357, 0.1018, 0.0471], device='cuda:6'), in_proj_covar=tensor([0.0367, 0.0396, 0.0390, 0.0369, 0.0437, 0.0409, 0.0504, 0.0328], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 03:44:59,533 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=146347.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:45:06,384 INFO [train.py:904] (6/8) Epoch 15, batch 4250, loss[loss=0.1664, simple_loss=0.2692, pruned_loss=0.03178, over 16759.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2782, pruned_loss=0.05416, over 3220435.73 frames. ], batch size: 89, lr: 4.55e-03, grad_scale: 4.0 2023-04-30 03:45:25,017 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.50 vs. limit=2.0 2023-04-30 03:45:38,842 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8366, 5.2346, 4.7350, 5.1103, 4.7901, 4.5212, 4.7412, 5.2683], device='cuda:6'), covar=tensor([0.2222, 0.1493, 0.2237, 0.1217, 0.1472, 0.1602, 0.2332, 0.1600], device='cuda:6'), in_proj_covar=tensor([0.0610, 0.0761, 0.0627, 0.0548, 0.0479, 0.0488, 0.0635, 0.0578], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:46:19,357 INFO [train.py:904] (6/8) Epoch 15, batch 4300, loss[loss=0.2026, simple_loss=0.301, pruned_loss=0.0521, over 16736.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2793, pruned_loss=0.05334, over 3199677.03 frames. ], batch size: 124, lr: 4.55e-03, grad_scale: 4.0 2023-04-30 03:46:23,348 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.412e+02 2.971e+02 3.359e+02 7.082e+02, threshold=5.941e+02, percent-clipped=4.0 2023-04-30 03:46:40,759 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-04-30 03:46:45,306 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5033, 4.4276, 4.2641, 2.8623, 3.8348, 4.3674, 3.8778, 2.3619], device='cuda:6'), covar=tensor([0.0431, 0.0020, 0.0037, 0.0324, 0.0075, 0.0070, 0.0063, 0.0354], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0074, 0.0074, 0.0130, 0.0088, 0.0098, 0.0086, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 03:46:47,716 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8810, 4.8951, 4.6794, 4.0642, 4.8446, 1.7159, 4.5751, 4.4772], device='cuda:6'), covar=tensor([0.0060, 0.0042, 0.0137, 0.0280, 0.0054, 0.2675, 0.0083, 0.0179], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0137, 0.0187, 0.0173, 0.0158, 0.0197, 0.0175, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:47:08,203 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5081, 4.5184, 4.3949, 3.2218, 4.4846, 1.4865, 4.1334, 3.9967], device='cuda:6'), covar=tensor([0.0117, 0.0091, 0.0194, 0.0590, 0.0099, 0.3695, 0.0162, 0.0362], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0137, 0.0187, 0.0173, 0.0158, 0.0197, 0.0175, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:47:31,145 INFO [train.py:904] (6/8) Epoch 15, batch 4350, loss[loss=0.2024, simple_loss=0.2871, pruned_loss=0.05883, over 17029.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2828, pruned_loss=0.05425, over 3202134.35 frames. ], batch size: 55, lr: 4.55e-03, grad_scale: 4.0 2023-04-30 03:48:08,878 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146477.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 03:48:45,757 INFO [train.py:904] (6/8) Epoch 15, batch 4400, loss[loss=0.192, simple_loss=0.2825, pruned_loss=0.05078, over 16475.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.2846, pruned_loss=0.05526, over 3210710.06 frames. ], batch size: 68, lr: 4.55e-03, grad_scale: 8.0 2023-04-30 03:48:50,393 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.716e+02 2.574e+02 2.972e+02 3.574e+02 6.742e+02, threshold=5.944e+02, percent-clipped=2.0 2023-04-30 03:48:51,428 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146505.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:49:04,248 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9830, 5.2721, 5.0272, 5.0453, 4.7617, 4.6372, 4.7006, 5.3702], device='cuda:6'), covar=tensor([0.1101, 0.0769, 0.0900, 0.0750, 0.0726, 0.0975, 0.0979, 0.0759], device='cuda:6'), in_proj_covar=tensor([0.0606, 0.0755, 0.0621, 0.0545, 0.0476, 0.0485, 0.0628, 0.0573], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:49:07,717 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3413, 5.6240, 5.3463, 5.4483, 5.1185, 4.8932, 5.0916, 5.7238], device='cuda:6'), covar=tensor([0.0970, 0.0685, 0.0905, 0.0680, 0.0682, 0.0757, 0.0925, 0.0736], device='cuda:6'), in_proj_covar=tensor([0.0605, 0.0754, 0.0621, 0.0545, 0.0475, 0.0485, 0.0627, 0.0573], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:49:21,074 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=146525.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 03:49:58,633 INFO [train.py:904] (6/8) Epoch 15, batch 4450, loss[loss=0.1934, simple_loss=0.2903, pruned_loss=0.04827, over 16856.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2873, pruned_loss=0.05606, over 3226696.05 frames. ], batch size: 102, lr: 4.55e-03, grad_scale: 8.0 2023-04-30 03:50:18,375 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7081, 2.7070, 2.6258, 4.2335, 3.3425, 4.0665, 1.5740, 3.1198], device='cuda:6'), covar=tensor([0.1265, 0.0723, 0.1100, 0.0129, 0.0232, 0.0306, 0.1523, 0.0669], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0165, 0.0185, 0.0168, 0.0200, 0.0210, 0.0187, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 03:50:57,172 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146591.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:51:06,381 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 03:51:12,438 INFO [train.py:904] (6/8) Epoch 15, batch 4500, loss[loss=0.1959, simple_loss=0.2845, pruned_loss=0.05363, over 16483.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2877, pruned_loss=0.05693, over 3230517.11 frames. ], batch size: 68, lr: 4.55e-03, grad_scale: 8.0 2023-04-30 03:51:16,068 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 1.915e+02 2.182e+02 2.470e+02 4.715e+02, threshold=4.363e+02, percent-clipped=0.0 2023-04-30 03:51:47,941 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146626.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:51:57,277 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-30 03:52:25,473 INFO [train.py:904] (6/8) Epoch 15, batch 4550, loss[loss=0.2229, simple_loss=0.3059, pruned_loss=0.06994, over 16211.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2887, pruned_loss=0.05771, over 3246561.10 frames. ], batch size: 165, lr: 4.55e-03, grad_scale: 8.0 2023-04-30 03:52:25,978 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146652.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:53:16,682 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146687.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 03:53:37,403 INFO [train.py:904] (6/8) Epoch 15, batch 4600, loss[loss=0.1831, simple_loss=0.2728, pruned_loss=0.04673, over 16901.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2898, pruned_loss=0.05837, over 3235388.15 frames. ], batch size: 90, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 03:53:41,728 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.606e+02 1.940e+02 2.192e+02 2.697e+02 4.440e+02, threshold=4.384e+02, percent-clipped=1.0 2023-04-30 03:54:21,490 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8797, 3.2684, 3.3448, 1.8535, 2.8175, 2.2001, 3.1729, 3.3191], device='cuda:6'), covar=tensor([0.0292, 0.0708, 0.0539, 0.1989, 0.0827, 0.0949, 0.0800, 0.0987], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0154, 0.0161, 0.0147, 0.0138, 0.0126, 0.0139, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 03:54:36,102 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 03:54:49,172 INFO [train.py:904] (6/8) Epoch 15, batch 4650, loss[loss=0.1935, simple_loss=0.2826, pruned_loss=0.0522, over 16506.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.289, pruned_loss=0.05868, over 3223911.80 frames. ], batch size: 75, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 03:55:02,740 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-30 03:55:23,941 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-04-30 03:56:03,159 INFO [train.py:904] (6/8) Epoch 15, batch 4700, loss[loss=0.1878, simple_loss=0.276, pruned_loss=0.04978, over 16445.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2864, pruned_loss=0.05769, over 3209574.83 frames. ], batch size: 146, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 03:56:07,885 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 1.906e+02 2.324e+02 2.777e+02 7.777e+02, threshold=4.648e+02, percent-clipped=3.0 2023-04-30 03:56:08,854 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146805.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:57:07,622 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146844.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:57:18,582 INFO [train.py:904] (6/8) Epoch 15, batch 4750, loss[loss=0.2019, simple_loss=0.2902, pruned_loss=0.05675, over 15619.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2826, pruned_loss=0.05581, over 3192606.09 frames. ], batch size: 190, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 03:57:20,576 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=146853.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:57:36,473 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.57 vs. limit=5.0 2023-04-30 03:58:22,068 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3869, 2.1326, 1.8254, 1.9082, 2.4605, 2.1171, 2.1987, 2.5991], device='cuda:6'), covar=tensor([0.0147, 0.0372, 0.0454, 0.0436, 0.0211, 0.0332, 0.0163, 0.0246], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0216, 0.0210, 0.0209, 0.0216, 0.0216, 0.0221, 0.0212], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 03:58:31,078 INFO [train.py:904] (6/8) Epoch 15, batch 4800, loss[loss=0.2147, simple_loss=0.3046, pruned_loss=0.06243, over 16891.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2796, pruned_loss=0.05399, over 3200250.32 frames. ], batch size: 116, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 03:58:36,176 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.051e+02 1.856e+02 2.167e+02 2.629e+02 4.962e+02, threshold=4.334e+02, percent-clipped=1.0 2023-04-30 03:58:36,795 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146905.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:59:40,651 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146947.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:59:47,611 INFO [train.py:904] (6/8) Epoch 15, batch 4850, loss[loss=0.1815, simple_loss=0.2656, pruned_loss=0.04866, over 17027.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2798, pruned_loss=0.05299, over 3177959.23 frames. ], batch size: 53, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:00:34,921 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146982.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 04:01:03,664 INFO [train.py:904] (6/8) Epoch 15, batch 4900, loss[loss=0.1821, simple_loss=0.2669, pruned_loss=0.04864, over 17041.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.2788, pruned_loss=0.05201, over 3155004.86 frames. ], batch size: 50, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:01:08,002 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.422e+02 1.989e+02 2.229e+02 2.704e+02 6.823e+02, threshold=4.458e+02, percent-clipped=4.0 2023-04-30 04:01:45,413 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 04:01:50,159 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-30 04:02:16,315 INFO [train.py:904] (6/8) Epoch 15, batch 4950, loss[loss=0.1726, simple_loss=0.2766, pruned_loss=0.03426, over 16389.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2784, pruned_loss=0.05101, over 3167750.15 frames. ], batch size: 146, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:02:34,076 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8832, 4.0000, 4.2664, 4.2293, 4.2401, 3.9899, 3.9739, 3.9399], device='cuda:6'), covar=tensor([0.0319, 0.0461, 0.0372, 0.0444, 0.0440, 0.0359, 0.0806, 0.0466], device='cuda:6'), in_proj_covar=tensor([0.0362, 0.0388, 0.0384, 0.0365, 0.0430, 0.0405, 0.0499, 0.0323], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 04:03:03,307 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-04-30 04:03:25,722 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3144, 2.2645, 2.2622, 4.1069, 2.1641, 2.6502, 2.3171, 2.4317], device='cuda:6'), covar=tensor([0.1122, 0.3109, 0.2491, 0.0446, 0.3582, 0.2176, 0.3020, 0.2822], device='cuda:6'), in_proj_covar=tensor([0.0378, 0.0417, 0.0345, 0.0320, 0.0421, 0.0478, 0.0381, 0.0487], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:03:26,618 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8612, 3.7893, 3.9797, 3.6778, 3.8584, 4.2796, 3.9255, 3.5854], device='cuda:6'), covar=tensor([0.2086, 0.2190, 0.1941, 0.2468, 0.2595, 0.1787, 0.1531, 0.2744], device='cuda:6'), in_proj_covar=tensor([0.0375, 0.0526, 0.0571, 0.0449, 0.0596, 0.0602, 0.0456, 0.0598], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 04:03:28,736 INFO [train.py:904] (6/8) Epoch 15, batch 5000, loss[loss=0.1809, simple_loss=0.276, pruned_loss=0.04288, over 16875.00 frames. ], tot_loss[loss=0.1906, simple_loss=0.2797, pruned_loss=0.05075, over 3174151.16 frames. ], batch size: 102, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:03:32,271 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.607e+02 2.177e+02 2.680e+02 3.002e+02 5.827e+02, threshold=5.360e+02, percent-clipped=3.0 2023-04-30 04:03:52,132 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8486, 3.9036, 2.2849, 4.7418, 2.8846, 4.5281, 2.4622, 2.9605], device='cuda:6'), covar=tensor([0.0240, 0.0282, 0.1573, 0.0069, 0.0794, 0.0342, 0.1447, 0.0734], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0170, 0.0190, 0.0143, 0.0168, 0.0209, 0.0197, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 04:04:12,532 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2023-04-30 04:04:39,304 INFO [train.py:904] (6/8) Epoch 15, batch 5050, loss[loss=0.1794, simple_loss=0.2809, pruned_loss=0.03899, over 16696.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2796, pruned_loss=0.05023, over 3193004.75 frames. ], batch size: 89, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:05:46,845 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147200.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:05:48,951 INFO [train.py:904] (6/8) Epoch 15, batch 5100, loss[loss=0.1674, simple_loss=0.2608, pruned_loss=0.037, over 16556.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.278, pruned_loss=0.04948, over 3203296.53 frames. ], batch size: 75, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:05:52,961 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.025e+02 2.374e+02 2.786e+02 3.985e+02, threshold=4.748e+02, percent-clipped=0.0 2023-04-30 04:06:30,091 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-30 04:06:45,600 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147241.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:06:49,109 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5045, 4.5257, 4.3401, 3.9541, 4.0033, 4.4164, 4.3260, 4.0989], device='cuda:6'), covar=tensor([0.0566, 0.0349, 0.0288, 0.0325, 0.1001, 0.0401, 0.0462, 0.0728], device='cuda:6'), in_proj_covar=tensor([0.0261, 0.0360, 0.0312, 0.0295, 0.0325, 0.0340, 0.0209, 0.0367], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:06:53,112 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147247.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:06:56,189 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1922, 4.1581, 4.1241, 3.3129, 4.0719, 1.5082, 3.8479, 3.7517], device='cuda:6'), covar=tensor([0.0089, 0.0082, 0.0120, 0.0377, 0.0090, 0.2698, 0.0132, 0.0229], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0131, 0.0178, 0.0167, 0.0151, 0.0189, 0.0167, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:06:59,914 INFO [train.py:904] (6/8) Epoch 15, batch 5150, loss[loss=0.1948, simple_loss=0.2918, pruned_loss=0.04893, over 16747.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2783, pruned_loss=0.04901, over 3184172.69 frames. ], batch size: 83, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:07:23,479 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7511, 1.8360, 2.3061, 2.7614, 2.6625, 3.0851, 1.9857, 3.0321], device='cuda:6'), covar=tensor([0.0156, 0.0420, 0.0305, 0.0226, 0.0245, 0.0140, 0.0418, 0.0098], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0181, 0.0166, 0.0171, 0.0180, 0.0137, 0.0183, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:07:45,017 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147282.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:08:03,062 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=147295.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:08:14,126 INFO [train.py:904] (6/8) Epoch 15, batch 5200, loss[loss=0.2315, simple_loss=0.3034, pruned_loss=0.07975, over 16208.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2774, pruned_loss=0.04913, over 3176915.47 frames. ], batch size: 35, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:08:14,586 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147302.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:08:17,787 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.189e+02 2.515e+02 3.060e+02 5.067e+02, threshold=5.031e+02, percent-clipped=2.0 2023-04-30 04:08:25,262 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147310.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:08:55,678 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=147330.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:09:27,024 INFO [train.py:904] (6/8) Epoch 15, batch 5250, loss[loss=0.1924, simple_loss=0.2789, pruned_loss=0.0529, over 16782.00 frames. ], tot_loss[loss=0.1859, simple_loss=0.2747, pruned_loss=0.04854, over 3183955.14 frames. ], batch size: 76, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:09:45,583 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1207, 1.4805, 1.8503, 2.1121, 2.1718, 2.3462, 1.7095, 2.2955], device='cuda:6'), covar=tensor([0.0198, 0.0439, 0.0243, 0.0296, 0.0266, 0.0170, 0.0430, 0.0111], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0182, 0.0168, 0.0173, 0.0182, 0.0138, 0.0185, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:09:54,963 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147371.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:10:37,512 INFO [train.py:904] (6/8) Epoch 15, batch 5300, loss[loss=0.1542, simple_loss=0.2376, pruned_loss=0.0354, over 17034.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2716, pruned_loss=0.04766, over 3172672.12 frames. ], batch size: 53, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:10:40,979 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.533e+02 1.977e+02 2.284e+02 2.754e+02 4.909e+02, threshold=4.569e+02, percent-clipped=0.0 2023-04-30 04:10:45,235 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147407.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:11:23,819 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147434.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:11:49,963 INFO [train.py:904] (6/8) Epoch 15, batch 5350, loss[loss=0.1845, simple_loss=0.2809, pruned_loss=0.04401, over 16889.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.2698, pruned_loss=0.04699, over 3182743.30 frames. ], batch size: 96, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:12:14,794 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147468.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:12:51,794 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5001, 4.4798, 4.8767, 4.8720, 4.8531, 4.5904, 4.5512, 4.3237], device='cuda:6'), covar=tensor([0.0244, 0.0515, 0.0335, 0.0321, 0.0396, 0.0294, 0.0767, 0.0441], device='cuda:6'), in_proj_covar=tensor([0.0360, 0.0386, 0.0381, 0.0362, 0.0430, 0.0402, 0.0498, 0.0323], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 04:12:53,669 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147495.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:13:01,222 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147500.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:13:03,281 INFO [train.py:904] (6/8) Epoch 15, batch 5400, loss[loss=0.1849, simple_loss=0.2782, pruned_loss=0.04586, over 15557.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2727, pruned_loss=0.04754, over 3187397.16 frames. ], batch size: 190, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:13:07,672 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 2.104e+02 2.571e+02 3.291e+02 5.861e+02, threshold=5.143e+02, percent-clipped=4.0 2023-04-30 04:14:08,457 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.60 vs. limit=2.0 2023-04-30 04:14:13,937 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=147548.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:14:20,735 INFO [train.py:904] (6/8) Epoch 15, batch 5450, loss[loss=0.1871, simple_loss=0.2824, pruned_loss=0.04593, over 16483.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2756, pruned_loss=0.04893, over 3180443.01 frames. ], batch size: 75, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:15:32,503 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147597.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:15:40,035 INFO [train.py:904] (6/8) Epoch 15, batch 5500, loss[loss=0.2145, simple_loss=0.2982, pruned_loss=0.06544, over 16443.00 frames. ], tot_loss[loss=0.195, simple_loss=0.283, pruned_loss=0.05351, over 3166335.98 frames. ], batch size: 68, lr: 4.53e-03, grad_scale: 4.0 2023-04-30 04:15:45,682 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.855e+02 2.456e+02 3.016e+02 4.198e+02 6.055e+02, threshold=6.032e+02, percent-clipped=7.0 2023-04-30 04:16:58,371 INFO [train.py:904] (6/8) Epoch 15, batch 5550, loss[loss=0.3029, simple_loss=0.3529, pruned_loss=0.1265, over 11064.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2905, pruned_loss=0.05906, over 3137690.46 frames. ], batch size: 247, lr: 4.53e-03, grad_scale: 4.0 2023-04-30 04:17:10,661 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147659.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:17:22,973 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147666.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:17:58,778 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4648, 1.7771, 2.1460, 2.4618, 2.5062, 2.8176, 1.8112, 2.6997], device='cuda:6'), covar=tensor([0.0170, 0.0416, 0.0275, 0.0266, 0.0245, 0.0150, 0.0441, 0.0106], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0181, 0.0167, 0.0171, 0.0180, 0.0137, 0.0184, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:18:21,668 INFO [train.py:904] (6/8) Epoch 15, batch 5600, loss[loss=0.3071, simple_loss=0.3507, pruned_loss=0.1317, over 10825.00 frames. ], tot_loss[loss=0.2121, simple_loss=0.2962, pruned_loss=0.06403, over 3106003.62 frames. ], batch size: 248, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:18:28,273 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.312e+02 3.387e+02 4.165e+02 5.212e+02 1.017e+03, threshold=8.330e+02, percent-clipped=15.0 2023-04-30 04:18:53,024 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147720.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:19:34,335 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1433, 1.9951, 2.6538, 3.1164, 2.9282, 3.5597, 2.1879, 3.4499], device='cuda:6'), covar=tensor([0.0141, 0.0401, 0.0221, 0.0188, 0.0208, 0.0110, 0.0410, 0.0088], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0181, 0.0166, 0.0170, 0.0178, 0.0136, 0.0183, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:19:46,181 INFO [train.py:904] (6/8) Epoch 15, batch 5650, loss[loss=0.2372, simple_loss=0.3204, pruned_loss=0.07704, over 16398.00 frames. ], tot_loss[loss=0.219, simple_loss=0.3014, pruned_loss=0.06831, over 3075887.53 frames. ], batch size: 146, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:20:04,641 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147763.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:20:47,700 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147790.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:21:05,460 INFO [train.py:904] (6/8) Epoch 15, batch 5700, loss[loss=0.196, simple_loss=0.2883, pruned_loss=0.05187, over 16573.00 frames. ], tot_loss[loss=0.2198, simple_loss=0.3017, pruned_loss=0.069, over 3089249.95 frames. ], batch size: 68, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:21:11,566 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.489e+02 3.643e+02 4.297e+02 4.973e+02 1.168e+03, threshold=8.593e+02, percent-clipped=2.0 2023-04-30 04:21:22,321 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-30 04:21:26,335 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147815.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:22:23,096 INFO [train.py:904] (6/8) Epoch 15, batch 5750, loss[loss=0.2092, simple_loss=0.2968, pruned_loss=0.0608, over 16396.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.3037, pruned_loss=0.06959, over 3094508.31 frames. ], batch size: 75, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:22:32,855 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147857.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:23:03,192 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147876.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:23:37,276 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147897.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:23:44,669 INFO [train.py:904] (6/8) Epoch 15, batch 5800, loss[loss=0.2217, simple_loss=0.2908, pruned_loss=0.07631, over 11987.00 frames. ], tot_loss[loss=0.2195, simple_loss=0.3027, pruned_loss=0.06814, over 3084788.65 frames. ], batch size: 248, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:23:51,421 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.104e+02 2.959e+02 3.306e+02 4.369e+02 1.266e+03, threshold=6.612e+02, percent-clipped=1.0 2023-04-30 04:23:53,957 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9121, 4.8471, 4.6617, 3.3095, 4.0801, 4.6770, 4.2679, 2.8139], device='cuda:6'), covar=tensor([0.0428, 0.0029, 0.0033, 0.0294, 0.0075, 0.0081, 0.0052, 0.0334], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0074, 0.0074, 0.0131, 0.0089, 0.0098, 0.0086, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 04:24:00,765 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147911.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:24:12,312 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147918.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:24:55,226 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=147945.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:25:05,474 INFO [train.py:904] (6/8) Epoch 15, batch 5850, loss[loss=0.2021, simple_loss=0.2773, pruned_loss=0.06345, over 11395.00 frames. ], tot_loss[loss=0.2174, simple_loss=0.3011, pruned_loss=0.06686, over 3085713.21 frames. ], batch size: 247, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:25:29,139 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147966.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:25:38,373 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147972.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:26:30,239 INFO [train.py:904] (6/8) Epoch 15, batch 5900, loss[loss=0.1759, simple_loss=0.2718, pruned_loss=0.03998, over 16853.00 frames. ], tot_loss[loss=0.2174, simple_loss=0.301, pruned_loss=0.06688, over 3085527.10 frames. ], batch size: 102, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:26:39,372 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.034e+02 2.736e+02 3.208e+02 4.005e+02 8.372e+02, threshold=6.416e+02, percent-clipped=2.0 2023-04-30 04:26:52,640 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148014.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:26:54,061 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148015.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:27:50,060 INFO [train.py:904] (6/8) Epoch 15, batch 5950, loss[loss=0.2027, simple_loss=0.2939, pruned_loss=0.05578, over 16948.00 frames. ], tot_loss[loss=0.2163, simple_loss=0.3014, pruned_loss=0.06561, over 3083448.59 frames. ], batch size: 109, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:28:09,455 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148063.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:28:49,024 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148090.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:29:08,854 INFO [train.py:904] (6/8) Epoch 15, batch 6000, loss[loss=0.1834, simple_loss=0.271, pruned_loss=0.04793, over 16503.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.3006, pruned_loss=0.06518, over 3099244.31 frames. ], batch size: 75, lr: 4.52e-03, grad_scale: 8.0 2023-04-30 04:29:08,854 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 04:29:15,347 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2003, 5.6415, 5.5655, 5.2981, 5.2606, 5.9132, 5.3476, 5.1663], device='cuda:6'), covar=tensor([0.0760, 0.1426, 0.1939, 0.1356, 0.2399, 0.0716, 0.1280, 0.2026], device='cuda:6'), in_proj_covar=tensor([0.0381, 0.0536, 0.0583, 0.0455, 0.0608, 0.0613, 0.0465, 0.0611], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 04:29:19,436 INFO [train.py:938] (6/8) Epoch 15, validation: loss=0.1559, simple_loss=0.2691, pruned_loss=0.0214, over 944034.00 frames. 2023-04-30 04:29:19,437 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17591MB 2023-04-30 04:29:26,128 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.824e+02 2.753e+02 3.314e+02 4.296e+02 7.936e+02, threshold=6.628e+02, percent-clipped=2.0 2023-04-30 04:29:27,909 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2023-04-30 04:29:33,717 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148111.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:30:15,355 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148138.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:30:15,629 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4049, 3.5285, 2.8226, 2.1412, 2.4395, 2.3330, 3.7835, 3.3132], device='cuda:6'), covar=tensor([0.3037, 0.0868, 0.1679, 0.2386, 0.2478, 0.1930, 0.0483, 0.1188], device='cuda:6'), in_proj_covar=tensor([0.0311, 0.0261, 0.0291, 0.0291, 0.0285, 0.0234, 0.0276, 0.0311], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:30:18,891 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.63 vs. limit=5.0 2023-04-30 04:30:36,858 INFO [train.py:904] (6/8) Epoch 15, batch 6050, loss[loss=0.1864, simple_loss=0.2865, pruned_loss=0.04314, over 16684.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2992, pruned_loss=0.06462, over 3105596.55 frames. ], batch size: 76, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:30:38,306 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9870, 2.4110, 2.5941, 1.8783, 2.6758, 2.7675, 2.4398, 2.3456], device='cuda:6'), covar=tensor([0.0745, 0.0216, 0.0197, 0.0943, 0.0105, 0.0222, 0.0421, 0.0424], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0104, 0.0090, 0.0136, 0.0072, 0.0114, 0.0122, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 04:31:08,627 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148171.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:31:45,717 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4879, 3.5419, 1.8254, 3.9662, 2.5562, 3.9252, 1.9988, 2.6494], device='cuda:6'), covar=tensor([0.0263, 0.0387, 0.1885, 0.0231, 0.0851, 0.0526, 0.1701, 0.0840], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0171, 0.0190, 0.0143, 0.0169, 0.0210, 0.0197, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 04:31:59,320 INFO [train.py:904] (6/8) Epoch 15, batch 6100, loss[loss=0.1742, simple_loss=0.2655, pruned_loss=0.04147, over 17046.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2986, pruned_loss=0.06376, over 3101722.30 frames. ], batch size: 53, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:32:04,599 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-30 04:32:08,601 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.685e+02 2.651e+02 3.170e+02 3.946e+02 8.387e+02, threshold=6.339e+02, percent-clipped=2.0 2023-04-30 04:32:18,624 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148213.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:32:36,387 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6896, 1.7540, 1.6283, 1.5708, 1.9124, 1.6000, 1.6439, 1.9394], device='cuda:6'), covar=tensor([0.0156, 0.0247, 0.0328, 0.0295, 0.0169, 0.0230, 0.0148, 0.0179], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0217, 0.0210, 0.0212, 0.0216, 0.0217, 0.0219, 0.0211], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:33:03,831 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4938, 3.6915, 4.1901, 1.9671, 4.4206, 4.3826, 3.0895, 3.1772], device='cuda:6'), covar=tensor([0.0849, 0.0219, 0.0152, 0.1169, 0.0048, 0.0102, 0.0398, 0.0434], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0104, 0.0090, 0.0138, 0.0073, 0.0115, 0.0123, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 04:33:18,245 INFO [train.py:904] (6/8) Epoch 15, batch 6150, loss[loss=0.1882, simple_loss=0.2843, pruned_loss=0.04609, over 16748.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2972, pruned_loss=0.0636, over 3076289.90 frames. ], batch size: 102, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:33:43,177 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148267.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:34:15,527 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148287.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:34:37,597 INFO [train.py:904] (6/8) Epoch 15, batch 6200, loss[loss=0.218, simple_loss=0.303, pruned_loss=0.0665, over 16381.00 frames. ], tot_loss[loss=0.212, simple_loss=0.2961, pruned_loss=0.06397, over 3059983.46 frames. ], batch size: 146, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:34:46,166 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.083e+02 3.031e+02 3.685e+02 4.615e+02 1.155e+03, threshold=7.370e+02, percent-clipped=8.0 2023-04-30 04:34:58,567 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148315.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:35:48,384 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148348.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:35:54,858 INFO [train.py:904] (6/8) Epoch 15, batch 6250, loss[loss=0.2062, simple_loss=0.2951, pruned_loss=0.05864, over 16406.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2952, pruned_loss=0.06356, over 3070023.46 frames. ], batch size: 146, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:36:11,839 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148363.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:37:11,929 INFO [train.py:904] (6/8) Epoch 15, batch 6300, loss[loss=0.1994, simple_loss=0.2861, pruned_loss=0.05629, over 16325.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2952, pruned_loss=0.06304, over 3076953.66 frames. ], batch size: 146, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:37:21,867 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.698e+02 2.809e+02 3.245e+02 4.323e+02 1.479e+03, threshold=6.491e+02, percent-clipped=2.0 2023-04-30 04:37:57,691 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148429.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:38:33,538 INFO [train.py:904] (6/8) Epoch 15, batch 6350, loss[loss=0.1827, simple_loss=0.2721, pruned_loss=0.04661, over 17253.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2954, pruned_loss=0.06369, over 3071423.79 frames. ], batch size: 52, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:39:03,798 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148471.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:39:34,180 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148490.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:39:45,792 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148498.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:39:51,469 INFO [train.py:904] (6/8) Epoch 15, batch 6400, loss[loss=0.2029, simple_loss=0.2826, pruned_loss=0.06157, over 16748.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2955, pruned_loss=0.06465, over 3088183.91 frames. ], batch size: 124, lr: 4.52e-03, grad_scale: 8.0 2023-04-30 04:39:59,983 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.110e+02 3.263e+02 3.782e+02 4.494e+02 8.205e+02, threshold=7.565e+02, percent-clipped=3.0 2023-04-30 04:40:09,512 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148513.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:40:18,146 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148519.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:41:07,997 INFO [train.py:904] (6/8) Epoch 15, batch 6450, loss[loss=0.2024, simple_loss=0.2927, pruned_loss=0.05602, over 16595.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2953, pruned_loss=0.06412, over 3071431.91 frames. ], batch size: 62, lr: 4.52e-03, grad_scale: 8.0 2023-04-30 04:41:19,874 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148559.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:41:22,167 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148561.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:41:30,692 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7679, 3.8627, 3.9358, 3.6947, 3.8370, 4.2288, 3.9475, 3.6820], device='cuda:6'), covar=tensor([0.2063, 0.2010, 0.2163, 0.2475, 0.2567, 0.1685, 0.1485, 0.2460], device='cuda:6'), in_proj_covar=tensor([0.0383, 0.0543, 0.0589, 0.0460, 0.0610, 0.0619, 0.0468, 0.0614], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 04:41:32,441 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148567.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:42:04,594 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6377, 3.9144, 2.7384, 2.2760, 2.5487, 2.2355, 3.9986, 3.2057], device='cuda:6'), covar=tensor([0.2921, 0.0748, 0.1922, 0.2395, 0.2638, 0.2165, 0.0531, 0.1337], device='cuda:6'), in_proj_covar=tensor([0.0307, 0.0259, 0.0288, 0.0289, 0.0282, 0.0233, 0.0273, 0.0309], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:42:26,999 INFO [train.py:904] (6/8) Epoch 15, batch 6500, loss[loss=0.216, simple_loss=0.2873, pruned_loss=0.07238, over 11623.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.2929, pruned_loss=0.06319, over 3086704.33 frames. ], batch size: 247, lr: 4.52e-03, grad_scale: 8.0 2023-04-30 04:42:36,999 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.948e+02 2.869e+02 3.331e+02 4.011e+02 8.248e+02, threshold=6.661e+02, percent-clipped=1.0 2023-04-30 04:42:47,764 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148615.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:43:33,196 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148643.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:43:47,240 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148651.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:43:47,985 INFO [train.py:904] (6/8) Epoch 15, batch 6550, loss[loss=0.2737, simple_loss=0.3314, pruned_loss=0.108, over 11539.00 frames. ], tot_loss[loss=0.212, simple_loss=0.2956, pruned_loss=0.06424, over 3084226.45 frames. ], batch size: 246, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:44:14,088 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7788, 3.8594, 4.3938, 2.0238, 4.5433, 4.5693, 3.2279, 3.3772], device='cuda:6'), covar=tensor([0.0709, 0.0182, 0.0131, 0.1103, 0.0048, 0.0111, 0.0347, 0.0382], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0105, 0.0091, 0.0137, 0.0073, 0.0114, 0.0123, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 04:44:47,684 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-30 04:45:05,098 INFO [train.py:904] (6/8) Epoch 15, batch 6600, loss[loss=0.262, simple_loss=0.3272, pruned_loss=0.09843, over 11894.00 frames. ], tot_loss[loss=0.2139, simple_loss=0.2979, pruned_loss=0.06498, over 3079518.26 frames. ], batch size: 249, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:45:13,945 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.578e+02 3.041e+02 3.810e+02 4.934e+02 1.256e+03, threshold=7.620e+02, percent-clipped=11.0 2023-04-30 04:45:19,897 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148712.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:45:26,258 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148716.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:46:00,324 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1925, 2.0592, 2.1584, 3.7507, 2.0927, 2.4707, 2.1856, 2.2113], device='cuda:6'), covar=tensor([0.1149, 0.3459, 0.2699, 0.0496, 0.3992, 0.2270, 0.3440, 0.3383], device='cuda:6'), in_proj_covar=tensor([0.0375, 0.0413, 0.0343, 0.0317, 0.0421, 0.0475, 0.0380, 0.0484], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:46:22,120 INFO [train.py:904] (6/8) Epoch 15, batch 6650, loss[loss=0.2068, simple_loss=0.2905, pruned_loss=0.06154, over 16716.00 frames. ], tot_loss[loss=0.2144, simple_loss=0.2979, pruned_loss=0.06541, over 3090100.42 frames. ], batch size: 76, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:46:34,220 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7175, 3.8441, 2.9298, 2.2142, 2.5626, 2.4091, 4.0801, 3.4091], device='cuda:6'), covar=tensor([0.2637, 0.0682, 0.1694, 0.2308, 0.2484, 0.1876, 0.0452, 0.1132], device='cuda:6'), in_proj_covar=tensor([0.0311, 0.0262, 0.0291, 0.0292, 0.0285, 0.0235, 0.0276, 0.0312], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 04:47:00,920 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148777.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:47:13,479 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148785.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:47:39,115 INFO [train.py:904] (6/8) Epoch 15, batch 6700, loss[loss=0.2863, simple_loss=0.3416, pruned_loss=0.1155, over 11787.00 frames. ], tot_loss[loss=0.2144, simple_loss=0.2973, pruned_loss=0.06579, over 3078358.03 frames. ], batch size: 248, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:47:47,145 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.925e+02 2.803e+02 3.506e+02 4.354e+02 9.571e+02, threshold=7.012e+02, percent-clipped=2.0 2023-04-30 04:47:57,622 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6266, 2.6153, 1.8159, 2.7428, 2.1031, 2.7707, 2.0884, 2.3464], device='cuda:6'), covar=tensor([0.0311, 0.0420, 0.1311, 0.0204, 0.0685, 0.0524, 0.1268, 0.0597], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0170, 0.0188, 0.0142, 0.0168, 0.0209, 0.0195, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 04:48:54,917 INFO [train.py:904] (6/8) Epoch 15, batch 6750, loss[loss=0.2021, simple_loss=0.292, pruned_loss=0.05608, over 16526.00 frames. ], tot_loss[loss=0.214, simple_loss=0.2968, pruned_loss=0.06566, over 3082041.34 frames. ], batch size: 68, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:48:58,963 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148854.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:49:01,703 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1817, 5.1380, 4.9999, 4.6686, 4.6416, 5.0173, 5.0177, 4.7352], device='cuda:6'), covar=tensor([0.0546, 0.0292, 0.0264, 0.0254, 0.0967, 0.0376, 0.0264, 0.0587], device='cuda:6'), in_proj_covar=tensor([0.0262, 0.0363, 0.0310, 0.0293, 0.0325, 0.0340, 0.0209, 0.0368], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:49:08,812 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2023-04-30 04:50:09,776 INFO [train.py:904] (6/8) Epoch 15, batch 6800, loss[loss=0.1932, simple_loss=0.2754, pruned_loss=0.05548, over 17127.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2969, pruned_loss=0.06577, over 3082237.23 frames. ], batch size: 47, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:50:21,236 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.768e+02 2.837e+02 3.542e+02 4.429e+02 7.153e+02, threshold=7.083e+02, percent-clipped=1.0 2023-04-30 04:51:15,912 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148943.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:51:28,452 INFO [train.py:904] (6/8) Epoch 15, batch 6850, loss[loss=0.2055, simple_loss=0.304, pruned_loss=0.05346, over 16923.00 frames. ], tot_loss[loss=0.214, simple_loss=0.2977, pruned_loss=0.06514, over 3104838.87 frames. ], batch size: 109, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:52:26,495 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148991.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:52:40,161 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9060, 2.1254, 2.3740, 3.1641, 2.1956, 2.3445, 2.3236, 2.2123], device='cuda:6'), covar=tensor([0.1150, 0.2962, 0.2144, 0.0632, 0.3681, 0.2060, 0.2840, 0.3046], device='cuda:6'), in_proj_covar=tensor([0.0373, 0.0410, 0.0342, 0.0315, 0.0419, 0.0472, 0.0377, 0.0479], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 04:52:43,265 INFO [train.py:904] (6/8) Epoch 15, batch 6900, loss[loss=0.2707, simple_loss=0.3415, pruned_loss=0.09995, over 15382.00 frames. ], tot_loss[loss=0.215, simple_loss=0.3, pruned_loss=0.06502, over 3100305.03 frames. ], batch size: 191, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:52:52,157 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149007.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:52:54,700 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.886e+02 2.687e+02 3.079e+02 3.976e+02 7.116e+02, threshold=6.157e+02, percent-clipped=1.0 2023-04-30 04:53:58,309 INFO [train.py:904] (6/8) Epoch 15, batch 6950, loss[loss=0.2119, simple_loss=0.2977, pruned_loss=0.06306, over 17214.00 frames. ], tot_loss[loss=0.2186, simple_loss=0.3024, pruned_loss=0.06739, over 3079374.67 frames. ], batch size: 44, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:54:27,426 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149072.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:54:45,951 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149085.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:55:10,868 INFO [train.py:904] (6/8) Epoch 15, batch 7000, loss[loss=0.237, simple_loss=0.308, pruned_loss=0.08302, over 11718.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.3027, pruned_loss=0.06679, over 3079374.84 frames. ], batch size: 248, lr: 4.51e-03, grad_scale: 2.0 2023-04-30 04:55:23,364 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.893e+02 3.767e+02 4.998e+02 1.151e+03, threshold=7.533e+02, percent-clipped=10.0 2023-04-30 04:55:55,533 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-30 04:55:56,337 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7731, 3.7325, 3.8845, 3.6569, 3.8168, 4.1951, 3.9099, 3.6289], device='cuda:6'), covar=tensor([0.2037, 0.2194, 0.2193, 0.2562, 0.2608, 0.1902, 0.1655, 0.2762], device='cuda:6'), in_proj_covar=tensor([0.0381, 0.0535, 0.0583, 0.0454, 0.0601, 0.0613, 0.0464, 0.0607], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 04:55:58,211 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=149133.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:56:25,570 INFO [train.py:904] (6/8) Epoch 15, batch 7050, loss[loss=0.2246, simple_loss=0.3057, pruned_loss=0.07174, over 16438.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.3037, pruned_loss=0.06699, over 3062226.07 frames. ], batch size: 146, lr: 4.51e-03, grad_scale: 2.0 2023-04-30 04:56:29,278 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149154.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:57:15,624 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9242, 5.5121, 5.6311, 5.3958, 5.3893, 5.9615, 5.4382, 5.2491], device='cuda:6'), covar=tensor([0.0950, 0.1637, 0.1965, 0.1953, 0.2375, 0.0980, 0.1472, 0.2407], device='cuda:6'), in_proj_covar=tensor([0.0380, 0.0534, 0.0582, 0.0452, 0.0601, 0.0611, 0.0464, 0.0604], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 04:57:40,354 INFO [train.py:904] (6/8) Epoch 15, batch 7100, loss[loss=0.1972, simple_loss=0.2817, pruned_loss=0.05634, over 16736.00 frames. ], tot_loss[loss=0.2195, simple_loss=0.3032, pruned_loss=0.06788, over 3022694.39 frames. ], batch size: 62, lr: 4.51e-03, grad_scale: 2.0 2023-04-30 04:57:40,784 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=149202.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:57:53,079 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.62 vs. limit=2.0 2023-04-30 04:57:53,981 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.949e+02 2.990e+02 3.583e+02 4.304e+02 1.223e+03, threshold=7.166e+02, percent-clipped=1.0 2023-04-30 04:58:55,144 INFO [train.py:904] (6/8) Epoch 15, batch 7150, loss[loss=0.1867, simple_loss=0.2788, pruned_loss=0.04736, over 16959.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.3017, pruned_loss=0.06787, over 3008995.81 frames. ], batch size: 96, lr: 4.51e-03, grad_scale: 2.0 2023-04-30 05:00:05,782 INFO [train.py:904] (6/8) Epoch 15, batch 7200, loss[loss=0.2188, simple_loss=0.2982, pruned_loss=0.06971, over 11805.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2982, pruned_loss=0.06501, over 3027738.69 frames. ], batch size: 246, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:00:13,303 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149307.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:00:17,916 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 2.693e+02 3.090e+02 3.791e+02 7.265e+02, threshold=6.181e+02, percent-clipped=1.0 2023-04-30 05:01:22,364 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2023-04-30 05:01:26,134 INFO [train.py:904] (6/8) Epoch 15, batch 7250, loss[loss=0.2477, simple_loss=0.307, pruned_loss=0.09423, over 11460.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2959, pruned_loss=0.06424, over 3013042.09 frames. ], batch size: 247, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:01:31,287 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=149355.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:01:57,374 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149372.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:02:42,286 INFO [train.py:904] (6/8) Epoch 15, batch 7300, loss[loss=0.2186, simple_loss=0.2892, pruned_loss=0.07399, over 11633.00 frames. ], tot_loss[loss=0.2111, simple_loss=0.295, pruned_loss=0.06365, over 3035615.15 frames. ], batch size: 246, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:02:55,637 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.231e+02 2.888e+02 3.399e+02 4.220e+02 6.338e+02, threshold=6.799e+02, percent-clipped=2.0 2023-04-30 05:03:10,459 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=149420.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:03:19,009 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-30 05:03:58,890 INFO [train.py:904] (6/8) Epoch 15, batch 7350, loss[loss=0.1983, simple_loss=0.2792, pruned_loss=0.05864, over 17171.00 frames. ], tot_loss[loss=0.213, simple_loss=0.296, pruned_loss=0.06497, over 3010119.41 frames. ], batch size: 46, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:04:01,170 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4157, 4.4413, 4.8993, 4.8497, 4.8570, 4.4630, 4.5137, 4.2703], device='cuda:6'), covar=tensor([0.0311, 0.0532, 0.0297, 0.0372, 0.0453, 0.0394, 0.0922, 0.0558], device='cuda:6'), in_proj_covar=tensor([0.0365, 0.0391, 0.0382, 0.0367, 0.0436, 0.0408, 0.0504, 0.0324], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 05:05:17,915 INFO [train.py:904] (6/8) Epoch 15, batch 7400, loss[loss=0.1962, simple_loss=0.2856, pruned_loss=0.05335, over 16808.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2972, pruned_loss=0.06555, over 3008025.81 frames. ], batch size: 39, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:05:32,172 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.760e+02 2.749e+02 3.166e+02 4.051e+02 9.173e+02, threshold=6.332e+02, percent-clipped=4.0 2023-04-30 05:05:37,345 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=149514.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:06:37,038 INFO [train.py:904] (6/8) Epoch 15, batch 7450, loss[loss=0.2193, simple_loss=0.3066, pruned_loss=0.06599, over 16893.00 frames. ], tot_loss[loss=0.2156, simple_loss=0.298, pruned_loss=0.06655, over 3008729.42 frames. ], batch size: 109, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:07:00,479 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=149565.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:07:16,432 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=149575.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:08:00,168 INFO [train.py:904] (6/8) Epoch 15, batch 7500, loss[loss=0.1776, simple_loss=0.2584, pruned_loss=0.04843, over 17117.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2976, pruned_loss=0.06531, over 3019776.33 frames. ], batch size: 47, lr: 4.50e-03, grad_scale: 2.0 2023-04-30 05:08:16,081 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.765e+02 2.987e+02 3.634e+02 4.459e+02 1.121e+03, threshold=7.267e+02, percent-clipped=5.0 2023-04-30 05:08:40,070 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=149626.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 05:09:18,806 INFO [train.py:904] (6/8) Epoch 15, batch 7550, loss[loss=0.1791, simple_loss=0.2701, pruned_loss=0.04406, over 16524.00 frames. ], tot_loss[loss=0.2138, simple_loss=0.2965, pruned_loss=0.06559, over 3025464.58 frames. ], batch size: 68, lr: 4.50e-03, grad_scale: 2.0 2023-04-30 05:10:35,809 INFO [train.py:904] (6/8) Epoch 15, batch 7600, loss[loss=0.2196, simple_loss=0.304, pruned_loss=0.06763, over 16507.00 frames. ], tot_loss[loss=0.2139, simple_loss=0.2962, pruned_loss=0.06577, over 3033029.63 frames. ], batch size: 75, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:10:50,663 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.987e+02 2.884e+02 3.508e+02 4.101e+02 6.446e+02, threshold=7.017e+02, percent-clipped=0.0 2023-04-30 05:10:57,952 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1558, 3.7691, 3.7756, 2.3654, 3.4462, 3.8119, 3.4794, 1.9968], device='cuda:6'), covar=tensor([0.0516, 0.0045, 0.0040, 0.0393, 0.0085, 0.0083, 0.0075, 0.0419], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0074, 0.0073, 0.0132, 0.0088, 0.0097, 0.0085, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 05:11:00,961 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-30 05:11:45,955 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=149746.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 05:11:55,062 INFO [train.py:904] (6/8) Epoch 15, batch 7650, loss[loss=0.2081, simple_loss=0.3077, pruned_loss=0.05427, over 16523.00 frames. ], tot_loss[loss=0.2151, simple_loss=0.297, pruned_loss=0.06655, over 3039936.24 frames. ], batch size: 75, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:13:13,692 INFO [train.py:904] (6/8) Epoch 15, batch 7700, loss[loss=0.2632, simple_loss=0.3233, pruned_loss=0.1015, over 11711.00 frames. ], tot_loss[loss=0.2153, simple_loss=0.297, pruned_loss=0.06679, over 3041894.25 frames. ], batch size: 248, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:13:22,694 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=149807.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 05:13:29,264 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.980e+02 3.235e+02 4.085e+02 5.006e+02 1.161e+03, threshold=8.169e+02, percent-clipped=3.0 2023-04-30 05:13:49,865 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6112, 2.5934, 1.9064, 2.7100, 2.1898, 2.7399, 2.0926, 2.4107], device='cuda:6'), covar=tensor([0.0295, 0.0364, 0.1203, 0.0235, 0.0645, 0.0505, 0.1060, 0.0571], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0172, 0.0192, 0.0144, 0.0171, 0.0212, 0.0199, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 05:14:19,770 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3159, 3.3058, 3.3621, 3.4567, 3.4900, 3.2513, 3.4573, 3.5294], device='cuda:6'), covar=tensor([0.1192, 0.0869, 0.1056, 0.0641, 0.0648, 0.1975, 0.1019, 0.0791], device='cuda:6'), in_proj_covar=tensor([0.0556, 0.0689, 0.0826, 0.0700, 0.0530, 0.0554, 0.0563, 0.0654], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 05:14:33,192 INFO [train.py:904] (6/8) Epoch 15, batch 7750, loss[loss=0.1854, simple_loss=0.2721, pruned_loss=0.04931, over 16741.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.2974, pruned_loss=0.06669, over 3052215.04 frames. ], batch size: 39, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:15:02,329 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149870.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:15:51,917 INFO [train.py:904] (6/8) Epoch 15, batch 7800, loss[loss=0.1738, simple_loss=0.2694, pruned_loss=0.03905, over 16882.00 frames. ], tot_loss[loss=0.2159, simple_loss=0.2981, pruned_loss=0.06687, over 3050091.56 frames. ], batch size: 96, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:16:07,343 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.059e+02 3.237e+02 3.890e+02 4.639e+02 7.802e+02, threshold=7.781e+02, percent-clipped=0.0 2023-04-30 05:16:09,310 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6125, 2.8071, 2.4047, 4.2962, 3.0358, 4.0543, 1.5108, 2.8600], device='cuda:6'), covar=tensor([0.1470, 0.0722, 0.1293, 0.0176, 0.0304, 0.0418, 0.1687, 0.0878], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0166, 0.0188, 0.0167, 0.0203, 0.0211, 0.0192, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 05:16:21,587 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149921.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 05:16:21,876 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-30 05:17:06,633 INFO [train.py:904] (6/8) Epoch 15, batch 7850, loss[loss=0.2627, simple_loss=0.3218, pruned_loss=0.1019, over 11532.00 frames. ], tot_loss[loss=0.215, simple_loss=0.298, pruned_loss=0.06602, over 3052754.17 frames. ], batch size: 248, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:17:41,591 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5055, 3.4691, 3.3978, 2.6637, 3.3646, 2.0620, 3.1118, 2.9291], device='cuda:6'), covar=tensor([0.0157, 0.0128, 0.0184, 0.0240, 0.0108, 0.2267, 0.0137, 0.0234], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0131, 0.0179, 0.0165, 0.0151, 0.0191, 0.0166, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 05:18:25,131 INFO [train.py:904] (6/8) Epoch 15, batch 7900, loss[loss=0.1874, simple_loss=0.278, pruned_loss=0.0484, over 16662.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2971, pruned_loss=0.06563, over 3048514.09 frames. ], batch size: 76, lr: 4.49e-03, grad_scale: 4.0 2023-04-30 05:18:40,305 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.219e+02 2.824e+02 3.475e+02 4.207e+02 7.504e+02, threshold=6.949e+02, percent-clipped=0.0 2023-04-30 05:19:24,940 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0194, 3.9977, 3.9584, 3.2447, 3.9523, 1.8633, 3.7304, 3.5515], device='cuda:6'), covar=tensor([0.0119, 0.0104, 0.0162, 0.0301, 0.0104, 0.2553, 0.0136, 0.0229], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0131, 0.0178, 0.0165, 0.0151, 0.0191, 0.0166, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 05:19:43,691 INFO [train.py:904] (6/8) Epoch 15, batch 7950, loss[loss=0.2146, simple_loss=0.2977, pruned_loss=0.06574, over 15475.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2972, pruned_loss=0.06555, over 3064700.95 frames. ], batch size: 191, lr: 4.49e-03, grad_scale: 4.0 2023-04-30 05:20:07,342 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3025, 4.3665, 4.1742, 3.9280, 3.8801, 4.2555, 3.9837, 3.9797], device='cuda:6'), covar=tensor([0.0607, 0.0586, 0.0296, 0.0285, 0.0872, 0.0514, 0.0708, 0.0649], device='cuda:6'), in_proj_covar=tensor([0.0257, 0.0354, 0.0301, 0.0287, 0.0317, 0.0334, 0.0208, 0.0360], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 05:20:20,299 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-30 05:20:25,884 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150079.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:20:58,677 INFO [train.py:904] (6/8) Epoch 15, batch 8000, loss[loss=0.2078, simple_loss=0.2999, pruned_loss=0.05787, over 16399.00 frames. ], tot_loss[loss=0.2153, simple_loss=0.2984, pruned_loss=0.06605, over 3067143.27 frames. ], batch size: 146, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:20:59,763 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150102.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 05:21:14,670 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.969e+02 2.932e+02 3.545e+02 4.036e+02 7.060e+02, threshold=7.089e+02, percent-clipped=1.0 2023-04-30 05:21:29,996 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2054, 4.0713, 4.3063, 4.4520, 4.5678, 4.1794, 4.5155, 4.6077], device='cuda:6'), covar=tensor([0.1844, 0.1221, 0.1479, 0.0701, 0.0606, 0.1101, 0.0820, 0.0626], device='cuda:6'), in_proj_covar=tensor([0.0562, 0.0696, 0.0831, 0.0703, 0.0538, 0.0557, 0.0569, 0.0659], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 05:21:55,973 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150140.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:22:14,254 INFO [train.py:904] (6/8) Epoch 15, batch 8050, loss[loss=0.2107, simple_loss=0.3009, pruned_loss=0.06022, over 16350.00 frames. ], tot_loss[loss=0.2139, simple_loss=0.2976, pruned_loss=0.06504, over 3089008.23 frames. ], batch size: 146, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:22:34,099 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2563, 3.3124, 3.7742, 1.7908, 3.9029, 3.9581, 2.9026, 2.8712], device='cuda:6'), covar=tensor([0.0819, 0.0237, 0.0167, 0.1173, 0.0064, 0.0131, 0.0411, 0.0438], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0104, 0.0090, 0.0137, 0.0073, 0.0114, 0.0122, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 05:22:41,289 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150170.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:23:19,939 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3245, 5.6941, 5.4117, 5.4567, 5.0887, 5.0361, 5.0907, 5.8124], device='cuda:6'), covar=tensor([0.1109, 0.0797, 0.0964, 0.0786, 0.0871, 0.0673, 0.1113, 0.0747], device='cuda:6'), in_proj_covar=tensor([0.0597, 0.0734, 0.0613, 0.0531, 0.0462, 0.0475, 0.0613, 0.0563], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 05:23:20,061 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4908, 2.5484, 2.0287, 2.3939, 2.9923, 2.6331, 3.1737, 3.2697], device='cuda:6'), covar=tensor([0.0093, 0.0374, 0.0493, 0.0391, 0.0227, 0.0332, 0.0203, 0.0215], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0215, 0.0210, 0.0210, 0.0214, 0.0216, 0.0217, 0.0209], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 05:23:30,578 INFO [train.py:904] (6/8) Epoch 15, batch 8100, loss[loss=0.1963, simple_loss=0.28, pruned_loss=0.05624, over 17122.00 frames. ], tot_loss[loss=0.2125, simple_loss=0.2968, pruned_loss=0.0641, over 3111761.65 frames. ], batch size: 48, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:23:45,512 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.847e+02 2.677e+02 3.415e+02 4.132e+02 1.188e+03, threshold=6.830e+02, percent-clipped=3.0 2023-04-30 05:23:56,038 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=150218.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:23:59,919 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150221.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:24:45,701 INFO [train.py:904] (6/8) Epoch 15, batch 8150, loss[loss=0.1956, simple_loss=0.2748, pruned_loss=0.05821, over 16450.00 frames. ], tot_loss[loss=0.2094, simple_loss=0.2937, pruned_loss=0.06252, over 3111600.49 frames. ], batch size: 68, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:25:11,110 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=150269.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:25:19,685 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150274.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:25:28,423 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8806, 4.8712, 4.7111, 4.3737, 4.3809, 4.7653, 4.7006, 4.4522], device='cuda:6'), covar=tensor([0.0578, 0.0427, 0.0272, 0.0296, 0.1008, 0.0471, 0.0368, 0.0735], device='cuda:6'), in_proj_covar=tensor([0.0259, 0.0357, 0.0304, 0.0288, 0.0320, 0.0336, 0.0210, 0.0363], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 05:25:54,030 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2993, 3.9767, 3.9405, 2.6892, 3.5571, 3.9451, 3.6097, 2.1424], device='cuda:6'), covar=tensor([0.0482, 0.0043, 0.0041, 0.0347, 0.0091, 0.0099, 0.0075, 0.0435], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0074, 0.0073, 0.0132, 0.0088, 0.0097, 0.0085, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 05:26:00,375 INFO [train.py:904] (6/8) Epoch 15, batch 8200, loss[loss=0.1828, simple_loss=0.2727, pruned_loss=0.0464, over 16800.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2915, pruned_loss=0.06227, over 3103526.27 frames. ], batch size: 83, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:26:16,347 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.058e+02 2.882e+02 3.369e+02 3.879e+02 7.748e+02, threshold=6.737e+02, percent-clipped=3.0 2023-04-30 05:26:41,629 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8890, 3.8319, 4.3058, 2.1644, 4.5447, 4.5228, 3.3209, 3.5143], device='cuda:6'), covar=tensor([0.0663, 0.0216, 0.0203, 0.1099, 0.0051, 0.0151, 0.0342, 0.0365], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0105, 0.0091, 0.0138, 0.0073, 0.0115, 0.0123, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 05:26:54,625 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150335.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:27:17,534 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-30 05:27:22,317 INFO [train.py:904] (6/8) Epoch 15, batch 8250, loss[loss=0.183, simple_loss=0.276, pruned_loss=0.04503, over 16833.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2909, pruned_loss=0.06015, over 3090146.17 frames. ], batch size: 42, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:28:45,002 INFO [train.py:904] (6/8) Epoch 15, batch 8300, loss[loss=0.1766, simple_loss=0.278, pruned_loss=0.03756, over 16834.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2884, pruned_loss=0.05752, over 3081491.29 frames. ], batch size: 102, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:28:45,640 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150402.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 05:28:53,126 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0497, 2.7344, 2.8748, 2.1378, 2.6617, 2.1900, 2.6872, 2.9307], device='cuda:6'), covar=tensor([0.0310, 0.0869, 0.0457, 0.1730, 0.0775, 0.0936, 0.0610, 0.0743], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0151, 0.0159, 0.0144, 0.0137, 0.0124, 0.0138, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 05:29:01,272 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.394e+02 2.913e+02 3.586e+02 6.520e+02, threshold=5.826e+02, percent-clipped=0.0 2023-04-30 05:29:10,455 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4421, 3.4485, 3.6937, 1.6458, 3.8377, 3.9571, 3.1320, 3.0571], device='cuda:6'), covar=tensor([0.0791, 0.0216, 0.0178, 0.1388, 0.0074, 0.0137, 0.0332, 0.0425], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0104, 0.0090, 0.0136, 0.0072, 0.0113, 0.0121, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 05:29:39,973 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150435.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:30:02,261 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1300, 2.4404, 2.6303, 2.0097, 2.6417, 2.8166, 2.5496, 2.5272], device='cuda:6'), covar=tensor([0.0634, 0.0195, 0.0240, 0.0871, 0.0085, 0.0251, 0.0374, 0.0394], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0104, 0.0090, 0.0136, 0.0072, 0.0113, 0.0121, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 05:30:05,309 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=150450.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 05:30:07,755 INFO [train.py:904] (6/8) Epoch 15, batch 8350, loss[loss=0.2125, simple_loss=0.2893, pruned_loss=0.06783, over 12105.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2879, pruned_loss=0.05575, over 3078671.37 frames. ], batch size: 246, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:31:29,420 INFO [train.py:904] (6/8) Epoch 15, batch 8400, loss[loss=0.1699, simple_loss=0.2711, pruned_loss=0.03431, over 16815.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2847, pruned_loss=0.05367, over 3058161.68 frames. ], batch size: 102, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:31:46,286 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.409e+02 2.912e+02 3.516e+02 5.302e+02, threshold=5.823e+02, percent-clipped=0.0 2023-04-30 05:32:27,398 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150537.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:32:49,402 INFO [train.py:904] (6/8) Epoch 15, batch 8450, loss[loss=0.2006, simple_loss=0.2842, pruned_loss=0.0585, over 12521.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2829, pruned_loss=0.05184, over 3058941.67 frames. ], batch size: 248, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:34:02,222 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150598.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:34:08,622 INFO [train.py:904] (6/8) Epoch 15, batch 8500, loss[loss=0.1845, simple_loss=0.2599, pruned_loss=0.05455, over 12015.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.2794, pruned_loss=0.04949, over 3072544.67 frames. ], batch size: 247, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:34:25,006 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 2.200e+02 2.703e+02 3.335e+02 6.908e+02, threshold=5.407e+02, percent-clipped=1.0 2023-04-30 05:34:54,218 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150630.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:35:10,991 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6456, 2.6937, 1.7264, 2.8034, 2.1255, 2.8068, 1.9287, 2.3624], device='cuda:6'), covar=tensor([0.0245, 0.0306, 0.1353, 0.0237, 0.0696, 0.0428, 0.1375, 0.0504], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0165, 0.0184, 0.0137, 0.0165, 0.0203, 0.0193, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 05:35:30,232 INFO [train.py:904] (6/8) Epoch 15, batch 8550, loss[loss=0.1856, simple_loss=0.2941, pruned_loss=0.03855, over 16865.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2775, pruned_loss=0.04853, over 3059818.16 frames. ], batch size: 96, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:35:43,160 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6300, 2.6769, 1.8614, 2.8064, 2.1311, 2.7719, 2.1278, 2.4156], device='cuda:6'), covar=tensor([0.0258, 0.0283, 0.1232, 0.0213, 0.0658, 0.0382, 0.1193, 0.0528], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0165, 0.0185, 0.0138, 0.0165, 0.0203, 0.0194, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 05:36:21,421 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-30 05:36:57,815 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-04-30 05:37:12,290 INFO [train.py:904] (6/8) Epoch 15, batch 8600, loss[loss=0.1756, simple_loss=0.2741, pruned_loss=0.03859, over 16743.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2773, pruned_loss=0.04753, over 3049733.06 frames. ], batch size: 83, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:37:32,421 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.969e+02 2.521e+02 3.034e+02 3.543e+02 4.971e+02, threshold=6.069e+02, percent-clipped=0.0 2023-04-30 05:38:14,990 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150732.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:38:20,288 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150735.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:38:51,966 INFO [train.py:904] (6/8) Epoch 15, batch 8650, loss[loss=0.1807, simple_loss=0.2667, pruned_loss=0.04732, over 12187.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2753, pruned_loss=0.04591, over 3052381.95 frames. ], batch size: 248, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:40:03,721 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=150783.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:40:23,674 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150793.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 05:40:39,983 INFO [train.py:904] (6/8) Epoch 15, batch 8700, loss[loss=0.1659, simple_loss=0.2626, pruned_loss=0.03457, over 15259.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2727, pruned_loss=0.04468, over 3057592.45 frames. ], batch size: 190, lr: 4.48e-03, grad_scale: 4.0 2023-04-30 05:41:01,900 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.169e+02 2.751e+02 3.261e+02 6.645e+02, threshold=5.502e+02, percent-clipped=1.0 2023-04-30 05:41:47,785 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7294, 2.0447, 1.7666, 1.8319, 2.3892, 2.0702, 2.3889, 2.5749], device='cuda:6'), covar=tensor([0.0143, 0.0399, 0.0471, 0.0458, 0.0276, 0.0363, 0.0163, 0.0238], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0212, 0.0206, 0.0206, 0.0211, 0.0211, 0.0209, 0.0203], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 05:42:16,603 INFO [train.py:904] (6/8) Epoch 15, batch 8750, loss[loss=0.2303, simple_loss=0.3286, pruned_loss=0.06605, over 16335.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2735, pruned_loss=0.04462, over 3066600.87 frames. ], batch size: 146, lr: 4.48e-03, grad_scale: 4.0 2023-04-30 05:43:52,661 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150893.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:44:06,382 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150900.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:44:08,757 INFO [train.py:904] (6/8) Epoch 15, batch 8800, loss[loss=0.1755, simple_loss=0.2713, pruned_loss=0.0399, over 16761.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2711, pruned_loss=0.04294, over 3067605.93 frames. ], batch size: 124, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:44:28,830 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.691e+02 2.483e+02 3.172e+02 3.684e+02 7.481e+02, threshold=6.344e+02, percent-clipped=2.0 2023-04-30 05:45:06,445 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150930.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:45:12,505 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1878, 3.1073, 3.4260, 1.5790, 3.5068, 3.6819, 2.7594, 2.7785], device='cuda:6'), covar=tensor([0.0841, 0.0272, 0.0202, 0.1350, 0.0084, 0.0128, 0.0454, 0.0431], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0100, 0.0086, 0.0133, 0.0069, 0.0109, 0.0117, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 05:45:50,522 INFO [train.py:904] (6/8) Epoch 15, batch 8850, loss[loss=0.1767, simple_loss=0.2778, pruned_loss=0.03777, over 15418.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.274, pruned_loss=0.04244, over 3060370.39 frames. ], batch size: 191, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:46:10,275 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150961.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:46:46,307 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=150978.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:47:17,194 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4546, 4.2231, 4.5081, 4.6210, 4.7618, 4.2649, 4.7815, 4.7803], device='cuda:6'), covar=tensor([0.1489, 0.1090, 0.1288, 0.0661, 0.0509, 0.1064, 0.0457, 0.0487], device='cuda:6'), in_proj_covar=tensor([0.0542, 0.0673, 0.0794, 0.0687, 0.0518, 0.0541, 0.0550, 0.0636], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 05:47:36,864 INFO [train.py:904] (6/8) Epoch 15, batch 8900, loss[loss=0.1931, simple_loss=0.2893, pruned_loss=0.04844, over 15301.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2739, pruned_loss=0.04196, over 3048537.29 frames. ], batch size: 191, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:47:59,490 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.662e+02 2.270e+02 2.672e+02 3.305e+02 7.174e+02, threshold=5.344e+02, percent-clipped=2.0 2023-04-30 05:48:24,554 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6106, 3.6756, 3.5021, 3.1858, 3.2933, 3.5934, 3.3149, 3.4225], device='cuda:6'), covar=tensor([0.0557, 0.0568, 0.0303, 0.0275, 0.0568, 0.0483, 0.1406, 0.0494], device='cuda:6'), in_proj_covar=tensor([0.0253, 0.0348, 0.0299, 0.0282, 0.0309, 0.0327, 0.0205, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 05:49:42,119 INFO [train.py:904] (6/8) Epoch 15, batch 8950, loss[loss=0.1559, simple_loss=0.2506, pruned_loss=0.03059, over 16405.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2729, pruned_loss=0.04204, over 3052213.62 frames. ], batch size: 68, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:50:30,078 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2621, 1.6130, 1.8963, 2.2020, 2.2871, 2.4356, 1.6193, 2.4827], device='cuda:6'), covar=tensor([0.0175, 0.0431, 0.0286, 0.0270, 0.0281, 0.0167, 0.0477, 0.0114], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0177, 0.0161, 0.0165, 0.0175, 0.0133, 0.0178, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:6') 2023-04-30 05:51:02,410 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=151088.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 05:51:31,003 INFO [train.py:904] (6/8) Epoch 15, batch 9000, loss[loss=0.1647, simple_loss=0.263, pruned_loss=0.03325, over 16850.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2701, pruned_loss=0.04105, over 3045625.83 frames. ], batch size: 90, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:51:31,004 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 05:51:40,827 INFO [train.py:938] (6/8) Epoch 15, validation: loss=0.15, simple_loss=0.2539, pruned_loss=0.02307, over 944034.00 frames. 2023-04-30 05:51:40,828 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 05:52:03,719 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.315e+02 2.164e+02 2.577e+02 3.241e+02 6.734e+02, threshold=5.154e+02, percent-clipped=2.0 2023-04-30 05:53:23,275 INFO [train.py:904] (6/8) Epoch 15, batch 9050, loss[loss=0.1824, simple_loss=0.2625, pruned_loss=0.05114, over 16843.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2709, pruned_loss=0.04169, over 3058069.87 frames. ], batch size: 116, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:54:28,387 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=151184.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:54:51,801 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151193.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:55:09,979 INFO [train.py:904] (6/8) Epoch 15, batch 9100, loss[loss=0.169, simple_loss=0.2561, pruned_loss=0.041, over 12370.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2707, pruned_loss=0.04217, over 3063197.32 frames. ], batch size: 247, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:55:31,070 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.276e+02 2.684e+02 3.281e+02 5.650e+02, threshold=5.369e+02, percent-clipped=3.0 2023-04-30 05:56:36,249 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8172, 2.1259, 1.7529, 1.9017, 2.4833, 2.1896, 2.4472, 2.6664], device='cuda:6'), covar=tensor([0.0123, 0.0386, 0.0473, 0.0451, 0.0252, 0.0326, 0.0171, 0.0214], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0213, 0.0207, 0.0207, 0.0211, 0.0212, 0.0210, 0.0203], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 05:56:44,650 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=151241.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:56:54,394 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=151245.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:57:08,959 INFO [train.py:904] (6/8) Epoch 15, batch 9150, loss[loss=0.164, simple_loss=0.2531, pruned_loss=0.03745, over 12307.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2707, pruned_loss=0.04164, over 3043822.70 frames. ], batch size: 250, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:57:18,577 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=151256.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:58:12,779 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=151282.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:58:39,360 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4075, 4.3285, 4.7854, 4.7659, 4.7668, 4.4525, 4.4288, 4.3329], device='cuda:6'), covar=tensor([0.0316, 0.0690, 0.0423, 0.0401, 0.0407, 0.0393, 0.0883, 0.0445], device='cuda:6'), in_proj_covar=tensor([0.0350, 0.0373, 0.0367, 0.0351, 0.0419, 0.0389, 0.0477, 0.0312], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 05:58:46,508 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=151298.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:58:51,677 INFO [train.py:904] (6/8) Epoch 15, batch 9200, loss[loss=0.1692, simple_loss=0.265, pruned_loss=0.03669, over 16879.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2661, pruned_loss=0.04044, over 3045663.62 frames. ], batch size: 96, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:59:12,238 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.606e+02 2.499e+02 3.414e+02 4.138e+02 7.678e+02, threshold=6.827e+02, percent-clipped=7.0 2023-04-30 06:00:09,402 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=151343.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:00:24,877 INFO [train.py:904] (6/8) Epoch 15, batch 9250, loss[loss=0.1558, simple_loss=0.2412, pruned_loss=0.03521, over 11895.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2656, pruned_loss=0.04049, over 3022254.75 frames. ], batch size: 248, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:00:41,060 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=151359.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:01:07,614 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2083, 4.2771, 4.0735, 3.7948, 3.8148, 4.2096, 3.9191, 3.9016], device='cuda:6'), covar=tensor([0.0527, 0.0503, 0.0293, 0.0279, 0.0715, 0.0418, 0.0668, 0.0638], device='cuda:6'), in_proj_covar=tensor([0.0252, 0.0347, 0.0298, 0.0282, 0.0307, 0.0326, 0.0205, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:01:43,152 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151388.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:02:14,983 INFO [train.py:904] (6/8) Epoch 15, batch 9300, loss[loss=0.1818, simple_loss=0.2591, pruned_loss=0.05226, over 12298.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2645, pruned_loss=0.04031, over 3027659.35 frames. ], batch size: 248, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:02:37,910 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.225e+02 2.604e+02 3.286e+02 5.862e+02, threshold=5.207e+02, percent-clipped=0.0 2023-04-30 06:03:16,629 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9336, 2.0659, 2.3689, 3.2374, 2.1502, 2.2743, 2.2684, 2.1746], device='cuda:6'), covar=tensor([0.1106, 0.3533, 0.2437, 0.0607, 0.4174, 0.2573, 0.3309, 0.3528], device='cuda:6'), in_proj_covar=tensor([0.0369, 0.0406, 0.0343, 0.0312, 0.0419, 0.0464, 0.0374, 0.0475], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:03:21,559 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-30 06:03:29,841 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=151436.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:03:59,110 INFO [train.py:904] (6/8) Epoch 15, batch 9350, loss[loss=0.1974, simple_loss=0.2863, pruned_loss=0.05426, over 16364.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2644, pruned_loss=0.04025, over 3026520.42 frames. ], batch size: 146, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:04:04,559 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0601, 3.2645, 3.1901, 2.2461, 2.9449, 3.2271, 3.1515, 1.9485], device='cuda:6'), covar=tensor([0.0426, 0.0042, 0.0042, 0.0355, 0.0085, 0.0079, 0.0071, 0.0445], device='cuda:6'), in_proj_covar=tensor([0.0129, 0.0072, 0.0072, 0.0130, 0.0086, 0.0094, 0.0084, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 06:05:20,872 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-30 06:05:36,966 INFO [train.py:904] (6/8) Epoch 15, batch 9400, loss[loss=0.1665, simple_loss=0.2689, pruned_loss=0.0321, over 16881.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.265, pruned_loss=0.04013, over 3039528.46 frames. ], batch size: 96, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:05:59,181 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.126e+02 2.557e+02 3.057e+02 4.455e+02, threshold=5.114e+02, percent-clipped=0.0 2023-04-30 06:06:23,568 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-30 06:06:55,071 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=151540.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:07:17,456 INFO [train.py:904] (6/8) Epoch 15, batch 9450, loss[loss=0.1756, simple_loss=0.2642, pruned_loss=0.04352, over 12497.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2666, pruned_loss=0.04, over 3046431.11 frames. ], batch size: 248, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:07:24,797 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151556.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:08:58,275 INFO [train.py:904] (6/8) Epoch 15, batch 9500, loss[loss=0.1614, simple_loss=0.2547, pruned_loss=0.03404, over 16288.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2651, pruned_loss=0.03918, over 3057692.34 frames. ], batch size: 146, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:09:04,140 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=151604.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:09:22,317 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.398e+02 2.144e+02 2.700e+02 3.388e+02 6.768e+02, threshold=5.400e+02, percent-clipped=4.0 2023-04-30 06:10:11,936 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=151638.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:10:44,450 INFO [train.py:904] (6/8) Epoch 15, batch 9550, loss[loss=0.1943, simple_loss=0.2871, pruned_loss=0.05074, over 16934.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2651, pruned_loss=0.03964, over 3051745.18 frames. ], batch size: 109, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:10:49,458 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=151654.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:11:59,339 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=151688.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:12:24,738 INFO [train.py:904] (6/8) Epoch 15, batch 9600, loss[loss=0.183, simple_loss=0.2787, pruned_loss=0.04364, over 16683.00 frames. ], tot_loss[loss=0.174, simple_loss=0.267, pruned_loss=0.04047, over 3054795.57 frames. ], batch size: 134, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:12:44,295 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.311e+02 2.682e+02 3.396e+02 6.074e+02, threshold=5.365e+02, percent-clipped=2.0 2023-04-30 06:14:04,288 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=151749.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:14:11,685 INFO [train.py:904] (6/8) Epoch 15, batch 9650, loss[loss=0.1648, simple_loss=0.2521, pruned_loss=0.03874, over 12457.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.269, pruned_loss=0.04075, over 3067094.42 frames. ], batch size: 248, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:14:41,222 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5999, 3.6435, 3.4701, 3.2157, 3.3161, 3.6042, 3.3826, 3.4059], device='cuda:6'), covar=tensor([0.0568, 0.0649, 0.0287, 0.0247, 0.0570, 0.0447, 0.1224, 0.0456], device='cuda:6'), in_proj_covar=tensor([0.0248, 0.0339, 0.0291, 0.0275, 0.0301, 0.0320, 0.0200, 0.0344], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-04-30 06:14:55,821 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2126, 3.5687, 3.8652, 2.0933, 3.0689, 2.4501, 3.6510, 3.5951], device='cuda:6'), covar=tensor([0.0220, 0.0725, 0.0432, 0.2040, 0.0760, 0.0940, 0.0661, 0.0925], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0145, 0.0157, 0.0144, 0.0136, 0.0123, 0.0136, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 06:15:07,629 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1633, 4.1615, 4.1170, 3.5794, 4.0974, 1.7210, 3.9049, 3.9463], device='cuda:6'), covar=tensor([0.0111, 0.0104, 0.0140, 0.0294, 0.0113, 0.2389, 0.0140, 0.0206], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0126, 0.0171, 0.0155, 0.0146, 0.0187, 0.0160, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:15:57,980 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4260, 4.4848, 4.6413, 4.4670, 4.5191, 5.0238, 4.5995, 4.2655], device='cuda:6'), covar=tensor([0.1328, 0.1945, 0.2042, 0.1958, 0.2505, 0.0981, 0.1364, 0.2244], device='cuda:6'), in_proj_covar=tensor([0.0359, 0.0509, 0.0554, 0.0426, 0.0568, 0.0588, 0.0438, 0.0565], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 06:15:58,106 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=151801.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:15:58,768 INFO [train.py:904] (6/8) Epoch 15, batch 9700, loss[loss=0.1825, simple_loss=0.2746, pruned_loss=0.04522, over 16105.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.268, pruned_loss=0.04055, over 3082542.87 frames. ], batch size: 165, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:16:19,905 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.752e+02 2.225e+02 2.684e+02 3.461e+02 6.863e+02, threshold=5.368e+02, percent-clipped=3.0 2023-04-30 06:17:18,831 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151840.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:17:39,268 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-30 06:17:41,643 INFO [train.py:904] (6/8) Epoch 15, batch 9750, loss[loss=0.1638, simple_loss=0.258, pruned_loss=0.03483, over 16728.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2662, pruned_loss=0.04028, over 3081804.86 frames. ], batch size: 76, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:18:01,668 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=151862.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:18:01,975 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-30 06:18:07,009 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-30 06:18:56,065 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=151888.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:19:18,821 INFO [train.py:904] (6/8) Epoch 15, batch 9800, loss[loss=0.1666, simple_loss=0.2716, pruned_loss=0.03082, over 16643.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2662, pruned_loss=0.03925, over 3097036.09 frames. ], batch size: 89, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:19:40,658 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.231e+02 2.655e+02 3.314e+02 7.232e+02, threshold=5.310e+02, percent-clipped=1.0 2023-04-30 06:20:01,430 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-30 06:20:08,785 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5907, 3.6780, 3.4463, 3.1665, 3.2863, 3.5806, 3.3182, 3.3924], device='cuda:6'), covar=tensor([0.0548, 0.0485, 0.0248, 0.0227, 0.0465, 0.0367, 0.1297, 0.0407], device='cuda:6'), in_proj_covar=tensor([0.0248, 0.0339, 0.0292, 0.0275, 0.0301, 0.0319, 0.0201, 0.0343], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-04-30 06:20:30,612 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151938.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:21:03,075 INFO [train.py:904] (6/8) Epoch 15, batch 9850, loss[loss=0.1999, simple_loss=0.2881, pruned_loss=0.05589, over 16761.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2675, pruned_loss=0.03929, over 3082462.88 frames. ], batch size: 134, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:21:08,475 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151954.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:21:13,125 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4954, 3.4503, 2.7122, 2.1350, 2.1292, 2.2129, 3.5958, 2.9827], device='cuda:6'), covar=tensor([0.2732, 0.0612, 0.1633, 0.2734, 0.2580, 0.1967, 0.0379, 0.1336], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0251, 0.0283, 0.0282, 0.0267, 0.0228, 0.0266, 0.0298], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:22:17,652 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=151986.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:22:57,937 INFO [train.py:904] (6/8) Epoch 15, batch 9900, loss[loss=0.1695, simple_loss=0.2699, pruned_loss=0.03451, over 15229.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2678, pruned_loss=0.03947, over 3060634.39 frames. ], batch size: 191, lr: 4.46e-03, grad_scale: 8.0 2023-04-30 06:22:58,858 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=152002.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:23:24,818 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.244e+02 2.678e+02 3.249e+02 7.284e+02, threshold=5.355e+02, percent-clipped=2.0 2023-04-30 06:24:37,950 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152044.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:24:49,592 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3549, 3.3619, 1.8477, 3.6858, 2.3579, 3.6793, 2.0871, 2.7487], device='cuda:6'), covar=tensor([0.0239, 0.0379, 0.1692, 0.0246, 0.0888, 0.0545, 0.1599, 0.0702], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0161, 0.0182, 0.0135, 0.0165, 0.0197, 0.0193, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 06:24:55,574 INFO [train.py:904] (6/8) Epoch 15, batch 9950, loss[loss=0.1785, simple_loss=0.2698, pruned_loss=0.04357, over 16232.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2696, pruned_loss=0.03961, over 3067524.82 frames. ], batch size: 166, lr: 4.46e-03, grad_scale: 8.0 2023-04-30 06:25:30,388 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-30 06:26:56,977 INFO [train.py:904] (6/8) Epoch 15, batch 10000, loss[loss=0.147, simple_loss=0.2383, pruned_loss=0.02782, over 17255.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2679, pruned_loss=0.03906, over 3070701.47 frames. ], batch size: 52, lr: 4.46e-03, grad_scale: 8.0 2023-04-30 06:27:18,766 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.235e+02 2.836e+02 3.494e+02 9.282e+02, threshold=5.672e+02, percent-clipped=5.0 2023-04-30 06:27:28,989 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-30 06:28:35,911 INFO [train.py:904] (6/8) Epoch 15, batch 10050, loss[loss=0.1894, simple_loss=0.2811, pruned_loss=0.04883, over 16957.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.268, pruned_loss=0.03921, over 3086401.36 frames. ], batch size: 109, lr: 4.46e-03, grad_scale: 4.0 2023-04-30 06:28:45,789 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152157.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:29:19,593 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6981, 4.6740, 4.4169, 4.0199, 4.5020, 1.6847, 4.3367, 4.3243], device='cuda:6'), covar=tensor([0.0081, 0.0084, 0.0191, 0.0272, 0.0119, 0.2528, 0.0132, 0.0196], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0128, 0.0172, 0.0154, 0.0148, 0.0188, 0.0162, 0.0153], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:29:29,072 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6735, 2.6585, 1.9257, 2.8074, 2.0673, 2.8028, 2.0491, 2.3602], device='cuda:6'), covar=tensor([0.0272, 0.0389, 0.1218, 0.0222, 0.0699, 0.0646, 0.1187, 0.0612], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0161, 0.0183, 0.0135, 0.0166, 0.0198, 0.0194, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 06:30:08,503 INFO [train.py:904] (6/8) Epoch 15, batch 10100, loss[loss=0.1989, simple_loss=0.2921, pruned_loss=0.05289, over 16875.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.268, pruned_loss=0.03923, over 3090413.45 frames. ], batch size: 116, lr: 4.46e-03, grad_scale: 4.0 2023-04-30 06:30:28,197 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.142e+02 2.576e+02 3.219e+02 1.174e+03, threshold=5.152e+02, percent-clipped=3.0 2023-04-30 06:30:55,440 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=152225.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:31:01,556 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9483, 4.2391, 4.0874, 4.0819, 3.7140, 3.8505, 3.8470, 4.2374], device='cuda:6'), covar=tensor([0.1117, 0.0889, 0.0911, 0.0739, 0.0885, 0.1547, 0.0909, 0.0991], device='cuda:6'), in_proj_covar=tensor([0.0574, 0.0711, 0.0575, 0.0509, 0.0448, 0.0460, 0.0587, 0.0543], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:31:53,399 INFO [train.py:904] (6/8) Epoch 16, batch 0, loss[loss=0.1904, simple_loss=0.2762, pruned_loss=0.05233, over 16394.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2762, pruned_loss=0.05233, over 16394.00 frames. ], batch size: 36, lr: 4.32e-03, grad_scale: 8.0 2023-04-30 06:31:53,399 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 06:32:00,888 INFO [train.py:938] (6/8) Epoch 16, validation: loss=0.1492, simple_loss=0.2525, pruned_loss=0.02288, over 944034.00 frames. 2023-04-30 06:32:00,888 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 06:32:12,343 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2532, 5.2723, 5.0655, 4.7492, 5.0339, 2.0783, 4.8432, 4.9295], device='cuda:6'), covar=tensor([0.0059, 0.0056, 0.0158, 0.0234, 0.0082, 0.2101, 0.0117, 0.0171], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0127, 0.0171, 0.0153, 0.0147, 0.0187, 0.0161, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:32:29,799 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=152273.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:32:30,276 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-30 06:32:48,529 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=152286.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:33:09,962 INFO [train.py:904] (6/8) Epoch 16, batch 50, loss[loss=0.1679, simple_loss=0.2586, pruned_loss=0.03856, over 17251.00 frames. ], tot_loss[loss=0.1962, simple_loss=0.2789, pruned_loss=0.05673, over 759360.90 frames. ], batch size: 45, lr: 4.32e-03, grad_scale: 1.0 2023-04-30 06:33:29,875 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 2.475e+02 3.034e+02 3.841e+02 6.460e+02, threshold=6.067e+02, percent-clipped=5.0 2023-04-30 06:33:53,586 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=152334.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:34:08,294 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=152344.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:34:18,730 INFO [train.py:904] (6/8) Epoch 16, batch 100, loss[loss=0.1978, simple_loss=0.2734, pruned_loss=0.06106, over 16708.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2722, pruned_loss=0.05124, over 1331227.63 frames. ], batch size: 134, lr: 4.32e-03, grad_scale: 1.0 2023-04-30 06:34:55,624 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-30 06:35:14,871 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=152392.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:35:26,676 INFO [train.py:904] (6/8) Epoch 16, batch 150, loss[loss=0.1778, simple_loss=0.2567, pruned_loss=0.04944, over 16816.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2694, pruned_loss=0.04931, over 1777866.57 frames. ], batch size: 102, lr: 4.32e-03, grad_scale: 1.0 2023-04-30 06:35:48,055 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.712e+02 2.454e+02 2.817e+02 3.397e+02 1.160e+03, threshold=5.634e+02, percent-clipped=3.0 2023-04-30 06:36:08,508 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-04-30 06:36:24,995 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=152445.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:36:35,176 INFO [train.py:904] (6/8) Epoch 16, batch 200, loss[loss=0.2187, simple_loss=0.2892, pruned_loss=0.07409, over 16911.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2711, pruned_loss=0.05067, over 2123374.88 frames. ], batch size: 96, lr: 4.32e-03, grad_scale: 1.0 2023-04-30 06:36:39,734 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0274, 4.0050, 4.3802, 4.3537, 4.4175, 4.0659, 4.0885, 3.9844], device='cuda:6'), covar=tensor([0.0327, 0.0562, 0.0396, 0.0406, 0.0464, 0.0415, 0.0834, 0.0609], device='cuda:6'), in_proj_covar=tensor([0.0350, 0.0378, 0.0370, 0.0355, 0.0420, 0.0393, 0.0479, 0.0315], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 06:36:41,887 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9456, 2.1503, 2.4567, 2.7788, 2.6634, 3.3621, 2.1466, 3.2827], device='cuda:6'), covar=tensor([0.0205, 0.0388, 0.0280, 0.0293, 0.0289, 0.0151, 0.0416, 0.0147], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0179, 0.0164, 0.0168, 0.0177, 0.0134, 0.0180, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:36:42,933 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=152457.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:36:59,198 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5408, 3.6134, 3.3473, 3.0070, 3.2314, 3.5162, 3.3052, 3.3337], device='cuda:6'), covar=tensor([0.0614, 0.0560, 0.0265, 0.0249, 0.0520, 0.0402, 0.1228, 0.0470], device='cuda:6'), in_proj_covar=tensor([0.0260, 0.0358, 0.0305, 0.0290, 0.0318, 0.0335, 0.0210, 0.0364], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:37:44,242 INFO [train.py:904] (6/8) Epoch 16, batch 250, loss[loss=0.1535, simple_loss=0.2472, pruned_loss=0.02989, over 17223.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.27, pruned_loss=0.0507, over 2380695.95 frames. ], batch size: 45, lr: 4.31e-03, grad_scale: 1.0 2023-04-30 06:37:48,060 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=152505.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:37:50,048 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=152506.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:38:05,709 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 2.341e+02 2.836e+02 3.695e+02 5.924e+02, threshold=5.672e+02, percent-clipped=1.0 2023-04-30 06:38:53,036 INFO [train.py:904] (6/8) Epoch 16, batch 300, loss[loss=0.1639, simple_loss=0.2545, pruned_loss=0.03669, over 17222.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2671, pruned_loss=0.04917, over 2588015.78 frames. ], batch size: 45, lr: 4.31e-03, grad_scale: 1.0 2023-04-30 06:39:33,806 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152581.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:39:35,087 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=152582.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:39:43,089 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=152589.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:40:01,546 INFO [train.py:904] (6/8) Epoch 16, batch 350, loss[loss=0.1878, simple_loss=0.2829, pruned_loss=0.04635, over 17047.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2639, pruned_loss=0.04807, over 2750667.35 frames. ], batch size: 50, lr: 4.31e-03, grad_scale: 1.0 2023-04-30 06:40:20,729 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.239e+02 2.608e+02 3.067e+02 7.666e+02, threshold=5.216e+02, percent-clipped=3.0 2023-04-30 06:40:38,465 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152629.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:40:58,679 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=152643.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:41:07,029 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=152650.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:41:08,822 INFO [train.py:904] (6/8) Epoch 16, batch 400, loss[loss=0.1956, simple_loss=0.2874, pruned_loss=0.05191, over 17033.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2623, pruned_loss=0.04754, over 2867771.12 frames. ], batch size: 50, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:41:48,220 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1418, 5.7361, 5.8936, 5.6422, 5.7328, 6.2700, 5.8174, 5.5720], device='cuda:6'), covar=tensor([0.0849, 0.1931, 0.2426, 0.2011, 0.2688, 0.1083, 0.1483, 0.2431], device='cuda:6'), in_proj_covar=tensor([0.0386, 0.0547, 0.0596, 0.0460, 0.0611, 0.0634, 0.0468, 0.0609], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 06:41:49,685 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8377, 3.0030, 2.8672, 5.0872, 4.2559, 4.5089, 1.6820, 3.2800], device='cuda:6'), covar=tensor([0.1250, 0.0696, 0.1083, 0.0169, 0.0196, 0.0356, 0.1498, 0.0664], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0164, 0.0186, 0.0163, 0.0192, 0.0208, 0.0189, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 06:41:52,767 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2752, 2.1157, 2.3261, 3.9288, 2.1537, 2.5317, 2.2237, 2.3336], device='cuda:6'), covar=tensor([0.1251, 0.3697, 0.2613, 0.0604, 0.3869, 0.2336, 0.3666, 0.3035], device='cuda:6'), in_proj_covar=tensor([0.0378, 0.0415, 0.0350, 0.0320, 0.0425, 0.0475, 0.0384, 0.0486], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:42:16,089 INFO [train.py:904] (6/8) Epoch 16, batch 450, loss[loss=0.1557, simple_loss=0.2495, pruned_loss=0.03095, over 17133.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2602, pruned_loss=0.04677, over 2966028.21 frames. ], batch size: 48, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:42:36,230 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.633e+02 2.224e+02 2.572e+02 2.956e+02 7.682e+02, threshold=5.144e+02, percent-clipped=1.0 2023-04-30 06:42:50,542 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=152726.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:43:25,828 INFO [train.py:904] (6/8) Epoch 16, batch 500, loss[loss=0.1731, simple_loss=0.2598, pruned_loss=0.04318, over 17215.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2588, pruned_loss=0.04584, over 3041676.97 frames. ], batch size: 45, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:43:39,618 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0087, 3.9695, 3.9586, 3.4239, 3.9386, 1.9415, 3.7500, 3.4445], device='cuda:6'), covar=tensor([0.0134, 0.0116, 0.0169, 0.0246, 0.0102, 0.2420, 0.0127, 0.0224], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0134, 0.0180, 0.0164, 0.0155, 0.0195, 0.0169, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:44:14,782 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=152787.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:44:33,481 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152801.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:44:34,214 INFO [train.py:904] (6/8) Epoch 16, batch 550, loss[loss=0.2132, simple_loss=0.2818, pruned_loss=0.07233, over 16675.00 frames. ], tot_loss[loss=0.175, simple_loss=0.259, pruned_loss=0.04554, over 3110359.19 frames. ], batch size: 134, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:44:55,529 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.310e+02 2.218e+02 2.743e+02 3.343e+02 5.376e+02, threshold=5.487e+02, percent-clipped=2.0 2023-04-30 06:45:35,407 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8926, 3.7580, 4.2810, 1.9508, 4.4690, 4.5854, 3.2842, 3.3336], device='cuda:6'), covar=tensor([0.0687, 0.0232, 0.0201, 0.1216, 0.0065, 0.0140, 0.0377, 0.0406], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0105, 0.0091, 0.0139, 0.0073, 0.0115, 0.0124, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 06:45:46,473 INFO [train.py:904] (6/8) Epoch 16, batch 600, loss[loss=0.1844, simple_loss=0.2711, pruned_loss=0.04882, over 17036.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2591, pruned_loss=0.0459, over 3147392.35 frames. ], batch size: 50, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:46:05,147 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5420, 2.9533, 2.8504, 1.8757, 2.5129, 1.9803, 3.0631, 3.1305], device='cuda:6'), covar=tensor([0.0268, 0.0840, 0.0696, 0.2236, 0.1105, 0.1147, 0.0641, 0.0848], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0152, 0.0161, 0.0147, 0.0139, 0.0126, 0.0139, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 06:46:25,179 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=152881.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:46:53,568 INFO [train.py:904] (6/8) Epoch 16, batch 650, loss[loss=0.1658, simple_loss=0.2429, pruned_loss=0.04434, over 12100.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2575, pruned_loss=0.04518, over 3188109.79 frames. ], batch size: 246, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:47:14,336 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.748e+02 2.133e+02 2.528e+02 3.105e+02 5.746e+02, threshold=5.056e+02, percent-clipped=1.0 2023-04-30 06:47:30,633 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=152929.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:47:30,700 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=152929.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:47:43,478 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152938.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:47:52,158 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152945.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:48:00,886 INFO [train.py:904] (6/8) Epoch 16, batch 700, loss[loss=0.1713, simple_loss=0.2482, pruned_loss=0.04721, over 15558.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2566, pruned_loss=0.0445, over 3221365.20 frames. ], batch size: 190, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:48:29,322 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 06:48:37,710 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=152977.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:49:10,745 INFO [train.py:904] (6/8) Epoch 16, batch 750, loss[loss=0.1463, simple_loss=0.2319, pruned_loss=0.03038, over 15893.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2569, pruned_loss=0.04465, over 3248551.96 frames. ], batch size: 35, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:49:31,098 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.303e+02 2.598e+02 3.090e+02 5.870e+02, threshold=5.196e+02, percent-clipped=1.0 2023-04-30 06:50:06,315 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-30 06:50:17,790 INFO [train.py:904] (6/8) Epoch 16, batch 800, loss[loss=0.1735, simple_loss=0.2553, pruned_loss=0.04583, over 16742.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2563, pruned_loss=0.04427, over 3270604.66 frames. ], batch size: 89, lr: 4.31e-03, grad_scale: 4.0 2023-04-30 06:51:00,656 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=153082.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:51:25,613 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=153101.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:51:26,408 INFO [train.py:904] (6/8) Epoch 16, batch 850, loss[loss=0.1591, simple_loss=0.2472, pruned_loss=0.03547, over 17216.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2558, pruned_loss=0.04413, over 3283116.37 frames. ], batch size: 45, lr: 4.31e-03, grad_scale: 4.0 2023-04-30 06:51:46,528 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 2.045e+02 2.529e+02 2.957e+02 4.458e+02, threshold=5.058e+02, percent-clipped=0.0 2023-04-30 06:51:46,922 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=153117.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:52:03,061 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5569, 3.5872, 3.3289, 3.0321, 3.1776, 3.5206, 3.2860, 3.3066], device='cuda:6'), covar=tensor([0.0567, 0.0611, 0.0314, 0.0313, 0.0613, 0.0467, 0.1163, 0.0536], device='cuda:6'), in_proj_covar=tensor([0.0277, 0.0380, 0.0325, 0.0311, 0.0341, 0.0359, 0.0222, 0.0389], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:52:32,206 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=153149.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:52:35,389 INFO [train.py:904] (6/8) Epoch 16, batch 900, loss[loss=0.1593, simple_loss=0.2532, pruned_loss=0.03269, over 17106.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2544, pruned_loss=0.04359, over 3289195.07 frames. ], batch size: 49, lr: 4.31e-03, grad_scale: 4.0 2023-04-30 06:52:52,754 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-30 06:52:56,758 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 06:53:11,079 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=153178.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:53:36,569 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4960, 5.3751, 5.2657, 4.8066, 4.8819, 5.3431, 5.3330, 4.9434], device='cuda:6'), covar=tensor([0.0518, 0.0444, 0.0317, 0.0322, 0.1162, 0.0400, 0.0253, 0.0813], device='cuda:6'), in_proj_covar=tensor([0.0280, 0.0383, 0.0327, 0.0314, 0.0344, 0.0362, 0.0223, 0.0392], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:53:43,178 INFO [train.py:904] (6/8) Epoch 16, batch 950, loss[loss=0.187, simple_loss=0.2586, pruned_loss=0.05766, over 16886.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2546, pruned_loss=0.04355, over 3298642.71 frames. ], batch size: 90, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 06:54:04,602 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.182e+02 2.578e+02 3.357e+02 5.954e+02, threshold=5.156e+02, percent-clipped=2.0 2023-04-30 06:54:33,213 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=153238.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:54:34,352 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2226, 5.2555, 5.6892, 5.6422, 5.6449, 5.3379, 5.2626, 5.0896], device='cuda:6'), covar=tensor([0.0297, 0.0481, 0.0324, 0.0410, 0.0448, 0.0323, 0.0874, 0.0370], device='cuda:6'), in_proj_covar=tensor([0.0375, 0.0406, 0.0394, 0.0378, 0.0446, 0.0419, 0.0511, 0.0333], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 06:54:42,478 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=153245.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:54:53,047 INFO [train.py:904] (6/8) Epoch 16, batch 1000, loss[loss=0.153, simple_loss=0.2404, pruned_loss=0.03276, over 17224.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2538, pruned_loss=0.04362, over 3297688.16 frames. ], batch size: 44, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 06:55:29,861 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=153278.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:55:40,267 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=153286.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:55:51,054 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=153293.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:56:02,849 INFO [train.py:904] (6/8) Epoch 16, batch 1050, loss[loss=0.165, simple_loss=0.2494, pruned_loss=0.04025, over 17179.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2533, pruned_loss=0.04373, over 3308278.27 frames. ], batch size: 46, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 06:56:24,074 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0405, 2.1417, 2.5858, 2.8984, 2.7693, 3.3026, 2.3474, 3.4373], device='cuda:6'), covar=tensor([0.0183, 0.0382, 0.0260, 0.0253, 0.0248, 0.0193, 0.0366, 0.0123], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0184, 0.0170, 0.0173, 0.0182, 0.0140, 0.0185, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 06:56:24,685 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.258e+02 2.836e+02 3.297e+02 6.244e+02, threshold=5.672e+02, percent-clipped=2.0 2023-04-30 06:56:55,475 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=153339.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:56:55,558 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8080, 3.8669, 2.9373, 2.2694, 2.5294, 2.3606, 3.9698, 3.4550], device='cuda:6'), covar=tensor([0.2452, 0.0543, 0.1565, 0.2842, 0.2554, 0.1906, 0.0479, 0.1311], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0263, 0.0294, 0.0291, 0.0282, 0.0237, 0.0278, 0.0315], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 06:57:12,820 INFO [train.py:904] (6/8) Epoch 16, batch 1100, loss[loss=0.1659, simple_loss=0.2601, pruned_loss=0.03585, over 17133.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2531, pruned_loss=0.04379, over 3299367.98 frames. ], batch size: 49, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 06:57:54,292 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=153382.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:58:21,562 INFO [train.py:904] (6/8) Epoch 16, batch 1150, loss[loss=0.1583, simple_loss=0.2397, pruned_loss=0.03847, over 15516.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2525, pruned_loss=0.04351, over 3296008.26 frames. ], batch size: 190, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 06:58:42,013 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.113e+02 2.608e+02 3.086e+02 6.684e+02, threshold=5.217e+02, percent-clipped=1.0 2023-04-30 06:58:46,832 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=153421.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:58:58,552 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=153430.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:59:27,881 INFO [train.py:904] (6/8) Epoch 16, batch 1200, loss[loss=0.1704, simple_loss=0.2423, pruned_loss=0.04926, over 16455.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2526, pruned_loss=0.04296, over 3298660.74 frames. ], batch size: 75, lr: 4.30e-03, grad_scale: 8.0 2023-04-30 06:59:57,478 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=153473.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:00:10,715 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=153482.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:00:34,435 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2815, 4.0114, 4.1378, 4.4407, 4.4845, 4.1134, 4.4317, 4.5264], device='cuda:6'), covar=tensor([0.1482, 0.1410, 0.1803, 0.0878, 0.0808, 0.1254, 0.1933, 0.0864], device='cuda:6'), in_proj_covar=tensor([0.0611, 0.0759, 0.0899, 0.0770, 0.0578, 0.0603, 0.0618, 0.0712], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 07:00:37,162 INFO [train.py:904] (6/8) Epoch 16, batch 1250, loss[loss=0.1764, simple_loss=0.263, pruned_loss=0.04487, over 16642.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.252, pruned_loss=0.04336, over 3298799.78 frames. ], batch size: 57, lr: 4.30e-03, grad_scale: 8.0 2023-04-30 07:00:57,395 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.717e+02 2.233e+02 2.659e+02 3.231e+02 7.143e+02, threshold=5.319e+02, percent-clipped=2.0 2023-04-30 07:01:11,983 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9651, 3.5038, 2.9851, 5.1116, 4.2461, 4.5012, 1.6791, 3.2674], device='cuda:6'), covar=tensor([0.1262, 0.0579, 0.1004, 0.0179, 0.0229, 0.0391, 0.1529, 0.0726], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0166, 0.0187, 0.0168, 0.0197, 0.0212, 0.0190, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 07:01:43,332 INFO [train.py:904] (6/8) Epoch 16, batch 1300, loss[loss=0.1603, simple_loss=0.2373, pruned_loss=0.04168, over 16212.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2518, pruned_loss=0.04326, over 3287119.51 frames. ], batch size: 165, lr: 4.30e-03, grad_scale: 8.0 2023-04-30 07:02:38,350 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2853, 2.2396, 2.3678, 4.1053, 2.2448, 2.6210, 2.2756, 2.4332], device='cuda:6'), covar=tensor([0.1334, 0.3410, 0.2722, 0.0525, 0.3815, 0.2422, 0.3673, 0.3037], device='cuda:6'), in_proj_covar=tensor([0.0384, 0.0420, 0.0353, 0.0325, 0.0428, 0.0484, 0.0389, 0.0493], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 07:02:43,875 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.68 vs. limit=5.0 2023-04-30 07:02:52,642 INFO [train.py:904] (6/8) Epoch 16, batch 1350, loss[loss=0.1942, simple_loss=0.2845, pruned_loss=0.05191, over 17043.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2525, pruned_loss=0.04386, over 3290077.37 frames. ], batch size: 55, lr: 4.30e-03, grad_scale: 8.0 2023-04-30 07:03:12,844 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.319e+02 2.697e+02 3.051e+02 7.500e+02, threshold=5.394e+02, percent-clipped=1.0 2023-04-30 07:03:33,324 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-30 07:03:37,064 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=153634.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:04:02,079 INFO [train.py:904] (6/8) Epoch 16, batch 1400, loss[loss=0.1706, simple_loss=0.2438, pruned_loss=0.04871, over 16704.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2526, pruned_loss=0.04366, over 3293711.51 frames. ], batch size: 134, lr: 4.30e-03, grad_scale: 8.0 2023-04-30 07:04:38,486 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6041, 3.6634, 4.1071, 2.1348, 4.2212, 4.2725, 3.1557, 3.2130], device='cuda:6'), covar=tensor([0.0748, 0.0220, 0.0185, 0.1098, 0.0082, 0.0165, 0.0401, 0.0412], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0105, 0.0092, 0.0138, 0.0073, 0.0117, 0.0124, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 07:04:47,756 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-30 07:05:12,110 INFO [train.py:904] (6/8) Epoch 16, batch 1450, loss[loss=0.1637, simple_loss=0.259, pruned_loss=0.03417, over 17045.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2523, pruned_loss=0.04354, over 3296858.91 frames. ], batch size: 50, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 07:05:34,100 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.274e+02 2.601e+02 3.378e+02 7.170e+02, threshold=5.202e+02, percent-clipped=2.0 2023-04-30 07:05:54,632 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5215, 2.4640, 2.0446, 2.2429, 2.8851, 2.6011, 3.2563, 3.1816], device='cuda:6'), covar=tensor([0.0138, 0.0419, 0.0559, 0.0461, 0.0267, 0.0373, 0.0238, 0.0253], device='cuda:6'), in_proj_covar=tensor([0.0186, 0.0227, 0.0219, 0.0220, 0.0228, 0.0229, 0.0232, 0.0223], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 07:06:10,660 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-30 07:06:22,460 INFO [train.py:904] (6/8) Epoch 16, batch 1500, loss[loss=0.1886, simple_loss=0.2621, pruned_loss=0.05752, over 16557.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2523, pruned_loss=0.04342, over 3303100.03 frames. ], batch size: 68, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 07:06:38,220 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-30 07:06:47,726 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0109, 2.0312, 2.2236, 3.5215, 2.0532, 2.3093, 2.1726, 2.1592], device='cuda:6'), covar=tensor([0.1299, 0.3735, 0.2592, 0.0632, 0.3943, 0.2527, 0.3489, 0.3415], device='cuda:6'), in_proj_covar=tensor([0.0385, 0.0421, 0.0353, 0.0325, 0.0428, 0.0485, 0.0391, 0.0495], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 07:06:51,053 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=153773.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:06:56,072 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=153777.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:07:26,437 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1904, 5.1959, 5.7526, 5.6775, 5.7043, 5.3208, 5.2440, 5.1395], device='cuda:6'), covar=tensor([0.0293, 0.0512, 0.0326, 0.0477, 0.0466, 0.0340, 0.0964, 0.0367], device='cuda:6'), in_proj_covar=tensor([0.0382, 0.0413, 0.0401, 0.0383, 0.0453, 0.0428, 0.0521, 0.0340], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 07:07:30,670 INFO [train.py:904] (6/8) Epoch 16, batch 1550, loss[loss=0.2069, simple_loss=0.2741, pruned_loss=0.06987, over 16413.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2536, pruned_loss=0.04458, over 3305125.68 frames. ], batch size: 146, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 07:07:53,742 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.189e+02 2.739e+02 3.070e+02 6.481e+02, threshold=5.478e+02, percent-clipped=4.0 2023-04-30 07:07:58,990 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=153821.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:08:39,792 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8115, 3.9826, 2.9849, 2.2423, 2.7322, 2.4381, 4.4738, 3.5445], device='cuda:6'), covar=tensor([0.2710, 0.0824, 0.1726, 0.2669, 0.2636, 0.1928, 0.0396, 0.1320], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0263, 0.0293, 0.0291, 0.0282, 0.0237, 0.0279, 0.0315], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:08:40,377 INFO [train.py:904] (6/8) Epoch 16, batch 1600, loss[loss=0.1937, simple_loss=0.2901, pruned_loss=0.0487, over 17023.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2552, pruned_loss=0.04525, over 3308334.49 frames. ], batch size: 55, lr: 4.30e-03, grad_scale: 8.0 2023-04-30 07:09:07,629 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0551, 4.1337, 4.6485, 2.3169, 4.7197, 4.8395, 3.5012, 3.7063], device='cuda:6'), covar=tensor([0.0669, 0.0185, 0.0162, 0.1101, 0.0093, 0.0166, 0.0363, 0.0359], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0105, 0.0092, 0.0137, 0.0073, 0.0117, 0.0124, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 07:09:47,765 INFO [train.py:904] (6/8) Epoch 16, batch 1650, loss[loss=0.1743, simple_loss=0.259, pruned_loss=0.04485, over 16793.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2561, pruned_loss=0.04496, over 3312983.07 frames. ], batch size: 39, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:10:09,057 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.581e+02 2.549e+02 2.974e+02 3.703e+02 7.236e+02, threshold=5.949e+02, percent-clipped=5.0 2023-04-30 07:10:15,988 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=153923.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:10:17,428 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-30 07:10:32,362 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=153934.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:10:47,396 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-30 07:10:56,045 INFO [train.py:904] (6/8) Epoch 16, batch 1700, loss[loss=0.1833, simple_loss=0.2747, pruned_loss=0.04592, over 17061.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.259, pruned_loss=0.04559, over 3309223.29 frames. ], batch size: 55, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:11:38,410 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=153982.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:11:40,929 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=153984.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:12:09,295 INFO [train.py:904] (6/8) Epoch 16, batch 1750, loss[loss=0.1841, simple_loss=0.259, pruned_loss=0.05462, over 16752.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2598, pruned_loss=0.04561, over 3311594.35 frames. ], batch size: 124, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:12:33,136 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.607e+02 2.261e+02 2.728e+02 3.333e+02 9.430e+02, threshold=5.456e+02, percent-clipped=2.0 2023-04-30 07:13:15,147 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154049.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 07:13:18,923 INFO [train.py:904] (6/8) Epoch 16, batch 1800, loss[loss=0.1825, simple_loss=0.2766, pruned_loss=0.04418, over 17049.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2606, pruned_loss=0.04538, over 3318133.73 frames. ], batch size: 53, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:13:53,233 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154076.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:13:54,372 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154077.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:14:03,106 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154084.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:14:28,062 INFO [train.py:904] (6/8) Epoch 16, batch 1850, loss[loss=0.1731, simple_loss=0.2675, pruned_loss=0.03936, over 17106.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2618, pruned_loss=0.04578, over 3307673.14 frames. ], batch size: 49, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:14:39,353 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154110.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 07:14:50,217 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.254e+02 2.561e+02 3.083e+02 7.849e+02, threshold=5.121e+02, percent-clipped=3.0 2023-04-30 07:14:59,481 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154125.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:15:15,709 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154137.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 07:15:25,839 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154145.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:15:33,361 INFO [train.py:904] (6/8) Epoch 16, batch 1900, loss[loss=0.172, simple_loss=0.2463, pruned_loss=0.04887, over 16856.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2609, pruned_loss=0.04524, over 3314235.80 frames. ], batch size: 96, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:15:55,678 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1230, 5.7544, 5.9232, 5.6132, 5.8199, 6.2352, 5.8005, 5.4581], device='cuda:6'), covar=tensor([0.0924, 0.1755, 0.2317, 0.2056, 0.2419, 0.0935, 0.1435, 0.2490], device='cuda:6'), in_proj_covar=tensor([0.0394, 0.0564, 0.0611, 0.0472, 0.0632, 0.0645, 0.0479, 0.0626], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:16:41,971 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6786, 4.5817, 4.5754, 4.3209, 4.3076, 4.6627, 4.4870, 4.3962], device='cuda:6'), covar=tensor([0.0641, 0.0700, 0.0312, 0.0298, 0.0839, 0.0479, 0.0454, 0.0648], device='cuda:6'), in_proj_covar=tensor([0.0284, 0.0393, 0.0333, 0.0321, 0.0350, 0.0370, 0.0228, 0.0401], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:16:44,655 INFO [train.py:904] (6/8) Epoch 16, batch 1950, loss[loss=0.1887, simple_loss=0.2704, pruned_loss=0.0535, over 16469.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2608, pruned_loss=0.04496, over 3314530.02 frames. ], batch size: 146, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:17:04,525 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.235e+02 2.551e+02 3.026e+02 6.742e+02, threshold=5.103e+02, percent-clipped=2.0 2023-04-30 07:17:10,507 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154221.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:17:51,473 INFO [train.py:904] (6/8) Epoch 16, batch 2000, loss[loss=0.1855, simple_loss=0.2515, pruned_loss=0.05971, over 16861.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2601, pruned_loss=0.04508, over 3307284.06 frames. ], batch size: 109, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:18:27,771 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154279.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:18:33,400 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154282.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 07:18:38,787 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154286.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:18:44,565 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154291.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:18:59,380 INFO [train.py:904] (6/8) Epoch 16, batch 2050, loss[loss=0.1725, simple_loss=0.265, pruned_loss=0.04005, over 17109.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2597, pruned_loss=0.04503, over 3314765.66 frames. ], batch size: 47, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:19:17,354 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2023-04-30 07:19:19,608 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.703e+02 2.241e+02 2.674e+02 3.116e+02 4.900e+02, threshold=5.347e+02, percent-clipped=0.0 2023-04-30 07:19:31,201 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 07:19:46,281 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154337.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 07:20:01,041 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154347.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:20:06,432 INFO [train.py:904] (6/8) Epoch 16, batch 2100, loss[loss=0.1774, simple_loss=0.2729, pruned_loss=0.04099, over 17129.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2608, pruned_loss=0.0452, over 3318726.27 frames. ], batch size: 47, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:20:06,950 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154352.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:20:08,290 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7361, 2.6834, 2.3644, 2.6345, 2.9842, 2.8710, 3.4139, 3.2610], device='cuda:6'), covar=tensor([0.0135, 0.0363, 0.0458, 0.0389, 0.0263, 0.0353, 0.0215, 0.0230], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0225, 0.0218, 0.0218, 0.0227, 0.0229, 0.0232, 0.0223], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 07:20:57,852 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2996, 3.9845, 4.5390, 2.2503, 4.7086, 4.8094, 3.4986, 3.7892], device='cuda:6'), covar=tensor([0.0607, 0.0262, 0.0209, 0.1129, 0.0064, 0.0139, 0.0392, 0.0330], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0106, 0.0093, 0.0138, 0.0074, 0.0119, 0.0126, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 07:21:09,683 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154398.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 07:21:14,694 INFO [train.py:904] (6/8) Epoch 16, batch 2150, loss[loss=0.1603, simple_loss=0.2549, pruned_loss=0.03289, over 17109.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2622, pruned_loss=0.04635, over 3312112.61 frames. ], batch size: 49, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:21:16,337 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8312, 4.2130, 4.1112, 3.0072, 3.5621, 4.1065, 3.9112, 2.0441], device='cuda:6'), covar=tensor([0.0500, 0.0097, 0.0074, 0.0406, 0.0161, 0.0162, 0.0129, 0.0629], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0077, 0.0076, 0.0134, 0.0091, 0.0100, 0.0089, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:21:19,319 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154405.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 07:21:36,518 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.770e+02 2.339e+02 2.706e+02 3.339e+02 5.041e+02, threshold=5.411e+02, percent-clipped=0.0 2023-04-30 07:21:57,912 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154432.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 07:22:07,255 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154440.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:22:14,930 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.83 vs. limit=5.0 2023-04-30 07:22:25,081 INFO [train.py:904] (6/8) Epoch 16, batch 2200, loss[loss=0.1953, simple_loss=0.2887, pruned_loss=0.05099, over 16620.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2626, pruned_loss=0.04625, over 3317377.57 frames. ], batch size: 62, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:22:55,732 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154474.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:23:32,242 INFO [train.py:904] (6/8) Epoch 16, batch 2250, loss[loss=0.2018, simple_loss=0.2861, pruned_loss=0.05875, over 16616.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2636, pruned_loss=0.04639, over 3314957.35 frames. ], batch size: 57, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:23:55,133 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.236e+02 2.597e+02 3.012e+02 8.725e+02, threshold=5.194e+02, percent-clipped=2.0 2023-04-30 07:24:18,880 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154535.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:24:25,928 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-04-30 07:24:31,594 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6326, 4.4750, 4.6675, 4.8289, 4.9619, 4.4045, 4.8653, 4.9584], device='cuda:6'), covar=tensor([0.1564, 0.1302, 0.1380, 0.0666, 0.0605, 0.1106, 0.1236, 0.0768], device='cuda:6'), in_proj_covar=tensor([0.0613, 0.0762, 0.0904, 0.0769, 0.0577, 0.0603, 0.0614, 0.0714], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 07:24:40,018 INFO [train.py:904] (6/8) Epoch 16, batch 2300, loss[loss=0.2371, simple_loss=0.3092, pruned_loss=0.08255, over 11782.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2635, pruned_loss=0.0462, over 3319457.89 frames. ], batch size: 247, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:25:16,755 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154577.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 07:25:20,026 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154579.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:25:51,249 INFO [train.py:904] (6/8) Epoch 16, batch 2350, loss[loss=0.1733, simple_loss=0.2643, pruned_loss=0.0412, over 17116.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2642, pruned_loss=0.04701, over 3310095.02 frames. ], batch size: 47, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:26:11,866 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 2.266e+02 2.563e+02 3.089e+02 4.962e+02, threshold=5.126e+02, percent-clipped=0.0 2023-04-30 07:26:25,329 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154627.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:26:26,726 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154628.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:26:45,870 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154642.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:26:52,320 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154647.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:26:57,937 INFO [train.py:904] (6/8) Epoch 16, batch 2400, loss[loss=0.1959, simple_loss=0.2694, pruned_loss=0.06116, over 16804.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2652, pruned_loss=0.04764, over 3315851.98 frames. ], batch size: 83, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:27:51,022 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154689.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:27:55,555 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154693.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 07:28:06,455 INFO [train.py:904] (6/8) Epoch 16, batch 2450, loss[loss=0.1745, simple_loss=0.2548, pruned_loss=0.04704, over 16675.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2661, pruned_loss=0.04741, over 3314082.08 frames. ], batch size: 134, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:28:12,098 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154705.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 07:28:28,917 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.609e+02 2.204e+02 2.505e+02 3.030e+02 6.585e+02, threshold=5.010e+02, percent-clipped=3.0 2023-04-30 07:28:31,778 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154720.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:28:50,330 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154732.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:28:59,894 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154740.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:29:16,066 INFO [train.py:904] (6/8) Epoch 16, batch 2500, loss[loss=0.223, simple_loss=0.2959, pruned_loss=0.07508, over 16520.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2655, pruned_loss=0.04741, over 3320750.71 frames. ], batch size: 146, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:29:18,086 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154753.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 07:29:53,790 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154780.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:29:55,079 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154781.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:30:05,087 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154788.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:30:24,142 INFO [train.py:904] (6/8) Epoch 16, batch 2550, loss[loss=0.1953, simple_loss=0.2731, pruned_loss=0.05875, over 16709.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.265, pruned_loss=0.04722, over 3326415.22 frames. ], batch size: 134, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:30:47,017 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.496e+02 2.136e+02 2.491e+02 3.165e+02 6.684e+02, threshold=4.981e+02, percent-clipped=5.0 2023-04-30 07:31:02,556 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154830.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:31:14,627 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0593, 3.0464, 3.0878, 2.1610, 2.8940, 3.2426, 2.9826, 1.9213], device='cuda:6'), covar=tensor([0.0452, 0.0119, 0.0064, 0.0370, 0.0119, 0.0088, 0.0106, 0.0403], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0077, 0.0077, 0.0133, 0.0090, 0.0099, 0.0089, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:31:32,683 INFO [train.py:904] (6/8) Epoch 16, batch 2600, loss[loss=0.1473, simple_loss=0.2416, pruned_loss=0.02648, over 17141.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2655, pruned_loss=0.04717, over 3310001.45 frames. ], batch size: 47, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:32:07,965 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154877.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 07:32:18,098 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0473, 5.4426, 5.6370, 5.2818, 5.3414, 5.9791, 5.4953, 5.2165], device='cuda:6'), covar=tensor([0.1030, 0.1808, 0.1927, 0.2046, 0.2659, 0.0963, 0.1296, 0.2185], device='cuda:6'), in_proj_covar=tensor([0.0398, 0.0567, 0.0614, 0.0480, 0.0635, 0.0648, 0.0484, 0.0632], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:32:40,408 INFO [train.py:904] (6/8) Epoch 16, batch 2650, loss[loss=0.1852, simple_loss=0.2742, pruned_loss=0.04813, over 16248.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2662, pruned_loss=0.04637, over 3318567.52 frames. ], batch size: 165, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:33:01,394 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.264e+02 2.671e+02 3.159e+02 6.185e+02, threshold=5.341e+02, percent-clipped=3.0 2023-04-30 07:33:12,794 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154925.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:33:35,841 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154942.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:33:43,642 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154947.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:33:49,086 INFO [train.py:904] (6/8) Epoch 16, batch 2700, loss[loss=0.1469, simple_loss=0.241, pruned_loss=0.02641, over 17186.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2662, pruned_loss=0.04588, over 3313302.51 frames. ], batch size: 44, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:34:08,793 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.14 vs. limit=5.0 2023-04-30 07:34:22,470 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154976.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:34:32,700 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154984.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:34:41,400 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154990.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:34:45,813 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154993.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 07:34:47,557 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-30 07:34:47,984 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154995.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:34:56,798 INFO [train.py:904] (6/8) Epoch 16, batch 2750, loss[loss=0.1868, simple_loss=0.2623, pruned_loss=0.05565, over 16810.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2651, pruned_loss=0.04556, over 3318086.68 frames. ], batch size: 102, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:35:18,258 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.250e+02 2.149e+02 2.425e+02 2.902e+02 6.172e+02, threshold=4.850e+02, percent-clipped=1.0 2023-04-30 07:35:46,204 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=155037.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:35:50,514 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=155041.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 07:36:04,754 INFO [train.py:904] (6/8) Epoch 16, batch 2800, loss[loss=0.1751, simple_loss=0.2584, pruned_loss=0.04588, over 16865.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2648, pruned_loss=0.04519, over 3327815.65 frames. ], batch size: 96, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:36:29,066 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=155069.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:36:38,305 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=155076.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:37:15,724 INFO [train.py:904] (6/8) Epoch 16, batch 2850, loss[loss=0.1827, simple_loss=0.2798, pruned_loss=0.0428, over 17112.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2642, pruned_loss=0.04508, over 3330764.23 frames. ], batch size: 48, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:37:36,420 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.487e+02 2.238e+02 2.628e+02 3.275e+02 9.061e+02, threshold=5.256e+02, percent-clipped=3.0 2023-04-30 07:37:54,365 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=155130.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:37:54,493 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=155130.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:37:59,596 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1042, 5.0469, 4.9525, 4.6002, 4.5778, 5.0339, 4.9683, 4.6922], device='cuda:6'), covar=tensor([0.0584, 0.0439, 0.0294, 0.0322, 0.1012, 0.0409, 0.0329, 0.0720], device='cuda:6'), in_proj_covar=tensor([0.0288, 0.0399, 0.0340, 0.0328, 0.0357, 0.0378, 0.0232, 0.0407], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:38:23,945 INFO [train.py:904] (6/8) Epoch 16, batch 2900, loss[loss=0.2658, simple_loss=0.3133, pruned_loss=0.1091, over 11634.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2632, pruned_loss=0.04527, over 3323487.83 frames. ], batch size: 246, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:39:01,145 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=155178.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:39:21,389 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6943, 3.7460, 2.8538, 2.2544, 2.4738, 2.3131, 3.8391, 3.3396], device='cuda:6'), covar=tensor([0.2386, 0.0598, 0.1605, 0.2817, 0.2568, 0.1865, 0.0481, 0.1283], device='cuda:6'), in_proj_covar=tensor([0.0311, 0.0264, 0.0292, 0.0292, 0.0285, 0.0238, 0.0279, 0.0318], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:39:33,218 INFO [train.py:904] (6/8) Epoch 16, batch 2950, loss[loss=0.1692, simple_loss=0.2572, pruned_loss=0.04058, over 16048.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2622, pruned_loss=0.04594, over 3322636.26 frames. ], batch size: 35, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:39:54,114 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.648e+02 2.372e+02 2.734e+02 3.300e+02 7.430e+02, threshold=5.468e+02, percent-clipped=3.0 2023-04-30 07:40:20,096 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2023-04-30 07:40:40,759 INFO [train.py:904] (6/8) Epoch 16, batch 3000, loss[loss=0.1797, simple_loss=0.2788, pruned_loss=0.04031, over 17122.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2627, pruned_loss=0.04671, over 3319402.42 frames. ], batch size: 48, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:40:40,760 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 07:40:49,854 INFO [train.py:938] (6/8) Epoch 16, validation: loss=0.1368, simple_loss=0.2429, pruned_loss=0.01541, over 944034.00 frames. 2023-04-30 07:40:49,854 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 07:41:00,851 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5048, 5.8794, 5.6228, 5.7288, 5.3118, 5.3077, 5.3004, 6.0454], device='cuda:6'), covar=tensor([0.1313, 0.0939, 0.0979, 0.0748, 0.0959, 0.0680, 0.1117, 0.0848], device='cuda:6'), in_proj_covar=tensor([0.0641, 0.0791, 0.0644, 0.0570, 0.0499, 0.0505, 0.0658, 0.0606], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 07:41:34,674 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=155284.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:41:59,650 INFO [train.py:904] (6/8) Epoch 16, batch 3050, loss[loss=0.1704, simple_loss=0.2494, pruned_loss=0.04571, over 16783.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.2624, pruned_loss=0.04653, over 3328628.62 frames. ], batch size: 83, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:42:03,030 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4460, 4.4316, 4.3872, 4.1554, 4.0898, 4.4462, 4.2334, 4.2006], device='cuda:6'), covar=tensor([0.0631, 0.0604, 0.0315, 0.0314, 0.0912, 0.0477, 0.0516, 0.0637], device='cuda:6'), in_proj_covar=tensor([0.0286, 0.0396, 0.0337, 0.0326, 0.0355, 0.0377, 0.0231, 0.0405], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:42:21,041 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.286e+02 2.843e+02 3.408e+02 8.038e+02, threshold=5.686e+02, percent-clipped=2.0 2023-04-30 07:42:42,353 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=155332.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:42:42,368 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=155332.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:43:09,367 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9351, 3.8384, 4.3014, 2.0441, 4.4909, 4.5051, 3.2671, 3.3309], device='cuda:6'), covar=tensor([0.0726, 0.0241, 0.0245, 0.1236, 0.0067, 0.0180, 0.0424, 0.0445], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0106, 0.0093, 0.0139, 0.0075, 0.0120, 0.0126, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 07:43:10,099 INFO [train.py:904] (6/8) Epoch 16, batch 3100, loss[loss=0.1663, simple_loss=0.2595, pruned_loss=0.03651, over 17055.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2612, pruned_loss=0.04609, over 3333872.64 frames. ], batch size: 55, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:43:42,855 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=155376.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:44:01,987 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-30 07:44:17,693 INFO [train.py:904] (6/8) Epoch 16, batch 3150, loss[loss=0.1542, simple_loss=0.243, pruned_loss=0.03271, over 17243.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2607, pruned_loss=0.04545, over 3335715.22 frames. ], batch size: 44, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:44:39,886 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.743e+02 2.373e+02 2.704e+02 3.278e+02 4.749e+02, threshold=5.408e+02, percent-clipped=0.0 2023-04-30 07:44:47,892 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=155424.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:44:49,092 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=155425.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:45:27,256 INFO [train.py:904] (6/8) Epoch 16, batch 3200, loss[loss=0.1654, simple_loss=0.2605, pruned_loss=0.03512, over 17052.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2601, pruned_loss=0.04509, over 3341795.93 frames. ], batch size: 50, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:45:34,186 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=155457.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:46:36,135 INFO [train.py:904] (6/8) Epoch 16, batch 3250, loss[loss=0.1445, simple_loss=0.2337, pruned_loss=0.02765, over 17200.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2603, pruned_loss=0.04541, over 3329551.30 frames. ], batch size: 44, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:46:58,460 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.619e+02 2.206e+02 2.546e+02 3.042e+02 5.764e+02, threshold=5.092e+02, percent-clipped=1.0 2023-04-30 07:46:58,913 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=155518.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:47:45,922 INFO [train.py:904] (6/8) Epoch 16, batch 3300, loss[loss=0.1679, simple_loss=0.2489, pruned_loss=0.04345, over 16695.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2617, pruned_loss=0.04607, over 3320639.55 frames. ], batch size: 89, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:48:19,976 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1734, 3.0593, 3.3180, 1.7846, 3.4360, 3.4275, 2.7648, 2.5416], device='cuda:6'), covar=tensor([0.0847, 0.0268, 0.0211, 0.1175, 0.0106, 0.0229, 0.0437, 0.0495], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0107, 0.0094, 0.0139, 0.0075, 0.0121, 0.0127, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 07:48:22,798 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0017, 5.3465, 5.1196, 5.1414, 4.9281, 4.7795, 4.7445, 5.4835], device='cuda:6'), covar=tensor([0.1169, 0.0884, 0.1000, 0.0834, 0.0780, 0.0954, 0.1084, 0.0917], device='cuda:6'), in_proj_covar=tensor([0.0640, 0.0796, 0.0642, 0.0573, 0.0497, 0.0506, 0.0660, 0.0609], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 07:48:32,642 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7549, 3.9424, 3.0609, 2.3235, 2.6109, 2.4275, 4.1190, 3.5177], device='cuda:6'), covar=tensor([0.2611, 0.0551, 0.1569, 0.2625, 0.2506, 0.1847, 0.0431, 0.1226], device='cuda:6'), in_proj_covar=tensor([0.0315, 0.0267, 0.0294, 0.0295, 0.0290, 0.0241, 0.0281, 0.0322], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:48:56,952 INFO [train.py:904] (6/8) Epoch 16, batch 3350, loss[loss=0.1595, simple_loss=0.2581, pruned_loss=0.03041, over 17032.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2612, pruned_loss=0.0454, over 3327382.32 frames. ], batch size: 55, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:48:58,894 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-30 07:49:01,210 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=155605.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:49:19,483 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.838e+02 2.398e+02 2.739e+02 3.162e+02 6.710e+02, threshold=5.477e+02, percent-clipped=3.0 2023-04-30 07:49:27,132 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7582, 3.9388, 2.1257, 4.4283, 2.9598, 4.3844, 2.3760, 3.0547], device='cuda:6'), covar=tensor([0.0262, 0.0337, 0.1764, 0.0266, 0.0764, 0.0457, 0.1486, 0.0724], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0175, 0.0193, 0.0156, 0.0174, 0.0218, 0.0202, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 07:49:39,719 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=155632.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:49:58,892 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2271, 4.1884, 4.1603, 3.6095, 4.1711, 1.6959, 3.9326, 3.7755], device='cuda:6'), covar=tensor([0.0104, 0.0097, 0.0143, 0.0276, 0.0086, 0.2713, 0.0126, 0.0206], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0144, 0.0192, 0.0177, 0.0165, 0.0203, 0.0181, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 07:50:08,445 INFO [train.py:904] (6/8) Epoch 16, batch 3400, loss[loss=0.1385, simple_loss=0.2302, pruned_loss=0.0234, over 17205.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2617, pruned_loss=0.04522, over 3323361.16 frames. ], batch size: 46, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:50:16,629 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=155657.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:50:28,050 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=155666.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:50:49,342 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=155680.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:51:19,270 INFO [train.py:904] (6/8) Epoch 16, batch 3450, loss[loss=0.181, simple_loss=0.2658, pruned_loss=0.04813, over 16628.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2611, pruned_loss=0.04547, over 3322271.60 frames. ], batch size: 75, lr: 4.27e-03, grad_scale: 16.0 2023-04-30 07:51:41,374 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.296e+02 2.647e+02 3.295e+02 8.789e+02, threshold=5.295e+02, percent-clipped=1.0 2023-04-30 07:51:41,807 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=155718.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:51:52,663 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=155725.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:52:28,014 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2132, 4.0516, 4.2370, 4.3924, 4.4802, 4.0241, 4.2126, 4.5010], device='cuda:6'), covar=tensor([0.1465, 0.1065, 0.1263, 0.0654, 0.0568, 0.1323, 0.2340, 0.0573], device='cuda:6'), in_proj_covar=tensor([0.0633, 0.0785, 0.0926, 0.0798, 0.0595, 0.0624, 0.0629, 0.0739], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 07:52:28,777 INFO [train.py:904] (6/8) Epoch 16, batch 3500, loss[loss=0.2133, simple_loss=0.2849, pruned_loss=0.07088, over 16690.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2604, pruned_loss=0.04542, over 3318870.48 frames. ], batch size: 134, lr: 4.27e-03, grad_scale: 16.0 2023-04-30 07:52:29,291 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8237, 4.1325, 3.0481, 2.2364, 2.7544, 2.4944, 4.4974, 3.5625], device='cuda:6'), covar=tensor([0.2749, 0.0722, 0.1710, 0.2770, 0.2746, 0.2001, 0.0431, 0.1423], device='cuda:6'), in_proj_covar=tensor([0.0315, 0.0266, 0.0295, 0.0295, 0.0289, 0.0241, 0.0282, 0.0322], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:52:49,882 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.95 vs. limit=5.0 2023-04-30 07:52:58,253 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=155773.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:53:14,794 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=155784.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:53:25,675 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.13 vs. limit=5.0 2023-04-30 07:53:40,573 INFO [train.py:904] (6/8) Epoch 16, batch 3550, loss[loss=0.1764, simple_loss=0.2566, pruned_loss=0.04809, over 16577.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2591, pruned_loss=0.04476, over 3314943.55 frames. ], batch size: 75, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:53:47,226 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3752, 5.2722, 5.2134, 4.7960, 4.8311, 5.2546, 5.1584, 4.9298], device='cuda:6'), covar=tensor([0.0507, 0.0528, 0.0292, 0.0308, 0.1102, 0.0444, 0.0307, 0.0746], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0404, 0.0342, 0.0331, 0.0359, 0.0382, 0.0234, 0.0412], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:53:56,147 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=155813.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:54:04,791 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.461e+02 2.310e+02 2.739e+02 3.224e+02 5.856e+02, threshold=5.477e+02, percent-clipped=2.0 2023-04-30 07:54:11,804 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4060, 4.2451, 4.4213, 4.5676, 4.6684, 4.2262, 4.4497, 4.6799], device='cuda:6'), covar=tensor([0.1320, 0.1061, 0.1240, 0.0661, 0.0535, 0.1079, 0.2245, 0.0573], device='cuda:6'), in_proj_covar=tensor([0.0633, 0.0785, 0.0927, 0.0799, 0.0594, 0.0625, 0.0631, 0.0740], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 07:54:42,501 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=155845.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 07:54:52,021 INFO [train.py:904] (6/8) Epoch 16, batch 3600, loss[loss=0.1711, simple_loss=0.2531, pruned_loss=0.04453, over 16811.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2573, pruned_loss=0.04402, over 3320564.19 frames. ], batch size: 102, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:55:48,982 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2023-04-30 07:56:03,885 INFO [train.py:904] (6/8) Epoch 16, batch 3650, loss[loss=0.1737, simple_loss=0.2429, pruned_loss=0.05219, over 16786.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2569, pruned_loss=0.04472, over 3306639.21 frames. ], batch size: 83, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:56:05,001 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1732, 3.2191, 3.4302, 2.1777, 3.0187, 2.4127, 3.6617, 3.5222], device='cuda:6'), covar=tensor([0.0200, 0.0880, 0.0529, 0.1733, 0.0764, 0.0896, 0.0530, 0.0822], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0159, 0.0163, 0.0150, 0.0141, 0.0127, 0.0143, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 07:56:28,564 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.258e+02 2.651e+02 3.350e+02 6.145e+02, threshold=5.302e+02, percent-clipped=1.0 2023-04-30 07:57:17,739 INFO [train.py:904] (6/8) Epoch 16, batch 3700, loss[loss=0.1539, simple_loss=0.2322, pruned_loss=0.03778, over 16886.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2551, pruned_loss=0.04592, over 3297236.36 frames. ], batch size: 96, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:57:32,537 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=155961.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:57:38,822 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-30 07:58:36,927 INFO [train.py:904] (6/8) Epoch 16, batch 3750, loss[loss=0.1673, simple_loss=0.2591, pruned_loss=0.03769, over 17120.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2556, pruned_loss=0.04723, over 3272955.18 frames. ], batch size: 49, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:58:53,071 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=156013.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:59:02,682 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.696e+02 2.313e+02 2.777e+02 3.166e+02 7.362e+02, threshold=5.555e+02, percent-clipped=3.0 2023-04-30 07:59:07,829 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6598, 4.5467, 4.5713, 4.3458, 4.3289, 4.6359, 4.4398, 4.4320], device='cuda:6'), covar=tensor([0.0702, 0.0803, 0.0318, 0.0267, 0.0820, 0.0539, 0.0462, 0.0599], device='cuda:6'), in_proj_covar=tensor([0.0286, 0.0397, 0.0336, 0.0325, 0.0352, 0.0376, 0.0229, 0.0404], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 07:59:16,202 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 07:59:51,245 INFO [train.py:904] (6/8) Epoch 16, batch 3800, loss[loss=0.1994, simple_loss=0.2808, pruned_loss=0.05899, over 17025.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2563, pruned_loss=0.04824, over 3279350.47 frames. ], batch size: 55, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 08:00:17,857 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=156070.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:00:30,665 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-30 08:01:03,105 INFO [train.py:904] (6/8) Epoch 16, batch 3850, loss[loss=0.1885, simple_loss=0.2612, pruned_loss=0.05786, over 16751.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2569, pruned_loss=0.04916, over 3270642.00 frames. ], batch size: 124, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:01:09,490 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7348, 1.9303, 2.3564, 2.6779, 2.6312, 2.6307, 1.9329, 2.8866], device='cuda:6'), covar=tensor([0.0141, 0.0376, 0.0271, 0.0215, 0.0254, 0.0251, 0.0432, 0.0109], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0186, 0.0173, 0.0176, 0.0188, 0.0144, 0.0185, 0.0137], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:01:20,118 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156113.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:01:28,704 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.683e+02 2.283e+02 2.734e+02 3.408e+02 7.535e+02, threshold=5.468e+02, percent-clipped=3.0 2023-04-30 08:01:47,451 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=156131.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 08:01:59,099 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=156140.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 08:02:16,234 INFO [train.py:904] (6/8) Epoch 16, batch 3900, loss[loss=0.1613, simple_loss=0.2366, pruned_loss=0.04304, over 16813.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2565, pruned_loss=0.04952, over 3273973.78 frames. ], batch size: 102, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:02:29,484 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156161.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:03:26,069 INFO [train.py:904] (6/8) Epoch 16, batch 3950, loss[loss=0.1815, simple_loss=0.2522, pruned_loss=0.05539, over 16780.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.256, pruned_loss=0.05008, over 3276605.11 frames. ], batch size: 102, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:03:47,404 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=156216.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:03:50,338 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.580e+02 2.349e+02 2.934e+02 3.426e+02 7.028e+02, threshold=5.868e+02, percent-clipped=1.0 2023-04-30 08:04:37,101 INFO [train.py:904] (6/8) Epoch 16, batch 4000, loss[loss=0.1767, simple_loss=0.2594, pruned_loss=0.04706, over 16592.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2563, pruned_loss=0.05043, over 3278795.62 frames. ], batch size: 62, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:04:49,942 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156261.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:05:12,217 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=156277.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:05:31,894 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=156290.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:05:38,034 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4903, 3.5533, 1.8567, 3.7462, 2.6249, 3.8638, 1.9666, 2.7445], device='cuda:6'), covar=tensor([0.0272, 0.0363, 0.1899, 0.0208, 0.0810, 0.0396, 0.1732, 0.0752], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0176, 0.0193, 0.0155, 0.0174, 0.0217, 0.0202, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 08:05:49,241 INFO [train.py:904] (6/8) Epoch 16, batch 4050, loss[loss=0.1744, simple_loss=0.2562, pruned_loss=0.04626, over 16466.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.257, pruned_loss=0.05007, over 3283514.16 frames. ], batch size: 75, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:05:59,641 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156309.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:06:05,055 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156313.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:06:14,225 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.383e+02 1.898e+02 2.203e+02 2.661e+02 4.268e+02, threshold=4.405e+02, percent-clipped=0.0 2023-04-30 08:06:46,135 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6476, 4.9315, 4.6848, 4.7628, 4.4124, 4.3936, 4.3550, 5.0071], device='cuda:6'), covar=tensor([0.1065, 0.0793, 0.0882, 0.0671, 0.0760, 0.1009, 0.1002, 0.0800], device='cuda:6'), in_proj_covar=tensor([0.0633, 0.0785, 0.0640, 0.0568, 0.0492, 0.0504, 0.0653, 0.0602], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:06:55,837 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-04-30 08:07:01,320 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=156351.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:07:01,986 INFO [train.py:904] (6/8) Epoch 16, batch 4100, loss[loss=0.1943, simple_loss=0.2811, pruned_loss=0.05377, over 16726.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2579, pruned_loss=0.049, over 3295360.77 frames. ], batch size: 83, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:07:15,199 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156361.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:08:15,984 INFO [train.py:904] (6/8) Epoch 16, batch 4150, loss[loss=0.2444, simple_loss=0.3142, pruned_loss=0.08731, over 11512.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2653, pruned_loss=0.05152, over 3259701.45 frames. ], batch size: 246, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:08:40,339 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.507e+02 2.118e+02 2.582e+02 3.252e+02 7.284e+02, threshold=5.164e+02, percent-clipped=7.0 2023-04-30 08:08:52,190 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=156426.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 08:09:13,260 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156440.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 08:09:17,987 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2470, 4.3250, 3.3931, 2.6439, 3.1653, 3.0410, 4.8399, 3.9298], device='cuda:6'), covar=tensor([0.2242, 0.0629, 0.1427, 0.2243, 0.2423, 0.1490, 0.0364, 0.0968], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0264, 0.0295, 0.0295, 0.0292, 0.0240, 0.0281, 0.0320], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 08:09:30,691 INFO [train.py:904] (6/8) Epoch 16, batch 4200, loss[loss=0.2079, simple_loss=0.3016, pruned_loss=0.0571, over 16648.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.2724, pruned_loss=0.05297, over 3232797.10 frames. ], batch size: 134, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:10:24,402 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156488.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:10:43,697 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-30 08:10:43,949 INFO [train.py:904] (6/8) Epoch 16, batch 4250, loss[loss=0.1943, simple_loss=0.2861, pruned_loss=0.05122, over 16657.00 frames. ], tot_loss[loss=0.1913, simple_loss=0.2763, pruned_loss=0.05314, over 3211763.03 frames. ], batch size: 57, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:10:53,187 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-30 08:11:09,161 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.668e+02 2.212e+02 2.674e+02 3.089e+02 6.979e+02, threshold=5.347e+02, percent-clipped=2.0 2023-04-30 08:11:56,444 INFO [train.py:904] (6/8) Epoch 16, batch 4300, loss[loss=0.2299, simple_loss=0.3081, pruned_loss=0.07584, over 11838.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.278, pruned_loss=0.05216, over 3213182.82 frames. ], batch size: 247, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:12:19,689 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-30 08:12:27,167 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=156572.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:13:09,409 INFO [train.py:904] (6/8) Epoch 16, batch 4350, loss[loss=0.1929, simple_loss=0.2895, pruned_loss=0.04812, over 16893.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2811, pruned_loss=0.05322, over 3209679.51 frames. ], batch size: 96, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:13:34,596 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.777e+02 2.270e+02 2.623e+02 3.147e+02 7.539e+02, threshold=5.245e+02, percent-clipped=2.0 2023-04-30 08:14:14,561 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=156646.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:14:22,116 INFO [train.py:904] (6/8) Epoch 16, batch 4400, loss[loss=0.2037, simple_loss=0.2883, pruned_loss=0.05951, over 16688.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2829, pruned_loss=0.05451, over 3192146.46 frames. ], batch size: 134, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:14:45,283 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4173, 5.3891, 5.1830, 4.5701, 5.3972, 1.8255, 5.0808, 4.9203], device='cuda:6'), covar=tensor([0.0039, 0.0033, 0.0106, 0.0267, 0.0037, 0.2707, 0.0068, 0.0151], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0139, 0.0186, 0.0172, 0.0159, 0.0198, 0.0175, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:15:01,959 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7803, 2.8189, 2.9840, 5.0534, 4.0303, 4.4562, 1.6476, 3.5074], device='cuda:6'), covar=tensor([0.1251, 0.0766, 0.0968, 0.0111, 0.0290, 0.0275, 0.1520, 0.0622], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0168, 0.0187, 0.0172, 0.0202, 0.0212, 0.0190, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 08:15:32,098 INFO [train.py:904] (6/8) Epoch 16, batch 4450, loss[loss=0.1933, simple_loss=0.2885, pruned_loss=0.04905, over 16793.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2869, pruned_loss=0.05615, over 3200183.54 frames. ], batch size: 116, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:15:57,563 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 2.034e+02 2.328e+02 2.907e+02 4.699e+02, threshold=4.656e+02, percent-clipped=0.0 2023-04-30 08:16:01,090 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8305, 3.2390, 3.1370, 2.1003, 2.9543, 3.1826, 3.0513, 1.8476], device='cuda:6'), covar=tensor([0.0547, 0.0048, 0.0054, 0.0401, 0.0096, 0.0105, 0.0089, 0.0464], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0075, 0.0076, 0.0130, 0.0089, 0.0099, 0.0088, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 08:16:09,708 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156726.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 08:16:41,454 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=156749.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:16:44,971 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1057, 2.3656, 2.4733, 2.6112, 2.1508, 3.2239, 1.8647, 2.8279], device='cuda:6'), covar=tensor([0.0973, 0.0529, 0.0900, 0.0146, 0.0133, 0.0301, 0.1259, 0.0574], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0168, 0.0187, 0.0172, 0.0202, 0.0212, 0.0190, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 08:16:45,639 INFO [train.py:904] (6/8) Epoch 16, batch 4500, loss[loss=0.2036, simple_loss=0.2747, pruned_loss=0.06618, over 11275.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2879, pruned_loss=0.05712, over 3196097.14 frames. ], batch size: 246, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:17:01,362 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8024, 3.0533, 3.1270, 5.0011, 3.9491, 4.3597, 1.6839, 3.5059], device='cuda:6'), covar=tensor([0.1249, 0.0682, 0.0910, 0.0087, 0.0306, 0.0298, 0.1466, 0.0635], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0168, 0.0187, 0.0172, 0.0203, 0.0213, 0.0190, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 08:17:16,191 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156774.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:17:54,953 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 08:17:56,547 INFO [train.py:904] (6/8) Epoch 16, batch 4550, loss[loss=0.2186, simple_loss=0.2905, pruned_loss=0.07335, over 11846.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2885, pruned_loss=0.0578, over 3212913.39 frames. ], batch size: 247, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:18:07,944 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=156810.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:18:19,417 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.005e+02 2.274e+02 2.770e+02 4.941e+02, threshold=4.548e+02, percent-clipped=1.0 2023-04-30 08:19:06,330 INFO [train.py:904] (6/8) Epoch 16, batch 4600, loss[loss=0.2129, simple_loss=0.2983, pruned_loss=0.0637, over 16363.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.289, pruned_loss=0.0576, over 3210109.35 frames. ], batch size: 35, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:19:18,062 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1495, 3.0283, 3.2016, 1.6215, 3.3249, 3.4003, 2.7356, 2.5847], device='cuda:6'), covar=tensor([0.0883, 0.0243, 0.0223, 0.1300, 0.0087, 0.0160, 0.0489, 0.0501], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0105, 0.0092, 0.0137, 0.0073, 0.0117, 0.0125, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 08:19:36,315 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156872.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:19:55,723 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=156885.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:20:05,366 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5360, 3.5929, 3.3869, 2.9661, 3.1840, 3.4991, 3.2729, 3.3009], device='cuda:6'), covar=tensor([0.0497, 0.0455, 0.0258, 0.0252, 0.0455, 0.0345, 0.1511, 0.0439], device='cuda:6'), in_proj_covar=tensor([0.0267, 0.0369, 0.0316, 0.0304, 0.0331, 0.0350, 0.0215, 0.0377], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:20:19,275 INFO [train.py:904] (6/8) Epoch 16, batch 4650, loss[loss=0.1929, simple_loss=0.2819, pruned_loss=0.05197, over 16375.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2874, pruned_loss=0.0573, over 3211251.62 frames. ], batch size: 146, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:20:45,036 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 1.893e+02 2.205e+02 2.654e+02 4.694e+02, threshold=4.410e+02, percent-clipped=1.0 2023-04-30 08:20:46,966 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156920.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:21:25,719 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156946.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:21:25,812 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=156946.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:21:33,080 INFO [train.py:904] (6/8) Epoch 16, batch 4700, loss[loss=0.185, simple_loss=0.2652, pruned_loss=0.05244, over 11745.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2851, pruned_loss=0.05624, over 3200649.65 frames. ], batch size: 248, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:21:41,184 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9745, 2.0002, 2.1877, 3.5430, 1.9970, 2.3845, 2.1825, 2.1495], device='cuda:6'), covar=tensor([0.1358, 0.3456, 0.2697, 0.0579, 0.4082, 0.2394, 0.3426, 0.3393], device='cuda:6'), in_proj_covar=tensor([0.0382, 0.0423, 0.0351, 0.0322, 0.0428, 0.0490, 0.0391, 0.0495], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:22:36,836 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156994.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:22:47,019 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-30 08:22:47,496 INFO [train.py:904] (6/8) Epoch 16, batch 4750, loss[loss=0.1677, simple_loss=0.2568, pruned_loss=0.03934, over 16661.00 frames. ], tot_loss[loss=0.1942, simple_loss=0.2807, pruned_loss=0.05387, over 3208774.09 frames. ], batch size: 76, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:22:59,840 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157010.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:23:11,418 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.334e+02 1.885e+02 2.255e+02 2.719e+02 5.243e+02, threshold=4.511e+02, percent-clipped=2.0 2023-04-30 08:23:59,602 INFO [train.py:904] (6/8) Epoch 16, batch 4800, loss[loss=0.1996, simple_loss=0.2811, pruned_loss=0.05902, over 11535.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.277, pruned_loss=0.05174, over 3220795.32 frames. ], batch size: 247, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:24:29,067 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157071.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:25:14,778 INFO [train.py:904] (6/8) Epoch 16, batch 4850, loss[loss=0.176, simple_loss=0.2745, pruned_loss=0.03877, over 16752.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2783, pruned_loss=0.0512, over 3204946.02 frames. ], batch size: 89, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:25:20,717 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157105.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:25:41,557 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.08 vs. limit=5.0 2023-04-30 08:25:41,918 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 1.885e+02 2.190e+02 2.625e+02 3.825e+02, threshold=4.379e+02, percent-clipped=0.0 2023-04-30 08:26:17,031 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157142.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:26:30,906 INFO [train.py:904] (6/8) Epoch 16, batch 4900, loss[loss=0.2059, simple_loss=0.2856, pruned_loss=0.06312, over 12101.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2772, pruned_loss=0.05037, over 3185448.73 frames. ], batch size: 246, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:26:53,726 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9795, 4.2438, 4.0140, 4.1029, 3.7590, 3.8641, 3.7990, 4.2272], device='cuda:6'), covar=tensor([0.1065, 0.0821, 0.0892, 0.0660, 0.0737, 0.1517, 0.0924, 0.0945], device='cuda:6'), in_proj_covar=tensor([0.0599, 0.0740, 0.0603, 0.0536, 0.0466, 0.0475, 0.0615, 0.0570], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:27:23,166 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5122, 4.3100, 4.4569, 4.7297, 4.8578, 4.4927, 4.8299, 4.8460], device='cuda:6'), covar=tensor([0.1454, 0.1395, 0.1772, 0.0720, 0.0580, 0.0836, 0.0653, 0.0682], device='cuda:6'), in_proj_covar=tensor([0.0589, 0.0725, 0.0855, 0.0737, 0.0548, 0.0579, 0.0585, 0.0682], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:27:26,850 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-04-30 08:27:45,015 INFO [train.py:904] (6/8) Epoch 16, batch 4950, loss[loss=0.1915, simple_loss=0.2806, pruned_loss=0.05122, over 12013.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2766, pruned_loss=0.04957, over 3178185.92 frames. ], batch size: 248, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:27:46,834 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157203.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:28:08,806 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 2.143e+02 2.414e+02 2.690e+02 6.197e+02, threshold=4.827e+02, percent-clipped=2.0 2023-04-30 08:28:41,078 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157241.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:28:57,711 INFO [train.py:904] (6/8) Epoch 16, batch 5000, loss[loss=0.1858, simple_loss=0.2796, pruned_loss=0.046, over 16841.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2783, pruned_loss=0.0493, over 3209208.04 frames. ], batch size: 116, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:29:57,807 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 08:30:10,090 INFO [train.py:904] (6/8) Epoch 16, batch 5050, loss[loss=0.1814, simple_loss=0.2643, pruned_loss=0.04931, over 16271.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2793, pruned_loss=0.04939, over 3210751.50 frames. ], batch size: 35, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:30:33,957 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.633e+02 2.188e+02 2.518e+02 3.068e+02 6.843e+02, threshold=5.037e+02, percent-clipped=3.0 2023-04-30 08:31:22,540 INFO [train.py:904] (6/8) Epoch 16, batch 5100, loss[loss=0.1557, simple_loss=0.2407, pruned_loss=0.03536, over 16705.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2775, pruned_loss=0.04874, over 3211792.92 frames. ], batch size: 89, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:31:29,170 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7045, 4.4788, 4.7005, 4.8852, 5.0704, 4.5568, 5.0477, 5.0380], device='cuda:6'), covar=tensor([0.1483, 0.1136, 0.1556, 0.0668, 0.0440, 0.0844, 0.0472, 0.0546], device='cuda:6'), in_proj_covar=tensor([0.0588, 0.0725, 0.0855, 0.0735, 0.0548, 0.0580, 0.0585, 0.0682], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:31:29,203 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157356.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:31:44,277 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157366.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:32:07,474 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157382.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:32:38,193 INFO [train.py:904] (6/8) Epoch 16, batch 5150, loss[loss=0.1914, simple_loss=0.2854, pruned_loss=0.04868, over 16539.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2773, pruned_loss=0.0484, over 3197117.71 frames. ], batch size: 75, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:32:42,294 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157404.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 08:32:44,008 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=157405.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:33:02,259 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157417.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:33:04,659 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.305e+02 1.981e+02 2.319e+02 2.667e+02 4.266e+02, threshold=4.638e+02, percent-clipped=0.0 2023-04-30 08:33:40,868 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157443.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:33:53,087 INFO [train.py:904] (6/8) Epoch 16, batch 5200, loss[loss=0.1838, simple_loss=0.2673, pruned_loss=0.05013, over 16698.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2762, pruned_loss=0.04811, over 3189447.94 frames. ], batch size: 89, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:33:54,777 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=157453.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:34:13,162 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157465.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 08:34:50,666 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1035, 2.7562, 2.8554, 2.0527, 2.7007, 2.1807, 2.7087, 2.9447], device='cuda:6'), covar=tensor([0.0255, 0.0697, 0.0534, 0.1544, 0.0752, 0.0846, 0.0554, 0.0649], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0155, 0.0161, 0.0148, 0.0139, 0.0126, 0.0140, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 08:35:01,939 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157498.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:35:08,180 INFO [train.py:904] (6/8) Epoch 16, batch 5250, loss[loss=0.1756, simple_loss=0.2632, pruned_loss=0.04401, over 16362.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2735, pruned_loss=0.04754, over 3197339.81 frames. ], batch size: 68, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:35:14,399 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157506.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 08:35:32,952 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.377e+02 2.109e+02 2.306e+02 2.689e+02 3.834e+02, threshold=4.611e+02, percent-clipped=0.0 2023-04-30 08:36:06,426 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=157541.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:36:21,502 INFO [train.py:904] (6/8) Epoch 16, batch 5300, loss[loss=0.2047, simple_loss=0.2802, pruned_loss=0.06462, over 11993.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.2699, pruned_loss=0.04643, over 3193515.29 frames. ], batch size: 247, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:36:45,036 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157567.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 08:36:46,186 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157568.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:37:17,229 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=157589.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:37:35,559 INFO [train.py:904] (6/8) Epoch 16, batch 5350, loss[loss=0.1875, simple_loss=0.2788, pruned_loss=0.04808, over 16791.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2688, pruned_loss=0.04599, over 3200031.46 frames. ], batch size: 83, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:38:00,207 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.392e+02 1.990e+02 2.338e+02 2.844e+02 4.721e+02, threshold=4.676e+02, percent-clipped=1.0 2023-04-30 08:38:12,371 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5854, 3.4952, 4.0904, 1.7481, 4.2460, 4.2570, 3.0112, 3.0604], device='cuda:6'), covar=tensor([0.0788, 0.0278, 0.0151, 0.1385, 0.0052, 0.0111, 0.0402, 0.0493], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0104, 0.0091, 0.0136, 0.0072, 0.0115, 0.0123, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 08:38:16,668 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157629.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:38:49,042 INFO [train.py:904] (6/8) Epoch 16, batch 5400, loss[loss=0.1818, simple_loss=0.2786, pruned_loss=0.04251, over 15622.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2722, pruned_loss=0.04712, over 3190336.80 frames. ], batch size: 190, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:39:09,637 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=157666.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:39:44,581 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4956, 3.4751, 3.4495, 2.7451, 3.3364, 2.0092, 3.1423, 2.8597], device='cuda:6'), covar=tensor([0.0142, 0.0115, 0.0155, 0.0234, 0.0097, 0.2227, 0.0130, 0.0221], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0136, 0.0183, 0.0171, 0.0155, 0.0194, 0.0171, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:39:46,883 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8116, 3.6838, 3.9051, 3.6036, 3.7953, 4.2743, 3.9017, 3.5944], device='cuda:6'), covar=tensor([0.1940, 0.2015, 0.1902, 0.2369, 0.2686, 0.1593, 0.1467, 0.2326], device='cuda:6'), in_proj_covar=tensor([0.0384, 0.0538, 0.0582, 0.0452, 0.0609, 0.0618, 0.0460, 0.0609], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 08:40:04,937 INFO [train.py:904] (6/8) Epoch 16, batch 5450, loss[loss=0.1953, simple_loss=0.2789, pruned_loss=0.05583, over 17126.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2751, pruned_loss=0.04849, over 3187715.21 frames. ], batch size: 49, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:40:20,461 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157712.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:40:23,494 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=157714.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:40:29,842 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.390e+02 2.375e+02 2.743e+02 3.382e+02 6.991e+02, threshold=5.486e+02, percent-clipped=5.0 2023-04-30 08:41:00,228 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157738.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:41:00,412 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4339, 3.5378, 1.9149, 3.9889, 2.5643, 3.9134, 2.2748, 2.7277], device='cuda:6'), covar=tensor([0.0267, 0.0359, 0.1735, 0.0168, 0.0811, 0.0507, 0.1342, 0.0722], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0172, 0.0192, 0.0147, 0.0172, 0.0210, 0.0200, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 08:41:20,984 INFO [train.py:904] (6/8) Epoch 16, batch 5500, loss[loss=0.2353, simple_loss=0.3193, pruned_loss=0.07561, over 16624.00 frames. ], tot_loss[loss=0.1942, simple_loss=0.2824, pruned_loss=0.05306, over 3172123.72 frames. ], batch size: 134, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:41:24,158 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6473, 4.6658, 4.5158, 4.1902, 4.1576, 4.5978, 4.4327, 4.2987], device='cuda:6'), covar=tensor([0.0539, 0.0418, 0.0250, 0.0306, 0.0897, 0.0457, 0.0377, 0.0605], device='cuda:6'), in_proj_covar=tensor([0.0271, 0.0379, 0.0319, 0.0309, 0.0335, 0.0359, 0.0218, 0.0381], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:41:33,682 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157760.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 08:41:46,628 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9581, 2.3198, 2.3312, 3.0328, 2.2035, 3.2456, 1.7695, 2.6719], device='cuda:6'), covar=tensor([0.1256, 0.0631, 0.1044, 0.0204, 0.0155, 0.0416, 0.1562, 0.0704], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0169, 0.0189, 0.0172, 0.0203, 0.0214, 0.0192, 0.0188], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 08:42:30,904 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=157798.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:42:36,592 INFO [train.py:904] (6/8) Epoch 16, batch 5550, loss[loss=0.2209, simple_loss=0.3017, pruned_loss=0.07009, over 16699.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2891, pruned_loss=0.05821, over 3134462.94 frames. ], batch size: 57, lr: 4.24e-03, grad_scale: 16.0 2023-04-30 08:42:42,824 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3817, 3.1888, 2.4245, 2.1160, 2.2093, 1.9884, 3.1792, 2.9891], device='cuda:6'), covar=tensor([0.2939, 0.0889, 0.2065, 0.2686, 0.2611, 0.2430, 0.0744, 0.1425], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0262, 0.0295, 0.0295, 0.0287, 0.0238, 0.0280, 0.0316], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 08:42:47,366 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4955, 3.4830, 2.7072, 2.0950, 2.3832, 2.2504, 3.6635, 3.3585], device='cuda:6'), covar=tensor([0.2755, 0.0615, 0.1662, 0.2626, 0.2323, 0.1941, 0.0494, 0.1050], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0262, 0.0295, 0.0295, 0.0287, 0.0238, 0.0280, 0.0316], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 08:43:04,449 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.012e+02 3.202e+02 3.935e+02 4.988e+02 8.755e+02, threshold=7.870e+02, percent-clipped=18.0 2023-04-30 08:43:31,570 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0584, 4.0072, 4.0092, 2.8711, 4.0549, 1.5520, 3.7567, 3.5183], device='cuda:6'), covar=tensor([0.0180, 0.0151, 0.0220, 0.0645, 0.0129, 0.3446, 0.0195, 0.0408], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0137, 0.0184, 0.0173, 0.0156, 0.0195, 0.0171, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:43:47,866 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=157846.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:43:57,047 INFO [train.py:904] (6/8) Epoch 16, batch 5600, loss[loss=0.2113, simple_loss=0.2986, pruned_loss=0.06199, over 16810.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2945, pruned_loss=0.06256, over 3118663.80 frames. ], batch size: 116, lr: 4.24e-03, grad_scale: 16.0 2023-04-30 08:44:00,074 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157853.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:44:09,936 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0065, 2.8511, 2.8313, 2.1376, 2.6797, 2.2406, 2.7143, 2.9699], device='cuda:6'), covar=tensor([0.0261, 0.0612, 0.0456, 0.1498, 0.0681, 0.0870, 0.0481, 0.0598], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0155, 0.0162, 0.0148, 0.0139, 0.0126, 0.0140, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 08:44:14,250 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157862.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 08:44:40,589 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0398, 2.4870, 2.5874, 1.9229, 2.6436, 2.7716, 2.4350, 2.3816], device='cuda:6'), covar=tensor([0.0671, 0.0201, 0.0191, 0.0876, 0.0099, 0.0241, 0.0412, 0.0407], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0104, 0.0090, 0.0136, 0.0072, 0.0116, 0.0123, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 08:45:21,486 INFO [train.py:904] (6/8) Epoch 16, batch 5650, loss[loss=0.2202, simple_loss=0.3124, pruned_loss=0.06397, over 16853.00 frames. ], tot_loss[loss=0.216, simple_loss=0.2992, pruned_loss=0.06644, over 3101672.61 frames. ], batch size: 96, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:45:41,612 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157914.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:45:50,669 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.188e+02 3.549e+02 4.142e+02 4.836e+02 1.033e+03, threshold=8.283e+02, percent-clipped=2.0 2023-04-30 08:45:56,919 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157924.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:46:39,301 INFO [train.py:904] (6/8) Epoch 16, batch 5700, loss[loss=0.2307, simple_loss=0.3163, pruned_loss=0.07258, over 16319.00 frames. ], tot_loss[loss=0.2183, simple_loss=0.3006, pruned_loss=0.06801, over 3082650.89 frames. ], batch size: 146, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:48:01,931 INFO [train.py:904] (6/8) Epoch 16, batch 5750, loss[loss=0.2493, simple_loss=0.3171, pruned_loss=0.09073, over 10890.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.3032, pruned_loss=0.06974, over 3045081.02 frames. ], batch size: 247, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:48:17,320 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158012.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:48:30,520 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.050e+02 3.002e+02 3.593e+02 4.729e+02 9.320e+02, threshold=7.185e+02, percent-clipped=2.0 2023-04-30 08:49:00,701 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158038.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:49:22,774 INFO [train.py:904] (6/8) Epoch 16, batch 5800, loss[loss=0.1911, simple_loss=0.2873, pruned_loss=0.04744, over 16826.00 frames. ], tot_loss[loss=0.2211, simple_loss=0.3035, pruned_loss=0.06933, over 3027651.04 frames. ], batch size: 102, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:49:33,841 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1800, 2.4233, 2.0882, 2.0966, 2.7994, 2.4836, 2.9255, 3.0512], device='cuda:6'), covar=tensor([0.0146, 0.0400, 0.0478, 0.0485, 0.0260, 0.0369, 0.0226, 0.0235], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0218, 0.0211, 0.0210, 0.0218, 0.0220, 0.0223, 0.0214], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:49:35,950 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158060.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:49:36,123 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158060.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 08:50:16,663 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158086.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:50:41,299 INFO [train.py:904] (6/8) Epoch 16, batch 5850, loss[loss=0.198, simple_loss=0.2893, pruned_loss=0.0533, over 16790.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.3002, pruned_loss=0.06671, over 3048882.46 frames. ], batch size: 83, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:50:51,676 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158108.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 08:51:08,527 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.631e+02 2.851e+02 3.457e+02 4.318e+02 8.019e+02, threshold=6.915e+02, percent-clipped=1.0 2023-04-30 08:51:13,064 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=158122.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:52:03,755 INFO [train.py:904] (6/8) Epoch 16, batch 5900, loss[loss=0.1912, simple_loss=0.2795, pruned_loss=0.05147, over 16273.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.3004, pruned_loss=0.0671, over 3058411.16 frames. ], batch size: 165, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:52:07,441 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.19 vs. limit=5.0 2023-04-30 08:52:24,455 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158162.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 08:52:29,790 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5893, 3.9885, 3.8575, 2.1847, 3.3057, 2.5944, 4.1103, 4.0626], device='cuda:6'), covar=tensor([0.0208, 0.0595, 0.0558, 0.1815, 0.0714, 0.0910, 0.0490, 0.0695], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0154, 0.0160, 0.0147, 0.0138, 0.0125, 0.0139, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 08:52:55,974 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=158183.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:53:01,079 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3665, 2.9304, 2.6296, 2.2244, 2.2506, 2.1904, 2.9103, 2.8543], device='cuda:6'), covar=tensor([0.2475, 0.0734, 0.1506, 0.2483, 0.2179, 0.1938, 0.0520, 0.1117], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0260, 0.0293, 0.0294, 0.0285, 0.0237, 0.0279, 0.0313], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 08:53:25,265 INFO [train.py:904] (6/8) Epoch 16, batch 5950, loss[loss=0.2222, simple_loss=0.3132, pruned_loss=0.06562, over 15466.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.3015, pruned_loss=0.06582, over 3067831.13 frames. ], batch size: 190, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:53:36,985 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=158209.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:53:38,263 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158210.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 08:53:55,265 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.924e+02 3.226e+02 4.367e+02 1.058e+03, threshold=6.453e+02, percent-clipped=4.0 2023-04-30 08:54:01,943 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158224.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:54:45,444 INFO [train.py:904] (6/8) Epoch 16, batch 6000, loss[loss=0.1889, simple_loss=0.276, pruned_loss=0.0509, over 15358.00 frames. ], tot_loss[loss=0.2163, simple_loss=0.3008, pruned_loss=0.06583, over 3057111.76 frames. ], batch size: 190, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:54:45,444 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 08:54:56,481 INFO [train.py:938] (6/8) Epoch 16, validation: loss=0.1553, simple_loss=0.2682, pruned_loss=0.0212, over 944034.00 frames. 2023-04-30 08:54:56,481 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 08:54:59,528 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5975, 3.6577, 2.7060, 2.2207, 2.4583, 2.2400, 3.8285, 3.3952], device='cuda:6'), covar=tensor([0.2683, 0.0654, 0.1783, 0.2425, 0.2402, 0.2024, 0.0475, 0.1103], device='cuda:6'), in_proj_covar=tensor([0.0314, 0.0260, 0.0294, 0.0294, 0.0286, 0.0238, 0.0279, 0.0314], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 08:55:27,066 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158272.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:56:01,639 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1000, 1.7428, 2.5273, 3.0294, 2.9690, 3.3702, 1.7219, 3.4314], device='cuda:6'), covar=tensor([0.0146, 0.0522, 0.0280, 0.0230, 0.0212, 0.0124, 0.0634, 0.0096], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0183, 0.0169, 0.0172, 0.0183, 0.0140, 0.0183, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 08:56:13,210 INFO [train.py:904] (6/8) Epoch 16, batch 6050, loss[loss=0.1787, simple_loss=0.2895, pruned_loss=0.03399, over 16724.00 frames. ], tot_loss[loss=0.2147, simple_loss=0.2993, pruned_loss=0.06509, over 3065475.92 frames. ], batch size: 89, lr: 4.23e-03, grad_scale: 8.0 2023-04-30 08:56:40,246 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.935e+02 2.678e+02 3.324e+02 4.253e+02 8.310e+02, threshold=6.647e+02, percent-clipped=4.0 2023-04-30 08:57:31,990 INFO [train.py:904] (6/8) Epoch 16, batch 6100, loss[loss=0.1878, simple_loss=0.2767, pruned_loss=0.04943, over 16526.00 frames. ], tot_loss[loss=0.213, simple_loss=0.2985, pruned_loss=0.06379, over 3089789.87 frames. ], batch size: 75, lr: 4.23e-03, grad_scale: 8.0 2023-04-30 08:58:24,093 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=158384.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:58:49,711 INFO [train.py:904] (6/8) Epoch 16, batch 6150, loss[loss=0.1833, simple_loss=0.2733, pruned_loss=0.04672, over 16434.00 frames. ], tot_loss[loss=0.2119, simple_loss=0.2969, pruned_loss=0.06348, over 3089495.58 frames. ], batch size: 68, lr: 4.23e-03, grad_scale: 8.0 2023-04-30 08:59:18,316 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.901e+02 2.793e+02 3.279e+02 3.951e+02 7.811e+02, threshold=6.558e+02, percent-clipped=1.0 2023-04-30 08:59:57,775 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=158445.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:00:07,998 INFO [train.py:904] (6/8) Epoch 16, batch 6200, loss[loss=0.2074, simple_loss=0.2919, pruned_loss=0.06147, over 16266.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2951, pruned_loss=0.06311, over 3088972.42 frames. ], batch size: 165, lr: 4.23e-03, grad_scale: 8.0 2023-04-30 09:00:27,488 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9232, 5.1840, 4.8864, 4.9314, 4.7598, 4.6806, 4.5737, 5.2672], device='cuda:6'), covar=tensor([0.1093, 0.0835, 0.0995, 0.0826, 0.0763, 0.0880, 0.1078, 0.0882], device='cuda:6'), in_proj_covar=tensor([0.0597, 0.0739, 0.0604, 0.0540, 0.0462, 0.0472, 0.0613, 0.0565], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:00:38,245 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4674, 2.9358, 2.6340, 2.2429, 2.2895, 2.2982, 2.9303, 2.8924], device='cuda:6'), covar=tensor([0.2293, 0.0802, 0.1577, 0.2331, 0.2096, 0.2017, 0.0481, 0.1335], device='cuda:6'), in_proj_covar=tensor([0.0316, 0.0263, 0.0297, 0.0297, 0.0290, 0.0240, 0.0282, 0.0319], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:00:44,893 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2216, 4.0743, 4.2969, 4.4110, 4.5469, 4.1615, 4.5137, 4.5408], device='cuda:6'), covar=tensor([0.1629, 0.1175, 0.1324, 0.0651, 0.0518, 0.1147, 0.0635, 0.0642], device='cuda:6'), in_proj_covar=tensor([0.0584, 0.0721, 0.0846, 0.0729, 0.0546, 0.0572, 0.0580, 0.0678], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:00:48,553 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=158478.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:01:21,944 INFO [train.py:904] (6/8) Epoch 16, batch 6250, loss[loss=0.1928, simple_loss=0.2853, pruned_loss=0.05014, over 17098.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2945, pruned_loss=0.06259, over 3099363.26 frames. ], batch size: 49, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:01:34,601 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158509.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:01:50,691 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-04-30 09:01:50,902 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.774e+02 2.618e+02 3.109e+02 4.242e+02 1.066e+03, threshold=6.218e+02, percent-clipped=4.0 2023-04-30 09:01:57,760 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9977, 2.7879, 2.7501, 2.0961, 2.5959, 2.1589, 2.7819, 2.9099], device='cuda:6'), covar=tensor([0.0281, 0.0669, 0.0517, 0.1706, 0.0790, 0.0902, 0.0528, 0.0757], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0155, 0.0161, 0.0148, 0.0139, 0.0125, 0.0140, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 09:02:38,738 INFO [train.py:904] (6/8) Epoch 16, batch 6300, loss[loss=0.2129, simple_loss=0.2998, pruned_loss=0.06295, over 15303.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2943, pruned_loss=0.06174, over 3113500.36 frames. ], batch size: 190, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:02:45,575 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158557.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:03:55,407 INFO [train.py:904] (6/8) Epoch 16, batch 6350, loss[loss=0.2038, simple_loss=0.283, pruned_loss=0.06233, over 17215.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.2954, pruned_loss=0.06385, over 3082566.83 frames. ], batch size: 44, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:04:24,044 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.016e+02 2.970e+02 3.577e+02 4.444e+02 9.031e+02, threshold=7.154e+02, percent-clipped=4.0 2023-04-30 09:05:11,915 INFO [train.py:904] (6/8) Epoch 16, batch 6400, loss[loss=0.1888, simple_loss=0.2703, pruned_loss=0.05369, over 16303.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2959, pruned_loss=0.06519, over 3072139.37 frames. ], batch size: 35, lr: 4.23e-03, grad_scale: 8.0 2023-04-30 09:06:27,727 INFO [train.py:904] (6/8) Epoch 16, batch 6450, loss[loss=0.215, simple_loss=0.2811, pruned_loss=0.07443, over 11971.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2953, pruned_loss=0.06415, over 3068876.22 frames. ], batch size: 247, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:06:56,385 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.032e+02 2.821e+02 3.663e+02 4.597e+02 9.189e+02, threshold=7.326e+02, percent-clipped=6.0 2023-04-30 09:07:26,485 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=158740.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:07:43,536 INFO [train.py:904] (6/8) Epoch 16, batch 6500, loss[loss=0.2096, simple_loss=0.2912, pruned_loss=0.064, over 16801.00 frames. ], tot_loss[loss=0.2094, simple_loss=0.2931, pruned_loss=0.06283, over 3099857.50 frames. ], batch size: 124, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:08:21,826 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=158777.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:08:22,858 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158778.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:08:31,639 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=158783.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:09:01,975 INFO [train.py:904] (6/8) Epoch 16, batch 6550, loss[loss=0.1959, simple_loss=0.3019, pruned_loss=0.04493, over 16417.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.295, pruned_loss=0.06332, over 3101315.64 frames. ], batch size: 75, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:09:33,175 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.010e+02 2.836e+02 3.301e+02 3.944e+02 7.483e+02, threshold=6.603e+02, percent-clipped=1.0 2023-04-30 09:09:39,691 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158826.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:09:41,303 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0406, 3.4425, 3.3578, 1.9916, 2.9719, 2.2526, 3.5050, 3.5678], device='cuda:6'), covar=tensor([0.0272, 0.0710, 0.0634, 0.2014, 0.0853, 0.0987, 0.0593, 0.0899], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0156, 0.0162, 0.0149, 0.0140, 0.0126, 0.0141, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 09:09:53,209 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=158835.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 09:09:58,205 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=158838.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:10:06,265 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=158844.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:10:18,011 INFO [train.py:904] (6/8) Epoch 16, batch 6600, loss[loss=0.2527, simple_loss=0.3243, pruned_loss=0.09059, over 11828.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2973, pruned_loss=0.06357, over 3105453.83 frames. ], batch size: 246, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:11:25,985 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=158896.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 09:11:34,213 INFO [train.py:904] (6/8) Epoch 16, batch 6650, loss[loss=0.1939, simple_loss=0.2844, pruned_loss=0.05168, over 16848.00 frames. ], tot_loss[loss=0.2128, simple_loss=0.2973, pruned_loss=0.0641, over 3112615.06 frames. ], batch size: 102, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:12:04,649 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.962e+02 3.031e+02 3.577e+02 4.481e+02 9.334e+02, threshold=7.154e+02, percent-clipped=3.0 2023-04-30 09:12:50,515 INFO [train.py:904] (6/8) Epoch 16, batch 6700, loss[loss=0.2453, simple_loss=0.3115, pruned_loss=0.08959, over 11225.00 frames. ], tot_loss[loss=0.2125, simple_loss=0.2965, pruned_loss=0.06424, over 3086884.01 frames. ], batch size: 247, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:14:07,729 INFO [train.py:904] (6/8) Epoch 16, batch 6750, loss[loss=0.2092, simple_loss=0.291, pruned_loss=0.06369, over 16497.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2951, pruned_loss=0.06384, over 3097144.48 frames. ], batch size: 75, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:14:37,801 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.101e+02 2.962e+02 3.496e+02 4.058e+02 1.383e+03, threshold=6.992e+02, percent-clipped=2.0 2023-04-30 09:14:39,069 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4552, 2.7161, 2.3395, 2.3319, 3.0014, 2.6461, 3.1530, 3.1689], device='cuda:6'), covar=tensor([0.0093, 0.0324, 0.0402, 0.0378, 0.0192, 0.0330, 0.0188, 0.0222], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0218, 0.0212, 0.0212, 0.0218, 0.0219, 0.0222, 0.0215], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:15:05,826 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159040.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:15:23,304 INFO [train.py:904] (6/8) Epoch 16, batch 6800, loss[loss=0.2053, simple_loss=0.2977, pruned_loss=0.05647, over 16711.00 frames. ], tot_loss[loss=0.2111, simple_loss=0.295, pruned_loss=0.06358, over 3089158.27 frames. ], batch size: 89, lr: 4.22e-03, grad_scale: 8.0 2023-04-30 09:15:50,676 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=159069.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:16:21,090 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159088.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:16:40,957 INFO [train.py:904] (6/8) Epoch 16, batch 6850, loss[loss=0.2055, simple_loss=0.3068, pruned_loss=0.05211, over 16597.00 frames. ], tot_loss[loss=0.2126, simple_loss=0.2967, pruned_loss=0.06422, over 3085334.80 frames. ], batch size: 62, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:16:55,887 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.63 vs. limit=5.0 2023-04-30 09:17:12,960 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.918e+02 2.851e+02 3.389e+02 4.147e+02 9.414e+02, threshold=6.778e+02, percent-clipped=4.0 2023-04-30 09:17:13,625 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5659, 4.0370, 4.0082, 2.2835, 3.4487, 2.5691, 4.2414, 4.0772], device='cuda:6'), covar=tensor([0.0206, 0.0584, 0.0520, 0.1847, 0.0684, 0.0953, 0.0430, 0.0730], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0157, 0.0163, 0.0150, 0.0141, 0.0126, 0.0141, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 09:17:17,412 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7745, 4.3166, 4.4327, 3.1393, 4.1102, 4.4860, 4.2242, 2.5288], device='cuda:6'), covar=tensor([0.0405, 0.0040, 0.0029, 0.0295, 0.0052, 0.0068, 0.0041, 0.0385], device='cuda:6'), in_proj_covar=tensor([0.0128, 0.0074, 0.0075, 0.0128, 0.0087, 0.0097, 0.0085, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:17:22,525 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=159130.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:17:26,685 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159133.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:17:30,606 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.62 vs. limit=5.0 2023-04-30 09:17:35,436 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159139.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:17:54,571 INFO [train.py:904] (6/8) Epoch 16, batch 6900, loss[loss=0.2323, simple_loss=0.3179, pruned_loss=0.07334, over 16387.00 frames. ], tot_loss[loss=0.2132, simple_loss=0.2988, pruned_loss=0.06381, over 3100104.22 frames. ], batch size: 146, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:18:07,171 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=159159.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:18:57,090 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159191.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 09:19:13,870 INFO [train.py:904] (6/8) Epoch 16, batch 6950, loss[loss=0.2457, simple_loss=0.312, pruned_loss=0.08964, over 11681.00 frames. ], tot_loss[loss=0.2157, simple_loss=0.3005, pruned_loss=0.06543, over 3098052.95 frames. ], batch size: 247, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:19:34,235 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5393, 4.4669, 4.3535, 3.0406, 3.8284, 4.3487, 3.8687, 2.5273], device='cuda:6'), covar=tensor([0.0485, 0.0029, 0.0036, 0.0334, 0.0074, 0.0098, 0.0072, 0.0390], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0075, 0.0076, 0.0130, 0.0088, 0.0098, 0.0087, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:19:42,687 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=159220.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:19:48,719 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 3.353e+02 4.093e+02 5.038e+02 1.287e+03, threshold=8.186e+02, percent-clipped=11.0 2023-04-30 09:20:28,035 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9013, 4.4904, 3.2720, 2.5128, 3.3950, 2.8965, 4.6211, 3.9503], device='cuda:6'), covar=tensor([0.3029, 0.0618, 0.1776, 0.2604, 0.2278, 0.1755, 0.0555, 0.1086], device='cuda:6'), in_proj_covar=tensor([0.0315, 0.0262, 0.0296, 0.0298, 0.0288, 0.0239, 0.0281, 0.0317], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:20:29,895 INFO [train.py:904] (6/8) Epoch 16, batch 7000, loss[loss=0.2078, simple_loss=0.286, pruned_loss=0.06484, over 11616.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.3004, pruned_loss=0.06502, over 3082556.89 frames. ], batch size: 246, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:21:28,000 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=159290.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:21:45,219 INFO [train.py:904] (6/8) Epoch 16, batch 7050, loss[loss=0.2126, simple_loss=0.3056, pruned_loss=0.05982, over 16377.00 frames. ], tot_loss[loss=0.215, simple_loss=0.3011, pruned_loss=0.06445, over 3096679.78 frames. ], batch size: 146, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:22:18,893 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.833e+02 2.988e+02 3.653e+02 4.362e+02 9.002e+02, threshold=7.306e+02, percent-clipped=3.0 2023-04-30 09:23:00,617 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=159351.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:23:01,353 INFO [train.py:904] (6/8) Epoch 16, batch 7100, loss[loss=0.1884, simple_loss=0.2818, pruned_loss=0.0475, over 16821.00 frames. ], tot_loss[loss=0.2138, simple_loss=0.2995, pruned_loss=0.064, over 3102144.38 frames. ], batch size: 102, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:23:09,963 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8906, 1.8838, 2.3939, 2.8694, 2.6706, 3.2366, 1.9474, 3.2465], device='cuda:6'), covar=tensor([0.0176, 0.0433, 0.0307, 0.0237, 0.0276, 0.0142, 0.0468, 0.0096], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0186, 0.0170, 0.0174, 0.0185, 0.0142, 0.0184, 0.0135], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:23:43,776 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0871, 4.8571, 5.0729, 5.2751, 5.4815, 4.7972, 5.4761, 5.4232], device='cuda:6'), covar=tensor([0.1830, 0.1285, 0.1597, 0.0645, 0.0498, 0.0773, 0.0485, 0.0670], device='cuda:6'), in_proj_covar=tensor([0.0579, 0.0713, 0.0837, 0.0720, 0.0541, 0.0566, 0.0580, 0.0675], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:24:17,480 INFO [train.py:904] (6/8) Epoch 16, batch 7150, loss[loss=0.2379, simple_loss=0.3002, pruned_loss=0.08785, over 11407.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2985, pruned_loss=0.06492, over 3079833.27 frames. ], batch size: 248, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:24:19,223 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1923, 4.1983, 4.1041, 3.3774, 4.1249, 1.6114, 3.9159, 3.7246], device='cuda:6'), covar=tensor([0.0104, 0.0097, 0.0172, 0.0322, 0.0092, 0.2806, 0.0121, 0.0218], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0136, 0.0183, 0.0168, 0.0154, 0.0194, 0.0169, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:24:49,380 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 3.206e+02 4.048e+02 4.680e+02 7.501e+02, threshold=8.096e+02, percent-clipped=1.0 2023-04-30 09:24:49,921 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6588, 2.3775, 2.4028, 3.4871, 2.6454, 3.7728, 1.4279, 2.7310], device='cuda:6'), covar=tensor([0.1378, 0.0836, 0.1221, 0.0173, 0.0229, 0.0399, 0.1698, 0.0839], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0167, 0.0189, 0.0170, 0.0204, 0.0212, 0.0193, 0.0189], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 09:24:51,548 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159425.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:25:02,969 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159433.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:25:11,783 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159439.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:25:20,966 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8630, 1.9212, 2.3001, 3.1307, 2.1373, 2.1625, 2.1407, 2.0334], device='cuda:6'), covar=tensor([0.1261, 0.4051, 0.2369, 0.0673, 0.4222, 0.2670, 0.3582, 0.3930], device='cuda:6'), in_proj_covar=tensor([0.0378, 0.0418, 0.0347, 0.0317, 0.0424, 0.0481, 0.0388, 0.0487], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:25:30,663 INFO [train.py:904] (6/8) Epoch 16, batch 7200, loss[loss=0.1891, simple_loss=0.2772, pruned_loss=0.05052, over 16427.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2959, pruned_loss=0.06277, over 3081986.71 frames. ], batch size: 146, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:26:15,869 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159481.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:26:26,897 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159487.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:26:33,431 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159491.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 09:26:50,664 INFO [train.py:904] (6/8) Epoch 16, batch 7250, loss[loss=0.1963, simple_loss=0.2733, pruned_loss=0.05971, over 16594.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2942, pruned_loss=0.06215, over 3060508.28 frames. ], batch size: 62, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:27:11,617 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159515.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:27:23,390 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.928e+02 2.845e+02 3.560e+02 4.449e+02 7.164e+02, threshold=7.119e+02, percent-clipped=0.0 2023-04-30 09:27:47,288 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159539.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 09:27:57,331 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 09:28:06,272 INFO [train.py:904] (6/8) Epoch 16, batch 7300, loss[loss=0.2457, simple_loss=0.3348, pruned_loss=0.07834, over 16851.00 frames. ], tot_loss[loss=0.209, simple_loss=0.2935, pruned_loss=0.06227, over 3055672.50 frames. ], batch size: 116, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:28:33,039 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5839, 3.6299, 2.7244, 2.1613, 2.5058, 2.2891, 3.8803, 3.3449], device='cuda:6'), covar=tensor([0.2861, 0.0643, 0.1833, 0.2421, 0.2659, 0.2101, 0.0413, 0.1108], device='cuda:6'), in_proj_covar=tensor([0.0317, 0.0263, 0.0297, 0.0298, 0.0289, 0.0240, 0.0282, 0.0318], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:29:22,389 INFO [train.py:904] (6/8) Epoch 16, batch 7350, loss[loss=0.2076, simple_loss=0.2941, pruned_loss=0.06055, over 17017.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.294, pruned_loss=0.06258, over 3066826.65 frames. ], batch size: 55, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:29:56,736 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 3.169e+02 3.768e+02 4.450e+02 9.358e+02, threshold=7.536e+02, percent-clipped=3.0 2023-04-30 09:30:07,816 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=159631.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:30:24,583 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2003, 4.8227, 4.7710, 3.4248, 4.1022, 4.7642, 4.1420, 3.0122], device='cuda:6'), covar=tensor([0.0409, 0.0039, 0.0032, 0.0307, 0.0081, 0.0098, 0.0069, 0.0338], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0075, 0.0076, 0.0131, 0.0089, 0.0100, 0.0088, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:30:31,262 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159646.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:30:39,217 INFO [train.py:904] (6/8) Epoch 16, batch 7400, loss[loss=0.2495, simple_loss=0.3138, pruned_loss=0.09258, over 11441.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2956, pruned_loss=0.06401, over 3034109.51 frames. ], batch size: 247, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:30:51,400 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0441, 5.5812, 5.7916, 5.5158, 5.5750, 6.0904, 5.6269, 5.3928], device='cuda:6'), covar=tensor([0.0840, 0.1876, 0.2048, 0.1836, 0.2192, 0.0957, 0.1337, 0.2415], device='cuda:6'), in_proj_covar=tensor([0.0384, 0.0546, 0.0596, 0.0461, 0.0611, 0.0629, 0.0470, 0.0617], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:30:59,156 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6896, 3.9587, 3.1336, 2.3287, 2.8118, 2.5423, 4.2272, 3.5707], device='cuda:6'), covar=tensor([0.2905, 0.0637, 0.1598, 0.2482, 0.2419, 0.1917, 0.0454, 0.1159], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0264, 0.0298, 0.0298, 0.0290, 0.0241, 0.0283, 0.0320], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:31:41,290 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=159692.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:31:56,845 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1513, 2.3118, 1.6955, 1.9993, 2.7003, 2.4353, 2.9184, 3.0349], device='cuda:6'), covar=tensor([0.0162, 0.0497, 0.0682, 0.0523, 0.0280, 0.0397, 0.0237, 0.0246], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0216, 0.0210, 0.0211, 0.0216, 0.0216, 0.0219, 0.0212], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:31:57,477 INFO [train.py:904] (6/8) Epoch 16, batch 7450, loss[loss=0.2087, simple_loss=0.3024, pruned_loss=0.05749, over 16862.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2961, pruned_loss=0.06428, over 3054142.15 frames. ], batch size: 83, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:32:05,505 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 09:32:33,424 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.461e+02 3.124e+02 3.664e+02 4.558e+02 8.463e+02, threshold=7.327e+02, percent-clipped=4.0 2023-04-30 09:32:36,498 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159725.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:32:53,215 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1321, 3.5112, 3.6652, 2.3202, 3.3102, 3.6375, 3.3849, 2.1794], device='cuda:6'), covar=tensor([0.0516, 0.0070, 0.0052, 0.0392, 0.0084, 0.0107, 0.0087, 0.0388], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0075, 0.0076, 0.0130, 0.0088, 0.0099, 0.0087, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:33:17,744 INFO [train.py:904] (6/8) Epoch 16, batch 7500, loss[loss=0.2526, simple_loss=0.3195, pruned_loss=0.09287, over 11403.00 frames. ], tot_loss[loss=0.2121, simple_loss=0.2966, pruned_loss=0.06385, over 3051289.03 frames. ], batch size: 247, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:33:51,103 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159773.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:34:37,012 INFO [train.py:904] (6/8) Epoch 16, batch 7550, loss[loss=0.2143, simple_loss=0.2978, pruned_loss=0.06542, over 15449.00 frames. ], tot_loss[loss=0.2121, simple_loss=0.2957, pruned_loss=0.06427, over 3034861.23 frames. ], batch size: 191, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:34:58,295 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159815.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:35:10,142 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.970e+02 2.719e+02 3.178e+02 3.820e+02 7.109e+02, threshold=6.356e+02, percent-clipped=0.0 2023-04-30 09:35:53,836 INFO [train.py:904] (6/8) Epoch 16, batch 7600, loss[loss=0.2148, simple_loss=0.2952, pruned_loss=0.06723, over 16366.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2951, pruned_loss=0.06472, over 3015885.18 frames. ], batch size: 35, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:35:59,113 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=159855.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:36:10,414 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159863.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:37:09,858 INFO [train.py:904] (6/8) Epoch 16, batch 7650, loss[loss=0.21, simple_loss=0.2968, pruned_loss=0.06162, over 16783.00 frames. ], tot_loss[loss=0.2134, simple_loss=0.2959, pruned_loss=0.06547, over 3013631.50 frames. ], batch size: 124, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:37:21,778 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=159910.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:37:25,948 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-30 09:37:30,474 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=159916.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:37:43,257 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.780e+02 3.207e+02 4.068e+02 5.310e+02 2.281e+03, threshold=8.135e+02, percent-clipped=15.0 2023-04-30 09:38:13,587 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159946.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:38:22,085 INFO [train.py:904] (6/8) Epoch 16, batch 7700, loss[loss=0.2307, simple_loss=0.3099, pruned_loss=0.07572, over 15308.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2956, pruned_loss=0.06532, over 3029382.06 frames. ], batch size: 191, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:38:51,313 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=159971.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:38:54,601 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.04 vs. limit=5.0 2023-04-30 09:39:15,437 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159987.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:39:26,409 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159994.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:39:41,051 INFO [train.py:904] (6/8) Epoch 16, batch 7750, loss[loss=0.1794, simple_loss=0.2694, pruned_loss=0.04467, over 17150.00 frames. ], tot_loss[loss=0.2129, simple_loss=0.2955, pruned_loss=0.06509, over 3032290.96 frames. ], batch size: 46, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:40:13,566 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.949e+02 3.102e+02 3.551e+02 4.185e+02 8.391e+02, threshold=7.102e+02, percent-clipped=1.0 2023-04-30 09:40:14,472 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-30 09:40:21,627 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4611, 4.3075, 4.5195, 4.6915, 4.8229, 4.4035, 4.8225, 4.7976], device='cuda:6'), covar=tensor([0.1774, 0.1183, 0.1422, 0.0655, 0.0583, 0.0980, 0.0600, 0.0661], device='cuda:6'), in_proj_covar=tensor([0.0576, 0.0709, 0.0833, 0.0713, 0.0538, 0.0566, 0.0578, 0.0674], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:40:53,467 INFO [train.py:904] (6/8) Epoch 16, batch 7800, loss[loss=0.2112, simple_loss=0.2982, pruned_loss=0.06205, over 17244.00 frames. ], tot_loss[loss=0.2138, simple_loss=0.2964, pruned_loss=0.06564, over 3035686.56 frames. ], batch size: 52, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:41:19,768 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1661, 3.1953, 1.9420, 3.4079, 2.4827, 3.4749, 2.1423, 2.6472], device='cuda:6'), covar=tensor([0.0278, 0.0380, 0.1582, 0.0267, 0.0761, 0.0654, 0.1437, 0.0707], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0169, 0.0191, 0.0147, 0.0172, 0.0209, 0.0200, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 09:41:28,374 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3018, 1.8357, 2.6584, 3.0357, 2.8652, 3.5323, 2.2203, 3.5376], device='cuda:6'), covar=tensor([0.0153, 0.0497, 0.0277, 0.0241, 0.0299, 0.0156, 0.0434, 0.0126], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0183, 0.0167, 0.0172, 0.0183, 0.0140, 0.0183, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:42:08,841 INFO [train.py:904] (6/8) Epoch 16, batch 7850, loss[loss=0.1945, simple_loss=0.2847, pruned_loss=0.0522, over 16755.00 frames. ], tot_loss[loss=0.2143, simple_loss=0.2978, pruned_loss=0.06536, over 3042862.62 frames. ], batch size: 83, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:42:43,406 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.920e+02 2.758e+02 3.305e+02 4.051e+02 1.069e+03, threshold=6.609e+02, percent-clipped=3.0 2023-04-30 09:42:46,907 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7911, 4.6134, 4.8723, 5.0302, 5.1948, 4.7071, 5.2036, 5.1577], device='cuda:6'), covar=tensor([0.1822, 0.1203, 0.1487, 0.0643, 0.0551, 0.0742, 0.0497, 0.0634], device='cuda:6'), in_proj_covar=tensor([0.0578, 0.0712, 0.0838, 0.0716, 0.0541, 0.0567, 0.0580, 0.0676], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:42:56,813 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2023-04-30 09:43:25,067 INFO [train.py:904] (6/8) Epoch 16, batch 7900, loss[loss=0.2043, simple_loss=0.2912, pruned_loss=0.05871, over 16690.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2959, pruned_loss=0.06424, over 3046565.10 frames. ], batch size: 134, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:43:56,095 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1618, 3.5652, 3.6470, 2.3298, 3.3332, 3.6274, 3.4313, 1.9614], device='cuda:6'), covar=tensor([0.0506, 0.0054, 0.0048, 0.0382, 0.0084, 0.0106, 0.0079, 0.0425], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0073, 0.0074, 0.0128, 0.0087, 0.0097, 0.0086, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:44:15,643 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-30 09:44:43,663 INFO [train.py:904] (6/8) Epoch 16, batch 7950, loss[loss=0.1997, simple_loss=0.2895, pruned_loss=0.05489, over 16747.00 frames. ], tot_loss[loss=0.2126, simple_loss=0.2963, pruned_loss=0.06447, over 3062777.94 frames. ], batch size: 83, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:44:56,921 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=160211.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:45:16,536 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.952e+02 2.700e+02 3.245e+02 4.105e+02 1.062e+03, threshold=6.489e+02, percent-clipped=6.0 2023-04-30 09:45:34,455 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9289, 1.8594, 2.1373, 3.4214, 1.8450, 2.1009, 1.9855, 1.9770], device='cuda:6'), covar=tensor([0.1472, 0.4226, 0.2854, 0.0686, 0.5305, 0.3248, 0.3939, 0.3999], device='cuda:6'), in_proj_covar=tensor([0.0377, 0.0420, 0.0346, 0.0317, 0.0426, 0.0483, 0.0388, 0.0488], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:45:40,453 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6302, 3.6011, 1.8044, 4.1956, 2.6565, 4.1183, 1.8737, 2.7818], device='cuda:6'), covar=tensor([0.0254, 0.0371, 0.2114, 0.0181, 0.0848, 0.0417, 0.2093, 0.0818], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0169, 0.0192, 0.0147, 0.0172, 0.0210, 0.0201, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 09:45:56,353 INFO [train.py:904] (6/8) Epoch 16, batch 8000, loss[loss=0.2263, simple_loss=0.3102, pruned_loss=0.07119, over 16750.00 frames. ], tot_loss[loss=0.214, simple_loss=0.2972, pruned_loss=0.06543, over 3055719.35 frames. ], batch size: 89, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:46:18,473 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=160266.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:46:50,246 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=160287.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:47:12,589 INFO [train.py:904] (6/8) Epoch 16, batch 8050, loss[loss=0.2041, simple_loss=0.2989, pruned_loss=0.05468, over 16789.00 frames. ], tot_loss[loss=0.2132, simple_loss=0.2968, pruned_loss=0.06482, over 3075823.80 frames. ], batch size: 83, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:47:16,323 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6685, 3.6298, 2.8142, 2.1833, 2.5361, 2.3024, 3.9404, 3.3656], device='cuda:6'), covar=tensor([0.2778, 0.0764, 0.1789, 0.2623, 0.2704, 0.2134, 0.0491, 0.1225], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0263, 0.0297, 0.0297, 0.0290, 0.0240, 0.0281, 0.0319], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:47:20,997 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=160307.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 09:47:47,809 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.763e+02 2.953e+02 3.727e+02 4.663e+02 8.321e+02, threshold=7.453e+02, percent-clipped=3.0 2023-04-30 09:48:04,059 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=160335.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:48:29,600 INFO [train.py:904] (6/8) Epoch 16, batch 8100, loss[loss=0.2178, simple_loss=0.3011, pruned_loss=0.0673, over 16689.00 frames. ], tot_loss[loss=0.212, simple_loss=0.2958, pruned_loss=0.06408, over 3081854.63 frames. ], batch size: 62, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:48:54,835 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=160368.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 09:49:17,630 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3914, 3.6473, 3.7721, 2.2134, 3.1502, 2.4881, 3.7687, 3.8325], device='cuda:6'), covar=tensor([0.0233, 0.0712, 0.0557, 0.1886, 0.0793, 0.0941, 0.0531, 0.0951], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0156, 0.0163, 0.0149, 0.0142, 0.0126, 0.0142, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 09:49:35,537 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4396, 2.9185, 2.9790, 1.9366, 2.6498, 2.0783, 3.0402, 3.1374], device='cuda:6'), covar=tensor([0.0288, 0.0760, 0.0629, 0.1997, 0.0920, 0.1011, 0.0646, 0.0914], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0157, 0.0163, 0.0149, 0.0142, 0.0126, 0.0142, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 09:49:46,003 INFO [train.py:904] (6/8) Epoch 16, batch 8150, loss[loss=0.1899, simple_loss=0.2733, pruned_loss=0.05321, over 16865.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2934, pruned_loss=0.06295, over 3082640.85 frames. ], batch size: 116, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:49:49,424 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5421, 3.5246, 2.7465, 2.1853, 2.3034, 2.3235, 3.6683, 3.1668], device='cuda:6'), covar=tensor([0.2917, 0.0694, 0.1785, 0.2626, 0.2764, 0.2038, 0.0521, 0.1333], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0263, 0.0296, 0.0297, 0.0289, 0.0240, 0.0281, 0.0319], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:50:21,735 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.821e+02 3.128e+02 3.748e+02 4.647e+02 9.176e+02, threshold=7.497e+02, percent-clipped=1.0 2023-04-30 09:50:58,985 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-30 09:51:05,066 INFO [train.py:904] (6/8) Epoch 16, batch 8200, loss[loss=0.2018, simple_loss=0.2839, pruned_loss=0.0598, over 16595.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.29, pruned_loss=0.0618, over 3102138.97 frames. ], batch size: 62, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:52:16,485 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7371, 2.8731, 2.3611, 4.1708, 2.6767, 4.0312, 1.7248, 2.9162], device='cuda:6'), covar=tensor([0.1228, 0.0642, 0.1132, 0.0173, 0.0141, 0.0429, 0.1384, 0.0677], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0166, 0.0188, 0.0170, 0.0203, 0.0211, 0.0192, 0.0188], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 09:52:27,310 INFO [train.py:904] (6/8) Epoch 16, batch 8250, loss[loss=0.2017, simple_loss=0.2968, pruned_loss=0.05331, over 16220.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2894, pruned_loss=0.05959, over 3090088.68 frames. ], batch size: 165, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:52:42,494 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=160511.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:52:44,954 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0279, 3.0972, 1.8734, 3.2895, 2.3632, 3.3239, 2.1403, 2.6388], device='cuda:6'), covar=tensor([0.0277, 0.0344, 0.1492, 0.0257, 0.0745, 0.0580, 0.1428, 0.0621], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0170, 0.0192, 0.0147, 0.0172, 0.0210, 0.0201, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 09:53:04,606 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.122e+02 2.811e+02 3.275e+02 4.165e+02 1.278e+03, threshold=6.551e+02, percent-clipped=3.0 2023-04-30 09:53:49,472 INFO [train.py:904] (6/8) Epoch 16, batch 8300, loss[loss=0.1709, simple_loss=0.2669, pruned_loss=0.03744, over 16221.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2865, pruned_loss=0.05692, over 3066712.18 frames. ], batch size: 165, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:54:01,473 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=160559.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:54:13,102 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=160566.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:55:09,921 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6484, 3.7687, 3.8250, 2.8477, 3.4295, 3.8190, 3.6203, 2.3023], device='cuda:6'), covar=tensor([0.0359, 0.0047, 0.0037, 0.0273, 0.0087, 0.0073, 0.0065, 0.0402], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0074, 0.0074, 0.0129, 0.0087, 0.0097, 0.0086, 0.0121], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 09:55:10,565 INFO [train.py:904] (6/8) Epoch 16, batch 8350, loss[loss=0.1581, simple_loss=0.2621, pruned_loss=0.02701, over 16799.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2864, pruned_loss=0.05496, over 3088850.73 frames. ], batch size: 102, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 09:55:27,737 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.36 vs. limit=5.0 2023-04-30 09:55:30,913 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=160614.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:55:48,785 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.217e+02 2.608e+02 3.326e+02 6.898e+02, threshold=5.216e+02, percent-clipped=1.0 2023-04-30 09:56:08,601 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3983, 3.5823, 3.2454, 3.0386, 2.9337, 3.4230, 3.2290, 3.2137], device='cuda:6'), covar=tensor([0.0896, 0.0722, 0.0430, 0.0391, 0.1078, 0.0556, 0.2301, 0.0645], device='cuda:6'), in_proj_covar=tensor([0.0263, 0.0369, 0.0307, 0.0296, 0.0319, 0.0346, 0.0213, 0.0368], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 09:56:33,054 INFO [train.py:904] (6/8) Epoch 16, batch 8400, loss[loss=0.1981, simple_loss=0.2796, pruned_loss=0.05831, over 11908.00 frames. ], tot_loss[loss=0.1946, simple_loss=0.2837, pruned_loss=0.05274, over 3085434.15 frames. ], batch size: 247, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 09:56:51,248 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=160663.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 09:57:54,560 INFO [train.py:904] (6/8) Epoch 16, batch 8450, loss[loss=0.1746, simple_loss=0.2705, pruned_loss=0.03938, over 15184.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.282, pruned_loss=0.0512, over 3079693.71 frames. ], batch size: 190, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 09:57:55,712 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 09:58:31,816 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 2.257e+02 2.696e+02 3.215e+02 7.413e+02, threshold=5.391e+02, percent-clipped=3.0 2023-04-30 09:59:15,543 INFO [train.py:904] (6/8) Epoch 16, batch 8500, loss[loss=0.1794, simple_loss=0.2563, pruned_loss=0.05125, over 12050.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2781, pruned_loss=0.04878, over 3069012.62 frames. ], batch size: 246, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:00:07,379 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8368, 3.2807, 3.5637, 2.1451, 2.8925, 2.2443, 3.3715, 3.4802], device='cuda:6'), covar=tensor([0.0302, 0.0853, 0.0439, 0.1885, 0.0823, 0.1026, 0.0688, 0.0970], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0152, 0.0159, 0.0145, 0.0138, 0.0124, 0.0138, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 10:00:08,767 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0206, 2.3592, 2.3956, 3.1024, 2.0358, 3.3296, 1.7975, 2.8299], device='cuda:6'), covar=tensor([0.1062, 0.0588, 0.0924, 0.0167, 0.0099, 0.0349, 0.1370, 0.0588], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0163, 0.0185, 0.0166, 0.0199, 0.0208, 0.0190, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 10:00:42,444 INFO [train.py:904] (6/8) Epoch 16, batch 8550, loss[loss=0.2053, simple_loss=0.3064, pruned_loss=0.05217, over 16890.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2758, pruned_loss=0.04754, over 3054502.33 frames. ], batch size: 109, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:00:54,162 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6650, 2.6355, 1.8908, 2.7833, 2.1281, 2.7973, 2.1222, 2.4348], device='cuda:6'), covar=tensor([0.0289, 0.0313, 0.1217, 0.0269, 0.0663, 0.0421, 0.1165, 0.0579], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0165, 0.0186, 0.0143, 0.0168, 0.0203, 0.0196, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 10:01:07,649 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7822, 3.3118, 3.4441, 1.8249, 2.7511, 2.2819, 3.2600, 3.5123], device='cuda:6'), covar=tensor([0.0345, 0.0815, 0.0570, 0.2254, 0.1003, 0.1044, 0.0830, 0.0978], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0152, 0.0159, 0.0145, 0.0138, 0.0124, 0.0138, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 10:01:27,155 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.280e+02 2.208e+02 2.559e+02 3.052e+02 5.214e+02, threshold=5.118e+02, percent-clipped=0.0 2023-04-30 10:02:22,609 INFO [train.py:904] (6/8) Epoch 16, batch 8600, loss[loss=0.1686, simple_loss=0.2625, pruned_loss=0.03733, over 16493.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2759, pruned_loss=0.04662, over 3042525.66 frames. ], batch size: 68, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:02:23,520 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9828, 3.8462, 4.0351, 4.1293, 4.2633, 3.8235, 4.2116, 4.2548], device='cuda:6'), covar=tensor([0.1467, 0.1013, 0.1235, 0.0662, 0.0485, 0.1737, 0.0614, 0.0618], device='cuda:6'), in_proj_covar=tensor([0.0557, 0.0690, 0.0810, 0.0696, 0.0526, 0.0550, 0.0560, 0.0658], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:03:48,834 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.12 vs. limit=5.0 2023-04-30 10:04:02,534 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=160901.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:04:03,213 INFO [train.py:904] (6/8) Epoch 16, batch 8650, loss[loss=0.1546, simple_loss=0.2538, pruned_loss=0.02765, over 15388.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2743, pruned_loss=0.04517, over 3035322.28 frames. ], batch size: 191, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:04:22,436 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-30 10:04:56,266 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.325e+02 2.764e+02 3.397e+02 5.575e+02, threshold=5.528e+02, percent-clipped=1.0 2023-04-30 10:05:07,530 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7331, 2.6079, 2.4420, 3.7423, 2.3083, 3.8676, 1.5739, 2.9549], device='cuda:6'), covar=tensor([0.1392, 0.0719, 0.1098, 0.0180, 0.0098, 0.0314, 0.1621, 0.0668], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0163, 0.0186, 0.0166, 0.0197, 0.0208, 0.0190, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 10:05:52,474 INFO [train.py:904] (6/8) Epoch 16, batch 8700, loss[loss=0.1695, simple_loss=0.2606, pruned_loss=0.03925, over 16902.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2716, pruned_loss=0.04407, over 3042436.08 frames. ], batch size: 116, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:06:13,607 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=160962.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:06:15,135 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=160963.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 10:06:18,604 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=160965.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:06:27,644 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6582, 3.0951, 3.2793, 1.9171, 2.6861, 2.0481, 3.1864, 3.3101], device='cuda:6'), covar=tensor([0.0279, 0.0786, 0.0518, 0.1987, 0.0873, 0.1075, 0.0702, 0.0845], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0150, 0.0157, 0.0144, 0.0137, 0.0123, 0.0137, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 10:06:47,588 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5216, 3.4866, 3.4751, 2.5164, 3.4153, 1.8920, 3.2055, 2.7865], device='cuda:6'), covar=tensor([0.0186, 0.0167, 0.0212, 0.0382, 0.0137, 0.3053, 0.0214, 0.0367], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0130, 0.0175, 0.0160, 0.0148, 0.0188, 0.0163, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:06:50,116 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8689, 1.3422, 1.6994, 1.7396, 1.8725, 1.9397, 1.5619, 1.9039], device='cuda:6'), covar=tensor([0.0244, 0.0374, 0.0196, 0.0255, 0.0264, 0.0181, 0.0399, 0.0115], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0181, 0.0167, 0.0169, 0.0182, 0.0138, 0.0182, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:07:21,593 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5206, 3.6144, 2.8254, 2.0554, 2.3083, 2.2661, 3.8478, 3.1786], device='cuda:6'), covar=tensor([0.2919, 0.0569, 0.1672, 0.2954, 0.2693, 0.2029, 0.0428, 0.1256], device='cuda:6'), in_proj_covar=tensor([0.0309, 0.0255, 0.0288, 0.0289, 0.0277, 0.0234, 0.0273, 0.0308], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:07:29,335 INFO [train.py:904] (6/8) Epoch 16, batch 8750, loss[loss=0.1656, simple_loss=0.2559, pruned_loss=0.03768, over 12300.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2709, pruned_loss=0.0436, over 3028787.99 frames. ], batch size: 246, lr: 4.20e-03, grad_scale: 4.0 2023-04-30 10:07:53,658 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=161011.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 10:08:12,535 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1172, 5.0709, 4.7949, 4.3224, 4.9635, 1.9931, 4.6807, 4.7368], device='cuda:6'), covar=tensor([0.0056, 0.0056, 0.0151, 0.0237, 0.0069, 0.2208, 0.0100, 0.0131], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0130, 0.0175, 0.0159, 0.0148, 0.0188, 0.0162, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:08:19,622 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8388, 3.7503, 3.9336, 3.9953, 4.0913, 3.6819, 4.0672, 4.1197], device='cuda:6'), covar=tensor([0.1403, 0.1007, 0.1089, 0.0674, 0.0521, 0.1847, 0.0712, 0.0663], device='cuda:6'), in_proj_covar=tensor([0.0552, 0.0680, 0.0797, 0.0690, 0.0520, 0.0545, 0.0552, 0.0650], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:08:27,683 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.172e+02 2.627e+02 3.227e+02 5.767e+02, threshold=5.254e+02, percent-clipped=1.0 2023-04-30 10:08:28,274 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=161026.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:08:48,483 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1426, 1.5190, 1.8309, 2.0824, 2.2223, 2.3511, 1.7915, 2.2962], device='cuda:6'), covar=tensor([0.0222, 0.0448, 0.0247, 0.0285, 0.0298, 0.0184, 0.0398, 0.0144], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0181, 0.0167, 0.0170, 0.0182, 0.0138, 0.0182, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:09:20,775 INFO [train.py:904] (6/8) Epoch 16, batch 8800, loss[loss=0.2101, simple_loss=0.2884, pruned_loss=0.06592, over 12678.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2694, pruned_loss=0.04224, over 3044054.67 frames. ], batch size: 246, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:09:50,989 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-30 10:11:05,639 INFO [train.py:904] (6/8) Epoch 16, batch 8850, loss[loss=0.1855, simple_loss=0.2861, pruned_loss=0.04246, over 16688.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2717, pruned_loss=0.04172, over 3035986.65 frames. ], batch size: 134, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:11:55,942 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.310e+02 2.876e+02 3.561e+02 5.593e+02, threshold=5.752e+02, percent-clipped=2.0 2023-04-30 10:12:53,428 INFO [train.py:904] (6/8) Epoch 16, batch 8900, loss[loss=0.1578, simple_loss=0.2547, pruned_loss=0.0305, over 16723.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2723, pruned_loss=0.04141, over 3046718.11 frames. ], batch size: 83, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:14:49,814 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4722, 4.5175, 4.6628, 4.4967, 4.5101, 5.0392, 4.5896, 4.3372], device='cuda:6'), covar=tensor([0.1095, 0.1759, 0.1798, 0.1898, 0.2262, 0.0931, 0.1413, 0.2297], device='cuda:6'), in_proj_covar=tensor([0.0366, 0.0531, 0.0575, 0.0443, 0.0588, 0.0611, 0.0455, 0.0595], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 10:14:59,677 INFO [train.py:904] (6/8) Epoch 16, batch 8950, loss[loss=0.1675, simple_loss=0.2631, pruned_loss=0.03589, over 16400.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.272, pruned_loss=0.0417, over 3051579.45 frames. ], batch size: 146, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:15:30,195 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=161216.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:15:47,456 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.408e+02 2.276e+02 2.813e+02 3.338e+02 5.711e+02, threshold=5.626e+02, percent-clipped=0.0 2023-04-30 10:16:01,826 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=161231.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:16:07,620 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4998, 1.6346, 2.0368, 2.4252, 2.4162, 2.6465, 1.7081, 2.7145], device='cuda:6'), covar=tensor([0.0199, 0.0501, 0.0329, 0.0263, 0.0301, 0.0204, 0.0514, 0.0130], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0181, 0.0167, 0.0168, 0.0181, 0.0137, 0.0181, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:16:46,884 INFO [train.py:904] (6/8) Epoch 16, batch 9000, loss[loss=0.1816, simple_loss=0.2648, pruned_loss=0.04922, over 11842.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2693, pruned_loss=0.0404, over 3056486.81 frames. ], batch size: 246, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:16:46,885 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 10:16:56,909 INFO [train.py:938] (6/8) Epoch 16, validation: loss=0.1491, simple_loss=0.2531, pruned_loss=0.02259, over 944034.00 frames. 2023-04-30 10:16:56,910 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 10:17:08,071 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=161257.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:17:16,036 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-30 10:17:49,250 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=161277.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:18:20,838 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=161292.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:18:40,788 INFO [train.py:904] (6/8) Epoch 16, batch 9050, loss[loss=0.1687, simple_loss=0.2565, pruned_loss=0.04052, over 16228.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2703, pruned_loss=0.04065, over 3062870.80 frames. ], batch size: 165, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:18:49,649 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8219, 3.1671, 2.6949, 4.7965, 3.6487, 4.3963, 1.6949, 3.2994], device='cuda:6'), covar=tensor([0.1305, 0.0633, 0.1120, 0.0151, 0.0191, 0.0309, 0.1508, 0.0618], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0162, 0.0185, 0.0165, 0.0194, 0.0207, 0.0190, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 10:19:19,443 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=161321.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:19:25,515 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-30 10:19:27,691 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.317e+02 2.661e+02 3.212e+02 6.846e+02, threshold=5.322e+02, percent-clipped=5.0 2023-04-30 10:20:22,808 INFO [train.py:904] (6/8) Epoch 16, batch 9100, loss[loss=0.1785, simple_loss=0.2782, pruned_loss=0.03935, over 16370.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2697, pruned_loss=0.04089, over 3079684.83 frames. ], batch size: 146, lr: 4.19e-03, grad_scale: 4.0 2023-04-30 10:22:19,256 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=161401.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:22:19,974 INFO [train.py:904] (6/8) Epoch 16, batch 9150, loss[loss=0.1636, simple_loss=0.2609, pruned_loss=0.03311, over 15370.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2699, pruned_loss=0.04047, over 3070420.05 frames. ], batch size: 192, lr: 4.19e-03, grad_scale: 4.0 2023-04-30 10:23:13,972 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.610e+02 2.279e+02 2.699e+02 3.193e+02 4.519e+02, threshold=5.398e+02, percent-clipped=0.0 2023-04-30 10:24:04,517 INFO [train.py:904] (6/8) Epoch 16, batch 9200, loss[loss=0.1818, simple_loss=0.2774, pruned_loss=0.04307, over 15400.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2649, pruned_loss=0.039, over 3085471.08 frames. ], batch size: 191, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:24:20,931 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=161460.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:24:24,483 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=161462.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:25:42,692 INFO [train.py:904] (6/8) Epoch 16, batch 9250, loss[loss=0.1524, simple_loss=0.2547, pruned_loss=0.0251, over 16823.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2653, pruned_loss=0.03961, over 3080251.06 frames. ], batch size: 102, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:26:23,376 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=161521.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:26:36,793 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.444e+02 2.938e+02 3.576e+02 6.685e+02, threshold=5.875e+02, percent-clipped=5.0 2023-04-30 10:27:34,800 INFO [train.py:904] (6/8) Epoch 16, batch 9300, loss[loss=0.1621, simple_loss=0.2576, pruned_loss=0.03327, over 16391.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2639, pruned_loss=0.03909, over 3081585.90 frames. ], batch size: 146, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:27:45,401 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=161557.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:28:23,204 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=161572.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:28:54,783 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=161587.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:28:55,101 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.92 vs. limit=5.0 2023-04-30 10:29:21,686 INFO [train.py:904] (6/8) Epoch 16, batch 9350, loss[loss=0.1752, simple_loss=0.2686, pruned_loss=0.04088, over 16910.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2635, pruned_loss=0.0389, over 3087587.16 frames. ], batch size: 90, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:29:28,430 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=161605.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:29:54,065 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7572, 3.7377, 2.4033, 4.4075, 2.8252, 4.3157, 2.3982, 3.0700], device='cuda:6'), covar=tensor([0.0268, 0.0390, 0.1529, 0.0186, 0.0834, 0.0465, 0.1610, 0.0713], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0164, 0.0185, 0.0141, 0.0166, 0.0200, 0.0195, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 10:30:03,209 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=161621.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:30:12,560 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.368e+02 2.307e+02 2.790e+02 3.265e+02 7.040e+02, threshold=5.579e+02, percent-clipped=1.0 2023-04-30 10:31:04,000 INFO [train.py:904] (6/8) Epoch 16, batch 9400, loss[loss=0.1545, simple_loss=0.2424, pruned_loss=0.03325, over 12322.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2633, pruned_loss=0.03904, over 3059539.02 frames. ], batch size: 247, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:31:39,351 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=161669.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:32:06,412 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6772, 4.9571, 4.7609, 4.7841, 4.4531, 4.4803, 4.4206, 5.0134], device='cuda:6'), covar=tensor([0.1100, 0.0898, 0.0874, 0.0712, 0.0827, 0.1081, 0.1161, 0.0874], device='cuda:6'), in_proj_covar=tensor([0.0584, 0.0717, 0.0583, 0.0527, 0.0453, 0.0466, 0.0596, 0.0549], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:32:06,963 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-30 10:32:44,154 INFO [train.py:904] (6/8) Epoch 16, batch 9450, loss[loss=0.1742, simple_loss=0.2658, pruned_loss=0.04132, over 16973.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2652, pruned_loss=0.03928, over 3060081.57 frames. ], batch size: 109, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:33:33,849 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.284e+02 2.576e+02 3.195e+02 6.155e+02, threshold=5.152e+02, percent-clipped=3.0 2023-04-30 10:34:23,901 INFO [train.py:904] (6/8) Epoch 16, batch 9500, loss[loss=0.1727, simple_loss=0.2693, pruned_loss=0.03808, over 15368.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2649, pruned_loss=0.03902, over 3064483.30 frames. ], batch size: 190, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:34:37,544 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=161757.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:36:00,149 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-30 10:36:08,214 INFO [train.py:904] (6/8) Epoch 16, batch 9550, loss[loss=0.193, simple_loss=0.2938, pruned_loss=0.04608, over 16449.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2651, pruned_loss=0.03935, over 3079369.70 frames. ], batch size: 147, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:36:38,443 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=161815.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:36:40,813 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=161816.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:37:00,496 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.628e+02 2.146e+02 2.559e+02 3.093e+02 5.125e+02, threshold=5.118e+02, percent-clipped=0.0 2023-04-30 10:37:29,466 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3261, 4.2959, 4.1735, 3.6424, 4.2052, 1.7466, 4.0330, 3.9341], device='cuda:6'), covar=tensor([0.0089, 0.0079, 0.0170, 0.0260, 0.0103, 0.2489, 0.0127, 0.0226], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0128, 0.0172, 0.0155, 0.0147, 0.0186, 0.0160, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:37:51,435 INFO [train.py:904] (6/8) Epoch 16, batch 9600, loss[loss=0.2001, simple_loss=0.2888, pruned_loss=0.05574, over 16586.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2667, pruned_loss=0.03999, over 3080487.42 frames. ], batch size: 68, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:38:29,816 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=161872.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:38:38,451 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=161876.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:39:02,414 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=161887.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:39:18,803 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4480, 4.4151, 4.2877, 3.7377, 4.2897, 1.6427, 4.1175, 4.1543], device='cuda:6'), covar=tensor([0.0084, 0.0082, 0.0179, 0.0295, 0.0106, 0.2541, 0.0123, 0.0203], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0128, 0.0172, 0.0156, 0.0147, 0.0187, 0.0160, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:39:37,497 INFO [train.py:904] (6/8) Epoch 16, batch 9650, loss[loss=0.1826, simple_loss=0.2743, pruned_loss=0.04552, over 16909.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2682, pruned_loss=0.0401, over 3089504.67 frames. ], batch size: 116, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:39:44,248 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7073, 3.2921, 3.2620, 1.8802, 2.8050, 2.2017, 3.1747, 3.3134], device='cuda:6'), covar=tensor([0.0355, 0.0691, 0.0566, 0.2063, 0.0823, 0.0975, 0.0812, 0.0974], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0147, 0.0158, 0.0145, 0.0136, 0.0123, 0.0137, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 10:39:59,826 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1228, 2.5625, 2.6567, 1.8923, 2.7741, 2.8448, 2.5207, 2.4475], device='cuda:6'), covar=tensor([0.0674, 0.0241, 0.0243, 0.1022, 0.0093, 0.0224, 0.0455, 0.0415], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0103, 0.0088, 0.0136, 0.0071, 0.0113, 0.0123, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 10:40:22,072 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=161920.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:40:36,068 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.427e+02 2.311e+02 2.824e+02 3.460e+02 5.730e+02, threshold=5.647e+02, percent-clipped=1.0 2023-04-30 10:40:49,470 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=161935.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:41:23,666 INFO [train.py:904] (6/8) Epoch 16, batch 9700, loss[loss=0.1845, simple_loss=0.2748, pruned_loss=0.04708, over 16463.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.267, pruned_loss=0.03987, over 3074376.91 frames. ], batch size: 146, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:42:34,607 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1865, 2.0999, 2.0991, 3.7196, 2.0121, 2.4025, 2.2158, 2.2585], device='cuda:6'), covar=tensor([0.1066, 0.3615, 0.2882, 0.0467, 0.4291, 0.2416, 0.3657, 0.3139], device='cuda:6'), in_proj_covar=tensor([0.0368, 0.0406, 0.0341, 0.0308, 0.0415, 0.0465, 0.0377, 0.0471], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:42:39,749 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2927, 4.2687, 4.4311, 4.3175, 4.3517, 4.8253, 4.3851, 4.1376], device='cuda:6'), covar=tensor([0.1395, 0.2124, 0.2543, 0.1908, 0.2302, 0.1071, 0.1525, 0.2387], device='cuda:6'), in_proj_covar=tensor([0.0356, 0.0517, 0.0565, 0.0432, 0.0574, 0.0599, 0.0446, 0.0579], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 10:43:08,495 INFO [train.py:904] (6/8) Epoch 16, batch 9750, loss[loss=0.1776, simple_loss=0.2698, pruned_loss=0.04275, over 16712.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2659, pruned_loss=0.04005, over 3058523.61 frames. ], batch size: 134, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:43:13,352 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0262, 3.3856, 3.3701, 2.1867, 3.0530, 3.3681, 3.2929, 1.9602], device='cuda:6'), covar=tensor([0.0511, 0.0039, 0.0049, 0.0408, 0.0095, 0.0081, 0.0065, 0.0455], device='cuda:6'), in_proj_covar=tensor([0.0130, 0.0073, 0.0074, 0.0129, 0.0088, 0.0096, 0.0085, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 10:43:58,494 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.156e+02 2.615e+02 3.105e+02 5.589e+02, threshold=5.230e+02, percent-clipped=0.0 2023-04-30 10:44:46,279 INFO [train.py:904] (6/8) Epoch 16, batch 9800, loss[loss=0.189, simple_loss=0.2943, pruned_loss=0.04184, over 15501.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2657, pruned_loss=0.03886, over 3078361.08 frames. ], batch size: 191, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:44:56,998 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=162057.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:45:34,282 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=162077.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:45:56,288 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-04-30 10:46:31,255 INFO [train.py:904] (6/8) Epoch 16, batch 9850, loss[loss=0.1713, simple_loss=0.2551, pruned_loss=0.04374, over 12458.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2671, pruned_loss=0.03911, over 3075341.54 frames. ], batch size: 248, lr: 4.18e-03, grad_scale: 8.0 2023-04-30 10:46:38,614 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=162105.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:46:59,897 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=162116.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:47:23,357 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 2.292e+02 2.812e+02 3.288e+02 4.754e+02, threshold=5.623e+02, percent-clipped=0.0 2023-04-30 10:47:51,618 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=162138.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:48:22,256 INFO [train.py:904] (6/8) Epoch 16, batch 9900, loss[loss=0.18, simple_loss=0.2837, pruned_loss=0.03815, over 16886.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2672, pruned_loss=0.03898, over 3065993.48 frames. ], batch size: 116, lr: 4.18e-03, grad_scale: 8.0 2023-04-30 10:48:52,743 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=162164.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:49:08,595 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=162171.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:49:21,676 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1787, 2.1109, 2.0702, 3.8491, 1.9350, 2.4282, 2.1997, 2.2322], device='cuda:6'), covar=tensor([0.1174, 0.3442, 0.2866, 0.0471, 0.4275, 0.2435, 0.3263, 0.3548], device='cuda:6'), in_proj_covar=tensor([0.0368, 0.0405, 0.0341, 0.0307, 0.0413, 0.0464, 0.0376, 0.0470], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:50:22,154 INFO [train.py:904] (6/8) Epoch 16, batch 9950, loss[loss=0.1761, simple_loss=0.2678, pruned_loss=0.04219, over 16861.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2696, pruned_loss=0.03925, over 3074232.57 frames. ], batch size: 116, lr: 4.18e-03, grad_scale: 8.0 2023-04-30 10:50:43,898 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9512, 3.3606, 3.2973, 1.9263, 2.8160, 2.3091, 3.3957, 3.5846], device='cuda:6'), covar=tensor([0.0240, 0.0641, 0.0666, 0.1915, 0.0820, 0.0914, 0.0661, 0.0722], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0146, 0.0157, 0.0145, 0.0136, 0.0123, 0.0136, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 10:50:48,920 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7914, 3.8004, 3.9660, 3.8106, 3.8916, 4.3173, 3.9638, 3.6949], device='cuda:6'), covar=tensor([0.2658, 0.2499, 0.2122, 0.2516, 0.2879, 0.1771, 0.1571, 0.2838], device='cuda:6'), in_proj_covar=tensor([0.0355, 0.0513, 0.0562, 0.0432, 0.0573, 0.0597, 0.0446, 0.0578], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 10:51:26,523 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.161e+02 2.630e+02 3.300e+02 8.357e+02, threshold=5.260e+02, percent-clipped=1.0 2023-04-30 10:51:36,407 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=162231.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:52:11,700 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 10:52:23,901 INFO [train.py:904] (6/8) Epoch 16, batch 10000, loss[loss=0.1685, simple_loss=0.2693, pruned_loss=0.03382, over 15217.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2685, pruned_loss=0.0391, over 3077663.64 frames. ], batch size: 190, lr: 4.18e-03, grad_scale: 8.0 2023-04-30 10:52:25,061 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-04-30 10:53:10,770 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6838, 2.8781, 2.7424, 4.7177, 3.4633, 4.4332, 1.5147, 3.1490], device='cuda:6'), covar=tensor([0.1410, 0.0718, 0.1088, 0.0124, 0.0188, 0.0285, 0.1678, 0.0678], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0161, 0.0184, 0.0161, 0.0188, 0.0203, 0.0189, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:6') 2023-04-30 10:53:40,262 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 10:53:46,679 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=162292.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:54:04,077 INFO [train.py:904] (6/8) Epoch 16, batch 10050, loss[loss=0.176, simple_loss=0.2763, pruned_loss=0.0378, over 16683.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2689, pruned_loss=0.03888, over 3084707.98 frames. ], batch size: 134, lr: 4.18e-03, grad_scale: 8.0 2023-04-30 10:54:37,496 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-30 10:54:54,671 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 2.225e+02 2.711e+02 3.158e+02 7.314e+02, threshold=5.421e+02, percent-clipped=6.0 2023-04-30 10:55:38,478 INFO [train.py:904] (6/8) Epoch 16, batch 10100, loss[loss=0.1653, simple_loss=0.2489, pruned_loss=0.04083, over 12563.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2688, pruned_loss=0.03895, over 3077186.29 frames. ], batch size: 248, lr: 4.18e-03, grad_scale: 8.0 2023-04-30 10:57:23,029 INFO [train.py:904] (6/8) Epoch 17, batch 0, loss[loss=0.2189, simple_loss=0.3058, pruned_loss=0.06602, over 16635.00 frames. ], tot_loss[loss=0.2189, simple_loss=0.3058, pruned_loss=0.06602, over 16635.00 frames. ], batch size: 62, lr: 4.05e-03, grad_scale: 8.0 2023-04-30 10:57:23,030 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 10:57:30,749 INFO [train.py:938] (6/8) Epoch 17, validation: loss=0.1481, simple_loss=0.2518, pruned_loss=0.02217, over 944034.00 frames. 2023-04-30 10:57:30,750 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 10:57:39,302 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4209, 5.8036, 5.4977, 5.5756, 5.1787, 5.1105, 5.2173, 5.8529], device='cuda:6'), covar=tensor([0.1288, 0.0876, 0.1318, 0.0783, 0.0891, 0.0741, 0.1064, 0.0969], device='cuda:6'), in_proj_covar=tensor([0.0582, 0.0715, 0.0578, 0.0526, 0.0452, 0.0464, 0.0595, 0.0553], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:58:09,502 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.615e+02 2.232e+02 2.721e+02 3.359e+02 8.921e+02, threshold=5.442e+02, percent-clipped=3.0 2023-04-30 10:58:14,184 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=162433.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:58:39,678 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-30 10:58:39,851 INFO [train.py:904] (6/8) Epoch 17, batch 50, loss[loss=0.1895, simple_loss=0.2827, pruned_loss=0.04815, over 16766.00 frames. ], tot_loss[loss=0.189, simple_loss=0.275, pruned_loss=0.05147, over 757032.42 frames. ], batch size: 57, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 10:59:06,474 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=162471.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:59:15,557 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0847, 2.6019, 2.1368, 2.3355, 2.9991, 2.7706, 3.1148, 3.1085], device='cuda:6'), covar=tensor([0.0194, 0.0315, 0.0419, 0.0389, 0.0194, 0.0278, 0.0223, 0.0228], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0218, 0.0212, 0.0211, 0.0216, 0.0217, 0.0217, 0.0208], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 10:59:47,953 INFO [train.py:904] (6/8) Epoch 17, batch 100, loss[loss=0.16, simple_loss=0.2501, pruned_loss=0.03499, over 15862.00 frames. ], tot_loss[loss=0.1858, simple_loss=0.2708, pruned_loss=0.05039, over 1330744.23 frames. ], batch size: 35, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 11:00:00,670 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=162511.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:00:12,326 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=162519.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:00:26,385 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.355e+02 2.682e+02 3.198e+02 6.473e+02, threshold=5.364e+02, percent-clipped=1.0 2023-04-30 11:00:56,544 INFO [train.py:904] (6/8) Epoch 17, batch 150, loss[loss=0.1847, simple_loss=0.2857, pruned_loss=0.04184, over 17099.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.267, pruned_loss=0.04879, over 1777188.11 frames. ], batch size: 49, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 11:01:23,595 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=162572.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 11:01:44,473 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=162587.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:01:54,942 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0011, 3.2266, 3.2002, 2.1572, 2.8022, 2.2537, 3.4245, 3.4816], device='cuda:6'), covar=tensor([0.0258, 0.0800, 0.0660, 0.1785, 0.0872, 0.0992, 0.0579, 0.0817], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0150, 0.0160, 0.0148, 0.0139, 0.0125, 0.0139, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 11:01:59,170 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-30 11:02:05,940 INFO [train.py:904] (6/8) Epoch 17, batch 200, loss[loss=0.1697, simple_loss=0.2607, pruned_loss=0.0393, over 16655.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2675, pruned_loss=0.04894, over 2122150.29 frames. ], batch size: 57, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 11:02:43,590 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.407e+02 2.768e+02 3.167e+02 5.394e+02, threshold=5.535e+02, percent-clipped=0.0 2023-04-30 11:02:58,835 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-30 11:03:12,317 INFO [train.py:904] (6/8) Epoch 17, batch 250, loss[loss=0.1547, simple_loss=0.2459, pruned_loss=0.03175, over 16861.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2667, pruned_loss=0.04934, over 2380379.66 frames. ], batch size: 42, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 11:03:47,994 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9977, 2.0878, 2.5591, 2.9331, 2.8573, 3.1149, 2.1298, 3.1415], device='cuda:6'), covar=tensor([0.0181, 0.0402, 0.0287, 0.0257, 0.0263, 0.0175, 0.0441, 0.0138], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0184, 0.0171, 0.0173, 0.0185, 0.0141, 0.0185, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:04:14,959 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7624, 4.8834, 5.0454, 4.8996, 4.9166, 5.5733, 5.0503, 4.7618], device='cuda:6'), covar=tensor([0.1275, 0.2211, 0.2560, 0.2239, 0.3097, 0.1085, 0.1581, 0.2313], device='cuda:6'), in_proj_covar=tensor([0.0376, 0.0546, 0.0600, 0.0458, 0.0610, 0.0632, 0.0473, 0.0610], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 11:04:20,330 INFO [train.py:904] (6/8) Epoch 17, batch 300, loss[loss=0.1714, simple_loss=0.263, pruned_loss=0.03994, over 17241.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2649, pruned_loss=0.0485, over 2594868.94 frames. ], batch size: 52, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 11:04:59,733 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 2.237e+02 2.646e+02 3.133e+02 7.879e+02, threshold=5.293e+02, percent-clipped=2.0 2023-04-30 11:05:03,720 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=162733.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:05:29,583 INFO [train.py:904] (6/8) Epoch 17, batch 350, loss[loss=0.1702, simple_loss=0.256, pruned_loss=0.04219, over 15961.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2612, pruned_loss=0.04616, over 2757556.18 frames. ], batch size: 35, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 11:05:53,415 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.77 vs. limit=5.0 2023-04-30 11:06:07,974 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=162781.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:06:36,763 INFO [train.py:904] (6/8) Epoch 17, batch 400, loss[loss=0.1862, simple_loss=0.262, pruned_loss=0.05513, over 16934.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2599, pruned_loss=0.04606, over 2889725.33 frames. ], batch size: 90, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:07:11,028 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=162827.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:07:15,899 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.145e+02 2.685e+02 3.328e+02 1.079e+03, threshold=5.370e+02, percent-clipped=6.0 2023-04-30 11:07:46,582 INFO [train.py:904] (6/8) Epoch 17, batch 450, loss[loss=0.1808, simple_loss=0.2773, pruned_loss=0.0422, over 17012.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2595, pruned_loss=0.04539, over 2986711.24 frames. ], batch size: 55, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:08:06,774 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=162867.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 11:08:16,850 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0537, 5.0142, 4.7819, 4.3063, 4.8547, 1.9328, 4.6240, 4.7059], device='cuda:6'), covar=tensor([0.0086, 0.0087, 0.0205, 0.0369, 0.0110, 0.2715, 0.0141, 0.0205], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0136, 0.0182, 0.0164, 0.0156, 0.0197, 0.0169, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:08:34,229 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=162887.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:08:35,603 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=162888.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:08:55,330 INFO [train.py:904] (6/8) Epoch 17, batch 500, loss[loss=0.1671, simple_loss=0.2614, pruned_loss=0.03639, over 17111.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2583, pruned_loss=0.04427, over 3071051.96 frames. ], batch size: 47, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:09:00,360 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9111, 3.2246, 2.8897, 5.0893, 4.2896, 4.5810, 1.8930, 3.3434], device='cuda:6'), covar=tensor([0.1296, 0.0651, 0.1106, 0.0174, 0.0240, 0.0367, 0.1424, 0.0685], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0164, 0.0186, 0.0168, 0.0195, 0.0209, 0.0193, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 11:09:32,593 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.133e+02 2.499e+02 3.161e+02 5.064e+02, threshold=4.998e+02, percent-clipped=0.0 2023-04-30 11:09:39,868 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=162935.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:09:46,960 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1887, 3.1846, 3.4196, 2.4405, 3.1859, 3.5201, 3.2306, 1.9336], device='cuda:6'), covar=tensor([0.0511, 0.0155, 0.0061, 0.0371, 0.0102, 0.0101, 0.0101, 0.0495], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0078, 0.0077, 0.0133, 0.0091, 0.0100, 0.0089, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 11:10:01,803 INFO [train.py:904] (6/8) Epoch 17, batch 550, loss[loss=0.1583, simple_loss=0.2625, pruned_loss=0.02704, over 17123.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2575, pruned_loss=0.04418, over 3135551.96 frames. ], batch size: 47, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:11:10,218 INFO [train.py:904] (6/8) Epoch 17, batch 600, loss[loss=0.1589, simple_loss=0.2455, pruned_loss=0.03615, over 17224.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2567, pruned_loss=0.04443, over 3173187.00 frames. ], batch size: 45, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:11:47,292 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.692e+02 2.256e+02 2.602e+02 3.226e+02 5.329e+02, threshold=5.204e+02, percent-clipped=2.0 2023-04-30 11:12:16,979 INFO [train.py:904] (6/8) Epoch 17, batch 650, loss[loss=0.1653, simple_loss=0.2459, pruned_loss=0.04231, over 16882.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.256, pruned_loss=0.0442, over 3214212.26 frames. ], batch size: 109, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:12:49,883 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6346, 4.9331, 5.3257, 5.2481, 5.2994, 4.9223, 4.6357, 4.6756], device='cuda:6'), covar=tensor([0.0679, 0.0833, 0.0604, 0.0882, 0.0682, 0.0638, 0.1354, 0.0610], device='cuda:6'), in_proj_covar=tensor([0.0379, 0.0411, 0.0401, 0.0377, 0.0447, 0.0423, 0.0516, 0.0336], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 11:13:09,404 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=163090.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:13:19,397 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-30 11:13:25,534 INFO [train.py:904] (6/8) Epoch 17, batch 700, loss[loss=0.19, simple_loss=0.2647, pruned_loss=0.05764, over 16844.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.256, pruned_loss=0.04369, over 3237565.71 frames. ], batch size: 90, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:13:31,205 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9698, 1.9049, 2.5008, 2.8330, 2.6343, 3.2704, 1.9626, 3.1981], device='cuda:6'), covar=tensor([0.0209, 0.0493, 0.0293, 0.0258, 0.0299, 0.0169, 0.0483, 0.0155], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0185, 0.0172, 0.0176, 0.0186, 0.0142, 0.0186, 0.0134], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:14:04,487 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.287e+02 2.752e+02 3.270e+02 5.446e+02, threshold=5.503e+02, percent-clipped=2.0 2023-04-30 11:14:34,402 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=163151.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:14:35,142 INFO [train.py:904] (6/8) Epoch 17, batch 750, loss[loss=0.1637, simple_loss=0.2539, pruned_loss=0.03672, over 16694.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2563, pruned_loss=0.04356, over 3264934.64 frames. ], batch size: 57, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:14:57,206 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=163167.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 11:15:18,763 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=163183.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:15:29,036 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6846, 2.4419, 2.3943, 3.5613, 2.8687, 3.7764, 1.5053, 2.7476], device='cuda:6'), covar=tensor([0.1425, 0.0735, 0.1162, 0.0219, 0.0146, 0.0384, 0.1611, 0.0845], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0162, 0.0185, 0.0169, 0.0194, 0.0209, 0.0191, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 11:15:44,387 INFO [train.py:904] (6/8) Epoch 17, batch 800, loss[loss=0.1679, simple_loss=0.2643, pruned_loss=0.03571, over 17107.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2561, pruned_loss=0.0437, over 3276113.74 frames. ], batch size: 47, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:16:03,278 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=163215.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:16:23,806 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.209e+02 2.648e+02 3.284e+02 7.557e+02, threshold=5.297e+02, percent-clipped=2.0 2023-04-30 11:16:36,555 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=163240.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:16:44,846 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-30 11:16:53,792 INFO [train.py:904] (6/8) Epoch 17, batch 850, loss[loss=0.188, simple_loss=0.2754, pruned_loss=0.0503, over 16809.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2557, pruned_loss=0.04423, over 3275850.66 frames. ], batch size: 57, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:18:01,114 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=163301.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:18:01,854 INFO [train.py:904] (6/8) Epoch 17, batch 900, loss[loss=0.1765, simple_loss=0.2526, pruned_loss=0.05021, over 12236.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2542, pruned_loss=0.04313, over 3279618.96 frames. ], batch size: 247, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:18:40,393 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.201e+02 2.666e+02 3.227e+02 4.456e+02, threshold=5.332e+02, percent-clipped=0.0 2023-04-30 11:19:09,573 INFO [train.py:904] (6/8) Epoch 17, batch 950, loss[loss=0.1796, simple_loss=0.2505, pruned_loss=0.05438, over 16715.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2548, pruned_loss=0.04316, over 3292218.54 frames. ], batch size: 134, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:19:20,560 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0444, 4.1041, 4.4458, 2.1442, 4.5896, 4.6799, 3.4624, 3.5679], device='cuda:6'), covar=tensor([0.0652, 0.0202, 0.0196, 0.1151, 0.0060, 0.0142, 0.0368, 0.0370], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0106, 0.0093, 0.0139, 0.0074, 0.0119, 0.0126, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 11:20:17,960 INFO [train.py:904] (6/8) Epoch 17, batch 1000, loss[loss=0.1533, simple_loss=0.2391, pruned_loss=0.03373, over 16827.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2537, pruned_loss=0.04283, over 3295593.17 frames. ], batch size: 39, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:20:42,126 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8036, 3.0413, 2.7471, 5.0534, 4.1375, 4.5710, 1.6556, 3.3531], device='cuda:6'), covar=tensor([0.1297, 0.0694, 0.1166, 0.0185, 0.0307, 0.0402, 0.1575, 0.0719], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0163, 0.0185, 0.0169, 0.0194, 0.0209, 0.0191, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 11:20:54,826 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.618e+02 2.252e+02 2.741e+02 3.139e+02 5.661e+02, threshold=5.483e+02, percent-clipped=2.0 2023-04-30 11:20:57,361 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8316, 4.1640, 4.2838, 3.1149, 3.5864, 4.2285, 3.9543, 2.6465], device='cuda:6'), covar=tensor([0.0437, 0.0073, 0.0041, 0.0305, 0.0110, 0.0090, 0.0075, 0.0366], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0078, 0.0078, 0.0133, 0.0091, 0.0101, 0.0090, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 11:21:18,976 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=163446.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:21:26,450 INFO [train.py:904] (6/8) Epoch 17, batch 1050, loss[loss=0.1705, simple_loss=0.2592, pruned_loss=0.04087, over 17046.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2533, pruned_loss=0.04254, over 3299451.81 frames. ], batch size: 53, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:22:10,656 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=163483.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:22:20,596 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3476, 3.0959, 3.3212, 1.7005, 3.4559, 3.4282, 2.8019, 2.6755], device='cuda:6'), covar=tensor([0.0734, 0.0234, 0.0211, 0.1249, 0.0101, 0.0241, 0.0449, 0.0444], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0106, 0.0093, 0.0139, 0.0074, 0.0119, 0.0126, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 11:22:36,966 INFO [train.py:904] (6/8) Epoch 17, batch 1100, loss[loss=0.1427, simple_loss=0.2299, pruned_loss=0.02769, over 15872.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2526, pruned_loss=0.04253, over 3290091.70 frames. ], batch size: 35, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:22:43,652 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5434, 4.4297, 4.4751, 4.1063, 4.1732, 4.5010, 4.3016, 4.2548], device='cuda:6'), covar=tensor([0.0582, 0.0668, 0.0289, 0.0328, 0.0858, 0.0470, 0.0486, 0.0660], device='cuda:6'), in_proj_covar=tensor([0.0281, 0.0391, 0.0328, 0.0318, 0.0340, 0.0369, 0.0225, 0.0395], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:23:14,823 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.362e+02 2.061e+02 2.430e+02 2.825e+02 4.071e+02, threshold=4.861e+02, percent-clipped=0.0 2023-04-30 11:23:16,254 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=163531.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:23:34,569 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0878, 5.0277, 4.9399, 4.4122, 4.5287, 4.9960, 4.8672, 4.6311], device='cuda:6'), covar=tensor([0.0540, 0.0516, 0.0301, 0.0362, 0.1094, 0.0461, 0.0409, 0.0712], device='cuda:6'), in_proj_covar=tensor([0.0282, 0.0392, 0.0329, 0.0319, 0.0341, 0.0370, 0.0226, 0.0396], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:23:43,925 INFO [train.py:904] (6/8) Epoch 17, batch 1150, loss[loss=0.1703, simple_loss=0.2427, pruned_loss=0.04893, over 16709.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2521, pruned_loss=0.04256, over 3287715.09 frames. ], batch size: 134, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:24:10,820 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-30 11:24:43,130 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=163596.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:24:44,736 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-30 11:24:52,261 INFO [train.py:904] (6/8) Epoch 17, batch 1200, loss[loss=0.1554, simple_loss=0.2363, pruned_loss=0.03729, over 16468.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2506, pruned_loss=0.04166, over 3290496.82 frames. ], batch size: 75, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:25:29,955 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.305e+02 2.671e+02 3.522e+02 8.114e+02, threshold=5.342e+02, percent-clipped=7.0 2023-04-30 11:25:58,589 INFO [train.py:904] (6/8) Epoch 17, batch 1250, loss[loss=0.1656, simple_loss=0.2437, pruned_loss=0.04373, over 16693.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2512, pruned_loss=0.04286, over 3294809.28 frames. ], batch size: 134, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:27:04,040 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-30 11:27:05,466 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.80 vs. limit=5.0 2023-04-30 11:27:06,150 INFO [train.py:904] (6/8) Epoch 17, batch 1300, loss[loss=0.1636, simple_loss=0.2548, pruned_loss=0.03621, over 16824.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2514, pruned_loss=0.0429, over 3304459.61 frames. ], batch size: 62, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:27:44,996 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.213e+02 2.483e+02 3.083e+02 7.276e+02, threshold=4.967e+02, percent-clipped=1.0 2023-04-30 11:28:03,662 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1829, 5.3159, 5.6991, 5.6441, 5.6972, 5.3287, 5.2758, 5.1336], device='cuda:6'), covar=tensor([0.0389, 0.0571, 0.0420, 0.0496, 0.0457, 0.0388, 0.0932, 0.0383], device='cuda:6'), in_proj_covar=tensor([0.0389, 0.0420, 0.0411, 0.0386, 0.0458, 0.0433, 0.0528, 0.0343], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 11:28:08,419 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=163746.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:28:16,453 INFO [train.py:904] (6/8) Epoch 17, batch 1350, loss[loss=0.1511, simple_loss=0.2379, pruned_loss=0.03216, over 17259.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.251, pruned_loss=0.04306, over 3303991.23 frames. ], batch size: 43, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:29:15,218 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=163794.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:29:25,092 INFO [train.py:904] (6/8) Epoch 17, batch 1400, loss[loss=0.1761, simple_loss=0.2463, pruned_loss=0.05292, over 16296.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2505, pruned_loss=0.04218, over 3307369.96 frames. ], batch size: 165, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:30:05,128 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.111e+02 2.569e+02 3.362e+02 6.310e+02, threshold=5.138e+02, percent-clipped=5.0 2023-04-30 11:30:06,280 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7601, 1.7748, 1.5932, 1.5092, 1.9427, 1.6340, 1.6904, 1.9548], device='cuda:6'), covar=tensor([0.0204, 0.0288, 0.0387, 0.0357, 0.0189, 0.0262, 0.0205, 0.0197], device='cuda:6'), in_proj_covar=tensor([0.0191, 0.0230, 0.0220, 0.0220, 0.0228, 0.0229, 0.0233, 0.0223], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:30:36,590 INFO [train.py:904] (6/8) Epoch 17, batch 1450, loss[loss=0.158, simple_loss=0.2347, pruned_loss=0.04064, over 16812.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2498, pruned_loss=0.04204, over 3300596.08 frames. ], batch size: 102, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:30:49,402 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9809, 4.8972, 4.8877, 4.4567, 4.5008, 4.9051, 4.8054, 4.6109], device='cuda:6'), covar=tensor([0.0647, 0.0785, 0.0309, 0.0328, 0.1037, 0.0599, 0.0457, 0.0775], device='cuda:6'), in_proj_covar=tensor([0.0284, 0.0395, 0.0330, 0.0322, 0.0344, 0.0372, 0.0227, 0.0397], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:30:57,265 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7698, 4.0490, 3.0282, 2.3453, 2.7806, 2.6739, 4.3667, 3.6411], device='cuda:6'), covar=tensor([0.2819, 0.0644, 0.1716, 0.2683, 0.2567, 0.1824, 0.0458, 0.1262], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0264, 0.0297, 0.0299, 0.0288, 0.0243, 0.0283, 0.0323], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 11:31:38,036 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=163896.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:31:45,956 INFO [train.py:904] (6/8) Epoch 17, batch 1500, loss[loss=0.1523, simple_loss=0.244, pruned_loss=0.03033, over 17221.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2505, pruned_loss=0.04212, over 3309269.02 frames. ], batch size: 46, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:32:24,619 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.269e+02 2.188e+02 2.579e+02 3.200e+02 4.820e+02, threshold=5.157e+02, percent-clipped=0.0 2023-04-30 11:32:39,317 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2023-04-30 11:32:45,102 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=163944.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:32:56,204 INFO [train.py:904] (6/8) Epoch 17, batch 1550, loss[loss=0.1765, simple_loss=0.2674, pruned_loss=0.04283, over 17130.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2517, pruned_loss=0.04276, over 3320292.41 frames. ], batch size: 49, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:34:07,077 INFO [train.py:904] (6/8) Epoch 17, batch 1600, loss[loss=0.1455, simple_loss=0.2315, pruned_loss=0.02972, over 16977.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2538, pruned_loss=0.04332, over 3317382.82 frames. ], batch size: 41, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:34:13,472 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-30 11:34:45,103 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 2.274e+02 2.815e+02 3.289e+02 5.501e+02, threshold=5.629e+02, percent-clipped=1.0 2023-04-30 11:34:51,219 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3942, 5.3704, 5.1889, 4.3084, 5.2830, 1.9633, 4.9687, 5.0757], device='cuda:6'), covar=tensor([0.0092, 0.0087, 0.0193, 0.0482, 0.0106, 0.2820, 0.0151, 0.0221], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0142, 0.0190, 0.0172, 0.0164, 0.0202, 0.0178, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:35:15,602 INFO [train.py:904] (6/8) Epoch 17, batch 1650, loss[loss=0.1844, simple_loss=0.2672, pruned_loss=0.0508, over 16471.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2551, pruned_loss=0.0438, over 3321123.86 frames. ], batch size: 68, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:35:34,107 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9832, 2.8055, 2.8019, 1.9889, 2.6076, 2.0708, 2.7336, 2.9573], device='cuda:6'), covar=tensor([0.0302, 0.0760, 0.0541, 0.1844, 0.0884, 0.0982, 0.0549, 0.0769], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0157, 0.0163, 0.0150, 0.0141, 0.0127, 0.0141, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 11:36:08,246 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164090.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:36:24,909 INFO [train.py:904] (6/8) Epoch 17, batch 1700, loss[loss=0.2074, simple_loss=0.2871, pruned_loss=0.06388, over 15480.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2568, pruned_loss=0.04442, over 3317704.70 frames. ], batch size: 191, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:37:01,948 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.646e+02 2.209e+02 2.796e+02 3.472e+02 6.333e+02, threshold=5.591e+02, percent-clipped=2.0 2023-04-30 11:37:20,686 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6804, 4.6740, 5.0713, 5.0102, 5.0836, 4.7337, 4.7201, 4.5331], device='cuda:6'), covar=tensor([0.0336, 0.0734, 0.0412, 0.0515, 0.0559, 0.0463, 0.0992, 0.0572], device='cuda:6'), in_proj_covar=tensor([0.0392, 0.0424, 0.0413, 0.0389, 0.0460, 0.0437, 0.0534, 0.0345], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 11:37:32,254 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164151.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:37:32,928 INFO [train.py:904] (6/8) Epoch 17, batch 1750, loss[loss=0.1916, simple_loss=0.2812, pruned_loss=0.05099, over 17075.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2574, pruned_loss=0.04483, over 3315820.01 frames. ], batch size: 55, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:37:36,755 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164155.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:38:41,937 INFO [train.py:904] (6/8) Epoch 17, batch 1800, loss[loss=0.1768, simple_loss=0.2769, pruned_loss=0.03833, over 17043.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2586, pruned_loss=0.04469, over 3321413.22 frames. ], batch size: 50, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:38:46,348 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 11:38:58,196 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1792, 4.9275, 5.1675, 5.3667, 5.6036, 4.7634, 5.5411, 5.5292], device='cuda:6'), covar=tensor([0.1811, 0.1256, 0.1864, 0.0800, 0.0505, 0.0865, 0.0473, 0.0591], device='cuda:6'), in_proj_covar=tensor([0.0621, 0.0771, 0.0916, 0.0773, 0.0583, 0.0620, 0.0626, 0.0730], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:39:01,661 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164216.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:39:19,764 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.509e+02 2.179e+02 2.567e+02 3.142e+02 4.810e+02, threshold=5.135e+02, percent-clipped=0.0 2023-04-30 11:39:49,955 INFO [train.py:904] (6/8) Epoch 17, batch 1850, loss[loss=0.1957, simple_loss=0.2936, pruned_loss=0.04891, over 16668.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2598, pruned_loss=0.04473, over 3318710.30 frames. ], batch size: 57, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:41:01,367 INFO [train.py:904] (6/8) Epoch 17, batch 1900, loss[loss=0.1484, simple_loss=0.2314, pruned_loss=0.03276, over 15865.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2593, pruned_loss=0.04437, over 3312874.04 frames. ], batch size: 35, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:41:20,279 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-30 11:41:41,203 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.257e+02 2.647e+02 3.150e+02 4.785e+02, threshold=5.294e+02, percent-clipped=0.0 2023-04-30 11:42:12,315 INFO [train.py:904] (6/8) Epoch 17, batch 1950, loss[loss=0.176, simple_loss=0.2701, pruned_loss=0.04097, over 17080.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2594, pruned_loss=0.04427, over 3320282.53 frames. ], batch size: 53, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:42:24,747 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164360.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:43:23,967 INFO [train.py:904] (6/8) Epoch 17, batch 2000, loss[loss=0.1365, simple_loss=0.2189, pruned_loss=0.02706, over 16818.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2589, pruned_loss=0.04422, over 3313050.48 frames. ], batch size: 39, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:43:51,053 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164421.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:44:02,063 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.498e+02 2.950e+02 3.427e+02 6.648e+02, threshold=5.900e+02, percent-clipped=3.0 2023-04-30 11:44:25,524 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=164446.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:44:32,478 INFO [train.py:904] (6/8) Epoch 17, batch 2050, loss[loss=0.1557, simple_loss=0.2427, pruned_loss=0.03431, over 17180.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2586, pruned_loss=0.04417, over 3316894.99 frames. ], batch size: 46, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:45:41,561 INFO [train.py:904] (6/8) Epoch 17, batch 2100, loss[loss=0.1575, simple_loss=0.2453, pruned_loss=0.03486, over 15992.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2596, pruned_loss=0.04454, over 3314158.09 frames. ], batch size: 35, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:45:48,111 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.21 vs. limit=5.0 2023-04-30 11:45:55,655 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=164511.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:46:20,646 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.320e+02 2.783e+02 3.356e+02 6.919e+02, threshold=5.567e+02, percent-clipped=1.0 2023-04-30 11:46:40,423 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6884, 3.9439, 2.6585, 4.5109, 3.0499, 4.5201, 2.5296, 3.1898], device='cuda:6'), covar=tensor([0.0321, 0.0365, 0.1340, 0.0409, 0.0740, 0.0502, 0.1458, 0.0692], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0176, 0.0195, 0.0158, 0.0175, 0.0219, 0.0204, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 11:46:43,404 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9744, 2.0765, 2.6044, 2.8748, 2.7509, 3.4558, 2.3852, 3.3336], device='cuda:6'), covar=tensor([0.0211, 0.0416, 0.0296, 0.0306, 0.0297, 0.0132, 0.0383, 0.0150], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0187, 0.0174, 0.0178, 0.0188, 0.0145, 0.0187, 0.0138], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:46:50,964 INFO [train.py:904] (6/8) Epoch 17, batch 2150, loss[loss=0.1778, simple_loss=0.2664, pruned_loss=0.04463, over 16771.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2604, pruned_loss=0.04493, over 3319237.51 frames. ], batch size: 57, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:47:07,436 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5932, 2.5522, 2.0288, 2.3141, 2.8891, 2.6159, 3.3484, 3.1891], device='cuda:6'), covar=tensor([0.0132, 0.0440, 0.0597, 0.0511, 0.0317, 0.0445, 0.0221, 0.0288], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0230, 0.0221, 0.0220, 0.0231, 0.0231, 0.0236, 0.0225], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:47:13,635 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164568.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:47:27,054 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164578.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:47:58,403 INFO [train.py:904] (6/8) Epoch 17, batch 2200, loss[loss=0.167, simple_loss=0.2498, pruned_loss=0.04213, over 16821.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2612, pruned_loss=0.0453, over 3323996.78 frames. ], batch size: 96, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:48:36,580 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164629.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:48:37,254 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.246e+02 2.692e+02 3.358e+02 7.856e+02, threshold=5.383e+02, percent-clipped=4.0 2023-04-30 11:48:50,099 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164639.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 11:49:06,805 INFO [train.py:904] (6/8) Epoch 17, batch 2250, loss[loss=0.1711, simple_loss=0.2429, pruned_loss=0.04969, over 16725.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2616, pruned_loss=0.04546, over 3322760.98 frames. ], batch size: 83, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:49:15,654 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164658.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:50:15,835 INFO [train.py:904] (6/8) Epoch 17, batch 2300, loss[loss=0.1739, simple_loss=0.2597, pruned_loss=0.04402, over 15540.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2613, pruned_loss=0.04493, over 3317842.97 frames. ], batch size: 190, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:50:35,075 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=164716.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:50:38,843 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164719.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:50:53,239 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.267e+02 2.732e+02 3.248e+02 6.270e+02, threshold=5.465e+02, percent-clipped=1.0 2023-04-30 11:51:00,127 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2557, 5.2298, 5.0190, 4.4272, 5.0307, 1.8449, 4.8458, 4.9304], device='cuda:6'), covar=tensor([0.0082, 0.0084, 0.0186, 0.0430, 0.0109, 0.2826, 0.0141, 0.0215], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0144, 0.0191, 0.0174, 0.0165, 0.0202, 0.0180, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:51:16,779 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=164746.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:51:16,798 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164746.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:51:24,576 INFO [train.py:904] (6/8) Epoch 17, batch 2350, loss[loss=0.1922, simple_loss=0.2626, pruned_loss=0.06086, over 16772.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2629, pruned_loss=0.04606, over 3302311.25 frames. ], batch size: 124, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:52:23,526 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=164794.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:52:34,376 INFO [train.py:904] (6/8) Epoch 17, batch 2400, loss[loss=0.1443, simple_loss=0.229, pruned_loss=0.02983, over 16816.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.263, pruned_loss=0.04601, over 3310733.72 frames. ], batch size: 39, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:52:41,616 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164807.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:52:46,692 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=164811.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:52:48,664 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8147, 3.9781, 2.5263, 4.5661, 3.1382, 4.5909, 2.6236, 3.3408], device='cuda:6'), covar=tensor([0.0292, 0.0333, 0.1484, 0.0256, 0.0753, 0.0423, 0.1364, 0.0677], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0175, 0.0194, 0.0157, 0.0173, 0.0217, 0.0202, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 11:53:12,672 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 2.216e+02 2.538e+02 3.112e+02 6.074e+02, threshold=5.077e+02, percent-clipped=1.0 2023-04-30 11:53:41,596 INFO [train.py:904] (6/8) Epoch 17, batch 2450, loss[loss=0.1669, simple_loss=0.2617, pruned_loss=0.03603, over 17132.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2642, pruned_loss=0.04577, over 3309615.99 frames. ], batch size: 47, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 11:53:51,211 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=164859.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:53:57,922 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2452, 3.6795, 3.9097, 2.0797, 3.1509, 2.4699, 3.8196, 3.7851], device='cuda:6'), covar=tensor([0.0311, 0.0792, 0.0497, 0.1957, 0.0793, 0.0934, 0.0591, 0.0994], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0158, 0.0163, 0.0150, 0.0140, 0.0127, 0.0140, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 11:54:01,794 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.05 vs. limit=5.0 2023-04-30 11:54:04,041 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.14 vs. limit=5.0 2023-04-30 11:54:46,720 INFO [train.py:904] (6/8) Epoch 17, batch 2500, loss[loss=0.1492, simple_loss=0.2308, pruned_loss=0.03378, over 16759.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.2638, pruned_loss=0.0458, over 3312314.11 frames. ], batch size: 39, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 11:55:17,540 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=164924.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:55:26,643 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.613e+02 2.192e+02 2.472e+02 2.956e+02 6.595e+02, threshold=4.945e+02, percent-clipped=3.0 2023-04-30 11:55:30,928 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=164934.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 11:55:55,231 INFO [train.py:904] (6/8) Epoch 17, batch 2550, loss[loss=0.1983, simple_loss=0.278, pruned_loss=0.05927, over 16529.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2633, pruned_loss=0.04528, over 3327497.20 frames. ], batch size: 75, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 11:56:01,645 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5396, 2.6663, 2.2207, 2.4811, 2.9357, 2.6949, 3.2055, 3.1596], device='cuda:6'), covar=tensor([0.0141, 0.0361, 0.0480, 0.0416, 0.0256, 0.0356, 0.0247, 0.0232], device='cuda:6'), in_proj_covar=tensor([0.0195, 0.0230, 0.0221, 0.0221, 0.0232, 0.0231, 0.0237, 0.0226], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:56:57,120 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6903, 4.4742, 4.7554, 4.9083, 5.0393, 4.4873, 5.0284, 5.0503], device='cuda:6'), covar=tensor([0.1649, 0.1390, 0.1650, 0.0770, 0.0573, 0.1054, 0.0785, 0.0654], device='cuda:6'), in_proj_covar=tensor([0.0627, 0.0778, 0.0929, 0.0787, 0.0589, 0.0626, 0.0630, 0.0734], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 11:57:02,085 INFO [train.py:904] (6/8) Epoch 17, batch 2600, loss[loss=0.1707, simple_loss=0.2686, pruned_loss=0.03637, over 17270.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.2631, pruned_loss=0.04476, over 3324719.67 frames. ], batch size: 52, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 11:57:17,927 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=165014.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:57:20,588 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=165016.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:57:41,409 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 2.187e+02 2.619e+02 3.229e+02 7.768e+02, threshold=5.237e+02, percent-clipped=3.0 2023-04-30 11:57:48,825 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2060, 3.2539, 3.1911, 5.1339, 4.0859, 4.5779, 2.1562, 3.4835], device='cuda:6'), covar=tensor([0.1187, 0.0724, 0.1042, 0.0153, 0.0306, 0.0359, 0.1425, 0.0727], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0166, 0.0187, 0.0174, 0.0200, 0.0212, 0.0192, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 11:58:08,463 INFO [train.py:904] (6/8) Epoch 17, batch 2650, loss[loss=0.2118, simple_loss=0.2981, pruned_loss=0.06272, over 16137.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2639, pruned_loss=0.04459, over 3329401.13 frames. ], batch size: 35, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 11:58:26,279 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=165064.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:59:18,027 INFO [train.py:904] (6/8) Epoch 17, batch 2700, loss[loss=0.1701, simple_loss=0.2662, pruned_loss=0.03697, over 17035.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2643, pruned_loss=0.04425, over 3331764.79 frames. ], batch size: 50, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 11:59:18,300 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=165102.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:59:47,978 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=165123.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:59:59,035 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.045e+02 2.516e+02 2.952e+02 6.082e+02, threshold=5.032e+02, percent-clipped=3.0 2023-04-30 11:59:59,488 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0019, 4.2411, 4.2071, 3.1487, 3.5831, 4.1958, 3.9299, 2.5672], device='cuda:6'), covar=tensor([0.0351, 0.0074, 0.0043, 0.0299, 0.0112, 0.0083, 0.0076, 0.0372], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0076, 0.0076, 0.0129, 0.0090, 0.0099, 0.0088, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 12:00:25,218 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-30 12:00:28,481 INFO [train.py:904] (6/8) Epoch 17, batch 2750, loss[loss=0.1869, simple_loss=0.285, pruned_loss=0.04443, over 17033.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2644, pruned_loss=0.04389, over 3333179.97 frames. ], batch size: 50, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:01:13,236 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=165184.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:01:38,386 INFO [train.py:904] (6/8) Epoch 17, batch 2800, loss[loss=0.1935, simple_loss=0.2777, pruned_loss=0.0546, over 16747.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2647, pruned_loss=0.0443, over 3331443.43 frames. ], batch size: 124, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:02:10,269 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=165224.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:02:15,517 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0295, 3.0943, 3.2518, 2.1524, 2.8830, 2.2864, 3.5215, 3.4819], device='cuda:6'), covar=tensor([0.0220, 0.0956, 0.0576, 0.1748, 0.0806, 0.0962, 0.0510, 0.0811], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0158, 0.0164, 0.0150, 0.0140, 0.0127, 0.0140, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 12:02:19,320 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.573e+02 2.251e+02 2.601e+02 3.544e+02 6.990e+02, threshold=5.203e+02, percent-clipped=6.0 2023-04-30 12:02:24,135 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=165234.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 12:02:26,849 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 12:02:33,686 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2382, 3.2802, 3.1791, 5.2078, 4.4619, 4.6739, 2.1066, 3.6026], device='cuda:6'), covar=tensor([0.1136, 0.0639, 0.0959, 0.0165, 0.0213, 0.0338, 0.1341, 0.0637], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0165, 0.0186, 0.0175, 0.0199, 0.0212, 0.0192, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 12:02:48,647 INFO [train.py:904] (6/8) Epoch 17, batch 2850, loss[loss=0.1726, simple_loss=0.2708, pruned_loss=0.03724, over 16705.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.2631, pruned_loss=0.04329, over 3332266.68 frames. ], batch size: 57, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:03:18,271 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=165272.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:03:31,438 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=165282.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:03:58,713 INFO [train.py:904] (6/8) Epoch 17, batch 2900, loss[loss=0.1699, simple_loss=0.2656, pruned_loss=0.03708, over 17058.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2621, pruned_loss=0.0448, over 3322702.74 frames. ], batch size: 50, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:04:16,996 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=165314.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:04:39,535 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.654e+02 2.311e+02 2.666e+02 3.214e+02 5.469e+02, threshold=5.331e+02, percent-clipped=1.0 2023-04-30 12:05:09,243 INFO [train.py:904] (6/8) Epoch 17, batch 2950, loss[loss=0.1777, simple_loss=0.2764, pruned_loss=0.03951, over 17049.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2619, pruned_loss=0.04519, over 3323817.55 frames. ], batch size: 50, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:05:16,346 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7277, 4.2361, 4.1975, 2.9058, 3.5642, 4.1692, 3.8583, 2.3710], device='cuda:6'), covar=tensor([0.0440, 0.0083, 0.0045, 0.0353, 0.0114, 0.0097, 0.0081, 0.0453], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0077, 0.0077, 0.0131, 0.0091, 0.0101, 0.0089, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 12:05:23,904 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=165362.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:06:20,613 INFO [train.py:904] (6/8) Epoch 17, batch 3000, loss[loss=0.1508, simple_loss=0.2401, pruned_loss=0.03069, over 17234.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2619, pruned_loss=0.04546, over 3332716.12 frames. ], batch size: 44, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:06:20,614 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 12:06:29,129 INFO [train.py:938] (6/8) Epoch 17, validation: loss=0.1364, simple_loss=0.2422, pruned_loss=0.0153, over 944034.00 frames. 2023-04-30 12:06:29,130 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 12:06:29,443 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=165402.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:06:58,264 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-30 12:07:06,050 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9195, 4.6954, 4.9422, 5.1291, 5.3345, 4.7014, 5.3310, 5.3456], device='cuda:6'), covar=tensor([0.1803, 0.1428, 0.1770, 0.0796, 0.0597, 0.1005, 0.0583, 0.0654], device='cuda:6'), in_proj_covar=tensor([0.0635, 0.0784, 0.0934, 0.0793, 0.0596, 0.0631, 0.0636, 0.0741], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 12:07:09,774 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.715e+02 2.395e+02 2.828e+02 3.269e+02 8.319e+02, threshold=5.656e+02, percent-clipped=3.0 2023-04-30 12:07:36,649 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=165450.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:07:38,650 INFO [train.py:904] (6/8) Epoch 17, batch 3050, loss[loss=0.167, simple_loss=0.2588, pruned_loss=0.03758, over 17131.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.2616, pruned_loss=0.04548, over 3328377.92 frames. ], batch size: 49, lr: 4.02e-03, grad_scale: 4.0 2023-04-30 12:08:15,874 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=165479.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:08:45,828 INFO [train.py:904] (6/8) Epoch 17, batch 3100, loss[loss=0.1848, simple_loss=0.2709, pruned_loss=0.0493, over 16594.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2615, pruned_loss=0.04583, over 3322792.31 frames. ], batch size: 68, lr: 4.02e-03, grad_scale: 4.0 2023-04-30 12:09:27,659 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.216e+02 2.589e+02 3.162e+02 6.656e+02, threshold=5.179e+02, percent-clipped=4.0 2023-04-30 12:09:55,001 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=165551.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:09:55,707 INFO [train.py:904] (6/8) Epoch 17, batch 3150, loss[loss=0.1628, simple_loss=0.2648, pruned_loss=0.0304, over 17034.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2608, pruned_loss=0.04547, over 3317122.05 frames. ], batch size: 50, lr: 4.02e-03, grad_scale: 4.0 2023-04-30 12:11:06,273 INFO [train.py:904] (6/8) Epoch 17, batch 3200, loss[loss=0.1398, simple_loss=0.2246, pruned_loss=0.02752, over 16987.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2603, pruned_loss=0.0448, over 3326036.72 frames. ], batch size: 41, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:11:21,079 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=165612.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:11:49,641 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 2.164e+02 2.494e+02 3.096e+02 4.677e+02, threshold=4.988e+02, percent-clipped=0.0 2023-04-30 12:12:15,439 INFO [train.py:904] (6/8) Epoch 17, batch 3250, loss[loss=0.1887, simple_loss=0.2712, pruned_loss=0.0531, over 16437.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2606, pruned_loss=0.04526, over 3310954.16 frames. ], batch size: 75, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:13:23,329 INFO [train.py:904] (6/8) Epoch 17, batch 3300, loss[loss=0.2104, simple_loss=0.2914, pruned_loss=0.06472, over 15565.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2617, pruned_loss=0.04601, over 3306187.43 frames. ], batch size: 191, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:14:06,777 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.766e+02 2.306e+02 2.673e+02 3.210e+02 6.560e+02, threshold=5.346e+02, percent-clipped=3.0 2023-04-30 12:14:30,228 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=165750.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:14:32,750 INFO [train.py:904] (6/8) Epoch 17, batch 3350, loss[loss=0.1492, simple_loss=0.2347, pruned_loss=0.03189, over 16752.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2623, pruned_loss=0.04594, over 3301634.54 frames. ], batch size: 39, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:14:39,816 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.03 vs. limit=5.0 2023-04-30 12:15:11,017 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=165779.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:15:42,358 INFO [train.py:904] (6/8) Epoch 17, batch 3400, loss[loss=0.1524, simple_loss=0.2337, pruned_loss=0.03556, over 16833.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2625, pruned_loss=0.04567, over 3293517.86 frames. ], batch size: 42, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:15:44,560 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=165803.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:15:56,246 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=165811.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:16:19,114 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=165827.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:16:27,736 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.228e+02 2.640e+02 3.185e+02 6.119e+02, threshold=5.280e+02, percent-clipped=1.0 2023-04-30 12:16:28,414 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-30 12:16:34,233 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6962, 3.8042, 2.2893, 4.1140, 2.9371, 4.0426, 2.4731, 3.0302], device='cuda:6'), covar=tensor([0.0276, 0.0383, 0.1495, 0.0281, 0.0735, 0.0669, 0.1347, 0.0666], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0175, 0.0194, 0.0159, 0.0174, 0.0218, 0.0202, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 12:16:54,366 INFO [train.py:904] (6/8) Epoch 17, batch 3450, loss[loss=0.1811, simple_loss=0.273, pruned_loss=0.04458, over 16690.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2598, pruned_loss=0.04453, over 3301103.18 frames. ], batch size: 57, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:17:11,567 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=165864.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:17:14,563 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4434, 4.2573, 4.4333, 4.6024, 4.7362, 4.2588, 4.5302, 4.7119], device='cuda:6'), covar=tensor([0.1614, 0.1140, 0.1439, 0.0729, 0.0623, 0.1197, 0.2068, 0.0795], device='cuda:6'), in_proj_covar=tensor([0.0640, 0.0791, 0.0946, 0.0803, 0.0602, 0.0639, 0.0641, 0.0748], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 12:17:33,934 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6997, 4.7633, 4.9347, 4.8038, 4.7718, 5.3924, 4.9371, 4.6435], device='cuda:6'), covar=tensor([0.1461, 0.2040, 0.2025, 0.1999, 0.2741, 0.1053, 0.1578, 0.2569], device='cuda:6'), in_proj_covar=tensor([0.0397, 0.0573, 0.0629, 0.0485, 0.0651, 0.0662, 0.0496, 0.0649], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 12:18:05,607 INFO [train.py:904] (6/8) Epoch 17, batch 3500, loss[loss=0.1724, simple_loss=0.251, pruned_loss=0.04685, over 16790.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2576, pruned_loss=0.04417, over 3306860.51 frames. ], batch size: 124, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:18:13,075 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=165907.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:18:49,909 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.135e+02 2.523e+02 3.067e+02 7.185e+02, threshold=5.046e+02, percent-clipped=1.0 2023-04-30 12:19:16,033 INFO [train.py:904] (6/8) Epoch 17, batch 3550, loss[loss=0.1498, simple_loss=0.2361, pruned_loss=0.03175, over 17223.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2566, pruned_loss=0.04394, over 3296289.04 frames. ], batch size: 44, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:20:04,639 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3606, 5.3066, 5.0609, 4.4930, 5.1619, 2.0219, 4.8971, 5.0811], device='cuda:6'), covar=tensor([0.0079, 0.0078, 0.0194, 0.0423, 0.0099, 0.2560, 0.0142, 0.0178], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0143, 0.0191, 0.0175, 0.0165, 0.0199, 0.0180, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 12:20:27,926 INFO [train.py:904] (6/8) Epoch 17, batch 3600, loss[loss=0.1569, simple_loss=0.2334, pruned_loss=0.04026, over 16407.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2563, pruned_loss=0.04357, over 3293909.18 frames. ], batch size: 68, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:21:12,106 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.506e+02 2.087e+02 2.490e+02 2.982e+02 5.501e+02, threshold=4.979e+02, percent-clipped=1.0 2023-04-30 12:21:30,622 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-30 12:21:40,963 INFO [train.py:904] (6/8) Epoch 17, batch 3650, loss[loss=0.185, simple_loss=0.2549, pruned_loss=0.05757, over 11443.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2558, pruned_loss=0.04415, over 3296067.93 frames. ], batch size: 246, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:21:44,528 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166054.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 12:22:55,841 INFO [train.py:904] (6/8) Epoch 17, batch 3700, loss[loss=0.1817, simple_loss=0.2645, pruned_loss=0.0494, over 16327.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2543, pruned_loss=0.04558, over 3278457.38 frames. ], batch size: 35, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:23:02,843 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166106.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:23:16,034 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166115.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 12:23:31,403 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3961, 4.4314, 4.7438, 4.7137, 4.7476, 4.4578, 4.4747, 4.2219], device='cuda:6'), covar=tensor([0.0326, 0.0577, 0.0359, 0.0434, 0.0521, 0.0382, 0.0745, 0.0646], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0427, 0.0416, 0.0394, 0.0466, 0.0439, 0.0536, 0.0349], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 12:23:42,388 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.852e+02 2.227e+02 2.592e+02 3.005e+02 5.986e+02, threshold=5.184e+02, percent-clipped=3.0 2023-04-30 12:24:04,278 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166147.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:24:10,327 INFO [train.py:904] (6/8) Epoch 17, batch 3750, loss[loss=0.208, simple_loss=0.2952, pruned_loss=0.06043, over 16679.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2554, pruned_loss=0.04704, over 3262212.40 frames. ], batch size: 57, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:24:21,421 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166159.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:25:24,426 INFO [train.py:904] (6/8) Epoch 17, batch 3800, loss[loss=0.169, simple_loss=0.2389, pruned_loss=0.0495, over 16932.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2566, pruned_loss=0.04809, over 3267758.28 frames. ], batch size: 109, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:25:32,420 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=166207.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:25:33,706 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166208.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:26:10,618 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.840e+02 2.250e+02 2.656e+02 3.137e+02 5.830e+02, threshold=5.312e+02, percent-clipped=1.0 2023-04-30 12:26:38,503 INFO [train.py:904] (6/8) Epoch 17, batch 3850, loss[loss=0.185, simple_loss=0.2665, pruned_loss=0.05177, over 16642.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2566, pruned_loss=0.04898, over 3271259.90 frames. ], batch size: 57, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:26:44,002 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=166255.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:27:30,879 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5570, 2.6526, 1.8927, 2.1630, 3.0048, 2.5253, 3.3045, 3.2285], device='cuda:6'), covar=tensor([0.0072, 0.0363, 0.0606, 0.0525, 0.0266, 0.0418, 0.0195, 0.0230], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0228, 0.0219, 0.0220, 0.0230, 0.0229, 0.0234, 0.0225], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 12:27:50,726 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-30 12:27:52,935 INFO [train.py:904] (6/8) Epoch 17, batch 3900, loss[loss=0.1657, simple_loss=0.2511, pruned_loss=0.04013, over 16213.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2564, pruned_loss=0.0494, over 3273305.53 frames. ], batch size: 35, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:28:01,087 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166307.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:28:37,861 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.433e+02 2.279e+02 2.722e+02 3.230e+02 5.998e+02, threshold=5.443e+02, percent-clipped=1.0 2023-04-30 12:29:07,065 INFO [train.py:904] (6/8) Epoch 17, batch 3950, loss[loss=0.1928, simple_loss=0.2587, pruned_loss=0.06339, over 16826.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2565, pruned_loss=0.04997, over 3280062.59 frames. ], batch size: 116, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:29:23,678 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-30 12:29:29,783 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.94 vs. limit=5.0 2023-04-30 12:29:30,699 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166368.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:30:18,532 INFO [train.py:904] (6/8) Epoch 17, batch 4000, loss[loss=0.1757, simple_loss=0.2585, pruned_loss=0.04645, over 16464.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2568, pruned_loss=0.05054, over 3274573.12 frames. ], batch size: 68, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:30:25,151 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=166406.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:30:31,161 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166410.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 12:31:03,720 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.351e+02 2.185e+02 2.579e+02 2.981e+02 5.548e+02, threshold=5.157e+02, percent-clipped=1.0 2023-04-30 12:31:31,403 INFO [train.py:904] (6/8) Epoch 17, batch 4050, loss[loss=0.2126, simple_loss=0.2825, pruned_loss=0.07132, over 12361.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2572, pruned_loss=0.04936, over 3281748.46 frames. ], batch size: 248, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:31:34,118 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=166454.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:31:42,538 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=166459.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:32:14,653 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166481.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:32:44,111 INFO [train.py:904] (6/8) Epoch 17, batch 4100, loss[loss=0.1754, simple_loss=0.2692, pruned_loss=0.04083, over 16703.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2586, pruned_loss=0.04878, over 3272611.45 frames. ], batch size: 89, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:32:46,307 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166503.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:32:51,061 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=166507.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:32:54,697 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166509.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:33:26,438 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-30 12:33:29,330 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166531.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:33:31,184 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4722, 3.5942, 2.5500, 2.1769, 2.4300, 2.3018, 3.8091, 3.2973], device='cuda:6'), covar=tensor([0.2947, 0.0689, 0.1940, 0.2430, 0.2510, 0.1994, 0.0516, 0.1153], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0264, 0.0298, 0.0301, 0.0293, 0.0244, 0.0285, 0.0325], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 12:33:31,669 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.012e+02 2.257e+02 2.846e+02 6.984e+02, threshold=4.513e+02, percent-clipped=1.0 2023-04-30 12:33:45,794 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166542.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:33:59,691 INFO [train.py:904] (6/8) Epoch 17, batch 4150, loss[loss=0.1905, simple_loss=0.2792, pruned_loss=0.05088, over 17260.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2648, pruned_loss=0.05058, over 3257638.20 frames. ], batch size: 52, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:34:28,415 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166570.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:34:41,153 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-30 12:35:01,043 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166592.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:35:14,002 INFO [train.py:904] (6/8) Epoch 17, batch 4200, loss[loss=0.2207, simple_loss=0.3124, pruned_loss=0.06445, over 16238.00 frames. ], tot_loss[loss=0.1887, simple_loss=0.2724, pruned_loss=0.05249, over 3245789.12 frames. ], batch size: 165, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:36:00,016 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.639e+02 2.403e+02 2.795e+02 3.348e+02 5.640e+02, threshold=5.591e+02, percent-clipped=5.0 2023-04-30 12:36:27,721 INFO [train.py:904] (6/8) Epoch 17, batch 4250, loss[loss=0.181, simple_loss=0.2767, pruned_loss=0.04267, over 16794.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.2758, pruned_loss=0.05259, over 3218970.92 frames. ], batch size: 83, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:36:43,547 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166663.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:37:39,001 INFO [train.py:904] (6/8) Epoch 17, batch 4300, loss[loss=0.2173, simple_loss=0.3089, pruned_loss=0.06289, over 16692.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.2774, pruned_loss=0.052, over 3206131.87 frames. ], batch size: 134, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:37:51,461 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=166710.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 12:38:24,693 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.116e+02 2.589e+02 3.113e+02 5.603e+02, threshold=5.178e+02, percent-clipped=2.0 2023-04-30 12:38:28,965 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166735.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:38:52,934 INFO [train.py:904] (6/8) Epoch 17, batch 4350, loss[loss=0.2003, simple_loss=0.2944, pruned_loss=0.05312, over 16244.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2808, pruned_loss=0.05285, over 3227544.60 frames. ], batch size: 165, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:39:01,861 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=166758.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 12:39:10,447 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166764.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:39:25,570 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3605, 2.3291, 2.4280, 4.2511, 2.2230, 2.7282, 2.4162, 2.5071], device='cuda:6'), covar=tensor([0.1112, 0.3136, 0.2415, 0.0418, 0.3720, 0.2100, 0.2880, 0.3127], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0432, 0.0355, 0.0327, 0.0430, 0.0500, 0.0399, 0.0505], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 12:39:57,560 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166796.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:40:05,343 INFO [train.py:904] (6/8) Epoch 17, batch 4400, loss[loss=0.2115, simple_loss=0.3001, pruned_loss=0.06145, over 16827.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.2831, pruned_loss=0.05422, over 3205269.59 frames. ], batch size: 116, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:40:06,840 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=166803.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:40:36,684 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8021, 3.7740, 4.2388, 1.9558, 4.5710, 4.5907, 3.1797, 3.3339], device='cuda:6'), covar=tensor([0.0727, 0.0244, 0.0186, 0.1126, 0.0054, 0.0081, 0.0404, 0.0378], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0106, 0.0094, 0.0137, 0.0075, 0.0120, 0.0126, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 12:40:37,932 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166825.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:40:49,145 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.232e+02 2.626e+02 3.012e+02 5.313e+02, threshold=5.252e+02, percent-clipped=1.0 2023-04-30 12:40:55,203 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166837.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:41:02,584 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4229, 5.4319, 5.3109, 4.9638, 4.9699, 5.3528, 5.1656, 5.0583], device='cuda:6'), covar=tensor([0.0443, 0.0215, 0.0197, 0.0202, 0.0805, 0.0254, 0.0246, 0.0544], device='cuda:6'), in_proj_covar=tensor([0.0278, 0.0385, 0.0329, 0.0316, 0.0338, 0.0365, 0.0223, 0.0392], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 12:41:15,725 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=166851.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:41:16,588 INFO [train.py:904] (6/8) Epoch 17, batch 4450, loss[loss=0.2213, simple_loss=0.3012, pruned_loss=0.07063, over 16586.00 frames. ], tot_loss[loss=0.1987, simple_loss=0.2867, pruned_loss=0.05535, over 3214637.29 frames. ], batch size: 57, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:41:19,691 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 12:41:36,437 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166865.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:42:07,492 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166887.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:42:21,371 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-30 12:42:28,857 INFO [train.py:904] (6/8) Epoch 17, batch 4500, loss[loss=0.1913, simple_loss=0.2851, pruned_loss=0.04877, over 16680.00 frames. ], tot_loss[loss=0.1987, simple_loss=0.2867, pruned_loss=0.05534, over 3237683.84 frames. ], batch size: 134, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:43:07,248 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166929.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:43:12,782 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.487e+02 1.901e+02 2.204e+02 2.572e+02 5.344e+02, threshold=4.409e+02, percent-clipped=1.0 2023-04-30 12:43:40,951 INFO [train.py:904] (6/8) Epoch 17, batch 4550, loss[loss=0.1972, simple_loss=0.2915, pruned_loss=0.05142, over 16724.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2876, pruned_loss=0.05612, over 3252639.55 frames. ], batch size: 89, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:43:57,315 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=166963.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:44:35,649 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166990.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:44:52,530 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8198, 3.8198, 4.3386, 1.9346, 4.6837, 4.6859, 3.1448, 3.4603], device='cuda:6'), covar=tensor([0.0806, 0.0225, 0.0155, 0.1259, 0.0044, 0.0086, 0.0429, 0.0393], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0105, 0.0093, 0.0136, 0.0074, 0.0119, 0.0125, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 12:44:53,170 INFO [train.py:904] (6/8) Epoch 17, batch 4600, loss[loss=0.2065, simple_loss=0.2942, pruned_loss=0.05946, over 17159.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2881, pruned_loss=0.05652, over 3251743.86 frames. ], batch size: 46, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:45:01,648 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0709, 5.3725, 5.1600, 5.1806, 4.9231, 4.6589, 4.7552, 5.4833], device='cuda:6'), covar=tensor([0.1092, 0.0714, 0.0919, 0.0787, 0.0715, 0.0992, 0.1141, 0.0791], device='cuda:6'), in_proj_covar=tensor([0.0630, 0.0776, 0.0632, 0.0576, 0.0492, 0.0497, 0.0646, 0.0599], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 12:45:07,106 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167011.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:45:38,255 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 1.923e+02 2.253e+02 2.642e+02 5.543e+02, threshold=4.506e+02, percent-clipped=2.0 2023-04-30 12:46:05,379 INFO [train.py:904] (6/8) Epoch 17, batch 4650, loss[loss=0.1814, simple_loss=0.2606, pruned_loss=0.05111, over 17275.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2871, pruned_loss=0.05657, over 3247443.65 frames. ], batch size: 52, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:46:33,873 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6705, 2.3691, 2.3235, 3.1765, 2.3585, 3.5490, 1.4320, 2.8047], device='cuda:6'), covar=tensor([0.1369, 0.0786, 0.1182, 0.0146, 0.0155, 0.0339, 0.1675, 0.0745], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0168, 0.0190, 0.0176, 0.0205, 0.0213, 0.0194, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 12:47:00,699 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=167091.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:47:13,841 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=167100.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:47:16,412 INFO [train.py:904] (6/8) Epoch 17, batch 4700, loss[loss=0.1682, simple_loss=0.2643, pruned_loss=0.03606, over 16760.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2838, pruned_loss=0.05552, over 3224167.37 frames. ], batch size: 89, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:47:17,175 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 12:47:18,987 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=167104.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:47:42,085 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=167120.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:48:00,930 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.373e+02 1.982e+02 2.244e+02 2.685e+02 3.929e+02, threshold=4.489e+02, percent-clipped=0.0 2023-04-30 12:48:06,765 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167137.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:48:28,394 INFO [train.py:904] (6/8) Epoch 17, batch 4750, loss[loss=0.1731, simple_loss=0.257, pruned_loss=0.04464, over 16601.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2794, pruned_loss=0.05346, over 3225671.37 frames. ], batch size: 68, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:48:42,096 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=167161.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:48:48,026 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167165.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:48:48,137 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=167165.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:49:16,408 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167185.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:49:18,962 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167187.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:49:40,730 INFO [train.py:904] (6/8) Epoch 17, batch 4800, loss[loss=0.1841, simple_loss=0.2712, pruned_loss=0.04846, over 17012.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2754, pruned_loss=0.05132, over 3229430.87 frames. ], batch size: 55, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:49:58,980 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167213.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:50:27,972 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 2.008e+02 2.170e+02 2.594e+02 4.937e+02, threshold=4.340e+02, percent-clipped=1.0 2023-04-30 12:50:32,464 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167235.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:50:54,312 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9351, 2.0656, 2.1523, 3.4487, 2.0703, 2.3511, 2.1508, 2.2121], device='cuda:6'), covar=tensor([0.1362, 0.3571, 0.2710, 0.0586, 0.3920, 0.2536, 0.3537, 0.3193], device='cuda:6'), in_proj_covar=tensor([0.0387, 0.0427, 0.0352, 0.0322, 0.0427, 0.0493, 0.0395, 0.0498], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 12:50:58,073 INFO [train.py:904] (6/8) Epoch 17, batch 4850, loss[loss=0.1844, simple_loss=0.276, pruned_loss=0.04638, over 16517.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2769, pruned_loss=0.05088, over 3210792.19 frames. ], batch size: 75, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:51:46,663 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=167285.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:52:12,193 INFO [train.py:904] (6/8) Epoch 17, batch 4900, loss[loss=0.1915, simple_loss=0.2708, pruned_loss=0.05606, over 11794.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.2761, pruned_loss=0.04962, over 3187676.00 frames. ], batch size: 248, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:52:30,125 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6467, 1.7882, 1.6256, 1.5156, 1.9174, 1.6726, 1.5732, 1.9647], device='cuda:6'), covar=tensor([0.0138, 0.0288, 0.0363, 0.0347, 0.0186, 0.0244, 0.0163, 0.0196], device='cuda:6'), in_proj_covar=tensor([0.0189, 0.0224, 0.0217, 0.0217, 0.0226, 0.0226, 0.0228, 0.0222], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 12:52:55,965 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.440e+02 1.938e+02 2.305e+02 2.710e+02 4.454e+02, threshold=4.611e+02, percent-clipped=1.0 2023-04-30 12:53:24,988 INFO [train.py:904] (6/8) Epoch 17, batch 4950, loss[loss=0.1985, simple_loss=0.2879, pruned_loss=0.05458, over 16898.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2758, pruned_loss=0.04887, over 3199016.15 frames. ], batch size: 116, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:54:09,064 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4186, 3.3575, 3.4102, 3.5194, 3.5734, 3.2704, 3.5487, 3.6234], device='cuda:6'), covar=tensor([0.1149, 0.0912, 0.1088, 0.0627, 0.0557, 0.2438, 0.0946, 0.0650], device='cuda:6'), in_proj_covar=tensor([0.0594, 0.0732, 0.0866, 0.0744, 0.0559, 0.0592, 0.0591, 0.0693], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 12:54:22,544 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167391.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:54:37,753 INFO [train.py:904] (6/8) Epoch 17, batch 5000, loss[loss=0.1821, simple_loss=0.2716, pruned_loss=0.04627, over 17226.00 frames. ], tot_loss[loss=0.1874, simple_loss=0.2773, pruned_loss=0.04877, over 3198821.99 frames. ], batch size: 44, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:54:46,617 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8565, 2.0791, 2.3543, 3.1070, 2.1386, 2.2691, 2.2264, 2.2005], device='cuda:6'), covar=tensor([0.1249, 0.3174, 0.2244, 0.0632, 0.3830, 0.2397, 0.3132, 0.3021], device='cuda:6'), in_proj_covar=tensor([0.0387, 0.0428, 0.0353, 0.0322, 0.0427, 0.0494, 0.0396, 0.0499], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 12:55:02,809 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167420.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:55:19,900 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=167432.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:55:20,678 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.629e+02 2.024e+02 2.500e+02 2.980e+02 7.048e+02, threshold=5.000e+02, percent-clipped=1.0 2023-04-30 12:55:29,660 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167439.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:55:46,692 INFO [train.py:904] (6/8) Epoch 17, batch 5050, loss[loss=0.2033, simple_loss=0.2861, pruned_loss=0.06028, over 11831.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2783, pruned_loss=0.04894, over 3189575.56 frames. ], batch size: 248, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:55:51,610 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=167456.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:55:57,237 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=167460.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:56:08,481 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167468.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:56:16,430 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-30 12:56:44,115 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=167493.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:56:56,776 INFO [train.py:904] (6/8) Epoch 17, batch 5100, loss[loss=0.1713, simple_loss=0.2582, pruned_loss=0.04218, over 16685.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2762, pruned_loss=0.0481, over 3203471.27 frames. ], batch size: 57, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:57:01,820 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5921, 3.5105, 4.0343, 1.8135, 4.2286, 4.1949, 3.0341, 3.0656], device='cuda:6'), covar=tensor([0.0830, 0.0270, 0.0156, 0.1323, 0.0051, 0.0113, 0.0403, 0.0465], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0106, 0.0093, 0.0137, 0.0075, 0.0119, 0.0126, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 12:57:06,909 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5886, 4.7490, 4.9747, 4.6690, 4.7734, 5.3291, 4.8139, 4.4933], device='cuda:6'), covar=tensor([0.1236, 0.1738, 0.1435, 0.1827, 0.2162, 0.0823, 0.1323, 0.2380], device='cuda:6'), in_proj_covar=tensor([0.0383, 0.0548, 0.0593, 0.0459, 0.0615, 0.0639, 0.0471, 0.0616], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 12:57:28,012 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9915, 3.6682, 3.6617, 2.2590, 3.2594, 3.5913, 3.3662, 1.9484], device='cuda:6'), covar=tensor([0.0580, 0.0046, 0.0047, 0.0439, 0.0094, 0.0118, 0.0114, 0.0532], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0076, 0.0077, 0.0131, 0.0091, 0.0100, 0.0089, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 12:57:39,791 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.439e+02 2.010e+02 2.231e+02 2.542e+02 5.876e+02, threshold=4.463e+02, percent-clipped=1.0 2023-04-30 12:58:08,928 INFO [train.py:904] (6/8) Epoch 17, batch 5150, loss[loss=0.175, simple_loss=0.2735, pruned_loss=0.03823, over 16832.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2767, pruned_loss=0.04787, over 3191661.51 frames. ], batch size: 83, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:58:38,915 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-30 12:58:56,779 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167585.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:59:21,203 INFO [train.py:904] (6/8) Epoch 17, batch 5200, loss[loss=0.1738, simple_loss=0.2553, pruned_loss=0.04621, over 17243.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2751, pruned_loss=0.04741, over 3199449.84 frames. ], batch size: 52, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:00:07,260 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 2.038e+02 2.260e+02 2.754e+02 5.460e+02, threshold=4.520e+02, percent-clipped=3.0 2023-04-30 13:00:08,172 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167633.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:00:31,966 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0958, 5.0359, 4.9323, 4.2354, 4.9771, 1.8470, 4.7425, 4.7896], device='cuda:6'), covar=tensor([0.0091, 0.0079, 0.0177, 0.0452, 0.0109, 0.2566, 0.0138, 0.0212], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0139, 0.0185, 0.0170, 0.0160, 0.0196, 0.0174, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:00:36,319 INFO [train.py:904] (6/8) Epoch 17, batch 5250, loss[loss=0.1825, simple_loss=0.2774, pruned_loss=0.04376, over 16652.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2726, pruned_loss=0.04686, over 3213767.98 frames. ], batch size: 62, lr: 3.99e-03, grad_scale: 16.0 2023-04-30 13:00:57,474 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-30 13:01:48,347 INFO [train.py:904] (6/8) Epoch 17, batch 5300, loss[loss=0.166, simple_loss=0.2518, pruned_loss=0.0401, over 16546.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2693, pruned_loss=0.04564, over 3207857.11 frames. ], batch size: 62, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:02:12,843 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=167719.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:02:35,320 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.320e+02 2.142e+02 2.405e+02 3.044e+02 5.207e+02, threshold=4.809e+02, percent-clipped=2.0 2023-04-30 13:03:01,788 INFO [train.py:904] (6/8) Epoch 17, batch 5350, loss[loss=0.1808, simple_loss=0.2728, pruned_loss=0.04437, over 16476.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2674, pruned_loss=0.04468, over 3218097.94 frames. ], batch size: 75, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:03:08,794 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167756.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:03:09,242 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 13:03:13,821 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167760.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:03:23,264 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=167767.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:03:41,680 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=167780.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:03:53,663 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=167788.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:04:14,425 INFO [train.py:904] (6/8) Epoch 17, batch 5400, loss[loss=0.1939, simple_loss=0.2897, pruned_loss=0.04906, over 16689.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2702, pruned_loss=0.04546, over 3205779.55 frames. ], batch size: 134, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:04:17,675 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167804.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:04:24,146 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167808.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:04:48,954 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9089, 1.2503, 1.7680, 1.7879, 1.9113, 2.0238, 1.5010, 1.9355], device='cuda:6'), covar=tensor([0.0189, 0.0384, 0.0170, 0.0255, 0.0212, 0.0147, 0.0429, 0.0101], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0187, 0.0174, 0.0178, 0.0186, 0.0143, 0.0188, 0.0138], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:04:54,048 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=167828.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:05:02,680 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.037e+02 2.360e+02 2.749e+02 5.264e+02, threshold=4.720e+02, percent-clipped=1.0 2023-04-30 13:05:31,682 INFO [train.py:904] (6/8) Epoch 17, batch 5450, loss[loss=0.2112, simple_loss=0.2847, pruned_loss=0.06884, over 11907.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2728, pruned_loss=0.04693, over 3204627.78 frames. ], batch size: 247, lr: 3.99e-03, grad_scale: 4.0 2023-04-30 13:06:48,592 INFO [train.py:904] (6/8) Epoch 17, batch 5500, loss[loss=0.2362, simple_loss=0.3184, pruned_loss=0.07701, over 16632.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.2802, pruned_loss=0.05166, over 3178155.25 frames. ], batch size: 35, lr: 3.99e-03, grad_scale: 4.0 2023-04-30 13:07:01,278 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8011, 1.8536, 1.6118, 1.5640, 1.9701, 1.6566, 1.6792, 1.9666], device='cuda:6'), covar=tensor([0.0148, 0.0233, 0.0320, 0.0294, 0.0175, 0.0229, 0.0178, 0.0159], device='cuda:6'), in_proj_covar=tensor([0.0186, 0.0223, 0.0216, 0.0217, 0.0225, 0.0224, 0.0225, 0.0220], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:07:39,670 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.878e+02 2.929e+02 3.465e+02 4.184e+02 8.083e+02, threshold=6.929e+02, percent-clipped=17.0 2023-04-30 13:08:05,266 INFO [train.py:904] (6/8) Epoch 17, batch 5550, loss[loss=0.2207, simple_loss=0.3072, pruned_loss=0.06713, over 15530.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.288, pruned_loss=0.05746, over 3135164.47 frames. ], batch size: 191, lr: 3.99e-03, grad_scale: 4.0 2023-04-30 13:09:31,455 INFO [train.py:904] (6/8) Epoch 17, batch 5600, loss[loss=0.2729, simple_loss=0.3265, pruned_loss=0.1097, over 11065.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.293, pruned_loss=0.06158, over 3110742.19 frames. ], batch size: 247, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:10:00,837 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-30 13:10:27,347 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.908e+02 3.389e+02 4.058e+02 4.967e+02 1.149e+03, threshold=8.116e+02, percent-clipped=3.0 2023-04-30 13:10:28,065 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=168035.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:10:54,415 INFO [train.py:904] (6/8) Epoch 17, batch 5650, loss[loss=0.3027, simple_loss=0.3532, pruned_loss=0.126, over 11117.00 frames. ], tot_loss[loss=0.2147, simple_loss=0.2978, pruned_loss=0.06579, over 3079213.43 frames. ], batch size: 247, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:11:30,929 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=168075.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:11:50,815 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=168088.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:12:04,707 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=168096.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 13:12:11,681 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2727, 3.5506, 3.9075, 1.5728, 4.1682, 4.1859, 2.9836, 2.8602], device='cuda:6'), covar=tensor([0.1130, 0.0212, 0.0201, 0.1408, 0.0064, 0.0131, 0.0437, 0.0568], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0104, 0.0092, 0.0136, 0.0074, 0.0118, 0.0124, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 13:12:12,308 INFO [train.py:904] (6/8) Epoch 17, batch 5700, loss[loss=0.2105, simple_loss=0.3073, pruned_loss=0.05681, over 16591.00 frames. ], tot_loss[loss=0.2167, simple_loss=0.299, pruned_loss=0.06724, over 3065163.12 frames. ], batch size: 68, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:12:46,747 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=168123.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:13:04,293 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.185e+02 3.066e+02 3.948e+02 4.929e+02 1.585e+03, threshold=7.895e+02, percent-clipped=5.0 2023-04-30 13:13:06,078 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=168136.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:13:31,681 INFO [train.py:904] (6/8) Epoch 17, batch 5750, loss[loss=0.203, simple_loss=0.2914, pruned_loss=0.05733, over 16276.00 frames. ], tot_loss[loss=0.2207, simple_loss=0.3022, pruned_loss=0.06956, over 3026359.55 frames. ], batch size: 165, lr: 3.98e-03, grad_scale: 8.0 2023-04-30 13:14:28,410 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-30 13:14:50,313 INFO [train.py:904] (6/8) Epoch 17, batch 5800, loss[loss=0.1914, simple_loss=0.2797, pruned_loss=0.05154, over 16636.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.3007, pruned_loss=0.06733, over 3038390.72 frames. ], batch size: 57, lr: 3.98e-03, grad_scale: 8.0 2023-04-30 13:15:41,127 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=168233.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:15:43,217 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.888e+02 3.447e+02 4.230e+02 9.306e+02, threshold=6.893e+02, percent-clipped=1.0 2023-04-30 13:15:46,872 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-30 13:16:09,621 INFO [train.py:904] (6/8) Epoch 17, batch 5850, loss[loss=0.2323, simple_loss=0.313, pruned_loss=0.07585, over 15379.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.2987, pruned_loss=0.06586, over 3035864.93 frames. ], batch size: 191, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:17:18,287 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.50 vs. limit=2.0 2023-04-30 13:17:19,344 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=168294.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:17:32,944 INFO [train.py:904] (6/8) Epoch 17, batch 5900, loss[loss=0.2157, simple_loss=0.3002, pruned_loss=0.06562, over 16404.00 frames. ], tot_loss[loss=0.2156, simple_loss=0.2987, pruned_loss=0.06621, over 3036586.85 frames. ], batch size: 35, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:17:49,067 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5513, 3.5959, 2.0714, 4.0010, 2.6617, 3.9925, 2.1535, 2.7543], device='cuda:6'), covar=tensor([0.0248, 0.0323, 0.1666, 0.0220, 0.0873, 0.0519, 0.1633, 0.0790], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0172, 0.0191, 0.0150, 0.0172, 0.0210, 0.0200, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 13:18:05,652 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.05 vs. limit=5.0 2023-04-30 13:18:26,518 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-30 13:18:28,194 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.786e+02 3.302e+02 4.334e+02 9.819e+02, threshold=6.604e+02, percent-clipped=5.0 2023-04-30 13:18:52,631 INFO [train.py:904] (6/8) Epoch 17, batch 5950, loss[loss=0.1961, simple_loss=0.2866, pruned_loss=0.05276, over 16669.00 frames. ], tot_loss[loss=0.2149, simple_loss=0.2995, pruned_loss=0.06513, over 3039620.52 frames. ], batch size: 89, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:19:01,908 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-30 13:19:29,629 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=168375.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:19:51,841 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0028, 4.0817, 3.9162, 3.6593, 3.6570, 4.0136, 3.6851, 3.7774], device='cuda:6'), covar=tensor([0.0634, 0.0609, 0.0288, 0.0268, 0.0775, 0.0498, 0.1034, 0.0648], device='cuda:6'), in_proj_covar=tensor([0.0270, 0.0379, 0.0320, 0.0309, 0.0329, 0.0360, 0.0219, 0.0382], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:19:53,443 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=168391.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 13:20:12,034 INFO [train.py:904] (6/8) Epoch 17, batch 6000, loss[loss=0.1912, simple_loss=0.2772, pruned_loss=0.05264, over 16850.00 frames. ], tot_loss[loss=0.2151, simple_loss=0.2995, pruned_loss=0.06531, over 3050440.08 frames. ], batch size: 102, lr: 3.98e-03, grad_scale: 8.0 2023-04-30 13:20:12,035 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 13:20:21,980 INFO [train.py:938] (6/8) Epoch 17, validation: loss=0.1535, simple_loss=0.2667, pruned_loss=0.0202, over 944034.00 frames. 2023-04-30 13:20:21,980 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 13:20:53,932 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=168423.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:20:54,022 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=168423.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:21:15,436 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.147e+02 2.768e+02 3.388e+02 4.185e+02 7.743e+02, threshold=6.777e+02, percent-clipped=1.0 2023-04-30 13:21:39,928 INFO [train.py:904] (6/8) Epoch 17, batch 6050, loss[loss=0.2287, simple_loss=0.3165, pruned_loss=0.07046, over 16665.00 frames. ], tot_loss[loss=0.2119, simple_loss=0.297, pruned_loss=0.06337, over 3099163.13 frames. ], batch size: 134, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:21:45,941 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.64 vs. limit=5.0 2023-04-30 13:21:53,981 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4479, 2.3435, 2.4176, 4.3489, 2.2666, 2.7508, 2.4142, 2.5140], device='cuda:6'), covar=tensor([0.1178, 0.3467, 0.2614, 0.0384, 0.3857, 0.2326, 0.3438, 0.3194], device='cuda:6'), in_proj_covar=tensor([0.0383, 0.0423, 0.0351, 0.0318, 0.0425, 0.0489, 0.0393, 0.0496], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:21:58,599 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0796, 2.4297, 2.6046, 1.8782, 2.6588, 2.7928, 2.4139, 2.3937], device='cuda:6'), covar=tensor([0.0706, 0.0231, 0.0213, 0.0940, 0.0099, 0.0248, 0.0424, 0.0434], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0104, 0.0092, 0.0136, 0.0074, 0.0118, 0.0124, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 13:22:09,545 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=168471.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:22:59,844 INFO [train.py:904] (6/8) Epoch 17, batch 6100, loss[loss=0.2398, simple_loss=0.3029, pruned_loss=0.08841, over 11672.00 frames. ], tot_loss[loss=0.2102, simple_loss=0.2959, pruned_loss=0.06223, over 3097991.78 frames. ], batch size: 247, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:23:55,308 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.024e+02 2.750e+02 3.352e+02 4.017e+02 8.288e+02, threshold=6.704e+02, percent-clipped=2.0 2023-04-30 13:24:18,356 INFO [train.py:904] (6/8) Epoch 17, batch 6150, loss[loss=0.1872, simple_loss=0.2667, pruned_loss=0.05383, over 16418.00 frames. ], tot_loss[loss=0.2082, simple_loss=0.294, pruned_loss=0.06121, over 3110477.74 frames. ], batch size: 68, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:25:17,263 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=168589.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:25:36,300 INFO [train.py:904] (6/8) Epoch 17, batch 6200, loss[loss=0.1801, simple_loss=0.2715, pruned_loss=0.04434, over 16618.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.2926, pruned_loss=0.06116, over 3115039.44 frames. ], batch size: 75, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:26:31,026 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.813e+02 2.748e+02 3.467e+02 4.001e+02 6.501e+02, threshold=6.934e+02, percent-clipped=0.0 2023-04-30 13:26:48,932 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9514, 4.8957, 4.7375, 4.0430, 4.8250, 1.6800, 4.5707, 4.4835], device='cuda:6'), covar=tensor([0.0086, 0.0090, 0.0185, 0.0355, 0.0104, 0.2788, 0.0164, 0.0201], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0138, 0.0184, 0.0170, 0.0158, 0.0195, 0.0172, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:26:53,706 INFO [train.py:904] (6/8) Epoch 17, batch 6250, loss[loss=0.2029, simple_loss=0.296, pruned_loss=0.05493, over 17110.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2925, pruned_loss=0.06087, over 3118294.89 frames. ], batch size: 48, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:27:53,831 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=168691.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:28:10,638 INFO [train.py:904] (6/8) Epoch 17, batch 6300, loss[loss=0.1975, simple_loss=0.2842, pruned_loss=0.05543, over 17044.00 frames. ], tot_loss[loss=0.206, simple_loss=0.2922, pruned_loss=0.05987, over 3128382.53 frames. ], batch size: 55, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:28:16,326 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1131, 2.0684, 1.7019, 1.7970, 2.3064, 1.9602, 2.0255, 2.3927], device='cuda:6'), covar=tensor([0.0172, 0.0321, 0.0422, 0.0392, 0.0197, 0.0279, 0.0186, 0.0208], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0223, 0.0215, 0.0217, 0.0224, 0.0223, 0.0225, 0.0218], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:28:16,703 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2023-04-30 13:29:06,429 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.898e+02 2.739e+02 3.263e+02 4.043e+02 7.710e+02, threshold=6.525e+02, percent-clipped=1.0 2023-04-30 13:29:09,982 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=168739.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:29:29,592 INFO [train.py:904] (6/8) Epoch 17, batch 6350, loss[loss=0.2096, simple_loss=0.2923, pruned_loss=0.0634, over 16408.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2925, pruned_loss=0.06066, over 3148827.74 frames. ], batch size: 146, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:29:47,456 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2663, 2.0144, 1.5434, 1.6842, 2.3109, 2.0081, 2.2430, 2.4969], device='cuda:6'), covar=tensor([0.0203, 0.0470, 0.0617, 0.0549, 0.0278, 0.0426, 0.0250, 0.0299], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0223, 0.0216, 0.0217, 0.0225, 0.0223, 0.0224, 0.0218], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:29:57,461 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.77 vs. limit=5.0 2023-04-30 13:30:21,359 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9175, 5.3873, 5.6211, 5.3755, 5.3690, 5.9573, 5.4253, 5.2611], device='cuda:6'), covar=tensor([0.0965, 0.1715, 0.2229, 0.1611, 0.2593, 0.0982, 0.1635, 0.2231], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0560, 0.0614, 0.0470, 0.0632, 0.0647, 0.0487, 0.0632], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 13:30:46,693 INFO [train.py:904] (6/8) Epoch 17, batch 6400, loss[loss=0.1795, simple_loss=0.2729, pruned_loss=0.04303, over 16814.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2929, pruned_loss=0.06187, over 3121077.16 frames. ], batch size: 102, lr: 3.98e-03, grad_scale: 8.0 2023-04-30 13:31:42,103 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.043e+02 2.957e+02 3.621e+02 4.188e+02 7.946e+02, threshold=7.242e+02, percent-clipped=4.0 2023-04-30 13:32:03,553 INFO [train.py:904] (6/8) Epoch 17, batch 6450, loss[loss=0.1881, simple_loss=0.2782, pruned_loss=0.04898, over 16535.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2918, pruned_loss=0.06021, over 3143991.87 frames. ], batch size: 68, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:33:02,253 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=168889.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:33:20,845 INFO [train.py:904] (6/8) Epoch 17, batch 6500, loss[loss=0.2012, simple_loss=0.2939, pruned_loss=0.05424, over 16748.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2903, pruned_loss=0.05991, over 3137374.10 frames. ], batch size: 83, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:34:15,419 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=168937.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:34:16,252 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.781e+02 2.744e+02 3.257e+02 4.082e+02 7.063e+02, threshold=6.513e+02, percent-clipped=0.0 2023-04-30 13:34:40,483 INFO [train.py:904] (6/8) Epoch 17, batch 6550, loss[loss=0.2116, simple_loss=0.3085, pruned_loss=0.05739, over 16666.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2937, pruned_loss=0.06157, over 3110391.27 frames. ], batch size: 57, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:34:48,114 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-04-30 13:35:03,811 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=168967.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:35:06,322 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=168969.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 13:35:58,162 INFO [train.py:904] (6/8) Epoch 17, batch 6600, loss[loss=0.2003, simple_loss=0.2889, pruned_loss=0.05586, over 16470.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2962, pruned_loss=0.06226, over 3101705.90 frames. ], batch size: 68, lr: 3.97e-03, grad_scale: 4.0 2023-04-30 13:36:37,549 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169028.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:36:40,799 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169030.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 13:36:50,873 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.775e+02 2.872e+02 3.571e+02 4.497e+02 9.620e+02, threshold=7.142e+02, percent-clipped=5.0 2023-04-30 13:36:56,500 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0421, 5.4974, 5.7828, 5.4277, 5.6358, 6.1116, 5.6257, 5.4149], device='cuda:6'), covar=tensor([0.0907, 0.1863, 0.2009, 0.1813, 0.2067, 0.0884, 0.1377, 0.2036], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0560, 0.0612, 0.0470, 0.0632, 0.0646, 0.0486, 0.0631], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 13:37:13,177 INFO [train.py:904] (6/8) Epoch 17, batch 6650, loss[loss=0.2013, simple_loss=0.2895, pruned_loss=0.0566, over 16383.00 frames. ], tot_loss[loss=0.2129, simple_loss=0.2976, pruned_loss=0.06414, over 3084037.53 frames. ], batch size: 146, lr: 3.97e-03, grad_scale: 4.0 2023-04-30 13:37:13,724 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7986, 3.8227, 4.1106, 4.0798, 4.1095, 3.8506, 3.8886, 3.8889], device='cuda:6'), covar=tensor([0.0353, 0.0611, 0.0431, 0.0468, 0.0495, 0.0428, 0.0860, 0.0499], device='cuda:6'), in_proj_covar=tensor([0.0382, 0.0417, 0.0407, 0.0383, 0.0454, 0.0430, 0.0527, 0.0341], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 13:37:59,822 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169082.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:38:25,000 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2267, 4.1940, 4.1109, 3.3708, 4.1448, 1.6583, 3.9393, 3.7268], device='cuda:6'), covar=tensor([0.0100, 0.0083, 0.0168, 0.0295, 0.0085, 0.2663, 0.0119, 0.0222], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0138, 0.0184, 0.0169, 0.0158, 0.0195, 0.0172, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:38:30,528 INFO [train.py:904] (6/8) Epoch 17, batch 6700, loss[loss=0.2405, simple_loss=0.307, pruned_loss=0.08704, over 11473.00 frames. ], tot_loss[loss=0.2126, simple_loss=0.2966, pruned_loss=0.06427, over 3071596.51 frames. ], batch size: 246, lr: 3.97e-03, grad_scale: 4.0 2023-04-30 13:38:53,667 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169116.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:38:56,449 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5450, 2.5959, 1.7155, 2.6327, 2.1140, 2.7569, 1.9775, 2.3104], device='cuda:6'), covar=tensor([0.0289, 0.0378, 0.1384, 0.0196, 0.0649, 0.0565, 0.1267, 0.0563], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0170, 0.0191, 0.0150, 0.0171, 0.0210, 0.0198, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 13:39:25,873 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3863, 2.9355, 2.9847, 2.0216, 2.7744, 2.1663, 3.0843, 3.1840], device='cuda:6'), covar=tensor([0.0300, 0.0822, 0.0645, 0.1959, 0.0841, 0.0998, 0.0661, 0.0845], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0158, 0.0164, 0.0150, 0.0142, 0.0128, 0.0142, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 13:39:26,519 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.893e+02 2.876e+02 3.506e+02 4.097e+02 7.829e+02, threshold=7.011e+02, percent-clipped=1.0 2023-04-30 13:39:34,977 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169143.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:39:37,384 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7640, 4.8009, 5.2164, 5.1621, 5.2091, 4.8425, 4.8255, 4.6029], device='cuda:6'), covar=tensor([0.0305, 0.0522, 0.0296, 0.0400, 0.0488, 0.0343, 0.0919, 0.0494], device='cuda:6'), in_proj_covar=tensor([0.0380, 0.0415, 0.0404, 0.0381, 0.0452, 0.0428, 0.0525, 0.0339], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 13:39:47,624 INFO [train.py:904] (6/8) Epoch 17, batch 6750, loss[loss=0.1901, simple_loss=0.2691, pruned_loss=0.0555, over 16656.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2946, pruned_loss=0.06367, over 3071703.83 frames. ], batch size: 62, lr: 3.97e-03, grad_scale: 4.0 2023-04-30 13:39:58,654 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169159.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:40:07,106 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-30 13:40:26,730 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169177.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:41:04,383 INFO [train.py:904] (6/8) Epoch 17, batch 6800, loss[loss=0.2366, simple_loss=0.3227, pruned_loss=0.07528, over 16911.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2951, pruned_loss=0.06348, over 3086616.98 frames. ], batch size: 109, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:41:33,654 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169220.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:42:02,686 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.174e+02 2.920e+02 3.483e+02 4.404e+02 7.991e+02, threshold=6.967e+02, percent-clipped=1.0 2023-04-30 13:42:11,504 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6316, 2.6235, 1.8632, 2.6754, 2.1647, 2.7921, 2.1323, 2.4433], device='cuda:6'), covar=tensor([0.0335, 0.0438, 0.1360, 0.0254, 0.0732, 0.0474, 0.1217, 0.0608], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0171, 0.0192, 0.0150, 0.0171, 0.0210, 0.0199, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 13:42:23,105 INFO [train.py:904] (6/8) Epoch 17, batch 6850, loss[loss=0.2323, simple_loss=0.3238, pruned_loss=0.07042, over 16267.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2963, pruned_loss=0.0642, over 3077409.91 frames. ], batch size: 165, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:42:58,143 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3273, 3.5325, 3.6200, 2.5321, 3.4251, 3.6958, 3.5374, 2.0376], device='cuda:6'), covar=tensor([0.0451, 0.0050, 0.0045, 0.0329, 0.0068, 0.0077, 0.0062, 0.0456], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0076, 0.0078, 0.0133, 0.0090, 0.0102, 0.0090, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 13:43:37,898 INFO [train.py:904] (6/8) Epoch 17, batch 6900, loss[loss=0.2185, simple_loss=0.302, pruned_loss=0.06748, over 15240.00 frames. ], tot_loss[loss=0.2134, simple_loss=0.2986, pruned_loss=0.06409, over 3082708.31 frames. ], batch size: 190, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:43:58,755 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1536, 3.6033, 3.6209, 2.3962, 3.3131, 3.6253, 3.3901, 2.0063], device='cuda:6'), covar=tensor([0.0523, 0.0045, 0.0047, 0.0364, 0.0084, 0.0099, 0.0083, 0.0448], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0076, 0.0078, 0.0133, 0.0090, 0.0102, 0.0090, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 13:44:11,064 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=169323.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:44:14,329 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=169325.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 13:44:33,264 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.645e+02 2.634e+02 3.268e+02 4.133e+02 7.362e+02, threshold=6.535e+02, percent-clipped=1.0 2023-04-30 13:44:49,707 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0341, 5.6613, 5.8172, 5.5103, 5.6257, 6.1592, 5.5549, 5.4281], device='cuda:6'), covar=tensor([0.0873, 0.1723, 0.1794, 0.1859, 0.2347, 0.0893, 0.1511, 0.2121], device='cuda:6'), in_proj_covar=tensor([0.0388, 0.0559, 0.0611, 0.0467, 0.0631, 0.0646, 0.0488, 0.0632], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 13:44:55,623 INFO [train.py:904] (6/8) Epoch 17, batch 6950, loss[loss=0.2141, simple_loss=0.3044, pruned_loss=0.06186, over 16755.00 frames. ], tot_loss[loss=0.2157, simple_loss=0.3001, pruned_loss=0.06561, over 3076591.35 frames. ], batch size: 124, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:46:11,848 INFO [train.py:904] (6/8) Epoch 17, batch 7000, loss[loss=0.1817, simple_loss=0.2848, pruned_loss=0.03933, over 16815.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.3002, pruned_loss=0.06507, over 3066527.27 frames. ], batch size: 102, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:47:07,099 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.910e+02 3.013e+02 3.558e+02 4.321e+02 1.050e+03, threshold=7.116e+02, percent-clipped=3.0 2023-04-30 13:47:07,488 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=169438.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:47:28,661 INFO [train.py:904] (6/8) Epoch 17, batch 7050, loss[loss=0.2335, simple_loss=0.3163, pruned_loss=0.07535, over 15313.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.3001, pruned_loss=0.0644, over 3052987.33 frames. ], batch size: 190, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:48:00,524 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=169472.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:48:47,182 INFO [train.py:904] (6/8) Epoch 17, batch 7100, loss[loss=0.2437, simple_loss=0.3093, pruned_loss=0.08905, over 11605.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2984, pruned_loss=0.06391, over 3052796.80 frames. ], batch size: 246, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:49:08,085 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=169515.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:49:21,811 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6409, 2.2771, 1.8340, 2.1018, 2.6139, 2.3106, 2.4960, 2.7528], device='cuda:6'), covar=tensor([0.0166, 0.0370, 0.0483, 0.0394, 0.0222, 0.0325, 0.0207, 0.0225], device='cuda:6'), in_proj_covar=tensor([0.0183, 0.0223, 0.0214, 0.0216, 0.0223, 0.0221, 0.0223, 0.0217], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:49:42,143 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 2.876e+02 3.484e+02 4.198e+02 1.044e+03, threshold=6.968e+02, percent-clipped=3.0 2023-04-30 13:49:43,905 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9836, 5.0279, 4.8565, 4.4956, 4.4789, 4.9369, 4.8181, 4.6182], device='cuda:6'), covar=tensor([0.0568, 0.0406, 0.0252, 0.0272, 0.0948, 0.0421, 0.0300, 0.0617], device='cuda:6'), in_proj_covar=tensor([0.0268, 0.0378, 0.0318, 0.0305, 0.0327, 0.0354, 0.0216, 0.0379], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:50:02,979 INFO [train.py:904] (6/8) Epoch 17, batch 7150, loss[loss=0.2085, simple_loss=0.2955, pruned_loss=0.06079, over 16821.00 frames. ], tot_loss[loss=0.212, simple_loss=0.2964, pruned_loss=0.06375, over 3057188.94 frames. ], batch size: 83, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:50:55,801 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8332, 2.7122, 2.6320, 1.8803, 2.5543, 2.7087, 2.6189, 1.8798], device='cuda:6'), covar=tensor([0.0411, 0.0074, 0.0068, 0.0348, 0.0118, 0.0112, 0.0106, 0.0388], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0077, 0.0078, 0.0132, 0.0090, 0.0102, 0.0089, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 13:51:02,239 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1376, 2.1005, 2.1406, 3.8236, 2.1004, 2.5109, 2.2343, 2.2832], device='cuda:6'), covar=tensor([0.1288, 0.3640, 0.2939, 0.0505, 0.4067, 0.2605, 0.3505, 0.3398], device='cuda:6'), in_proj_covar=tensor([0.0381, 0.0420, 0.0349, 0.0316, 0.0424, 0.0487, 0.0391, 0.0491], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:51:19,558 INFO [train.py:904] (6/8) Epoch 17, batch 7200, loss[loss=0.1852, simple_loss=0.2764, pruned_loss=0.04701, over 15379.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2935, pruned_loss=0.06153, over 3076975.87 frames. ], batch size: 190, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:51:48,686 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169621.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:51:51,913 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=169623.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:51:55,212 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=169625.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 13:52:16,450 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.648e+02 2.818e+02 3.396e+02 4.262e+02 6.944e+02, threshold=6.791e+02, percent-clipped=0.0 2023-04-30 13:52:38,960 INFO [train.py:904] (6/8) Epoch 17, batch 7250, loss[loss=0.1903, simple_loss=0.2703, pruned_loss=0.05517, over 16853.00 frames. ], tot_loss[loss=0.2063, simple_loss=0.2914, pruned_loss=0.06058, over 3070067.27 frames. ], batch size: 116, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:53:08,592 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=169671.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:53:12,522 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=169673.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 13:53:16,725 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2782, 1.5558, 1.9984, 2.2149, 2.3265, 2.5360, 1.6503, 2.4240], device='cuda:6'), covar=tensor([0.0206, 0.0485, 0.0284, 0.0302, 0.0287, 0.0168, 0.0502, 0.0134], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0185, 0.0171, 0.0174, 0.0184, 0.0142, 0.0187, 0.0136], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:53:25,517 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169682.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:53:57,114 INFO [train.py:904] (6/8) Epoch 17, batch 7300, loss[loss=0.182, simple_loss=0.2735, pruned_loss=0.04523, over 17178.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2906, pruned_loss=0.05987, over 3100531.57 frames. ], batch size: 44, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:54:41,500 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169729.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:54:55,061 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 2.965e+02 3.498e+02 4.419e+02 8.107e+02, threshold=6.996e+02, percent-clipped=3.0 2023-04-30 13:54:55,514 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=169738.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:55:10,714 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169747.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:55:17,629 INFO [train.py:904] (6/8) Epoch 17, batch 7350, loss[loss=0.2213, simple_loss=0.3075, pruned_loss=0.0675, over 15379.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2913, pruned_loss=0.06048, over 3092907.24 frames. ], batch size: 190, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:55:49,910 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=169772.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:55:58,397 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8112, 3.7206, 3.8645, 3.9777, 4.0532, 3.6528, 4.0009, 4.0769], device='cuda:6'), covar=tensor([0.1459, 0.1128, 0.1272, 0.0645, 0.0620, 0.1865, 0.0787, 0.0640], device='cuda:6'), in_proj_covar=tensor([0.0579, 0.0716, 0.0853, 0.0731, 0.0554, 0.0582, 0.0592, 0.0683], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:56:12,138 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=169786.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:56:12,320 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169786.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:56:19,783 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169790.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:56:37,953 INFO [train.py:904] (6/8) Epoch 17, batch 7400, loss[loss=0.2, simple_loss=0.2847, pruned_loss=0.05759, over 17006.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2922, pruned_loss=0.06121, over 3074867.29 frames. ], batch size: 53, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:56:49,192 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169808.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:56:59,857 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=169815.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:57:07,986 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=169820.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:57:36,683 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.712e+02 2.775e+02 3.366e+02 3.998e+02 8.087e+02, threshold=6.732e+02, percent-clipped=3.0 2023-04-30 13:57:52,236 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169847.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:57:59,463 INFO [train.py:904] (6/8) Epoch 17, batch 7450, loss[loss=0.2381, simple_loss=0.3219, pruned_loss=0.07714, over 16798.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2935, pruned_loss=0.06186, over 3083484.67 frames. ], batch size: 124, lr: 3.96e-03, grad_scale: 8.0 2023-04-30 13:57:59,966 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6999, 4.9625, 4.7359, 4.7780, 4.4953, 4.4741, 4.4453, 5.0138], device='cuda:6'), covar=tensor([0.1094, 0.0880, 0.0976, 0.0857, 0.0815, 0.1109, 0.1089, 0.1040], device='cuda:6'), in_proj_covar=tensor([0.0623, 0.0758, 0.0621, 0.0563, 0.0478, 0.0489, 0.0627, 0.0588], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:58:19,039 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=169863.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:58:38,636 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0235, 2.0235, 2.1667, 3.6095, 2.0315, 2.3605, 2.1114, 2.1551], device='cuda:6'), covar=tensor([0.1270, 0.3643, 0.2827, 0.0533, 0.4177, 0.2655, 0.3746, 0.3436], device='cuda:6'), in_proj_covar=tensor([0.0378, 0.0421, 0.0348, 0.0316, 0.0425, 0.0486, 0.0392, 0.0490], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 13:59:20,321 INFO [train.py:904] (6/8) Epoch 17, batch 7500, loss[loss=0.1703, simple_loss=0.2627, pruned_loss=0.03895, over 16891.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.2935, pruned_loss=0.06116, over 3084519.00 frames. ], batch size: 96, lr: 3.96e-03, grad_scale: 8.0 2023-04-30 14:00:17,626 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.828e+02 2.669e+02 3.210e+02 3.850e+02 6.683e+02, threshold=6.420e+02, percent-clipped=0.0 2023-04-30 14:00:39,248 INFO [train.py:904] (6/8) Epoch 17, batch 7550, loss[loss=0.1823, simple_loss=0.2742, pruned_loss=0.04518, over 16805.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2934, pruned_loss=0.0618, over 3064919.73 frames. ], batch size: 83, lr: 3.96e-03, grad_scale: 8.0 2023-04-30 14:01:01,193 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.66 vs. limit=5.0 2023-04-30 14:01:19,635 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=169977.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:01:58,035 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3565, 5.5961, 5.3467, 5.4166, 5.1117, 5.0219, 5.0503, 5.7437], device='cuda:6'), covar=tensor([0.1065, 0.0827, 0.0931, 0.0791, 0.0818, 0.0704, 0.1077, 0.0776], device='cuda:6'), in_proj_covar=tensor([0.0620, 0.0757, 0.0619, 0.0560, 0.0475, 0.0487, 0.0625, 0.0583], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 14:02:00,832 INFO [train.py:904] (6/8) Epoch 17, batch 7600, loss[loss=0.1985, simple_loss=0.2828, pruned_loss=0.05708, over 16907.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2927, pruned_loss=0.06148, over 3086907.17 frames. ], batch size: 109, lr: 3.96e-03, grad_scale: 8.0 2023-04-30 14:02:31,999 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3880, 2.1569, 1.7100, 1.8884, 2.3871, 2.0898, 2.1626, 2.5444], device='cuda:6'), covar=tensor([0.0149, 0.0365, 0.0490, 0.0430, 0.0223, 0.0357, 0.0201, 0.0234], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0220, 0.0214, 0.0215, 0.0220, 0.0219, 0.0221, 0.0214], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 14:02:58,189 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.022e+02 2.819e+02 3.511e+02 4.461e+02 7.625e+02, threshold=7.022e+02, percent-clipped=5.0 2023-04-30 14:03:08,030 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=170045.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:03:17,366 INFO [train.py:904] (6/8) Epoch 17, batch 7650, loss[loss=0.1905, simple_loss=0.2803, pruned_loss=0.05031, over 17194.00 frames. ], tot_loss[loss=0.2082, simple_loss=0.293, pruned_loss=0.06171, over 3107637.67 frames. ], batch size: 46, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:04:07,610 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=170085.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:04:31,773 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-30 14:04:33,128 INFO [train.py:904] (6/8) Epoch 17, batch 7700, loss[loss=0.1931, simple_loss=0.2804, pruned_loss=0.05294, over 16214.00 frames. ], tot_loss[loss=0.2091, simple_loss=0.2931, pruned_loss=0.06253, over 3083583.81 frames. ], batch size: 165, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:04:35,296 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=170103.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:04:39,672 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=170106.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:05:29,557 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.892e+02 2.959e+02 3.624e+02 4.617e+02 9.999e+02, threshold=7.248e+02, percent-clipped=4.0 2023-04-30 14:05:34,957 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=170142.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:05:50,028 INFO [train.py:904] (6/8) Epoch 17, batch 7750, loss[loss=0.1895, simple_loss=0.2844, pruned_loss=0.04729, over 16752.00 frames. ], tot_loss[loss=0.2094, simple_loss=0.2934, pruned_loss=0.06268, over 3083016.11 frames. ], batch size: 83, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:06:11,343 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6706, 1.6370, 2.2178, 2.5926, 2.5392, 3.0166, 1.6596, 2.9536], device='cuda:6'), covar=tensor([0.0186, 0.0508, 0.0296, 0.0268, 0.0292, 0.0160, 0.0556, 0.0117], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0184, 0.0170, 0.0173, 0.0183, 0.0142, 0.0186, 0.0136], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 14:06:42,169 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3782, 3.3752, 3.4172, 3.4995, 3.5237, 3.2947, 3.4943, 3.5666], device='cuda:6'), covar=tensor([0.1305, 0.0940, 0.1080, 0.0643, 0.0714, 0.2334, 0.1061, 0.0776], device='cuda:6'), in_proj_covar=tensor([0.0586, 0.0724, 0.0861, 0.0737, 0.0560, 0.0587, 0.0598, 0.0692], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 14:07:07,218 INFO [train.py:904] (6/8) Epoch 17, batch 7800, loss[loss=0.2192, simple_loss=0.3016, pruned_loss=0.06836, over 15276.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2937, pruned_loss=0.063, over 3077026.45 frames. ], batch size: 190, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:07:40,113 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2076, 4.1816, 4.1053, 3.3346, 4.1641, 1.7289, 3.8911, 3.7497], device='cuda:6'), covar=tensor([0.0129, 0.0112, 0.0183, 0.0301, 0.0103, 0.2613, 0.0160, 0.0213], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0136, 0.0183, 0.0167, 0.0156, 0.0193, 0.0170, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 14:08:03,351 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.635e+02 2.884e+02 3.642e+02 4.850e+02 8.293e+02, threshold=7.285e+02, percent-clipped=4.0 2023-04-30 14:08:06,908 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1928, 5.8218, 6.0220, 5.6656, 5.7523, 6.3341, 5.7258, 5.5403], device='cuda:6'), covar=tensor([0.0936, 0.1727, 0.2151, 0.2117, 0.2423, 0.0849, 0.1567, 0.2391], device='cuda:6'), in_proj_covar=tensor([0.0388, 0.0560, 0.0616, 0.0469, 0.0632, 0.0648, 0.0488, 0.0630], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 14:08:23,348 INFO [train.py:904] (6/8) Epoch 17, batch 7850, loss[loss=0.235, simple_loss=0.3019, pruned_loss=0.08403, over 11587.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2943, pruned_loss=0.06261, over 3088471.42 frames. ], batch size: 249, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:08:36,822 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3199, 4.3203, 4.2031, 2.7321, 3.8106, 4.2529, 3.7478, 2.2014], device='cuda:6'), covar=tensor([0.0581, 0.0033, 0.0045, 0.0402, 0.0084, 0.0096, 0.0090, 0.0469], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0077, 0.0078, 0.0132, 0.0091, 0.0103, 0.0090, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 14:09:00,965 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=170277.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:09:34,587 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5189, 3.5442, 1.9913, 3.9900, 2.5842, 3.9711, 2.2087, 2.8850], device='cuda:6'), covar=tensor([0.0275, 0.0382, 0.1733, 0.0225, 0.0875, 0.0598, 0.1490, 0.0700], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0171, 0.0193, 0.0150, 0.0173, 0.0212, 0.0200, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 14:09:38,277 INFO [train.py:904] (6/8) Epoch 17, batch 7900, loss[loss=0.1935, simple_loss=0.2781, pruned_loss=0.05449, over 16654.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2934, pruned_loss=0.0622, over 3082050.44 frames. ], batch size: 62, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:10:13,070 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=170325.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:10:13,177 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8080, 3.7591, 3.8954, 3.9975, 4.0697, 3.6682, 4.0072, 4.0897], device='cuda:6'), covar=tensor([0.1573, 0.1074, 0.1213, 0.0652, 0.0620, 0.1729, 0.0827, 0.0721], device='cuda:6'), in_proj_covar=tensor([0.0584, 0.0722, 0.0856, 0.0733, 0.0557, 0.0583, 0.0596, 0.0690], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 14:10:36,113 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.088e+02 2.800e+02 3.465e+02 4.175e+02 8.586e+02, threshold=6.929e+02, percent-clipped=4.0 2023-04-30 14:10:55,565 INFO [train.py:904] (6/8) Epoch 17, batch 7950, loss[loss=0.2415, simple_loss=0.3087, pruned_loss=0.0872, over 11675.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2945, pruned_loss=0.06339, over 3069135.95 frames. ], batch size: 248, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:11:49,330 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=170385.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:12:09,033 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9871, 4.3301, 3.3026, 2.4630, 2.9958, 2.6658, 4.7924, 3.8844], device='cuda:6'), covar=tensor([0.2716, 0.0664, 0.1648, 0.2564, 0.2878, 0.1888, 0.0360, 0.1192], device='cuda:6'), in_proj_covar=tensor([0.0320, 0.0260, 0.0297, 0.0299, 0.0288, 0.0242, 0.0285, 0.0320], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 14:12:13,319 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=170401.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:12:14,139 INFO [train.py:904] (6/8) Epoch 17, batch 8000, loss[loss=0.2178, simple_loss=0.31, pruned_loss=0.06277, over 16879.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2952, pruned_loss=0.06379, over 3073864.86 frames. ], batch size: 116, lr: 3.96e-03, grad_scale: 8.0 2023-04-30 14:12:15,840 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=170403.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:12:48,323 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4650, 5.8047, 5.5179, 5.5294, 5.2071, 5.1192, 5.1957, 5.8633], device='cuda:6'), covar=tensor([0.1260, 0.0860, 0.0958, 0.0809, 0.0896, 0.0774, 0.1099, 0.0909], device='cuda:6'), in_proj_covar=tensor([0.0625, 0.0763, 0.0626, 0.0567, 0.0479, 0.0492, 0.0632, 0.0592], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 14:12:59,221 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.99 vs. limit=5.0 2023-04-30 14:13:00,994 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=170433.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:13:11,711 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.959e+02 2.905e+02 3.477e+02 4.096e+02 7.787e+02, threshold=6.954e+02, percent-clipped=1.0 2023-04-30 14:13:15,198 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=170442.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:13:29,061 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=170451.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:13:29,848 INFO [train.py:904] (6/8) Epoch 17, batch 8050, loss[loss=0.1787, simple_loss=0.2752, pruned_loss=0.04104, over 16701.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.2956, pruned_loss=0.06382, over 3070399.95 frames. ], batch size: 124, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:13:44,242 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=170461.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:13:47,961 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7109, 2.6495, 2.4730, 4.4059, 3.2988, 4.0428, 1.5906, 2.9402], device='cuda:6'), covar=tensor([0.1341, 0.0782, 0.1251, 0.0208, 0.0340, 0.0431, 0.1637, 0.0844], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0169, 0.0190, 0.0175, 0.0204, 0.0213, 0.0195, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 14:14:28,181 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=170490.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:14:46,136 INFO [train.py:904] (6/8) Epoch 17, batch 8100, loss[loss=0.2119, simple_loss=0.2953, pruned_loss=0.06423, over 16428.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2958, pruned_loss=0.06386, over 3054944.38 frames. ], batch size: 146, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:15:16,571 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=170522.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 14:15:42,385 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.803e+02 2.570e+02 3.195e+02 3.934e+02 7.205e+02, threshold=6.390e+02, percent-clipped=3.0 2023-04-30 14:16:00,978 INFO [train.py:904] (6/8) Epoch 17, batch 8150, loss[loss=0.2128, simple_loss=0.2891, pruned_loss=0.06825, over 16678.00 frames. ], tot_loss[loss=0.2091, simple_loss=0.2929, pruned_loss=0.06263, over 3074422.89 frames. ], batch size: 134, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:16:27,643 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=170569.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:17:13,675 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-30 14:17:13,738 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 14:17:18,490 INFO [train.py:904] (6/8) Epoch 17, batch 8200, loss[loss=0.1811, simple_loss=0.2626, pruned_loss=0.0498, over 16329.00 frames. ], tot_loss[loss=0.2067, simple_loss=0.2903, pruned_loss=0.0615, over 3093175.64 frames. ], batch size: 35, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:18:05,906 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=170630.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:18:21,306 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.868e+02 2.599e+02 3.232e+02 3.801e+02 6.958e+02, threshold=6.463e+02, percent-clipped=1.0 2023-04-30 14:18:41,069 INFO [train.py:904] (6/8) Epoch 17, batch 8250, loss[loss=0.1825, simple_loss=0.269, pruned_loss=0.04802, over 12348.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2897, pruned_loss=0.05953, over 3053450.96 frames. ], batch size: 246, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:19:02,603 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5179, 3.3135, 3.5258, 1.7180, 3.6179, 3.7774, 3.0632, 2.8719], device='cuda:6'), covar=tensor([0.0713, 0.0223, 0.0192, 0.1276, 0.0100, 0.0163, 0.0353, 0.0453], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0103, 0.0091, 0.0135, 0.0073, 0.0116, 0.0122, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 14:20:04,592 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=170701.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:20:05,299 INFO [train.py:904] (6/8) Epoch 17, batch 8300, loss[loss=0.1712, simple_loss=0.2582, pruned_loss=0.04204, over 12000.00 frames. ], tot_loss[loss=0.199, simple_loss=0.286, pruned_loss=0.05599, over 3056635.07 frames. ], batch size: 247, lr: 3.95e-03, grad_scale: 4.0 2023-04-30 14:20:24,426 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1120, 3.4999, 3.7247, 2.1469, 3.1365, 2.3892, 3.5789, 3.6272], device='cuda:6'), covar=tensor([0.0255, 0.0767, 0.0489, 0.1925, 0.0773, 0.0999, 0.0615, 0.1013], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0154, 0.0160, 0.0147, 0.0139, 0.0124, 0.0138, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 14:21:09,571 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 2.327e+02 2.774e+02 3.132e+02 7.452e+02, threshold=5.547e+02, percent-clipped=1.0 2023-04-30 14:21:26,073 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=170749.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:21:29,853 INFO [train.py:904] (6/8) Epoch 17, batch 8350, loss[loss=0.2, simple_loss=0.2912, pruned_loss=0.05438, over 15327.00 frames. ], tot_loss[loss=0.1969, simple_loss=0.2856, pruned_loss=0.05408, over 3063820.40 frames. ], batch size: 190, lr: 3.95e-03, grad_scale: 4.0 2023-04-30 14:22:42,538 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-30 14:22:51,857 INFO [train.py:904] (6/8) Epoch 17, batch 8400, loss[loss=0.1853, simple_loss=0.2721, pruned_loss=0.04928, over 12235.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2829, pruned_loss=0.05224, over 3050877.61 frames. ], batch size: 246, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:23:06,034 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0061, 3.1126, 3.1158, 2.2390, 2.9157, 3.1190, 3.1322, 1.9469], device='cuda:6'), covar=tensor([0.0470, 0.0060, 0.0062, 0.0365, 0.0109, 0.0100, 0.0083, 0.0462], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0075, 0.0076, 0.0130, 0.0089, 0.0101, 0.0088, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 14:23:16,981 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=170817.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 14:23:31,669 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9636, 2.3003, 2.3323, 3.0932, 1.9291, 3.2943, 1.7724, 2.8253], device='cuda:6'), covar=tensor([0.1202, 0.0641, 0.0976, 0.0187, 0.0091, 0.0405, 0.1455, 0.0642], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0167, 0.0188, 0.0173, 0.0201, 0.0210, 0.0193, 0.0188], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 14:23:52,183 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.262e+02 2.769e+02 3.288e+02 8.219e+02, threshold=5.539e+02, percent-clipped=2.0 2023-04-30 14:24:10,747 INFO [train.py:904] (6/8) Epoch 17, batch 8450, loss[loss=0.1812, simple_loss=0.2684, pruned_loss=0.04699, over 12434.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.281, pruned_loss=0.05061, over 3048189.71 frames. ], batch size: 248, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:25:32,473 INFO [train.py:904] (6/8) Epoch 17, batch 8500, loss[loss=0.1615, simple_loss=0.2572, pruned_loss=0.03294, over 15118.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2773, pruned_loss=0.04808, over 3056911.03 frames. ], batch size: 190, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:25:57,468 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.55 vs. limit=5.0 2023-04-30 14:26:10,252 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=170925.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:26:14,632 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=170928.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:26:34,258 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.118e+02 2.662e+02 3.144e+02 4.893e+02, threshold=5.323e+02, percent-clipped=0.0 2023-04-30 14:26:56,008 INFO [train.py:904] (6/8) Epoch 17, batch 8550, loss[loss=0.1678, simple_loss=0.25, pruned_loss=0.04274, over 11825.00 frames. ], tot_loss[loss=0.1841, simple_loss=0.2748, pruned_loss=0.0467, over 3050808.90 frames. ], batch size: 247, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:28:08,419 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=170989.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:28:33,700 INFO [train.py:904] (6/8) Epoch 17, batch 8600, loss[loss=0.1752, simple_loss=0.2591, pruned_loss=0.04566, over 12303.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2747, pruned_loss=0.0454, over 3054444.78 frames. ], batch size: 248, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:29:38,151 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-30 14:29:52,152 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.454e+02 2.845e+02 3.449e+02 5.171e+02, threshold=5.691e+02, percent-clipped=0.0 2023-04-30 14:30:15,670 INFO [train.py:904] (6/8) Epoch 17, batch 8650, loss[loss=0.1556, simple_loss=0.2578, pruned_loss=0.02664, over 16827.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2725, pruned_loss=0.04368, over 3053156.54 frames. ], batch size: 102, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:31:44,047 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=171092.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:31:52,033 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0529, 3.0856, 1.8773, 3.2976, 2.3100, 3.2863, 2.0964, 2.6289], device='cuda:6'), covar=tensor([0.0273, 0.0329, 0.1508, 0.0188, 0.0805, 0.0510, 0.1381, 0.0634], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0167, 0.0188, 0.0146, 0.0169, 0.0205, 0.0196, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 14:32:02,416 INFO [train.py:904] (6/8) Epoch 17, batch 8700, loss[loss=0.1615, simple_loss=0.2499, pruned_loss=0.0366, over 12468.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2695, pruned_loss=0.0425, over 3046097.38 frames. ], batch size: 247, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:32:31,781 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=171117.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 14:33:13,274 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.121e+02 2.631e+02 3.199e+02 5.449e+02, threshold=5.263e+02, percent-clipped=0.0 2023-04-30 14:33:38,782 INFO [train.py:904] (6/8) Epoch 17, batch 8750, loss[loss=0.1739, simple_loss=0.2742, pruned_loss=0.03674, over 16764.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2697, pruned_loss=0.04227, over 3035218.94 frames. ], batch size: 76, lr: 3.95e-03, grad_scale: 4.0 2023-04-30 14:33:43,350 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=171153.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:34:00,267 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3125, 3.7892, 3.8267, 2.6361, 3.4235, 3.7908, 3.5899, 2.1844], device='cuda:6'), covar=tensor([0.0487, 0.0034, 0.0035, 0.0357, 0.0084, 0.0072, 0.0071, 0.0447], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0074, 0.0075, 0.0129, 0.0088, 0.0099, 0.0087, 0.0122], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 14:34:13,579 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=171165.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:35:21,518 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=171196.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:35:33,202 INFO [train.py:904] (6/8) Epoch 17, batch 8800, loss[loss=0.1772, simple_loss=0.274, pruned_loss=0.04022, over 16623.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2685, pruned_loss=0.04145, over 3050041.96 frames. ], batch size: 89, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:36:22,766 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=171225.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:36:54,974 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.623e+02 2.375e+02 2.916e+02 3.471e+02 5.213e+02, threshold=5.833e+02, percent-clipped=0.0 2023-04-30 14:37:18,415 INFO [train.py:904] (6/8) Epoch 17, batch 8850, loss[loss=0.1702, simple_loss=0.2791, pruned_loss=0.03067, over 16478.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2712, pruned_loss=0.04095, over 3044125.72 frames. ], batch size: 62, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:37:29,138 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=171257.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:37:40,157 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-30 14:38:03,244 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=171273.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:38:07,028 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9663, 2.6212, 2.8719, 2.0237, 2.6710, 2.0386, 2.6204, 2.7655], device='cuda:6'), covar=tensor([0.0327, 0.0997, 0.0557, 0.1894, 0.0825, 0.1020, 0.0725, 0.0971], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0151, 0.0159, 0.0145, 0.0137, 0.0124, 0.0137, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 14:38:09,070 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3705, 3.7333, 3.8620, 2.1737, 3.2613, 2.5155, 3.6901, 3.8450], device='cuda:6'), covar=tensor([0.0262, 0.0769, 0.0515, 0.1882, 0.0751, 0.0893, 0.0633, 0.0970], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0151, 0.0159, 0.0145, 0.0137, 0.0124, 0.0137, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 14:38:27,514 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=171284.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:39:01,944 INFO [train.py:904] (6/8) Epoch 17, batch 8900, loss[loss=0.1847, simple_loss=0.2794, pruned_loss=0.04495, over 15459.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2707, pruned_loss=0.0402, over 3024933.00 frames. ], batch size: 191, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:40:38,363 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 2.247e+02 2.656e+02 3.091e+02 5.659e+02, threshold=5.312e+02, percent-clipped=0.0 2023-04-30 14:40:51,395 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=171345.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:41:04,765 INFO [train.py:904] (6/8) Epoch 17, batch 8950, loss[loss=0.1669, simple_loss=0.2609, pruned_loss=0.03644, over 16970.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2699, pruned_loss=0.04022, over 3043008.11 frames. ], batch size: 116, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:42:53,004 INFO [train.py:904] (6/8) Epoch 17, batch 9000, loss[loss=0.1351, simple_loss=0.2244, pruned_loss=0.02294, over 16342.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2665, pruned_loss=0.03897, over 3039892.04 frames. ], batch size: 35, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:42:53,005 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 14:43:02,949 INFO [train.py:938] (6/8) Epoch 17, validation: loss=0.1478, simple_loss=0.2519, pruned_loss=0.02187, over 944034.00 frames. 2023-04-30 14:43:02,949 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 14:43:11,116 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=171406.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:44:23,316 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.259e+02 2.012e+02 2.627e+02 3.134e+02 6.797e+02, threshold=5.255e+02, percent-clipped=3.0 2023-04-30 14:44:41,565 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=171448.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:44:47,386 INFO [train.py:904] (6/8) Epoch 17, batch 9050, loss[loss=0.1653, simple_loss=0.2507, pruned_loss=0.03998, over 16179.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.268, pruned_loss=0.03941, over 3074944.01 frames. ], batch size: 165, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:46:24,690 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=171498.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:46:30,495 INFO [train.py:904] (6/8) Epoch 17, batch 9100, loss[loss=0.182, simple_loss=0.2737, pruned_loss=0.04522, over 16110.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2674, pruned_loss=0.04009, over 3058865.24 frames. ], batch size: 35, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:46:49,760 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-30 14:46:55,558 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-30 14:48:02,850 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 2.178e+02 2.662e+02 3.383e+02 6.301e+02, threshold=5.324e+02, percent-clipped=5.0 2023-04-30 14:48:28,381 INFO [train.py:904] (6/8) Epoch 17, batch 9150, loss[loss=0.1536, simple_loss=0.2537, pruned_loss=0.0267, over 16685.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2674, pruned_loss=0.03959, over 3058602.16 frames. ], batch size: 76, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:48:29,559 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=171552.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:48:44,328 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=171559.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:49:36,969 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=171584.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:49:56,806 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4688, 3.3527, 3.6075, 1.7930, 3.6901, 3.7927, 2.9965, 2.8803], device='cuda:6'), covar=tensor([0.0709, 0.0204, 0.0178, 0.1158, 0.0067, 0.0116, 0.0348, 0.0415], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0100, 0.0087, 0.0132, 0.0071, 0.0111, 0.0119, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 14:50:13,334 INFO [train.py:904] (6/8) Epoch 17, batch 9200, loss[loss=0.1618, simple_loss=0.2564, pruned_loss=0.03361, over 15255.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2637, pruned_loss=0.039, over 3058535.15 frames. ], batch size: 190, lr: 3.94e-03, grad_scale: 8.0 2023-04-30 14:50:31,711 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2302, 4.2607, 4.4268, 4.1649, 4.3283, 4.8353, 4.3590, 4.0166], device='cuda:6'), covar=tensor([0.1581, 0.2015, 0.1951, 0.2311, 0.2696, 0.1061, 0.1577, 0.2524], device='cuda:6'), in_proj_covar=tensor([0.0365, 0.0534, 0.0585, 0.0447, 0.0597, 0.0620, 0.0465, 0.0599], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 14:51:10,342 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=171632.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:51:27,838 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.614e+02 2.277e+02 2.610e+02 3.229e+02 7.994e+02, threshold=5.220e+02, percent-clipped=2.0 2023-04-30 14:51:50,814 INFO [train.py:904] (6/8) Epoch 17, batch 9250, loss[loss=0.1771, simple_loss=0.2699, pruned_loss=0.04214, over 16156.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2637, pruned_loss=0.03887, over 3059987.54 frames. ], batch size: 165, lr: 3.94e-03, grad_scale: 8.0 2023-04-30 14:51:54,401 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.19 vs. limit=5.0 2023-04-30 14:53:07,826 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8082, 3.7646, 3.9407, 3.6919, 3.9272, 4.3075, 3.9299, 3.6493], device='cuda:6'), covar=tensor([0.2086, 0.2338, 0.2179, 0.2470, 0.2675, 0.1503, 0.1442, 0.2313], device='cuda:6'), in_proj_covar=tensor([0.0361, 0.0526, 0.0577, 0.0441, 0.0590, 0.0613, 0.0460, 0.0590], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 14:53:42,074 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=171701.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:53:42,881 INFO [train.py:904] (6/8) Epoch 17, batch 9300, loss[loss=0.1569, simple_loss=0.2499, pruned_loss=0.03194, over 16122.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.262, pruned_loss=0.0384, over 3060720.20 frames. ], batch size: 165, lr: 3.94e-03, grad_scale: 8.0 2023-04-30 14:54:27,457 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-30 14:54:52,776 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7742, 3.7616, 3.9577, 3.7300, 3.8819, 4.3046, 3.8935, 3.5721], device='cuda:6'), covar=tensor([0.2105, 0.2218, 0.1988, 0.2750, 0.2853, 0.1450, 0.1686, 0.2818], device='cuda:6'), in_proj_covar=tensor([0.0364, 0.0532, 0.0582, 0.0446, 0.0596, 0.0619, 0.0464, 0.0594], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 14:55:09,819 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.061e+02 2.530e+02 2.950e+02 5.223e+02, threshold=5.060e+02, percent-clipped=1.0 2023-04-30 14:55:21,284 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=171748.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:55:27,769 INFO [train.py:904] (6/8) Epoch 17, batch 9350, loss[loss=0.1672, simple_loss=0.2577, pruned_loss=0.03831, over 16913.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.262, pruned_loss=0.0383, over 3059359.06 frames. ], batch size: 109, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 14:56:58,064 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=171796.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:57:07,773 INFO [train.py:904] (6/8) Epoch 17, batch 9400, loss[loss=0.1489, simple_loss=0.2384, pruned_loss=0.02968, over 12690.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2622, pruned_loss=0.03835, over 3059429.17 frames. ], batch size: 246, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 14:57:45,754 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0791, 3.2374, 3.1864, 2.2059, 2.9221, 3.2433, 3.1022, 1.9929], device='cuda:6'), covar=tensor([0.0464, 0.0044, 0.0055, 0.0362, 0.0112, 0.0074, 0.0083, 0.0429], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0074, 0.0075, 0.0130, 0.0089, 0.0099, 0.0087, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 14:58:29,671 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.619e+02 2.242e+02 2.701e+02 3.369e+02 7.708e+02, threshold=5.403e+02, percent-clipped=2.0 2023-04-30 14:58:48,378 INFO [train.py:904] (6/8) Epoch 17, batch 9450, loss[loss=0.1823, simple_loss=0.2746, pruned_loss=0.04503, over 16179.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2645, pruned_loss=0.03854, over 3070032.08 frames. ], batch size: 165, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 14:58:49,659 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=171852.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:58:52,188 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=171854.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:00:27,218 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=171900.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:00:31,023 INFO [train.py:904] (6/8) Epoch 17, batch 9500, loss[loss=0.1807, simple_loss=0.2723, pruned_loss=0.04461, over 16893.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2632, pruned_loss=0.03798, over 3069729.98 frames. ], batch size: 116, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:01:06,793 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7584, 3.1326, 3.0898, 1.7495, 2.6476, 1.8670, 3.2373, 3.3158], device='cuda:6'), covar=tensor([0.0256, 0.0805, 0.0661, 0.2470, 0.1007, 0.1413, 0.0636, 0.0915], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0149, 0.0158, 0.0146, 0.0137, 0.0123, 0.0136, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 15:01:41,248 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=171937.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:01:43,976 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0606, 2.7537, 2.8551, 2.1244, 2.6677, 2.1247, 2.6926, 2.9153], device='cuda:6'), covar=tensor([0.0339, 0.0872, 0.0557, 0.1746, 0.0770, 0.0961, 0.0641, 0.0880], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0148, 0.0158, 0.0145, 0.0136, 0.0123, 0.0136, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 15:01:51,140 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.446e+02 2.125e+02 2.584e+02 3.095e+02 6.778e+02, threshold=5.168e+02, percent-clipped=1.0 2023-04-30 15:02:14,537 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-30 15:02:14,949 INFO [train.py:904] (6/8) Epoch 17, batch 9550, loss[loss=0.1841, simple_loss=0.2695, pruned_loss=0.04933, over 12522.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2625, pruned_loss=0.03836, over 3048996.61 frames. ], batch size: 248, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:02:25,432 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5843, 3.0328, 3.2490, 1.9560, 2.8222, 2.0774, 3.1858, 3.1759], device='cuda:6'), covar=tensor([0.0308, 0.0892, 0.0571, 0.2063, 0.0823, 0.1106, 0.0691, 0.1062], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0148, 0.0158, 0.0145, 0.0136, 0.0123, 0.0136, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 15:03:20,040 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9940, 3.2344, 3.2238, 2.1539, 3.0657, 3.2793, 3.1587, 1.5609], device='cuda:6'), covar=tensor([0.0608, 0.0074, 0.0091, 0.0466, 0.0124, 0.0135, 0.0135, 0.0702], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0075, 0.0076, 0.0131, 0.0090, 0.0100, 0.0088, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 15:03:49,332 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=171998.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:03:57,414 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=172001.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:03:58,206 INFO [train.py:904] (6/8) Epoch 17, batch 9600, loss[loss=0.1843, simple_loss=0.2655, pruned_loss=0.05154, over 12681.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2643, pruned_loss=0.03887, over 3054903.09 frames. ], batch size: 246, lr: 3.94e-03, grad_scale: 8.0 2023-04-30 15:05:04,824 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 15:05:20,542 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.253e+02 2.956e+02 3.661e+02 6.195e+02, threshold=5.912e+02, percent-clipped=3.0 2023-04-30 15:05:38,438 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=172049.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:05:44,197 INFO [train.py:904] (6/8) Epoch 17, batch 9650, loss[loss=0.1716, simple_loss=0.2617, pruned_loss=0.04072, over 12256.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2656, pruned_loss=0.03906, over 3051787.55 frames. ], batch size: 246, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:06:19,369 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-30 15:06:25,197 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 15:06:35,683 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8173, 1.8530, 2.3040, 2.8021, 2.6188, 3.0710, 1.9209, 3.0619], device='cuda:6'), covar=tensor([0.0198, 0.0474, 0.0324, 0.0256, 0.0289, 0.0162, 0.0498, 0.0121], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0180, 0.0166, 0.0168, 0.0179, 0.0137, 0.0182, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:07:32,482 INFO [train.py:904] (6/8) Epoch 17, batch 9700, loss[loss=0.1673, simple_loss=0.2659, pruned_loss=0.03433, over 16747.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2648, pruned_loss=0.03889, over 3058666.00 frames. ], batch size: 83, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:07:36,947 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.22 vs. limit=5.0 2023-04-30 15:08:06,084 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=172118.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:08:11,553 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4182, 3.3439, 3.4843, 3.5418, 3.5996, 3.3211, 3.5751, 3.6433], device='cuda:6'), covar=tensor([0.1217, 0.0882, 0.1040, 0.0639, 0.0569, 0.2212, 0.0750, 0.0702], device='cuda:6'), in_proj_covar=tensor([0.0559, 0.0688, 0.0809, 0.0706, 0.0529, 0.0555, 0.0567, 0.0661], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:08:57,398 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.657e+02 2.330e+02 2.721e+02 3.165e+02 5.976e+02, threshold=5.442e+02, percent-clipped=1.0 2023-04-30 15:09:14,186 INFO [train.py:904] (6/8) Epoch 17, batch 9750, loss[loss=0.1634, simple_loss=0.2492, pruned_loss=0.03878, over 12523.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2636, pruned_loss=0.03878, over 3066581.61 frames. ], batch size: 248, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:09:18,892 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=172154.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:10:10,086 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=172179.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:10:30,422 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8912, 3.3037, 3.4841, 1.9855, 2.8954, 2.2334, 3.5104, 3.4601], device='cuda:6'), covar=tensor([0.0261, 0.0796, 0.0549, 0.2016, 0.0784, 0.0982, 0.0621, 0.1005], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0148, 0.0159, 0.0146, 0.0137, 0.0123, 0.0136, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 15:10:54,091 INFO [train.py:904] (6/8) Epoch 17, batch 9800, loss[loss=0.1916, simple_loss=0.2871, pruned_loss=0.04805, over 16918.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2639, pruned_loss=0.03813, over 3080824.93 frames. ], batch size: 116, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:10:54,783 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=172202.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:12:17,139 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.406e+02 2.047e+02 2.504e+02 3.250e+02 5.993e+02, threshold=5.009e+02, percent-clipped=1.0 2023-04-30 15:12:38,993 INFO [train.py:904] (6/8) Epoch 17, batch 9850, loss[loss=0.1588, simple_loss=0.2477, pruned_loss=0.03496, over 12379.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2651, pruned_loss=0.03793, over 3080033.48 frames. ], batch size: 248, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:14:14,441 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=172293.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:14:32,014 INFO [train.py:904] (6/8) Epoch 17, batch 9900, loss[loss=0.1804, simple_loss=0.2794, pruned_loss=0.04067, over 15294.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2654, pruned_loss=0.03802, over 3066880.06 frames. ], batch size: 192, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:16:10,690 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.496e+02 2.206e+02 2.589e+02 2.968e+02 8.308e+02, threshold=5.178e+02, percent-clipped=1.0 2023-04-30 15:16:31,646 INFO [train.py:904] (6/8) Epoch 17, batch 9950, loss[loss=0.1583, simple_loss=0.2581, pruned_loss=0.02928, over 16658.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2671, pruned_loss=0.03846, over 3054094.36 frames. ], batch size: 89, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:18:12,333 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.91 vs. limit=5.0 2023-04-30 15:18:33,054 INFO [train.py:904] (6/8) Epoch 17, batch 10000, loss[loss=0.1735, simple_loss=0.2719, pruned_loss=0.03757, over 16636.00 frames. ], tot_loss[loss=0.171, simple_loss=0.266, pruned_loss=0.03799, over 3072775.94 frames. ], batch size: 134, lr: 3.94e-03, grad_scale: 8.0 2023-04-30 15:18:47,107 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2081, 3.4199, 3.4078, 2.3897, 3.2084, 3.4171, 3.3053, 1.9737], device='cuda:6'), covar=tensor([0.0473, 0.0037, 0.0042, 0.0338, 0.0089, 0.0064, 0.0062, 0.0456], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0074, 0.0075, 0.0130, 0.0089, 0.0098, 0.0086, 0.0123], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 15:19:56,010 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.112e+02 2.435e+02 3.073e+02 5.359e+02, threshold=4.869e+02, percent-clipped=1.0 2023-04-30 15:20:13,519 INFO [train.py:904] (6/8) Epoch 17, batch 10050, loss[loss=0.1926, simple_loss=0.2868, pruned_loss=0.04917, over 16874.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2661, pruned_loss=0.03809, over 3084526.21 frames. ], batch size: 116, lr: 3.93e-03, grad_scale: 8.0 2023-04-30 15:20:58,583 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=172474.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:21:47,149 INFO [train.py:904] (6/8) Epoch 17, batch 10100, loss[loss=0.1778, simple_loss=0.2645, pruned_loss=0.04554, over 16831.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2667, pruned_loss=0.03837, over 3082618.81 frames. ], batch size: 116, lr: 3.93e-03, grad_scale: 8.0 2023-04-30 15:22:36,453 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9940, 2.2938, 2.4006, 2.9776, 2.0608, 3.3163, 1.7272, 2.8807], device='cuda:6'), covar=tensor([0.1113, 0.0595, 0.0908, 0.0145, 0.0085, 0.0335, 0.1354, 0.0541], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0163, 0.0185, 0.0165, 0.0189, 0.0204, 0.0189, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:6') 2023-04-30 15:22:57,875 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.137e+02 2.474e+02 3.171e+02 5.916e+02, threshold=4.949e+02, percent-clipped=1.0 2023-04-30 15:23:00,745 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.24 vs. limit=5.0 2023-04-30 15:23:33,082 INFO [train.py:904] (6/8) Epoch 18, batch 0, loss[loss=0.2374, simple_loss=0.3141, pruned_loss=0.08037, over 16750.00 frames. ], tot_loss[loss=0.2374, simple_loss=0.3141, pruned_loss=0.08037, over 16750.00 frames. ], batch size: 124, lr: 3.82e-03, grad_scale: 8.0 2023-04-30 15:23:33,083 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 15:23:40,343 INFO [train.py:938] (6/8) Epoch 18, validation: loss=0.148, simple_loss=0.2515, pruned_loss=0.02223, over 944034.00 frames. 2023-04-30 15:23:40,343 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 15:24:36,440 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=172593.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:24:44,247 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=172598.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:24:50,173 INFO [train.py:904] (6/8) Epoch 18, batch 50, loss[loss=0.1677, simple_loss=0.2494, pruned_loss=0.04301, over 16765.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2814, pruned_loss=0.05911, over 740920.17 frames. ], batch size: 89, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:25:21,764 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=172625.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:25:42,961 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=172641.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:25:48,599 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 2.462e+02 2.938e+02 3.654e+02 9.185e+02, threshold=5.877e+02, percent-clipped=7.0 2023-04-30 15:25:54,434 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0173, 2.2901, 2.2115, 2.7313, 1.9534, 3.2459, 1.7393, 2.6655], device='cuda:6'), covar=tensor([0.1059, 0.0602, 0.1052, 0.0157, 0.0104, 0.0413, 0.1291, 0.0708], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0165, 0.0187, 0.0168, 0.0192, 0.0207, 0.0191, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 15:25:56,033 INFO [train.py:904] (6/8) Epoch 18, batch 100, loss[loss=0.1646, simple_loss=0.2465, pruned_loss=0.04136, over 16848.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2732, pruned_loss=0.05348, over 1311165.60 frames. ], batch size: 42, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:26:05,958 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=172659.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:26:42,125 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-30 15:26:44,064 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=172686.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:26:46,400 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2792, 3.4668, 3.4077, 2.3683, 3.0019, 2.4018, 3.7159, 3.7599], device='cuda:6'), covar=tensor([0.0228, 0.0730, 0.0644, 0.1644, 0.0780, 0.0996, 0.0453, 0.0759], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0151, 0.0160, 0.0148, 0.0138, 0.0125, 0.0138, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 15:27:02,717 INFO [train.py:904] (6/8) Epoch 18, batch 150, loss[loss=0.2038, simple_loss=0.2953, pruned_loss=0.05616, over 16717.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2691, pruned_loss=0.05101, over 1759182.51 frames. ], batch size: 57, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:27:13,798 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0359, 4.0420, 3.9434, 3.6617, 3.7037, 4.0217, 3.6996, 3.8025], device='cuda:6'), covar=tensor([0.0637, 0.0618, 0.0287, 0.0257, 0.0650, 0.0455, 0.0919, 0.0581], device='cuda:6'), in_proj_covar=tensor([0.0264, 0.0372, 0.0312, 0.0299, 0.0320, 0.0348, 0.0213, 0.0369], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:27:46,406 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=172734.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:27:52,936 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-04-30 15:28:01,449 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.661e+02 2.271e+02 2.721e+02 3.424e+02 7.505e+02, threshold=5.441e+02, percent-clipped=3.0 2023-04-30 15:28:09,984 INFO [train.py:904] (6/8) Epoch 18, batch 200, loss[loss=0.1503, simple_loss=0.2296, pruned_loss=0.03555, over 16768.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2695, pruned_loss=0.0505, over 2097182.41 frames. ], batch size: 39, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:28:36,188 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1804, 3.0515, 3.2914, 2.3769, 3.0918, 3.4216, 3.1531, 2.1015], device='cuda:6'), covar=tensor([0.0497, 0.0163, 0.0060, 0.0370, 0.0111, 0.0094, 0.0095, 0.0427], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0077, 0.0077, 0.0132, 0.0091, 0.0100, 0.0088, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 15:28:41,345 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=172774.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:29:10,036 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=172795.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:29:10,429 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-30 15:29:18,639 INFO [train.py:904] (6/8) Epoch 18, batch 250, loss[loss=0.1782, simple_loss=0.2699, pruned_loss=0.0432, over 16713.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2675, pruned_loss=0.05008, over 2360364.40 frames. ], batch size: 62, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:29:29,848 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4410, 5.3657, 5.1909, 4.7690, 5.2568, 2.1862, 4.9927, 5.1899], device='cuda:6'), covar=tensor([0.0077, 0.0084, 0.0175, 0.0310, 0.0081, 0.2282, 0.0124, 0.0155], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0138, 0.0181, 0.0164, 0.0157, 0.0196, 0.0170, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:29:47,728 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=172822.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:30:18,760 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=172845.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:30:19,548 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.240e+02 2.726e+02 3.152e+02 7.045e+02, threshold=5.451e+02, percent-clipped=2.0 2023-04-30 15:30:23,658 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3289, 3.5293, 3.9400, 2.2936, 3.2660, 2.4565, 3.7656, 3.7659], device='cuda:6'), covar=tensor([0.0292, 0.0850, 0.0454, 0.1869, 0.0762, 0.0917, 0.0606, 0.1036], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0154, 0.0163, 0.0150, 0.0141, 0.0127, 0.0140, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 15:30:28,991 INFO [train.py:904] (6/8) Epoch 18, batch 300, loss[loss=0.1665, simple_loss=0.2608, pruned_loss=0.03613, over 17098.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2643, pruned_loss=0.04784, over 2556289.38 frames. ], batch size: 49, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:31:32,733 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0233, 2.1096, 2.5884, 2.9838, 2.8077, 3.4329, 2.3495, 3.3397], device='cuda:6'), covar=tensor([0.0229, 0.0446, 0.0310, 0.0278, 0.0304, 0.0172, 0.0460, 0.0170], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0185, 0.0170, 0.0172, 0.0182, 0.0141, 0.0186, 0.0135], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:31:39,798 INFO [train.py:904] (6/8) Epoch 18, batch 350, loss[loss=0.1781, simple_loss=0.2574, pruned_loss=0.04937, over 12258.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2612, pruned_loss=0.04596, over 2728382.77 frames. ], batch size: 246, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:31:45,358 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=172906.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:31:50,189 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8320, 2.8565, 2.5374, 2.7172, 3.2118, 2.9661, 3.5586, 3.3930], device='cuda:6'), covar=tensor([0.0116, 0.0364, 0.0450, 0.0407, 0.0260, 0.0339, 0.0196, 0.0225], device='cuda:6'), in_proj_covar=tensor([0.0189, 0.0229, 0.0220, 0.0222, 0.0228, 0.0227, 0.0228, 0.0219], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:31:55,008 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0247, 2.0353, 2.5376, 2.9145, 2.7854, 3.3707, 2.2963, 3.3044], device='cuda:6'), covar=tensor([0.0225, 0.0474, 0.0303, 0.0277, 0.0304, 0.0190, 0.0458, 0.0182], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0185, 0.0170, 0.0173, 0.0183, 0.0141, 0.0186, 0.0135], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:32:26,204 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7953, 2.7867, 2.3936, 2.6909, 3.1207, 2.9233, 3.4959, 3.2536], device='cuda:6'), covar=tensor([0.0115, 0.0342, 0.0454, 0.0364, 0.0245, 0.0319, 0.0212, 0.0236], device='cuda:6'), in_proj_covar=tensor([0.0189, 0.0229, 0.0220, 0.0222, 0.0228, 0.0227, 0.0228, 0.0219], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:32:40,483 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.138e+02 2.555e+02 2.961e+02 4.551e+02, threshold=5.109e+02, percent-clipped=0.0 2023-04-30 15:32:47,897 INFO [train.py:904] (6/8) Epoch 18, batch 400, loss[loss=0.2096, simple_loss=0.2781, pruned_loss=0.0705, over 12608.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2601, pruned_loss=0.0456, over 2861148.59 frames. ], batch size: 247, lr: 3.82e-03, grad_scale: 2.0 2023-04-30 15:32:51,729 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=172954.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:33:23,273 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-30 15:33:30,476 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=172981.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:33:59,531 INFO [train.py:904] (6/8) Epoch 18, batch 450, loss[loss=0.1526, simple_loss=0.2403, pruned_loss=0.03247, over 16633.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2584, pruned_loss=0.04491, over 2967704.73 frames. ], batch size: 62, lr: 3.82e-03, grad_scale: 2.0 2023-04-30 15:35:00,082 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.073e+02 2.463e+02 2.995e+02 6.633e+02, threshold=4.925e+02, percent-clipped=1.0 2023-04-30 15:35:08,315 INFO [train.py:904] (6/8) Epoch 18, batch 500, loss[loss=0.1617, simple_loss=0.2495, pruned_loss=0.03696, over 16373.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2571, pruned_loss=0.04429, over 3054338.24 frames. ], batch size: 75, lr: 3.82e-03, grad_scale: 2.0 2023-04-30 15:35:44,506 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-30 15:35:59,652 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=173090.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:36:14,708 INFO [train.py:904] (6/8) Epoch 18, batch 550, loss[loss=0.1551, simple_loss=0.2506, pruned_loss=0.02982, over 17040.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2551, pruned_loss=0.04295, over 3120232.96 frames. ], batch size: 50, lr: 3.82e-03, grad_scale: 2.0 2023-04-30 15:37:14,571 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.114e+02 2.403e+02 3.161e+02 6.470e+02, threshold=4.805e+02, percent-clipped=1.0 2023-04-30 15:37:20,898 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=173150.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:37:22,686 INFO [train.py:904] (6/8) Epoch 18, batch 600, loss[loss=0.1744, simple_loss=0.2447, pruned_loss=0.05206, over 16730.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2544, pruned_loss=0.04303, over 3156963.33 frames. ], batch size: 124, lr: 3.81e-03, grad_scale: 2.0 2023-04-30 15:37:37,218 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-30 15:38:30,576 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=173201.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:38:31,376 INFO [train.py:904] (6/8) Epoch 18, batch 650, loss[loss=0.1726, simple_loss=0.246, pruned_loss=0.04956, over 16847.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2531, pruned_loss=0.04239, over 3195915.23 frames. ], batch size: 96, lr: 3.81e-03, grad_scale: 2.0 2023-04-30 15:38:45,553 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=173211.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:39:32,445 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.711e+02 2.266e+02 2.757e+02 3.739e+02 8.562e+02, threshold=5.513e+02, percent-clipped=7.0 2023-04-30 15:39:40,833 INFO [train.py:904] (6/8) Epoch 18, batch 700, loss[loss=0.1679, simple_loss=0.2605, pruned_loss=0.03763, over 17145.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2542, pruned_loss=0.04224, over 3228279.38 frames. ], batch size: 47, lr: 3.81e-03, grad_scale: 2.0 2023-04-30 15:39:44,177 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=173254.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:40:22,721 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=173281.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:40:46,896 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8428, 4.0098, 2.5841, 4.5337, 3.1130, 4.5111, 2.6933, 3.2843], device='cuda:6'), covar=tensor([0.0292, 0.0329, 0.1531, 0.0286, 0.0822, 0.0484, 0.1443, 0.0699], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0173, 0.0195, 0.0155, 0.0174, 0.0213, 0.0203, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 15:40:50,345 INFO [train.py:904] (6/8) Epoch 18, batch 750, loss[loss=0.1616, simple_loss=0.2594, pruned_loss=0.03196, over 17048.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2546, pruned_loss=0.04224, over 3254045.94 frames. ], batch size: 50, lr: 3.81e-03, grad_scale: 2.0 2023-04-30 15:40:51,630 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=173302.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:41:00,059 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=173309.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:41:26,735 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=173329.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:41:50,586 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.162e+02 2.529e+02 2.992e+02 5.194e+02, threshold=5.057e+02, percent-clipped=0.0 2023-04-30 15:41:58,280 INFO [train.py:904] (6/8) Epoch 18, batch 800, loss[loss=0.1539, simple_loss=0.2386, pruned_loss=0.03464, over 16849.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2539, pruned_loss=0.04197, over 3269512.65 frames. ], batch size: 42, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:42:22,663 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=173370.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:42:30,225 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8992, 3.0415, 2.7292, 2.9677, 3.3105, 3.0762, 3.6186, 3.4829], device='cuda:6'), covar=tensor([0.0117, 0.0345, 0.0416, 0.0368, 0.0228, 0.0332, 0.0214, 0.0210], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0231, 0.0221, 0.0223, 0.0231, 0.0230, 0.0231, 0.0221], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:42:43,154 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2023-04-30 15:42:49,280 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=173390.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:43:05,169 INFO [train.py:904] (6/8) Epoch 18, batch 850, loss[loss=0.1689, simple_loss=0.2468, pruned_loss=0.04552, over 16712.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2529, pruned_loss=0.04164, over 3281081.09 frames. ], batch size: 76, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:43:55,074 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=173438.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:44:03,719 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4454, 2.3669, 2.3754, 4.1461, 2.3565, 2.6756, 2.4326, 2.4935], device='cuda:6'), covar=tensor([0.1254, 0.3456, 0.2875, 0.0592, 0.4008, 0.2612, 0.3391, 0.3590], device='cuda:6'), in_proj_covar=tensor([0.0386, 0.0426, 0.0355, 0.0322, 0.0427, 0.0490, 0.0396, 0.0498], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:44:07,463 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.076e+02 2.484e+02 3.047e+02 7.232e+02, threshold=4.969e+02, percent-clipped=3.0 2023-04-30 15:44:15,656 INFO [train.py:904] (6/8) Epoch 18, batch 900, loss[loss=0.1896, simple_loss=0.2637, pruned_loss=0.05773, over 16771.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2527, pruned_loss=0.04123, over 3287840.59 frames. ], batch size: 124, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:44:40,167 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7843, 2.9813, 2.7492, 5.1109, 4.2061, 4.4904, 1.7096, 3.2408], device='cuda:6'), covar=tensor([0.1342, 0.0719, 0.1158, 0.0196, 0.0247, 0.0406, 0.1583, 0.0734], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0167, 0.0189, 0.0175, 0.0198, 0.0213, 0.0194, 0.0189], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 15:45:11,337 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=173491.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 15:45:24,865 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=173501.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:45:25,648 INFO [train.py:904] (6/8) Epoch 18, batch 950, loss[loss=0.1896, simple_loss=0.2783, pruned_loss=0.05048, over 17060.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2528, pruned_loss=0.04185, over 3304524.11 frames. ], batch size: 53, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:45:31,768 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=173506.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:45:58,317 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6423, 2.5322, 2.0492, 2.4363, 2.9234, 2.7675, 3.1809, 3.1934], device='cuda:6'), covar=tensor([0.0128, 0.0471, 0.0592, 0.0492, 0.0299, 0.0366, 0.0295, 0.0259], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0234, 0.0224, 0.0225, 0.0234, 0.0232, 0.0233, 0.0224], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:46:26,060 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.297e+02 2.589e+02 3.111e+02 6.149e+02, threshold=5.178e+02, percent-clipped=2.0 2023-04-30 15:46:31,946 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=173549.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:46:35,237 INFO [train.py:904] (6/8) Epoch 18, batch 1000, loss[loss=0.1763, simple_loss=0.2676, pruned_loss=0.04249, over 16994.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2524, pruned_loss=0.04198, over 3301331.06 frames. ], batch size: 55, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:46:35,685 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=173552.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 15:47:44,517 INFO [train.py:904] (6/8) Epoch 18, batch 1050, loss[loss=0.1619, simple_loss=0.2581, pruned_loss=0.03288, over 17287.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2515, pruned_loss=0.042, over 3301021.39 frames. ], batch size: 52, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:48:02,478 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=173614.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:48:32,918 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.67 vs. limit=5.0 2023-04-30 15:48:46,277 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.302e+02 2.654e+02 3.076e+02 5.985e+02, threshold=5.308e+02, percent-clipped=1.0 2023-04-30 15:48:54,975 INFO [train.py:904] (6/8) Epoch 18, batch 1100, loss[loss=0.1582, simple_loss=0.2418, pruned_loss=0.03733, over 16542.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2505, pruned_loss=0.04145, over 3294235.49 frames. ], batch size: 68, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:49:12,499 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=173665.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:49:26,659 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=173675.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 15:49:37,002 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-30 15:50:01,540 INFO [train.py:904] (6/8) Epoch 18, batch 1150, loss[loss=0.1553, simple_loss=0.246, pruned_loss=0.03235, over 17177.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2501, pruned_loss=0.0408, over 3307460.91 frames. ], batch size: 46, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:50:15,896 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-30 15:51:02,212 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 2.041e+02 2.391e+02 2.834e+02 4.820e+02, threshold=4.781e+02, percent-clipped=0.0 2023-04-30 15:51:11,451 INFO [train.py:904] (6/8) Epoch 18, batch 1200, loss[loss=0.1515, simple_loss=0.2339, pruned_loss=0.03452, over 16726.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2494, pruned_loss=0.04025, over 3305134.99 frames. ], batch size: 39, lr: 3.81e-03, grad_scale: 8.0 2023-04-30 15:51:26,961 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8555, 5.2331, 4.9949, 4.9583, 4.6508, 4.6743, 4.7138, 5.3246], device='cuda:6'), covar=tensor([0.1401, 0.0974, 0.1082, 0.1015, 0.0925, 0.1064, 0.1169, 0.0984], device='cuda:6'), in_proj_covar=tensor([0.0653, 0.0799, 0.0646, 0.0593, 0.0504, 0.0505, 0.0663, 0.0616], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:52:19,913 INFO [train.py:904] (6/8) Epoch 18, batch 1250, loss[loss=0.1822, simple_loss=0.2848, pruned_loss=0.03985, over 17110.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2498, pruned_loss=0.04077, over 3306522.20 frames. ], batch size: 48, lr: 3.81e-03, grad_scale: 8.0 2023-04-30 15:52:24,838 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=173806.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:52:42,394 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8870, 3.7717, 3.9482, 4.0438, 4.0984, 3.6870, 3.9654, 4.1187], device='cuda:6'), covar=tensor([0.1297, 0.1023, 0.1106, 0.0600, 0.0538, 0.1884, 0.1631, 0.0654], device='cuda:6'), in_proj_covar=tensor([0.0622, 0.0767, 0.0903, 0.0777, 0.0580, 0.0615, 0.0628, 0.0733], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 15:53:21,075 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.313e+02 2.602e+02 3.076e+02 5.284e+02, threshold=5.204e+02, percent-clipped=3.0 2023-04-30 15:53:22,611 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=173847.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 15:53:22,793 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9644, 3.9860, 4.4622, 2.3240, 4.6921, 4.6864, 3.5225, 3.7916], device='cuda:6'), covar=tensor([0.0781, 0.0229, 0.0213, 0.1130, 0.0068, 0.0160, 0.0370, 0.0352], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0107, 0.0094, 0.0140, 0.0076, 0.0121, 0.0126, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 15:53:28,943 INFO [train.py:904] (6/8) Epoch 18, batch 1300, loss[loss=0.1647, simple_loss=0.2617, pruned_loss=0.03384, over 16683.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2501, pruned_loss=0.04081, over 3311207.88 frames. ], batch size: 57, lr: 3.81e-03, grad_scale: 8.0 2023-04-30 15:53:32,735 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=173854.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:53:39,609 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1168, 4.1073, 4.4857, 4.4738, 4.5055, 4.2106, 4.2282, 4.1362], device='cuda:6'), covar=tensor([0.0401, 0.0768, 0.0457, 0.0429, 0.0471, 0.0461, 0.0818, 0.0628], device='cuda:6'), in_proj_covar=tensor([0.0396, 0.0430, 0.0420, 0.0394, 0.0464, 0.0441, 0.0535, 0.0349], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 15:54:08,203 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 15:54:37,697 INFO [train.py:904] (6/8) Epoch 18, batch 1350, loss[loss=0.1623, simple_loss=0.256, pruned_loss=0.03426, over 17135.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2508, pruned_loss=0.0412, over 3315017.80 frames. ], batch size: 48, lr: 3.81e-03, grad_scale: 8.0 2023-04-30 15:54:39,534 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-30 15:55:37,173 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 2.229e+02 2.739e+02 3.130e+02 5.090e+02, threshold=5.477e+02, percent-clipped=0.0 2023-04-30 15:55:44,243 INFO [train.py:904] (6/8) Epoch 18, batch 1400, loss[loss=0.1774, simple_loss=0.2631, pruned_loss=0.04586, over 16717.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2511, pruned_loss=0.04109, over 3321427.41 frames. ], batch size: 57, lr: 3.81e-03, grad_scale: 8.0 2023-04-30 15:56:04,294 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=173965.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:56:09,764 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=173970.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 15:56:37,874 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-30 15:56:56,560 INFO [train.py:904] (6/8) Epoch 18, batch 1450, loss[loss=0.172, simple_loss=0.2697, pruned_loss=0.03715, over 17046.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2498, pruned_loss=0.04119, over 3317272.49 frames. ], batch size: 55, lr: 3.81e-03, grad_scale: 8.0 2023-04-30 15:57:11,462 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174013.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:57:14,682 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174015.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:57:16,920 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174017.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:57:54,753 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 2.047e+02 2.425e+02 3.116e+02 5.904e+02, threshold=4.850e+02, percent-clipped=1.0 2023-04-30 15:58:03,185 INFO [train.py:904] (6/8) Epoch 18, batch 1500, loss[loss=0.1435, simple_loss=0.2392, pruned_loss=0.02386, over 17243.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2503, pruned_loss=0.04229, over 3311913.24 frames. ], batch size: 45, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 15:58:36,091 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174076.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 15:58:38,883 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174078.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:58:51,801 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6852, 4.7682, 4.8852, 4.7451, 4.6741, 5.3166, 4.8814, 4.5878], device='cuda:6'), covar=tensor([0.1440, 0.1933, 0.2209, 0.2027, 0.2889, 0.1145, 0.1592, 0.2678], device='cuda:6'), in_proj_covar=tensor([0.0395, 0.0577, 0.0632, 0.0477, 0.0649, 0.0661, 0.0497, 0.0649], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 15:59:10,942 INFO [train.py:904] (6/8) Epoch 18, batch 1550, loss[loss=0.1353, simple_loss=0.2205, pruned_loss=0.02501, over 16835.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2518, pruned_loss=0.04371, over 3314580.56 frames. ], batch size: 42, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:00:00,502 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174139.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:00:09,837 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.219e+02 2.746e+02 3.375e+02 7.585e+02, threshold=5.492e+02, percent-clipped=5.0 2023-04-30 16:00:11,922 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174147.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 16:00:17,417 INFO [train.py:904] (6/8) Epoch 18, batch 1600, loss[loss=0.1575, simple_loss=0.2442, pruned_loss=0.03545, over 16831.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2537, pruned_loss=0.0444, over 3318832.54 frames. ], batch size: 42, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:01:16,801 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174195.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 16:01:23,754 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174200.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:01:25,623 INFO [train.py:904] (6/8) Epoch 18, batch 1650, loss[loss=0.1576, simple_loss=0.2434, pruned_loss=0.03596, over 17008.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2553, pruned_loss=0.04464, over 3313773.19 frames. ], batch size: 41, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:01:55,660 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9338, 4.6900, 5.0053, 5.1851, 5.4338, 4.7704, 5.3704, 5.3753], device='cuda:6'), covar=tensor([0.1870, 0.1397, 0.1829, 0.0803, 0.0557, 0.0930, 0.0573, 0.0555], device='cuda:6'), in_proj_covar=tensor([0.0620, 0.0767, 0.0903, 0.0778, 0.0577, 0.0614, 0.0628, 0.0735], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:02:03,966 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6751, 4.8762, 5.2722, 5.1978, 5.2321, 4.8987, 4.7227, 4.6517], device='cuda:6'), covar=tensor([0.0493, 0.0660, 0.0433, 0.0543, 0.0657, 0.0491, 0.1342, 0.0562], device='cuda:6'), in_proj_covar=tensor([0.0400, 0.0436, 0.0424, 0.0399, 0.0473, 0.0448, 0.0541, 0.0355], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 16:02:23,838 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.798e+02 2.550e+02 2.934e+02 3.695e+02 6.371e+02, threshold=5.868e+02, percent-clipped=1.0 2023-04-30 16:02:32,758 INFO [train.py:904] (6/8) Epoch 18, batch 1700, loss[loss=0.2081, simple_loss=0.29, pruned_loss=0.06311, over 15433.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2575, pruned_loss=0.04519, over 3323247.26 frames. ], batch size: 191, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:02:37,425 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2963, 2.2388, 2.3256, 4.0917, 2.1937, 2.6303, 2.2344, 2.3745], device='cuda:6'), covar=tensor([0.1286, 0.3643, 0.2808, 0.0509, 0.3919, 0.2419, 0.3651, 0.3131], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0430, 0.0359, 0.0327, 0.0431, 0.0495, 0.0400, 0.0504], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:02:37,746 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 16:02:57,247 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174270.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 16:03:06,449 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-30 16:03:15,644 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174283.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:03:25,058 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174291.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:03:40,505 INFO [train.py:904] (6/8) Epoch 18, batch 1750, loss[loss=0.1565, simple_loss=0.2397, pruned_loss=0.03662, over 16852.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.258, pruned_loss=0.0447, over 3325447.69 frames. ], batch size: 42, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:03:59,115 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9444, 2.0151, 2.5580, 3.0053, 2.7231, 3.3713, 2.3901, 3.4091], device='cuda:6'), covar=tensor([0.0209, 0.0448, 0.0282, 0.0246, 0.0271, 0.0168, 0.0422, 0.0122], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0189, 0.0175, 0.0179, 0.0187, 0.0147, 0.0192, 0.0140], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:04:01,308 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174318.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:04:36,728 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174344.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:04:38,536 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.659e+02 2.402e+02 2.710e+02 3.242e+02 7.254e+02, threshold=5.419e+02, percent-clipped=1.0 2023-04-30 16:04:41,349 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8885, 2.0368, 2.4757, 3.0232, 2.7301, 3.4026, 2.4160, 3.4171], device='cuda:6'), covar=tensor([0.0251, 0.0492, 0.0332, 0.0273, 0.0305, 0.0176, 0.0436, 0.0152], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0189, 0.0175, 0.0179, 0.0187, 0.0147, 0.0192, 0.0140], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:04:46,157 INFO [train.py:904] (6/8) Epoch 18, batch 1800, loss[loss=0.1509, simple_loss=0.2384, pruned_loss=0.03165, over 17218.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2596, pruned_loss=0.04496, over 3317338.38 frames. ], batch size: 44, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:04:46,568 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174352.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 16:05:10,945 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=174371.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 16:05:13,916 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=174373.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:05:33,438 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174388.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:05:51,520 INFO [train.py:904] (6/8) Epoch 18, batch 1850, loss[loss=0.1788, simple_loss=0.2623, pruned_loss=0.04764, over 16159.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2596, pruned_loss=0.04442, over 3323290.88 frames. ], batch size: 165, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:06:51,029 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.281e+02 2.184e+02 2.475e+02 2.961e+02 4.707e+02, threshold=4.951e+02, percent-clipped=0.0 2023-04-30 16:06:55,609 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174449.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:06:58,395 INFO [train.py:904] (6/8) Epoch 18, batch 1900, loss[loss=0.1681, simple_loss=0.263, pruned_loss=0.0366, over 17129.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2584, pruned_loss=0.04404, over 3314446.67 frames. ], batch size: 48, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:07:12,746 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-30 16:07:16,497 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1856, 5.7809, 5.9186, 5.5640, 5.6493, 6.2530, 5.7946, 5.4853], device='cuda:6'), covar=tensor([0.0977, 0.1800, 0.2251, 0.2076, 0.2769, 0.0980, 0.1377, 0.2340], device='cuda:6'), in_proj_covar=tensor([0.0398, 0.0581, 0.0638, 0.0481, 0.0654, 0.0665, 0.0502, 0.0652], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 16:07:57,453 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=174495.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:08:06,705 INFO [train.py:904] (6/8) Epoch 18, batch 1950, loss[loss=0.1742, simple_loss=0.2655, pruned_loss=0.04142, over 16557.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2577, pruned_loss=0.04328, over 3310909.35 frames. ], batch size: 68, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:09:05,466 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.435e+02 2.251e+02 2.513e+02 3.165e+02 5.916e+02, threshold=5.027e+02, percent-clipped=3.0 2023-04-30 16:09:14,567 INFO [train.py:904] (6/8) Epoch 18, batch 2000, loss[loss=0.1609, simple_loss=0.2606, pruned_loss=0.03062, over 17136.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2569, pruned_loss=0.04308, over 3319280.64 frames. ], batch size: 47, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:09:24,285 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7818, 2.6072, 2.6391, 1.9059, 2.5647, 2.6570, 2.5635, 1.8583], device='cuda:6'), covar=tensor([0.0445, 0.0114, 0.0079, 0.0365, 0.0123, 0.0128, 0.0123, 0.0377], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0080, 0.0080, 0.0134, 0.0094, 0.0105, 0.0092, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 16:09:46,916 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-30 16:09:59,038 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6393, 2.9570, 3.0229, 2.0762, 2.6619, 2.1514, 3.2025, 3.2773], device='cuda:6'), covar=tensor([0.0267, 0.0857, 0.0630, 0.1862, 0.0922, 0.1053, 0.0606, 0.0879], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0158, 0.0164, 0.0151, 0.0142, 0.0127, 0.0141, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 16:10:23,407 INFO [train.py:904] (6/8) Epoch 18, batch 2050, loss[loss=0.1781, simple_loss=0.2556, pruned_loss=0.05035, over 16875.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2578, pruned_loss=0.04321, over 3327336.24 frames. ], batch size: 109, lr: 3.80e-03, grad_scale: 16.0 2023-04-30 16:10:30,720 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-30 16:10:35,630 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8447, 1.8967, 2.4167, 2.7691, 2.6290, 3.2420, 2.2391, 3.2018], device='cuda:6'), covar=tensor([0.0236, 0.0529, 0.0316, 0.0342, 0.0327, 0.0185, 0.0476, 0.0166], device='cuda:6'), in_proj_covar=tensor([0.0182, 0.0190, 0.0176, 0.0181, 0.0188, 0.0148, 0.0193, 0.0141], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:10:59,574 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2177, 5.9138, 6.0245, 5.6690, 5.7952, 6.3570, 5.8722, 5.5292], device='cuda:6'), covar=tensor([0.0767, 0.1529, 0.1947, 0.1894, 0.2624, 0.0848, 0.1351, 0.2244], device='cuda:6'), in_proj_covar=tensor([0.0396, 0.0581, 0.0638, 0.0481, 0.0654, 0.0664, 0.0500, 0.0652], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 16:11:00,780 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7334, 4.8971, 5.2119, 5.1672, 5.2185, 4.8980, 4.8393, 4.6340], device='cuda:6'), covar=tensor([0.0359, 0.0497, 0.0395, 0.0446, 0.0449, 0.0382, 0.0839, 0.0498], device='cuda:6'), in_proj_covar=tensor([0.0398, 0.0433, 0.0420, 0.0394, 0.0467, 0.0443, 0.0537, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 16:11:17,527 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=174639.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:11:26,172 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-30 16:11:26,649 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.708e+02 2.224e+02 2.527e+02 3.035e+02 6.825e+02, threshold=5.053e+02, percent-clipped=2.0 2023-04-30 16:11:28,696 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=174647.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 16:11:35,057 INFO [train.py:904] (6/8) Epoch 18, batch 2100, loss[loss=0.1855, simple_loss=0.2839, pruned_loss=0.04359, over 17160.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2589, pruned_loss=0.0442, over 3318316.20 frames. ], batch size: 48, lr: 3.80e-03, grad_scale: 16.0 2023-04-30 16:12:02,035 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174671.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 16:12:06,185 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174673.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:12:45,802 INFO [train.py:904] (6/8) Epoch 18, batch 2150, loss[loss=0.1836, simple_loss=0.2713, pruned_loss=0.04799, over 15408.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2594, pruned_loss=0.04433, over 3314540.91 frames. ], batch size: 190, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:12:50,357 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174705.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:13:10,642 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174719.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:13:13,523 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174721.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:13:45,452 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=174744.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:13:45,546 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9903, 3.8931, 4.0755, 4.2057, 4.2565, 3.8461, 4.0558, 4.2592], device='cuda:6'), covar=tensor([0.1579, 0.1056, 0.1161, 0.0596, 0.0594, 0.1479, 0.2255, 0.0678], device='cuda:6'), in_proj_covar=tensor([0.0625, 0.0772, 0.0913, 0.0788, 0.0583, 0.0622, 0.0637, 0.0743], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:13:49,518 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.235e+02 2.673e+02 3.216e+02 6.695e+02, threshold=5.347e+02, percent-clipped=2.0 2023-04-30 16:13:56,940 INFO [train.py:904] (6/8) Epoch 18, batch 2200, loss[loss=0.1816, simple_loss=0.2725, pruned_loss=0.0453, over 16710.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2599, pruned_loss=0.04403, over 3318519.87 frames. ], batch size: 62, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:14:16,369 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174766.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:14:56,044 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174795.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:15:06,365 INFO [train.py:904] (6/8) Epoch 18, batch 2250, loss[loss=0.1825, simple_loss=0.2723, pruned_loss=0.04634, over 17113.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2612, pruned_loss=0.04431, over 3323396.47 frames. ], batch size: 49, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:15:23,853 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7486, 1.7460, 1.5165, 1.4385, 1.8837, 1.5844, 1.5939, 1.8752], device='cuda:6'), covar=tensor([0.0189, 0.0294, 0.0430, 0.0381, 0.0202, 0.0272, 0.0186, 0.0211], device='cuda:6'), in_proj_covar=tensor([0.0199, 0.0233, 0.0222, 0.0223, 0.0233, 0.0231, 0.0235, 0.0225], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:16:02,779 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174843.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:16:07,689 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.616e+02 2.164e+02 2.481e+02 3.118e+02 6.907e+02, threshold=4.963e+02, percent-clipped=2.0 2023-04-30 16:16:14,116 INFO [train.py:904] (6/8) Epoch 18, batch 2300, loss[loss=0.1862, simple_loss=0.2775, pruned_loss=0.04746, over 17068.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.2624, pruned_loss=0.04505, over 3311836.84 frames. ], batch size: 50, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:17:22,134 INFO [train.py:904] (6/8) Epoch 18, batch 2350, loss[loss=0.171, simple_loss=0.2507, pruned_loss=0.04565, over 16389.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.261, pruned_loss=0.04493, over 3318933.96 frames. ], batch size: 146, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:18:14,659 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174939.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:18:23,315 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-30 16:18:25,282 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 2.298e+02 2.659e+02 3.331e+02 6.396e+02, threshold=5.319e+02, percent-clipped=2.0 2023-04-30 16:18:25,679 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174947.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 16:18:32,029 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5182, 4.4386, 4.4155, 3.3948, 4.4822, 1.7815, 4.1120, 4.0821], device='cuda:6'), covar=tensor([0.0185, 0.0167, 0.0283, 0.0676, 0.0164, 0.3342, 0.0259, 0.0394], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0147, 0.0194, 0.0175, 0.0169, 0.0204, 0.0183, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:18:32,764 INFO [train.py:904] (6/8) Epoch 18, batch 2400, loss[loss=0.1854, simple_loss=0.2705, pruned_loss=0.05011, over 16711.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2617, pruned_loss=0.04511, over 3322570.84 frames. ], batch size: 83, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:19:21,462 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174987.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:19:33,419 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174995.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:19:42,218 INFO [train.py:904] (6/8) Epoch 18, batch 2450, loss[loss=0.2089, simple_loss=0.2928, pruned_loss=0.06253, over 16288.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2624, pruned_loss=0.04457, over 3327520.01 frames. ], batch size: 165, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:20:12,767 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9538, 1.9950, 2.6155, 3.0075, 2.7735, 3.4830, 2.3772, 3.4869], device='cuda:6'), covar=tensor([0.0246, 0.0510, 0.0298, 0.0288, 0.0312, 0.0178, 0.0461, 0.0143], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0191, 0.0177, 0.0182, 0.0190, 0.0149, 0.0194, 0.0143], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:20:24,378 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5329, 3.6067, 3.2441, 2.9307, 3.1597, 3.4722, 3.3009, 3.2663], device='cuda:6'), covar=tensor([0.0629, 0.0621, 0.0283, 0.0258, 0.0512, 0.0442, 0.1177, 0.0490], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0413, 0.0341, 0.0333, 0.0355, 0.0385, 0.0236, 0.0411], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 16:20:40,526 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=175044.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:20:43,743 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.297e+02 2.808e+02 3.204e+02 6.791e+02, threshold=5.617e+02, percent-clipped=2.0 2023-04-30 16:20:51,500 INFO [train.py:904] (6/8) Epoch 18, batch 2500, loss[loss=0.164, simple_loss=0.2601, pruned_loss=0.03398, over 17138.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2617, pruned_loss=0.04379, over 3330052.23 frames. ], batch size: 49, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:21:04,688 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=175061.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:21:46,594 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=175092.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:22:00,355 INFO [train.py:904] (6/8) Epoch 18, batch 2550, loss[loss=0.1594, simple_loss=0.2551, pruned_loss=0.03183, over 17227.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2622, pruned_loss=0.04424, over 3320653.47 frames. ], batch size: 46, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:22:30,180 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3653, 2.2154, 2.4299, 4.1632, 2.2357, 2.6609, 2.2762, 2.4584], device='cuda:6'), covar=tensor([0.1388, 0.3740, 0.2710, 0.0539, 0.3957, 0.2560, 0.3637, 0.2919], device='cuda:6'), in_proj_covar=tensor([0.0393, 0.0432, 0.0361, 0.0327, 0.0432, 0.0500, 0.0401, 0.0505], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:22:34,744 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 16:22:52,020 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175139.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:23:03,050 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.483e+02 2.226e+02 2.530e+02 3.220e+02 6.959e+02, threshold=5.059e+02, percent-clipped=1.0 2023-04-30 16:23:09,311 INFO [train.py:904] (6/8) Epoch 18, batch 2600, loss[loss=0.2074, simple_loss=0.2889, pruned_loss=0.06299, over 12256.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2607, pruned_loss=0.04392, over 3318098.22 frames. ], batch size: 246, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:24:17,736 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175200.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:24:19,525 INFO [train.py:904] (6/8) Epoch 18, batch 2650, loss[loss=0.1918, simple_loss=0.2746, pruned_loss=0.05448, over 16881.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2619, pruned_loss=0.04425, over 3314502.04 frames. ], batch size: 116, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:24:27,548 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175207.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:25:22,161 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.085e+02 2.405e+02 2.864e+02 4.837e+02, threshold=4.811e+02, percent-clipped=0.0 2023-04-30 16:25:27,540 INFO [train.py:904] (6/8) Epoch 18, batch 2700, loss[loss=0.1794, simple_loss=0.2756, pruned_loss=0.04165, over 17124.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.262, pruned_loss=0.04367, over 3328738.20 frames. ], batch size: 47, lr: 3.79e-03, grad_scale: 4.0 2023-04-30 16:25:50,121 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175268.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:26:36,863 INFO [train.py:904] (6/8) Epoch 18, batch 2750, loss[loss=0.1527, simple_loss=0.2457, pruned_loss=0.02987, over 16995.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2617, pruned_loss=0.04317, over 3329864.65 frames. ], batch size: 41, lr: 3.79e-03, grad_scale: 4.0 2023-04-30 16:27:40,024 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.507e+02 2.063e+02 2.386e+02 2.912e+02 4.096e+02, threshold=4.773e+02, percent-clipped=0.0 2023-04-30 16:27:45,205 INFO [train.py:904] (6/8) Epoch 18, batch 2800, loss[loss=0.1625, simple_loss=0.2434, pruned_loss=0.0408, over 16739.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2613, pruned_loss=0.04266, over 3334225.74 frames. ], batch size: 134, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:27:58,755 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=175361.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:28:11,812 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175370.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:28:41,460 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2476, 2.1069, 1.6863, 1.8254, 2.3307, 2.0470, 2.2167, 2.4470], device='cuda:6'), covar=tensor([0.0216, 0.0312, 0.0459, 0.0414, 0.0199, 0.0324, 0.0197, 0.0214], device='cuda:6'), in_proj_covar=tensor([0.0200, 0.0235, 0.0224, 0.0224, 0.0234, 0.0233, 0.0238, 0.0227], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:28:45,718 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175394.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:28:56,404 INFO [train.py:904] (6/8) Epoch 18, batch 2850, loss[loss=0.1812, simple_loss=0.2524, pruned_loss=0.055, over 16737.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2608, pruned_loss=0.04259, over 3330075.71 frames. ], batch size: 134, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:29:05,422 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=175409.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:29:35,280 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175431.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:29:58,183 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4482, 3.6549, 4.0363, 2.3912, 3.1356, 2.4999, 3.9309, 3.9066], device='cuda:6'), covar=tensor([0.0291, 0.0843, 0.0458, 0.1784, 0.0814, 0.0924, 0.0607, 0.0968], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0159, 0.0164, 0.0150, 0.0141, 0.0126, 0.0141, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 16:29:58,776 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.293e+02 2.737e+02 3.416e+02 1.541e+03, threshold=5.475e+02, percent-clipped=7.0 2023-04-30 16:30:00,566 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7767, 3.6683, 3.9753, 2.1106, 4.0884, 4.1090, 3.2436, 2.9770], device='cuda:6'), covar=tensor([0.0680, 0.0199, 0.0136, 0.1125, 0.0076, 0.0156, 0.0373, 0.0440], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0106, 0.0094, 0.0138, 0.0077, 0.0122, 0.0125, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 16:30:04,918 INFO [train.py:904] (6/8) Epoch 18, batch 2900, loss[loss=0.1629, simple_loss=0.2527, pruned_loss=0.03656, over 17226.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2591, pruned_loss=0.04307, over 3337615.09 frames. ], batch size: 45, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:30:09,618 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175455.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:30:45,634 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175481.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:31:04,459 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=175495.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:31:13,024 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4287, 5.4229, 5.1998, 4.7069, 5.3340, 2.1932, 5.0531, 5.2126], device='cuda:6'), covar=tensor([0.0085, 0.0087, 0.0194, 0.0359, 0.0092, 0.2348, 0.0133, 0.0161], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0147, 0.0195, 0.0176, 0.0169, 0.0203, 0.0184, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:31:13,711 INFO [train.py:904] (6/8) Epoch 18, batch 2950, loss[loss=0.1609, simple_loss=0.2445, pruned_loss=0.03863, over 15764.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2582, pruned_loss=0.04349, over 3336358.83 frames. ], batch size: 35, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:32:09,690 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175542.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:32:17,355 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.593e+02 2.280e+02 2.587e+02 3.034e+02 5.964e+02, threshold=5.174e+02, percent-clipped=1.0 2023-04-30 16:32:23,313 INFO [train.py:904] (6/8) Epoch 18, batch 3000, loss[loss=0.1628, simple_loss=0.2576, pruned_loss=0.03406, over 17195.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2589, pruned_loss=0.04401, over 3333669.53 frames. ], batch size: 46, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:32:23,314 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 16:32:32,127 INFO [train.py:938] (6/8) Epoch 18, validation: loss=0.1359, simple_loss=0.2415, pruned_loss=0.01516, over 944034.00 frames. 2023-04-30 16:32:32,128 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 16:32:48,742 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=175563.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:33:12,045 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 16:33:42,215 INFO [train.py:904] (6/8) Epoch 18, batch 3050, loss[loss=0.1484, simple_loss=0.2355, pruned_loss=0.03063, over 16847.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2593, pruned_loss=0.04442, over 3327845.67 frames. ], batch size: 42, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:33:53,831 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 16:34:02,906 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-04-30 16:34:30,989 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2182, 5.8865, 5.9853, 5.6508, 5.8150, 6.3696, 5.8315, 5.4899], device='cuda:6'), covar=tensor([0.0899, 0.1958, 0.1940, 0.2205, 0.2585, 0.0956, 0.1310, 0.2415], device='cuda:6'), in_proj_covar=tensor([0.0401, 0.0587, 0.0643, 0.0488, 0.0658, 0.0669, 0.0506, 0.0656], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 16:34:46,186 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 2.265e+02 2.697e+02 3.394e+02 4.892e+02, threshold=5.395e+02, percent-clipped=0.0 2023-04-30 16:34:51,893 INFO [train.py:904] (6/8) Epoch 18, batch 3100, loss[loss=0.1816, simple_loss=0.2549, pruned_loss=0.05416, over 16745.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2584, pruned_loss=0.04445, over 3337441.72 frames. ], batch size: 124, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:35:06,987 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-30 16:36:01,368 INFO [train.py:904] (6/8) Epoch 18, batch 3150, loss[loss=0.1673, simple_loss=0.2409, pruned_loss=0.04686, over 16890.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2576, pruned_loss=0.04392, over 3342632.81 frames. ], batch size: 96, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:36:13,048 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175710.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:36:36,410 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=175726.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:37:05,951 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.459e+02 2.210e+02 2.531e+02 3.035e+02 7.463e+02, threshold=5.061e+02, percent-clipped=4.0 2023-04-30 16:37:09,285 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=175750.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:37:11,226 INFO [train.py:904] (6/8) Epoch 18, batch 3200, loss[loss=0.1522, simple_loss=0.2399, pruned_loss=0.03224, over 15845.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2569, pruned_loss=0.04336, over 3335197.10 frames. ], batch size: 35, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:37:37,869 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175771.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:38:11,045 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=175795.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:38:16,163 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-30 16:38:20,684 INFO [train.py:904] (6/8) Epoch 18, batch 3250, loss[loss=0.1982, simple_loss=0.2787, pruned_loss=0.05885, over 16547.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2576, pruned_loss=0.04394, over 3336929.51 frames. ], batch size: 68, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:38:36,884 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8112, 2.7466, 2.3398, 2.6399, 3.0777, 2.8444, 3.4551, 3.3380], device='cuda:6'), covar=tensor([0.0119, 0.0386, 0.0496, 0.0382, 0.0257, 0.0349, 0.0236, 0.0220], device='cuda:6'), in_proj_covar=tensor([0.0201, 0.0235, 0.0225, 0.0224, 0.0235, 0.0233, 0.0240, 0.0228], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:38:52,168 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7746, 2.5169, 1.9535, 2.3124, 2.8649, 2.6457, 2.9349, 2.9966], device='cuda:6'), covar=tensor([0.0212, 0.0385, 0.0538, 0.0404, 0.0219, 0.0322, 0.0210, 0.0221], device='cuda:6'), in_proj_covar=tensor([0.0202, 0.0235, 0.0225, 0.0224, 0.0235, 0.0233, 0.0240, 0.0228], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:39:10,115 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=175837.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:39:17,732 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=175843.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:39:23,828 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.233e+02 2.626e+02 2.962e+02 5.772e+02, threshold=5.253e+02, percent-clipped=1.0 2023-04-30 16:39:29,883 INFO [train.py:904] (6/8) Epoch 18, batch 3300, loss[loss=0.1807, simple_loss=0.276, pruned_loss=0.04277, over 17113.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2587, pruned_loss=0.04382, over 3343107.94 frames. ], batch size: 49, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:39:45,178 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=175863.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:39:49,324 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.7181, 6.1052, 5.8137, 5.8374, 5.4118, 5.4827, 5.4542, 6.1941], device='cuda:6'), covar=tensor([0.1264, 0.0878, 0.0990, 0.0820, 0.0926, 0.0649, 0.1173, 0.0895], device='cuda:6'), in_proj_covar=tensor([0.0661, 0.0816, 0.0657, 0.0604, 0.0512, 0.0514, 0.0676, 0.0629], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:40:11,012 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175881.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:40:38,824 INFO [train.py:904] (6/8) Epoch 18, batch 3350, loss[loss=0.2159, simple_loss=0.2994, pruned_loss=0.06622, over 12051.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2594, pruned_loss=0.04397, over 3323710.52 frames. ], batch size: 246, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:40:51,560 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=175911.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:41:34,570 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175942.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:41:38,199 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175945.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:41:41,869 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 2.233e+02 2.642e+02 3.041e+02 9.151e+02, threshold=5.284e+02, percent-clipped=6.0 2023-04-30 16:41:47,202 INFO [train.py:904] (6/8) Epoch 18, batch 3400, loss[loss=0.2062, simple_loss=0.2911, pruned_loss=0.06059, over 12119.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.26, pruned_loss=0.04403, over 3315772.32 frames. ], batch size: 246, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:42:52,011 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-30 16:43:01,858 INFO [train.py:904] (6/8) Epoch 18, batch 3450, loss[loss=0.1599, simple_loss=0.2545, pruned_loss=0.03265, over 17277.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2587, pruned_loss=0.04309, over 3331368.96 frames. ], batch size: 52, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:43:08,043 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176006.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:43:12,405 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-30 16:43:35,510 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176026.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:44:06,111 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.158e+02 2.512e+02 2.983e+02 6.813e+02, threshold=5.023e+02, percent-clipped=2.0 2023-04-30 16:44:09,483 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176050.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:44:11,307 INFO [train.py:904] (6/8) Epoch 18, batch 3500, loss[loss=0.1813, simple_loss=0.2753, pruned_loss=0.04368, over 17117.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2578, pruned_loss=0.0431, over 3326093.61 frames. ], batch size: 49, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:44:31,791 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176066.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:44:41,808 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176074.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:45:15,799 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176098.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:45:19,968 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2324, 3.1687, 3.3661, 2.3998, 3.1097, 3.4432, 3.0986, 2.0311], device='cuda:6'), covar=tensor([0.0426, 0.0115, 0.0062, 0.0337, 0.0103, 0.0081, 0.0110, 0.0403], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0080, 0.0080, 0.0133, 0.0094, 0.0104, 0.0091, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 16:45:20,518 INFO [train.py:904] (6/8) Epoch 18, batch 3550, loss[loss=0.1715, simple_loss=0.2516, pruned_loss=0.04576, over 16286.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2566, pruned_loss=0.04278, over 3315438.23 frames. ], batch size: 165, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:46:10,539 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176137.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:46:26,481 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 2.026e+02 2.447e+02 2.938e+02 5.668e+02, threshold=4.895e+02, percent-clipped=1.0 2023-04-30 16:46:32,329 INFO [train.py:904] (6/8) Epoch 18, batch 3600, loss[loss=0.159, simple_loss=0.2408, pruned_loss=0.03861, over 16900.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2548, pruned_loss=0.04234, over 3314153.16 frames. ], batch size: 90, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:47:18,726 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176185.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:47:43,325 INFO [train.py:904] (6/8) Epoch 18, batch 3650, loss[loss=0.1769, simple_loss=0.2507, pruned_loss=0.05153, over 16534.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2541, pruned_loss=0.04281, over 3299387.42 frames. ], batch size: 146, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:48:30,675 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7507, 2.3699, 2.4274, 3.2063, 2.7097, 3.6209, 1.5532, 2.7502], device='cuda:6'), covar=tensor([0.1337, 0.0796, 0.1165, 0.0200, 0.0151, 0.0375, 0.1587, 0.0794], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0169, 0.0189, 0.0182, 0.0203, 0.0215, 0.0194, 0.0188], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 16:48:35,858 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176237.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:48:43,996 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=176242.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:48:47,991 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0925, 3.1560, 3.2984, 2.1732, 2.8717, 2.3010, 3.6495, 3.5421], device='cuda:6'), covar=tensor([0.0230, 0.0935, 0.0657, 0.1853, 0.0845, 0.1044, 0.0485, 0.0853], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0162, 0.0166, 0.0152, 0.0143, 0.0128, 0.0144, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 16:48:53,036 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.679e+02 2.298e+02 2.711e+02 3.292e+02 6.434e+02, threshold=5.422e+02, percent-clipped=3.0 2023-04-30 16:48:58,528 INFO [train.py:904] (6/8) Epoch 18, batch 3700, loss[loss=0.1768, simple_loss=0.2514, pruned_loss=0.05113, over 11402.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2533, pruned_loss=0.04452, over 3285405.33 frames. ], batch size: 247, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:49:07,016 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=176257.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:50:10,835 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176301.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:50:11,691 INFO [train.py:904] (6/8) Epoch 18, batch 3750, loss[loss=0.1633, simple_loss=0.2372, pruned_loss=0.0447, over 16723.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2542, pruned_loss=0.04584, over 3262954.58 frames. ], batch size: 83, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:50:13,726 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176303.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:50:35,180 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176318.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:51:17,331 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.625e+02 2.173e+02 2.603e+02 3.284e+02 5.784e+02, threshold=5.206e+02, percent-clipped=1.0 2023-04-30 16:51:23,241 INFO [train.py:904] (6/8) Epoch 18, batch 3800, loss[loss=0.1992, simple_loss=0.2873, pruned_loss=0.05554, over 12477.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2552, pruned_loss=0.04695, over 3261879.01 frames. ], batch size: 246, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:51:44,214 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176366.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:52:15,856 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0017, 2.1005, 2.6149, 2.9814, 2.8379, 2.9893, 2.1707, 3.2213], device='cuda:6'), covar=tensor([0.0157, 0.0408, 0.0289, 0.0214, 0.0255, 0.0228, 0.0442, 0.0120], device='cuda:6'), in_proj_covar=tensor([0.0182, 0.0187, 0.0176, 0.0180, 0.0187, 0.0149, 0.0191, 0.0143], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:52:31,398 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5734, 3.6079, 2.3069, 3.8095, 2.9580, 3.8021, 2.3869, 2.9140], device='cuda:6'), covar=tensor([0.0213, 0.0366, 0.1272, 0.0258, 0.0588, 0.0718, 0.1171, 0.0610], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0176, 0.0193, 0.0160, 0.0174, 0.0219, 0.0202, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 16:52:35,499 INFO [train.py:904] (6/8) Epoch 18, batch 3850, loss[loss=0.1708, simple_loss=0.2488, pruned_loss=0.04646, over 16322.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2555, pruned_loss=0.04783, over 3258485.91 frames. ], batch size: 68, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:52:37,977 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-04-30 16:52:53,512 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176414.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:53:42,933 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.342e+02 2.782e+02 3.184e+02 4.336e+02, threshold=5.564e+02, percent-clipped=0.0 2023-04-30 16:53:49,441 INFO [train.py:904] (6/8) Epoch 18, batch 3900, loss[loss=0.1744, simple_loss=0.2467, pruned_loss=0.05101, over 16862.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.255, pruned_loss=0.04823, over 3267776.56 frames. ], batch size: 96, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:53:51,584 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 16:54:11,199 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=176466.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:54:13,881 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 16:55:00,904 INFO [train.py:904] (6/8) Epoch 18, batch 3950, loss[loss=0.1786, simple_loss=0.2544, pruned_loss=0.05137, over 16340.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.255, pruned_loss=0.04901, over 3272630.22 frames. ], batch size: 146, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:55:35,374 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176527.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:55:50,457 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176537.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:56:05,763 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.227e+02 2.521e+02 3.093e+02 7.043e+02, threshold=5.041e+02, percent-clipped=2.0 2023-04-30 16:56:12,320 INFO [train.py:904] (6/8) Epoch 18, batch 4000, loss[loss=0.1769, simple_loss=0.2622, pruned_loss=0.0458, over 17130.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2545, pruned_loss=0.04915, over 3284724.83 frames. ], batch size: 47, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:56:33,457 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6944, 2.3472, 1.8252, 2.0865, 2.7056, 2.3784, 2.6869, 2.8600], device='cuda:6'), covar=tensor([0.0227, 0.0367, 0.0548, 0.0460, 0.0208, 0.0348, 0.0179, 0.0225], device='cuda:6'), in_proj_covar=tensor([0.0201, 0.0233, 0.0223, 0.0223, 0.0234, 0.0231, 0.0238, 0.0226], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:56:34,307 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8257, 3.7677, 3.9360, 3.7032, 3.8256, 4.2646, 3.9074, 3.5939], device='cuda:6'), covar=tensor([0.2166, 0.2389, 0.2090, 0.2529, 0.2882, 0.1863, 0.1612, 0.2550], device='cuda:6'), in_proj_covar=tensor([0.0403, 0.0588, 0.0641, 0.0491, 0.0659, 0.0669, 0.0511, 0.0657], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 16:56:37,454 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3564, 4.2674, 4.4069, 4.5656, 4.6782, 4.2421, 4.5744, 4.7112], device='cuda:6'), covar=tensor([0.1658, 0.1010, 0.1325, 0.0696, 0.0548, 0.1202, 0.1279, 0.0543], device='cuda:6'), in_proj_covar=tensor([0.0633, 0.0784, 0.0918, 0.0796, 0.0586, 0.0631, 0.0642, 0.0746], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:57:00,438 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176585.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:57:18,438 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176598.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:57:23,546 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176601.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:57:24,370 INFO [train.py:904] (6/8) Epoch 18, batch 4050, loss[loss=0.1702, simple_loss=0.2521, pruned_loss=0.04408, over 17229.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2554, pruned_loss=0.04833, over 3273996.26 frames. ], batch size: 45, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:57:39,761 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176613.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:57:40,011 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1471, 3.1717, 3.5054, 2.1653, 2.9710, 2.2877, 3.5802, 3.5616], device='cuda:6'), covar=tensor([0.0198, 0.0812, 0.0559, 0.1847, 0.0814, 0.0935, 0.0505, 0.0815], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0161, 0.0165, 0.0151, 0.0142, 0.0127, 0.0143, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 16:57:58,519 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=176625.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:58:30,484 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 1.835e+02 2.096e+02 2.511e+02 3.496e+02, threshold=4.192e+02, percent-clipped=0.0 2023-04-30 16:58:33,596 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176649.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:58:36,778 INFO [train.py:904] (6/8) Epoch 18, batch 4100, loss[loss=0.1777, simple_loss=0.2714, pruned_loss=0.04195, over 16784.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2564, pruned_loss=0.04753, over 3282664.07 frames. ], batch size: 83, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:58:57,228 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=176665.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:59:26,858 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6090, 3.6087, 2.6773, 2.2324, 2.5621, 2.4195, 3.8958, 3.3785], device='cuda:6'), covar=tensor([0.2937, 0.0828, 0.2079, 0.2706, 0.2461, 0.2049, 0.0553, 0.1291], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0265, 0.0300, 0.0304, 0.0294, 0.0248, 0.0287, 0.0329], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 16:59:28,815 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176686.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:59:44,869 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3973, 1.7905, 2.1671, 2.4105, 2.4820, 2.7203, 1.7929, 2.6032], device='cuda:6'), covar=tensor([0.0186, 0.0408, 0.0274, 0.0253, 0.0268, 0.0168, 0.0469, 0.0117], device='cuda:6'), in_proj_covar=tensor([0.0182, 0.0188, 0.0177, 0.0180, 0.0188, 0.0148, 0.0191, 0.0143], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 16:59:51,167 INFO [train.py:904] (6/8) Epoch 18, batch 4150, loss[loss=0.2199, simple_loss=0.3051, pruned_loss=0.06738, over 16881.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2642, pruned_loss=0.0502, over 3248256.89 frames. ], batch size: 109, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 17:00:27,794 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176726.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:00:51,103 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8353, 5.0811, 5.2493, 4.9821, 5.0227, 5.6053, 5.1281, 4.8186], device='cuda:6'), covar=tensor([0.0853, 0.1652, 0.1546, 0.1778, 0.2402, 0.0900, 0.1321, 0.2245], device='cuda:6'), in_proj_covar=tensor([0.0396, 0.0579, 0.0631, 0.0485, 0.0649, 0.0661, 0.0502, 0.0649], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 17:01:00,609 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.553e+02 2.334e+02 2.630e+02 3.373e+02 5.770e+02, threshold=5.259e+02, percent-clipped=9.0 2023-04-30 17:01:06,244 INFO [train.py:904] (6/8) Epoch 18, batch 4200, loss[loss=0.2215, simple_loss=0.314, pruned_loss=0.06449, over 16974.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2712, pruned_loss=0.05172, over 3217212.75 frames. ], batch size: 41, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 17:02:08,516 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.58 vs. limit=5.0 2023-04-30 17:02:21,462 INFO [train.py:904] (6/8) Epoch 18, batch 4250, loss[loss=0.1826, simple_loss=0.2769, pruned_loss=0.04415, over 16465.00 frames. ], tot_loss[loss=0.1895, simple_loss=0.2749, pruned_loss=0.05206, over 3181190.82 frames. ], batch size: 68, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 17:02:51,825 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176822.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:03:30,835 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.197e+02 2.694e+02 3.150e+02 7.622e+02, threshold=5.388e+02, percent-clipped=1.0 2023-04-30 17:03:35,707 INFO [train.py:904] (6/8) Epoch 18, batch 4300, loss[loss=0.1772, simple_loss=0.2799, pruned_loss=0.03726, over 16832.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2759, pruned_loss=0.05108, over 3188889.64 frames. ], batch size: 102, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:03:38,923 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=176853.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:04:44,764 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176898.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:04:49,703 INFO [train.py:904] (6/8) Epoch 18, batch 4350, loss[loss=0.2098, simple_loss=0.2991, pruned_loss=0.06021, over 16404.00 frames. ], tot_loss[loss=0.1921, simple_loss=0.2796, pruned_loss=0.05232, over 3191460.65 frames. ], batch size: 146, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:05:06,434 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176913.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:05:07,743 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176914.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 17:05:21,608 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5302, 4.5043, 4.3307, 3.7328, 4.4591, 1.7007, 4.1860, 4.0190], device='cuda:6'), covar=tensor([0.0077, 0.0064, 0.0169, 0.0299, 0.0071, 0.2901, 0.0116, 0.0225], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0148, 0.0195, 0.0177, 0.0169, 0.0204, 0.0185, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:05:51,732 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-30 17:05:55,206 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176946.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:05:57,198 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.674e+02 2.251e+02 2.629e+02 3.304e+02 5.605e+02, threshold=5.258e+02, percent-clipped=2.0 2023-04-30 17:06:03,297 INFO [train.py:904] (6/8) Epoch 18, batch 4400, loss[loss=0.2071, simple_loss=0.2946, pruned_loss=0.05982, over 16190.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2816, pruned_loss=0.05348, over 3188084.89 frames. ], batch size: 165, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:06:07,455 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4638, 4.3825, 4.5226, 4.6646, 4.7767, 4.3694, 4.7525, 4.7958], device='cuda:6'), covar=tensor([0.1585, 0.1119, 0.1334, 0.0616, 0.0471, 0.0959, 0.0584, 0.0556], device='cuda:6'), in_proj_covar=tensor([0.0607, 0.0753, 0.0883, 0.0767, 0.0565, 0.0606, 0.0618, 0.0717], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:06:17,417 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176961.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:06:28,891 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-04-30 17:06:45,622 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176981.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:07:01,808 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9404, 2.7258, 2.8094, 2.0501, 2.6595, 2.1594, 2.6935, 2.9102], device='cuda:6'), covar=tensor([0.0242, 0.0678, 0.0549, 0.1770, 0.0821, 0.0857, 0.0595, 0.0597], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0160, 0.0164, 0.0149, 0.0141, 0.0126, 0.0141, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 17:07:16,202 INFO [train.py:904] (6/8) Epoch 18, batch 4450, loss[loss=0.2129, simple_loss=0.3043, pruned_loss=0.06081, over 17038.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.285, pruned_loss=0.05482, over 3184653.25 frames. ], batch size: 53, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:07:29,076 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8661, 2.1354, 2.3943, 3.1550, 2.1821, 2.3248, 2.3008, 2.2340], device='cuda:6'), covar=tensor([0.1198, 0.2827, 0.2204, 0.0627, 0.3668, 0.2094, 0.2708, 0.3159], device='cuda:6'), in_proj_covar=tensor([0.0389, 0.0432, 0.0355, 0.0324, 0.0427, 0.0500, 0.0401, 0.0504], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:07:44,705 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=177021.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:08:24,607 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.309e+02 1.969e+02 2.287e+02 2.678e+02 4.361e+02, threshold=4.575e+02, percent-clipped=0.0 2023-04-30 17:08:30,150 INFO [train.py:904] (6/8) Epoch 18, batch 4500, loss[loss=0.1814, simple_loss=0.2718, pruned_loss=0.04552, over 16760.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2851, pruned_loss=0.05503, over 3193676.35 frames. ], batch size: 89, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:09:00,329 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1377, 2.9688, 3.2327, 1.6541, 3.4514, 3.4755, 2.7747, 2.5745], device='cuda:6'), covar=tensor([0.0903, 0.0287, 0.0214, 0.1270, 0.0082, 0.0156, 0.0485, 0.0514], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0106, 0.0094, 0.0138, 0.0076, 0.0122, 0.0125, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 17:09:43,612 INFO [train.py:904] (6/8) Epoch 18, batch 4550, loss[loss=0.2152, simple_loss=0.3005, pruned_loss=0.06501, over 16271.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2854, pruned_loss=0.05556, over 3208752.65 frames. ], batch size: 165, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:09:47,363 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3071, 3.1113, 3.3898, 1.7793, 3.6056, 3.5965, 2.8342, 2.7263], device='cuda:6'), covar=tensor([0.0815, 0.0255, 0.0193, 0.1169, 0.0069, 0.0145, 0.0461, 0.0469], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0106, 0.0094, 0.0138, 0.0076, 0.0122, 0.0125, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 17:09:52,967 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=177109.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:09:55,574 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=177111.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:10:11,671 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=177122.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:10:48,535 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.278e+02 1.819e+02 2.100e+02 2.524e+02 4.531e+02, threshold=4.200e+02, percent-clipped=0.0 2023-04-30 17:10:53,841 INFO [train.py:904] (6/8) Epoch 18, batch 4600, loss[loss=0.19, simple_loss=0.2796, pruned_loss=0.05022, over 17003.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2867, pruned_loss=0.05594, over 3217567.57 frames. ], batch size: 50, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:11:19,982 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=177170.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:11:20,245 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=177170.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:11:24,022 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=177172.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:12:05,776 INFO [train.py:904] (6/8) Epoch 18, batch 4650, loss[loss=0.2007, simple_loss=0.2744, pruned_loss=0.06347, over 16426.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2864, pruned_loss=0.05645, over 3216703.74 frames. ], batch size: 35, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:12:16,113 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=177209.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 17:13:10,932 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 1.973e+02 2.204e+02 2.634e+02 3.745e+02, threshold=4.408e+02, percent-clipped=0.0 2023-04-30 17:13:16,216 INFO [train.py:904] (6/8) Epoch 18, batch 4700, loss[loss=0.1709, simple_loss=0.2624, pruned_loss=0.0397, over 16889.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2831, pruned_loss=0.0549, over 3214315.16 frames. ], batch size: 102, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:13:59,447 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=177281.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:14:17,223 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9665, 4.8138, 4.9827, 5.1923, 5.4087, 4.8060, 5.3705, 5.3784], device='cuda:6'), covar=tensor([0.1724, 0.1167, 0.1622, 0.0681, 0.0479, 0.0816, 0.0532, 0.0513], device='cuda:6'), in_proj_covar=tensor([0.0603, 0.0744, 0.0875, 0.0763, 0.0560, 0.0601, 0.0613, 0.0709], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:14:28,060 INFO [train.py:904] (6/8) Epoch 18, batch 4750, loss[loss=0.1672, simple_loss=0.2576, pruned_loss=0.03837, over 16325.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2786, pruned_loss=0.05258, over 3228047.14 frames. ], batch size: 165, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:14:56,853 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=177321.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:15:07,310 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=177329.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:15:27,399 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2297, 3.1131, 3.3943, 1.6693, 3.6342, 3.6161, 2.8404, 2.6175], device='cuda:6'), covar=tensor([0.0841, 0.0264, 0.0180, 0.1298, 0.0067, 0.0139, 0.0400, 0.0531], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0107, 0.0095, 0.0139, 0.0076, 0.0123, 0.0126, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 17:15:34,632 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.255e+02 1.837e+02 2.097e+02 2.577e+02 4.933e+02, threshold=4.195e+02, percent-clipped=1.0 2023-04-30 17:15:40,551 INFO [train.py:904] (6/8) Epoch 18, batch 4800, loss[loss=0.2111, simple_loss=0.3013, pruned_loss=0.06049, over 16695.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2754, pruned_loss=0.05052, over 3231977.05 frames. ], batch size: 134, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:15:47,077 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-30 17:16:07,489 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=177369.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:16:57,032 INFO [train.py:904] (6/8) Epoch 18, batch 4850, loss[loss=0.1643, simple_loss=0.2432, pruned_loss=0.0427, over 17027.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2764, pruned_loss=0.04989, over 3209045.16 frames. ], batch size: 53, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:17:19,832 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3047, 4.3776, 4.6639, 4.6280, 4.6391, 4.3142, 4.3127, 4.1662], device='cuda:6'), covar=tensor([0.0269, 0.0516, 0.0339, 0.0352, 0.0392, 0.0366, 0.0840, 0.0469], device='cuda:6'), in_proj_covar=tensor([0.0379, 0.0412, 0.0406, 0.0380, 0.0451, 0.0425, 0.0520, 0.0338], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 17:17:28,969 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4143, 3.3801, 1.9807, 3.8306, 2.5033, 3.8094, 2.1766, 2.7628], device='cuda:6'), covar=tensor([0.0288, 0.0395, 0.1677, 0.0115, 0.0936, 0.0519, 0.1514, 0.0784], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0171, 0.0188, 0.0150, 0.0171, 0.0210, 0.0197, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 17:17:36,071 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4452, 4.4110, 4.3580, 3.6488, 4.3954, 1.6647, 4.0788, 4.0384], device='cuda:6'), covar=tensor([0.0095, 0.0097, 0.0143, 0.0353, 0.0086, 0.2677, 0.0141, 0.0220], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0144, 0.0188, 0.0173, 0.0163, 0.0199, 0.0179, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:18:05,878 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 1.952e+02 2.254e+02 2.679e+02 4.935e+02, threshold=4.508e+02, percent-clipped=3.0 2023-04-30 17:18:11,209 INFO [train.py:904] (6/8) Epoch 18, batch 4900, loss[loss=0.1673, simple_loss=0.2556, pruned_loss=0.03952, over 17040.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2754, pruned_loss=0.04847, over 3189827.90 frames. ], batch size: 50, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:18:29,232 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1001, 4.0978, 2.6553, 5.0431, 3.1926, 4.8866, 2.9481, 3.4015], device='cuda:6'), covar=tensor([0.0219, 0.0323, 0.1491, 0.0081, 0.0795, 0.0316, 0.1213, 0.0634], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0170, 0.0188, 0.0149, 0.0170, 0.0209, 0.0196, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 17:18:30,786 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=177465.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:18:33,658 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=177467.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:18:44,815 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-30 17:19:07,844 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 17:19:25,467 INFO [train.py:904] (6/8) Epoch 18, batch 4950, loss[loss=0.1945, simple_loss=0.2832, pruned_loss=0.05286, over 16978.00 frames. ], tot_loss[loss=0.1858, simple_loss=0.2754, pruned_loss=0.04811, over 3211366.11 frames. ], batch size: 41, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:19:34,706 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=177509.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 17:20:14,882 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6735, 4.8955, 4.5555, 4.3261, 3.8667, 4.7963, 4.7953, 4.3780], device='cuda:6'), covar=tensor([0.0941, 0.0528, 0.0536, 0.0428, 0.1975, 0.0537, 0.0342, 0.0693], device='cuda:6'), in_proj_covar=tensor([0.0278, 0.0394, 0.0328, 0.0318, 0.0339, 0.0369, 0.0223, 0.0389], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:20:30,519 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.064e+02 2.400e+02 2.988e+02 6.295e+02, threshold=4.800e+02, percent-clipped=2.0 2023-04-30 17:20:35,819 INFO [train.py:904] (6/8) Epoch 18, batch 5000, loss[loss=0.1903, simple_loss=0.2973, pruned_loss=0.0416, over 16740.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2773, pruned_loss=0.04808, over 3216262.23 frames. ], batch size: 89, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:20:37,497 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=177553.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:20:41,866 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=177557.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:21:47,168 INFO [train.py:904] (6/8) Epoch 18, batch 5050, loss[loss=0.1921, simple_loss=0.2825, pruned_loss=0.05081, over 12142.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2774, pruned_loss=0.04791, over 3205711.94 frames. ], batch size: 247, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:21:49,882 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6599, 4.7376, 4.9434, 4.7018, 4.7540, 5.3255, 4.8259, 4.5248], device='cuda:6'), covar=tensor([0.1119, 0.1726, 0.1756, 0.1952, 0.2506, 0.0918, 0.1324, 0.2377], device='cuda:6'), in_proj_covar=tensor([0.0391, 0.0562, 0.0615, 0.0474, 0.0633, 0.0645, 0.0485, 0.0635], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 17:22:04,624 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=177614.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:22:54,210 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.030e+02 2.394e+02 2.756e+02 4.573e+02, threshold=4.789e+02, percent-clipped=0.0 2023-04-30 17:22:58,431 INFO [train.py:904] (6/8) Epoch 18, batch 5100, loss[loss=0.1943, simple_loss=0.2692, pruned_loss=0.05968, over 16459.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2754, pruned_loss=0.04766, over 3215433.61 frames. ], batch size: 35, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:23:00,731 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-30 17:23:16,229 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=177664.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:23:36,912 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1962, 3.1017, 3.4291, 1.7186, 3.6414, 3.6224, 2.8166, 2.6715], device='cuda:6'), covar=tensor([0.0870, 0.0277, 0.0162, 0.1261, 0.0064, 0.0129, 0.0401, 0.0504], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0105, 0.0093, 0.0137, 0.0075, 0.0120, 0.0124, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 17:23:43,615 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2023-04-30 17:23:46,597 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8460, 4.2362, 3.1419, 2.3996, 2.8228, 2.5929, 4.4737, 3.7800], device='cuda:6'), covar=tensor([0.2526, 0.0590, 0.1635, 0.2666, 0.2446, 0.1703, 0.0407, 0.0982], device='cuda:6'), in_proj_covar=tensor([0.0320, 0.0266, 0.0301, 0.0304, 0.0294, 0.0247, 0.0289, 0.0327], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 17:24:05,196 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-30 17:24:10,486 INFO [train.py:904] (6/8) Epoch 18, batch 5150, loss[loss=0.1737, simple_loss=0.2629, pruned_loss=0.04229, over 17030.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2756, pruned_loss=0.0473, over 3191957.52 frames. ], batch size: 50, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:24:38,711 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9633, 5.0034, 5.3871, 5.3604, 5.3353, 4.9730, 4.9014, 4.6592], device='cuda:6'), covar=tensor([0.0272, 0.0436, 0.0284, 0.0297, 0.0434, 0.0327, 0.0939, 0.0448], device='cuda:6'), in_proj_covar=tensor([0.0379, 0.0413, 0.0407, 0.0380, 0.0453, 0.0427, 0.0522, 0.0340], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 17:24:46,072 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=177725.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:25:04,649 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2968, 3.4702, 3.6370, 3.6085, 3.5892, 3.4217, 3.4334, 3.4901], device='cuda:6'), covar=tensor([0.0388, 0.0586, 0.0399, 0.0419, 0.0608, 0.0488, 0.0817, 0.0447], device='cuda:6'), in_proj_covar=tensor([0.0380, 0.0413, 0.0408, 0.0381, 0.0453, 0.0427, 0.0522, 0.0341], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 17:25:20,503 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.353e+02 1.888e+02 2.241e+02 2.680e+02 5.710e+02, threshold=4.482e+02, percent-clipped=2.0 2023-04-30 17:25:24,673 INFO [train.py:904] (6/8) Epoch 18, batch 5200, loss[loss=0.1764, simple_loss=0.2555, pruned_loss=0.04867, over 16510.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2744, pruned_loss=0.04667, over 3189515.46 frames. ], batch size: 62, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:25:44,165 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=177765.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:25:47,833 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=177767.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:26:06,686 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=177780.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:26:07,073 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 17:26:22,002 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-30 17:26:31,594 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=177797.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:26:39,156 INFO [train.py:904] (6/8) Epoch 18, batch 5250, loss[loss=0.1967, simple_loss=0.2763, pruned_loss=0.05853, over 12400.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2717, pruned_loss=0.04634, over 3202078.54 frames. ], batch size: 246, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:26:55,361 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=177813.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:26:57,990 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1684, 4.2470, 4.0339, 3.7610, 3.7309, 4.1568, 3.8564, 3.9212], device='cuda:6'), covar=tensor([0.0585, 0.0486, 0.0332, 0.0320, 0.0907, 0.0488, 0.0715, 0.0568], device='cuda:6'), in_proj_covar=tensor([0.0280, 0.0395, 0.0329, 0.0320, 0.0340, 0.0370, 0.0224, 0.0389], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:26:58,860 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=177815.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:27:37,122 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2127, 5.7934, 5.8883, 5.6419, 5.7764, 6.2268, 5.8082, 5.5474], device='cuda:6'), covar=tensor([0.0730, 0.1656, 0.1241, 0.1838, 0.2032, 0.0798, 0.1177, 0.2208], device='cuda:6'), in_proj_covar=tensor([0.0391, 0.0561, 0.0614, 0.0473, 0.0632, 0.0643, 0.0485, 0.0633], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 17:27:37,263 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=177841.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:27:48,912 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 1.963e+02 2.196e+02 2.536e+02 4.359e+02, threshold=4.392e+02, percent-clipped=1.0 2023-04-30 17:27:52,849 INFO [train.py:904] (6/8) Epoch 18, batch 5300, loss[loss=0.1565, simple_loss=0.2571, pruned_loss=0.02788, over 16791.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2683, pruned_loss=0.04535, over 3196968.69 frames. ], batch size: 102, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:28:01,116 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6755, 2.2993, 2.3628, 3.1783, 2.3177, 3.6298, 1.5686, 2.7830], device='cuda:6'), covar=tensor([0.1344, 0.0815, 0.1199, 0.0164, 0.0171, 0.0373, 0.1599, 0.0771], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0171, 0.0192, 0.0181, 0.0205, 0.0214, 0.0197, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 17:28:02,161 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=177858.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:28:29,012 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1891, 3.3547, 3.5385, 2.0650, 2.9859, 2.4197, 3.5559, 3.6379], device='cuda:6'), covar=tensor([0.0218, 0.0733, 0.0529, 0.1904, 0.0801, 0.0885, 0.0558, 0.0810], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0160, 0.0166, 0.0150, 0.0142, 0.0127, 0.0142, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 17:29:05,781 INFO [train.py:904] (6/8) Epoch 18, batch 5350, loss[loss=0.1573, simple_loss=0.2469, pruned_loss=0.03378, over 17101.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2661, pruned_loss=0.0445, over 3184658.38 frames. ], batch size: 47, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:29:17,490 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=177909.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:30:14,903 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-30 17:30:15,240 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.427e+02 1.886e+02 2.232e+02 2.628e+02 4.605e+02, threshold=4.463e+02, percent-clipped=1.0 2023-04-30 17:30:19,769 INFO [train.py:904] (6/8) Epoch 18, batch 5400, loss[loss=0.1907, simple_loss=0.2811, pruned_loss=0.0501, over 16288.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2686, pruned_loss=0.0451, over 3178879.39 frames. ], batch size: 165, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:30:36,690 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-30 17:31:39,766 INFO [train.py:904] (6/8) Epoch 18, batch 5450, loss[loss=0.1694, simple_loss=0.2648, pruned_loss=0.037, over 17112.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2722, pruned_loss=0.0467, over 3187679.86 frames. ], batch size: 49, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:31:44,599 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4681, 3.7477, 2.7550, 2.2418, 2.5179, 2.3016, 3.9709, 3.3745], device='cuda:6'), covar=tensor([0.2876, 0.0652, 0.1767, 0.2475, 0.2499, 0.1934, 0.0454, 0.1139], device='cuda:6'), in_proj_covar=tensor([0.0315, 0.0261, 0.0295, 0.0299, 0.0288, 0.0243, 0.0285, 0.0323], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 17:32:08,838 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=178020.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:32:12,132 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1112, 2.4185, 2.6061, 1.9288, 2.7602, 2.7871, 2.4642, 2.4146], device='cuda:6'), covar=tensor([0.0660, 0.0236, 0.0223, 0.0868, 0.0107, 0.0307, 0.0415, 0.0397], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0105, 0.0093, 0.0137, 0.0076, 0.0121, 0.0124, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 17:32:38,190 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4454, 4.4844, 4.3059, 4.0582, 3.9999, 4.4238, 4.1832, 4.1421], device='cuda:6'), covar=tensor([0.0616, 0.0580, 0.0326, 0.0311, 0.0910, 0.0485, 0.0601, 0.0621], device='cuda:6'), in_proj_covar=tensor([0.0284, 0.0401, 0.0333, 0.0324, 0.0344, 0.0376, 0.0225, 0.0394], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:32:52,432 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.664e+02 2.290e+02 2.936e+02 3.716e+02 7.248e+02, threshold=5.872e+02, percent-clipped=9.0 2023-04-30 17:32:56,916 INFO [train.py:904] (6/8) Epoch 18, batch 5500, loss[loss=0.2303, simple_loss=0.3167, pruned_loss=0.07195, over 16872.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2795, pruned_loss=0.05107, over 3164324.08 frames. ], batch size: 102, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:34:14,385 INFO [train.py:904] (6/8) Epoch 18, batch 5550, loss[loss=0.2306, simple_loss=0.3147, pruned_loss=0.07323, over 15407.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2864, pruned_loss=0.05585, over 3148722.06 frames. ], batch size: 191, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:35:09,859 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=178136.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:35:31,150 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.108e+02 3.250e+02 3.857e+02 5.113e+02 1.112e+03, threshold=7.715e+02, percent-clipped=10.0 2023-04-30 17:35:34,173 INFO [train.py:904] (6/8) Epoch 18, batch 5600, loss[loss=0.2984, simple_loss=0.3595, pruned_loss=0.1187, over 11526.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2907, pruned_loss=0.05979, over 3125936.70 frames. ], batch size: 247, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:35:36,602 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=178153.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:36:58,587 INFO [train.py:904] (6/8) Epoch 18, batch 5650, loss[loss=0.3082, simple_loss=0.3512, pruned_loss=0.1326, over 11385.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2963, pruned_loss=0.06491, over 3061499.23 frames. ], batch size: 247, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:37:09,801 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=178209.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:37:54,297 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3714, 5.3936, 5.1270, 4.2594, 5.3143, 1.7543, 4.9785, 4.9215], device='cuda:6'), covar=tensor([0.0091, 0.0070, 0.0198, 0.0474, 0.0089, 0.2856, 0.0143, 0.0228], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0143, 0.0188, 0.0173, 0.0163, 0.0198, 0.0178, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:38:05,073 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6599, 1.7917, 1.6123, 1.5618, 1.9425, 1.5935, 1.5864, 1.8686], device='cuda:6'), covar=tensor([0.0188, 0.0259, 0.0355, 0.0316, 0.0193, 0.0248, 0.0170, 0.0202], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0227, 0.0220, 0.0218, 0.0228, 0.0227, 0.0228, 0.0221], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:38:16,450 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.765e+02 3.197e+02 3.818e+02 4.692e+02 7.543e+02, threshold=7.636e+02, percent-clipped=0.0 2023-04-30 17:38:17,793 INFO [train.py:904] (6/8) Epoch 18, batch 5700, loss[loss=0.2594, simple_loss=0.3176, pruned_loss=0.1006, over 11307.00 frames. ], tot_loss[loss=0.2139, simple_loss=0.2967, pruned_loss=0.06556, over 3065131.97 frames. ], batch size: 248, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:38:25,290 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=178257.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:39:39,090 INFO [train.py:904] (6/8) Epoch 18, batch 5750, loss[loss=0.249, simple_loss=0.3164, pruned_loss=0.09081, over 11410.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.2999, pruned_loss=0.06736, over 3040750.41 frames. ], batch size: 246, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:40:08,724 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=178320.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:40:59,315 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 3.189e+02 3.915e+02 4.745e+02 1.114e+03, threshold=7.831e+02, percent-clipped=3.0 2023-04-30 17:41:00,704 INFO [train.py:904] (6/8) Epoch 18, batch 5800, loss[loss=0.2117, simple_loss=0.2853, pruned_loss=0.06901, over 11739.00 frames. ], tot_loss[loss=0.2165, simple_loss=0.3, pruned_loss=0.06644, over 3035333.96 frames. ], batch size: 247, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:41:28,158 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=178368.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:42:19,739 INFO [train.py:904] (6/8) Epoch 18, batch 5850, loss[loss=0.2187, simple_loss=0.2948, pruned_loss=0.07131, over 11490.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2976, pruned_loss=0.06436, over 3052402.16 frames. ], batch size: 246, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:43:16,267 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=178436.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:43:39,873 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.729e+02 2.654e+02 3.099e+02 3.804e+02 9.220e+02, threshold=6.197e+02, percent-clipped=1.0 2023-04-30 17:43:41,465 INFO [train.py:904] (6/8) Epoch 18, batch 5900, loss[loss=0.2074, simple_loss=0.2905, pruned_loss=0.0622, over 17180.00 frames. ], tot_loss[loss=0.213, simple_loss=0.2972, pruned_loss=0.06441, over 3050727.66 frames. ], batch size: 46, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:43:42,240 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=178452.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:43:44,113 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=178453.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:44:32,739 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=178484.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:44:51,612 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0880, 3.3598, 3.4885, 2.2338, 3.2333, 3.5197, 3.3243, 1.9419], device='cuda:6'), covar=tensor([0.0537, 0.0073, 0.0060, 0.0427, 0.0100, 0.0104, 0.0085, 0.0450], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0079, 0.0080, 0.0132, 0.0094, 0.0105, 0.0091, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 17:44:58,683 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=178501.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:44:59,546 INFO [train.py:904] (6/8) Epoch 18, batch 5950, loss[loss=0.2154, simple_loss=0.298, pruned_loss=0.06645, over 17036.00 frames. ], tot_loss[loss=0.212, simple_loss=0.2978, pruned_loss=0.06306, over 3063274.80 frames. ], batch size: 53, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:45:18,609 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=178513.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:45:21,353 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0233, 3.0161, 1.9597, 3.2491, 2.3365, 3.2922, 2.0778, 2.5477], device='cuda:6'), covar=tensor([0.0280, 0.0393, 0.1432, 0.0270, 0.0815, 0.0661, 0.1403, 0.0678], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0171, 0.0190, 0.0151, 0.0173, 0.0211, 0.0198, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 17:45:29,126 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3959, 4.4972, 4.6392, 4.5086, 4.5416, 5.0301, 4.5736, 4.3635], device='cuda:6'), covar=tensor([0.1414, 0.1904, 0.2072, 0.1998, 0.2468, 0.1033, 0.1660, 0.2651], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0561, 0.0616, 0.0471, 0.0631, 0.0642, 0.0488, 0.0636], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 17:46:17,845 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 2.727e+02 3.457e+02 4.156e+02 1.353e+03, threshold=6.914e+02, percent-clipped=3.0 2023-04-30 17:46:19,085 INFO [train.py:904] (6/8) Epoch 18, batch 6000, loss[loss=0.2296, simple_loss=0.2998, pruned_loss=0.07971, over 11396.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.2968, pruned_loss=0.06239, over 3065375.30 frames. ], batch size: 246, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:46:19,086 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 17:46:29,951 INFO [train.py:938] (6/8) Epoch 18, validation: loss=0.1531, simple_loss=0.2661, pruned_loss=0.02007, over 944034.00 frames. 2023-04-30 17:46:29,952 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 17:47:48,044 INFO [train.py:904] (6/8) Epoch 18, batch 6050, loss[loss=0.1962, simple_loss=0.29, pruned_loss=0.05123, over 17135.00 frames. ], tot_loss[loss=0.2091, simple_loss=0.2951, pruned_loss=0.06157, over 3085802.81 frames. ], batch size: 48, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:48:14,319 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=178619.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:48:55,300 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6113, 2.2446, 1.7735, 2.0105, 2.6007, 2.2268, 2.4129, 2.7130], device='cuda:6'), covar=tensor([0.0174, 0.0375, 0.0511, 0.0435, 0.0230, 0.0373, 0.0196, 0.0236], device='cuda:6'), in_proj_covar=tensor([0.0191, 0.0224, 0.0217, 0.0216, 0.0226, 0.0224, 0.0226, 0.0220], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:49:06,484 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.761e+02 2.787e+02 3.191e+02 3.956e+02 7.476e+02, threshold=6.381e+02, percent-clipped=1.0 2023-04-30 17:49:06,499 INFO [train.py:904] (6/8) Epoch 18, batch 6100, loss[loss=0.1924, simple_loss=0.2802, pruned_loss=0.05229, over 16521.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2946, pruned_loss=0.06032, over 3115991.17 frames. ], batch size: 75, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:49:24,131 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7859, 2.8047, 2.8437, 4.8195, 3.7159, 4.2845, 1.6331, 3.1401], device='cuda:6'), covar=tensor([0.1365, 0.0789, 0.1111, 0.0134, 0.0309, 0.0396, 0.1707, 0.0809], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0170, 0.0191, 0.0180, 0.0206, 0.0215, 0.0196, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 17:49:51,604 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=178680.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:50:13,089 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=178694.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:50:16,787 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=178697.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 17:50:18,826 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1208, 2.8485, 3.0834, 1.7181, 3.2952, 3.3319, 2.6981, 2.5611], device='cuda:6'), covar=tensor([0.0837, 0.0301, 0.0188, 0.1210, 0.0081, 0.0198, 0.0454, 0.0462], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0106, 0.0095, 0.0139, 0.0076, 0.0122, 0.0125, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 17:50:23,986 INFO [train.py:904] (6/8) Epoch 18, batch 6150, loss[loss=0.191, simple_loss=0.2744, pruned_loss=0.05381, over 16885.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2931, pruned_loss=0.06024, over 3094113.15 frames. ], batch size: 116, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 17:51:35,035 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.11 vs. limit=5.0 2023-04-30 17:51:39,656 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.052e+02 2.816e+02 3.285e+02 4.160e+02 8.412e+02, threshold=6.570e+02, percent-clipped=4.0 2023-04-30 17:51:39,671 INFO [train.py:904] (6/8) Epoch 18, batch 6200, loss[loss=0.2084, simple_loss=0.2853, pruned_loss=0.06578, over 11532.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2909, pruned_loss=0.05965, over 3084725.24 frames. ], batch size: 248, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 17:51:45,771 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=178755.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 17:51:49,740 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=178758.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 17:52:09,029 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=178770.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:52:12,928 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0518, 5.0882, 5.5536, 5.4627, 5.4961, 5.1147, 5.0455, 4.8158], device='cuda:6'), covar=tensor([0.0313, 0.0524, 0.0313, 0.0432, 0.0477, 0.0367, 0.1075, 0.0470], device='cuda:6'), in_proj_covar=tensor([0.0386, 0.0421, 0.0412, 0.0387, 0.0460, 0.0434, 0.0531, 0.0346], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 17:52:52,115 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-30 17:52:56,568 INFO [train.py:904] (6/8) Epoch 18, batch 6250, loss[loss=0.1952, simple_loss=0.3045, pruned_loss=0.04294, over 16840.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2907, pruned_loss=0.05931, over 3099662.89 frames. ], batch size: 102, lr: 3.75e-03, grad_scale: 2.0 2023-04-30 17:53:07,990 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=178808.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:53:42,347 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=178831.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:54:15,683 INFO [train.py:904] (6/8) Epoch 18, batch 6300, loss[loss=0.1939, simple_loss=0.2912, pruned_loss=0.04833, over 16735.00 frames. ], tot_loss[loss=0.205, simple_loss=0.2912, pruned_loss=0.05944, over 3090016.02 frames. ], batch size: 83, lr: 3.75e-03, grad_scale: 2.0 2023-04-30 17:54:17,523 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.018e+02 2.791e+02 3.273e+02 3.863e+02 9.337e+02, threshold=6.545e+02, percent-clipped=2.0 2023-04-30 17:54:45,824 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9351, 3.9649, 4.2984, 4.2641, 4.2646, 3.9824, 3.9855, 3.9786], device='cuda:6'), covar=tensor([0.0368, 0.0637, 0.0413, 0.0432, 0.0507, 0.0495, 0.1013, 0.0560], device='cuda:6'), in_proj_covar=tensor([0.0386, 0.0420, 0.0412, 0.0387, 0.0460, 0.0434, 0.0531, 0.0346], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 17:55:22,840 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=178894.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 17:55:34,243 INFO [train.py:904] (6/8) Epoch 18, batch 6350, loss[loss=0.2156, simple_loss=0.3006, pruned_loss=0.06527, over 16174.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2924, pruned_loss=0.06105, over 3069263.64 frames. ], batch size: 165, lr: 3.75e-03, grad_scale: 2.0 2023-04-30 17:56:52,036 INFO [train.py:904] (6/8) Epoch 18, batch 6400, loss[loss=0.2133, simple_loss=0.2928, pruned_loss=0.06687, over 17118.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.2923, pruned_loss=0.06177, over 3069336.16 frames. ], batch size: 49, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 17:56:53,836 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.717e+02 3.036e+02 3.548e+02 4.325e+02 9.483e+02, threshold=7.097e+02, percent-clipped=4.0 2023-04-30 17:56:57,535 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=178955.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 17:57:06,746 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5339, 5.5434, 5.4070, 4.6980, 5.4579, 2.0878, 5.1994, 5.1689], device='cuda:6'), covar=tensor([0.0070, 0.0065, 0.0154, 0.0367, 0.0082, 0.2608, 0.0118, 0.0156], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0142, 0.0187, 0.0172, 0.0162, 0.0197, 0.0177, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:57:20,950 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.34 vs. limit=5.0 2023-04-30 17:57:28,084 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=178975.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:57:30,119 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7005, 4.5107, 4.7393, 4.9184, 5.0831, 4.5930, 5.0339, 5.0578], device='cuda:6'), covar=tensor([0.1948, 0.1352, 0.1628, 0.0752, 0.0606, 0.0908, 0.0734, 0.0619], device='cuda:6'), in_proj_covar=tensor([0.0598, 0.0737, 0.0865, 0.0754, 0.0560, 0.0594, 0.0608, 0.0700], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:58:03,869 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.43 vs. limit=5.0 2023-04-30 17:58:07,507 INFO [train.py:904] (6/8) Epoch 18, batch 6450, loss[loss=0.2114, simple_loss=0.2805, pruned_loss=0.07112, over 11626.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2927, pruned_loss=0.06164, over 3066444.28 frames. ], batch size: 248, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 17:58:34,644 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6114, 1.7751, 2.2646, 2.4879, 2.5248, 2.8716, 1.8654, 2.7877], device='cuda:6'), covar=tensor([0.0195, 0.0478, 0.0290, 0.0359, 0.0283, 0.0179, 0.0496, 0.0137], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0189, 0.0174, 0.0178, 0.0186, 0.0145, 0.0191, 0.0141], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 17:59:12,403 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-30 17:59:20,221 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-30 17:59:24,077 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179050.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 17:59:26,156 INFO [train.py:904] (6/8) Epoch 18, batch 6500, loss[loss=0.1857, simple_loss=0.279, pruned_loss=0.04623, over 16870.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2912, pruned_loss=0.06099, over 3078078.49 frames. ], batch size: 102, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 17:59:27,316 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.795e+02 2.918e+02 3.772e+02 4.451e+02 8.133e+02, threshold=7.544e+02, percent-clipped=2.0 2023-04-30 17:59:28,356 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179053.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 17:59:40,649 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7698, 4.7950, 4.6434, 4.3401, 4.2969, 4.7457, 4.6415, 4.4298], device='cuda:6'), covar=tensor([0.0584, 0.0516, 0.0260, 0.0289, 0.0950, 0.0410, 0.0390, 0.0605], device='cuda:6'), in_proj_covar=tensor([0.0278, 0.0394, 0.0323, 0.0315, 0.0336, 0.0368, 0.0221, 0.0387], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:00:32,565 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-30 18:00:44,327 INFO [train.py:904] (6/8) Epoch 18, batch 6550, loss[loss=0.1977, simple_loss=0.2996, pruned_loss=0.04793, over 17029.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2935, pruned_loss=0.06092, over 3094101.76 frames. ], batch size: 50, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:00:54,416 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=179108.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:00:56,027 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-30 18:01:03,329 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 18:01:20,177 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179126.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:01:58,901 INFO [train.py:904] (6/8) Epoch 18, batch 6600, loss[loss=0.2294, simple_loss=0.3199, pruned_loss=0.06943, over 16293.00 frames. ], tot_loss[loss=0.2094, simple_loss=0.2958, pruned_loss=0.06154, over 3085898.28 frames. ], batch size: 165, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:02:00,668 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 2.804e+02 3.237e+02 3.890e+02 6.888e+02, threshold=6.473e+02, percent-clipped=0.0 2023-04-30 18:02:05,163 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=179156.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:02:20,059 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-30 18:02:35,447 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-30 18:03:17,504 INFO [train.py:904] (6/8) Epoch 18, batch 6650, loss[loss=0.194, simple_loss=0.2784, pruned_loss=0.05483, over 16452.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.296, pruned_loss=0.06263, over 3085043.30 frames. ], batch size: 146, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:03:24,248 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4207, 2.2068, 1.8101, 1.9879, 2.4526, 2.1474, 2.2174, 2.5960], device='cuda:6'), covar=tensor([0.0173, 0.0342, 0.0498, 0.0418, 0.0250, 0.0366, 0.0220, 0.0245], device='cuda:6'), in_proj_covar=tensor([0.0189, 0.0223, 0.0218, 0.0217, 0.0226, 0.0224, 0.0226, 0.0220], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:04:30,597 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179250.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:04:32,474 INFO [train.py:904] (6/8) Epoch 18, batch 6700, loss[loss=0.2476, simple_loss=0.3124, pruned_loss=0.0914, over 11657.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2952, pruned_loss=0.06303, over 3075890.12 frames. ], batch size: 249, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:04:34,184 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.091e+02 2.690e+02 3.432e+02 4.163e+02 9.246e+02, threshold=6.864e+02, percent-clipped=3.0 2023-04-30 18:05:09,375 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=179275.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:05:48,861 INFO [train.py:904] (6/8) Epoch 18, batch 6750, loss[loss=0.206, simple_loss=0.284, pruned_loss=0.06399, over 16655.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2932, pruned_loss=0.06215, over 3101488.15 frames. ], batch size: 62, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:06:20,235 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=179323.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:06:23,379 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1809, 3.5095, 3.7107, 2.0377, 3.0757, 2.4262, 3.6579, 3.7383], device='cuda:6'), covar=tensor([0.0248, 0.0768, 0.0500, 0.1937, 0.0793, 0.0909, 0.0575, 0.0866], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0159, 0.0164, 0.0149, 0.0142, 0.0127, 0.0141, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 18:06:44,114 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9471, 4.9186, 4.7117, 3.7582, 4.8687, 1.6914, 4.5409, 4.3568], device='cuda:6'), covar=tensor([0.0098, 0.0088, 0.0205, 0.0526, 0.0103, 0.3156, 0.0164, 0.0304], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0142, 0.0188, 0.0172, 0.0163, 0.0198, 0.0177, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:07:00,958 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=179350.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:07:03,095 INFO [train.py:904] (6/8) Epoch 18, batch 6800, loss[loss=0.2112, simple_loss=0.3023, pruned_loss=0.06011, over 16924.00 frames. ], tot_loss[loss=0.2094, simple_loss=0.2941, pruned_loss=0.06234, over 3117611.97 frames. ], batch size: 109, lr: 3.75e-03, grad_scale: 8.0 2023-04-30 18:07:04,928 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.988e+02 2.962e+02 3.412e+02 4.074e+02 1.001e+03, threshold=6.824e+02, percent-clipped=4.0 2023-04-30 18:07:05,952 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=179353.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 18:07:18,271 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=179361.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:07:36,771 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5064, 3.5418, 3.3148, 3.0176, 3.1895, 3.4720, 3.3179, 3.2985], device='cuda:6'), covar=tensor([0.0563, 0.0607, 0.0284, 0.0264, 0.0524, 0.0468, 0.1193, 0.0513], device='cuda:6'), in_proj_covar=tensor([0.0277, 0.0394, 0.0323, 0.0314, 0.0336, 0.0367, 0.0221, 0.0388], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:08:16,241 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=179398.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:08:20,457 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=179401.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:08:21,224 INFO [train.py:904] (6/8) Epoch 18, batch 6850, loss[loss=0.2565, simple_loss=0.3127, pruned_loss=0.1001, over 11747.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2955, pruned_loss=0.06326, over 3095812.67 frames. ], batch size: 248, lr: 3.75e-03, grad_scale: 8.0 2023-04-30 18:08:47,923 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-30 18:08:50,212 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=179422.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:08:56,299 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=179426.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:09:24,169 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=179444.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 18:09:34,744 INFO [train.py:904] (6/8) Epoch 18, batch 6900, loss[loss=0.2897, simple_loss=0.3418, pruned_loss=0.1187, over 11711.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2979, pruned_loss=0.06319, over 3087126.64 frames. ], batch size: 248, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:09:38,468 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.677e+02 2.748e+02 3.355e+02 3.841e+02 5.597e+02, threshold=6.710e+02, percent-clipped=0.0 2023-04-30 18:09:47,396 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1945, 4.2038, 4.5181, 4.5034, 4.5043, 4.2159, 4.2299, 4.1544], device='cuda:6'), covar=tensor([0.0328, 0.0723, 0.0450, 0.0412, 0.0459, 0.0467, 0.0900, 0.0518], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0425, 0.0414, 0.0391, 0.0464, 0.0436, 0.0534, 0.0350], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 18:10:10,836 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=179474.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:10:48,242 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0012, 5.0332, 5.4338, 5.3994, 5.3978, 5.0289, 4.9747, 4.7407], device='cuda:6'), covar=tensor([0.0301, 0.0497, 0.0325, 0.0351, 0.0442, 0.0399, 0.0966, 0.0500], device='cuda:6'), in_proj_covar=tensor([0.0393, 0.0429, 0.0417, 0.0394, 0.0468, 0.0440, 0.0539, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 18:10:53,461 INFO [train.py:904] (6/8) Epoch 18, batch 6950, loss[loss=0.2232, simple_loss=0.3041, pruned_loss=0.07111, over 15298.00 frames. ], tot_loss[loss=0.2144, simple_loss=0.2995, pruned_loss=0.06465, over 3082871.82 frames. ], batch size: 191, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:10:58,840 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=179505.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 18:11:34,856 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7156, 2.4742, 2.1670, 3.1684, 2.0482, 3.5599, 1.4268, 2.6206], device='cuda:6'), covar=tensor([0.1459, 0.0785, 0.1425, 0.0227, 0.0164, 0.0431, 0.1899, 0.0882], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0167, 0.0189, 0.0176, 0.0203, 0.0211, 0.0194, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 18:12:07,073 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=179550.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:12:09,855 INFO [train.py:904] (6/8) Epoch 18, batch 7000, loss[loss=0.2043, simple_loss=0.3054, pruned_loss=0.05155, over 16590.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.2999, pruned_loss=0.06454, over 3059432.68 frames. ], batch size: 62, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:12:12,174 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.911e+02 2.791e+02 3.383e+02 4.252e+02 7.699e+02, threshold=6.767e+02, percent-clipped=2.0 2023-04-30 18:12:56,553 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-30 18:13:16,966 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=179598.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:13:21,687 INFO [train.py:904] (6/8) Epoch 18, batch 7050, loss[loss=0.2053, simple_loss=0.2907, pruned_loss=0.05999, over 16450.00 frames. ], tot_loss[loss=0.214, simple_loss=0.2999, pruned_loss=0.06408, over 3059263.58 frames. ], batch size: 75, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:13:36,676 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5434, 3.5121, 2.7289, 2.1752, 2.4652, 2.2904, 3.8031, 3.3014], device='cuda:6'), covar=tensor([0.2933, 0.0799, 0.1792, 0.2613, 0.2407, 0.2052, 0.0461, 0.1218], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0265, 0.0300, 0.0306, 0.0292, 0.0249, 0.0287, 0.0329], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 18:14:12,917 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=179636.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:14:37,716 INFO [train.py:904] (6/8) Epoch 18, batch 7100, loss[loss=0.1998, simple_loss=0.288, pruned_loss=0.05573, over 16168.00 frames. ], tot_loss[loss=0.2128, simple_loss=0.2985, pruned_loss=0.06358, over 3065204.53 frames. ], batch size: 165, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:14:40,299 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.117e+02 2.863e+02 3.448e+02 4.199e+02 1.001e+03, threshold=6.895e+02, percent-clipped=2.0 2023-04-30 18:15:33,589 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7731, 1.2881, 1.7037, 1.6172, 1.7488, 1.9553, 1.5837, 1.7511], device='cuda:6'), covar=tensor([0.0258, 0.0386, 0.0228, 0.0278, 0.0268, 0.0173, 0.0425, 0.0131], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0189, 0.0175, 0.0178, 0.0187, 0.0145, 0.0191, 0.0141], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:15:46,859 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=179697.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:15:53,677 INFO [train.py:904] (6/8) Epoch 18, batch 7150, loss[loss=0.2104, simple_loss=0.2881, pruned_loss=0.06638, over 16605.00 frames. ], tot_loss[loss=0.2111, simple_loss=0.2962, pruned_loss=0.063, over 3077053.27 frames. ], batch size: 57, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:15:55,559 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.24 vs. limit=5.0 2023-04-30 18:16:09,051 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3464, 2.9388, 2.6565, 2.2955, 2.2688, 2.2701, 2.9052, 2.8845], device='cuda:6'), covar=tensor([0.2226, 0.0706, 0.1481, 0.2213, 0.2124, 0.1920, 0.0488, 0.1139], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0265, 0.0299, 0.0305, 0.0292, 0.0249, 0.0287, 0.0328], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 18:16:16,355 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179717.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:16:27,330 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.51 vs. limit=2.0 2023-04-30 18:17:00,427 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4789, 3.4692, 2.7224, 2.1267, 2.2482, 2.2298, 3.5931, 3.1251], device='cuda:6'), covar=tensor([0.3031, 0.0713, 0.1826, 0.2798, 0.2723, 0.2231, 0.0528, 0.1318], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0265, 0.0299, 0.0305, 0.0291, 0.0249, 0.0286, 0.0328], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 18:17:08,207 INFO [train.py:904] (6/8) Epoch 18, batch 7200, loss[loss=0.1871, simple_loss=0.2761, pruned_loss=0.04907, over 16437.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2927, pruned_loss=0.0605, over 3090672.84 frames. ], batch size: 35, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:17:10,636 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.743e+02 2.765e+02 3.366e+02 4.196e+02 7.871e+02, threshold=6.733e+02, percent-clipped=4.0 2023-04-30 18:17:29,469 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-30 18:17:30,262 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4257, 2.1351, 1.8072, 1.9386, 2.4521, 2.1054, 2.3134, 2.5840], device='cuda:6'), covar=tensor([0.0176, 0.0406, 0.0523, 0.0444, 0.0251, 0.0361, 0.0221, 0.0247], device='cuda:6'), in_proj_covar=tensor([0.0188, 0.0222, 0.0217, 0.0217, 0.0225, 0.0222, 0.0225, 0.0219], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:17:54,647 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.95 vs. limit=5.0 2023-04-30 18:18:17,993 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=179796.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:18:24,741 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179800.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 18:18:27,049 INFO [train.py:904] (6/8) Epoch 18, batch 7250, loss[loss=0.1892, simple_loss=0.2723, pruned_loss=0.0531, over 16245.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2904, pruned_loss=0.05933, over 3086361.92 frames. ], batch size: 165, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:18:29,387 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7721, 3.8850, 4.1440, 4.0999, 4.1113, 3.8742, 3.9156, 3.8562], device='cuda:6'), covar=tensor([0.0320, 0.0545, 0.0410, 0.0426, 0.0447, 0.0411, 0.0757, 0.0488], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0426, 0.0416, 0.0393, 0.0467, 0.0438, 0.0534, 0.0351], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 18:19:41,479 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0454, 2.0329, 2.6630, 2.9265, 2.8457, 3.4938, 2.1765, 3.4574], device='cuda:6'), covar=tensor([0.0180, 0.0457, 0.0260, 0.0251, 0.0267, 0.0135, 0.0488, 0.0127], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0186, 0.0173, 0.0176, 0.0185, 0.0143, 0.0189, 0.0139], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:19:42,142 INFO [train.py:904] (6/8) Epoch 18, batch 7300, loss[loss=0.1832, simple_loss=0.2742, pruned_loss=0.04613, over 17263.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2895, pruned_loss=0.05887, over 3090736.95 frames. ], batch size: 52, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:19:45,256 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.839e+02 2.793e+02 3.453e+02 4.292e+02 8.148e+02, threshold=6.907e+02, percent-clipped=1.0 2023-04-30 18:19:50,992 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=179857.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:20:33,713 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0800, 5.7176, 5.9285, 5.6635, 5.7316, 6.2241, 5.6982, 5.4531], device='cuda:6'), covar=tensor([0.0823, 0.1522, 0.1619, 0.1553, 0.1793, 0.0793, 0.1328, 0.2252], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0559, 0.0618, 0.0470, 0.0628, 0.0644, 0.0486, 0.0631], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 18:20:58,393 INFO [train.py:904] (6/8) Epoch 18, batch 7350, loss[loss=0.1993, simple_loss=0.2874, pruned_loss=0.05554, over 16166.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2907, pruned_loss=0.06011, over 3077061.84 frames. ], batch size: 165, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:21:05,532 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=179906.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 18:21:32,178 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=179924.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:22:16,601 INFO [train.py:904] (6/8) Epoch 18, batch 7400, loss[loss=0.2273, simple_loss=0.3068, pruned_loss=0.07393, over 16711.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2924, pruned_loss=0.06089, over 3078677.13 frames. ], batch size: 134, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:22:19,965 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.039e+02 2.864e+02 3.444e+02 4.186e+02 8.589e+02, threshold=6.889e+02, percent-clipped=1.0 2023-04-30 18:22:30,134 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0429, 2.3698, 1.9855, 2.1182, 2.7496, 2.4021, 2.7644, 2.9401], device='cuda:6'), covar=tensor([0.0143, 0.0405, 0.0510, 0.0470, 0.0252, 0.0395, 0.0204, 0.0253], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0222, 0.0217, 0.0217, 0.0225, 0.0222, 0.0225, 0.0218], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:22:41,141 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=179967.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:23:09,725 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=179985.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:23:21,184 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179992.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:23:39,778 INFO [train.py:904] (6/8) Epoch 18, batch 7450, loss[loss=0.1905, simple_loss=0.2806, pruned_loss=0.05021, over 16489.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2934, pruned_loss=0.06193, over 3077573.63 frames. ], batch size: 68, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:24:05,904 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180017.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:24:10,838 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3435, 3.4369, 3.7538, 1.7341, 3.9360, 3.9268, 2.9338, 2.7829], device='cuda:6'), covar=tensor([0.0886, 0.0249, 0.0210, 0.1358, 0.0077, 0.0230, 0.0429, 0.0533], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0105, 0.0094, 0.0138, 0.0075, 0.0120, 0.0124, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 18:24:19,813 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=180026.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:24:37,412 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0596, 1.9644, 2.6125, 3.0770, 2.9035, 3.4519, 2.1466, 3.4589], device='cuda:6'), covar=tensor([0.0197, 0.0508, 0.0322, 0.0236, 0.0254, 0.0144, 0.0522, 0.0108], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0187, 0.0173, 0.0175, 0.0185, 0.0143, 0.0189, 0.0139], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:24:50,309 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4455, 3.2557, 2.6391, 2.1421, 2.2244, 2.2173, 3.2911, 3.0360], device='cuda:6'), covar=tensor([0.2829, 0.0756, 0.1787, 0.2634, 0.2568, 0.2147, 0.0531, 0.1338], device='cuda:6'), in_proj_covar=tensor([0.0320, 0.0263, 0.0299, 0.0304, 0.0291, 0.0247, 0.0285, 0.0327], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 18:25:01,487 INFO [train.py:904] (6/8) Epoch 18, batch 7500, loss[loss=0.2225, simple_loss=0.2943, pruned_loss=0.07532, over 11554.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.2935, pruned_loss=0.0612, over 3059135.09 frames. ], batch size: 246, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:25:04,519 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.958e+02 3.002e+02 3.628e+02 4.895e+02 8.341e+02, threshold=7.256e+02, percent-clipped=3.0 2023-04-30 18:25:22,816 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180065.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:25:33,712 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9950, 4.0669, 3.8961, 3.6482, 3.6217, 3.9867, 3.6642, 3.7433], device='cuda:6'), covar=tensor([0.0612, 0.0542, 0.0291, 0.0292, 0.0777, 0.0462, 0.1111, 0.0657], device='cuda:6'), in_proj_covar=tensor([0.0272, 0.0387, 0.0317, 0.0309, 0.0331, 0.0360, 0.0218, 0.0381], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:25:56,694 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=180087.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:26:16,488 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180100.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:26:19,023 INFO [train.py:904] (6/8) Epoch 18, batch 7550, loss[loss=0.2068, simple_loss=0.2845, pruned_loss=0.06458, over 17028.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.293, pruned_loss=0.06222, over 3028520.82 frames. ], batch size: 55, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:26:47,544 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.24 vs. limit=5.0 2023-04-30 18:27:18,248 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5379, 4.7358, 4.8829, 4.6931, 4.7083, 5.2630, 4.6856, 4.4596], device='cuda:6'), covar=tensor([0.1234, 0.1724, 0.2257, 0.1917, 0.2366, 0.1035, 0.1797, 0.2568], device='cuda:6'), in_proj_covar=tensor([0.0387, 0.0559, 0.0619, 0.0469, 0.0627, 0.0644, 0.0488, 0.0631], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 18:27:30,331 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180148.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 18:27:36,545 INFO [train.py:904] (6/8) Epoch 18, batch 7600, loss[loss=0.1923, simple_loss=0.2783, pruned_loss=0.05321, over 17252.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2915, pruned_loss=0.06146, over 3051945.09 frames. ], batch size: 52, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:27:37,483 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180152.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:27:39,406 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.837e+02 2.698e+02 3.315e+02 4.368e+02 9.372e+02, threshold=6.630e+02, percent-clipped=3.0 2023-04-30 18:28:55,649 INFO [train.py:904] (6/8) Epoch 18, batch 7650, loss[loss=0.2167, simple_loss=0.3036, pruned_loss=0.06491, over 16760.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.2924, pruned_loss=0.06264, over 3053169.70 frames. ], batch size: 124, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:29:00,838 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=180205.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:29:05,858 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0492, 2.3910, 2.3530, 2.9024, 1.9801, 3.2131, 1.7902, 2.6932], device='cuda:6'), covar=tensor([0.1155, 0.0620, 0.1010, 0.0188, 0.0113, 0.0366, 0.1454, 0.0695], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0170, 0.0191, 0.0178, 0.0206, 0.0214, 0.0196, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 18:30:13,287 INFO [train.py:904] (6/8) Epoch 18, batch 7700, loss[loss=0.1937, simple_loss=0.2791, pruned_loss=0.05418, over 16417.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2924, pruned_loss=0.0631, over 3041494.88 frames. ], batch size: 146, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:30:18,205 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.849e+02 3.108e+02 3.616e+02 4.495e+02 6.527e+02, threshold=7.232e+02, percent-clipped=0.0 2023-04-30 18:30:20,092 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5618, 3.7057, 3.9255, 1.8686, 4.0902, 4.2049, 3.1916, 3.0570], device='cuda:6'), covar=tensor([0.0918, 0.0230, 0.0223, 0.1424, 0.0112, 0.0168, 0.0431, 0.0493], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0105, 0.0094, 0.0139, 0.0075, 0.0121, 0.0125, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 18:30:29,288 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180262.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 18:30:35,622 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=180266.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:30:57,473 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180280.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:31:02,300 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-04-30 18:31:15,950 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180292.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:31:30,812 INFO [train.py:904] (6/8) Epoch 18, batch 7750, loss[loss=0.1957, simple_loss=0.2885, pruned_loss=0.05141, over 16462.00 frames. ], tot_loss[loss=0.209, simple_loss=0.2926, pruned_loss=0.06275, over 3049331.50 frames. ], batch size: 75, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:31:31,623 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 18:31:43,391 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4287, 2.9385, 2.6818, 2.2850, 2.2643, 2.3023, 2.9723, 2.9006], device='cuda:6'), covar=tensor([0.2324, 0.0763, 0.1623, 0.2456, 0.2249, 0.2018, 0.0503, 0.1280], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0264, 0.0300, 0.0306, 0.0292, 0.0249, 0.0287, 0.0328], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 18:32:23,807 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=180336.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:32:29,343 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180340.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:32:46,606 INFO [train.py:904] (6/8) Epoch 18, batch 7800, loss[loss=0.192, simple_loss=0.284, pruned_loss=0.05001, over 16856.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2937, pruned_loss=0.0636, over 3058155.60 frames. ], batch size: 76, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:32:51,024 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.798e+02 2.996e+02 3.455e+02 4.260e+02 9.369e+02, threshold=6.911e+02, percent-clipped=1.0 2023-04-30 18:32:55,365 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2571, 5.2949, 5.1120, 4.2710, 5.1113, 1.8516, 4.8514, 4.9723], device='cuda:6'), covar=tensor([0.0134, 0.0123, 0.0219, 0.0522, 0.0135, 0.2789, 0.0218, 0.0220], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0138, 0.0184, 0.0168, 0.0159, 0.0195, 0.0173, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:33:33,618 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180382.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:33:55,570 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=180397.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:33:56,909 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1850, 4.0011, 4.4815, 2.2910, 4.7489, 4.7571, 3.4324, 3.5538], device='cuda:6'), covar=tensor([0.0626, 0.0211, 0.0165, 0.1096, 0.0047, 0.0120, 0.0355, 0.0384], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0105, 0.0094, 0.0138, 0.0075, 0.0120, 0.0124, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 18:34:02,022 INFO [train.py:904] (6/8) Epoch 18, batch 7850, loss[loss=0.2045, simple_loss=0.3087, pruned_loss=0.05015, over 16852.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2951, pruned_loss=0.06345, over 3055533.99 frames. ], batch size: 96, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:34:10,433 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.7250, 6.0652, 5.6796, 5.8328, 5.4124, 5.3114, 5.3960, 6.1621], device='cuda:6'), covar=tensor([0.1119, 0.0736, 0.1097, 0.0829, 0.0858, 0.0659, 0.1162, 0.0897], device='cuda:6'), in_proj_covar=tensor([0.0618, 0.0760, 0.0624, 0.0569, 0.0475, 0.0490, 0.0636, 0.0591], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:34:59,673 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8431, 5.1291, 4.8716, 4.9181, 4.6859, 4.6091, 4.4926, 5.2163], device='cuda:6'), covar=tensor([0.1166, 0.0786, 0.0976, 0.0806, 0.0732, 0.0958, 0.1161, 0.0880], device='cuda:6'), in_proj_covar=tensor([0.0619, 0.0761, 0.0626, 0.0570, 0.0475, 0.0490, 0.0637, 0.0591], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:35:16,217 INFO [train.py:904] (6/8) Epoch 18, batch 7900, loss[loss=0.1958, simple_loss=0.2927, pruned_loss=0.04945, over 16902.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2936, pruned_loss=0.06213, over 3074705.67 frames. ], batch size: 90, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:35:17,229 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180452.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:35:20,362 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.862e+02 2.825e+02 3.445e+02 4.253e+02 7.547e+02, threshold=6.890e+02, percent-clipped=0.0 2023-04-30 18:36:25,776 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 18:36:32,646 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180500.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:36:34,744 INFO [train.py:904] (6/8) Epoch 18, batch 7950, loss[loss=0.1913, simple_loss=0.2687, pruned_loss=0.0569, over 17070.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2942, pruned_loss=0.06272, over 3086948.49 frames. ], batch size: 53, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:37:35,932 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-30 18:37:52,805 INFO [train.py:904] (6/8) Epoch 18, batch 8000, loss[loss=0.2819, simple_loss=0.3346, pruned_loss=0.1146, over 11707.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2951, pruned_loss=0.06345, over 3097335.45 frames. ], batch size: 247, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:37:57,084 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.803e+02 2.710e+02 3.198e+02 3.720e+02 8.207e+02, threshold=6.397e+02, percent-clipped=2.0 2023-04-30 18:38:07,695 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180561.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:38:08,933 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180562.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 18:38:21,356 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=180570.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:38:36,571 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180580.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:39:10,314 INFO [train.py:904] (6/8) Epoch 18, batch 8050, loss[loss=0.2182, simple_loss=0.3029, pruned_loss=0.06672, over 16877.00 frames. ], tot_loss[loss=0.2105, simple_loss=0.295, pruned_loss=0.06303, over 3091843.24 frames. ], batch size: 90, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:39:11,892 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 18:39:23,126 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180610.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:39:50,350 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180628.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:39:54,564 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=180631.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:40:09,064 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2216, 2.4943, 2.0273, 2.2318, 2.8256, 2.4959, 2.9485, 3.0369], device='cuda:6'), covar=tensor([0.0128, 0.0372, 0.0513, 0.0437, 0.0237, 0.0372, 0.0215, 0.0236], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0222, 0.0217, 0.0217, 0.0225, 0.0222, 0.0225, 0.0218], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:40:26,588 INFO [train.py:904] (6/8) Epoch 18, batch 8100, loss[loss=0.1901, simple_loss=0.275, pruned_loss=0.05261, over 17252.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2948, pruned_loss=0.06259, over 3085201.56 frames. ], batch size: 52, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:40:32,029 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.097e+02 2.839e+02 3.233e+02 3.954e+02 8.532e+02, threshold=6.466e+02, percent-clipped=3.0 2023-04-30 18:40:42,081 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4878, 4.4913, 4.3733, 3.6011, 4.4052, 1.6117, 4.1659, 4.0573], device='cuda:6'), covar=tensor([0.0092, 0.0080, 0.0185, 0.0340, 0.0094, 0.2776, 0.0138, 0.0238], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0139, 0.0185, 0.0170, 0.0160, 0.0196, 0.0174, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:41:00,902 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6305, 3.6318, 2.8668, 2.2753, 2.4892, 2.4186, 3.9839, 3.2775], device='cuda:6'), covar=tensor([0.3128, 0.0932, 0.1964, 0.2824, 0.2902, 0.2069, 0.0563, 0.1490], device='cuda:6'), in_proj_covar=tensor([0.0323, 0.0265, 0.0301, 0.0306, 0.0294, 0.0250, 0.0288, 0.0328], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 18:41:12,133 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180682.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:41:26,939 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180692.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:41:41,226 INFO [train.py:904] (6/8) Epoch 18, batch 8150, loss[loss=0.1757, simple_loss=0.2642, pruned_loss=0.04365, over 16911.00 frames. ], tot_loss[loss=0.2082, simple_loss=0.2925, pruned_loss=0.06198, over 3081469.66 frames. ], batch size: 96, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:42:03,563 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.09 vs. limit=5.0 2023-04-30 18:42:24,164 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180730.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:42:25,736 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.15 vs. limit=5.0 2023-04-30 18:42:26,668 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=180732.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:42:56,045 INFO [train.py:904] (6/8) Epoch 18, batch 8200, loss[loss=0.1828, simple_loss=0.2733, pruned_loss=0.04619, over 16735.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2899, pruned_loss=0.06099, over 3096645.74 frames. ], batch size: 76, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:43:02,072 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.971e+02 2.834e+02 3.334e+02 4.232e+02 1.685e+03, threshold=6.669e+02, percent-clipped=1.0 2023-04-30 18:43:12,712 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3399, 1.6515, 2.0725, 2.3072, 2.3898, 2.5794, 1.8676, 2.4845], device='cuda:6'), covar=tensor([0.0197, 0.0468, 0.0292, 0.0311, 0.0287, 0.0182, 0.0426, 0.0140], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0187, 0.0173, 0.0175, 0.0184, 0.0143, 0.0188, 0.0139], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:43:51,645 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5590, 3.6066, 3.3976, 3.1424, 3.2092, 3.5230, 3.3212, 3.3532], device='cuda:6'), covar=tensor([0.0586, 0.0574, 0.0288, 0.0251, 0.0537, 0.0417, 0.1149, 0.0471], device='cuda:6'), in_proj_covar=tensor([0.0275, 0.0392, 0.0320, 0.0310, 0.0332, 0.0362, 0.0220, 0.0382], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:44:01,962 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=180793.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:44:04,029 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5649, 3.6084, 3.3970, 3.1373, 3.1956, 3.5211, 3.3188, 3.3594], device='cuda:6'), covar=tensor([0.0588, 0.0548, 0.0284, 0.0261, 0.0539, 0.0426, 0.1272, 0.0496], device='cuda:6'), in_proj_covar=tensor([0.0275, 0.0392, 0.0320, 0.0310, 0.0332, 0.0362, 0.0220, 0.0382], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:44:10,027 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.42 vs. limit=5.0 2023-04-30 18:44:14,882 INFO [train.py:904] (6/8) Epoch 18, batch 8250, loss[loss=0.1867, simple_loss=0.2907, pruned_loss=0.04133, over 16659.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2885, pruned_loss=0.05829, over 3089934.26 frames. ], batch size: 134, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:44:44,115 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7743, 3.1265, 3.4466, 1.9815, 2.9135, 2.1682, 3.3972, 3.3027], device='cuda:6'), covar=tensor([0.0263, 0.0824, 0.0527, 0.2128, 0.0802, 0.1051, 0.0650, 0.0927], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0158, 0.0163, 0.0149, 0.0141, 0.0127, 0.0140, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 18:44:53,336 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=180825.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:45:23,855 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5492, 3.5447, 3.4908, 2.6693, 3.3654, 1.9958, 3.1786, 2.8709], device='cuda:6'), covar=tensor([0.0141, 0.0113, 0.0160, 0.0210, 0.0110, 0.2334, 0.0128, 0.0197], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0139, 0.0184, 0.0169, 0.0159, 0.0195, 0.0173, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:45:37,334 INFO [train.py:904] (6/8) Epoch 18, batch 8300, loss[loss=0.1766, simple_loss=0.2707, pruned_loss=0.04125, over 16700.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2859, pruned_loss=0.05519, over 3091091.30 frames. ], batch size: 57, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:45:43,843 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.692e+02 2.270e+02 2.749e+02 3.299e+02 7.574e+02, threshold=5.499e+02, percent-clipped=1.0 2023-04-30 18:45:52,511 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180861.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:46:32,572 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=180886.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:46:58,357 INFO [train.py:904] (6/8) Epoch 18, batch 8350, loss[loss=0.1817, simple_loss=0.2757, pruned_loss=0.04389, over 15272.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2856, pruned_loss=0.05318, over 3103502.16 frames. ], batch size: 191, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:47:09,894 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180909.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:47:36,514 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180926.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:47:39,771 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9370, 3.7887, 3.9861, 4.1246, 4.2123, 3.8288, 4.1903, 4.2473], device='cuda:6'), covar=tensor([0.1783, 0.1386, 0.1589, 0.0806, 0.0684, 0.1622, 0.0764, 0.0802], device='cuda:6'), in_proj_covar=tensor([0.0590, 0.0729, 0.0853, 0.0744, 0.0556, 0.0587, 0.0606, 0.0701], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:48:16,300 INFO [train.py:904] (6/8) Epoch 18, batch 8400, loss[loss=0.1682, simple_loss=0.2616, pruned_loss=0.03746, over 15378.00 frames. ], tot_loss[loss=0.1924, simple_loss=0.2827, pruned_loss=0.05106, over 3104103.96 frames. ], batch size: 191, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 18:48:22,150 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.789e+02 2.220e+02 2.634e+02 3.245e+02 6.969e+02, threshold=5.268e+02, percent-clipped=3.0 2023-04-30 18:48:43,639 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6405, 3.7479, 3.8254, 2.9625, 3.3787, 3.7750, 3.5815, 2.2935], device='cuda:6'), covar=tensor([0.0399, 0.0054, 0.0042, 0.0255, 0.0104, 0.0083, 0.0072, 0.0417], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0078, 0.0078, 0.0131, 0.0092, 0.0103, 0.0090, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 18:49:16,481 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180992.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:49:24,228 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3697, 3.0224, 3.1483, 1.8874, 3.3184, 3.3095, 2.8326, 2.7429], device='cuda:6'), covar=tensor([0.0641, 0.0236, 0.0222, 0.1040, 0.0075, 0.0192, 0.0389, 0.0382], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0103, 0.0091, 0.0135, 0.0073, 0.0117, 0.0121, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 18:49:31,280 INFO [train.py:904] (6/8) Epoch 18, batch 8450, loss[loss=0.1737, simple_loss=0.2606, pruned_loss=0.04345, over 12290.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2809, pruned_loss=0.04966, over 3085232.65 frames. ], batch size: 248, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 18:49:59,142 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 18:50:24,266 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0917, 3.1134, 1.9451, 3.3214, 2.3805, 3.3435, 2.1604, 2.6264], device='cuda:6'), covar=tensor([0.0260, 0.0342, 0.1435, 0.0227, 0.0754, 0.0448, 0.1420, 0.0687], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0169, 0.0187, 0.0147, 0.0170, 0.0205, 0.0195, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 18:50:31,761 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=181040.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:50:50,496 INFO [train.py:904] (6/8) Epoch 18, batch 8500, loss[loss=0.1628, simple_loss=0.2564, pruned_loss=0.03463, over 16471.00 frames. ], tot_loss[loss=0.186, simple_loss=0.277, pruned_loss=0.04753, over 3063307.07 frames. ], batch size: 75, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:50:58,701 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.304e+02 3.083e+02 3.822e+02 7.945e+02, threshold=6.166e+02, percent-clipped=7.0 2023-04-30 18:51:49,227 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=181088.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:51:57,532 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5448, 3.4482, 2.6888, 2.1353, 2.1860, 2.3632, 3.6460, 3.1364], device='cuda:6'), covar=tensor([0.2892, 0.0737, 0.1794, 0.2937, 0.2955, 0.2085, 0.0483, 0.1331], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0256, 0.0291, 0.0297, 0.0284, 0.0242, 0.0279, 0.0318], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 18:52:13,295 INFO [train.py:904] (6/8) Epoch 18, batch 8550, loss[loss=0.1731, simple_loss=0.2582, pruned_loss=0.04398, over 12317.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2743, pruned_loss=0.04585, over 3055536.81 frames. ], batch size: 248, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:53:50,705 INFO [train.py:904] (6/8) Epoch 18, batch 8600, loss[loss=0.1659, simple_loss=0.2568, pruned_loss=0.03757, over 16656.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2739, pruned_loss=0.04463, over 3057318.86 frames. ], batch size: 62, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:54:01,233 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.718e+02 2.216e+02 2.624e+02 3.286e+02 8.410e+02, threshold=5.248e+02, percent-clipped=1.0 2023-04-30 18:54:41,194 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4020, 1.6144, 2.0089, 2.3415, 2.3718, 2.6250, 1.8254, 2.5653], device='cuda:6'), covar=tensor([0.0215, 0.0516, 0.0308, 0.0312, 0.0296, 0.0178, 0.0491, 0.0147], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0185, 0.0170, 0.0172, 0.0182, 0.0140, 0.0186, 0.0136], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 18:54:49,175 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-30 18:54:50,091 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=181181.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:55:29,852 INFO [train.py:904] (6/8) Epoch 18, batch 8650, loss[loss=0.1737, simple_loss=0.2652, pruned_loss=0.04112, over 12181.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.272, pruned_loss=0.04361, over 3037089.24 frames. ], batch size: 247, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:56:18,769 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9186, 2.3231, 2.3148, 2.9113, 1.8833, 3.3331, 1.6794, 2.7806], device='cuda:6'), covar=tensor([0.1295, 0.0623, 0.1080, 0.0154, 0.0087, 0.0355, 0.1491, 0.0670], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0167, 0.0189, 0.0175, 0.0201, 0.0211, 0.0193, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 18:56:25,678 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=181226.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:57:15,662 INFO [train.py:904] (6/8) Epoch 18, batch 8700, loss[loss=0.1585, simple_loss=0.2538, pruned_loss=0.0316, over 16916.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2695, pruned_loss=0.04243, over 3053409.59 frames. ], batch size: 90, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:57:25,074 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.665e+02 2.245e+02 2.629e+02 3.109e+02 5.522e+02, threshold=5.258e+02, percent-clipped=1.0 2023-04-30 18:57:56,765 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=181274.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:58:39,642 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=181297.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:58:48,987 INFO [train.py:904] (6/8) Epoch 18, batch 8750, loss[loss=0.2064, simple_loss=0.2992, pruned_loss=0.05683, over 16344.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2697, pruned_loss=0.04231, over 3042133.92 frames. ], batch size: 146, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:59:04,426 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5807, 3.6379, 3.4116, 3.0719, 3.2564, 3.5546, 3.3094, 3.3867], device='cuda:6'), covar=tensor([0.0574, 0.0527, 0.0263, 0.0254, 0.0510, 0.0444, 0.1268, 0.0441], device='cuda:6'), in_proj_covar=tensor([0.0270, 0.0384, 0.0315, 0.0305, 0.0324, 0.0357, 0.0218, 0.0376], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 19:00:06,912 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-30 19:00:41,584 INFO [train.py:904] (6/8) Epoch 18, batch 8800, loss[loss=0.18, simple_loss=0.2756, pruned_loss=0.04222, over 16231.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2682, pruned_loss=0.04118, over 3061705.51 frames. ], batch size: 165, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 19:00:51,211 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.344e+02 2.185e+02 2.552e+02 3.082e+02 8.545e+02, threshold=5.104e+02, percent-clipped=2.0 2023-04-30 19:00:54,117 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=181358.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:01:57,981 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=181388.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:02:27,190 INFO [train.py:904] (6/8) Epoch 18, batch 8850, loss[loss=0.1875, simple_loss=0.2916, pruned_loss=0.04173, over 16816.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2702, pruned_loss=0.04035, over 3057652.26 frames. ], batch size: 124, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 19:03:01,480 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6050, 3.8546, 2.7152, 2.1983, 2.3939, 2.3569, 4.0888, 3.2853], device='cuda:6'), covar=tensor([0.2838, 0.0617, 0.1899, 0.2848, 0.2788, 0.2028, 0.0380, 0.1260], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0255, 0.0291, 0.0294, 0.0280, 0.0241, 0.0278, 0.0315], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 19:03:14,700 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7582, 4.5350, 4.7971, 4.9411, 5.0738, 4.5201, 5.0737, 5.0974], device='cuda:6'), covar=tensor([0.1799, 0.1131, 0.1533, 0.0652, 0.0513, 0.0873, 0.0513, 0.0515], device='cuda:6'), in_proj_covar=tensor([0.0575, 0.0709, 0.0830, 0.0727, 0.0541, 0.0574, 0.0588, 0.0679], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 19:03:44,883 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=181436.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:04:15,168 INFO [train.py:904] (6/8) Epoch 18, batch 8900, loss[loss=0.1656, simple_loss=0.2666, pruned_loss=0.03226, over 16656.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2706, pruned_loss=0.03969, over 3048356.88 frames. ], batch size: 89, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 19:04:25,752 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.184e+02 2.557e+02 3.261e+02 8.427e+02, threshold=5.113e+02, percent-clipped=4.0 2023-04-30 19:04:47,905 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-30 19:05:23,238 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=181481.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:06:18,288 INFO [train.py:904] (6/8) Epoch 18, batch 8950, loss[loss=0.1621, simple_loss=0.2579, pruned_loss=0.03321, over 16249.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2704, pruned_loss=0.03998, over 3066354.51 frames. ], batch size: 165, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 19:07:17,084 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=181529.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:07:19,906 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=181530.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:07:41,462 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5728, 3.5681, 3.5483, 2.7733, 3.4703, 2.0210, 3.2805, 3.0357], device='cuda:6'), covar=tensor([0.0116, 0.0097, 0.0136, 0.0162, 0.0091, 0.2191, 0.0126, 0.0210], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0136, 0.0178, 0.0163, 0.0156, 0.0192, 0.0170, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 19:08:03,982 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7190, 4.0580, 2.9797, 2.3009, 2.5580, 2.4414, 4.2818, 3.3719], device='cuda:6'), covar=tensor([0.2752, 0.0559, 0.1705, 0.2744, 0.2822, 0.2031, 0.0383, 0.1342], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0256, 0.0291, 0.0295, 0.0280, 0.0242, 0.0279, 0.0316], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 19:08:08,280 INFO [train.py:904] (6/8) Epoch 18, batch 9000, loss[loss=0.1769, simple_loss=0.2767, pruned_loss=0.03857, over 16424.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2675, pruned_loss=0.0385, over 3093720.25 frames. ], batch size: 147, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 19:08:08,280 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 19:08:17,835 INFO [train.py:938] (6/8) Epoch 18, validation: loss=0.1475, simple_loss=0.2516, pruned_loss=0.02169, over 944034.00 frames. 2023-04-30 19:08:17,836 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 19:08:27,942 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.232e+02 2.270e+02 2.606e+02 3.112e+02 6.247e+02, threshold=5.212e+02, percent-clipped=1.0 2023-04-30 19:09:40,019 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=181591.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:10:01,194 INFO [train.py:904] (6/8) Epoch 18, batch 9050, loss[loss=0.1627, simple_loss=0.2495, pruned_loss=0.03801, over 16265.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2682, pruned_loss=0.03893, over 3097441.15 frames. ], batch size: 166, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:11:46,764 INFO [train.py:904] (6/8) Epoch 18, batch 9100, loss[loss=0.1682, simple_loss=0.257, pruned_loss=0.03965, over 12243.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2682, pruned_loss=0.03947, over 3089338.03 frames. ], batch size: 247, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:11:50,071 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=181653.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:11:55,852 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.522e+02 2.401e+02 2.871e+02 3.524e+02 6.480e+02, threshold=5.743e+02, percent-clipped=6.0 2023-04-30 19:13:43,587 INFO [train.py:904] (6/8) Epoch 18, batch 9150, loss[loss=0.173, simple_loss=0.2609, pruned_loss=0.04252, over 16477.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2685, pruned_loss=0.0389, over 3090090.79 frames. ], batch size: 147, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:13:51,982 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9519, 1.7846, 1.6249, 1.3665, 1.9379, 1.5538, 1.5060, 1.8418], device='cuda:6'), covar=tensor([0.0150, 0.0287, 0.0385, 0.0378, 0.0211, 0.0267, 0.0139, 0.0184], device='cuda:6'), in_proj_covar=tensor([0.0183, 0.0221, 0.0215, 0.0215, 0.0222, 0.0220, 0.0221, 0.0213], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 19:15:27,777 INFO [train.py:904] (6/8) Epoch 18, batch 9200, loss[loss=0.1706, simple_loss=0.2668, pruned_loss=0.03723, over 16677.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2643, pruned_loss=0.03828, over 3084154.84 frames. ], batch size: 134, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:15:36,945 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.469e+02 2.118e+02 2.449e+02 2.968e+02 5.097e+02, threshold=4.898e+02, percent-clipped=0.0 2023-04-30 19:15:44,506 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9836, 4.0898, 3.8805, 3.5841, 3.6302, 4.0309, 3.6387, 3.7473], device='cuda:6'), covar=tensor([0.0666, 0.0677, 0.0352, 0.0342, 0.0789, 0.0626, 0.1022, 0.0660], device='cuda:6'), in_proj_covar=tensor([0.0266, 0.0379, 0.0313, 0.0301, 0.0318, 0.0351, 0.0217, 0.0372], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 19:16:19,385 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9005, 4.9757, 5.3269, 5.2849, 5.3194, 5.0012, 4.9226, 4.7092], device='cuda:6'), covar=tensor([0.0283, 0.0463, 0.0381, 0.0422, 0.0349, 0.0348, 0.0842, 0.0409], device='cuda:6'), in_proj_covar=tensor([0.0371, 0.0402, 0.0395, 0.0372, 0.0439, 0.0415, 0.0504, 0.0334], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 19:17:05,940 INFO [train.py:904] (6/8) Epoch 18, batch 9250, loss[loss=0.1457, simple_loss=0.2283, pruned_loss=0.03159, over 12292.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2641, pruned_loss=0.03845, over 3087901.79 frames. ], batch size: 248, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:18:57,072 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-30 19:18:57,425 INFO [train.py:904] (6/8) Epoch 18, batch 9300, loss[loss=0.1727, simple_loss=0.256, pruned_loss=0.04472, over 12040.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2627, pruned_loss=0.03836, over 3069760.93 frames. ], batch size: 250, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:19:06,077 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.346e+02 2.677e+02 3.465e+02 6.012e+02, threshold=5.355e+02, percent-clipped=4.0 2023-04-30 19:20:12,871 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=181886.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:20:41,433 INFO [train.py:904] (6/8) Epoch 18, batch 9350, loss[loss=0.1728, simple_loss=0.2571, pruned_loss=0.0443, over 12718.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2625, pruned_loss=0.03835, over 3070059.28 frames. ], batch size: 250, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:21:42,754 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0985, 1.4240, 1.9004, 2.0177, 2.1840, 2.3284, 1.6607, 2.2323], device='cuda:6'), covar=tensor([0.0257, 0.0501, 0.0292, 0.0332, 0.0333, 0.0214, 0.0497, 0.0132], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0183, 0.0170, 0.0170, 0.0182, 0.0139, 0.0185, 0.0134], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 19:22:24,029 INFO [train.py:904] (6/8) Epoch 18, batch 9400, loss[loss=0.1516, simple_loss=0.2412, pruned_loss=0.03103, over 12882.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2624, pruned_loss=0.03812, over 3067617.03 frames. ], batch size: 248, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:22:28,041 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=181953.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:22:28,358 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.68 vs. limit=5.0 2023-04-30 19:22:33,105 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.435e+02 2.111e+02 2.455e+02 3.021e+02 7.291e+02, threshold=4.911e+02, percent-clipped=2.0 2023-04-30 19:22:58,093 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-30 19:23:16,339 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=181978.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:23:31,281 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1222, 2.5820, 2.7027, 1.8913, 2.8480, 2.8709, 2.5046, 2.4469], device='cuda:6'), covar=tensor([0.0614, 0.0239, 0.0206, 0.0939, 0.0089, 0.0223, 0.0428, 0.0413], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0101, 0.0089, 0.0134, 0.0072, 0.0114, 0.0120, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 19:24:05,243 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=182001.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:24:06,043 INFO [train.py:904] (6/8) Epoch 18, batch 9450, loss[loss=0.1471, simple_loss=0.2399, pruned_loss=0.02717, over 16584.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2644, pruned_loss=0.03811, over 3076511.32 frames. ], batch size: 62, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:25:19,200 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=182039.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:25:39,764 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7520, 2.2068, 1.8657, 1.9487, 2.5243, 2.2421, 2.2931, 2.6162], device='cuda:6'), covar=tensor([0.0174, 0.0404, 0.0510, 0.0460, 0.0254, 0.0320, 0.0194, 0.0257], device='cuda:6'), in_proj_covar=tensor([0.0183, 0.0221, 0.0215, 0.0215, 0.0222, 0.0220, 0.0219, 0.0213], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 19:25:41,959 INFO [train.py:904] (6/8) Epoch 18, batch 9500, loss[loss=0.159, simple_loss=0.2678, pruned_loss=0.02517, over 16859.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2632, pruned_loss=0.03746, over 3081266.94 frames. ], batch size: 102, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:25:55,088 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.398e+02 2.182e+02 2.603e+02 3.134e+02 5.554e+02, threshold=5.207e+02, percent-clipped=1.0 2023-04-30 19:26:14,270 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5426, 3.5301, 3.5242, 2.8166, 3.4598, 1.9846, 3.2419, 2.8957], device='cuda:6'), covar=tensor([0.0124, 0.0123, 0.0161, 0.0171, 0.0099, 0.2353, 0.0113, 0.0224], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0135, 0.0177, 0.0160, 0.0154, 0.0191, 0.0167, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 19:27:04,341 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-30 19:27:27,071 INFO [train.py:904] (6/8) Epoch 18, batch 9550, loss[loss=0.1583, simple_loss=0.2505, pruned_loss=0.03303, over 12291.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2628, pruned_loss=0.03729, over 3086634.08 frames. ], batch size: 247, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:27:47,589 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-30 19:29:06,814 INFO [train.py:904] (6/8) Epoch 18, batch 9600, loss[loss=0.1775, simple_loss=0.2785, pruned_loss=0.0382, over 15553.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2646, pruned_loss=0.03789, over 3082302.35 frames. ], batch size: 191, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:29:15,380 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.236e+02 2.461e+02 2.915e+02 3.491e+02 6.374e+02, threshold=5.830e+02, percent-clipped=5.0 2023-04-30 19:30:14,124 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=182186.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:30:52,020 INFO [train.py:904] (6/8) Epoch 18, batch 9650, loss[loss=0.1671, simple_loss=0.2736, pruned_loss=0.03031, over 16857.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2667, pruned_loss=0.03839, over 3091660.85 frames. ], batch size: 102, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:31:40,521 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7065, 1.6504, 2.2255, 2.5287, 2.4663, 2.8330, 1.5008, 2.8471], device='cuda:6'), covar=tensor([0.0173, 0.0556, 0.0300, 0.0288, 0.0307, 0.0180, 0.0724, 0.0136], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0180, 0.0167, 0.0168, 0.0179, 0.0137, 0.0183, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 19:31:59,408 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2080, 4.3499, 4.4812, 4.2828, 4.3662, 4.8578, 4.3748, 4.0604], device='cuda:6'), covar=tensor([0.1678, 0.1824, 0.2117, 0.2001, 0.2611, 0.0963, 0.1502, 0.2474], device='cuda:6'), in_proj_covar=tensor([0.0366, 0.0530, 0.0589, 0.0447, 0.0594, 0.0617, 0.0466, 0.0596], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 19:32:00,673 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=182234.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:32:37,737 INFO [train.py:904] (6/8) Epoch 18, batch 9700, loss[loss=0.1634, simple_loss=0.256, pruned_loss=0.03538, over 17032.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2661, pruned_loss=0.03859, over 3089562.99 frames. ], batch size: 55, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:32:45,807 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.506e+02 2.275e+02 2.741e+02 3.246e+02 5.302e+02, threshold=5.483e+02, percent-clipped=0.0 2023-04-30 19:33:37,410 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=182281.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:33:50,418 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2920, 5.3447, 5.1377, 4.6663, 4.8027, 5.2435, 5.1707, 4.8732], device='cuda:6'), covar=tensor([0.0607, 0.0631, 0.0286, 0.0314, 0.0984, 0.0556, 0.0245, 0.0707], device='cuda:6'), in_proj_covar=tensor([0.0264, 0.0373, 0.0310, 0.0297, 0.0316, 0.0347, 0.0214, 0.0367], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-04-30 19:34:17,896 INFO [train.py:904] (6/8) Epoch 18, batch 9750, loss[loss=0.1576, simple_loss=0.2589, pruned_loss=0.02812, over 15376.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2647, pruned_loss=0.03854, over 3087926.40 frames. ], batch size: 191, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:35:19,766 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0730, 3.3725, 3.5883, 2.1269, 3.1004, 2.2735, 3.5476, 3.4735], device='cuda:6'), covar=tensor([0.0273, 0.0810, 0.0516, 0.1969, 0.0731, 0.1045, 0.0645, 0.0977], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0151, 0.0160, 0.0147, 0.0138, 0.0124, 0.0137, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 19:35:24,365 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=182334.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:35:27,465 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.22 vs. limit=5.0 2023-04-30 19:35:39,145 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=182342.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:35:56,048 INFO [train.py:904] (6/8) Epoch 18, batch 9800, loss[loss=0.1711, simple_loss=0.2777, pruned_loss=0.03222, over 16303.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.264, pruned_loss=0.0377, over 3072176.88 frames. ], batch size: 165, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:36:05,429 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 1.961e+02 2.421e+02 2.847e+02 6.201e+02, threshold=4.843e+02, percent-clipped=1.0 2023-04-30 19:37:39,078 INFO [train.py:904] (6/8) Epoch 18, batch 9850, loss[loss=0.1598, simple_loss=0.2485, pruned_loss=0.03557, over 12381.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2648, pruned_loss=0.03749, over 3057906.22 frames. ], batch size: 250, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:39:29,930 INFO [train.py:904] (6/8) Epoch 18, batch 9900, loss[loss=0.181, simple_loss=0.278, pruned_loss=0.04202, over 16296.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2651, pruned_loss=0.03719, over 3063915.14 frames. ], batch size: 166, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:39:40,680 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.546e+02 2.082e+02 2.451e+02 2.864e+02 7.377e+02, threshold=4.903e+02, percent-clipped=2.0 2023-04-30 19:39:41,951 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2345, 4.3737, 4.4701, 4.2423, 4.3117, 4.8126, 4.4309, 4.1616], device='cuda:6'), covar=tensor([0.1618, 0.1635, 0.1911, 0.2106, 0.2412, 0.1027, 0.1403, 0.2300], device='cuda:6'), in_proj_covar=tensor([0.0360, 0.0524, 0.0581, 0.0442, 0.0586, 0.0612, 0.0459, 0.0589], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 19:40:07,187 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8316, 5.0941, 4.8859, 4.8683, 4.6565, 4.6194, 4.6263, 5.1915], device='cuda:6'), covar=tensor([0.1143, 0.0917, 0.1113, 0.0834, 0.0827, 0.1041, 0.1120, 0.1002], device='cuda:6'), in_proj_covar=tensor([0.0608, 0.0746, 0.0609, 0.0554, 0.0468, 0.0481, 0.0619, 0.0577], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 19:40:57,147 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3545, 3.4343, 3.6622, 3.6208, 3.6455, 3.4742, 3.4904, 3.5366], device='cuda:6'), covar=tensor([0.0396, 0.0774, 0.0493, 0.0491, 0.0548, 0.0601, 0.0865, 0.0473], device='cuda:6'), in_proj_covar=tensor([0.0360, 0.0391, 0.0383, 0.0363, 0.0428, 0.0403, 0.0489, 0.0325], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 19:41:28,490 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-30 19:41:28,673 INFO [train.py:904] (6/8) Epoch 18, batch 9950, loss[loss=0.1644, simple_loss=0.2661, pruned_loss=0.03136, over 16396.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2668, pruned_loss=0.03726, over 3072435.22 frames. ], batch size: 146, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:41:58,940 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=182514.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:42:21,635 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9528, 4.9286, 5.3499, 5.2972, 5.3099, 5.0537, 4.9731, 4.7964], device='cuda:6'), covar=tensor([0.0289, 0.0505, 0.0309, 0.0305, 0.0383, 0.0339, 0.0799, 0.0367], device='cuda:6'), in_proj_covar=tensor([0.0361, 0.0392, 0.0383, 0.0364, 0.0428, 0.0404, 0.0489, 0.0326], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 19:43:31,282 INFO [train.py:904] (6/8) Epoch 18, batch 10000, loss[loss=0.1657, simple_loss=0.2699, pruned_loss=0.0308, over 16391.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2651, pruned_loss=0.03667, over 3103026.95 frames. ], batch size: 146, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:43:39,457 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6711, 2.4886, 2.4080, 3.7252, 2.1528, 3.8339, 1.4338, 2.8126], device='cuda:6'), covar=tensor([0.1426, 0.0758, 0.1178, 0.0127, 0.0132, 0.0345, 0.1672, 0.0746], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0166, 0.0188, 0.0172, 0.0194, 0.0208, 0.0192, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 19:43:42,213 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.446e+02 2.117e+02 2.346e+02 2.932e+02 5.510e+02, threshold=4.691e+02, percent-clipped=3.0 2023-04-30 19:43:52,874 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-30 19:44:17,933 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=182575.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:45:14,020 INFO [train.py:904] (6/8) Epoch 18, batch 10050, loss[loss=0.1797, simple_loss=0.2662, pruned_loss=0.04656, over 11941.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.265, pruned_loss=0.03675, over 3084002.95 frames. ], batch size: 247, lr: 3.71e-03, grad_scale: 8.0 2023-04-30 19:46:14,973 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=182634.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:46:20,791 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=182637.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:46:39,692 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0733, 3.1043, 1.9615, 3.3522, 2.3893, 3.3139, 2.1238, 2.6020], device='cuda:6'), covar=tensor([0.0301, 0.0409, 0.1617, 0.0219, 0.0888, 0.0620, 0.1480, 0.0751], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0164, 0.0184, 0.0142, 0.0167, 0.0198, 0.0192, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 19:46:46,716 INFO [train.py:904] (6/8) Epoch 18, batch 10100, loss[loss=0.1612, simple_loss=0.2567, pruned_loss=0.03288, over 16328.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2657, pruned_loss=0.03698, over 3092667.26 frames. ], batch size: 165, lr: 3.71e-03, grad_scale: 8.0 2023-04-30 19:46:55,747 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.437e+02 2.131e+02 2.554e+02 3.108e+02 6.507e+02, threshold=5.108e+02, percent-clipped=7.0 2023-04-30 19:47:44,865 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=182682.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:48:32,141 INFO [train.py:904] (6/8) Epoch 19, batch 0, loss[loss=0.2491, simple_loss=0.3093, pruned_loss=0.09441, over 16892.00 frames. ], tot_loss[loss=0.2491, simple_loss=0.3093, pruned_loss=0.09441, over 16892.00 frames. ], batch size: 116, lr: 3.61e-03, grad_scale: 8.0 2023-04-30 19:48:32,141 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 19:48:39,775 INFO [train.py:938] (6/8) Epoch 19, validation: loss=0.1468, simple_loss=0.2504, pruned_loss=0.0216, over 944034.00 frames. 2023-04-30 19:48:39,776 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 19:49:37,167 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6114, 6.0201, 5.7487, 5.7521, 5.3420, 5.2652, 5.4142, 6.1343], device='cuda:6'), covar=tensor([0.1253, 0.0856, 0.1066, 0.0895, 0.0867, 0.0763, 0.1192, 0.0846], device='cuda:6'), in_proj_covar=tensor([0.0612, 0.0750, 0.0615, 0.0557, 0.0471, 0.0483, 0.0626, 0.0580], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 19:49:50,015 INFO [train.py:904] (6/8) Epoch 19, batch 50, loss[loss=0.1913, simple_loss=0.283, pruned_loss=0.04984, over 16633.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2733, pruned_loss=0.05086, over 752943.46 frames. ], batch size: 62, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:49:59,029 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.571e+02 3.042e+02 3.790e+02 6.505e+02, threshold=6.085e+02, percent-clipped=6.0 2023-04-30 19:50:55,406 INFO [train.py:904] (6/8) Epoch 19, batch 100, loss[loss=0.1762, simple_loss=0.2623, pruned_loss=0.04511, over 16673.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2701, pruned_loss=0.04894, over 1313503.65 frames. ], batch size: 62, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:52:03,139 INFO [train.py:904] (6/8) Epoch 19, batch 150, loss[loss=0.1831, simple_loss=0.2552, pruned_loss=0.05546, over 16758.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2674, pruned_loss=0.04812, over 1757507.63 frames. ], batch size: 83, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:52:10,237 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=182856.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:52:15,181 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 2.297e+02 2.720e+02 3.223e+02 8.471e+02, threshold=5.441e+02, percent-clipped=1.0 2023-04-30 19:52:18,047 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=182862.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:52:28,189 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=182870.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:53:13,450 INFO [train.py:904] (6/8) Epoch 19, batch 200, loss[loss=0.1874, simple_loss=0.265, pruned_loss=0.05493, over 16643.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2662, pruned_loss=0.04799, over 2112105.83 frames. ], batch size: 134, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:53:31,178 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3590, 4.3484, 4.2323, 4.0101, 3.6569, 4.4756, 4.2918, 4.0663], device='cuda:6'), covar=tensor([0.1146, 0.1314, 0.0666, 0.0536, 0.1714, 0.0759, 0.0706, 0.0873], device='cuda:6'), in_proj_covar=tensor([0.0271, 0.0387, 0.0320, 0.0307, 0.0327, 0.0358, 0.0218, 0.0378], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 19:53:34,363 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=182917.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:53:42,614 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=182923.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:54:00,397 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=182937.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:54:21,115 INFO [train.py:904] (6/8) Epoch 19, batch 250, loss[loss=0.1821, simple_loss=0.2821, pruned_loss=0.04104, over 17113.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2636, pruned_loss=0.04686, over 2387311.74 frames. ], batch size: 48, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:54:32,985 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.641e+02 2.306e+02 2.752e+02 3.234e+02 1.372e+03, threshold=5.503e+02, percent-clipped=2.0 2023-04-30 19:55:08,834 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=182985.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:55:30,302 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-30 19:55:30,551 INFO [train.py:904] (6/8) Epoch 19, batch 300, loss[loss=0.1611, simple_loss=0.2427, pruned_loss=0.03973, over 16248.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2622, pruned_loss=0.04637, over 2599991.29 frames. ], batch size: 165, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:56:07,193 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6366, 3.6192, 2.1333, 3.8964, 2.8357, 3.8717, 2.3577, 2.9484], device='cuda:6'), covar=tensor([0.0265, 0.0455, 0.1660, 0.0384, 0.0776, 0.0799, 0.1462, 0.0707], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0172, 0.0192, 0.0152, 0.0174, 0.0209, 0.0200, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 19:56:41,129 INFO [train.py:904] (6/8) Epoch 19, batch 350, loss[loss=0.1695, simple_loss=0.2517, pruned_loss=0.04369, over 16768.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2601, pruned_loss=0.04514, over 2753584.78 frames. ], batch size: 102, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:56:52,337 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.344e+02 2.193e+02 2.670e+02 3.333e+02 1.052e+03, threshold=5.340e+02, percent-clipped=4.0 2023-04-30 19:57:51,264 INFO [train.py:904] (6/8) Epoch 19, batch 400, loss[loss=0.158, simple_loss=0.2438, pruned_loss=0.03606, over 16637.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2578, pruned_loss=0.04505, over 2876855.89 frames. ], batch size: 57, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 19:58:18,199 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2031, 5.8625, 5.9258, 5.6215, 5.7377, 6.3117, 5.7884, 5.4796], device='cuda:6'), covar=tensor([0.0860, 0.1821, 0.2030, 0.2402, 0.2628, 0.0913, 0.1508, 0.2545], device='cuda:6'), in_proj_covar=tensor([0.0386, 0.0560, 0.0623, 0.0472, 0.0628, 0.0653, 0.0491, 0.0631], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 19:59:03,157 INFO [train.py:904] (6/8) Epoch 19, batch 450, loss[loss=0.1751, simple_loss=0.2472, pruned_loss=0.05151, over 16739.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2562, pruned_loss=0.04402, over 2983830.05 frames. ], batch size: 83, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 19:59:14,111 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.162e+02 2.492e+02 3.004e+02 8.634e+02, threshold=4.984e+02, percent-clipped=1.0 2023-04-30 19:59:28,307 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=183170.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:00:07,981 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-30 20:00:12,293 INFO [train.py:904] (6/8) Epoch 19, batch 500, loss[loss=0.157, simple_loss=0.2458, pruned_loss=0.03416, over 16991.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2551, pruned_loss=0.04325, over 3049520.81 frames. ], batch size: 41, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 20:00:28,367 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=183212.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:00:34,240 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-30 20:00:35,914 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=183218.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:00:35,968 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=183218.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:00:36,099 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8548, 4.2774, 4.4093, 3.1277, 3.5548, 4.2669, 3.8860, 2.5182], device='cuda:6'), covar=tensor([0.0457, 0.0065, 0.0037, 0.0342, 0.0145, 0.0092, 0.0079, 0.0458], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0079, 0.0078, 0.0130, 0.0093, 0.0103, 0.0090, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 20:01:21,077 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-30 20:01:23,246 INFO [train.py:904] (6/8) Epoch 19, batch 550, loss[loss=0.1838, simple_loss=0.2724, pruned_loss=0.04764, over 17087.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2537, pruned_loss=0.04213, over 3112147.71 frames. ], batch size: 53, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 20:01:34,915 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.167e+02 2.560e+02 2.885e+02 5.610e+02, threshold=5.120e+02, percent-clipped=1.0 2023-04-30 20:02:02,379 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-30 20:02:32,735 INFO [train.py:904] (6/8) Epoch 19, batch 600, loss[loss=0.1644, simple_loss=0.2355, pruned_loss=0.04667, over 16915.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2533, pruned_loss=0.04242, over 3158165.35 frames. ], batch size: 109, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 20:02:47,544 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8980, 4.6412, 4.9488, 5.1538, 5.3480, 4.6367, 5.3793, 5.3579], device='cuda:6'), covar=tensor([0.1855, 0.1424, 0.1781, 0.0829, 0.0584, 0.0993, 0.0494, 0.0635], device='cuda:6'), in_proj_covar=tensor([0.0610, 0.0750, 0.0876, 0.0768, 0.0569, 0.0601, 0.0624, 0.0717], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:02:50,966 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0196, 4.4310, 3.1155, 2.3305, 2.7492, 2.5965, 4.7129, 3.6935], device='cuda:6'), covar=tensor([0.2618, 0.0530, 0.1783, 0.2935, 0.2953, 0.2036, 0.0365, 0.1250], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0264, 0.0299, 0.0304, 0.0289, 0.0249, 0.0287, 0.0327], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 20:03:42,735 INFO [train.py:904] (6/8) Epoch 19, batch 650, loss[loss=0.1819, simple_loss=0.27, pruned_loss=0.0469, over 17041.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.252, pruned_loss=0.04254, over 3200474.77 frames. ], batch size: 53, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 20:03:54,610 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 2.241e+02 2.597e+02 3.196e+02 6.464e+02, threshold=5.194e+02, percent-clipped=2.0 2023-04-30 20:04:53,075 INFO [train.py:904] (6/8) Epoch 19, batch 700, loss[loss=0.158, simple_loss=0.2501, pruned_loss=0.03297, over 17243.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2517, pruned_loss=0.04197, over 3234471.60 frames. ], batch size: 45, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 20:05:01,027 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=183408.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:05:59,948 INFO [train.py:904] (6/8) Epoch 19, batch 750, loss[loss=0.1621, simple_loss=0.247, pruned_loss=0.03856, over 16500.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2527, pruned_loss=0.04242, over 3250503.72 frames. ], batch size: 68, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 20:06:11,898 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.150e+02 2.519e+02 3.057e+02 7.230e+02, threshold=5.039e+02, percent-clipped=2.0 2023-04-30 20:06:25,293 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=183469.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:07:10,059 INFO [train.py:904] (6/8) Epoch 19, batch 800, loss[loss=0.1445, simple_loss=0.2265, pruned_loss=0.03127, over 16831.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2528, pruned_loss=0.04234, over 3264649.41 frames. ], batch size: 39, lr: 3.61e-03, grad_scale: 4.0 2023-04-30 20:07:24,166 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=183512.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:07:32,487 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=183518.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:08:04,824 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-30 20:08:19,002 INFO [train.py:904] (6/8) Epoch 19, batch 850, loss[loss=0.1512, simple_loss=0.2457, pruned_loss=0.02836, over 17190.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2522, pruned_loss=0.04177, over 3282894.96 frames. ], batch size: 46, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:08:29,786 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.473e+02 2.149e+02 2.547e+02 3.022e+02 7.911e+02, threshold=5.094e+02, percent-clipped=1.0 2023-04-30 20:08:30,864 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=183560.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:08:38,498 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=183566.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:09:17,110 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2133, 5.1963, 5.0799, 4.5841, 4.6606, 5.1334, 5.0463, 4.7086], device='cuda:6'), covar=tensor([0.0567, 0.0496, 0.0261, 0.0317, 0.1070, 0.0449, 0.0327, 0.0695], device='cuda:6'), in_proj_covar=tensor([0.0288, 0.0411, 0.0338, 0.0328, 0.0348, 0.0382, 0.0231, 0.0402], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:09:24,491 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 20:09:28,459 INFO [train.py:904] (6/8) Epoch 19, batch 900, loss[loss=0.1692, simple_loss=0.2475, pruned_loss=0.04542, over 16796.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2512, pruned_loss=0.04114, over 3295642.10 frames. ], batch size: 83, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:09:44,219 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=183614.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:10:35,713 INFO [train.py:904] (6/8) Epoch 19, batch 950, loss[loss=0.1567, simple_loss=0.2351, pruned_loss=0.03913, over 16670.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2513, pruned_loss=0.04124, over 3302192.96 frames. ], batch size: 89, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:10:45,887 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 1.990e+02 2.288e+02 2.656e+02 5.552e+02, threshold=4.575e+02, percent-clipped=1.0 2023-04-30 20:11:08,261 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=183675.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:11:12,983 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3875, 4.3027, 4.2889, 4.0445, 4.0460, 4.3772, 4.1287, 4.1324], device='cuda:6'), covar=tensor([0.0556, 0.0685, 0.0277, 0.0234, 0.0682, 0.0417, 0.0602, 0.0539], device='cuda:6'), in_proj_covar=tensor([0.0289, 0.0412, 0.0338, 0.0328, 0.0348, 0.0383, 0.0232, 0.0402], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:11:24,517 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-30 20:11:42,836 INFO [train.py:904] (6/8) Epoch 19, batch 1000, loss[loss=0.1544, simple_loss=0.248, pruned_loss=0.03041, over 17263.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2503, pruned_loss=0.04116, over 3315533.88 frames. ], batch size: 52, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:11:43,256 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6004, 3.5577, 3.9388, 2.0839, 4.0502, 4.0878, 3.3086, 2.9695], device='cuda:6'), covar=tensor([0.0803, 0.0228, 0.0177, 0.1181, 0.0095, 0.0185, 0.0336, 0.0415], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0107, 0.0095, 0.0141, 0.0077, 0.0123, 0.0127, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 20:11:45,295 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.63 vs. limit=5.0 2023-04-30 20:11:56,725 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2234, 5.7517, 5.8430, 5.6307, 5.5972, 6.2107, 5.6979, 5.4272], device='cuda:6'), covar=tensor([0.0846, 0.2028, 0.2442, 0.2019, 0.2961, 0.0973, 0.1557, 0.2485], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0576, 0.0644, 0.0484, 0.0648, 0.0673, 0.0501, 0.0649], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 20:12:02,997 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7347, 3.7789, 2.4053, 4.2969, 2.8732, 4.3133, 2.4895, 3.1065], device='cuda:6'), covar=tensor([0.0270, 0.0392, 0.1538, 0.0357, 0.0862, 0.0537, 0.1468, 0.0723], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0174, 0.0193, 0.0156, 0.0174, 0.0212, 0.0200, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 20:12:50,597 INFO [train.py:904] (6/8) Epoch 19, batch 1050, loss[loss=0.1847, simple_loss=0.2686, pruned_loss=0.0504, over 17039.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2507, pruned_loss=0.04148, over 3316583.94 frames. ], batch size: 55, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:13:01,769 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.633e+02 2.214e+02 2.595e+02 3.147e+02 6.979e+02, threshold=5.191e+02, percent-clipped=4.0 2023-04-30 20:13:06,564 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=183764.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:14:00,539 INFO [train.py:904] (6/8) Epoch 19, batch 1100, loss[loss=0.1645, simple_loss=0.2478, pruned_loss=0.04055, over 16536.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2496, pruned_loss=0.04104, over 3313784.25 frames. ], batch size: 68, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:14:55,868 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=183842.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:15:08,304 INFO [train.py:904] (6/8) Epoch 19, batch 1150, loss[loss=0.1689, simple_loss=0.2447, pruned_loss=0.04656, over 16772.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2498, pruned_loss=0.04077, over 3317500.99 frames. ], batch size: 83, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:15:20,305 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.320e+02 2.057e+02 2.490e+02 3.144e+02 5.373e+02, threshold=4.980e+02, percent-clipped=1.0 2023-04-30 20:16:04,900 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2674, 5.1752, 5.1221, 4.6204, 4.6905, 5.1376, 5.1219, 4.7380], device='cuda:6'), covar=tensor([0.0553, 0.0563, 0.0302, 0.0348, 0.1112, 0.0472, 0.0332, 0.0803], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0416, 0.0342, 0.0331, 0.0352, 0.0387, 0.0235, 0.0406], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:16:18,543 INFO [train.py:904] (6/8) Epoch 19, batch 1200, loss[loss=0.1698, simple_loss=0.2653, pruned_loss=0.03712, over 17139.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.25, pruned_loss=0.0403, over 3320314.78 frames. ], batch size: 48, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:16:19,034 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5948, 2.4943, 2.0344, 2.3252, 2.9075, 2.7504, 3.1878, 3.2218], device='cuda:6'), covar=tensor([0.0159, 0.0483, 0.0676, 0.0536, 0.0330, 0.0395, 0.0338, 0.0298], device='cuda:6'), in_proj_covar=tensor([0.0202, 0.0235, 0.0226, 0.0227, 0.0236, 0.0234, 0.0238, 0.0229], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:16:20,037 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=183903.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:17:25,272 INFO [train.py:904] (6/8) Epoch 19, batch 1250, loss[loss=0.1545, simple_loss=0.2464, pruned_loss=0.03134, over 17151.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2501, pruned_loss=0.04072, over 3319193.24 frames. ], batch size: 46, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:17:29,266 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3279, 4.1782, 4.3670, 4.5176, 4.6087, 4.1321, 4.3801, 4.6142], device='cuda:6'), covar=tensor([0.1701, 0.1127, 0.1327, 0.0723, 0.0586, 0.1298, 0.2764, 0.0730], device='cuda:6'), in_proj_covar=tensor([0.0628, 0.0773, 0.0904, 0.0792, 0.0586, 0.0619, 0.0640, 0.0737], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:17:35,896 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.556e+02 2.248e+02 2.515e+02 3.055e+02 4.782e+02, threshold=5.031e+02, percent-clipped=0.0 2023-04-30 20:17:50,866 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=183970.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:17:54,577 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7883, 2.4376, 1.9356, 2.2825, 2.8220, 2.6570, 2.9026, 2.9251], device='cuda:6'), covar=tensor([0.0228, 0.0383, 0.0522, 0.0418, 0.0256, 0.0309, 0.0210, 0.0264], device='cuda:6'), in_proj_covar=tensor([0.0202, 0.0235, 0.0226, 0.0227, 0.0236, 0.0233, 0.0237, 0.0228], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:18:02,048 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-30 20:18:17,058 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5741, 5.9598, 5.7327, 5.7788, 5.3377, 5.3706, 5.3004, 6.0961], device='cuda:6'), covar=tensor([0.1418, 0.0860, 0.1078, 0.0898, 0.1065, 0.0696, 0.1285, 0.0875], device='cuda:6'), in_proj_covar=tensor([0.0664, 0.0814, 0.0670, 0.0605, 0.0510, 0.0520, 0.0677, 0.0632], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:18:37,489 INFO [train.py:904] (6/8) Epoch 19, batch 1300, loss[loss=0.1621, simple_loss=0.2584, pruned_loss=0.0329, over 16672.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2509, pruned_loss=0.04093, over 3319671.32 frames. ], batch size: 62, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:18:47,832 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 20:19:44,241 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=184050.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:19:46,322 INFO [train.py:904] (6/8) Epoch 19, batch 1350, loss[loss=0.1588, simple_loss=0.2395, pruned_loss=0.03904, over 16676.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2504, pruned_loss=0.0403, over 3320160.99 frames. ], batch size: 134, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:19:55,434 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.519e+02 2.201e+02 2.541e+02 3.098e+02 6.218e+02, threshold=5.081e+02, percent-clipped=6.0 2023-04-30 20:20:03,911 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=184064.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:20:21,131 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.21 vs. limit=5.0 2023-04-30 20:20:27,047 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-30 20:20:55,152 INFO [train.py:904] (6/8) Epoch 19, batch 1400, loss[loss=0.175, simple_loss=0.2496, pruned_loss=0.05024, over 16505.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2496, pruned_loss=0.04016, over 3317797.65 frames. ], batch size: 68, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:21:09,598 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=184111.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:21:10,486 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=184112.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:21:11,864 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7943, 3.6300, 3.9869, 2.0887, 4.1566, 4.2103, 3.2228, 3.1596], device='cuda:6'), covar=tensor([0.0731, 0.0251, 0.0225, 0.1184, 0.0087, 0.0190, 0.0422, 0.0444], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0107, 0.0095, 0.0139, 0.0077, 0.0123, 0.0126, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 20:22:05,080 INFO [train.py:904] (6/8) Epoch 19, batch 1450, loss[loss=0.1675, simple_loss=0.245, pruned_loss=0.04495, over 16760.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2485, pruned_loss=0.04004, over 3322545.05 frames. ], batch size: 124, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:22:15,457 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.293e+02 2.595e+02 3.219e+02 5.738e+02, threshold=5.191e+02, percent-clipped=1.0 2023-04-30 20:22:33,102 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7302, 3.7037, 2.9163, 2.2287, 2.4139, 2.3154, 3.8636, 3.2237], device='cuda:6'), covar=tensor([0.2667, 0.0638, 0.1658, 0.2966, 0.2842, 0.2179, 0.0520, 0.1665], device='cuda:6'), in_proj_covar=tensor([0.0324, 0.0268, 0.0302, 0.0306, 0.0294, 0.0253, 0.0290, 0.0332], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 20:22:40,250 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4150, 3.7473, 3.9519, 2.1366, 3.1790, 2.5252, 4.0439, 3.9111], device='cuda:6'), covar=tensor([0.0245, 0.0796, 0.0491, 0.1948, 0.0795, 0.0955, 0.0518, 0.0980], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0159, 0.0164, 0.0150, 0.0142, 0.0127, 0.0142, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 20:22:42,530 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3130, 4.2883, 4.2347, 3.6253, 4.2713, 1.8726, 4.0218, 3.8374], device='cuda:6'), covar=tensor([0.0122, 0.0096, 0.0177, 0.0263, 0.0097, 0.2573, 0.0138, 0.0239], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0148, 0.0192, 0.0176, 0.0170, 0.0205, 0.0184, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:23:08,679 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=184198.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:23:13,290 INFO [train.py:904] (6/8) Epoch 19, batch 1500, loss[loss=0.1553, simple_loss=0.2402, pruned_loss=0.03519, over 17196.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2481, pruned_loss=0.03994, over 3313302.46 frames. ], batch size: 44, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:23:46,751 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.69 vs. limit=5.0 2023-04-30 20:24:05,592 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-30 20:24:22,394 INFO [train.py:904] (6/8) Epoch 19, batch 1550, loss[loss=0.1914, simple_loss=0.2622, pruned_loss=0.06029, over 16824.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2489, pruned_loss=0.04092, over 3314959.38 frames. ], batch size: 96, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:24:34,823 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.543e+02 2.353e+02 2.741e+02 3.128e+02 4.649e+02, threshold=5.482e+02, percent-clipped=0.0 2023-04-30 20:24:47,470 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=184270.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:25:31,172 INFO [train.py:904] (6/8) Epoch 19, batch 1600, loss[loss=0.1604, simple_loss=0.2443, pruned_loss=0.03828, over 17005.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2512, pruned_loss=0.04208, over 3301268.74 frames. ], batch size: 41, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:25:53,349 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=184318.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:26:15,606 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-30 20:26:39,948 INFO [train.py:904] (6/8) Epoch 19, batch 1650, loss[loss=0.1704, simple_loss=0.2554, pruned_loss=0.04271, over 15934.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2526, pruned_loss=0.04267, over 3301527.61 frames. ], batch size: 35, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:26:45,869 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6944, 3.9090, 2.2912, 4.4873, 2.8937, 4.4330, 2.5958, 3.1774], device='cuda:6'), covar=tensor([0.0334, 0.0387, 0.1681, 0.0248, 0.0897, 0.0561, 0.1403, 0.0753], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0177, 0.0195, 0.0160, 0.0175, 0.0215, 0.0202, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 20:26:50,246 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=184359.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:26:52,312 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.549e+02 2.220e+02 2.674e+02 3.187e+02 5.456e+02, threshold=5.348e+02, percent-clipped=0.0 2023-04-30 20:27:45,793 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.48 vs. limit=2.0 2023-04-30 20:27:49,665 INFO [train.py:904] (6/8) Epoch 19, batch 1700, loss[loss=0.1832, simple_loss=0.2616, pruned_loss=0.05237, over 16420.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2551, pruned_loss=0.04372, over 3297752.93 frames. ], batch size: 146, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:27:55,024 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=184406.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:28:03,278 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7218, 4.7502, 5.1419, 5.1510, 5.1504, 4.8000, 4.7709, 4.6102], device='cuda:6'), covar=tensor([0.0383, 0.0620, 0.0496, 0.0462, 0.0667, 0.0547, 0.1088, 0.0526], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0440, 0.0427, 0.0402, 0.0476, 0.0450, 0.0542, 0.0358], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 20:28:14,044 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=184420.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:28:32,351 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.04 vs. limit=5.0 2023-04-30 20:28:58,450 INFO [train.py:904] (6/8) Epoch 19, batch 1750, loss[loss=0.1781, simple_loss=0.2807, pruned_loss=0.03779, over 17229.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2572, pruned_loss=0.04407, over 3304822.80 frames. ], batch size: 52, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:29:10,989 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.342e+02 2.772e+02 3.206e+02 7.183e+02, threshold=5.544e+02, percent-clipped=2.0 2023-04-30 20:29:45,758 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=184486.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:30:00,605 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7280, 4.8166, 5.1709, 5.1778, 5.1860, 4.8546, 4.8015, 4.6312], device='cuda:6'), covar=tensor([0.0355, 0.0510, 0.0423, 0.0415, 0.0496, 0.0443, 0.0933, 0.0466], device='cuda:6'), in_proj_covar=tensor([0.0404, 0.0441, 0.0429, 0.0404, 0.0479, 0.0452, 0.0546, 0.0360], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 20:30:02,694 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=184498.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:30:07,526 INFO [train.py:904] (6/8) Epoch 19, batch 1800, loss[loss=0.1486, simple_loss=0.239, pruned_loss=0.02904, over 16796.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2579, pruned_loss=0.04368, over 3308480.84 frames. ], batch size: 83, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:30:47,183 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.72 vs. limit=5.0 2023-04-30 20:31:07,262 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=184546.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:31:09,212 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=184547.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:31:16,704 INFO [train.py:904] (6/8) Epoch 19, batch 1850, loss[loss=0.1557, simple_loss=0.2518, pruned_loss=0.02981, over 17144.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2584, pruned_loss=0.04316, over 3316338.67 frames. ], batch size: 47, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:31:21,506 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 20:31:29,852 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.319e+02 2.167e+02 2.563e+02 3.002e+02 5.511e+02, threshold=5.126e+02, percent-clipped=0.0 2023-04-30 20:32:26,186 INFO [train.py:904] (6/8) Epoch 19, batch 1900, loss[loss=0.1798, simple_loss=0.2753, pruned_loss=0.04219, over 17065.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2579, pruned_loss=0.04268, over 3303624.40 frames. ], batch size: 53, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:32:53,766 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7918, 4.2304, 4.2562, 3.0867, 3.6660, 4.1698, 3.7999, 2.5025], device='cuda:6'), covar=tensor([0.0433, 0.0068, 0.0046, 0.0336, 0.0121, 0.0097, 0.0084, 0.0431], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0082, 0.0081, 0.0133, 0.0096, 0.0106, 0.0092, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 20:33:25,463 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2989, 5.6465, 5.2143, 5.6319, 5.1345, 5.0040, 5.1741, 5.7602], device='cuda:6'), covar=tensor([0.2494, 0.1750, 0.2344, 0.1526, 0.1592, 0.1299, 0.2354, 0.2086], device='cuda:6'), in_proj_covar=tensor([0.0658, 0.0811, 0.0666, 0.0602, 0.0508, 0.0516, 0.0673, 0.0627], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:33:35,921 INFO [train.py:904] (6/8) Epoch 19, batch 1950, loss[loss=0.18, simple_loss=0.2565, pruned_loss=0.05169, over 16828.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2583, pruned_loss=0.04264, over 3306480.88 frames. ], batch size: 116, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:33:48,870 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 2.129e+02 2.542e+02 2.951e+02 4.781e+02, threshold=5.085e+02, percent-clipped=0.0 2023-04-30 20:34:15,212 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0501, 2.2199, 2.7634, 2.9226, 2.7871, 3.4786, 2.4530, 3.4589], device='cuda:6'), covar=tensor([0.0229, 0.0437, 0.0305, 0.0348, 0.0309, 0.0183, 0.0451, 0.0168], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0193, 0.0179, 0.0182, 0.0192, 0.0149, 0.0195, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:34:47,042 INFO [train.py:904] (6/8) Epoch 19, batch 2000, loss[loss=0.1875, simple_loss=0.2723, pruned_loss=0.05135, over 16688.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2583, pruned_loss=0.04282, over 3306723.24 frames. ], batch size: 62, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:34:53,108 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=184706.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:35:06,466 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=184715.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:35:24,001 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-30 20:35:56,926 INFO [train.py:904] (6/8) Epoch 19, batch 2050, loss[loss=0.1929, simple_loss=0.2695, pruned_loss=0.05808, over 16654.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2584, pruned_loss=0.04331, over 3312972.97 frames. ], batch size: 134, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:35:59,935 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=184754.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:36:10,010 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.173e+02 2.577e+02 2.933e+02 5.913e+02, threshold=5.153e+02, percent-clipped=1.0 2023-04-30 20:36:16,187 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2248, 4.3003, 4.5868, 4.5673, 4.6304, 4.3087, 4.3415, 4.1724], device='cuda:6'), covar=tensor([0.0376, 0.0594, 0.0462, 0.0497, 0.0472, 0.0485, 0.0788, 0.0707], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0440, 0.0428, 0.0404, 0.0475, 0.0453, 0.0543, 0.0359], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 20:37:10,296 INFO [train.py:904] (6/8) Epoch 19, batch 2100, loss[loss=0.1857, simple_loss=0.2624, pruned_loss=0.05455, over 16715.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2589, pruned_loss=0.04364, over 3312565.90 frames. ], batch size: 134, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:38:05,006 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=184842.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:38:18,978 INFO [train.py:904] (6/8) Epoch 19, batch 2150, loss[loss=0.1482, simple_loss=0.2349, pruned_loss=0.03072, over 17201.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2592, pruned_loss=0.04353, over 3320422.00 frames. ], batch size: 44, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:38:30,739 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.463e+02 2.154e+02 2.741e+02 3.179e+02 7.758e+02, threshold=5.482e+02, percent-clipped=2.0 2023-04-30 20:38:38,899 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=184866.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:39:27,103 INFO [train.py:904] (6/8) Epoch 19, batch 2200, loss[loss=0.2017, simple_loss=0.2756, pruned_loss=0.06393, over 16854.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2601, pruned_loss=0.04422, over 3320521.22 frames. ], batch size: 116, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:39:40,869 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4987, 3.3459, 2.7150, 2.0951, 2.2492, 2.2030, 3.5115, 3.0293], device='cuda:6'), covar=tensor([0.2750, 0.0843, 0.1743, 0.2898, 0.2728, 0.2213, 0.0586, 0.1622], device='cuda:6'), in_proj_covar=tensor([0.0321, 0.0266, 0.0299, 0.0303, 0.0292, 0.0251, 0.0288, 0.0330], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 20:39:46,132 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5476, 1.8048, 2.2058, 2.4230, 2.5000, 2.5252, 1.9434, 2.6610], device='cuda:6'), covar=tensor([0.0177, 0.0456, 0.0300, 0.0264, 0.0284, 0.0291, 0.0448, 0.0161], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0191, 0.0178, 0.0181, 0.0191, 0.0149, 0.0194, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:40:01,223 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=184927.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:40:17,935 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=184939.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:40:35,942 INFO [train.py:904] (6/8) Epoch 19, batch 2250, loss[loss=0.1705, simple_loss=0.2738, pruned_loss=0.03356, over 17253.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2618, pruned_loss=0.045, over 3321085.80 frames. ], batch size: 45, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:40:48,440 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.653e+02 2.330e+02 2.636e+02 3.046e+02 4.750e+02, threshold=5.271e+02, percent-clipped=0.0 2023-04-30 20:41:45,046 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=185000.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:41:46,800 INFO [train.py:904] (6/8) Epoch 19, batch 2300, loss[loss=0.1497, simple_loss=0.2266, pruned_loss=0.03637, over 16806.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.261, pruned_loss=0.04473, over 3320674.19 frames. ], batch size: 102, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:41:54,980 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-30 20:42:05,823 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=185015.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:42:20,105 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-30 20:42:56,744 INFO [train.py:904] (6/8) Epoch 19, batch 2350, loss[loss=0.1936, simple_loss=0.2614, pruned_loss=0.06286, over 16916.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2605, pruned_loss=0.04506, over 3319329.01 frames. ], batch size: 109, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:42:57,566 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-30 20:43:08,878 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=185060.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:43:09,638 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.650e+02 2.245e+02 2.599e+02 3.315e+02 5.469e+02, threshold=5.198e+02, percent-clipped=1.0 2023-04-30 20:43:12,099 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=185063.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:44:06,533 INFO [train.py:904] (6/8) Epoch 19, batch 2400, loss[loss=0.1659, simple_loss=0.2549, pruned_loss=0.0384, over 17227.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2608, pruned_loss=0.04462, over 3316781.39 frames. ], batch size: 46, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:44:33,435 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=185121.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 20:44:43,137 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=185129.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:45:02,255 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=185142.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:45:15,152 INFO [train.py:904] (6/8) Epoch 19, batch 2450, loss[loss=0.1775, simple_loss=0.2696, pruned_loss=0.04272, over 17106.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2617, pruned_loss=0.04429, over 3314332.03 frames. ], batch size: 48, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:45:27,049 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.487e+02 2.259e+02 2.760e+02 3.178e+02 5.977e+02, threshold=5.520e+02, percent-clipped=3.0 2023-04-30 20:45:47,591 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7891, 4.1420, 3.1023, 2.3799, 2.7688, 2.5763, 4.5572, 3.5602], device='cuda:6'), covar=tensor([0.2774, 0.0623, 0.1632, 0.2585, 0.2767, 0.1882, 0.0353, 0.1313], device='cuda:6'), in_proj_covar=tensor([0.0321, 0.0266, 0.0300, 0.0303, 0.0293, 0.0251, 0.0287, 0.0331], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 20:46:01,177 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8146, 2.4238, 1.9457, 2.1999, 2.8501, 2.6121, 2.9309, 2.9377], device='cuda:6'), covar=tensor([0.0176, 0.0397, 0.0523, 0.0452, 0.0229, 0.0325, 0.0225, 0.0271], device='cuda:6'), in_proj_covar=tensor([0.0205, 0.0238, 0.0227, 0.0229, 0.0239, 0.0237, 0.0242, 0.0232], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:46:02,685 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 20:46:08,234 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=185190.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:46:08,415 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=185190.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:46:13,169 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8665, 2.0277, 2.5579, 2.8365, 2.7384, 3.2399, 2.1318, 3.2976], device='cuda:6'), covar=tensor([0.0212, 0.0462, 0.0294, 0.0283, 0.0288, 0.0172, 0.0485, 0.0142], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0192, 0.0179, 0.0181, 0.0191, 0.0150, 0.0195, 0.0145], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:46:23,967 INFO [train.py:904] (6/8) Epoch 19, batch 2500, loss[loss=0.1586, simple_loss=0.2512, pruned_loss=0.03298, over 17237.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2623, pruned_loss=0.04451, over 3311519.55 frames. ], batch size: 52, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:46:51,984 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=185222.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:47:32,256 INFO [train.py:904] (6/8) Epoch 19, batch 2550, loss[loss=0.1576, simple_loss=0.2408, pruned_loss=0.03715, over 17027.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2623, pruned_loss=0.04436, over 3321009.13 frames. ], batch size: 41, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:47:44,040 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.145e+02 2.534e+02 2.988e+02 8.366e+02, threshold=5.067e+02, percent-clipped=1.0 2023-04-30 20:48:31,965 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=185295.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:48:40,161 INFO [train.py:904] (6/8) Epoch 19, batch 2600, loss[loss=0.1585, simple_loss=0.2498, pruned_loss=0.0336, over 16804.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2628, pruned_loss=0.04411, over 3322907.44 frames. ], batch size: 39, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:49:10,690 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9141, 4.1968, 2.9667, 2.3873, 2.7515, 2.5665, 4.5892, 3.6534], device='cuda:6'), covar=tensor([0.2703, 0.0638, 0.1824, 0.2692, 0.2932, 0.1951, 0.0398, 0.1249], device='cuda:6'), in_proj_covar=tensor([0.0320, 0.0265, 0.0300, 0.0303, 0.0294, 0.0251, 0.0288, 0.0331], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 20:49:23,730 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5413, 5.9041, 5.6518, 5.7432, 5.3424, 5.2732, 5.2441, 6.0771], device='cuda:6'), covar=tensor([0.1332, 0.1032, 0.1211, 0.0894, 0.0941, 0.0705, 0.1315, 0.1036], device='cuda:6'), in_proj_covar=tensor([0.0666, 0.0821, 0.0678, 0.0608, 0.0514, 0.0521, 0.0682, 0.0634], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:49:50,076 INFO [train.py:904] (6/8) Epoch 19, batch 2650, loss[loss=0.1781, simple_loss=0.266, pruned_loss=0.04513, over 17225.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2635, pruned_loss=0.04401, over 3326330.71 frames. ], batch size: 45, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:49:55,050 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-30 20:50:03,485 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.113e+02 2.462e+02 3.036e+02 8.000e+02, threshold=4.924e+02, percent-clipped=5.0 2023-04-30 20:50:37,320 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1557, 3.5004, 3.8264, 2.0954, 3.0078, 2.4111, 3.5277, 3.6601], device='cuda:6'), covar=tensor([0.0359, 0.0935, 0.0539, 0.1963, 0.0914, 0.0971, 0.0808, 0.1088], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0160, 0.0164, 0.0150, 0.0142, 0.0127, 0.0142, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 20:50:59,019 INFO [train.py:904] (6/8) Epoch 19, batch 2700, loss[loss=0.1505, simple_loss=0.2502, pruned_loss=0.02542, over 17114.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2628, pruned_loss=0.04349, over 3328193.97 frames. ], batch size: 49, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:51:19,152 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=185416.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 20:51:33,119 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6659, 2.4000, 2.4550, 4.5899, 2.3749, 2.8394, 2.5008, 2.6284], device='cuda:6'), covar=tensor([0.1170, 0.3655, 0.2819, 0.0418, 0.3922, 0.2541, 0.3252, 0.3625], device='cuda:6'), in_proj_covar=tensor([0.0395, 0.0438, 0.0361, 0.0326, 0.0432, 0.0504, 0.0407, 0.0511], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:52:08,642 INFO [train.py:904] (6/8) Epoch 19, batch 2750, loss[loss=0.1754, simple_loss=0.2641, pruned_loss=0.04333, over 16448.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2628, pruned_loss=0.04298, over 3330960.15 frames. ], batch size: 68, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:52:20,528 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.496e+02 2.075e+02 2.462e+02 3.061e+02 7.702e+02, threshold=4.923e+02, percent-clipped=3.0 2023-04-30 20:52:30,804 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=185468.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:52:37,303 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6045, 4.9715, 5.3115, 5.2679, 5.2882, 4.9308, 4.6189, 4.7200], device='cuda:6'), covar=tensor([0.0676, 0.0623, 0.0516, 0.0618, 0.0772, 0.0611, 0.1597, 0.0569], device='cuda:6'), in_proj_covar=tensor([0.0403, 0.0444, 0.0431, 0.0402, 0.0476, 0.0453, 0.0548, 0.0359], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 20:52:54,424 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=185485.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:53:18,343 INFO [train.py:904] (6/8) Epoch 19, batch 2800, loss[loss=0.1767, simple_loss=0.2552, pruned_loss=0.04912, over 16293.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2619, pruned_loss=0.04267, over 3332604.94 frames. ], batch size: 165, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:53:46,637 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=185522.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:53:56,349 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=185529.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:54:28,302 INFO [train.py:904] (6/8) Epoch 19, batch 2850, loss[loss=0.2079, simple_loss=0.2842, pruned_loss=0.06576, over 15466.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2615, pruned_loss=0.04274, over 3327943.44 frames. ], batch size: 191, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:54:41,475 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 2.166e+02 2.620e+02 3.152e+02 5.145e+02, threshold=5.240e+02, percent-clipped=1.0 2023-04-30 20:54:53,466 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=185570.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:55:25,994 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3680, 5.6992, 5.5154, 5.5705, 5.1471, 5.1745, 5.1708, 5.8813], device='cuda:6'), covar=tensor([0.1370, 0.0969, 0.1042, 0.0784, 0.0925, 0.0689, 0.1238, 0.0857], device='cuda:6'), in_proj_covar=tensor([0.0674, 0.0828, 0.0684, 0.0614, 0.0521, 0.0528, 0.0690, 0.0640], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 20:55:28,899 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=185595.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:55:39,056 INFO [train.py:904] (6/8) Epoch 19, batch 2900, loss[loss=0.1773, simple_loss=0.2641, pruned_loss=0.04523, over 17022.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2599, pruned_loss=0.04271, over 3330066.99 frames. ], batch size: 53, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 20:56:32,496 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-30 20:56:36,423 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=185643.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:56:47,750 INFO [train.py:904] (6/8) Epoch 19, batch 2950, loss[loss=0.1589, simple_loss=0.2553, pruned_loss=0.0313, over 17245.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2597, pruned_loss=0.04323, over 3329330.50 frames. ], batch size: 52, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 20:57:00,933 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.894e+02 2.308e+02 2.753e+02 3.431e+02 9.141e+02, threshold=5.506e+02, percent-clipped=4.0 2023-04-30 20:57:24,279 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4955, 3.6198, 4.0828, 2.2057, 3.1945, 2.5097, 3.9173, 3.8431], device='cuda:6'), covar=tensor([0.0267, 0.0915, 0.0414, 0.1904, 0.0772, 0.0942, 0.0583, 0.1002], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0159, 0.0163, 0.0149, 0.0141, 0.0126, 0.0141, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 20:57:58,854 INFO [train.py:904] (6/8) Epoch 19, batch 3000, loss[loss=0.1886, simple_loss=0.2841, pruned_loss=0.0465, over 16735.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.26, pruned_loss=0.04416, over 3327717.44 frames. ], batch size: 57, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 20:57:58,854 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 20:58:04,307 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6091, 3.8164, 3.8801, 2.9119, 3.5290, 3.9223, 3.7084, 2.4937], device='cuda:6'), covar=tensor([0.0434, 0.0071, 0.0051, 0.0318, 0.0114, 0.0080, 0.0080, 0.0421], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0082, 0.0082, 0.0133, 0.0096, 0.0107, 0.0093, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 20:58:07,639 INFO [train.py:938] (6/8) Epoch 19, validation: loss=0.1362, simple_loss=0.2416, pruned_loss=0.01538, over 944034.00 frames. 2023-04-30 20:58:07,639 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 20:58:28,378 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=185716.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 20:58:28,449 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7694, 2.7161, 2.7525, 4.8231, 3.6285, 4.3326, 1.6146, 3.1583], device='cuda:6'), covar=tensor([0.1458, 0.0944, 0.1242, 0.0247, 0.0280, 0.0387, 0.1752, 0.0827], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0172, 0.0192, 0.0185, 0.0205, 0.0217, 0.0197, 0.0189], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 20:59:16,277 INFO [train.py:904] (6/8) Epoch 19, batch 3050, loss[loss=0.1639, simple_loss=0.2651, pruned_loss=0.03137, over 17087.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2597, pruned_loss=0.04418, over 3331933.24 frames. ], batch size: 55, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 20:59:28,122 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=185759.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:59:30,009 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.398e+02 2.137e+02 2.455e+02 2.756e+02 4.328e+02, threshold=4.910e+02, percent-clipped=0.0 2023-04-30 20:59:35,050 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=185764.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:00:04,052 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=185785.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:00:23,318 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=185799.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 21:00:26,857 INFO [train.py:904] (6/8) Epoch 19, batch 3100, loss[loss=0.1888, simple_loss=0.2591, pruned_loss=0.05931, over 16732.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2596, pruned_loss=0.04456, over 3333656.95 frames. ], batch size: 124, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:00:51,365 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=185820.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 21:00:57,011 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=185824.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:01:09,059 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=185833.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:01:16,185 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-30 21:01:34,487 INFO [train.py:904] (6/8) Epoch 19, batch 3150, loss[loss=0.2191, simple_loss=0.282, pruned_loss=0.07812, over 16888.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2586, pruned_loss=0.04442, over 3339821.66 frames. ], batch size: 109, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:01:46,537 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=185860.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 21:01:47,101 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.254e+02 2.659e+02 3.271e+02 6.332e+02, threshold=5.317e+02, percent-clipped=3.0 2023-04-30 21:01:50,550 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9013, 4.6568, 4.8669, 5.0903, 5.2734, 4.5808, 5.2661, 5.2608], device='cuda:6'), covar=tensor([0.1785, 0.1323, 0.1839, 0.0831, 0.0593, 0.0943, 0.0609, 0.0610], device='cuda:6'), in_proj_covar=tensor([0.0653, 0.0807, 0.0949, 0.0823, 0.0614, 0.0645, 0.0662, 0.0771], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:02:43,073 INFO [train.py:904] (6/8) Epoch 19, batch 3200, loss[loss=0.1847, simple_loss=0.2585, pruned_loss=0.05547, over 16878.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2591, pruned_loss=0.04407, over 3325702.54 frames. ], batch size: 116, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:03:41,041 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6000, 2.3744, 2.4020, 4.4294, 2.2975, 2.8004, 2.4729, 2.5657], device='cuda:6'), covar=tensor([0.1155, 0.3540, 0.2895, 0.0464, 0.4170, 0.2427, 0.3433, 0.3464], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0441, 0.0364, 0.0329, 0.0435, 0.0507, 0.0410, 0.0515], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:03:53,534 INFO [train.py:904] (6/8) Epoch 19, batch 3250, loss[loss=0.1538, simple_loss=0.2393, pruned_loss=0.03414, over 16811.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2577, pruned_loss=0.04303, over 3321405.59 frames. ], batch size: 39, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:04:06,304 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.264e+02 2.586e+02 3.072e+02 8.632e+02, threshold=5.173e+02, percent-clipped=3.0 2023-04-30 21:04:06,810 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4960, 3.3743, 3.8757, 1.8751, 3.9538, 3.9246, 3.1901, 2.8062], device='cuda:6'), covar=tensor([0.0833, 0.0247, 0.0140, 0.1217, 0.0084, 0.0189, 0.0353, 0.0487], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0107, 0.0096, 0.0139, 0.0078, 0.0124, 0.0126, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 21:05:07,473 INFO [train.py:904] (6/8) Epoch 19, batch 3300, loss[loss=0.1789, simple_loss=0.2675, pruned_loss=0.04518, over 17093.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2581, pruned_loss=0.04293, over 3322961.82 frames. ], batch size: 53, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:06:15,806 INFO [train.py:904] (6/8) Epoch 19, batch 3350, loss[loss=0.1573, simple_loss=0.2505, pruned_loss=0.03204, over 17155.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2588, pruned_loss=0.04279, over 3316767.94 frames. ], batch size: 46, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:06:27,998 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 2.170e+02 2.521e+02 2.996e+02 5.638e+02, threshold=5.041e+02, percent-clipped=4.0 2023-04-30 21:06:42,050 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0420, 4.0846, 4.3935, 4.4021, 4.4174, 4.1083, 4.1645, 4.0958], device='cuda:6'), covar=tensor([0.0416, 0.0630, 0.0409, 0.0372, 0.0497, 0.0502, 0.0727, 0.0555], device='cuda:6'), in_proj_covar=tensor([0.0409, 0.0450, 0.0436, 0.0407, 0.0482, 0.0460, 0.0555, 0.0365], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 21:07:23,735 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186100.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:07:25,691 INFO [train.py:904] (6/8) Epoch 19, batch 3400, loss[loss=0.1586, simple_loss=0.244, pruned_loss=0.0366, over 16880.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2584, pruned_loss=0.04233, over 3320597.62 frames. ], batch size: 102, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:07:44,692 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186115.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 21:07:57,633 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=186124.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:08:34,722 INFO [train.py:904] (6/8) Epoch 19, batch 3450, loss[loss=0.1715, simple_loss=0.244, pruned_loss=0.04946, over 16725.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2571, pruned_loss=0.0422, over 3320420.33 frames. ], batch size: 124, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:08:39,229 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186155.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 21:08:44,174 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186158.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:08:47,843 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.616e+02 2.301e+02 2.793e+02 3.380e+02 7.495e+02, threshold=5.586e+02, percent-clipped=5.0 2023-04-30 21:08:48,279 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186161.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:09:03,003 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=186172.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:09:27,193 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-30 21:09:45,133 INFO [train.py:904] (6/8) Epoch 19, batch 3500, loss[loss=0.1896, simple_loss=0.2708, pruned_loss=0.0542, over 15295.00 frames. ], tot_loss[loss=0.17, simple_loss=0.256, pruned_loss=0.04203, over 3316391.51 frames. ], batch size: 190, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:09:50,146 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-30 21:10:08,900 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186219.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:10:17,721 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2023-04-30 21:10:55,186 INFO [train.py:904] (6/8) Epoch 19, batch 3550, loss[loss=0.1387, simple_loss=0.2282, pruned_loss=0.02454, over 16867.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2545, pruned_loss=0.04143, over 3318339.91 frames. ], batch size: 42, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:11:05,309 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186259.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:11:06,977 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 2.073e+02 2.370e+02 2.856e+02 5.323e+02, threshold=4.740e+02, percent-clipped=0.0 2023-04-30 21:12:03,398 INFO [train.py:904] (6/8) Epoch 19, batch 3600, loss[loss=0.131, simple_loss=0.2157, pruned_loss=0.02319, over 16741.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2541, pruned_loss=0.04131, over 3302294.73 frames. ], batch size: 39, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:12:03,857 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0844, 5.4245, 5.1038, 5.2081, 4.9945, 4.8964, 4.8849, 5.5148], device='cuda:6'), covar=tensor([0.1272, 0.0934, 0.1211, 0.0915, 0.0805, 0.0997, 0.1213, 0.0936], device='cuda:6'), in_proj_covar=tensor([0.0677, 0.0832, 0.0687, 0.0619, 0.0524, 0.0530, 0.0696, 0.0643], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:12:28,393 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186320.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:13:13,926 INFO [train.py:904] (6/8) Epoch 19, batch 3650, loss[loss=0.1843, simple_loss=0.2543, pruned_loss=0.05718, over 16902.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2536, pruned_loss=0.04211, over 3304560.44 frames. ], batch size: 109, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:13:27,661 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.643e+02 2.244e+02 2.696e+02 3.128e+02 5.239e+02, threshold=5.393e+02, percent-clipped=3.0 2023-04-30 21:14:28,525 INFO [train.py:904] (6/8) Epoch 19, batch 3700, loss[loss=0.1515, simple_loss=0.2396, pruned_loss=0.03166, over 15624.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2519, pruned_loss=0.04306, over 3292545.92 frames. ], batch size: 191, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:14:48,214 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-30 21:14:48,826 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=186415.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:14:57,121 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186421.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:15:41,466 INFO [train.py:904] (6/8) Epoch 19, batch 3750, loss[loss=0.1885, simple_loss=0.2519, pruned_loss=0.06257, over 16907.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2527, pruned_loss=0.045, over 3269546.10 frames. ], batch size: 116, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:15:46,647 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=186455.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 21:15:47,856 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186456.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:15:55,022 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.254e+02 2.571e+02 3.173e+02 4.680e+02, threshold=5.142e+02, percent-clipped=0.0 2023-04-30 21:15:57,662 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=186463.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:16:15,871 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186475.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:16:26,550 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186482.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:16:53,853 INFO [train.py:904] (6/8) Epoch 19, batch 3800, loss[loss=0.1773, simple_loss=0.2645, pruned_loss=0.04509, over 17001.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2541, pruned_loss=0.04654, over 3263528.81 frames. ], batch size: 55, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:16:55,980 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=186503.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 21:17:11,649 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186514.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:17:15,604 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186517.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:17:43,737 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186536.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:17:47,923 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186539.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:17:52,048 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5591, 4.5455, 4.5204, 3.8703, 4.5599, 1.7094, 4.3036, 4.1109], device='cuda:6'), covar=tensor([0.0153, 0.0106, 0.0187, 0.0393, 0.0105, 0.2843, 0.0152, 0.0254], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0149, 0.0194, 0.0178, 0.0172, 0.0203, 0.0186, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:18:05,819 INFO [train.py:904] (6/8) Epoch 19, batch 3850, loss[loss=0.1765, simple_loss=0.2589, pruned_loss=0.04703, over 16118.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2541, pruned_loss=0.04695, over 3273494.03 frames. ], batch size: 35, lr: 3.58e-03, grad_scale: 4.0 2023-04-30 21:18:22,161 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.100e+02 2.475e+02 3.097e+02 5.903e+02, threshold=4.949e+02, percent-clipped=1.0 2023-04-30 21:18:28,994 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186567.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:18:44,989 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186578.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:18:52,239 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6795, 3.7174, 2.7865, 2.2473, 2.5666, 2.3732, 3.6686, 3.3772], device='cuda:6'), covar=tensor([0.2748, 0.0649, 0.1799, 0.2924, 0.2673, 0.2128, 0.0673, 0.1377], device='cuda:6'), in_proj_covar=tensor([0.0323, 0.0268, 0.0302, 0.0306, 0.0298, 0.0254, 0.0291, 0.0333], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 21:18:56,112 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186586.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:19:16,960 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186600.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:19:18,740 INFO [train.py:904] (6/8) Epoch 19, batch 3900, loss[loss=0.1887, simple_loss=0.2678, pruned_loss=0.05481, over 17007.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2543, pruned_loss=0.04749, over 3277287.12 frames. ], batch size: 41, lr: 3.58e-03, grad_scale: 4.0 2023-04-30 21:19:37,487 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186615.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:19:55,110 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186628.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:20:05,061 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1953, 4.2614, 4.5656, 4.5540, 4.6008, 4.2890, 4.3087, 4.2068], device='cuda:6'), covar=tensor([0.0401, 0.0579, 0.0411, 0.0403, 0.0461, 0.0436, 0.0794, 0.0563], device='cuda:6'), in_proj_covar=tensor([0.0405, 0.0444, 0.0431, 0.0402, 0.0474, 0.0454, 0.0547, 0.0361], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 21:20:21,584 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2719, 3.3141, 1.9889, 3.4029, 2.5787, 3.4933, 2.1442, 2.6833], device='cuda:6'), covar=tensor([0.0258, 0.0348, 0.1463, 0.0251, 0.0715, 0.0650, 0.1343, 0.0677], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0177, 0.0194, 0.0162, 0.0176, 0.0218, 0.0201, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 21:20:21,600 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186647.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:20:28,330 INFO [train.py:904] (6/8) Epoch 19, batch 3950, loss[loss=0.1903, simple_loss=0.2684, pruned_loss=0.05608, over 16478.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2543, pruned_loss=0.04812, over 3282039.54 frames. ], batch size: 146, lr: 3.57e-03, grad_scale: 4.0 2023-04-30 21:20:42,972 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.601e+02 2.399e+02 2.822e+02 3.355e+02 8.994e+02, threshold=5.644e+02, percent-clipped=5.0 2023-04-30 21:21:38,417 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3003, 3.5707, 3.6980, 2.6858, 3.4621, 3.8301, 3.5162, 2.1100], device='cuda:6'), covar=tensor([0.0472, 0.0109, 0.0050, 0.0330, 0.0096, 0.0088, 0.0080, 0.0457], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0079, 0.0080, 0.0131, 0.0095, 0.0104, 0.0092, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 21:21:39,185 INFO [train.py:904] (6/8) Epoch 19, batch 4000, loss[loss=0.1736, simple_loss=0.2508, pruned_loss=0.04816, over 16907.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2544, pruned_loss=0.04835, over 3285894.57 frames. ], batch size: 116, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:21:49,118 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186709.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:22:42,385 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0774, 3.2791, 3.5406, 1.9630, 2.9631, 2.2777, 3.5816, 3.5928], device='cuda:6'), covar=tensor([0.0195, 0.0765, 0.0521, 0.1929, 0.0781, 0.0920, 0.0475, 0.0780], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0163, 0.0167, 0.0153, 0.0144, 0.0129, 0.0144, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 21:22:49,923 INFO [train.py:904] (6/8) Epoch 19, batch 4050, loss[loss=0.1718, simple_loss=0.2551, pruned_loss=0.04423, over 16496.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2547, pruned_loss=0.04752, over 3279052.54 frames. ], batch size: 146, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:22:57,347 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=186756.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:23:06,009 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.223e+02 1.873e+02 2.216e+02 2.589e+02 4.495e+02, threshold=4.432e+02, percent-clipped=0.0 2023-04-30 21:23:16,255 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186770.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:23:27,094 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186777.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:23:48,465 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5895, 4.6295, 4.4194, 3.1482, 3.9055, 4.5099, 3.8865, 2.4972], device='cuda:6'), covar=tensor([0.0476, 0.0023, 0.0040, 0.0314, 0.0091, 0.0076, 0.0082, 0.0406], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0080, 0.0081, 0.0132, 0.0095, 0.0105, 0.0092, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 21:23:58,577 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1436, 5.5410, 5.7527, 5.4934, 5.5371, 6.1180, 5.6749, 5.3148], device='cuda:6'), covar=tensor([0.0831, 0.1711, 0.1538, 0.1842, 0.2163, 0.0881, 0.1382, 0.2252], device='cuda:6'), in_proj_covar=tensor([0.0406, 0.0587, 0.0646, 0.0490, 0.0655, 0.0684, 0.0508, 0.0654], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 21:24:03,613 INFO [train.py:904] (6/8) Epoch 19, batch 4100, loss[loss=0.1744, simple_loss=0.2625, pruned_loss=0.04311, over 16493.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2557, pruned_loss=0.04668, over 3273447.07 frames. ], batch size: 68, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:24:07,124 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=186804.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:24:13,901 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9168, 3.2904, 3.2004, 2.0847, 3.1656, 3.2777, 3.1502, 1.5497], device='cuda:6'), covar=tensor([0.0597, 0.0061, 0.0092, 0.0507, 0.0109, 0.0142, 0.0105, 0.0647], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0080, 0.0081, 0.0132, 0.0095, 0.0105, 0.0092, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 21:24:21,943 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=186814.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:24:33,290 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9146, 4.9989, 5.2830, 5.2501, 5.3111, 4.9656, 4.9131, 4.7102], device='cuda:6'), covar=tensor([0.0285, 0.0366, 0.0314, 0.0357, 0.0398, 0.0323, 0.0829, 0.0425], device='cuda:6'), in_proj_covar=tensor([0.0404, 0.0445, 0.0432, 0.0404, 0.0475, 0.0456, 0.0549, 0.0361], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 21:24:46,727 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186831.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:24:59,827 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186840.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:25:08,839 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2506, 2.2420, 2.2757, 3.8372, 2.1993, 2.5815, 2.3868, 2.4091], device='cuda:6'), covar=tensor([0.1262, 0.3479, 0.2719, 0.0548, 0.3976, 0.2351, 0.3024, 0.3401], device='cuda:6'), in_proj_covar=tensor([0.0396, 0.0440, 0.0360, 0.0326, 0.0432, 0.0507, 0.0408, 0.0514], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:25:14,224 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0602, 4.3500, 4.5667, 4.5441, 4.5710, 4.2401, 4.0038, 4.1842], device='cuda:6'), covar=tensor([0.0617, 0.0646, 0.0539, 0.0602, 0.0693, 0.0638, 0.1601, 0.0598], device='cuda:6'), in_proj_covar=tensor([0.0400, 0.0440, 0.0428, 0.0399, 0.0470, 0.0452, 0.0544, 0.0357], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 21:25:17,893 INFO [train.py:904] (6/8) Epoch 19, batch 4150, loss[loss=0.2003, simple_loss=0.2864, pruned_loss=0.05708, over 16881.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2628, pruned_loss=0.04914, over 3242874.82 frames. ], batch size: 116, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:25:34,161 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=186862.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:25:34,923 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.347e+02 2.016e+02 2.433e+02 2.890e+02 5.187e+02, threshold=4.866e+02, percent-clipped=4.0 2023-04-30 21:25:49,776 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186873.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:25:53,412 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-30 21:26:10,514 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7455, 2.7655, 2.7751, 4.7696, 3.7314, 4.1108, 1.8119, 2.9308], device='cuda:6'), covar=tensor([0.1372, 0.0833, 0.1127, 0.0154, 0.0316, 0.0432, 0.1571, 0.0862], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0172, 0.0191, 0.0184, 0.0205, 0.0215, 0.0196, 0.0189], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 21:26:23,060 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186895.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:26:31,998 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186901.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 21:26:32,668 INFO [train.py:904] (6/8) Epoch 19, batch 4200, loss[loss=0.2272, simple_loss=0.3143, pruned_loss=0.06999, over 15299.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2697, pruned_loss=0.05089, over 3207781.55 frames. ], batch size: 191, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:26:47,056 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186911.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:26:48,309 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5188, 3.5961, 3.2533, 2.9976, 3.1608, 3.4710, 3.3070, 3.2868], device='cuda:6'), covar=tensor([0.0550, 0.0513, 0.0300, 0.0275, 0.0577, 0.0442, 0.1404, 0.0491], device='cuda:6'), in_proj_covar=tensor([0.0290, 0.0414, 0.0342, 0.0331, 0.0352, 0.0385, 0.0233, 0.0406], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:26:50,151 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9419, 5.3175, 5.5121, 5.2433, 5.2781, 5.8578, 5.4003, 5.0872], device='cuda:6'), covar=tensor([0.0882, 0.1714, 0.1522, 0.1591, 0.2053, 0.0779, 0.1347, 0.2097], device='cuda:6'), in_proj_covar=tensor([0.0401, 0.0580, 0.0638, 0.0486, 0.0645, 0.0676, 0.0501, 0.0647], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 21:26:53,390 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=186915.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:27:06,397 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186923.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:27:34,737 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186942.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:27:49,658 INFO [train.py:904] (6/8) Epoch 19, batch 4250, loss[loss=0.1846, simple_loss=0.2823, pruned_loss=0.04342, over 16414.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2735, pruned_loss=0.05118, over 3182747.93 frames. ], batch size: 68, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:27:52,546 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6044, 2.4048, 2.4162, 3.3327, 2.2545, 3.6616, 1.5338, 2.7281], device='cuda:6'), covar=tensor([0.1317, 0.0824, 0.1172, 0.0145, 0.0165, 0.0358, 0.1629, 0.0825], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0171, 0.0190, 0.0182, 0.0203, 0.0213, 0.0195, 0.0188], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 21:28:05,619 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.577e+02 2.275e+02 2.568e+02 3.131e+02 7.157e+02, threshold=5.135e+02, percent-clipped=7.0 2023-04-30 21:28:06,530 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=186963.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:28:19,208 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186972.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:28:47,127 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9327, 4.9335, 4.7005, 3.9618, 4.8384, 1.9637, 4.5538, 4.4000], device='cuda:6'), covar=tensor([0.0074, 0.0069, 0.0184, 0.0331, 0.0073, 0.2624, 0.0122, 0.0242], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0147, 0.0193, 0.0176, 0.0170, 0.0203, 0.0184, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:29:02,897 INFO [train.py:904] (6/8) Epoch 19, batch 4300, loss[loss=0.1961, simple_loss=0.2889, pruned_loss=0.05169, over 16565.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2751, pruned_loss=0.05058, over 3188980.60 frames. ], batch size: 68, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:29:37,770 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=187025.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:29:56,695 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=187038.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:30:10,226 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0269, 5.2624, 5.0215, 5.0283, 4.8213, 4.7163, 4.6607, 5.3310], device='cuda:6'), covar=tensor([0.0848, 0.0685, 0.0902, 0.0749, 0.0630, 0.0800, 0.1016, 0.0710], device='cuda:6'), in_proj_covar=tensor([0.0653, 0.0804, 0.0660, 0.0600, 0.0505, 0.0513, 0.0671, 0.0619], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:30:16,621 INFO [train.py:904] (6/8) Epoch 19, batch 4350, loss[loss=0.1984, simple_loss=0.2867, pruned_loss=0.05507, over 16806.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2782, pruned_loss=0.05122, over 3188482.57 frames. ], batch size: 39, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:30:31,211 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.85 vs. limit=5.0 2023-04-30 21:30:32,929 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.681e+02 2.200e+02 2.562e+02 2.922e+02 5.186e+02, threshold=5.123e+02, percent-clipped=1.0 2023-04-30 21:30:35,925 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187065.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:30:54,493 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187077.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:31:08,009 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=187086.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 21:31:26,237 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=187099.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:31:31,214 INFO [train.py:904] (6/8) Epoch 19, batch 4400, loss[loss=0.2118, simple_loss=0.2889, pruned_loss=0.06733, over 11726.00 frames. ], tot_loss[loss=0.1924, simple_loss=0.2802, pruned_loss=0.0523, over 3176617.63 frames. ], batch size: 247, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:31:42,577 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.08 vs. limit=5.0 2023-04-30 21:32:05,653 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187125.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:32:13,521 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187131.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:32:13,670 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4284, 3.4081, 2.0460, 3.9821, 2.5786, 3.9940, 2.1841, 2.6736], device='cuda:6'), covar=tensor([0.0302, 0.0415, 0.1754, 0.0136, 0.0913, 0.0388, 0.1601, 0.0852], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0176, 0.0194, 0.0158, 0.0175, 0.0216, 0.0199, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 21:32:43,364 INFO [train.py:904] (6/8) Epoch 19, batch 4450, loss[loss=0.2012, simple_loss=0.2895, pruned_loss=0.05651, over 16298.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2833, pruned_loss=0.05315, over 3198802.55 frames. ], batch size: 35, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:32:47,771 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8978, 4.7656, 4.8993, 5.0709, 5.2107, 4.6666, 5.2267, 5.2513], device='cuda:6'), covar=tensor([0.1518, 0.1064, 0.1402, 0.0628, 0.0449, 0.0751, 0.0474, 0.0493], device='cuda:6'), in_proj_covar=tensor([0.0620, 0.0767, 0.0900, 0.0784, 0.0585, 0.0612, 0.0630, 0.0733], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:33:00,621 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.070e+02 2.320e+02 2.761e+02 4.804e+02, threshold=4.641e+02, percent-clipped=0.0 2023-04-30 21:33:15,045 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187173.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:33:24,883 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187179.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:33:35,456 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5671, 3.6430, 3.3438, 3.0906, 3.2290, 3.5266, 3.3293, 3.3603], device='cuda:6'), covar=tensor([0.0486, 0.0365, 0.0241, 0.0223, 0.0498, 0.0339, 0.1233, 0.0430], device='cuda:6'), in_proj_covar=tensor([0.0285, 0.0404, 0.0335, 0.0325, 0.0345, 0.0376, 0.0229, 0.0397], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:33:48,144 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187195.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:33:49,992 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187196.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 21:33:57,506 INFO [train.py:904] (6/8) Epoch 19, batch 4500, loss[loss=0.2243, simple_loss=0.3056, pruned_loss=0.07151, over 12042.00 frames. ], tot_loss[loss=0.195, simple_loss=0.2833, pruned_loss=0.0533, over 3214087.96 frames. ], batch size: 247, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:34:24,640 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187221.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:34:27,211 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187223.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:34:54,244 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187242.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:34:55,422 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187243.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:35:08,490 INFO [train.py:904] (6/8) Epoch 19, batch 4550, loss[loss=0.199, simple_loss=0.2922, pruned_loss=0.05293, over 16383.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2849, pruned_loss=0.05464, over 3212464.52 frames. ], batch size: 68, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:35:24,435 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 1.917e+02 2.180e+02 2.712e+02 4.446e+02, threshold=4.359e+02, percent-clipped=0.0 2023-04-30 21:35:30,860 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187267.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:35:36,696 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187271.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:35:43,959 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3068, 2.4451, 2.4836, 4.2879, 2.2205, 2.8425, 2.4727, 2.5566], device='cuda:6'), covar=tensor([0.1276, 0.3283, 0.2550, 0.0413, 0.4073, 0.2185, 0.3064, 0.3227], device='cuda:6'), in_proj_covar=tensor([0.0395, 0.0438, 0.0358, 0.0323, 0.0432, 0.0505, 0.0407, 0.0511], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:36:01,890 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2023-04-30 21:36:05,135 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187290.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:36:20,300 INFO [train.py:904] (6/8) Epoch 19, batch 4600, loss[loss=0.1727, simple_loss=0.2611, pruned_loss=0.04211, over 16462.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2858, pruned_loss=0.05534, over 3200453.81 frames. ], batch size: 75, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:37:33,043 INFO [train.py:904] (6/8) Epoch 19, batch 4650, loss[loss=0.223, simple_loss=0.2963, pruned_loss=0.07483, over 17022.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2853, pruned_loss=0.05539, over 3207432.68 frames. ], batch size: 53, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:37:49,309 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.228e+02 1.930e+02 2.188e+02 2.546e+02 5.637e+02, threshold=4.377e+02, percent-clipped=1.0 2023-04-30 21:37:52,239 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187365.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:38:01,627 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=187372.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:38:15,556 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187381.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 21:38:33,295 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187394.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:38:43,312 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.98 vs. limit=5.0 2023-04-30 21:38:44,846 INFO [train.py:904] (6/8) Epoch 19, batch 4700, loss[loss=0.1628, simple_loss=0.2588, pruned_loss=0.03343, over 16395.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.2825, pruned_loss=0.05414, over 3210896.36 frames. ], batch size: 146, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:38:53,622 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6378, 4.6569, 4.5289, 3.7328, 4.5768, 1.7910, 4.3576, 4.2438], device='cuda:6'), covar=tensor([0.0132, 0.0136, 0.0152, 0.0475, 0.0119, 0.2678, 0.0146, 0.0259], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0146, 0.0191, 0.0175, 0.0168, 0.0201, 0.0183, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:38:59,621 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-30 21:39:01,765 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187413.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:39:29,995 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=187433.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:39:34,156 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2016, 4.3708, 4.5583, 4.5359, 4.6089, 4.3093, 4.1903, 4.1973], device='cuda:6'), covar=tensor([0.0450, 0.0593, 0.0538, 0.0587, 0.0562, 0.0539, 0.1306, 0.0537], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0428, 0.0417, 0.0390, 0.0460, 0.0440, 0.0530, 0.0347], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 21:39:55,445 INFO [train.py:904] (6/8) Epoch 19, batch 4750, loss[loss=0.1805, simple_loss=0.274, pruned_loss=0.04352, over 15391.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.2786, pruned_loss=0.05209, over 3206660.42 frames. ], batch size: 191, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:40:11,092 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 1.903e+02 2.161e+02 2.490e+02 3.592e+02, threshold=4.322e+02, percent-clipped=0.0 2023-04-30 21:40:57,690 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187496.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:41:05,658 INFO [train.py:904] (6/8) Epoch 19, batch 4800, loss[loss=0.1705, simple_loss=0.2628, pruned_loss=0.03915, over 16524.00 frames. ], tot_loss[loss=0.1874, simple_loss=0.2749, pruned_loss=0.04992, over 3212067.81 frames. ], batch size: 68, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:41:21,429 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=187512.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:41:28,244 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1062, 3.4002, 3.4690, 1.9555, 2.8722, 2.3589, 3.5355, 3.5685], device='cuda:6'), covar=tensor([0.0249, 0.0732, 0.0601, 0.2075, 0.0912, 0.0940, 0.0641, 0.0883], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0160, 0.0165, 0.0150, 0.0142, 0.0127, 0.0142, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 21:41:30,322 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=187519.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:41:37,745 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=187523.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:41:40,843 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5363, 3.5169, 3.4974, 2.7381, 3.3477, 1.9884, 3.1744, 2.8814], device='cuda:6'), covar=tensor([0.0152, 0.0171, 0.0157, 0.0298, 0.0117, 0.2385, 0.0152, 0.0267], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0145, 0.0190, 0.0174, 0.0167, 0.0200, 0.0181, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:42:08,454 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187544.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:42:08,935 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-30 21:42:19,531 INFO [train.py:904] (6/8) Epoch 19, batch 4850, loss[loss=0.1866, simple_loss=0.2743, pruned_loss=0.04945, over 16746.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2755, pruned_loss=0.04911, over 3196386.85 frames. ], batch size: 57, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:42:24,936 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2118, 2.9646, 3.0737, 1.7955, 3.2584, 3.3003, 2.7792, 2.5051], device='cuda:6'), covar=tensor([0.0794, 0.0244, 0.0194, 0.1170, 0.0072, 0.0159, 0.0386, 0.0522], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0107, 0.0096, 0.0137, 0.0077, 0.0122, 0.0126, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 21:42:36,203 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 1.857e+02 2.110e+02 2.587e+02 7.641e+02, threshold=4.221e+02, percent-clipped=2.0 2023-04-30 21:42:42,007 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187567.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:42:51,143 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=187573.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:43:02,094 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=187580.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:43:07,618 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=187584.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:43:18,365 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2068, 4.1753, 4.1633, 2.6047, 3.5943, 4.1096, 3.6453, 2.1033], device='cuda:6'), covar=tensor([0.0603, 0.0047, 0.0041, 0.0430, 0.0106, 0.0125, 0.0117, 0.0526], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0078, 0.0079, 0.0131, 0.0094, 0.0104, 0.0091, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 21:43:32,170 INFO [train.py:904] (6/8) Epoch 19, batch 4900, loss[loss=0.171, simple_loss=0.2696, pruned_loss=0.03622, over 16713.00 frames. ], tot_loss[loss=0.1848, simple_loss=0.2742, pruned_loss=0.04767, over 3201335.04 frames. ], batch size: 134, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:43:51,078 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187615.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:44:25,630 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4748, 3.6000, 2.0763, 3.9461, 2.6633, 3.9075, 2.1584, 2.7929], device='cuda:6'), covar=tensor([0.0278, 0.0303, 0.1634, 0.0141, 0.0868, 0.0562, 0.1649, 0.0742], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0173, 0.0190, 0.0154, 0.0172, 0.0211, 0.0196, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 21:44:43,103 INFO [train.py:904] (6/8) Epoch 19, batch 4950, loss[loss=0.1809, simple_loss=0.2768, pruned_loss=0.04245, over 15364.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2738, pruned_loss=0.04713, over 3205153.19 frames. ], batch size: 191, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:44:58,341 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 2.032e+02 2.342e+02 2.691e+02 5.269e+02, threshold=4.685e+02, percent-clipped=1.0 2023-04-30 21:45:24,872 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187681.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 21:45:39,735 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5245, 4.5279, 4.8047, 4.7822, 4.8062, 4.4890, 4.4563, 4.3516], device='cuda:6'), covar=tensor([0.0268, 0.0445, 0.0363, 0.0385, 0.0521, 0.0337, 0.0918, 0.0424], device='cuda:6'), in_proj_covar=tensor([0.0385, 0.0425, 0.0414, 0.0386, 0.0458, 0.0435, 0.0522, 0.0343], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 21:45:43,947 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187694.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:45:55,107 INFO [train.py:904] (6/8) Epoch 19, batch 5000, loss[loss=0.2061, simple_loss=0.2885, pruned_loss=0.06183, over 12219.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2755, pruned_loss=0.04754, over 3198721.32 frames. ], batch size: 246, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:46:32,202 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187728.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:46:33,445 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187729.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:46:52,689 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187742.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:47:07,085 INFO [train.py:904] (6/8) Epoch 19, batch 5050, loss[loss=0.216, simple_loss=0.3021, pruned_loss=0.06492, over 12288.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2755, pruned_loss=0.04691, over 3208042.63 frames. ], batch size: 246, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:47:21,720 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.446e+02 2.150e+02 2.460e+02 2.787e+02 5.516e+02, threshold=4.919e+02, percent-clipped=2.0 2023-04-30 21:47:35,301 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5095, 3.4728, 3.4439, 2.7653, 3.3008, 2.1366, 3.1221, 2.7874], device='cuda:6'), covar=tensor([0.0160, 0.0137, 0.0162, 0.0302, 0.0115, 0.2181, 0.0158, 0.0252], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0144, 0.0188, 0.0174, 0.0166, 0.0199, 0.0180, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:48:17,850 INFO [train.py:904] (6/8) Epoch 19, batch 5100, loss[loss=0.1707, simple_loss=0.2735, pruned_loss=0.03398, over 16766.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2743, pruned_loss=0.04627, over 3204687.92 frames. ], batch size: 89, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:49:06,723 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9307, 4.7479, 4.9537, 5.1402, 5.3406, 4.7646, 5.3312, 5.3130], device='cuda:6'), covar=tensor([0.1604, 0.1312, 0.1568, 0.0702, 0.0412, 0.0738, 0.0410, 0.0550], device='cuda:6'), in_proj_covar=tensor([0.0608, 0.0749, 0.0880, 0.0768, 0.0573, 0.0599, 0.0615, 0.0718], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:49:30,691 INFO [train.py:904] (6/8) Epoch 19, batch 5150, loss[loss=0.1752, simple_loss=0.271, pruned_loss=0.03967, over 16923.00 frames. ], tot_loss[loss=0.1831, simple_loss=0.2744, pruned_loss=0.04586, over 3186598.28 frames. ], batch size: 96, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:49:47,461 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.389e+02 2.013e+02 2.291e+02 2.768e+02 7.535e+02, threshold=4.582e+02, percent-clipped=1.0 2023-04-30 21:49:54,446 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187868.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:50:04,808 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187875.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:50:11,357 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187879.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:50:41,939 INFO [train.py:904] (6/8) Epoch 19, batch 5200, loss[loss=0.1758, simple_loss=0.2614, pruned_loss=0.04512, over 16482.00 frames. ], tot_loss[loss=0.182, simple_loss=0.273, pruned_loss=0.04547, over 3188189.03 frames. ], batch size: 68, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:50:50,320 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6106, 3.5362, 2.7197, 2.2087, 2.3786, 2.3852, 3.6460, 3.2654], device='cuda:6'), covar=tensor([0.2592, 0.0643, 0.1795, 0.2653, 0.2363, 0.1906, 0.0578, 0.1146], device='cuda:6'), in_proj_covar=tensor([0.0321, 0.0264, 0.0299, 0.0304, 0.0292, 0.0249, 0.0287, 0.0328], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 21:51:53,322 INFO [train.py:904] (6/8) Epoch 19, batch 5250, loss[loss=0.1793, simple_loss=0.2637, pruned_loss=0.04746, over 16683.00 frames. ], tot_loss[loss=0.1809, simple_loss=0.2711, pruned_loss=0.04533, over 3196987.75 frames. ], batch size: 62, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:52:08,308 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 1.924e+02 2.321e+02 2.643e+02 4.137e+02, threshold=4.643e+02, percent-clipped=0.0 2023-04-30 21:53:07,116 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-30 21:53:08,598 INFO [train.py:904] (6/8) Epoch 19, batch 5300, loss[loss=0.1535, simple_loss=0.2392, pruned_loss=0.03384, over 16754.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.267, pruned_loss=0.04412, over 3214898.61 frames. ], batch size: 83, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:53:46,805 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=188028.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:53:52,665 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8027, 4.0714, 3.7405, 3.5444, 3.1858, 3.9447, 3.6061, 3.6305], device='cuda:6'), covar=tensor([0.0895, 0.0618, 0.0512, 0.0421, 0.1502, 0.0566, 0.1576, 0.0712], device='cuda:6'), in_proj_covar=tensor([0.0283, 0.0405, 0.0334, 0.0324, 0.0345, 0.0379, 0.0228, 0.0396], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:54:04,370 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-30 21:54:21,263 INFO [train.py:904] (6/8) Epoch 19, batch 5350, loss[loss=0.1867, simple_loss=0.2718, pruned_loss=0.05078, over 16150.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2651, pruned_loss=0.04339, over 3211681.96 frames. ], batch size: 35, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:54:24,804 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5595, 3.5217, 3.4878, 2.7710, 3.3903, 2.0123, 3.1800, 2.8340], device='cuda:6'), covar=tensor([0.0134, 0.0122, 0.0154, 0.0286, 0.0102, 0.2446, 0.0139, 0.0263], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0144, 0.0188, 0.0173, 0.0165, 0.0199, 0.0180, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:54:28,924 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6120, 3.8441, 2.6519, 2.2118, 2.4541, 2.2809, 4.0438, 3.3849], device='cuda:6'), covar=tensor([0.2840, 0.0658, 0.2058, 0.2553, 0.2606, 0.2121, 0.0512, 0.1135], device='cuda:6'), in_proj_covar=tensor([0.0319, 0.0263, 0.0297, 0.0302, 0.0290, 0.0248, 0.0286, 0.0326], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 21:54:38,013 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 1.961e+02 2.159e+02 2.589e+02 4.705e+02, threshold=4.317e+02, percent-clipped=1.0 2023-04-30 21:54:57,086 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=188076.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:55:35,753 INFO [train.py:904] (6/8) Epoch 19, batch 5400, loss[loss=0.1841, simple_loss=0.28, pruned_loss=0.04411, over 16512.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.268, pruned_loss=0.04437, over 3221196.71 frames. ], batch size: 68, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:56:43,153 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-30 21:56:53,444 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7451, 2.5177, 2.3955, 3.3239, 2.2292, 3.5778, 1.5873, 2.6527], device='cuda:6'), covar=tensor([0.1263, 0.0717, 0.1126, 0.0186, 0.0170, 0.0421, 0.1582, 0.0808], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0172, 0.0192, 0.0182, 0.0203, 0.0213, 0.0196, 0.0189], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 21:56:54,045 INFO [train.py:904] (6/8) Epoch 19, batch 5450, loss[loss=0.2479, simple_loss=0.3225, pruned_loss=0.08661, over 15299.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2711, pruned_loss=0.04565, over 3217247.79 frames. ], batch size: 190, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:57:02,852 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3730, 3.3605, 3.3737, 3.4626, 3.4889, 3.2717, 3.4782, 3.5411], device='cuda:6'), covar=tensor([0.1405, 0.1016, 0.1214, 0.0738, 0.0727, 0.2605, 0.1183, 0.0977], device='cuda:6'), in_proj_covar=tensor([0.0617, 0.0759, 0.0894, 0.0780, 0.0581, 0.0609, 0.0624, 0.0728], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:57:11,918 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 1.998e+02 2.677e+02 3.507e+02 6.066e+02, threshold=5.353e+02, percent-clipped=8.0 2023-04-30 21:57:20,290 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=188168.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:57:32,262 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=188175.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:57:38,528 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=188179.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:58:14,499 INFO [train.py:904] (6/8) Epoch 19, batch 5500, loss[loss=0.2438, simple_loss=0.3261, pruned_loss=0.08071, over 16815.00 frames. ], tot_loss[loss=0.1895, simple_loss=0.2783, pruned_loss=0.05034, over 3167181.26 frames. ], batch size: 116, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:58:20,584 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7290, 1.4240, 1.7335, 1.6761, 1.8182, 1.9036, 1.5805, 1.8739], device='cuda:6'), covar=tensor([0.0218, 0.0319, 0.0173, 0.0234, 0.0240, 0.0147, 0.0341, 0.0112], device='cuda:6'), in_proj_covar=tensor([0.0182, 0.0189, 0.0177, 0.0180, 0.0189, 0.0149, 0.0192, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:58:38,577 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=188216.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:58:48,994 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=188223.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:58:52,405 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1654, 2.2080, 2.2233, 3.8714, 2.0663, 2.6630, 2.3126, 2.3708], device='cuda:6'), covar=tensor([0.1179, 0.3212, 0.2598, 0.0481, 0.3870, 0.2047, 0.3088, 0.3102], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0431, 0.0356, 0.0319, 0.0426, 0.0498, 0.0402, 0.0502], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:58:55,304 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=188227.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:59:03,236 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6571, 4.9520, 4.7160, 4.7335, 4.4572, 4.4040, 4.3955, 5.0050], device='cuda:6'), covar=tensor([0.1207, 0.0826, 0.1011, 0.0827, 0.0780, 0.1233, 0.1161, 0.0860], device='cuda:6'), in_proj_covar=tensor([0.0643, 0.0794, 0.0653, 0.0589, 0.0499, 0.0504, 0.0657, 0.0611], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:59:26,343 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7127, 1.7884, 2.3915, 2.6380, 2.5962, 3.0349, 1.8625, 3.0180], device='cuda:6'), covar=tensor([0.0203, 0.0482, 0.0281, 0.0311, 0.0264, 0.0177, 0.0530, 0.0141], device='cuda:6'), in_proj_covar=tensor([0.0182, 0.0189, 0.0177, 0.0181, 0.0190, 0.0149, 0.0192, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:59:35,425 INFO [train.py:904] (6/8) Epoch 19, batch 5550, loss[loss=0.2909, simple_loss=0.3557, pruned_loss=0.1131, over 11607.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.2853, pruned_loss=0.05513, over 3149614.72 frames. ], batch size: 247, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:59:52,923 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7668, 1.8026, 1.5303, 1.4966, 1.9685, 1.5905, 1.6816, 1.9416], device='cuda:6'), covar=tensor([0.0180, 0.0284, 0.0392, 0.0304, 0.0203, 0.0236, 0.0175, 0.0180], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0228, 0.0220, 0.0220, 0.0229, 0.0227, 0.0230, 0.0223], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 21:59:53,551 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.484e+02 3.263e+02 3.931e+02 4.573e+02 1.048e+03, threshold=7.861e+02, percent-clipped=6.0 2023-04-30 21:59:56,524 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6151, 1.7990, 2.2848, 2.5254, 2.5575, 2.8357, 1.8494, 2.8368], device='cuda:6'), covar=tensor([0.0191, 0.0440, 0.0288, 0.0295, 0.0272, 0.0184, 0.0504, 0.0113], device='cuda:6'), in_proj_covar=tensor([0.0182, 0.0189, 0.0177, 0.0180, 0.0190, 0.0149, 0.0192, 0.0144], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:00:26,048 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=188283.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:00:55,098 INFO [train.py:904] (6/8) Epoch 19, batch 5600, loss[loss=0.2182, simple_loss=0.2984, pruned_loss=0.06902, over 16690.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2901, pruned_loss=0.05894, over 3125375.71 frames. ], batch size: 57, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:01:25,715 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.18 vs. limit=5.0 2023-04-30 22:01:55,804 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6713, 2.7211, 2.6506, 4.2873, 3.0844, 3.9445, 1.5811, 2.8203], device='cuda:6'), covar=tensor([0.1409, 0.0766, 0.1142, 0.0189, 0.0256, 0.0426, 0.1633, 0.0837], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0172, 0.0192, 0.0183, 0.0204, 0.0214, 0.0197, 0.0189], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 22:02:07,200 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=188344.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:02:19,465 INFO [train.py:904] (6/8) Epoch 19, batch 5650, loss[loss=0.2691, simple_loss=0.3287, pruned_loss=0.1047, over 11222.00 frames. ], tot_loss[loss=0.2127, simple_loss=0.2965, pruned_loss=0.06445, over 3062486.42 frames. ], batch size: 248, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:02:36,991 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.386e+02 3.211e+02 3.758e+02 5.159e+02 1.051e+03, threshold=7.516e+02, percent-clipped=3.0 2023-04-30 22:03:04,676 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4761, 4.5923, 4.7098, 4.5549, 4.5830, 5.1507, 4.6638, 4.4259], device='cuda:6'), covar=tensor([0.1380, 0.1941, 0.2458, 0.2143, 0.2608, 0.1098, 0.1777, 0.2599], device='cuda:6'), in_proj_covar=tensor([0.0398, 0.0569, 0.0628, 0.0480, 0.0638, 0.0662, 0.0492, 0.0643], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 22:03:36,659 INFO [train.py:904] (6/8) Epoch 19, batch 5700, loss[loss=0.1881, simple_loss=0.2878, pruned_loss=0.04421, over 16795.00 frames. ], tot_loss[loss=0.2139, simple_loss=0.2974, pruned_loss=0.06521, over 3076586.91 frames. ], batch size: 83, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:04:28,486 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3869, 3.1302, 2.9692, 1.9921, 2.6175, 2.1612, 2.8467, 3.3173], device='cuda:6'), covar=tensor([0.0428, 0.0696, 0.0631, 0.2091, 0.1033, 0.1050, 0.1012, 0.0909], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0160, 0.0166, 0.0151, 0.0144, 0.0128, 0.0143, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 22:04:55,839 INFO [train.py:904] (6/8) Epoch 19, batch 5750, loss[loss=0.2456, simple_loss=0.3096, pruned_loss=0.09077, over 11449.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.2999, pruned_loss=0.06686, over 3041444.70 frames. ], batch size: 248, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:05:12,679 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.963e+02 2.939e+02 3.429e+02 4.149e+02 7.338e+02, threshold=6.857e+02, percent-clipped=0.0 2023-04-30 22:05:45,226 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=188481.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:06:19,317 INFO [train.py:904] (6/8) Epoch 19, batch 5800, loss[loss=0.1835, simple_loss=0.2693, pruned_loss=0.04882, over 16192.00 frames. ], tot_loss[loss=0.2147, simple_loss=0.2992, pruned_loss=0.06509, over 3066687.97 frames. ], batch size: 35, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:07:24,045 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=188542.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:07:38,949 INFO [train.py:904] (6/8) Epoch 19, batch 5850, loss[loss=0.2301, simple_loss=0.3117, pruned_loss=0.07423, over 16687.00 frames. ], tot_loss[loss=0.2126, simple_loss=0.2975, pruned_loss=0.06384, over 3058842.30 frames. ], batch size: 134, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:07:57,309 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.796e+02 2.681e+02 3.202e+02 3.753e+02 8.228e+02, threshold=6.404e+02, percent-clipped=1.0 2023-04-30 22:08:59,599 INFO [train.py:904] (6/8) Epoch 19, batch 5900, loss[loss=0.2067, simple_loss=0.2955, pruned_loss=0.05897, over 16871.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.2968, pruned_loss=0.06317, over 3072672.96 frames. ], batch size: 90, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:09:58,739 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7581, 1.7214, 2.3198, 2.6484, 2.5783, 3.0193, 1.8921, 3.0085], device='cuda:6'), covar=tensor([0.0175, 0.0488, 0.0293, 0.0294, 0.0264, 0.0168, 0.0502, 0.0131], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0188, 0.0175, 0.0178, 0.0189, 0.0148, 0.0191, 0.0142], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:10:01,929 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=188639.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:10:18,957 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0902, 5.6475, 5.7836, 5.4565, 5.5910, 6.1306, 5.5730, 5.3503], device='cuda:6'), covar=tensor([0.0847, 0.1802, 0.2349, 0.1825, 0.2166, 0.0842, 0.1438, 0.2137], device='cuda:6'), in_proj_covar=tensor([0.0397, 0.0572, 0.0631, 0.0480, 0.0636, 0.0664, 0.0491, 0.0642], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 22:10:21,848 INFO [train.py:904] (6/8) Epoch 19, batch 5950, loss[loss=0.2241, simple_loss=0.306, pruned_loss=0.07112, over 15331.00 frames. ], tot_loss[loss=0.2101, simple_loss=0.2971, pruned_loss=0.06151, over 3082688.34 frames. ], batch size: 191, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:10:40,535 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.478e+02 2.950e+02 3.832e+02 6.444e+02, threshold=5.900e+02, percent-clipped=1.0 2023-04-30 22:11:00,160 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3646, 2.1302, 1.6487, 1.9186, 2.4498, 2.1132, 2.1976, 2.5518], device='cuda:6'), covar=tensor([0.0165, 0.0367, 0.0504, 0.0436, 0.0228, 0.0385, 0.0179, 0.0235], device='cuda:6'), in_proj_covar=tensor([0.0195, 0.0226, 0.0219, 0.0219, 0.0227, 0.0226, 0.0228, 0.0221], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:11:41,786 INFO [train.py:904] (6/8) Epoch 19, batch 6000, loss[loss=0.2161, simple_loss=0.2991, pruned_loss=0.06649, over 16935.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2956, pruned_loss=0.06049, over 3119275.12 frames. ], batch size: 109, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:11:41,786 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 22:11:52,558 INFO [train.py:938] (6/8) Epoch 19, validation: loss=0.1523, simple_loss=0.2653, pruned_loss=0.01968, over 944034.00 frames. 2023-04-30 22:11:52,559 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 22:13:13,862 INFO [train.py:904] (6/8) Epoch 19, batch 6050, loss[loss=0.1692, simple_loss=0.2673, pruned_loss=0.03551, over 17045.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2943, pruned_loss=0.06017, over 3123860.09 frames. ], batch size: 55, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:13:33,104 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 2.719e+02 3.422e+02 3.983e+02 7.831e+02, threshold=6.844e+02, percent-clipped=4.0 2023-04-30 22:14:19,840 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7977, 2.6535, 2.5008, 1.9583, 2.5262, 2.5820, 2.5648, 1.9056], device='cuda:6'), covar=tensor([0.0428, 0.0086, 0.0089, 0.0364, 0.0141, 0.0141, 0.0113, 0.0384], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0079, 0.0080, 0.0132, 0.0095, 0.0105, 0.0092, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 22:14:24,847 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0349, 5.1578, 5.5539, 5.4904, 5.5330, 5.1650, 5.0920, 4.8702], device='cuda:6'), covar=tensor([0.0350, 0.0504, 0.0320, 0.0395, 0.0416, 0.0364, 0.0999, 0.0453], device='cuda:6'), in_proj_covar=tensor([0.0392, 0.0431, 0.0417, 0.0392, 0.0464, 0.0441, 0.0533, 0.0350], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-04-30 22:14:32,941 INFO [train.py:904] (6/8) Epoch 19, batch 6100, loss[loss=0.2099, simple_loss=0.2913, pruned_loss=0.06427, over 15429.00 frames. ], tot_loss[loss=0.2063, simple_loss=0.2938, pruned_loss=0.05945, over 3120797.00 frames. ], batch size: 190, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:14:42,814 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-30 22:15:33,829 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=188837.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:15:50,308 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5440, 3.6579, 2.7680, 2.1490, 2.3589, 2.4117, 3.8973, 3.3540], device='cuda:6'), covar=tensor([0.2932, 0.0639, 0.1832, 0.2739, 0.2585, 0.1963, 0.0457, 0.1162], device='cuda:6'), in_proj_covar=tensor([0.0323, 0.0266, 0.0299, 0.0305, 0.0293, 0.0250, 0.0289, 0.0328], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 22:15:57,514 INFO [train.py:904] (6/8) Epoch 19, batch 6150, loss[loss=0.2032, simple_loss=0.2884, pruned_loss=0.05899, over 16894.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2923, pruned_loss=0.05928, over 3088889.10 frames. ], batch size: 116, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:16:16,882 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.816e+02 2.888e+02 3.402e+02 4.182e+02 1.052e+03, threshold=6.804e+02, percent-clipped=3.0 2023-04-30 22:16:24,538 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5268, 4.3653, 4.2249, 3.0319, 3.8248, 4.3068, 3.8343, 2.4411], device='cuda:6'), covar=tensor([0.0475, 0.0036, 0.0040, 0.0318, 0.0093, 0.0096, 0.0077, 0.0397], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0079, 0.0080, 0.0132, 0.0095, 0.0105, 0.0091, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 22:17:16,331 INFO [train.py:904] (6/8) Epoch 19, batch 6200, loss[loss=0.2181, simple_loss=0.2965, pruned_loss=0.06988, over 12250.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2895, pruned_loss=0.05863, over 3101560.19 frames. ], batch size: 248, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:17:58,845 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=188929.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:18:01,851 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=188931.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:18:14,985 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=188939.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:18:34,123 INFO [train.py:904] (6/8) Epoch 19, batch 6250, loss[loss=0.1881, simple_loss=0.2861, pruned_loss=0.04506, over 16860.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2895, pruned_loss=0.05895, over 3094417.58 frames. ], batch size: 102, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:18:52,888 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.881e+02 2.827e+02 3.410e+02 4.287e+02 1.283e+03, threshold=6.821e+02, percent-clipped=4.0 2023-04-30 22:19:28,072 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=188987.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:19:32,877 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=188990.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:19:36,042 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=188992.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:19:45,211 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8948, 5.1914, 4.9507, 4.9653, 4.7374, 4.6310, 4.6518, 5.2701], device='cuda:6'), covar=tensor([0.1216, 0.0839, 0.0966, 0.0826, 0.0886, 0.1022, 0.1163, 0.0853], device='cuda:6'), in_proj_covar=tensor([0.0635, 0.0781, 0.0641, 0.0582, 0.0489, 0.0500, 0.0650, 0.0599], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:19:51,018 INFO [train.py:904] (6/8) Epoch 19, batch 6300, loss[loss=0.2314, simple_loss=0.2997, pruned_loss=0.08158, over 11693.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2896, pruned_loss=0.0589, over 3090204.24 frames. ], batch size: 247, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:19:59,765 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2489, 2.3061, 2.3372, 4.1348, 2.1920, 2.6011, 2.3819, 2.4520], device='cuda:6'), covar=tensor([0.1292, 0.3401, 0.2741, 0.0471, 0.4009, 0.2517, 0.3359, 0.3150], device='cuda:6'), in_proj_covar=tensor([0.0389, 0.0431, 0.0355, 0.0317, 0.0428, 0.0497, 0.0402, 0.0502], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:20:14,076 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-30 22:20:35,820 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189030.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:21:09,035 INFO [train.py:904] (6/8) Epoch 19, batch 6350, loss[loss=0.2024, simple_loss=0.2827, pruned_loss=0.06107, over 16166.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2908, pruned_loss=0.06026, over 3073543.33 frames. ], batch size: 165, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:21:27,080 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.790e+02 2.691e+02 3.246e+02 4.132e+02 1.028e+03, threshold=6.492e+02, percent-clipped=3.0 2023-04-30 22:21:28,340 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189064.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:21:51,911 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189079.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:22:11,126 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189091.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:22:26,881 INFO [train.py:904] (6/8) Epoch 19, batch 6400, loss[loss=0.1711, simple_loss=0.2675, pruned_loss=0.03736, over 16823.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2905, pruned_loss=0.06081, over 3072611.35 frames. ], batch size: 102, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:23:02,727 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189125.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:23:21,551 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189137.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:23:25,660 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189140.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:23:43,274 INFO [train.py:904] (6/8) Epoch 19, batch 6450, loss[loss=0.1875, simple_loss=0.2746, pruned_loss=0.05024, over 16875.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2901, pruned_loss=0.05938, over 3101870.94 frames. ], batch size: 116, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:24:00,575 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-30 22:24:01,000 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.834e+02 2.902e+02 3.401e+02 4.159e+02 9.749e+02, threshold=6.803e+02, percent-clipped=3.0 2023-04-30 22:24:19,168 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2023-04-30 22:24:34,447 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189185.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:25:00,513 INFO [train.py:904] (6/8) Epoch 19, batch 6500, loss[loss=0.1976, simple_loss=0.269, pruned_loss=0.06316, over 11476.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2881, pruned_loss=0.05879, over 3096730.36 frames. ], batch size: 248, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:26:14,655 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189248.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:26:19,681 INFO [train.py:904] (6/8) Epoch 19, batch 6550, loss[loss=0.2265, simple_loss=0.3227, pruned_loss=0.06515, over 16348.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2904, pruned_loss=0.05942, over 3104786.30 frames. ], batch size: 146, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:26:37,017 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.957e+02 2.752e+02 3.137e+02 4.130e+02 8.401e+02, threshold=6.273e+02, percent-clipped=2.0 2023-04-30 22:26:38,442 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.00 vs. limit=5.0 2023-04-30 22:26:41,316 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3342, 2.0982, 1.6822, 1.8226, 2.3823, 2.0213, 2.1473, 2.4676], device='cuda:6'), covar=tensor([0.0236, 0.0411, 0.0503, 0.0492, 0.0260, 0.0391, 0.0235, 0.0285], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0224, 0.0217, 0.0217, 0.0225, 0.0223, 0.0225, 0.0220], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:26:48,776 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7405, 2.7814, 2.4579, 4.1868, 2.9663, 3.9533, 1.4161, 2.8829], device='cuda:6'), covar=tensor([0.1349, 0.0750, 0.1295, 0.0199, 0.0288, 0.0492, 0.1783, 0.0838], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0171, 0.0193, 0.0183, 0.0205, 0.0214, 0.0197, 0.0189], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 22:27:09,637 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189285.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:27:12,562 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189287.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:27:34,958 INFO [train.py:904] (6/8) Epoch 19, batch 6600, loss[loss=0.2229, simple_loss=0.3018, pruned_loss=0.07201, over 16194.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2934, pruned_loss=0.06042, over 3094018.53 frames. ], batch size: 165, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:27:45,691 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189309.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:27:52,345 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0820, 2.4217, 2.5992, 1.9619, 2.6955, 2.7954, 2.3846, 2.3627], device='cuda:6'), covar=tensor([0.0688, 0.0254, 0.0238, 0.0904, 0.0113, 0.0281, 0.0435, 0.0442], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0107, 0.0097, 0.0139, 0.0078, 0.0122, 0.0126, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 22:28:44,313 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189347.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:28:50,942 INFO [train.py:904] (6/8) Epoch 19, batch 6650, loss[loss=0.1962, simple_loss=0.2808, pruned_loss=0.05586, over 16487.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2941, pruned_loss=0.06168, over 3075043.83 frames. ], batch size: 68, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:29:08,220 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.792e+02 2.870e+02 3.524e+02 4.277e+02 9.308e+02, threshold=7.047e+02, percent-clipped=5.0 2023-04-30 22:29:17,762 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-30 22:29:42,458 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189386.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:30:06,163 INFO [train.py:904] (6/8) Epoch 19, batch 6700, loss[loss=0.2121, simple_loss=0.3004, pruned_loss=0.06189, over 16434.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2925, pruned_loss=0.06158, over 3084867.33 frames. ], batch size: 146, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:30:15,390 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189408.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:30:27,987 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5238, 3.5276, 3.4810, 2.7506, 3.3954, 2.0418, 3.1460, 2.7861], device='cuda:6'), covar=tensor([0.0147, 0.0130, 0.0174, 0.0220, 0.0097, 0.2269, 0.0127, 0.0251], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0145, 0.0191, 0.0174, 0.0166, 0.0200, 0.0180, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:30:33,778 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189420.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:30:56,678 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189435.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:31:21,912 INFO [train.py:904] (6/8) Epoch 19, batch 6750, loss[loss=0.2086, simple_loss=0.3015, pruned_loss=0.05782, over 16729.00 frames. ], tot_loss[loss=0.2067, simple_loss=0.2915, pruned_loss=0.06099, over 3098876.74 frames. ], batch size: 89, lr: 3.55e-03, grad_scale: 4.0 2023-04-30 22:31:42,235 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 2.911e+02 3.292e+02 3.890e+02 7.012e+02, threshold=6.585e+02, percent-clipped=0.0 2023-04-30 22:32:01,418 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6841, 4.6847, 4.4735, 3.8223, 4.5993, 1.7423, 4.3499, 4.1605], device='cuda:6'), covar=tensor([0.0085, 0.0069, 0.0178, 0.0364, 0.0081, 0.2715, 0.0118, 0.0261], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0145, 0.0191, 0.0174, 0.0166, 0.0199, 0.0180, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:32:35,703 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 22:32:38,375 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189501.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:32:39,012 INFO [train.py:904] (6/8) Epoch 19, batch 6800, loss[loss=0.2169, simple_loss=0.303, pruned_loss=0.06545, over 16905.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2918, pruned_loss=0.06125, over 3108299.69 frames. ], batch size: 109, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:33:24,343 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4290, 4.6877, 4.4320, 4.4818, 4.2484, 4.1922, 4.2196, 4.7120], device='cuda:6'), covar=tensor([0.1113, 0.0863, 0.1076, 0.0832, 0.0765, 0.1394, 0.1091, 0.0951], device='cuda:6'), in_proj_covar=tensor([0.0641, 0.0782, 0.0646, 0.0584, 0.0490, 0.0503, 0.0652, 0.0604], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:33:58,198 INFO [train.py:904] (6/8) Epoch 19, batch 6850, loss[loss=0.2123, simple_loss=0.3162, pruned_loss=0.05421, over 16823.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2928, pruned_loss=0.06094, over 3113762.31 frames. ], batch size: 116, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:34:13,636 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189562.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:34:17,024 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.892e+02 2.732e+02 3.346e+02 4.467e+02 6.913e+02, threshold=6.693e+02, percent-clipped=4.0 2023-04-30 22:34:47,772 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189585.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:34:50,362 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189587.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:35:07,976 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.63 vs. limit=5.0 2023-04-30 22:35:12,995 INFO [train.py:904] (6/8) Epoch 19, batch 6900, loss[loss=0.2138, simple_loss=0.3123, pruned_loss=0.05766, over 16739.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2955, pruned_loss=0.06144, over 3106274.08 frames. ], batch size: 124, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:35:17,326 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189604.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:36:00,957 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189633.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:36:04,231 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189635.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:36:12,743 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189640.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:36:29,233 INFO [train.py:904] (6/8) Epoch 19, batch 6950, loss[loss=0.2388, simple_loss=0.3107, pruned_loss=0.08346, over 11423.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2967, pruned_loss=0.06287, over 3084342.33 frames. ], batch size: 246, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:36:48,879 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.017e+02 3.156e+02 3.874e+02 4.949e+02 9.528e+02, threshold=7.748e+02, percent-clipped=6.0 2023-04-30 22:37:21,366 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189686.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:37:43,816 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189701.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:37:44,541 INFO [train.py:904] (6/8) Epoch 19, batch 7000, loss[loss=0.21, simple_loss=0.2872, pruned_loss=0.06637, over 11357.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2971, pruned_loss=0.06267, over 3080642.97 frames. ], batch size: 246, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:37:45,791 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1867, 2.1915, 2.1738, 3.8837, 2.0926, 2.5581, 2.2234, 2.3290], device='cuda:6'), covar=tensor([0.1342, 0.3779, 0.3030, 0.0500, 0.4255, 0.2524, 0.3902, 0.3216], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0430, 0.0355, 0.0317, 0.0429, 0.0496, 0.0403, 0.0503], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:37:47,738 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189703.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:37:49,127 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189704.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:38:13,451 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189720.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:38:25,543 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189728.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:38:31,978 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9700, 2.2828, 2.2931, 2.8077, 2.0097, 3.2143, 1.7399, 2.7203], device='cuda:6'), covar=tensor([0.1197, 0.0678, 0.1095, 0.0173, 0.0137, 0.0353, 0.1482, 0.0725], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0173, 0.0195, 0.0185, 0.0207, 0.0215, 0.0198, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 22:38:34,807 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189734.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:38:36,101 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189735.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:39:01,654 INFO [train.py:904] (6/8) Epoch 19, batch 7050, loss[loss=0.2048, simple_loss=0.2928, pruned_loss=0.05839, over 16381.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2976, pruned_loss=0.06246, over 3085518.53 frames. ], batch size: 146, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:39:22,079 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.118e+02 2.966e+02 3.403e+02 4.255e+02 9.294e+02, threshold=6.806e+02, percent-clipped=3.0 2023-04-30 22:39:23,296 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189765.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:39:27,447 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189768.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:39:50,863 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189783.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:40:01,316 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189789.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:40:19,986 INFO [train.py:904] (6/8) Epoch 19, batch 7100, loss[loss=0.2451, simple_loss=0.3167, pruned_loss=0.0867, over 11330.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.2964, pruned_loss=0.06214, over 3082982.45 frames. ], batch size: 247, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:40:33,320 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0861, 4.9087, 5.1031, 5.2687, 5.4814, 4.8348, 5.4664, 5.4472], device='cuda:6'), covar=tensor([0.2016, 0.1229, 0.1688, 0.0745, 0.0515, 0.0854, 0.0532, 0.0600], device='cuda:6'), in_proj_covar=tensor([0.0607, 0.0746, 0.0880, 0.0768, 0.0576, 0.0604, 0.0622, 0.0719], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:41:01,449 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-30 22:41:38,323 INFO [train.py:904] (6/8) Epoch 19, batch 7150, loss[loss=0.2116, simple_loss=0.2942, pruned_loss=0.0645, over 16707.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2939, pruned_loss=0.06151, over 3098623.66 frames. ], batch size: 134, lr: 3.54e-03, grad_scale: 4.0 2023-04-30 22:41:45,621 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189857.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:41:58,949 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.902e+02 2.755e+02 3.301e+02 3.879e+02 9.632e+02, threshold=6.601e+02, percent-clipped=3.0 2023-04-30 22:42:49,039 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1981, 4.1849, 4.0583, 3.3040, 4.1105, 1.7024, 3.9146, 3.6904], device='cuda:6'), covar=tensor([0.0111, 0.0084, 0.0179, 0.0326, 0.0102, 0.2766, 0.0139, 0.0257], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0144, 0.0190, 0.0173, 0.0165, 0.0199, 0.0179, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:42:51,645 INFO [train.py:904] (6/8) Epoch 19, batch 7200, loss[loss=0.2003, simple_loss=0.2866, pruned_loss=0.05696, over 16415.00 frames. ], tot_loss[loss=0.206, simple_loss=0.2919, pruned_loss=0.06003, over 3097153.60 frames. ], batch size: 146, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:42:55,776 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189904.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:43:10,506 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5697, 3.0278, 3.1970, 1.9139, 2.7070, 2.0705, 3.2408, 3.2768], device='cuda:6'), covar=tensor([0.0290, 0.0800, 0.0618, 0.2084, 0.0876, 0.1080, 0.0635, 0.0957], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0161, 0.0167, 0.0152, 0.0145, 0.0129, 0.0144, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 22:43:40,344 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-30 22:44:12,492 INFO [train.py:904] (6/8) Epoch 19, batch 7250, loss[loss=0.1852, simple_loss=0.2731, pruned_loss=0.04865, over 16315.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2894, pruned_loss=0.05862, over 3095113.18 frames. ], batch size: 146, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:44:12,829 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189952.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:44:34,219 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.477e+02 2.873e+02 3.623e+02 8.553e+02, threshold=5.746e+02, percent-clipped=4.0 2023-04-30 22:44:53,733 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0990, 5.3904, 5.1277, 5.1295, 4.8874, 4.7637, 4.7865, 5.4609], device='cuda:6'), covar=tensor([0.1130, 0.0882, 0.0985, 0.0864, 0.0850, 0.0992, 0.1197, 0.0994], device='cuda:6'), in_proj_covar=tensor([0.0636, 0.0776, 0.0642, 0.0579, 0.0486, 0.0499, 0.0647, 0.0599], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:45:19,909 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189996.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:45:32,081 INFO [train.py:904] (6/8) Epoch 19, batch 7300, loss[loss=0.2018, simple_loss=0.2929, pruned_loss=0.05538, over 16594.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2888, pruned_loss=0.05843, over 3103903.04 frames. ], batch size: 62, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:45:33,920 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190003.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:45:47,227 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-30 22:46:11,319 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2343, 4.0445, 3.8211, 2.4777, 3.5453, 3.9450, 3.5522, 2.1769], device='cuda:6'), covar=tensor([0.0536, 0.0037, 0.0053, 0.0433, 0.0099, 0.0092, 0.0091, 0.0422], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0079, 0.0079, 0.0132, 0.0093, 0.0105, 0.0091, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 22:46:48,532 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190051.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:46:49,311 INFO [train.py:904] (6/8) Epoch 19, batch 7350, loss[loss=0.2144, simple_loss=0.294, pruned_loss=0.06745, over 15374.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2894, pruned_loss=0.05913, over 3087165.48 frames. ], batch size: 191, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:47:01,714 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190060.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:47:10,690 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.253e+02 3.016e+02 3.386e+02 4.070e+02 1.285e+03, threshold=6.773e+02, percent-clipped=7.0 2023-04-30 22:47:41,029 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190084.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:48:08,858 INFO [train.py:904] (6/8) Epoch 19, batch 7400, loss[loss=0.1838, simple_loss=0.2807, pruned_loss=0.04343, over 16823.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.291, pruned_loss=0.06013, over 3072983.70 frames. ], batch size: 102, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:48:47,023 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8031, 3.7425, 3.8835, 3.9798, 4.0535, 3.6668, 3.9993, 4.0946], device='cuda:6'), covar=tensor([0.1600, 0.1115, 0.1253, 0.0682, 0.0578, 0.1717, 0.0860, 0.0683], device='cuda:6'), in_proj_covar=tensor([0.0603, 0.0742, 0.0873, 0.0762, 0.0572, 0.0600, 0.0618, 0.0713], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:49:05,314 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-30 22:49:09,908 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190139.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 22:49:16,188 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-30 22:49:30,171 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190151.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:49:30,836 INFO [train.py:904] (6/8) Epoch 19, batch 7450, loss[loss=0.2384, simple_loss=0.3086, pruned_loss=0.0841, over 11235.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2919, pruned_loss=0.06087, over 3071812.49 frames. ], batch size: 247, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:49:41,008 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190157.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:49:55,755 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.975e+02 2.938e+02 3.493e+02 4.428e+02 1.014e+03, threshold=6.986e+02, percent-clipped=6.0 2023-04-30 22:50:44,204 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-30 22:50:51,756 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190200.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 22:50:54,216 INFO [train.py:904] (6/8) Epoch 19, batch 7500, loss[loss=0.1975, simple_loss=0.2753, pruned_loss=0.0598, over 16448.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2924, pruned_loss=0.06044, over 3078928.05 frames. ], batch size: 68, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:50:58,212 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4839, 4.6774, 4.8083, 4.5925, 4.6518, 5.1723, 4.6987, 4.4606], device='cuda:6'), covar=tensor([0.1319, 0.1792, 0.2103, 0.1997, 0.2410, 0.1010, 0.1567, 0.2349], device='cuda:6'), in_proj_covar=tensor([0.0400, 0.0574, 0.0632, 0.0478, 0.0638, 0.0663, 0.0495, 0.0645], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 22:50:59,307 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190205.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:51:10,864 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190212.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:51:26,235 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1514, 2.0379, 1.6508, 1.8106, 2.3151, 1.9869, 2.0477, 2.3886], device='cuda:6'), covar=tensor([0.0183, 0.0374, 0.0470, 0.0408, 0.0232, 0.0342, 0.0182, 0.0229], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0226, 0.0218, 0.0218, 0.0226, 0.0225, 0.0227, 0.0220], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 22:51:30,392 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-30 22:52:13,611 INFO [train.py:904] (6/8) Epoch 19, batch 7550, loss[loss=0.2422, simple_loss=0.3122, pruned_loss=0.08609, over 11391.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2916, pruned_loss=0.06103, over 3065911.15 frames. ], batch size: 246, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:52:34,942 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.804e+02 2.643e+02 3.256e+02 4.054e+02 9.532e+02, threshold=6.511e+02, percent-clipped=2.0 2023-04-30 22:53:22,624 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190296.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:53:31,291 INFO [train.py:904] (6/8) Epoch 19, batch 7600, loss[loss=0.2393, simple_loss=0.3193, pruned_loss=0.07961, over 16255.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2912, pruned_loss=0.06201, over 3052349.07 frames. ], batch size: 165, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:53:34,992 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9051, 2.9681, 3.2281, 1.5571, 3.2602, 3.4723, 2.6799, 2.3776], device='cuda:6'), covar=tensor([0.1144, 0.0244, 0.0196, 0.1401, 0.0089, 0.0168, 0.0483, 0.0662], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0105, 0.0095, 0.0137, 0.0077, 0.0121, 0.0125, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 22:53:53,569 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190316.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:54:37,076 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190344.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:54:48,809 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 22:54:48,951 INFO [train.py:904] (6/8) Epoch 19, batch 7650, loss[loss=0.2174, simple_loss=0.3078, pruned_loss=0.0635, over 16505.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.2914, pruned_loss=0.06171, over 3068023.20 frames. ], batch size: 68, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:55:02,268 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190360.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:55:10,292 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.023e+02 2.803e+02 3.556e+02 4.194e+02 6.713e+02, threshold=7.113e+02, percent-clipped=1.0 2023-04-30 22:55:27,572 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190377.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:55:33,892 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-04-30 22:55:38,608 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190384.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:56:02,445 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190399.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:56:06,245 INFO [train.py:904] (6/8) Epoch 19, batch 7700, loss[loss=0.2409, simple_loss=0.3209, pruned_loss=0.08044, over 11599.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2911, pruned_loss=0.06173, over 3073375.60 frames. ], batch size: 247, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:56:15,128 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190408.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:56:50,414 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190432.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:57:21,407 INFO [train.py:904] (6/8) Epoch 19, batch 7750, loss[loss=0.2342, simple_loss=0.2994, pruned_loss=0.08445, over 11458.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2916, pruned_loss=0.06203, over 3069612.91 frames. ], batch size: 246, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:57:35,227 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190460.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:57:43,808 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.766e+02 2.725e+02 3.462e+02 4.075e+02 8.569e+02, threshold=6.924e+02, percent-clipped=2.0 2023-04-30 22:57:51,494 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7217, 4.9770, 5.1271, 4.9523, 4.9763, 5.5425, 4.9935, 4.7855], device='cuda:6'), covar=tensor([0.1081, 0.1970, 0.2484, 0.2072, 0.2523, 0.0999, 0.1701, 0.2448], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0572, 0.0632, 0.0479, 0.0638, 0.0663, 0.0496, 0.0643], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 22:58:30,671 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190495.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 22:58:39,266 INFO [train.py:904] (6/8) Epoch 19, batch 7800, loss[loss=0.2683, simple_loss=0.3378, pruned_loss=0.09937, over 11537.00 frames. ], tot_loss[loss=0.208, simple_loss=0.292, pruned_loss=0.06196, over 3078823.01 frames. ], batch size: 248, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:58:47,780 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190507.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:59:57,134 INFO [train.py:904] (6/8) Epoch 19, batch 7850, loss[loss=0.2684, simple_loss=0.3262, pruned_loss=0.1053, over 11872.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2927, pruned_loss=0.06174, over 3072895.87 frames. ], batch size: 248, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:00:17,836 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.643e+02 2.838e+02 3.506e+02 4.307e+02 8.316e+02, threshold=7.012e+02, percent-clipped=1.0 2023-04-30 23:00:56,910 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190591.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 23:01:12,371 INFO [train.py:904] (6/8) Epoch 19, batch 7900, loss[loss=0.2202, simple_loss=0.3168, pruned_loss=0.06179, over 16766.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2918, pruned_loss=0.0614, over 3062909.24 frames. ], batch size: 124, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:01:12,950 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5825, 5.5795, 5.3059, 4.6066, 5.4127, 2.1472, 5.1947, 5.1723], device='cuda:6'), covar=tensor([0.0089, 0.0068, 0.0190, 0.0399, 0.0094, 0.2425, 0.0127, 0.0179], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0143, 0.0189, 0.0172, 0.0164, 0.0198, 0.0178, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:01:20,092 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 23:01:23,402 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190609.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:02:32,481 INFO [train.py:904] (6/8) Epoch 19, batch 7950, loss[loss=0.2086, simple_loss=0.2904, pruned_loss=0.06342, over 17221.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2915, pruned_loss=0.06089, over 3090313.77 frames. ], batch size: 44, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:02:33,072 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190652.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 23:02:54,091 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.686e+02 2.695e+02 3.262e+02 3.966e+02 7.542e+02, threshold=6.523e+02, percent-clipped=2.0 2023-04-30 23:03:01,430 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190670.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 23:03:04,352 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190672.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:03:08,643 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190675.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:03:49,817 INFO [train.py:904] (6/8) Epoch 19, batch 8000, loss[loss=0.1973, simple_loss=0.2788, pruned_loss=0.05791, over 16641.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2925, pruned_loss=0.06174, over 3076675.74 frames. ], batch size: 62, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:04:37,714 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4460, 2.6094, 2.0569, 2.2698, 2.9531, 2.5765, 3.0845, 3.1785], device='cuda:6'), covar=tensor([0.0106, 0.0334, 0.0496, 0.0431, 0.0237, 0.0354, 0.0213, 0.0191], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0225, 0.0218, 0.0218, 0.0228, 0.0225, 0.0227, 0.0220], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:04:40,828 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2787, 3.3316, 2.0628, 3.5765, 2.5502, 3.6315, 2.1636, 2.6437], device='cuda:6'), covar=tensor([0.0265, 0.0380, 0.1671, 0.0221, 0.0782, 0.0606, 0.1529, 0.0767], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0174, 0.0193, 0.0155, 0.0173, 0.0213, 0.0199, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 23:04:44,489 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190736.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:05:07,643 INFO [train.py:904] (6/8) Epoch 19, batch 8050, loss[loss=0.2313, simple_loss=0.2966, pruned_loss=0.08304, over 11433.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.2923, pruned_loss=0.0617, over 3071480.34 frames. ], batch size: 249, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:05:12,148 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190755.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:05:28,126 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.906e+02 2.693e+02 3.232e+02 4.056e+02 6.686e+02, threshold=6.463e+02, percent-clipped=1.0 2023-04-30 23:06:07,272 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-04-30 23:06:12,424 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190795.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 23:06:22,312 INFO [train.py:904] (6/8) Epoch 19, batch 8100, loss[loss=0.2243, simple_loss=0.3051, pruned_loss=0.07182, over 15254.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2918, pruned_loss=0.06097, over 3077821.83 frames. ], batch size: 191, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:06:31,296 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190807.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:07:25,606 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190843.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 23:07:38,789 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1913, 2.3307, 2.3804, 4.1378, 2.2331, 2.6239, 2.3882, 2.4628], device='cuda:6'), covar=tensor([0.1350, 0.3492, 0.2801, 0.0462, 0.3954, 0.2376, 0.3373, 0.3141], device='cuda:6'), in_proj_covar=tensor([0.0389, 0.0433, 0.0355, 0.0319, 0.0431, 0.0497, 0.0404, 0.0505], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:07:39,349 INFO [train.py:904] (6/8) Epoch 19, batch 8150, loss[loss=0.1664, simple_loss=0.2562, pruned_loss=0.03829, over 16470.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2896, pruned_loss=0.0597, over 3085806.89 frames. ], batch size: 75, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:07:44,593 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190855.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:08:01,318 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.839e+02 2.760e+02 3.332e+02 4.142e+02 6.995e+02, threshold=6.664e+02, percent-clipped=2.0 2023-04-30 23:08:09,542 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7715, 2.4690, 2.2785, 3.2088, 2.2856, 3.5956, 1.4325, 2.7064], device='cuda:6'), covar=tensor([0.1386, 0.0786, 0.1316, 0.0189, 0.0208, 0.0413, 0.1811, 0.0868], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0174, 0.0194, 0.0185, 0.0209, 0.0215, 0.0199, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 23:08:57,387 INFO [train.py:904] (6/8) Epoch 19, batch 8200, loss[loss=0.193, simple_loss=0.2853, pruned_loss=0.05039, over 16279.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2863, pruned_loss=0.0586, over 3103907.73 frames. ], batch size: 165, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:10:12,363 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190947.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 23:10:19,459 INFO [train.py:904] (6/8) Epoch 19, batch 8250, loss[loss=0.166, simple_loss=0.2708, pruned_loss=0.03065, over 16873.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2855, pruned_loss=0.05605, over 3076879.08 frames. ], batch size: 96, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:10:40,452 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190965.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 23:10:41,154 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.606e+02 3.069e+02 3.589e+02 7.796e+02, threshold=6.137e+02, percent-clipped=2.0 2023-04-30 23:10:51,848 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190972.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:11:39,003 INFO [train.py:904] (6/8) Epoch 19, batch 8300, loss[loss=0.1944, simple_loss=0.292, pruned_loss=0.04843, over 16602.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2831, pruned_loss=0.05328, over 3077121.09 frames. ], batch size: 134, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:12:10,264 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191020.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:12:27,701 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=191031.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:13:00,831 INFO [train.py:904] (6/8) Epoch 19, batch 8350, loss[loss=0.2003, simple_loss=0.2881, pruned_loss=0.05621, over 16724.00 frames. ], tot_loss[loss=0.1925, simple_loss=0.2824, pruned_loss=0.0513, over 3070176.40 frames. ], batch size: 124, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:13:08,036 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191055.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:13:24,343 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.343e+02 2.792e+02 3.340e+02 8.341e+02, threshold=5.585e+02, percent-clipped=1.0 2023-04-30 23:13:28,660 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=191068.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:14:22,694 INFO [train.py:904] (6/8) Epoch 19, batch 8400, loss[loss=0.1675, simple_loss=0.2636, pruned_loss=0.03574, over 15257.00 frames. ], tot_loss[loss=0.1897, simple_loss=0.2802, pruned_loss=0.04964, over 3062504.23 frames. ], batch size: 191, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:14:24,515 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191103.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:15:06,937 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=191129.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:15:11,040 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=191132.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:15:18,308 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2187, 3.2884, 2.1288, 3.5080, 2.4580, 3.5085, 2.1097, 2.7070], device='cuda:6'), covar=tensor([0.0301, 0.0377, 0.1414, 0.0275, 0.0772, 0.0622, 0.1581, 0.0664], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0171, 0.0189, 0.0152, 0.0171, 0.0208, 0.0197, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-04-30 23:15:39,508 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3150, 3.1668, 3.3045, 1.8258, 3.4734, 3.5423, 2.8583, 2.8804], device='cuda:6'), covar=tensor([0.0705, 0.0233, 0.0200, 0.1173, 0.0073, 0.0157, 0.0395, 0.0378], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0105, 0.0094, 0.0136, 0.0076, 0.0119, 0.0124, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-04-30 23:15:42,687 INFO [train.py:904] (6/8) Epoch 19, batch 8450, loss[loss=0.1631, simple_loss=0.2671, pruned_loss=0.02956, over 16749.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2789, pruned_loss=0.0481, over 3062955.31 frames. ], batch size: 89, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:16:06,334 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.247e+02 2.599e+02 3.228e+02 6.628e+02, threshold=5.197e+02, percent-clipped=1.0 2023-04-30 23:16:37,000 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=191185.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:16:50,124 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=191193.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:17:04,704 INFO [train.py:904] (6/8) Epoch 19, batch 8500, loss[loss=0.1692, simple_loss=0.2471, pruned_loss=0.04562, over 11936.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2754, pruned_loss=0.04613, over 3047064.77 frames. ], batch size: 248, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:17:32,773 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6365, 2.3784, 2.3415, 4.6230, 2.3251, 2.7931, 2.4103, 2.5616], device='cuda:6'), covar=tensor([0.1060, 0.3726, 0.3011, 0.0339, 0.4308, 0.2492, 0.3798, 0.3316], device='cuda:6'), in_proj_covar=tensor([0.0382, 0.0424, 0.0350, 0.0313, 0.0422, 0.0487, 0.0396, 0.0495], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:18:22,823 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=191246.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:18:24,994 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191247.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 23:18:32,469 INFO [train.py:904] (6/8) Epoch 19, batch 8550, loss[loss=0.1711, simple_loss=0.2545, pruned_loss=0.04389, over 12040.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2731, pruned_loss=0.04526, over 3021612.32 frames. ], batch size: 247, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:18:44,558 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3884, 2.8826, 3.2522, 1.9071, 2.8485, 2.1296, 2.9647, 3.0528], device='cuda:6'), covar=tensor([0.0331, 0.0923, 0.0480, 0.2083, 0.0772, 0.1029, 0.0702, 0.0906], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0156, 0.0162, 0.0147, 0.0141, 0.0125, 0.0140, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 23:18:57,566 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191265.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:19:00,615 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.506e+02 2.304e+02 2.724e+02 3.153e+02 7.683e+02, threshold=5.448e+02, percent-clipped=1.0 2023-04-30 23:19:57,152 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-30 23:19:57,190 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-30 23:20:00,161 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191295.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 23:20:13,439 INFO [train.py:904] (6/8) Epoch 19, batch 8600, loss[loss=0.1724, simple_loss=0.2596, pruned_loss=0.04261, over 12606.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2734, pruned_loss=0.04426, over 3028955.23 frames. ], batch size: 247, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:20:37,122 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191313.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:21:12,146 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191331.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:21:52,935 INFO [train.py:904] (6/8) Epoch 19, batch 8650, loss[loss=0.1654, simple_loss=0.2629, pruned_loss=0.03392, over 16798.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.272, pruned_loss=0.04314, over 3029776.57 frames. ], batch size: 124, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:22:25,672 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=191365.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:22:29,946 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.144e+02 2.571e+02 3.162e+02 5.065e+02, threshold=5.143e+02, percent-clipped=0.0 2023-04-30 23:22:55,735 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191379.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:23:04,961 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9538, 4.5178, 3.3552, 2.3355, 2.7883, 2.5813, 4.8146, 3.7651], device='cuda:6'), covar=tensor([0.2799, 0.0465, 0.1642, 0.3014, 0.2825, 0.2114, 0.0312, 0.1124], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0259, 0.0294, 0.0298, 0.0286, 0.0246, 0.0281, 0.0321], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-04-30 23:23:39,020 INFO [train.py:904] (6/8) Epoch 19, batch 8700, loss[loss=0.158, simple_loss=0.2554, pruned_loss=0.03028, over 16829.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2696, pruned_loss=0.04169, over 3067850.01 frames. ], batch size: 102, lr: 3.53e-03, grad_scale: 2.0 2023-04-30 23:23:55,510 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-30 23:24:20,726 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=191424.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:24:23,674 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=191426.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:25:14,445 INFO [train.py:904] (6/8) Epoch 19, batch 8750, loss[loss=0.1645, simple_loss=0.2675, pruned_loss=0.0307, over 16623.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2686, pruned_loss=0.04094, over 3060486.67 frames. ], batch size: 62, lr: 3.53e-03, grad_scale: 2.0 2023-04-30 23:25:26,020 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4060, 3.7414, 3.7377, 1.8807, 2.9907, 2.1333, 3.7847, 3.9788], device='cuda:6'), covar=tensor([0.0211, 0.0713, 0.0562, 0.2237, 0.0896, 0.1145, 0.0596, 0.0824], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0154, 0.0161, 0.0147, 0.0140, 0.0125, 0.0139, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 23:25:58,340 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.289e+02 2.109e+02 2.725e+02 3.287e+02 5.129e+02, threshold=5.449e+02, percent-clipped=0.0 2023-04-30 23:26:40,134 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=191488.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:27:07,634 INFO [train.py:904] (6/8) Epoch 19, batch 8800, loss[loss=0.1721, simple_loss=0.2769, pruned_loss=0.03363, over 16875.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2672, pruned_loss=0.03983, over 3087250.83 frames. ], batch size: 96, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:28:25,989 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-30 23:28:29,246 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=191541.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:28:51,736 INFO [train.py:904] (6/8) Epoch 19, batch 8850, loss[loss=0.1691, simple_loss=0.2571, pruned_loss=0.04055, over 12393.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2696, pruned_loss=0.03913, over 3081924.97 frames. ], batch size: 248, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:29:28,627 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.662e+02 2.105e+02 2.539e+02 3.345e+02 5.950e+02, threshold=5.078e+02, percent-clipped=2.0 2023-04-30 23:29:56,525 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2717, 2.3347, 2.0222, 2.0355, 2.7439, 2.3795, 2.7873, 2.9411], device='cuda:6'), covar=tensor([0.0144, 0.0412, 0.0503, 0.0449, 0.0242, 0.0375, 0.0190, 0.0235], device='cuda:6'), in_proj_covar=tensor([0.0188, 0.0224, 0.0215, 0.0216, 0.0224, 0.0221, 0.0222, 0.0216], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:30:39,776 INFO [train.py:904] (6/8) Epoch 19, batch 8900, loss[loss=0.1685, simple_loss=0.2666, pruned_loss=0.03525, over 15268.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2699, pruned_loss=0.03893, over 3072348.47 frames. ], batch size: 190, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:32:04,686 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-30 23:32:32,349 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7135, 2.6775, 2.3040, 2.4602, 3.1260, 2.8447, 3.2221, 3.3446], device='cuda:6'), covar=tensor([0.0097, 0.0397, 0.0502, 0.0432, 0.0240, 0.0325, 0.0248, 0.0227], device='cuda:6'), in_proj_covar=tensor([0.0188, 0.0224, 0.0216, 0.0216, 0.0225, 0.0222, 0.0222, 0.0216], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:32:45,267 INFO [train.py:904] (6/8) Epoch 19, batch 8950, loss[loss=0.18, simple_loss=0.278, pruned_loss=0.04097, over 15207.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2693, pruned_loss=0.03896, over 3087972.17 frames. ], batch size: 190, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:33:21,157 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.162e+02 2.456e+02 2.877e+02 5.499e+02, threshold=4.911e+02, percent-clipped=2.0 2023-04-30 23:34:29,380 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6267, 4.6428, 4.4775, 4.1058, 4.1704, 4.5326, 4.3718, 4.1963], device='cuda:6'), covar=tensor([0.0571, 0.0694, 0.0353, 0.0320, 0.0899, 0.0633, 0.0475, 0.0705], device='cuda:6'), in_proj_covar=tensor([0.0271, 0.0392, 0.0318, 0.0309, 0.0325, 0.0361, 0.0221, 0.0379], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:34:34,392 INFO [train.py:904] (6/8) Epoch 19, batch 9000, loss[loss=0.161, simple_loss=0.2603, pruned_loss=0.03088, over 16672.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2662, pruned_loss=0.03753, over 3093278.19 frames. ], batch size: 134, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:34:34,393 INFO [train.py:929] (6/8) Computing validation loss 2023-04-30 23:34:44,207 INFO [train.py:938] (6/8) Epoch 19, validation: loss=0.1462, simple_loss=0.2506, pruned_loss=0.02087, over 944034.00 frames. 2023-04-30 23:34:44,208 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-04-30 23:35:26,227 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=191721.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:35:31,566 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191724.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:36:27,795 INFO [train.py:904] (6/8) Epoch 19, batch 9050, loss[loss=0.1704, simple_loss=0.2557, pruned_loss=0.04251, over 16415.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2667, pruned_loss=0.03802, over 3096436.80 frames. ], batch size: 146, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:37:04,254 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.085e+02 2.454e+02 3.075e+02 7.905e+02, threshold=4.907e+02, percent-clipped=4.0 2023-04-30 23:37:09,313 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191772.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:37:11,130 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-30 23:37:39,905 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191788.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:38:06,023 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7395, 1.2883, 1.7421, 1.6000, 1.8046, 1.8640, 1.5923, 1.8137], device='cuda:6'), covar=tensor([0.0274, 0.0434, 0.0224, 0.0292, 0.0308, 0.0221, 0.0441, 0.0139], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0184, 0.0170, 0.0172, 0.0185, 0.0143, 0.0187, 0.0137], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:38:10,475 INFO [train.py:904] (6/8) Epoch 19, batch 9100, loss[loss=0.1488, simple_loss=0.2396, pruned_loss=0.02896, over 12296.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2666, pruned_loss=0.03893, over 3087134.41 frames. ], batch size: 247, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:38:20,104 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3021, 3.3173, 1.8701, 3.6597, 2.4242, 3.6299, 2.1004, 2.7749], device='cuda:6'), covar=tensor([0.0274, 0.0371, 0.1747, 0.0215, 0.0897, 0.0483, 0.1590, 0.0728], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0168, 0.0187, 0.0149, 0.0169, 0.0204, 0.0195, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 23:39:32,434 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191836.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:39:43,923 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191841.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:40:05,584 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8410, 3.8810, 2.5740, 4.5216, 2.9347, 4.4305, 2.4697, 3.2253], device='cuda:6'), covar=tensor([0.0270, 0.0348, 0.1468, 0.0187, 0.0891, 0.0390, 0.1623, 0.0716], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0167, 0.0187, 0.0148, 0.0169, 0.0203, 0.0194, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-04-30 23:40:08,191 INFO [train.py:904] (6/8) Epoch 19, batch 9150, loss[loss=0.1428, simple_loss=0.2371, pruned_loss=0.02426, over 16441.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2669, pruned_loss=0.03857, over 3072263.78 frames. ], batch size: 68, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:40:46,341 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.543e+02 2.243e+02 2.682e+02 3.704e+02 5.734e+02, threshold=5.364e+02, percent-clipped=7.0 2023-04-30 23:41:30,830 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191889.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:41:50,023 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0440, 4.0463, 3.9505, 3.3062, 3.9988, 1.7764, 3.8011, 3.5904], device='cuda:6'), covar=tensor([0.0099, 0.0090, 0.0183, 0.0236, 0.0106, 0.2620, 0.0119, 0.0220], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0141, 0.0183, 0.0165, 0.0161, 0.0196, 0.0174, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:41:52,732 INFO [train.py:904] (6/8) Epoch 19, batch 9200, loss[loss=0.149, simple_loss=0.2368, pruned_loss=0.03054, over 12376.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2625, pruned_loss=0.03751, over 3067002.55 frames. ], batch size: 247, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:42:11,266 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-30 23:42:37,483 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2911, 2.4678, 2.0666, 2.2061, 2.8531, 2.5194, 2.8442, 3.0461], device='cuda:6'), covar=tensor([0.0119, 0.0399, 0.0521, 0.0469, 0.0237, 0.0363, 0.0184, 0.0218], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0223, 0.0215, 0.0216, 0.0224, 0.0221, 0.0221, 0.0215], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:42:40,688 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7195, 4.1020, 3.6933, 4.0199, 3.6679, 3.7089, 3.7074, 4.0987], device='cuda:6'), covar=tensor([0.2747, 0.2004, 0.2953, 0.1567, 0.1763, 0.3339, 0.2617, 0.2262], device='cuda:6'), in_proj_covar=tensor([0.0625, 0.0760, 0.0624, 0.0566, 0.0479, 0.0489, 0.0637, 0.0585], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:43:29,333 INFO [train.py:904] (6/8) Epoch 19, batch 9250, loss[loss=0.155, simple_loss=0.2342, pruned_loss=0.03793, over 12154.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2622, pruned_loss=0.0376, over 3060358.40 frames. ], batch size: 247, lr: 3.52e-03, grad_scale: 8.0 2023-04-30 23:43:35,337 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-30 23:43:46,417 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3809, 4.2184, 4.4662, 4.5784, 4.7422, 4.2960, 4.7589, 4.7756], device='cuda:6'), covar=tensor([0.1810, 0.1332, 0.1470, 0.0727, 0.0478, 0.0976, 0.0474, 0.0638], device='cuda:6'), in_proj_covar=tensor([0.0582, 0.0722, 0.0844, 0.0741, 0.0555, 0.0580, 0.0596, 0.0695], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:44:05,910 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.629e+02 2.331e+02 2.686e+02 3.317e+02 7.389e+02, threshold=5.371e+02, percent-clipped=4.0 2023-04-30 23:45:23,367 INFO [train.py:904] (6/8) Epoch 19, batch 9300, loss[loss=0.1463, simple_loss=0.238, pruned_loss=0.02732, over 16504.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2604, pruned_loss=0.03708, over 3042527.24 frames. ], batch size: 68, lr: 3.52e-03, grad_scale: 4.0 2023-04-30 23:46:09,746 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=192021.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:46:14,071 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-30 23:47:05,515 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2072, 1.6484, 1.9320, 2.1176, 2.2469, 2.2875, 1.7625, 2.3029], device='cuda:6'), covar=tensor([0.0195, 0.0445, 0.0274, 0.0324, 0.0310, 0.0217, 0.0474, 0.0155], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0183, 0.0168, 0.0171, 0.0183, 0.0141, 0.0185, 0.0136], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:47:06,711 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6615, 4.9759, 4.7755, 4.7607, 4.4704, 4.4979, 4.4112, 5.0443], device='cuda:6'), covar=tensor([0.1059, 0.0945, 0.0915, 0.0790, 0.0830, 0.1053, 0.1140, 0.0937], device='cuda:6'), in_proj_covar=tensor([0.0623, 0.0759, 0.0621, 0.0565, 0.0479, 0.0487, 0.0636, 0.0584], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:47:09,782 INFO [train.py:904] (6/8) Epoch 19, batch 9350, loss[loss=0.1794, simple_loss=0.2727, pruned_loss=0.043, over 16438.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2601, pruned_loss=0.03703, over 3045586.45 frames. ], batch size: 146, lr: 3.52e-03, grad_scale: 4.0 2023-04-30 23:47:46,413 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=192069.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:47:47,130 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 1.981e+02 2.500e+02 3.039e+02 5.486e+02, threshold=4.999e+02, percent-clipped=1.0 2023-04-30 23:47:59,167 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8967, 2.1450, 1.6996, 1.8044, 2.4940, 2.1931, 2.5265, 2.8125], device='cuda:6'), covar=tensor([0.0204, 0.0595, 0.0783, 0.0672, 0.0352, 0.0528, 0.0196, 0.0286], device='cuda:6'), in_proj_covar=tensor([0.0189, 0.0223, 0.0216, 0.0217, 0.0225, 0.0222, 0.0222, 0.0215], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:48:49,310 INFO [train.py:904] (6/8) Epoch 19, batch 9400, loss[loss=0.1664, simple_loss=0.2702, pruned_loss=0.03134, over 16680.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2598, pruned_loss=0.03657, over 3038441.24 frames. ], batch size: 89, lr: 3.52e-03, grad_scale: 4.0 2023-04-30 23:48:50,436 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192102.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:49:02,787 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.15 vs. limit=5.0 2023-04-30 23:49:37,560 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1099, 4.0961, 4.0289, 3.0344, 4.0490, 1.4837, 3.7496, 3.6723], device='cuda:6'), covar=tensor([0.0132, 0.0118, 0.0220, 0.0499, 0.0132, 0.3390, 0.0186, 0.0380], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0139, 0.0181, 0.0162, 0.0159, 0.0193, 0.0171, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:49:47,438 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3080, 1.4050, 2.0245, 2.1110, 2.1729, 2.3292, 1.6486, 2.3635], device='cuda:6'), covar=tensor([0.0220, 0.0561, 0.0259, 0.0312, 0.0334, 0.0235, 0.0570, 0.0161], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0183, 0.0168, 0.0171, 0.0184, 0.0141, 0.0185, 0.0136], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:50:17,987 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3703, 4.3472, 4.1639, 3.6407, 4.2520, 1.6914, 4.0419, 3.9304], device='cuda:6'), covar=tensor([0.0081, 0.0084, 0.0190, 0.0260, 0.0101, 0.2496, 0.0121, 0.0220], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0139, 0.0181, 0.0162, 0.0159, 0.0193, 0.0171, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:50:29,966 INFO [train.py:904] (6/8) Epoch 19, batch 9450, loss[loss=0.1547, simple_loss=0.2561, pruned_loss=0.02671, over 15182.00 frames. ], tot_loss[loss=0.168, simple_loss=0.262, pruned_loss=0.03702, over 3041344.79 frames. ], batch size: 190, lr: 3.52e-03, grad_scale: 4.0 2023-04-30 23:50:35,902 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192155.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:50:51,119 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192163.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:51:05,238 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.652e+02 2.168e+02 2.516e+02 3.148e+02 5.649e+02, threshold=5.031e+02, percent-clipped=1.0 2023-04-30 23:52:10,463 INFO [train.py:904] (6/8) Epoch 19, batch 9500, loss[loss=0.1665, simple_loss=0.2588, pruned_loss=0.03711, over 16673.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.262, pruned_loss=0.03709, over 3048599.77 frames. ], batch size: 134, lr: 3.52e-03, grad_scale: 4.0 2023-04-30 23:52:39,638 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192216.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:52:45,082 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192219.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:52:49,935 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192221.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:53:55,147 INFO [train.py:904] (6/8) Epoch 19, batch 9550, loss[loss=0.1698, simple_loss=0.2562, pruned_loss=0.0417, over 12507.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2621, pruned_loss=0.03749, over 3060836.64 frames. ], batch size: 250, lr: 3.52e-03, grad_scale: 4.0 2023-04-30 23:54:22,862 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3398, 3.7839, 3.8451, 2.6983, 3.3696, 3.8123, 3.6053, 2.2654], device='cuda:6'), covar=tensor([0.0516, 0.0047, 0.0043, 0.0375, 0.0111, 0.0089, 0.0069, 0.0478], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0078, 0.0078, 0.0132, 0.0093, 0.0104, 0.0089, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-04-30 23:54:34,513 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.443e+02 2.006e+02 2.351e+02 2.778e+02 5.702e+02, threshold=4.701e+02, percent-clipped=1.0 2023-04-30 23:54:37,058 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9442, 4.2469, 4.0789, 4.1004, 3.7398, 3.8410, 3.8774, 4.2334], device='cuda:6'), covar=tensor([0.1064, 0.0992, 0.0875, 0.0795, 0.0810, 0.1685, 0.0920, 0.0934], device='cuda:6'), in_proj_covar=tensor([0.0617, 0.0753, 0.0614, 0.0560, 0.0475, 0.0482, 0.0630, 0.0578], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:54:55,611 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192280.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:55:01,222 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192282.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:55:23,877 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9148, 2.6653, 2.9376, 2.0528, 2.7232, 2.1466, 2.6357, 2.8593], device='cuda:6'), covar=tensor([0.0296, 0.0897, 0.0449, 0.1767, 0.0705, 0.0917, 0.0598, 0.0906], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0151, 0.0159, 0.0146, 0.0138, 0.0123, 0.0138, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-04-30 23:55:38,437 INFO [train.py:904] (6/8) Epoch 19, batch 9600, loss[loss=0.1768, simple_loss=0.271, pruned_loss=0.04124, over 16559.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.263, pruned_loss=0.03758, over 3072773.50 frames. ], batch size: 75, lr: 3.52e-03, grad_scale: 8.0 2023-04-30 23:57:08,014 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-30 23:57:23,954 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-30 23:57:27,639 INFO [train.py:904] (6/8) Epoch 19, batch 9650, loss[loss=0.1729, simple_loss=0.2694, pruned_loss=0.0382, over 16218.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2647, pruned_loss=0.03779, over 3066644.76 frames. ], batch size: 165, lr: 3.52e-03, grad_scale: 8.0 2023-04-30 23:58:09,763 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.165e+02 2.448e+02 2.941e+02 5.888e+02, threshold=4.896e+02, percent-clipped=1.0 2023-04-30 23:58:35,385 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5919, 3.6409, 3.4925, 3.2056, 3.2757, 3.6006, 3.3356, 3.4057], device='cuda:6'), covar=tensor([0.0571, 0.0680, 0.0311, 0.0257, 0.0607, 0.0529, 0.1296, 0.0556], device='cuda:6'), in_proj_covar=tensor([0.0266, 0.0381, 0.0310, 0.0301, 0.0317, 0.0354, 0.0214, 0.0370], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-04-30 23:58:47,363 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5795, 3.7327, 2.7723, 2.1167, 2.2835, 2.2574, 3.8589, 3.2203], device='cuda:6'), covar=tensor([0.2870, 0.0512, 0.1771, 0.2986, 0.2754, 0.2194, 0.0368, 0.1198], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0255, 0.0289, 0.0294, 0.0277, 0.0243, 0.0277, 0.0315], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-04-30 23:58:54,558 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192392.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:59:09,441 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192399.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:59:15,171 INFO [train.py:904] (6/8) Epoch 19, batch 9700, loss[loss=0.1724, simple_loss=0.2701, pruned_loss=0.0374, over 16186.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2636, pruned_loss=0.03743, over 3064795.83 frames. ], batch size: 165, lr: 3.52e-03, grad_scale: 8.0 2023-04-30 23:59:34,946 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4816, 4.6001, 4.7770, 4.5994, 4.7325, 5.1784, 4.7598, 4.4308], device='cuda:6'), covar=tensor([0.1340, 0.2075, 0.2404, 0.2097, 0.2531, 0.1029, 0.1465, 0.2410], device='cuda:6'), in_proj_covar=tensor([0.0377, 0.0544, 0.0603, 0.0452, 0.0605, 0.0631, 0.0471, 0.0604], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 00:00:19,373 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192433.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:00:23,794 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8535, 5.1873, 5.3911, 5.2136, 5.2459, 5.7600, 5.3440, 5.0572], device='cuda:6'), covar=tensor([0.0921, 0.1821, 0.2495, 0.1624, 0.2117, 0.0916, 0.1396, 0.2015], device='cuda:6'), in_proj_covar=tensor([0.0375, 0.0542, 0.0601, 0.0451, 0.0602, 0.0630, 0.0469, 0.0602], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 00:00:57,137 INFO [train.py:904] (6/8) Epoch 19, batch 9750, loss[loss=0.1686, simple_loss=0.2648, pruned_loss=0.03623, over 16829.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2622, pruned_loss=0.03747, over 3075288.86 frames. ], batch size: 124, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:00:59,788 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192453.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:01:09,805 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192458.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:01:14,081 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192460.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:01:31,653 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.106e+02 2.517e+02 3.092e+02 5.794e+02, threshold=5.033e+02, percent-clipped=3.0 2023-05-01 00:02:23,687 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192494.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:02:37,067 INFO [train.py:904] (6/8) Epoch 19, batch 9800, loss[loss=0.1542, simple_loss=0.2431, pruned_loss=0.03264, over 12344.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2629, pruned_loss=0.03681, over 3080909.51 frames. ], batch size: 247, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:02:55,412 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192511.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:04:23,519 INFO [train.py:904] (6/8) Epoch 19, batch 9850, loss[loss=0.1691, simple_loss=0.2676, pruned_loss=0.0353, over 16392.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.264, pruned_loss=0.03667, over 3092367.13 frames. ], batch size: 146, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:05:00,568 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 1.961e+02 2.506e+02 2.957e+02 4.630e+02, threshold=5.011e+02, percent-clipped=0.0 2023-05-01 00:05:11,851 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192575.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:05:15,470 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192577.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:05:22,096 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.73 vs. limit=5.0 2023-05-01 00:06:14,628 INFO [train.py:904] (6/8) Epoch 19, batch 9900, loss[loss=0.1808, simple_loss=0.2615, pruned_loss=0.05007, over 12421.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2635, pruned_loss=0.03647, over 3056537.84 frames. ], batch size: 246, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:06:47,305 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192616.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:07:31,350 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-01 00:08:13,281 INFO [train.py:904] (6/8) Epoch 19, batch 9950, loss[loss=0.1615, simple_loss=0.2555, pruned_loss=0.03371, over 16430.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2661, pruned_loss=0.03689, over 3074912.10 frames. ], batch size: 68, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:08:54,783 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.121e+02 2.565e+02 3.086e+02 6.326e+02, threshold=5.129e+02, percent-clipped=1.0 2023-05-01 00:09:04,436 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192673.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:09:07,681 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-05-01 00:09:15,105 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192677.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:09:32,184 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6812, 4.9993, 4.7874, 4.7921, 4.5203, 4.4986, 4.3600, 5.0530], device='cuda:6'), covar=tensor([0.1204, 0.0828, 0.0960, 0.0742, 0.0774, 0.1132, 0.1103, 0.0850], device='cuda:6'), in_proj_covar=tensor([0.0615, 0.0757, 0.0615, 0.0561, 0.0473, 0.0482, 0.0629, 0.0579], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:10:12,824 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-01 00:10:14,415 INFO [train.py:904] (6/8) Epoch 19, batch 10000, loss[loss=0.1857, simple_loss=0.2821, pruned_loss=0.04462, over 16696.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2646, pruned_loss=0.03651, over 3100044.71 frames. ], batch size: 134, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:11:22,372 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192734.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:11:25,780 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192735.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:11:50,857 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192748.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:11:56,958 INFO [train.py:904] (6/8) Epoch 19, batch 10050, loss[loss=0.1767, simple_loss=0.2725, pruned_loss=0.04049, over 16446.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2649, pruned_loss=0.03633, over 3096144.60 frames. ], batch size: 68, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:11:57,711 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3481, 4.6779, 4.4845, 4.4818, 4.1629, 4.2090, 4.1357, 4.7169], device='cuda:6'), covar=tensor([0.1203, 0.0894, 0.0988, 0.0775, 0.0879, 0.1429, 0.1101, 0.0805], device='cuda:6'), in_proj_covar=tensor([0.0613, 0.0754, 0.0614, 0.0560, 0.0473, 0.0482, 0.0628, 0.0578], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:12:02,953 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192755.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:12:09,740 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=192758.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:12:32,889 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.427e+02 2.015e+02 2.519e+02 2.909e+02 8.183e+02, threshold=5.037e+02, percent-clipped=1.0 2023-05-01 00:13:08,087 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192789.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:13:09,905 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 00:13:21,300 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192796.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:13:30,581 INFO [train.py:904] (6/8) Epoch 19, batch 10100, loss[loss=0.1736, simple_loss=0.2664, pruned_loss=0.0404, over 17058.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2653, pruned_loss=0.0366, over 3102020.94 frames. ], batch size: 53, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:13:39,004 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=192806.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:13:48,331 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=192811.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:14:28,787 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192831.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:14:36,409 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 00:15:13,688 INFO [train.py:904] (6/8) Epoch 20, batch 0, loss[loss=0.2123, simple_loss=0.3036, pruned_loss=0.06053, over 17059.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.3036, pruned_loss=0.06053, over 17059.00 frames. ], batch size: 55, lr: 3.43e-03, grad_scale: 8.0 2023-05-01 00:15:13,689 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 00:15:21,160 INFO [train.py:938] (6/8) Epoch 20, validation: loss=0.146, simple_loss=0.2496, pruned_loss=0.02121, over 944034.00 frames. 2023-05-01 00:15:21,161 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 00:15:32,534 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=192859.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:15:49,198 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 2.412e+02 2.946e+02 3.638e+02 7.163e+02, threshold=5.893e+02, percent-clipped=6.0 2023-05-01 00:15:53,779 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=192875.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:15:56,034 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=192877.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:16:17,633 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192892.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 00:16:31,046 INFO [train.py:904] (6/8) Epoch 20, batch 50, loss[loss=0.1544, simple_loss=0.2459, pruned_loss=0.03141, over 17108.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2717, pruned_loss=0.05065, over 757515.12 frames. ], batch size: 47, lr: 3.43e-03, grad_scale: 2.0 2023-05-01 00:17:00,245 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=192923.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:17:03,080 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=192925.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:17:21,328 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3842, 5.3241, 5.1611, 4.8632, 5.1994, 2.2034, 5.0091, 5.1698], device='cuda:6'), covar=tensor([0.0092, 0.0087, 0.0215, 0.0291, 0.0112, 0.2357, 0.0118, 0.0170], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0142, 0.0184, 0.0163, 0.0161, 0.0198, 0.0174, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:17:38,938 INFO [train.py:904] (6/8) Epoch 20, batch 100, loss[loss=0.1655, simple_loss=0.2628, pruned_loss=0.03408, over 17100.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2667, pruned_loss=0.04822, over 1329322.78 frames. ], batch size: 48, lr: 3.43e-03, grad_scale: 2.0 2023-05-01 00:17:48,067 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2345, 5.5800, 5.2828, 5.3177, 5.0045, 4.9841, 4.9606, 5.6906], device='cuda:6'), covar=tensor([0.1108, 0.0944, 0.1426, 0.0990, 0.0929, 0.0876, 0.1224, 0.0987], device='cuda:6'), in_proj_covar=tensor([0.0627, 0.0774, 0.0627, 0.0573, 0.0484, 0.0494, 0.0645, 0.0590], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:18:07,333 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.214e+02 2.624e+02 3.260e+02 6.598e+02, threshold=5.249e+02, percent-clipped=3.0 2023-05-01 00:18:07,671 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192972.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:18:34,902 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1809, 5.0977, 5.0286, 4.4758, 4.6434, 5.0537, 4.9960, 4.6468], device='cuda:6'), covar=tensor([0.0574, 0.0465, 0.0307, 0.0379, 0.1078, 0.0453, 0.0324, 0.0740], device='cuda:6'), in_proj_covar=tensor([0.0271, 0.0387, 0.0316, 0.0306, 0.0324, 0.0359, 0.0218, 0.0377], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:18:37,613 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 00:18:41,873 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7050, 3.7191, 2.1472, 4.0733, 2.8198, 3.9712, 2.3908, 3.0102], device='cuda:6'), covar=tensor([0.0264, 0.0406, 0.1651, 0.0352, 0.0800, 0.0799, 0.1395, 0.0677], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0171, 0.0189, 0.0152, 0.0172, 0.0207, 0.0198, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 00:18:48,432 INFO [train.py:904] (6/8) Epoch 20, batch 150, loss[loss=0.1883, simple_loss=0.28, pruned_loss=0.04831, over 16499.00 frames. ], tot_loss[loss=0.18, simple_loss=0.265, pruned_loss=0.04749, over 1773826.31 frames. ], batch size: 75, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:19:01,758 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-05-01 00:19:27,944 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193029.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:19:43,987 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193041.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:19:52,855 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193048.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:19:58,179 INFO [train.py:904] (6/8) Epoch 20, batch 200, loss[loss=0.1835, simple_loss=0.2703, pruned_loss=0.04835, over 16310.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2637, pruned_loss=0.04647, over 2119935.23 frames. ], batch size: 36, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:20:03,484 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193055.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:20:05,130 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-05-01 00:20:27,493 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.769e+02 2.323e+02 2.688e+02 3.539e+02 1.444e+03, threshold=5.377e+02, percent-clipped=5.0 2023-05-01 00:20:50,741 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193089.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:20:53,795 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193091.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:21:00,957 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193096.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:21:08,806 INFO [train.py:904] (6/8) Epoch 20, batch 250, loss[loss=0.1552, simple_loss=0.2515, pruned_loss=0.0295, over 17142.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2616, pruned_loss=0.04576, over 2392038.69 frames. ], batch size: 48, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:21:09,277 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193102.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:21:10,866 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193103.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:21:58,201 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193137.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:22:17,118 INFO [train.py:904] (6/8) Epoch 20, batch 300, loss[loss=0.1587, simple_loss=0.2519, pruned_loss=0.03271, over 17077.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2587, pruned_loss=0.04392, over 2598931.06 frames. ], batch size: 55, lr: 3.42e-03, grad_scale: 1.0 2023-05-01 00:22:43,878 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2073, 4.3384, 4.3925, 4.2558, 4.2466, 4.8223, 4.2915, 4.0060], device='cuda:6'), covar=tensor([0.1797, 0.2188, 0.2472, 0.2248, 0.3116, 0.1306, 0.1944, 0.2728], device='cuda:6'), in_proj_covar=tensor([0.0394, 0.0570, 0.0631, 0.0475, 0.0635, 0.0658, 0.0495, 0.0631], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 00:22:46,393 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.636e+02 2.161e+02 2.496e+02 2.884e+02 5.011e+02, threshold=4.993e+02, percent-clipped=0.0 2023-05-01 00:23:05,594 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193187.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 00:23:10,451 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3694, 3.3398, 2.1207, 3.5449, 2.6125, 3.5052, 2.2087, 2.7043], device='cuda:6'), covar=tensor([0.0268, 0.0477, 0.1410, 0.0384, 0.0791, 0.0850, 0.1343, 0.0698], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0176, 0.0193, 0.0157, 0.0176, 0.0212, 0.0202, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 00:23:28,161 INFO [train.py:904] (6/8) Epoch 20, batch 350, loss[loss=0.1499, simple_loss=0.2271, pruned_loss=0.03634, over 16561.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2569, pruned_loss=0.04327, over 2756685.92 frames. ], batch size: 68, lr: 3.42e-03, grad_scale: 1.0 2023-05-01 00:23:39,614 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193211.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:24:37,894 INFO [train.py:904] (6/8) Epoch 20, batch 400, loss[loss=0.1575, simple_loss=0.2594, pruned_loss=0.02777, over 16707.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2561, pruned_loss=0.04315, over 2884031.98 frames. ], batch size: 57, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:25:00,585 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2086, 3.3768, 3.0572, 5.1906, 4.4890, 4.6585, 2.0834, 3.5089], device='cuda:6'), covar=tensor([0.1221, 0.0659, 0.1088, 0.0215, 0.0231, 0.0384, 0.1445, 0.0660], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0171, 0.0192, 0.0181, 0.0199, 0.0212, 0.0197, 0.0189], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 00:25:05,931 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193272.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:25:06,036 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193272.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:25:06,669 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.085e+02 2.558e+02 2.961e+02 5.266e+02, threshold=5.116e+02, percent-clipped=1.0 2023-05-01 00:25:20,659 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4816, 2.3377, 2.3352, 4.2489, 2.2950, 2.7094, 2.4474, 2.4607], device='cuda:6'), covar=tensor([0.1209, 0.3713, 0.3085, 0.0498, 0.4152, 0.2562, 0.3418, 0.3678], device='cuda:6'), in_proj_covar=tensor([0.0393, 0.0436, 0.0362, 0.0321, 0.0432, 0.0500, 0.0406, 0.0509], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:25:23,453 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193284.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:25:26,107 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-01 00:25:46,656 INFO [train.py:904] (6/8) Epoch 20, batch 450, loss[loss=0.1728, simple_loss=0.2696, pruned_loss=0.03796, over 17066.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2547, pruned_loss=0.04221, over 2978749.70 frames. ], batch size: 50, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:26:01,973 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2283, 4.4827, 4.5362, 3.4693, 3.7503, 4.4768, 4.0268, 2.7561], device='cuda:6'), covar=tensor([0.0404, 0.0056, 0.0034, 0.0283, 0.0124, 0.0083, 0.0089, 0.0419], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0081, 0.0080, 0.0134, 0.0095, 0.0106, 0.0092, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 00:26:14,330 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193320.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:26:26,077 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193329.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:26:48,809 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193345.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:26:57,283 INFO [train.py:904] (6/8) Epoch 20, batch 500, loss[loss=0.1625, simple_loss=0.2397, pruned_loss=0.0427, over 16873.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2544, pruned_loss=0.04165, over 3061369.66 frames. ], batch size: 116, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:27:26,048 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.231e+02 2.613e+02 3.252e+02 5.197e+02, threshold=5.226e+02, percent-clipped=2.0 2023-05-01 00:27:31,762 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193377.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:27:51,532 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193391.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:27:59,424 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193397.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:28:07,092 INFO [train.py:904] (6/8) Epoch 20, batch 550, loss[loss=0.1646, simple_loss=0.2465, pruned_loss=0.04135, over 16714.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2533, pruned_loss=0.04105, over 3124432.51 frames. ], batch size: 89, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:28:32,364 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193421.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:28:45,493 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-01 00:28:46,876 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0317, 2.1844, 2.5977, 2.9624, 2.8178, 3.4242, 2.4811, 3.4176], device='cuda:6'), covar=tensor([0.0225, 0.0461, 0.0330, 0.0316, 0.0324, 0.0188, 0.0424, 0.0153], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0189, 0.0174, 0.0177, 0.0190, 0.0147, 0.0191, 0.0141], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:28:56,785 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193439.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:28:58,190 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9326, 4.6891, 4.9801, 5.1807, 5.3929, 4.6855, 5.3904, 5.3702], device='cuda:6'), covar=tensor([0.2078, 0.1422, 0.1788, 0.0841, 0.0521, 0.1129, 0.0521, 0.0636], device='cuda:6'), in_proj_covar=tensor([0.0627, 0.0778, 0.0909, 0.0795, 0.0594, 0.0621, 0.0644, 0.0743], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:29:14,799 INFO [train.py:904] (6/8) Epoch 20, batch 600, loss[loss=0.1656, simple_loss=0.2577, pruned_loss=0.03677, over 17239.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2522, pruned_loss=0.04049, over 3173141.35 frames. ], batch size: 43, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:29:18,247 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193454.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 00:29:38,286 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-05-01 00:29:43,210 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.108e+02 2.456e+02 2.826e+02 6.166e+02, threshold=4.913e+02, percent-clipped=1.0 2023-05-01 00:29:55,243 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193482.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:30:03,368 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193487.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 00:30:21,022 INFO [train.py:904] (6/8) Epoch 20, batch 650, loss[loss=0.1759, simple_loss=0.2521, pruned_loss=0.04985, over 16423.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2515, pruned_loss=0.04042, over 3202884.60 frames. ], batch size: 146, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:30:39,986 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193515.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:31:07,550 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193535.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:31:11,794 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9075, 3.0181, 3.2922, 2.1454, 2.8544, 2.2192, 3.5003, 3.2905], device='cuda:6'), covar=tensor([0.0256, 0.1039, 0.0614, 0.1884, 0.0869, 0.1027, 0.0524, 0.1067], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0158, 0.0165, 0.0151, 0.0144, 0.0127, 0.0142, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 00:31:29,636 INFO [train.py:904] (6/8) Epoch 20, batch 700, loss[loss=0.1748, simple_loss=0.2651, pruned_loss=0.04229, over 16721.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2517, pruned_loss=0.04041, over 3237373.11 frames. ], batch size: 62, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:31:48,972 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193567.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 00:31:57,460 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.109e+02 2.476e+02 3.086e+02 5.855e+02, threshold=4.951e+02, percent-clipped=1.0 2023-05-01 00:32:35,677 INFO [train.py:904] (6/8) Epoch 20, batch 750, loss[loss=0.1764, simple_loss=0.2551, pruned_loss=0.04881, over 16721.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2534, pruned_loss=0.04128, over 3253360.77 frames. ], batch size: 134, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:32:54,636 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193615.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:33:04,529 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7275, 3.1119, 3.0609, 2.0804, 2.6824, 2.2809, 3.2561, 3.3474], device='cuda:6'), covar=tensor([0.0287, 0.0847, 0.0706, 0.1867, 0.0946, 0.0975, 0.0626, 0.0921], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0158, 0.0165, 0.0151, 0.0144, 0.0127, 0.0143, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 00:33:28,481 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193640.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:33:42,946 INFO [train.py:904] (6/8) Epoch 20, batch 800, loss[loss=0.1648, simple_loss=0.265, pruned_loss=0.03229, over 17141.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2529, pruned_loss=0.04141, over 3268375.06 frames. ], batch size: 49, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:33:52,535 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-05-01 00:34:05,433 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3116, 4.6754, 4.6477, 3.4370, 3.9123, 4.5754, 4.0756, 2.8616], device='cuda:6'), covar=tensor([0.0427, 0.0064, 0.0045, 0.0351, 0.0126, 0.0114, 0.0100, 0.0447], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0082, 0.0081, 0.0135, 0.0096, 0.0107, 0.0093, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0004], device='cuda:6') 2023-05-01 00:34:10,660 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 2.187e+02 2.610e+02 3.095e+02 1.121e+03, threshold=5.220e+02, percent-clipped=2.0 2023-05-01 00:34:16,609 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193676.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:34:26,360 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9587, 1.9360, 2.4859, 2.8504, 2.8642, 2.9268, 2.1744, 3.0702], device='cuda:6'), covar=tensor([0.0178, 0.0486, 0.0330, 0.0228, 0.0268, 0.0226, 0.0442, 0.0174], device='cuda:6'), in_proj_covar=tensor([0.0182, 0.0191, 0.0176, 0.0178, 0.0192, 0.0149, 0.0191, 0.0143], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:34:45,108 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193697.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:34:45,541 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-01 00:34:52,172 INFO [train.py:904] (6/8) Epoch 20, batch 850, loss[loss=0.1611, simple_loss=0.2523, pruned_loss=0.03496, over 17160.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2517, pruned_loss=0.04078, over 3272956.86 frames. ], batch size: 46, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:35:02,674 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9502, 3.9818, 4.3141, 2.3034, 4.5924, 4.6669, 3.3344, 3.6094], device='cuda:6'), covar=tensor([0.0727, 0.0224, 0.0229, 0.1057, 0.0072, 0.0141, 0.0391, 0.0376], device='cuda:6'), in_proj_covar=tensor([0.0146, 0.0107, 0.0095, 0.0138, 0.0078, 0.0121, 0.0126, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 00:35:17,865 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9857, 5.0708, 5.4546, 5.4723, 5.4651, 5.1616, 5.0655, 4.8838], device='cuda:6'), covar=tensor([0.0365, 0.0550, 0.0459, 0.0420, 0.0514, 0.0405, 0.1014, 0.0479], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0441, 0.0427, 0.0399, 0.0476, 0.0449, 0.0541, 0.0363], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 00:35:51,453 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193745.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:35:59,338 INFO [train.py:904] (6/8) Epoch 20, batch 900, loss[loss=0.1696, simple_loss=0.2639, pruned_loss=0.03766, over 17129.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2507, pruned_loss=0.04024, over 3284605.30 frames. ], batch size: 49, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:36:28,230 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.015e+02 2.385e+02 2.719e+02 5.424e+02, threshold=4.769e+02, percent-clipped=3.0 2023-05-01 00:36:33,729 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193777.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:37:09,296 INFO [train.py:904] (6/8) Epoch 20, batch 950, loss[loss=0.1916, simple_loss=0.2726, pruned_loss=0.05526, over 16781.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2512, pruned_loss=0.04058, over 3295666.82 frames. ], batch size: 124, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:37:20,475 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193810.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:38:01,179 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2750, 4.1154, 4.3479, 4.4778, 4.5398, 4.1246, 4.3429, 4.5217], device='cuda:6'), covar=tensor([0.1548, 0.1220, 0.1231, 0.0703, 0.0648, 0.1242, 0.2335, 0.0965], device='cuda:6'), in_proj_covar=tensor([0.0634, 0.0785, 0.0920, 0.0802, 0.0599, 0.0626, 0.0650, 0.0749], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:38:17,860 INFO [train.py:904] (6/8) Epoch 20, batch 1000, loss[loss=0.1496, simple_loss=0.2301, pruned_loss=0.0346, over 16585.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2501, pruned_loss=0.04075, over 3296157.77 frames. ], batch size: 75, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:38:22,238 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-05-01 00:38:39,183 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6362, 4.7629, 4.8980, 4.7135, 4.7655, 5.3744, 4.8613, 4.5419], device='cuda:6'), covar=tensor([0.1538, 0.2061, 0.2778, 0.2235, 0.2907, 0.1135, 0.1824, 0.2626], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0584, 0.0645, 0.0486, 0.0650, 0.0674, 0.0503, 0.0647], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 00:38:39,287 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193867.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 00:38:45,939 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.639e+02 2.059e+02 2.451e+02 3.046e+02 5.383e+02, threshold=4.901e+02, percent-clipped=2.0 2023-05-01 00:39:00,683 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193884.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:39:24,779 INFO [train.py:904] (6/8) Epoch 20, batch 1050, loss[loss=0.1622, simple_loss=0.2571, pruned_loss=0.03366, over 17249.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2504, pruned_loss=0.04091, over 3300172.87 frames. ], batch size: 52, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:39:43,715 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193915.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:40:03,358 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5401, 2.3584, 1.9409, 2.1639, 2.7240, 2.4971, 2.6351, 2.7938], device='cuda:6'), covar=tensor([0.0237, 0.0388, 0.0543, 0.0441, 0.0214, 0.0345, 0.0215, 0.0291], device='cuda:6'), in_proj_covar=tensor([0.0205, 0.0237, 0.0225, 0.0229, 0.0239, 0.0237, 0.0238, 0.0230], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:40:18,192 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193940.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:40:25,191 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193945.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:40:35,455 INFO [train.py:904] (6/8) Epoch 20, batch 1100, loss[loss=0.1511, simple_loss=0.2329, pruned_loss=0.03463, over 16833.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2493, pruned_loss=0.04016, over 3307105.27 frames. ], batch size: 96, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:41:01,746 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193971.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:41:04,218 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.207e+02 2.619e+02 3.357e+02 2.000e+03, threshold=5.237e+02, percent-clipped=3.0 2023-05-01 00:41:25,201 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193988.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:41:46,951 INFO [train.py:904] (6/8) Epoch 20, batch 1150, loss[loss=0.1583, simple_loss=0.2483, pruned_loss=0.03416, over 17229.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2494, pruned_loss=0.04006, over 3306622.60 frames. ], batch size: 44, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:42:14,705 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-01 00:42:56,045 INFO [train.py:904] (6/8) Epoch 20, batch 1200, loss[loss=0.162, simple_loss=0.2354, pruned_loss=0.04428, over 16803.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2481, pruned_loss=0.0395, over 3303859.73 frames. ], batch size: 83, lr: 3.42e-03, grad_scale: 8.0 2023-05-01 00:43:07,386 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8658, 2.8089, 2.6891, 4.4456, 3.6141, 4.1951, 1.7168, 2.9800], device='cuda:6'), covar=tensor([0.1399, 0.0759, 0.1175, 0.0206, 0.0207, 0.0419, 0.1564, 0.0854], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0173, 0.0193, 0.0184, 0.0201, 0.0214, 0.0199, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 00:43:25,069 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=194072.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:43:25,793 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.319e+02 2.146e+02 2.588e+02 3.131e+02 5.620e+02, threshold=5.177e+02, percent-clipped=2.0 2023-05-01 00:43:33,068 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=194077.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 00:44:06,932 INFO [train.py:904] (6/8) Epoch 20, batch 1250, loss[loss=0.1514, simple_loss=0.2304, pruned_loss=0.03616, over 16886.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2486, pruned_loss=0.04015, over 3316154.71 frames. ], batch size: 96, lr: 3.42e-03, grad_scale: 8.0 2023-05-01 00:44:17,551 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=194110.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 00:44:38,182 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=194125.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:44:49,888 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=194133.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:45:15,923 INFO [train.py:904] (6/8) Epoch 20, batch 1300, loss[loss=0.1539, simple_loss=0.2479, pruned_loss=0.02997, over 17241.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2491, pruned_loss=0.04001, over 3317787.02 frames. ], batch size: 45, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 00:45:26,572 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=194158.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 00:45:27,870 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6623, 2.5484, 2.2894, 2.4505, 3.0200, 2.6909, 3.2578, 3.2209], device='cuda:6'), covar=tensor([0.0136, 0.0431, 0.0500, 0.0456, 0.0263, 0.0411, 0.0234, 0.0251], device='cuda:6'), in_proj_covar=tensor([0.0205, 0.0238, 0.0225, 0.0229, 0.0239, 0.0237, 0.0237, 0.0230], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:45:46,012 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5461, 3.5726, 2.3072, 3.7933, 2.7950, 3.7776, 2.2025, 2.9201], device='cuda:6'), covar=tensor([0.0271, 0.0504, 0.1513, 0.0371, 0.0763, 0.0756, 0.1587, 0.0673], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0180, 0.0196, 0.0163, 0.0179, 0.0218, 0.0205, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 00:45:46,644 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.297e+02 2.207e+02 2.541e+02 3.000e+02 4.873e+02, threshold=5.083e+02, percent-clipped=0.0 2023-05-01 00:46:27,312 INFO [train.py:904] (6/8) Epoch 20, batch 1350, loss[loss=0.167, simple_loss=0.2455, pruned_loss=0.04428, over 16231.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2492, pruned_loss=0.03949, over 3321660.26 frames. ], batch size: 164, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:47:21,329 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=194240.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:47:36,939 INFO [train.py:904] (6/8) Epoch 20, batch 1400, loss[loss=0.1959, simple_loss=0.2686, pruned_loss=0.06159, over 16882.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2502, pruned_loss=0.03967, over 3316798.77 frames. ], batch size: 109, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:47:52,209 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-05-01 00:48:03,373 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=194271.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:48:06,429 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.050e+02 2.370e+02 3.098e+02 5.438e+02, threshold=4.739e+02, percent-clipped=2.0 2023-05-01 00:48:44,297 INFO [train.py:904] (6/8) Epoch 20, batch 1450, loss[loss=0.1653, simple_loss=0.2322, pruned_loss=0.04921, over 16536.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2492, pruned_loss=0.03973, over 3309560.95 frames. ], batch size: 146, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:49:06,979 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=194318.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:49:08,665 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=194319.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:49:54,001 INFO [train.py:904] (6/8) Epoch 20, batch 1500, loss[loss=0.1861, simple_loss=0.2588, pruned_loss=0.05673, over 16733.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2482, pruned_loss=0.03971, over 3303203.95 frames. ], batch size: 134, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:49:55,522 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=194353.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:50:07,939 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=194362.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:50:24,391 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.182e+02 2.566e+02 3.379e+02 8.536e+02, threshold=5.133e+02, percent-clipped=4.0 2023-05-01 00:50:31,285 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=194379.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:50:31,321 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4646, 2.8719, 3.1244, 2.0820, 2.7399, 2.2227, 3.1052, 3.0903], device='cuda:6'), covar=tensor([0.0295, 0.0922, 0.0588, 0.1850, 0.0894, 0.1010, 0.0631, 0.0880], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0162, 0.0167, 0.0153, 0.0145, 0.0129, 0.0145, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 00:50:41,587 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3075, 5.2724, 5.1659, 4.6192, 4.7307, 5.2284, 5.2171, 4.8423], device='cuda:6'), covar=tensor([0.0633, 0.0464, 0.0304, 0.0372, 0.1267, 0.0421, 0.0329, 0.0772], device='cuda:6'), in_proj_covar=tensor([0.0298, 0.0426, 0.0345, 0.0339, 0.0357, 0.0397, 0.0239, 0.0414], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:51:03,653 INFO [train.py:904] (6/8) Epoch 20, batch 1550, loss[loss=0.142, simple_loss=0.2235, pruned_loss=0.0302, over 16233.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2495, pruned_loss=0.04072, over 3315192.33 frames. ], batch size: 36, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:51:20,876 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=194414.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:51:34,142 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=194423.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 00:51:40,647 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=194428.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:52:13,192 INFO [train.py:904] (6/8) Epoch 20, batch 1600, loss[loss=0.1709, simple_loss=0.2461, pruned_loss=0.04783, over 16708.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.252, pruned_loss=0.04189, over 3308262.39 frames. ], batch size: 124, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 00:52:20,923 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5729, 2.3482, 2.3698, 4.3814, 2.2047, 2.7389, 2.4110, 2.4687], device='cuda:6'), covar=tensor([0.1205, 0.3632, 0.3060, 0.0485, 0.4339, 0.2616, 0.3529, 0.3690], device='cuda:6'), in_proj_covar=tensor([0.0397, 0.0440, 0.0364, 0.0326, 0.0435, 0.0506, 0.0409, 0.0515], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:52:43,858 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.299e+02 2.983e+02 3.595e+02 1.383e+03, threshold=5.966e+02, percent-clipped=6.0 2023-05-01 00:53:22,707 INFO [train.py:904] (6/8) Epoch 20, batch 1650, loss[loss=0.1785, simple_loss=0.2592, pruned_loss=0.04892, over 16506.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2538, pruned_loss=0.04244, over 3305223.86 frames. ], batch size: 146, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 00:54:16,621 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=194540.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:54:18,910 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9242, 3.9307, 4.3678, 2.1930, 4.6505, 4.7217, 3.3344, 3.6517], device='cuda:6'), covar=tensor([0.0770, 0.0293, 0.0243, 0.1224, 0.0076, 0.0135, 0.0421, 0.0380], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0109, 0.0098, 0.0141, 0.0080, 0.0125, 0.0128, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 00:54:32,831 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7735, 3.8270, 3.0525, 2.2921, 2.6331, 2.5125, 4.0137, 3.4459], device='cuda:6'), covar=tensor([0.2669, 0.0734, 0.1722, 0.2965, 0.2586, 0.2019, 0.0567, 0.1509], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0265, 0.0299, 0.0302, 0.0291, 0.0252, 0.0287, 0.0330], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 00:54:33,460 INFO [train.py:904] (6/8) Epoch 20, batch 1700, loss[loss=0.1727, simple_loss=0.2583, pruned_loss=0.04352, over 16665.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2555, pruned_loss=0.04319, over 3302018.74 frames. ], batch size: 89, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:54:45,447 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3898, 2.3071, 1.7533, 2.0602, 2.6272, 2.3403, 2.4488, 2.7120], device='cuda:6'), covar=tensor([0.0233, 0.0360, 0.0575, 0.0447, 0.0228, 0.0321, 0.0225, 0.0243], device='cuda:6'), in_proj_covar=tensor([0.0208, 0.0239, 0.0228, 0.0231, 0.0241, 0.0239, 0.0239, 0.0232], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:54:48,055 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.66 vs. limit=5.0 2023-05-01 00:55:05,695 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.259e+02 2.640e+02 3.354e+02 1.280e+03, threshold=5.281e+02, percent-clipped=2.0 2023-05-01 00:55:25,664 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=194588.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:55:43,649 INFO [train.py:904] (6/8) Epoch 20, batch 1750, loss[loss=0.1852, simple_loss=0.2743, pruned_loss=0.04808, over 16251.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2564, pruned_loss=0.04298, over 3308625.01 frames. ], batch size: 164, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:55:44,026 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0127, 4.7320, 4.8343, 5.2060, 5.3904, 4.6710, 5.4236, 5.3786], device='cuda:6'), covar=tensor([0.1886, 0.1578, 0.2275, 0.0994, 0.0795, 0.0946, 0.0659, 0.0871], device='cuda:6'), in_proj_covar=tensor([0.0648, 0.0803, 0.0940, 0.0823, 0.0612, 0.0642, 0.0663, 0.0766], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:56:27,146 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=194634.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:56:51,745 INFO [train.py:904] (6/8) Epoch 20, batch 1800, loss[loss=0.1802, simple_loss=0.2668, pruned_loss=0.04681, over 16278.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2578, pruned_loss=0.04328, over 3308899.35 frames. ], batch size: 165, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:57:06,049 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7183, 3.7597, 1.9946, 4.2478, 2.9216, 4.1927, 2.2107, 2.9102], device='cuda:6'), covar=tensor([0.0302, 0.0382, 0.1935, 0.0347, 0.0784, 0.0452, 0.1834, 0.0819], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0180, 0.0198, 0.0165, 0.0179, 0.0220, 0.0205, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 00:57:10,219 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4403, 4.4982, 4.8644, 4.8374, 4.8788, 4.5454, 4.5531, 4.4428], device='cuda:6'), covar=tensor([0.0366, 0.0630, 0.0400, 0.0429, 0.0531, 0.0451, 0.0933, 0.0588], device='cuda:6'), in_proj_covar=tensor([0.0411, 0.0452, 0.0435, 0.0410, 0.0488, 0.0462, 0.0553, 0.0370], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 00:57:22,529 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=194674.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:57:23,424 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.509e+02 2.398e+02 2.901e+02 3.350e+02 9.637e+02, threshold=5.801e+02, percent-clipped=10.0 2023-05-01 00:57:37,285 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0191, 4.3185, 4.3917, 3.2423, 3.5728, 4.3265, 3.8034, 2.6691], device='cuda:6'), covar=tensor([0.0431, 0.0058, 0.0038, 0.0313, 0.0150, 0.0091, 0.0088, 0.0420], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0082, 0.0082, 0.0135, 0.0097, 0.0108, 0.0094, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0004], device='cuda:6') 2023-05-01 00:57:50,499 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=194695.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:57:58,019 INFO [train.py:904] (6/8) Epoch 20, batch 1850, loss[loss=0.1519, simple_loss=0.246, pruned_loss=0.02892, over 17226.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.259, pruned_loss=0.0434, over 3303472.14 frames. ], batch size: 43, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:58:08,688 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=194709.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:58:21,443 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=194718.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 00:58:34,465 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=194728.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:58:56,386 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2312, 5.5675, 5.3095, 5.4015, 4.9809, 5.0292, 4.9738, 5.7109], device='cuda:6'), covar=tensor([0.1213, 0.0971, 0.1122, 0.0838, 0.0902, 0.0912, 0.1202, 0.0914], device='cuda:6'), in_proj_covar=tensor([0.0672, 0.0828, 0.0675, 0.0616, 0.0518, 0.0525, 0.0693, 0.0635], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 00:59:06,818 INFO [train.py:904] (6/8) Epoch 20, batch 1900, loss[loss=0.1489, simple_loss=0.2361, pruned_loss=0.03084, over 16805.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.258, pruned_loss=0.04231, over 3310385.47 frames. ], batch size: 39, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:59:38,382 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.030e+02 2.433e+02 3.010e+02 6.888e+02, threshold=4.866e+02, percent-clipped=1.0 2023-05-01 00:59:41,138 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=194776.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:00:16,322 INFO [train.py:904] (6/8) Epoch 20, batch 1950, loss[loss=0.1851, simple_loss=0.2833, pruned_loss=0.04341, over 16655.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2585, pruned_loss=0.04213, over 3306393.78 frames. ], batch size: 57, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 01:00:20,150 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 01:00:44,486 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-01 01:00:49,250 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2620, 5.2132, 5.1336, 4.6301, 4.7362, 5.1768, 5.1599, 4.7959], device='cuda:6'), covar=tensor([0.0664, 0.0484, 0.0311, 0.0380, 0.1129, 0.0487, 0.0324, 0.0750], device='cuda:6'), in_proj_covar=tensor([0.0298, 0.0428, 0.0346, 0.0340, 0.0357, 0.0399, 0.0239, 0.0414], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:01:17,155 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=194847.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:01:23,588 INFO [train.py:904] (6/8) Epoch 20, batch 2000, loss[loss=0.1438, simple_loss=0.2304, pruned_loss=0.02853, over 16834.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2578, pruned_loss=0.04191, over 3307252.40 frames. ], batch size: 39, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:01:54,983 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.120e+02 2.565e+02 3.052e+02 4.628e+02, threshold=5.130e+02, percent-clipped=0.0 2023-05-01 01:02:03,874 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1347, 2.1389, 2.2840, 3.8746, 2.1593, 2.4728, 2.2209, 2.3098], device='cuda:6'), covar=tensor([0.1341, 0.3593, 0.2869, 0.0582, 0.3750, 0.2556, 0.3789, 0.3087], device='cuda:6'), in_proj_covar=tensor([0.0397, 0.0439, 0.0363, 0.0326, 0.0433, 0.0505, 0.0408, 0.0514], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:02:32,416 INFO [train.py:904] (6/8) Epoch 20, batch 2050, loss[loss=0.1427, simple_loss=0.2305, pruned_loss=0.02741, over 17102.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2567, pruned_loss=0.04177, over 3314299.64 frames. ], batch size: 41, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:02:41,481 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=194908.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:03:41,679 INFO [train.py:904] (6/8) Epoch 20, batch 2100, loss[loss=0.1818, simple_loss=0.271, pruned_loss=0.0463, over 17067.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.258, pruned_loss=0.0426, over 3310717.42 frames. ], batch size: 55, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:04:12,872 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=194974.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:04:13,685 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.067e+02 2.444e+02 3.012e+02 5.275e+02, threshold=4.887e+02, percent-clipped=1.0 2023-05-01 01:04:34,697 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=194990.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:04:50,814 INFO [train.py:904] (6/8) Epoch 20, batch 2150, loss[loss=0.2376, simple_loss=0.3121, pruned_loss=0.0816, over 11865.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2586, pruned_loss=0.04295, over 3313140.84 frames. ], batch size: 246, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:05:01,384 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195009.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:05:04,327 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=195011.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:05:13,464 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195018.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 01:05:19,276 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195022.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:05:50,387 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=195044.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 01:05:55,266 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4337, 2.3794, 2.3307, 4.2293, 2.1969, 2.7341, 2.3734, 2.5503], device='cuda:6'), covar=tensor([0.1286, 0.3643, 0.2988, 0.0562, 0.4260, 0.2609, 0.3512, 0.3465], device='cuda:6'), in_proj_covar=tensor([0.0397, 0.0440, 0.0363, 0.0326, 0.0433, 0.0506, 0.0409, 0.0515], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:06:01,906 INFO [train.py:904] (6/8) Epoch 20, batch 2200, loss[loss=0.1492, simple_loss=0.2355, pruned_loss=0.03142, over 16879.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2586, pruned_loss=0.04328, over 3307386.60 frames. ], batch size: 42, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:06:09,187 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195057.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:06:21,178 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195066.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 01:06:28,885 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=195071.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:06:30,143 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=195072.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 01:06:33,763 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.892e+02 2.410e+02 2.761e+02 3.300e+02 5.006e+02, threshold=5.521e+02, percent-clipped=1.0 2023-05-01 01:07:10,884 INFO [train.py:904] (6/8) Epoch 20, batch 2250, loss[loss=0.1966, simple_loss=0.2715, pruned_loss=0.06089, over 16182.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2592, pruned_loss=0.0434, over 3314735.74 frames. ], batch size: 165, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:07:11,387 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9173, 2.8679, 2.5599, 4.4408, 3.6129, 4.1899, 1.6999, 3.1227], device='cuda:6'), covar=tensor([0.1305, 0.0702, 0.1219, 0.0163, 0.0228, 0.0435, 0.1500, 0.0782], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0174, 0.0192, 0.0187, 0.0203, 0.0215, 0.0198, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 01:07:15,445 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=195105.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 01:07:37,923 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7777, 2.4358, 1.9329, 2.2326, 2.8032, 2.5980, 2.8426, 2.9363], device='cuda:6'), covar=tensor([0.0220, 0.0409, 0.0549, 0.0479, 0.0254, 0.0356, 0.0210, 0.0271], device='cuda:6'), in_proj_covar=tensor([0.0208, 0.0240, 0.0227, 0.0230, 0.0240, 0.0239, 0.0240, 0.0232], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:07:54,824 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=195132.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:07:58,453 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 01:08:21,143 INFO [train.py:904] (6/8) Epoch 20, batch 2300, loss[loss=0.1744, simple_loss=0.2604, pruned_loss=0.04424, over 15442.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2591, pruned_loss=0.04312, over 3315988.75 frames. ], batch size: 190, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:08:51,383 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.574e+02 2.280e+02 2.663e+02 3.175e+02 5.300e+02, threshold=5.327e+02, percent-clipped=0.0 2023-05-01 01:08:55,897 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1558, 1.9141, 2.7327, 3.1080, 2.8949, 3.6424, 2.2531, 3.6352], device='cuda:6'), covar=tensor([0.0207, 0.0613, 0.0293, 0.0272, 0.0322, 0.0195, 0.0578, 0.0158], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0195, 0.0180, 0.0182, 0.0196, 0.0154, 0.0196, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:09:29,599 INFO [train.py:904] (6/8) Epoch 20, batch 2350, loss[loss=0.1622, simple_loss=0.2629, pruned_loss=0.03077, over 17026.00 frames. ], tot_loss[loss=0.174, simple_loss=0.26, pruned_loss=0.04399, over 3312627.04 frames. ], batch size: 50, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:09:31,698 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=195203.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:10:36,693 INFO [train.py:904] (6/8) Epoch 20, batch 2400, loss[loss=0.1886, simple_loss=0.284, pruned_loss=0.04658, over 17109.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2609, pruned_loss=0.04408, over 3310015.11 frames. ], batch size: 47, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:11:07,016 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.253e+02 2.639e+02 3.077e+02 5.540e+02, threshold=5.278e+02, percent-clipped=1.0 2023-05-01 01:11:29,084 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195290.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:11:45,805 INFO [train.py:904] (6/8) Epoch 20, batch 2450, loss[loss=0.1658, simple_loss=0.2465, pruned_loss=0.04254, over 16872.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.261, pruned_loss=0.04384, over 3304728.83 frames. ], batch size: 96, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:12:35,686 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195338.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:12:54,075 INFO [train.py:904] (6/8) Epoch 20, batch 2500, loss[loss=0.153, simple_loss=0.2411, pruned_loss=0.03245, over 17230.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2605, pruned_loss=0.04304, over 3298036.24 frames. ], batch size: 45, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:13:16,709 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=195367.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 01:13:26,924 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.185e+02 2.562e+02 3.023e+02 6.708e+02, threshold=5.124e+02, percent-clipped=4.0 2023-05-01 01:13:44,580 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6576, 3.6868, 2.0625, 3.9946, 2.8856, 3.9234, 1.9940, 2.9030], device='cuda:6'), covar=tensor([0.0242, 0.0358, 0.1680, 0.0293, 0.0731, 0.0574, 0.1872, 0.0684], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0181, 0.0197, 0.0166, 0.0179, 0.0221, 0.0205, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 01:13:50,864 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=195392.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:14:01,550 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=195400.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 01:14:04,077 INFO [train.py:904] (6/8) Epoch 20, batch 2550, loss[loss=0.1678, simple_loss=0.2597, pruned_loss=0.03796, over 16556.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2611, pruned_loss=0.04364, over 3297240.18 frames. ], batch size: 62, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:14:38,108 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=195427.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:15:11,784 INFO [train.py:904] (6/8) Epoch 20, batch 2600, loss[loss=0.1662, simple_loss=0.254, pruned_loss=0.03915, over 16329.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.261, pruned_loss=0.04273, over 3303960.84 frames. ], batch size: 36, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:15:13,356 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=195453.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:15:16,297 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7462, 2.6551, 2.4615, 4.0436, 3.2503, 3.9879, 1.5524, 2.8267], device='cuda:6'), covar=tensor([0.1336, 0.0698, 0.1153, 0.0161, 0.0135, 0.0357, 0.1533, 0.0852], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0173, 0.0191, 0.0186, 0.0203, 0.0214, 0.0197, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 01:15:42,988 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.574e+02 2.224e+02 2.625e+02 3.381e+02 7.141e+02, threshold=5.251e+02, percent-clipped=4.0 2023-05-01 01:15:58,008 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0222, 2.9482, 2.5834, 2.8561, 3.2934, 3.0536, 3.5668, 3.4952], device='cuda:6'), covar=tensor([0.0134, 0.0391, 0.0489, 0.0396, 0.0264, 0.0335, 0.0302, 0.0247], device='cuda:6'), in_proj_covar=tensor([0.0210, 0.0242, 0.0229, 0.0232, 0.0243, 0.0241, 0.0243, 0.0235], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:16:20,745 INFO [train.py:904] (6/8) Epoch 20, batch 2650, loss[loss=0.1803, simple_loss=0.2611, pruned_loss=0.04977, over 15658.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2614, pruned_loss=0.04223, over 3315213.74 frames. ], batch size: 191, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:16:22,200 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195503.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:17:28,733 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195551.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:17:29,628 INFO [train.py:904] (6/8) Epoch 20, batch 2700, loss[loss=0.1805, simple_loss=0.27, pruned_loss=0.04553, over 16165.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2602, pruned_loss=0.04125, over 3324113.45 frames. ], batch size: 35, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:18:00,661 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.486e+02 2.029e+02 2.571e+02 3.039e+02 8.808e+02, threshold=5.142e+02, percent-clipped=4.0 2023-05-01 01:18:25,983 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9129, 1.8877, 2.4354, 2.8057, 2.7792, 3.1001, 2.1479, 3.1717], device='cuda:6'), covar=tensor([0.0232, 0.0509, 0.0350, 0.0326, 0.0314, 0.0240, 0.0510, 0.0160], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0193, 0.0178, 0.0182, 0.0195, 0.0153, 0.0195, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:18:39,437 INFO [train.py:904] (6/8) Epoch 20, batch 2750, loss[loss=0.1747, simple_loss=0.2541, pruned_loss=0.04767, over 16289.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2607, pruned_loss=0.04131, over 3332650.35 frames. ], batch size: 165, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:19:09,851 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7930, 2.4323, 1.8481, 2.1645, 2.8105, 2.5661, 2.8452, 2.8829], device='cuda:6'), covar=tensor([0.0207, 0.0416, 0.0590, 0.0487, 0.0252, 0.0371, 0.0263, 0.0304], device='cuda:6'), in_proj_covar=tensor([0.0210, 0.0241, 0.0228, 0.0231, 0.0242, 0.0240, 0.0241, 0.0235], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:19:47,489 INFO [train.py:904] (6/8) Epoch 20, batch 2800, loss[loss=0.1899, simple_loss=0.2709, pruned_loss=0.05444, over 12310.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2607, pruned_loss=0.04175, over 3330852.98 frames. ], batch size: 247, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:20:07,215 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195667.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 01:20:18,613 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.320e+02 2.150e+02 2.557e+02 3.074e+02 6.713e+02, threshold=5.114e+02, percent-clipped=2.0 2023-05-01 01:20:52,465 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195700.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 01:20:54,302 INFO [train.py:904] (6/8) Epoch 20, batch 2850, loss[loss=0.1377, simple_loss=0.2263, pruned_loss=0.0245, over 16857.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2601, pruned_loss=0.0417, over 3328017.22 frames. ], batch size: 42, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:21:05,749 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8342, 1.9451, 2.5157, 2.7949, 2.6782, 3.1990, 2.2543, 3.2714], device='cuda:6'), covar=tensor([0.0245, 0.0470, 0.0335, 0.0309, 0.0337, 0.0201, 0.0463, 0.0128], device='cuda:6'), in_proj_covar=tensor([0.0186, 0.0193, 0.0178, 0.0182, 0.0195, 0.0153, 0.0194, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:21:05,879 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.80 vs. limit=5.0 2023-05-01 01:21:13,262 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195715.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:21:29,672 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195727.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:21:44,038 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3595, 3.4703, 3.6697, 2.5543, 3.3442, 3.7556, 3.4869, 2.0403], device='cuda:6'), covar=tensor([0.0515, 0.0126, 0.0059, 0.0376, 0.0130, 0.0098, 0.0102, 0.0512], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0082, 0.0082, 0.0134, 0.0097, 0.0109, 0.0094, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0004], device='cuda:6') 2023-05-01 01:21:57,091 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195748.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 01:21:57,100 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=195748.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:22:02,487 INFO [train.py:904] (6/8) Epoch 20, batch 2900, loss[loss=0.182, simple_loss=0.2649, pruned_loss=0.04952, over 16460.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2598, pruned_loss=0.04233, over 3324511.18 frames. ], batch size: 68, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:22:16,655 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8026, 1.8898, 2.4246, 2.8607, 2.6868, 3.3085, 2.1592, 3.2993], device='cuda:6'), covar=tensor([0.0287, 0.0512, 0.0349, 0.0314, 0.0333, 0.0196, 0.0494, 0.0149], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0194, 0.0179, 0.0182, 0.0195, 0.0154, 0.0195, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:22:33,088 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.294e+02 2.743e+02 3.539e+02 8.268e+02, threshold=5.486e+02, percent-clipped=5.0 2023-05-01 01:22:34,111 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195775.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:22:50,709 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1313, 3.8825, 4.3196, 2.3058, 4.6011, 4.6318, 3.3788, 3.5438], device='cuda:6'), covar=tensor([0.0669, 0.0243, 0.0211, 0.1009, 0.0066, 0.0140, 0.0380, 0.0405], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0109, 0.0098, 0.0140, 0.0080, 0.0125, 0.0128, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 01:23:10,989 INFO [train.py:904] (6/8) Epoch 20, batch 2950, loss[loss=0.1909, simple_loss=0.2786, pruned_loss=0.05164, over 16730.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2597, pruned_loss=0.04243, over 3331216.09 frames. ], batch size: 57, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:24:18,942 INFO [train.py:904] (6/8) Epoch 20, batch 3000, loss[loss=0.1754, simple_loss=0.2582, pruned_loss=0.04629, over 16633.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2599, pruned_loss=0.04313, over 3321530.34 frames. ], batch size: 68, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:24:18,942 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 01:24:27,132 INFO [train.py:938] (6/8) Epoch 20, validation: loss=0.1354, simple_loss=0.2409, pruned_loss=0.01492, over 944034.00 frames. 2023-05-01 01:24:27,133 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 01:24:29,621 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4234, 5.3572, 5.1425, 4.5940, 5.1909, 1.8793, 4.9050, 5.0815], device='cuda:6'), covar=tensor([0.0068, 0.0069, 0.0204, 0.0398, 0.0102, 0.2662, 0.0151, 0.0198], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0154, 0.0199, 0.0180, 0.0176, 0.0208, 0.0189, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:24:37,125 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7853, 5.0611, 5.3635, 5.0796, 5.0873, 5.7442, 5.1598, 4.8841], device='cuda:6'), covar=tensor([0.1330, 0.2113, 0.2219, 0.1994, 0.2699, 0.1107, 0.1688, 0.2529], device='cuda:6'), in_proj_covar=tensor([0.0417, 0.0608, 0.0666, 0.0507, 0.0674, 0.0699, 0.0517, 0.0677], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 01:24:58,716 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.268e+02 2.640e+02 3.074e+02 4.713e+02, threshold=5.280e+02, percent-clipped=0.0 2023-05-01 01:25:22,441 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1836, 2.1479, 2.3566, 3.8758, 2.1938, 2.4836, 2.2651, 2.3439], device='cuda:6'), covar=tensor([0.1351, 0.3763, 0.2830, 0.0637, 0.3813, 0.2671, 0.3760, 0.3074], device='cuda:6'), in_proj_covar=tensor([0.0398, 0.0441, 0.0364, 0.0327, 0.0434, 0.0508, 0.0411, 0.0518], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:25:37,523 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7411, 3.5191, 3.8160, 1.9833, 3.9437, 3.9392, 3.1539, 2.9884], device='cuda:6'), covar=tensor([0.0670, 0.0223, 0.0170, 0.1120, 0.0094, 0.0174, 0.0383, 0.0453], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0110, 0.0099, 0.0141, 0.0081, 0.0126, 0.0129, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 01:25:38,259 INFO [train.py:904] (6/8) Epoch 20, batch 3050, loss[loss=0.169, simple_loss=0.2454, pruned_loss=0.04627, over 16799.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2593, pruned_loss=0.04323, over 3316013.66 frames. ], batch size: 102, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:26:46,789 INFO [train.py:904] (6/8) Epoch 20, batch 3100, loss[loss=0.1984, simple_loss=0.2668, pruned_loss=0.065, over 16867.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2587, pruned_loss=0.04368, over 3309454.54 frames. ], batch size: 116, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:27:16,965 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.170e+02 2.517e+02 3.010e+02 4.589e+02, threshold=5.034e+02, percent-clipped=0.0 2023-05-01 01:27:55,506 INFO [train.py:904] (6/8) Epoch 20, batch 3150, loss[loss=0.1808, simple_loss=0.2638, pruned_loss=0.04893, over 16366.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2581, pruned_loss=0.0436, over 3312319.10 frames. ], batch size: 165, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:28:02,807 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-01 01:28:15,929 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-01 01:28:18,486 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8697, 4.7315, 4.7072, 4.3759, 4.3929, 4.8082, 4.5968, 4.4837], device='cuda:6'), covar=tensor([0.0719, 0.0850, 0.0356, 0.0378, 0.0959, 0.0508, 0.0461, 0.0768], device='cuda:6'), in_proj_covar=tensor([0.0303, 0.0434, 0.0352, 0.0347, 0.0364, 0.0404, 0.0242, 0.0421], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 01:28:20,179 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 01:28:57,185 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=196048.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:29:03,217 INFO [train.py:904] (6/8) Epoch 20, batch 3200, loss[loss=0.1665, simple_loss=0.2615, pruned_loss=0.03573, over 17261.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2575, pruned_loss=0.04277, over 3326059.69 frames. ], batch size: 52, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:29:35,518 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.205e+02 2.543e+02 3.042e+02 4.560e+02, threshold=5.087e+02, percent-clipped=0.0 2023-05-01 01:30:04,479 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=196096.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:30:11,568 INFO [train.py:904] (6/8) Epoch 20, batch 3250, loss[loss=0.1798, simple_loss=0.264, pruned_loss=0.04776, over 17171.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2569, pruned_loss=0.04251, over 3316112.68 frames. ], batch size: 46, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:30:28,453 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6081, 2.3360, 2.3104, 4.5917, 2.2590, 2.7374, 2.4149, 2.4952], device='cuda:6'), covar=tensor([0.1221, 0.3618, 0.3184, 0.0434, 0.4221, 0.2569, 0.3701, 0.3484], device='cuda:6'), in_proj_covar=tensor([0.0400, 0.0442, 0.0366, 0.0328, 0.0436, 0.0511, 0.0413, 0.0519], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:31:08,078 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6070, 2.4095, 2.4616, 4.5735, 2.4385, 2.7507, 2.4816, 2.6455], device='cuda:6'), covar=tensor([0.1238, 0.3756, 0.3042, 0.0487, 0.4197, 0.2781, 0.3564, 0.3731], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0440, 0.0364, 0.0327, 0.0434, 0.0509, 0.0411, 0.0516], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:31:16,842 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5277, 2.2045, 1.6584, 1.9197, 2.5633, 2.2829, 2.4987, 2.6756], device='cuda:6'), covar=tensor([0.0236, 0.0435, 0.0628, 0.0499, 0.0256, 0.0386, 0.0240, 0.0298], device='cuda:6'), in_proj_covar=tensor([0.0211, 0.0242, 0.0228, 0.0232, 0.0242, 0.0240, 0.0244, 0.0236], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:31:19,978 INFO [train.py:904] (6/8) Epoch 20, batch 3300, loss[loss=0.1537, simple_loss=0.2438, pruned_loss=0.0318, over 17200.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2572, pruned_loss=0.04238, over 3315317.54 frames. ], batch size: 44, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:31:29,410 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3956, 4.3262, 4.3387, 3.3780, 4.3939, 1.5770, 4.0102, 3.8586], device='cuda:6'), covar=tensor([0.0172, 0.0155, 0.0216, 0.0564, 0.0128, 0.3564, 0.0215, 0.0346], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0154, 0.0199, 0.0179, 0.0176, 0.0208, 0.0189, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:31:52,354 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.123e+02 2.550e+02 3.106e+02 4.546e+02, threshold=5.100e+02, percent-clipped=0.0 2023-05-01 01:32:28,249 INFO [train.py:904] (6/8) Epoch 20, batch 3350, loss[loss=0.1507, simple_loss=0.2445, pruned_loss=0.02843, over 17102.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2573, pruned_loss=0.04247, over 3314519.47 frames. ], batch size: 47, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:32:50,017 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1437, 3.4064, 3.3769, 2.1563, 2.8509, 2.4126, 3.4949, 3.6648], device='cuda:6'), covar=tensor([0.0290, 0.0864, 0.0613, 0.1752, 0.0861, 0.0925, 0.0616, 0.0821], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0163, 0.0165, 0.0150, 0.0143, 0.0128, 0.0144, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 01:33:04,714 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-05-01 01:33:24,676 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6658, 4.6961, 4.8693, 4.6999, 4.7037, 5.3217, 4.7944, 4.5127], device='cuda:6'), covar=tensor([0.1344, 0.2067, 0.2122, 0.2285, 0.2679, 0.1048, 0.1588, 0.2541], device='cuda:6'), in_proj_covar=tensor([0.0419, 0.0607, 0.0667, 0.0508, 0.0673, 0.0703, 0.0518, 0.0678], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 01:33:34,071 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3144, 2.2991, 2.3920, 4.1337, 2.2519, 2.7397, 2.3599, 2.5319], device='cuda:6'), covar=tensor([0.1442, 0.3758, 0.2922, 0.0591, 0.4119, 0.2499, 0.3883, 0.3208], device='cuda:6'), in_proj_covar=tensor([0.0401, 0.0442, 0.0366, 0.0329, 0.0437, 0.0511, 0.0413, 0.0519], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:33:35,776 INFO [train.py:904] (6/8) Epoch 20, batch 3400, loss[loss=0.1691, simple_loss=0.2415, pruned_loss=0.04836, over 16819.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2568, pruned_loss=0.04213, over 3322263.76 frames. ], batch size: 116, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:34:06,838 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.504e+02 2.175e+02 2.675e+02 3.072e+02 5.322e+02, threshold=5.351e+02, percent-clipped=1.0 2023-05-01 01:34:26,674 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2018, 3.2784, 3.4147, 2.1196, 2.9043, 2.4017, 3.6703, 3.5737], device='cuda:6'), covar=tensor([0.0230, 0.0924, 0.0626, 0.1993, 0.0850, 0.1031, 0.0507, 0.0955], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0163, 0.0165, 0.0150, 0.0143, 0.0128, 0.0143, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 01:34:44,341 INFO [train.py:904] (6/8) Epoch 20, batch 3450, loss[loss=0.1445, simple_loss=0.2305, pruned_loss=0.02921, over 17027.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2562, pruned_loss=0.04171, over 3319225.87 frames. ], batch size: 41, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:35:10,238 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=196321.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:35:50,041 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.28 vs. limit=5.0 2023-05-01 01:35:50,467 INFO [train.py:904] (6/8) Epoch 20, batch 3500, loss[loss=0.156, simple_loss=0.2404, pruned_loss=0.03582, over 15484.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.255, pruned_loss=0.04099, over 3318890.39 frames. ], batch size: 191, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:36:03,318 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=196360.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:36:23,663 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 2.052e+02 2.465e+02 2.837e+02 5.038e+02, threshold=4.930e+02, percent-clipped=0.0 2023-05-01 01:36:35,804 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=196382.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:37:01,667 INFO [train.py:904] (6/8) Epoch 20, batch 3550, loss[loss=0.1851, simple_loss=0.2609, pruned_loss=0.05464, over 16893.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2544, pruned_loss=0.04121, over 3305513.76 frames. ], batch size: 109, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:37:28,429 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=196421.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:37:37,976 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2023-05-01 01:38:06,978 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8019, 4.8935, 5.1134, 4.8360, 4.9524, 5.5522, 5.0390, 4.6564], device='cuda:6'), covar=tensor([0.1249, 0.2053, 0.2028, 0.2182, 0.2808, 0.1161, 0.1768, 0.2845], device='cuda:6'), in_proj_covar=tensor([0.0417, 0.0606, 0.0665, 0.0507, 0.0671, 0.0701, 0.0516, 0.0676], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 01:38:10,214 INFO [train.py:904] (6/8) Epoch 20, batch 3600, loss[loss=0.1456, simple_loss=0.2228, pruned_loss=0.03416, over 16796.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2531, pruned_loss=0.04091, over 3305095.02 frames. ], batch size: 83, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 01:38:41,980 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.511e+02 2.224e+02 2.585e+02 3.008e+02 4.978e+02, threshold=5.169e+02, percent-clipped=1.0 2023-05-01 01:39:20,690 INFO [train.py:904] (6/8) Epoch 20, batch 3650, loss[loss=0.1834, simple_loss=0.2763, pruned_loss=0.04524, over 16739.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2531, pruned_loss=0.04136, over 3297386.81 frames. ], batch size: 57, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 01:40:01,969 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2056, 4.1759, 4.1911, 3.6322, 4.1720, 1.7613, 3.9697, 3.6437], device='cuda:6'), covar=tensor([0.0145, 0.0117, 0.0180, 0.0260, 0.0094, 0.2805, 0.0132, 0.0232], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0152, 0.0198, 0.0178, 0.0175, 0.0206, 0.0187, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:40:28,905 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0618, 4.0165, 4.0362, 3.3869, 3.9657, 1.8540, 3.7870, 3.3737], device='cuda:6'), covar=tensor([0.0152, 0.0116, 0.0187, 0.0242, 0.0094, 0.2727, 0.0125, 0.0245], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0152, 0.0198, 0.0178, 0.0175, 0.0206, 0.0187, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:40:32,641 INFO [train.py:904] (6/8) Epoch 20, batch 3700, loss[loss=0.1574, simple_loss=0.2301, pruned_loss=0.0424, over 16875.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2519, pruned_loss=0.04284, over 3268033.06 frames. ], batch size: 96, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:41:07,135 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.139e+02 2.467e+02 2.956e+02 4.680e+02, threshold=4.935e+02, percent-clipped=0.0 2023-05-01 01:41:11,837 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=196578.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:41:47,078 INFO [train.py:904] (6/8) Epoch 20, batch 3750, loss[loss=0.1752, simple_loss=0.2527, pruned_loss=0.04884, over 16493.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2524, pruned_loss=0.0444, over 3269018.72 frames. ], batch size: 146, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:42:33,683 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-05-01 01:42:39,631 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=196639.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:42:58,321 INFO [train.py:904] (6/8) Epoch 20, batch 3800, loss[loss=0.1947, simple_loss=0.2676, pruned_loss=0.06086, over 16901.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.254, pruned_loss=0.04558, over 3259583.29 frames. ], batch size: 96, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:43:31,146 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.679e+02 2.212e+02 2.504e+02 2.913e+02 5.553e+02, threshold=5.009e+02, percent-clipped=1.0 2023-05-01 01:43:35,321 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=196677.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:44:10,772 INFO [train.py:904] (6/8) Epoch 20, batch 3850, loss[loss=0.1892, simple_loss=0.2621, pruned_loss=0.05812, over 16383.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2533, pruned_loss=0.04618, over 3270185.37 frames. ], batch size: 146, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:44:32,368 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=196716.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:45:23,536 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5650, 1.7574, 2.2405, 2.3256, 2.5242, 2.5220, 1.8258, 2.6643], device='cuda:6'), covar=tensor([0.0169, 0.0481, 0.0302, 0.0299, 0.0292, 0.0292, 0.0494, 0.0139], device='cuda:6'), in_proj_covar=tensor([0.0186, 0.0192, 0.0178, 0.0182, 0.0195, 0.0153, 0.0194, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:45:24,168 INFO [train.py:904] (6/8) Epoch 20, batch 3900, loss[loss=0.1888, simple_loss=0.2622, pruned_loss=0.05773, over 16443.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2533, pruned_loss=0.04678, over 3267399.13 frames. ], batch size: 146, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:45:57,455 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.527e+02 2.027e+02 2.552e+02 2.919e+02 4.484e+02, threshold=5.103e+02, percent-clipped=0.0 2023-05-01 01:46:04,934 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=196779.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:46:19,633 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0617, 4.0769, 3.8993, 3.6378, 3.7098, 4.0598, 3.6966, 3.8448], device='cuda:6'), covar=tensor([0.0635, 0.0610, 0.0272, 0.0248, 0.0588, 0.0412, 0.1068, 0.0490], device='cuda:6'), in_proj_covar=tensor([0.0301, 0.0432, 0.0348, 0.0344, 0.0361, 0.0400, 0.0240, 0.0417], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:46:36,465 INFO [train.py:904] (6/8) Epoch 20, batch 3950, loss[loss=0.1853, simple_loss=0.2563, pruned_loss=0.0571, over 16817.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2528, pruned_loss=0.04713, over 3272174.44 frames. ], batch size: 102, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:47:16,083 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2261, 4.1860, 4.1083, 3.8712, 3.8996, 4.2245, 3.8836, 4.0103], device='cuda:6'), covar=tensor([0.0680, 0.0737, 0.0309, 0.0258, 0.0689, 0.0457, 0.0800, 0.0536], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0432, 0.0349, 0.0344, 0.0361, 0.0401, 0.0240, 0.0418], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:47:32,162 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=196840.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:47:44,612 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4098, 3.3917, 2.1644, 3.5485, 2.7308, 3.5850, 2.2809, 2.7690], device='cuda:6'), covar=tensor([0.0250, 0.0406, 0.1456, 0.0286, 0.0691, 0.0765, 0.1336, 0.0671], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0178, 0.0195, 0.0164, 0.0177, 0.0219, 0.0203, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 01:47:44,705 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0613, 2.1251, 2.3044, 3.7344, 2.1254, 2.4268, 2.2261, 2.2975], device='cuda:6'), covar=tensor([0.1480, 0.3707, 0.2871, 0.0620, 0.3790, 0.2552, 0.3717, 0.3122], device='cuda:6'), in_proj_covar=tensor([0.0403, 0.0446, 0.0368, 0.0330, 0.0437, 0.0515, 0.0415, 0.0523], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:47:46,097 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6983, 2.6462, 2.4834, 4.6315, 2.4536, 3.0163, 2.5946, 2.8106], device='cuda:6'), covar=tensor([0.1062, 0.2964, 0.2646, 0.0356, 0.3634, 0.2029, 0.3083, 0.2783], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0445, 0.0368, 0.0330, 0.0437, 0.0515, 0.0415, 0.0523], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:47:48,961 INFO [train.py:904] (6/8) Epoch 20, batch 4000, loss[loss=0.1586, simple_loss=0.2415, pruned_loss=0.03781, over 17037.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2528, pruned_loss=0.0474, over 3270271.20 frames. ], batch size: 53, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:47:52,409 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=196854.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:48:15,885 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4074, 5.7135, 5.4595, 5.5029, 5.1486, 5.0347, 5.1141, 5.8209], device='cuda:6'), covar=tensor([0.1160, 0.0703, 0.0961, 0.0763, 0.0781, 0.0690, 0.1047, 0.0777], device='cuda:6'), in_proj_covar=tensor([0.0670, 0.0824, 0.0675, 0.0618, 0.0520, 0.0524, 0.0693, 0.0636], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:48:21,943 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.289e+02 2.579e+02 3.040e+02 5.791e+02, threshold=5.158e+02, percent-clipped=1.0 2023-05-01 01:48:32,446 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=196881.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:49:01,021 INFO [train.py:904] (6/8) Epoch 20, batch 4050, loss[loss=0.1774, simple_loss=0.2542, pruned_loss=0.05029, over 16327.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2538, pruned_loss=0.04683, over 3272770.50 frames. ], batch size: 35, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:49:19,937 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=196915.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:49:26,460 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4352, 1.6945, 2.1887, 2.4400, 2.5331, 2.7829, 1.8325, 2.6147], device='cuda:6'), covar=tensor([0.0227, 0.0514, 0.0313, 0.0344, 0.0311, 0.0184, 0.0532, 0.0119], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0191, 0.0177, 0.0182, 0.0194, 0.0152, 0.0193, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:49:35,428 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-05-01 01:49:36,804 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-01 01:49:47,743 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=196934.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:49:57,987 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=196942.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:50:12,375 INFO [train.py:904] (6/8) Epoch 20, batch 4100, loss[loss=0.1941, simple_loss=0.2794, pruned_loss=0.05442, over 16903.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2554, pruned_loss=0.0465, over 3264765.50 frames. ], batch size: 109, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:50:43,905 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8017, 2.3552, 1.8138, 2.0201, 2.6250, 2.2563, 2.5638, 2.7698], device='cuda:6'), covar=tensor([0.0157, 0.0372, 0.0564, 0.0494, 0.0248, 0.0385, 0.0224, 0.0245], device='cuda:6'), in_proj_covar=tensor([0.0205, 0.0234, 0.0223, 0.0225, 0.0234, 0.0233, 0.0236, 0.0230], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:50:44,593 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.448e+02 1.746e+02 2.102e+02 2.519e+02 4.672e+02, threshold=4.204e+02, percent-clipped=0.0 2023-05-01 01:50:48,506 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=196977.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:51:23,807 INFO [train.py:904] (6/8) Epoch 20, batch 4150, loss[loss=0.2178, simple_loss=0.2913, pruned_loss=0.07217, over 17076.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2626, pruned_loss=0.04847, over 3260632.39 frames. ], batch size: 53, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:51:45,578 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=197016.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:52:00,055 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=197025.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:52:40,714 INFO [train.py:904] (6/8) Epoch 20, batch 4200, loss[loss=0.2102, simple_loss=0.3085, pruned_loss=0.05596, over 16702.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2699, pruned_loss=0.05058, over 3216931.54 frames. ], batch size: 89, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:52:58,850 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=197064.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:53:12,969 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.581e+02 2.329e+02 2.798e+02 3.249e+02 9.837e+02, threshold=5.595e+02, percent-clipped=2.0 2023-05-01 01:53:51,289 INFO [train.py:904] (6/8) Epoch 20, batch 4250, loss[loss=0.1863, simple_loss=0.2823, pruned_loss=0.04519, over 16526.00 frames. ], tot_loss[loss=0.1874, simple_loss=0.2738, pruned_loss=0.05053, over 3182195.66 frames. ], batch size: 68, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:54:40,507 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=197135.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:54:58,307 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4097, 4.4864, 4.2996, 4.0135, 4.0233, 4.4383, 4.0765, 4.1191], device='cuda:6'), covar=tensor([0.0570, 0.0423, 0.0280, 0.0283, 0.0756, 0.0355, 0.0670, 0.0558], device='cuda:6'), in_proj_covar=tensor([0.0295, 0.0423, 0.0342, 0.0338, 0.0353, 0.0391, 0.0235, 0.0409], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 01:55:04,298 INFO [train.py:904] (6/8) Epoch 20, batch 4300, loss[loss=0.1684, simple_loss=0.2624, pruned_loss=0.03719, over 16470.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.275, pruned_loss=0.04935, over 3208468.96 frames. ], batch size: 68, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:55:07,099 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-01 01:55:37,987 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.176e+02 2.515e+02 3.039e+02 5.119e+02, threshold=5.030e+02, percent-clipped=0.0 2023-05-01 01:56:12,640 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4261, 3.2181, 3.5870, 1.7405, 3.8022, 3.8535, 2.8254, 2.8921], device='cuda:6'), covar=tensor([0.0800, 0.0284, 0.0233, 0.1209, 0.0078, 0.0131, 0.0470, 0.0441], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0109, 0.0097, 0.0139, 0.0079, 0.0124, 0.0127, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 01:56:18,918 INFO [train.py:904] (6/8) Epoch 20, batch 4350, loss[loss=0.1986, simple_loss=0.2909, pruned_loss=0.05318, over 16515.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2775, pruned_loss=0.05032, over 3191716.80 frames. ], batch size: 68, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:56:21,151 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-05-01 01:56:29,938 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=197210.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:57:05,410 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=197234.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:57:09,110 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=197237.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:57:31,882 INFO [train.py:904] (6/8) Epoch 20, batch 4400, loss[loss=0.1941, simple_loss=0.2838, pruned_loss=0.05221, over 16224.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2796, pruned_loss=0.05141, over 3184022.20 frames. ], batch size: 35, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:57:48,689 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0178, 3.0189, 1.9037, 3.3199, 2.2721, 3.3406, 2.0588, 2.5279], device='cuda:6'), covar=tensor([0.0301, 0.0365, 0.1574, 0.0168, 0.0848, 0.0433, 0.1458, 0.0730], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0176, 0.0193, 0.0160, 0.0175, 0.0215, 0.0200, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 01:58:04,301 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.492e+02 2.097e+02 2.441e+02 2.762e+02 5.334e+02, threshold=4.883e+02, percent-clipped=1.0 2023-05-01 01:58:04,988 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2023-05-01 01:58:14,499 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=197282.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:58:42,795 INFO [train.py:904] (6/8) Epoch 20, batch 4450, loss[loss=0.2141, simple_loss=0.3106, pruned_loss=0.05882, over 16743.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.2835, pruned_loss=0.05308, over 3194554.22 frames. ], batch size: 89, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:59:55,984 INFO [train.py:904] (6/8) Epoch 20, batch 4500, loss[loss=0.1901, simple_loss=0.2778, pruned_loss=0.05123, over 16693.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2843, pruned_loss=0.05385, over 3200781.74 frames. ], batch size: 83, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 02:00:15,392 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5819, 2.7651, 2.3364, 2.5399, 3.1272, 2.7382, 3.1737, 3.2282], device='cuda:6'), covar=tensor([0.0081, 0.0330, 0.0440, 0.0375, 0.0195, 0.0320, 0.0179, 0.0218], device='cuda:6'), in_proj_covar=tensor([0.0202, 0.0231, 0.0222, 0.0223, 0.0233, 0.0231, 0.0233, 0.0228], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:00:30,609 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 1.873e+02 2.163e+02 2.505e+02 3.448e+02, threshold=4.325e+02, percent-clipped=0.0 2023-05-01 02:01:08,123 INFO [train.py:904] (6/8) Epoch 20, batch 4550, loss[loss=0.1976, simple_loss=0.2934, pruned_loss=0.05095, over 16486.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.2854, pruned_loss=0.05507, over 3217748.97 frames. ], batch size: 75, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 02:01:27,221 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=197415.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:01:56,104 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=197435.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:01:57,390 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1683, 3.9391, 3.7518, 2.4125, 3.5585, 3.8941, 3.5025, 2.1035], device='cuda:6'), covar=tensor([0.0597, 0.0034, 0.0052, 0.0445, 0.0083, 0.0081, 0.0084, 0.0476], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0082, 0.0081, 0.0133, 0.0096, 0.0107, 0.0094, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 02:02:18,326 INFO [train.py:904] (6/8) Epoch 20, batch 4600, loss[loss=0.2065, simple_loss=0.2937, pruned_loss=0.05962, over 16261.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2859, pruned_loss=0.05493, over 3221853.44 frames. ], batch size: 165, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 02:02:51,855 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 02:02:52,207 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 1.798e+02 2.178e+02 2.547e+02 5.479e+02, threshold=4.357e+02, percent-clipped=1.0 2023-05-01 02:02:52,816 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=197476.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:03:03,000 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=197483.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:03:14,057 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.57 vs. limit=2.0 2023-05-01 02:03:30,131 INFO [train.py:904] (6/8) Epoch 20, batch 4650, loss[loss=0.1654, simple_loss=0.2592, pruned_loss=0.0358, over 16921.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2842, pruned_loss=0.05444, over 3218119.02 frames. ], batch size: 102, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 02:03:41,240 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=197510.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:03:49,590 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=197516.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:04:20,756 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=197537.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:04:32,513 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=197545.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:04:42,157 INFO [train.py:904] (6/8) Epoch 20, batch 4700, loss[loss=0.1798, simple_loss=0.2623, pruned_loss=0.04866, over 16688.00 frames. ], tot_loss[loss=0.1934, simple_loss=0.2807, pruned_loss=0.05305, over 3210478.14 frames. ], batch size: 57, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 02:04:52,429 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=197558.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:04:53,020 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-05-01 02:05:18,233 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 1.893e+02 2.227e+02 2.649e+02 4.562e+02, threshold=4.454e+02, percent-clipped=1.0 2023-05-01 02:05:19,868 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=197577.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 02:05:31,365 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=197585.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:05:55,395 INFO [train.py:904] (6/8) Epoch 20, batch 4750, loss[loss=0.1615, simple_loss=0.2499, pruned_loss=0.03651, over 16707.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2769, pruned_loss=0.05135, over 3211388.22 frames. ], batch size: 89, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:06:01,903 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=197606.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:06:11,967 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.76 vs. limit=5.0 2023-05-01 02:07:08,960 INFO [train.py:904] (6/8) Epoch 20, batch 4800, loss[loss=0.173, simple_loss=0.2647, pruned_loss=0.04064, over 16873.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2737, pruned_loss=0.04971, over 3203078.64 frames. ], batch size: 116, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:07:45,094 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 1.779e+02 2.011e+02 2.336e+02 4.336e+02, threshold=4.022e+02, percent-clipped=0.0 2023-05-01 02:08:24,127 INFO [train.py:904] (6/8) Epoch 20, batch 4850, loss[loss=0.1674, simple_loss=0.2742, pruned_loss=0.03032, over 16697.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2748, pruned_loss=0.04909, over 3188555.16 frames. ], batch size: 134, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:08:56,074 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6330, 2.3899, 2.2592, 3.5291, 2.0464, 3.7095, 1.4474, 2.7919], device='cuda:6'), covar=tensor([0.1442, 0.0870, 0.1342, 0.0156, 0.0142, 0.0394, 0.1755, 0.0811], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0172, 0.0191, 0.0184, 0.0203, 0.0211, 0.0197, 0.0189], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 02:09:20,979 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3181, 4.3892, 4.2107, 3.9010, 3.8892, 4.3151, 4.0272, 4.0141], device='cuda:6'), covar=tensor([0.0550, 0.0419, 0.0283, 0.0291, 0.0859, 0.0431, 0.0675, 0.0594], device='cuda:6'), in_proj_covar=tensor([0.0285, 0.0409, 0.0331, 0.0327, 0.0342, 0.0378, 0.0228, 0.0396], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:09:21,252 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-01 02:09:40,086 INFO [train.py:904] (6/8) Epoch 20, batch 4900, loss[loss=0.1672, simple_loss=0.2635, pruned_loss=0.03539, over 16868.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2745, pruned_loss=0.0482, over 3181573.83 frames. ], batch size: 116, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:10:08,658 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=197771.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:10:15,934 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.320e+02 1.929e+02 2.192e+02 2.700e+02 5.551e+02, threshold=4.384e+02, percent-clipped=1.0 2023-05-01 02:10:52,823 INFO [train.py:904] (6/8) Epoch 20, batch 4950, loss[loss=0.1834, simple_loss=0.2824, pruned_loss=0.04216, over 16509.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2733, pruned_loss=0.04691, over 3184144.56 frames. ], batch size: 75, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:11:29,053 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0946, 3.1673, 2.8234, 5.1518, 3.7920, 4.2286, 2.2752, 3.2563], device='cuda:6'), covar=tensor([0.1278, 0.0762, 0.1230, 0.0124, 0.0422, 0.0417, 0.1372, 0.0914], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0173, 0.0192, 0.0184, 0.0204, 0.0213, 0.0198, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 02:12:04,523 INFO [train.py:904] (6/8) Epoch 20, batch 5000, loss[loss=0.1758, simple_loss=0.2678, pruned_loss=0.04192, over 16541.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2752, pruned_loss=0.04698, over 3202278.55 frames. ], batch size: 68, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:12:33,414 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=197872.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 02:12:38,856 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.376e+02 1.953e+02 2.351e+02 2.789e+02 5.042e+02, threshold=4.702e+02, percent-clipped=2.0 2023-05-01 02:13:01,271 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=197892.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:13:13,690 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=197901.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:13:14,487 INFO [train.py:904] (6/8) Epoch 20, batch 5050, loss[loss=0.17, simple_loss=0.2581, pruned_loss=0.04098, over 16630.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2753, pruned_loss=0.04669, over 3217313.36 frames. ], batch size: 62, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:13:35,043 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3814, 3.3458, 3.4040, 3.5008, 3.5505, 3.3106, 3.5219, 3.5969], device='cuda:6'), covar=tensor([0.1202, 0.0871, 0.1022, 0.0622, 0.0584, 0.2390, 0.0865, 0.0677], device='cuda:6'), in_proj_covar=tensor([0.0625, 0.0773, 0.0900, 0.0790, 0.0588, 0.0618, 0.0633, 0.0732], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:13:39,611 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-01 02:14:25,041 INFO [train.py:904] (6/8) Epoch 20, batch 5100, loss[loss=0.1717, simple_loss=0.2647, pruned_loss=0.03932, over 16707.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2731, pruned_loss=0.04576, over 3226206.78 frames. ], batch size: 89, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:14:27,941 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=197953.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:15:00,606 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 1.957e+02 2.274e+02 2.606e+02 4.454e+02, threshold=4.549e+02, percent-clipped=0.0 2023-05-01 02:15:41,431 INFO [train.py:904] (6/8) Epoch 20, batch 5150, loss[loss=0.1771, simple_loss=0.2707, pruned_loss=0.04177, over 16754.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2733, pruned_loss=0.04518, over 3211637.15 frames. ], batch size: 83, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:16:52,361 INFO [train.py:904] (6/8) Epoch 20, batch 5200, loss[loss=0.1704, simple_loss=0.2619, pruned_loss=0.03946, over 16678.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2716, pruned_loss=0.0448, over 3212161.26 frames. ], batch size: 89, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:17:06,616 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-05-01 02:17:18,576 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0873, 5.1958, 5.5443, 5.4689, 5.4986, 5.1742, 5.0872, 4.8690], device='cuda:6'), covar=tensor([0.0277, 0.0480, 0.0274, 0.0365, 0.0418, 0.0322, 0.0861, 0.0413], device='cuda:6'), in_proj_covar=tensor([0.0389, 0.0431, 0.0415, 0.0388, 0.0461, 0.0439, 0.0528, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 02:17:20,263 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=198071.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 02:17:27,748 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 1.904e+02 2.236e+02 2.771e+02 4.329e+02, threshold=4.472e+02, percent-clipped=0.0 2023-05-01 02:17:35,545 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=198081.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:18:03,836 INFO [train.py:904] (6/8) Epoch 20, batch 5250, loss[loss=0.1857, simple_loss=0.2787, pruned_loss=0.04636, over 16330.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2692, pruned_loss=0.04435, over 3207826.14 frames. ], batch size: 165, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:18:28,861 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=198119.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:19:01,603 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6216, 3.6072, 4.2270, 2.1262, 4.4074, 4.4299, 3.0588, 3.1218], device='cuda:6'), covar=tensor([0.0828, 0.0301, 0.0157, 0.1221, 0.0050, 0.0089, 0.0433, 0.0500], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0110, 0.0098, 0.0140, 0.0080, 0.0124, 0.0129, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 02:19:01,624 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=198142.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:19:15,056 INFO [train.py:904] (6/8) Epoch 20, batch 5300, loss[loss=0.1513, simple_loss=0.2433, pruned_loss=0.02959, over 16498.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2663, pruned_loss=0.04362, over 3196464.72 frames. ], batch size: 146, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:19:38,447 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1361, 2.4497, 1.9325, 2.2186, 2.8162, 2.5166, 2.7339, 3.0477], device='cuda:6'), covar=tensor([0.0153, 0.0449, 0.0672, 0.0505, 0.0274, 0.0406, 0.0230, 0.0286], device='cuda:6'), in_proj_covar=tensor([0.0200, 0.0229, 0.0222, 0.0222, 0.0232, 0.0231, 0.0231, 0.0228], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:19:44,946 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=198172.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:19:49,806 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 1.825e+02 2.153e+02 2.534e+02 4.591e+02, threshold=4.307e+02, percent-clipped=1.0 2023-05-01 02:20:03,582 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1187, 3.4199, 3.5295, 2.0633, 2.9809, 2.4598, 3.4904, 3.7392], device='cuda:6'), covar=tensor([0.0288, 0.0762, 0.0604, 0.2079, 0.0919, 0.0916, 0.0718, 0.0875], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0163, 0.0167, 0.0151, 0.0144, 0.0129, 0.0144, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 02:20:27,599 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=198201.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:20:28,366 INFO [train.py:904] (6/8) Epoch 20, batch 5350, loss[loss=0.192, simple_loss=0.2847, pruned_loss=0.04968, over 15320.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2643, pruned_loss=0.043, over 3195774.88 frames. ], batch size: 190, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:20:53,641 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0135, 5.0432, 5.4125, 5.3736, 5.4109, 5.0609, 4.9621, 4.7084], device='cuda:6'), covar=tensor([0.0276, 0.0518, 0.0323, 0.0396, 0.0490, 0.0325, 0.0989, 0.0460], device='cuda:6'), in_proj_covar=tensor([0.0391, 0.0434, 0.0418, 0.0390, 0.0465, 0.0442, 0.0531, 0.0354], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 02:20:54,795 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=198220.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:20:59,201 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8385, 5.0234, 5.2737, 5.2422, 5.2893, 4.9979, 4.7632, 4.6787], device='cuda:6'), covar=tensor([0.0431, 0.0604, 0.0419, 0.0468, 0.0600, 0.0418, 0.1286, 0.0487], device='cuda:6'), in_proj_covar=tensor([0.0391, 0.0434, 0.0418, 0.0390, 0.0465, 0.0442, 0.0531, 0.0354], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 02:21:35,323 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=198248.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:21:37,093 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=198249.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:21:40,980 INFO [train.py:904] (6/8) Epoch 20, batch 5400, loss[loss=0.2045, simple_loss=0.2918, pruned_loss=0.05863, over 16989.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2672, pruned_loss=0.04394, over 3209666.74 frames. ], batch size: 41, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:22:15,985 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.461e+02 1.945e+02 2.236e+02 2.644e+02 4.931e+02, threshold=4.472e+02, percent-clipped=2.0 2023-05-01 02:22:57,918 INFO [train.py:904] (6/8) Epoch 20, batch 5450, loss[loss=0.2097, simple_loss=0.2958, pruned_loss=0.06178, over 16594.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2702, pruned_loss=0.04526, over 3230384.95 frames. ], batch size: 57, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:23:04,385 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5667, 3.7306, 2.8394, 2.3050, 2.4775, 2.4154, 3.9277, 3.4130], device='cuda:6'), covar=tensor([0.2995, 0.0647, 0.1772, 0.2645, 0.2654, 0.1944, 0.0464, 0.1128], device='cuda:6'), in_proj_covar=tensor([0.0323, 0.0268, 0.0302, 0.0306, 0.0294, 0.0252, 0.0292, 0.0331], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 02:23:12,776 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-01 02:23:18,428 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9766, 4.0265, 4.2959, 4.2687, 4.3021, 4.0287, 4.0344, 4.0266], device='cuda:6'), covar=tensor([0.0297, 0.0580, 0.0391, 0.0409, 0.0414, 0.0428, 0.0778, 0.0451], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0432, 0.0416, 0.0388, 0.0463, 0.0441, 0.0529, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 02:23:37,581 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0844, 2.8597, 3.1304, 1.7457, 3.2563, 3.3282, 2.7312, 2.5208], device='cuda:6'), covar=tensor([0.0844, 0.0298, 0.0194, 0.1170, 0.0081, 0.0200, 0.0439, 0.0502], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0110, 0.0098, 0.0140, 0.0080, 0.0124, 0.0129, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 02:24:01,152 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3323, 4.1919, 4.2900, 4.5111, 4.6271, 4.2713, 4.6246, 4.6577], device='cuda:6'), covar=tensor([0.1750, 0.1318, 0.1862, 0.0819, 0.0682, 0.1103, 0.0690, 0.0764], device='cuda:6'), in_proj_covar=tensor([0.0622, 0.0767, 0.0893, 0.0779, 0.0586, 0.0614, 0.0626, 0.0726], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:24:14,758 INFO [train.py:904] (6/8) Epoch 20, batch 5500, loss[loss=0.2231, simple_loss=0.3073, pruned_loss=0.06944, over 16886.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2773, pruned_loss=0.04984, over 3166220.35 frames. ], batch size: 116, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:24:38,134 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4554, 4.5061, 4.3419, 4.0533, 4.0221, 4.4355, 4.1958, 4.1716], device='cuda:6'), covar=tensor([0.0668, 0.0679, 0.0296, 0.0300, 0.0865, 0.0560, 0.0577, 0.0606], device='cuda:6'), in_proj_covar=tensor([0.0290, 0.0421, 0.0339, 0.0334, 0.0350, 0.0390, 0.0233, 0.0405], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:24:51,692 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.837e+02 2.873e+02 3.524e+02 4.449e+02 7.452e+02, threshold=7.049e+02, percent-clipped=24.0 2023-05-01 02:25:34,175 INFO [train.py:904] (6/8) Epoch 20, batch 5550, loss[loss=0.2095, simple_loss=0.2976, pruned_loss=0.06066, over 16319.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2845, pruned_loss=0.05481, over 3130749.66 frames. ], batch size: 146, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:25:51,906 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-01 02:26:30,787 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=198437.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:26:53,774 INFO [train.py:904] (6/8) Epoch 20, batch 5600, loss[loss=0.2897, simple_loss=0.3494, pruned_loss=0.115, over 11627.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2894, pruned_loss=0.05908, over 3094929.64 frames. ], batch size: 247, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:27:34,781 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.278e+02 3.283e+02 3.655e+02 4.380e+02 7.107e+02, threshold=7.309e+02, percent-clipped=1.0 2023-05-01 02:28:14,188 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=198499.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:28:18,605 INFO [train.py:904] (6/8) Epoch 20, batch 5650, loss[loss=0.2322, simple_loss=0.3095, pruned_loss=0.07745, over 15259.00 frames. ], tot_loss[loss=0.2077, simple_loss=0.2931, pruned_loss=0.06111, over 3103192.69 frames. ], batch size: 191, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:29:12,402 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-05-01 02:29:32,555 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=198548.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:29:36,995 INFO [train.py:904] (6/8) Epoch 20, batch 5700, loss[loss=0.2236, simple_loss=0.3151, pruned_loss=0.06607, over 15358.00 frames. ], tot_loss[loss=0.2105, simple_loss=0.2949, pruned_loss=0.06312, over 3073082.38 frames. ], batch size: 190, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:29:37,492 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9010, 4.1657, 3.9995, 4.0298, 3.7124, 3.7468, 3.7961, 4.1542], device='cuda:6'), covar=tensor([0.1126, 0.0938, 0.0978, 0.0830, 0.0769, 0.1810, 0.1038, 0.0985], device='cuda:6'), in_proj_covar=tensor([0.0645, 0.0787, 0.0649, 0.0590, 0.0496, 0.0505, 0.0661, 0.0609], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:29:50,868 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=198560.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:30:14,597 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.943e+02 3.352e+02 4.045e+02 5.075e+02 1.137e+03, threshold=8.090e+02, percent-clipped=5.0 2023-05-01 02:30:47,280 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=198596.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:30:55,562 INFO [train.py:904] (6/8) Epoch 20, batch 5750, loss[loss=0.186, simple_loss=0.2709, pruned_loss=0.05053, over 16727.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.298, pruned_loss=0.06546, over 3036348.23 frames. ], batch size: 62, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:31:03,176 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6089, 2.5231, 1.8752, 2.7104, 2.1326, 2.7383, 2.1703, 2.4037], device='cuda:6'), covar=tensor([0.0306, 0.0337, 0.1275, 0.0233, 0.0624, 0.0423, 0.1206, 0.0575], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0175, 0.0193, 0.0157, 0.0175, 0.0213, 0.0200, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 02:31:08,909 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-01 02:31:17,295 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=198615.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:32:08,155 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 02:32:16,951 INFO [train.py:904] (6/8) Epoch 20, batch 5800, loss[loss=0.1828, simple_loss=0.2747, pruned_loss=0.04547, over 16314.00 frames. ], tot_loss[loss=0.2138, simple_loss=0.2979, pruned_loss=0.06484, over 3016156.87 frames. ], batch size: 146, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:32:53,794 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.915e+02 2.730e+02 3.384e+02 4.113e+02 5.841e+02, threshold=6.768e+02, percent-clipped=0.0 2023-05-01 02:32:54,361 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=198676.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:33:34,960 INFO [train.py:904] (6/8) Epoch 20, batch 5850, loss[loss=0.1899, simple_loss=0.2803, pruned_loss=0.04979, over 17127.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2954, pruned_loss=0.06286, over 3041645.68 frames. ], batch size: 48, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:34:05,312 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6160, 4.8834, 5.0025, 4.7837, 4.7967, 5.4124, 4.8401, 4.5795], device='cuda:6'), covar=tensor([0.1341, 0.1810, 0.2073, 0.1915, 0.2599, 0.0982, 0.1769, 0.2737], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0581, 0.0635, 0.0481, 0.0641, 0.0670, 0.0498, 0.0649], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 02:34:27,432 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5041, 4.5110, 4.8900, 4.8674, 4.8734, 4.5636, 4.5363, 4.4247], device='cuda:6'), covar=tensor([0.0355, 0.0610, 0.0406, 0.0421, 0.0464, 0.0434, 0.0975, 0.0529], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0442, 0.0426, 0.0396, 0.0474, 0.0449, 0.0538, 0.0360], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 02:34:30,878 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=198737.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:34:53,518 INFO [train.py:904] (6/8) Epoch 20, batch 5900, loss[loss=0.2453, simple_loss=0.312, pruned_loss=0.08926, over 11346.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.295, pruned_loss=0.06219, over 3064249.12 frames. ], batch size: 246, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:35:24,997 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3717, 3.3410, 3.4145, 3.4998, 3.5262, 3.2814, 3.5023, 3.5779], device='cuda:6'), covar=tensor([0.1205, 0.0914, 0.0894, 0.0576, 0.0600, 0.2343, 0.0909, 0.0735], device='cuda:6'), in_proj_covar=tensor([0.0619, 0.0763, 0.0886, 0.0775, 0.0584, 0.0611, 0.0626, 0.0721], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:35:26,428 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6901, 4.0522, 3.1361, 2.2999, 2.7622, 2.6647, 4.3120, 3.6495], device='cuda:6'), covar=tensor([0.2914, 0.0552, 0.1639, 0.2663, 0.2489, 0.1844, 0.0416, 0.1137], device='cuda:6'), in_proj_covar=tensor([0.0324, 0.0267, 0.0302, 0.0305, 0.0294, 0.0252, 0.0292, 0.0331], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 02:35:27,621 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8824, 3.9277, 2.2597, 4.4976, 3.0169, 4.3655, 2.4865, 3.1288], device='cuda:6'), covar=tensor([0.0254, 0.0376, 0.1760, 0.0258, 0.0812, 0.0551, 0.1540, 0.0771], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0176, 0.0193, 0.0158, 0.0175, 0.0213, 0.0201, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 02:35:34,249 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.541e+02 2.837e+02 3.258e+02 4.009e+02 8.301e+02, threshold=6.515e+02, percent-clipped=1.0 2023-05-01 02:35:48,536 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=198785.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:36:14,567 INFO [train.py:904] (6/8) Epoch 20, batch 5950, loss[loss=0.1881, simple_loss=0.2784, pruned_loss=0.04894, over 16491.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2957, pruned_loss=0.06134, over 3062596.20 frames. ], batch size: 75, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:37:02,293 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0472, 5.5135, 5.6549, 5.3951, 5.4740, 6.0240, 5.5385, 5.3157], device='cuda:6'), covar=tensor([0.0925, 0.1890, 0.2498, 0.2167, 0.2616, 0.1081, 0.1661, 0.2561], device='cuda:6'), in_proj_covar=tensor([0.0404, 0.0584, 0.0639, 0.0484, 0.0644, 0.0673, 0.0502, 0.0653], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 02:37:31,021 INFO [train.py:904] (6/8) Epoch 20, batch 6000, loss[loss=0.1853, simple_loss=0.2763, pruned_loss=0.04717, over 16444.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2948, pruned_loss=0.06091, over 3076401.11 frames. ], batch size: 68, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:37:31,021 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 02:37:41,832 INFO [train.py:938] (6/8) Epoch 20, validation: loss=0.1516, simple_loss=0.2644, pruned_loss=0.01942, over 944034.00 frames. 2023-05-01 02:37:41,832 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 02:37:42,221 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=198852.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:37:46,544 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=198855.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:37:58,546 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-01 02:38:17,161 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.795e+02 2.764e+02 3.257e+02 3.996e+02 6.001e+02, threshold=6.515e+02, percent-clipped=0.0 2023-05-01 02:38:28,082 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=198883.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:38:56,337 INFO [train.py:904] (6/8) Epoch 20, batch 6050, loss[loss=0.1799, simple_loss=0.2744, pruned_loss=0.04277, over 16352.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2932, pruned_loss=0.06, over 3099843.08 frames. ], batch size: 146, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:39:14,653 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=198913.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:39:36,227 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=198927.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:39:47,162 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8405, 1.4177, 1.7196, 1.7002, 1.8364, 1.9381, 1.6462, 1.8432], device='cuda:6'), covar=tensor([0.0245, 0.0381, 0.0191, 0.0305, 0.0248, 0.0171, 0.0394, 0.0136], device='cuda:6'), in_proj_covar=tensor([0.0182, 0.0191, 0.0175, 0.0179, 0.0191, 0.0149, 0.0192, 0.0145], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:40:05,078 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=198944.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:40:17,598 INFO [train.py:904] (6/8) Epoch 20, batch 6100, loss[loss=0.1868, simple_loss=0.2724, pruned_loss=0.05059, over 16418.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.292, pruned_loss=0.05871, over 3104808.79 frames. ], batch size: 35, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:40:46,261 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=198971.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:40:52,992 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2023-05-01 02:40:53,272 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.654e+02 2.636e+02 3.021e+02 3.731e+02 7.049e+02, threshold=6.042e+02, percent-clipped=1.0 2023-05-01 02:41:12,513 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=198988.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:41:16,032 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 02:41:22,052 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0357, 3.5092, 3.4803, 2.3177, 3.2707, 3.5459, 3.2625, 1.9387], device='cuda:6'), covar=tensor([0.0552, 0.0053, 0.0059, 0.0390, 0.0098, 0.0115, 0.0099, 0.0476], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0081, 0.0080, 0.0132, 0.0095, 0.0107, 0.0092, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 02:41:32,340 INFO [train.py:904] (6/8) Epoch 20, batch 6150, loss[loss=0.1932, simple_loss=0.2762, pruned_loss=0.05514, over 16472.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2898, pruned_loss=0.05768, over 3127549.09 frames. ], batch size: 75, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:42:49,579 INFO [train.py:904] (6/8) Epoch 20, batch 6200, loss[loss=0.2139, simple_loss=0.2864, pruned_loss=0.07069, over 11711.00 frames. ], tot_loss[loss=0.2013, simple_loss=0.2878, pruned_loss=0.05742, over 3113218.56 frames. ], batch size: 248, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 02:43:28,028 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.886e+02 2.934e+02 3.552e+02 4.446e+02 1.093e+03, threshold=7.104e+02, percent-clipped=2.0 2023-05-01 02:44:06,523 INFO [train.py:904] (6/8) Epoch 20, batch 6250, loss[loss=0.2273, simple_loss=0.3009, pruned_loss=0.07692, over 11785.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2878, pruned_loss=0.05767, over 3101786.66 frames. ], batch size: 247, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 02:45:21,381 INFO [train.py:904] (6/8) Epoch 20, batch 6300, loss[loss=0.2433, simple_loss=0.3115, pruned_loss=0.08758, over 11617.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2881, pruned_loss=0.05717, over 3125390.30 frames. ], batch size: 246, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 02:45:24,389 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7481, 4.9776, 5.1386, 4.8956, 4.9711, 5.5378, 4.9927, 4.7700], device='cuda:6'), covar=tensor([0.1114, 0.1788, 0.2455, 0.2140, 0.2547, 0.0980, 0.1636, 0.2342], device='cuda:6'), in_proj_covar=tensor([0.0403, 0.0584, 0.0643, 0.0486, 0.0644, 0.0674, 0.0502, 0.0652], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 02:45:26,460 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=199155.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:45:59,333 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=199176.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:46:00,084 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.793e+02 2.632e+02 3.175e+02 3.943e+02 7.347e+02, threshold=6.350e+02, percent-clipped=2.0 2023-05-01 02:46:38,965 INFO [train.py:904] (6/8) Epoch 20, batch 6350, loss[loss=0.2249, simple_loss=0.3001, pruned_loss=0.07486, over 16654.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2898, pruned_loss=0.05873, over 3126170.61 frames. ], batch size: 134, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 02:46:40,593 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=199203.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:46:48,061 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199208.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:47:00,923 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9086, 4.8934, 4.7782, 4.4384, 4.4323, 4.8501, 4.7327, 4.5364], device='cuda:6'), covar=tensor([0.0577, 0.0456, 0.0282, 0.0301, 0.0953, 0.0417, 0.0344, 0.0633], device='cuda:6'), in_proj_covar=tensor([0.0283, 0.0409, 0.0329, 0.0324, 0.0339, 0.0377, 0.0227, 0.0393], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:47:30,915 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=199237.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 02:47:34,236 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199239.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:47:36,802 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-05-01 02:47:50,399 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9994, 3.1503, 2.8583, 5.2772, 4.2125, 4.5567, 2.0138, 3.1243], device='cuda:6'), covar=tensor([0.1292, 0.0743, 0.1206, 0.0148, 0.0388, 0.0361, 0.1557, 0.0941], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0175, 0.0195, 0.0186, 0.0207, 0.0214, 0.0201, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 02:47:52,862 INFO [train.py:904] (6/8) Epoch 20, batch 6400, loss[loss=0.2036, simple_loss=0.2986, pruned_loss=0.05436, over 16669.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2902, pruned_loss=0.06006, over 3108100.30 frames. ], batch size: 134, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:48:21,653 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=199271.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:48:29,164 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.234e+02 2.937e+02 3.403e+02 4.348e+02 9.192e+02, threshold=6.807e+02, percent-clipped=3.0 2023-05-01 02:48:32,143 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5337, 2.5221, 2.4699, 3.5626, 2.6926, 3.7020, 1.3061, 2.9173], device='cuda:6'), covar=tensor([0.1437, 0.0773, 0.1175, 0.0193, 0.0219, 0.0451, 0.1798, 0.0754], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0174, 0.0194, 0.0186, 0.0207, 0.0214, 0.0201, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 02:48:39,508 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199283.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:49:00,509 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-05-01 02:49:07,103 INFO [train.py:904] (6/8) Epoch 20, batch 6450, loss[loss=0.1898, simple_loss=0.2815, pruned_loss=0.0491, over 16475.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2895, pruned_loss=0.05872, over 3118990.60 frames. ], batch size: 68, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:49:33,334 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=199319.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:50:00,204 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2023-05-01 02:50:06,634 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.27 vs. limit=5.0 2023-05-01 02:50:24,668 INFO [train.py:904] (6/8) Epoch 20, batch 6500, loss[loss=0.2013, simple_loss=0.2851, pruned_loss=0.0588, over 16191.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2878, pruned_loss=0.05836, over 3106827.37 frames. ], batch size: 165, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:51:02,007 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.751e+02 2.867e+02 3.442e+02 4.204e+02 8.359e+02, threshold=6.884e+02, percent-clipped=2.0 2023-05-01 02:51:39,130 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=199400.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:51:41,756 INFO [train.py:904] (6/8) Epoch 20, batch 6550, loss[loss=0.2057, simple_loss=0.3057, pruned_loss=0.05289, over 16333.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2904, pruned_loss=0.05928, over 3089389.99 frames. ], batch size: 146, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:52:56,160 INFO [train.py:904] (6/8) Epoch 20, batch 6600, loss[loss=0.2673, simple_loss=0.3287, pruned_loss=0.103, over 11862.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2929, pruned_loss=0.06005, over 3091737.15 frames. ], batch size: 246, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:53:05,801 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=199458.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:53:10,081 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=199461.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 02:53:33,378 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.746e+02 2.722e+02 3.309e+02 3.917e+02 9.551e+02, threshold=6.618e+02, percent-clipped=2.0 2023-05-01 02:53:36,308 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3063, 3.5545, 3.2991, 3.1652, 2.9641, 3.4778, 3.2372, 3.2030], device='cuda:6'), covar=tensor([0.1034, 0.0727, 0.0489, 0.0416, 0.1089, 0.0577, 0.2300, 0.0770], device='cuda:6'), in_proj_covar=tensor([0.0284, 0.0409, 0.0329, 0.0324, 0.0338, 0.0378, 0.0228, 0.0394], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:53:59,424 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=199494.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:54:11,564 INFO [train.py:904] (6/8) Epoch 20, batch 6650, loss[loss=0.2512, simple_loss=0.3142, pruned_loss=0.09414, over 11090.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2937, pruned_loss=0.06149, over 3075977.32 frames. ], batch size: 247, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:54:19,071 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3493, 2.3523, 2.3951, 4.1639, 2.3216, 2.7417, 2.4233, 2.5394], device='cuda:6'), covar=tensor([0.1296, 0.3368, 0.2799, 0.0462, 0.3778, 0.2376, 0.3314, 0.3133], device='cuda:6'), in_proj_covar=tensor([0.0394, 0.0437, 0.0359, 0.0321, 0.0431, 0.0504, 0.0407, 0.0513], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:54:20,133 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=199508.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:54:37,269 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=199519.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 02:54:56,458 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199532.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 02:55:00,504 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=199535.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:55:06,355 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4935, 4.5628, 4.8881, 4.8569, 4.8509, 4.5590, 4.5040, 4.3786], device='cuda:6'), covar=tensor([0.0318, 0.0444, 0.0349, 0.0368, 0.0491, 0.0399, 0.0982, 0.0545], device='cuda:6'), in_proj_covar=tensor([0.0397, 0.0437, 0.0422, 0.0396, 0.0473, 0.0447, 0.0539, 0.0359], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 02:55:06,378 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=199539.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:55:08,895 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0796, 2.8749, 3.0845, 1.8167, 3.2305, 3.3297, 2.6161, 2.5227], device='cuda:6'), covar=tensor([0.0955, 0.0289, 0.0219, 0.1262, 0.0108, 0.0222, 0.0506, 0.0558], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0108, 0.0097, 0.0139, 0.0079, 0.0123, 0.0128, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 02:55:25,176 INFO [train.py:904] (6/8) Epoch 20, batch 6700, loss[loss=0.2068, simple_loss=0.2926, pruned_loss=0.06047, over 16739.00 frames. ], tot_loss[loss=0.2077, simple_loss=0.2926, pruned_loss=0.06143, over 3077187.35 frames. ], batch size: 76, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:55:31,431 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=199555.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 02:55:32,678 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=199556.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:56:02,758 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.649e+02 2.838e+02 3.650e+02 4.570e+02 8.139e+02, threshold=7.299e+02, percent-clipped=7.0 2023-05-01 02:56:13,404 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=199583.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:56:18,217 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=199587.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:56:31,546 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=199596.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 02:56:38,605 INFO [train.py:904] (6/8) Epoch 20, batch 6750, loss[loss=0.2243, simple_loss=0.3026, pruned_loss=0.07304, over 15263.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2915, pruned_loss=0.0611, over 3089187.95 frames. ], batch size: 191, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:56:51,597 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7875, 3.8688, 4.1270, 4.0992, 4.0977, 3.8625, 3.8575, 3.8942], device='cuda:6'), covar=tensor([0.0385, 0.0614, 0.0414, 0.0421, 0.0538, 0.0485, 0.0930, 0.0514], device='cuda:6'), in_proj_covar=tensor([0.0400, 0.0440, 0.0425, 0.0399, 0.0476, 0.0450, 0.0541, 0.0361], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 02:57:20,748 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=199631.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:57:32,610 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2561, 4.0965, 4.3238, 4.4665, 4.5832, 4.1804, 4.5400, 4.6105], device='cuda:6'), covar=tensor([0.1701, 0.1282, 0.1457, 0.0678, 0.0576, 0.1219, 0.0763, 0.0648], device='cuda:6'), in_proj_covar=tensor([0.0617, 0.0759, 0.0888, 0.0772, 0.0586, 0.0606, 0.0624, 0.0723], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 02:57:53,090 INFO [train.py:904] (6/8) Epoch 20, batch 6800, loss[loss=0.1969, simple_loss=0.2943, pruned_loss=0.04978, over 16866.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2913, pruned_loss=0.06054, over 3106484.84 frames. ], batch size: 96, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:58:29,264 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.005e+02 2.711e+02 3.238e+02 3.835e+02 6.988e+02, threshold=6.476e+02, percent-clipped=0.0 2023-05-01 02:59:05,436 INFO [train.py:904] (6/8) Epoch 20, batch 6850, loss[loss=0.1842, simple_loss=0.2848, pruned_loss=0.0418, over 16713.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2921, pruned_loss=0.06079, over 3104476.07 frames. ], batch size: 83, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 03:00:20,622 INFO [train.py:904] (6/8) Epoch 20, batch 6900, loss[loss=0.2104, simple_loss=0.2989, pruned_loss=0.06099, over 16422.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2947, pruned_loss=0.06029, over 3123149.36 frames. ], batch size: 146, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 03:00:26,881 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199756.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:00:59,795 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.774e+02 2.558e+02 3.121e+02 3.929e+02 6.424e+02, threshold=6.242e+02, percent-clipped=0.0 2023-05-01 03:01:36,153 INFO [train.py:904] (6/8) Epoch 20, batch 6950, loss[loss=0.2113, simple_loss=0.297, pruned_loss=0.06278, over 16646.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.295, pruned_loss=0.06076, over 3139598.25 frames. ], batch size: 57, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 03:01:38,469 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6590, 3.6714, 2.2815, 4.2012, 2.9159, 4.1605, 2.3390, 2.9315], device='cuda:6'), covar=tensor([0.0263, 0.0358, 0.1688, 0.0249, 0.0768, 0.0552, 0.1623, 0.0792], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0175, 0.0194, 0.0158, 0.0175, 0.0214, 0.0201, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 03:01:55,025 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199814.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:01:56,675 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-01 03:02:05,636 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9147, 4.1638, 3.9801, 4.0298, 3.7066, 3.8048, 3.8440, 4.1575], device='cuda:6'), covar=tensor([0.1112, 0.0909, 0.1054, 0.0813, 0.0808, 0.1637, 0.0946, 0.1000], device='cuda:6'), in_proj_covar=tensor([0.0646, 0.0790, 0.0655, 0.0595, 0.0500, 0.0509, 0.0664, 0.0613], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 03:02:22,395 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=199832.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:02:47,649 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199850.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 03:02:49,629 INFO [train.py:904] (6/8) Epoch 20, batch 7000, loss[loss=0.2046, simple_loss=0.3057, pruned_loss=0.05175, over 16440.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2953, pruned_loss=0.06036, over 3142082.35 frames. ], batch size: 68, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 03:02:54,370 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 03:03:29,995 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.723e+02 2.829e+02 3.376e+02 4.341e+02 9.860e+02, threshold=6.751e+02, percent-clipped=7.0 2023-05-01 03:03:33,462 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=199880.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:03:49,928 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199891.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 03:04:07,200 INFO [train.py:904] (6/8) Epoch 20, batch 7050, loss[loss=0.195, simple_loss=0.2832, pruned_loss=0.05336, over 16765.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2963, pruned_loss=0.06101, over 3125117.36 frames. ], batch size: 39, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 03:04:18,528 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=199909.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:04:20,658 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-05-01 03:05:25,770 INFO [train.py:904] (6/8) Epoch 20, batch 7100, loss[loss=0.2024, simple_loss=0.2846, pruned_loss=0.06017, over 16408.00 frames. ], tot_loss[loss=0.2077, simple_loss=0.2944, pruned_loss=0.06049, over 3112127.47 frames. ], batch size: 146, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:05:54,818 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=199970.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:05:56,375 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 03:06:05,961 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.882e+02 2.782e+02 3.561e+02 4.163e+02 7.041e+02, threshold=7.121e+02, percent-clipped=1.0 2023-05-01 03:06:12,042 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9701, 4.0571, 3.8813, 3.6637, 3.6080, 4.0219, 3.6545, 3.7783], device='cuda:6'), covar=tensor([0.0686, 0.0557, 0.0347, 0.0328, 0.0841, 0.0467, 0.1012, 0.0598], device='cuda:6'), in_proj_covar=tensor([0.0279, 0.0403, 0.0323, 0.0319, 0.0334, 0.0372, 0.0224, 0.0388], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 03:06:30,929 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2265, 3.9181, 3.8454, 2.3970, 3.4801, 3.8884, 3.5279, 2.1894], device='cuda:6'), covar=tensor([0.0586, 0.0051, 0.0052, 0.0457, 0.0112, 0.0109, 0.0103, 0.0478], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0081, 0.0080, 0.0133, 0.0095, 0.0107, 0.0092, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 03:06:45,331 INFO [train.py:904] (6/8) Epoch 20, batch 7150, loss[loss=0.208, simple_loss=0.2922, pruned_loss=0.06187, over 16178.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2928, pruned_loss=0.06054, over 3100611.02 frames. ], batch size: 165, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:06:54,938 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7927, 5.0712, 5.2294, 5.0380, 5.1432, 5.6146, 5.0982, 4.8568], device='cuda:6'), covar=tensor([0.1103, 0.1742, 0.1945, 0.1746, 0.2053, 0.0854, 0.1488, 0.2318], device='cuda:6'), in_proj_covar=tensor([0.0401, 0.0580, 0.0638, 0.0479, 0.0638, 0.0667, 0.0497, 0.0648], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 03:07:59,653 INFO [train.py:904] (6/8) Epoch 20, batch 7200, loss[loss=0.1967, simple_loss=0.2925, pruned_loss=0.05042, over 16400.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2907, pruned_loss=0.05905, over 3079031.10 frames. ], batch size: 146, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:08:06,595 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200056.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:08:40,489 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.812e+02 2.598e+02 3.072e+02 3.513e+02 6.350e+02, threshold=6.144e+02, percent-clipped=0.0 2023-05-01 03:09:19,964 INFO [train.py:904] (6/8) Epoch 20, batch 7250, loss[loss=0.2164, simple_loss=0.2927, pruned_loss=0.07003, over 16397.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2882, pruned_loss=0.05767, over 3084329.36 frames. ], batch size: 35, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:09:23,330 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200104.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:09:37,483 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200114.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:09:57,269 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2023-05-01 03:10:32,021 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200150.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 03:10:35,115 INFO [train.py:904] (6/8) Epoch 20, batch 7300, loss[loss=0.1839, simple_loss=0.2723, pruned_loss=0.04775, over 17046.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2875, pruned_loss=0.05718, over 3100958.16 frames. ], batch size: 53, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:10:48,109 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3457, 3.1127, 2.5422, 2.1003, 2.1831, 2.1806, 3.2894, 2.9164], device='cuda:6'), covar=tensor([0.2998, 0.0782, 0.1965, 0.2725, 0.2864, 0.2283, 0.0481, 0.1303], device='cuda:6'), in_proj_covar=tensor([0.0328, 0.0269, 0.0304, 0.0310, 0.0297, 0.0256, 0.0294, 0.0334], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 03:10:50,130 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-01 03:10:50,982 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200162.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:10:57,716 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5731, 3.6554, 2.3093, 4.3074, 2.9226, 4.2270, 2.3873, 2.8832], device='cuda:6'), covar=tensor([0.0302, 0.0343, 0.1690, 0.0157, 0.0785, 0.0442, 0.1522, 0.0783], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0174, 0.0193, 0.0156, 0.0174, 0.0213, 0.0200, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 03:10:57,933 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 03:11:07,536 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0844, 4.0113, 4.1694, 4.2887, 4.3681, 3.9770, 4.3596, 4.4017], device='cuda:6'), covar=tensor([0.1637, 0.0976, 0.1197, 0.0575, 0.0512, 0.1391, 0.0621, 0.0591], device='cuda:6'), in_proj_covar=tensor([0.0619, 0.0762, 0.0890, 0.0775, 0.0586, 0.0608, 0.0625, 0.0723], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 03:11:15,311 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=200177.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:11:15,880 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 2.904e+02 3.413e+02 4.614e+02 7.836e+02, threshold=6.825e+02, percent-clipped=9.0 2023-05-01 03:11:37,008 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200191.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 03:11:47,792 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200198.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 03:11:51,867 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8445, 3.7914, 3.9128, 4.0177, 4.0836, 3.7133, 4.0497, 4.1178], device='cuda:6'), covar=tensor([0.1558, 0.1030, 0.1241, 0.0588, 0.0568, 0.1839, 0.0739, 0.0657], device='cuda:6'), in_proj_covar=tensor([0.0617, 0.0759, 0.0888, 0.0773, 0.0584, 0.0606, 0.0623, 0.0721], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 03:11:52,585 INFO [train.py:904] (6/8) Epoch 20, batch 7350, loss[loss=0.2128, simple_loss=0.2989, pruned_loss=0.06332, over 15342.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2887, pruned_loss=0.05834, over 3087991.78 frames. ], batch size: 190, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:12:51,313 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=200238.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 03:12:52,207 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200239.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:13:12,411 INFO [train.py:904] (6/8) Epoch 20, batch 7400, loss[loss=0.2183, simple_loss=0.3076, pruned_loss=0.06448, over 16714.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2905, pruned_loss=0.0594, over 3079646.18 frames. ], batch size: 57, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:13:33,057 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=200265.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:13:52,749 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.844e+02 2.893e+02 3.356e+02 4.091e+02 7.315e+02, threshold=6.713e+02, percent-clipped=2.0 2023-05-01 03:13:55,421 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=200279.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:14:32,167 INFO [train.py:904] (6/8) Epoch 20, batch 7450, loss[loss=0.2136, simple_loss=0.2879, pruned_loss=0.0697, over 11684.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2915, pruned_loss=0.0606, over 3063618.66 frames. ], batch size: 247, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:14:51,271 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4557, 2.0416, 1.7170, 1.8439, 2.3386, 2.0449, 2.0069, 2.4743], device='cuda:6'), covar=tensor([0.0234, 0.0456, 0.0591, 0.0498, 0.0246, 0.0389, 0.0227, 0.0271], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0227, 0.0220, 0.0220, 0.0229, 0.0228, 0.0228, 0.0224], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 03:15:22,242 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-01 03:15:27,177 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.24 vs. limit=5.0 2023-05-01 03:15:36,335 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=200340.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:15:43,790 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5130, 3.5379, 3.3473, 3.0270, 3.1661, 3.4775, 3.3337, 3.3225], device='cuda:6'), covar=tensor([0.0627, 0.0716, 0.0299, 0.0300, 0.0494, 0.0472, 0.1213, 0.0517], device='cuda:6'), in_proj_covar=tensor([0.0279, 0.0402, 0.0323, 0.0320, 0.0332, 0.0372, 0.0224, 0.0387], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 03:15:54,169 INFO [train.py:904] (6/8) Epoch 20, batch 7500, loss[loss=0.2007, simple_loss=0.2867, pruned_loss=0.05737, over 16763.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2922, pruned_loss=0.06029, over 3056260.99 frames. ], batch size: 83, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:16:04,821 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 03:16:12,596 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7161, 4.9616, 4.7553, 4.7595, 4.4939, 4.4729, 4.4468, 5.0474], device='cuda:6'), covar=tensor([0.1138, 0.0873, 0.1037, 0.0953, 0.0802, 0.1194, 0.1192, 0.0860], device='cuda:6'), in_proj_covar=tensor([0.0643, 0.0789, 0.0651, 0.0593, 0.0496, 0.0509, 0.0661, 0.0609], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 03:16:34,099 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.040e+02 2.766e+02 3.444e+02 4.330e+02 6.961e+02, threshold=6.888e+02, percent-clipped=1.0 2023-05-01 03:17:11,459 INFO [train.py:904] (6/8) Epoch 20, batch 7550, loss[loss=0.2394, simple_loss=0.3102, pruned_loss=0.0843, over 11743.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.291, pruned_loss=0.06023, over 3057379.78 frames. ], batch size: 250, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:18:26,295 INFO [train.py:904] (6/8) Epoch 20, batch 7600, loss[loss=0.2354, simple_loss=0.2933, pruned_loss=0.08873, over 11335.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2901, pruned_loss=0.05985, over 3077427.90 frames. ], batch size: 247, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:18:37,141 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5070, 4.6302, 4.7753, 4.6160, 4.7041, 5.1781, 4.6638, 4.4570], device='cuda:6'), covar=tensor([0.1261, 0.1752, 0.2283, 0.1907, 0.2316, 0.1010, 0.1663, 0.2399], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0582, 0.0640, 0.0480, 0.0639, 0.0671, 0.0502, 0.0649], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 03:19:06,493 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 2.777e+02 3.348e+02 4.097e+02 9.491e+02, threshold=6.697e+02, percent-clipped=3.0 2023-05-01 03:19:44,056 INFO [train.py:904] (6/8) Epoch 20, batch 7650, loss[loss=0.2411, simple_loss=0.3081, pruned_loss=0.08701, over 11646.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2907, pruned_loss=0.06056, over 3077685.03 frames. ], batch size: 248, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:20:33,324 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=200533.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:21:02,175 INFO [train.py:904] (6/8) Epoch 20, batch 7700, loss[loss=0.1806, simple_loss=0.2717, pruned_loss=0.04471, over 16757.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.291, pruned_loss=0.06102, over 3056678.87 frames. ], batch size: 83, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:21:22,872 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200565.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:21:44,265 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.047e+02 2.918e+02 3.681e+02 4.507e+02 8.362e+02, threshold=7.363e+02, percent-clipped=4.0 2023-05-01 03:22:21,772 INFO [train.py:904] (6/8) Epoch 20, batch 7750, loss[loss=0.1995, simple_loss=0.2871, pruned_loss=0.05602, over 16697.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.291, pruned_loss=0.06062, over 3080595.95 frames. ], batch size: 124, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:22:32,857 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 03:22:38,650 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200613.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:23:06,568 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1133, 4.1981, 4.5391, 4.4937, 4.5039, 4.2320, 4.2195, 4.1346], device='cuda:6'), covar=tensor([0.0382, 0.0639, 0.0378, 0.0422, 0.0514, 0.0408, 0.0920, 0.0522], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0439, 0.0425, 0.0396, 0.0474, 0.0447, 0.0538, 0.0360], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 03:23:11,976 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=200635.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:23:28,172 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=200645.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:23:36,962 INFO [train.py:904] (6/8) Epoch 20, batch 7800, loss[loss=0.2847, simple_loss=0.3362, pruned_loss=0.1165, over 11418.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.291, pruned_loss=0.06015, over 3098345.25 frames. ], batch size: 248, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:24:18,679 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.931e+02 2.831e+02 3.422e+02 4.035e+02 8.624e+02, threshold=6.845e+02, percent-clipped=1.0 2023-05-01 03:24:23,088 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2907, 3.1528, 3.5656, 1.7522, 3.7050, 3.7505, 2.8698, 2.7217], device='cuda:6'), covar=tensor([0.0819, 0.0281, 0.0167, 0.1230, 0.0072, 0.0172, 0.0416, 0.0486], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0107, 0.0096, 0.0138, 0.0078, 0.0122, 0.0126, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 03:24:53,417 INFO [train.py:904] (6/8) Epoch 20, batch 7850, loss[loss=0.1995, simple_loss=0.287, pruned_loss=0.05599, over 16487.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2922, pruned_loss=0.06035, over 3094795.94 frames. ], batch size: 68, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:25:00,326 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=200706.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:25:39,780 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-05-01 03:26:08,718 INFO [train.py:904] (6/8) Epoch 20, batch 7900, loss[loss=0.2364, simple_loss=0.3092, pruned_loss=0.08177, over 11724.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2914, pruned_loss=0.05998, over 3078979.48 frames. ], batch size: 248, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:26:49,221 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.098e+02 2.835e+02 3.346e+02 4.054e+02 6.354e+02, threshold=6.693e+02, percent-clipped=0.0 2023-05-01 03:27:27,127 INFO [train.py:904] (6/8) Epoch 20, batch 7950, loss[loss=0.198, simple_loss=0.2848, pruned_loss=0.05565, over 16198.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2914, pruned_loss=0.05993, over 3089253.31 frames. ], batch size: 165, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:28:09,357 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-05-01 03:28:16,241 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200833.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 03:28:44,603 INFO [train.py:904] (6/8) Epoch 20, batch 8000, loss[loss=0.2708, simple_loss=0.3251, pruned_loss=0.1083, over 11518.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2924, pruned_loss=0.06099, over 3075656.45 frames. ], batch size: 247, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:29:24,820 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.028e+02 3.046e+02 3.376e+02 4.028e+02 7.694e+02, threshold=6.753e+02, percent-clipped=2.0 2023-05-01 03:29:28,342 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200881.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 03:29:59,852 INFO [train.py:904] (6/8) Epoch 20, batch 8050, loss[loss=0.1905, simple_loss=0.2769, pruned_loss=0.05205, over 16622.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2917, pruned_loss=0.06026, over 3090177.50 frames. ], batch size: 57, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:30:08,190 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6409, 4.7310, 4.5178, 4.2474, 4.1798, 4.6497, 4.4563, 4.3426], device='cuda:6'), covar=tensor([0.0709, 0.0661, 0.0345, 0.0318, 0.0998, 0.0512, 0.0471, 0.0677], device='cuda:6'), in_proj_covar=tensor([0.0282, 0.0408, 0.0329, 0.0324, 0.0338, 0.0376, 0.0227, 0.0394], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 03:30:12,833 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 03:30:50,051 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200935.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:31:15,303 INFO [train.py:904] (6/8) Epoch 20, batch 8100, loss[loss=0.1915, simple_loss=0.2894, pruned_loss=0.04677, over 16418.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2909, pruned_loss=0.05923, over 3113634.45 frames. ], batch size: 35, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:31:57,146 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.634e+02 3.129e+02 3.819e+02 6.896e+02, threshold=6.259e+02, percent-clipped=1.0 2023-05-01 03:32:04,948 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200983.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:32:31,895 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=201001.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:32:33,316 INFO [train.py:904] (6/8) Epoch 20, batch 8150, loss[loss=0.1868, simple_loss=0.2765, pruned_loss=0.0486, over 16881.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2888, pruned_loss=0.05847, over 3110628.63 frames. ], batch size: 116, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:33:49,646 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4427, 2.6343, 2.1745, 2.3562, 3.0152, 2.6335, 3.0299, 3.1896], device='cuda:6'), covar=tensor([0.0150, 0.0362, 0.0510, 0.0424, 0.0228, 0.0382, 0.0278, 0.0237], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0227, 0.0219, 0.0220, 0.0228, 0.0228, 0.0227, 0.0224], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 03:33:52,112 INFO [train.py:904] (6/8) Epoch 20, batch 8200, loss[loss=0.1942, simple_loss=0.2864, pruned_loss=0.05097, over 15411.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2862, pruned_loss=0.05739, over 3126569.15 frames. ], batch size: 191, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:34:24,438 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201071.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:34:36,686 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.058e+02 2.732e+02 3.390e+02 4.593e+02 1.143e+03, threshold=6.781e+02, percent-clipped=6.0 2023-05-01 03:35:15,217 INFO [train.py:904] (6/8) Epoch 20, batch 8250, loss[loss=0.1767, simple_loss=0.2631, pruned_loss=0.04519, over 12506.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2852, pruned_loss=0.05498, over 3104889.61 frames. ], batch size: 248, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:35:41,859 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-05-01 03:35:53,256 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5929, 4.6400, 4.9678, 4.9707, 4.9677, 4.7048, 4.6124, 4.5615], device='cuda:6'), covar=tensor([0.0362, 0.0616, 0.0432, 0.0400, 0.0470, 0.0401, 0.1030, 0.0425], device='cuda:6'), in_proj_covar=tensor([0.0403, 0.0445, 0.0430, 0.0400, 0.0479, 0.0453, 0.0546, 0.0362], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 03:36:06,048 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=201132.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:36:38,968 INFO [train.py:904] (6/8) Epoch 20, batch 8300, loss[loss=0.1831, simple_loss=0.2802, pruned_loss=0.04297, over 16211.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.2824, pruned_loss=0.05241, over 3080007.97 frames. ], batch size: 165, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:37:22,448 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.604e+02 2.376e+02 2.857e+02 3.605e+02 7.537e+02, threshold=5.714e+02, percent-clipped=2.0 2023-05-01 03:38:00,750 INFO [train.py:904] (6/8) Epoch 20, batch 8350, loss[loss=0.1781, simple_loss=0.2766, pruned_loss=0.03981, over 16836.00 frames. ], tot_loss[loss=0.1916, simple_loss=0.2817, pruned_loss=0.05073, over 3071393.43 frames. ], batch size: 116, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:39:23,230 INFO [train.py:904] (6/8) Epoch 20, batch 8400, loss[loss=0.1605, simple_loss=0.2591, pruned_loss=0.03094, over 15585.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2791, pruned_loss=0.04923, over 3046713.50 frames. ], batch size: 191, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:39:26,491 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201253.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:40:00,817 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201274.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:40:08,501 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.430e+02 2.319e+02 2.651e+02 3.336e+02 5.393e+02, threshold=5.301e+02, percent-clipped=0.0 2023-05-01 03:40:44,407 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=201301.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:40:45,026 INFO [train.py:904] (6/8) Epoch 20, batch 8450, loss[loss=0.1964, simple_loss=0.2782, pruned_loss=0.0573, over 12462.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.2774, pruned_loss=0.0477, over 3035713.10 frames. ], batch size: 250, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:41:05,912 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=201314.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:41:20,194 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.74 vs. limit=5.0 2023-05-01 03:41:29,576 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9666, 4.0155, 4.3167, 4.2893, 4.2999, 4.0606, 4.0180, 4.0868], device='cuda:6'), covar=tensor([0.0426, 0.0900, 0.0485, 0.0514, 0.0529, 0.0578, 0.1194, 0.0570], device='cuda:6'), in_proj_covar=tensor([0.0403, 0.0445, 0.0430, 0.0400, 0.0478, 0.0451, 0.0542, 0.0363], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 03:41:37,915 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-01 03:41:39,438 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=201335.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:41:52,185 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5176, 3.0567, 2.7623, 2.2644, 2.1930, 2.3264, 2.9187, 2.8287], device='cuda:6'), covar=tensor([0.2384, 0.0711, 0.1527, 0.2717, 0.2777, 0.2146, 0.0443, 0.1322], device='cuda:6'), in_proj_covar=tensor([0.0320, 0.0262, 0.0297, 0.0302, 0.0289, 0.0250, 0.0286, 0.0324], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 03:42:02,486 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=201349.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:42:07,348 INFO [train.py:904] (6/8) Epoch 20, batch 8500, loss[loss=0.186, simple_loss=0.2724, pruned_loss=0.04985, over 16764.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2738, pruned_loss=0.04556, over 3039046.91 frames. ], batch size: 124, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:42:50,125 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.494e+02 2.179e+02 2.765e+02 3.393e+02 6.239e+02, threshold=5.531e+02, percent-clipped=4.0 2023-05-01 03:43:31,052 INFO [train.py:904] (6/8) Epoch 20, batch 8550, loss[loss=0.1785, simple_loss=0.2782, pruned_loss=0.03943, over 16364.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2716, pruned_loss=0.04471, over 3014082.03 frames. ], batch size: 146, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:43:41,262 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-01 03:44:19,131 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=201427.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:44:24,514 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2023-05-01 03:45:09,639 INFO [train.py:904] (6/8) Epoch 20, batch 8600, loss[loss=0.1752, simple_loss=0.2587, pruned_loss=0.04589, over 12252.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2718, pruned_loss=0.04342, over 3030013.38 frames. ], batch size: 246, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:46:02,940 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 2.180e+02 2.549e+02 3.038e+02 6.489e+02, threshold=5.098e+02, percent-clipped=1.0 2023-05-01 03:46:09,241 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6476, 4.4504, 4.7036, 4.8264, 5.0114, 4.5350, 4.9919, 4.9921], device='cuda:6'), covar=tensor([0.1998, 0.1270, 0.1671, 0.0718, 0.0558, 0.0957, 0.0504, 0.0637], device='cuda:6'), in_proj_covar=tensor([0.0604, 0.0747, 0.0874, 0.0760, 0.0577, 0.0599, 0.0616, 0.0713], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 03:46:48,552 INFO [train.py:904] (6/8) Epoch 20, batch 8650, loss[loss=0.1699, simple_loss=0.2603, pruned_loss=0.03972, over 11917.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2701, pruned_loss=0.04201, over 3032703.99 frames. ], batch size: 248, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 03:47:23,835 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-05-01 03:48:14,843 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-01 03:48:36,194 INFO [train.py:904] (6/8) Epoch 20, batch 8700, loss[loss=0.1791, simple_loss=0.2833, pruned_loss=0.03747, over 15344.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2674, pruned_loss=0.04077, over 3051019.34 frames. ], batch size: 190, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 03:49:28,969 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.047e+02 2.487e+02 3.171e+02 4.995e+02, threshold=4.975e+02, percent-clipped=0.0 2023-05-01 03:50:00,267 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6707, 2.7076, 2.3240, 3.8724, 2.3317, 3.7936, 1.4643, 2.8308], device='cuda:6'), covar=tensor([0.1404, 0.0696, 0.1211, 0.0165, 0.0108, 0.0370, 0.1780, 0.0775], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0169, 0.0190, 0.0181, 0.0202, 0.0209, 0.0197, 0.0188], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 03:50:13,600 INFO [train.py:904] (6/8) Epoch 20, batch 8750, loss[loss=0.1663, simple_loss=0.2523, pruned_loss=0.04018, over 12327.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2669, pruned_loss=0.04034, over 3032831.54 frames. ], batch size: 250, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 03:50:31,548 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=201609.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:50:53,786 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-01 03:51:13,444 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-01 03:51:22,016 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=201630.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:51:51,019 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201644.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:52:00,056 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6450, 3.9438, 2.8752, 2.2643, 2.4486, 2.5315, 4.1630, 3.3633], device='cuda:6'), covar=tensor([0.3038, 0.0556, 0.1830, 0.2780, 0.2684, 0.1973, 0.0380, 0.1286], device='cuda:6'), in_proj_covar=tensor([0.0319, 0.0263, 0.0297, 0.0301, 0.0287, 0.0250, 0.0285, 0.0324], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 03:52:07,673 INFO [train.py:904] (6/8) Epoch 20, batch 8800, loss[loss=0.1721, simple_loss=0.2648, pruned_loss=0.03965, over 16733.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2656, pruned_loss=0.03961, over 3031818.29 frames. ], batch size: 124, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 03:52:25,357 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201661.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:53:06,514 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.243e+02 2.623e+02 3.212e+02 5.763e+02, threshold=5.246e+02, percent-clipped=4.0 2023-05-01 03:53:51,633 INFO [train.py:904] (6/8) Epoch 20, batch 8850, loss[loss=0.1532, simple_loss=0.245, pruned_loss=0.03071, over 12264.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2677, pruned_loss=0.03871, over 3032679.33 frames. ], batch size: 248, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 03:53:57,907 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=201705.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:54:00,620 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201706.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:54:34,712 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=201722.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:54:39,862 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.11 vs. limit=5.0 2023-05-01 03:54:45,908 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=201727.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:55:26,624 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-01 03:55:38,109 INFO [train.py:904] (6/8) Epoch 20, batch 8900, loss[loss=0.1487, simple_loss=0.2387, pruned_loss=0.02938, over 12563.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.268, pruned_loss=0.03805, over 3040107.40 frames. ], batch size: 248, lr: 3.35e-03, grad_scale: 2.0 2023-05-01 03:56:08,882 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=201767.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:56:27,848 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=201775.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:56:46,606 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.216e+02 2.705e+02 3.264e+02 5.822e+02, threshold=5.410e+02, percent-clipped=1.0 2023-05-01 03:57:13,659 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2386, 3.1985, 3.4117, 1.7213, 3.5752, 3.6685, 2.8992, 2.8049], device='cuda:6'), covar=tensor([0.0836, 0.0252, 0.0197, 0.1299, 0.0082, 0.0148, 0.0418, 0.0477], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0104, 0.0092, 0.0134, 0.0075, 0.0117, 0.0123, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 03:57:31,499 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9474, 4.2574, 4.0996, 4.1232, 3.8366, 3.8376, 3.8315, 4.2537], device='cuda:6'), covar=tensor([0.1172, 0.0920, 0.0947, 0.0754, 0.0750, 0.1812, 0.1037, 0.0920], device='cuda:6'), in_proj_covar=tensor([0.0632, 0.0773, 0.0636, 0.0581, 0.0490, 0.0501, 0.0647, 0.0598], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 03:57:41,706 INFO [train.py:904] (6/8) Epoch 20, batch 8950, loss[loss=0.1597, simple_loss=0.2505, pruned_loss=0.0345, over 16948.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2675, pruned_loss=0.03843, over 3042453.73 frames. ], batch size: 109, lr: 3.35e-03, grad_scale: 2.0 2023-05-01 03:59:08,610 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.59 vs. limit=5.0 2023-05-01 03:59:30,703 INFO [train.py:904] (6/8) Epoch 20, batch 9000, loss[loss=0.1697, simple_loss=0.2591, pruned_loss=0.04013, over 16813.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2647, pruned_loss=0.03744, over 3044062.62 frames. ], batch size: 124, lr: 3.35e-03, grad_scale: 2.0 2023-05-01 03:59:30,703 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 03:59:41,113 INFO [train.py:938] (6/8) Epoch 20, validation: loss=0.1464, simple_loss=0.2502, pruned_loss=0.02125, over 944034.00 frames. 2023-05-01 03:59:41,113 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 03:59:48,657 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3656, 4.6777, 4.4836, 4.4909, 4.2003, 4.1860, 4.2063, 4.7136], device='cuda:6'), covar=tensor([0.1151, 0.0869, 0.0958, 0.0771, 0.0750, 0.1465, 0.1075, 0.0850], device='cuda:6'), in_proj_covar=tensor([0.0626, 0.0767, 0.0630, 0.0576, 0.0486, 0.0496, 0.0640, 0.0594], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 04:00:41,799 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.083e+02 2.552e+02 3.280e+02 1.556e+03, threshold=5.104e+02, percent-clipped=3.0 2023-05-01 04:01:24,589 INFO [train.py:904] (6/8) Epoch 20, batch 9050, loss[loss=0.1573, simple_loss=0.2526, pruned_loss=0.03104, over 16870.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2657, pruned_loss=0.03804, over 3045528.92 frames. ], batch size: 102, lr: 3.35e-03, grad_scale: 2.0 2023-05-01 04:01:40,824 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=201909.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:02:21,240 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=201930.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:03:12,543 INFO [train.py:904] (6/8) Epoch 20, batch 9100, loss[loss=0.1911, simple_loss=0.2885, pruned_loss=0.04681, over 16097.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2652, pruned_loss=0.03856, over 3049278.36 frames. ], batch size: 165, lr: 3.35e-03, grad_scale: 2.0 2023-05-01 04:03:22,655 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=201957.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:04:03,897 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0921, 5.0773, 4.8403, 4.4254, 4.9235, 1.9217, 4.6915, 4.6385], device='cuda:6'), covar=tensor([0.0085, 0.0097, 0.0193, 0.0339, 0.0106, 0.2602, 0.0140, 0.0219], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0145, 0.0185, 0.0167, 0.0164, 0.0199, 0.0176, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 04:04:14,332 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=201978.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:04:23,055 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 2.025e+02 2.515e+02 2.947e+02 5.747e+02, threshold=5.029e+02, percent-clipped=1.0 2023-05-01 04:04:43,672 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201990.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:05:11,355 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=202000.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:05:14,326 INFO [train.py:904] (6/8) Epoch 20, batch 9150, loss[loss=0.1481, simple_loss=0.2446, pruned_loss=0.02582, over 15364.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2663, pruned_loss=0.03857, over 3049025.00 frames. ], batch size: 191, lr: 3.35e-03, grad_scale: 2.0 2023-05-01 04:05:48,734 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=202017.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:06:59,071 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=202051.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:06:59,884 INFO [train.py:904] (6/8) Epoch 20, batch 9200, loss[loss=0.1837, simple_loss=0.2726, pruned_loss=0.04741, over 16421.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2623, pruned_loss=0.0377, over 3065871.22 frames. ], batch size: 147, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 04:07:20,314 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=202062.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:07:55,480 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.302e+02 2.670e+02 3.574e+02 1.158e+03, threshold=5.341e+02, percent-clipped=6.0 2023-05-01 04:08:37,606 INFO [train.py:904] (6/8) Epoch 20, batch 9250, loss[loss=0.1544, simple_loss=0.2493, pruned_loss=0.02971, over 16866.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2615, pruned_loss=0.0376, over 3048318.42 frames. ], batch size: 96, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 04:09:53,841 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-01 04:10:29,962 INFO [train.py:904] (6/8) Epoch 20, batch 9300, loss[loss=0.1569, simple_loss=0.2498, pruned_loss=0.03197, over 15353.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2601, pruned_loss=0.03717, over 3054252.74 frames. ], batch size: 191, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 04:11:18,484 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5112, 3.6756, 2.6802, 2.0921, 2.2260, 2.3298, 3.8722, 3.2036], device='cuda:6'), covar=tensor([0.3094, 0.0636, 0.1857, 0.2866, 0.2888, 0.2184, 0.0395, 0.1239], device='cuda:6'), in_proj_covar=tensor([0.0315, 0.0260, 0.0294, 0.0298, 0.0281, 0.0248, 0.0282, 0.0320], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 04:11:37,344 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.443e+02 1.963e+02 2.278e+02 2.874e+02 5.860e+02, threshold=4.556e+02, percent-clipped=2.0 2023-05-01 04:12:16,143 INFO [train.py:904] (6/8) Epoch 20, batch 9350, loss[loss=0.1974, simple_loss=0.2965, pruned_loss=0.04914, over 16401.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.26, pruned_loss=0.03718, over 3056708.43 frames. ], batch size: 146, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 04:13:59,184 INFO [train.py:904] (6/8) Epoch 20, batch 9400, loss[loss=0.1737, simple_loss=0.2758, pruned_loss=0.03577, over 16179.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2607, pruned_loss=0.03687, over 3066288.54 frames. ], batch size: 165, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 04:14:45,268 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-01 04:15:00,973 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.064e+02 2.476e+02 2.960e+02 4.580e+02, threshold=4.951e+02, percent-clipped=1.0 2023-05-01 04:15:39,405 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=202300.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:15:41,530 INFO [train.py:904] (6/8) Epoch 20, batch 9450, loss[loss=0.1675, simple_loss=0.2705, pruned_loss=0.03224, over 16218.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2624, pruned_loss=0.03705, over 3072493.93 frames. ], batch size: 165, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 04:15:55,005 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0860, 3.9359, 4.1294, 4.2365, 4.3693, 3.9673, 4.3546, 4.3848], device='cuda:6'), covar=tensor([0.1874, 0.1334, 0.1626, 0.0835, 0.0655, 0.1397, 0.0731, 0.0750], device='cuda:6'), in_proj_covar=tensor([0.0594, 0.0730, 0.0851, 0.0750, 0.0566, 0.0587, 0.0602, 0.0697], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 04:16:13,580 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=202317.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:17:14,533 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=202346.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:17:18,058 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=202348.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:17:25,743 INFO [train.py:904] (6/8) Epoch 20, batch 9500, loss[loss=0.1691, simple_loss=0.2684, pruned_loss=0.03488, over 16906.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2616, pruned_loss=0.03675, over 3062737.41 frames. ], batch size: 96, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:17:49,681 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=202362.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:17:55,125 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=202365.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:18:26,628 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 2.130e+02 2.550e+02 3.364e+02 1.443e+03, threshold=5.099e+02, percent-clipped=7.0 2023-05-01 04:19:13,220 INFO [train.py:904] (6/8) Epoch 20, batch 9550, loss[loss=0.1638, simple_loss=0.2569, pruned_loss=0.03533, over 12705.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2609, pruned_loss=0.03678, over 3058715.84 frames. ], batch size: 248, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:19:32,678 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=202410.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:20:55,234 INFO [train.py:904] (6/8) Epoch 20, batch 9600, loss[loss=0.1883, simple_loss=0.2672, pruned_loss=0.05471, over 12328.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2626, pruned_loss=0.0375, over 3045794.38 frames. ], batch size: 248, lr: 3.34e-03, grad_scale: 8.0 2023-05-01 04:21:53,893 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.693e+02 2.243e+02 2.599e+02 3.368e+02 6.147e+02, threshold=5.198e+02, percent-clipped=3.0 2023-05-01 04:22:41,347 INFO [train.py:904] (6/8) Epoch 20, batch 9650, loss[loss=0.1573, simple_loss=0.2602, pruned_loss=0.02722, over 16942.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.264, pruned_loss=0.03749, over 3056488.76 frames. ], batch size: 102, lr: 3.34e-03, grad_scale: 8.0 2023-05-01 04:23:36,109 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.26 vs. limit=5.0 2023-05-01 04:24:31,592 INFO [train.py:904] (6/8) Epoch 20, batch 9700, loss[loss=0.183, simple_loss=0.2834, pruned_loss=0.0413, over 16366.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.263, pruned_loss=0.03707, over 3060101.07 frames. ], batch size: 146, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:25:37,513 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.240e+02 2.594e+02 3.217e+02 6.091e+02, threshold=5.188e+02, percent-clipped=1.0 2023-05-01 04:26:15,967 INFO [train.py:904] (6/8) Epoch 20, batch 9750, loss[loss=0.1672, simple_loss=0.2662, pruned_loss=0.03408, over 16932.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2618, pruned_loss=0.03721, over 3084390.00 frames. ], batch size: 116, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:26:18,912 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3892, 3.1389, 2.5982, 2.2188, 2.1174, 2.1074, 3.0127, 2.8143], device='cuda:6'), covar=tensor([0.2627, 0.0787, 0.1906, 0.2820, 0.3035, 0.2592, 0.0518, 0.1445], device='cuda:6'), in_proj_covar=tensor([0.0314, 0.0259, 0.0294, 0.0297, 0.0280, 0.0248, 0.0281, 0.0319], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 04:27:47,569 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=202646.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:27:57,087 INFO [train.py:904] (6/8) Epoch 20, batch 9800, loss[loss=0.1673, simple_loss=0.2704, pruned_loss=0.03206, over 16813.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2616, pruned_loss=0.03605, over 3077445.69 frames. ], batch size: 124, lr: 3.34e-03, grad_scale: 2.0 2023-05-01 04:28:06,719 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3693, 4.0865, 4.1317, 2.8445, 3.6178, 4.0688, 3.7706, 2.2838], device='cuda:6'), covar=tensor([0.0599, 0.0036, 0.0037, 0.0374, 0.0102, 0.0091, 0.0068, 0.0519], device='cuda:6'), in_proj_covar=tensor([0.0131, 0.0078, 0.0077, 0.0129, 0.0094, 0.0103, 0.0088, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 04:28:57,878 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.120e+02 2.499e+02 2.833e+02 7.051e+02, threshold=4.998e+02, percent-clipped=2.0 2023-05-01 04:29:26,658 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=202694.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:29:42,362 INFO [train.py:904] (6/8) Epoch 20, batch 9850, loss[loss=0.1564, simple_loss=0.2614, pruned_loss=0.02575, over 16891.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2627, pruned_loss=0.03588, over 3068170.90 frames. ], batch size: 102, lr: 3.34e-03, grad_scale: 2.0 2023-05-01 04:31:34,887 INFO [train.py:904] (6/8) Epoch 20, batch 9900, loss[loss=0.1743, simple_loss=0.2729, pruned_loss=0.03783, over 16977.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2638, pruned_loss=0.03622, over 3047515.45 frames. ], batch size: 109, lr: 3.34e-03, grad_scale: 2.0 2023-05-01 04:31:38,478 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5521, 3.5632, 3.5308, 2.9110, 3.3962, 1.9629, 3.2728, 2.9398], device='cuda:6'), covar=tensor([0.0135, 0.0137, 0.0170, 0.0222, 0.0125, 0.2482, 0.0139, 0.0267], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0144, 0.0183, 0.0164, 0.0164, 0.0198, 0.0175, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 04:32:49,189 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.306e+02 2.190e+02 2.483e+02 3.109e+02 6.841e+02, threshold=4.966e+02, percent-clipped=3.0 2023-05-01 04:33:30,684 INFO [train.py:904] (6/8) Epoch 20, batch 9950, loss[loss=0.1527, simple_loss=0.2506, pruned_loss=0.0274, over 16658.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2655, pruned_loss=0.03652, over 3034342.84 frames. ], batch size: 62, lr: 3.34e-03, grad_scale: 2.0 2023-05-01 04:33:32,517 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-01 04:33:41,365 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1773, 2.9452, 3.1873, 1.8350, 3.3490, 3.4140, 2.8012, 2.6109], device='cuda:6'), covar=tensor([0.0784, 0.0268, 0.0148, 0.1190, 0.0086, 0.0142, 0.0372, 0.0482], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0105, 0.0091, 0.0135, 0.0076, 0.0117, 0.0124, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 04:33:57,559 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9597, 2.9745, 2.5743, 2.7670, 3.3914, 3.1873, 3.4679, 3.6057], device='cuda:6'), covar=tensor([0.0129, 0.0403, 0.0501, 0.0420, 0.0253, 0.0316, 0.0326, 0.0207], device='cuda:6'), in_proj_covar=tensor([0.0191, 0.0225, 0.0217, 0.0217, 0.0225, 0.0225, 0.0220, 0.0218], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 04:34:42,231 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=202830.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 04:35:32,299 INFO [train.py:904] (6/8) Epoch 20, batch 10000, loss[loss=0.1899, simple_loss=0.283, pruned_loss=0.04835, over 16661.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2644, pruned_loss=0.03645, over 3052735.87 frames. ], batch size: 134, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:36:36,180 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.432e+02 2.233e+02 2.660e+02 3.205e+02 6.054e+02, threshold=5.319e+02, percent-clipped=4.0 2023-05-01 04:36:52,307 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=202891.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 04:37:11,275 INFO [train.py:904] (6/8) Epoch 20, batch 10050, loss[loss=0.1528, simple_loss=0.249, pruned_loss=0.02828, over 16601.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2644, pruned_loss=0.0365, over 3048010.23 frames. ], batch size: 62, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:38:05,709 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.22 vs. limit=5.0 2023-05-01 04:38:17,149 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0330, 2.2619, 1.9507, 2.0756, 2.6629, 2.3231, 2.4651, 2.7711], device='cuda:6'), covar=tensor([0.0154, 0.0470, 0.0558, 0.0511, 0.0288, 0.0464, 0.0190, 0.0282], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0227, 0.0218, 0.0218, 0.0226, 0.0227, 0.0221, 0.0219], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 04:38:41,705 INFO [train.py:904] (6/8) Epoch 20, batch 10100, loss[loss=0.1534, simple_loss=0.2487, pruned_loss=0.02909, over 15433.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2642, pruned_loss=0.03634, over 3053308.94 frames. ], batch size: 190, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:39:40,380 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.271e+02 2.682e+02 3.301e+02 6.522e+02, threshold=5.364e+02, percent-clipped=2.0 2023-05-01 04:40:23,119 INFO [train.py:904] (6/8) Epoch 21, batch 0, loss[loss=0.1561, simple_loss=0.2367, pruned_loss=0.03776, over 16746.00 frames. ], tot_loss[loss=0.1561, simple_loss=0.2367, pruned_loss=0.03776, over 16746.00 frames. ], batch size: 39, lr: 3.26e-03, grad_scale: 8.0 2023-05-01 04:40:23,119 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 04:40:30,888 INFO [train.py:938] (6/8) Epoch 21, validation: loss=0.1451, simple_loss=0.2491, pruned_loss=0.02058, over 944034.00 frames. 2023-05-01 04:40:30,888 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 04:41:07,237 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0396, 3.8994, 4.0894, 4.1787, 4.2550, 3.8895, 4.1267, 4.2571], device='cuda:6'), covar=tensor([0.1543, 0.1166, 0.1223, 0.0702, 0.0635, 0.1327, 0.1394, 0.0855], device='cuda:6'), in_proj_covar=tensor([0.0594, 0.0726, 0.0847, 0.0750, 0.0565, 0.0582, 0.0603, 0.0699], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 04:41:19,714 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203038.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:41:36,907 INFO [train.py:904] (6/8) Epoch 21, batch 50, loss[loss=0.1691, simple_loss=0.2503, pruned_loss=0.04393, over 16465.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2728, pruned_loss=0.05373, over 758471.59 frames. ], batch size: 75, lr: 3.26e-03, grad_scale: 2.0 2023-05-01 04:42:25,364 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203085.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:42:26,045 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.693e+02 2.424e+02 2.967e+02 3.842e+02 7.470e+02, threshold=5.934e+02, percent-clipped=2.0 2023-05-01 04:42:42,662 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8205, 2.7861, 2.6934, 4.8826, 3.7445, 4.2118, 1.5680, 3.0345], device='cuda:6'), covar=tensor([0.1455, 0.0922, 0.1344, 0.0270, 0.0288, 0.0490, 0.1873, 0.0915], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0171, 0.0191, 0.0181, 0.0199, 0.0210, 0.0198, 0.0188], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 04:42:43,785 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203099.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:42:48,525 INFO [train.py:904] (6/8) Epoch 21, batch 100, loss[loss=0.1763, simple_loss=0.2719, pruned_loss=0.04042, over 17138.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.27, pruned_loss=0.04987, over 1324009.02 frames. ], batch size: 49, lr: 3.26e-03, grad_scale: 2.0 2023-05-01 04:43:50,317 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203146.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:43:52,347 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-05-01 04:43:57,341 INFO [train.py:904] (6/8) Epoch 21, batch 150, loss[loss=0.1704, simple_loss=0.271, pruned_loss=0.03488, over 17110.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2662, pruned_loss=0.04737, over 1766557.30 frames. ], batch size: 47, lr: 3.26e-03, grad_scale: 2.0 2023-05-01 04:44:44,682 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.604e+02 2.249e+02 2.595e+02 3.084e+02 6.012e+02, threshold=5.190e+02, percent-clipped=1.0 2023-05-01 04:44:45,001 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=203186.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 04:44:54,636 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-05-01 04:45:04,855 INFO [train.py:904] (6/8) Epoch 21, batch 200, loss[loss=0.1669, simple_loss=0.2576, pruned_loss=0.03812, over 16508.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2655, pruned_loss=0.04747, over 2106090.05 frames. ], batch size: 68, lr: 3.26e-03, grad_scale: 2.0 2023-05-01 04:45:33,080 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2602, 3.3215, 3.5579, 1.9361, 3.7147, 3.7748, 2.9151, 2.6602], device='cuda:6'), covar=tensor([0.1112, 0.0232, 0.0205, 0.1267, 0.0099, 0.0170, 0.0474, 0.0626], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0106, 0.0093, 0.0137, 0.0077, 0.0119, 0.0125, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 04:45:59,831 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1214, 5.1912, 4.9502, 4.5374, 4.4823, 5.0716, 5.1176, 4.6066], device='cuda:6'), covar=tensor([0.0735, 0.0570, 0.0462, 0.0441, 0.1367, 0.0486, 0.0314, 0.0928], device='cuda:6'), in_proj_covar=tensor([0.0281, 0.0402, 0.0325, 0.0320, 0.0331, 0.0370, 0.0223, 0.0388], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 04:46:15,293 INFO [train.py:904] (6/8) Epoch 21, batch 250, loss[loss=0.1855, simple_loss=0.2635, pruned_loss=0.0538, over 12138.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.262, pruned_loss=0.04614, over 2373001.00 frames. ], batch size: 247, lr: 3.26e-03, grad_scale: 2.0 2023-05-01 04:46:38,893 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8947, 4.2987, 2.9369, 2.3315, 2.7319, 2.4786, 4.7106, 3.6075], device='cuda:6'), covar=tensor([0.2796, 0.0615, 0.1925, 0.2820, 0.2844, 0.2136, 0.0354, 0.1297], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0266, 0.0301, 0.0305, 0.0287, 0.0254, 0.0289, 0.0330], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 04:47:00,829 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.546e+02 2.163e+02 2.580e+02 3.064e+02 7.723e+02, threshold=5.160e+02, percent-clipped=4.0 2023-05-01 04:47:23,329 INFO [train.py:904] (6/8) Epoch 21, batch 300, loss[loss=0.1685, simple_loss=0.2619, pruned_loss=0.03754, over 17243.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2602, pruned_loss=0.04578, over 2586469.95 frames. ], batch size: 45, lr: 3.26e-03, grad_scale: 2.0 2023-05-01 04:48:04,133 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3280, 2.3297, 2.3730, 4.1173, 2.3095, 2.6543, 2.4036, 2.4868], device='cuda:6'), covar=tensor([0.1410, 0.3687, 0.3071, 0.0577, 0.4053, 0.2542, 0.3673, 0.3461], device='cuda:6'), in_proj_covar=tensor([0.0395, 0.0440, 0.0364, 0.0322, 0.0432, 0.0503, 0.0410, 0.0513], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 04:48:32,603 INFO [train.py:904] (6/8) Epoch 21, batch 350, loss[loss=0.1696, simple_loss=0.2619, pruned_loss=0.03862, over 17125.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.259, pruned_loss=0.04493, over 2754674.73 frames. ], batch size: 49, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:48:55,301 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7379, 2.6460, 2.8310, 4.7298, 2.6166, 2.9320, 2.7106, 2.8233], device='cuda:6'), covar=tensor([0.1149, 0.3401, 0.2667, 0.0411, 0.3847, 0.2467, 0.3414, 0.3262], device='cuda:6'), in_proj_covar=tensor([0.0395, 0.0441, 0.0364, 0.0323, 0.0432, 0.0503, 0.0410, 0.0513], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 04:49:17,934 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.171e+02 2.523e+02 3.052e+02 8.873e+02, threshold=5.047e+02, percent-clipped=2.0 2023-05-01 04:49:30,141 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=203394.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:49:41,953 INFO [train.py:904] (6/8) Epoch 21, batch 400, loss[loss=0.1944, simple_loss=0.2711, pruned_loss=0.05886, over 16428.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2577, pruned_loss=0.0446, over 2872945.94 frames. ], batch size: 146, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 04:50:13,260 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2023-05-01 04:50:19,397 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2122, 4.0350, 4.4936, 2.5725, 4.7378, 4.7865, 3.3570, 3.7457], device='cuda:6'), covar=tensor([0.0697, 0.0259, 0.0221, 0.1047, 0.0072, 0.0136, 0.0452, 0.0381], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0107, 0.0095, 0.0139, 0.0078, 0.0121, 0.0127, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 04:50:37,061 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=203441.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:50:48,793 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-05-01 04:50:50,734 INFO [train.py:904] (6/8) Epoch 21, batch 450, loss[loss=0.1655, simple_loss=0.2576, pruned_loss=0.03668, over 17182.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2556, pruned_loss=0.04327, over 2967537.35 frames. ], batch size: 46, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 04:50:55,490 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7161, 3.9504, 3.0319, 2.2542, 2.5448, 2.4502, 4.0501, 3.4044], device='cuda:6'), covar=tensor([0.2832, 0.0561, 0.1696, 0.3030, 0.2757, 0.2030, 0.0514, 0.1455], device='cuda:6'), in_proj_covar=tensor([0.0324, 0.0267, 0.0303, 0.0308, 0.0291, 0.0256, 0.0291, 0.0333], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 04:51:10,952 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-01 04:51:11,839 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203467.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:51:37,694 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.511e+02 2.094e+02 2.458e+02 3.011e+02 6.519e+02, threshold=4.916e+02, percent-clipped=3.0 2023-05-01 04:51:38,087 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=203486.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 04:51:59,114 INFO [train.py:904] (6/8) Epoch 21, batch 500, loss[loss=0.1677, simple_loss=0.2658, pruned_loss=0.03483, over 17242.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2553, pruned_loss=0.04205, over 3049848.51 frames. ], batch size: 52, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:52:36,253 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203528.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:52:43,126 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=203534.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 04:53:08,035 INFO [train.py:904] (6/8) Epoch 21, batch 550, loss[loss=0.186, simple_loss=0.2673, pruned_loss=0.05237, over 16441.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2539, pruned_loss=0.04165, over 3099645.16 frames. ], batch size: 75, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:53:09,635 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8958, 2.2651, 2.4614, 1.8878, 2.6774, 2.7251, 2.3920, 2.2152], device='cuda:6'), covar=tensor([0.0939, 0.0271, 0.0277, 0.1090, 0.0135, 0.0265, 0.0501, 0.0585], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0108, 0.0095, 0.0139, 0.0078, 0.0122, 0.0127, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 04:53:52,820 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-01 04:53:56,913 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.204e+02 2.652e+02 3.296e+02 1.402e+03, threshold=5.303e+02, percent-clipped=5.0 2023-05-01 04:54:17,584 INFO [train.py:904] (6/8) Epoch 21, batch 600, loss[loss=0.1432, simple_loss=0.2313, pruned_loss=0.0276, over 15948.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2533, pruned_loss=0.04232, over 3153090.92 frames. ], batch size: 35, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:55:27,189 INFO [train.py:904] (6/8) Epoch 21, batch 650, loss[loss=0.1677, simple_loss=0.245, pruned_loss=0.04524, over 16716.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2521, pruned_loss=0.04225, over 3186163.81 frames. ], batch size: 134, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:55:50,929 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 04:56:01,791 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6345, 4.7826, 4.9708, 4.7620, 4.7690, 5.3892, 4.9504, 4.6358], device='cuda:6'), covar=tensor([0.1560, 0.2139, 0.2443, 0.2338, 0.3173, 0.1153, 0.1764, 0.2564], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0584, 0.0641, 0.0483, 0.0644, 0.0674, 0.0503, 0.0646], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 04:56:14,236 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 2.016e+02 2.395e+02 3.216e+02 7.414e+02, threshold=4.790e+02, percent-clipped=2.0 2023-05-01 04:56:22,626 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.40 vs. limit=5.0 2023-05-01 04:56:24,919 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=203694.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:56:34,644 INFO [train.py:904] (6/8) Epoch 21, batch 700, loss[loss=0.1611, simple_loss=0.2518, pruned_loss=0.03521, over 17218.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2514, pruned_loss=0.04162, over 3223309.08 frames. ], batch size: 44, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:57:30,237 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=203741.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:57:31,298 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=203742.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:57:38,317 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203747.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:57:44,448 INFO [train.py:904] (6/8) Epoch 21, batch 750, loss[loss=0.1681, simple_loss=0.2634, pruned_loss=0.03637, over 17107.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2519, pruned_loss=0.04196, over 3241744.57 frames. ], batch size: 48, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:58:03,811 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203766.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:58:33,073 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.172e+02 2.611e+02 3.083e+02 3.128e+03, threshold=5.221e+02, percent-clipped=7.0 2023-05-01 04:58:36,275 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=203789.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:58:54,752 INFO [train.py:904] (6/8) Epoch 21, batch 800, loss[loss=0.1501, simple_loss=0.2367, pruned_loss=0.03175, over 16813.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2518, pruned_loss=0.04172, over 3249426.21 frames. ], batch size: 42, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 04:59:03,712 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203808.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:59:24,767 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=203823.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:59:29,165 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.14 vs. limit=5.0 2023-05-01 04:59:30,037 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203827.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:59:58,242 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203847.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:00:04,931 INFO [train.py:904] (6/8) Epoch 21, batch 850, loss[loss=0.1597, simple_loss=0.2465, pruned_loss=0.0365, over 17214.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2512, pruned_loss=0.04163, over 3254381.31 frames. ], batch size: 44, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:00:10,468 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8685, 5.1952, 5.3117, 5.1355, 5.1413, 5.7626, 5.2686, 4.9681], device='cuda:6'), covar=tensor([0.1342, 0.2150, 0.2528, 0.2159, 0.2975, 0.1275, 0.1734, 0.2500], device='cuda:6'), in_proj_covar=tensor([0.0401, 0.0589, 0.0649, 0.0487, 0.0650, 0.0680, 0.0509, 0.0653], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 05:00:29,564 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-05-01 05:00:53,307 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.368e+02 1.964e+02 2.295e+02 2.674e+02 5.967e+02, threshold=4.591e+02, percent-clipped=3.0 2023-05-01 05:01:11,316 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203900.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 05:01:12,977 INFO [train.py:904] (6/8) Epoch 21, batch 900, loss[loss=0.1544, simple_loss=0.2468, pruned_loss=0.03099, over 17158.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2505, pruned_loss=0.04103, over 3271043.90 frames. ], batch size: 46, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:01:21,673 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203908.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 05:01:30,240 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.08 vs. limit=5.0 2023-05-01 05:02:01,431 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5502, 2.3904, 1.9629, 2.2196, 2.7468, 2.5071, 2.6673, 2.7995], device='cuda:6'), covar=tensor([0.0218, 0.0402, 0.0512, 0.0404, 0.0202, 0.0309, 0.0211, 0.0274], device='cuda:6'), in_proj_covar=tensor([0.0206, 0.0238, 0.0227, 0.0227, 0.0237, 0.0235, 0.0238, 0.0233], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:02:07,279 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-05-01 05:02:21,433 INFO [train.py:904] (6/8) Epoch 21, batch 950, loss[loss=0.1577, simple_loss=0.2505, pruned_loss=0.03245, over 17032.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2506, pruned_loss=0.04083, over 3288985.57 frames. ], batch size: 55, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:02:33,965 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4829, 3.4691, 2.1818, 3.6870, 2.7493, 3.6787, 2.2466, 2.7854], device='cuda:6'), covar=tensor([0.0305, 0.0491, 0.1653, 0.0402, 0.0825, 0.0800, 0.1482, 0.0800], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0178, 0.0195, 0.0162, 0.0178, 0.0215, 0.0204, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 05:02:35,161 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203961.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 05:02:40,876 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3520, 3.8135, 3.5786, 1.7700, 2.8924, 2.2506, 3.7189, 3.9885], device='cuda:6'), covar=tensor([0.0310, 0.0832, 0.0745, 0.2591, 0.1138, 0.1279, 0.0729, 0.0946], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0162, 0.0166, 0.0153, 0.0144, 0.0130, 0.0144, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 05:03:10,185 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.138e+02 2.475e+02 2.994e+02 7.321e+02, threshold=4.951e+02, percent-clipped=5.0 2023-05-01 05:03:34,202 INFO [train.py:904] (6/8) Epoch 21, batch 1000, loss[loss=0.1641, simple_loss=0.2414, pruned_loss=0.04339, over 16494.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2499, pruned_loss=0.04069, over 3297952.26 frames. ], batch size: 146, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:03:44,775 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3011, 1.6498, 2.0193, 2.1152, 2.2443, 2.3312, 1.7839, 2.3095], device='cuda:6'), covar=tensor([0.0236, 0.0508, 0.0291, 0.0325, 0.0335, 0.0307, 0.0499, 0.0196], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0192, 0.0178, 0.0182, 0.0194, 0.0152, 0.0195, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:04:10,881 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8758, 1.9547, 2.3778, 2.6639, 2.7280, 2.7014, 1.9282, 2.9250], device='cuda:6'), covar=tensor([0.0176, 0.0501, 0.0346, 0.0314, 0.0298, 0.0306, 0.0542, 0.0172], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0192, 0.0178, 0.0182, 0.0194, 0.0152, 0.0195, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:04:37,398 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2512, 5.1612, 5.0710, 4.5482, 4.6987, 5.0658, 5.0685, 4.6924], device='cuda:6'), covar=tensor([0.0588, 0.0565, 0.0333, 0.0355, 0.1126, 0.0505, 0.0284, 0.0880], device='cuda:6'), in_proj_covar=tensor([0.0293, 0.0419, 0.0339, 0.0335, 0.0348, 0.0390, 0.0233, 0.0408], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:04:43,010 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4629, 3.4387, 2.0429, 3.6103, 2.6859, 3.6003, 2.0615, 2.7315], device='cuda:6'), covar=tensor([0.0253, 0.0404, 0.1670, 0.0380, 0.0789, 0.0722, 0.1576, 0.0753], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0177, 0.0195, 0.0162, 0.0177, 0.0214, 0.0203, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 05:04:43,739 INFO [train.py:904] (6/8) Epoch 21, batch 1050, loss[loss=0.176, simple_loss=0.2515, pruned_loss=0.05024, over 16720.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2493, pruned_loss=0.04017, over 3300344.89 frames. ], batch size: 134, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:05:09,428 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-05-01 05:05:10,433 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8709, 4.4721, 4.4563, 3.2043, 3.6999, 4.3691, 3.9694, 2.5680], device='cuda:6'), covar=tensor([0.0482, 0.0058, 0.0045, 0.0336, 0.0125, 0.0101, 0.0087, 0.0474], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0084, 0.0083, 0.0135, 0.0099, 0.0110, 0.0094, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0004], device='cuda:6') 2023-05-01 05:05:19,347 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=204077.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:05:34,276 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 2.155e+02 2.486e+02 2.944e+02 8.933e+02, threshold=4.972e+02, percent-clipped=2.0 2023-05-01 05:05:54,729 INFO [train.py:904] (6/8) Epoch 21, batch 1100, loss[loss=0.1644, simple_loss=0.2546, pruned_loss=0.03715, over 16755.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2485, pruned_loss=0.03991, over 3306350.60 frames. ], batch size: 62, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:05:56,312 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=204103.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:06:00,784 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6154, 1.7852, 2.1998, 2.3609, 2.5193, 2.4876, 1.8203, 2.6084], device='cuda:6'), covar=tensor([0.0181, 0.0504, 0.0320, 0.0307, 0.0282, 0.0309, 0.0521, 0.0188], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0192, 0.0179, 0.0183, 0.0194, 0.0152, 0.0195, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:06:24,058 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=204122.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:06:25,277 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204123.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:06:45,691 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=204138.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 05:07:05,256 INFO [train.py:904] (6/8) Epoch 21, batch 1150, loss[loss=0.1573, simple_loss=0.2434, pruned_loss=0.03561, over 16831.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2486, pruned_loss=0.03963, over 3316877.94 frames. ], batch size: 42, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:07:21,014 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=204163.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:07:32,274 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204171.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:07:53,760 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.065e+02 2.500e+02 2.935e+02 5.015e+02, threshold=5.000e+02, percent-clipped=1.0 2023-05-01 05:08:07,189 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2497, 5.6161, 5.3756, 5.4156, 5.1071, 5.0958, 5.0946, 5.7149], device='cuda:6'), covar=tensor([0.1357, 0.1044, 0.1085, 0.1004, 0.0888, 0.0891, 0.1229, 0.0975], device='cuda:6'), in_proj_covar=tensor([0.0678, 0.0826, 0.0679, 0.0623, 0.0526, 0.0533, 0.0696, 0.0642], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:08:13,780 INFO [train.py:904] (6/8) Epoch 21, batch 1200, loss[loss=0.1769, simple_loss=0.2515, pruned_loss=0.05117, over 16699.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2478, pruned_loss=0.03906, over 3316120.25 frames. ], batch size: 124, lr: 3.25e-03, grad_scale: 8.0 2023-05-01 05:08:15,241 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=204203.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 05:08:15,371 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9354, 4.6538, 4.9851, 5.1476, 5.3549, 4.6684, 5.3202, 5.3176], device='cuda:6'), covar=tensor([0.1951, 0.1432, 0.1793, 0.0753, 0.0607, 0.0995, 0.0614, 0.0696], device='cuda:6'), in_proj_covar=tensor([0.0649, 0.0794, 0.0926, 0.0815, 0.0610, 0.0634, 0.0657, 0.0757], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:08:29,165 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5984, 4.6489, 4.7922, 4.6405, 4.6878, 5.2484, 4.7341, 4.4503], device='cuda:6'), covar=tensor([0.1580, 0.2203, 0.2563, 0.2380, 0.3035, 0.1111, 0.1807, 0.2433], device='cuda:6'), in_proj_covar=tensor([0.0404, 0.0594, 0.0654, 0.0491, 0.0656, 0.0683, 0.0515, 0.0659], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 05:08:44,943 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=204224.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:09:06,284 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 05:09:24,352 INFO [train.py:904] (6/8) Epoch 21, batch 1250, loss[loss=0.1647, simple_loss=0.2433, pruned_loss=0.04305, over 15596.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2482, pruned_loss=0.03969, over 3306577.90 frames. ], batch size: 191, lr: 3.25e-03, grad_scale: 8.0 2023-05-01 05:09:30,702 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=204256.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 05:10:01,398 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8331, 2.8659, 2.4752, 2.8122, 3.1473, 2.9347, 3.4451, 3.3189], device='cuda:6'), covar=tensor([0.0153, 0.0403, 0.0524, 0.0396, 0.0289, 0.0387, 0.0273, 0.0307], device='cuda:6'), in_proj_covar=tensor([0.0208, 0.0238, 0.0228, 0.0228, 0.0237, 0.0236, 0.0238, 0.0234], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:10:04,709 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9844, 2.0762, 2.4692, 2.8413, 2.7717, 3.1331, 2.1772, 3.1575], device='cuda:6'), covar=tensor([0.0216, 0.0457, 0.0339, 0.0315, 0.0311, 0.0242, 0.0475, 0.0163], device='cuda:6'), in_proj_covar=tensor([0.0186, 0.0194, 0.0180, 0.0184, 0.0196, 0.0153, 0.0196, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:10:12,294 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.450e+02 2.352e+02 2.804e+02 3.486e+02 9.364e+02, threshold=5.607e+02, percent-clipped=5.0 2023-05-01 05:10:21,139 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1238, 2.6127, 2.1370, 2.3553, 2.9690, 2.6817, 2.9962, 3.0807], device='cuda:6'), covar=tensor([0.0239, 0.0408, 0.0534, 0.0443, 0.0258, 0.0371, 0.0256, 0.0284], device='cuda:6'), in_proj_covar=tensor([0.0208, 0.0237, 0.0227, 0.0227, 0.0237, 0.0236, 0.0238, 0.0233], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:10:33,931 INFO [train.py:904] (6/8) Epoch 21, batch 1300, loss[loss=0.1577, simple_loss=0.2512, pruned_loss=0.03215, over 17250.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2479, pruned_loss=0.03955, over 3302697.83 frames. ], batch size: 52, lr: 3.25e-03, grad_scale: 8.0 2023-05-01 05:11:17,960 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4241, 3.7400, 3.9934, 2.2023, 3.2819, 2.4967, 3.9354, 3.8893], device='cuda:6'), covar=tensor([0.0310, 0.0842, 0.0477, 0.1939, 0.0768, 0.0939, 0.0587, 0.0980], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0163, 0.0167, 0.0153, 0.0145, 0.0130, 0.0145, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 05:11:42,899 INFO [train.py:904] (6/8) Epoch 21, batch 1350, loss[loss=0.1878, simple_loss=0.2573, pruned_loss=0.05916, over 16903.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2484, pruned_loss=0.03932, over 3314591.02 frames. ], batch size: 90, lr: 3.25e-03, grad_scale: 8.0 2023-05-01 05:11:48,259 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1953, 4.0483, 4.2647, 4.3865, 4.4841, 4.0588, 4.2227, 4.4648], device='cuda:6'), covar=tensor([0.1660, 0.1090, 0.1286, 0.0695, 0.0601, 0.1267, 0.2618, 0.0783], device='cuda:6'), in_proj_covar=tensor([0.0650, 0.0795, 0.0928, 0.0818, 0.0611, 0.0635, 0.0657, 0.0759], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:12:18,546 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-01 05:12:31,608 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.473e+02 2.188e+02 2.581e+02 2.996e+02 9.160e+02, threshold=5.162e+02, percent-clipped=1.0 2023-05-01 05:12:52,517 INFO [train.py:904] (6/8) Epoch 21, batch 1400, loss[loss=0.142, simple_loss=0.2242, pruned_loss=0.02984, over 15734.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2485, pruned_loss=0.03935, over 3313805.29 frames. ], batch size: 191, lr: 3.25e-03, grad_scale: 8.0 2023-05-01 05:12:54,574 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204403.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:13:20,716 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204422.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:13:35,303 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=204433.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 05:14:00,513 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204451.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:14:01,405 INFO [train.py:904] (6/8) Epoch 21, batch 1450, loss[loss=0.1752, simple_loss=0.2798, pruned_loss=0.03529, over 17028.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2478, pruned_loss=0.03857, over 3311705.88 frames. ], batch size: 50, lr: 3.25e-03, grad_scale: 8.0 2023-05-01 05:14:26,854 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204470.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:14:28,965 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9852, 2.1944, 2.3345, 2.6277, 2.1474, 3.2203, 1.7926, 2.7488], device='cuda:6'), covar=tensor([0.1154, 0.0789, 0.1152, 0.0200, 0.0146, 0.0433, 0.1500, 0.0724], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0175, 0.0195, 0.0189, 0.0205, 0.0216, 0.0202, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 05:14:50,905 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.087e+02 2.507e+02 3.018e+02 6.138e+02, threshold=5.015e+02, percent-clipped=2.0 2023-05-01 05:14:57,741 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6560, 6.0823, 5.7798, 5.8716, 5.4039, 5.5584, 5.4614, 6.1832], device='cuda:6'), covar=tensor([0.1461, 0.1015, 0.1104, 0.0909, 0.1018, 0.0724, 0.1178, 0.0956], device='cuda:6'), in_proj_covar=tensor([0.0681, 0.0831, 0.0680, 0.0627, 0.0529, 0.0535, 0.0699, 0.0644], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:15:10,559 INFO [train.py:904] (6/8) Epoch 21, batch 1500, loss[loss=0.1455, simple_loss=0.2387, pruned_loss=0.02613, over 17230.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2475, pruned_loss=0.03818, over 3319987.63 frames. ], batch size: 44, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:15:12,808 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204503.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:15:33,847 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=204519.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:16:17,197 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204551.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:16:18,101 INFO [train.py:904] (6/8) Epoch 21, batch 1550, loss[loss=0.1499, simple_loss=0.2454, pruned_loss=0.02719, over 17180.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2498, pruned_loss=0.03982, over 3316957.42 frames. ], batch size: 45, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:16:23,646 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204556.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 05:16:26,437 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-05-01 05:17:07,000 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.626e+02 2.391e+02 2.852e+02 3.354e+02 7.624e+02, threshold=5.704e+02, percent-clipped=5.0 2023-05-01 05:17:08,256 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 05:17:26,279 INFO [train.py:904] (6/8) Epoch 21, batch 1600, loss[loss=0.1545, simple_loss=0.2399, pruned_loss=0.03455, over 17025.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2519, pruned_loss=0.04086, over 3308922.62 frames. ], batch size: 41, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:17:29,638 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204604.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 05:17:31,131 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-05-01 05:17:49,613 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-01 05:18:35,817 INFO [train.py:904] (6/8) Epoch 21, batch 1650, loss[loss=0.1912, simple_loss=0.2611, pruned_loss=0.0607, over 16703.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2527, pruned_loss=0.04112, over 3315812.77 frames. ], batch size: 124, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:18:38,340 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.30 vs. limit=5.0 2023-05-01 05:19:25,720 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-01 05:19:25,766 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.742e+02 2.251e+02 2.625e+02 3.143e+02 5.667e+02, threshold=5.251e+02, percent-clipped=0.0 2023-05-01 05:19:45,537 INFO [train.py:904] (6/8) Epoch 21, batch 1700, loss[loss=0.1582, simple_loss=0.2462, pruned_loss=0.03512, over 17200.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2541, pruned_loss=0.04127, over 3322062.80 frames. ], batch size: 44, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:20:30,022 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204733.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 05:20:50,042 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8966, 4.1136, 3.1592, 2.4204, 2.7195, 2.5775, 4.3560, 3.5462], device='cuda:6'), covar=tensor([0.2658, 0.0623, 0.1695, 0.2901, 0.2842, 0.2004, 0.0443, 0.1457], device='cuda:6'), in_proj_covar=tensor([0.0325, 0.0270, 0.0303, 0.0308, 0.0294, 0.0256, 0.0293, 0.0335], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 05:20:55,679 INFO [train.py:904] (6/8) Epoch 21, batch 1750, loss[loss=0.1561, simple_loss=0.2402, pruned_loss=0.03596, over 16709.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2551, pruned_loss=0.04173, over 3326000.20 frames. ], batch size: 39, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:21:26,942 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-01 05:21:37,074 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204781.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:21:46,292 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.620e+02 2.363e+02 2.773e+02 3.345e+02 7.836e+02, threshold=5.546e+02, percent-clipped=6.0 2023-05-01 05:22:06,990 INFO [train.py:904] (6/8) Epoch 21, batch 1800, loss[loss=0.1498, simple_loss=0.239, pruned_loss=0.03032, over 16187.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2554, pruned_loss=0.04131, over 3328641.73 frames. ], batch size: 36, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:22:30,028 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204819.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:22:32,920 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=204821.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:22:34,690 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9897, 5.4712, 5.6255, 5.3361, 5.4089, 6.0254, 5.4598, 5.1460], device='cuda:6'), covar=tensor([0.1088, 0.1990, 0.2244, 0.2192, 0.2703, 0.0926, 0.1606, 0.2489], device='cuda:6'), in_proj_covar=tensor([0.0409, 0.0599, 0.0660, 0.0495, 0.0661, 0.0689, 0.0521, 0.0664], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 05:23:03,718 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3749, 4.6776, 4.4964, 4.5138, 4.2177, 4.1921, 4.2088, 4.7200], device='cuda:6'), covar=tensor([0.1248, 0.0846, 0.0991, 0.0793, 0.0778, 0.1424, 0.1036, 0.0902], device='cuda:6'), in_proj_covar=tensor([0.0678, 0.0826, 0.0679, 0.0624, 0.0526, 0.0531, 0.0697, 0.0640], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:23:15,229 INFO [train.py:904] (6/8) Epoch 21, batch 1850, loss[loss=0.1587, simple_loss=0.2548, pruned_loss=0.03126, over 17240.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2564, pruned_loss=0.04156, over 3329985.75 frames. ], batch size: 52, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:23:37,466 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204867.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:23:58,555 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=204882.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:24:06,222 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 2.082e+02 2.497e+02 3.026e+02 6.100e+02, threshold=4.995e+02, percent-clipped=2.0 2023-05-01 05:24:26,173 INFO [train.py:904] (6/8) Epoch 21, batch 1900, loss[loss=0.1529, simple_loss=0.2329, pruned_loss=0.03646, over 17020.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2551, pruned_loss=0.04076, over 3327977.49 frames. ], batch size: 41, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:25:35,962 INFO [train.py:904] (6/8) Epoch 21, batch 1950, loss[loss=0.1664, simple_loss=0.2673, pruned_loss=0.03276, over 17035.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2555, pruned_loss=0.04075, over 3316459.62 frames. ], batch size: 50, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:26:26,412 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.621e+02 2.195e+02 2.614e+02 3.172e+02 4.787e+02, threshold=5.228e+02, percent-clipped=0.0 2023-05-01 05:26:36,817 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9345, 3.9844, 2.6294, 4.6594, 3.1563, 4.5850, 2.6687, 3.2769], device='cuda:6'), covar=tensor([0.0302, 0.0409, 0.1428, 0.0248, 0.0773, 0.0524, 0.1447, 0.0699], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0179, 0.0196, 0.0166, 0.0178, 0.0219, 0.0205, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 05:26:44,684 INFO [train.py:904] (6/8) Epoch 21, batch 2000, loss[loss=0.1953, simple_loss=0.286, pruned_loss=0.05231, over 17091.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2555, pruned_loss=0.04077, over 3316939.42 frames. ], batch size: 55, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:27:55,522 INFO [train.py:904] (6/8) Epoch 21, batch 2050, loss[loss=0.1665, simple_loss=0.2667, pruned_loss=0.03312, over 16640.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.256, pruned_loss=0.04127, over 3298028.54 frames. ], batch size: 62, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:28:18,892 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9221, 4.7425, 4.9531, 5.1553, 5.3534, 4.6968, 5.3176, 5.3413], device='cuda:6'), covar=tensor([0.1852, 0.1270, 0.1784, 0.0777, 0.0578, 0.1008, 0.0554, 0.0570], device='cuda:6'), in_proj_covar=tensor([0.0656, 0.0804, 0.0944, 0.0828, 0.0617, 0.0647, 0.0667, 0.0770], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:28:44,309 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.164e+02 2.478e+02 3.098e+02 5.896e+02, threshold=4.956e+02, percent-clipped=1.0 2023-05-01 05:28:50,353 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-01 05:29:04,143 INFO [train.py:904] (6/8) Epoch 21, batch 2100, loss[loss=0.1678, simple_loss=0.2493, pruned_loss=0.04312, over 16839.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2567, pruned_loss=0.0415, over 3293064.28 frames. ], batch size: 96, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:29:14,607 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=205109.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:30:14,923 INFO [train.py:904] (6/8) Epoch 21, batch 2150, loss[loss=0.2504, simple_loss=0.3214, pruned_loss=0.0897, over 12008.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2582, pruned_loss=0.04224, over 3285002.45 frames. ], batch size: 248, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:30:27,008 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7600, 2.4027, 2.3573, 3.3305, 2.7063, 3.5836, 1.6321, 2.7627], device='cuda:6'), covar=tensor([0.1318, 0.0743, 0.1173, 0.0229, 0.0166, 0.0440, 0.1538, 0.0802], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0175, 0.0194, 0.0190, 0.0205, 0.0216, 0.0201, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 05:30:39,946 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=205170.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 05:30:50,421 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=205177.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:31:04,863 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.289e+02 2.743e+02 3.268e+02 8.955e+02, threshold=5.486e+02, percent-clipped=2.0 2023-05-01 05:31:19,712 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=205198.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:31:24,886 INFO [train.py:904] (6/8) Epoch 21, batch 2200, loss[loss=0.1661, simple_loss=0.253, pruned_loss=0.03961, over 16813.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2578, pruned_loss=0.04211, over 3303554.31 frames. ], batch size: 102, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:32:22,479 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-05-01 05:32:34,214 INFO [train.py:904] (6/8) Epoch 21, batch 2250, loss[loss=0.1595, simple_loss=0.2467, pruned_loss=0.03618, over 16790.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2574, pruned_loss=0.04172, over 3313933.33 frames. ], batch size: 102, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:32:45,037 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=205259.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:33:23,477 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.631e+02 2.149e+02 2.507e+02 2.983e+02 5.754e+02, threshold=5.014e+02, percent-clipped=1.0 2023-05-01 05:33:44,278 INFO [train.py:904] (6/8) Epoch 21, batch 2300, loss[loss=0.1733, simple_loss=0.2714, pruned_loss=0.03759, over 17133.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2575, pruned_loss=0.04187, over 3311326.41 frames. ], batch size: 48, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:34:44,464 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-05-01 05:34:53,180 INFO [train.py:904] (6/8) Epoch 21, batch 2350, loss[loss=0.1903, simple_loss=0.2926, pruned_loss=0.044, over 17121.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2582, pruned_loss=0.0427, over 3315345.69 frames. ], batch size: 49, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:34:58,768 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3677, 4.3963, 4.7637, 4.7498, 4.7992, 4.4534, 4.4820, 4.2715], device='cuda:6'), covar=tensor([0.0411, 0.0748, 0.0459, 0.0441, 0.0511, 0.0510, 0.0853, 0.0767], device='cuda:6'), in_proj_covar=tensor([0.0409, 0.0451, 0.0439, 0.0408, 0.0484, 0.0460, 0.0547, 0.0369], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 05:35:42,800 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 2.163e+02 2.457e+02 2.980e+02 4.846e+02, threshold=4.914e+02, percent-clipped=0.0 2023-05-01 05:36:02,959 INFO [train.py:904] (6/8) Epoch 21, batch 2400, loss[loss=0.1634, simple_loss=0.2649, pruned_loss=0.03101, over 17125.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2588, pruned_loss=0.04297, over 3317449.15 frames. ], batch size: 48, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:36:06,368 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4905, 4.3042, 4.5362, 4.6752, 4.8069, 4.3625, 4.6974, 4.7670], device='cuda:6'), covar=tensor([0.1730, 0.1325, 0.1538, 0.0797, 0.0631, 0.1221, 0.1771, 0.0910], device='cuda:6'), in_proj_covar=tensor([0.0659, 0.0810, 0.0949, 0.0832, 0.0618, 0.0651, 0.0670, 0.0776], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:37:10,842 INFO [train.py:904] (6/8) Epoch 21, batch 2450, loss[loss=0.1824, simple_loss=0.2768, pruned_loss=0.04394, over 16697.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2601, pruned_loss=0.04299, over 3318436.05 frames. ], batch size: 57, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:37:29,569 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=205465.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 05:37:45,995 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=205477.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:38:00,842 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 2.346e+02 2.648e+02 3.284e+02 5.369e+02, threshold=5.296e+02, percent-clipped=1.0 2023-05-01 05:38:10,488 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5951, 3.6074, 2.7743, 2.1399, 2.3391, 2.2287, 3.7359, 3.2680], device='cuda:6'), covar=tensor([0.2760, 0.0620, 0.1739, 0.2971, 0.2760, 0.2153, 0.0547, 0.1385], device='cuda:6'), in_proj_covar=tensor([0.0328, 0.0272, 0.0305, 0.0310, 0.0298, 0.0258, 0.0296, 0.0338], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 05:38:21,260 INFO [train.py:904] (6/8) Epoch 21, batch 2500, loss[loss=0.1864, simple_loss=0.2676, pruned_loss=0.05258, over 16839.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.26, pruned_loss=0.0426, over 3319534.53 frames. ], batch size: 96, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:38:53,421 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=205525.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:39:30,352 INFO [train.py:904] (6/8) Epoch 21, batch 2550, loss[loss=0.1529, simple_loss=0.2373, pruned_loss=0.03425, over 16946.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.26, pruned_loss=0.04254, over 3306639.74 frames. ], batch size: 41, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:39:33,578 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=205554.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:40:19,333 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.328e+02 2.679e+02 3.297e+02 1.189e+03, threshold=5.358e+02, percent-clipped=3.0 2023-05-01 05:40:38,679 INFO [train.py:904] (6/8) Epoch 21, batch 2600, loss[loss=0.1819, simple_loss=0.2864, pruned_loss=0.03869, over 17029.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2603, pruned_loss=0.04305, over 3299709.12 frames. ], batch size: 50, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:40:39,303 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7149, 2.4204, 2.4412, 4.5750, 2.3891, 2.8544, 2.4990, 2.5912], device='cuda:6'), covar=tensor([0.1216, 0.3746, 0.3055, 0.0449, 0.4112, 0.2676, 0.3631, 0.3676], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0448, 0.0370, 0.0330, 0.0436, 0.0515, 0.0417, 0.0525], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:41:49,800 INFO [train.py:904] (6/8) Epoch 21, batch 2650, loss[loss=0.1881, simple_loss=0.2702, pruned_loss=0.05304, over 16691.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2605, pruned_loss=0.04269, over 3303770.24 frames. ], batch size: 134, lr: 3.24e-03, grad_scale: 4.0 2023-05-01 05:41:54,624 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=205655.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:42:40,190 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.048e+02 2.349e+02 2.910e+02 5.870e+02, threshold=4.699e+02, percent-clipped=1.0 2023-05-01 05:43:00,011 INFO [train.py:904] (6/8) Epoch 21, batch 2700, loss[loss=0.1905, simple_loss=0.2658, pruned_loss=0.05759, over 16910.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2605, pruned_loss=0.04228, over 3307978.42 frames. ], batch size: 109, lr: 3.24e-03, grad_scale: 4.0 2023-05-01 05:43:18,694 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=205716.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:44:00,225 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=205744.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:44:09,872 INFO [train.py:904] (6/8) Epoch 21, batch 2750, loss[loss=0.1574, simple_loss=0.2532, pruned_loss=0.03084, over 16702.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2609, pruned_loss=0.04169, over 3310848.87 frames. ], batch size: 76, lr: 3.24e-03, grad_scale: 4.0 2023-05-01 05:44:13,734 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8055, 3.9114, 3.0646, 2.3589, 2.6102, 2.5218, 4.1402, 3.4379], device='cuda:6'), covar=tensor([0.2595, 0.0620, 0.1685, 0.2876, 0.2635, 0.1918, 0.0516, 0.1377], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0270, 0.0304, 0.0309, 0.0297, 0.0257, 0.0294, 0.0337], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 05:44:29,207 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=205765.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:45:02,029 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.167e+02 2.431e+02 2.778e+02 3.742e+02, threshold=4.863e+02, percent-clipped=0.0 2023-05-01 05:45:19,234 INFO [train.py:904] (6/8) Epoch 21, batch 2800, loss[loss=0.163, simple_loss=0.2528, pruned_loss=0.03661, over 16066.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2602, pruned_loss=0.04106, over 3319841.67 frames. ], batch size: 35, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:45:22,646 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1759, 5.1455, 4.9057, 4.4112, 5.0125, 1.7271, 4.7270, 4.7988], device='cuda:6'), covar=tensor([0.0084, 0.0079, 0.0215, 0.0365, 0.0090, 0.2972, 0.0138, 0.0203], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0157, 0.0199, 0.0180, 0.0178, 0.0210, 0.0190, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:45:24,431 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=205805.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:45:35,897 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=205813.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:46:29,085 INFO [train.py:904] (6/8) Epoch 21, batch 2850, loss[loss=0.1334, simple_loss=0.2253, pruned_loss=0.02078, over 17196.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2588, pruned_loss=0.04015, over 3320570.77 frames. ], batch size: 44, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:46:31,699 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=205854.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:47:17,820 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.397e+02 2.169e+02 2.532e+02 3.008e+02 5.585e+02, threshold=5.064e+02, percent-clipped=2.0 2023-05-01 05:47:19,309 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=205890.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:47:19,889 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.52 vs. limit=5.0 2023-05-01 05:47:36,338 INFO [train.py:904] (6/8) Epoch 21, batch 2900, loss[loss=0.1669, simple_loss=0.2434, pruned_loss=0.04521, over 16470.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2578, pruned_loss=0.04051, over 3329078.09 frames. ], batch size: 146, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:47:36,636 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=205902.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:48:44,795 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=205951.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:48:45,569 INFO [train.py:904] (6/8) Epoch 21, batch 2950, loss[loss=0.1725, simple_loss=0.2492, pruned_loss=0.0479, over 16929.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2569, pruned_loss=0.04101, over 3326867.21 frames. ], batch size: 96, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:49:22,958 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8542, 2.9564, 2.7785, 4.6262, 3.8146, 4.2817, 1.7544, 3.0848], device='cuda:6'), covar=tensor([0.1296, 0.0669, 0.1052, 0.0193, 0.0208, 0.0384, 0.1482, 0.0804], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0175, 0.0194, 0.0192, 0.0207, 0.0217, 0.0202, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 05:49:36,947 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.251e+02 2.659e+02 3.342e+02 9.466e+02, threshold=5.318e+02, percent-clipped=4.0 2023-05-01 05:49:58,057 INFO [train.py:904] (6/8) Epoch 21, batch 3000, loss[loss=0.1898, simple_loss=0.2717, pruned_loss=0.05397, over 16544.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2581, pruned_loss=0.04209, over 3332720.70 frames. ], batch size: 75, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:49:58,057 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 05:50:06,474 INFO [train.py:938] (6/8) Epoch 21, validation: loss=0.1356, simple_loss=0.2408, pruned_loss=0.01521, over 944034.00 frames. 2023-05-01 05:50:06,475 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 05:50:07,382 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-05-01 05:50:18,442 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=206011.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:51:14,522 INFO [train.py:904] (6/8) Epoch 21, batch 3050, loss[loss=0.1518, simple_loss=0.2429, pruned_loss=0.03036, over 17220.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2581, pruned_loss=0.04245, over 3321096.98 frames. ], batch size: 44, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:51:21,071 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=206056.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:51:43,736 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8481, 1.8792, 2.4351, 2.8086, 2.6842, 3.2352, 2.1317, 3.2208], device='cuda:6'), covar=tensor([0.0263, 0.0583, 0.0353, 0.0322, 0.0355, 0.0195, 0.0526, 0.0169], device='cuda:6'), in_proj_covar=tensor([0.0189, 0.0195, 0.0181, 0.0186, 0.0197, 0.0155, 0.0197, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 05:51:55,891 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 05:52:05,512 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 2.000e+02 2.421e+02 3.043e+02 5.222e+02, threshold=4.843e+02, percent-clipped=0.0 2023-05-01 05:52:22,007 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=206100.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:52:24,073 INFO [train.py:904] (6/8) Epoch 21, batch 3100, loss[loss=0.1812, simple_loss=0.2541, pruned_loss=0.05416, over 16749.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2575, pruned_loss=0.0426, over 3325974.29 frames. ], batch size: 124, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:52:30,853 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3811, 4.1214, 4.4928, 2.2624, 4.7493, 4.6900, 3.5288, 3.7231], device='cuda:6'), covar=tensor([0.0603, 0.0219, 0.0231, 0.1095, 0.0072, 0.0191, 0.0369, 0.0364], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0109, 0.0098, 0.0139, 0.0080, 0.0125, 0.0129, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 05:52:43,757 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=206117.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:53:30,989 INFO [train.py:904] (6/8) Epoch 21, batch 3150, loss[loss=0.1559, simple_loss=0.2493, pruned_loss=0.03122, over 17247.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2571, pruned_loss=0.0425, over 3334820.28 frames. ], batch size: 52, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:54:22,972 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.110e+02 2.360e+02 2.759e+02 5.694e+02, threshold=4.721e+02, percent-clipped=1.0 2023-05-01 05:54:35,058 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=206197.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:54:41,550 INFO [train.py:904] (6/8) Epoch 21, batch 3200, loss[loss=0.1446, simple_loss=0.2376, pruned_loss=0.02584, over 17227.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2562, pruned_loss=0.04211, over 3331764.15 frames. ], batch size: 45, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:55:43,201 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=206246.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:55:51,280 INFO [train.py:904] (6/8) Epoch 21, batch 3250, loss[loss=0.1859, simple_loss=0.2689, pruned_loss=0.05143, over 16822.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2563, pruned_loss=0.04202, over 3336538.17 frames. ], batch size: 102, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:55:59,461 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=206258.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:56:42,634 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.332e+02 2.138e+02 2.458e+02 2.946e+02 5.794e+02, threshold=4.917e+02, percent-clipped=2.0 2023-05-01 05:57:00,584 INFO [train.py:904] (6/8) Epoch 21, batch 3300, loss[loss=0.2013, simple_loss=0.2803, pruned_loss=0.06111, over 15404.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2568, pruned_loss=0.04218, over 3326394.70 frames. ], batch size: 190, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:57:13,899 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=206311.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:57:41,259 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4470, 5.4078, 5.2828, 4.7509, 4.8760, 5.3693, 5.2404, 4.9370], device='cuda:6'), covar=tensor([0.0601, 0.0430, 0.0299, 0.0336, 0.1074, 0.0451, 0.0292, 0.0752], device='cuda:6'), in_proj_covar=tensor([0.0307, 0.0440, 0.0357, 0.0352, 0.0365, 0.0408, 0.0245, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 05:58:09,889 INFO [train.py:904] (6/8) Epoch 21, batch 3350, loss[loss=0.1724, simple_loss=0.2495, pruned_loss=0.0476, over 16771.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2569, pruned_loss=0.0418, over 3327641.04 frames. ], batch size: 83, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:58:20,166 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=206359.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:58:39,133 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2603, 3.3706, 3.5232, 1.8324, 3.6783, 3.7595, 2.9616, 2.6524], device='cuda:6'), covar=tensor([0.1027, 0.0196, 0.0201, 0.1267, 0.0117, 0.0166, 0.0444, 0.0554], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0109, 0.0098, 0.0140, 0.0081, 0.0126, 0.0129, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 05:59:00,090 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.620e+02 2.208e+02 2.654e+02 3.136e+02 4.728e+02, threshold=5.309e+02, percent-clipped=0.0 2023-05-01 05:59:16,460 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=206400.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:59:18,363 INFO [train.py:904] (6/8) Epoch 21, batch 3400, loss[loss=0.1906, simple_loss=0.2621, pruned_loss=0.05949, over 16671.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.257, pruned_loss=0.04202, over 3327723.27 frames. ], batch size: 89, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:59:32,347 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=206412.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:59:44,282 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8645, 4.6771, 4.9183, 5.1389, 5.3345, 4.7543, 5.3125, 5.3481], device='cuda:6'), covar=tensor([0.2138, 0.1468, 0.2191, 0.0955, 0.0749, 0.0962, 0.0658, 0.0689], device='cuda:6'), in_proj_covar=tensor([0.0665, 0.0815, 0.0957, 0.0842, 0.0624, 0.0653, 0.0673, 0.0779], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:00:21,657 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=206448.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:00:26,881 INFO [train.py:904] (6/8) Epoch 21, batch 3450, loss[loss=0.1903, simple_loss=0.2648, pruned_loss=0.0579, over 16660.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.256, pruned_loss=0.04137, over 3316163.12 frames. ], batch size: 134, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:01:17,175 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.080e+02 2.437e+02 2.944e+02 7.361e+02, threshold=4.875e+02, percent-clipped=3.0 2023-05-01 06:01:36,849 INFO [train.py:904] (6/8) Epoch 21, batch 3500, loss[loss=0.1734, simple_loss=0.2556, pruned_loss=0.04563, over 16856.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2543, pruned_loss=0.04094, over 3300107.16 frames. ], batch size: 96, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:02:37,899 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=206546.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:02:44,997 INFO [train.py:904] (6/8) Epoch 21, batch 3550, loss[loss=0.1445, simple_loss=0.2355, pruned_loss=0.02679, over 16774.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2527, pruned_loss=0.04032, over 3311600.69 frames. ], batch size: 39, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:02:47,121 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=206553.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:03:37,043 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.127e+02 2.401e+02 2.794e+02 4.680e+02, threshold=4.802e+02, percent-clipped=0.0 2023-05-01 06:03:44,325 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=206594.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:03:55,931 INFO [train.py:904] (6/8) Epoch 21, batch 3600, loss[loss=0.1471, simple_loss=0.2275, pruned_loss=0.03338, over 16749.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2517, pruned_loss=0.03991, over 3310004.88 frames. ], batch size: 89, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:04:04,943 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7908, 2.4964, 2.4263, 3.3057, 2.7580, 3.6292, 1.6502, 2.7681], device='cuda:6'), covar=tensor([0.1308, 0.0749, 0.1133, 0.0225, 0.0188, 0.0383, 0.1558, 0.0819], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0175, 0.0194, 0.0192, 0.0207, 0.0217, 0.0201, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:05:06,985 INFO [train.py:904] (6/8) Epoch 21, batch 3650, loss[loss=0.1531, simple_loss=0.2276, pruned_loss=0.03925, over 16405.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2511, pruned_loss=0.04023, over 3310013.21 frames. ], batch size: 146, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:05:59,786 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.607e+02 2.224e+02 2.566e+02 3.274e+02 6.534e+02, threshold=5.132e+02, percent-clipped=5.0 2023-05-01 06:06:18,975 INFO [train.py:904] (6/8) Epoch 21, batch 3700, loss[loss=0.1618, simple_loss=0.2438, pruned_loss=0.03988, over 16537.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2498, pruned_loss=0.04121, over 3306618.77 frames. ], batch size: 146, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:06:34,617 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=206712.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:06:37,724 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6225, 4.7915, 4.9271, 4.7641, 4.7815, 5.3802, 4.9066, 4.5529], device='cuda:6'), covar=tensor([0.1605, 0.2021, 0.2113, 0.2066, 0.2686, 0.1074, 0.1622, 0.2622], device='cuda:6'), in_proj_covar=tensor([0.0416, 0.0612, 0.0669, 0.0508, 0.0674, 0.0703, 0.0525, 0.0674], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 06:07:32,564 INFO [train.py:904] (6/8) Epoch 21, batch 3750, loss[loss=0.1794, simple_loss=0.272, pruned_loss=0.04345, over 16506.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2506, pruned_loss=0.04304, over 3276962.16 frames. ], batch size: 62, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:07:45,684 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=206760.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:08:26,978 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.616e+02 2.267e+02 2.726e+02 3.357e+02 9.074e+02, threshold=5.452e+02, percent-clipped=3.0 2023-05-01 06:08:45,459 INFO [train.py:904] (6/8) Epoch 21, batch 3800, loss[loss=0.1843, simple_loss=0.2681, pruned_loss=0.05028, over 16565.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.252, pruned_loss=0.04439, over 3273234.04 frames. ], batch size: 62, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:09:32,785 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=206834.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:09:41,169 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.59 vs. limit=5.0 2023-05-01 06:09:58,313 INFO [train.py:904] (6/8) Epoch 21, batch 3850, loss[loss=0.2017, simple_loss=0.2789, pruned_loss=0.06224, over 12458.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2531, pruned_loss=0.04549, over 3260691.01 frames. ], batch size: 246, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:10:00,773 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=206853.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:10:52,982 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.673e+02 2.267e+02 2.812e+02 3.321e+02 1.032e+03, threshold=5.623e+02, percent-clipped=3.0 2023-05-01 06:11:00,968 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=206895.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 06:11:09,938 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=206901.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:11:10,712 INFO [train.py:904] (6/8) Epoch 21, batch 3900, loss[loss=0.1775, simple_loss=0.2557, pruned_loss=0.04972, over 15518.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2525, pruned_loss=0.04608, over 3273218.76 frames. ], batch size: 190, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:11:21,382 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-01 06:11:54,872 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0127, 2.1908, 2.6992, 2.9838, 2.9463, 3.3109, 2.0521, 3.3446], device='cuda:6'), covar=tensor([0.0210, 0.0474, 0.0269, 0.0302, 0.0268, 0.0182, 0.0542, 0.0109], device='cuda:6'), in_proj_covar=tensor([0.0189, 0.0194, 0.0180, 0.0186, 0.0198, 0.0156, 0.0197, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:12:24,775 INFO [train.py:904] (6/8) Epoch 21, batch 3950, loss[loss=0.1832, simple_loss=0.2769, pruned_loss=0.04477, over 17106.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2525, pruned_loss=0.04666, over 3280755.38 frames. ], batch size: 47, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:12:41,668 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8511, 4.8514, 4.5876, 3.2148, 4.0868, 4.6597, 3.9648, 2.5747], device='cuda:6'), covar=tensor([0.0548, 0.0028, 0.0040, 0.0370, 0.0086, 0.0105, 0.0102, 0.0464], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0084, 0.0083, 0.0134, 0.0098, 0.0110, 0.0095, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 06:12:46,558 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-05-01 06:13:16,329 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.684e+02 2.243e+02 2.750e+02 3.370e+02 5.125e+02, threshold=5.500e+02, percent-clipped=0.0 2023-05-01 06:13:35,019 INFO [train.py:904] (6/8) Epoch 21, batch 4000, loss[loss=0.1661, simple_loss=0.2432, pruned_loss=0.04449, over 16870.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2532, pruned_loss=0.04738, over 3273207.70 frames. ], batch size: 116, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:14:45,528 INFO [train.py:904] (6/8) Epoch 21, batch 4050, loss[loss=0.1909, simple_loss=0.2702, pruned_loss=0.05582, over 12211.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2543, pruned_loss=0.04672, over 3274550.34 frames. ], batch size: 247, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:15:37,537 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.164e+02 1.811e+02 2.022e+02 2.362e+02 4.277e+02, threshold=4.045e+02, percent-clipped=0.0 2023-05-01 06:15:55,998 INFO [train.py:904] (6/8) Epoch 21, batch 4100, loss[loss=0.1776, simple_loss=0.27, pruned_loss=0.04265, over 16723.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2552, pruned_loss=0.04554, over 3267895.76 frames. ], batch size: 124, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:16:39,041 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-05-01 06:17:10,042 INFO [train.py:904] (6/8) Epoch 21, batch 4150, loss[loss=0.1974, simple_loss=0.293, pruned_loss=0.05093, over 16606.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2617, pruned_loss=0.04807, over 3226301.30 frames. ], batch size: 62, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:17:45,661 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=207176.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:17:51,423 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5722, 3.8078, 2.8040, 2.2608, 2.5391, 2.6103, 4.0337, 3.2995], device='cuda:6'), covar=tensor([0.2897, 0.0645, 0.1831, 0.2607, 0.2792, 0.1856, 0.0495, 0.1338], device='cuda:6'), in_proj_covar=tensor([0.0328, 0.0272, 0.0307, 0.0314, 0.0301, 0.0260, 0.0298, 0.0341], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 06:18:04,514 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.144e+02 2.646e+02 3.287e+02 6.191e+02, threshold=5.292e+02, percent-clipped=8.0 2023-05-01 06:18:06,772 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=207190.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 06:18:24,675 INFO [train.py:904] (6/8) Epoch 21, batch 4200, loss[loss=0.2442, simple_loss=0.3135, pruned_loss=0.08745, over 11363.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2679, pruned_loss=0.04945, over 3185695.86 frames. ], batch size: 246, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:19:18,469 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=207237.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:19:40,048 INFO [train.py:904] (6/8) Epoch 21, batch 4250, loss[loss=0.1912, simple_loss=0.2912, pruned_loss=0.04562, over 16807.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2712, pruned_loss=0.04927, over 3175380.14 frames. ], batch size: 102, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:20:31,170 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-01 06:20:35,820 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.204e+02 2.470e+02 3.047e+02 4.735e+02, threshold=4.941e+02, percent-clipped=0.0 2023-05-01 06:20:55,788 INFO [train.py:904] (6/8) Epoch 21, batch 4300, loss[loss=0.2031, simple_loss=0.2914, pruned_loss=0.05738, over 16635.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2731, pruned_loss=0.04859, over 3174442.09 frames. ], batch size: 62, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:22:07,506 INFO [train.py:904] (6/8) Epoch 21, batch 4350, loss[loss=0.1856, simple_loss=0.2811, pruned_loss=0.04502, over 16698.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2764, pruned_loss=0.04962, over 3176358.82 frames. ], batch size: 89, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:22:21,093 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2023-05-01 06:23:02,764 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.219e+02 2.598e+02 3.036e+02 7.307e+02, threshold=5.195e+02, percent-clipped=1.0 2023-05-01 06:23:22,045 INFO [train.py:904] (6/8) Epoch 21, batch 4400, loss[loss=0.1945, simple_loss=0.2736, pruned_loss=0.05767, over 16642.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2788, pruned_loss=0.05092, over 3175307.56 frames. ], batch size: 57, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:24:37,205 INFO [train.py:904] (6/8) Epoch 21, batch 4450, loss[loss=0.2151, simple_loss=0.3066, pruned_loss=0.0618, over 16303.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2824, pruned_loss=0.05211, over 3203350.05 frames. ], batch size: 165, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:24:40,525 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.8289, 6.1731, 5.8728, 6.0540, 5.6059, 5.3721, 5.7289, 6.3259], device='cuda:6'), covar=tensor([0.1275, 0.0756, 0.0876, 0.0727, 0.0717, 0.0649, 0.0981, 0.0695], device='cuda:6'), in_proj_covar=tensor([0.0669, 0.0817, 0.0676, 0.0620, 0.0519, 0.0527, 0.0687, 0.0638], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:25:16,061 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9963, 1.9563, 2.5550, 2.9067, 2.8381, 3.3112, 2.0598, 3.2962], device='cuda:6'), covar=tensor([0.0204, 0.0551, 0.0308, 0.0305, 0.0287, 0.0177, 0.0575, 0.0147], device='cuda:6'), in_proj_covar=tensor([0.0188, 0.0193, 0.0178, 0.0185, 0.0196, 0.0153, 0.0197, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:25:31,511 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.026e+02 2.353e+02 2.743e+02 3.990e+02, threshold=4.706e+02, percent-clipped=0.0 2023-05-01 06:25:33,775 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=207490.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 06:25:50,489 INFO [train.py:904] (6/8) Epoch 21, batch 4500, loss[loss=0.1829, simple_loss=0.2761, pruned_loss=0.04483, over 16748.00 frames. ], tot_loss[loss=0.194, simple_loss=0.2828, pruned_loss=0.0526, over 3215325.81 frames. ], batch size: 89, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:26:35,687 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=207532.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:26:43,762 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=207538.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:27:03,812 INFO [train.py:904] (6/8) Epoch 21, batch 4550, loss[loss=0.1976, simple_loss=0.2788, pruned_loss=0.05817, over 16593.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.2833, pruned_loss=0.05353, over 3217280.17 frames. ], batch size: 62, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:27:28,713 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8512, 4.8392, 4.5674, 3.7622, 4.7207, 1.7331, 4.4705, 4.2037], device='cuda:6'), covar=tensor([0.0052, 0.0046, 0.0139, 0.0270, 0.0057, 0.2973, 0.0094, 0.0211], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0155, 0.0197, 0.0178, 0.0175, 0.0207, 0.0187, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:27:57,547 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.324e+02 1.874e+02 2.187e+02 2.648e+02 6.639e+02, threshold=4.374e+02, percent-clipped=1.0 2023-05-01 06:28:16,259 INFO [train.py:904] (6/8) Epoch 21, batch 4600, loss[loss=0.1935, simple_loss=0.2843, pruned_loss=0.05129, over 16409.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2846, pruned_loss=0.05414, over 3207059.64 frames. ], batch size: 146, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:28:39,187 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7223, 2.8659, 2.4979, 4.2816, 3.2674, 3.9777, 1.6352, 3.0136], device='cuda:6'), covar=tensor([0.1304, 0.0752, 0.1212, 0.0121, 0.0249, 0.0342, 0.1673, 0.0739], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0175, 0.0194, 0.0189, 0.0207, 0.0214, 0.0200, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:29:29,137 INFO [train.py:904] (6/8) Epoch 21, batch 4650, loss[loss=0.173, simple_loss=0.2591, pruned_loss=0.04346, over 17218.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2837, pruned_loss=0.05405, over 3213903.42 frames. ], batch size: 44, lr: 3.22e-03, grad_scale: 16.0 2023-05-01 06:30:23,479 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 1.925e+02 2.221e+02 2.572e+02 4.933e+02, threshold=4.441e+02, percent-clipped=1.0 2023-05-01 06:30:26,904 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7092, 4.9477, 5.1314, 4.8794, 4.9731, 5.5311, 5.0137, 4.7627], device='cuda:6'), covar=tensor([0.1024, 0.1670, 0.1800, 0.1977, 0.2333, 0.0815, 0.1291, 0.2286], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0583, 0.0634, 0.0486, 0.0646, 0.0671, 0.0498, 0.0648], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 06:30:42,426 INFO [train.py:904] (6/8) Epoch 21, batch 4700, loss[loss=0.1801, simple_loss=0.2644, pruned_loss=0.04787, over 17242.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2801, pruned_loss=0.05252, over 3221287.31 frames. ], batch size: 52, lr: 3.22e-03, grad_scale: 16.0 2023-05-01 06:30:52,036 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3014, 3.4147, 3.5710, 1.9650, 2.9256, 2.3129, 3.5479, 3.6628], device='cuda:6'), covar=tensor([0.0235, 0.0853, 0.0603, 0.2214, 0.0934, 0.1004, 0.0624, 0.0976], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0164, 0.0166, 0.0151, 0.0144, 0.0129, 0.0142, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:31:26,271 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5967, 2.4400, 2.1936, 3.2765, 2.1084, 3.4976, 1.4720, 2.6017], device='cuda:6'), covar=tensor([0.1595, 0.0897, 0.1419, 0.0198, 0.0163, 0.0413, 0.2039, 0.0953], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0175, 0.0193, 0.0189, 0.0207, 0.0214, 0.0200, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:31:49,510 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3561, 2.5559, 2.2083, 2.2449, 2.8036, 2.4227, 2.8212, 2.9617], device='cuda:6'), covar=tensor([0.0113, 0.0419, 0.0543, 0.0514, 0.0308, 0.0465, 0.0229, 0.0318], device='cuda:6'), in_proj_covar=tensor([0.0208, 0.0234, 0.0225, 0.0226, 0.0236, 0.0234, 0.0237, 0.0233], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:31:56,744 INFO [train.py:904] (6/8) Epoch 21, batch 4750, loss[loss=0.161, simple_loss=0.2492, pruned_loss=0.0364, over 16458.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.275, pruned_loss=0.05002, over 3234178.89 frames. ], batch size: 75, lr: 3.22e-03, grad_scale: 16.0 2023-05-01 06:32:28,201 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0808, 1.5865, 1.9205, 2.0717, 2.2350, 2.3202, 1.6458, 2.1959], device='cuda:6'), covar=tensor([0.0249, 0.0482, 0.0292, 0.0359, 0.0322, 0.0231, 0.0586, 0.0160], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0193, 0.0178, 0.0185, 0.0195, 0.0153, 0.0197, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:32:50,130 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 1.797e+02 2.146e+02 2.444e+02 4.649e+02, threshold=4.292e+02, percent-clipped=1.0 2023-05-01 06:33:11,157 INFO [train.py:904] (6/8) Epoch 21, batch 4800, loss[loss=0.1789, simple_loss=0.2777, pruned_loss=0.04002, over 15402.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2717, pruned_loss=0.04791, over 3229377.72 frames. ], batch size: 190, lr: 3.22e-03, grad_scale: 16.0 2023-05-01 06:33:31,287 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.17 vs. limit=5.0 2023-05-01 06:33:55,477 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=207832.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:34:01,016 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8083, 5.0678, 4.8403, 4.9017, 4.6460, 4.5493, 4.4454, 5.1366], device='cuda:6'), covar=tensor([0.1123, 0.0775, 0.0923, 0.0778, 0.0738, 0.1042, 0.1162, 0.0785], device='cuda:6'), in_proj_covar=tensor([0.0661, 0.0805, 0.0670, 0.0612, 0.0513, 0.0520, 0.0678, 0.0630], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:34:05,899 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9105, 2.3356, 2.3306, 2.8388, 2.0233, 3.2537, 1.7502, 2.7420], device='cuda:6'), covar=tensor([0.1190, 0.0698, 0.1125, 0.0152, 0.0110, 0.0349, 0.1479, 0.0728], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0174, 0.0193, 0.0189, 0.0206, 0.0213, 0.0200, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:34:24,539 INFO [train.py:904] (6/8) Epoch 21, batch 4850, loss[loss=0.185, simple_loss=0.2828, pruned_loss=0.04363, over 16151.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.273, pruned_loss=0.04737, over 3218035.52 frames. ], batch size: 165, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:35:08,243 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=207880.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:35:22,068 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 1.907e+02 2.222e+02 2.565e+02 4.310e+02, threshold=4.444e+02, percent-clipped=1.0 2023-05-01 06:35:40,335 INFO [train.py:904] (6/8) Epoch 21, batch 4900, loss[loss=0.1749, simple_loss=0.2591, pruned_loss=0.04532, over 16536.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2727, pruned_loss=0.04663, over 3198167.23 frames. ], batch size: 68, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:36:40,253 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2806, 3.8150, 4.0215, 1.8965, 4.2215, 4.2929, 3.0943, 2.8312], device='cuda:6'), covar=tensor([0.1201, 0.0170, 0.0159, 0.1368, 0.0068, 0.0113, 0.0424, 0.0629], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0109, 0.0098, 0.0140, 0.0081, 0.0125, 0.0130, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:36:52,780 INFO [train.py:904] (6/8) Epoch 21, batch 4950, loss[loss=0.1868, simple_loss=0.2765, pruned_loss=0.04851, over 11933.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.2725, pruned_loss=0.04626, over 3195334.75 frames. ], batch size: 246, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:37:02,001 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7258, 3.8517, 2.3408, 4.4891, 2.9157, 4.3406, 2.5357, 2.9945], device='cuda:6'), covar=tensor([0.0297, 0.0339, 0.1749, 0.0113, 0.0807, 0.0520, 0.1453, 0.0772], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0176, 0.0194, 0.0161, 0.0177, 0.0215, 0.0201, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:37:16,868 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-05-01 06:37:29,609 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-01 06:37:47,241 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5705, 2.8982, 3.0552, 1.9145, 2.6131, 2.0068, 3.1162, 3.1362], device='cuda:6'), covar=tensor([0.0244, 0.0793, 0.0684, 0.2027, 0.0932, 0.1035, 0.0626, 0.0843], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0164, 0.0167, 0.0152, 0.0145, 0.0130, 0.0144, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:37:47,807 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 1.964e+02 2.322e+02 2.960e+02 4.980e+02, threshold=4.644e+02, percent-clipped=1.0 2023-05-01 06:37:49,789 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-05-01 06:38:08,406 INFO [train.py:904] (6/8) Epoch 21, batch 5000, loss[loss=0.1891, simple_loss=0.2853, pruned_loss=0.04642, over 16462.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2744, pruned_loss=0.04639, over 3212354.41 frames. ], batch size: 146, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:38:52,058 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=208032.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:39:07,130 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3543, 3.5212, 3.5929, 2.1175, 2.9983, 2.4078, 3.6465, 3.6008], device='cuda:6'), covar=tensor([0.0213, 0.0744, 0.0564, 0.1907, 0.0861, 0.0883, 0.0579, 0.0962], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0164, 0.0168, 0.0152, 0.0145, 0.0130, 0.0144, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:39:21,554 INFO [train.py:904] (6/8) Epoch 21, batch 5050, loss[loss=0.1863, simple_loss=0.2744, pruned_loss=0.04914, over 17045.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2747, pruned_loss=0.04621, over 3227760.35 frames. ], batch size: 55, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:40:18,553 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 1.998e+02 2.352e+02 2.773e+02 5.877e+02, threshold=4.704e+02, percent-clipped=4.0 2023-05-01 06:40:23,289 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=208093.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 06:40:29,509 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-05-01 06:40:35,283 INFO [train.py:904] (6/8) Epoch 21, batch 5100, loss[loss=0.177, simple_loss=0.2677, pruned_loss=0.04316, over 16789.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2728, pruned_loss=0.04524, over 3227041.33 frames. ], batch size: 83, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:41:17,397 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8197, 1.2846, 1.6625, 1.7322, 1.8219, 1.9311, 1.6091, 1.8443], device='cuda:6'), covar=tensor([0.0251, 0.0383, 0.0218, 0.0304, 0.0254, 0.0187, 0.0411, 0.0136], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0193, 0.0179, 0.0185, 0.0195, 0.0152, 0.0197, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:41:48,640 INFO [train.py:904] (6/8) Epoch 21, batch 5150, loss[loss=0.1806, simple_loss=0.2796, pruned_loss=0.04083, over 16288.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.273, pruned_loss=0.04495, over 3204588.23 frames. ], batch size: 165, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:41:52,848 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2023-05-01 06:42:32,232 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1754, 3.2203, 3.5614, 1.7302, 3.6726, 3.7137, 2.8437, 2.5812], device='cuda:6'), covar=tensor([0.0961, 0.0282, 0.0155, 0.1304, 0.0071, 0.0167, 0.0432, 0.0578], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0109, 0.0097, 0.0139, 0.0080, 0.0124, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:42:43,673 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 1.920e+02 2.264e+02 2.595e+02 3.716e+02, threshold=4.527e+02, percent-clipped=0.0 2023-05-01 06:42:44,171 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3007, 3.2931, 1.9285, 3.5635, 2.5276, 3.6016, 2.1567, 2.6363], device='cuda:6'), covar=tensor([0.0277, 0.0354, 0.1747, 0.0208, 0.0885, 0.0558, 0.1560, 0.0808], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0174, 0.0191, 0.0159, 0.0175, 0.0212, 0.0198, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:42:49,595 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9640, 4.9323, 4.8906, 3.4493, 4.9286, 1.6951, 4.5044, 4.6261], device='cuda:6'), covar=tensor([0.0185, 0.0137, 0.0258, 0.0903, 0.0160, 0.3520, 0.0237, 0.0342], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0152, 0.0194, 0.0175, 0.0172, 0.0204, 0.0183, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:42:54,459 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2205, 3.9578, 3.9749, 2.3663, 3.5487, 3.9937, 3.5081, 2.1341], device='cuda:6'), covar=tensor([0.0604, 0.0050, 0.0042, 0.0452, 0.0091, 0.0096, 0.0100, 0.0494], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0082, 0.0082, 0.0133, 0.0097, 0.0108, 0.0093, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 06:43:01,060 INFO [train.py:904] (6/8) Epoch 21, batch 5200, loss[loss=0.1674, simple_loss=0.2499, pruned_loss=0.04244, over 16692.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2715, pruned_loss=0.04418, over 3221877.71 frames. ], batch size: 57, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:43:55,251 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7324, 4.8426, 5.1653, 5.1166, 5.1292, 4.8180, 4.7956, 4.6114], device='cuda:6'), covar=tensor([0.0335, 0.0487, 0.0301, 0.0359, 0.0450, 0.0332, 0.0901, 0.0452], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0445, 0.0432, 0.0402, 0.0475, 0.0454, 0.0541, 0.0364], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 06:44:11,865 INFO [train.py:904] (6/8) Epoch 21, batch 5250, loss[loss=0.1758, simple_loss=0.2631, pruned_loss=0.04426, over 16539.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2693, pruned_loss=0.04418, over 3207312.87 frames. ], batch size: 75, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:44:31,759 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=208265.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:44:37,089 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-01 06:45:07,262 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 1.864e+02 2.120e+02 2.516e+02 5.355e+02, threshold=4.239e+02, percent-clipped=2.0 2023-05-01 06:45:17,329 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 06:45:25,338 INFO [train.py:904] (6/8) Epoch 21, batch 5300, loss[loss=0.182, simple_loss=0.2679, pruned_loss=0.04805, over 12125.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2659, pruned_loss=0.0429, over 3199227.81 frames. ], batch size: 246, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:45:28,745 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9375, 5.1805, 4.9538, 5.0004, 4.7411, 4.6989, 4.5961, 5.2651], device='cuda:6'), covar=tensor([0.1206, 0.0773, 0.0919, 0.0768, 0.0750, 0.0928, 0.1095, 0.0812], device='cuda:6'), in_proj_covar=tensor([0.0661, 0.0807, 0.0671, 0.0611, 0.0513, 0.0519, 0.0676, 0.0629], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:45:38,360 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 06:45:48,592 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 06:46:00,421 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=208326.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:46:04,860 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4843, 3.7237, 2.7509, 2.1895, 2.4114, 2.3712, 3.9887, 3.2887], device='cuda:6'), covar=tensor([0.2914, 0.0623, 0.1854, 0.2784, 0.2639, 0.1934, 0.0481, 0.1177], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0270, 0.0304, 0.0310, 0.0296, 0.0257, 0.0297, 0.0335], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 06:46:21,945 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.8004, 6.1165, 5.7755, 5.9367, 5.5312, 5.4678, 5.5621, 6.1653], device='cuda:6'), covar=tensor([0.1123, 0.0736, 0.0872, 0.0781, 0.0796, 0.0601, 0.0978, 0.0825], device='cuda:6'), in_proj_covar=tensor([0.0661, 0.0807, 0.0671, 0.0611, 0.0512, 0.0518, 0.0675, 0.0628], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:46:37,967 INFO [train.py:904] (6/8) Epoch 21, batch 5350, loss[loss=0.1895, simple_loss=0.267, pruned_loss=0.05596, over 12209.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2642, pruned_loss=0.0421, over 3206550.17 frames. ], batch size: 248, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:46:52,029 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4321, 3.3593, 3.7801, 1.9211, 3.9483, 3.9829, 3.0204, 2.8450], device='cuda:6'), covar=tensor([0.0872, 0.0277, 0.0167, 0.1214, 0.0062, 0.0123, 0.0406, 0.0525], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0109, 0.0096, 0.0139, 0.0080, 0.0123, 0.0129, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:47:33,144 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=208388.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 06:47:34,964 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 1.963e+02 2.291e+02 2.599e+02 4.161e+02, threshold=4.581e+02, percent-clipped=0.0 2023-05-01 06:47:53,657 INFO [train.py:904] (6/8) Epoch 21, batch 5400, loss[loss=0.1811, simple_loss=0.2799, pruned_loss=0.04117, over 15215.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.267, pruned_loss=0.04261, over 3220761.90 frames. ], batch size: 190, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:47:57,029 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2770, 4.5278, 4.6975, 4.6607, 4.7093, 4.4074, 4.1386, 4.3283], device='cuda:6'), covar=tensor([0.0515, 0.0742, 0.0554, 0.0616, 0.0700, 0.0560, 0.1440, 0.0548], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0446, 0.0433, 0.0403, 0.0476, 0.0455, 0.0543, 0.0365], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 06:49:10,972 INFO [train.py:904] (6/8) Epoch 21, batch 5450, loss[loss=0.195, simple_loss=0.2791, pruned_loss=0.05549, over 16612.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2699, pruned_loss=0.04402, over 3220647.36 frames. ], batch size: 57, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:49:36,478 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-05-01 06:49:41,191 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-05-01 06:50:09,156 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.336e+02 2.337e+02 2.736e+02 3.594e+02 9.529e+02, threshold=5.472e+02, percent-clipped=14.0 2023-05-01 06:50:19,130 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-05-01 06:50:28,301 INFO [train.py:904] (6/8) Epoch 21, batch 5500, loss[loss=0.1943, simple_loss=0.2852, pruned_loss=0.05171, over 17126.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2776, pruned_loss=0.04929, over 3162423.88 frames. ], batch size: 47, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:51:47,645 INFO [train.py:904] (6/8) Epoch 21, batch 5550, loss[loss=0.2388, simple_loss=0.3231, pruned_loss=0.0772, over 15270.00 frames. ], tot_loss[loss=0.1963, simple_loss=0.2846, pruned_loss=0.05403, over 3144603.19 frames. ], batch size: 190, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:52:04,225 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2482, 3.6360, 3.5085, 2.0886, 2.9482, 2.3932, 3.5161, 3.8911], device='cuda:6'), covar=tensor([0.0301, 0.0776, 0.0671, 0.2159, 0.1008, 0.1056, 0.0780, 0.1025], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0163, 0.0166, 0.0151, 0.0144, 0.0130, 0.0143, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:52:27,790 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=208576.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:52:49,286 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.296e+02 3.115e+02 3.921e+02 4.748e+02 1.231e+03, threshold=7.841e+02, percent-clipped=12.0 2023-05-01 06:53:07,710 INFO [train.py:904] (6/8) Epoch 21, batch 5600, loss[loss=0.198, simple_loss=0.2849, pruned_loss=0.05553, over 16729.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2897, pruned_loss=0.05864, over 3085649.73 frames. ], batch size: 124, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:53:21,094 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5066, 3.5087, 3.4945, 2.7674, 3.3814, 2.0721, 3.2017, 2.8845], device='cuda:6'), covar=tensor([0.0161, 0.0134, 0.0187, 0.0243, 0.0107, 0.2299, 0.0152, 0.0244], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0151, 0.0193, 0.0175, 0.0172, 0.0202, 0.0183, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:53:39,436 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=208621.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:53:47,620 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0107, 2.9016, 2.8513, 2.1607, 2.6636, 2.1806, 2.7753, 3.0464], device='cuda:6'), covar=tensor([0.0310, 0.0630, 0.0461, 0.1597, 0.0756, 0.0894, 0.0512, 0.0626], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0163, 0.0166, 0.0151, 0.0144, 0.0129, 0.0143, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:53:55,554 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7998, 1.3910, 1.7398, 1.7035, 1.7980, 1.8975, 1.6133, 1.7988], device='cuda:6'), covar=tensor([0.0216, 0.0341, 0.0183, 0.0262, 0.0234, 0.0158, 0.0376, 0.0127], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0192, 0.0178, 0.0184, 0.0195, 0.0152, 0.0196, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:54:07,052 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=208637.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:54:08,454 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1738, 1.8375, 2.6827, 3.0634, 2.9004, 3.4375, 1.9073, 3.4801], device='cuda:6'), covar=tensor([0.0170, 0.0592, 0.0292, 0.0254, 0.0246, 0.0140, 0.0704, 0.0108], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0192, 0.0178, 0.0184, 0.0195, 0.0152, 0.0196, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:54:29,857 INFO [train.py:904] (6/8) Epoch 21, batch 5650, loss[loss=0.2428, simple_loss=0.3151, pruned_loss=0.08522, over 15317.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2943, pruned_loss=0.06253, over 3067388.74 frames. ], batch size: 190, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:55:28,121 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=208688.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:55:30,664 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.148e+02 3.160e+02 3.845e+02 4.644e+02 8.553e+02, threshold=7.690e+02, percent-clipped=3.0 2023-05-01 06:55:48,393 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0567, 2.3385, 2.3884, 2.8139, 2.0980, 3.2011, 1.8530, 2.7543], device='cuda:6'), covar=tensor([0.1131, 0.0626, 0.1113, 0.0197, 0.0132, 0.0403, 0.1488, 0.0758], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0174, 0.0194, 0.0189, 0.0207, 0.0214, 0.0201, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 06:55:50,854 INFO [train.py:904] (6/8) Epoch 21, batch 5700, loss[loss=0.2505, simple_loss=0.3103, pruned_loss=0.09533, over 11722.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2956, pruned_loss=0.06385, over 3077455.23 frames. ], batch size: 248, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:56:45,734 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=208736.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:57:10,969 INFO [train.py:904] (6/8) Epoch 21, batch 5750, loss[loss=0.2049, simple_loss=0.2946, pruned_loss=0.05762, over 16946.00 frames. ], tot_loss[loss=0.2146, simple_loss=0.2985, pruned_loss=0.06538, over 3062161.60 frames. ], batch size: 96, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:58:13,920 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.275e+02 2.983e+02 3.733e+02 4.569e+02 7.128e+02, threshold=7.466e+02, percent-clipped=0.0 2023-05-01 06:58:33,851 INFO [train.py:904] (6/8) Epoch 21, batch 5800, loss[loss=0.2316, simple_loss=0.3025, pruned_loss=0.08032, over 11879.00 frames. ], tot_loss[loss=0.213, simple_loss=0.298, pruned_loss=0.06403, over 3049235.77 frames. ], batch size: 246, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:59:36,350 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5458, 2.5525, 2.5604, 4.3737, 2.3304, 2.8968, 2.5219, 2.6507], device='cuda:6'), covar=tensor([0.1207, 0.3193, 0.2622, 0.0430, 0.3991, 0.2193, 0.3186, 0.3007], device='cuda:6'), in_proj_covar=tensor([0.0397, 0.0442, 0.0362, 0.0323, 0.0431, 0.0508, 0.0411, 0.0516], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 06:59:52,593 INFO [train.py:904] (6/8) Epoch 21, batch 5850, loss[loss=0.1889, simple_loss=0.2806, pruned_loss=0.04858, over 16968.00 frames. ], tot_loss[loss=0.2101, simple_loss=0.2955, pruned_loss=0.06231, over 3060132.42 frames. ], batch size: 109, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:00:41,449 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3515, 5.3225, 5.0841, 4.4195, 5.2359, 1.7850, 4.9797, 4.8388], device='cuda:6'), covar=tensor([0.0097, 0.0082, 0.0174, 0.0374, 0.0080, 0.2804, 0.0161, 0.0209], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0150, 0.0192, 0.0173, 0.0170, 0.0200, 0.0181, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:00:53,525 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.655e+02 2.740e+02 3.068e+02 3.692e+02 6.527e+02, threshold=6.136e+02, percent-clipped=0.0 2023-05-01 07:00:57,592 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 07:01:12,710 INFO [train.py:904] (6/8) Epoch 21, batch 5900, loss[loss=0.2375, simple_loss=0.3016, pruned_loss=0.08671, over 11333.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2958, pruned_loss=0.06268, over 3061868.37 frames. ], batch size: 246, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:01:43,971 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5037, 3.5634, 3.3406, 3.0115, 3.1887, 3.4644, 3.3167, 3.2348], device='cuda:6'), covar=tensor([0.0572, 0.0544, 0.0258, 0.0231, 0.0470, 0.0439, 0.1309, 0.0440], device='cuda:6'), in_proj_covar=tensor([0.0287, 0.0416, 0.0336, 0.0332, 0.0344, 0.0386, 0.0230, 0.0403], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:01:48,534 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=208921.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:02:05,302 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=208932.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:02:36,220 INFO [train.py:904] (6/8) Epoch 21, batch 5950, loss[loss=0.2045, simple_loss=0.2921, pruned_loss=0.05845, over 16891.00 frames. ], tot_loss[loss=0.2091, simple_loss=0.296, pruned_loss=0.0611, over 3073580.97 frames. ], batch size: 116, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:03:02,963 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=208969.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:03:36,546 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 2.651e+02 3.328e+02 3.727e+02 9.744e+02, threshold=6.656e+02, percent-clipped=4.0 2023-05-01 07:03:56,981 INFO [train.py:904] (6/8) Epoch 21, batch 6000, loss[loss=0.2032, simple_loss=0.2912, pruned_loss=0.05757, over 16898.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2948, pruned_loss=0.06062, over 3072438.66 frames. ], batch size: 109, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:03:56,981 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 07:04:08,270 INFO [train.py:938] (6/8) Epoch 21, validation: loss=0.1512, simple_loss=0.2639, pruned_loss=0.01924, over 944034.00 frames. 2023-05-01 07:04:08,271 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 07:05:25,680 INFO [train.py:904] (6/8) Epoch 21, batch 6050, loss[loss=0.1946, simple_loss=0.2871, pruned_loss=0.05099, over 15406.00 frames. ], tot_loss[loss=0.206, simple_loss=0.2929, pruned_loss=0.05958, over 3075656.62 frames. ], batch size: 191, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:06:26,990 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.899e+02 2.646e+02 3.170e+02 4.163e+02 9.783e+02, threshold=6.340e+02, percent-clipped=1.0 2023-05-01 07:06:41,602 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-05-01 07:06:45,918 INFO [train.py:904] (6/8) Epoch 21, batch 6100, loss[loss=0.1986, simple_loss=0.2858, pruned_loss=0.05569, over 16922.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2928, pruned_loss=0.05894, over 3093587.80 frames. ], batch size: 109, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:07:29,430 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8603, 3.7601, 3.8957, 4.0025, 4.0757, 3.6791, 4.0392, 4.1063], device='cuda:6'), covar=tensor([0.1507, 0.1147, 0.1274, 0.0671, 0.0604, 0.1912, 0.0912, 0.0704], device='cuda:6'), in_proj_covar=tensor([0.0622, 0.0767, 0.0896, 0.0782, 0.0588, 0.0618, 0.0636, 0.0736], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:07:55,703 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4680, 3.5653, 2.5932, 2.1405, 2.3564, 2.3058, 3.7567, 3.1958], device='cuda:6'), covar=tensor([0.3017, 0.0662, 0.1937, 0.2635, 0.2571, 0.2035, 0.0482, 0.1268], device='cuda:6'), in_proj_covar=tensor([0.0326, 0.0269, 0.0303, 0.0309, 0.0295, 0.0256, 0.0294, 0.0333], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 07:08:04,715 INFO [train.py:904] (6/8) Epoch 21, batch 6150, loss[loss=0.1958, simple_loss=0.2871, pruned_loss=0.05222, over 16982.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2905, pruned_loss=0.05801, over 3101270.02 frames. ], batch size: 55, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:08:10,710 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6152, 3.5613, 3.8285, 2.0884, 3.3594, 2.4203, 3.9513, 3.8969], device='cuda:6'), covar=tensor([0.0208, 0.0849, 0.0527, 0.2133, 0.0763, 0.0946, 0.0569, 0.0972], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0163, 0.0167, 0.0152, 0.0144, 0.0129, 0.0142, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 07:09:04,296 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.750e+02 2.685e+02 3.365e+02 4.144e+02 7.779e+02, threshold=6.731e+02, percent-clipped=2.0 2023-05-01 07:09:23,293 INFO [train.py:904] (6/8) Epoch 21, batch 6200, loss[loss=0.1872, simple_loss=0.2783, pruned_loss=0.0481, over 17042.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2894, pruned_loss=0.05816, over 3106776.22 frames. ], batch size: 55, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:09:23,820 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209202.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:10:10,033 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=209232.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:10:39,932 INFO [train.py:904] (6/8) Epoch 21, batch 6250, loss[loss=0.1825, simple_loss=0.2821, pruned_loss=0.04147, over 16820.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2893, pruned_loss=0.05769, over 3106939.30 frames. ], batch size: 102, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:10:46,195 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209255.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:10:57,734 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209263.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:11:22,157 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=209280.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:11:35,898 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.929e+02 2.681e+02 3.167e+02 3.896e+02 8.805e+02, threshold=6.333e+02, percent-clipped=2.0 2023-05-01 07:11:50,467 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209298.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:11:54,987 INFO [train.py:904] (6/8) Epoch 21, batch 6300, loss[loss=0.1748, simple_loss=0.2693, pruned_loss=0.04019, over 16892.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2886, pruned_loss=0.05717, over 3105624.38 frames. ], batch size: 90, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:12:17,943 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209316.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:13:13,581 INFO [train.py:904] (6/8) Epoch 21, batch 6350, loss[loss=0.2532, simple_loss=0.3086, pruned_loss=0.09893, over 11244.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2893, pruned_loss=0.05824, over 3094762.92 frames. ], batch size: 246, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:13:24,966 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209359.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:14:13,987 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.652e+02 3.041e+02 3.551e+02 4.346e+02 9.300e+02, threshold=7.102e+02, percent-clipped=4.0 2023-05-01 07:14:31,817 INFO [train.py:904] (6/8) Epoch 21, batch 6400, loss[loss=0.2513, simple_loss=0.3183, pruned_loss=0.09211, over 11630.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2893, pruned_loss=0.05921, over 3089156.33 frames. ], batch size: 247, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:15:02,634 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6570, 1.8276, 2.1815, 2.5543, 2.5861, 2.8787, 2.0344, 2.8532], device='cuda:6'), covar=tensor([0.0220, 0.0514, 0.0385, 0.0353, 0.0324, 0.0216, 0.0515, 0.0158], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0194, 0.0178, 0.0184, 0.0196, 0.0152, 0.0196, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:15:35,876 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6518, 2.5799, 2.0020, 2.7051, 2.2271, 2.7694, 2.1415, 2.3921], device='cuda:6'), covar=tensor([0.0343, 0.0503, 0.1235, 0.0326, 0.0726, 0.0644, 0.1319, 0.0608], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0177, 0.0194, 0.0161, 0.0176, 0.0216, 0.0201, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 07:15:41,986 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6498, 2.5746, 2.3963, 3.6942, 2.7855, 3.8642, 1.4048, 2.9526], device='cuda:6'), covar=tensor([0.1412, 0.0793, 0.1306, 0.0166, 0.0230, 0.0474, 0.1864, 0.0793], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0174, 0.0194, 0.0189, 0.0206, 0.0215, 0.0201, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 07:15:48,008 INFO [train.py:904] (6/8) Epoch 21, batch 6450, loss[loss=0.233, simple_loss=0.3174, pruned_loss=0.07427, over 17033.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2902, pruned_loss=0.0591, over 3099273.33 frames. ], batch size: 41, lr: 3.21e-03, grad_scale: 4.0 2023-05-01 07:16:10,771 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209466.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:16:15,927 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 07:16:33,187 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 07:16:52,762 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.025e+02 2.825e+02 3.422e+02 4.079e+02 7.372e+02, threshold=6.844e+02, percent-clipped=3.0 2023-05-01 07:17:08,456 INFO [train.py:904] (6/8) Epoch 21, batch 6500, loss[loss=0.1976, simple_loss=0.2802, pruned_loss=0.05753, over 16610.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2881, pruned_loss=0.05834, over 3116122.97 frames. ], batch size: 62, lr: 3.21e-03, grad_scale: 4.0 2023-05-01 07:17:21,400 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209510.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:17:48,105 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209527.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:18:28,631 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5606, 4.3513, 4.5998, 4.7453, 4.9483, 4.4384, 4.8989, 4.9217], device='cuda:6'), covar=tensor([0.1930, 0.1341, 0.1630, 0.0824, 0.0657, 0.1155, 0.0769, 0.0660], device='cuda:6'), in_proj_covar=tensor([0.0624, 0.0768, 0.0898, 0.0782, 0.0585, 0.0616, 0.0636, 0.0736], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:18:29,351 INFO [train.py:904] (6/8) Epoch 21, batch 6550, loss[loss=0.1962, simple_loss=0.2986, pruned_loss=0.04688, over 16708.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.2905, pruned_loss=0.05882, over 3128857.36 frames. ], batch size: 83, lr: 3.21e-03, grad_scale: 4.0 2023-05-01 07:18:40,050 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=209558.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:19:01,160 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209571.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:19:22,721 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-01 07:19:33,742 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 2.646e+02 3.318e+02 3.875e+02 1.016e+03, threshold=6.636e+02, percent-clipped=2.0 2023-05-01 07:19:49,520 INFO [train.py:904] (6/8) Epoch 21, batch 6600, loss[loss=0.225, simple_loss=0.3077, pruned_loss=0.07119, over 16692.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2925, pruned_loss=0.05926, over 3122130.29 frames. ], batch size: 124, lr: 3.21e-03, grad_scale: 4.0 2023-05-01 07:19:56,045 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209606.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:20:03,696 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=209611.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:21:07,687 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-05-01 07:21:07,995 INFO [train.py:904] (6/8) Epoch 21, batch 6650, loss[loss=0.2374, simple_loss=0.3053, pruned_loss=0.08476, over 11390.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2935, pruned_loss=0.0606, over 3093371.96 frames. ], batch size: 248, lr: 3.21e-03, grad_scale: 2.0 2023-05-01 07:21:12,193 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=209654.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:21:33,103 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209667.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:22:05,640 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209688.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:22:11,939 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.980e+02 3.003e+02 3.834e+02 5.149e+02 1.142e+03, threshold=7.669e+02, percent-clipped=10.0 2023-05-01 07:22:25,654 INFO [train.py:904] (6/8) Epoch 21, batch 6700, loss[loss=0.2239, simple_loss=0.2912, pruned_loss=0.07828, over 11579.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2913, pruned_loss=0.0599, over 3104712.58 frames. ], batch size: 248, lr: 3.21e-03, grad_scale: 2.0 2023-05-01 07:22:31,748 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 07:23:11,440 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7678, 2.5164, 2.3040, 3.2447, 2.2892, 3.5919, 1.5883, 2.7652], device='cuda:6'), covar=tensor([0.1381, 0.0744, 0.1353, 0.0186, 0.0200, 0.0416, 0.1766, 0.0869], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0174, 0.0195, 0.0188, 0.0206, 0.0214, 0.0201, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 07:23:40,222 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209749.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:23:44,112 INFO [train.py:904] (6/8) Epoch 21, batch 6750, loss[loss=0.2171, simple_loss=0.306, pruned_loss=0.06412, over 15422.00 frames. ], tot_loss[loss=0.206, simple_loss=0.291, pruned_loss=0.06046, over 3094217.18 frames. ], batch size: 191, lr: 3.20e-03, grad_scale: 2.0 2023-05-01 07:24:47,246 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.958e+02 2.606e+02 3.292e+02 3.909e+02 7.681e+02, threshold=6.584e+02, percent-clipped=1.0 2023-05-01 07:25:01,547 INFO [train.py:904] (6/8) Epoch 21, batch 6800, loss[loss=0.1815, simple_loss=0.2716, pruned_loss=0.04567, over 16462.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2906, pruned_loss=0.05999, over 3083146.65 frames. ], batch size: 68, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:25:33,965 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=209822.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:26:19,716 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1886, 4.3177, 2.6896, 4.8554, 3.1969, 4.7217, 2.7522, 3.3823], device='cuda:6'), covar=tensor([0.0223, 0.0315, 0.1537, 0.0209, 0.0743, 0.0619, 0.1498, 0.0681], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0176, 0.0194, 0.0161, 0.0176, 0.0215, 0.0201, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 07:26:20,406 INFO [train.py:904] (6/8) Epoch 21, batch 6850, loss[loss=0.1857, simple_loss=0.2846, pruned_loss=0.04343, over 16766.00 frames. ], tot_loss[loss=0.2067, simple_loss=0.292, pruned_loss=0.06073, over 3073336.27 frames. ], batch size: 39, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:26:29,604 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=209858.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:26:30,233 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.53 vs. limit=5.0 2023-05-01 07:26:42,224 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=209866.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:27:01,896 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5183, 2.5132, 2.0495, 2.2081, 3.0178, 2.6979, 3.1523, 3.3194], device='cuda:6'), covar=tensor([0.0154, 0.0588, 0.0721, 0.0544, 0.0305, 0.0393, 0.0252, 0.0267], device='cuda:6'), in_proj_covar=tensor([0.0203, 0.0231, 0.0223, 0.0224, 0.0232, 0.0230, 0.0231, 0.0227], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:27:21,614 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 2.648e+02 3.249e+02 3.781e+02 8.526e+02, threshold=6.499e+02, percent-clipped=1.0 2023-05-01 07:27:34,654 INFO [train.py:904] (6/8) Epoch 21, batch 6900, loss[loss=0.1898, simple_loss=0.2849, pruned_loss=0.04732, over 17044.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2941, pruned_loss=0.06021, over 3083107.04 frames. ], batch size: 50, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:27:41,303 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=209906.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:27:48,381 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=209911.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:27:50,441 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3873, 2.8661, 2.6472, 2.2627, 2.3036, 2.2957, 2.8679, 2.8621], device='cuda:6'), covar=tensor([0.2188, 0.0691, 0.1465, 0.2190, 0.2008, 0.1966, 0.0505, 0.1250], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0271, 0.0303, 0.0311, 0.0296, 0.0257, 0.0294, 0.0334], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 07:27:50,624 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-05-01 07:28:51,081 INFO [train.py:904] (6/8) Epoch 21, batch 6950, loss[loss=0.2216, simple_loss=0.306, pruned_loss=0.06858, over 16722.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.296, pruned_loss=0.06243, over 3066009.69 frames. ], batch size: 89, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:28:54,470 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=209954.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:29:02,106 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=209959.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:29:06,368 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=209962.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:29:51,931 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.035e+02 3.251e+02 3.766e+02 4.564e+02 9.580e+02, threshold=7.532e+02, percent-clipped=9.0 2023-05-01 07:30:09,079 INFO [train.py:904] (6/8) Epoch 21, batch 7000, loss[loss=0.2021, simple_loss=0.2955, pruned_loss=0.05431, over 16966.00 frames. ], tot_loss[loss=0.2102, simple_loss=0.2963, pruned_loss=0.06203, over 3058402.55 frames. ], batch size: 41, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:30:09,335 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=210002.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:30:49,946 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0626, 2.1169, 2.1583, 3.6299, 2.0945, 2.4979, 2.2278, 2.2613], device='cuda:6'), covar=tensor([0.1345, 0.3462, 0.3030, 0.0578, 0.4139, 0.2390, 0.3503, 0.3322], device='cuda:6'), in_proj_covar=tensor([0.0398, 0.0444, 0.0364, 0.0324, 0.0434, 0.0510, 0.0414, 0.0518], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:31:11,217 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=210044.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:31:22,810 INFO [train.py:904] (6/8) Epoch 21, batch 7050, loss[loss=0.17, simple_loss=0.262, pruned_loss=0.03896, over 16308.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.2965, pruned_loss=0.06138, over 3062652.16 frames. ], batch size: 35, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:31:30,579 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=210057.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:32:24,158 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.868e+02 2.885e+02 3.523e+02 4.096e+02 8.338e+02, threshold=7.047e+02, percent-clipped=1.0 2023-05-01 07:32:37,544 INFO [train.py:904] (6/8) Epoch 21, batch 7100, loss[loss=0.1988, simple_loss=0.2826, pruned_loss=0.05748, over 16208.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2949, pruned_loss=0.061, over 3056341.74 frames. ], batch size: 165, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:33:03,349 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=210118.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:33:08,747 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-01 07:33:09,971 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=210122.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:33:11,671 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-05-01 07:33:55,440 INFO [train.py:904] (6/8) Epoch 21, batch 7150, loss[loss=0.1991, simple_loss=0.2839, pruned_loss=0.05711, over 16248.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2937, pruned_loss=0.06131, over 3039526.49 frames. ], batch size: 165, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:34:16,439 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=210166.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:34:21,275 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=210170.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:34:48,213 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-01 07:34:53,944 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.076e+02 2.584e+02 3.186e+02 4.207e+02 8.537e+02, threshold=6.371e+02, percent-clipped=2.0 2023-05-01 07:35:07,915 INFO [train.py:904] (6/8) Epoch 21, batch 7200, loss[loss=0.1983, simple_loss=0.2764, pruned_loss=0.06005, over 11390.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.291, pruned_loss=0.05934, over 3044350.96 frames. ], batch size: 247, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:35:13,896 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0537, 4.1510, 4.4153, 4.3914, 4.3923, 4.1372, 4.1279, 4.0573], device='cuda:6'), covar=tensor([0.0344, 0.0584, 0.0390, 0.0407, 0.0484, 0.0400, 0.0919, 0.0565], device='cuda:6'), in_proj_covar=tensor([0.0400, 0.0444, 0.0433, 0.0402, 0.0480, 0.0455, 0.0540, 0.0364], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 07:35:27,061 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=210214.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:35:36,885 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4581, 5.7886, 5.5169, 5.5686, 5.1420, 5.1900, 5.2074, 5.9134], device='cuda:6'), covar=tensor([0.1200, 0.0775, 0.0981, 0.0908, 0.0836, 0.0733, 0.1108, 0.0759], device='cuda:6'), in_proj_covar=tensor([0.0662, 0.0800, 0.0670, 0.0610, 0.0510, 0.0520, 0.0674, 0.0629], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:36:28,283 INFO [train.py:904] (6/8) Epoch 21, batch 7250, loss[loss=0.174, simple_loss=0.262, pruned_loss=0.04303, over 16821.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2885, pruned_loss=0.05796, over 3045062.06 frames. ], batch size: 102, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:36:43,143 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-01 07:36:43,769 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=210262.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:37:31,771 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.805e+02 2.676e+02 3.072e+02 3.848e+02 7.717e+02, threshold=6.145e+02, percent-clipped=2.0 2023-05-01 07:37:44,655 INFO [train.py:904] (6/8) Epoch 21, batch 7300, loss[loss=0.1798, simple_loss=0.277, pruned_loss=0.04129, over 16989.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2871, pruned_loss=0.05755, over 3051818.85 frames. ], batch size: 41, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:37:51,079 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-01 07:37:58,862 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=210310.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:38:45,241 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0613, 2.3779, 2.3530, 2.6067, 1.9271, 3.1285, 1.8771, 2.6834], device='cuda:6'), covar=tensor([0.1001, 0.0587, 0.1010, 0.0173, 0.0120, 0.0365, 0.1284, 0.0702], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0174, 0.0196, 0.0189, 0.0207, 0.0215, 0.0202, 0.0194], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 07:38:48,857 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=210344.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:38:59,602 INFO [train.py:904] (6/8) Epoch 21, batch 7350, loss[loss=0.1909, simple_loss=0.2806, pruned_loss=0.05055, over 16561.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2891, pruned_loss=0.05904, over 3036308.22 frames. ], batch size: 68, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:40:00,159 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=210392.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:40:02,081 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.675e+02 2.648e+02 3.204e+02 3.841e+02 6.410e+02, threshold=6.409e+02, percent-clipped=2.0 2023-05-01 07:40:11,061 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4319, 4.3081, 4.5002, 4.6271, 4.7809, 4.3463, 4.7460, 4.7971], device='cuda:6'), covar=tensor([0.1799, 0.1239, 0.1498, 0.0722, 0.0647, 0.1100, 0.0805, 0.0633], device='cuda:6'), in_proj_covar=tensor([0.0618, 0.0762, 0.0882, 0.0777, 0.0584, 0.0612, 0.0632, 0.0727], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:40:14,853 INFO [train.py:904] (6/8) Epoch 21, batch 7400, loss[loss=0.2047, simple_loss=0.296, pruned_loss=0.05674, over 16327.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2898, pruned_loss=0.05884, over 3075838.34 frames. ], batch size: 146, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:40:27,103 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2070, 5.2653, 5.1056, 4.7331, 4.7421, 5.1900, 5.1044, 4.8445], device='cuda:6'), covar=tensor([0.0682, 0.0595, 0.0297, 0.0309, 0.1036, 0.0562, 0.0331, 0.0685], device='cuda:6'), in_proj_covar=tensor([0.0280, 0.0406, 0.0327, 0.0322, 0.0337, 0.0376, 0.0226, 0.0392], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:40:32,778 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=210413.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:40:44,386 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5051, 2.4777, 2.5202, 4.4613, 2.3161, 2.9469, 2.5692, 2.7192], device='cuda:6'), covar=tensor([0.1215, 0.3292, 0.2599, 0.0420, 0.3811, 0.2178, 0.3225, 0.2802], device='cuda:6'), in_proj_covar=tensor([0.0397, 0.0445, 0.0362, 0.0324, 0.0435, 0.0512, 0.0414, 0.0518], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:41:02,827 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.00 vs. limit=5.0 2023-05-01 07:41:32,231 INFO [train.py:904] (6/8) Epoch 21, batch 7450, loss[loss=0.2056, simple_loss=0.2958, pruned_loss=0.05764, over 16877.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2905, pruned_loss=0.05987, over 3083895.52 frames. ], batch size: 116, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:41:58,373 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2063, 3.5294, 3.6295, 2.1656, 3.1305, 2.3956, 3.5670, 3.7238], device='cuda:6'), covar=tensor([0.0278, 0.0762, 0.0549, 0.2012, 0.0806, 0.0970, 0.0615, 0.0908], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0163, 0.0167, 0.0152, 0.0144, 0.0129, 0.0143, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 07:42:42,640 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.331e+02 3.072e+02 3.553e+02 4.443e+02 7.195e+02, threshold=7.106e+02, percent-clipped=1.0 2023-05-01 07:42:48,816 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0401, 4.0511, 3.9733, 3.2138, 3.9779, 1.8534, 3.7840, 3.5866], device='cuda:6'), covar=tensor([0.0137, 0.0112, 0.0187, 0.0323, 0.0104, 0.2679, 0.0137, 0.0254], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0149, 0.0192, 0.0173, 0.0170, 0.0202, 0.0181, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:42:53,254 INFO [train.py:904] (6/8) Epoch 21, batch 7500, loss[loss=0.2007, simple_loss=0.2868, pruned_loss=0.05737, over 16945.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2911, pruned_loss=0.05959, over 3085740.57 frames. ], batch size: 109, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:43:25,814 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-01 07:43:47,670 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8924, 2.6368, 2.6151, 1.9238, 2.5359, 2.6716, 2.5191, 1.9149], device='cuda:6'), covar=tensor([0.0432, 0.0094, 0.0091, 0.0374, 0.0135, 0.0133, 0.0128, 0.0392], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0082, 0.0083, 0.0134, 0.0096, 0.0107, 0.0093, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 07:44:09,156 INFO [train.py:904] (6/8) Epoch 21, batch 7550, loss[loss=0.206, simple_loss=0.2859, pruned_loss=0.06304, over 15259.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2905, pruned_loss=0.06013, over 3067695.91 frames. ], batch size: 191, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:45:11,346 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.786e+02 2.775e+02 3.392e+02 4.108e+02 6.854e+02, threshold=6.785e+02, percent-clipped=0.0 2023-05-01 07:45:23,229 INFO [train.py:904] (6/8) Epoch 21, batch 7600, loss[loss=0.1833, simple_loss=0.263, pruned_loss=0.05177, over 16799.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2896, pruned_loss=0.06027, over 3069754.86 frames. ], batch size: 39, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:45:47,400 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.92 vs. limit=5.0 2023-05-01 07:46:02,499 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2823, 2.1120, 2.6922, 3.1254, 2.9964, 3.7929, 2.2751, 3.5794], device='cuda:6'), covar=tensor([0.0207, 0.0525, 0.0342, 0.0299, 0.0285, 0.0135, 0.0549, 0.0155], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0191, 0.0176, 0.0182, 0.0194, 0.0150, 0.0194, 0.0148], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:46:08,307 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=210632.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:46:37,228 INFO [train.py:904] (6/8) Epoch 21, batch 7650, loss[loss=0.2706, simple_loss=0.3286, pruned_loss=0.1063, over 11931.00 frames. ], tot_loss[loss=0.2058, simple_loss=0.2904, pruned_loss=0.06063, over 3089986.72 frames. ], batch size: 248, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:46:59,517 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=210666.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:47:25,959 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=210682.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:47:42,141 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=210693.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:47:42,756 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.936e+02 3.052e+02 3.513e+02 4.198e+02 7.732e+02, threshold=7.025e+02, percent-clipped=2.0 2023-05-01 07:47:44,132 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6948, 2.5577, 2.1637, 3.3975, 2.3646, 3.6672, 1.4270, 2.8222], device='cuda:6'), covar=tensor([0.1439, 0.0808, 0.1505, 0.0207, 0.0201, 0.0412, 0.1885, 0.0816], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0175, 0.0196, 0.0190, 0.0208, 0.0216, 0.0202, 0.0195], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 07:47:55,351 INFO [train.py:904] (6/8) Epoch 21, batch 7700, loss[loss=0.2194, simple_loss=0.2934, pruned_loss=0.07276, over 11702.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2909, pruned_loss=0.06164, over 3060071.51 frames. ], batch size: 247, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:48:12,312 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=210713.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:48:34,189 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=210727.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:48:59,584 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=210743.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:49:12,160 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.94 vs. limit=5.0 2023-05-01 07:49:12,532 INFO [train.py:904] (6/8) Epoch 21, batch 7750, loss[loss=0.2133, simple_loss=0.2926, pruned_loss=0.06703, over 16495.00 frames. ], tot_loss[loss=0.2063, simple_loss=0.2908, pruned_loss=0.06088, over 3059764.39 frames. ], batch size: 146, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:49:27,500 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=210761.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:49:33,009 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=210764.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:49:36,102 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3307, 4.4100, 4.2341, 3.9715, 3.9444, 4.3369, 4.0384, 4.0729], device='cuda:6'), covar=tensor([0.0588, 0.0491, 0.0278, 0.0264, 0.0756, 0.0426, 0.0612, 0.0581], device='cuda:6'), in_proj_covar=tensor([0.0279, 0.0405, 0.0326, 0.0320, 0.0335, 0.0375, 0.0226, 0.0391], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 07:49:36,579 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-05-01 07:50:18,619 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.950e+02 2.795e+02 3.327e+02 4.093e+02 7.547e+02, threshold=6.654e+02, percent-clipped=2.0 2023-05-01 07:50:31,004 INFO [train.py:904] (6/8) Epoch 21, batch 7800, loss[loss=0.1714, simple_loss=0.2721, pruned_loss=0.03534, over 16915.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2914, pruned_loss=0.06106, over 3077988.98 frames. ], batch size: 96, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:51:07,534 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=210825.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:51:48,910 INFO [train.py:904] (6/8) Epoch 21, batch 7850, loss[loss=0.2076, simple_loss=0.3012, pruned_loss=0.05699, over 16885.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2922, pruned_loss=0.06101, over 3061661.85 frames. ], batch size: 42, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:52:54,099 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.040e+02 2.722e+02 3.190e+02 3.938e+02 6.813e+02, threshold=6.379e+02, percent-clipped=1.0 2023-05-01 07:53:05,677 INFO [train.py:904] (6/8) Epoch 21, batch 7900, loss[loss=0.1999, simple_loss=0.2884, pruned_loss=0.05574, over 16763.00 frames. ], tot_loss[loss=0.2058, simple_loss=0.2909, pruned_loss=0.06037, over 3069376.97 frames. ], batch size: 124, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:53:59,073 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-01 07:54:03,614 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2023-05-01 07:54:24,386 INFO [train.py:904] (6/8) Epoch 21, batch 7950, loss[loss=0.2288, simple_loss=0.3072, pruned_loss=0.07517, over 15348.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.2923, pruned_loss=0.06122, over 3068684.63 frames. ], batch size: 191, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:54:33,143 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7979, 3.8716, 4.1254, 4.0976, 4.1191, 3.8680, 3.8749, 3.8987], device='cuda:6'), covar=tensor([0.0342, 0.0674, 0.0400, 0.0452, 0.0467, 0.0482, 0.0872, 0.0522], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0443, 0.0431, 0.0401, 0.0478, 0.0453, 0.0539, 0.0361], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 07:55:14,457 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=210984.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:55:20,869 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=210988.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:55:29,111 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.961e+02 3.312e+02 4.005e+02 9.237e+02, threshold=6.624e+02, percent-clipped=4.0 2023-05-01 07:55:41,323 INFO [train.py:904] (6/8) Epoch 21, batch 8000, loss[loss=0.209, simple_loss=0.3066, pruned_loss=0.05577, over 16359.00 frames. ], tot_loss[loss=0.2077, simple_loss=0.2925, pruned_loss=0.06142, over 3075335.60 frames. ], batch size: 146, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:56:12,594 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=211022.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:56:35,948 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=211038.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:56:45,507 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=211045.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:56:56,538 INFO [train.py:904] (6/8) Epoch 21, batch 8050, loss[loss=0.2215, simple_loss=0.2932, pruned_loss=0.07487, over 11727.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.2925, pruned_loss=0.0616, over 3052513.63 frames. ], batch size: 246, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 07:57:57,290 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2407, 3.8837, 4.4396, 2.0308, 4.7101, 4.6981, 3.3038, 3.4218], device='cuda:6'), covar=tensor([0.0635, 0.0261, 0.0200, 0.1268, 0.0057, 0.0124, 0.0427, 0.0425], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0107, 0.0096, 0.0137, 0.0079, 0.0123, 0.0128, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 07:57:59,279 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.685e+02 2.757e+02 3.257e+02 3.943e+02 6.625e+02, threshold=6.515e+02, percent-clipped=2.0 2023-05-01 07:58:08,124 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6181, 3.6767, 2.2722, 4.1582, 2.7887, 4.1103, 2.2699, 2.9346], device='cuda:6'), covar=tensor([0.0278, 0.0402, 0.1603, 0.0181, 0.0794, 0.0510, 0.1537, 0.0724], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0174, 0.0192, 0.0158, 0.0174, 0.0214, 0.0199, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 07:58:10,496 INFO [train.py:904] (6/8) Epoch 21, batch 8100, loss[loss=0.2404, simple_loss=0.323, pruned_loss=0.07891, over 15385.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2919, pruned_loss=0.06104, over 3051617.50 frames. ], batch size: 190, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 07:58:38,286 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=211120.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:59:05,839 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0115, 3.3527, 3.4022, 2.0903, 3.1142, 3.3945, 3.1453, 1.9330], device='cuda:6'), covar=tensor([0.0582, 0.0061, 0.0066, 0.0477, 0.0116, 0.0118, 0.0112, 0.0501], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0082, 0.0083, 0.0134, 0.0096, 0.0108, 0.0093, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 07:59:22,869 INFO [train.py:904] (6/8) Epoch 21, batch 8150, loss[loss=0.2164, simple_loss=0.2897, pruned_loss=0.07154, over 11891.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2898, pruned_loss=0.05998, over 3063659.25 frames. ], batch size: 247, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 07:59:36,770 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 08:00:19,516 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3995, 3.3342, 3.4493, 3.5237, 3.5545, 3.2966, 3.4820, 3.5940], device='cuda:6'), covar=tensor([0.1295, 0.0944, 0.0985, 0.0606, 0.0663, 0.2328, 0.1046, 0.0767], device='cuda:6'), in_proj_covar=tensor([0.0616, 0.0759, 0.0879, 0.0772, 0.0581, 0.0611, 0.0631, 0.0728], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:00:27,467 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.932e+02 2.753e+02 3.326e+02 4.060e+02 8.278e+02, threshold=6.652e+02, percent-clipped=2.0 2023-05-01 08:00:34,151 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2488, 5.2600, 5.0538, 4.3184, 5.1509, 1.9798, 4.9012, 4.7996], device='cuda:6'), covar=tensor([0.0097, 0.0084, 0.0211, 0.0451, 0.0102, 0.2593, 0.0142, 0.0222], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0151, 0.0195, 0.0175, 0.0172, 0.0204, 0.0182, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:00:40,743 INFO [train.py:904] (6/8) Epoch 21, batch 8200, loss[loss=0.2402, simple_loss=0.3018, pruned_loss=0.08929, over 11299.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2877, pruned_loss=0.05958, over 3062129.46 frames. ], batch size: 248, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:01:59,672 INFO [train.py:904] (6/8) Epoch 21, batch 8250, loss[loss=0.1766, simple_loss=0.2758, pruned_loss=0.0387, over 15360.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2861, pruned_loss=0.05668, over 3066410.46 frames. ], batch size: 191, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:02:37,462 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-01 08:02:56,870 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=211288.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:03:06,691 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.662e+02 2.346e+02 2.848e+02 3.670e+02 8.296e+02, threshold=5.695e+02, percent-clipped=3.0 2023-05-01 08:03:18,674 INFO [train.py:904] (6/8) Epoch 21, batch 8300, loss[loss=0.1743, simple_loss=0.271, pruned_loss=0.03883, over 15289.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.283, pruned_loss=0.05371, over 3042843.59 frames. ], batch size: 191, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:03:51,795 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=211322.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:04:12,349 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=211336.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:04:15,982 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=211338.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:04:18,540 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=211340.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:04:36,926 INFO [train.py:904] (6/8) Epoch 21, batch 8350, loss[loss=0.2106, simple_loss=0.3082, pruned_loss=0.05653, over 15256.00 frames. ], tot_loss[loss=0.1935, simple_loss=0.283, pruned_loss=0.05201, over 3049728.74 frames. ], batch size: 190, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:05:05,460 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=211370.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:05:30,342 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=211386.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:05:30,468 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=211386.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:05:43,485 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.192e+02 2.619e+02 3.085e+02 8.186e+02, threshold=5.238e+02, percent-clipped=3.0 2023-05-01 08:05:55,504 INFO [train.py:904] (6/8) Epoch 21, batch 8400, loss[loss=0.1631, simple_loss=0.2583, pruned_loss=0.03397, over 16249.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2802, pruned_loss=0.0502, over 3020063.79 frames. ], batch size: 165, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:06:10,919 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=211412.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:06:22,626 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=211420.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:07:02,208 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2203, 4.0251, 4.3047, 4.4229, 4.6030, 4.1703, 4.5379, 4.6123], device='cuda:6'), covar=tensor([0.1793, 0.1296, 0.1583, 0.0838, 0.0539, 0.1133, 0.0709, 0.0708], device='cuda:6'), in_proj_covar=tensor([0.0604, 0.0747, 0.0868, 0.0762, 0.0573, 0.0603, 0.0621, 0.0718], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:07:04,484 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=211447.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:07:09,261 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 08:07:10,964 INFO [train.py:904] (6/8) Epoch 21, batch 8450, loss[loss=0.1738, simple_loss=0.2687, pruned_loss=0.03946, over 16716.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2782, pruned_loss=0.04819, over 3041870.80 frames. ], batch size: 124, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:07:36,211 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=211468.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:07:45,055 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=211473.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:08:18,590 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.143e+02 2.501e+02 3.093e+02 5.428e+02, threshold=5.001e+02, percent-clipped=1.0 2023-05-01 08:08:29,989 INFO [train.py:904] (6/8) Epoch 21, batch 8500, loss[loss=0.1768, simple_loss=0.2614, pruned_loss=0.04612, over 16138.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2741, pruned_loss=0.04574, over 3034479.00 frames. ], batch size: 165, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:08:41,778 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1459, 3.9936, 4.2049, 4.3445, 4.4835, 4.0259, 4.4329, 4.4855], device='cuda:6'), covar=tensor([0.1688, 0.1177, 0.1487, 0.0754, 0.0590, 0.1391, 0.0744, 0.0765], device='cuda:6'), in_proj_covar=tensor([0.0606, 0.0749, 0.0870, 0.0764, 0.0576, 0.0605, 0.0624, 0.0722], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:09:54,385 INFO [train.py:904] (6/8) Epoch 21, batch 8550, loss[loss=0.1667, simple_loss=0.2509, pruned_loss=0.04119, over 11782.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2717, pruned_loss=0.04455, over 3030604.85 frames. ], batch size: 247, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:10:34,216 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8709, 3.3348, 3.5347, 2.0698, 2.9396, 2.2903, 3.3599, 3.4827], device='cuda:6'), covar=tensor([0.0323, 0.0812, 0.0518, 0.2118, 0.0832, 0.1034, 0.0720, 0.0982], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0159, 0.0163, 0.0150, 0.0141, 0.0126, 0.0140, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 08:11:18,416 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.272e+02 2.648e+02 3.114e+02 4.500e+02, threshold=5.297e+02, percent-clipped=0.0 2023-05-01 08:11:28,055 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 08:11:32,286 INFO [train.py:904] (6/8) Epoch 21, batch 8600, loss[loss=0.1854, simple_loss=0.2819, pruned_loss=0.04448, over 16622.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2718, pruned_loss=0.04373, over 3018354.98 frames. ], batch size: 134, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:11:41,100 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8870, 2.8536, 2.4980, 4.5655, 3.1045, 4.2509, 1.6191, 3.1184], device='cuda:6'), covar=tensor([0.1285, 0.0748, 0.1240, 0.0149, 0.0130, 0.0294, 0.1595, 0.0695], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0172, 0.0193, 0.0186, 0.0203, 0.0212, 0.0199, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 08:11:43,229 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5667, 4.8751, 4.6595, 4.6581, 4.3901, 4.3929, 4.3526, 4.9038], device='cuda:6'), covar=tensor([0.1218, 0.0888, 0.0941, 0.0892, 0.0874, 0.1416, 0.1085, 0.0932], device='cuda:6'), in_proj_covar=tensor([0.0646, 0.0779, 0.0651, 0.0596, 0.0496, 0.0510, 0.0657, 0.0616], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:12:00,659 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0976, 2.4165, 2.0802, 2.1105, 2.6943, 2.3693, 2.6300, 2.9355], device='cuda:6'), covar=tensor([0.0152, 0.0408, 0.0510, 0.0516, 0.0302, 0.0413, 0.0191, 0.0255], device='cuda:6'), in_proj_covar=tensor([0.0197, 0.0225, 0.0218, 0.0218, 0.0225, 0.0224, 0.0224, 0.0220], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:12:30,583 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3958, 3.3072, 2.7088, 2.1221, 2.1197, 2.3143, 3.4146, 2.9374], device='cuda:6'), covar=tensor([0.3125, 0.0667, 0.1856, 0.3295, 0.2991, 0.2241, 0.0522, 0.1475], device='cuda:6'), in_proj_covar=tensor([0.0323, 0.0264, 0.0301, 0.0308, 0.0292, 0.0255, 0.0290, 0.0330], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 08:12:48,793 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=211640.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:13:09,685 INFO [train.py:904] (6/8) Epoch 21, batch 8650, loss[loss=0.1725, simple_loss=0.2692, pruned_loss=0.03787, over 12177.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.27, pruned_loss=0.04162, over 3047271.73 frames. ], batch size: 247, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:14:30,913 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=211688.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:14:43,483 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.069e+02 2.535e+02 3.155e+02 6.979e+02, threshold=5.069e+02, percent-clipped=2.0 2023-05-01 08:14:57,319 INFO [train.py:904] (6/8) Epoch 21, batch 8700, loss[loss=0.1684, simple_loss=0.2639, pruned_loss=0.03649, over 12753.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2676, pruned_loss=0.04069, over 3057460.42 frames. ], batch size: 247, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:15:21,115 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7176, 5.0105, 4.7889, 4.7997, 4.5548, 4.5536, 4.4705, 5.0598], device='cuda:6'), covar=tensor([0.1132, 0.0846, 0.0950, 0.0805, 0.0817, 0.1123, 0.1156, 0.0899], device='cuda:6'), in_proj_covar=tensor([0.0647, 0.0781, 0.0653, 0.0598, 0.0498, 0.0510, 0.0658, 0.0617], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:16:03,007 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4760, 4.8770, 4.4163, 4.7181, 4.4169, 4.3792, 4.3902, 4.8945], device='cuda:6'), covar=tensor([0.2445, 0.1571, 0.2379, 0.1420, 0.1583, 0.1962, 0.2265, 0.1700], device='cuda:6'), in_proj_covar=tensor([0.0648, 0.0782, 0.0654, 0.0599, 0.0498, 0.0511, 0.0658, 0.0617], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:16:10,366 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=211742.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:16:31,418 INFO [train.py:904] (6/8) Epoch 21, batch 8750, loss[loss=0.1754, simple_loss=0.2779, pruned_loss=0.03647, over 16916.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2665, pruned_loss=0.03986, over 3041294.29 frames. ], batch size: 109, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:17:13,833 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=211768.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:18:09,820 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.610e+02 2.163e+02 2.564e+02 3.220e+02 5.959e+02, threshold=5.128e+02, percent-clipped=1.0 2023-05-01 08:18:23,762 INFO [train.py:904] (6/8) Epoch 21, batch 8800, loss[loss=0.1763, simple_loss=0.2634, pruned_loss=0.04454, over 12478.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2654, pruned_loss=0.03919, over 3042862.12 frames. ], batch size: 246, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:19:10,029 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9347, 2.2964, 2.2823, 3.1101, 1.8407, 3.2246, 1.7213, 2.6618], device='cuda:6'), covar=tensor([0.1342, 0.0738, 0.1235, 0.0176, 0.0091, 0.0392, 0.1620, 0.0794], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0170, 0.0190, 0.0183, 0.0200, 0.0210, 0.0198, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 08:20:07,833 INFO [train.py:904] (6/8) Epoch 21, batch 8850, loss[loss=0.1858, simple_loss=0.289, pruned_loss=0.04131, over 16233.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2681, pruned_loss=0.0386, over 3039105.74 frames. ], batch size: 165, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:21:38,908 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.506e+02 2.129e+02 2.519e+02 2.912e+02 9.426e+02, threshold=5.038e+02, percent-clipped=1.0 2023-05-01 08:21:54,319 INFO [train.py:904] (6/8) Epoch 21, batch 8900, loss[loss=0.164, simple_loss=0.2596, pruned_loss=0.03419, over 16728.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2681, pruned_loss=0.03787, over 3041103.66 frames. ], batch size: 83, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:23:57,813 INFO [train.py:904] (6/8) Epoch 21, batch 8950, loss[loss=0.1626, simple_loss=0.2607, pruned_loss=0.03224, over 16288.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2676, pruned_loss=0.0378, over 3074932.42 frames. ], batch size: 146, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:24:11,401 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2180, 3.9243, 4.3856, 2.2069, 4.5519, 4.5981, 3.5199, 3.6127], device='cuda:6'), covar=tensor([0.0556, 0.0196, 0.0138, 0.1074, 0.0046, 0.0093, 0.0303, 0.0326], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0104, 0.0092, 0.0134, 0.0076, 0.0118, 0.0124, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 08:25:05,295 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-01 08:25:26,989 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8251, 3.1413, 3.4232, 1.8437, 2.9647, 2.2435, 3.3433, 3.3438], device='cuda:6'), covar=tensor([0.0259, 0.0825, 0.0546, 0.2182, 0.0790, 0.0971, 0.0682, 0.0938], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0157, 0.0162, 0.0149, 0.0141, 0.0126, 0.0139, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 08:25:29,293 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.092e+02 2.341e+02 2.927e+02 4.703e+02, threshold=4.682e+02, percent-clipped=0.0 2023-05-01 08:25:46,899 INFO [train.py:904] (6/8) Epoch 21, batch 9000, loss[loss=0.1638, simple_loss=0.261, pruned_loss=0.03333, over 16415.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2645, pruned_loss=0.0368, over 3064570.20 frames. ], batch size: 147, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:25:46,899 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 08:25:57,432 INFO [train.py:938] (6/8) Epoch 21, validation: loss=0.1457, simple_loss=0.2498, pruned_loss=0.02077, over 944034.00 frames. 2023-05-01 08:25:57,433 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 08:26:21,712 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8163, 3.7595, 3.9269, 3.7668, 3.9272, 4.3057, 3.9662, 3.6884], device='cuda:6'), covar=tensor([0.2003, 0.2375, 0.2307, 0.2560, 0.2707, 0.1715, 0.1721, 0.2672], device='cuda:6'), in_proj_covar=tensor([0.0383, 0.0559, 0.0614, 0.0466, 0.0617, 0.0648, 0.0488, 0.0627], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 08:27:20,314 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=212042.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:27:39,212 INFO [train.py:904] (6/8) Epoch 21, batch 9050, loss[loss=0.1615, simple_loss=0.2504, pruned_loss=0.03634, over 16578.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2649, pruned_loss=0.03741, over 3061270.11 frames. ], batch size: 134, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:28:14,864 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=212068.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:28:58,995 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=212090.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:29:10,423 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.640e+02 2.192e+02 2.458e+02 2.901e+02 5.041e+02, threshold=4.916e+02, percent-clipped=1.0 2023-05-01 08:29:20,826 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9945, 4.2905, 4.1219, 4.1444, 3.8076, 3.8728, 3.9094, 4.2791], device='cuda:6'), covar=tensor([0.1112, 0.0865, 0.0993, 0.0823, 0.0801, 0.1731, 0.0954, 0.0995], device='cuda:6'), in_proj_covar=tensor([0.0644, 0.0782, 0.0652, 0.0596, 0.0497, 0.0508, 0.0657, 0.0613], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:29:26,246 INFO [train.py:904] (6/8) Epoch 21, batch 9100, loss[loss=0.1818, simple_loss=0.2826, pruned_loss=0.04052, over 15283.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2651, pruned_loss=0.03822, over 3059622.59 frames. ], batch size: 191, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:29:30,957 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=212104.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 08:29:53,537 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=212116.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:30:40,912 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 08:31:18,932 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-05-01 08:31:22,738 INFO [train.py:904] (6/8) Epoch 21, batch 9150, loss[loss=0.1677, simple_loss=0.259, pruned_loss=0.03821, over 16847.00 frames. ], tot_loss[loss=0.17, simple_loss=0.265, pruned_loss=0.0375, over 3062723.99 frames. ], batch size: 124, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:31:40,566 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 08:31:45,838 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1436, 2.5143, 2.6208, 1.8757, 2.7634, 2.8583, 2.4941, 2.4591], device='cuda:6'), covar=tensor([0.0655, 0.0236, 0.0217, 0.0992, 0.0114, 0.0231, 0.0432, 0.0421], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0104, 0.0092, 0.0134, 0.0076, 0.0118, 0.0124, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 08:31:52,679 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=212165.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 08:32:05,904 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-05-01 08:32:57,277 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.621e+02 2.079e+02 2.542e+02 3.144e+02 5.396e+02, threshold=5.084e+02, percent-clipped=1.0 2023-05-01 08:33:09,950 INFO [train.py:904] (6/8) Epoch 21, batch 9200, loss[loss=0.1396, simple_loss=0.231, pruned_loss=0.02413, over 17062.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2608, pruned_loss=0.03647, over 3059352.34 frames. ], batch size: 50, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:33:29,143 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2023-05-01 08:34:48,818 INFO [train.py:904] (6/8) Epoch 21, batch 9250, loss[loss=0.1435, simple_loss=0.2312, pruned_loss=0.02785, over 12255.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2608, pruned_loss=0.03654, over 3067938.16 frames. ], batch size: 250, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:34:51,564 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=212253.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:36:25,633 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.441e+02 2.162e+02 2.598e+02 3.159e+02 6.088e+02, threshold=5.195e+02, percent-clipped=3.0 2023-05-01 08:36:39,742 INFO [train.py:904] (6/8) Epoch 21, batch 9300, loss[loss=0.1431, simple_loss=0.2404, pruned_loss=0.02292, over 16935.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2596, pruned_loss=0.03614, over 3069581.18 frames. ], batch size: 102, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:37:07,698 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=212314.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 08:37:28,622 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0218, 2.2373, 1.9789, 2.1408, 2.6202, 2.3301, 2.5776, 2.7987], device='cuda:6'), covar=tensor([0.0195, 0.0449, 0.0512, 0.0479, 0.0282, 0.0424, 0.0212, 0.0280], device='cuda:6'), in_proj_covar=tensor([0.0195, 0.0225, 0.0218, 0.0218, 0.0225, 0.0224, 0.0222, 0.0218], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:37:30,334 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-01 08:38:22,831 INFO [train.py:904] (6/8) Epoch 21, batch 9350, loss[loss=0.1885, simple_loss=0.2783, pruned_loss=0.0494, over 15360.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2596, pruned_loss=0.03603, over 3074610.69 frames. ], batch size: 191, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:38:25,889 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4045, 4.5485, 4.6944, 4.4269, 4.5519, 5.0431, 4.5657, 4.2352], device='cuda:6'), covar=tensor([0.1362, 0.1788, 0.1958, 0.2144, 0.2356, 0.0949, 0.1550, 0.2456], device='cuda:6'), in_proj_covar=tensor([0.0381, 0.0561, 0.0618, 0.0466, 0.0616, 0.0651, 0.0488, 0.0625], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 08:39:47,854 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.212e+02 2.571e+02 3.078e+02 5.703e+02, threshold=5.142e+02, percent-clipped=1.0 2023-05-01 08:40:02,993 INFO [train.py:904] (6/8) Epoch 21, batch 9400, loss[loss=0.1814, simple_loss=0.2815, pruned_loss=0.04067, over 16106.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2594, pruned_loss=0.03561, over 3080603.49 frames. ], batch size: 165, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:41:18,510 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=212439.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:41:42,917 INFO [train.py:904] (6/8) Epoch 21, batch 9450, loss[loss=0.1772, simple_loss=0.2763, pruned_loss=0.03905, over 16339.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2611, pruned_loss=0.03575, over 3073994.43 frames. ], batch size: 146, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:41:58,733 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=212460.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 08:43:11,247 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 1.974e+02 2.463e+02 3.006e+02 5.675e+02, threshold=4.927e+02, percent-clipped=1.0 2023-05-01 08:43:21,176 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=212500.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:43:25,062 INFO [train.py:904] (6/8) Epoch 21, batch 9500, loss[loss=0.1578, simple_loss=0.2562, pruned_loss=0.02972, over 16864.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2608, pruned_loss=0.03541, over 3100792.73 frames. ], batch size: 96, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:44:22,790 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5306, 4.6637, 4.4385, 4.1015, 4.1438, 4.5600, 4.3371, 4.2392], device='cuda:6'), covar=tensor([0.0606, 0.0489, 0.0334, 0.0311, 0.0867, 0.0466, 0.0556, 0.0687], device='cuda:6'), in_proj_covar=tensor([0.0274, 0.0396, 0.0321, 0.0316, 0.0327, 0.0367, 0.0222, 0.0383], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-05-01 08:45:12,071 INFO [train.py:904] (6/8) Epoch 21, batch 9550, loss[loss=0.1897, simple_loss=0.2853, pruned_loss=0.04705, over 16759.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2606, pruned_loss=0.03551, over 3107155.91 frames. ], batch size: 134, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:45:43,446 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3225, 5.3039, 5.0193, 4.6473, 5.1724, 2.1312, 4.8739, 4.8786], device='cuda:6'), covar=tensor([0.0080, 0.0083, 0.0202, 0.0268, 0.0097, 0.2379, 0.0122, 0.0177], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0145, 0.0185, 0.0164, 0.0164, 0.0197, 0.0173, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:46:39,908 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.109e+02 2.597e+02 3.170e+02 5.717e+02, threshold=5.194e+02, percent-clipped=4.0 2023-05-01 08:46:50,960 INFO [train.py:904] (6/8) Epoch 21, batch 9600, loss[loss=0.1854, simple_loss=0.2865, pruned_loss=0.04217, over 16638.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2615, pruned_loss=0.03624, over 3069039.39 frames. ], batch size: 134, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:47:06,245 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=212609.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 08:48:38,608 INFO [train.py:904] (6/8) Epoch 21, batch 9650, loss[loss=0.1725, simple_loss=0.267, pruned_loss=0.03899, over 16645.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2634, pruned_loss=0.03686, over 3047628.98 frames. ], batch size: 134, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:50:11,919 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.547e+02 2.239e+02 2.609e+02 3.201e+02 8.250e+02, threshold=5.217e+02, percent-clipped=3.0 2023-05-01 08:50:21,241 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=212698.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:50:28,029 INFO [train.py:904] (6/8) Epoch 21, batch 9700, loss[loss=0.1697, simple_loss=0.2693, pruned_loss=0.035, over 15286.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2634, pruned_loss=0.03699, over 3060755.77 frames. ], batch size: 191, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:50:46,062 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=212711.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:51:43,507 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-01 08:52:10,270 INFO [train.py:904] (6/8) Epoch 21, batch 9750, loss[loss=0.1689, simple_loss=0.2541, pruned_loss=0.0418, over 12289.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2625, pruned_loss=0.03727, over 3058999.83 frames. ], batch size: 248, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:52:21,296 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9853, 5.2476, 5.0650, 5.0618, 4.8093, 4.7008, 4.6121, 5.3360], device='cuda:6'), covar=tensor([0.1295, 0.0956, 0.0973, 0.0839, 0.0836, 0.1000, 0.1300, 0.0995], device='cuda:6'), in_proj_covar=tensor([0.0638, 0.0776, 0.0644, 0.0588, 0.0494, 0.0503, 0.0648, 0.0609], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:52:25,404 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=212759.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:52:27,195 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=212760.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 08:52:49,269 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=212772.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:52:53,755 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=212774.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:53:36,933 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 2.010e+02 2.340e+02 2.880e+02 4.703e+02, threshold=4.680e+02, percent-clipped=0.0 2023-05-01 08:53:37,559 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=212795.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:53:46,219 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9226, 4.2331, 4.0575, 4.0517, 3.7698, 3.8340, 3.8353, 4.2087], device='cuda:6'), covar=tensor([0.1184, 0.0845, 0.0984, 0.0851, 0.0830, 0.1638, 0.0976, 0.1040], device='cuda:6'), in_proj_covar=tensor([0.0640, 0.0778, 0.0646, 0.0589, 0.0496, 0.0505, 0.0650, 0.0610], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 08:53:48,982 INFO [train.py:904] (6/8) Epoch 21, batch 9800, loss[loss=0.1832, simple_loss=0.2751, pruned_loss=0.04571, over 16938.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2623, pruned_loss=0.03651, over 3075725.52 frames. ], batch size: 116, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:54:01,428 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=212808.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 08:54:33,841 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 08:54:51,510 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8401, 3.8738, 4.1379, 4.1150, 4.1327, 3.9035, 3.8912, 3.9556], device='cuda:6'), covar=tensor([0.0349, 0.0760, 0.0441, 0.0442, 0.0483, 0.0489, 0.0846, 0.0436], device='cuda:6'), in_proj_covar=tensor([0.0384, 0.0428, 0.0415, 0.0387, 0.0462, 0.0436, 0.0513, 0.0350], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 08:54:51,602 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=212835.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:55:10,511 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2782, 2.9770, 3.1152, 1.7374, 3.2987, 3.3950, 2.7646, 2.7141], device='cuda:6'), covar=tensor([0.0704, 0.0260, 0.0198, 0.1198, 0.0086, 0.0159, 0.0439, 0.0390], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0103, 0.0090, 0.0133, 0.0076, 0.0117, 0.0123, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 08:55:31,550 INFO [train.py:904] (6/8) Epoch 21, batch 9850, loss[loss=0.1793, simple_loss=0.2615, pruned_loss=0.04853, over 12088.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.263, pruned_loss=0.03639, over 3057838.84 frames. ], batch size: 248, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:57:07,909 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.176e+02 2.812e+02 3.349e+02 5.684e+02, threshold=5.624e+02, percent-clipped=7.0 2023-05-01 08:57:21,369 INFO [train.py:904] (6/8) Epoch 21, batch 9900, loss[loss=0.1688, simple_loss=0.2841, pruned_loss=0.02674, over 16872.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2633, pruned_loss=0.03643, over 3037419.55 frames. ], batch size: 102, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:57:38,626 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=212909.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:58:49,802 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-05-01 08:59:17,571 INFO [train.py:904] (6/8) Epoch 21, batch 9950, loss[loss=0.1816, simple_loss=0.279, pruned_loss=0.04209, over 16336.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2657, pruned_loss=0.03683, over 3041975.01 frames. ], batch size: 146, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:59:29,578 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=212957.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:00:58,645 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.203e+02 2.698e+02 3.794e+02 1.048e+03, threshold=5.396e+02, percent-clipped=2.0 2023-05-01 09:01:17,069 INFO [train.py:904] (6/8) Epoch 21, batch 10000, loss[loss=0.1709, simple_loss=0.2746, pruned_loss=0.03364, over 16193.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2643, pruned_loss=0.03621, over 3058908.63 frames. ], batch size: 165, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 09:01:51,274 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3432, 3.0137, 2.6512, 2.2066, 2.1440, 2.2755, 3.0336, 2.7768], device='cuda:6'), covar=tensor([0.2803, 0.0618, 0.1667, 0.2610, 0.2665, 0.2193, 0.0493, 0.1260], device='cuda:6'), in_proj_covar=tensor([0.0315, 0.0258, 0.0295, 0.0300, 0.0280, 0.0249, 0.0282, 0.0319], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 09:02:58,011 INFO [train.py:904] (6/8) Epoch 21, batch 10050, loss[loss=0.1853, simple_loss=0.2832, pruned_loss=0.04373, over 15522.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.264, pruned_loss=0.03586, over 3064296.72 frames. ], batch size: 192, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 09:03:02,982 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=213054.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:03:28,731 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=213067.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:04:21,049 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 2.063e+02 2.495e+02 3.076e+02 5.792e+02, threshold=4.990e+02, percent-clipped=2.0 2023-05-01 09:04:21,597 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=213095.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:04:32,167 INFO [train.py:904] (6/8) Epoch 21, batch 10100, loss[loss=0.1645, simple_loss=0.2534, pruned_loss=0.03778, over 16898.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2645, pruned_loss=0.03635, over 3063542.26 frames. ], batch size: 116, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 09:05:27,296 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=213130.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:05:42,671 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=213143.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:05:51,223 INFO [train.py:904] (6/8) Epoch 21, batch 10150, loss[loss=0.1509, simple_loss=0.2378, pruned_loss=0.03196, over 12599.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2636, pruned_loss=0.03641, over 3056831.44 frames. ], batch size: 248, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 09:06:16,171 INFO [train.py:904] (6/8) Epoch 22, batch 0, loss[loss=0.2265, simple_loss=0.3192, pruned_loss=0.06694, over 16757.00 frames. ], tot_loss[loss=0.2265, simple_loss=0.3192, pruned_loss=0.06694, over 16757.00 frames. ], batch size: 57, lr: 3.10e-03, grad_scale: 8.0 2023-05-01 09:06:16,171 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 09:06:23,635 INFO [train.py:938] (6/8) Epoch 22, validation: loss=0.1457, simple_loss=0.2489, pruned_loss=0.0212, over 944034.00 frames. 2023-05-01 09:06:23,635 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 09:07:26,339 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.831e+02 2.320e+02 2.870e+02 3.515e+02 6.456e+02, threshold=5.740e+02, percent-clipped=5.0 2023-05-01 09:07:34,021 INFO [train.py:904] (6/8) Epoch 22, batch 50, loss[loss=0.1681, simple_loss=0.2604, pruned_loss=0.03791, over 16766.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2721, pruned_loss=0.05061, over 746078.26 frames. ], batch size: 57, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:08:41,847 INFO [train.py:904] (6/8) Epoch 22, batch 100, loss[loss=0.1537, simple_loss=0.2365, pruned_loss=0.03543, over 16799.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2673, pruned_loss=0.04708, over 1319548.15 frames. ], batch size: 39, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:09:44,707 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.194e+02 2.643e+02 3.119e+02 6.629e+02, threshold=5.286e+02, percent-clipped=1.0 2023-05-01 09:09:51,952 INFO [train.py:904] (6/8) Epoch 22, batch 150, loss[loss=0.2026, simple_loss=0.2807, pruned_loss=0.0622, over 16692.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.264, pruned_loss=0.04522, over 1772290.32 frames. ], batch size: 134, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:11:00,679 INFO [train.py:904] (6/8) Epoch 22, batch 200, loss[loss=0.1695, simple_loss=0.2464, pruned_loss=0.04632, over 16751.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2647, pruned_loss=0.04628, over 2115940.72 frames. ], batch size: 83, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:11:02,275 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=213354.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:11:15,191 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-05-01 09:11:19,275 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5602, 4.5512, 4.4501, 3.9184, 4.4727, 1.6986, 4.1953, 4.1009], device='cuda:6'), covar=tensor([0.0153, 0.0107, 0.0186, 0.0303, 0.0116, 0.2877, 0.0164, 0.0255], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0147, 0.0187, 0.0164, 0.0166, 0.0199, 0.0176, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:11:20,449 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=213367.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:11:55,838 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.88 vs. limit=5.0 2023-05-01 09:12:00,918 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.471e+02 2.291e+02 2.770e+02 3.341e+02 7.349e+02, threshold=5.539e+02, percent-clipped=1.0 2023-05-01 09:12:07,194 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=213402.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:12:08,109 INFO [train.py:904] (6/8) Epoch 22, batch 250, loss[loss=0.1514, simple_loss=0.2362, pruned_loss=0.03331, over 15849.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2628, pruned_loss=0.04583, over 2385474.04 frames. ], batch size: 35, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:12:25,570 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=213415.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:12:46,418 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=213430.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:13:01,713 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-01 09:13:17,487 INFO [train.py:904] (6/8) Epoch 22, batch 300, loss[loss=0.171, simple_loss=0.2641, pruned_loss=0.03891, over 16607.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2599, pruned_loss=0.0441, over 2595195.30 frames. ], batch size: 68, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:13:31,952 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-01 09:13:52,063 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=213478.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:14:04,034 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4614, 4.3025, 4.4881, 4.6915, 4.8161, 4.4264, 4.7518, 4.8118], device='cuda:6'), covar=tensor([0.1854, 0.1424, 0.1801, 0.0904, 0.0762, 0.1049, 0.1509, 0.0781], device='cuda:6'), in_proj_covar=tensor([0.0614, 0.0757, 0.0880, 0.0769, 0.0582, 0.0611, 0.0634, 0.0730], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:14:19,526 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.474e+02 2.218e+02 2.593e+02 2.976e+02 4.668e+02, threshold=5.187e+02, percent-clipped=0.0 2023-05-01 09:14:25,241 INFO [train.py:904] (6/8) Epoch 22, batch 350, loss[loss=0.1662, simple_loss=0.2416, pruned_loss=0.04543, over 16880.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2575, pruned_loss=0.0432, over 2759106.59 frames. ], batch size: 96, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:15:13,689 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1853, 6.0069, 6.0594, 5.7781, 5.9037, 6.4515, 5.9036, 5.5762], device='cuda:6'), covar=tensor([0.0961, 0.1829, 0.2187, 0.2105, 0.2672, 0.1027, 0.1576, 0.2222], device='cuda:6'), in_proj_covar=tensor([0.0405, 0.0588, 0.0652, 0.0491, 0.0650, 0.0685, 0.0513, 0.0655], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 09:15:34,171 INFO [train.py:904] (6/8) Epoch 22, batch 400, loss[loss=0.1818, simple_loss=0.2634, pruned_loss=0.05011, over 16149.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2563, pruned_loss=0.04269, over 2891886.57 frames. ], batch size: 165, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:15:52,126 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=213566.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:16:03,834 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2820, 2.3194, 2.4575, 4.0479, 2.3490, 2.6739, 2.3981, 2.5018], device='cuda:6'), covar=tensor([0.1393, 0.3616, 0.2851, 0.0654, 0.3838, 0.2512, 0.3751, 0.3024], device='cuda:6'), in_proj_covar=tensor([0.0397, 0.0445, 0.0367, 0.0326, 0.0437, 0.0509, 0.0416, 0.0521], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:16:36,629 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.747e+02 2.265e+02 2.651e+02 3.318e+02 1.993e+03, threshold=5.302e+02, percent-clipped=3.0 2023-05-01 09:16:43,116 INFO [train.py:904] (6/8) Epoch 22, batch 450, loss[loss=0.1796, simple_loss=0.2578, pruned_loss=0.05068, over 12155.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2551, pruned_loss=0.04262, over 2974975.88 frames. ], batch size: 247, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:16:59,441 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.52 vs. limit=5.0 2023-05-01 09:17:16,820 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=213627.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:17:22,747 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1169, 3.9075, 4.4441, 2.2531, 4.6505, 4.7240, 3.3752, 3.6671], device='cuda:6'), covar=tensor([0.0684, 0.0256, 0.0200, 0.1169, 0.0066, 0.0157, 0.0423, 0.0384], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0107, 0.0095, 0.0139, 0.0079, 0.0124, 0.0128, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 09:17:41,464 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 09:17:47,451 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-01 09:17:48,267 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7502, 3.7395, 4.0967, 1.9476, 4.2489, 4.4445, 3.1710, 3.3991], device='cuda:6'), covar=tensor([0.0859, 0.0261, 0.0282, 0.1366, 0.0123, 0.0191, 0.0485, 0.0455], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0107, 0.0095, 0.0139, 0.0079, 0.0124, 0.0128, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 09:17:53,028 INFO [train.py:904] (6/8) Epoch 22, batch 500, loss[loss=0.1628, simple_loss=0.2603, pruned_loss=0.03263, over 17101.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.253, pruned_loss=0.04159, over 3057902.65 frames. ], batch size: 47, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:18:28,377 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7146, 2.8858, 2.7768, 5.1350, 4.2984, 4.5534, 1.6214, 3.4299], device='cuda:6'), covar=tensor([0.1422, 0.0810, 0.1260, 0.0210, 0.0229, 0.0413, 0.1743, 0.0721], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0173, 0.0193, 0.0187, 0.0201, 0.0213, 0.0202, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 09:18:54,885 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.471e+02 2.187e+02 2.448e+02 2.877e+02 6.308e+02, threshold=4.896e+02, percent-clipped=1.0 2023-05-01 09:19:01,577 INFO [train.py:904] (6/8) Epoch 22, batch 550, loss[loss=0.1458, simple_loss=0.2315, pruned_loss=0.03002, over 16865.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2518, pruned_loss=0.04088, over 3121575.61 frames. ], batch size: 42, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:20:10,672 INFO [train.py:904] (6/8) Epoch 22, batch 600, loss[loss=0.1692, simple_loss=0.2635, pruned_loss=0.03744, over 17064.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2517, pruned_loss=0.04131, over 3172600.69 frames. ], batch size: 55, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:20:20,352 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0568, 2.2531, 2.7294, 2.9800, 2.8639, 3.4322, 2.5507, 3.3850], device='cuda:6'), covar=tensor([0.0282, 0.0506, 0.0323, 0.0349, 0.0346, 0.0208, 0.0462, 0.0209], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0191, 0.0177, 0.0182, 0.0196, 0.0151, 0.0194, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:20:49,096 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5985, 2.4542, 2.4285, 4.4858, 2.3635, 2.8393, 2.4706, 2.5932], device='cuda:6'), covar=tensor([0.1221, 0.3638, 0.3070, 0.0468, 0.4155, 0.2646, 0.3696, 0.3721], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0445, 0.0367, 0.0327, 0.0437, 0.0509, 0.0417, 0.0522], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:21:13,515 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.232e+02 2.565e+02 3.000e+02 5.415e+02, threshold=5.129e+02, percent-clipped=2.0 2023-05-01 09:21:21,578 INFO [train.py:904] (6/8) Epoch 22, batch 650, loss[loss=0.1752, simple_loss=0.2533, pruned_loss=0.04855, over 16710.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2507, pruned_loss=0.04097, over 3207678.88 frames. ], batch size: 124, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:21:51,285 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8796, 2.8634, 2.6804, 4.5081, 3.7193, 4.2817, 1.7402, 3.0301], device='cuda:6'), covar=tensor([0.1397, 0.0718, 0.1122, 0.0177, 0.0163, 0.0382, 0.1562, 0.0823], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0173, 0.0193, 0.0187, 0.0201, 0.0214, 0.0201, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 09:22:08,155 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7112, 2.5604, 2.3793, 3.9784, 3.1825, 3.9925, 1.5974, 2.7910], device='cuda:6'), covar=tensor([0.1529, 0.0810, 0.1332, 0.0194, 0.0162, 0.0381, 0.1749, 0.0893], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0173, 0.0193, 0.0187, 0.0200, 0.0213, 0.0201, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 09:22:16,661 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.53 vs. limit=5.0 2023-05-01 09:22:30,249 INFO [train.py:904] (6/8) Epoch 22, batch 700, loss[loss=0.1612, simple_loss=0.2488, pruned_loss=0.03677, over 17235.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2515, pruned_loss=0.04093, over 3240145.62 frames. ], batch size: 44, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:23:28,135 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1284, 4.6249, 4.6332, 3.3765, 3.8196, 4.5313, 4.0569, 2.9504], device='cuda:6'), covar=tensor([0.0421, 0.0050, 0.0039, 0.0321, 0.0125, 0.0084, 0.0089, 0.0388], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0083, 0.0083, 0.0135, 0.0098, 0.0108, 0.0094, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 09:23:35,483 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.408e+02 2.030e+02 2.447e+02 3.057e+02 6.587e+02, threshold=4.894e+02, percent-clipped=3.0 2023-05-01 09:23:41,788 INFO [train.py:904] (6/8) Epoch 22, batch 750, loss[loss=0.1654, simple_loss=0.251, pruned_loss=0.0399, over 16481.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2512, pruned_loss=0.04062, over 3254744.11 frames. ], batch size: 68, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:24:08,342 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=213922.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:24:31,648 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9396, 2.1675, 2.4855, 2.8742, 2.8459, 2.9177, 2.0611, 3.0836], device='cuda:6'), covar=tensor([0.0176, 0.0443, 0.0335, 0.0267, 0.0291, 0.0272, 0.0535, 0.0150], device='cuda:6'), in_proj_covar=tensor([0.0186, 0.0191, 0.0177, 0.0181, 0.0195, 0.0151, 0.0193, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:24:53,399 INFO [train.py:904] (6/8) Epoch 22, batch 800, loss[loss=0.1463, simple_loss=0.232, pruned_loss=0.03032, over 16848.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.251, pruned_loss=0.04065, over 3273842.45 frames. ], batch size: 42, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:25:04,744 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1279, 5.0744, 4.9702, 4.4738, 4.6136, 4.9931, 4.9822, 4.6481], device='cuda:6'), covar=tensor([0.0586, 0.0550, 0.0329, 0.0359, 0.1147, 0.0482, 0.0376, 0.0822], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0422, 0.0339, 0.0337, 0.0347, 0.0392, 0.0233, 0.0408], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:25:40,740 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.53 vs. limit=5.0 2023-05-01 09:25:56,893 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.094e+02 2.441e+02 2.760e+02 5.264e+02, threshold=4.882e+02, percent-clipped=1.0 2023-05-01 09:26:06,616 INFO [train.py:904] (6/8) Epoch 22, batch 850, loss[loss=0.1647, simple_loss=0.2501, pruned_loss=0.03965, over 16898.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2511, pruned_loss=0.04011, over 3292283.24 frames. ], batch size: 96, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:26:20,153 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1060, 3.9462, 4.1737, 4.2957, 4.3844, 3.9828, 4.1696, 4.3887], device='cuda:6'), covar=tensor([0.1601, 0.1198, 0.1301, 0.0678, 0.0576, 0.1351, 0.2631, 0.0720], device='cuda:6'), in_proj_covar=tensor([0.0639, 0.0786, 0.0916, 0.0800, 0.0604, 0.0635, 0.0658, 0.0763], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:27:07,739 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0139, 5.5163, 5.5808, 5.3913, 5.3172, 6.0303, 5.4893, 5.2428], device='cuda:6'), covar=tensor([0.1046, 0.2105, 0.2470, 0.1999, 0.3228, 0.1045, 0.1538, 0.2266], device='cuda:6'), in_proj_covar=tensor([0.0409, 0.0599, 0.0660, 0.0498, 0.0660, 0.0694, 0.0519, 0.0662], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 09:27:17,240 INFO [train.py:904] (6/8) Epoch 22, batch 900, loss[loss=0.1612, simple_loss=0.2457, pruned_loss=0.03831, over 17207.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2508, pruned_loss=0.03965, over 3303868.12 frames. ], batch size: 44, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:27:57,841 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1494, 4.1345, 4.0705, 3.4323, 4.1081, 1.6996, 3.8645, 3.5970], device='cuda:6'), covar=tensor([0.0170, 0.0132, 0.0220, 0.0329, 0.0121, 0.2923, 0.0176, 0.0295], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0154, 0.0196, 0.0174, 0.0174, 0.0207, 0.0185, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:28:19,799 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.353e+02 1.995e+02 2.382e+02 2.694e+02 6.707e+02, threshold=4.763e+02, percent-clipped=1.0 2023-05-01 09:28:27,535 INFO [train.py:904] (6/8) Epoch 22, batch 950, loss[loss=0.1701, simple_loss=0.2485, pruned_loss=0.04584, over 16426.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2498, pruned_loss=0.03959, over 3307675.43 frames. ], batch size: 146, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:29:03,298 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=214129.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:29:35,585 INFO [train.py:904] (6/8) Epoch 22, batch 1000, loss[loss=0.1551, simple_loss=0.2363, pruned_loss=0.03695, over 16454.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2483, pruned_loss=0.03918, over 3308889.35 frames. ], batch size: 68, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:30:29,078 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=214190.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:30:39,708 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.141e+02 2.430e+02 2.890e+02 5.407e+02, threshold=4.860e+02, percent-clipped=2.0 2023-05-01 09:30:46,635 INFO [train.py:904] (6/8) Epoch 22, batch 1050, loss[loss=0.1577, simple_loss=0.2562, pruned_loss=0.02962, over 17112.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2483, pruned_loss=0.03916, over 3314334.67 frames. ], batch size: 47, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:31:07,806 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=214218.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:31:13,753 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=214222.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:31:56,091 INFO [train.py:904] (6/8) Epoch 22, batch 1100, loss[loss=0.1502, simple_loss=0.2443, pruned_loss=0.02806, over 17214.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2479, pruned_loss=0.03899, over 3315448.21 frames. ], batch size: 46, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:32:04,146 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-05-01 09:32:19,738 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=214270.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:32:31,820 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=214279.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:32:57,752 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.499e+02 2.112e+02 2.466e+02 2.921e+02 8.908e+02, threshold=4.932e+02, percent-clipped=7.0 2023-05-01 09:33:03,839 INFO [train.py:904] (6/8) Epoch 22, batch 1150, loss[loss=0.1729, simple_loss=0.2667, pruned_loss=0.03949, over 16635.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2481, pruned_loss=0.03908, over 3320604.11 frames. ], batch size: 57, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:33:21,520 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8586, 4.4622, 3.2148, 2.3812, 2.8271, 2.7424, 4.7344, 3.7238], device='cuda:6'), covar=tensor([0.2827, 0.0522, 0.1720, 0.2790, 0.2783, 0.2000, 0.0315, 0.1297], device='cuda:6'), in_proj_covar=tensor([0.0326, 0.0270, 0.0306, 0.0312, 0.0296, 0.0260, 0.0295, 0.0337], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 09:34:15,635 INFO [train.py:904] (6/8) Epoch 22, batch 1200, loss[loss=0.1744, simple_loss=0.2481, pruned_loss=0.05032, over 16474.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2473, pruned_loss=0.03872, over 3314339.14 frames. ], batch size: 68, lr: 3.10e-03, grad_scale: 8.0 2023-05-01 09:35:18,108 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 2.085e+02 2.414e+02 2.922e+02 4.636e+02, threshold=4.828e+02, percent-clipped=0.0 2023-05-01 09:35:25,088 INFO [train.py:904] (6/8) Epoch 22, batch 1250, loss[loss=0.167, simple_loss=0.2596, pruned_loss=0.03717, over 17105.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.248, pruned_loss=0.03875, over 3318201.00 frames. ], batch size: 49, lr: 3.10e-03, grad_scale: 8.0 2023-05-01 09:36:35,039 INFO [train.py:904] (6/8) Epoch 22, batch 1300, loss[loss=0.1808, simple_loss=0.2536, pruned_loss=0.05398, over 16873.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2483, pruned_loss=0.03919, over 3312741.30 frames. ], batch size: 96, lr: 3.10e-03, grad_scale: 8.0 2023-05-01 09:37:12,577 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5429, 3.5142, 3.4597, 2.7994, 3.3217, 2.1263, 3.1196, 2.7503], device='cuda:6'), covar=tensor([0.0145, 0.0139, 0.0185, 0.0230, 0.0108, 0.2242, 0.0144, 0.0247], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0156, 0.0199, 0.0176, 0.0177, 0.0208, 0.0188, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:37:18,479 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=214485.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:37:34,614 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.092e+02 2.483e+02 2.893e+02 4.900e+02, threshold=4.965e+02, percent-clipped=1.0 2023-05-01 09:37:42,345 INFO [train.py:904] (6/8) Epoch 22, batch 1350, loss[loss=0.1705, simple_loss=0.2586, pruned_loss=0.04117, over 17037.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2485, pruned_loss=0.03923, over 3303182.91 frames. ], batch size: 53, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:38:48,216 INFO [train.py:904] (6/8) Epoch 22, batch 1400, loss[loss=0.1509, simple_loss=0.2258, pruned_loss=0.03801, over 16883.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2483, pruned_loss=0.03931, over 3310288.61 frames. ], batch size: 90, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:39:17,609 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=214574.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:39:50,086 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.086e+02 2.334e+02 2.820e+02 4.887e+02, threshold=4.669e+02, percent-clipped=0.0 2023-05-01 09:39:57,055 INFO [train.py:904] (6/8) Epoch 22, batch 1450, loss[loss=0.1524, simple_loss=0.2352, pruned_loss=0.03484, over 16366.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2474, pruned_loss=0.0393, over 3310743.71 frames. ], batch size: 36, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:41:07,194 INFO [train.py:904] (6/8) Epoch 22, batch 1500, loss[loss=0.1778, simple_loss=0.2485, pruned_loss=0.05357, over 16906.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2469, pruned_loss=0.03941, over 3317575.17 frames. ], batch size: 90, lr: 3.09e-03, grad_scale: 4.0 2023-05-01 09:41:09,957 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1676, 4.1393, 4.1375, 3.5747, 4.1521, 1.8259, 3.9247, 3.7797], device='cuda:6'), covar=tensor([0.0177, 0.0144, 0.0182, 0.0315, 0.0118, 0.2703, 0.0169, 0.0226], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0156, 0.0199, 0.0177, 0.0177, 0.0209, 0.0188, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:41:48,007 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-05-01 09:42:09,954 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-05-01 09:42:10,190 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.149e+02 2.472e+02 3.063e+02 5.864e+02, threshold=4.945e+02, percent-clipped=3.0 2023-05-01 09:42:16,353 INFO [train.py:904] (6/8) Epoch 22, batch 1550, loss[loss=0.1769, simple_loss=0.2706, pruned_loss=0.04159, over 16718.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2481, pruned_loss=0.04069, over 3310270.11 frames. ], batch size: 62, lr: 3.09e-03, grad_scale: 4.0 2023-05-01 09:43:28,003 INFO [train.py:904] (6/8) Epoch 22, batch 1600, loss[loss=0.1691, simple_loss=0.2572, pruned_loss=0.04053, over 15971.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2508, pruned_loss=0.04124, over 3319085.01 frames. ], batch size: 35, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:44:12,524 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=214785.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:44:32,065 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.512e+02 2.192e+02 2.616e+02 3.002e+02 5.148e+02, threshold=5.232e+02, percent-clipped=2.0 2023-05-01 09:44:37,869 INFO [train.py:904] (6/8) Epoch 22, batch 1650, loss[loss=0.1676, simple_loss=0.2583, pruned_loss=0.0384, over 17084.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2521, pruned_loss=0.04143, over 3309537.74 frames. ], batch size: 53, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:45:20,795 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=214833.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:45:34,975 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 09:45:47,643 INFO [train.py:904] (6/8) Epoch 22, batch 1700, loss[loss=0.1723, simple_loss=0.2488, pruned_loss=0.04789, over 16700.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2533, pruned_loss=0.04185, over 3315512.04 frames. ], batch size: 76, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:46:18,494 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=214874.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:46:23,075 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-05-01 09:46:30,912 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0494, 4.0494, 3.9475, 3.6667, 3.7437, 4.0500, 3.6866, 3.8364], device='cuda:6'), covar=tensor([0.0639, 0.0640, 0.0327, 0.0307, 0.0670, 0.0471, 0.1007, 0.0559], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0438, 0.0353, 0.0351, 0.0362, 0.0406, 0.0241, 0.0422], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:46:39,720 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3265, 5.2910, 5.1592, 4.6428, 4.7958, 5.2148, 5.1298, 4.7884], device='cuda:6'), covar=tensor([0.0636, 0.0510, 0.0351, 0.0385, 0.1221, 0.0492, 0.0357, 0.0795], device='cuda:6'), in_proj_covar=tensor([0.0301, 0.0438, 0.0353, 0.0350, 0.0361, 0.0406, 0.0241, 0.0421], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:46:53,104 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.696e+02 2.472e+02 2.965e+02 3.692e+02 1.718e+03, threshold=5.930e+02, percent-clipped=4.0 2023-05-01 09:46:58,548 INFO [train.py:904] (6/8) Epoch 22, batch 1750, loss[loss=0.1892, simple_loss=0.2705, pruned_loss=0.05402, over 16568.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2539, pruned_loss=0.04174, over 3318865.20 frames. ], batch size: 75, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:47:01,135 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 09:47:25,705 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=214922.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:48:06,729 INFO [train.py:904] (6/8) Epoch 22, batch 1800, loss[loss=0.1472, simple_loss=0.2362, pruned_loss=0.02913, over 17200.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2556, pruned_loss=0.04205, over 3320527.75 frames. ], batch size: 44, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:49:13,316 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.476e+02 2.061e+02 2.528e+02 2.863e+02 4.992e+02, threshold=5.055e+02, percent-clipped=0.0 2023-05-01 09:49:18,204 INFO [train.py:904] (6/8) Epoch 22, batch 1850, loss[loss=0.1844, simple_loss=0.2821, pruned_loss=0.04338, over 16653.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2566, pruned_loss=0.04186, over 3325629.45 frames. ], batch size: 62, lr: 3.09e-03, grad_scale: 4.0 2023-05-01 09:49:50,204 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3378, 3.5613, 3.8800, 2.0765, 3.2166, 2.5417, 3.8357, 3.7069], device='cuda:6'), covar=tensor([0.0280, 0.0934, 0.0531, 0.2090, 0.0765, 0.0981, 0.0569, 0.1099], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0163, 0.0166, 0.0154, 0.0144, 0.0130, 0.0143, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 09:50:04,389 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.08 vs. limit=5.0 2023-05-01 09:50:27,231 INFO [train.py:904] (6/8) Epoch 22, batch 1900, loss[loss=0.1721, simple_loss=0.2557, pruned_loss=0.04427, over 16417.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.256, pruned_loss=0.04105, over 3329939.87 frames. ], batch size: 146, lr: 3.09e-03, grad_scale: 4.0 2023-05-01 09:51:31,792 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.086e+02 2.418e+02 2.993e+02 8.338e+02, threshold=4.836e+02, percent-clipped=2.0 2023-05-01 09:51:36,123 INFO [train.py:904] (6/8) Epoch 22, batch 1950, loss[loss=0.1714, simple_loss=0.269, pruned_loss=0.03686, over 17085.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2563, pruned_loss=0.04095, over 3326819.14 frames. ], batch size: 55, lr: 3.09e-03, grad_scale: 4.0 2023-05-01 09:52:44,782 INFO [train.py:904] (6/8) Epoch 22, batch 2000, loss[loss=0.1744, simple_loss=0.2537, pruned_loss=0.04758, over 16409.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2564, pruned_loss=0.04097, over 3319012.33 frames. ], batch size: 146, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:53:46,002 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-01 09:53:48,739 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.559e+02 2.173e+02 2.667e+02 3.241e+02 6.066e+02, threshold=5.334e+02, percent-clipped=3.0 2023-05-01 09:53:53,918 INFO [train.py:904] (6/8) Epoch 22, batch 2050, loss[loss=0.1811, simple_loss=0.2565, pruned_loss=0.05285, over 16811.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2555, pruned_loss=0.041, over 3326315.28 frames. ], batch size: 102, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:54:31,499 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215230.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 09:54:45,177 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4378, 2.3170, 2.2612, 4.3873, 2.3588, 2.6901, 2.4301, 2.5149], device='cuda:6'), covar=tensor([0.1283, 0.3744, 0.3242, 0.0486, 0.4173, 0.2768, 0.3626, 0.3756], device='cuda:6'), in_proj_covar=tensor([0.0403, 0.0451, 0.0371, 0.0331, 0.0438, 0.0517, 0.0422, 0.0528], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:55:01,196 INFO [train.py:904] (6/8) Epoch 22, batch 2100, loss[loss=0.1743, simple_loss=0.2684, pruned_loss=0.04012, over 17074.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2559, pruned_loss=0.04173, over 3334954.21 frames. ], batch size: 55, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:55:10,086 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5359, 3.4996, 2.8358, 2.1152, 2.2795, 2.2576, 3.5848, 3.0367], device='cuda:6'), covar=tensor([0.2857, 0.0667, 0.1673, 0.2965, 0.2899, 0.2238, 0.0534, 0.1608], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0271, 0.0307, 0.0314, 0.0298, 0.0262, 0.0295, 0.0339], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 09:55:54,411 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215291.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 09:56:04,628 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.033e+02 2.346e+02 2.893e+02 7.245e+02, threshold=4.692e+02, percent-clipped=2.0 2023-05-01 09:56:09,027 INFO [train.py:904] (6/8) Epoch 22, batch 2150, loss[loss=0.1864, simple_loss=0.2649, pruned_loss=0.05388, over 16777.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2573, pruned_loss=0.04241, over 3321396.54 frames. ], batch size: 102, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:56:39,630 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215325.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 09:56:51,069 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215333.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:57:18,609 INFO [train.py:904] (6/8) Epoch 22, batch 2200, loss[loss=0.1905, simple_loss=0.2775, pruned_loss=0.05177, over 16571.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2575, pruned_loss=0.04238, over 3322802.22 frames. ], batch size: 68, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:58:02,287 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215386.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 09:58:03,521 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0106, 2.0998, 2.2222, 3.5855, 2.0815, 2.3636, 2.2139, 2.2393], device='cuda:6'), covar=tensor([0.1465, 0.3750, 0.3024, 0.0677, 0.4123, 0.2712, 0.3633, 0.3389], device='cuda:6'), in_proj_covar=tensor([0.0404, 0.0452, 0.0372, 0.0331, 0.0439, 0.0519, 0.0422, 0.0530], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 09:58:10,383 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3143, 3.3681, 3.7007, 2.2620, 3.0583, 2.4346, 3.7897, 3.7266], device='cuda:6'), covar=tensor([0.0244, 0.0943, 0.0574, 0.2000, 0.0890, 0.1010, 0.0519, 0.0926], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0164, 0.0167, 0.0155, 0.0145, 0.0131, 0.0144, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 09:58:14,376 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215394.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 09:58:20,772 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.171e+02 2.606e+02 3.111e+02 9.974e+02, threshold=5.211e+02, percent-clipped=4.0 2023-05-01 09:58:24,008 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5008, 3.6088, 4.1956, 2.3429, 3.2964, 2.5976, 4.0551, 3.8185], device='cuda:6'), covar=tensor([0.0284, 0.0940, 0.0453, 0.1915, 0.0803, 0.0967, 0.0537, 0.1119], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0164, 0.0167, 0.0155, 0.0145, 0.0131, 0.0144, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 09:58:24,658 INFO [train.py:904] (6/8) Epoch 22, batch 2250, loss[loss=0.1574, simple_loss=0.2421, pruned_loss=0.03631, over 16999.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2571, pruned_loss=0.04221, over 3326191.76 frames. ], batch size: 41, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:59:36,620 INFO [train.py:904] (6/8) Epoch 22, batch 2300, loss[loss=0.1554, simple_loss=0.2374, pruned_loss=0.03666, over 16768.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.257, pruned_loss=0.04188, over 3327170.64 frames. ], batch size: 124, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:00:42,879 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.161e+02 2.569e+02 3.134e+02 6.590e+02, threshold=5.137e+02, percent-clipped=2.0 2023-05-01 10:00:46,496 INFO [train.py:904] (6/8) Epoch 22, batch 2350, loss[loss=0.1806, simple_loss=0.2509, pruned_loss=0.05513, over 16790.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2576, pruned_loss=0.04253, over 3325295.86 frames. ], batch size: 83, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:00:46,946 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215503.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:01:00,058 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-05-01 10:01:48,866 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215548.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:01:55,001 INFO [train.py:904] (6/8) Epoch 22, batch 2400, loss[loss=0.1648, simple_loss=0.2572, pruned_loss=0.0362, over 17128.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2587, pruned_loss=0.04293, over 3313911.04 frames. ], batch size: 48, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:02:11,570 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215564.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:02:12,687 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215565.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:02:21,015 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6516, 3.8387, 2.4509, 4.3832, 2.9208, 4.3255, 2.5734, 3.1216], device='cuda:6'), covar=tensor([0.0307, 0.0344, 0.1532, 0.0279, 0.0849, 0.0544, 0.1488, 0.0712], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0179, 0.0196, 0.0167, 0.0179, 0.0222, 0.0204, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 10:02:41,710 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=215586.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 10:02:59,707 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.274e+02 2.111e+02 2.344e+02 3.035e+02 7.498e+02, threshold=4.688e+02, percent-clipped=4.0 2023-05-01 10:03:04,614 INFO [train.py:904] (6/8) Epoch 22, batch 2450, loss[loss=0.1854, simple_loss=0.2783, pruned_loss=0.04625, over 11955.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2593, pruned_loss=0.04231, over 3316746.24 frames. ], batch size: 246, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:03:08,683 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 10:03:12,025 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215609.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:03:21,180 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-01 10:03:35,698 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215626.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:04:11,703 INFO [train.py:904] (6/8) Epoch 22, batch 2500, loss[loss=0.1638, simple_loss=0.2623, pruned_loss=0.03263, over 17247.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.259, pruned_loss=0.04211, over 3317585.64 frames. ], batch size: 52, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:04:51,431 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=215681.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 10:05:02,692 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=215689.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 10:05:12,962 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1702, 5.1578, 4.9054, 4.3955, 4.9633, 1.9010, 4.7268, 4.7217], device='cuda:6'), covar=tensor([0.0088, 0.0086, 0.0225, 0.0401, 0.0117, 0.3028, 0.0155, 0.0269], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0158, 0.0201, 0.0179, 0.0179, 0.0210, 0.0190, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:05:15,510 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.601e+02 2.254e+02 2.666e+02 2.993e+02 6.188e+02, threshold=5.333e+02, percent-clipped=1.0 2023-05-01 10:05:20,562 INFO [train.py:904] (6/8) Epoch 22, batch 2550, loss[loss=0.165, simple_loss=0.2664, pruned_loss=0.03185, over 17047.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2592, pruned_loss=0.04246, over 3323416.16 frames. ], batch size: 50, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:06:30,382 INFO [train.py:904] (6/8) Epoch 22, batch 2600, loss[loss=0.1484, simple_loss=0.2371, pruned_loss=0.02985, over 16786.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2591, pruned_loss=0.04231, over 3318672.16 frames. ], batch size: 39, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:07:01,157 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8224, 2.7552, 2.3446, 2.7192, 3.1257, 2.9348, 3.4777, 3.3435], device='cuda:6'), covar=tensor([0.0150, 0.0419, 0.0548, 0.0431, 0.0286, 0.0383, 0.0237, 0.0284], device='cuda:6'), in_proj_covar=tensor([0.0218, 0.0241, 0.0231, 0.0232, 0.0242, 0.0241, 0.0244, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:07:14,019 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4118, 3.5824, 3.7845, 2.6240, 3.4592, 3.8248, 3.5425, 2.2400], device='cuda:6'), covar=tensor([0.0521, 0.0157, 0.0056, 0.0415, 0.0123, 0.0109, 0.0105, 0.0458], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0084, 0.0084, 0.0133, 0.0099, 0.0110, 0.0095, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 10:07:33,127 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2237, 5.1747, 4.9401, 4.4542, 5.0396, 1.8629, 4.7736, 4.8241], device='cuda:6'), covar=tensor([0.0102, 0.0098, 0.0224, 0.0421, 0.0107, 0.2980, 0.0155, 0.0243], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0158, 0.0202, 0.0179, 0.0180, 0.0211, 0.0191, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:07:36,118 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.223e+02 2.543e+02 2.941e+02 8.287e+02, threshold=5.085e+02, percent-clipped=5.0 2023-05-01 10:07:39,998 INFO [train.py:904] (6/8) Epoch 22, batch 2650, loss[loss=0.1949, simple_loss=0.2954, pruned_loss=0.04718, over 16604.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2601, pruned_loss=0.04226, over 3321955.32 frames. ], batch size: 62, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:07:42,368 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-01 10:08:06,489 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-05-01 10:08:16,662 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215829.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:08:48,040 INFO [train.py:904] (6/8) Epoch 22, batch 2700, loss[loss=0.1643, simple_loss=0.264, pruned_loss=0.0323, over 17153.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.26, pruned_loss=0.04178, over 3323127.08 frames. ], batch size: 47, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:08:55,235 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7243, 4.6515, 4.6128, 4.3093, 4.3447, 4.7105, 4.5339, 4.4058], device='cuda:6'), covar=tensor([0.0712, 0.0815, 0.0351, 0.0317, 0.0871, 0.0516, 0.0439, 0.0720], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0447, 0.0358, 0.0358, 0.0369, 0.0413, 0.0245, 0.0431], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 10:08:57,511 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=215859.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:09:01,763 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-05-01 10:09:34,740 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=215886.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:09:39,296 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9399, 3.8022, 4.3430, 2.1447, 4.5509, 4.6325, 3.2529, 3.4745], device='cuda:6'), covar=tensor([0.0742, 0.0243, 0.0230, 0.1137, 0.0071, 0.0182, 0.0422, 0.0410], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0109, 0.0099, 0.0139, 0.0080, 0.0128, 0.0130, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 10:09:39,323 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215890.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:09:53,920 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.480e+02 2.142e+02 2.465e+02 2.972e+02 5.160e+02, threshold=4.929e+02, percent-clipped=1.0 2023-05-01 10:09:57,374 INFO [train.py:904] (6/8) Epoch 22, batch 2750, loss[loss=0.1543, simple_loss=0.256, pruned_loss=0.02631, over 17018.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2606, pruned_loss=0.04135, over 3329298.38 frames. ], batch size: 50, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:09:59,451 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=215904.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:10:21,991 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=215921.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:10:33,185 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8122, 4.7263, 4.7078, 4.3806, 4.4108, 4.7742, 4.5805, 4.4906], device='cuda:6'), covar=tensor([0.0576, 0.0668, 0.0298, 0.0296, 0.0933, 0.0435, 0.0420, 0.0656], device='cuda:6'), in_proj_covar=tensor([0.0306, 0.0446, 0.0356, 0.0356, 0.0368, 0.0412, 0.0245, 0.0430], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 10:10:38,794 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=215934.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:11:05,010 INFO [train.py:904] (6/8) Epoch 22, batch 2800, loss[loss=0.1722, simple_loss=0.2601, pruned_loss=0.04211, over 15425.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2607, pruned_loss=0.04115, over 3324328.06 frames. ], batch size: 191, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:11:11,334 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-01 10:11:42,338 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=215981.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:11:53,968 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=215989.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:12:07,420 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.162e+02 2.449e+02 3.092e+02 5.987e+02, threshold=4.899e+02, percent-clipped=3.0 2023-05-01 10:12:14,947 INFO [train.py:904] (6/8) Epoch 22, batch 2850, loss[loss=0.163, simple_loss=0.2602, pruned_loss=0.03291, over 17173.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2596, pruned_loss=0.04101, over 3325545.08 frames. ], batch size: 46, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:12:51,538 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=216028.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:12:52,546 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216029.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:13:04,675 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216037.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:13:24,261 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4680, 5.4414, 5.3653, 4.8864, 4.9600, 5.3925, 5.3373, 4.9865], device='cuda:6'), covar=tensor([0.0578, 0.0544, 0.0291, 0.0321, 0.1159, 0.0520, 0.0259, 0.0792], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0449, 0.0359, 0.0360, 0.0371, 0.0416, 0.0247, 0.0434], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 10:13:26,222 INFO [train.py:904] (6/8) Epoch 22, batch 2900, loss[loss=0.1602, simple_loss=0.2534, pruned_loss=0.03348, over 17122.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2584, pruned_loss=0.04117, over 3326317.08 frames. ], batch size: 49, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:13:33,688 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6282, 3.7166, 2.2679, 4.0193, 2.8773, 3.9219, 2.4072, 2.9893], device='cuda:6'), covar=tensor([0.0290, 0.0426, 0.1577, 0.0308, 0.0781, 0.0783, 0.1385, 0.0683], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0180, 0.0197, 0.0168, 0.0180, 0.0223, 0.0205, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 10:14:17,848 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=216089.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:14:31,569 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.264e+02 2.773e+02 3.437e+02 6.159e+02, threshold=5.547e+02, percent-clipped=1.0 2023-05-01 10:14:35,745 INFO [train.py:904] (6/8) Epoch 22, batch 2950, loss[loss=0.1615, simple_loss=0.26, pruned_loss=0.03152, over 17227.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.257, pruned_loss=0.04133, over 3319316.60 frames. ], batch size: 52, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:15:43,432 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8627, 4.8972, 5.2819, 5.2800, 5.3284, 4.9739, 4.9127, 4.7752], device='cuda:6'), covar=tensor([0.0335, 0.0652, 0.0500, 0.0434, 0.0447, 0.0438, 0.0937, 0.0451], device='cuda:6'), in_proj_covar=tensor([0.0419, 0.0467, 0.0453, 0.0421, 0.0500, 0.0478, 0.0561, 0.0381], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 10:15:45,385 INFO [train.py:904] (6/8) Epoch 22, batch 3000, loss[loss=0.1709, simple_loss=0.2636, pruned_loss=0.03915, over 17256.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2581, pruned_loss=0.04247, over 3313303.35 frames. ], batch size: 52, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:15:45,385 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 10:15:54,106 INFO [train.py:938] (6/8) Epoch 22, validation: loss=0.1347, simple_loss=0.2399, pruned_loss=0.0148, over 944034.00 frames. 2023-05-01 10:15:54,107 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 10:16:02,789 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=216159.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:16:11,899 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 10:16:13,025 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.36 vs. limit=5.0 2023-05-01 10:16:22,931 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3376, 5.3207, 5.0580, 4.5007, 5.1419, 1.9630, 4.9040, 4.9701], device='cuda:6'), covar=tensor([0.0092, 0.0103, 0.0216, 0.0439, 0.0129, 0.2864, 0.0155, 0.0233], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0159, 0.0202, 0.0180, 0.0181, 0.0211, 0.0192, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:16:38,786 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=216185.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:16:59,839 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.653e+02 2.146e+02 2.550e+02 3.086e+02 1.158e+03, threshold=5.100e+02, percent-clipped=1.0 2023-05-01 10:17:03,725 INFO [train.py:904] (6/8) Epoch 22, batch 3050, loss[loss=0.1892, simple_loss=0.2792, pruned_loss=0.04957, over 16682.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2576, pruned_loss=0.04201, over 3316794.36 frames. ], batch size: 57, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:17:05,250 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=216204.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 10:17:09,668 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216207.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:17:09,862 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2669, 1.5680, 2.0532, 2.0985, 2.3122, 2.3219, 1.7443, 2.3592], device='cuda:6'), covar=tensor([0.0254, 0.0507, 0.0293, 0.0352, 0.0291, 0.0308, 0.0514, 0.0181], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0196, 0.0182, 0.0187, 0.0199, 0.0157, 0.0198, 0.0153], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:17:29,416 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=216221.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:18:11,439 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216252.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 10:18:12,223 INFO [train.py:904] (6/8) Epoch 22, batch 3100, loss[loss=0.1497, simple_loss=0.2359, pruned_loss=0.03171, over 16761.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2572, pruned_loss=0.04179, over 3318813.97 frames. ], batch size: 39, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:18:33,928 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216269.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:18:46,439 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-05-01 10:19:16,924 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.355e+02 2.131e+02 2.523e+02 3.440e+02 6.792e+02, threshold=5.047e+02, percent-clipped=4.0 2023-05-01 10:19:21,077 INFO [train.py:904] (6/8) Epoch 22, batch 3150, loss[loss=0.1384, simple_loss=0.2308, pruned_loss=0.02303, over 17225.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2565, pruned_loss=0.04143, over 3309674.40 frames. ], batch size: 45, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:19:24,340 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4621, 3.1983, 3.4594, 1.8351, 3.5652, 3.5707, 2.8825, 2.7172], device='cuda:6'), covar=tensor([0.0775, 0.0255, 0.0247, 0.1187, 0.0126, 0.0210, 0.0469, 0.0438], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0109, 0.0099, 0.0140, 0.0081, 0.0129, 0.0131, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 10:20:03,403 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-01 10:20:18,373 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-05-01 10:20:29,468 INFO [train.py:904] (6/8) Epoch 22, batch 3200, loss[loss=0.1519, simple_loss=0.2409, pruned_loss=0.03146, over 16000.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2548, pruned_loss=0.04099, over 3314141.78 frames. ], batch size: 35, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:21:13,901 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=216384.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:21:36,408 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.593e+02 2.182e+02 2.419e+02 2.918e+02 5.281e+02, threshold=4.837e+02, percent-clipped=1.0 2023-05-01 10:21:37,316 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3424, 3.3560, 2.1826, 3.5783, 2.6755, 3.5634, 2.2758, 2.8519], device='cuda:6'), covar=tensor([0.0291, 0.0420, 0.1457, 0.0322, 0.0729, 0.0832, 0.1345, 0.0636], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0181, 0.0197, 0.0169, 0.0180, 0.0224, 0.0206, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 10:21:40,321 INFO [train.py:904] (6/8) Epoch 22, batch 3250, loss[loss=0.1623, simple_loss=0.2508, pruned_loss=0.03688, over 17215.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2556, pruned_loss=0.04145, over 3317352.66 frames. ], batch size: 44, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:22:05,176 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 10:22:52,219 INFO [train.py:904] (6/8) Epoch 22, batch 3300, loss[loss=0.1599, simple_loss=0.2485, pruned_loss=0.03559, over 16772.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2558, pruned_loss=0.0411, over 3323691.57 frames. ], batch size: 76, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:23:33,686 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0206, 3.6807, 4.2514, 2.2164, 4.4290, 4.4891, 3.2969, 3.4565], device='cuda:6'), covar=tensor([0.0680, 0.0267, 0.0215, 0.1095, 0.0074, 0.0205, 0.0428, 0.0396], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0110, 0.0100, 0.0141, 0.0082, 0.0130, 0.0132, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 10:23:36,657 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=216485.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:23:56,628 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.388e+02 2.110e+02 2.633e+02 3.149e+02 8.538e+02, threshold=5.267e+02, percent-clipped=2.0 2023-05-01 10:24:00,075 INFO [train.py:904] (6/8) Epoch 22, batch 3350, loss[loss=0.1803, simple_loss=0.2578, pruned_loss=0.05144, over 16748.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2563, pruned_loss=0.04134, over 3330140.44 frames. ], batch size: 89, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:24:42,589 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216533.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:25:10,643 INFO [train.py:904] (6/8) Epoch 22, batch 3400, loss[loss=0.1401, simple_loss=0.2313, pruned_loss=0.02438, over 16817.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2562, pruned_loss=0.04142, over 3316616.94 frames. ], batch size: 42, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:26:02,889 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=216591.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:26:15,739 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.115e+02 2.508e+02 2.941e+02 4.033e+02, threshold=5.017e+02, percent-clipped=0.0 2023-05-01 10:26:19,664 INFO [train.py:904] (6/8) Epoch 22, batch 3450, loss[loss=0.1526, simple_loss=0.2313, pruned_loss=0.03698, over 16843.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.255, pruned_loss=0.041, over 3322000.20 frames. ], batch size: 96, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:26:48,244 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2857, 5.6502, 5.3854, 5.4294, 5.1488, 4.9707, 5.0187, 5.7298], device='cuda:6'), covar=tensor([0.1311, 0.0859, 0.0994, 0.0908, 0.0881, 0.0930, 0.1361, 0.0886], device='cuda:6'), in_proj_covar=tensor([0.0696, 0.0854, 0.0703, 0.0646, 0.0540, 0.0546, 0.0715, 0.0667], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:27:01,589 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1988, 5.1149, 4.9338, 3.9977, 5.0147, 1.8064, 4.7126, 4.7879], device='cuda:6'), covar=tensor([0.0107, 0.0125, 0.0279, 0.0617, 0.0157, 0.3371, 0.0197, 0.0316], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0161, 0.0205, 0.0183, 0.0183, 0.0212, 0.0194, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:27:25,747 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-05-01 10:27:26,959 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-05-01 10:27:29,239 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=216652.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:27:29,984 INFO [train.py:904] (6/8) Epoch 22, batch 3500, loss[loss=0.1675, simple_loss=0.2526, pruned_loss=0.04121, over 16623.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2542, pruned_loss=0.04078, over 3324139.30 frames. ], batch size: 62, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:27:32,823 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=216655.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:28:12,644 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=216684.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:28:17,024 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4491, 4.3551, 4.3869, 4.1198, 4.1554, 4.4491, 4.1174, 4.2293], device='cuda:6'), covar=tensor([0.0665, 0.0872, 0.0303, 0.0282, 0.0800, 0.0519, 0.0684, 0.0628], device='cuda:6'), in_proj_covar=tensor([0.0310, 0.0451, 0.0360, 0.0361, 0.0373, 0.0419, 0.0247, 0.0435], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 10:28:35,763 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 1.992e+02 2.337e+02 2.852e+02 6.760e+02, threshold=4.673e+02, percent-clipped=2.0 2023-05-01 10:28:39,279 INFO [train.py:904] (6/8) Epoch 22, batch 3550, loss[loss=0.1422, simple_loss=0.2231, pruned_loss=0.03068, over 16791.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2532, pruned_loss=0.04019, over 3312298.57 frames. ], batch size: 39, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:28:42,520 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=216705.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:28:57,375 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=216716.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:29:20,538 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216732.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:29:32,775 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8973, 2.1300, 2.3549, 1.9009, 2.5125, 2.6203, 2.3074, 2.1964], device='cuda:6'), covar=tensor([0.0968, 0.0288, 0.0316, 0.1148, 0.0156, 0.0349, 0.0608, 0.0579], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0109, 0.0100, 0.0140, 0.0082, 0.0129, 0.0131, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 10:29:49,503 INFO [train.py:904] (6/8) Epoch 22, batch 3600, loss[loss=0.1658, simple_loss=0.2459, pruned_loss=0.04279, over 15289.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2516, pruned_loss=0.03977, over 3305367.52 frames. ], batch size: 190, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:30:08,402 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=216766.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:30:25,232 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8722, 2.6033, 2.5848, 1.9704, 2.5909, 2.7137, 2.5796, 1.9347], device='cuda:6'), covar=tensor([0.0444, 0.0113, 0.0100, 0.0391, 0.0138, 0.0143, 0.0134, 0.0402], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0084, 0.0084, 0.0133, 0.0099, 0.0110, 0.0095, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 10:30:49,062 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3657, 3.3403, 3.3824, 3.4783, 3.5354, 3.2917, 3.5141, 3.6127], device='cuda:6'), covar=tensor([0.1177, 0.0876, 0.1084, 0.0647, 0.0639, 0.2218, 0.1079, 0.0725], device='cuda:6'), in_proj_covar=tensor([0.0670, 0.0832, 0.0969, 0.0845, 0.0636, 0.0668, 0.0690, 0.0800], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:31:00,613 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.119e+02 2.478e+02 2.922e+02 5.214e+02, threshold=4.955e+02, percent-clipped=3.0 2023-05-01 10:31:03,547 INFO [train.py:904] (6/8) Epoch 22, batch 3650, loss[loss=0.1782, simple_loss=0.251, pruned_loss=0.05271, over 16750.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2512, pruned_loss=0.04108, over 3282118.24 frames. ], batch size: 124, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:32:09,298 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0900, 3.2733, 3.4884, 2.2137, 3.0157, 2.3059, 3.7268, 3.6770], device='cuda:6'), covar=tensor([0.0217, 0.0902, 0.0574, 0.1870, 0.0814, 0.1039, 0.0439, 0.0729], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0166, 0.0167, 0.0155, 0.0146, 0.0131, 0.0145, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 10:32:18,422 INFO [train.py:904] (6/8) Epoch 22, batch 3700, loss[loss=0.1888, simple_loss=0.2716, pruned_loss=0.053, over 15460.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2504, pruned_loss=0.04229, over 3269652.19 frames. ], batch size: 190, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:32:31,728 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-01 10:33:19,823 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7327, 3.6500, 4.2306, 2.1163, 4.4914, 4.4977, 3.1959, 3.2877], device='cuda:6'), covar=tensor([0.0919, 0.0304, 0.0225, 0.1303, 0.0085, 0.0148, 0.0467, 0.0509], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0110, 0.0100, 0.0140, 0.0082, 0.0129, 0.0131, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 10:33:31,442 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.238e+02 2.459e+02 3.071e+02 7.166e+02, threshold=4.918e+02, percent-clipped=1.0 2023-05-01 10:33:32,645 INFO [train.py:904] (6/8) Epoch 22, batch 3750, loss[loss=0.1928, simple_loss=0.2927, pruned_loss=0.04646, over 17085.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2509, pruned_loss=0.04328, over 3265243.72 frames. ], batch size: 53, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:33:35,137 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-01 10:34:36,639 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=216947.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:34:44,479 INFO [train.py:904] (6/8) Epoch 22, batch 3800, loss[loss=0.17, simple_loss=0.2518, pruned_loss=0.04408, over 16861.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2516, pruned_loss=0.04429, over 3273651.39 frames. ], batch size: 96, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:35:03,029 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=216965.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:35:55,032 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.604e+02 2.325e+02 2.685e+02 3.161e+02 5.857e+02, threshold=5.370e+02, percent-clipped=3.0 2023-05-01 10:35:56,826 INFO [train.py:904] (6/8) Epoch 22, batch 3850, loss[loss=0.167, simple_loss=0.2493, pruned_loss=0.04234, over 16573.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2524, pruned_loss=0.04572, over 3276605.75 frames. ], batch size: 62, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:36:08,459 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=217011.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 10:36:13,975 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9557, 3.0823, 2.4968, 2.8194, 3.3871, 3.1202, 3.5166, 3.5058], device='cuda:6'), covar=tensor([0.0069, 0.0358, 0.0515, 0.0417, 0.0225, 0.0348, 0.0196, 0.0219], device='cuda:6'), in_proj_covar=tensor([0.0220, 0.0240, 0.0230, 0.0231, 0.0241, 0.0241, 0.0244, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:36:29,815 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=217026.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:36:57,290 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-01 10:37:09,496 INFO [train.py:904] (6/8) Epoch 22, batch 3900, loss[loss=0.1834, simple_loss=0.261, pruned_loss=0.05296, over 16526.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.252, pruned_loss=0.04626, over 3271523.08 frames. ], batch size: 68, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:37:22,071 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=217061.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:38:04,677 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5957, 4.6419, 4.8664, 4.6645, 4.6906, 5.2607, 4.7851, 4.4701], device='cuda:6'), covar=tensor([0.1472, 0.2099, 0.2038, 0.1936, 0.2694, 0.1062, 0.1575, 0.2573], device='cuda:6'), in_proj_covar=tensor([0.0420, 0.0608, 0.0666, 0.0504, 0.0673, 0.0701, 0.0523, 0.0673], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 10:38:19,552 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.01 vs. limit=5.0 2023-05-01 10:38:21,557 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.630e+02 2.294e+02 2.626e+02 3.091e+02 8.707e+02, threshold=5.252e+02, percent-clipped=1.0 2023-05-01 10:38:22,861 INFO [train.py:904] (6/8) Epoch 22, batch 3950, loss[loss=0.204, simple_loss=0.2822, pruned_loss=0.06291, over 12661.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2514, pruned_loss=0.04655, over 3271744.42 frames. ], batch size: 246, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:39:35,757 INFO [train.py:904] (6/8) Epoch 22, batch 4000, loss[loss=0.1858, simple_loss=0.2741, pruned_loss=0.04876, over 16855.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2518, pruned_loss=0.04687, over 3271797.82 frames. ], batch size: 83, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:40:48,053 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 1.995e+02 2.386e+02 2.991e+02 7.237e+02, threshold=4.771e+02, percent-clipped=2.0 2023-05-01 10:40:49,969 INFO [train.py:904] (6/8) Epoch 22, batch 4050, loss[loss=0.1577, simple_loss=0.2463, pruned_loss=0.03458, over 16673.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2523, pruned_loss=0.04595, over 3274278.64 frames. ], batch size: 89, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:41:07,048 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0316, 2.0933, 2.7509, 2.9161, 2.9596, 3.4985, 2.2969, 3.3791], device='cuda:6'), covar=tensor([0.0216, 0.0500, 0.0282, 0.0295, 0.0296, 0.0137, 0.0525, 0.0116], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0195, 0.0181, 0.0187, 0.0199, 0.0156, 0.0198, 0.0153], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:41:51,997 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0468, 3.2730, 3.1156, 1.7166, 2.7151, 1.8488, 3.5480, 3.6145], device='cuda:6'), covar=tensor([0.0251, 0.0889, 0.0866, 0.2769, 0.1190, 0.1462, 0.0626, 0.1096], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0165, 0.0166, 0.0153, 0.0145, 0.0130, 0.0144, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 10:41:55,312 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=217247.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:41:57,504 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7867, 3.9665, 2.9326, 2.4268, 2.7094, 2.6608, 4.3695, 3.5883], device='cuda:6'), covar=tensor([0.2792, 0.0620, 0.1857, 0.2515, 0.2606, 0.1885, 0.0388, 0.1085], device='cuda:6'), in_proj_covar=tensor([0.0328, 0.0272, 0.0308, 0.0316, 0.0301, 0.0263, 0.0297, 0.0342], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 10:42:04,913 INFO [train.py:904] (6/8) Epoch 22, batch 4100, loss[loss=0.2039, simple_loss=0.2949, pruned_loss=0.05641, over 16662.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2545, pruned_loss=0.04571, over 3265251.65 frames. ], batch size: 134, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:43:02,124 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-05-01 10:43:04,930 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6660, 3.2737, 3.2713, 2.0342, 2.8319, 2.2071, 3.1834, 3.4140], device='cuda:6'), covar=tensor([0.0325, 0.0745, 0.0602, 0.2058, 0.0917, 0.0996, 0.0784, 0.0931], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0165, 0.0167, 0.0154, 0.0146, 0.0131, 0.0144, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 10:43:11,030 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=217295.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:43:21,263 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 2.082e+02 2.531e+02 2.908e+02 7.878e+02, threshold=5.062e+02, percent-clipped=6.0 2023-05-01 10:43:23,202 INFO [train.py:904] (6/8) Epoch 22, batch 4150, loss[loss=0.1959, simple_loss=0.285, pruned_loss=0.05338, over 16701.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2617, pruned_loss=0.04781, over 3246256.44 frames. ], batch size: 57, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:43:36,150 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=217311.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 10:43:52,137 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=217321.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:44:39,665 INFO [train.py:904] (6/8) Epoch 22, batch 4200, loss[loss=0.2066, simple_loss=0.286, pruned_loss=0.06361, over 11478.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2683, pruned_loss=0.04979, over 3201780.61 frames. ], batch size: 246, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:44:50,072 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=217359.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:44:52,804 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=217361.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:45:26,648 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7325, 1.7645, 1.6135, 1.3522, 1.8519, 1.5477, 1.6011, 1.8063], device='cuda:6'), covar=tensor([0.0192, 0.0261, 0.0356, 0.0337, 0.0205, 0.0252, 0.0150, 0.0197], device='cuda:6'), in_proj_covar=tensor([0.0217, 0.0238, 0.0229, 0.0230, 0.0239, 0.0238, 0.0242, 0.0236], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:45:32,876 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=217388.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:45:39,629 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 10:45:53,906 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 2.187e+02 2.459e+02 3.018e+02 5.838e+02, threshold=4.919e+02, percent-clipped=1.0 2023-05-01 10:45:55,231 INFO [train.py:904] (6/8) Epoch 22, batch 4250, loss[loss=0.1858, simple_loss=0.263, pruned_loss=0.05428, over 12166.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2717, pruned_loss=0.04957, over 3180597.52 frames. ], batch size: 247, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:46:04,663 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=217409.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:47:04,201 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=217449.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:47:09,152 INFO [train.py:904] (6/8) Epoch 22, batch 4300, loss[loss=0.2026, simple_loss=0.2999, pruned_loss=0.05258, over 16865.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2726, pruned_loss=0.04908, over 3139815.46 frames. ], batch size: 102, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:48:04,597 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-05-01 10:48:07,882 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=217492.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:48:13,701 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.67 vs. limit=5.0 2023-05-01 10:48:23,053 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.190e+02 2.513e+02 2.861e+02 5.953e+02, threshold=5.026e+02, percent-clipped=1.0 2023-05-01 10:48:24,299 INFO [train.py:904] (6/8) Epoch 22, batch 4350, loss[loss=0.2114, simple_loss=0.2963, pruned_loss=0.0632, over 15518.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2754, pruned_loss=0.04982, over 3143812.12 frames. ], batch size: 191, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:49:39,978 INFO [train.py:904] (6/8) Epoch 22, batch 4400, loss[loss=0.1869, simple_loss=0.2722, pruned_loss=0.05083, over 16659.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2777, pruned_loss=0.05076, over 3174419.54 frames. ], batch size: 134, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:49:41,120 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=217553.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:50:52,392 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.401e+02 2.074e+02 2.354e+02 2.786e+02 4.691e+02, threshold=4.709e+02, percent-clipped=0.0 2023-05-01 10:50:53,484 INFO [train.py:904] (6/8) Epoch 22, batch 4450, loss[loss=0.2252, simple_loss=0.3132, pruned_loss=0.06856, over 16401.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2815, pruned_loss=0.05215, over 3184093.24 frames. ], batch size: 146, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:51:20,384 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=217621.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:51:44,638 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 10:52:08,356 INFO [train.py:904] (6/8) Epoch 22, batch 4500, loss[loss=0.2201, simple_loss=0.2989, pruned_loss=0.07067, over 16833.00 frames. ], tot_loss[loss=0.1935, simple_loss=0.2818, pruned_loss=0.05256, over 3202480.35 frames. ], batch size: 116, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:52:24,146 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=217664.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:52:32,278 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=217669.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:53:18,212 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 1.861e+02 2.083e+02 2.379e+02 4.476e+02, threshold=4.167e+02, percent-clipped=0.0 2023-05-01 10:53:19,292 INFO [train.py:904] (6/8) Epoch 22, batch 4550, loss[loss=0.2199, simple_loss=0.3067, pruned_loss=0.06653, over 16471.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.283, pruned_loss=0.05357, over 3206790.28 frames. ], batch size: 75, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:53:53,252 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=217725.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:53:59,555 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2633, 1.4273, 1.9602, 2.1001, 2.1999, 2.4636, 1.6747, 2.3560], device='cuda:6'), covar=tensor([0.0227, 0.0580, 0.0322, 0.0324, 0.0325, 0.0204, 0.0608, 0.0144], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0195, 0.0182, 0.0188, 0.0200, 0.0156, 0.0199, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:54:09,270 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 10:54:19,879 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=217744.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:54:31,799 INFO [train.py:904] (6/8) Epoch 22, batch 4600, loss[loss=0.1952, simple_loss=0.286, pruned_loss=0.05218, over 17172.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2842, pruned_loss=0.054, over 3205655.08 frames. ], batch size: 46, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:54:55,269 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3174, 5.2847, 5.1521, 4.8157, 4.8564, 5.2120, 5.0197, 4.8784], device='cuda:6'), covar=tensor([0.0453, 0.0296, 0.0215, 0.0245, 0.0787, 0.0274, 0.0255, 0.0568], device='cuda:6'), in_proj_covar=tensor([0.0295, 0.0428, 0.0344, 0.0344, 0.0352, 0.0396, 0.0236, 0.0411], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:55:25,203 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9038, 5.3430, 5.5761, 5.3002, 5.3367, 5.9233, 5.4155, 5.1077], device='cuda:6'), covar=tensor([0.1032, 0.1828, 0.1970, 0.1736, 0.2475, 0.0899, 0.1387, 0.2377], device='cuda:6'), in_proj_covar=tensor([0.0409, 0.0587, 0.0642, 0.0486, 0.0648, 0.0680, 0.0506, 0.0653], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 10:55:25,298 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8202, 4.6668, 4.8385, 5.0010, 5.1581, 4.6693, 5.1406, 5.1835], device='cuda:6'), covar=tensor([0.1537, 0.1104, 0.1435, 0.0638, 0.0454, 0.0787, 0.0534, 0.0488], device='cuda:6'), in_proj_covar=tensor([0.0638, 0.0791, 0.0919, 0.0802, 0.0608, 0.0636, 0.0658, 0.0760], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:55:41,527 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 1.780e+02 2.045e+02 2.386e+02 4.197e+02, threshold=4.091e+02, percent-clipped=1.0 2023-05-01 10:55:42,804 INFO [train.py:904] (6/8) Epoch 22, batch 4650, loss[loss=0.1718, simple_loss=0.2623, pruned_loss=0.04067, over 16831.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2832, pruned_loss=0.05402, over 3220885.75 frames. ], batch size: 83, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:55:55,921 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1076, 5.0787, 4.8080, 4.2682, 5.0356, 1.9013, 4.7497, 4.4118], device='cuda:6'), covar=tensor([0.0056, 0.0043, 0.0152, 0.0282, 0.0050, 0.2814, 0.0080, 0.0211], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0158, 0.0203, 0.0181, 0.0180, 0.0211, 0.0192, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:56:46,911 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=217848.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:56:52,991 INFO [train.py:904] (6/8) Epoch 22, batch 4700, loss[loss=0.1906, simple_loss=0.2751, pruned_loss=0.05305, over 16563.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2807, pruned_loss=0.05296, over 3209448.78 frames. ], batch size: 62, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:57:32,349 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3617, 4.3746, 4.3191, 3.4973, 4.3233, 1.6855, 4.0616, 4.0049], device='cuda:6'), covar=tensor([0.0142, 0.0152, 0.0170, 0.0537, 0.0117, 0.2915, 0.0161, 0.0275], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0159, 0.0203, 0.0182, 0.0180, 0.0212, 0.0192, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 10:57:44,865 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5897, 2.4765, 2.3048, 3.5993, 2.2849, 3.8017, 1.5158, 2.7294], device='cuda:6'), covar=tensor([0.1469, 0.0944, 0.1393, 0.0209, 0.0207, 0.0438, 0.1779, 0.0893], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0173, 0.0194, 0.0191, 0.0204, 0.0215, 0.0201, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 10:57:59,278 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=217898.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:58:04,665 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.316e+02 1.895e+02 2.114e+02 2.418e+02 3.922e+02, threshold=4.227e+02, percent-clipped=0.0 2023-05-01 10:58:05,929 INFO [train.py:904] (6/8) Epoch 22, batch 4750, loss[loss=0.1655, simple_loss=0.2565, pruned_loss=0.03722, over 16745.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2765, pruned_loss=0.05065, over 3204456.05 frames. ], batch size: 89, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:59:17,329 INFO [train.py:904] (6/8) Epoch 22, batch 4800, loss[loss=0.1628, simple_loss=0.2534, pruned_loss=0.03604, over 16757.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2732, pruned_loss=0.04851, over 3200497.06 frames. ], batch size: 89, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:59:27,577 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=217959.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 11:00:09,055 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6623, 2.5542, 2.5058, 4.1378, 2.6293, 3.9351, 1.4531, 2.9743], device='cuda:6'), covar=tensor([0.1367, 0.0842, 0.1187, 0.0120, 0.0148, 0.0343, 0.1713, 0.0730], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0173, 0.0193, 0.0191, 0.0204, 0.0214, 0.0200, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 11:00:36,342 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.164e+02 1.772e+02 2.165e+02 2.536e+02 4.360e+02, threshold=4.330e+02, percent-clipped=2.0 2023-05-01 11:00:36,358 INFO [train.py:904] (6/8) Epoch 22, batch 4850, loss[loss=0.2123, simple_loss=0.2967, pruned_loss=0.06392, over 11668.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2743, pruned_loss=0.04797, over 3169280.11 frames. ], batch size: 247, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:01:01,605 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218020.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:01:28,235 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7523, 3.3540, 3.3244, 2.1414, 3.1001, 3.3539, 3.0933, 1.7033], device='cuda:6'), covar=tensor([0.0719, 0.0073, 0.0071, 0.0486, 0.0116, 0.0130, 0.0133, 0.0638], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0081, 0.0083, 0.0131, 0.0097, 0.0108, 0.0093, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 11:01:38,644 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218044.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:01:44,483 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.70 vs. limit=5.0 2023-05-01 11:01:51,558 INFO [train.py:904] (6/8) Epoch 22, batch 4900, loss[loss=0.1978, simple_loss=0.2775, pruned_loss=0.05901, over 12255.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2738, pruned_loss=0.04735, over 3134717.83 frames. ], batch size: 248, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:02:17,289 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7430, 2.7259, 2.4123, 4.6098, 3.1771, 4.0091, 1.6311, 2.9687], device='cuda:6'), covar=tensor([0.1414, 0.0893, 0.1412, 0.0158, 0.0332, 0.0444, 0.1742, 0.0903], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0173, 0.0193, 0.0190, 0.0203, 0.0214, 0.0200, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 11:02:35,689 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5374, 3.6048, 3.4017, 3.0548, 3.2083, 3.4725, 3.2873, 3.3494], device='cuda:6'), covar=tensor([0.0541, 0.0497, 0.0279, 0.0260, 0.0557, 0.0440, 0.1390, 0.0436], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0426, 0.0341, 0.0341, 0.0350, 0.0393, 0.0234, 0.0407], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 11:02:49,302 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218092.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:03:01,678 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218100.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:03:05,447 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.042e+02 2.302e+02 2.921e+02 6.655e+02, threshold=4.605e+02, percent-clipped=4.0 2023-05-01 11:03:05,462 INFO [train.py:904] (6/8) Epoch 22, batch 4950, loss[loss=0.1775, simple_loss=0.277, pruned_loss=0.03898, over 16264.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.2727, pruned_loss=0.04627, over 3161378.56 frames. ], batch size: 165, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:03:29,631 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6908, 1.9214, 2.3229, 2.6584, 2.7137, 3.1405, 1.9792, 3.0300], device='cuda:6'), covar=tensor([0.0254, 0.0542, 0.0364, 0.0366, 0.0313, 0.0192, 0.0622, 0.0142], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0196, 0.0181, 0.0187, 0.0200, 0.0155, 0.0199, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 11:03:47,259 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 11:04:11,450 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218148.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:04:18,894 INFO [train.py:904] (6/8) Epoch 22, batch 5000, loss[loss=0.1888, simple_loss=0.2766, pruned_loss=0.05049, over 16597.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.274, pruned_loss=0.04634, over 3182035.28 frames. ], batch size: 62, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:04:30,120 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=218161.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:05:01,564 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218182.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:05:22,062 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218196.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:05:32,358 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.007e+02 2.287e+02 2.605e+02 5.161e+02, threshold=4.575e+02, percent-clipped=1.0 2023-05-01 11:05:32,373 INFO [train.py:904] (6/8) Epoch 22, batch 5050, loss[loss=0.2036, simple_loss=0.2843, pruned_loss=0.06149, over 12400.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2743, pruned_loss=0.04627, over 3183376.68 frames. ], batch size: 248, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:05:42,380 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218210.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:06:30,496 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=218243.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:06:44,147 INFO [train.py:904] (6/8) Epoch 22, batch 5100, loss[loss=0.1809, simple_loss=0.2678, pruned_loss=0.04701, over 16754.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2724, pruned_loss=0.04538, over 3190988.57 frames. ], batch size: 124, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:06:45,775 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218254.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 11:07:00,184 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.25 vs. limit=5.0 2023-05-01 11:07:10,553 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=218271.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 11:07:39,095 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9314, 5.2900, 5.5564, 5.2196, 5.3513, 5.8864, 5.3396, 5.0218], device='cuda:6'), covar=tensor([0.0963, 0.1720, 0.1596, 0.1796, 0.2146, 0.0783, 0.1293, 0.2326], device='cuda:6'), in_proj_covar=tensor([0.0408, 0.0580, 0.0635, 0.0484, 0.0645, 0.0673, 0.0506, 0.0652], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 11:07:57,332 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 1.925e+02 2.150e+02 2.531e+02 3.604e+02, threshold=4.300e+02, percent-clipped=0.0 2023-05-01 11:07:57,347 INFO [train.py:904] (6/8) Epoch 22, batch 5150, loss[loss=0.1613, simple_loss=0.2625, pruned_loss=0.03004, over 16795.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2722, pruned_loss=0.0446, over 3190622.77 frames. ], batch size: 83, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:08:23,164 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218320.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:08:31,402 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.81 vs. limit=5.0 2023-05-01 11:09:11,242 INFO [train.py:904] (6/8) Epoch 22, batch 5200, loss[loss=0.1731, simple_loss=0.2626, pruned_loss=0.04179, over 16686.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.271, pruned_loss=0.04422, over 3205107.00 frames. ], batch size: 76, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:09:33,120 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218368.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:10:23,968 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 1.955e+02 2.305e+02 2.711e+02 5.108e+02, threshold=4.610e+02, percent-clipped=2.0 2023-05-01 11:10:23,983 INFO [train.py:904] (6/8) Epoch 22, batch 5250, loss[loss=0.1685, simple_loss=0.2614, pruned_loss=0.03782, over 16594.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2687, pruned_loss=0.04373, over 3204359.28 frames. ], batch size: 62, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:11:16,904 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 11:11:37,150 INFO [train.py:904] (6/8) Epoch 22, batch 5300, loss[loss=0.1519, simple_loss=0.241, pruned_loss=0.03143, over 16520.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2657, pruned_loss=0.04289, over 3201499.34 frames. ], batch size: 68, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:11:41,562 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218456.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:12:51,254 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.460e+02 1.955e+02 2.262e+02 2.726e+02 6.747e+02, threshold=4.525e+02, percent-clipped=3.0 2023-05-01 11:12:51,269 INFO [train.py:904] (6/8) Epoch 22, batch 5350, loss[loss=0.2088, simple_loss=0.3082, pruned_loss=0.0547, over 16893.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2645, pruned_loss=0.04271, over 3208526.90 frames. ], batch size: 116, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:13:01,190 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218509.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:13:43,230 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218538.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:14:03,262 INFO [train.py:904] (6/8) Epoch 22, batch 5400, loss[loss=0.1732, simple_loss=0.2712, pruned_loss=0.03758, over 16767.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2666, pruned_loss=0.04307, over 3206151.14 frames. ], batch size: 83, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:14:06,173 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218554.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 11:14:23,174 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218566.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 11:14:29,555 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=218570.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:14:34,753 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218574.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:14:35,079 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 11:15:18,766 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218602.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 11:15:19,489 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.606e+02 2.152e+02 2.360e+02 2.669e+02 5.554e+02, threshold=4.720e+02, percent-clipped=2.0 2023-05-01 11:15:19,504 INFO [train.py:904] (6/8) Epoch 22, batch 5450, loss[loss=0.1813, simple_loss=0.273, pruned_loss=0.04483, over 16959.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2693, pruned_loss=0.04441, over 3181950.91 frames. ], batch size: 90, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:15:30,200 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9019, 2.7900, 2.8343, 2.1153, 2.6871, 2.1719, 2.7604, 2.9446], device='cuda:6'), covar=tensor([0.0261, 0.0730, 0.0489, 0.1629, 0.0751, 0.0869, 0.0513, 0.0649], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0164, 0.0167, 0.0153, 0.0145, 0.0130, 0.0143, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 11:16:11,022 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=218635.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:16:37,990 INFO [train.py:904] (6/8) Epoch 22, batch 5500, loss[loss=0.1952, simple_loss=0.2905, pruned_loss=0.04999, over 16789.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.276, pruned_loss=0.0484, over 3155302.88 frames. ], batch size: 83, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:17:21,472 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 11:17:57,928 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.055e+02 2.950e+02 3.517e+02 4.298e+02 7.348e+02, threshold=7.033e+02, percent-clipped=15.0 2023-05-01 11:17:57,943 INFO [train.py:904] (6/8) Epoch 22, batch 5550, loss[loss=0.234, simple_loss=0.3123, pruned_loss=0.07786, over 16279.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2829, pruned_loss=0.05283, over 3154771.03 frames. ], batch size: 165, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:18:32,322 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 11:18:55,024 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-01 11:19:20,447 INFO [train.py:904] (6/8) Epoch 22, batch 5600, loss[loss=0.2802, simple_loss=0.3359, pruned_loss=0.1123, over 11274.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2883, pruned_loss=0.05751, over 3092492.83 frames. ], batch size: 248, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:19:25,852 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218756.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:20:41,870 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.037e+02 3.163e+02 3.914e+02 4.827e+02 8.104e+02, threshold=7.827e+02, percent-clipped=4.0 2023-05-01 11:20:41,885 INFO [train.py:904] (6/8) Epoch 22, batch 5650, loss[loss=0.2144, simple_loss=0.2999, pruned_loss=0.06448, over 16290.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2926, pruned_loss=0.06096, over 3078894.40 frames. ], batch size: 165, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:20:44,087 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218804.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:20:49,358 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0840, 5.0705, 4.9695, 4.5999, 4.6135, 4.9984, 4.8709, 4.7214], device='cuda:6'), covar=tensor([0.0585, 0.0456, 0.0286, 0.0292, 0.0941, 0.0483, 0.0358, 0.0621], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0426, 0.0340, 0.0339, 0.0349, 0.0393, 0.0234, 0.0406], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 11:21:10,468 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7333, 4.9995, 5.1853, 4.9726, 4.9951, 5.5472, 5.0309, 4.7774], device='cuda:6'), covar=tensor([0.1115, 0.1847, 0.2213, 0.1755, 0.2306, 0.0969, 0.1699, 0.2497], device='cuda:6'), in_proj_covar=tensor([0.0407, 0.0581, 0.0641, 0.0486, 0.0646, 0.0675, 0.0507, 0.0652], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 11:21:27,625 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-01 11:21:35,466 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218838.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:21:56,891 INFO [train.py:904] (6/8) Epoch 22, batch 5700, loss[loss=0.2074, simple_loss=0.3021, pruned_loss=0.05631, over 16887.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.2939, pruned_loss=0.06264, over 3078540.05 frames. ], batch size: 116, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:22:12,744 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-05-01 11:22:16,244 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218865.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:22:18,187 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218866.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:22:48,658 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218886.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:23:13,875 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.920e+02 2.926e+02 3.484e+02 4.272e+02 7.645e+02, threshold=6.969e+02, percent-clipped=0.0 2023-05-01 11:23:13,890 INFO [train.py:904] (6/8) Epoch 22, batch 5750, loss[loss=0.2143, simple_loss=0.3036, pruned_loss=0.0625, over 16416.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2965, pruned_loss=0.06402, over 3061426.60 frames. ], batch size: 146, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:23:32,713 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218914.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:23:55,949 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9116, 2.8059, 2.8523, 2.2340, 2.7069, 2.1849, 2.7941, 2.9435], device='cuda:6'), covar=tensor([0.0259, 0.0731, 0.0590, 0.1589, 0.0772, 0.1030, 0.0528, 0.0660], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0163, 0.0166, 0.0153, 0.0145, 0.0130, 0.0143, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 11:23:57,590 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218930.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:24:34,384 INFO [train.py:904] (6/8) Epoch 22, batch 5800, loss[loss=0.2402, simple_loss=0.3019, pruned_loss=0.08924, over 12160.00 frames. ], tot_loss[loss=0.2119, simple_loss=0.2967, pruned_loss=0.06353, over 3050390.01 frames. ], batch size: 248, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:24:54,917 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218965.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:25:05,514 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 11:25:15,101 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8160, 2.5970, 2.6066, 1.9308, 2.5122, 2.7245, 2.5867, 1.8719], device='cuda:6'), covar=tensor([0.0471, 0.0134, 0.0102, 0.0413, 0.0147, 0.0136, 0.0131, 0.0433], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0083, 0.0084, 0.0133, 0.0098, 0.0110, 0.0095, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 11:25:24,835 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218984.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:25:53,553 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.200e+02 2.954e+02 3.399e+02 3.875e+02 5.875e+02, threshold=6.799e+02, percent-clipped=0.0 2023-05-01 11:25:53,575 INFO [train.py:904] (6/8) Epoch 22, batch 5850, loss[loss=0.2043, simple_loss=0.2779, pruned_loss=0.06532, over 11889.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2945, pruned_loss=0.06167, over 3054280.87 frames. ], batch size: 248, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:26:21,593 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219021.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:26:30,044 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219026.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:27:02,383 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219045.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:27:14,926 INFO [train.py:904] (6/8) Epoch 22, batch 5900, loss[loss=0.2103, simple_loss=0.2859, pruned_loss=0.06736, over 16643.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2935, pruned_loss=0.06136, over 3044018.42 frames. ], batch size: 134, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:27:45,193 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7898, 2.8262, 2.8833, 2.1469, 2.6930, 2.1581, 2.7657, 2.9093], device='cuda:6'), covar=tensor([0.0231, 0.0646, 0.0461, 0.1585, 0.0755, 0.0869, 0.0508, 0.0642], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0163, 0.0166, 0.0153, 0.0145, 0.0130, 0.0143, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 11:27:49,458 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219072.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:28:04,709 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219082.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:28:35,884 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.790e+02 2.529e+02 2.968e+02 3.874e+02 6.567e+02, threshold=5.937e+02, percent-clipped=0.0 2023-05-01 11:28:35,899 INFO [train.py:904] (6/8) Epoch 22, batch 5950, loss[loss=0.2145, simple_loss=0.2914, pruned_loss=0.06874, over 11901.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2945, pruned_loss=0.06087, over 3027795.62 frames. ], batch size: 248, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:29:24,528 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219133.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:29:57,662 INFO [train.py:904] (6/8) Epoch 22, batch 6000, loss[loss=0.1963, simple_loss=0.2784, pruned_loss=0.05714, over 16793.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2936, pruned_loss=0.06024, over 3044821.78 frames. ], batch size: 39, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:29:57,663 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 11:30:07,622 INFO [train.py:938] (6/8) Epoch 22, validation: loss=0.1507, simple_loss=0.2632, pruned_loss=0.01907, over 944034.00 frames. 2023-05-01 11:30:07,623 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 11:30:27,696 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219165.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:31:28,922 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.979e+02 2.852e+02 3.411e+02 4.220e+02 7.320e+02, threshold=6.821e+02, percent-clipped=6.0 2023-05-01 11:31:28,937 INFO [train.py:904] (6/8) Epoch 22, batch 6050, loss[loss=0.2194, simple_loss=0.2945, pruned_loss=0.07215, over 11512.00 frames. ], tot_loss[loss=0.205, simple_loss=0.2922, pruned_loss=0.05893, over 3080277.92 frames. ], batch size: 246, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:31:29,608 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219203.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:31:44,912 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219213.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:32:06,223 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5235, 3.4383, 3.4434, 2.7457, 3.3680, 2.1149, 3.1211, 2.7512], device='cuda:6'), covar=tensor([0.0167, 0.0153, 0.0201, 0.0251, 0.0118, 0.2237, 0.0160, 0.0268], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0156, 0.0198, 0.0178, 0.0175, 0.0206, 0.0187, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 11:32:11,193 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219230.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:32:30,313 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8852, 3.7765, 4.3235, 1.9418, 4.4903, 4.5339, 3.2480, 3.2691], device='cuda:6'), covar=tensor([0.0751, 0.0246, 0.0167, 0.1292, 0.0065, 0.0130, 0.0415, 0.0461], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0110, 0.0099, 0.0140, 0.0082, 0.0128, 0.0131, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 11:32:46,223 INFO [train.py:904] (6/8) Epoch 22, batch 6100, loss[loss=0.179, simple_loss=0.2708, pruned_loss=0.04366, over 16520.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2913, pruned_loss=0.0578, over 3094098.31 frames. ], batch size: 75, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:33:05,438 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219264.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:33:26,347 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219278.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:34:04,107 INFO [train.py:904] (6/8) Epoch 22, batch 6150, loss[loss=0.2106, simple_loss=0.285, pruned_loss=0.06815, over 11613.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2896, pruned_loss=0.05748, over 3090623.93 frames. ], batch size: 246, lr: 3.06e-03, grad_scale: 4.0 2023-05-01 11:34:05,864 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.700e+02 2.531e+02 2.938e+02 3.735e+02 5.885e+02, threshold=5.877e+02, percent-clipped=0.0 2023-05-01 11:34:29,469 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5000, 3.5627, 3.3193, 2.9327, 3.1772, 3.4582, 3.3503, 3.2517], device='cuda:6'), covar=tensor([0.0556, 0.0590, 0.0278, 0.0267, 0.0513, 0.0447, 0.1181, 0.0442], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0427, 0.0341, 0.0338, 0.0347, 0.0393, 0.0234, 0.0406], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 11:34:33,720 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219321.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:34:33,813 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4038, 3.3964, 3.4367, 3.5076, 3.5392, 3.3183, 3.5211, 3.5883], device='cuda:6'), covar=tensor([0.1291, 0.0941, 0.1034, 0.0625, 0.0678, 0.2166, 0.1019, 0.0844], device='cuda:6'), in_proj_covar=tensor([0.0628, 0.0779, 0.0903, 0.0790, 0.0597, 0.0624, 0.0650, 0.0749], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 11:35:03,737 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219340.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:35:06,921 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219342.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 11:35:22,806 INFO [train.py:904] (6/8) Epoch 22, batch 6200, loss[loss=0.2378, simple_loss=0.3064, pruned_loss=0.08458, over 11606.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2879, pruned_loss=0.05718, over 3078437.88 frames. ], batch size: 248, lr: 3.06e-03, grad_scale: 4.0 2023-05-01 11:35:47,180 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9696, 2.7641, 2.8239, 2.1248, 2.6316, 2.1347, 2.7684, 2.9597], device='cuda:6'), covar=tensor([0.0314, 0.0806, 0.0583, 0.1791, 0.0899, 0.1002, 0.0624, 0.0738], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0164, 0.0167, 0.0153, 0.0145, 0.0130, 0.0143, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 11:36:02,559 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219377.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:36:27,898 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8388, 5.0622, 5.2713, 4.9444, 5.0040, 5.6335, 5.0919, 4.8198], device='cuda:6'), covar=tensor([0.1113, 0.1996, 0.2511, 0.2085, 0.2668, 0.0978, 0.1688, 0.2653], device='cuda:6'), in_proj_covar=tensor([0.0411, 0.0589, 0.0651, 0.0493, 0.0656, 0.0681, 0.0515, 0.0661], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 11:36:42,474 INFO [train.py:904] (6/8) Epoch 22, batch 6250, loss[loss=0.2033, simple_loss=0.2913, pruned_loss=0.05768, over 16224.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2877, pruned_loss=0.05682, over 3089846.85 frames. ], batch size: 35, lr: 3.06e-03, grad_scale: 4.0 2023-05-01 11:36:43,096 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219403.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 11:36:43,733 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.827e+02 2.722e+02 3.215e+02 3.900e+02 7.497e+02, threshold=6.430e+02, percent-clipped=6.0 2023-05-01 11:37:20,975 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219428.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:37:57,562 INFO [train.py:904] (6/8) Epoch 22, batch 6300, loss[loss=0.1958, simple_loss=0.2821, pruned_loss=0.05476, over 16528.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2882, pruned_loss=0.05683, over 3101050.77 frames. ], batch size: 68, lr: 3.06e-03, grad_scale: 4.0 2023-05-01 11:38:55,982 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219490.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:39:13,658 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.01 vs. limit=5.0 2023-05-01 11:39:15,298 INFO [train.py:904] (6/8) Epoch 22, batch 6350, loss[loss=0.2244, simple_loss=0.2922, pruned_loss=0.07829, over 11553.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2893, pruned_loss=0.05862, over 3061774.16 frames. ], batch size: 248, lr: 3.06e-03, grad_scale: 4.0 2023-05-01 11:39:16,415 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.781e+02 2.840e+02 3.669e+02 4.695e+02 9.321e+02, threshold=7.339e+02, percent-clipped=9.0 2023-05-01 11:40:07,785 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9260, 4.9944, 5.3418, 5.2900, 5.3348, 5.0047, 4.9544, 4.6953], device='cuda:6'), covar=tensor([0.0310, 0.0524, 0.0302, 0.0397, 0.0460, 0.0352, 0.0902, 0.0506], device='cuda:6'), in_proj_covar=tensor([0.0405, 0.0450, 0.0435, 0.0404, 0.0483, 0.0458, 0.0542, 0.0366], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 11:40:28,663 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219551.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:40:31,084 INFO [train.py:904] (6/8) Epoch 22, batch 6400, loss[loss=0.2021, simple_loss=0.2892, pruned_loss=0.05747, over 16616.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2888, pruned_loss=0.05914, over 3066341.20 frames. ], batch size: 134, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:40:40,453 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219559.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:40:43,791 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0154, 3.4861, 3.4076, 2.1029, 3.2364, 3.5398, 3.2525, 2.0108], device='cuda:6'), covar=tensor([0.0613, 0.0065, 0.0075, 0.0505, 0.0114, 0.0112, 0.0113, 0.0506], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0083, 0.0084, 0.0132, 0.0098, 0.0109, 0.0094, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 11:41:45,410 INFO [train.py:904] (6/8) Epoch 22, batch 6450, loss[loss=0.2107, simple_loss=0.2934, pruned_loss=0.064, over 11297.00 frames. ], tot_loss[loss=0.202, simple_loss=0.288, pruned_loss=0.05804, over 3075828.34 frames. ], batch size: 246, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:41:47,189 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.783e+02 2.795e+02 3.528e+02 4.334e+02 7.123e+02, threshold=7.056e+02, percent-clipped=0.0 2023-05-01 11:42:13,403 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219621.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:42:45,443 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219640.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:42:56,603 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7894, 3.8817, 2.3555, 4.5862, 3.0560, 4.4706, 2.5323, 3.1736], device='cuda:6'), covar=tensor([0.0291, 0.0421, 0.1773, 0.0184, 0.0824, 0.0593, 0.1489, 0.0757], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0175, 0.0194, 0.0161, 0.0176, 0.0216, 0.0200, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 11:43:04,572 INFO [train.py:904] (6/8) Epoch 22, batch 6500, loss[loss=0.2093, simple_loss=0.2855, pruned_loss=0.06654, over 11542.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2861, pruned_loss=0.05698, over 3114950.10 frames. ], batch size: 248, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:43:30,546 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219669.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:43:43,536 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219677.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:43:47,967 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5717, 2.5128, 2.1443, 4.2512, 2.4941, 3.8499, 1.6748, 2.5542], device='cuda:6'), covar=tensor([0.1741, 0.1048, 0.1657, 0.0256, 0.0382, 0.0538, 0.1971, 0.1130], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0174, 0.0195, 0.0190, 0.0207, 0.0216, 0.0203, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 11:43:57,078 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5785, 2.6656, 2.2640, 3.9213, 2.7510, 3.9199, 1.4106, 2.7582], device='cuda:6'), covar=tensor([0.1499, 0.0800, 0.1448, 0.0180, 0.0254, 0.0435, 0.1892, 0.0921], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0174, 0.0195, 0.0190, 0.0207, 0.0216, 0.0203, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 11:44:01,334 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219688.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:44:13,894 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.8855, 6.2450, 5.9587, 6.0890, 5.6497, 5.5607, 5.5800, 6.3752], device='cuda:6'), covar=tensor([0.1345, 0.0832, 0.1028, 0.0759, 0.0797, 0.0537, 0.1264, 0.0843], device='cuda:6'), in_proj_covar=tensor([0.0671, 0.0817, 0.0678, 0.0622, 0.0516, 0.0527, 0.0685, 0.0639], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 11:44:19,768 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219698.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 11:44:26,639 INFO [train.py:904] (6/8) Epoch 22, batch 6550, loss[loss=0.2149, simple_loss=0.3133, pruned_loss=0.05827, over 15395.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2892, pruned_loss=0.05795, over 3103963.19 frames. ], batch size: 191, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:44:28,427 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 2.583e+02 3.105e+02 3.738e+02 8.268e+02, threshold=6.210e+02, percent-clipped=2.0 2023-05-01 11:44:29,500 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3745, 4.3788, 4.7276, 4.7314, 4.7060, 4.4715, 4.4357, 4.3622], device='cuda:6'), covar=tensor([0.0392, 0.0714, 0.0568, 0.0522, 0.0479, 0.0554, 0.0875, 0.0529], device='cuda:6'), in_proj_covar=tensor([0.0407, 0.0452, 0.0436, 0.0405, 0.0485, 0.0459, 0.0543, 0.0367], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 11:44:31,192 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.63 vs. limit=2.0 2023-05-01 11:45:03,215 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219725.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:45:08,807 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219728.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:45:38,190 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2945, 4.3812, 4.1942, 3.9450, 3.9386, 4.3174, 4.0226, 4.0536], device='cuda:6'), covar=tensor([0.0588, 0.0554, 0.0296, 0.0277, 0.0732, 0.0444, 0.0696, 0.0578], device='cuda:6'), in_proj_covar=tensor([0.0290, 0.0427, 0.0339, 0.0337, 0.0346, 0.0391, 0.0234, 0.0405], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 11:45:46,184 INFO [train.py:904] (6/8) Epoch 22, batch 6600, loss[loss=0.1957, simple_loss=0.2848, pruned_loss=0.05329, over 16895.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2917, pruned_loss=0.05899, over 3091665.66 frames. ], batch size: 116, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:46:02,178 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219763.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:46:23,295 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219776.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:46:30,341 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-01 11:47:06,508 INFO [train.py:904] (6/8) Epoch 22, batch 6650, loss[loss=0.2033, simple_loss=0.2904, pruned_loss=0.05804, over 15413.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2923, pruned_loss=0.06, over 3073039.95 frames. ], batch size: 191, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:47:07,640 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.837e+02 2.943e+02 3.397e+02 4.500e+02 8.184e+02, threshold=6.793e+02, percent-clipped=7.0 2023-05-01 11:47:39,805 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219824.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:48:10,752 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219846.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:48:20,373 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8046, 3.8526, 3.9371, 3.7292, 3.8746, 4.2631, 3.9017, 3.6335], device='cuda:6'), covar=tensor([0.2035, 0.2074, 0.2455, 0.2236, 0.2442, 0.1610, 0.1581, 0.2418], device='cuda:6'), in_proj_covar=tensor([0.0407, 0.0586, 0.0649, 0.0488, 0.0651, 0.0676, 0.0510, 0.0656], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 11:48:21,239 INFO [train.py:904] (6/8) Epoch 22, batch 6700, loss[loss=0.2346, simple_loss=0.3049, pruned_loss=0.08213, over 11418.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2908, pruned_loss=0.06015, over 3059504.34 frames. ], batch size: 246, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:48:27,613 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219857.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:48:30,130 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219859.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:49:04,116 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.62 vs. limit=5.0 2023-05-01 11:49:36,044 INFO [train.py:904] (6/8) Epoch 22, batch 6750, loss[loss=0.1778, simple_loss=0.2656, pruned_loss=0.04506, over 17271.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2891, pruned_loss=0.05925, over 3080857.30 frames. ], batch size: 52, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:49:37,872 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.993e+02 2.924e+02 3.493e+02 4.502e+02 8.889e+02, threshold=6.985e+02, percent-clipped=2.0 2023-05-01 11:49:43,893 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219907.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:50:00,742 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219918.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:50:54,124 INFO [train.py:904] (6/8) Epoch 22, batch 6800, loss[loss=0.179, simple_loss=0.2666, pruned_loss=0.04575, over 16585.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2882, pruned_loss=0.05861, over 3090985.89 frames. ], batch size: 68, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:51:53,311 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0879, 5.1626, 4.9358, 4.5560, 4.5505, 5.0455, 4.9270, 4.7024], device='cuda:6'), covar=tensor([0.0788, 0.0913, 0.0444, 0.0468, 0.1143, 0.0661, 0.0528, 0.0872], device='cuda:6'), in_proj_covar=tensor([0.0290, 0.0428, 0.0341, 0.0338, 0.0347, 0.0393, 0.0235, 0.0406], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 11:52:04,945 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219998.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 11:52:15,008 INFO [train.py:904] (6/8) Epoch 22, batch 6850, loss[loss=0.2103, simple_loss=0.3185, pruned_loss=0.05103, over 16861.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2902, pruned_loss=0.05973, over 3075091.28 frames. ], batch size: 116, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:52:16,804 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.941e+02 2.883e+02 3.357e+02 3.930e+02 6.765e+02, threshold=6.713e+02, percent-clipped=0.0 2023-05-01 11:53:20,649 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220046.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 11:53:31,113 INFO [train.py:904] (6/8) Epoch 22, batch 6900, loss[loss=0.2109, simple_loss=0.301, pruned_loss=0.06035, over 16424.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2919, pruned_loss=0.05878, over 3108182.40 frames. ], batch size: 146, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:54:24,739 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220087.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:54:41,863 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220098.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:54:49,991 INFO [train.py:904] (6/8) Epoch 22, batch 6950, loss[loss=0.1854, simple_loss=0.2767, pruned_loss=0.04703, over 16669.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2935, pruned_loss=0.06, over 3120683.42 frames. ], batch size: 83, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:54:51,079 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.857e+02 2.861e+02 3.409e+02 4.251e+02 1.328e+03, threshold=6.818e+02, percent-clipped=1.0 2023-05-01 11:55:14,107 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220119.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:55:54,751 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220146.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:55:57,597 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220148.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:56:03,772 INFO [train.py:904] (6/8) Epoch 22, batch 7000, loss[loss=0.178, simple_loss=0.2858, pruned_loss=0.03507, over 16935.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2936, pruned_loss=0.05944, over 3114366.95 frames. ], batch size: 90, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 11:56:14,316 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220159.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:57:05,684 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220194.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:57:18,841 INFO [train.py:904] (6/8) Epoch 22, batch 7050, loss[loss=0.249, simple_loss=0.3083, pruned_loss=0.09482, over 11256.00 frames. ], tot_loss[loss=0.206, simple_loss=0.2944, pruned_loss=0.0588, over 3134267.88 frames. ], batch size: 248, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 11:57:21,954 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.068e+02 2.577e+02 3.204e+02 3.781e+02 6.030e+02, threshold=6.408e+02, percent-clipped=0.0 2023-05-01 11:57:35,243 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220213.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:57:59,366 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220229.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:58:36,505 INFO [train.py:904] (6/8) Epoch 22, batch 7100, loss[loss=0.2274, simple_loss=0.2919, pruned_loss=0.08144, over 11172.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2927, pruned_loss=0.05844, over 3119578.91 frames. ], batch size: 246, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 11:58:45,427 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220258.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:58:46,720 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5398, 3.5958, 3.3001, 2.9698, 3.2091, 3.5027, 3.2963, 3.3378], device='cuda:6'), covar=tensor([0.0608, 0.0724, 0.0317, 0.0288, 0.0515, 0.0491, 0.1454, 0.0487], device='cuda:6'), in_proj_covar=tensor([0.0288, 0.0425, 0.0339, 0.0335, 0.0344, 0.0390, 0.0233, 0.0403], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 11:59:35,833 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220290.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:59:54,480 INFO [train.py:904] (6/8) Epoch 22, batch 7150, loss[loss=0.234, simple_loss=0.2984, pruned_loss=0.08482, over 11671.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2908, pruned_loss=0.05875, over 3096573.41 frames. ], batch size: 246, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 11:59:58,136 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.701e+02 3.416e+02 4.420e+02 1.024e+03, threshold=6.833e+02, percent-clipped=4.0 2023-05-01 12:00:20,943 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220319.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:00:48,275 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1685, 3.1872, 1.9707, 3.4494, 2.4171, 3.4988, 2.0622, 2.6106], device='cuda:6'), covar=tensor([0.0322, 0.0430, 0.1748, 0.0208, 0.0954, 0.0664, 0.1741, 0.0869], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0174, 0.0193, 0.0160, 0.0175, 0.0215, 0.0200, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 12:01:07,951 INFO [train.py:904] (6/8) Epoch 22, batch 7200, loss[loss=0.179, simple_loss=0.2731, pruned_loss=0.04249, over 16818.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2889, pruned_loss=0.05768, over 3069559.14 frames. ], batch size: 83, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:01:35,467 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0199, 4.0928, 3.9132, 3.6396, 3.6818, 4.0260, 3.6889, 3.7985], device='cuda:6'), covar=tensor([0.0614, 0.0700, 0.0293, 0.0282, 0.0683, 0.0545, 0.1065, 0.0592], device='cuda:6'), in_proj_covar=tensor([0.0286, 0.0423, 0.0337, 0.0334, 0.0342, 0.0387, 0.0232, 0.0401], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:02:00,852 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8268, 1.9221, 2.3742, 2.6682, 2.6538, 3.0853, 1.9415, 3.0339], device='cuda:6'), covar=tensor([0.0243, 0.0557, 0.0361, 0.0345, 0.0365, 0.0212, 0.0645, 0.0158], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0192, 0.0178, 0.0182, 0.0196, 0.0152, 0.0195, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:02:13,637 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-01 12:02:28,580 INFO [train.py:904] (6/8) Epoch 22, batch 7250, loss[loss=0.1787, simple_loss=0.2661, pruned_loss=0.04568, over 16912.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2867, pruned_loss=0.05664, over 3065425.33 frames. ], batch size: 116, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:02:30,895 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 2.548e+02 3.054e+02 3.716e+02 9.083e+02, threshold=6.108e+02, percent-clipped=2.0 2023-05-01 12:02:53,160 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220419.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:03:30,263 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220443.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:03:45,059 INFO [train.py:904] (6/8) Epoch 22, batch 7300, loss[loss=0.1924, simple_loss=0.2868, pruned_loss=0.04899, over 16359.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2859, pruned_loss=0.05639, over 3074791.46 frames. ], batch size: 146, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:03:47,414 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220454.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:04:01,606 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-01 12:04:07,574 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220467.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:04:29,909 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2537, 5.2691, 5.0867, 4.7419, 4.8421, 5.1730, 5.0112, 4.8597], device='cuda:6'), covar=tensor([0.0513, 0.0267, 0.0239, 0.0249, 0.0766, 0.0335, 0.0299, 0.0531], device='cuda:6'), in_proj_covar=tensor([0.0287, 0.0422, 0.0336, 0.0333, 0.0342, 0.0387, 0.0231, 0.0401], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:04:42,948 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-01 12:04:48,678 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-05-01 12:05:02,409 INFO [train.py:904] (6/8) Epoch 22, batch 7350, loss[loss=0.2184, simple_loss=0.3072, pruned_loss=0.06477, over 16715.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.287, pruned_loss=0.05736, over 3059290.44 frames. ], batch size: 124, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:05:05,570 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.460e+02 2.713e+02 3.260e+02 3.989e+02 1.125e+03, threshold=6.520e+02, percent-clipped=5.0 2023-05-01 12:05:18,393 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220513.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:06:20,489 INFO [train.py:904] (6/8) Epoch 22, batch 7400, loss[loss=0.1948, simple_loss=0.288, pruned_loss=0.05077, over 17119.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2882, pruned_loss=0.05787, over 3067305.50 frames. ], batch size: 49, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:06:25,403 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220555.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:06:35,403 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220561.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:06:57,766 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4796, 3.5298, 2.7579, 2.1655, 2.3250, 2.2574, 3.7322, 3.2086], device='cuda:6'), covar=tensor([0.3115, 0.0645, 0.1850, 0.2958, 0.2802, 0.2292, 0.0463, 0.1302], device='cuda:6'), in_proj_covar=tensor([0.0328, 0.0269, 0.0306, 0.0315, 0.0299, 0.0260, 0.0296, 0.0337], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 12:07:01,329 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1594, 5.1759, 4.9808, 4.5717, 4.6426, 5.0630, 5.0130, 4.7650], device='cuda:6'), covar=tensor([0.0593, 0.0457, 0.0320, 0.0335, 0.0986, 0.0427, 0.0295, 0.0709], device='cuda:6'), in_proj_covar=tensor([0.0286, 0.0421, 0.0336, 0.0332, 0.0342, 0.0386, 0.0231, 0.0400], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:07:13,076 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220585.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:07:13,315 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3852, 3.2890, 2.6819, 2.1458, 2.1858, 2.2022, 3.4276, 3.0175], device='cuda:6'), covar=tensor([0.2990, 0.0654, 0.1777, 0.2818, 0.2602, 0.2326, 0.0512, 0.1427], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0268, 0.0305, 0.0314, 0.0298, 0.0260, 0.0296, 0.0337], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 12:07:41,479 INFO [train.py:904] (6/8) Epoch 22, batch 7450, loss[loss=0.2092, simple_loss=0.3048, pruned_loss=0.05685, over 16351.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2889, pruned_loss=0.05845, over 3089322.28 frames. ], batch size: 146, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:07:43,942 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.847e+02 2.745e+02 3.256e+02 3.859e+02 7.938e+02, threshold=6.512e+02, percent-clipped=1.0 2023-05-01 12:07:46,527 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0914, 1.4808, 1.9111, 2.0925, 2.2144, 2.3550, 1.6895, 2.2582], device='cuda:6'), covar=tensor([0.0237, 0.0512, 0.0287, 0.0331, 0.0309, 0.0225, 0.0531, 0.0163], device='cuda:6'), in_proj_covar=tensor([0.0186, 0.0192, 0.0178, 0.0182, 0.0195, 0.0152, 0.0195, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:08:01,208 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220614.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:08:04,667 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220616.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 12:09:01,337 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220651.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:09:03,161 INFO [train.py:904] (6/8) Epoch 22, batch 7500, loss[loss=0.2152, simple_loss=0.3056, pruned_loss=0.06235, over 16276.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2903, pruned_loss=0.05861, over 3073269.02 frames. ], batch size: 165, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:10:21,148 INFO [train.py:904] (6/8) Epoch 22, batch 7550, loss[loss=0.1903, simple_loss=0.2781, pruned_loss=0.05126, over 16806.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2896, pruned_loss=0.05892, over 3067078.13 frames. ], batch size: 83, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:10:24,494 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.979e+02 2.918e+02 3.506e+02 4.599e+02 8.060e+02, threshold=7.013e+02, percent-clipped=6.0 2023-05-01 12:10:36,064 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220712.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:10:48,118 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8481, 2.0222, 2.3519, 3.1171, 2.1842, 2.2273, 2.2797, 2.1639], device='cuda:6'), covar=tensor([0.1363, 0.3560, 0.2413, 0.0655, 0.3984, 0.2487, 0.3264, 0.3483], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0447, 0.0364, 0.0324, 0.0435, 0.0514, 0.0418, 0.0520], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:11:23,752 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220743.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:11:37,871 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7634, 3.8659, 1.9418, 4.6013, 2.8507, 4.4625, 2.0884, 2.9743], device='cuda:6'), covar=tensor([0.0286, 0.0425, 0.2288, 0.0181, 0.0886, 0.0484, 0.2100, 0.0924], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0174, 0.0192, 0.0160, 0.0174, 0.0214, 0.0200, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 12:11:38,541 INFO [train.py:904] (6/8) Epoch 22, batch 7600, loss[loss=0.1812, simple_loss=0.2683, pruned_loss=0.04702, over 17284.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2886, pruned_loss=0.05905, over 3054123.09 frames. ], batch size: 52, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:11:40,699 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220754.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:12:37,572 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220791.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:12:55,479 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220802.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:12:56,364 INFO [train.py:904] (6/8) Epoch 22, batch 7650, loss[loss=0.2128, simple_loss=0.3025, pruned_loss=0.06157, over 16362.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2894, pruned_loss=0.05923, over 3058635.42 frames. ], batch size: 146, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:12:59,171 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.956e+02 2.818e+02 3.454e+02 4.152e+02 9.180e+02, threshold=6.907e+02, percent-clipped=1.0 2023-05-01 12:13:27,063 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220822.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:13:47,975 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3420, 3.4147, 3.7493, 2.0265, 3.0626, 2.4655, 3.6712, 3.6849], device='cuda:6'), covar=tensor([0.0228, 0.0941, 0.0534, 0.2293, 0.0913, 0.0968, 0.0621, 0.1049], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0165, 0.0167, 0.0154, 0.0145, 0.0130, 0.0143, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 12:14:11,869 INFO [train.py:904] (6/8) Epoch 22, batch 7700, loss[loss=0.1978, simple_loss=0.2834, pruned_loss=0.05616, over 15299.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2902, pruned_loss=0.06052, over 3038294.72 frames. ], batch size: 190, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:14:30,136 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7980, 3.8711, 4.1214, 4.0947, 4.1178, 3.8857, 3.8881, 3.8809], device='cuda:6'), covar=tensor([0.0392, 0.0729, 0.0444, 0.0457, 0.0504, 0.0492, 0.0931, 0.0587], device='cuda:6'), in_proj_covar=tensor([0.0407, 0.0452, 0.0436, 0.0406, 0.0484, 0.0460, 0.0545, 0.0369], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 12:14:49,258 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-05-01 12:14:58,201 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220883.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:15:01,841 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220885.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:15:29,151 INFO [train.py:904] (6/8) Epoch 22, batch 7750, loss[loss=0.1672, simple_loss=0.266, pruned_loss=0.03414, over 16830.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2907, pruned_loss=0.0602, over 3068527.46 frames. ], batch size: 102, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:15:30,880 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220904.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:15:32,047 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.793e+02 2.706e+02 3.407e+02 4.136e+02 9.876e+02, threshold=6.815e+02, percent-clipped=5.0 2023-05-01 12:15:32,711 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3752, 3.0351, 2.5998, 2.2873, 2.2352, 2.1603, 3.0231, 2.8317], device='cuda:6'), covar=tensor([0.2449, 0.0791, 0.1846, 0.2570, 0.2734, 0.2499, 0.0611, 0.1364], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0268, 0.0305, 0.0314, 0.0298, 0.0260, 0.0295, 0.0336], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 12:15:40,128 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220911.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 12:15:45,839 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220914.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:16:15,124 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220933.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:16:42,220 INFO [train.py:904] (6/8) Epoch 22, batch 7800, loss[loss=0.2657, simple_loss=0.3281, pruned_loss=0.1017, over 11341.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2916, pruned_loss=0.0611, over 3062487.81 frames. ], batch size: 246, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:16:46,925 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-01 12:16:54,234 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4395, 3.2540, 3.6053, 1.8153, 3.7354, 3.7912, 2.9152, 2.8154], device='cuda:6'), covar=tensor([0.0797, 0.0285, 0.0215, 0.1252, 0.0085, 0.0202, 0.0473, 0.0469], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0108, 0.0098, 0.0138, 0.0080, 0.0125, 0.0129, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 12:16:56,582 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220962.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:17:00,314 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220965.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:17:55,496 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-05-01 12:17:55,724 INFO [train.py:904] (6/8) Epoch 22, batch 7850, loss[loss=0.1907, simple_loss=0.2794, pruned_loss=0.05102, over 16871.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2921, pruned_loss=0.06048, over 3082018.28 frames. ], batch size: 116, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:17:58,013 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.636e+02 2.766e+02 3.554e+02 4.297e+02 6.446e+02, threshold=7.108e+02, percent-clipped=0.0 2023-05-01 12:18:02,912 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=221007.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:18:56,084 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5135, 2.2059, 1.8950, 1.9615, 2.5309, 2.1819, 2.3208, 2.6541], device='cuda:6'), covar=tensor([0.0213, 0.0401, 0.0514, 0.0435, 0.0248, 0.0373, 0.0225, 0.0263], device='cuda:6'), in_proj_covar=tensor([0.0206, 0.0232, 0.0224, 0.0222, 0.0233, 0.0230, 0.0231, 0.0226], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:19:09,450 INFO [train.py:904] (6/8) Epoch 22, batch 7900, loss[loss=0.2195, simple_loss=0.2879, pruned_loss=0.07559, over 11017.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2906, pruned_loss=0.05946, over 3077724.82 frames. ], batch size: 247, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:19:55,998 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6009, 2.5262, 2.4708, 3.8886, 2.8592, 3.8281, 1.4066, 2.9033], device='cuda:6'), covar=tensor([0.1391, 0.0804, 0.1198, 0.0185, 0.0231, 0.0345, 0.1748, 0.0762], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0174, 0.0194, 0.0189, 0.0207, 0.0214, 0.0202, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 12:20:27,100 INFO [train.py:904] (6/8) Epoch 22, batch 7950, loss[loss=0.1965, simple_loss=0.2772, pruned_loss=0.05791, over 16322.00 frames. ], tot_loss[loss=0.205, simple_loss=0.2903, pruned_loss=0.05988, over 3069275.29 frames. ], batch size: 68, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:20:32,010 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.765e+02 2.674e+02 3.039e+02 3.526e+02 5.687e+02, threshold=6.078e+02, percent-clipped=0.0 2023-05-01 12:21:00,136 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2999, 5.6321, 5.4004, 5.4485, 5.1034, 4.9931, 5.0352, 5.7570], device='cuda:6'), covar=tensor([0.1242, 0.0834, 0.0954, 0.0801, 0.0820, 0.0804, 0.1217, 0.0838], device='cuda:6'), in_proj_covar=tensor([0.0668, 0.0811, 0.0674, 0.0617, 0.0512, 0.0526, 0.0681, 0.0636], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:21:41,818 INFO [train.py:904] (6/8) Epoch 22, batch 8000, loss[loss=0.2417, simple_loss=0.3086, pruned_loss=0.08742, over 10871.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2918, pruned_loss=0.06127, over 3048877.85 frames. ], batch size: 246, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:21:51,291 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0688, 3.9723, 4.1273, 4.2670, 4.3821, 3.9315, 4.2893, 4.3880], device='cuda:6'), covar=tensor([0.1732, 0.1204, 0.1400, 0.0703, 0.0606, 0.1595, 0.0948, 0.0760], device='cuda:6'), in_proj_covar=tensor([0.0631, 0.0778, 0.0903, 0.0784, 0.0598, 0.0625, 0.0651, 0.0752], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:22:19,111 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=221178.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:22:23,860 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 12:22:32,712 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 12:22:35,015 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-01 12:22:55,167 INFO [train.py:904] (6/8) Epoch 22, batch 8050, loss[loss=0.2078, simple_loss=0.298, pruned_loss=0.05877, over 16207.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2914, pruned_loss=0.0607, over 3062121.99 frames. ], batch size: 165, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:23:01,251 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.760e+02 2.816e+02 3.381e+02 4.131e+02 1.220e+03, threshold=6.763e+02, percent-clipped=7.0 2023-05-01 12:23:07,982 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=221211.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:23:30,503 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3950, 3.4212, 2.4005, 2.1313, 2.1366, 2.0576, 3.5021, 2.9598], device='cuda:6'), covar=tensor([0.3353, 0.0776, 0.2371, 0.2937, 0.3141, 0.2708, 0.0681, 0.1511], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0267, 0.0304, 0.0313, 0.0297, 0.0260, 0.0295, 0.0335], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 12:24:01,679 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3942, 4.5488, 4.6690, 4.4940, 4.5339, 5.0526, 4.5705, 4.3104], device='cuda:6'), covar=tensor([0.1438, 0.1863, 0.2298, 0.2012, 0.2517, 0.1038, 0.1661, 0.2375], device='cuda:6'), in_proj_covar=tensor([0.0411, 0.0591, 0.0654, 0.0491, 0.0653, 0.0684, 0.0515, 0.0659], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 12:24:08,783 INFO [train.py:904] (6/8) Epoch 22, batch 8100, loss[loss=0.2173, simple_loss=0.2954, pruned_loss=0.06956, over 11751.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2906, pruned_loss=0.05986, over 3078567.76 frames. ], batch size: 248, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:24:17,302 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=221259.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:24:19,975 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=221260.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:24:33,083 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4395, 4.2573, 4.3953, 4.6925, 4.8490, 4.3881, 4.8786, 4.8426], device='cuda:6'), covar=tensor([0.2188, 0.1546, 0.2308, 0.0964, 0.0838, 0.1326, 0.0878, 0.0966], device='cuda:6'), in_proj_covar=tensor([0.0631, 0.0777, 0.0903, 0.0783, 0.0598, 0.0625, 0.0651, 0.0752], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:25:24,332 INFO [train.py:904] (6/8) Epoch 22, batch 8150, loss[loss=0.2184, simple_loss=0.2857, pruned_loss=0.07555, over 11693.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2886, pruned_loss=0.05923, over 3065908.53 frames. ], batch size: 246, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:25:31,013 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.995e+02 2.655e+02 3.146e+02 3.866e+02 6.459e+02, threshold=6.292e+02, percent-clipped=0.0 2023-05-01 12:25:32,117 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=221307.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:26:41,083 INFO [train.py:904] (6/8) Epoch 22, batch 8200, loss[loss=0.1651, simple_loss=0.2472, pruned_loss=0.04151, over 16454.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2856, pruned_loss=0.05807, over 3075777.63 frames. ], batch size: 35, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:26:44,114 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=221355.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:27:00,205 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6513, 1.8935, 2.2839, 2.6060, 2.6188, 2.9810, 2.0038, 2.9361], device='cuda:6'), covar=tensor([0.0235, 0.0530, 0.0367, 0.0324, 0.0332, 0.0215, 0.0546, 0.0182], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0190, 0.0176, 0.0181, 0.0194, 0.0151, 0.0193, 0.0149], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:27:58,678 INFO [train.py:904] (6/8) Epoch 22, batch 8250, loss[loss=0.1684, simple_loss=0.2717, pruned_loss=0.0325, over 16663.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2844, pruned_loss=0.05571, over 3048272.42 frames. ], batch size: 134, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:28:05,605 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.564e+02 2.638e+02 3.060e+02 3.657e+02 7.056e+02, threshold=6.120e+02, percent-clipped=1.0 2023-05-01 12:28:06,885 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8180, 3.7204, 3.8846, 4.0020, 4.1071, 3.7104, 4.0292, 4.1141], device='cuda:6'), covar=tensor([0.1784, 0.1254, 0.1425, 0.0752, 0.0621, 0.1799, 0.0838, 0.0781], device='cuda:6'), in_proj_covar=tensor([0.0629, 0.0776, 0.0900, 0.0783, 0.0597, 0.0625, 0.0650, 0.0752], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:28:29,299 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5532, 4.4922, 4.3541, 3.5649, 4.3667, 1.6617, 4.1443, 4.1641], device='cuda:6'), covar=tensor([0.0104, 0.0116, 0.0194, 0.0381, 0.0123, 0.2948, 0.0142, 0.0246], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0154, 0.0197, 0.0176, 0.0173, 0.0206, 0.0185, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:29:10,645 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2023-05-01 12:29:17,813 INFO [train.py:904] (6/8) Epoch 22, batch 8300, loss[loss=0.2022, simple_loss=0.2948, pruned_loss=0.05476, over 15417.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2816, pruned_loss=0.0525, over 3066200.54 frames. ], batch size: 190, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:29:57,608 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0435, 5.0588, 4.9012, 4.4848, 4.5556, 4.9791, 4.9203, 4.6471], device='cuda:6'), covar=tensor([0.0556, 0.0538, 0.0314, 0.0333, 0.1033, 0.0504, 0.0263, 0.0744], device='cuda:6'), in_proj_covar=tensor([0.0287, 0.0424, 0.0337, 0.0332, 0.0342, 0.0388, 0.0232, 0.0402], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:29:57,617 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=221478.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:30:38,296 INFO [train.py:904] (6/8) Epoch 22, batch 8350, loss[loss=0.2169, simple_loss=0.2984, pruned_loss=0.06768, over 12259.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2822, pruned_loss=0.05113, over 3078371.72 frames. ], batch size: 247, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:30:43,698 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.406e+02 2.123e+02 2.436e+02 3.235e+02 5.612e+02, threshold=4.873e+02, percent-clipped=0.0 2023-05-01 12:30:44,322 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=221507.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:30:47,604 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 12:30:57,019 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2457, 4.2831, 4.4467, 4.2531, 4.3627, 4.8519, 4.4168, 4.0703], device='cuda:6'), covar=tensor([0.1676, 0.2152, 0.2088, 0.2171, 0.2434, 0.0999, 0.1590, 0.2606], device='cuda:6'), in_proj_covar=tensor([0.0405, 0.0581, 0.0644, 0.0482, 0.0641, 0.0674, 0.0508, 0.0649], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 12:31:15,660 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=221526.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:31:22,585 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 12:31:56,139 INFO [train.py:904] (6/8) Epoch 22, batch 8400, loss[loss=0.1804, simple_loss=0.2657, pruned_loss=0.04756, over 12381.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.2796, pruned_loss=0.04945, over 3062844.18 frames. ], batch size: 248, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:32:08,922 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=221560.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:32:21,118 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=221568.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:33:07,575 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4522, 4.4241, 4.2261, 3.4635, 4.3132, 1.7920, 4.0876, 4.0598], device='cuda:6'), covar=tensor([0.0096, 0.0113, 0.0213, 0.0310, 0.0114, 0.2699, 0.0149, 0.0242], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0152, 0.0195, 0.0174, 0.0171, 0.0204, 0.0184, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:33:17,379 INFO [train.py:904] (6/8) Epoch 22, batch 8450, loss[loss=0.1989, simple_loss=0.2867, pruned_loss=0.05559, over 16097.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2772, pruned_loss=0.04708, over 3064797.85 frames. ], batch size: 35, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:33:24,323 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.610e+02 2.296e+02 2.720e+02 3.439e+02 5.542e+02, threshold=5.440e+02, percent-clipped=2.0 2023-05-01 12:33:26,676 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=221608.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:33:46,241 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6778, 2.7508, 2.4736, 4.0511, 2.4236, 4.0061, 1.5669, 2.8070], device='cuda:6'), covar=tensor([0.1511, 0.0824, 0.1223, 0.0220, 0.0169, 0.0383, 0.1850, 0.0796], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0172, 0.0192, 0.0186, 0.0204, 0.0212, 0.0200, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 12:34:09,543 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4149, 2.3695, 2.3655, 4.1600, 2.2169, 2.6661, 2.4497, 2.5122], device='cuda:6'), covar=tensor([0.1134, 0.3585, 0.3070, 0.0458, 0.4287, 0.2611, 0.3604, 0.3419], device='cuda:6'), in_proj_covar=tensor([0.0393, 0.0440, 0.0361, 0.0319, 0.0429, 0.0507, 0.0413, 0.0514], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:34:38,835 INFO [train.py:904] (6/8) Epoch 22, batch 8500, loss[loss=0.1642, simple_loss=0.2587, pruned_loss=0.03485, over 16753.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.2737, pruned_loss=0.04507, over 3077643.17 frames. ], batch size: 124, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:34:46,869 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7117, 3.1353, 3.4674, 2.1465, 2.9072, 2.2070, 3.2924, 3.3520], device='cuda:6'), covar=tensor([0.0302, 0.0923, 0.0465, 0.2118, 0.0857, 0.1088, 0.0675, 0.0944], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0160, 0.0162, 0.0150, 0.0142, 0.0127, 0.0139, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 12:35:04,943 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8240, 4.1369, 4.1523, 3.0111, 3.7355, 4.1803, 3.7770, 2.5614], device='cuda:6'), covar=tensor([0.0424, 0.0055, 0.0039, 0.0308, 0.0090, 0.0088, 0.0073, 0.0450], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0082, 0.0083, 0.0131, 0.0096, 0.0108, 0.0093, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 12:36:02,531 INFO [train.py:904] (6/8) Epoch 22, batch 8550, loss[loss=0.1897, simple_loss=0.2857, pruned_loss=0.04689, over 16157.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2715, pruned_loss=0.04393, over 3093356.79 frames. ], batch size: 165, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:36:10,060 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.073e+02 2.527e+02 2.950e+02 5.682e+02, threshold=5.053e+02, percent-clipped=1.0 2023-05-01 12:37:41,422 INFO [train.py:904] (6/8) Epoch 22, batch 8600, loss[loss=0.1758, simple_loss=0.2776, pruned_loss=0.03696, over 16676.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2713, pruned_loss=0.04318, over 3054883.59 frames. ], batch size: 57, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:37:53,159 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=221758.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:38:32,774 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6075, 3.6908, 3.4981, 3.2092, 3.3047, 3.6103, 3.3411, 3.4499], device='cuda:6'), covar=tensor([0.0606, 0.0597, 0.0325, 0.0286, 0.0555, 0.0513, 0.1421, 0.0512], device='cuda:6'), in_proj_covar=tensor([0.0284, 0.0418, 0.0333, 0.0328, 0.0337, 0.0382, 0.0229, 0.0397], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:39:21,614 INFO [train.py:904] (6/8) Epoch 22, batch 8650, loss[loss=0.1465, simple_loss=0.2431, pruned_loss=0.02499, over 11680.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2695, pruned_loss=0.04166, over 3035821.44 frames. ], batch size: 247, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:39:30,559 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.618e+02 2.228e+02 2.506e+02 3.024e+02 6.242e+02, threshold=5.012e+02, percent-clipped=2.0 2023-05-01 12:39:47,474 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7831, 3.8801, 3.9700, 3.7600, 3.9144, 4.3270, 3.9663, 3.6421], device='cuda:6'), covar=tensor([0.2216, 0.2066, 0.2359, 0.2406, 0.2875, 0.1538, 0.1591, 0.2552], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0575, 0.0637, 0.0478, 0.0634, 0.0668, 0.0504, 0.0641], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 12:39:51,178 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0992, 5.3852, 5.1238, 5.1734, 4.9163, 4.8800, 4.7113, 5.4838], device='cuda:6'), covar=tensor([0.1144, 0.0804, 0.1122, 0.0811, 0.0753, 0.0846, 0.1205, 0.0845], device='cuda:6'), in_proj_covar=tensor([0.0656, 0.0799, 0.0661, 0.0606, 0.0503, 0.0517, 0.0667, 0.0624], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:39:58,978 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=221819.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:40:41,441 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9570, 2.1466, 2.3913, 3.2235, 2.2588, 2.3347, 2.3448, 2.2900], device='cuda:6'), covar=tensor([0.1367, 0.3961, 0.2761, 0.0706, 0.4250, 0.2811, 0.3733, 0.3536], device='cuda:6'), in_proj_covar=tensor([0.0392, 0.0439, 0.0359, 0.0318, 0.0428, 0.0505, 0.0411, 0.0511], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:40:45,124 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-05-01 12:40:59,862 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7348, 2.5670, 2.3488, 3.7256, 1.8850, 3.8499, 1.4828, 2.8515], device='cuda:6'), covar=tensor([0.1468, 0.0883, 0.1245, 0.0172, 0.0084, 0.0320, 0.1851, 0.0771], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0173, 0.0192, 0.0185, 0.0203, 0.0212, 0.0200, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 12:41:06,645 INFO [train.py:904] (6/8) Epoch 22, batch 8700, loss[loss=0.1604, simple_loss=0.2542, pruned_loss=0.03327, over 16462.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2667, pruned_loss=0.04049, over 3051657.90 frames. ], batch size: 68, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:41:27,796 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=221863.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:41:48,569 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2359, 1.5882, 1.9346, 2.2250, 2.2792, 2.4915, 1.8528, 2.4096], device='cuda:6'), covar=tensor([0.0259, 0.0540, 0.0342, 0.0348, 0.0370, 0.0229, 0.0515, 0.0151], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0189, 0.0175, 0.0180, 0.0192, 0.0149, 0.0192, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:41:51,519 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 12:42:16,918 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=221890.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:42:42,867 INFO [train.py:904] (6/8) Epoch 22, batch 8750, loss[loss=0.1689, simple_loss=0.2718, pruned_loss=0.03295, over 16465.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2671, pruned_loss=0.04015, over 3058611.70 frames. ], batch size: 68, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:42:53,174 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.058e+02 2.561e+02 3.126e+02 5.404e+02, threshold=5.122e+02, percent-clipped=2.0 2023-05-01 12:44:31,203 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=221951.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:44:34,618 INFO [train.py:904] (6/8) Epoch 22, batch 8800, loss[loss=0.1683, simple_loss=0.2653, pruned_loss=0.03572, over 16351.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2658, pruned_loss=0.03945, over 3057135.13 frames. ], batch size: 146, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:45:37,736 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3988, 1.6173, 1.9733, 2.3284, 2.2978, 2.6677, 1.8484, 2.6467], device='cuda:6'), covar=tensor([0.0278, 0.0599, 0.0428, 0.0406, 0.0404, 0.0248, 0.0588, 0.0165], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0188, 0.0175, 0.0179, 0.0192, 0.0149, 0.0192, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:46:22,312 INFO [train.py:904] (6/8) Epoch 22, batch 8850, loss[loss=0.1687, simple_loss=0.2738, pruned_loss=0.03181, over 16244.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2682, pruned_loss=0.03859, over 3051729.80 frames. ], batch size: 165, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:46:28,904 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.701e+02 2.177e+02 2.540e+02 3.054e+02 9.241e+02, threshold=5.079e+02, percent-clipped=4.0 2023-05-01 12:46:35,818 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222009.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:48:07,872 INFO [train.py:904] (6/8) Epoch 22, batch 8900, loss[loss=0.1697, simple_loss=0.2627, pruned_loss=0.03833, over 16864.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2681, pruned_loss=0.03784, over 3046801.28 frames. ], batch size: 116, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 12:48:43,913 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=222070.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:50:11,688 INFO [train.py:904] (6/8) Epoch 22, batch 8950, loss[loss=0.1456, simple_loss=0.2463, pruned_loss=0.02247, over 16863.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2675, pruned_loss=0.03788, over 3061125.95 frames. ], batch size: 96, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 12:50:23,608 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.486e+02 2.055e+02 2.537e+02 3.242e+02 5.018e+02, threshold=5.074e+02, percent-clipped=0.0 2023-05-01 12:50:37,720 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222114.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:52:03,799 INFO [train.py:904] (6/8) Epoch 22, batch 9000, loss[loss=0.1545, simple_loss=0.2529, pruned_loss=0.02806, over 15329.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2647, pruned_loss=0.03661, over 3076888.25 frames. ], batch size: 191, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 12:52:03,799 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 12:52:14,709 INFO [train.py:938] (6/8) Epoch 22, validation: loss=0.1453, simple_loss=0.2492, pruned_loss=0.0207, over 944034.00 frames. 2023-05-01 12:52:14,710 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 12:52:36,573 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=222163.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:53:22,612 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222185.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:53:58,628 INFO [train.py:904] (6/8) Epoch 22, batch 9050, loss[loss=0.1612, simple_loss=0.25, pruned_loss=0.0362, over 16708.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2654, pruned_loss=0.03718, over 3076615.10 frames. ], batch size: 134, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 12:54:04,329 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222205.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:54:09,040 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.330e+02 2.762e+02 3.325e+02 5.542e+02, threshold=5.524e+02, percent-clipped=4.0 2023-05-01 12:54:17,175 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=222211.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:55:21,922 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8317, 2.8999, 2.5452, 4.6926, 3.2815, 4.2903, 1.5103, 3.1451], device='cuda:6'), covar=tensor([0.1346, 0.0745, 0.1202, 0.0160, 0.0159, 0.0314, 0.1713, 0.0668], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0172, 0.0192, 0.0184, 0.0201, 0.0212, 0.0200, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 12:55:31,423 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222246.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:55:31,566 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=222246.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 12:55:44,664 INFO [train.py:904] (6/8) Epoch 22, batch 9100, loss[loss=0.1799, simple_loss=0.2802, pruned_loss=0.03983, over 16200.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2652, pruned_loss=0.03754, over 3073134.36 frames. ], batch size: 166, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 12:55:56,093 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6691, 2.4537, 2.1938, 3.4775, 1.9552, 3.6069, 1.4559, 2.6955], device='cuda:6'), covar=tensor([0.1520, 0.0810, 0.1351, 0.0181, 0.0116, 0.0405, 0.1822, 0.0843], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0172, 0.0192, 0.0184, 0.0201, 0.0211, 0.0200, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 12:56:11,330 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=222266.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:56:33,936 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-01 12:56:46,282 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-01 12:57:12,071 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0536, 4.0292, 3.9015, 3.1864, 3.9653, 1.7371, 3.7572, 3.4779], device='cuda:6'), covar=tensor([0.0107, 0.0117, 0.0195, 0.0262, 0.0107, 0.2866, 0.0135, 0.0287], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0152, 0.0193, 0.0171, 0.0171, 0.0204, 0.0182, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:57:42,764 INFO [train.py:904] (6/8) Epoch 22, batch 9150, loss[loss=0.1567, simple_loss=0.2516, pruned_loss=0.0309, over 15542.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2645, pruned_loss=0.03712, over 3035223.17 frames. ], batch size: 192, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 12:57:53,762 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.620e+02 2.177e+02 2.613e+02 3.280e+02 4.905e+02, threshold=5.227e+02, percent-clipped=0.0 2023-05-01 12:59:11,106 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9765, 1.8310, 1.7000, 1.5209, 2.0089, 1.6386, 1.6302, 1.9159], device='cuda:6'), covar=tensor([0.0160, 0.0333, 0.0443, 0.0342, 0.0230, 0.0286, 0.0149, 0.0215], device='cuda:6'), in_proj_covar=tensor([0.0200, 0.0228, 0.0220, 0.0219, 0.0228, 0.0226, 0.0224, 0.0220], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 12:59:28,939 INFO [train.py:904] (6/8) Epoch 22, batch 9200, loss[loss=0.1617, simple_loss=0.2602, pruned_loss=0.03162, over 15533.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2603, pruned_loss=0.0361, over 3036537.23 frames. ], batch size: 190, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:59:40,665 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 12:59:52,286 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222365.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:00:15,384 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6718, 4.0228, 2.9971, 2.2363, 2.5023, 2.5160, 4.3224, 3.4272], device='cuda:6'), covar=tensor([0.2957, 0.0524, 0.1762, 0.3064, 0.2904, 0.2132, 0.0329, 0.1294], device='cuda:6'), in_proj_covar=tensor([0.0320, 0.0261, 0.0298, 0.0307, 0.0287, 0.0255, 0.0288, 0.0327], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 13:01:05,892 INFO [train.py:904] (6/8) Epoch 22, batch 9250, loss[loss=0.141, simple_loss=0.2284, pruned_loss=0.02682, over 12191.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2597, pruned_loss=0.03587, over 3040966.35 frames. ], batch size: 248, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:01:16,244 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.499e+02 2.130e+02 2.525e+02 3.050e+02 5.911e+02, threshold=5.049e+02, percent-clipped=2.0 2023-05-01 13:01:28,808 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=222414.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:02:15,260 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9657, 4.2499, 4.0582, 4.0971, 3.7327, 3.8649, 3.8319, 4.2420], device='cuda:6'), covar=tensor([0.1178, 0.0963, 0.1139, 0.0829, 0.0891, 0.1566, 0.1105, 0.0956], device='cuda:6'), in_proj_covar=tensor([0.0650, 0.0791, 0.0652, 0.0601, 0.0497, 0.0512, 0.0662, 0.0619], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 13:02:56,973 INFO [train.py:904] (6/8) Epoch 22, batch 9300, loss[loss=0.1426, simple_loss=0.236, pruned_loss=0.02465, over 16496.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.258, pruned_loss=0.03537, over 3022548.95 frames. ], batch size: 68, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:03:08,434 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-01 13:03:13,680 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-01 13:03:17,562 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=222462.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:04:10,124 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3233, 5.8976, 6.0109, 5.6399, 5.8094, 6.3072, 5.7225, 5.3780], device='cuda:6'), covar=tensor([0.0702, 0.1614, 0.1802, 0.2040, 0.2147, 0.0763, 0.1404, 0.2152], device='cuda:6'), in_proj_covar=tensor([0.0392, 0.0567, 0.0628, 0.0470, 0.0627, 0.0655, 0.0495, 0.0632], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 13:04:40,773 INFO [train.py:904] (6/8) Epoch 22, batch 9350, loss[loss=0.1639, simple_loss=0.2535, pruned_loss=0.0372, over 15377.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.258, pruned_loss=0.03555, over 3050626.72 frames. ], batch size: 190, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:04:49,911 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 1.994e+02 2.436e+02 3.069e+02 5.740e+02, threshold=4.871e+02, percent-clipped=2.0 2023-05-01 13:05:19,692 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2883, 2.4295, 2.0685, 2.2851, 2.8466, 2.4760, 2.8032, 3.0092], device='cuda:6'), covar=tensor([0.0141, 0.0440, 0.0556, 0.0483, 0.0278, 0.0433, 0.0245, 0.0258], device='cuda:6'), in_proj_covar=tensor([0.0201, 0.0228, 0.0221, 0.0220, 0.0228, 0.0227, 0.0224, 0.0220], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 13:05:23,311 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222523.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:05:58,169 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222541.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 13:06:08,601 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=222546.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:06:20,190 INFO [train.py:904] (6/8) Epoch 22, batch 9400, loss[loss=0.1552, simple_loss=0.2432, pruned_loss=0.03367, over 12428.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2582, pruned_loss=0.03501, over 3059822.59 frames. ], batch size: 247, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:06:38,160 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222561.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:06:55,320 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222570.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:07:24,493 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=222584.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 13:07:43,445 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=222594.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:08:00,581 INFO [train.py:904] (6/8) Epoch 22, batch 9450, loss[loss=0.1728, simple_loss=0.262, pruned_loss=0.04175, over 16741.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.26, pruned_loss=0.03543, over 3059439.05 frames. ], batch size: 83, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:08:08,412 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 2.238e+02 2.730e+02 3.189e+02 9.782e+02, threshold=5.460e+02, percent-clipped=5.0 2023-05-01 13:08:28,959 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-05-01 13:08:56,858 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=222631.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:09:40,758 INFO [train.py:904] (6/8) Epoch 22, batch 9500, loss[loss=0.1462, simple_loss=0.2328, pruned_loss=0.02981, over 12797.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2597, pruned_loss=0.03539, over 3072884.92 frames. ], batch size: 248, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:09:53,516 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-05-01 13:10:07,022 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=222665.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:10:09,974 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4002, 4.9023, 4.8329, 3.8066, 4.0758, 4.7557, 4.2144, 3.2289], device='cuda:6'), covar=tensor([0.0370, 0.0028, 0.0038, 0.0242, 0.0094, 0.0086, 0.0056, 0.0354], device='cuda:6'), in_proj_covar=tensor([0.0132, 0.0080, 0.0081, 0.0130, 0.0095, 0.0106, 0.0091, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 13:11:22,395 INFO [train.py:904] (6/8) Epoch 22, batch 9550, loss[loss=0.1647, simple_loss=0.2547, pruned_loss=0.03737, over 12141.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2597, pruned_loss=0.03559, over 3065864.39 frames. ], batch size: 246, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:11:31,439 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0244, 1.8425, 1.6814, 1.5053, 2.0203, 1.7089, 1.6349, 1.9818], device='cuda:6'), covar=tensor([0.0186, 0.0336, 0.0459, 0.0438, 0.0253, 0.0293, 0.0190, 0.0232], device='cuda:6'), in_proj_covar=tensor([0.0201, 0.0228, 0.0220, 0.0220, 0.0228, 0.0227, 0.0224, 0.0219], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 13:11:34,506 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.117e+02 2.369e+02 3.001e+02 4.954e+02, threshold=4.737e+02, percent-clipped=0.0 2023-05-01 13:11:44,579 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=222713.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:13:00,674 INFO [train.py:904] (6/8) Epoch 22, batch 9600, loss[loss=0.161, simple_loss=0.2508, pruned_loss=0.03557, over 12336.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2608, pruned_loss=0.03632, over 3045724.58 frames. ], batch size: 247, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:13:43,206 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.63 vs. limit=5.0 2023-05-01 13:14:44,326 INFO [train.py:904] (6/8) Epoch 22, batch 9650, loss[loss=0.192, simple_loss=0.2824, pruned_loss=0.05082, over 16476.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2622, pruned_loss=0.03648, over 3055301.71 frames. ], batch size: 146, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 13:14:58,744 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 2.257e+02 2.597e+02 3.466e+02 1.012e+03, threshold=5.195e+02, percent-clipped=6.0 2023-05-01 13:15:12,939 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8027, 2.7349, 2.8875, 2.0851, 2.6754, 2.0360, 2.6797, 2.7522], device='cuda:6'), covar=tensor([0.0320, 0.0878, 0.0512, 0.1992, 0.0822, 0.1014, 0.0634, 0.0872], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0158, 0.0163, 0.0150, 0.0141, 0.0127, 0.0139, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 13:15:26,326 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-05-01 13:16:03,718 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=222841.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 13:16:27,402 INFO [train.py:904] (6/8) Epoch 22, batch 9700, loss[loss=0.1605, simple_loss=0.2574, pruned_loss=0.03179, over 16143.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2608, pruned_loss=0.03618, over 3054818.65 frames. ], batch size: 165, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 13:16:43,422 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=222861.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:17:23,177 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222879.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 13:17:42,990 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=222889.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:18:08,540 INFO [train.py:904] (6/8) Epoch 22, batch 9750, loss[loss=0.1577, simple_loss=0.2554, pruned_loss=0.03003, over 16197.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2596, pruned_loss=0.03622, over 3046577.15 frames. ], batch size: 165, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 13:18:18,157 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.037e+02 2.441e+02 3.104e+02 5.362e+02, threshold=4.883e+02, percent-clipped=3.0 2023-05-01 13:18:19,214 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=222909.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:18:52,837 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222926.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:19:24,018 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222940.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:19:45,653 INFO [train.py:904] (6/8) Epoch 22, batch 9800, loss[loss=0.1711, simple_loss=0.2725, pruned_loss=0.03483, over 16909.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2593, pruned_loss=0.03542, over 3036907.07 frames. ], batch size: 116, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 13:20:15,202 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222969.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:20:17,825 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3719, 3.4209, 1.9210, 3.8545, 2.5480, 3.7849, 2.0607, 2.8255], device='cuda:6'), covar=tensor([0.0310, 0.0449, 0.1801, 0.0273, 0.0855, 0.0608, 0.1695, 0.0771], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0172, 0.0188, 0.0156, 0.0173, 0.0209, 0.0199, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 13:20:40,840 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222984.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:21:00,649 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2574, 3.2718, 1.9233, 3.6316, 2.3961, 3.5827, 2.1124, 2.7871], device='cuda:6'), covar=tensor([0.0326, 0.0445, 0.1748, 0.0249, 0.0909, 0.0596, 0.1655, 0.0759], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0171, 0.0188, 0.0156, 0.0172, 0.0208, 0.0199, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 13:21:23,117 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=223001.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:21:26,492 INFO [train.py:904] (6/8) Epoch 22, batch 9850, loss[loss=0.1841, simple_loss=0.2779, pruned_loss=0.04512, over 15414.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2608, pruned_loss=0.03515, over 3045262.78 frames. ], batch size: 190, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 13:21:37,440 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.461e+02 2.181e+02 2.565e+02 3.075e+02 7.070e+02, threshold=5.130e+02, percent-clipped=2.0 2023-05-01 13:22:19,885 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=223030.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:22:43,351 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9698, 1.7431, 1.5476, 1.3950, 1.8935, 1.5729, 1.5689, 1.9462], device='cuda:6'), covar=tensor([0.0216, 0.0434, 0.0611, 0.0507, 0.0308, 0.0393, 0.0171, 0.0310], device='cuda:6'), in_proj_covar=tensor([0.0200, 0.0228, 0.0221, 0.0220, 0.0229, 0.0227, 0.0223, 0.0220], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 13:23:01,421 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=223045.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:23:15,937 INFO [train.py:904] (6/8) Epoch 22, batch 9900, loss[loss=0.1614, simple_loss=0.2502, pruned_loss=0.03626, over 13101.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2613, pruned_loss=0.0352, over 3030591.50 frames. ], batch size: 250, lr: 3.03e-03, grad_scale: 4.0 2023-05-01 13:23:44,377 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=223064.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:24:16,647 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8097, 3.8253, 3.9609, 3.7328, 3.8812, 4.3015, 3.9352, 3.6552], device='cuda:6'), covar=tensor([0.1835, 0.2212, 0.2248, 0.2347, 0.2639, 0.1383, 0.1553, 0.2494], device='cuda:6'), in_proj_covar=tensor([0.0384, 0.0558, 0.0616, 0.0459, 0.0616, 0.0642, 0.0486, 0.0617], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 13:25:13,509 INFO [train.py:904] (6/8) Epoch 22, batch 9950, loss[loss=0.16, simple_loss=0.259, pruned_loss=0.03048, over 17162.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2635, pruned_loss=0.03559, over 3027219.79 frames. ], batch size: 48, lr: 3.03e-03, grad_scale: 4.0 2023-05-01 13:25:27,891 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.209e+02 2.588e+02 3.009e+02 4.332e+02, threshold=5.177e+02, percent-clipped=0.0 2023-05-01 13:26:07,209 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=223125.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:27:13,200 INFO [train.py:904] (6/8) Epoch 22, batch 10000, loss[loss=0.1784, simple_loss=0.279, pruned_loss=0.03888, over 16463.00 frames. ], tot_loss[loss=0.166, simple_loss=0.262, pruned_loss=0.03499, over 3065850.54 frames. ], batch size: 147, lr: 3.03e-03, grad_scale: 8.0 2023-05-01 13:27:26,221 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2023-05-01 13:27:48,969 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1279, 3.9316, 4.4135, 2.1303, 4.5946, 4.6775, 3.5081, 3.5449], device='cuda:6'), covar=tensor([0.0621, 0.0206, 0.0146, 0.1219, 0.0045, 0.0097, 0.0301, 0.0408], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0103, 0.0092, 0.0134, 0.0077, 0.0119, 0.0123, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 13:27:52,327 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=223173.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 13:28:04,989 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223179.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 13:28:45,588 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2175, 3.3495, 3.7603, 2.2057, 3.1101, 2.3141, 3.6160, 3.5547], device='cuda:6'), covar=tensor([0.0235, 0.0924, 0.0474, 0.1946, 0.0773, 0.0960, 0.0608, 0.1034], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0155, 0.0161, 0.0149, 0.0140, 0.0126, 0.0138, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 13:28:49,441 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2778, 3.6219, 3.6476, 2.4080, 3.3242, 3.6774, 3.4831, 1.8755], device='cuda:6'), covar=tensor([0.0557, 0.0066, 0.0070, 0.0457, 0.0124, 0.0119, 0.0098, 0.0681], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0081, 0.0082, 0.0131, 0.0096, 0.0106, 0.0092, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 13:28:54,091 INFO [train.py:904] (6/8) Epoch 22, batch 10050, loss[loss=0.1697, simple_loss=0.2606, pruned_loss=0.03937, over 12093.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2623, pruned_loss=0.03493, over 3076906.00 frames. ], batch size: 248, lr: 3.03e-03, grad_scale: 8.0 2023-05-01 13:29:04,328 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.317e+02 2.078e+02 2.413e+02 2.769e+02 4.519e+02, threshold=4.827e+02, percent-clipped=0.0 2023-05-01 13:29:05,355 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6026, 3.6323, 2.4629, 4.1849, 2.7131, 4.0990, 2.4375, 3.0745], device='cuda:6'), covar=tensor([0.0284, 0.0387, 0.1375, 0.0205, 0.0827, 0.0480, 0.1403, 0.0729], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0170, 0.0186, 0.0155, 0.0171, 0.0207, 0.0197, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-01 13:29:08,032 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3727, 2.9737, 2.6588, 2.2107, 2.1695, 2.2768, 3.0055, 2.8048], device='cuda:6'), covar=tensor([0.2595, 0.0642, 0.1572, 0.2638, 0.2468, 0.2007, 0.0445, 0.1357], device='cuda:6'), in_proj_covar=tensor([0.0315, 0.0257, 0.0294, 0.0301, 0.0281, 0.0250, 0.0283, 0.0322], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 13:29:40,793 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223226.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:29:42,095 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223227.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:29:54,715 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=223234.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 13:30:27,329 INFO [train.py:904] (6/8) Epoch 22, batch 10100, loss[loss=0.1553, simple_loss=0.2511, pruned_loss=0.02979, over 16687.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2627, pruned_loss=0.0353, over 3068681.46 frames. ], batch size: 134, lr: 3.03e-03, grad_scale: 8.0 2023-05-01 13:31:10,810 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223274.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:31:39,358 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=223296.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:32:13,437 INFO [train.py:904] (6/8) Epoch 23, batch 0, loss[loss=0.2249, simple_loss=0.2943, pruned_loss=0.07775, over 16779.00 frames. ], tot_loss[loss=0.2249, simple_loss=0.2943, pruned_loss=0.07775, over 16779.00 frames. ], batch size: 124, lr: 2.97e-03, grad_scale: 8.0 2023-05-01 13:32:13,437 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 13:32:20,849 INFO [train.py:938] (6/8) Epoch 23, validation: loss=0.1454, simple_loss=0.2487, pruned_loss=0.02111, over 944034.00 frames. 2023-05-01 13:32:20,850 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 13:32:28,407 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 2.627e+02 3.041e+02 3.739e+02 7.225e+02, threshold=6.083e+02, percent-clipped=7.0 2023-05-01 13:32:40,818 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8515, 4.1155, 4.3567, 4.3302, 4.3698, 4.0916, 3.8118, 4.0763], device='cuda:6'), covar=tensor([0.0743, 0.1125, 0.0811, 0.0930, 0.0895, 0.0845, 0.1727, 0.0746], device='cuda:6'), in_proj_covar=tensor([0.0392, 0.0436, 0.0426, 0.0393, 0.0467, 0.0445, 0.0524, 0.0357], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 13:32:49,533 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=223325.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:33:09,279 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=223340.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:33:26,108 INFO [train.py:904] (6/8) Epoch 23, batch 50, loss[loss=0.1925, simple_loss=0.275, pruned_loss=0.05498, over 16660.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2724, pruned_loss=0.05116, over 741881.54 frames. ], batch size: 134, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:34:32,069 INFO [train.py:904] (6/8) Epoch 23, batch 100, loss[loss=0.1649, simple_loss=0.2615, pruned_loss=0.03414, over 17082.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2669, pruned_loss=0.04709, over 1313246.22 frames. ], batch size: 53, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:34:34,326 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7529, 4.9782, 5.1002, 4.8943, 4.9066, 5.5434, 5.0242, 4.6874], device='cuda:6'), covar=tensor([0.1344, 0.2064, 0.2413, 0.2379, 0.2878, 0.0990, 0.1720, 0.2782], device='cuda:6'), in_proj_covar=tensor([0.0390, 0.0567, 0.0627, 0.0468, 0.0629, 0.0653, 0.0494, 0.0627], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 13:34:42,058 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.242e+02 2.764e+02 3.330e+02 6.883e+02, threshold=5.527e+02, percent-clipped=1.0 2023-05-01 13:34:55,444 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=223420.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:35:22,732 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1205, 5.0841, 4.8250, 4.4624, 4.8956, 1.9492, 4.6885, 4.6774], device='cuda:6'), covar=tensor([0.0089, 0.0089, 0.0230, 0.0337, 0.0104, 0.2707, 0.0138, 0.0250], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0152, 0.0192, 0.0168, 0.0170, 0.0205, 0.0182, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 13:35:33,491 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 13:35:36,014 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-05-01 13:35:38,698 INFO [train.py:904] (6/8) Epoch 23, batch 150, loss[loss=0.1717, simple_loss=0.261, pruned_loss=0.04125, over 16791.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2656, pruned_loss=0.04667, over 1758075.47 frames. ], batch size: 83, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:35:40,447 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9581, 1.8468, 2.4565, 2.8181, 2.7061, 3.2210, 2.1092, 3.2874], device='cuda:6'), covar=tensor([0.0264, 0.0568, 0.0382, 0.0321, 0.0380, 0.0216, 0.0602, 0.0183], device='cuda:6'), in_proj_covar=tensor([0.0186, 0.0189, 0.0176, 0.0180, 0.0193, 0.0150, 0.0193, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 13:36:47,607 INFO [train.py:904] (6/8) Epoch 23, batch 200, loss[loss=0.178, simple_loss=0.2624, pruned_loss=0.04678, over 16789.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2645, pruned_loss=0.04596, over 2095840.36 frames. ], batch size: 102, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:36:57,903 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.401e+02 2.279e+02 2.532e+02 3.033e+02 5.752e+02, threshold=5.065e+02, percent-clipped=1.0 2023-05-01 13:37:21,188 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=223529.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 13:37:52,716 INFO [train.py:904] (6/8) Epoch 23, batch 250, loss[loss=0.1738, simple_loss=0.249, pruned_loss=0.04931, over 16860.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2623, pruned_loss=0.04545, over 2358232.91 frames. ], batch size: 90, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:38:29,140 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=223578.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:38:45,622 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 13:38:54,845 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223596.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:39:04,044 INFO [train.py:904] (6/8) Epoch 23, batch 300, loss[loss=0.1803, simple_loss=0.2556, pruned_loss=0.05246, over 16877.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2603, pruned_loss=0.04442, over 2567482.97 frames. ], batch size: 116, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:39:14,755 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.219e+02 2.672e+02 3.069e+02 6.824e+02, threshold=5.344e+02, percent-clipped=3.0 2023-05-01 13:39:35,537 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223625.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:39:35,751 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7555, 2.4227, 2.5080, 4.5467, 2.4326, 2.8302, 2.5434, 2.7053], device='cuda:6'), covar=tensor([0.1219, 0.3817, 0.3086, 0.0486, 0.4218, 0.2573, 0.3518, 0.3512], device='cuda:6'), in_proj_covar=tensor([0.0398, 0.0447, 0.0368, 0.0324, 0.0436, 0.0513, 0.0419, 0.0521], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 13:39:54,173 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=223639.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:39:55,278 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223640.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:40:01,751 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223644.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:40:13,646 INFO [train.py:904] (6/8) Epoch 23, batch 350, loss[loss=0.1847, simple_loss=0.2768, pruned_loss=0.04628, over 17138.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.257, pruned_loss=0.04327, over 2737050.36 frames. ], batch size: 53, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:40:38,143 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6515, 4.6825, 5.0239, 5.0211, 5.0407, 4.7281, 4.7056, 4.5713], device='cuda:6'), covar=tensor([0.0385, 0.0733, 0.0400, 0.0396, 0.0506, 0.0502, 0.0910, 0.0558], device='cuda:6'), in_proj_covar=tensor([0.0405, 0.0451, 0.0439, 0.0405, 0.0482, 0.0460, 0.0540, 0.0368], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 13:40:42,278 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223673.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:41:03,084 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223688.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:41:15,679 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3700, 3.4034, 2.1520, 3.5983, 2.7238, 3.5841, 2.2083, 2.7482], device='cuda:6'), covar=tensor([0.0311, 0.0473, 0.1570, 0.0432, 0.0813, 0.0845, 0.1496, 0.0785], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0177, 0.0193, 0.0163, 0.0177, 0.0215, 0.0203, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 13:41:22,666 INFO [train.py:904] (6/8) Epoch 23, batch 400, loss[loss=0.1634, simple_loss=0.2544, pruned_loss=0.0362, over 17199.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2548, pruned_loss=0.04245, over 2867470.95 frames. ], batch size: 46, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:41:24,801 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-05-01 13:41:34,906 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.389e+02 2.103e+02 2.400e+02 2.968e+02 1.328e+03, threshold=4.800e+02, percent-clipped=3.0 2023-05-01 13:41:40,060 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4294, 3.5439, 3.6720, 3.6563, 3.6790, 3.5090, 3.5255, 3.5569], device='cuda:6'), covar=tensor([0.0434, 0.0794, 0.0504, 0.0495, 0.0545, 0.0574, 0.0825, 0.0508], device='cuda:6'), in_proj_covar=tensor([0.0410, 0.0456, 0.0443, 0.0409, 0.0486, 0.0465, 0.0546, 0.0372], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 13:41:47,228 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223720.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:42:32,054 INFO [train.py:904] (6/8) Epoch 23, batch 450, loss[loss=0.139, simple_loss=0.2305, pruned_loss=0.02372, over 16863.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2538, pruned_loss=0.04201, over 2964224.33 frames. ], batch size: 42, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:42:52,158 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223768.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:43:30,218 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-01 13:43:40,978 INFO [train.py:904] (6/8) Epoch 23, batch 500, loss[loss=0.156, simple_loss=0.2426, pruned_loss=0.03472, over 17030.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.253, pruned_loss=0.04123, over 3042002.62 frames. ], batch size: 41, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:43:50,475 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.13 vs. limit=5.0 2023-05-01 13:43:52,984 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.054e+02 2.358e+02 2.805e+02 6.007e+02, threshold=4.715e+02, percent-clipped=4.0 2023-05-01 13:44:16,877 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223829.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 13:44:49,713 INFO [train.py:904] (6/8) Epoch 23, batch 550, loss[loss=0.2117, simple_loss=0.2938, pruned_loss=0.06473, over 15429.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2521, pruned_loss=0.04069, over 3109992.71 frames. ], batch size: 190, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:45:13,134 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.51 vs. limit=5.0 2023-05-01 13:45:17,153 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4638, 5.4353, 5.2948, 4.7418, 4.9419, 5.3707, 5.2497, 4.9635], device='cuda:6'), covar=tensor([0.0543, 0.0459, 0.0318, 0.0337, 0.1079, 0.0441, 0.0260, 0.0755], device='cuda:6'), in_proj_covar=tensor([0.0294, 0.0431, 0.0344, 0.0340, 0.0350, 0.0397, 0.0235, 0.0410], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 13:45:24,265 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223877.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 13:45:50,309 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9878, 2.5934, 2.6670, 4.2334, 3.3984, 4.1586, 1.6974, 2.9376], device='cuda:6'), covar=tensor([0.1414, 0.0848, 0.1172, 0.0204, 0.0178, 0.0416, 0.1708, 0.0870], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0174, 0.0193, 0.0188, 0.0201, 0.0215, 0.0203, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 13:45:59,109 INFO [train.py:904] (6/8) Epoch 23, batch 600, loss[loss=0.1294, simple_loss=0.2212, pruned_loss=0.0188, over 16849.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2515, pruned_loss=0.04077, over 3152992.89 frames. ], batch size: 42, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:46:10,991 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.619e+02 2.055e+02 2.460e+02 3.168e+02 5.428e+02, threshold=4.920e+02, percent-clipped=4.0 2023-05-01 13:46:20,884 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7874, 3.1032, 3.2625, 2.2136, 2.8347, 2.3176, 3.2983, 3.3482], device='cuda:6'), covar=tensor([0.0372, 0.1047, 0.0591, 0.1931, 0.0936, 0.1017, 0.0728, 0.1090], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0163, 0.0166, 0.0153, 0.0145, 0.0129, 0.0142, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 13:46:42,828 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=223934.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:47:08,240 INFO [train.py:904] (6/8) Epoch 23, batch 650, loss[loss=0.1828, simple_loss=0.2556, pruned_loss=0.05501, over 16940.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2512, pruned_loss=0.0406, over 3190324.44 frames. ], batch size: 109, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:48:22,024 INFO [train.py:904] (6/8) Epoch 23, batch 700, loss[loss=0.1663, simple_loss=0.2537, pruned_loss=0.03948, over 17287.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2504, pruned_loss=0.04039, over 3223013.73 frames. ], batch size: 52, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:48:35,479 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.336e+02 2.741e+02 3.391e+02 8.377e+02, threshold=5.482e+02, percent-clipped=4.0 2023-05-01 13:48:38,558 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-01 13:48:41,646 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.62 vs. limit=2.0 2023-05-01 13:49:24,299 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3834, 4.6905, 4.5243, 4.4906, 4.2239, 4.2149, 4.2460, 4.7331], device='cuda:6'), covar=tensor([0.1204, 0.0953, 0.1084, 0.0945, 0.0903, 0.1574, 0.1145, 0.1056], device='cuda:6'), in_proj_covar=tensor([0.0683, 0.0830, 0.0683, 0.0630, 0.0523, 0.0535, 0.0699, 0.0649], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 13:49:33,344 INFO [train.py:904] (6/8) Epoch 23, batch 750, loss[loss=0.135, simple_loss=0.22, pruned_loss=0.02499, over 16828.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2511, pruned_loss=0.04085, over 3238883.18 frames. ], batch size: 39, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:50:41,982 INFO [train.py:904] (6/8) Epoch 23, batch 800, loss[loss=0.1559, simple_loss=0.229, pruned_loss=0.04143, over 16711.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2506, pruned_loss=0.04058, over 3256777.38 frames. ], batch size: 83, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:50:54,817 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 2.213e+02 2.619e+02 3.114e+02 4.585e+02, threshold=5.238e+02, percent-clipped=0.0 2023-05-01 13:51:51,762 INFO [train.py:904] (6/8) Epoch 23, batch 850, loss[loss=0.1668, simple_loss=0.2564, pruned_loss=0.03854, over 17087.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.25, pruned_loss=0.03979, over 3269067.41 frames. ], batch size: 55, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:52:36,184 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7449, 2.7837, 2.4690, 2.7306, 3.0900, 2.8844, 3.3884, 3.3245], device='cuda:6'), covar=tensor([0.0160, 0.0447, 0.0537, 0.0450, 0.0293, 0.0423, 0.0273, 0.0277], device='cuda:6'), in_proj_covar=tensor([0.0218, 0.0241, 0.0232, 0.0232, 0.0242, 0.0241, 0.0241, 0.0235], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 13:53:00,745 INFO [train.py:904] (6/8) Epoch 23, batch 900, loss[loss=0.1862, simple_loss=0.2558, pruned_loss=0.05835, over 16517.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2491, pruned_loss=0.03962, over 3268466.00 frames. ], batch size: 146, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:53:14,895 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.103e+02 2.438e+02 2.948e+02 5.610e+02, threshold=4.876e+02, percent-clipped=1.0 2023-05-01 13:53:15,321 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=224212.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:53:25,981 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 13:53:46,030 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=224234.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:54:10,198 INFO [train.py:904] (6/8) Epoch 23, batch 950, loss[loss=0.1752, simple_loss=0.2701, pruned_loss=0.04017, over 17184.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.249, pruned_loss=0.03903, over 3284094.78 frames. ], batch size: 46, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:54:38,551 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=224273.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:54:51,190 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=224282.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:55:20,379 INFO [train.py:904] (6/8) Epoch 23, batch 1000, loss[loss=0.1652, simple_loss=0.2555, pruned_loss=0.0375, over 16750.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2478, pruned_loss=0.03936, over 3288048.45 frames. ], batch size: 57, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:55:33,115 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.05 vs. limit=5.0 2023-05-01 13:55:33,529 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.242e+02 2.591e+02 2.992e+02 5.709e+02, threshold=5.183e+02, percent-clipped=2.0 2023-05-01 13:56:17,304 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 13:56:31,385 INFO [train.py:904] (6/8) Epoch 23, batch 1050, loss[loss=0.157, simple_loss=0.2344, pruned_loss=0.03984, over 12752.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2481, pruned_loss=0.03916, over 3291763.35 frames. ], batch size: 247, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:57:42,154 INFO [train.py:904] (6/8) Epoch 23, batch 1100, loss[loss=0.1518, simple_loss=0.2468, pruned_loss=0.02842, over 17164.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2478, pruned_loss=0.03889, over 3295720.55 frames. ], batch size: 46, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:57:54,066 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.340e+02 2.008e+02 2.334e+02 2.683e+02 4.688e+02, threshold=4.667e+02, percent-clipped=0.0 2023-05-01 13:58:51,564 INFO [train.py:904] (6/8) Epoch 23, batch 1150, loss[loss=0.1537, simple_loss=0.2357, pruned_loss=0.03585, over 16714.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2466, pruned_loss=0.0383, over 3298304.50 frames. ], batch size: 76, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:59:32,757 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.30 vs. limit=5.0 2023-05-01 14:00:00,970 INFO [train.py:904] (6/8) Epoch 23, batch 1200, loss[loss=0.1395, simple_loss=0.2167, pruned_loss=0.0311, over 16834.00 frames. ], tot_loss[loss=0.1612, simple_loss=0.2464, pruned_loss=0.03793, over 3304270.05 frames. ], batch size: 102, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:00:14,523 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.279e+02 2.643e+02 3.361e+02 1.197e+03, threshold=5.285e+02, percent-clipped=8.0 2023-05-01 14:01:10,499 INFO [train.py:904] (6/8) Epoch 23, batch 1250, loss[loss=0.1566, simple_loss=0.2461, pruned_loss=0.0336, over 17197.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.2458, pruned_loss=0.03822, over 3310301.79 frames. ], batch size: 46, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:01:12,832 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6824, 3.7058, 2.8891, 2.2813, 2.3667, 2.3286, 3.8316, 3.2544], device='cuda:6'), covar=tensor([0.2767, 0.0614, 0.1698, 0.3011, 0.2833, 0.2171, 0.0478, 0.1573], device='cuda:6'), in_proj_covar=tensor([0.0329, 0.0269, 0.0307, 0.0315, 0.0297, 0.0263, 0.0298, 0.0341], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 14:01:22,910 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=224562.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:01:31,867 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=224568.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:02:20,413 INFO [train.py:904] (6/8) Epoch 23, batch 1300, loss[loss=0.171, simple_loss=0.2677, pruned_loss=0.03715, over 17129.00 frames. ], tot_loss[loss=0.1609, simple_loss=0.2458, pruned_loss=0.03805, over 3313944.31 frames. ], batch size: 48, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:02:28,057 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5577, 3.5427, 2.2040, 3.8251, 2.7468, 3.7689, 2.3075, 2.8336], device='cuda:6'), covar=tensor([0.0271, 0.0460, 0.1685, 0.0343, 0.0821, 0.0881, 0.1546, 0.0790], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0181, 0.0197, 0.0169, 0.0180, 0.0221, 0.0207, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 14:02:33,657 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.195e+02 2.523e+02 3.004e+02 8.579e+02, threshold=5.045e+02, percent-clipped=3.0 2023-05-01 14:02:49,053 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=224623.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:03:29,364 INFO [train.py:904] (6/8) Epoch 23, batch 1350, loss[loss=0.1409, simple_loss=0.2298, pruned_loss=0.026, over 17195.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.2465, pruned_loss=0.03816, over 3311547.93 frames. ], batch size: 46, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:04:40,086 INFO [train.py:904] (6/8) Epoch 23, batch 1400, loss[loss=0.1882, simple_loss=0.2627, pruned_loss=0.05679, over 16888.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2467, pruned_loss=0.03819, over 3319216.47 frames. ], batch size: 109, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:04:52,698 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.431e+02 2.135e+02 2.533e+02 2.912e+02 4.513e+02, threshold=5.066e+02, percent-clipped=0.0 2023-05-01 14:05:13,985 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=224726.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:05:33,810 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=224740.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:05:50,555 INFO [train.py:904] (6/8) Epoch 23, batch 1450, loss[loss=0.1692, simple_loss=0.2471, pruned_loss=0.04568, over 16520.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2471, pruned_loss=0.0383, over 3323539.65 frames. ], batch size: 75, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:06:39,666 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=224787.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:07:00,032 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=224801.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:07:01,746 INFO [train.py:904] (6/8) Epoch 23, batch 1500, loss[loss=0.1552, simple_loss=0.2473, pruned_loss=0.03158, over 17045.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2465, pruned_loss=0.03806, over 3331766.55 frames. ], batch size: 41, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:07:16,530 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.459e+02 2.087e+02 2.500e+02 3.086e+02 8.479e+02, threshold=5.000e+02, percent-clipped=3.0 2023-05-01 14:07:17,144 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9408, 3.1644, 3.2867, 2.2081, 2.7840, 2.3312, 3.4351, 3.4832], device='cuda:6'), covar=tensor([0.0250, 0.0879, 0.0627, 0.1864, 0.0946, 0.0991, 0.0570, 0.0822], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0165, 0.0168, 0.0154, 0.0146, 0.0130, 0.0144, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 14:08:14,248 INFO [train.py:904] (6/8) Epoch 23, batch 1550, loss[loss=0.1752, simple_loss=0.2477, pruned_loss=0.05131, over 16421.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2475, pruned_loss=0.03881, over 3324733.28 frames. ], batch size: 146, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 14:08:35,300 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=224868.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:09:23,916 INFO [train.py:904] (6/8) Epoch 23, batch 1600, loss[loss=0.2084, simple_loss=0.2847, pruned_loss=0.06601, over 12153.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2499, pruned_loss=0.0401, over 3317798.92 frames. ], batch size: 248, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:09:36,820 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.147e+02 2.697e+02 3.461e+02 9.919e+02, threshold=5.394e+02, percent-clipped=4.0 2023-05-01 14:09:41,291 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=224916.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:09:43,756 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=224918.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:10:32,920 INFO [train.py:904] (6/8) Epoch 23, batch 1650, loss[loss=0.1331, simple_loss=0.2169, pruned_loss=0.0246, over 16783.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2514, pruned_loss=0.04063, over 3322741.43 frames. ], batch size: 39, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:11:10,129 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9065, 3.0342, 2.9329, 5.0771, 4.1355, 4.4249, 1.7153, 3.2574], device='cuda:6'), covar=tensor([0.1306, 0.0753, 0.1129, 0.0193, 0.0243, 0.0419, 0.1646, 0.0803], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0176, 0.0195, 0.0192, 0.0204, 0.0217, 0.0204, 0.0194], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 14:11:23,508 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-05-01 14:11:41,679 INFO [train.py:904] (6/8) Epoch 23, batch 1700, loss[loss=0.1536, simple_loss=0.2499, pruned_loss=0.02868, over 17241.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2528, pruned_loss=0.04108, over 3315929.91 frames. ], batch size: 52, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:11:56,160 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.319e+02 2.651e+02 3.431e+02 9.066e+02, threshold=5.301e+02, percent-clipped=1.0 2023-05-01 14:12:13,283 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9138, 2.6060, 2.0893, 2.4637, 2.9534, 2.7695, 3.0048, 3.0606], device='cuda:6'), covar=tensor([0.0207, 0.0404, 0.0530, 0.0424, 0.0240, 0.0295, 0.0241, 0.0251], device='cuda:6'), in_proj_covar=tensor([0.0220, 0.0242, 0.0232, 0.0232, 0.0243, 0.0242, 0.0244, 0.0237], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:12:52,528 INFO [train.py:904] (6/8) Epoch 23, batch 1750, loss[loss=0.1626, simple_loss=0.2589, pruned_loss=0.03312, over 17178.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2537, pruned_loss=0.04081, over 3321184.77 frames. ], batch size: 46, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:13:33,046 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225082.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:13:34,922 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5598, 4.4911, 4.4689, 4.1406, 4.2026, 4.5221, 4.2873, 4.2385], device='cuda:6'), covar=tensor([0.0648, 0.0847, 0.0354, 0.0337, 0.0897, 0.0560, 0.0589, 0.0699], device='cuda:6'), in_proj_covar=tensor([0.0305, 0.0448, 0.0357, 0.0354, 0.0362, 0.0414, 0.0242, 0.0426], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:13:52,620 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225096.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:14:01,698 INFO [train.py:904] (6/8) Epoch 23, batch 1800, loss[loss=0.1762, simple_loss=0.2584, pruned_loss=0.04694, over 16925.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2554, pruned_loss=0.04132, over 3325705.21 frames. ], batch size: 96, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:14:02,764 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225103.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:14:15,807 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.405e+02 2.184e+02 2.489e+02 3.045e+02 6.303e+02, threshold=4.978e+02, percent-clipped=2.0 2023-05-01 14:14:35,773 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7530, 4.5757, 4.8643, 5.0007, 5.2124, 4.6018, 5.1882, 5.2136], device='cuda:6'), covar=tensor([0.2045, 0.1493, 0.1814, 0.0842, 0.0561, 0.1030, 0.0649, 0.0633], device='cuda:6'), in_proj_covar=tensor([0.0674, 0.0832, 0.0964, 0.0842, 0.0636, 0.0668, 0.0692, 0.0803], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:14:57,328 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225142.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:15:12,126 INFO [train.py:904] (6/8) Epoch 23, batch 1850, loss[loss=0.1364, simple_loss=0.2299, pruned_loss=0.0215, over 16842.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2558, pruned_loss=0.04091, over 3322007.91 frames. ], batch size: 42, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:15:27,774 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225164.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:16:22,247 INFO [train.py:904] (6/8) Epoch 23, batch 1900, loss[loss=0.1877, simple_loss=0.2775, pruned_loss=0.04896, over 17055.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2548, pruned_loss=0.0404, over 3320885.23 frames. ], batch size: 55, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:16:22,787 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225203.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:16:36,681 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 2.136e+02 2.472e+02 2.864e+02 6.129e+02, threshold=4.945e+02, percent-clipped=2.0 2023-05-01 14:16:43,985 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=225218.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:17:11,972 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-01 14:17:31,819 INFO [train.py:904] (6/8) Epoch 23, batch 1950, loss[loss=0.1751, simple_loss=0.2652, pruned_loss=0.04248, over 12381.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2545, pruned_loss=0.03981, over 3314690.59 frames. ], batch size: 247, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:17:37,083 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 14:17:44,497 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7990, 2.7461, 2.5484, 4.8998, 3.9416, 4.3641, 1.6022, 3.1584], device='cuda:6'), covar=tensor([0.1395, 0.0830, 0.1331, 0.0207, 0.0212, 0.0389, 0.1682, 0.0771], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0176, 0.0197, 0.0193, 0.0205, 0.0218, 0.0205, 0.0194], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 14:17:49,237 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4976, 2.2595, 2.2796, 4.2680, 2.2098, 2.6336, 2.3697, 2.3991], device='cuda:6'), covar=tensor([0.1241, 0.3998, 0.3332, 0.0519, 0.4481, 0.2797, 0.3816, 0.4048], device='cuda:6'), in_proj_covar=tensor([0.0407, 0.0456, 0.0374, 0.0332, 0.0441, 0.0526, 0.0428, 0.0532], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:17:50,058 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=225266.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:18:24,631 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8260, 4.6476, 4.8936, 5.0571, 5.2980, 4.6238, 5.2672, 5.2841], device='cuda:6'), covar=tensor([0.2304, 0.1484, 0.2063, 0.0941, 0.0626, 0.1090, 0.0622, 0.0633], device='cuda:6'), in_proj_covar=tensor([0.0677, 0.0838, 0.0971, 0.0844, 0.0639, 0.0672, 0.0694, 0.0805], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:18:37,349 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7045, 2.5388, 2.2284, 2.4884, 2.9143, 2.7175, 3.2239, 3.1305], device='cuda:6'), covar=tensor([0.0159, 0.0504, 0.0595, 0.0501, 0.0340, 0.0432, 0.0336, 0.0319], device='cuda:6'), in_proj_covar=tensor([0.0221, 0.0242, 0.0232, 0.0232, 0.0243, 0.0242, 0.0243, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:18:42,383 INFO [train.py:904] (6/8) Epoch 23, batch 2000, loss[loss=0.1646, simple_loss=0.2522, pruned_loss=0.03851, over 16569.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.254, pruned_loss=0.03927, over 3319719.16 frames. ], batch size: 68, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:18:56,051 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.089e+02 2.408e+02 2.851e+02 8.100e+02, threshold=4.816e+02, percent-clipped=1.0 2023-05-01 14:19:52,847 INFO [train.py:904] (6/8) Epoch 23, batch 2050, loss[loss=0.1937, simple_loss=0.2748, pruned_loss=0.05632, over 16859.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2547, pruned_loss=0.04015, over 3315645.39 frames. ], batch size: 116, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:20:02,878 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-05-01 14:20:35,691 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=225382.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:20:54,923 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=225396.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:21:04,272 INFO [train.py:904] (6/8) Epoch 23, batch 2100, loss[loss=0.1709, simple_loss=0.2601, pruned_loss=0.04084, over 16771.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2549, pruned_loss=0.04041, over 3322084.27 frames. ], batch size: 83, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:21:09,867 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8736, 1.8888, 2.4963, 2.7916, 2.7723, 3.2177, 2.2074, 3.2518], device='cuda:6'), covar=tensor([0.0279, 0.0642, 0.0374, 0.0373, 0.0356, 0.0230, 0.0557, 0.0235], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0195, 0.0183, 0.0188, 0.0200, 0.0158, 0.0199, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:21:18,934 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.459e+02 2.181e+02 2.818e+02 3.430e+02 7.396e+02, threshold=5.635e+02, percent-clipped=7.0 2023-05-01 14:21:29,406 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-05-01 14:21:44,080 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=225430.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:21:57,143 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225439.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:22:03,548 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=225444.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:22:15,762 INFO [train.py:904] (6/8) Epoch 23, batch 2150, loss[loss=0.174, simple_loss=0.2635, pruned_loss=0.04228, over 16494.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2544, pruned_loss=0.04015, over 3323271.38 frames. ], batch size: 75, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:22:25,346 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225459.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:22:25,412 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2285, 5.2126, 5.1103, 4.5667, 4.7105, 5.1160, 5.0790, 4.7712], device='cuda:6'), covar=tensor([0.0605, 0.0537, 0.0345, 0.0394, 0.1189, 0.0521, 0.0304, 0.0804], device='cuda:6'), in_proj_covar=tensor([0.0311, 0.0456, 0.0363, 0.0361, 0.0367, 0.0421, 0.0246, 0.0433], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 14:22:29,367 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225462.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:22:57,580 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1590, 3.4347, 3.5042, 2.3036, 2.9978, 2.4481, 3.7105, 3.7074], device='cuda:6'), covar=tensor([0.0271, 0.0826, 0.0637, 0.1924, 0.0873, 0.0997, 0.0519, 0.0844], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0166, 0.0168, 0.0154, 0.0146, 0.0130, 0.0144, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 14:23:19,352 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225498.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:23:22,102 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225500.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:23:26,035 INFO [train.py:904] (6/8) Epoch 23, batch 2200, loss[loss=0.1542, simple_loss=0.2534, pruned_loss=0.02746, over 17130.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2552, pruned_loss=0.04019, over 3326228.75 frames. ], batch size: 47, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:23:29,406 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225505.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:23:40,045 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.652e+02 2.156e+02 2.513e+02 2.761e+02 5.563e+02, threshold=5.025e+02, percent-clipped=0.0 2023-05-01 14:23:53,558 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225522.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:23:55,320 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225523.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:24:36,368 INFO [train.py:904] (6/8) Epoch 23, batch 2250, loss[loss=0.151, simple_loss=0.2459, pruned_loss=0.02806, over 17215.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2553, pruned_loss=0.04068, over 3321128.05 frames. ], batch size: 46, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:24:40,102 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7473, 1.8687, 2.3015, 2.5014, 2.6530, 2.6601, 1.9257, 2.8479], device='cuda:6'), covar=tensor([0.0200, 0.0526, 0.0351, 0.0346, 0.0320, 0.0304, 0.0547, 0.0161], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0196, 0.0183, 0.0188, 0.0200, 0.0157, 0.0199, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:24:45,158 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225559.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 14:24:54,957 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225566.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:25:18,799 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225583.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:25:46,973 INFO [train.py:904] (6/8) Epoch 23, batch 2300, loss[loss=0.1613, simple_loss=0.2465, pruned_loss=0.03804, over 16667.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2558, pruned_loss=0.04081, over 3317845.26 frames. ], batch size: 89, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:25:56,969 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 14:26:01,554 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.669e+02 2.061e+02 2.466e+02 3.026e+02 4.196e+02, threshold=4.932e+02, percent-clipped=0.0 2023-05-01 14:26:11,648 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225620.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 14:26:58,883 INFO [train.py:904] (6/8) Epoch 23, batch 2350, loss[loss=0.1954, simple_loss=0.2696, pruned_loss=0.06058, over 16785.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2564, pruned_loss=0.04111, over 3316172.85 frames. ], batch size: 124, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:27:49,308 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.75 vs. limit=5.0 2023-05-01 14:28:10,319 INFO [train.py:904] (6/8) Epoch 23, batch 2400, loss[loss=0.1728, simple_loss=0.2604, pruned_loss=0.04255, over 16704.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2577, pruned_loss=0.04121, over 3319423.76 frames. ], batch size: 89, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:28:23,215 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.701e+02 2.227e+02 2.621e+02 3.138e+02 6.803e+02, threshold=5.242e+02, percent-clipped=3.0 2023-05-01 14:29:17,848 INFO [train.py:904] (6/8) Epoch 23, batch 2450, loss[loss=0.1673, simple_loss=0.2626, pruned_loss=0.036, over 17113.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2588, pruned_loss=0.04145, over 3312445.86 frames. ], batch size: 49, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:29:26,308 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=225759.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:30:02,753 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3723, 2.3390, 2.4162, 4.0903, 2.3109, 2.6438, 2.3919, 2.5320], device='cuda:6'), covar=tensor([0.1453, 0.3994, 0.3073, 0.0656, 0.4397, 0.2932, 0.3873, 0.3445], device='cuda:6'), in_proj_covar=tensor([0.0409, 0.0459, 0.0376, 0.0335, 0.0443, 0.0529, 0.0430, 0.0537], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:30:17,163 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225795.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:30:22,195 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=225798.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:30:29,832 INFO [train.py:904] (6/8) Epoch 23, batch 2500, loss[loss=0.1706, simple_loss=0.266, pruned_loss=0.03762, over 16727.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2585, pruned_loss=0.04099, over 3320209.75 frames. ], batch size: 62, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:30:36,123 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=225807.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:30:43,415 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.105e+02 2.544e+02 3.075e+02 7.140e+02, threshold=5.088e+02, percent-clipped=2.0 2023-05-01 14:30:51,421 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225818.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:31:29,064 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=225846.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:31:38,345 INFO [train.py:904] (6/8) Epoch 23, batch 2550, loss[loss=0.1563, simple_loss=0.2597, pruned_loss=0.02642, over 17252.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2578, pruned_loss=0.04115, over 3318160.76 frames. ], batch size: 52, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:31:42,629 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5234, 5.9083, 5.6770, 5.7276, 5.2768, 5.3032, 5.3086, 6.0439], device='cuda:6'), covar=tensor([0.1506, 0.0956, 0.1045, 0.0860, 0.1027, 0.0789, 0.1263, 0.1034], device='cuda:6'), in_proj_covar=tensor([0.0702, 0.0855, 0.0702, 0.0649, 0.0540, 0.0548, 0.0717, 0.0668], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:31:49,566 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225861.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:31:50,892 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225862.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:32:08,454 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-05-01 14:32:14,110 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225878.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:32:16,577 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0827, 2.2155, 2.7148, 2.9260, 2.8847, 3.5479, 2.5089, 3.4437], device='cuda:6'), covar=tensor([0.0260, 0.0494, 0.0326, 0.0350, 0.0340, 0.0181, 0.0449, 0.0166], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0195, 0.0182, 0.0188, 0.0200, 0.0158, 0.0198, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:32:38,049 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225895.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:32:49,281 INFO [train.py:904] (6/8) Epoch 23, batch 2600, loss[loss=0.1569, simple_loss=0.2403, pruned_loss=0.0367, over 16679.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2576, pruned_loss=0.04083, over 3325629.66 frames. ], batch size: 89, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:33:03,046 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.463e+02 2.265e+02 2.588e+02 2.921e+02 5.309e+02, threshold=5.176e+02, percent-clipped=1.0 2023-05-01 14:33:06,087 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225915.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 14:33:16,591 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225923.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:33:56,756 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3326, 4.1879, 4.4084, 4.5375, 4.6562, 4.2047, 4.4955, 4.6516], device='cuda:6'), covar=tensor([0.1877, 0.1344, 0.1452, 0.0750, 0.0668, 0.1308, 0.2098, 0.0759], device='cuda:6'), in_proj_covar=tensor([0.0673, 0.0835, 0.0965, 0.0843, 0.0640, 0.0668, 0.0693, 0.0803], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:33:58,791 INFO [train.py:904] (6/8) Epoch 23, batch 2650, loss[loss=0.1844, simple_loss=0.2788, pruned_loss=0.04496, over 16461.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2584, pruned_loss=0.04089, over 3328953.22 frames. ], batch size: 75, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:34:04,252 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225956.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:34:22,250 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4934, 3.2729, 3.6214, 1.9666, 3.6791, 3.6716, 3.0555, 2.6977], device='cuda:6'), covar=tensor([0.0764, 0.0245, 0.0163, 0.1152, 0.0108, 0.0221, 0.0387, 0.0492], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0110, 0.0100, 0.0141, 0.0082, 0.0128, 0.0129, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 14:34:43,658 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7130, 3.8292, 2.4800, 4.4487, 2.9774, 4.4442, 2.5772, 3.1355], device='cuda:6'), covar=tensor([0.0341, 0.0434, 0.1622, 0.0430, 0.0902, 0.0511, 0.1556, 0.0822], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0183, 0.0198, 0.0172, 0.0181, 0.0224, 0.0208, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 14:35:12,278 INFO [train.py:904] (6/8) Epoch 23, batch 2700, loss[loss=0.1761, simple_loss=0.2604, pruned_loss=0.04595, over 16676.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.258, pruned_loss=0.04059, over 3333876.16 frames. ], batch size: 134, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:35:25,728 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.491e+02 2.127e+02 2.428e+02 2.889e+02 7.591e+02, threshold=4.856e+02, percent-clipped=2.0 2023-05-01 14:36:06,269 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=226041.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:36:23,069 INFO [train.py:904] (6/8) Epoch 23, batch 2750, loss[loss=0.1708, simple_loss=0.2684, pruned_loss=0.03665, over 17060.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2585, pruned_loss=0.0402, over 3333995.43 frames. ], batch size: 55, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:36:38,024 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5574, 3.4901, 3.4751, 2.7437, 3.3216, 2.0471, 3.1475, 2.7836], device='cuda:6'), covar=tensor([0.0146, 0.0144, 0.0195, 0.0240, 0.0118, 0.2564, 0.0138, 0.0276], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0165, 0.0207, 0.0184, 0.0185, 0.0216, 0.0197, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:36:49,890 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3368, 2.3782, 2.4426, 4.1291, 2.1767, 2.7151, 2.4291, 2.4733], device='cuda:6'), covar=tensor([0.1364, 0.3648, 0.3076, 0.0581, 0.4448, 0.2667, 0.3500, 0.3669], device='cuda:6'), in_proj_covar=tensor([0.0409, 0.0458, 0.0375, 0.0335, 0.0443, 0.0528, 0.0429, 0.0537], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:36:53,270 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.7502, 6.0608, 5.7720, 5.9169, 5.5809, 5.4103, 5.4449, 6.2144], device='cuda:6'), covar=tensor([0.1219, 0.0874, 0.1073, 0.0776, 0.0799, 0.0700, 0.1192, 0.0867], device='cuda:6'), in_proj_covar=tensor([0.0706, 0.0860, 0.0707, 0.0653, 0.0544, 0.0549, 0.0722, 0.0672], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:37:13,212 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 14:37:21,881 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226095.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:37:32,004 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=226102.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:37:32,648 INFO [train.py:904] (6/8) Epoch 23, batch 2800, loss[loss=0.1869, simple_loss=0.2692, pruned_loss=0.0523, over 16746.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.259, pruned_loss=0.04023, over 3331078.91 frames. ], batch size: 124, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:37:47,345 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.139e+02 2.068e+02 2.362e+02 2.950e+02 8.233e+02, threshold=4.725e+02, percent-clipped=4.0 2023-05-01 14:37:54,705 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226118.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:38:29,330 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226143.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:38:42,389 INFO [train.py:904] (6/8) Epoch 23, batch 2850, loss[loss=0.1546, simple_loss=0.2336, pruned_loss=0.03781, over 16825.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2572, pruned_loss=0.03988, over 3324671.45 frames. ], batch size: 102, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:38:54,320 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226161.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:39:00,760 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226166.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:39:18,191 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226178.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:39:22,550 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 14:39:51,834 INFO [train.py:904] (6/8) Epoch 23, batch 2900, loss[loss=0.1341, simple_loss=0.2256, pruned_loss=0.02134, over 17013.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2566, pruned_loss=0.04055, over 3323376.72 frames. ], batch size: 41, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:39:59,226 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9870, 4.8621, 4.8691, 4.4625, 4.5762, 4.9069, 4.7493, 4.5969], device='cuda:6'), covar=tensor([0.0591, 0.0787, 0.0326, 0.0356, 0.0941, 0.0509, 0.0413, 0.0773], device='cuda:6'), in_proj_covar=tensor([0.0310, 0.0458, 0.0364, 0.0363, 0.0370, 0.0422, 0.0246, 0.0436], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 14:40:00,214 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226209.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:40:05,815 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.036e+02 2.506e+02 2.935e+02 4.421e+02, threshold=5.013e+02, percent-clipped=0.0 2023-05-01 14:40:09,385 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226215.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 14:40:13,512 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=226218.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:40:13,877 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 14:40:24,106 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226226.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:40:34,203 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=226233.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:40:48,856 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8452, 4.2617, 4.1295, 2.9617, 3.6012, 4.2168, 3.8461, 2.1547], device='cuda:6'), covar=tensor([0.0572, 0.0101, 0.0088, 0.0484, 0.0190, 0.0160, 0.0163, 0.0680], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0085, 0.0087, 0.0135, 0.0100, 0.0112, 0.0098, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 14:40:58,438 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=226251.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:41:00,395 INFO [train.py:904] (6/8) Epoch 23, batch 2950, loss[loss=0.1744, simple_loss=0.275, pruned_loss=0.03692, over 17069.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2562, pruned_loss=0.04077, over 3325131.47 frames. ], batch size: 53, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:41:14,674 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226263.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 14:41:43,494 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6696, 3.7989, 2.7954, 2.2317, 2.5409, 2.3239, 3.8805, 3.3294], device='cuda:6'), covar=tensor([0.2835, 0.0630, 0.1788, 0.2983, 0.2724, 0.2205, 0.0567, 0.1393], device='cuda:6'), in_proj_covar=tensor([0.0329, 0.0271, 0.0308, 0.0317, 0.0301, 0.0265, 0.0300, 0.0346], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 14:41:56,380 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=226294.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:42:08,786 INFO [train.py:904] (6/8) Epoch 23, batch 3000, loss[loss=0.1781, simple_loss=0.2642, pruned_loss=0.04598, over 16537.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2568, pruned_loss=0.0414, over 3327365.93 frames. ], batch size: 75, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:42:08,786 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 14:42:17,866 INFO [train.py:938] (6/8) Epoch 23, validation: loss=0.1344, simple_loss=0.2397, pruned_loss=0.01456, over 944034.00 frames. 2023-05-01 14:42:17,867 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 14:42:30,994 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.698e+02 2.313e+02 2.743e+02 3.282e+02 6.136e+02, threshold=5.486e+02, percent-clipped=4.0 2023-05-01 14:43:26,883 INFO [train.py:904] (6/8) Epoch 23, batch 3050, loss[loss=0.1682, simple_loss=0.2544, pruned_loss=0.04096, over 16393.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2567, pruned_loss=0.04184, over 3319084.82 frames. ], batch size: 68, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:43:37,938 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7070, 3.8544, 2.2096, 4.4350, 2.9876, 4.3779, 1.9879, 2.9270], device='cuda:6'), covar=tensor([0.0333, 0.0404, 0.1888, 0.0326, 0.0850, 0.0466, 0.2228, 0.0916], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0182, 0.0196, 0.0171, 0.0180, 0.0224, 0.0207, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 14:43:57,211 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4368, 4.1587, 4.6050, 2.3850, 4.7549, 4.8258, 3.6089, 3.8051], device='cuda:6'), covar=tensor([0.0589, 0.0199, 0.0189, 0.1132, 0.0077, 0.0149, 0.0361, 0.0371], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0109, 0.0100, 0.0140, 0.0082, 0.0128, 0.0129, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 14:44:29,373 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=226397.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:44:37,979 INFO [train.py:904] (6/8) Epoch 23, batch 3100, loss[loss=0.1311, simple_loss=0.2142, pruned_loss=0.024, over 16833.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2559, pruned_loss=0.04188, over 3320509.03 frames. ], batch size: 102, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:44:50,759 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2489, 5.8276, 5.9697, 5.6040, 5.7887, 6.2437, 5.8153, 5.5551], device='cuda:6'), covar=tensor([0.0870, 0.1795, 0.2201, 0.1921, 0.2352, 0.0922, 0.1425, 0.2083], device='cuda:6'), in_proj_covar=tensor([0.0422, 0.0616, 0.0681, 0.0506, 0.0681, 0.0703, 0.0531, 0.0678], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 14:44:51,600 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.140e+02 2.727e+02 3.303e+02 6.006e+02, threshold=5.454e+02, percent-clipped=2.0 2023-05-01 14:45:22,168 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6745, 3.6583, 2.8724, 2.2860, 2.3303, 2.3434, 3.8319, 3.2196], device='cuda:6'), covar=tensor([0.2709, 0.0655, 0.1725, 0.2848, 0.2795, 0.2143, 0.0483, 0.1582], device='cuda:6'), in_proj_covar=tensor([0.0328, 0.0271, 0.0308, 0.0317, 0.0301, 0.0264, 0.0300, 0.0346], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 14:45:37,641 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7140, 3.8984, 2.5173, 4.4608, 3.0756, 4.4747, 2.6781, 3.1006], device='cuda:6'), covar=tensor([0.0337, 0.0432, 0.1598, 0.0415, 0.0880, 0.0503, 0.1456, 0.0848], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0181, 0.0196, 0.0171, 0.0180, 0.0223, 0.0207, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 14:45:47,148 INFO [train.py:904] (6/8) Epoch 23, batch 3150, loss[loss=0.1331, simple_loss=0.2251, pruned_loss=0.02054, over 16729.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2547, pruned_loss=0.04148, over 3322463.01 frames. ], batch size: 39, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:45:56,950 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6105, 2.3660, 1.9129, 2.2122, 2.6819, 2.4682, 2.6296, 2.7864], device='cuda:6'), covar=tensor([0.0212, 0.0426, 0.0580, 0.0439, 0.0251, 0.0350, 0.0219, 0.0295], device='cuda:6'), in_proj_covar=tensor([0.0225, 0.0245, 0.0233, 0.0235, 0.0245, 0.0243, 0.0246, 0.0241], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:46:20,583 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 14:46:54,851 INFO [train.py:904] (6/8) Epoch 23, batch 3200, loss[loss=0.151, simple_loss=0.2409, pruned_loss=0.03058, over 16819.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2545, pruned_loss=0.04109, over 3311266.31 frames. ], batch size: 39, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:47:09,850 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.604e+02 2.052e+02 2.422e+02 2.820e+02 4.460e+02, threshold=4.844e+02, percent-clipped=0.0 2023-05-01 14:47:15,829 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226518.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:47:54,089 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3836, 5.7482, 5.4994, 5.5519, 5.1278, 5.1309, 5.1235, 5.8934], device='cuda:6'), covar=tensor([0.1345, 0.1030, 0.1153, 0.0864, 0.0955, 0.0805, 0.1280, 0.0889], device='cuda:6'), in_proj_covar=tensor([0.0704, 0.0857, 0.0706, 0.0650, 0.0542, 0.0548, 0.0720, 0.0669], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:47:58,126 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3757, 3.7218, 4.0946, 2.2878, 3.2298, 2.4994, 3.9268, 3.9163], device='cuda:6'), covar=tensor([0.0307, 0.0895, 0.0464, 0.2002, 0.0801, 0.0986, 0.0578, 0.0966], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0169, 0.0170, 0.0156, 0.0147, 0.0132, 0.0146, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-01 14:48:02,022 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226551.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:48:02,161 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7197, 2.6160, 1.8998, 2.7893, 2.1568, 2.8848, 2.1785, 2.4320], device='cuda:6'), covar=tensor([0.0324, 0.0394, 0.1358, 0.0273, 0.0734, 0.0496, 0.1190, 0.0642], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0181, 0.0195, 0.0170, 0.0179, 0.0223, 0.0206, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 14:48:04,025 INFO [train.py:904] (6/8) Epoch 23, batch 3250, loss[loss=0.1391, simple_loss=0.2245, pruned_loss=0.02691, over 16993.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2553, pruned_loss=0.04089, over 3314783.56 frames. ], batch size: 41, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:48:22,282 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226566.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:48:54,452 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=226589.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:49:08,848 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226599.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:49:14,394 INFO [train.py:904] (6/8) Epoch 23, batch 3300, loss[loss=0.1841, simple_loss=0.2573, pruned_loss=0.05548, over 16856.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2562, pruned_loss=0.04139, over 3312322.85 frames. ], batch size: 109, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:49:16,477 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-05-01 14:49:27,924 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.422e+02 2.105e+02 2.619e+02 3.003e+02 4.974e+02, threshold=5.238e+02, percent-clipped=1.0 2023-05-01 14:50:23,120 INFO [train.py:904] (6/8) Epoch 23, batch 3350, loss[loss=0.1626, simple_loss=0.2604, pruned_loss=0.03242, over 17134.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2566, pruned_loss=0.04107, over 3320879.91 frames. ], batch size: 49, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:51:25,658 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226697.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:51:33,990 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0023, 2.2624, 2.3322, 2.7145, 2.1407, 3.2176, 1.7781, 2.7091], device='cuda:6'), covar=tensor([0.1081, 0.0698, 0.1033, 0.0171, 0.0120, 0.0341, 0.1394, 0.0721], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0175, 0.0196, 0.0195, 0.0206, 0.0218, 0.0204, 0.0194], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 14:51:34,603 INFO [train.py:904] (6/8) Epoch 23, batch 3400, loss[loss=0.182, simple_loss=0.2691, pruned_loss=0.0475, over 12469.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2563, pruned_loss=0.04076, over 3315517.51 frames. ], batch size: 247, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:51:47,765 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 2.087e+02 2.446e+02 3.167e+02 4.781e+02, threshold=4.893e+02, percent-clipped=0.0 2023-05-01 14:52:33,287 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226745.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:52:43,440 INFO [train.py:904] (6/8) Epoch 23, batch 3450, loss[loss=0.1428, simple_loss=0.2289, pruned_loss=0.02835, over 16854.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2551, pruned_loss=0.04039, over 3315585.84 frames. ], batch size: 39, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:53:47,344 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2340, 5.8145, 5.9619, 5.6069, 5.8887, 6.2957, 5.7944, 5.5127], device='cuda:6'), covar=tensor([0.0953, 0.2055, 0.2402, 0.2199, 0.2265, 0.0983, 0.1739, 0.2624], device='cuda:6'), in_proj_covar=tensor([0.0427, 0.0620, 0.0686, 0.0511, 0.0685, 0.0709, 0.0538, 0.0685], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 14:53:52,946 INFO [train.py:904] (6/8) Epoch 23, batch 3500, loss[loss=0.185, simple_loss=0.2672, pruned_loss=0.05134, over 16197.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2536, pruned_loss=0.03968, over 3314829.36 frames. ], batch size: 165, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:54:07,220 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.312e+02 2.101e+02 2.445e+02 3.055e+02 4.723e+02, threshold=4.890e+02, percent-clipped=0.0 2023-05-01 14:55:03,885 INFO [train.py:904] (6/8) Epoch 23, batch 3550, loss[loss=0.1586, simple_loss=0.2478, pruned_loss=0.03469, over 17199.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2526, pruned_loss=0.03962, over 3312906.85 frames. ], batch size: 44, lr: 2.94e-03, grad_scale: 16.0 2023-05-01 14:55:26,861 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7606, 3.8731, 4.0823, 2.9667, 3.6488, 4.1472, 3.7446, 2.4202], device='cuda:6'), covar=tensor([0.0457, 0.0203, 0.0056, 0.0352, 0.0114, 0.0098, 0.0096, 0.0480], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0085, 0.0087, 0.0135, 0.0100, 0.0112, 0.0098, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 14:55:52,948 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226889.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:56:06,170 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.76 vs. limit=5.0 2023-05-01 14:56:12,760 INFO [train.py:904] (6/8) Epoch 23, batch 3600, loss[loss=0.1607, simple_loss=0.2392, pruned_loss=0.04108, over 16387.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2514, pruned_loss=0.03924, over 3320310.76 frames. ], batch size: 146, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:56:28,214 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.070e+02 2.545e+02 2.955e+02 4.911e+02, threshold=5.089e+02, percent-clipped=1.0 2023-05-01 14:56:29,950 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0006, 5.3625, 5.0995, 5.0865, 4.8623, 4.8072, 4.8149, 5.4422], device='cuda:6'), covar=tensor([0.1387, 0.1052, 0.1217, 0.1417, 0.0996, 0.1196, 0.1340, 0.1147], device='cuda:6'), in_proj_covar=tensor([0.0703, 0.0857, 0.0704, 0.0650, 0.0540, 0.0547, 0.0719, 0.0670], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 14:57:03,052 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226937.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:57:24,672 INFO [train.py:904] (6/8) Epoch 23, batch 3650, loss[loss=0.1567, simple_loss=0.2374, pruned_loss=0.038, over 16551.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2503, pruned_loss=0.03986, over 3311490.19 frames. ], batch size: 146, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:58:19,635 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=226989.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:58:40,559 INFO [train.py:904] (6/8) Epoch 23, batch 3700, loss[loss=0.1731, simple_loss=0.245, pruned_loss=0.05056, over 16847.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2492, pruned_loss=0.0411, over 3297527.84 frames. ], batch size: 109, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:58:56,577 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 2.184e+02 2.522e+02 2.938e+02 6.329e+02, threshold=5.043e+02, percent-clipped=1.0 2023-05-01 14:59:04,673 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5073, 3.3603, 2.7684, 2.2252, 2.2710, 2.3489, 3.4857, 2.9842], device='cuda:6'), covar=tensor([0.2880, 0.0668, 0.1727, 0.2696, 0.2845, 0.2147, 0.0501, 0.1595], device='cuda:6'), in_proj_covar=tensor([0.0331, 0.0273, 0.0310, 0.0319, 0.0303, 0.0266, 0.0302, 0.0346], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 14:59:50,771 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=227050.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:59:53,687 INFO [train.py:904] (6/8) Epoch 23, batch 3750, loss[loss=0.2077, simple_loss=0.2836, pruned_loss=0.06591, over 11690.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2507, pruned_loss=0.04273, over 3279801.03 frames. ], batch size: 246, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:00:04,972 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6589, 3.6947, 2.9243, 2.2513, 2.4100, 2.4264, 3.7943, 3.2236], device='cuda:6'), covar=tensor([0.2803, 0.0569, 0.1651, 0.3251, 0.2901, 0.2086, 0.0521, 0.1524], device='cuda:6'), in_proj_covar=tensor([0.0330, 0.0272, 0.0309, 0.0319, 0.0302, 0.0266, 0.0301, 0.0345], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 15:01:07,855 INFO [train.py:904] (6/8) Epoch 23, batch 3800, loss[loss=0.168, simple_loss=0.2526, pruned_loss=0.04171, over 16689.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.252, pruned_loss=0.04387, over 3273941.92 frames. ], batch size: 134, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:01:25,256 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.697e+02 2.146e+02 2.630e+02 3.388e+02 6.503e+02, threshold=5.260e+02, percent-clipped=3.0 2023-05-01 15:02:08,503 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3137, 3.4476, 3.6570, 2.4569, 3.3298, 3.7852, 3.4268, 2.1416], device='cuda:6'), covar=tensor([0.0506, 0.0110, 0.0053, 0.0405, 0.0123, 0.0089, 0.0095, 0.0485], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0084, 0.0086, 0.0134, 0.0099, 0.0111, 0.0097, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 15:02:21,617 INFO [train.py:904] (6/8) Epoch 23, batch 3850, loss[loss=0.1906, simple_loss=0.2633, pruned_loss=0.05897, over 16355.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2523, pruned_loss=0.04492, over 3261367.50 frames. ], batch size: 165, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:03:21,878 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6896, 4.7472, 4.9307, 4.7358, 4.7548, 5.3289, 4.8605, 4.5581], device='cuda:6'), covar=tensor([0.1649, 0.2064, 0.2134, 0.2296, 0.2824, 0.1062, 0.1775, 0.2817], device='cuda:6'), in_proj_covar=tensor([0.0422, 0.0616, 0.0680, 0.0508, 0.0679, 0.0702, 0.0533, 0.0681], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 15:03:34,968 INFO [train.py:904] (6/8) Epoch 23, batch 3900, loss[loss=0.1729, simple_loss=0.2482, pruned_loss=0.04876, over 16203.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2521, pruned_loss=0.04548, over 3263152.78 frames. ], batch size: 165, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:03:51,667 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.230e+02 2.530e+02 3.049e+02 6.179e+02, threshold=5.060e+02, percent-clipped=1.0 2023-05-01 15:04:41,909 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9009, 4.1372, 2.5006, 4.7703, 3.2786, 4.8122, 2.6943, 3.2667], device='cuda:6'), covar=tensor([0.0291, 0.0317, 0.1570, 0.0101, 0.0681, 0.0250, 0.1490, 0.0713], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0179, 0.0195, 0.0169, 0.0178, 0.0221, 0.0204, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 15:04:47,721 INFO [train.py:904] (6/8) Epoch 23, batch 3950, loss[loss=0.1683, simple_loss=0.2361, pruned_loss=0.05027, over 16874.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2514, pruned_loss=0.04548, over 3268500.91 frames. ], batch size: 109, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:06:00,635 INFO [train.py:904] (6/8) Epoch 23, batch 4000, loss[loss=0.1849, simple_loss=0.2576, pruned_loss=0.0561, over 16386.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2515, pruned_loss=0.04584, over 3272338.62 frames. ], batch size: 146, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:06:15,221 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7132, 2.6662, 2.6760, 4.9831, 3.8941, 4.2738, 1.6932, 2.9694], device='cuda:6'), covar=tensor([0.1395, 0.0852, 0.1305, 0.0114, 0.0364, 0.0353, 0.1680, 0.0951], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0175, 0.0194, 0.0193, 0.0205, 0.0216, 0.0203, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 15:06:17,148 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.633e+02 2.172e+02 2.478e+02 2.977e+02 5.073e+02, threshold=4.957e+02, percent-clipped=1.0 2023-05-01 15:06:20,185 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9685, 2.2388, 2.2430, 2.6982, 1.9826, 3.1884, 1.7639, 2.6415], device='cuda:6'), covar=tensor([0.1114, 0.0710, 0.1118, 0.0146, 0.0122, 0.0307, 0.1430, 0.0785], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0175, 0.0194, 0.0193, 0.0205, 0.0216, 0.0203, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 15:07:01,807 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=227345.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:07:13,384 INFO [train.py:904] (6/8) Epoch 23, batch 4050, loss[loss=0.1747, simple_loss=0.2608, pruned_loss=0.04427, over 16765.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2531, pruned_loss=0.0456, over 3257564.89 frames. ], batch size: 124, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:08:27,045 INFO [train.py:904] (6/8) Epoch 23, batch 4100, loss[loss=0.1709, simple_loss=0.2569, pruned_loss=0.04242, over 16724.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2548, pruned_loss=0.04525, over 3255887.50 frames. ], batch size: 89, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:08:39,180 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=227411.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:08:42,769 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 1.752e+02 1.921e+02 2.258e+02 5.560e+02, threshold=3.841e+02, percent-clipped=1.0 2023-05-01 15:09:45,391 INFO [train.py:904] (6/8) Epoch 23, batch 4150, loss[loss=0.1884, simple_loss=0.287, pruned_loss=0.04488, over 16548.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2622, pruned_loss=0.04787, over 3214666.61 frames. ], batch size: 75, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:10:16,217 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=227472.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:10:58,827 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.92 vs. limit=5.0 2023-05-01 15:11:03,932 INFO [train.py:904] (6/8) Epoch 23, batch 4200, loss[loss=0.2098, simple_loss=0.3076, pruned_loss=0.05597, over 16399.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2684, pruned_loss=0.04928, over 3190601.97 frames. ], batch size: 146, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:11:20,462 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.273e+02 2.688e+02 3.076e+02 6.689e+02, threshold=5.376e+02, percent-clipped=9.0 2023-05-01 15:12:19,943 INFO [train.py:904] (6/8) Epoch 23, batch 4250, loss[loss=0.1859, simple_loss=0.2783, pruned_loss=0.04679, over 17132.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2718, pruned_loss=0.04936, over 3167882.07 frames. ], batch size: 48, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:13:36,401 INFO [train.py:904] (6/8) Epoch 23, batch 4300, loss[loss=0.1749, simple_loss=0.2727, pruned_loss=0.03852, over 16705.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2729, pruned_loss=0.04813, over 3184211.85 frames. ], batch size: 89, lr: 2.94e-03, grad_scale: 4.0 2023-05-01 15:13:55,081 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.568e+02 2.112e+02 2.458e+02 3.022e+02 6.068e+02, threshold=4.917e+02, percent-clipped=1.0 2023-05-01 15:14:42,487 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=227645.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:14:53,543 INFO [train.py:904] (6/8) Epoch 23, batch 4350, loss[loss=0.192, simple_loss=0.2881, pruned_loss=0.04799, over 16669.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2757, pruned_loss=0.04894, over 3186389.18 frames. ], batch size: 89, lr: 2.94e-03, grad_scale: 4.0 2023-05-01 15:15:45,908 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-01 15:15:55,221 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=227693.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:15:55,352 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=227693.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:16:09,817 INFO [train.py:904] (6/8) Epoch 23, batch 4400, loss[loss=0.1904, simple_loss=0.2784, pruned_loss=0.05118, over 15475.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2779, pruned_loss=0.05012, over 3180471.20 frames. ], batch size: 190, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:16:27,158 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.099e+02 2.601e+02 2.882e+02 6.298e+02, threshold=5.202e+02, percent-clipped=2.0 2023-05-01 15:16:41,842 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8477, 2.1560, 2.3924, 3.0982, 2.1800, 2.3392, 2.3444, 2.2411], device='cuda:6'), covar=tensor([0.1345, 0.3010, 0.2456, 0.0724, 0.3803, 0.2352, 0.2865, 0.3103], device='cuda:6'), in_proj_covar=tensor([0.0407, 0.0457, 0.0372, 0.0332, 0.0439, 0.0527, 0.0426, 0.0534], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 15:17:22,034 INFO [train.py:904] (6/8) Epoch 23, batch 4450, loss[loss=0.1871, simple_loss=0.2675, pruned_loss=0.05337, over 12186.00 frames. ], tot_loss[loss=0.1923, simple_loss=0.2814, pruned_loss=0.0516, over 3183033.64 frames. ], batch size: 248, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:17:23,623 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=227754.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:17:41,689 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=227767.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:18:23,641 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 15:18:35,087 INFO [train.py:904] (6/8) Epoch 23, batch 4500, loss[loss=0.1885, simple_loss=0.278, pruned_loss=0.04944, over 16736.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2814, pruned_loss=0.05204, over 3181904.24 frames. ], batch size: 124, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:18:52,359 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 1.777e+02 2.085e+02 2.523e+02 4.139e+02, threshold=4.170e+02, percent-clipped=0.0 2023-05-01 15:19:11,712 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=227827.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:19:48,152 INFO [train.py:904] (6/8) Epoch 23, batch 4550, loss[loss=0.2064, simple_loss=0.29, pruned_loss=0.06143, over 16368.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2828, pruned_loss=0.05335, over 3181104.07 frames. ], batch size: 35, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:20:39,353 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=227888.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:21:00,778 INFO [train.py:904] (6/8) Epoch 23, batch 4600, loss[loss=0.187, simple_loss=0.2815, pruned_loss=0.04628, over 17020.00 frames. ], tot_loss[loss=0.1946, simple_loss=0.2834, pruned_loss=0.05292, over 3192649.77 frames. ], batch size: 50, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:21:18,053 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1990, 4.9577, 4.7257, 3.3871, 4.2273, 4.7694, 4.1025, 3.0844], device='cuda:6'), covar=tensor([0.0422, 0.0019, 0.0033, 0.0332, 0.0073, 0.0079, 0.0080, 0.0353], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0084, 0.0086, 0.0134, 0.0099, 0.0111, 0.0096, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 15:21:18,786 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.195e+02 1.806e+02 2.082e+02 2.426e+02 3.731e+02, threshold=4.163e+02, percent-clipped=0.0 2023-05-01 15:22:14,200 INFO [train.py:904] (6/8) Epoch 23, batch 4650, loss[loss=0.205, simple_loss=0.2879, pruned_loss=0.06107, over 16714.00 frames. ], tot_loss[loss=0.1942, simple_loss=0.2824, pruned_loss=0.05302, over 3204677.78 frames. ], batch size: 57, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:23:28,868 INFO [train.py:904] (6/8) Epoch 23, batch 4700, loss[loss=0.1863, simple_loss=0.2739, pruned_loss=0.04931, over 16746.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2799, pruned_loss=0.05194, over 3209137.91 frames. ], batch size: 83, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:23:34,967 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=228007.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:23:45,412 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 1.844e+02 2.189e+02 2.555e+02 4.222e+02, threshold=4.378e+02, percent-clipped=1.0 2023-05-01 15:24:34,874 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228049.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:24:41,825 INFO [train.py:904] (6/8) Epoch 23, batch 4750, loss[loss=0.1616, simple_loss=0.2555, pruned_loss=0.03388, over 16763.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2761, pruned_loss=0.05016, over 3200128.14 frames. ], batch size: 83, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:24:45,270 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4844, 4.5010, 4.4289, 3.4108, 4.4793, 1.5533, 4.1380, 3.9257], device='cuda:6'), covar=tensor([0.0142, 0.0152, 0.0200, 0.0641, 0.0156, 0.3261, 0.0175, 0.0401], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0162, 0.0203, 0.0180, 0.0180, 0.0209, 0.0191, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 15:25:02,199 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228067.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:25:03,428 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=228068.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:25:11,746 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-05-01 15:25:19,542 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=228080.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:25:54,062 INFO [train.py:904] (6/8) Epoch 23, batch 4800, loss[loss=0.1599, simple_loss=0.2574, pruned_loss=0.03122, over 16840.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2724, pruned_loss=0.04801, over 3197329.81 frames. ], batch size: 96, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:26:10,773 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.116e+02 1.852e+02 2.092e+02 2.435e+02 6.400e+02, threshold=4.184e+02, percent-clipped=1.0 2023-05-01 15:26:11,948 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=228115.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:26:25,191 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-01 15:26:50,649 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=228141.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:27:07,818 INFO [train.py:904] (6/8) Epoch 23, batch 4850, loss[loss=0.1965, simple_loss=0.2866, pruned_loss=0.0532, over 15420.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2728, pruned_loss=0.04763, over 3173554.42 frames. ], batch size: 190, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:27:54,530 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228183.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:28:19,207 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8886, 3.8015, 4.4873, 1.8455, 4.6703, 4.6852, 3.2838, 3.3426], device='cuda:6'), covar=tensor([0.0749, 0.0270, 0.0133, 0.1329, 0.0052, 0.0101, 0.0383, 0.0462], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0108, 0.0099, 0.0139, 0.0081, 0.0126, 0.0128, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 15:28:23,601 INFO [train.py:904] (6/8) Epoch 23, batch 4900, loss[loss=0.1544, simple_loss=0.2513, pruned_loss=0.02878, over 16802.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.2719, pruned_loss=0.04655, over 3166628.87 frames. ], batch size: 102, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:28:42,121 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.282e+02 1.946e+02 2.242e+02 2.632e+02 4.949e+02, threshold=4.484e+02, percent-clipped=2.0 2023-05-01 15:29:36,712 INFO [train.py:904] (6/8) Epoch 23, batch 4950, loss[loss=0.1916, simple_loss=0.2809, pruned_loss=0.05119, over 17054.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.2715, pruned_loss=0.04615, over 3165300.29 frames. ], batch size: 55, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:30:32,436 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2298, 2.4376, 2.4054, 4.0914, 2.2294, 2.7444, 2.4273, 2.6270], device='cuda:6'), covar=tensor([0.1447, 0.3332, 0.2790, 0.0488, 0.3771, 0.2408, 0.3393, 0.2938], device='cuda:6'), in_proj_covar=tensor([0.0407, 0.0456, 0.0372, 0.0331, 0.0438, 0.0525, 0.0426, 0.0532], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 15:30:51,766 INFO [train.py:904] (6/8) Epoch 23, batch 5000, loss[loss=0.1834, simple_loss=0.2785, pruned_loss=0.04419, over 15257.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2727, pruned_loss=0.04584, over 3173322.48 frames. ], batch size: 190, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:30:54,344 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.03 vs. limit=5.0 2023-05-01 15:31:10,001 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 1.949e+02 2.305e+02 2.743e+02 3.734e+02, threshold=4.611e+02, percent-clipped=0.0 2023-05-01 15:31:17,539 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9757, 4.9230, 4.8519, 4.0656, 4.8769, 1.8521, 4.6310, 4.5510], device='cuda:6'), covar=tensor([0.0098, 0.0105, 0.0160, 0.0503, 0.0121, 0.2742, 0.0136, 0.0245], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0160, 0.0200, 0.0177, 0.0178, 0.0207, 0.0189, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 15:31:19,852 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0512, 3.9527, 3.9209, 2.2881, 3.5223, 3.9389, 3.4663, 2.0624], device='cuda:6'), covar=tensor([0.0652, 0.0046, 0.0044, 0.0471, 0.0099, 0.0086, 0.0123, 0.0485], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0085, 0.0087, 0.0135, 0.0100, 0.0112, 0.0096, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 15:31:30,404 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=228329.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:31:36,929 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=228333.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:31:59,970 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228349.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:32:04,592 INFO [train.py:904] (6/8) Epoch 23, batch 5050, loss[loss=0.1932, simple_loss=0.2813, pruned_loss=0.05258, over 16895.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2735, pruned_loss=0.04566, over 3186749.01 frames. ], batch size: 116, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:32:19,753 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228363.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:32:19,875 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4099, 4.4916, 4.3248, 3.9954, 3.9909, 4.4022, 4.1264, 4.1620], device='cuda:6'), covar=tensor([0.0609, 0.0478, 0.0297, 0.0299, 0.0911, 0.0506, 0.0625, 0.0603], device='cuda:6'), in_proj_covar=tensor([0.0297, 0.0439, 0.0347, 0.0345, 0.0351, 0.0404, 0.0237, 0.0413], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 15:32:58,986 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=228390.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:33:04,509 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=228394.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:33:09,565 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=228397.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:33:17,239 INFO [train.py:904] (6/8) Epoch 23, batch 5100, loss[loss=0.1741, simple_loss=0.2644, pruned_loss=0.04188, over 16752.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2725, pruned_loss=0.04552, over 3192155.10 frames. ], batch size: 124, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:33:34,756 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 1.954e+02 2.359e+02 2.853e+02 6.747e+02, threshold=4.718e+02, percent-clipped=3.0 2023-05-01 15:33:55,595 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=228429.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 15:34:06,505 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228436.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:34:30,926 INFO [train.py:904] (6/8) Epoch 23, batch 5150, loss[loss=0.1676, simple_loss=0.2704, pruned_loss=0.03241, over 16872.00 frames. ], tot_loss[loss=0.1809, simple_loss=0.2725, pruned_loss=0.04464, over 3194568.85 frames. ], batch size: 102, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:35:15,408 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228483.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:35:25,946 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=228490.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 15:35:44,365 INFO [train.py:904] (6/8) Epoch 23, batch 5200, loss[loss=0.1595, simple_loss=0.2429, pruned_loss=0.03805, over 16240.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2705, pruned_loss=0.04361, over 3200021.96 frames. ], batch size: 35, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:36:01,315 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.301e+02 2.022e+02 2.336e+02 2.717e+02 6.475e+02, threshold=4.673e+02, percent-clipped=1.0 2023-05-01 15:36:25,678 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=228531.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:36:28,953 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9992, 3.0707, 1.8784, 3.2597, 2.4071, 3.3151, 2.1135, 2.5839], device='cuda:6'), covar=tensor([0.0273, 0.0354, 0.1557, 0.0195, 0.0783, 0.0511, 0.1466, 0.0746], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0175, 0.0192, 0.0163, 0.0174, 0.0215, 0.0200, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 15:36:33,203 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7088, 3.7653, 2.6848, 2.2718, 2.5682, 2.3836, 3.8034, 3.3524], device='cuda:6'), covar=tensor([0.2992, 0.0649, 0.2082, 0.2874, 0.2652, 0.2097, 0.0680, 0.1296], device='cuda:6'), in_proj_covar=tensor([0.0329, 0.0271, 0.0307, 0.0316, 0.0299, 0.0262, 0.0299, 0.0341], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 15:36:57,628 INFO [train.py:904] (6/8) Epoch 23, batch 5250, loss[loss=0.188, simple_loss=0.2794, pruned_loss=0.04832, over 16678.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2684, pruned_loss=0.04369, over 3199369.75 frames. ], batch size: 134, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:38:10,575 INFO [train.py:904] (6/8) Epoch 23, batch 5300, loss[loss=0.155, simple_loss=0.2413, pruned_loss=0.03435, over 16477.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.2648, pruned_loss=0.04243, over 3206519.63 frames. ], batch size: 146, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:38:13,179 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2023-05-01 15:38:28,435 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 1.965e+02 2.226e+02 2.604e+02 5.338e+02, threshold=4.452e+02, percent-clipped=3.0 2023-05-01 15:39:23,395 INFO [train.py:904] (6/8) Epoch 23, batch 5350, loss[loss=0.1881, simple_loss=0.2787, pruned_loss=0.0487, over 16695.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2631, pruned_loss=0.04183, over 3217739.26 frames. ], batch size: 134, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:39:34,153 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4930, 3.6579, 2.1633, 4.1684, 2.7768, 4.0605, 2.3123, 2.8977], device='cuda:6'), covar=tensor([0.0321, 0.0377, 0.1726, 0.0217, 0.0921, 0.0691, 0.1640, 0.0877], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0177, 0.0193, 0.0164, 0.0176, 0.0217, 0.0201, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 15:39:37,204 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8595, 4.8842, 5.2284, 5.1946, 5.2118, 4.9250, 4.8475, 4.6845], device='cuda:6'), covar=tensor([0.0281, 0.0551, 0.0323, 0.0404, 0.0415, 0.0310, 0.0919, 0.0443], device='cuda:6'), in_proj_covar=tensor([0.0409, 0.0454, 0.0441, 0.0411, 0.0487, 0.0461, 0.0547, 0.0371], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 15:39:38,494 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228663.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:40:10,707 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228685.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:40:10,953 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4271, 3.3973, 2.6178, 2.1626, 2.2386, 2.2589, 3.4949, 3.1147], device='cuda:6'), covar=tensor([0.3063, 0.0632, 0.1967, 0.2914, 0.2650, 0.2157, 0.0576, 0.1284], device='cuda:6'), in_proj_covar=tensor([0.0329, 0.0271, 0.0308, 0.0316, 0.0299, 0.0263, 0.0300, 0.0341], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 15:40:16,851 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228689.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:40:36,799 INFO [train.py:904] (6/8) Epoch 23, batch 5400, loss[loss=0.1814, simple_loss=0.275, pruned_loss=0.04392, over 16480.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2658, pruned_loss=0.04262, over 3209089.41 frames. ], batch size: 75, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:40:46,748 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1293, 2.2117, 2.1028, 3.8684, 2.1069, 2.6100, 2.3223, 2.4419], device='cuda:6'), covar=tensor([0.1463, 0.3726, 0.3148, 0.0547, 0.4054, 0.2527, 0.3640, 0.3125], device='cuda:6'), in_proj_covar=tensor([0.0405, 0.0453, 0.0371, 0.0330, 0.0436, 0.0522, 0.0424, 0.0529], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 15:40:48,793 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=228711.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:40:51,852 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 15:40:54,347 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.419e+02 1.938e+02 2.190e+02 2.538e+02 4.586e+02, threshold=4.379e+02, percent-clipped=1.0 2023-05-01 15:41:27,411 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228736.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:41:54,045 INFO [train.py:904] (6/8) Epoch 23, batch 5450, loss[loss=0.1701, simple_loss=0.2597, pruned_loss=0.04031, over 16421.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2679, pruned_loss=0.04351, over 3196405.58 frames. ], batch size: 68, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:41:56,900 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4115, 3.3625, 3.4288, 3.5144, 3.5564, 3.2965, 3.5187, 3.6166], device='cuda:6'), covar=tensor([0.1225, 0.0978, 0.1109, 0.0664, 0.0692, 0.2335, 0.1155, 0.0760], device='cuda:6'), in_proj_covar=tensor([0.0640, 0.0797, 0.0921, 0.0802, 0.0610, 0.0638, 0.0658, 0.0764], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 15:42:42,359 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=228784.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:42:44,360 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228785.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 15:43:12,599 INFO [train.py:904] (6/8) Epoch 23, batch 5500, loss[loss=0.2139, simple_loss=0.3039, pruned_loss=0.06195, over 17138.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2752, pruned_loss=0.04798, over 3167288.27 frames. ], batch size: 47, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:43:32,425 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.710e+02 2.602e+02 3.171e+02 3.709e+02 7.104e+02, threshold=6.343e+02, percent-clipped=12.0 2023-05-01 15:44:19,501 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.42 vs. limit=5.0 2023-05-01 15:44:31,572 INFO [train.py:904] (6/8) Epoch 23, batch 5550, loss[loss=0.2203, simple_loss=0.3046, pruned_loss=0.06798, over 15306.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2824, pruned_loss=0.05294, over 3138716.37 frames. ], batch size: 191, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:45:52,969 INFO [train.py:904] (6/8) Epoch 23, batch 5600, loss[loss=0.2645, simple_loss=0.3225, pruned_loss=0.1033, over 10721.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2873, pruned_loss=0.05751, over 3086823.98 frames. ], batch size: 248, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:46:10,853 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.072e+02 3.210e+02 3.734e+02 4.915e+02 1.270e+03, threshold=7.469e+02, percent-clipped=7.0 2023-05-01 15:46:29,716 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8905, 4.8867, 4.7653, 4.3991, 4.3975, 4.7961, 4.6782, 4.5417], device='cuda:6'), covar=tensor([0.0694, 0.0543, 0.0329, 0.0355, 0.1096, 0.0504, 0.0414, 0.0729], device='cuda:6'), in_proj_covar=tensor([0.0297, 0.0440, 0.0347, 0.0345, 0.0353, 0.0404, 0.0236, 0.0413], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 15:46:45,609 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=228935.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:47:13,516 INFO [train.py:904] (6/8) Epoch 23, batch 5650, loss[loss=0.2946, simple_loss=0.3468, pruned_loss=0.1212, over 11150.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2911, pruned_loss=0.05975, over 3088940.71 frames. ], batch size: 248, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:48:04,663 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228985.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:48:11,550 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228989.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:48:22,396 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=228996.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:48:32,719 INFO [train.py:904] (6/8) Epoch 23, batch 5700, loss[loss=0.21, simple_loss=0.3021, pruned_loss=0.05897, over 16760.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2924, pruned_loss=0.06091, over 3096596.26 frames. ], batch size: 39, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:48:50,700 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.137e+02 3.035e+02 3.772e+02 4.443e+02 6.175e+02, threshold=7.544e+02, percent-clipped=0.0 2023-05-01 15:49:18,826 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=229033.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:49:25,417 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=229037.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:49:50,897 INFO [train.py:904] (6/8) Epoch 23, batch 5750, loss[loss=0.2134, simple_loss=0.3011, pruned_loss=0.06291, over 16550.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2954, pruned_loss=0.0627, over 3073899.78 frames. ], batch size: 62, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:50:44,764 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=229085.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 15:51:13,969 INFO [train.py:904] (6/8) Epoch 23, batch 5800, loss[loss=0.2125, simple_loss=0.2918, pruned_loss=0.06663, over 12202.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2948, pruned_loss=0.06151, over 3073974.50 frames. ], batch size: 246, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:51:20,402 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-01 15:51:30,367 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3865, 4.3690, 4.7044, 4.6961, 4.7038, 4.4156, 4.3884, 4.3216], device='cuda:6'), covar=tensor([0.0342, 0.0607, 0.0451, 0.0427, 0.0469, 0.0428, 0.0891, 0.0530], device='cuda:6'), in_proj_covar=tensor([0.0413, 0.0460, 0.0445, 0.0414, 0.0490, 0.0467, 0.0551, 0.0372], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 15:51:32,214 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.995e+02 2.918e+02 3.323e+02 4.052e+02 8.500e+02, threshold=6.645e+02, percent-clipped=1.0 2023-05-01 15:51:44,230 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-01 15:52:01,486 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=229133.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 15:52:06,793 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-01 15:52:31,388 INFO [train.py:904] (6/8) Epoch 23, batch 5850, loss[loss=0.2041, simple_loss=0.2979, pruned_loss=0.05518, over 16280.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2933, pruned_loss=0.06066, over 3065811.32 frames. ], batch size: 165, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:53:43,908 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4380, 2.2232, 1.9392, 2.0174, 2.5193, 2.1777, 2.2595, 2.6503], device='cuda:6'), covar=tensor([0.0231, 0.0453, 0.0546, 0.0478, 0.0263, 0.0400, 0.0197, 0.0276], device='cuda:6'), in_proj_covar=tensor([0.0215, 0.0236, 0.0226, 0.0229, 0.0237, 0.0236, 0.0236, 0.0234], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 15:53:53,137 INFO [train.py:904] (6/8) Epoch 23, batch 5900, loss[loss=0.2271, simple_loss=0.2941, pruned_loss=0.08001, over 11505.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2929, pruned_loss=0.06047, over 3083163.30 frames. ], batch size: 246, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:54:16,065 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.674e+02 2.520e+02 3.106e+02 3.673e+02 6.247e+02, threshold=6.213e+02, percent-clipped=0.0 2023-05-01 15:55:17,140 INFO [train.py:904] (6/8) Epoch 23, batch 5950, loss[loss=0.1881, simple_loss=0.2838, pruned_loss=0.04618, over 17225.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2932, pruned_loss=0.05927, over 3073347.04 frames. ], batch size: 52, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:55:22,768 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3913, 3.3524, 3.4319, 3.5141, 3.5556, 3.2905, 3.5261, 3.6108], device='cuda:6'), covar=tensor([0.1210, 0.0909, 0.0991, 0.0649, 0.0670, 0.2506, 0.1075, 0.0821], device='cuda:6'), in_proj_covar=tensor([0.0635, 0.0789, 0.0906, 0.0794, 0.0604, 0.0632, 0.0654, 0.0758], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 15:56:18,281 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=229291.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:56:33,855 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229300.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:56:37,757 INFO [train.py:904] (6/8) Epoch 23, batch 6000, loss[loss=0.188, simple_loss=0.2799, pruned_loss=0.04808, over 16361.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2925, pruned_loss=0.059, over 3079192.65 frames. ], batch size: 35, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:56:37,758 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 15:56:49,494 INFO [train.py:938] (6/8) Epoch 23, validation: loss=0.1497, simple_loss=0.2623, pruned_loss=0.01859, over 944034.00 frames. 2023-05-01 15:56:49,494 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 15:57:07,747 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.923e+02 2.769e+02 3.211e+02 3.808e+02 9.716e+02, threshold=6.422e+02, percent-clipped=3.0 2023-05-01 15:57:30,466 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1822, 3.4721, 3.5120, 2.1731, 3.0823, 2.3629, 3.5996, 3.7423], device='cuda:6'), covar=tensor([0.0247, 0.0744, 0.0606, 0.1949, 0.0811, 0.0944, 0.0616, 0.0926], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0166, 0.0168, 0.0154, 0.0147, 0.0131, 0.0144, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 15:58:01,595 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5473, 1.6569, 2.1864, 2.5418, 2.5338, 2.8203, 1.9395, 2.8248], device='cuda:6'), covar=tensor([0.0218, 0.0621, 0.0392, 0.0343, 0.0345, 0.0219, 0.0580, 0.0157], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0193, 0.0180, 0.0185, 0.0199, 0.0156, 0.0197, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 15:58:06,098 INFO [train.py:904] (6/8) Epoch 23, batch 6050, loss[loss=0.1982, simple_loss=0.2922, pruned_loss=0.05206, over 16713.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2914, pruned_loss=0.05845, over 3084111.25 frames. ], batch size: 124, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:58:21,089 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=229361.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:58:53,231 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8997, 2.0880, 2.4356, 3.1430, 2.1795, 2.3508, 2.3318, 2.2682], device='cuda:6'), covar=tensor([0.1405, 0.3405, 0.2340, 0.0739, 0.4022, 0.2338, 0.3152, 0.3184], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0452, 0.0368, 0.0326, 0.0434, 0.0518, 0.0422, 0.0527], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 15:59:21,989 INFO [train.py:904] (6/8) Epoch 23, batch 6100, loss[loss=0.1895, simple_loss=0.2823, pruned_loss=0.04832, over 16225.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2907, pruned_loss=0.05713, over 3098678.55 frames. ], batch size: 165, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:59:40,508 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 2.664e+02 3.486e+02 4.048e+02 8.551e+02, threshold=6.973e+02, percent-clipped=3.0 2023-05-01 16:00:22,344 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229443.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:00:36,341 INFO [train.py:904] (6/8) Epoch 23, batch 6150, loss[loss=0.1849, simple_loss=0.271, pruned_loss=0.04941, over 17115.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2884, pruned_loss=0.05638, over 3111655.72 frames. ], batch size: 47, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 16:01:10,907 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-01 16:01:15,849 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229477.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:01:36,803 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4712, 2.5790, 2.5319, 4.4292, 2.4205, 2.9170, 2.5707, 2.7182], device='cuda:6'), covar=tensor([0.1243, 0.3122, 0.2652, 0.0434, 0.3720, 0.2212, 0.3116, 0.2992], device='cuda:6'), in_proj_covar=tensor([0.0404, 0.0453, 0.0370, 0.0327, 0.0436, 0.0520, 0.0423, 0.0529], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:01:53,539 INFO [train.py:904] (6/8) Epoch 23, batch 6200, loss[loss=0.168, simple_loss=0.2595, pruned_loss=0.0383, over 17065.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.287, pruned_loss=0.05629, over 3108373.27 frames. ], batch size: 55, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:01:55,777 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=229504.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:02:14,504 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.866e+02 2.788e+02 3.260e+02 3.956e+02 6.763e+02, threshold=6.520e+02, percent-clipped=0.0 2023-05-01 16:02:50,007 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=229538.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:03:11,717 INFO [train.py:904] (6/8) Epoch 23, batch 6250, loss[loss=0.1994, simple_loss=0.2831, pruned_loss=0.05787, over 15346.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2869, pruned_loss=0.05628, over 3108745.73 frames. ], batch size: 190, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:03:49,525 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229577.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:04:09,840 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=229591.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:04:28,569 INFO [train.py:904] (6/8) Epoch 23, batch 6300, loss[loss=0.2025, simple_loss=0.2838, pruned_loss=0.06056, over 16408.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2869, pruned_loss=0.05566, over 3116854.50 frames. ], batch size: 146, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:04:30,287 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6970, 4.7607, 4.5921, 4.2299, 4.2344, 4.6533, 4.4663, 4.3932], device='cuda:6'), covar=tensor([0.0637, 0.0502, 0.0296, 0.0334, 0.1013, 0.0461, 0.0486, 0.0615], device='cuda:6'), in_proj_covar=tensor([0.0293, 0.0436, 0.0343, 0.0340, 0.0348, 0.0399, 0.0233, 0.0408], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:04:50,640 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 2.619e+02 3.224e+02 4.167e+02 1.378e+03, threshold=6.449e+02, percent-clipped=8.0 2023-05-01 16:05:27,327 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=229638.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:05:28,547 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=229639.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:05:48,008 INFO [train.py:904] (6/8) Epoch 23, batch 6350, loss[loss=0.1845, simple_loss=0.2768, pruned_loss=0.0461, over 16692.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2865, pruned_loss=0.05593, over 3133210.43 frames. ], batch size: 89, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:05:53,552 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=229656.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:05:57,995 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229659.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 16:06:10,597 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3884, 4.6495, 4.4129, 4.4513, 4.2299, 4.1609, 4.1542, 4.6983], device='cuda:6'), covar=tensor([0.1133, 0.0845, 0.1068, 0.0892, 0.0822, 0.1678, 0.1120, 0.0943], device='cuda:6'), in_proj_covar=tensor([0.0673, 0.0818, 0.0677, 0.0624, 0.0518, 0.0527, 0.0688, 0.0643], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:06:53,467 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6073, 3.4403, 3.9064, 1.9827, 4.0236, 4.0840, 2.9901, 3.0238], device='cuda:6'), covar=tensor([0.0796, 0.0261, 0.0184, 0.1197, 0.0075, 0.0161, 0.0423, 0.0456], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0109, 0.0099, 0.0139, 0.0082, 0.0126, 0.0128, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 16:07:04,876 INFO [train.py:904] (6/8) Epoch 23, batch 6400, loss[loss=0.174, simple_loss=0.27, pruned_loss=0.03898, over 16868.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2864, pruned_loss=0.05677, over 3128562.75 frames. ], batch size: 102, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:07:14,763 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6983, 4.5517, 4.7773, 4.9385, 5.0784, 4.5321, 5.0716, 5.0881], device='cuda:6'), covar=tensor([0.1890, 0.1240, 0.1553, 0.0698, 0.0546, 0.0967, 0.0563, 0.0607], device='cuda:6'), in_proj_covar=tensor([0.0633, 0.0784, 0.0902, 0.0790, 0.0601, 0.0628, 0.0650, 0.0756], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:07:24,870 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.907e+02 2.905e+02 3.417e+02 4.252e+02 8.377e+02, threshold=6.835e+02, percent-clipped=2.0 2023-05-01 16:07:32,387 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=229720.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 16:07:57,586 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6244, 6.0196, 5.7215, 5.7873, 5.4227, 5.4235, 5.3473, 6.1099], device='cuda:6'), covar=tensor([0.1393, 0.0763, 0.1012, 0.0923, 0.0890, 0.0644, 0.1340, 0.0825], device='cuda:6'), in_proj_covar=tensor([0.0674, 0.0818, 0.0677, 0.0625, 0.0518, 0.0526, 0.0689, 0.0643], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:08:17,981 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4950, 1.7150, 2.2081, 2.4552, 2.5035, 2.7656, 1.9375, 2.6182], device='cuda:6'), covar=tensor([0.0235, 0.0532, 0.0321, 0.0343, 0.0352, 0.0188, 0.0513, 0.0164], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0193, 0.0179, 0.0184, 0.0199, 0.0156, 0.0198, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:08:21,137 INFO [train.py:904] (6/8) Epoch 23, batch 6450, loss[loss=0.1869, simple_loss=0.2793, pruned_loss=0.04725, over 15440.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2863, pruned_loss=0.05594, over 3132469.39 frames. ], batch size: 191, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:08:39,087 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4343, 3.1996, 3.6076, 1.7735, 3.6971, 3.7583, 2.8096, 2.7995], device='cuda:6'), covar=tensor([0.0816, 0.0288, 0.0191, 0.1309, 0.0090, 0.0193, 0.0483, 0.0502], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0108, 0.0098, 0.0139, 0.0082, 0.0126, 0.0127, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 16:09:34,332 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=229799.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:09:40,106 INFO [train.py:904] (6/8) Epoch 23, batch 6500, loss[loss=0.2018, simple_loss=0.2848, pruned_loss=0.05939, over 16227.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2852, pruned_loss=0.05622, over 3115853.69 frames. ], batch size: 165, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:09:59,295 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.807e+02 2.714e+02 3.280e+02 3.954e+02 7.541e+02, threshold=6.561e+02, percent-clipped=1.0 2023-05-01 16:10:25,443 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=229833.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:10:57,914 INFO [train.py:904] (6/8) Epoch 23, batch 6550, loss[loss=0.2152, simple_loss=0.3196, pruned_loss=0.05536, over 16223.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2881, pruned_loss=0.05736, over 3107101.12 frames. ], batch size: 165, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:12:02,467 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229894.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:12:15,578 INFO [train.py:904] (6/8) Epoch 23, batch 6600, loss[loss=0.2124, simple_loss=0.3017, pruned_loss=0.06154, over 16385.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2905, pruned_loss=0.05784, over 3116510.96 frames. ], batch size: 146, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:12:35,465 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.769e+02 2.706e+02 3.377e+02 4.261e+02 8.660e+02, threshold=6.754e+02, percent-clipped=4.0 2023-05-01 16:13:02,419 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=229933.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:13:33,860 INFO [train.py:904] (6/8) Epoch 23, batch 6650, loss[loss=0.174, simple_loss=0.2658, pruned_loss=0.04111, over 16862.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2908, pruned_loss=0.05882, over 3105198.91 frames. ], batch size: 90, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:13:37,587 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=229955.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:13:38,707 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=229956.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:14:26,144 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229987.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:14:52,371 INFO [train.py:904] (6/8) Epoch 23, batch 6700, loss[loss=0.196, simple_loss=0.2787, pruned_loss=0.05663, over 16833.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2899, pruned_loss=0.05878, over 3111585.91 frames. ], batch size: 116, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:14:54,788 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230004.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:15:07,363 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.87 vs. limit=5.0 2023-05-01 16:15:11,327 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230015.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 16:15:12,080 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 2.536e+02 3.063e+02 3.747e+02 8.042e+02, threshold=6.126e+02, percent-clipped=2.0 2023-05-01 16:16:02,759 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230048.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:16:09,691 INFO [train.py:904] (6/8) Epoch 23, batch 6750, loss[loss=0.2066, simple_loss=0.28, pruned_loss=0.06666, over 16995.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.29, pruned_loss=0.05974, over 3088035.13 frames. ], batch size: 41, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:16:42,471 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9400, 2.1311, 2.2758, 3.4121, 2.0758, 2.4530, 2.2645, 2.2674], device='cuda:6'), covar=tensor([0.1404, 0.3294, 0.2761, 0.0664, 0.4122, 0.2304, 0.3378, 0.3350], device='cuda:6'), in_proj_covar=tensor([0.0404, 0.0452, 0.0369, 0.0327, 0.0436, 0.0519, 0.0423, 0.0528], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:17:19,818 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230099.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:17:24,162 INFO [train.py:904] (6/8) Epoch 23, batch 6800, loss[loss=0.2109, simple_loss=0.2925, pruned_loss=0.06467, over 16742.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2907, pruned_loss=0.06022, over 3083309.35 frames. ], batch size: 124, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:17:43,640 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.754e+02 2.682e+02 3.194e+02 4.039e+02 7.212e+02, threshold=6.387e+02, percent-clipped=3.0 2023-05-01 16:17:53,818 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2329, 2.2131, 2.3414, 3.8707, 2.1927, 2.3922, 2.2914, 2.2950], device='cuda:6'), covar=tensor([0.1557, 0.4019, 0.3087, 0.0687, 0.4803, 0.3038, 0.3776, 0.4022], device='cuda:6'), in_proj_covar=tensor([0.0404, 0.0451, 0.0369, 0.0327, 0.0436, 0.0519, 0.0423, 0.0527], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:18:11,077 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230133.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:18:31,931 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230147.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:18:36,496 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230150.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:18:40,517 INFO [train.py:904] (6/8) Epoch 23, batch 6850, loss[loss=0.1995, simple_loss=0.2907, pruned_loss=0.05415, over 16371.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.292, pruned_loss=0.06079, over 3087857.00 frames. ], batch size: 165, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:19:20,587 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3128, 2.0427, 2.7450, 3.1809, 2.9103, 3.5414, 2.1077, 3.5591], device='cuda:6'), covar=tensor([0.0161, 0.0551, 0.0324, 0.0289, 0.0359, 0.0158, 0.0628, 0.0132], device='cuda:6'), in_proj_covar=tensor([0.0190, 0.0193, 0.0179, 0.0185, 0.0199, 0.0155, 0.0198, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:19:23,367 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230181.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:19:42,318 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0142, 2.9735, 2.6818, 2.7137, 3.4319, 2.9927, 3.4787, 3.6411], device='cuda:6'), covar=tensor([0.0128, 0.0532, 0.0549, 0.0506, 0.0279, 0.0401, 0.0293, 0.0246], device='cuda:6'), in_proj_covar=tensor([0.0213, 0.0235, 0.0226, 0.0229, 0.0237, 0.0235, 0.0235, 0.0233], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:19:56,865 INFO [train.py:904] (6/8) Epoch 23, batch 6900, loss[loss=0.2661, simple_loss=0.3288, pruned_loss=0.1016, over 11291.00 frames. ], tot_loss[loss=0.2067, simple_loss=0.294, pruned_loss=0.05972, over 3102897.15 frames. ], batch size: 246, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:20:11,446 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230211.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:20:20,076 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.486e+02 2.571e+02 3.126e+02 3.927e+02 7.395e+02, threshold=6.253e+02, percent-clipped=1.0 2023-05-01 16:20:45,472 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230233.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:20:59,016 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230241.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:21:13,679 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230250.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:21:17,999 INFO [train.py:904] (6/8) Epoch 23, batch 6950, loss[loss=0.1983, simple_loss=0.2845, pruned_loss=0.05598, over 16728.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2953, pruned_loss=0.06096, over 3093410.97 frames. ], batch size: 124, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:22:01,066 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230281.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:22:34,176 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230302.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:22:34,886 INFO [train.py:904] (6/8) Epoch 23, batch 7000, loss[loss=0.1963, simple_loss=0.2923, pruned_loss=0.05014, over 16638.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.2953, pruned_loss=0.05973, over 3110855.92 frames. ], batch size: 57, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:22:53,481 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230315.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 16:22:56,095 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.940e+02 2.703e+02 3.306e+02 4.193e+02 6.685e+02, threshold=6.612e+02, percent-clipped=2.0 2023-05-01 16:23:35,807 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230343.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:23:50,811 INFO [train.py:904] (6/8) Epoch 23, batch 7050, loss[loss=0.2163, simple_loss=0.2949, pruned_loss=0.06886, over 16927.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2956, pruned_loss=0.05926, over 3106274.75 frames. ], batch size: 109, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:24:05,842 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230363.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 16:25:07,481 INFO [train.py:904] (6/8) Epoch 23, batch 7100, loss[loss=0.2215, simple_loss=0.3026, pruned_loss=0.07016, over 15566.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2936, pruned_loss=0.05928, over 3091727.50 frames. ], batch size: 191, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:25:18,532 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0464, 4.0486, 3.9855, 3.2047, 3.9804, 1.7900, 3.7908, 3.4884], device='cuda:6'), covar=tensor([0.0121, 0.0113, 0.0191, 0.0287, 0.0095, 0.2831, 0.0134, 0.0300], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0159, 0.0199, 0.0176, 0.0175, 0.0205, 0.0187, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:25:30,796 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 2.925e+02 3.270e+02 3.973e+02 7.550e+02, threshold=6.541e+02, percent-clipped=2.0 2023-05-01 16:25:35,687 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5357, 3.3124, 3.8233, 1.7772, 3.8989, 4.0151, 2.9108, 2.8124], device='cuda:6'), covar=tensor([0.0853, 0.0322, 0.0199, 0.1402, 0.0087, 0.0170, 0.0518, 0.0554], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0108, 0.0099, 0.0138, 0.0082, 0.0126, 0.0127, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 16:26:24,958 INFO [train.py:904] (6/8) Epoch 23, batch 7150, loss[loss=0.2236, simple_loss=0.3135, pruned_loss=0.06687, over 15463.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2915, pruned_loss=0.05898, over 3085125.98 frames. ], batch size: 190, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:26:49,274 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230468.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:27:42,211 INFO [train.py:904] (6/8) Epoch 23, batch 7200, loss[loss=0.1915, simple_loss=0.2816, pruned_loss=0.05075, over 15386.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2899, pruned_loss=0.05805, over 3053848.56 frames. ], batch size: 191, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:27:47,860 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230506.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:28:03,476 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.863e+02 2.554e+02 3.131e+02 3.804e+02 6.997e+02, threshold=6.261e+02, percent-clipped=1.0 2023-05-01 16:28:16,791 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230525.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:28:23,162 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230529.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 16:28:30,066 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230533.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:28:58,987 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230550.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:29:02,529 INFO [train.py:904] (6/8) Epoch 23, batch 7250, loss[loss=0.1909, simple_loss=0.2759, pruned_loss=0.05301, over 16787.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2872, pruned_loss=0.05686, over 3056629.56 frames. ], batch size: 124, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:29:51,464 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230586.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:30:03,523 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230594.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:30:08,932 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230597.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:30:10,101 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230598.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:30:15,615 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9966, 3.2257, 3.2087, 2.0963, 2.9459, 3.2161, 3.0101, 2.0297], device='cuda:6'), covar=tensor([0.0556, 0.0068, 0.0075, 0.0474, 0.0133, 0.0135, 0.0125, 0.0455], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0086, 0.0087, 0.0136, 0.0100, 0.0112, 0.0097, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 16:30:18,070 INFO [train.py:904] (6/8) Epoch 23, batch 7300, loss[loss=0.1878, simple_loss=0.2832, pruned_loss=0.04621, over 16387.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2875, pruned_loss=0.05723, over 3056988.07 frames. ], batch size: 146, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:30:39,629 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.739e+02 2.620e+02 3.228e+02 4.011e+02 8.387e+02, threshold=6.456e+02, percent-clipped=2.0 2023-05-01 16:31:19,767 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230643.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:31:29,220 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 16:31:33,676 INFO [train.py:904] (6/8) Epoch 23, batch 7350, loss[loss=0.1841, simple_loss=0.2755, pruned_loss=0.04638, over 15431.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2886, pruned_loss=0.05806, over 3048392.18 frames. ], batch size: 191, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:31:39,856 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-05-01 16:32:32,620 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 16:32:33,478 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230691.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:32:53,495 INFO [train.py:904] (6/8) Epoch 23, batch 7400, loss[loss=0.2241, simple_loss=0.317, pruned_loss=0.06561, over 16855.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2896, pruned_loss=0.05885, over 3052888.30 frames. ], batch size: 42, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:33:16,062 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.016e+02 2.706e+02 3.245e+02 4.052e+02 5.999e+02, threshold=6.490e+02, percent-clipped=0.0 2023-05-01 16:33:53,032 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2189, 4.2988, 4.1148, 3.8512, 3.8506, 4.2250, 3.8894, 3.9868], device='cuda:6'), covar=tensor([0.0606, 0.0612, 0.0300, 0.0291, 0.0766, 0.0474, 0.0834, 0.0571], device='cuda:6'), in_proj_covar=tensor([0.0288, 0.0429, 0.0335, 0.0335, 0.0341, 0.0392, 0.0230, 0.0403], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:34:13,423 INFO [train.py:904] (6/8) Epoch 23, batch 7450, loss[loss=0.1951, simple_loss=0.2863, pruned_loss=0.05193, over 16925.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2905, pruned_loss=0.05955, over 3068350.91 frames. ], batch size: 109, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:35:19,291 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230792.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:35:35,338 INFO [train.py:904] (6/8) Epoch 23, batch 7500, loss[loss=0.1946, simple_loss=0.2823, pruned_loss=0.05347, over 15318.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2914, pruned_loss=0.05958, over 3043830.79 frames. ], batch size: 190, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:35:41,047 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230806.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:35:58,126 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.010e+02 2.839e+02 3.579e+02 4.414e+02 1.002e+03, threshold=7.158e+02, percent-clipped=2.0 2023-05-01 16:36:09,655 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230824.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 16:36:52,035 INFO [train.py:904] (6/8) Epoch 23, batch 7550, loss[loss=0.2338, simple_loss=0.3016, pruned_loss=0.08297, over 11460.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2905, pruned_loss=0.06, over 3023246.73 frames. ], batch size: 247, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:36:52,594 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230853.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:36:53,662 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230854.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:37:11,852 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6897, 3.6790, 2.0846, 4.3092, 2.7898, 4.2283, 2.3503, 2.9499], device='cuda:6'), covar=tensor([0.0284, 0.0414, 0.1867, 0.0214, 0.0888, 0.0554, 0.1592, 0.0836], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0176, 0.0194, 0.0164, 0.0176, 0.0216, 0.0202, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 16:37:33,348 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230881.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:37:45,765 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230889.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:37:47,091 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230890.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:37:47,324 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-05-01 16:37:58,363 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230897.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:38:06,491 INFO [train.py:904] (6/8) Epoch 23, batch 7600, loss[loss=0.1842, simple_loss=0.2765, pruned_loss=0.04593, over 16853.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2888, pruned_loss=0.0594, over 3034774.70 frames. ], batch size: 116, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:38:27,915 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.909e+02 2.696e+02 3.168e+02 3.892e+02 6.127e+02, threshold=6.336e+02, percent-clipped=0.0 2023-05-01 16:39:03,480 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3689, 4.6627, 4.4523, 4.4535, 4.2050, 4.1586, 4.1988, 4.7026], device='cuda:6'), covar=tensor([0.1178, 0.0845, 0.1058, 0.0905, 0.0847, 0.1626, 0.1204, 0.0967], device='cuda:6'), in_proj_covar=tensor([0.0675, 0.0816, 0.0678, 0.0629, 0.0520, 0.0530, 0.0690, 0.0647], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:39:10,345 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230945.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:39:19,862 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230951.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:39:21,933 INFO [train.py:904] (6/8) Epoch 23, batch 7650, loss[loss=0.2259, simple_loss=0.2967, pruned_loss=0.07753, over 11523.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2895, pruned_loss=0.06003, over 3035438.72 frames. ], batch size: 247, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:40:09,829 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3703, 3.3889, 2.0054, 3.7719, 2.5853, 3.7689, 2.2808, 2.7094], device='cuda:6'), covar=tensor([0.0296, 0.0412, 0.1738, 0.0204, 0.0856, 0.0555, 0.1447, 0.0835], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0176, 0.0195, 0.0164, 0.0176, 0.0217, 0.0203, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 16:40:35,943 INFO [train.py:904] (6/8) Epoch 23, batch 7700, loss[loss=0.1875, simple_loss=0.2756, pruned_loss=0.04972, over 17038.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2901, pruned_loss=0.0606, over 3041618.31 frames. ], batch size: 55, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:40:41,058 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.82 vs. limit=5.0 2023-05-01 16:40:57,684 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.025e+02 2.849e+02 3.493e+02 4.384e+02 8.642e+02, threshold=6.986e+02, percent-clipped=7.0 2023-05-01 16:41:26,960 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2470, 2.1578, 2.7482, 3.1517, 3.0132, 3.7697, 2.2936, 3.6029], device='cuda:6'), covar=tensor([0.0244, 0.0529, 0.0369, 0.0320, 0.0344, 0.0146, 0.0568, 0.0153], device='cuda:6'), in_proj_covar=tensor([0.0190, 0.0193, 0.0179, 0.0185, 0.0198, 0.0155, 0.0198, 0.0154], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:41:53,915 INFO [train.py:904] (6/8) Epoch 23, batch 7750, loss[loss=0.1943, simple_loss=0.2832, pruned_loss=0.0527, over 16695.00 frames. ], tot_loss[loss=0.205, simple_loss=0.2899, pruned_loss=0.06002, over 3064070.68 frames. ], batch size: 62, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:42:28,903 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 16:43:09,649 INFO [train.py:904] (6/8) Epoch 23, batch 7800, loss[loss=0.2582, simple_loss=0.3145, pruned_loss=0.101, over 11289.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2905, pruned_loss=0.06045, over 3062137.85 frames. ], batch size: 248, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:43:29,792 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6719, 3.9424, 2.9568, 2.2921, 2.6534, 2.5088, 4.2426, 3.4583], device='cuda:6'), covar=tensor([0.2928, 0.0598, 0.1845, 0.2850, 0.2614, 0.2119, 0.0403, 0.1268], device='cuda:6'), in_proj_covar=tensor([0.0331, 0.0270, 0.0308, 0.0317, 0.0299, 0.0263, 0.0299, 0.0340], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 16:43:30,324 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.091e+02 2.907e+02 3.422e+02 4.020e+02 9.124e+02, threshold=6.845e+02, percent-clipped=1.0 2023-05-01 16:43:41,685 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=231124.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 16:44:07,789 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9527, 3.4909, 3.0066, 5.3707, 4.1572, 4.5173, 2.0979, 3.3838], device='cuda:6'), covar=tensor([0.1334, 0.0617, 0.1075, 0.0181, 0.0357, 0.0384, 0.1489, 0.0776], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0176, 0.0197, 0.0193, 0.0208, 0.0216, 0.0205, 0.0195], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 16:44:17,268 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=231148.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:44:20,806 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7721, 4.0722, 2.9839, 2.3693, 2.7413, 2.6157, 4.5195, 3.4549], device='cuda:6'), covar=tensor([0.2889, 0.0616, 0.1903, 0.2844, 0.2710, 0.1970, 0.0375, 0.1372], device='cuda:6'), in_proj_covar=tensor([0.0332, 0.0270, 0.0308, 0.0318, 0.0300, 0.0263, 0.0300, 0.0341], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 16:44:24,895 INFO [train.py:904] (6/8) Epoch 23, batch 7850, loss[loss=0.178, simple_loss=0.2775, pruned_loss=0.0393, over 16873.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2905, pruned_loss=0.05944, over 3076722.92 frames. ], batch size: 96, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:44:37,493 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2023-05-01 16:44:52,238 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=231172.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:45:05,904 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=231181.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:45:16,796 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=231189.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:45:31,353 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-05-01 16:45:37,199 INFO [train.py:904] (6/8) Epoch 23, batch 7900, loss[loss=0.2134, simple_loss=0.2987, pruned_loss=0.06408, over 16587.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2897, pruned_loss=0.05936, over 3055621.79 frames. ], batch size: 57, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:45:43,924 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0719, 3.3355, 3.3677, 2.1478, 3.1503, 3.3868, 3.1646, 1.9012], device='cuda:6'), covar=tensor([0.0542, 0.0074, 0.0067, 0.0458, 0.0109, 0.0124, 0.0116, 0.0517], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0085, 0.0086, 0.0134, 0.0098, 0.0111, 0.0095, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 16:45:55,939 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-01 16:45:57,446 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.744e+02 2.682e+02 3.251e+02 4.238e+02 6.670e+02, threshold=6.501e+02, percent-clipped=0.0 2023-05-01 16:46:03,511 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0654, 4.1208, 4.4406, 4.3954, 4.4124, 4.1632, 4.1433, 4.1124], device='cuda:6'), covar=tensor([0.0357, 0.0562, 0.0387, 0.0402, 0.0463, 0.0403, 0.0901, 0.0508], device='cuda:6'), in_proj_covar=tensor([0.0412, 0.0458, 0.0444, 0.0413, 0.0490, 0.0466, 0.0551, 0.0372], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 16:46:16,886 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=231229.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:46:30,781 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 16:46:31,360 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=231237.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:46:44,780 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=231246.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:46:52,756 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=231251.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:46:54,844 INFO [train.py:904] (6/8) Epoch 23, batch 7950, loss[loss=0.2616, simple_loss=0.3144, pruned_loss=0.1044, over 11752.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2903, pruned_loss=0.05993, over 3059606.50 frames. ], batch size: 246, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:48:11,518 INFO [train.py:904] (6/8) Epoch 23, batch 8000, loss[loss=0.2366, simple_loss=0.3039, pruned_loss=0.0847, over 10895.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2913, pruned_loss=0.06071, over 3049875.94 frames. ], batch size: 247, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:48:26,478 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=231312.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:48:33,449 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.913e+02 2.670e+02 3.387e+02 4.030e+02 6.156e+02, threshold=6.774e+02, percent-clipped=0.0 2023-05-01 16:49:26,910 INFO [train.py:904] (6/8) Epoch 23, batch 8050, loss[loss=0.2041, simple_loss=0.2998, pruned_loss=0.05422, over 16708.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2909, pruned_loss=0.05995, over 3070617.00 frames. ], batch size: 124, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:50:04,406 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 16:50:31,860 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8164, 3.9305, 2.4567, 4.5353, 3.1261, 4.4512, 2.5863, 3.1497], device='cuda:6'), covar=tensor([0.0285, 0.0352, 0.1635, 0.0199, 0.0769, 0.0464, 0.1534, 0.0777], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0174, 0.0193, 0.0163, 0.0175, 0.0216, 0.0201, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 16:50:42,759 INFO [train.py:904] (6/8) Epoch 23, batch 8100, loss[loss=0.2378, simple_loss=0.3103, pruned_loss=0.0827, over 15317.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2902, pruned_loss=0.05937, over 3080512.93 frames. ], batch size: 191, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:51:04,319 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.823e+02 2.502e+02 3.046e+02 3.676e+02 7.002e+02, threshold=6.092e+02, percent-clipped=1.0 2023-05-01 16:51:51,832 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=231448.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:51:59,310 INFO [train.py:904] (6/8) Epoch 23, batch 8150, loss[loss=0.2282, simple_loss=0.2984, pruned_loss=0.07901, over 11581.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2878, pruned_loss=0.05827, over 3081675.13 frames. ], batch size: 248, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:52:20,933 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8301, 1.9205, 2.3866, 2.7277, 2.7282, 3.0917, 2.0016, 3.0168], device='cuda:6'), covar=tensor([0.0212, 0.0540, 0.0323, 0.0334, 0.0313, 0.0183, 0.0560, 0.0152], device='cuda:6'), in_proj_covar=tensor([0.0188, 0.0191, 0.0177, 0.0182, 0.0196, 0.0154, 0.0196, 0.0153], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:53:05,171 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=231496.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:53:14,921 INFO [train.py:904] (6/8) Epoch 23, batch 8200, loss[loss=0.1902, simple_loss=0.2893, pruned_loss=0.04556, over 16244.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2854, pruned_loss=0.05767, over 3078091.28 frames. ], batch size: 165, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:53:38,106 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.702e+02 2.780e+02 3.369e+02 4.151e+02 6.479e+02, threshold=6.737e+02, percent-clipped=3.0 2023-05-01 16:53:39,602 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.02 vs. limit=5.0 2023-05-01 16:53:39,794 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-05-01 16:54:26,248 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=231546.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:54:37,501 INFO [train.py:904] (6/8) Epoch 23, batch 8250, loss[loss=0.1961, simple_loss=0.2905, pruned_loss=0.05088, over 16260.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2844, pruned_loss=0.05536, over 3055106.63 frames. ], batch size: 165, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:55:05,033 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9042, 3.7166, 4.0358, 2.3036, 4.1963, 4.2765, 3.3006, 3.3971], device='cuda:6'), covar=tensor([0.0658, 0.0224, 0.0194, 0.1065, 0.0077, 0.0142, 0.0349, 0.0371], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0106, 0.0096, 0.0136, 0.0080, 0.0123, 0.0125, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 16:55:30,520 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5902, 2.9929, 3.2512, 1.9938, 2.8429, 2.1675, 3.1628, 3.2472], device='cuda:6'), covar=tensor([0.0352, 0.0897, 0.0570, 0.2207, 0.0872, 0.1087, 0.0712, 0.0867], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0164, 0.0167, 0.0153, 0.0145, 0.0129, 0.0142, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 16:55:43,800 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=231594.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:55:43,979 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1625, 2.4545, 2.5882, 2.0644, 2.6932, 2.7900, 2.5733, 2.5251], device='cuda:6'), covar=tensor([0.0594, 0.0214, 0.0220, 0.0874, 0.0118, 0.0281, 0.0384, 0.0388], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0106, 0.0096, 0.0136, 0.0080, 0.0123, 0.0124, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 16:55:57,538 INFO [train.py:904] (6/8) Epoch 23, batch 8300, loss[loss=0.156, simple_loss=0.2614, pruned_loss=0.02529, over 16854.00 frames. ], tot_loss[loss=0.1924, simple_loss=0.2812, pruned_loss=0.05182, over 3065795.75 frames. ], batch size: 102, lr: 2.91e-03, grad_scale: 4.0 2023-05-01 16:56:05,055 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=231607.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:56:21,698 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9438, 2.1408, 2.3872, 3.1473, 2.2122, 2.3417, 2.3338, 2.2299], device='cuda:6'), covar=tensor([0.1264, 0.3667, 0.2743, 0.0715, 0.4596, 0.2522, 0.3627, 0.3925], device='cuda:6'), in_proj_covar=tensor([0.0397, 0.0446, 0.0365, 0.0322, 0.0431, 0.0512, 0.0417, 0.0521], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:56:22,155 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.405e+02 2.236e+02 2.512e+02 2.973e+02 5.364e+02, threshold=5.024e+02, percent-clipped=0.0 2023-05-01 16:57:11,315 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=231647.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:57:19,287 INFO [train.py:904] (6/8) Epoch 23, batch 8350, loss[loss=0.2389, simple_loss=0.3067, pruned_loss=0.0855, over 11734.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2803, pruned_loss=0.05009, over 3050486.00 frames. ], batch size: 248, lr: 2.91e-03, grad_scale: 4.0 2023-05-01 16:57:56,683 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3245, 4.1488, 4.4072, 4.5244, 4.6934, 4.2125, 4.6461, 4.7140], device='cuda:6'), covar=tensor([0.2004, 0.1377, 0.1616, 0.0805, 0.0615, 0.1223, 0.0677, 0.0704], device='cuda:6'), in_proj_covar=tensor([0.0624, 0.0777, 0.0892, 0.0783, 0.0599, 0.0622, 0.0652, 0.0758], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 16:58:14,588 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-01 16:58:38,337 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-01 16:58:41,947 INFO [train.py:904] (6/8) Epoch 23, batch 8400, loss[loss=0.1639, simple_loss=0.2589, pruned_loss=0.03441, over 15363.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2777, pruned_loss=0.04821, over 3038267.34 frames. ], batch size: 190, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:58:50,828 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=231708.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 16:59:06,587 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 2.191e+02 2.741e+02 3.304e+02 6.516e+02, threshold=5.483e+02, percent-clipped=5.0 2023-05-01 16:59:20,999 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4876, 3.6654, 2.8188, 2.1085, 2.2079, 2.3378, 3.8181, 3.0911], device='cuda:6'), covar=tensor([0.3224, 0.0575, 0.1773, 0.3234, 0.3121, 0.2321, 0.0449, 0.1539], device='cuda:6'), in_proj_covar=tensor([0.0328, 0.0267, 0.0305, 0.0315, 0.0297, 0.0261, 0.0296, 0.0337], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 16:59:34,464 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3515, 4.3213, 4.1819, 3.4938, 4.2448, 1.7632, 4.0568, 3.9694], device='cuda:6'), covar=tensor([0.0114, 0.0133, 0.0211, 0.0343, 0.0126, 0.2949, 0.0150, 0.0245], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0159, 0.0200, 0.0177, 0.0176, 0.0208, 0.0188, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 17:00:04,853 INFO [train.py:904] (6/8) Epoch 23, batch 8450, loss[loss=0.1836, simple_loss=0.2833, pruned_loss=0.04197, over 16848.00 frames. ], tot_loss[loss=0.1841, simple_loss=0.2758, pruned_loss=0.04613, over 3055718.11 frames. ], batch size: 116, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:00:27,906 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-05-01 17:01:24,902 INFO [train.py:904] (6/8) Epoch 23, batch 8500, loss[loss=0.1541, simple_loss=0.2507, pruned_loss=0.02879, over 16236.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2721, pruned_loss=0.04402, over 3049402.96 frames. ], batch size: 165, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:01:48,516 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 2.174e+02 2.671e+02 3.276e+02 7.093e+02, threshold=5.341e+02, percent-clipped=2.0 2023-05-01 17:01:56,087 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-01 17:02:48,223 INFO [train.py:904] (6/8) Epoch 23, batch 8550, loss[loss=0.1996, simple_loss=0.2946, pruned_loss=0.05229, over 16771.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2697, pruned_loss=0.04279, over 3028206.24 frames. ], batch size: 124, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:03:04,688 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2256, 2.9896, 3.1298, 1.7977, 3.2549, 3.3285, 2.8336, 2.6845], device='cuda:6'), covar=tensor([0.0778, 0.0263, 0.0218, 0.1229, 0.0107, 0.0223, 0.0409, 0.0462], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0105, 0.0095, 0.0135, 0.0079, 0.0121, 0.0123, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 17:03:16,444 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-01 17:03:42,302 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7008, 2.0885, 1.5890, 1.7150, 2.3559, 2.0292, 2.2399, 2.5772], device='cuda:6'), covar=tensor([0.0276, 0.0598, 0.0793, 0.0749, 0.0379, 0.0589, 0.0243, 0.0383], device='cuda:6'), in_proj_covar=tensor([0.0207, 0.0230, 0.0221, 0.0223, 0.0231, 0.0229, 0.0229, 0.0226], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 17:04:28,270 INFO [train.py:904] (6/8) Epoch 23, batch 8600, loss[loss=0.1695, simple_loss=0.2576, pruned_loss=0.04069, over 12203.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2701, pruned_loss=0.04237, over 3025083.70 frames. ], batch size: 247, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:04:36,581 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=231907.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:04:56,343 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4495, 3.3918, 3.4918, 3.5575, 3.6058, 3.3373, 3.5693, 3.6585], device='cuda:6'), covar=tensor([0.1218, 0.0963, 0.1072, 0.0683, 0.0651, 0.2282, 0.0895, 0.0732], device='cuda:6'), in_proj_covar=tensor([0.0620, 0.0770, 0.0885, 0.0777, 0.0593, 0.0618, 0.0646, 0.0753], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 17:05:00,378 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 2.309e+02 2.625e+02 3.302e+02 5.565e+02, threshold=5.250e+02, percent-clipped=1.0 2023-05-01 17:05:19,028 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.07 vs. limit=5.0 2023-05-01 17:06:03,606 INFO [train.py:904] (6/8) Epoch 23, batch 8650, loss[loss=0.1642, simple_loss=0.2663, pruned_loss=0.03103, over 16693.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2682, pruned_loss=0.04097, over 3027416.33 frames. ], batch size: 89, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:06:07,765 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6618, 4.7533, 4.8915, 4.7088, 4.8192, 5.2778, 4.7811, 4.4739], device='cuda:6'), covar=tensor([0.1127, 0.1876, 0.2171, 0.1756, 0.2318, 0.0865, 0.1493, 0.2444], device='cuda:6'), in_proj_covar=tensor([0.0400, 0.0581, 0.0645, 0.0480, 0.0638, 0.0671, 0.0503, 0.0642], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 17:06:10,285 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=231955.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:07:53,209 INFO [train.py:904] (6/8) Epoch 23, batch 8700, loss[loss=0.1612, simple_loss=0.2587, pruned_loss=0.03181, over 15600.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2657, pruned_loss=0.0395, over 3046609.13 frames. ], batch size: 191, lr: 2.91e-03, grad_scale: 4.0 2023-05-01 17:07:54,842 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=232003.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 17:08:23,060 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.274e+02 2.142e+02 2.520e+02 3.226e+02 6.058e+02, threshold=5.039e+02, percent-clipped=2.0 2023-05-01 17:08:36,815 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.65 vs. limit=5.0 2023-05-01 17:09:05,628 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-01 17:09:27,783 INFO [train.py:904] (6/8) Epoch 23, batch 8750, loss[loss=0.1688, simple_loss=0.2693, pruned_loss=0.03412, over 15204.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2649, pruned_loss=0.03874, over 3041606.91 frames. ], batch size: 190, lr: 2.91e-03, grad_scale: 4.0 2023-05-01 17:10:56,261 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.47 vs. limit=2.0 2023-05-01 17:11:15,140 INFO [train.py:904] (6/8) Epoch 23, batch 8800, loss[loss=0.1838, simple_loss=0.278, pruned_loss=0.04483, over 16714.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2639, pruned_loss=0.03777, over 3058604.44 frames. ], batch size: 134, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:11:33,039 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4608, 4.4329, 4.2886, 3.6364, 4.3434, 1.7353, 4.1477, 4.0521], device='cuda:6'), covar=tensor([0.0094, 0.0095, 0.0183, 0.0313, 0.0118, 0.2743, 0.0129, 0.0242], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0158, 0.0198, 0.0175, 0.0174, 0.0206, 0.0186, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 17:11:46,651 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.145e+02 2.530e+02 3.020e+02 7.216e+02, threshold=5.060e+02, percent-clipped=1.0 2023-05-01 17:12:19,553 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-01 17:12:57,922 INFO [train.py:904] (6/8) Epoch 23, batch 8850, loss[loss=0.1884, simple_loss=0.2997, pruned_loss=0.03855, over 16345.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2667, pruned_loss=0.03749, over 3063184.80 frames. ], batch size: 165, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:14:17,186 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.05 vs. limit=5.0 2023-05-01 17:14:42,072 INFO [train.py:904] (6/8) Epoch 23, batch 8900, loss[loss=0.1664, simple_loss=0.2634, pruned_loss=0.03469, over 16354.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2672, pruned_loss=0.0368, over 3069176.18 frames. ], batch size: 146, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:15:12,042 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 2.232e+02 2.655e+02 3.319e+02 5.405e+02, threshold=5.309e+02, percent-clipped=1.0 2023-05-01 17:16:31,032 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7814, 3.5240, 3.8251, 1.9104, 4.0133, 4.0641, 3.2113, 3.2011], device='cuda:6'), covar=tensor([0.0614, 0.0223, 0.0212, 0.1161, 0.0064, 0.0109, 0.0311, 0.0373], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0103, 0.0093, 0.0133, 0.0077, 0.0119, 0.0121, 0.0124], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 17:16:33,772 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=232247.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:16:45,324 INFO [train.py:904] (6/8) Epoch 23, batch 8950, loss[loss=0.1937, simple_loss=0.2711, pruned_loss=0.05817, over 12501.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2665, pruned_loss=0.03702, over 3070540.07 frames. ], batch size: 250, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:17:43,179 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-01 17:18:31,849 INFO [train.py:904] (6/8) Epoch 23, batch 9000, loss[loss=0.1541, simple_loss=0.2484, pruned_loss=0.02996, over 16321.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2633, pruned_loss=0.03587, over 3070984.62 frames. ], batch size: 166, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:18:31,849 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 17:18:42,673 INFO [train.py:938] (6/8) Epoch 23, validation: loss=0.1452, simple_loss=0.249, pruned_loss=0.02066, over 944034.00 frames. 2023-05-01 17:18:42,673 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 17:18:43,575 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=232303.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 17:18:54,234 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=232308.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:19:18,068 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 1.963e+02 2.332e+02 2.758e+02 5.487e+02, threshold=4.665e+02, percent-clipped=1.0 2023-05-01 17:19:54,487 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8594, 2.6487, 2.8680, 2.0938, 2.6759, 2.1554, 2.6909, 2.8591], device='cuda:6'), covar=tensor([0.0299, 0.0977, 0.0555, 0.1897, 0.0890, 0.1016, 0.0660, 0.0920], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0160, 0.0164, 0.0151, 0.0142, 0.0127, 0.0140, 0.0171], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 17:20:23,201 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-05-01 17:20:24,510 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=232351.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:20:28,118 INFO [train.py:904] (6/8) Epoch 23, batch 9050, loss[loss=0.1656, simple_loss=0.2509, pruned_loss=0.04019, over 16692.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2635, pruned_loss=0.03622, over 3067555.50 frames. ], batch size: 134, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:20:52,979 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3128, 5.6742, 5.2160, 5.5979, 5.2351, 4.9078, 5.1985, 5.7691], device='cuda:6'), covar=tensor([0.1874, 0.1333, 0.2407, 0.1107, 0.1347, 0.1285, 0.1982, 0.1640], device='cuda:6'), in_proj_covar=tensor([0.0668, 0.0813, 0.0671, 0.0621, 0.0516, 0.0529, 0.0683, 0.0640], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 17:21:21,002 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-05-01 17:21:23,506 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=232381.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:22:12,827 INFO [train.py:904] (6/8) Epoch 23, batch 9100, loss[loss=0.1702, simple_loss=0.257, pruned_loss=0.04173, over 12049.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2624, pruned_loss=0.0366, over 3056374.53 frames. ], batch size: 246, lr: 2.91e-03, grad_scale: 4.0 2023-05-01 17:22:14,043 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7655, 2.2048, 2.2514, 3.4882, 1.7626, 3.5667, 1.4583, 2.8416], device='cuda:6'), covar=tensor([0.1471, 0.0965, 0.1329, 0.0183, 0.0116, 0.0414, 0.1866, 0.0772], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0173, 0.0194, 0.0188, 0.0201, 0.0212, 0.0202, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 17:22:46,301 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 2.157e+02 2.616e+02 3.171e+02 8.876e+02, threshold=5.231e+02, percent-clipped=3.0 2023-05-01 17:23:46,441 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=232442.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 17:24:09,372 INFO [train.py:904] (6/8) Epoch 23, batch 9150, loss[loss=0.1622, simple_loss=0.2616, pruned_loss=0.03137, over 17230.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2628, pruned_loss=0.03646, over 3057721.59 frames. ], batch size: 44, lr: 2.91e-03, grad_scale: 4.0 2023-05-01 17:25:00,522 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=232476.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:25:45,337 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3249, 2.1896, 2.2471, 3.9067, 2.2245, 2.5083, 2.3780, 2.2989], device='cuda:6'), covar=tensor([0.1253, 0.3704, 0.3167, 0.0539, 0.4347, 0.2814, 0.3467, 0.3894], device='cuda:6'), in_proj_covar=tensor([0.0393, 0.0443, 0.0364, 0.0318, 0.0428, 0.0506, 0.0414, 0.0514], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 17:25:52,923 INFO [train.py:904] (6/8) Epoch 23, batch 9200, loss[loss=0.1525, simple_loss=0.2449, pruned_loss=0.03003, over 16646.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.259, pruned_loss=0.03585, over 3037337.00 frames. ], batch size: 89, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:26:19,924 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9410, 5.2465, 5.0697, 5.0627, 4.7712, 4.7540, 4.6225, 5.3505], device='cuda:6'), covar=tensor([0.1196, 0.0889, 0.0870, 0.0758, 0.0799, 0.0901, 0.1149, 0.0909], device='cuda:6'), in_proj_covar=tensor([0.0664, 0.0806, 0.0666, 0.0617, 0.0512, 0.0526, 0.0678, 0.0636], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 17:26:21,311 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=232518.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:26:24,112 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.032e+02 2.532e+02 2.962e+02 8.451e+02, threshold=5.064e+02, percent-clipped=2.0 2023-05-01 17:26:57,068 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=232537.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:27:09,436 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 17:27:21,303 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1315, 2.5155, 2.5810, 1.9143, 2.6786, 2.8127, 2.5095, 2.4994], device='cuda:6'), covar=tensor([0.0609, 0.0235, 0.0246, 0.1016, 0.0127, 0.0257, 0.0398, 0.0407], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0104, 0.0094, 0.0134, 0.0078, 0.0120, 0.0122, 0.0125], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 17:27:27,733 INFO [train.py:904] (6/8) Epoch 23, batch 9250, loss[loss=0.173, simple_loss=0.2503, pruned_loss=0.04783, over 12294.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2588, pruned_loss=0.03588, over 3035511.05 frames. ], batch size: 248, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:27:29,775 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2023-05-01 17:28:23,250 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=232579.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:29:01,413 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6160, 2.4336, 2.3551, 3.8202, 2.0478, 3.8218, 1.4396, 2.8012], device='cuda:6'), covar=tensor([0.1559, 0.0955, 0.1279, 0.0198, 0.0119, 0.0423, 0.1923, 0.0764], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0173, 0.0194, 0.0186, 0.0200, 0.0211, 0.0201, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 17:29:17,461 INFO [train.py:904] (6/8) Epoch 23, batch 9300, loss[loss=0.1486, simple_loss=0.2427, pruned_loss=0.02727, over 16423.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2573, pruned_loss=0.0353, over 3019132.54 frames. ], batch size: 68, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:29:18,461 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=232603.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:29:31,613 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6718, 2.6782, 1.8552, 2.8323, 2.1075, 2.8137, 2.1532, 2.4311], device='cuda:6'), covar=tensor([0.0346, 0.0385, 0.1397, 0.0285, 0.0651, 0.0526, 0.1278, 0.0617], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0171, 0.0188, 0.0157, 0.0171, 0.0207, 0.0198, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 17:29:58,636 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.317e+02 2.074e+02 2.482e+02 2.986e+02 5.209e+02, threshold=4.963e+02, percent-clipped=1.0 2023-05-01 17:30:02,468 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7868, 2.3516, 2.2134, 3.5331, 1.9839, 3.6347, 1.5051, 2.9042], device='cuda:6'), covar=tensor([0.1380, 0.0873, 0.1313, 0.0178, 0.0110, 0.0385, 0.1772, 0.0709], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0173, 0.0194, 0.0186, 0.0199, 0.0211, 0.0201, 0.0191], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 17:30:41,613 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-01 17:31:04,804 INFO [train.py:904] (6/8) Epoch 23, batch 9350, loss[loss=0.1726, simple_loss=0.2534, pruned_loss=0.04585, over 12425.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2576, pruned_loss=0.03532, over 3029592.49 frames. ], batch size: 250, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:31:59,682 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0189, 2.6985, 2.9002, 2.1556, 2.7162, 2.1704, 2.7588, 2.9352], device='cuda:6'), covar=tensor([0.0334, 0.0970, 0.0498, 0.1903, 0.0880, 0.1020, 0.0668, 0.0842], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0158, 0.0163, 0.0150, 0.0142, 0.0126, 0.0139, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 17:32:47,938 INFO [train.py:904] (6/8) Epoch 23, batch 9400, loss[loss=0.1392, simple_loss=0.2287, pruned_loss=0.02489, over 12485.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2578, pruned_loss=0.03518, over 3026623.42 frames. ], batch size: 248, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:33:21,841 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.048e+02 2.308e+02 2.875e+02 5.038e+02, threshold=4.615e+02, percent-clipped=1.0 2023-05-01 17:33:57,944 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=232737.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 17:34:29,673 INFO [train.py:904] (6/8) Epoch 23, batch 9450, loss[loss=0.1661, simple_loss=0.2624, pruned_loss=0.03487, over 16651.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2599, pruned_loss=0.03568, over 3030661.99 frames. ], batch size: 134, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:36:00,004 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4877, 3.0745, 2.7235, 2.2547, 2.1834, 2.2852, 3.0938, 2.8261], device='cuda:6'), covar=tensor([0.2755, 0.0623, 0.1556, 0.2840, 0.2861, 0.2404, 0.0453, 0.1507], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0262, 0.0299, 0.0309, 0.0286, 0.0257, 0.0289, 0.0329], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 17:36:04,920 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1837, 5.2303, 5.0511, 4.5616, 4.6861, 5.0986, 5.0541, 4.7138], device='cuda:6'), covar=tensor([0.0536, 0.0458, 0.0292, 0.0336, 0.1010, 0.0489, 0.0277, 0.0703], device='cuda:6'), in_proj_covar=tensor([0.0284, 0.0418, 0.0330, 0.0327, 0.0332, 0.0383, 0.0226, 0.0394], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-05-01 17:36:09,232 INFO [train.py:904] (6/8) Epoch 23, batch 9500, loss[loss=0.1702, simple_loss=0.2493, pruned_loss=0.0455, over 12797.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2595, pruned_loss=0.03536, over 3042105.74 frames. ], batch size: 248, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:36:44,544 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.549e+02 2.194e+02 2.525e+02 3.286e+02 6.197e+02, threshold=5.051e+02, percent-clipped=6.0 2023-05-01 17:37:08,362 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=232832.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:37:54,041 INFO [train.py:904] (6/8) Epoch 23, batch 9550, loss[loss=0.1853, simple_loss=0.2745, pruned_loss=0.04809, over 12679.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2592, pruned_loss=0.03551, over 3044640.46 frames. ], batch size: 246, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:38:38,052 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=232874.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:38:38,277 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0037, 2.7125, 2.8761, 2.0912, 2.7074, 2.1485, 2.6930, 2.8816], device='cuda:6'), covar=tensor([0.0308, 0.0877, 0.0544, 0.1905, 0.0813, 0.0993, 0.0591, 0.0844], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0157, 0.0163, 0.0150, 0.0141, 0.0126, 0.0139, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 17:39:33,897 INFO [train.py:904] (6/8) Epoch 23, batch 9600, loss[loss=0.1772, simple_loss=0.2833, pruned_loss=0.03549, over 16129.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2609, pruned_loss=0.03604, over 3062324.77 frames. ], batch size: 165, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:39:34,702 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=232903.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:40:05,774 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.131e+02 2.488e+02 2.932e+02 6.217e+02, threshold=4.975e+02, percent-clipped=3.0 2023-05-01 17:41:17,234 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=232951.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:41:20,710 INFO [train.py:904] (6/8) Epoch 23, batch 9650, loss[loss=0.1478, simple_loss=0.2435, pruned_loss=0.02608, over 16687.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2626, pruned_loss=0.03646, over 3056939.77 frames. ], batch size: 57, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:41:22,223 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7894, 3.8436, 4.1333, 4.1208, 4.1054, 3.9100, 3.9015, 3.9390], device='cuda:6'), covar=tensor([0.0409, 0.0709, 0.0507, 0.0429, 0.0574, 0.0497, 0.0849, 0.0484], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0440, 0.0432, 0.0396, 0.0473, 0.0448, 0.0528, 0.0361], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 17:42:42,063 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=232990.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:43:00,242 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 17:43:09,179 INFO [train.py:904] (6/8) Epoch 23, batch 9700, loss[loss=0.1747, simple_loss=0.2702, pruned_loss=0.0396, over 16213.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2618, pruned_loss=0.03643, over 3055866.01 frames. ], batch size: 165, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:43:25,773 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9282, 2.0584, 2.4284, 2.8695, 2.7250, 3.2316, 2.1864, 3.2087], device='cuda:6'), covar=tensor([0.0223, 0.0549, 0.0360, 0.0291, 0.0342, 0.0196, 0.0513, 0.0147], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0188, 0.0174, 0.0177, 0.0193, 0.0150, 0.0192, 0.0147], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 17:43:40,115 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=233018.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:43:43,780 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.165e+02 2.462e+02 3.116e+02 5.593e+02, threshold=4.924e+02, percent-clipped=3.0 2023-05-01 17:44:22,585 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=233037.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:44:50,787 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=233051.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 17:44:53,543 INFO [train.py:904] (6/8) Epoch 23, batch 9750, loss[loss=0.1704, simple_loss=0.2673, pruned_loss=0.03678, over 15498.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2612, pruned_loss=0.03673, over 3059871.18 frames. ], batch size: 191, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:45:31,898 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-01 17:45:43,929 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=233079.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:45:59,986 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=233085.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:46:31,579 INFO [train.py:904] (6/8) Epoch 23, batch 9800, loss[loss=0.1435, simple_loss=0.2322, pruned_loss=0.02743, over 11937.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2605, pruned_loss=0.03566, over 3038193.93 frames. ], batch size: 247, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:47:03,691 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.589e+02 2.119e+02 2.583e+02 3.356e+02 7.260e+02, threshold=5.167e+02, percent-clipped=1.0 2023-05-01 17:47:08,340 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0577, 4.0501, 3.9140, 3.2907, 4.0071, 1.8166, 3.8178, 3.5381], device='cuda:6'), covar=tensor([0.0092, 0.0095, 0.0185, 0.0252, 0.0091, 0.2717, 0.0118, 0.0280], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0157, 0.0196, 0.0171, 0.0173, 0.0206, 0.0185, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 17:47:27,284 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=233132.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:48:16,814 INFO [train.py:904] (6/8) Epoch 23, batch 9850, loss[loss=0.1679, simple_loss=0.2679, pruned_loss=0.03394, over 16802.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2613, pruned_loss=0.0351, over 3060078.98 frames. ], batch size: 124, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:48:20,837 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8580, 3.8602, 4.0017, 3.8488, 3.9392, 4.3136, 3.9554, 3.6826], device='cuda:6'), covar=tensor([0.1969, 0.2341, 0.2241, 0.2382, 0.2786, 0.1534, 0.1602, 0.2592], device='cuda:6'), in_proj_covar=tensor([0.0388, 0.0570, 0.0631, 0.0471, 0.0628, 0.0657, 0.0491, 0.0627], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 17:48:29,398 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-05-01 17:48:32,430 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9363, 3.0069, 2.7298, 4.8056, 3.4934, 4.3065, 1.7220, 3.2270], device='cuda:6'), covar=tensor([0.1314, 0.0703, 0.1082, 0.0149, 0.0133, 0.0292, 0.1515, 0.0633], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0172, 0.0193, 0.0185, 0.0197, 0.0210, 0.0201, 0.0190], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 17:49:01,349 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=233174.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:49:13,852 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=233180.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:50:08,888 INFO [train.py:904] (6/8) Epoch 23, batch 9900, loss[loss=0.1535, simple_loss=0.2585, pruned_loss=0.02426, over 16763.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2614, pruned_loss=0.03522, over 3051660.21 frames. ], batch size: 83, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:50:34,300 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5063, 2.0527, 1.7977, 1.7126, 2.3144, 1.9197, 1.9453, 2.3234], device='cuda:6'), covar=tensor([0.0192, 0.0472, 0.0556, 0.0482, 0.0297, 0.0408, 0.0220, 0.0266], device='cuda:6'), in_proj_covar=tensor([0.0204, 0.0229, 0.0220, 0.0222, 0.0229, 0.0229, 0.0223, 0.0222], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 17:50:46,347 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.034e+02 2.487e+02 3.058e+02 5.815e+02, threshold=4.974e+02, percent-clipped=3.0 2023-05-01 17:50:52,784 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=233222.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:52:06,031 INFO [train.py:904] (6/8) Epoch 23, batch 9950, loss[loss=0.178, simple_loss=0.2746, pruned_loss=0.04067, over 16932.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2635, pruned_loss=0.03555, over 3055018.20 frames. ], batch size: 125, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:52:20,939 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=233259.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:54:07,452 INFO [train.py:904] (6/8) Epoch 23, batch 10000, loss[loss=0.1477, simple_loss=0.253, pruned_loss=0.02121, over 16874.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2626, pruned_loss=0.03506, over 3076175.78 frames. ], batch size: 96, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:54:40,344 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 2.128e+02 2.777e+02 3.344e+02 7.240e+02, threshold=5.555e+02, percent-clipped=2.0 2023-05-01 17:54:41,091 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=233320.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:55:36,787 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=233346.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 17:55:41,084 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0434, 3.0400, 1.9275, 3.3215, 2.2992, 3.2967, 2.0664, 2.5715], device='cuda:6'), covar=tensor([0.0333, 0.0442, 0.1662, 0.0273, 0.0903, 0.0624, 0.1649, 0.0798], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0168, 0.0186, 0.0155, 0.0169, 0.0205, 0.0197, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-01 17:55:47,972 INFO [train.py:904] (6/8) Epoch 23, batch 10050, loss[loss=0.182, simple_loss=0.2819, pruned_loss=0.04109, over 16743.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2629, pruned_loss=0.03477, over 3079107.45 frames. ], batch size: 134, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:56:31,056 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=233374.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:56:37,758 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8775, 2.2146, 1.8976, 1.9528, 2.5000, 2.1833, 2.2813, 2.6658], device='cuda:6'), covar=tensor([0.0197, 0.0524, 0.0602, 0.0548, 0.0327, 0.0465, 0.0262, 0.0313], device='cuda:6'), in_proj_covar=tensor([0.0204, 0.0229, 0.0221, 0.0223, 0.0229, 0.0230, 0.0224, 0.0223], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 17:57:20,830 INFO [train.py:904] (6/8) Epoch 23, batch 10100, loss[loss=0.1645, simple_loss=0.2511, pruned_loss=0.03895, over 16394.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2629, pruned_loss=0.0348, over 3094415.06 frames. ], batch size: 146, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:57:53,983 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.206e+02 2.577e+02 3.013e+02 4.747e+02, threshold=5.154e+02, percent-clipped=0.0 2023-05-01 17:59:06,334 INFO [train.py:904] (6/8) Epoch 24, batch 0, loss[loss=0.2215, simple_loss=0.3081, pruned_loss=0.06749, over 12231.00 frames. ], tot_loss[loss=0.2215, simple_loss=0.3081, pruned_loss=0.06749, over 12231.00 frames. ], batch size: 247, lr: 2.84e-03, grad_scale: 8.0 2023-05-01 17:59:06,334 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 17:59:14,244 INFO [train.py:938] (6/8) Epoch 24, validation: loss=0.145, simple_loss=0.2483, pruned_loss=0.02085, over 944034.00 frames. 2023-05-01 17:59:14,245 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 18:00:06,211 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9155, 2.4711, 2.1027, 2.2402, 2.8655, 2.5736, 2.8593, 2.9695], device='cuda:6'), covar=tensor([0.0213, 0.0459, 0.0533, 0.0461, 0.0247, 0.0340, 0.0206, 0.0284], device='cuda:6'), in_proj_covar=tensor([0.0207, 0.0231, 0.0222, 0.0224, 0.0230, 0.0231, 0.0225, 0.0224], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:00:15,701 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0102, 4.4454, 4.5308, 3.2853, 3.7379, 4.4418, 3.9796, 2.6145], device='cuda:6'), covar=tensor([0.0438, 0.0067, 0.0041, 0.0327, 0.0162, 0.0092, 0.0091, 0.0435], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0084, 0.0085, 0.0134, 0.0098, 0.0109, 0.0094, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 18:00:23,694 INFO [train.py:904] (6/8) Epoch 24, batch 50, loss[loss=0.1807, simple_loss=0.2793, pruned_loss=0.0411, over 17139.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2679, pruned_loss=0.0456, over 756179.11 frames. ], batch size: 48, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:00:29,643 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6096, 3.7375, 2.3015, 4.2358, 2.9438, 4.1945, 2.4305, 3.1000], device='cuda:6'), covar=tensor([0.0342, 0.0426, 0.1648, 0.0440, 0.0829, 0.0620, 0.1600, 0.0774], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0170, 0.0188, 0.0158, 0.0171, 0.0208, 0.0198, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 18:00:52,608 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.624e+02 2.410e+02 2.907e+02 3.501e+02 5.437e+02, threshold=5.814e+02, percent-clipped=4.0 2023-05-01 18:01:32,986 INFO [train.py:904] (6/8) Epoch 24, batch 100, loss[loss=0.1935, simple_loss=0.2815, pruned_loss=0.05271, over 16645.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2634, pruned_loss=0.04423, over 1331915.99 frames. ], batch size: 62, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:01:53,787 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6594, 3.8319, 2.3935, 4.4738, 3.0502, 4.3534, 2.4663, 3.1143], device='cuda:6'), covar=tensor([0.0391, 0.0450, 0.1702, 0.0360, 0.0866, 0.0561, 0.1673, 0.0848], device='cuda:6'), in_proj_covar=tensor([0.0166, 0.0171, 0.0189, 0.0159, 0.0172, 0.0210, 0.0199, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 18:02:40,658 INFO [train.py:904] (6/8) Epoch 24, batch 150, loss[loss=0.1697, simple_loss=0.2505, pruned_loss=0.04447, over 16865.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2627, pruned_loss=0.04337, over 1767154.86 frames. ], batch size: 96, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:02:56,760 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=233615.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:03:08,348 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.737e+02 2.160e+02 2.597e+02 3.268e+02 6.927e+02, threshold=5.194e+02, percent-clipped=3.0 2023-05-01 18:03:30,588 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=233639.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:03:39,455 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=233646.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 18:03:48,629 INFO [train.py:904] (6/8) Epoch 24, batch 200, loss[loss=0.1957, simple_loss=0.2729, pruned_loss=0.05928, over 16860.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2629, pruned_loss=0.0438, over 2103025.60 frames. ], batch size: 116, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:04:18,032 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=233674.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:04:45,846 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=233694.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:04:54,320 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=233700.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 18:04:58,753 INFO [train.py:904] (6/8) Epoch 24, batch 250, loss[loss=0.1669, simple_loss=0.2426, pruned_loss=0.04559, over 16916.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.26, pruned_loss=0.04382, over 2355292.80 frames. ], batch size: 96, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:05:07,751 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=233710.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:05:25,966 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=233722.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:05:26,869 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.709e+02 2.284e+02 2.752e+02 3.292e+02 5.130e+02, threshold=5.503e+02, percent-clipped=0.0 2023-05-01 18:06:08,092 INFO [train.py:904] (6/8) Epoch 24, batch 300, loss[loss=0.1609, simple_loss=0.2575, pruned_loss=0.03216, over 16726.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.258, pruned_loss=0.04287, over 2566411.50 frames. ], batch size: 62, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:06:33,462 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=233771.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:06:45,297 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2631, 2.2985, 2.3942, 4.0111, 2.2790, 2.6221, 2.3851, 2.4276], device='cuda:6'), covar=tensor([0.1461, 0.3617, 0.3171, 0.0679, 0.4088, 0.2693, 0.3865, 0.3320], device='cuda:6'), in_proj_covar=tensor([0.0400, 0.0450, 0.0372, 0.0324, 0.0435, 0.0513, 0.0422, 0.0524], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:07:04,785 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6984, 4.7655, 4.5961, 4.2136, 3.9446, 4.7627, 4.6189, 4.2853], device='cuda:6'), covar=tensor([0.0903, 0.0945, 0.0521, 0.0517, 0.1650, 0.0521, 0.0510, 0.0899], device='cuda:6'), in_proj_covar=tensor([0.0290, 0.0429, 0.0339, 0.0337, 0.0341, 0.0392, 0.0231, 0.0406], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:07:16,410 INFO [train.py:904] (6/8) Epoch 24, batch 350, loss[loss=0.1581, simple_loss=0.2558, pruned_loss=0.03024, over 17117.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2553, pruned_loss=0.04121, over 2733890.88 frames. ], batch size: 47, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:07:43,747 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.126e+02 2.409e+02 2.975e+02 6.590e+02, threshold=4.818e+02, percent-clipped=2.0 2023-05-01 18:08:25,442 INFO [train.py:904] (6/8) Epoch 24, batch 400, loss[loss=0.1934, simple_loss=0.2688, pruned_loss=0.05904, over 16890.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2537, pruned_loss=0.04016, over 2865593.70 frames. ], batch size: 90, lr: 2.84e-03, grad_scale: 2.0 2023-05-01 18:09:08,242 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5788, 5.9595, 5.6908, 5.7339, 5.3470, 5.4499, 5.3399, 6.0587], device='cuda:6'), covar=tensor([0.1385, 0.1002, 0.1096, 0.0873, 0.0938, 0.0689, 0.1333, 0.0991], device='cuda:6'), in_proj_covar=tensor([0.0681, 0.0825, 0.0680, 0.0633, 0.0524, 0.0532, 0.0698, 0.0650], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:09:32,974 INFO [train.py:904] (6/8) Epoch 24, batch 450, loss[loss=0.1617, simple_loss=0.2411, pruned_loss=0.04116, over 16879.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2519, pruned_loss=0.04003, over 2972665.80 frames. ], batch size: 90, lr: 2.84e-03, grad_scale: 2.0 2023-05-01 18:09:47,258 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1447, 3.7563, 4.2932, 2.2793, 4.4303, 4.5358, 3.3233, 3.4465], device='cuda:6'), covar=tensor([0.0680, 0.0291, 0.0233, 0.1190, 0.0095, 0.0191, 0.0436, 0.0418], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0107, 0.0097, 0.0139, 0.0081, 0.0125, 0.0126, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 18:09:50,203 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=233915.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:09:59,638 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.087e+02 2.389e+02 2.800e+02 4.652e+02, threshold=4.779e+02, percent-clipped=0.0 2023-05-01 18:10:38,995 INFO [train.py:904] (6/8) Epoch 24, batch 500, loss[loss=0.1544, simple_loss=0.2431, pruned_loss=0.03286, over 16872.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2507, pruned_loss=0.03958, over 3050838.09 frames. ], batch size: 96, lr: 2.83e-03, grad_scale: 2.0 2023-05-01 18:10:48,727 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-05-01 18:10:53,522 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=233963.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:11:20,666 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 18:11:22,705 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3007, 5.8737, 5.9588, 5.7170, 5.7485, 6.2928, 5.8369, 5.5440], device='cuda:6'), covar=tensor([0.0882, 0.1806, 0.2409, 0.1912, 0.2664, 0.0931, 0.1580, 0.2489], device='cuda:6'), in_proj_covar=tensor([0.0408, 0.0599, 0.0662, 0.0496, 0.0662, 0.0691, 0.0515, 0.0660], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 18:11:37,685 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=233995.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 18:11:50,847 INFO [train.py:904] (6/8) Epoch 24, batch 550, loss[loss=0.1683, simple_loss=0.2621, pruned_loss=0.03726, over 17080.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2499, pruned_loss=0.03895, over 3114009.92 frames. ], batch size: 55, lr: 2.83e-03, grad_scale: 2.0 2023-05-01 18:12:17,096 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.103e+02 2.436e+02 3.122e+02 5.008e+02, threshold=4.871e+02, percent-clipped=2.0 2023-05-01 18:12:20,390 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234025.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:12:36,702 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9780, 3.0403, 3.2626, 2.0989, 2.9273, 2.3077, 3.4316, 3.3464], device='cuda:6'), covar=tensor([0.0255, 0.0972, 0.0619, 0.2046, 0.0884, 0.1010, 0.0589, 0.1027], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0163, 0.0167, 0.0154, 0.0146, 0.0129, 0.0143, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 18:12:57,928 INFO [train.py:904] (6/8) Epoch 24, batch 600, loss[loss=0.1716, simple_loss=0.2458, pruned_loss=0.04873, over 16754.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2486, pruned_loss=0.03884, over 3162682.78 frames. ], batch size: 102, lr: 2.83e-03, grad_scale: 2.0 2023-05-01 18:13:17,453 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=234066.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:13:45,366 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234086.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:14:06,112 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 18:14:08,030 INFO [train.py:904] (6/8) Epoch 24, batch 650, loss[loss=0.1459, simple_loss=0.232, pruned_loss=0.02987, over 16480.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2479, pruned_loss=0.03825, over 3205239.24 frames. ], batch size: 68, lr: 2.83e-03, grad_scale: 2.0 2023-05-01 18:14:10,900 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-01 18:14:36,191 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 2.015e+02 2.288e+02 2.913e+02 5.815e+02, threshold=4.575e+02, percent-clipped=2.0 2023-05-01 18:15:16,420 INFO [train.py:904] (6/8) Epoch 24, batch 700, loss[loss=0.1895, simple_loss=0.2699, pruned_loss=0.05461, over 16305.00 frames. ], tot_loss[loss=0.1618, simple_loss=0.2477, pruned_loss=0.03797, over 3225380.33 frames. ], batch size: 165, lr: 2.83e-03, grad_scale: 2.0 2023-05-01 18:15:24,295 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5042, 5.9496, 5.6762, 5.7239, 5.2224, 5.3300, 5.3134, 6.0742], device='cuda:6'), covar=tensor([0.1598, 0.1153, 0.1111, 0.0939, 0.1065, 0.0768, 0.1324, 0.1038], device='cuda:6'), in_proj_covar=tensor([0.0685, 0.0833, 0.0686, 0.0638, 0.0529, 0.0536, 0.0705, 0.0655], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:16:23,747 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234202.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:16:24,497 INFO [train.py:904] (6/8) Epoch 24, batch 750, loss[loss=0.213, simple_loss=0.2908, pruned_loss=0.06754, over 12256.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2479, pruned_loss=0.03763, over 3255209.86 frames. ], batch size: 247, lr: 2.83e-03, grad_scale: 2.0 2023-05-01 18:16:44,340 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-01 18:16:45,126 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2933, 3.3322, 3.4566, 2.4136, 3.1760, 3.5493, 3.2497, 2.1121], device='cuda:6'), covar=tensor([0.0506, 0.0115, 0.0071, 0.0426, 0.0136, 0.0125, 0.0134, 0.0475], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0086, 0.0087, 0.0135, 0.0099, 0.0111, 0.0096, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 18:16:52,425 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.040e+02 2.434e+02 2.884e+02 6.768e+02, threshold=4.867e+02, percent-clipped=4.0 2023-05-01 18:17:04,947 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234231.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:17:33,977 INFO [train.py:904] (6/8) Epoch 24, batch 800, loss[loss=0.1535, simple_loss=0.2482, pruned_loss=0.02939, over 17036.00 frames. ], tot_loss[loss=0.1609, simple_loss=0.2474, pruned_loss=0.03724, over 3281047.80 frames. ], batch size: 50, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:17:48,309 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234263.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 18:18:00,244 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-01 18:18:29,141 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234292.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:18:33,017 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=234295.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 18:18:43,643 INFO [train.py:904] (6/8) Epoch 24, batch 850, loss[loss=0.1598, simple_loss=0.2493, pruned_loss=0.03516, over 16969.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.2469, pruned_loss=0.03736, over 3282839.92 frames. ], batch size: 41, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:19:11,879 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.010e+02 2.479e+02 2.918e+02 4.237e+02, threshold=4.958e+02, percent-clipped=0.0 2023-05-01 18:19:40,593 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=234343.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:19:49,569 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2819, 5.2919, 5.1664, 4.6520, 4.7726, 5.1799, 5.1228, 4.7973], device='cuda:6'), covar=tensor([0.0628, 0.0469, 0.0336, 0.0360, 0.1214, 0.0477, 0.0343, 0.0779], device='cuda:6'), in_proj_covar=tensor([0.0300, 0.0444, 0.0351, 0.0350, 0.0353, 0.0407, 0.0238, 0.0420], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:19:52,506 INFO [train.py:904] (6/8) Epoch 24, batch 900, loss[loss=0.1852, simple_loss=0.2592, pruned_loss=0.05559, over 15548.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.2467, pruned_loss=0.03739, over 3284668.20 frames. ], batch size: 190, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:20:11,527 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=234366.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:20:33,657 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=234381.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:21:02,819 INFO [train.py:904] (6/8) Epoch 24, batch 950, loss[loss=0.1704, simple_loss=0.2636, pruned_loss=0.03862, over 17089.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2475, pruned_loss=0.03815, over 3289317.80 frames. ], batch size: 53, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:21:17,861 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=234414.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:21:30,309 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.144e+02 2.493e+02 3.109e+02 1.071e+03, threshold=4.986e+02, percent-clipped=5.0 2023-05-01 18:21:37,396 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9632, 2.0908, 2.4952, 2.8459, 2.6842, 3.3476, 2.2873, 3.3790], device='cuda:6'), covar=tensor([0.0280, 0.0535, 0.0361, 0.0347, 0.0399, 0.0206, 0.0501, 0.0186], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0195, 0.0181, 0.0186, 0.0200, 0.0158, 0.0198, 0.0155], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:21:38,430 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1438, 5.2361, 5.6326, 5.5996, 5.6535, 5.2589, 5.2079, 5.0242], device='cuda:6'), covar=tensor([0.0400, 0.0567, 0.0453, 0.0487, 0.0506, 0.0444, 0.0987, 0.0462], device='cuda:6'), in_proj_covar=tensor([0.0427, 0.0473, 0.0462, 0.0424, 0.0503, 0.0483, 0.0565, 0.0386], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 18:22:10,607 INFO [train.py:904] (6/8) Epoch 24, batch 1000, loss[loss=0.1461, simple_loss=0.2284, pruned_loss=0.03189, over 16626.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2469, pruned_loss=0.03816, over 3301499.82 frames. ], batch size: 62, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:22:50,727 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8911, 4.1266, 2.8675, 4.6975, 3.2884, 4.5773, 2.9791, 3.3404], device='cuda:6'), covar=tensor([0.0360, 0.0395, 0.1467, 0.0321, 0.0816, 0.0506, 0.1431, 0.0789], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0179, 0.0196, 0.0169, 0.0178, 0.0220, 0.0205, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 18:23:22,529 INFO [train.py:904] (6/8) Epoch 24, batch 1050, loss[loss=0.1686, simple_loss=0.2693, pruned_loss=0.03397, over 17270.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.2469, pruned_loss=0.03762, over 3312650.77 frames. ], batch size: 52, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:23:28,481 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-05-01 18:23:50,904 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.512e+02 2.145e+02 2.609e+02 3.101e+02 1.491e+03, threshold=5.219e+02, percent-clipped=5.0 2023-05-01 18:24:30,871 INFO [train.py:904] (6/8) Epoch 24, batch 1100, loss[loss=0.1806, simple_loss=0.2543, pruned_loss=0.05346, over 16908.00 frames. ], tot_loss[loss=0.1604, simple_loss=0.246, pruned_loss=0.03741, over 3310145.07 frames. ], batch size: 116, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:24:37,560 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=234558.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 18:25:18,094 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=234587.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:25:38,660 INFO [train.py:904] (6/8) Epoch 24, batch 1150, loss[loss=0.1632, simple_loss=0.2347, pruned_loss=0.04583, over 16768.00 frames. ], tot_loss[loss=0.1601, simple_loss=0.2459, pruned_loss=0.03719, over 3318906.06 frames. ], batch size: 83, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:25:53,389 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2409, 3.5463, 3.7220, 2.2516, 3.2548, 2.5495, 3.7135, 3.7845], device='cuda:6'), covar=tensor([0.0273, 0.0912, 0.0613, 0.2073, 0.0844, 0.1039, 0.0552, 0.0844], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0165, 0.0168, 0.0155, 0.0146, 0.0130, 0.0144, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 18:26:06,112 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.437e+02 2.089e+02 2.435e+02 2.920e+02 5.927e+02, threshold=4.869e+02, percent-clipped=1.0 2023-05-01 18:26:33,574 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 18:26:46,784 INFO [train.py:904] (6/8) Epoch 24, batch 1200, loss[loss=0.1701, simple_loss=0.2528, pruned_loss=0.04368, over 16575.00 frames. ], tot_loss[loss=0.1592, simple_loss=0.2453, pruned_loss=0.03659, over 3323460.93 frames. ], batch size: 68, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:27:09,065 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234669.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:27:25,017 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=234681.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:27:53,727 INFO [train.py:904] (6/8) Epoch 24, batch 1250, loss[loss=0.1413, simple_loss=0.2271, pruned_loss=0.02779, over 16973.00 frames. ], tot_loss[loss=0.1591, simple_loss=0.2448, pruned_loss=0.03674, over 3322949.92 frames. ], batch size: 41, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:28:21,122 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.044e+02 2.334e+02 2.766e+02 4.867e+02, threshold=4.669e+02, percent-clipped=0.0 2023-05-01 18:28:25,817 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234726.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:28:30,423 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=234729.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:28:31,739 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234730.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 18:28:35,925 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4811, 3.5322, 4.0902, 2.3090, 3.2967, 2.5814, 3.8319, 3.7094], device='cuda:6'), covar=tensor([0.0248, 0.0938, 0.0427, 0.1980, 0.0788, 0.0967, 0.0594, 0.1034], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0166, 0.0169, 0.0155, 0.0147, 0.0131, 0.0145, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 18:28:37,193 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2023-05-01 18:28:39,575 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.04 vs. limit=5.0 2023-05-01 18:29:02,399 INFO [train.py:904] (6/8) Epoch 24, batch 1300, loss[loss=0.1764, simple_loss=0.2538, pruned_loss=0.04952, over 16718.00 frames. ], tot_loss[loss=0.1604, simple_loss=0.2454, pruned_loss=0.0377, over 3312959.67 frames. ], batch size: 124, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:29:27,412 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1588, 5.6908, 5.8279, 5.5644, 5.6264, 6.1934, 5.6629, 5.3114], device='cuda:6'), covar=tensor([0.0963, 0.1889, 0.2406, 0.1991, 0.2602, 0.0941, 0.1605, 0.2482], device='cuda:6'), in_proj_covar=tensor([0.0418, 0.0611, 0.0675, 0.0508, 0.0673, 0.0705, 0.0526, 0.0674], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 18:29:41,651 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234782.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:29:47,435 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234787.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:30:10,000 INFO [train.py:904] (6/8) Epoch 24, batch 1350, loss[loss=0.149, simple_loss=0.2289, pruned_loss=0.0345, over 16772.00 frames. ], tot_loss[loss=0.1604, simple_loss=0.2456, pruned_loss=0.03763, over 3318290.58 frames. ], batch size: 39, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:30:38,934 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.157e+02 2.428e+02 2.986e+02 5.268e+02, threshold=4.855e+02, percent-clipped=4.0 2023-05-01 18:30:41,875 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1741, 4.0025, 4.4123, 2.4617, 4.5893, 4.6747, 3.3829, 3.5364], device='cuda:6'), covar=tensor([0.0701, 0.0262, 0.0216, 0.1139, 0.0085, 0.0179, 0.0425, 0.0452], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0109, 0.0100, 0.0141, 0.0083, 0.0129, 0.0129, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 18:30:45,934 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-01 18:31:07,924 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234843.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:31:19,624 INFO [train.py:904] (6/8) Epoch 24, batch 1400, loss[loss=0.1541, simple_loss=0.227, pruned_loss=0.04064, over 15483.00 frames. ], tot_loss[loss=0.1603, simple_loss=0.2456, pruned_loss=0.03749, over 3319523.84 frames. ], batch size: 190, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:31:27,994 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=234858.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:31:59,535 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-05-01 18:32:00,156 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234882.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:32:07,376 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=234887.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:32:29,521 INFO [train.py:904] (6/8) Epoch 24, batch 1450, loss[loss=0.1416, simple_loss=0.2251, pruned_loss=0.02908, over 11982.00 frames. ], tot_loss[loss=0.1601, simple_loss=0.2449, pruned_loss=0.03767, over 3316691.97 frames. ], batch size: 248, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:32:34,694 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=234906.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:32:42,144 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-01 18:32:51,058 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-01 18:32:58,961 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 1.989e+02 2.305e+02 2.603e+02 6.556e+02, threshold=4.610e+02, percent-clipped=2.0 2023-05-01 18:33:14,529 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=234935.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:33:26,798 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234943.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:33:38,960 INFO [train.py:904] (6/8) Epoch 24, batch 1500, loss[loss=0.1808, simple_loss=0.2483, pruned_loss=0.05667, over 16367.00 frames. ], tot_loss[loss=0.1606, simple_loss=0.2452, pruned_loss=0.03796, over 3317939.45 frames. ], batch size: 146, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:33:44,888 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.89 vs. limit=5.0 2023-05-01 18:33:45,569 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234958.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:33:49,612 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234960.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:34:06,044 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-01 18:34:13,055 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8506, 1.9565, 2.3984, 2.6605, 2.7097, 2.6371, 1.9154, 2.9228], device='cuda:6'), covar=tensor([0.0207, 0.0530, 0.0378, 0.0323, 0.0332, 0.0399, 0.0608, 0.0210], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0198, 0.0183, 0.0189, 0.0203, 0.0161, 0.0201, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:34:28,921 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2971, 2.2824, 2.4723, 4.0750, 2.2583, 2.6742, 2.2998, 2.4863], device='cuda:6'), covar=tensor([0.1520, 0.3538, 0.2916, 0.0618, 0.3869, 0.2429, 0.3886, 0.2888], device='cuda:6'), in_proj_covar=tensor([0.0411, 0.0460, 0.0378, 0.0333, 0.0443, 0.0526, 0.0431, 0.0538], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:34:49,719 INFO [train.py:904] (6/8) Epoch 24, batch 1550, loss[loss=0.1654, simple_loss=0.2607, pruned_loss=0.03505, over 17289.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2473, pruned_loss=0.03914, over 3318117.59 frames. ], batch size: 52, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:35:12,967 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=235019.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:35:15,747 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=235021.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:35:18,854 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.608e+02 2.203e+02 2.509e+02 3.045e+02 6.004e+02, threshold=5.019e+02, percent-clipped=5.0 2023-05-01 18:35:20,353 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235025.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 18:35:21,533 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1777, 5.7125, 5.8700, 5.5644, 5.7194, 6.2485, 5.7976, 5.4185], device='cuda:6'), covar=tensor([0.0891, 0.1968, 0.2494, 0.2053, 0.2497, 0.0888, 0.1520, 0.2529], device='cuda:6'), in_proj_covar=tensor([0.0421, 0.0616, 0.0679, 0.0510, 0.0679, 0.0710, 0.0529, 0.0682], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 18:35:58,105 INFO [train.py:904] (6/8) Epoch 24, batch 1600, loss[loss=0.1793, simple_loss=0.2522, pruned_loss=0.05317, over 16773.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2492, pruned_loss=0.03934, over 3322976.78 frames. ], batch size: 124, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:36:37,992 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235082.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:37:06,937 INFO [train.py:904] (6/8) Epoch 24, batch 1650, loss[loss=0.1501, simple_loss=0.237, pruned_loss=0.03162, over 15751.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2506, pruned_loss=0.03984, over 3320354.00 frames. ], batch size: 35, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:37:35,278 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.609e+02 2.234e+02 2.602e+02 3.440e+02 9.714e+02, threshold=5.204e+02, percent-clipped=4.0 2023-05-01 18:37:36,717 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-01 18:37:55,436 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235138.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:38:16,050 INFO [train.py:904] (6/8) Epoch 24, batch 1700, loss[loss=0.1888, simple_loss=0.2765, pruned_loss=0.05048, over 16359.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2524, pruned_loss=0.04076, over 3309833.07 frames. ], batch size: 165, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:39:24,670 INFO [train.py:904] (6/8) Epoch 24, batch 1750, loss[loss=0.1677, simple_loss=0.2555, pruned_loss=0.04001, over 17074.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2531, pruned_loss=0.04074, over 3313435.32 frames. ], batch size: 55, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:39:52,420 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.615e+02 2.256e+02 2.911e+02 3.628e+02 6.601e+02, threshold=5.823e+02, percent-clipped=7.0 2023-05-01 18:40:12,948 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235238.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:40:19,733 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.90 vs. limit=5.0 2023-05-01 18:40:33,173 INFO [train.py:904] (6/8) Epoch 24, batch 1800, loss[loss=0.1548, simple_loss=0.251, pruned_loss=0.02928, over 17111.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.254, pruned_loss=0.04079, over 3321573.30 frames. ], batch size: 49, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:40:36,422 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=235255.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:40:57,005 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 18:41:30,313 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-05-01 18:41:34,357 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2569, 2.3250, 2.5240, 4.0904, 2.3183, 2.6981, 2.3967, 2.5460], device='cuda:6'), covar=tensor([0.1489, 0.3586, 0.2881, 0.0609, 0.3956, 0.2573, 0.3795, 0.3077], device='cuda:6'), in_proj_covar=tensor([0.0412, 0.0462, 0.0379, 0.0334, 0.0444, 0.0528, 0.0433, 0.0539], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:41:40,802 INFO [train.py:904] (6/8) Epoch 24, batch 1850, loss[loss=0.1452, simple_loss=0.2441, pruned_loss=0.02312, over 17287.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.255, pruned_loss=0.04061, over 3323669.69 frames. ], batch size: 52, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:41:56,544 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235314.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:41:59,487 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235316.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:41:59,632 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=235316.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:42:11,305 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.696e+02 2.161e+02 2.587e+02 3.187e+02 4.962e+02, threshold=5.175e+02, percent-clipped=0.0 2023-05-01 18:42:12,319 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235325.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:42:49,548 INFO [train.py:904] (6/8) Epoch 24, batch 1900, loss[loss=0.1318, simple_loss=0.2198, pruned_loss=0.02193, over 16764.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2547, pruned_loss=0.04033, over 3299302.98 frames. ], batch size: 39, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:43:01,050 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=235360.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 18:43:14,803 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2367, 5.0167, 5.2502, 5.4008, 5.6076, 4.8690, 5.5647, 5.5847], device='cuda:6'), covar=tensor([0.1775, 0.1292, 0.1618, 0.0816, 0.0518, 0.0999, 0.0672, 0.0594], device='cuda:6'), in_proj_covar=tensor([0.0677, 0.0840, 0.0963, 0.0849, 0.0644, 0.0667, 0.0700, 0.0814], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:43:17,590 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235373.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:43:31,589 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235382.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:44:00,955 INFO [train.py:904] (6/8) Epoch 24, batch 1950, loss[loss=0.1886, simple_loss=0.2827, pruned_loss=0.04723, over 12108.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2555, pruned_loss=0.04056, over 3292090.09 frames. ], batch size: 247, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:44:26,883 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=235421.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 18:44:31,114 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.100e+02 2.476e+02 2.993e+02 5.203e+02, threshold=4.952e+02, percent-clipped=1.0 2023-05-01 18:44:37,916 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235430.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:44:43,422 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.03 vs. limit=5.0 2023-05-01 18:44:49,016 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235438.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:45:00,460 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2326, 1.6434, 2.0326, 2.0979, 2.2849, 2.2890, 1.7103, 2.3083], device='cuda:6'), covar=tensor([0.0241, 0.0495, 0.0293, 0.0355, 0.0313, 0.0318, 0.0554, 0.0211], device='cuda:6'), in_proj_covar=tensor([0.0195, 0.0196, 0.0183, 0.0188, 0.0202, 0.0161, 0.0200, 0.0158], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:45:08,803 INFO [train.py:904] (6/8) Epoch 24, batch 2000, loss[loss=0.1725, simple_loss=0.2439, pruned_loss=0.05059, over 16778.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2556, pruned_loss=0.04042, over 3301413.87 frames. ], batch size: 124, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:45:53,862 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235486.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:46:08,444 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2012, 4.9489, 5.2101, 5.3665, 5.5692, 4.9006, 5.5337, 5.5712], device='cuda:6'), covar=tensor([0.1723, 0.1370, 0.1707, 0.0789, 0.0527, 0.0867, 0.0531, 0.0593], device='cuda:6'), in_proj_covar=tensor([0.0682, 0.0845, 0.0971, 0.0853, 0.0648, 0.0670, 0.0704, 0.0818], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:46:16,400 INFO [train.py:904] (6/8) Epoch 24, batch 2050, loss[loss=0.191, simple_loss=0.2709, pruned_loss=0.05558, over 15367.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2555, pruned_loss=0.04086, over 3298219.80 frames. ], batch size: 190, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:46:46,280 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.341e+02 2.111e+02 2.349e+02 3.134e+02 8.501e+02, threshold=4.699e+02, percent-clipped=2.0 2023-05-01 18:46:57,422 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-01 18:47:03,517 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235538.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:47:24,545 INFO [train.py:904] (6/8) Epoch 24, batch 2100, loss[loss=0.1484, simple_loss=0.2408, pruned_loss=0.02801, over 17205.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2553, pruned_loss=0.04109, over 3289557.07 frames. ], batch size: 46, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:47:34,480 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4185, 3.3939, 3.4605, 3.5225, 3.5620, 3.3123, 3.5087, 3.6316], device='cuda:6'), covar=tensor([0.1334, 0.0990, 0.1117, 0.0717, 0.0694, 0.2190, 0.1401, 0.0829], device='cuda:6'), in_proj_covar=tensor([0.0683, 0.0846, 0.0974, 0.0855, 0.0649, 0.0672, 0.0706, 0.0821], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:47:49,118 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1001, 2.3373, 2.7344, 3.1350, 2.9026, 3.5463, 2.5388, 3.4743], device='cuda:6'), covar=tensor([0.0262, 0.0513, 0.0345, 0.0357, 0.0393, 0.0213, 0.0479, 0.0196], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0198, 0.0184, 0.0189, 0.0204, 0.0162, 0.0201, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:48:10,074 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235586.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:48:31,075 INFO [train.py:904] (6/8) Epoch 24, batch 2150, loss[loss=0.1736, simple_loss=0.2535, pruned_loss=0.04684, over 16355.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2555, pruned_loss=0.04125, over 3292878.92 frames. ], batch size: 165, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:48:43,102 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235611.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:48:47,533 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235614.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:48:48,073 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 18:48:49,920 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235616.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:49:02,117 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.182e+02 2.677e+02 3.005e+02 4.991e+02, threshold=5.354e+02, percent-clipped=2.0 2023-05-01 18:49:41,311 INFO [train.py:904] (6/8) Epoch 24, batch 2200, loss[loss=0.1958, simple_loss=0.2807, pruned_loss=0.05543, over 15424.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2565, pruned_loss=0.04155, over 3296526.59 frames. ], batch size: 190, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:49:54,417 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235662.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:49:57,749 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235664.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:50:18,269 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=235679.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:50:50,798 INFO [train.py:904] (6/8) Epoch 24, batch 2250, loss[loss=0.1631, simple_loss=0.2457, pruned_loss=0.04027, over 16675.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2565, pruned_loss=0.04136, over 3307330.23 frames. ], batch size: 89, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:51:08,548 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235716.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 18:51:19,589 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9622, 4.5399, 3.0863, 2.4355, 2.9283, 2.6285, 4.8800, 3.6117], device='cuda:6'), covar=tensor([0.2952, 0.0572, 0.2046, 0.3007, 0.2918, 0.2234, 0.0365, 0.1561], device='cuda:6'), in_proj_covar=tensor([0.0331, 0.0273, 0.0311, 0.0319, 0.0301, 0.0268, 0.0300, 0.0343], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 18:51:20,155 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.445e+02 2.086e+02 2.517e+02 3.047e+02 4.896e+02, threshold=5.035e+02, percent-clipped=0.0 2023-05-01 18:51:41,261 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=235740.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:51:57,982 INFO [train.py:904] (6/8) Epoch 24, batch 2300, loss[loss=0.1509, simple_loss=0.2323, pruned_loss=0.03476, over 15860.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2568, pruned_loss=0.04183, over 3304557.81 frames. ], batch size: 35, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:52:50,296 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=235791.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:53:08,147 INFO [train.py:904] (6/8) Epoch 24, batch 2350, loss[loss=0.1929, simple_loss=0.261, pruned_loss=0.06242, over 16528.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2573, pruned_loss=0.04242, over 3302786.17 frames. ], batch size: 75, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:53:37,789 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.657e+02 2.284e+02 2.753e+02 3.332e+02 6.201e+02, threshold=5.507e+02, percent-clipped=2.0 2023-05-01 18:54:14,679 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=235852.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:54:15,310 INFO [train.py:904] (6/8) Epoch 24, batch 2400, loss[loss=0.1742, simple_loss=0.27, pruned_loss=0.03923, over 17044.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2588, pruned_loss=0.04239, over 3314704.00 frames. ], batch size: 53, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:54:40,612 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9935, 2.2644, 2.6524, 3.0589, 2.8892, 3.5211, 2.5351, 3.5314], device='cuda:6'), covar=tensor([0.0322, 0.0493, 0.0356, 0.0347, 0.0361, 0.0232, 0.0493, 0.0185], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0197, 0.0184, 0.0189, 0.0204, 0.0162, 0.0201, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:55:22,142 INFO [train.py:904] (6/8) Epoch 24, batch 2450, loss[loss=0.187, simple_loss=0.2817, pruned_loss=0.04616, over 17054.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2589, pruned_loss=0.04187, over 3315437.08 frames. ], batch size: 53, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:55:33,305 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235911.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:55:43,736 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1460, 5.1505, 4.8961, 4.3316, 4.9420, 1.8010, 4.7126, 4.7088], device='cuda:6'), covar=tensor([0.0089, 0.0096, 0.0252, 0.0481, 0.0123, 0.3159, 0.0165, 0.0297], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0166, 0.0208, 0.0183, 0.0183, 0.0214, 0.0197, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 18:55:51,879 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.168e+02 2.510e+02 2.858e+02 5.891e+02, threshold=5.019e+02, percent-clipped=1.0 2023-05-01 18:56:28,796 INFO [train.py:904] (6/8) Epoch 24, batch 2500, loss[loss=0.1646, simple_loss=0.2652, pruned_loss=0.03203, over 16708.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.258, pruned_loss=0.04123, over 3313878.91 frames. ], batch size: 62, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:56:37,274 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235959.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:57:41,345 INFO [train.py:904] (6/8) Epoch 24, batch 2550, loss[loss=0.1504, simple_loss=0.2344, pruned_loss=0.03316, over 16837.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2577, pruned_loss=0.04086, over 3315974.27 frames. ], batch size: 42, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:57:59,217 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=236016.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 18:58:01,844 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=236018.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:58:10,533 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.492e+02 2.266e+02 2.526e+02 3.100e+02 4.908e+02, threshold=5.053e+02, percent-clipped=0.0 2023-05-01 18:58:24,330 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=236035.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:58:49,149 INFO [train.py:904] (6/8) Epoch 24, batch 2600, loss[loss=0.1807, simple_loss=0.2719, pruned_loss=0.04475, over 16702.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2575, pruned_loss=0.04067, over 3317727.47 frames. ], batch size: 57, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:59:03,972 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=236064.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 18:59:25,500 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=236079.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:59:58,182 INFO [train.py:904] (6/8) Epoch 24, batch 2650, loss[loss=0.1542, simple_loss=0.2502, pruned_loss=0.02908, over 17158.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.258, pruned_loss=0.0408, over 3321735.07 frames. ], batch size: 46, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:00:27,606 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.167e+02 2.518e+02 3.194e+02 5.192e+02, threshold=5.036e+02, percent-clipped=2.0 2023-05-01 19:00:58,544 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=236147.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:01:05,271 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7168, 3.9362, 2.6650, 4.5904, 3.1324, 4.5141, 2.8398, 3.3379], device='cuda:6'), covar=tensor([0.0360, 0.0435, 0.1451, 0.0311, 0.0850, 0.0501, 0.1320, 0.0720], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0182, 0.0198, 0.0173, 0.0180, 0.0224, 0.0206, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 19:01:05,907 INFO [train.py:904] (6/8) Epoch 24, batch 2700, loss[loss=0.1651, simple_loss=0.2636, pruned_loss=0.03328, over 17133.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2585, pruned_loss=0.04066, over 3321044.00 frames. ], batch size: 48, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:01:27,884 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9648, 4.7352, 5.0407, 5.1858, 5.3950, 4.7176, 5.3854, 5.4126], device='cuda:6'), covar=tensor([0.2040, 0.1439, 0.1754, 0.0803, 0.0550, 0.0969, 0.0680, 0.0577], device='cuda:6'), in_proj_covar=tensor([0.0682, 0.0844, 0.0972, 0.0854, 0.0651, 0.0674, 0.0706, 0.0819], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:01:54,942 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4316, 5.4024, 5.2895, 4.8079, 4.9254, 5.3765, 5.3046, 4.9474], device='cuda:6'), covar=tensor([0.0610, 0.0420, 0.0286, 0.0345, 0.1027, 0.0416, 0.0266, 0.0748], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0464, 0.0364, 0.0364, 0.0369, 0.0422, 0.0248, 0.0440], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 19:02:15,469 INFO [train.py:904] (6/8) Epoch 24, batch 2750, loss[loss=0.16, simple_loss=0.2596, pruned_loss=0.03021, over 17261.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.258, pruned_loss=0.03977, over 3328922.57 frames. ], batch size: 52, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:02:44,667 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.298e+02 1.972e+02 2.253e+02 2.676e+02 4.118e+02, threshold=4.507e+02, percent-clipped=0.0 2023-05-01 19:03:22,962 INFO [train.py:904] (6/8) Epoch 24, batch 2800, loss[loss=0.1545, simple_loss=0.2509, pruned_loss=0.02904, over 17094.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2573, pruned_loss=0.03951, over 3328817.65 frames. ], batch size: 47, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:04:32,914 INFO [train.py:904] (6/8) Epoch 24, batch 2850, loss[loss=0.1834, simple_loss=0.2747, pruned_loss=0.04603, over 17052.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2573, pruned_loss=0.03944, over 3330373.02 frames. ], batch size: 50, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:05:03,604 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2023-05-01 19:05:04,031 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 2.044e+02 2.504e+02 3.029e+02 5.589e+02, threshold=5.008e+02, percent-clipped=4.0 2023-05-01 19:05:17,508 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=236335.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:05:43,131 INFO [train.py:904] (6/8) Epoch 24, batch 2900, loss[loss=0.1583, simple_loss=0.2568, pruned_loss=0.0299, over 17143.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2559, pruned_loss=0.03968, over 3327818.08 frames. ], batch size: 48, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:06:13,275 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=236374.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:06:25,531 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=236383.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:06:45,425 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3019, 5.2849, 5.1696, 4.6779, 4.7957, 5.2302, 5.1568, 4.8585], device='cuda:6'), covar=tensor([0.0639, 0.0477, 0.0308, 0.0376, 0.1124, 0.0479, 0.0356, 0.0784], device='cuda:6'), in_proj_covar=tensor([0.0314, 0.0466, 0.0366, 0.0366, 0.0371, 0.0424, 0.0250, 0.0442], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 19:06:53,578 INFO [train.py:904] (6/8) Epoch 24, batch 2950, loss[loss=0.1799, simple_loss=0.2517, pruned_loss=0.05409, over 16846.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2555, pruned_loss=0.04047, over 3325273.81 frames. ], batch size: 116, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:07:24,083 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.806e+02 2.404e+02 2.771e+02 3.339e+02 1.037e+03, threshold=5.543e+02, percent-clipped=6.0 2023-05-01 19:07:30,663 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0347, 4.4530, 4.3801, 3.2400, 3.7206, 4.4001, 3.8877, 2.5798], device='cuda:6'), covar=tensor([0.0475, 0.0071, 0.0060, 0.0375, 0.0145, 0.0112, 0.0106, 0.0490], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0088, 0.0088, 0.0136, 0.0101, 0.0113, 0.0097, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 19:07:54,772 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=236447.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:08:02,709 INFO [train.py:904] (6/8) Epoch 24, batch 3000, loss[loss=0.1551, simple_loss=0.2477, pruned_loss=0.03121, over 17130.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2558, pruned_loss=0.041, over 3333364.29 frames. ], batch size: 49, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:08:02,709 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 19:08:12,051 INFO [train.py:938] (6/8) Epoch 24, validation: loss=0.1342, simple_loss=0.2393, pruned_loss=0.0145, over 944034.00 frames. 2023-05-01 19:08:12,052 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 19:08:46,457 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4122, 3.5889, 3.2689, 2.9962, 3.0301, 3.4836, 3.1799, 3.2615], device='cuda:6'), covar=tensor([0.0873, 0.0760, 0.0475, 0.0419, 0.0902, 0.0570, 0.2308, 0.0666], device='cuda:6'), in_proj_covar=tensor([0.0314, 0.0466, 0.0367, 0.0366, 0.0371, 0.0425, 0.0250, 0.0443], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 19:09:12,138 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=236495.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:09:22,989 INFO [train.py:904] (6/8) Epoch 24, batch 3050, loss[loss=0.1662, simple_loss=0.249, pruned_loss=0.04169, over 16874.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2563, pruned_loss=0.04115, over 3321250.70 frames. ], batch size: 96, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:09:53,422 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.422e+02 2.094e+02 2.423e+02 2.785e+02 4.383e+02, threshold=4.846e+02, percent-clipped=1.0 2023-05-01 19:10:01,959 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2298, 4.0708, 4.2915, 4.4184, 4.4940, 4.0741, 4.2572, 4.4853], device='cuda:6'), covar=tensor([0.1550, 0.1168, 0.1232, 0.0659, 0.0626, 0.1481, 0.2210, 0.0789], device='cuda:6'), in_proj_covar=tensor([0.0686, 0.0850, 0.0978, 0.0854, 0.0652, 0.0679, 0.0707, 0.0823], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:10:32,486 INFO [train.py:904] (6/8) Epoch 24, batch 3100, loss[loss=0.1611, simple_loss=0.2537, pruned_loss=0.03423, over 17145.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2558, pruned_loss=0.04131, over 3327173.86 frames. ], batch size: 48, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:11:43,887 INFO [train.py:904] (6/8) Epoch 24, batch 3150, loss[loss=0.1903, simple_loss=0.2608, pruned_loss=0.05991, over 16409.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2557, pruned_loss=0.04148, over 3321831.73 frames. ], batch size: 146, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:12:13,746 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.103e+02 2.556e+02 2.857e+02 5.533e+02, threshold=5.112e+02, percent-clipped=2.0 2023-05-01 19:12:49,293 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7987, 4.9152, 5.1046, 4.9183, 4.9618, 5.5609, 5.0919, 4.7627], device='cuda:6'), covar=tensor([0.1537, 0.2099, 0.2636, 0.2254, 0.2856, 0.1117, 0.1601, 0.2516], device='cuda:6'), in_proj_covar=tensor([0.0424, 0.0624, 0.0687, 0.0517, 0.0689, 0.0718, 0.0538, 0.0685], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 19:12:52,317 INFO [train.py:904] (6/8) Epoch 24, batch 3200, loss[loss=0.1806, simple_loss=0.2602, pruned_loss=0.05043, over 16920.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2549, pruned_loss=0.04118, over 3322627.75 frames. ], batch size: 109, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:13:21,974 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=236674.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:13:34,884 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6639, 4.5101, 4.6799, 4.8422, 5.0129, 4.5155, 4.9301, 5.0100], device='cuda:6'), covar=tensor([0.1832, 0.1215, 0.1591, 0.0787, 0.0681, 0.1163, 0.1175, 0.0849], device='cuda:6'), in_proj_covar=tensor([0.0685, 0.0850, 0.0976, 0.0851, 0.0652, 0.0678, 0.0707, 0.0822], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:13:50,683 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-01 19:14:01,577 INFO [train.py:904] (6/8) Epoch 24, batch 3250, loss[loss=0.1611, simple_loss=0.2505, pruned_loss=0.03586, over 16810.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2543, pruned_loss=0.04083, over 3321485.56 frames. ], batch size: 102, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:14:27,661 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=236722.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:14:31,667 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.578e+02 2.232e+02 2.615e+02 3.014e+02 5.902e+02, threshold=5.231e+02, percent-clipped=1.0 2023-05-01 19:15:11,538 INFO [train.py:904] (6/8) Epoch 24, batch 3300, loss[loss=0.1574, simple_loss=0.2587, pruned_loss=0.028, over 17114.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2558, pruned_loss=0.04152, over 3309694.33 frames. ], batch size: 49, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:16:21,170 INFO [train.py:904] (6/8) Epoch 24, batch 3350, loss[loss=0.1336, simple_loss=0.221, pruned_loss=0.0231, over 16818.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2555, pruned_loss=0.04093, over 3307580.31 frames. ], batch size: 39, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:16:51,747 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 2.070e+02 2.526e+02 2.944e+02 6.359e+02, threshold=5.052e+02, percent-clipped=3.0 2023-05-01 19:16:54,205 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-01 19:17:04,627 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1924, 5.1717, 4.9636, 4.3592, 5.0397, 1.8155, 4.7985, 4.8540], device='cuda:6'), covar=tensor([0.0095, 0.0088, 0.0217, 0.0462, 0.0107, 0.3078, 0.0153, 0.0231], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0168, 0.0210, 0.0186, 0.0186, 0.0216, 0.0199, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:17:15,819 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-05-01 19:17:33,207 INFO [train.py:904] (6/8) Epoch 24, batch 3400, loss[loss=0.1447, simple_loss=0.2262, pruned_loss=0.03162, over 16840.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2551, pruned_loss=0.04075, over 3305182.89 frames. ], batch size: 42, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:17:40,358 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5545, 4.5952, 4.9408, 4.9165, 4.9568, 4.6552, 4.6351, 4.5008], device='cuda:6'), covar=tensor([0.0358, 0.0659, 0.0422, 0.0403, 0.0509, 0.0424, 0.0830, 0.0612], device='cuda:6'), in_proj_covar=tensor([0.0429, 0.0478, 0.0465, 0.0428, 0.0510, 0.0485, 0.0570, 0.0390], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 19:17:44,507 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9105, 5.2658, 5.0232, 5.0349, 4.7580, 4.6977, 4.7028, 5.3768], device='cuda:6'), covar=tensor([0.1266, 0.0851, 0.1113, 0.0861, 0.0856, 0.1159, 0.1248, 0.0826], device='cuda:6'), in_proj_covar=tensor([0.0715, 0.0871, 0.0717, 0.0670, 0.0555, 0.0557, 0.0731, 0.0680], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:18:31,042 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=236892.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:18:45,597 INFO [train.py:904] (6/8) Epoch 24, batch 3450, loss[loss=0.1605, simple_loss=0.2343, pruned_loss=0.0433, over 16424.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2539, pruned_loss=0.04023, over 3290322.22 frames. ], batch size: 146, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:19:09,335 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9362, 2.8773, 2.6614, 4.6470, 3.7414, 4.2448, 1.6641, 3.0664], device='cuda:6'), covar=tensor([0.1369, 0.0770, 0.1211, 0.0232, 0.0227, 0.0436, 0.1683, 0.0837], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0177, 0.0197, 0.0197, 0.0207, 0.0219, 0.0205, 0.0195], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 19:19:15,849 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.559e+02 2.094e+02 2.355e+02 2.776e+02 4.395e+02, threshold=4.710e+02, percent-clipped=0.0 2023-05-01 19:19:39,828 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0696, 4.5849, 3.1570, 2.4600, 2.7780, 2.7211, 4.9936, 3.6375], device='cuda:6'), covar=tensor([0.2868, 0.0552, 0.1903, 0.3083, 0.3206, 0.2131, 0.0342, 0.1599], device='cuda:6'), in_proj_covar=tensor([0.0330, 0.0272, 0.0309, 0.0320, 0.0302, 0.0267, 0.0298, 0.0344], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 19:19:55,271 INFO [train.py:904] (6/8) Epoch 24, batch 3500, loss[loss=0.2205, simple_loss=0.2972, pruned_loss=0.07192, over 11979.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2529, pruned_loss=0.04, over 3286458.71 frames. ], batch size: 246, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:19:56,883 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=236953.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:20:07,005 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8852, 4.7871, 4.7728, 4.4089, 4.4354, 4.8258, 4.6151, 4.5455], device='cuda:6'), covar=tensor([0.0653, 0.0894, 0.0326, 0.0344, 0.0958, 0.0473, 0.0485, 0.0712], device='cuda:6'), in_proj_covar=tensor([0.0316, 0.0469, 0.0368, 0.0368, 0.0373, 0.0426, 0.0251, 0.0445], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 19:21:06,713 INFO [train.py:904] (6/8) Epoch 24, batch 3550, loss[loss=0.1596, simple_loss=0.2402, pruned_loss=0.03952, over 16887.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2521, pruned_loss=0.03918, over 3289582.77 frames. ], batch size: 90, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:21:19,157 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5032, 5.8883, 5.5789, 5.6569, 5.2383, 5.2691, 5.1778, 5.9764], device='cuda:6'), covar=tensor([0.1325, 0.0853, 0.1035, 0.0849, 0.0893, 0.0693, 0.1333, 0.0887], device='cuda:6'), in_proj_covar=tensor([0.0718, 0.0873, 0.0719, 0.0673, 0.0557, 0.0558, 0.0734, 0.0682], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:21:24,771 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6620, 4.9571, 5.2643, 5.2064, 5.2919, 4.9513, 4.6324, 4.7241], device='cuda:6'), covar=tensor([0.0735, 0.0721, 0.0675, 0.0718, 0.0713, 0.0671, 0.1545, 0.0608], device='cuda:6'), in_proj_covar=tensor([0.0435, 0.0484, 0.0470, 0.0432, 0.0515, 0.0491, 0.0577, 0.0394], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 19:21:35,731 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 1.933e+02 2.243e+02 2.593e+02 4.523e+02, threshold=4.485e+02, percent-clipped=0.0 2023-05-01 19:22:15,045 INFO [train.py:904] (6/8) Epoch 24, batch 3600, loss[loss=0.1428, simple_loss=0.2355, pruned_loss=0.02509, over 17240.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2508, pruned_loss=0.0385, over 3296372.48 frames. ], batch size: 44, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:23:26,120 INFO [train.py:904] (6/8) Epoch 24, batch 3650, loss[loss=0.1601, simple_loss=0.2545, pruned_loss=0.03288, over 17169.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.25, pruned_loss=0.03894, over 3298194.94 frames. ], batch size: 46, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:23:58,727 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.153e+02 2.502e+02 3.114e+02 9.969e+02, threshold=5.004e+02, percent-clipped=5.0 2023-05-01 19:24:39,858 INFO [train.py:904] (6/8) Epoch 24, batch 3700, loss[loss=0.1707, simple_loss=0.2482, pruned_loss=0.04661, over 11770.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2487, pruned_loss=0.04079, over 3285280.32 frames. ], batch size: 248, lr: 2.82e-03, grad_scale: 4.0 2023-05-01 19:25:53,323 INFO [train.py:904] (6/8) Epoch 24, batch 3750, loss[loss=0.1615, simple_loss=0.2353, pruned_loss=0.04383, over 16846.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2499, pruned_loss=0.04224, over 3277387.62 frames. ], batch size: 96, lr: 2.82e-03, grad_scale: 4.0 2023-05-01 19:26:25,688 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.223e+02 2.609e+02 3.048e+02 5.921e+02, threshold=5.219e+02, percent-clipped=3.0 2023-05-01 19:26:57,501 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=237248.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 19:26:59,399 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4011, 5.4263, 5.2671, 4.8734, 4.9061, 5.3678, 5.0887, 5.0207], device='cuda:6'), covar=tensor([0.0492, 0.0347, 0.0245, 0.0271, 0.0871, 0.0312, 0.0312, 0.0555], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0466, 0.0365, 0.0364, 0.0368, 0.0421, 0.0247, 0.0440], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 19:27:03,606 INFO [train.py:904] (6/8) Epoch 24, batch 3800, loss[loss=0.1801, simple_loss=0.2636, pruned_loss=0.0483, over 16220.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2518, pruned_loss=0.04351, over 3276544.84 frames. ], batch size: 165, lr: 2.82e-03, grad_scale: 4.0 2023-05-01 19:27:11,149 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4117, 3.5596, 3.6879, 3.6444, 3.6723, 3.5034, 3.5448, 3.5278], device='cuda:6'), covar=tensor([0.0386, 0.0601, 0.0439, 0.0442, 0.0589, 0.0517, 0.0718, 0.0581], device='cuda:6'), in_proj_covar=tensor([0.0432, 0.0483, 0.0467, 0.0431, 0.0512, 0.0487, 0.0573, 0.0392], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 19:27:25,245 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3231, 2.4455, 2.5927, 4.2373, 2.3606, 2.7808, 2.4904, 2.6199], device='cuda:6'), covar=tensor([0.1394, 0.3582, 0.2742, 0.0516, 0.3748, 0.2415, 0.3625, 0.2922], device='cuda:6'), in_proj_covar=tensor([0.0416, 0.0465, 0.0381, 0.0337, 0.0444, 0.0534, 0.0435, 0.0544], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:28:20,060 INFO [train.py:904] (6/8) Epoch 24, batch 3850, loss[loss=0.1803, simple_loss=0.2607, pruned_loss=0.04993, over 16415.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2516, pruned_loss=0.04419, over 3277008.51 frames. ], batch size: 146, lr: 2.81e-03, grad_scale: 4.0 2023-05-01 19:28:52,885 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.317e+02 2.601e+02 3.052e+02 4.766e+02, threshold=5.202e+02, percent-clipped=0.0 2023-05-01 19:28:53,432 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4474, 3.0169, 3.3864, 1.8389, 3.4448, 3.4178, 2.9214, 2.6380], device='cuda:6'), covar=tensor([0.0704, 0.0281, 0.0171, 0.1138, 0.0119, 0.0219, 0.0400, 0.0424], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0110, 0.0100, 0.0140, 0.0083, 0.0129, 0.0129, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 19:29:00,977 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9575, 2.1027, 2.5353, 2.8861, 2.8875, 3.0104, 2.2873, 3.1758], device='cuda:6'), covar=tensor([0.0235, 0.0522, 0.0350, 0.0316, 0.0322, 0.0295, 0.0490, 0.0156], device='cuda:6'), in_proj_covar=tensor([0.0197, 0.0196, 0.0185, 0.0189, 0.0204, 0.0163, 0.0201, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:29:30,846 INFO [train.py:904] (6/8) Epoch 24, batch 3900, loss[loss=0.1712, simple_loss=0.241, pruned_loss=0.05069, over 16784.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2518, pruned_loss=0.04512, over 3280080.88 frames. ], batch size: 124, lr: 2.81e-03, grad_scale: 4.0 2023-05-01 19:29:57,990 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9791, 4.9090, 4.8638, 4.5276, 4.5701, 4.9224, 4.7114, 4.6648], device='cuda:6'), covar=tensor([0.0606, 0.0563, 0.0333, 0.0310, 0.0881, 0.0424, 0.0455, 0.0627], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0466, 0.0365, 0.0364, 0.0367, 0.0422, 0.0247, 0.0440], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 19:30:43,386 INFO [train.py:904] (6/8) Epoch 24, batch 3950, loss[loss=0.1584, simple_loss=0.2311, pruned_loss=0.04286, over 16796.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.251, pruned_loss=0.04557, over 3291972.95 frames. ], batch size: 89, lr: 2.81e-03, grad_scale: 4.0 2023-05-01 19:31:14,713 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 19:31:17,992 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 2.223e+02 2.645e+02 3.283e+02 6.262e+02, threshold=5.289e+02, percent-clipped=1.0 2023-05-01 19:31:57,071 INFO [train.py:904] (6/8) Epoch 24, batch 4000, loss[loss=0.1867, simple_loss=0.2694, pruned_loss=0.05202, over 16407.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2501, pruned_loss=0.04539, over 3301930.19 frames. ], batch size: 68, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:32:06,801 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 19:32:53,169 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=237491.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:33:10,919 INFO [train.py:904] (6/8) Epoch 24, batch 4050, loss[loss=0.1812, simple_loss=0.2642, pruned_loss=0.04905, over 17011.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2514, pruned_loss=0.04499, over 3304981.47 frames. ], batch size: 55, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:33:36,490 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6074, 2.2246, 2.2214, 3.1111, 1.9461, 3.4556, 1.4519, 2.5932], device='cuda:6'), covar=tensor([0.1497, 0.0984, 0.1335, 0.0191, 0.0156, 0.0357, 0.1869, 0.0946], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0177, 0.0196, 0.0196, 0.0206, 0.0217, 0.0205, 0.0195], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 19:33:43,734 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 1.828e+02 2.122e+02 2.508e+02 6.067e+02, threshold=4.243e+02, percent-clipped=1.0 2023-05-01 19:34:17,609 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=237548.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:34:23,846 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=237552.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:34:24,475 INFO [train.py:904] (6/8) Epoch 24, batch 4100, loss[loss=0.1785, simple_loss=0.2716, pruned_loss=0.04274, over 16390.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2536, pruned_loss=0.04475, over 3281914.44 frames. ], batch size: 146, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:34:38,922 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=237563.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 19:35:30,752 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=237596.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:35:35,885 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6581, 2.7372, 2.2352, 2.6137, 3.0238, 2.7486, 3.1557, 3.2890], device='cuda:6'), covar=tensor([0.0114, 0.0418, 0.0560, 0.0433, 0.0276, 0.0385, 0.0272, 0.0254], device='cuda:6'), in_proj_covar=tensor([0.0227, 0.0242, 0.0233, 0.0234, 0.0244, 0.0243, 0.0245, 0.0241], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:35:40,173 INFO [train.py:904] (6/8) Epoch 24, batch 4150, loss[loss=0.2032, simple_loss=0.3051, pruned_loss=0.05066, over 16330.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2601, pruned_loss=0.04738, over 3222506.32 frames. ], batch size: 146, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:36:14,612 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=237624.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 19:36:16,370 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.139e+02 2.639e+02 3.206e+02 5.465e+02, threshold=5.279e+02, percent-clipped=4.0 2023-05-01 19:36:40,065 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-01 19:36:56,392 INFO [train.py:904] (6/8) Epoch 24, batch 4200, loss[loss=0.2058, simple_loss=0.3033, pruned_loss=0.05414, over 16927.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2666, pruned_loss=0.04898, over 3177996.81 frames. ], batch size: 96, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:38:10,756 INFO [train.py:904] (6/8) Epoch 24, batch 4250, loss[loss=0.1781, simple_loss=0.2616, pruned_loss=0.04726, over 11964.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2698, pruned_loss=0.04863, over 3161504.45 frames. ], batch size: 246, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:38:45,326 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.496e+02 2.159e+02 2.541e+02 2.876e+02 4.427e+02, threshold=5.081e+02, percent-clipped=0.0 2023-05-01 19:39:26,152 INFO [train.py:904] (6/8) Epoch 24, batch 4300, loss[loss=0.1844, simple_loss=0.2827, pruned_loss=0.04305, over 16903.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2709, pruned_loss=0.04734, over 3167281.33 frames. ], batch size: 90, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:40:00,938 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1136, 3.0310, 2.9543, 5.2042, 4.0556, 4.3847, 1.9831, 3.3226], device='cuda:6'), covar=tensor([0.1180, 0.0745, 0.1162, 0.0181, 0.0455, 0.0386, 0.1481, 0.0829], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0177, 0.0195, 0.0194, 0.0205, 0.0215, 0.0203, 0.0194], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 19:40:23,411 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=237790.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:40:41,741 INFO [train.py:904] (6/8) Epoch 24, batch 4350, loss[loss=0.1974, simple_loss=0.2869, pruned_loss=0.05398, over 17099.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.274, pruned_loss=0.04818, over 3176351.45 frames. ], batch size: 47, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:41:03,170 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2669, 3.0517, 3.3840, 1.7219, 3.5218, 3.5376, 2.7046, 2.6761], device='cuda:6'), covar=tensor([0.0825, 0.0320, 0.0244, 0.1253, 0.0091, 0.0142, 0.0509, 0.0478], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0111, 0.0101, 0.0141, 0.0083, 0.0129, 0.0130, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 19:41:14,400 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.621e+02 2.331e+02 2.533e+02 3.016e+02 1.010e+03, threshold=5.065e+02, percent-clipped=1.0 2023-05-01 19:41:45,926 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=237847.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:41:49,047 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9587, 4.2018, 4.0332, 4.0650, 3.7851, 3.8072, 3.8164, 4.2090], device='cuda:6'), covar=tensor([0.1017, 0.0854, 0.0981, 0.0760, 0.0725, 0.1747, 0.0879, 0.0923], device='cuda:6'), in_proj_covar=tensor([0.0696, 0.0842, 0.0697, 0.0654, 0.0537, 0.0542, 0.0711, 0.0661], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:41:52,146 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=237851.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:41:54,770 INFO [train.py:904] (6/8) Epoch 24, batch 4400, loss[loss=0.2158, simple_loss=0.3011, pruned_loss=0.06523, over 17209.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.277, pruned_loss=0.04992, over 3169208.63 frames. ], batch size: 44, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:43:05,640 INFO [train.py:904] (6/8) Epoch 24, batch 4450, loss[loss=0.1776, simple_loss=0.2724, pruned_loss=0.04136, over 16845.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2807, pruned_loss=0.05145, over 3173835.38 frames. ], batch size: 102, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:43:18,876 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=237912.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:43:28,618 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=237919.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:43:38,899 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 1.917e+02 2.137e+02 2.680e+02 5.470e+02, threshold=4.273e+02, percent-clipped=1.0 2023-05-01 19:44:16,778 INFO [train.py:904] (6/8) Epoch 24, batch 4500, loss[loss=0.1973, simple_loss=0.28, pruned_loss=0.05727, over 17067.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2815, pruned_loss=0.05228, over 3176394.99 frames. ], batch size: 53, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:44:47,412 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=237973.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:44:59,987 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4277, 4.6802, 4.4717, 4.4929, 4.2319, 4.1927, 4.1781, 4.7332], device='cuda:6'), covar=tensor([0.1135, 0.0836, 0.1049, 0.0798, 0.0786, 0.1405, 0.1052, 0.0851], device='cuda:6'), in_proj_covar=tensor([0.0694, 0.0840, 0.0696, 0.0652, 0.0537, 0.0541, 0.0709, 0.0661], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:45:32,218 INFO [train.py:904] (6/8) Epoch 24, batch 4550, loss[loss=0.2085, simple_loss=0.3001, pruned_loss=0.0584, over 16583.00 frames. ], tot_loss[loss=0.194, simple_loss=0.2822, pruned_loss=0.05291, over 3195931.33 frames. ], batch size: 62, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:45:49,403 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.16 vs. limit=5.0 2023-05-01 19:46:04,569 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.356e+02 1.751e+02 2.009e+02 2.366e+02 4.725e+02, threshold=4.018e+02, percent-clipped=1.0 2023-05-01 19:46:20,467 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-05-01 19:46:44,383 INFO [train.py:904] (6/8) Epoch 24, batch 4600, loss[loss=0.1897, simple_loss=0.278, pruned_loss=0.05073, over 17048.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2833, pruned_loss=0.05317, over 3212536.09 frames. ], batch size: 55, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:46:49,516 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8868, 2.6190, 2.8125, 2.0966, 2.6658, 2.0925, 2.7362, 2.7739], device='cuda:6'), covar=tensor([0.0248, 0.0836, 0.0475, 0.1816, 0.0829, 0.0938, 0.0528, 0.0654], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0167, 0.0168, 0.0154, 0.0146, 0.0131, 0.0143, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 19:47:00,815 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=238064.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:47:15,255 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=238074.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:47:19,076 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 19:47:35,673 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2094, 5.4703, 5.2835, 5.2853, 5.0193, 4.8944, 4.8768, 5.5918], device='cuda:6'), covar=tensor([0.1187, 0.0765, 0.0942, 0.0758, 0.0798, 0.0806, 0.1104, 0.0800], device='cuda:6'), in_proj_covar=tensor([0.0691, 0.0837, 0.0694, 0.0650, 0.0535, 0.0539, 0.0707, 0.0659], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:47:43,951 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4867, 5.4301, 5.3584, 4.9889, 5.1002, 5.4137, 5.2900, 5.0793], device='cuda:6'), covar=tensor([0.0461, 0.0317, 0.0193, 0.0220, 0.0695, 0.0254, 0.0200, 0.0520], device='cuda:6'), in_proj_covar=tensor([0.0298, 0.0444, 0.0349, 0.0349, 0.0352, 0.0402, 0.0239, 0.0420], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:47:50,293 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1780, 3.1154, 1.5910, 3.4570, 2.2916, 3.4387, 1.8779, 2.5085], device='cuda:6'), covar=tensor([0.0331, 0.0443, 0.2258, 0.0224, 0.1047, 0.0482, 0.2035, 0.0939], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0178, 0.0194, 0.0169, 0.0177, 0.0220, 0.0203, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 19:47:53,659 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 19:47:56,650 INFO [train.py:904] (6/8) Epoch 24, batch 4650, loss[loss=0.1836, simple_loss=0.2719, pruned_loss=0.04767, over 16528.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2826, pruned_loss=0.05317, over 3206246.83 frames. ], batch size: 75, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:48:29,678 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=238125.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:48:30,376 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 1.898e+02 2.163e+02 2.416e+02 3.900e+02, threshold=4.325e+02, percent-clipped=0.0 2023-05-01 19:48:30,822 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=238126.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:48:43,751 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=238135.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:48:59,277 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=238146.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:49:00,460 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238147.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:49:09,298 INFO [train.py:904] (6/8) Epoch 24, batch 4700, loss[loss=0.1726, simple_loss=0.2592, pruned_loss=0.04306, over 16728.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2807, pruned_loss=0.05239, over 3195183.94 frames. ], batch size: 124, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:49:58,922 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=238187.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:50:06,582 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-01 19:50:09,168 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238195.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:50:20,662 INFO [train.py:904] (6/8) Epoch 24, batch 4750, loss[loss=0.1546, simple_loss=0.2456, pruned_loss=0.03175, over 16872.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2765, pruned_loss=0.05038, over 3184551.51 frames. ], batch size: 96, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:50:41,759 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2023-05-01 19:50:43,591 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238219.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:50:53,741 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 1.823e+02 2.174e+02 2.542e+02 5.265e+02, threshold=4.348e+02, percent-clipped=2.0 2023-05-01 19:51:31,473 INFO [train.py:904] (6/8) Epoch 24, batch 4800, loss[loss=0.171, simple_loss=0.2638, pruned_loss=0.0391, over 16569.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.2725, pruned_loss=0.04811, over 3193913.73 frames. ], batch size: 68, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:51:49,889 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5066, 3.5630, 2.1794, 4.1159, 2.7066, 4.0110, 2.4005, 2.8083], device='cuda:6'), covar=tensor([0.0324, 0.0423, 0.1792, 0.0168, 0.0918, 0.0579, 0.1590, 0.0865], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0179, 0.0195, 0.0169, 0.0177, 0.0220, 0.0203, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 19:51:52,869 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238267.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:51:54,666 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=238268.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:52:25,776 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1810, 2.2957, 2.3076, 3.9264, 2.2459, 2.6135, 2.3615, 2.4266], device='cuda:6'), covar=tensor([0.1403, 0.3521, 0.2947, 0.0554, 0.4041, 0.2473, 0.3681, 0.3116], device='cuda:6'), in_proj_covar=tensor([0.0409, 0.0460, 0.0375, 0.0331, 0.0440, 0.0528, 0.0429, 0.0537], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:52:46,483 INFO [train.py:904] (6/8) Epoch 24, batch 4850, loss[loss=0.1689, simple_loss=0.2627, pruned_loss=0.03762, over 16863.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2734, pruned_loss=0.04725, over 3194879.77 frames. ], batch size: 96, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:52:53,327 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6679, 4.8487, 5.0912, 5.0468, 5.0907, 4.8310, 4.6816, 4.5548], device='cuda:6'), covar=tensor([0.0426, 0.0586, 0.0446, 0.0473, 0.0572, 0.0424, 0.1277, 0.0507], device='cuda:6'), in_proj_covar=tensor([0.0413, 0.0460, 0.0449, 0.0413, 0.0495, 0.0468, 0.0552, 0.0376], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 19:53:22,732 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.326e+02 1.927e+02 2.251e+02 2.681e+02 4.052e+02, threshold=4.502e+02, percent-clipped=0.0 2023-05-01 19:54:04,041 INFO [train.py:904] (6/8) Epoch 24, batch 4900, loss[loss=0.17, simple_loss=0.2592, pruned_loss=0.04041, over 16604.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2722, pruned_loss=0.04613, over 3170830.62 frames. ], batch size: 62, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:55:17,228 INFO [train.py:904] (6/8) Epoch 24, batch 4950, loss[loss=0.1734, simple_loss=0.2703, pruned_loss=0.03829, over 16716.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2717, pruned_loss=0.04597, over 3157051.05 frames. ], batch size: 89, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:55:41,592 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=238420.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:55:41,724 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=238420.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 19:55:49,726 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.002e+02 2.407e+02 2.827e+02 4.284e+02, threshold=4.815e+02, percent-clipped=0.0 2023-05-01 19:55:57,438 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=238430.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:55:59,969 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4299, 4.4011, 4.3008, 3.2553, 4.3978, 1.5333, 4.1155, 3.9397], device='cuda:6'), covar=tensor([0.0162, 0.0132, 0.0239, 0.0622, 0.0148, 0.3443, 0.0191, 0.0359], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0163, 0.0204, 0.0181, 0.0180, 0.0209, 0.0192, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:56:20,276 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238446.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:56:29,587 INFO [train.py:904] (6/8) Epoch 24, batch 5000, loss[loss=0.1893, simple_loss=0.2737, pruned_loss=0.05241, over 16767.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2731, pruned_loss=0.04588, over 3173343.65 frames. ], batch size: 39, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:56:35,181 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5944, 4.4049, 4.6166, 4.7686, 4.9627, 4.5610, 4.9715, 4.9740], device='cuda:6'), covar=tensor([0.1604, 0.1281, 0.1630, 0.0774, 0.0479, 0.0841, 0.0546, 0.0590], device='cuda:6'), in_proj_covar=tensor([0.0644, 0.0797, 0.0918, 0.0806, 0.0615, 0.0636, 0.0665, 0.0775], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 19:57:11,128 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=238481.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:57:12,056 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=238482.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:57:31,131 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238494.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:57:42,730 INFO [train.py:904] (6/8) Epoch 24, batch 5050, loss[loss=0.204, simple_loss=0.2831, pruned_loss=0.06246, over 12207.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2739, pruned_loss=0.04605, over 3175457.58 frames. ], batch size: 246, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:58:15,454 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 2.006e+02 2.316e+02 2.692e+02 7.292e+02, threshold=4.632e+02, percent-clipped=1.0 2023-05-01 19:58:53,437 INFO [train.py:904] (6/8) Epoch 24, batch 5100, loss[loss=0.1652, simple_loss=0.2596, pruned_loss=0.03541, over 16709.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.2724, pruned_loss=0.04523, over 3174426.68 frames. ], batch size: 62, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:59:16,751 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238568.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:00:07,249 INFO [train.py:904] (6/8) Epoch 24, batch 5150, loss[loss=0.1702, simple_loss=0.2768, pruned_loss=0.03181, over 16866.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2725, pruned_loss=0.04416, over 3188846.90 frames. ], batch size: 116, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:00:26,118 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238616.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:00:36,906 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9201, 2.7697, 2.5792, 4.4496, 3.1906, 3.9511, 1.8136, 3.0656], device='cuda:6'), covar=tensor([0.1251, 0.0752, 0.1171, 0.0121, 0.0218, 0.0373, 0.1512, 0.0751], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0177, 0.0195, 0.0193, 0.0204, 0.0215, 0.0204, 0.0194], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:00:38,784 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.286e+02 1.999e+02 2.272e+02 2.615e+02 4.924e+02, threshold=4.544e+02, percent-clipped=1.0 2023-05-01 20:01:17,675 INFO [train.py:904] (6/8) Epoch 24, batch 5200, loss[loss=0.1631, simple_loss=0.2556, pruned_loss=0.03535, over 16374.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2704, pruned_loss=0.0431, over 3207693.28 frames. ], batch size: 146, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:01:19,724 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 20:01:57,543 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=238681.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:02:18,815 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-01 20:02:19,338 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2087, 3.0528, 3.3403, 1.7575, 3.4633, 3.4744, 2.8174, 2.5862], device='cuda:6'), covar=tensor([0.0830, 0.0254, 0.0173, 0.1202, 0.0081, 0.0148, 0.0401, 0.0537], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0109, 0.0100, 0.0138, 0.0082, 0.0126, 0.0129, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:02:28,268 INFO [train.py:904] (6/8) Epoch 24, batch 5250, loss[loss=0.1765, simple_loss=0.2718, pruned_loss=0.0406, over 16450.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2677, pruned_loss=0.04278, over 3217748.77 frames. ], batch size: 146, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:02:52,136 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238720.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:03:01,283 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.388e+02 1.983e+02 2.265e+02 2.655e+02 9.176e+02, threshold=4.530e+02, percent-clipped=2.0 2023-05-01 20:03:07,601 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238730.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:03:25,259 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=238742.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:03:34,658 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1263, 5.4189, 5.1937, 5.2414, 4.9767, 4.8892, 4.8045, 5.5159], device='cuda:6'), covar=tensor([0.1162, 0.0824, 0.0939, 0.0712, 0.0712, 0.0954, 0.1032, 0.0744], device='cuda:6'), in_proj_covar=tensor([0.0689, 0.0837, 0.0694, 0.0646, 0.0534, 0.0536, 0.0704, 0.0656], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:03:39,613 INFO [train.py:904] (6/8) Epoch 24, batch 5300, loss[loss=0.1538, simple_loss=0.2464, pruned_loss=0.03064, over 16754.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2635, pruned_loss=0.0415, over 3226397.19 frames. ], batch size: 89, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:03:40,301 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-01 20:04:01,107 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238768.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:04:12,325 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=238776.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 20:04:15,112 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238778.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:04:20,290 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238782.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:04:22,749 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3120, 3.8608, 3.8126, 2.3915, 3.3617, 3.8415, 3.3307, 1.9595], device='cuda:6'), covar=tensor([0.0587, 0.0046, 0.0048, 0.0439, 0.0119, 0.0087, 0.0124, 0.0550], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0086, 0.0086, 0.0134, 0.0100, 0.0111, 0.0097, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 20:04:35,037 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-01 20:04:50,564 INFO [train.py:904] (6/8) Epoch 24, batch 5350, loss[loss=0.169, simple_loss=0.2636, pruned_loss=0.03718, over 16835.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2621, pruned_loss=0.04112, over 3220467.29 frames. ], batch size: 83, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:04:56,360 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0663, 5.3426, 5.1152, 5.1476, 4.8719, 4.8315, 4.7822, 5.4472], device='cuda:6'), covar=tensor([0.1171, 0.0851, 0.0937, 0.0843, 0.0808, 0.0875, 0.1011, 0.0888], device='cuda:6'), in_proj_covar=tensor([0.0690, 0.0838, 0.0694, 0.0646, 0.0536, 0.0536, 0.0705, 0.0656], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:05:21,939 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.474e+02 1.926e+02 2.251e+02 2.610e+02 1.024e+03, threshold=4.501e+02, percent-clipped=4.0 2023-05-01 20:05:26,546 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4496, 5.4800, 5.2905, 4.6248, 5.4120, 2.1197, 5.1452, 5.1172], device='cuda:6'), covar=tensor([0.0082, 0.0072, 0.0173, 0.0476, 0.0099, 0.2774, 0.0110, 0.0200], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0164, 0.0205, 0.0182, 0.0180, 0.0210, 0.0192, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:05:27,686 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238830.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:05:40,008 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9648, 3.8863, 4.4931, 2.2760, 4.7574, 4.6891, 3.3404, 3.5669], device='cuda:6'), covar=tensor([0.0724, 0.0263, 0.0159, 0.1121, 0.0048, 0.0108, 0.0381, 0.0402], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0109, 0.0100, 0.0138, 0.0082, 0.0126, 0.0128, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:05:42,990 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7731, 2.4825, 2.5650, 4.9236, 3.6221, 4.1015, 1.6162, 2.9813], device='cuda:6'), covar=tensor([0.1327, 0.0885, 0.1234, 0.0140, 0.0279, 0.0417, 0.1633, 0.0824], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0178, 0.0196, 0.0194, 0.0206, 0.0216, 0.0205, 0.0195], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:05:54,980 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6710, 2.7155, 2.2180, 2.5072, 3.0328, 2.7273, 3.1585, 3.2623], device='cuda:6'), covar=tensor([0.0092, 0.0392, 0.0566, 0.0444, 0.0281, 0.0399, 0.0278, 0.0261], device='cuda:6'), in_proj_covar=tensor([0.0220, 0.0238, 0.0230, 0.0231, 0.0240, 0.0238, 0.0239, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:06:01,195 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0220, 4.0220, 3.9456, 3.1621, 3.9695, 1.7457, 3.7467, 3.4455], device='cuda:6'), covar=tensor([0.0109, 0.0111, 0.0179, 0.0333, 0.0094, 0.3053, 0.0134, 0.0270], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0163, 0.0204, 0.0182, 0.0180, 0.0210, 0.0192, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:06:01,799 INFO [train.py:904] (6/8) Epoch 24, batch 5400, loss[loss=0.1728, simple_loss=0.2678, pruned_loss=0.03891, over 15384.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2647, pruned_loss=0.04168, over 3225772.03 frames. ], batch size: 190, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:06:37,093 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8756, 3.9700, 4.0026, 3.8211, 3.8930, 4.3075, 3.9351, 3.6284], device='cuda:6'), covar=tensor([0.2045, 0.1734, 0.1962, 0.2109, 0.2392, 0.1641, 0.1440, 0.2369], device='cuda:6'), in_proj_covar=tensor([0.0407, 0.0594, 0.0651, 0.0491, 0.0654, 0.0685, 0.0513, 0.0652], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 20:07:17,450 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5413, 2.5424, 2.0594, 2.3058, 2.9047, 2.5321, 3.0575, 3.1580], device='cuda:6'), covar=tensor([0.0104, 0.0408, 0.0578, 0.0487, 0.0285, 0.0416, 0.0248, 0.0242], device='cuda:6'), in_proj_covar=tensor([0.0220, 0.0238, 0.0230, 0.0231, 0.0240, 0.0238, 0.0240, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:07:18,012 INFO [train.py:904] (6/8) Epoch 24, batch 5450, loss[loss=0.1997, simple_loss=0.2987, pruned_loss=0.05034, over 16473.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2669, pruned_loss=0.04274, over 3217166.08 frames. ], batch size: 68, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:07:39,609 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9096, 1.3697, 1.7571, 1.8015, 1.8615, 1.9888, 1.5256, 1.9262], device='cuda:6'), covar=tensor([0.0267, 0.0448, 0.0249, 0.0304, 0.0264, 0.0184, 0.0561, 0.0149], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0195, 0.0182, 0.0186, 0.0202, 0.0161, 0.0200, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:07:54,284 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.137e+02 2.656e+02 3.537e+02 7.468e+02, threshold=5.312e+02, percent-clipped=13.0 2023-05-01 20:08:10,993 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=238936.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 20:08:35,279 INFO [train.py:904] (6/8) Epoch 24, batch 5500, loss[loss=0.2011, simple_loss=0.2898, pruned_loss=0.05621, over 16633.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.274, pruned_loss=0.04737, over 3149105.78 frames. ], batch size: 62, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:09:44,166 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=238997.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 20:09:53,112 INFO [train.py:904] (6/8) Epoch 24, batch 5550, loss[loss=0.1894, simple_loss=0.2791, pruned_loss=0.04988, over 17014.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2819, pruned_loss=0.05288, over 3120769.32 frames. ], batch size: 41, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:10:30,505 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.205e+02 3.032e+02 3.628e+02 4.510e+02 7.943e+02, threshold=7.255e+02, percent-clipped=11.0 2023-05-01 20:10:48,342 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=239037.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:11:12,506 INFO [train.py:904] (6/8) Epoch 24, batch 5600, loss[loss=0.1884, simple_loss=0.2715, pruned_loss=0.05259, over 16550.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2861, pruned_loss=0.05654, over 3103509.93 frames. ], batch size: 57, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:11:52,279 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=239076.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 20:12:26,555 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-05-01 20:12:36,908 INFO [train.py:904] (6/8) Epoch 24, batch 5650, loss[loss=0.2521, simple_loss=0.3325, pruned_loss=0.08583, over 15388.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.2922, pruned_loss=0.0618, over 3052655.34 frames. ], batch size: 191, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:12:54,652 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.99 vs. limit=5.0 2023-05-01 20:13:11,117 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=239124.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 20:13:14,764 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.394e+02 3.339e+02 4.323e+02 5.614e+02 1.229e+03, threshold=8.646e+02, percent-clipped=9.0 2023-05-01 20:13:47,385 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1848, 4.3356, 4.4732, 4.2332, 4.3243, 4.8238, 4.3533, 4.0654], device='cuda:6'), covar=tensor([0.1652, 0.1987, 0.2298, 0.2045, 0.2392, 0.1127, 0.1698, 0.2601], device='cuda:6'), in_proj_covar=tensor([0.0411, 0.0602, 0.0661, 0.0496, 0.0660, 0.0692, 0.0519, 0.0659], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 20:13:55,705 INFO [train.py:904] (6/8) Epoch 24, batch 5700, loss[loss=0.198, simple_loss=0.2968, pruned_loss=0.04962, over 16380.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2943, pruned_loss=0.06323, over 3050412.39 frames. ], batch size: 146, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:15:14,246 INFO [train.py:904] (6/8) Epoch 24, batch 5750, loss[loss=0.2178, simple_loss=0.3014, pruned_loss=0.06716, over 15348.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2964, pruned_loss=0.06418, over 3043537.81 frames. ], batch size: 191, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:15:53,491 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.857e+02 3.000e+02 3.580e+02 4.341e+02 9.766e+02, threshold=7.159e+02, percent-clipped=1.0 2023-05-01 20:16:37,341 INFO [train.py:904] (6/8) Epoch 24, batch 5800, loss[loss=0.1728, simple_loss=0.2711, pruned_loss=0.0373, over 16832.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.2956, pruned_loss=0.06298, over 3036273.73 frames. ], batch size: 96, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:17:40,240 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=239292.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 20:17:56,784 INFO [train.py:904] (6/8) Epoch 24, batch 5850, loss[loss=0.2385, simple_loss=0.3105, pruned_loss=0.08323, over 11231.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2931, pruned_loss=0.06103, over 3052056.22 frames. ], batch size: 246, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:18:33,898 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.882e+02 2.891e+02 3.438e+02 4.490e+02 7.448e+02, threshold=6.875e+02, percent-clipped=1.0 2023-05-01 20:18:52,952 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=239337.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:19:19,097 INFO [train.py:904] (6/8) Epoch 24, batch 5900, loss[loss=0.2187, simple_loss=0.2914, pruned_loss=0.073, over 11649.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2926, pruned_loss=0.06083, over 3049872.73 frames. ], batch size: 246, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:19:50,737 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.37 vs. limit=5.0 2023-05-01 20:20:10,306 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5124, 3.4352, 3.4710, 2.6274, 3.3834, 2.0722, 3.2114, 2.7394], device='cuda:6'), covar=tensor([0.0176, 0.0152, 0.0201, 0.0223, 0.0112, 0.2503, 0.0156, 0.0272], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0164, 0.0205, 0.0182, 0.0180, 0.0210, 0.0192, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:20:14,709 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=239385.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:20:29,941 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7509, 1.7954, 1.6384, 1.4326, 1.9208, 1.5768, 1.6065, 1.8894], device='cuda:6'), covar=tensor([0.0226, 0.0295, 0.0406, 0.0375, 0.0224, 0.0283, 0.0201, 0.0214], device='cuda:6'), in_proj_covar=tensor([0.0219, 0.0237, 0.0230, 0.0230, 0.0239, 0.0237, 0.0239, 0.0236], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:20:42,777 INFO [train.py:904] (6/8) Epoch 24, batch 5950, loss[loss=0.2314, simple_loss=0.3047, pruned_loss=0.0791, over 11615.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2933, pruned_loss=0.05988, over 3036327.89 frames. ], batch size: 247, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:21:03,341 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5427, 3.5864, 2.2651, 4.1538, 2.8236, 4.0965, 2.2799, 2.8682], device='cuda:6'), covar=tensor([0.0320, 0.0422, 0.1618, 0.0245, 0.0808, 0.0615, 0.1655, 0.0877], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0179, 0.0195, 0.0168, 0.0178, 0.0218, 0.0204, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:21:12,947 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8301, 2.6713, 2.6251, 1.8464, 2.5611, 2.6671, 2.5255, 1.9117], device='cuda:6'), covar=tensor([0.0495, 0.0105, 0.0101, 0.0434, 0.0150, 0.0131, 0.0144, 0.0415], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0087, 0.0086, 0.0134, 0.0099, 0.0111, 0.0097, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 20:21:21,253 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.785e+02 2.477e+02 3.014e+02 3.724e+02 7.833e+02, threshold=6.029e+02, percent-clipped=2.0 2023-05-01 20:22:03,335 INFO [train.py:904] (6/8) Epoch 24, batch 6000, loss[loss=0.1904, simple_loss=0.2745, pruned_loss=0.05317, over 17049.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2922, pruned_loss=0.05935, over 3057765.83 frames. ], batch size: 53, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:22:03,335 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 20:22:14,269 INFO [train.py:938] (6/8) Epoch 24, validation: loss=0.1493, simple_loss=0.2618, pruned_loss=0.01837, over 944034.00 frames. 2023-05-01 20:22:14,269 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 20:22:36,002 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6261, 3.0950, 3.2473, 1.9826, 2.7934, 2.0867, 3.2783, 3.3168], device='cuda:6'), covar=tensor([0.0259, 0.0782, 0.0573, 0.2066, 0.0862, 0.1083, 0.0598, 0.0870], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0166, 0.0168, 0.0154, 0.0146, 0.0131, 0.0144, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:22:46,207 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3406, 4.3045, 4.1999, 3.4313, 4.3054, 1.7167, 4.0567, 3.8421], device='cuda:6'), covar=tensor([0.0127, 0.0101, 0.0206, 0.0334, 0.0100, 0.2939, 0.0148, 0.0280], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0163, 0.0204, 0.0180, 0.0179, 0.0209, 0.0191, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:22:48,288 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8176, 3.2313, 3.3020, 2.0057, 2.8031, 2.0713, 3.3483, 3.4386], device='cuda:6'), covar=tensor([0.0263, 0.0763, 0.0605, 0.2140, 0.0899, 0.1056, 0.0645, 0.0953], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0166, 0.0169, 0.0155, 0.0146, 0.0131, 0.0144, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:23:24,003 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9389, 3.2800, 3.4765, 2.0087, 2.9069, 2.2179, 3.4896, 3.5213], device='cuda:6'), covar=tensor([0.0253, 0.0794, 0.0634, 0.2127, 0.0867, 0.1063, 0.0560, 0.0892], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0166, 0.0168, 0.0155, 0.0146, 0.0131, 0.0144, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:23:24,371 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.00 vs. limit=5.0 2023-05-01 20:23:32,266 INFO [train.py:904] (6/8) Epoch 24, batch 6050, loss[loss=0.2068, simple_loss=0.2978, pruned_loss=0.05789, over 16880.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2907, pruned_loss=0.05809, over 3086350.82 frames. ], batch size: 109, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:24:09,878 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.866e+02 2.713e+02 3.363e+02 4.065e+02 7.660e+02, threshold=6.725e+02, percent-clipped=3.0 2023-05-01 20:24:21,732 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-01 20:24:49,385 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-01 20:24:51,441 INFO [train.py:904] (6/8) Epoch 24, batch 6100, loss[loss=0.1875, simple_loss=0.2778, pruned_loss=0.04859, over 16683.00 frames. ], tot_loss[loss=0.202, simple_loss=0.29, pruned_loss=0.05703, over 3087615.62 frames. ], batch size: 89, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:25:01,462 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4709, 4.0129, 4.0818, 2.6867, 3.6325, 4.0764, 3.6482, 2.3859], device='cuda:6'), covar=tensor([0.0545, 0.0060, 0.0052, 0.0405, 0.0112, 0.0123, 0.0108, 0.0440], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0087, 0.0087, 0.0134, 0.0099, 0.0111, 0.0096, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 20:25:29,266 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8376, 1.4511, 1.7346, 1.7337, 1.8025, 1.9425, 1.6482, 1.7931], device='cuda:6'), covar=tensor([0.0299, 0.0405, 0.0225, 0.0327, 0.0281, 0.0190, 0.0445, 0.0151], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0194, 0.0183, 0.0185, 0.0202, 0.0160, 0.0200, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:25:56,325 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=239592.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 20:26:14,091 INFO [train.py:904] (6/8) Epoch 24, batch 6150, loss[loss=0.1995, simple_loss=0.2926, pruned_loss=0.05324, over 16860.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2882, pruned_loss=0.05664, over 3094672.67 frames. ], batch size: 102, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:26:31,259 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9641, 3.8530, 4.0233, 4.1490, 4.2471, 3.8470, 4.1906, 4.2540], device='cuda:6'), covar=tensor([0.1713, 0.1288, 0.1461, 0.0691, 0.0563, 0.1662, 0.0956, 0.0739], device='cuda:6'), in_proj_covar=tensor([0.0644, 0.0791, 0.0912, 0.0801, 0.0611, 0.0631, 0.0664, 0.0771], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:26:45,900 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=239622.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:26:53,237 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.890e+02 2.573e+02 3.097e+02 3.625e+02 7.083e+02, threshold=6.193e+02, percent-clipped=1.0 2023-05-01 20:27:14,513 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=239640.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 20:27:34,955 INFO [train.py:904] (6/8) Epoch 24, batch 6200, loss[loss=0.1713, simple_loss=0.2667, pruned_loss=0.03795, over 16697.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2862, pruned_loss=0.05631, over 3074597.63 frames. ], batch size: 89, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:28:06,624 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3279, 3.4304, 3.5854, 3.5621, 3.5765, 3.4126, 3.4442, 3.4848], device='cuda:6'), covar=tensor([0.0424, 0.0664, 0.0448, 0.0431, 0.0514, 0.0515, 0.0824, 0.0505], device='cuda:6'), in_proj_covar=tensor([0.0420, 0.0470, 0.0457, 0.0418, 0.0503, 0.0475, 0.0559, 0.0380], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 20:28:22,554 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=239683.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:28:26,381 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9335, 4.7234, 4.9255, 5.1195, 5.3239, 4.7607, 5.3172, 5.3055], device='cuda:6'), covar=tensor([0.1900, 0.1294, 0.1853, 0.0799, 0.0554, 0.0835, 0.0677, 0.0645], device='cuda:6'), in_proj_covar=tensor([0.0642, 0.0790, 0.0912, 0.0801, 0.0611, 0.0631, 0.0663, 0.0770], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:28:26,626 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-05-01 20:28:52,025 INFO [train.py:904] (6/8) Epoch 24, batch 6250, loss[loss=0.1995, simple_loss=0.2856, pruned_loss=0.05665, over 15591.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2864, pruned_loss=0.05633, over 3083718.10 frames. ], batch size: 191, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:29:29,984 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.754e+02 2.614e+02 3.057e+02 3.659e+02 7.500e+02, threshold=6.114e+02, percent-clipped=2.0 2023-05-01 20:30:05,176 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9627, 4.2102, 4.0165, 4.0652, 3.7497, 3.8101, 3.8256, 4.1880], device='cuda:6'), covar=tensor([0.1101, 0.0887, 0.0970, 0.0882, 0.0830, 0.1624, 0.0971, 0.1049], device='cuda:6'), in_proj_covar=tensor([0.0689, 0.0829, 0.0689, 0.0643, 0.0530, 0.0533, 0.0699, 0.0651], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:30:05,922 INFO [train.py:904] (6/8) Epoch 24, batch 6300, loss[loss=0.2158, simple_loss=0.2915, pruned_loss=0.0701, over 11701.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2859, pruned_loss=0.05539, over 3099080.13 frames. ], batch size: 246, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:31:09,729 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 20:31:24,178 INFO [train.py:904] (6/8) Epoch 24, batch 6350, loss[loss=0.1844, simple_loss=0.2781, pruned_loss=0.04533, over 16665.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2869, pruned_loss=0.05637, over 3107605.03 frames. ], batch size: 134, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:32:03,926 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.770e+02 2.790e+02 3.396e+02 4.270e+02 1.026e+03, threshold=6.791e+02, percent-clipped=1.0 2023-05-01 20:32:42,127 INFO [train.py:904] (6/8) Epoch 24, batch 6400, loss[loss=0.2585, simple_loss=0.3264, pruned_loss=0.09527, over 11599.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2873, pruned_loss=0.05755, over 3094904.62 frames. ], batch size: 248, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:32:59,835 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2548, 3.0623, 3.4510, 1.8094, 3.5649, 3.5875, 2.8507, 2.6379], device='cuda:6'), covar=tensor([0.0891, 0.0315, 0.0198, 0.1276, 0.0086, 0.0209, 0.0452, 0.0559], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0110, 0.0100, 0.0139, 0.0083, 0.0129, 0.0129, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:33:58,190 INFO [train.py:904] (6/8) Epoch 24, batch 6450, loss[loss=0.2037, simple_loss=0.2936, pruned_loss=0.05686, over 17214.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2873, pruned_loss=0.0571, over 3105184.74 frames. ], batch size: 44, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:34:21,894 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7714, 1.8357, 2.3538, 2.6549, 2.6182, 3.0170, 1.9497, 2.9946], device='cuda:6'), covar=tensor([0.0207, 0.0536, 0.0346, 0.0336, 0.0338, 0.0181, 0.0574, 0.0142], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0193, 0.0181, 0.0184, 0.0201, 0.0159, 0.0199, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:34:37,436 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.694e+02 2.672e+02 3.087e+02 3.802e+02 7.341e+02, threshold=6.174e+02, percent-clipped=2.0 2023-05-01 20:35:16,099 INFO [train.py:904] (6/8) Epoch 24, batch 6500, loss[loss=0.1849, simple_loss=0.2742, pruned_loss=0.04774, over 16690.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2849, pruned_loss=0.0562, over 3105158.24 frames. ], batch size: 57, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:35:55,168 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=239978.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:36:27,571 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 20:36:39,174 INFO [train.py:904] (6/8) Epoch 24, batch 6550, loss[loss=0.1815, simple_loss=0.2786, pruned_loss=0.04219, over 17203.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2876, pruned_loss=0.05661, over 3104439.18 frames. ], batch size: 44, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:37:16,973 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 2.521e+02 3.197e+02 3.720e+02 9.360e+02, threshold=6.395e+02, percent-clipped=2.0 2023-05-01 20:37:36,973 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8383, 3.9792, 2.5632, 4.6877, 3.1190, 4.5600, 2.4852, 3.1842], device='cuda:6'), covar=tensor([0.0310, 0.0361, 0.1613, 0.0242, 0.0784, 0.0488, 0.1566, 0.0772], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0178, 0.0195, 0.0167, 0.0177, 0.0217, 0.0203, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:37:54,317 INFO [train.py:904] (6/8) Epoch 24, batch 6600, loss[loss=0.212, simple_loss=0.2969, pruned_loss=0.06358, over 16839.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2896, pruned_loss=0.05723, over 3099877.67 frames. ], batch size: 42, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:38:18,432 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5237, 3.4588, 3.4705, 2.7077, 3.3594, 2.0555, 3.1473, 2.7289], device='cuda:6'), covar=tensor([0.0163, 0.0154, 0.0189, 0.0241, 0.0112, 0.2416, 0.0148, 0.0253], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0165, 0.0207, 0.0182, 0.0181, 0.0212, 0.0193, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:39:11,701 INFO [train.py:904] (6/8) Epoch 24, batch 6650, loss[loss=0.204, simple_loss=0.2844, pruned_loss=0.06178, over 16532.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2903, pruned_loss=0.05833, over 3094853.66 frames. ], batch size: 57, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:39:50,364 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.847e+02 2.925e+02 3.532e+02 4.365e+02 9.170e+02, threshold=7.065e+02, percent-clipped=1.0 2023-05-01 20:40:28,847 INFO [train.py:904] (6/8) Epoch 24, batch 6700, loss[loss=0.1882, simple_loss=0.2816, pruned_loss=0.04737, over 16689.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2889, pruned_loss=0.05863, over 3086070.11 frames. ], batch size: 57, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:41:45,714 INFO [train.py:904] (6/8) Epoch 24, batch 6750, loss[loss=0.1971, simple_loss=0.287, pruned_loss=0.05365, over 16732.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2876, pruned_loss=0.05851, over 3072908.37 frames. ], batch size: 76, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:42:23,537 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.933e+02 2.874e+02 3.326e+02 4.186e+02 6.587e+02, threshold=6.652e+02, percent-clipped=0.0 2023-05-01 20:43:01,407 INFO [train.py:904] (6/8) Epoch 24, batch 6800, loss[loss=0.1995, simple_loss=0.2947, pruned_loss=0.05212, over 16739.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2867, pruned_loss=0.05771, over 3098597.75 frames. ], batch size: 76, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:43:33,042 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2023-05-01 20:43:42,194 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=240278.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:44:06,639 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5407, 3.4989, 3.4816, 2.7284, 3.4071, 2.0107, 3.1952, 2.8055], device='cuda:6'), covar=tensor([0.0177, 0.0157, 0.0210, 0.0274, 0.0120, 0.2636, 0.0152, 0.0283], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0165, 0.0207, 0.0183, 0.0181, 0.0212, 0.0194, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:44:07,898 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8497, 2.7381, 2.6018, 4.5464, 3.2231, 4.0798, 1.8294, 2.9914], device='cuda:6'), covar=tensor([0.1321, 0.0760, 0.1168, 0.0166, 0.0324, 0.0421, 0.1456, 0.0800], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0176, 0.0195, 0.0193, 0.0205, 0.0216, 0.0203, 0.0194], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:44:21,119 INFO [train.py:904] (6/8) Epoch 24, batch 6850, loss[loss=0.2002, simple_loss=0.2912, pruned_loss=0.0546, over 15413.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2873, pruned_loss=0.05778, over 3102789.97 frames. ], batch size: 190, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:44:56,066 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=240326.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:45:00,028 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.623e+02 3.086e+02 3.760e+02 7.465e+02, threshold=6.171e+02, percent-clipped=2.0 2023-05-01 20:45:17,539 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=240340.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:45:35,444 INFO [train.py:904] (6/8) Epoch 24, batch 6900, loss[loss=0.2236, simple_loss=0.312, pruned_loss=0.06763, over 16462.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2898, pruned_loss=0.05746, over 3106289.02 frames. ], batch size: 146, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:46:26,550 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-01 20:46:33,641 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.64 vs. limit=5.0 2023-05-01 20:46:50,877 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=240401.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 20:46:52,761 INFO [train.py:904] (6/8) Epoch 24, batch 6950, loss[loss=0.2578, simple_loss=0.321, pruned_loss=0.09726, over 11244.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.292, pruned_loss=0.05991, over 3082808.16 frames. ], batch size: 248, lr: 2.80e-03, grad_scale: 2.0 2023-05-01 20:47:33,339 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.139e+02 2.974e+02 3.564e+02 4.384e+02 8.202e+02, threshold=7.128e+02, percent-clipped=2.0 2023-05-01 20:48:07,679 INFO [train.py:904] (6/8) Epoch 24, batch 7000, loss[loss=0.1903, simple_loss=0.2852, pruned_loss=0.04767, over 17113.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2921, pruned_loss=0.05865, over 3104863.86 frames. ], batch size: 47, lr: 2.80e-03, grad_scale: 2.0 2023-05-01 20:48:32,181 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0822, 4.1175, 4.4252, 4.3787, 4.3978, 4.1538, 4.1247, 4.0750], device='cuda:6'), covar=tensor([0.0370, 0.0609, 0.0464, 0.0496, 0.0528, 0.0515, 0.0966, 0.0596], device='cuda:6'), in_proj_covar=tensor([0.0415, 0.0464, 0.0451, 0.0416, 0.0497, 0.0472, 0.0555, 0.0377], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 20:48:49,505 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6033, 3.2765, 3.7253, 1.9338, 3.8849, 3.9190, 2.9655, 2.8863], device='cuda:6'), covar=tensor([0.0750, 0.0280, 0.0221, 0.1217, 0.0082, 0.0154, 0.0462, 0.0487], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0109, 0.0100, 0.0138, 0.0082, 0.0128, 0.0129, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:49:14,332 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0087, 2.1832, 2.2457, 3.6315, 2.0798, 2.4805, 2.2510, 2.3107], device='cuda:6'), covar=tensor([0.1529, 0.3588, 0.3095, 0.0639, 0.4361, 0.2508, 0.3746, 0.3527], device='cuda:6'), in_proj_covar=tensor([0.0408, 0.0457, 0.0375, 0.0330, 0.0440, 0.0523, 0.0429, 0.0535], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:49:21,388 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0760, 5.0750, 4.8528, 4.1954, 5.0019, 1.8409, 4.7096, 4.5811], device='cuda:6'), covar=tensor([0.0104, 0.0092, 0.0216, 0.0440, 0.0098, 0.2857, 0.0145, 0.0248], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0165, 0.0207, 0.0183, 0.0181, 0.0213, 0.0194, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 20:49:23,902 INFO [train.py:904] (6/8) Epoch 24, batch 7050, loss[loss=0.2075, simple_loss=0.2916, pruned_loss=0.06174, over 15307.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2928, pruned_loss=0.05834, over 3116959.10 frames. ], batch size: 190, lr: 2.80e-03, grad_scale: 2.0 2023-05-01 20:49:24,955 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6515, 4.6873, 5.0451, 4.9894, 5.0275, 4.6908, 4.6629, 4.4934], device='cuda:6'), covar=tensor([0.0337, 0.0544, 0.0377, 0.0420, 0.0534, 0.0420, 0.1008, 0.0585], device='cuda:6'), in_proj_covar=tensor([0.0416, 0.0465, 0.0451, 0.0416, 0.0498, 0.0473, 0.0555, 0.0377], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 20:50:06,654 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.746e+02 2.536e+02 3.115e+02 4.008e+02 6.614e+02, threshold=6.229e+02, percent-clipped=0.0 2023-05-01 20:50:42,263 INFO [train.py:904] (6/8) Epoch 24, batch 7100, loss[loss=0.2375, simple_loss=0.3086, pruned_loss=0.08322, over 11744.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2913, pruned_loss=0.05825, over 3104025.28 frames. ], batch size: 246, lr: 2.80e-03, grad_scale: 2.0 2023-05-01 20:51:59,029 INFO [train.py:904] (6/8) Epoch 24, batch 7150, loss[loss=0.1974, simple_loss=0.2886, pruned_loss=0.05309, over 16681.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2893, pruned_loss=0.05768, over 3121893.69 frames. ], batch size: 83, lr: 2.80e-03, grad_scale: 2.0 2023-05-01 20:52:39,163 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.962e+02 2.941e+02 3.518e+02 4.081e+02 6.999e+02, threshold=7.036e+02, percent-clipped=1.0 2023-05-01 20:52:49,485 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-05-01 20:53:12,641 INFO [train.py:904] (6/8) Epoch 24, batch 7200, loss[loss=0.1779, simple_loss=0.2719, pruned_loss=0.0419, over 15453.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2875, pruned_loss=0.05663, over 3107967.94 frames. ], batch size: 191, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:54:21,085 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=240696.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 20:54:32,008 INFO [train.py:904] (6/8) Epoch 24, batch 7250, loss[loss=0.1916, simple_loss=0.2807, pruned_loss=0.05126, over 16941.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2858, pruned_loss=0.05602, over 3099675.28 frames. ], batch size: 109, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 20:54:47,675 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 20:55:12,201 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.642e+02 3.264e+02 4.366e+02 7.559e+02, threshold=6.528e+02, percent-clipped=1.0 2023-05-01 20:55:44,643 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-05-01 20:55:45,057 INFO [train.py:904] (6/8) Epoch 24, batch 7300, loss[loss=0.1888, simple_loss=0.2828, pruned_loss=0.04739, over 16696.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2858, pruned_loss=0.05638, over 3103311.29 frames. ], batch size: 134, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 20:56:43,335 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8483, 5.2367, 5.4292, 5.1722, 5.2617, 5.7603, 5.2382, 4.9964], device='cuda:6'), covar=tensor([0.1022, 0.1736, 0.2070, 0.1752, 0.2134, 0.0875, 0.1582, 0.2432], device='cuda:6'), in_proj_covar=tensor([0.0413, 0.0606, 0.0670, 0.0498, 0.0660, 0.0698, 0.0521, 0.0665], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 20:56:50,224 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8929, 2.7303, 2.6152, 1.9045, 2.5904, 2.6871, 2.5388, 1.8741], device='cuda:6'), covar=tensor([0.0453, 0.0091, 0.0102, 0.0407, 0.0147, 0.0128, 0.0139, 0.0424], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0086, 0.0086, 0.0134, 0.0098, 0.0110, 0.0095, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 20:57:02,849 INFO [train.py:904] (6/8) Epoch 24, batch 7350, loss[loss=0.2004, simple_loss=0.2873, pruned_loss=0.05678, over 15506.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2871, pruned_loss=0.05784, over 3052222.15 frames. ], batch size: 191, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 20:57:30,463 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-01 20:57:44,658 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.796e+02 2.735e+02 3.201e+02 4.098e+02 9.519e+02, threshold=6.402e+02, percent-clipped=6.0 2023-05-01 20:58:18,665 INFO [train.py:904] (6/8) Epoch 24, batch 7400, loss[loss=0.2557, simple_loss=0.3132, pruned_loss=0.09912, over 11241.00 frames. ], tot_loss[loss=0.2013, simple_loss=0.2876, pruned_loss=0.05749, over 3081355.90 frames. ], batch size: 248, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 20:58:19,383 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2023-05-01 20:58:59,128 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6212, 2.4886, 2.2209, 3.6702, 2.2147, 3.6681, 1.5082, 2.5303], device='cuda:6'), covar=tensor([0.1555, 0.0906, 0.1493, 0.0207, 0.0214, 0.0503, 0.1880, 0.1025], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0177, 0.0197, 0.0193, 0.0206, 0.0216, 0.0205, 0.0195], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:59:15,622 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1746, 3.1839, 2.0665, 3.4531, 2.4521, 3.4848, 1.9894, 2.5699], device='cuda:6'), covar=tensor([0.0334, 0.0412, 0.1564, 0.0250, 0.0847, 0.0607, 0.1772, 0.0927], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0177, 0.0195, 0.0166, 0.0178, 0.0217, 0.0203, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 20:59:34,862 INFO [train.py:904] (6/8) Epoch 24, batch 7450, loss[loss=0.1847, simple_loss=0.2676, pruned_loss=0.05093, over 17046.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2894, pruned_loss=0.05857, over 3078922.67 frames. ], batch size: 50, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:00:19,560 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.977e+02 2.843e+02 3.471e+02 4.367e+02 9.441e+02, threshold=6.943e+02, percent-clipped=5.0 2023-05-01 21:00:22,625 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=240932.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 21:00:55,587 INFO [train.py:904] (6/8) Epoch 24, batch 7500, loss[loss=0.1805, simple_loss=0.2734, pruned_loss=0.04381, over 16738.00 frames. ], tot_loss[loss=0.202, simple_loss=0.289, pruned_loss=0.0575, over 3084750.47 frames. ], batch size: 89, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:01:46,091 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=240986.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:01:57,402 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=240993.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 21:02:00,975 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=240996.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 21:02:11,647 INFO [train.py:904] (6/8) Epoch 24, batch 7550, loss[loss=0.2198, simple_loss=0.2871, pruned_loss=0.0762, over 11501.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2885, pruned_loss=0.05802, over 3071176.18 frames. ], batch size: 246, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:02:53,683 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.791e+02 2.697e+02 3.200e+02 3.848e+02 7.597e+02, threshold=6.400e+02, percent-clipped=1.0 2023-05-01 21:03:15,838 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=241044.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:03:20,300 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=241047.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:03:29,244 INFO [train.py:904] (6/8) Epoch 24, batch 7600, loss[loss=0.2321, simple_loss=0.3011, pruned_loss=0.08152, over 11507.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2877, pruned_loss=0.05844, over 3045371.34 frames. ], batch size: 247, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:04:47,281 INFO [train.py:904] (6/8) Epoch 24, batch 7650, loss[loss=0.1867, simple_loss=0.2789, pruned_loss=0.04722, over 16567.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2879, pruned_loss=0.05873, over 3058062.21 frames. ], batch size: 68, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:05:30,616 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.802e+02 2.751e+02 3.488e+02 4.821e+02 1.001e+03, threshold=6.976e+02, percent-clipped=5.0 2023-05-01 21:06:04,717 INFO [train.py:904] (6/8) Epoch 24, batch 7700, loss[loss=0.2049, simple_loss=0.2866, pruned_loss=0.06163, over 17042.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2874, pruned_loss=0.059, over 3053080.55 frames. ], batch size: 55, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:07:23,955 INFO [train.py:904] (6/8) Epoch 24, batch 7750, loss[loss=0.2184, simple_loss=0.3115, pruned_loss=0.06264, over 16355.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2875, pruned_loss=0.0585, over 3061026.68 frames. ], batch size: 146, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:08:06,386 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.032e+02 2.844e+02 3.408e+02 3.883e+02 6.664e+02, threshold=6.815e+02, percent-clipped=0.0 2023-05-01 21:08:38,514 INFO [train.py:904] (6/8) Epoch 24, batch 7800, loss[loss=0.1702, simple_loss=0.2628, pruned_loss=0.03879, over 16526.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2883, pruned_loss=0.05908, over 3073837.53 frames. ], batch size: 68, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:08:53,109 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6602, 2.8408, 2.6034, 4.4504, 3.1287, 4.0769, 1.6325, 2.8724], device='cuda:6'), covar=tensor([0.1507, 0.0773, 0.1242, 0.0204, 0.0277, 0.0398, 0.1760, 0.0881], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0177, 0.0197, 0.0193, 0.0205, 0.0216, 0.0204, 0.0195], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 21:09:34,093 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=241288.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 21:09:56,310 INFO [train.py:904] (6/8) Epoch 24, batch 7850, loss[loss=0.1932, simple_loss=0.2797, pruned_loss=0.05331, over 16739.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2893, pruned_loss=0.05865, over 3072962.78 frames. ], batch size: 124, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:10:01,954 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4461, 2.4845, 2.3734, 4.1361, 2.2529, 2.8340, 2.5335, 2.6506], device='cuda:6'), covar=tensor([0.1291, 0.3608, 0.2987, 0.0509, 0.4169, 0.2425, 0.3529, 0.3192], device='cuda:6'), in_proj_covar=tensor([0.0404, 0.0455, 0.0372, 0.0328, 0.0438, 0.0520, 0.0427, 0.0531], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 21:10:38,474 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.020e+02 2.608e+02 3.095e+02 3.566e+02 5.691e+02, threshold=6.191e+02, percent-clipped=0.0 2023-05-01 21:10:55,522 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=241342.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:11:08,786 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8935, 2.7175, 2.8333, 2.1411, 2.7413, 2.1451, 2.7033, 2.9020], device='cuda:6'), covar=tensor([0.0260, 0.0802, 0.0476, 0.1701, 0.0739, 0.0930, 0.0561, 0.0752], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0165, 0.0167, 0.0154, 0.0145, 0.0131, 0.0143, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 21:11:11,149 INFO [train.py:904] (6/8) Epoch 24, batch 7900, loss[loss=0.1967, simple_loss=0.2849, pruned_loss=0.05424, over 16734.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2882, pruned_loss=0.05798, over 3085614.65 frames. ], batch size: 62, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:11:39,033 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-01 21:12:29,004 INFO [train.py:904] (6/8) Epoch 24, batch 7950, loss[loss=0.1806, simple_loss=0.271, pruned_loss=0.04508, over 16844.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2885, pruned_loss=0.05848, over 3088393.72 frames. ], batch size: 116, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:12:35,055 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=241406.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:13:12,617 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.937e+02 2.663e+02 3.306e+02 3.929e+02 7.099e+02, threshold=6.611e+02, percent-clipped=2.0 2023-05-01 21:13:24,883 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9827, 5.5035, 5.6513, 5.3140, 5.4181, 5.9880, 5.3879, 5.1352], device='cuda:6'), covar=tensor([0.0936, 0.1539, 0.2046, 0.1952, 0.2392, 0.0883, 0.1590, 0.2426], device='cuda:6'), in_proj_covar=tensor([0.0415, 0.0611, 0.0676, 0.0501, 0.0662, 0.0697, 0.0524, 0.0668], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 21:13:31,992 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6316, 1.7909, 1.5235, 1.4268, 1.8210, 1.5157, 1.6014, 1.8826], device='cuda:6'), covar=tensor([0.0242, 0.0354, 0.0513, 0.0412, 0.0280, 0.0310, 0.0211, 0.0261], device='cuda:6'), in_proj_covar=tensor([0.0211, 0.0233, 0.0225, 0.0226, 0.0235, 0.0231, 0.0233, 0.0230], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 21:13:44,936 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0471, 3.0420, 1.9290, 3.2465, 2.3226, 3.3182, 2.1622, 2.5844], device='cuda:6'), covar=tensor([0.0303, 0.0425, 0.1667, 0.0265, 0.0888, 0.0639, 0.1463, 0.0804], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0179, 0.0196, 0.0167, 0.0179, 0.0219, 0.0204, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 21:13:46,678 INFO [train.py:904] (6/8) Epoch 24, batch 8000, loss[loss=0.1937, simple_loss=0.2871, pruned_loss=0.05016, over 16423.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2887, pruned_loss=0.05828, over 3104099.81 frames. ], batch size: 146, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:14:10,184 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=241467.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 21:15:04,288 INFO [train.py:904] (6/8) Epoch 24, batch 8050, loss[loss=0.1776, simple_loss=0.268, pruned_loss=0.04358, over 16702.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2888, pruned_loss=0.05806, over 3108438.77 frames. ], batch size: 124, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:15:47,625 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.925e+02 2.671e+02 3.058e+02 3.888e+02 6.821e+02, threshold=6.115e+02, percent-clipped=1.0 2023-05-01 21:16:07,754 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4303, 4.6802, 4.4799, 4.4516, 4.2131, 4.2210, 4.1767, 4.7254], device='cuda:6'), covar=tensor([0.1119, 0.0830, 0.0954, 0.0914, 0.0843, 0.1428, 0.1070, 0.0912], device='cuda:6'), in_proj_covar=tensor([0.0693, 0.0827, 0.0692, 0.0645, 0.0527, 0.0534, 0.0700, 0.0651], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 21:16:22,148 INFO [train.py:904] (6/8) Epoch 24, batch 8100, loss[loss=0.1757, simple_loss=0.2588, pruned_loss=0.04624, over 17267.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2878, pruned_loss=0.05744, over 3120451.07 frames. ], batch size: 52, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:17:14,358 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=241588.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 21:17:38,324 INFO [train.py:904] (6/8) Epoch 24, batch 8150, loss[loss=0.1784, simple_loss=0.2716, pruned_loss=0.0426, over 16401.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2861, pruned_loss=0.05678, over 3115001.02 frames. ], batch size: 146, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:17:55,348 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-01 21:18:06,817 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=241621.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:18:22,058 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.026e+02 2.691e+02 3.345e+02 4.318e+02 1.089e+03, threshold=6.691e+02, percent-clipped=5.0 2023-05-01 21:18:30,850 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=241636.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 21:18:39,504 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=241642.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:18:55,841 INFO [train.py:904] (6/8) Epoch 24, batch 8200, loss[loss=0.1927, simple_loss=0.2857, pruned_loss=0.04986, over 16261.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2843, pruned_loss=0.05643, over 3106953.37 frames. ], batch size: 165, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:19:38,373 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3676, 3.3202, 3.4341, 3.4999, 3.5634, 3.2922, 3.4915, 3.6034], device='cuda:6'), covar=tensor([0.1358, 0.1044, 0.1074, 0.0669, 0.0741, 0.2420, 0.1130, 0.0918], device='cuda:6'), in_proj_covar=tensor([0.0638, 0.0789, 0.0904, 0.0793, 0.0609, 0.0627, 0.0661, 0.0767], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 21:19:43,504 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=241682.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:19:53,121 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 21:19:56,098 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=241690.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:20:16,377 INFO [train.py:904] (6/8) Epoch 24, batch 8250, loss[loss=0.1828, simple_loss=0.2819, pruned_loss=0.04189, over 16686.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2831, pruned_loss=0.05413, over 3082839.39 frames. ], batch size: 134, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:21:03,231 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.441e+02 2.851e+02 3.404e+02 7.117e+02, threshold=5.702e+02, percent-clipped=1.0 2023-05-01 21:21:38,998 INFO [train.py:904] (6/8) Epoch 24, batch 8300, loss[loss=0.1888, simple_loss=0.2833, pruned_loss=0.0471, over 16737.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2807, pruned_loss=0.05147, over 3063356.70 frames. ], batch size: 124, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:21:52,731 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0286, 1.8554, 1.6779, 1.5319, 1.9787, 1.6504, 1.6237, 1.9498], device='cuda:6'), covar=tensor([0.0209, 0.0351, 0.0475, 0.0389, 0.0253, 0.0299, 0.0200, 0.0252], device='cuda:6'), in_proj_covar=tensor([0.0210, 0.0232, 0.0224, 0.0224, 0.0233, 0.0230, 0.0231, 0.0228], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 21:21:55,150 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=241762.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 21:22:52,856 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4257, 3.3368, 3.4817, 3.5537, 3.6138, 3.3330, 3.5470, 3.6582], device='cuda:6'), covar=tensor([0.1274, 0.1074, 0.1022, 0.0694, 0.0704, 0.2447, 0.1094, 0.0894], device='cuda:6'), in_proj_covar=tensor([0.0632, 0.0783, 0.0899, 0.0788, 0.0605, 0.0623, 0.0656, 0.0762], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 21:23:02,690 INFO [train.py:904] (6/8) Epoch 24, batch 8350, loss[loss=0.2272, simple_loss=0.301, pruned_loss=0.07671, over 12132.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2805, pruned_loss=0.04955, over 3070322.11 frames. ], batch size: 248, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:23:03,903 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=241803.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:23:48,197 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.476e+02 2.114e+02 2.559e+02 3.113e+02 5.727e+02, threshold=5.118e+02, percent-clipped=1.0 2023-05-01 21:23:48,794 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=241831.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 21:24:23,521 INFO [train.py:904] (6/8) Epoch 24, batch 8400, loss[loss=0.1819, simple_loss=0.272, pruned_loss=0.04591, over 12112.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.2776, pruned_loss=0.04758, over 3046457.99 frames. ], batch size: 246, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:24:42,829 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=241864.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:25:28,915 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=241892.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 21:25:45,954 INFO [train.py:904] (6/8) Epoch 24, batch 8450, loss[loss=0.1802, simple_loss=0.2767, pruned_loss=0.04181, over 11980.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2759, pruned_loss=0.04586, over 3048235.94 frames. ], batch size: 246, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:25:50,210 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8640, 4.0953, 3.8106, 3.5593, 3.3309, 4.0440, 3.6243, 3.7203], device='cuda:6'), covar=tensor([0.0886, 0.0731, 0.0508, 0.0450, 0.1300, 0.0511, 0.1538, 0.0765], device='cuda:6'), in_proj_covar=tensor([0.0290, 0.0435, 0.0338, 0.0338, 0.0341, 0.0392, 0.0234, 0.0409], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 21:26:31,269 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.437e+02 2.077e+02 2.419e+02 2.958e+02 5.415e+02, threshold=4.839e+02, percent-clipped=2.0 2023-05-01 21:27:07,287 INFO [train.py:904] (6/8) Epoch 24, batch 8500, loss[loss=0.1461, simple_loss=0.2446, pruned_loss=0.0238, over 16656.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2721, pruned_loss=0.04369, over 3043313.77 frames. ], batch size: 83, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:27:47,683 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=241977.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:28:33,821 INFO [train.py:904] (6/8) Epoch 24, batch 8550, loss[loss=0.1584, simple_loss=0.243, pruned_loss=0.03692, over 11769.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.2697, pruned_loss=0.04281, over 3014374.24 frames. ], batch size: 247, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:29:26,914 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.227e+02 2.590e+02 3.113e+02 4.507e+02, threshold=5.180e+02, percent-clipped=0.0 2023-05-01 21:30:13,213 INFO [train.py:904] (6/8) Epoch 24, batch 8600, loss[loss=0.1729, simple_loss=0.2506, pruned_loss=0.04763, over 12300.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2701, pruned_loss=0.04206, over 3017192.64 frames. ], batch size: 246, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:30:31,475 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242062.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:31:51,604 INFO [train.py:904] (6/8) Epoch 24, batch 8650, loss[loss=0.161, simple_loss=0.2563, pruned_loss=0.03281, over 12256.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2683, pruned_loss=0.04056, over 3020029.14 frames. ], batch size: 248, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:32:06,971 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4241, 3.3544, 2.7500, 2.1151, 2.0932, 2.3531, 3.3952, 2.9174], device='cuda:6'), covar=tensor([0.3172, 0.0646, 0.1842, 0.3355, 0.2985, 0.2281, 0.0470, 0.1461], device='cuda:6'), in_proj_covar=tensor([0.0325, 0.0266, 0.0303, 0.0313, 0.0294, 0.0263, 0.0294, 0.0336], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 21:32:10,578 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=242110.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:32:24,068 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5540, 3.3013, 3.5013, 1.8959, 3.7105, 3.7314, 2.9635, 2.9365], device='cuda:6'), covar=tensor([0.0690, 0.0281, 0.0218, 0.1203, 0.0082, 0.0163, 0.0435, 0.0427], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0106, 0.0096, 0.0134, 0.0079, 0.0123, 0.0124, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 21:32:56,837 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.159e+02 2.524e+02 3.190e+02 5.272e+02, threshold=5.049e+02, percent-clipped=1.0 2023-05-01 21:33:37,177 INFO [train.py:904] (6/8) Epoch 24, batch 8700, loss[loss=0.1525, simple_loss=0.2506, pruned_loss=0.02722, over 16881.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2651, pruned_loss=0.03904, over 3016360.84 frames. ], batch size: 96, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:33:38,236 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242153.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:33:50,559 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242159.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:34:05,498 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6853, 4.4538, 4.7564, 4.8369, 5.0124, 4.5342, 5.0438, 5.0069], device='cuda:6'), covar=tensor([0.1788, 0.1327, 0.1484, 0.0726, 0.0526, 0.0877, 0.0611, 0.0785], device='cuda:6'), in_proj_covar=tensor([0.0628, 0.0779, 0.0891, 0.0786, 0.0600, 0.0621, 0.0653, 0.0758], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 21:34:42,445 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242187.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 21:35:13,442 INFO [train.py:904] (6/8) Epoch 24, batch 8750, loss[loss=0.1815, simple_loss=0.2826, pruned_loss=0.04018, over 15479.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2647, pruned_loss=0.03814, over 3041338.29 frames. ], batch size: 191, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:35:42,282 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242214.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:36:10,691 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242226.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:36:21,483 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 2.077e+02 2.590e+02 3.155e+02 6.830e+02, threshold=5.181e+02, percent-clipped=2.0 2023-05-01 21:37:05,082 INFO [train.py:904] (6/8) Epoch 24, batch 8800, loss[loss=0.1508, simple_loss=0.2496, pruned_loss=0.02596, over 16456.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2634, pruned_loss=0.03702, over 3055687.69 frames. ], batch size: 75, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:37:56,142 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242277.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:38:09,314 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4595, 4.7290, 4.6082, 4.5730, 4.3149, 4.2064, 4.2351, 4.7677], device='cuda:6'), covar=tensor([0.1034, 0.0940, 0.0742, 0.0695, 0.0670, 0.1513, 0.1098, 0.0815], device='cuda:6'), in_proj_covar=tensor([0.0677, 0.0811, 0.0674, 0.0630, 0.0516, 0.0525, 0.0684, 0.0637], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 21:38:17,952 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242287.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:38:50,061 INFO [train.py:904] (6/8) Epoch 24, batch 8850, loss[loss=0.1658, simple_loss=0.2683, pruned_loss=0.03167, over 16253.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2659, pruned_loss=0.03663, over 3036972.50 frames. ], batch size: 165, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:39:03,894 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242309.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:39:30,579 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 21:39:38,584 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=242325.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:39:52,555 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 2.137e+02 2.486e+02 3.124e+02 5.230e+02, threshold=4.972e+02, percent-clipped=1.0 2023-05-01 21:40:38,017 INFO [train.py:904] (6/8) Epoch 24, batch 8900, loss[loss=0.1618, simple_loss=0.2614, pruned_loss=0.03104, over 16822.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2662, pruned_loss=0.03596, over 3050411.97 frames. ], batch size: 90, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:41:12,542 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242370.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:41:41,398 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7943, 1.9282, 2.3071, 2.8065, 2.6277, 3.0811, 2.1046, 3.0542], device='cuda:6'), covar=tensor([0.0215, 0.0567, 0.0398, 0.0274, 0.0366, 0.0206, 0.0577, 0.0166], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0190, 0.0177, 0.0180, 0.0195, 0.0155, 0.0194, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 21:42:44,441 INFO [train.py:904] (6/8) Epoch 24, batch 8950, loss[loss=0.1669, simple_loss=0.2568, pruned_loss=0.03855, over 12354.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2663, pruned_loss=0.0367, over 3060571.92 frames. ], batch size: 247, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 21:43:49,726 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.299e+02 2.560e+02 2.923e+02 8.049e+02, threshold=5.120e+02, percent-clipped=2.0 2023-05-01 21:44:35,817 INFO [train.py:904] (6/8) Epoch 24, batch 9000, loss[loss=0.1488, simple_loss=0.2375, pruned_loss=0.0301, over 11841.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2636, pruned_loss=0.03572, over 3075386.48 frames. ], batch size: 246, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 21:44:35,818 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 21:44:45,532 INFO [train.py:938] (6/8) Epoch 24, validation: loss=0.1445, simple_loss=0.2484, pruned_loss=0.02026, over 944034.00 frames. 2023-05-01 21:44:45,532 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 21:44:59,709 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242459.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:45:49,145 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242483.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:45:58,868 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242487.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 21:46:30,341 INFO [train.py:904] (6/8) Epoch 24, batch 9050, loss[loss=0.1533, simple_loss=0.2473, pruned_loss=0.02963, over 17109.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2643, pruned_loss=0.03603, over 3080647.53 frames. ], batch size: 49, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 21:46:40,246 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=242507.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:46:45,331 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242509.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:47:20,278 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.96 vs. limit=5.0 2023-05-01 21:47:30,072 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.227e+02 2.575e+02 3.137e+02 5.273e+02, threshold=5.150e+02, percent-clipped=1.0 2023-05-01 21:47:35,727 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=242535.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 21:48:00,806 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242544.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:48:17,897 INFO [train.py:904] (6/8) Epoch 24, batch 9100, loss[loss=0.1693, simple_loss=0.2693, pruned_loss=0.03466, over 16983.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2637, pruned_loss=0.03651, over 3065674.68 frames. ], batch size: 125, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 21:48:23,317 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242555.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:49:27,654 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242582.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:50:16,322 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242602.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:50:16,953 INFO [train.py:904] (6/8) Epoch 24, batch 9150, loss[loss=0.1534, simple_loss=0.2483, pruned_loss=0.02922, over 16662.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.264, pruned_loss=0.03626, over 3069366.22 frames. ], batch size: 134, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 21:50:47,185 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242616.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:51:21,273 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.667e+02 2.227e+02 2.578e+02 3.148e+02 6.071e+02, threshold=5.155e+02, percent-clipped=3.0 2023-05-01 21:52:01,742 INFO [train.py:904] (6/8) Epoch 24, batch 9200, loss[loss=0.1532, simple_loss=0.2453, pruned_loss=0.03055, over 17095.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2594, pruned_loss=0.03503, over 3076837.64 frames. ], batch size: 53, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 21:52:22,356 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242663.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:52:24,964 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242665.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:53:38,106 INFO [train.py:904] (6/8) Epoch 24, batch 9250, loss[loss=0.1551, simple_loss=0.257, pruned_loss=0.02664, over 16779.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2593, pruned_loss=0.03552, over 3055989.25 frames. ], batch size: 83, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 21:54:40,869 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.148e+02 2.469e+02 2.996e+02 5.681e+02, threshold=4.938e+02, percent-clipped=3.0 2023-05-01 21:54:50,830 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-05-01 21:55:27,519 INFO [train.py:904] (6/8) Epoch 24, batch 9300, loss[loss=0.1571, simple_loss=0.2467, pruned_loss=0.03381, over 12631.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.258, pruned_loss=0.03519, over 3027995.10 frames. ], batch size: 248, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 21:56:13,860 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242772.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:57:14,581 INFO [train.py:904] (6/8) Epoch 24, batch 9350, loss[loss=0.1637, simple_loss=0.2509, pruned_loss=0.03821, over 12178.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.258, pruned_loss=0.0347, over 3055493.52 frames. ], batch size: 246, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 21:57:27,690 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242809.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:58:13,360 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.321e+02 1.952e+02 2.311e+02 3.086e+02 6.417e+02, threshold=4.623e+02, percent-clipped=1.0 2023-05-01 21:58:14,229 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242833.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:58:25,510 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242839.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:58:54,546 INFO [train.py:904] (6/8) Epoch 24, batch 9400, loss[loss=0.164, simple_loss=0.2584, pruned_loss=0.03485, over 15269.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2582, pruned_loss=0.03451, over 3056137.01 frames. ], batch size: 190, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 21:59:03,718 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=242857.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:59:54,992 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242882.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:00:35,986 INFO [train.py:904] (6/8) Epoch 24, batch 9450, loss[loss=0.1628, simple_loss=0.2563, pruned_loss=0.03467, over 16568.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2597, pruned_loss=0.03458, over 3057501.83 frames. ], batch size: 68, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:00:51,517 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242911.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:01:08,960 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9695, 1.9041, 2.4340, 2.8802, 2.6491, 3.2071, 2.0593, 3.2139], device='cuda:6'), covar=tensor([0.0221, 0.0638, 0.0399, 0.0300, 0.0378, 0.0212, 0.0622, 0.0189], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0190, 0.0177, 0.0179, 0.0195, 0.0154, 0.0193, 0.0152], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:01:14,124 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242921.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:01:32,437 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=242930.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:01:37,830 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.228e+02 2.255e+02 2.567e+02 3.168e+02 7.490e+02, threshold=5.133e+02, percent-clipped=6.0 2023-05-01 22:01:45,304 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8922, 3.8344, 4.0147, 3.7400, 3.9765, 4.3335, 4.0081, 3.6688], device='cuda:6'), covar=tensor([0.1840, 0.2230, 0.2132, 0.2506, 0.2478, 0.1703, 0.1591, 0.2753], device='cuda:6'), in_proj_covar=tensor([0.0394, 0.0586, 0.0646, 0.0480, 0.0633, 0.0673, 0.0501, 0.0637], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 22:01:54,002 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 22:02:16,733 INFO [train.py:904] (6/8) Epoch 24, batch 9500, loss[loss=0.1594, simple_loss=0.2593, pruned_loss=0.0298, over 16714.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.259, pruned_loss=0.03436, over 3056992.34 frames. ], batch size: 83, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:02:18,018 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2777, 4.3415, 4.5106, 4.2169, 4.3736, 4.8576, 4.3954, 4.0447], device='cuda:6'), covar=tensor([0.1514, 0.1834, 0.2040, 0.2293, 0.2470, 0.1043, 0.1579, 0.2595], device='cuda:6'), in_proj_covar=tensor([0.0393, 0.0585, 0.0645, 0.0480, 0.0633, 0.0672, 0.0501, 0.0637], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 22:02:30,100 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242958.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:02:43,806 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242965.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:03:05,405 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7954, 2.6706, 2.3187, 4.3553, 2.6272, 4.0314, 1.5007, 2.8206], device='cuda:6'), covar=tensor([0.1272, 0.0746, 0.1252, 0.0160, 0.0134, 0.0424, 0.1636, 0.0808], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0173, 0.0193, 0.0187, 0.0197, 0.0211, 0.0202, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 22:03:16,258 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242982.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 22:03:31,267 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6391, 3.7047, 3.5206, 3.1587, 3.3307, 3.6272, 3.3713, 3.4627], device='cuda:6'), covar=tensor([0.0586, 0.0589, 0.0312, 0.0294, 0.0483, 0.0463, 0.1592, 0.0490], device='cuda:6'), in_proj_covar=tensor([0.0284, 0.0423, 0.0331, 0.0332, 0.0332, 0.0382, 0.0228, 0.0397], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-05-01 22:04:01,164 INFO [train.py:904] (6/8) Epoch 24, batch 9550, loss[loss=0.1884, simple_loss=0.2857, pruned_loss=0.0455, over 16967.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2592, pruned_loss=0.03481, over 3066440.96 frames. ], batch size: 109, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:04:24,732 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243013.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:05:06,080 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.055e+02 2.389e+02 2.872e+02 5.971e+02, threshold=4.777e+02, percent-clipped=1.0 2023-05-01 22:05:43,254 INFO [train.py:904] (6/8) Epoch 24, batch 9600, loss[loss=0.1763, simple_loss=0.2737, pruned_loss=0.03945, over 16715.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2607, pruned_loss=0.03557, over 3065630.09 frames. ], batch size: 83, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:06:12,749 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6767, 2.4830, 2.3229, 3.8772, 2.0894, 3.7523, 1.4157, 2.8158], device='cuda:6'), covar=tensor([0.1420, 0.0804, 0.1280, 0.0143, 0.0097, 0.0358, 0.1810, 0.0764], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0173, 0.0193, 0.0187, 0.0197, 0.0211, 0.0202, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 22:06:16,776 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9820, 2.7168, 2.9128, 2.0559, 2.7489, 2.1833, 2.7222, 2.8584], device='cuda:6'), covar=tensor([0.0254, 0.0882, 0.0452, 0.1822, 0.0761, 0.0863, 0.0578, 0.0821], device='cuda:6'), in_proj_covar=tensor([0.0152, 0.0158, 0.0163, 0.0150, 0.0141, 0.0127, 0.0139, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 22:07:31,348 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4428, 3.3393, 3.5072, 3.5814, 3.6143, 3.3659, 3.5904, 3.6549], device='cuda:6'), covar=tensor([0.1271, 0.1062, 0.1065, 0.0635, 0.0644, 0.2152, 0.0937, 0.0835], device='cuda:6'), in_proj_covar=tensor([0.0617, 0.0762, 0.0874, 0.0769, 0.0590, 0.0607, 0.0640, 0.0741], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:07:32,092 INFO [train.py:904] (6/8) Epoch 24, batch 9650, loss[loss=0.166, simple_loss=0.2543, pruned_loss=0.03888, over 12460.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2625, pruned_loss=0.03637, over 3035688.84 frames. ], batch size: 248, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:08:29,797 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=243128.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:08:37,734 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.193e+02 2.679e+02 3.134e+02 6.217e+02, threshold=5.359e+02, percent-clipped=4.0 2023-05-01 22:08:50,183 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=243139.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:09:19,549 INFO [train.py:904] (6/8) Epoch 24, batch 9700, loss[loss=0.1621, simple_loss=0.2571, pruned_loss=0.03354, over 16805.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2611, pruned_loss=0.03621, over 3025129.18 frames. ], batch size: 124, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:09:24,369 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1605, 2.1899, 2.5653, 3.0753, 2.7576, 3.4612, 2.4992, 3.4972], device='cuda:6'), covar=tensor([0.0220, 0.0531, 0.0390, 0.0275, 0.0355, 0.0186, 0.0466, 0.0169], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0188, 0.0176, 0.0177, 0.0193, 0.0153, 0.0192, 0.0151], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:09:41,130 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.69 vs. limit=5.0 2023-05-01 22:10:08,768 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0125, 2.3433, 2.3540, 3.0869, 1.8514, 3.2794, 1.8260, 2.8278], device='cuda:6'), covar=tensor([0.1196, 0.0691, 0.1043, 0.0165, 0.0084, 0.0356, 0.1471, 0.0673], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0173, 0.0192, 0.0186, 0.0196, 0.0210, 0.0202, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 22:10:30,927 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243187.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:11:01,229 INFO [train.py:904] (6/8) Epoch 24, batch 9750, loss[loss=0.1679, simple_loss=0.2666, pruned_loss=0.03459, over 15434.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2596, pruned_loss=0.0361, over 3029217.52 frames. ], batch size: 191, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 22:11:17,043 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=243211.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:12:03,356 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 1.977e+02 2.392e+02 3.038e+02 5.323e+02, threshold=4.783e+02, percent-clipped=0.0 2023-05-01 22:12:37,393 INFO [train.py:904] (6/8) Epoch 24, batch 9800, loss[loss=0.1582, simple_loss=0.2645, pruned_loss=0.02596, over 16928.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2597, pruned_loss=0.03484, over 3054390.78 frames. ], batch size: 96, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 22:12:49,025 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=243258.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:12:50,825 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243259.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:13:22,204 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=243277.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 22:13:58,673 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=243293.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:14:21,695 INFO [train.py:904] (6/8) Epoch 24, batch 9850, loss[loss=0.163, simple_loss=0.2517, pruned_loss=0.03716, over 12594.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2608, pruned_loss=0.03443, over 3060906.61 frames. ], batch size: 248, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 22:14:28,946 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243306.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:14:38,011 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4695, 4.5853, 4.3461, 4.0481, 3.9161, 4.5094, 4.2939, 4.0880], device='cuda:6'), covar=tensor([0.0683, 0.0604, 0.0416, 0.0384, 0.1202, 0.0563, 0.0497, 0.0786], device='cuda:6'), in_proj_covar=tensor([0.0285, 0.0422, 0.0330, 0.0331, 0.0331, 0.0382, 0.0228, 0.0396], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-05-01 22:15:22,936 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 1.939e+02 2.274e+02 2.876e+02 6.181e+02, threshold=4.547e+02, percent-clipped=2.0 2023-05-01 22:16:11,622 INFO [train.py:904] (6/8) Epoch 24, batch 9900, loss[loss=0.1636, simple_loss=0.2524, pruned_loss=0.03739, over 12386.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2615, pruned_loss=0.03452, over 3060979.47 frames. ], batch size: 247, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 22:16:15,599 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=243354.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:18:09,537 INFO [train.py:904] (6/8) Epoch 24, batch 9950, loss[loss=0.1585, simple_loss=0.2592, pruned_loss=0.02895, over 16750.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2638, pruned_loss=0.03532, over 3056919.02 frames. ], batch size: 76, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 22:19:11,961 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=243428.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:19:24,675 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.448e+02 2.110e+02 2.470e+02 3.179e+02 5.008e+02, threshold=4.941e+02, percent-clipped=4.0 2023-05-01 22:20:10,932 INFO [train.py:904] (6/8) Epoch 24, batch 10000, loss[loss=0.1609, simple_loss=0.2458, pruned_loss=0.03803, over 12870.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2618, pruned_loss=0.03474, over 3065081.61 frames. ], batch size: 250, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:20:25,967 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6301, 1.8499, 2.1558, 2.5829, 2.5371, 2.9097, 2.0361, 2.8496], device='cuda:6'), covar=tensor([0.0246, 0.0554, 0.0433, 0.0329, 0.0366, 0.0211, 0.0530, 0.0180], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0187, 0.0175, 0.0177, 0.0192, 0.0152, 0.0191, 0.0150], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:20:55,588 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243476.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:21:51,123 INFO [train.py:904] (6/8) Epoch 24, batch 10050, loss[loss=0.1645, simple_loss=0.2652, pruned_loss=0.03188, over 16910.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2625, pruned_loss=0.03451, over 3066312.01 frames. ], batch size: 96, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:22:03,440 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5022, 4.5135, 4.3137, 3.8505, 4.4361, 1.7723, 4.2133, 4.1417], device='cuda:6'), covar=tensor([0.0113, 0.0118, 0.0232, 0.0285, 0.0111, 0.2613, 0.0158, 0.0227], device='cuda:6'), in_proj_covar=tensor([0.0165, 0.0157, 0.0195, 0.0170, 0.0173, 0.0204, 0.0184, 0.0166], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:22:52,101 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.427e+02 2.109e+02 2.585e+02 3.087e+02 5.366e+02, threshold=5.170e+02, percent-clipped=3.0 2023-05-01 22:23:17,282 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2604, 4.3972, 4.5129, 4.3055, 4.4261, 4.8539, 4.3884, 4.0910], device='cuda:6'), covar=tensor([0.1559, 0.1693, 0.1914, 0.1969, 0.2126, 0.0924, 0.1619, 0.2476], device='cuda:6'), in_proj_covar=tensor([0.0388, 0.0577, 0.0638, 0.0474, 0.0624, 0.0660, 0.0494, 0.0627], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 22:23:25,157 INFO [train.py:904] (6/8) Epoch 24, batch 10100, loss[loss=0.1655, simple_loss=0.2551, pruned_loss=0.03795, over 16216.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2626, pruned_loss=0.03486, over 3058336.96 frames. ], batch size: 165, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:24:02,828 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.89 vs. limit=5.0 2023-05-01 22:24:16,129 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=243577.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 22:25:09,202 INFO [train.py:904] (6/8) Epoch 25, batch 0, loss[loss=0.217, simple_loss=0.2957, pruned_loss=0.06919, over 16765.00 frames. ], tot_loss[loss=0.217, simple_loss=0.2957, pruned_loss=0.06919, over 16765.00 frames. ], batch size: 124, lr: 2.72e-03, grad_scale: 8.0 2023-05-01 22:25:09,202 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 22:25:16,827 INFO [train.py:938] (6/8) Epoch 25, validation: loss=0.1443, simple_loss=0.2477, pruned_loss=0.02048, over 944034.00 frames. 2023-05-01 22:25:16,828 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 22:25:48,589 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243625.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:25:53,157 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4628, 4.4381, 4.8338, 4.8167, 4.8550, 4.5563, 4.5388, 4.3745], device='cuda:6'), covar=tensor([0.0413, 0.0835, 0.0461, 0.0456, 0.0565, 0.0476, 0.1000, 0.0657], device='cuda:6'), in_proj_covar=tensor([0.0397, 0.0443, 0.0435, 0.0400, 0.0476, 0.0454, 0.0528, 0.0364], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 22:26:03,176 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.517e+02 2.877e+02 3.521e+02 6.755e+02, threshold=5.755e+02, percent-clipped=6.0 2023-05-01 22:26:21,448 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=243649.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:26:26,915 INFO [train.py:904] (6/8) Epoch 25, batch 50, loss[loss=0.1997, simple_loss=0.2986, pruned_loss=0.05039, over 16763.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2687, pruned_loss=0.0494, over 745362.19 frames. ], batch size: 57, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:26:58,571 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1375, 5.0670, 4.9719, 4.5188, 4.6642, 5.0260, 4.9490, 4.6258], device='cuda:6'), covar=tensor([0.0574, 0.0556, 0.0339, 0.0401, 0.1097, 0.0491, 0.0361, 0.0817], device='cuda:6'), in_proj_covar=tensor([0.0285, 0.0424, 0.0331, 0.0332, 0.0333, 0.0382, 0.0228, 0.0398], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:6') 2023-05-01 22:27:20,363 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6139, 3.2624, 3.7105, 2.0289, 3.7657, 3.7714, 3.1003, 2.7695], device='cuda:6'), covar=tensor([0.0712, 0.0263, 0.0175, 0.1128, 0.0109, 0.0197, 0.0409, 0.0453], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0106, 0.0095, 0.0135, 0.0079, 0.0123, 0.0124, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-01 22:27:35,800 INFO [train.py:904] (6/8) Epoch 25, batch 100, loss[loss=0.1887, simple_loss=0.2784, pruned_loss=0.0495, over 15673.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2652, pruned_loss=0.04672, over 1318350.49 frames. ], batch size: 191, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:28:22,077 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.579e+02 2.266e+02 2.864e+02 3.484e+02 1.313e+03, threshold=5.728e+02, percent-clipped=5.0 2023-05-01 22:28:35,847 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2794, 3.3291, 3.6969, 2.1446, 3.0987, 2.2906, 3.8011, 3.6548], device='cuda:6'), covar=tensor([0.0247, 0.0978, 0.0562, 0.2142, 0.0855, 0.1093, 0.0492, 0.0919], device='cuda:6'), in_proj_covar=tensor([0.0154, 0.0160, 0.0165, 0.0152, 0.0142, 0.0128, 0.0141, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 22:28:41,013 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8633, 2.5479, 2.0324, 2.2338, 2.8816, 2.5979, 2.9147, 2.9656], device='cuda:6'), covar=tensor([0.0237, 0.0390, 0.0528, 0.0468, 0.0249, 0.0388, 0.0233, 0.0295], device='cuda:6'), in_proj_covar=tensor([0.0215, 0.0238, 0.0228, 0.0228, 0.0238, 0.0236, 0.0234, 0.0232], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:28:45,143 INFO [train.py:904] (6/8) Epoch 25, batch 150, loss[loss=0.1963, simple_loss=0.2709, pruned_loss=0.06084, over 16460.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2629, pruned_loss=0.04509, over 1757632.73 frames. ], batch size: 146, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:29:26,465 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-01 22:29:53,799 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5102, 1.6241, 2.1855, 2.3292, 2.4754, 2.4754, 1.7881, 2.5854], device='cuda:6'), covar=tensor([0.0194, 0.0613, 0.0318, 0.0334, 0.0281, 0.0278, 0.0619, 0.0182], device='cuda:6'), in_proj_covar=tensor([0.0188, 0.0190, 0.0178, 0.0181, 0.0196, 0.0155, 0.0194, 0.0153], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:29:55,570 INFO [train.py:904] (6/8) Epoch 25, batch 200, loss[loss=0.1841, simple_loss=0.26, pruned_loss=0.05412, over 16744.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2617, pruned_loss=0.04448, over 2099179.34 frames. ], batch size: 124, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:30:25,952 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4973, 5.9153, 5.6652, 5.6625, 5.3390, 5.3058, 5.2889, 6.0386], device='cuda:6'), covar=tensor([0.1751, 0.1089, 0.1175, 0.1060, 0.0945, 0.0819, 0.1401, 0.1097], device='cuda:6'), in_proj_covar=tensor([0.0685, 0.0824, 0.0679, 0.0638, 0.0522, 0.0529, 0.0691, 0.0646], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:30:40,581 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.064e+02 2.455e+02 3.172e+02 7.590e+02, threshold=4.909e+02, percent-clipped=1.0 2023-05-01 22:30:42,886 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6896, 4.4486, 4.7074, 4.8242, 4.9499, 4.4811, 4.8754, 4.9407], device='cuda:6'), covar=tensor([0.1791, 0.1312, 0.1435, 0.0834, 0.0656, 0.1188, 0.1958, 0.0879], device='cuda:6'), in_proj_covar=tensor([0.0633, 0.0781, 0.0895, 0.0789, 0.0605, 0.0621, 0.0655, 0.0759], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:31:04,162 INFO [train.py:904] (6/8) Epoch 25, batch 250, loss[loss=0.1941, simple_loss=0.2691, pruned_loss=0.05955, over 16847.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2588, pruned_loss=0.04327, over 2366903.41 frames. ], batch size: 96, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:32:14,684 INFO [train.py:904] (6/8) Epoch 25, batch 300, loss[loss=0.1607, simple_loss=0.263, pruned_loss=0.02923, over 17115.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2569, pruned_loss=0.04205, over 2573446.69 frames. ], batch size: 47, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:33:00,867 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.556e+02 2.231e+02 2.519e+02 2.986e+02 7.202e+02, threshold=5.038e+02, percent-clipped=2.0 2023-05-01 22:33:19,568 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=243949.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:33:24,546 INFO [train.py:904] (6/8) Epoch 25, batch 350, loss[loss=0.176, simple_loss=0.2597, pruned_loss=0.04618, over 16835.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2552, pruned_loss=0.04101, over 2743047.03 frames. ], batch size: 96, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:34:25,842 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243997.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:34:38,353 INFO [train.py:904] (6/8) Epoch 25, batch 400, loss[loss=0.1623, simple_loss=0.2479, pruned_loss=0.03834, over 16529.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.254, pruned_loss=0.04137, over 2869657.94 frames. ], batch size: 68, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:34:50,908 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1161, 3.1319, 2.9216, 5.1172, 4.2016, 4.2926, 1.9237, 3.1664], device='cuda:6'), covar=tensor([0.1231, 0.0741, 0.1143, 0.0211, 0.0239, 0.0619, 0.1550, 0.0820], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0176, 0.0195, 0.0191, 0.0200, 0.0213, 0.0204, 0.0194], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 22:34:54,317 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8519, 4.3048, 3.0059, 2.3176, 2.5912, 2.6003, 4.6858, 3.4358], device='cuda:6'), covar=tensor([0.2878, 0.0615, 0.1945, 0.3058, 0.3095, 0.2197, 0.0345, 0.1601], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0268, 0.0305, 0.0315, 0.0294, 0.0265, 0.0295, 0.0339], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 22:35:02,304 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-01 22:35:22,991 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.463e+02 2.193e+02 2.643e+02 3.297e+02 6.329e+02, threshold=5.285e+02, percent-clipped=2.0 2023-05-01 22:35:42,248 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-05-01 22:35:47,033 INFO [train.py:904] (6/8) Epoch 25, batch 450, loss[loss=0.1578, simple_loss=0.2346, pruned_loss=0.04048, over 16809.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2519, pruned_loss=0.04018, over 2980363.25 frames. ], batch size: 83, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:36:50,457 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7908, 4.7395, 5.1391, 5.1369, 5.1572, 4.9002, 4.8346, 4.6663], device='cuda:6'), covar=tensor([0.0346, 0.0747, 0.0391, 0.0376, 0.0435, 0.0406, 0.0860, 0.0531], device='cuda:6'), in_proj_covar=tensor([0.0411, 0.0459, 0.0447, 0.0412, 0.0491, 0.0470, 0.0546, 0.0375], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 22:36:55,203 INFO [train.py:904] (6/8) Epoch 25, batch 500, loss[loss=0.1747, simple_loss=0.2456, pruned_loss=0.05189, over 16790.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2504, pruned_loss=0.03919, over 3056875.52 frames. ], batch size: 124, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:37:18,530 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0568, 4.3111, 4.1198, 3.0459, 3.8059, 4.2482, 3.9723, 2.1325], device='cuda:6'), covar=tensor([0.0606, 0.0099, 0.0099, 0.0499, 0.0168, 0.0170, 0.0120, 0.0742], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0088, 0.0087, 0.0136, 0.0100, 0.0111, 0.0096, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 22:37:42,041 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.358e+02 2.128e+02 2.486e+02 2.973e+02 4.659e+02, threshold=4.972e+02, percent-clipped=0.0 2023-05-01 22:38:05,720 INFO [train.py:904] (6/8) Epoch 25, batch 550, loss[loss=0.155, simple_loss=0.2597, pruned_loss=0.02515, over 17137.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2499, pruned_loss=0.03903, over 3112992.84 frames. ], batch size: 48, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:38:13,026 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2128, 3.2551, 3.7167, 2.2432, 3.1002, 2.5154, 3.6729, 3.6684], device='cuda:6'), covar=tensor([0.0269, 0.0967, 0.0593, 0.2049, 0.0871, 0.0985, 0.0589, 0.0991], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0164, 0.0168, 0.0154, 0.0144, 0.0130, 0.0143, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-01 22:38:46,650 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5724, 3.6363, 3.4324, 3.0847, 3.2395, 3.5516, 3.2737, 3.3747], device='cuda:6'), covar=tensor([0.0616, 0.0709, 0.0349, 0.0326, 0.0593, 0.0500, 0.1679, 0.0583], device='cuda:6'), in_proj_covar=tensor([0.0299, 0.0446, 0.0346, 0.0349, 0.0350, 0.0402, 0.0239, 0.0418], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:39:05,129 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5307, 2.3807, 1.9290, 2.1286, 2.6966, 2.4615, 2.6238, 2.7603], device='cuda:6'), covar=tensor([0.0316, 0.0419, 0.0565, 0.0496, 0.0267, 0.0378, 0.0241, 0.0324], device='cuda:6'), in_proj_covar=tensor([0.0222, 0.0243, 0.0232, 0.0233, 0.0244, 0.0242, 0.0242, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:39:15,773 INFO [train.py:904] (6/8) Epoch 25, batch 600, loss[loss=0.1722, simple_loss=0.2506, pruned_loss=0.04684, over 16490.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2495, pruned_loss=0.03933, over 3157139.97 frames. ], batch size: 146, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:39:24,756 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5273, 4.5151, 4.8690, 4.8818, 4.8990, 4.6175, 4.6040, 4.4876], device='cuda:6'), covar=tensor([0.0445, 0.0916, 0.0481, 0.0477, 0.0515, 0.0481, 0.0896, 0.0634], device='cuda:6'), in_proj_covar=tensor([0.0415, 0.0462, 0.0450, 0.0416, 0.0494, 0.0474, 0.0549, 0.0378], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 22:39:38,951 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-05-01 22:40:02,986 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.116e+02 2.478e+02 2.900e+02 1.597e+03, threshold=4.957e+02, percent-clipped=3.0 2023-05-01 22:40:25,350 INFO [train.py:904] (6/8) Epoch 25, batch 650, loss[loss=0.1444, simple_loss=0.2306, pruned_loss=0.02904, over 17019.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2483, pruned_loss=0.03911, over 3188798.72 frames. ], batch size: 41, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:40:32,801 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5889, 2.4743, 2.4653, 4.4320, 2.4172, 2.8613, 2.5309, 2.6220], device='cuda:6'), covar=tensor([0.1290, 0.3603, 0.3204, 0.0517, 0.4218, 0.2504, 0.3732, 0.3748], device='cuda:6'), in_proj_covar=tensor([0.0411, 0.0459, 0.0378, 0.0332, 0.0443, 0.0524, 0.0432, 0.0537], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:40:53,559 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6872, 3.8105, 2.5546, 4.3437, 3.0229, 4.2939, 2.6157, 3.1415], device='cuda:6'), covar=tensor([0.0349, 0.0411, 0.1591, 0.0362, 0.0828, 0.0565, 0.1519, 0.0796], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0179, 0.0195, 0.0169, 0.0179, 0.0218, 0.0205, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 22:40:53,600 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0091, 2.7405, 2.7190, 4.7592, 3.6841, 4.2620, 1.6861, 3.1456], device='cuda:6'), covar=tensor([0.1317, 0.0891, 0.1248, 0.0247, 0.0272, 0.0441, 0.1728, 0.0806], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0175, 0.0194, 0.0191, 0.0199, 0.0213, 0.0204, 0.0194], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 22:41:08,459 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6647, 3.6884, 2.3466, 3.9415, 2.9486, 3.9088, 2.4938, 3.0498], device='cuda:6'), covar=tensor([0.0276, 0.0401, 0.1541, 0.0382, 0.0783, 0.0779, 0.1388, 0.0689], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0179, 0.0195, 0.0169, 0.0179, 0.0218, 0.0205, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 22:41:33,625 INFO [train.py:904] (6/8) Epoch 25, batch 700, loss[loss=0.1643, simple_loss=0.2553, pruned_loss=0.0367, over 17223.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2477, pruned_loss=0.03861, over 3221990.06 frames. ], batch size: 45, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:41:50,839 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2307, 4.3105, 4.3769, 4.2063, 4.2564, 4.8253, 4.3243, 4.0161], device='cuda:6'), covar=tensor([0.1804, 0.2253, 0.3028, 0.2340, 0.3086, 0.1187, 0.1904, 0.2837], device='cuda:6'), in_proj_covar=tensor([0.0413, 0.0612, 0.0677, 0.0502, 0.0665, 0.0698, 0.0525, 0.0665], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 22:41:55,146 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2853, 5.9177, 5.9895, 5.6783, 5.8762, 6.4020, 5.8989, 5.6115], device='cuda:6'), covar=tensor([0.0881, 0.2067, 0.2622, 0.2181, 0.2530, 0.0967, 0.1539, 0.2422], device='cuda:6'), in_proj_covar=tensor([0.0413, 0.0613, 0.0677, 0.0503, 0.0666, 0.0699, 0.0525, 0.0665], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 22:42:16,523 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4522, 5.8664, 5.6531, 5.6932, 5.2464, 5.3492, 5.2770, 5.9985], device='cuda:6'), covar=tensor([0.1552, 0.1121, 0.1012, 0.0930, 0.0974, 0.0675, 0.1269, 0.0881], device='cuda:6'), in_proj_covar=tensor([0.0700, 0.0845, 0.0692, 0.0651, 0.0535, 0.0541, 0.0710, 0.0661], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:42:20,953 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 2.146e+02 2.447e+02 2.875e+02 6.652e+02, threshold=4.894e+02, percent-clipped=4.0 2023-05-01 22:42:42,412 INFO [train.py:904] (6/8) Epoch 25, batch 750, loss[loss=0.1695, simple_loss=0.2704, pruned_loss=0.03429, over 17111.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2482, pruned_loss=0.03925, over 3243353.50 frames. ], batch size: 49, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:42:42,796 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=244353.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:43:11,789 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7555, 3.9515, 2.8495, 2.3309, 2.6390, 2.4327, 4.0048, 3.3919], device='cuda:6'), covar=tensor([0.3090, 0.0630, 0.2006, 0.3182, 0.2997, 0.2322, 0.0624, 0.1608], device='cuda:6'), in_proj_covar=tensor([0.0329, 0.0269, 0.0308, 0.0317, 0.0297, 0.0267, 0.0298, 0.0342], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 22:43:34,891 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8066, 4.7027, 4.6896, 4.2618, 4.3673, 4.7295, 4.5596, 4.4156], device='cuda:6'), covar=tensor([0.0681, 0.0981, 0.0389, 0.0381, 0.1059, 0.0569, 0.0504, 0.0839], device='cuda:6'), in_proj_covar=tensor([0.0303, 0.0451, 0.0351, 0.0353, 0.0355, 0.0407, 0.0242, 0.0424], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:43:52,150 INFO [train.py:904] (6/8) Epoch 25, batch 800, loss[loss=0.1566, simple_loss=0.242, pruned_loss=0.03562, over 17228.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2477, pruned_loss=0.03874, over 3261459.72 frames. ], batch size: 45, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:44:08,210 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=244414.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:44:13,677 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7645, 3.4227, 3.8767, 2.0165, 3.9470, 3.9974, 3.2040, 3.0322], device='cuda:6'), covar=tensor([0.0738, 0.0268, 0.0205, 0.1232, 0.0107, 0.0213, 0.0413, 0.0459], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0111, 0.0100, 0.0140, 0.0083, 0.0130, 0.0129, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 22:44:39,342 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.613e+02 2.216e+02 2.518e+02 3.114e+02 7.971e+02, threshold=5.036e+02, percent-clipped=2.0 2023-05-01 22:45:03,353 INFO [train.py:904] (6/8) Epoch 25, batch 850, loss[loss=0.1548, simple_loss=0.247, pruned_loss=0.03126, over 17185.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2478, pruned_loss=0.03887, over 3276901.15 frames. ], batch size: 46, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:46:12,666 INFO [train.py:904] (6/8) Epoch 25, batch 900, loss[loss=0.1755, simple_loss=0.2611, pruned_loss=0.04493, over 12548.00 frames. ], tot_loss[loss=0.1618, simple_loss=0.2472, pruned_loss=0.03816, over 3280532.43 frames. ], batch size: 246, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:47:00,726 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.397e+02 1.999e+02 2.403e+02 2.874e+02 4.869e+02, threshold=4.806e+02, percent-clipped=0.0 2023-05-01 22:47:23,394 INFO [train.py:904] (6/8) Epoch 25, batch 950, loss[loss=0.1653, simple_loss=0.2424, pruned_loss=0.04413, over 16427.00 frames. ], tot_loss[loss=0.1621, simple_loss=0.2474, pruned_loss=0.03842, over 3292498.75 frames. ], batch size: 146, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:47:47,340 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6531, 2.7213, 2.3132, 2.4949, 3.0549, 2.8031, 3.2707, 3.1929], device='cuda:6'), covar=tensor([0.0176, 0.0471, 0.0598, 0.0521, 0.0321, 0.0424, 0.0295, 0.0315], device='cuda:6'), in_proj_covar=tensor([0.0224, 0.0245, 0.0234, 0.0234, 0.0246, 0.0243, 0.0244, 0.0240], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:48:33,896 INFO [train.py:904] (6/8) Epoch 25, batch 1000, loss[loss=0.1647, simple_loss=0.2588, pruned_loss=0.03526, over 17088.00 frames. ], tot_loss[loss=0.1612, simple_loss=0.2458, pruned_loss=0.03823, over 3291552.24 frames. ], batch size: 47, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:49:21,002 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.583e+02 2.078e+02 2.468e+02 2.868e+02 8.684e+02, threshold=4.936e+02, percent-clipped=5.0 2023-05-01 22:49:42,536 INFO [train.py:904] (6/8) Epoch 25, batch 1050, loss[loss=0.1476, simple_loss=0.2404, pruned_loss=0.0274, over 17195.00 frames. ], tot_loss[loss=0.1602, simple_loss=0.2451, pruned_loss=0.03769, over 3296549.42 frames. ], batch size: 46, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:50:53,097 INFO [train.py:904] (6/8) Epoch 25, batch 1100, loss[loss=0.1616, simple_loss=0.2363, pruned_loss=0.04346, over 16711.00 frames. ], tot_loss[loss=0.1599, simple_loss=0.2445, pruned_loss=0.03762, over 3304009.46 frames. ], batch size: 124, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 22:51:01,256 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=244709.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:51:13,168 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=244717.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 22:51:39,965 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-01 22:51:40,295 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 2.068e+02 2.379e+02 2.713e+02 6.650e+02, threshold=4.759e+02, percent-clipped=1.0 2023-05-01 22:52:02,023 INFO [train.py:904] (6/8) Epoch 25, batch 1150, loss[loss=0.1552, simple_loss=0.2442, pruned_loss=0.03309, over 16977.00 frames. ], tot_loss[loss=0.159, simple_loss=0.2448, pruned_loss=0.03659, over 3309419.53 frames. ], batch size: 41, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 22:52:37,177 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=244778.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 22:53:11,752 INFO [train.py:904] (6/8) Epoch 25, batch 1200, loss[loss=0.1727, simple_loss=0.2484, pruned_loss=0.04848, over 16783.00 frames. ], tot_loss[loss=0.1576, simple_loss=0.2432, pruned_loss=0.036, over 3308783.55 frames. ], batch size: 102, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 22:53:24,006 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6478, 4.9503, 4.7587, 4.7437, 4.4700, 4.4306, 4.4382, 5.0136], device='cuda:6'), covar=tensor([0.1296, 0.0977, 0.1083, 0.0919, 0.0892, 0.1345, 0.1176, 0.0988], device='cuda:6'), in_proj_covar=tensor([0.0711, 0.0857, 0.0704, 0.0664, 0.0544, 0.0549, 0.0722, 0.0673], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:53:29,265 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9480, 2.6017, 2.1136, 2.3578, 2.9115, 2.6713, 2.8739, 3.0149], device='cuda:6'), covar=tensor([0.0278, 0.0430, 0.0577, 0.0550, 0.0279, 0.0402, 0.0299, 0.0388], device='cuda:6'), in_proj_covar=tensor([0.0226, 0.0245, 0.0235, 0.0235, 0.0247, 0.0245, 0.0247, 0.0242], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:53:43,745 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4714, 3.8453, 4.2692, 2.3939, 3.3578, 2.8370, 3.7989, 3.9372], device='cuda:6'), covar=tensor([0.0360, 0.0959, 0.0444, 0.1973, 0.0892, 0.0896, 0.0785, 0.1169], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0167, 0.0170, 0.0156, 0.0147, 0.0132, 0.0146, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 22:53:57,424 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.058e+02 2.429e+02 2.907e+02 5.779e+02, threshold=4.859e+02, percent-clipped=1.0 2023-05-01 22:54:19,233 INFO [train.py:904] (6/8) Epoch 25, batch 1250, loss[loss=0.1522, simple_loss=0.2304, pruned_loss=0.03701, over 16798.00 frames. ], tot_loss[loss=0.1576, simple_loss=0.2428, pruned_loss=0.03621, over 3313119.88 frames. ], batch size: 39, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 22:55:06,357 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=244887.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:55:26,306 INFO [train.py:904] (6/8) Epoch 25, batch 1300, loss[loss=0.1692, simple_loss=0.2485, pruned_loss=0.045, over 16398.00 frames. ], tot_loss[loss=0.1577, simple_loss=0.2427, pruned_loss=0.03635, over 3313074.00 frames. ], batch size: 75, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 22:56:12,138 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.186e+02 2.539e+02 3.062e+02 9.189e+02, threshold=5.078e+02, percent-clipped=4.0 2023-05-01 22:56:24,390 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 22:56:28,286 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=244948.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:56:35,068 INFO [train.py:904] (6/8) Epoch 25, batch 1350, loss[loss=0.177, simple_loss=0.2619, pruned_loss=0.04608, over 16859.00 frames. ], tot_loss[loss=0.1587, simple_loss=0.2438, pruned_loss=0.03679, over 3317734.82 frames. ], batch size: 96, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 22:56:36,705 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=244954.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:56:39,536 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-05-01 22:57:29,190 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-01 22:57:43,144 INFO [train.py:904] (6/8) Epoch 25, batch 1400, loss[loss=0.1555, simple_loss=0.2326, pruned_loss=0.03916, over 16858.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.245, pruned_loss=0.03725, over 3321905.67 frames. ], batch size: 96, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 22:57:52,385 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=245009.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:58:00,020 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=245015.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:58:17,251 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1856, 5.1884, 4.9366, 4.3972, 5.0102, 1.8773, 4.7344, 4.6834], device='cuda:6'), covar=tensor([0.0102, 0.0088, 0.0239, 0.0414, 0.0113, 0.2970, 0.0156, 0.0270], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0169, 0.0208, 0.0183, 0.0186, 0.0215, 0.0197, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 22:58:27,157 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 22:58:28,654 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.075e+02 2.428e+02 3.013e+02 6.709e+02, threshold=4.856e+02, percent-clipped=2.0 2023-05-01 22:58:51,281 INFO [train.py:904] (6/8) Epoch 25, batch 1450, loss[loss=0.1727, simple_loss=0.2634, pruned_loss=0.04094, over 17107.00 frames. ], tot_loss[loss=0.1599, simple_loss=0.2447, pruned_loss=0.03759, over 3327860.52 frames. ], batch size: 47, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 22:58:56,919 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=245057.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:59:18,775 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=245073.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 22:59:27,927 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.09 vs. limit=2.0 2023-05-01 23:00:00,084 INFO [train.py:904] (6/8) Epoch 25, batch 1500, loss[loss=0.1588, simple_loss=0.2506, pruned_loss=0.03347, over 17249.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.2444, pruned_loss=0.03752, over 3327214.80 frames. ], batch size: 52, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:00:40,617 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8118, 4.3431, 3.0598, 2.3804, 2.6800, 2.5610, 4.6316, 3.5392], device='cuda:6'), covar=tensor([0.2922, 0.0542, 0.1905, 0.2919, 0.2899, 0.2151, 0.0362, 0.1411], device='cuda:6'), in_proj_covar=tensor([0.0330, 0.0272, 0.0310, 0.0319, 0.0300, 0.0269, 0.0300, 0.0345], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 23:00:46,573 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.188e+02 2.473e+02 2.927e+02 4.606e+02, threshold=4.945e+02, percent-clipped=0.0 2023-05-01 23:01:08,931 INFO [train.py:904] (6/8) Epoch 25, batch 1550, loss[loss=0.2054, simple_loss=0.2834, pruned_loss=0.06367, over 12054.00 frames. ], tot_loss[loss=0.1612, simple_loss=0.2455, pruned_loss=0.03844, over 3330812.04 frames. ], batch size: 247, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:02:19,883 INFO [train.py:904] (6/8) Epoch 25, batch 1600, loss[loss=0.1547, simple_loss=0.2448, pruned_loss=0.0323, over 17164.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2479, pruned_loss=0.03911, over 3322984.57 frames. ], batch size: 46, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:03:06,948 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.282e+02 2.634e+02 3.263e+02 7.681e+02, threshold=5.268e+02, percent-clipped=4.0 2023-05-01 23:03:16,804 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=245243.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:03:29,744 INFO [train.py:904] (6/8) Epoch 25, batch 1650, loss[loss=0.178, simple_loss=0.2584, pruned_loss=0.04881, over 16766.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2492, pruned_loss=0.03937, over 3332838.38 frames. ], batch size: 134, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:03:59,975 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.38 vs. limit=5.0 2023-05-01 23:04:41,930 INFO [train.py:904] (6/8) Epoch 25, batch 1700, loss[loss=0.1607, simple_loss=0.2472, pruned_loss=0.03709, over 16864.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2514, pruned_loss=0.04029, over 3325973.36 frames. ], batch size: 96, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:04:51,854 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=245310.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:05:08,514 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5055, 3.6962, 3.8957, 2.6791, 3.5664, 3.9752, 3.6980, 2.2758], device='cuda:6'), covar=tensor([0.0544, 0.0238, 0.0062, 0.0421, 0.0122, 0.0107, 0.0101, 0.0496], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0088, 0.0088, 0.0136, 0.0100, 0.0112, 0.0097, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 23:05:22,517 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 23:05:30,605 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.070e+02 2.579e+02 3.059e+02 5.844e+02, threshold=5.158e+02, percent-clipped=3.0 2023-05-01 23:05:52,657 INFO [train.py:904] (6/8) Epoch 25, batch 1750, loss[loss=0.1748, simple_loss=0.263, pruned_loss=0.04334, over 16327.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2525, pruned_loss=0.0407, over 3311683.47 frames. ], batch size: 165, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:06:06,388 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 23:06:20,898 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=245373.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 23:07:03,955 INFO [train.py:904] (6/8) Epoch 25, batch 1800, loss[loss=0.1799, simple_loss=0.2627, pruned_loss=0.04851, over 16813.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2534, pruned_loss=0.04054, over 3308138.06 frames. ], batch size: 102, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:07:29,359 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=245421.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 23:07:35,765 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.48 vs. limit=5.0 2023-05-01 23:07:51,077 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.502e+02 2.188e+02 2.630e+02 2.978e+02 6.380e+02, threshold=5.260e+02, percent-clipped=2.0 2023-05-01 23:08:12,915 INFO [train.py:904] (6/8) Epoch 25, batch 1850, loss[loss=0.1515, simple_loss=0.2507, pruned_loss=0.02618, over 17123.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2534, pruned_loss=0.03969, over 3311796.32 frames. ], batch size: 48, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:08:24,204 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=245460.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:08:31,536 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-05-01 23:09:23,471 INFO [train.py:904] (6/8) Epoch 25, batch 1900, loss[loss=0.1782, simple_loss=0.2647, pruned_loss=0.04587, over 15513.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2532, pruned_loss=0.03929, over 3312396.72 frames. ], batch size: 192, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:09:49,764 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=245521.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:10:11,812 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.101e+02 2.496e+02 2.952e+02 1.304e+03, threshold=4.992e+02, percent-clipped=2.0 2023-05-01 23:10:21,016 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=245543.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:10:21,098 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7397, 3.8989, 2.5880, 4.5306, 3.0689, 4.4738, 2.6148, 3.2215], device='cuda:6'), covar=tensor([0.0400, 0.0487, 0.1680, 0.0334, 0.0912, 0.0542, 0.1621, 0.0841], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0181, 0.0197, 0.0173, 0.0180, 0.0222, 0.0206, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 23:10:25,697 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=245546.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:10:33,940 INFO [train.py:904] (6/8) Epoch 25, batch 1950, loss[loss=0.1723, simple_loss=0.2714, pruned_loss=0.03657, over 16632.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2528, pruned_loss=0.03885, over 3319475.22 frames. ], batch size: 62, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:11:07,138 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-01 23:11:26,898 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=245591.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:11:32,399 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3827, 3.8645, 4.3235, 2.2931, 4.5384, 4.6620, 3.4658, 3.5467], device='cuda:6'), covar=tensor([0.0592, 0.0282, 0.0314, 0.1165, 0.0099, 0.0187, 0.0433, 0.0429], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0111, 0.0101, 0.0139, 0.0084, 0.0131, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 23:11:41,387 INFO [train.py:904] (6/8) Epoch 25, batch 2000, loss[loss=0.1622, simple_loss=0.236, pruned_loss=0.0442, over 16788.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2526, pruned_loss=0.03895, over 3324476.28 frames. ], batch size: 102, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:11:49,074 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=245607.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 23:11:53,215 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=245610.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:11:53,252 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5455, 3.5813, 3.8398, 2.7486, 3.5145, 3.9011, 3.6128, 2.3024], device='cuda:6'), covar=tensor([0.0512, 0.0226, 0.0061, 0.0395, 0.0123, 0.0118, 0.0108, 0.0485], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0088, 0.0088, 0.0135, 0.0100, 0.0112, 0.0097, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 23:12:00,171 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-01 23:12:05,496 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6481, 4.5714, 4.5432, 3.7615, 4.5896, 1.7319, 4.3208, 4.1529], device='cuda:6'), covar=tensor([0.0211, 0.0181, 0.0238, 0.0504, 0.0170, 0.3433, 0.0217, 0.0339], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0169, 0.0209, 0.0183, 0.0186, 0.0215, 0.0199, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:12:31,549 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.177e+02 2.547e+02 3.013e+02 5.081e+02, threshold=5.093e+02, percent-clipped=1.0 2023-05-01 23:12:50,393 INFO [train.py:904] (6/8) Epoch 25, batch 2050, loss[loss=0.1597, simple_loss=0.2638, pruned_loss=0.02782, over 17182.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2524, pruned_loss=0.03866, over 3329939.82 frames. ], batch size: 46, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:12:57,318 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=245658.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:13:18,993 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8085, 4.7690, 4.6362, 3.7406, 4.7491, 1.6825, 4.4503, 4.2984], device='cuda:6'), covar=tensor([0.0220, 0.0159, 0.0286, 0.0637, 0.0162, 0.3637, 0.0238, 0.0353], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0171, 0.0210, 0.0185, 0.0187, 0.0217, 0.0200, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:13:34,858 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=245686.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 23:13:41,487 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 23:13:58,101 INFO [train.py:904] (6/8) Epoch 25, batch 2100, loss[loss=0.1547, simple_loss=0.2566, pruned_loss=0.02638, over 17178.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2535, pruned_loss=0.03934, over 3333892.12 frames. ], batch size: 46, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:14:12,321 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8212, 5.1073, 4.8692, 4.8986, 4.6501, 4.6297, 4.5869, 5.1682], device='cuda:6'), covar=tensor([0.1284, 0.1037, 0.1173, 0.0913, 0.0875, 0.1222, 0.1218, 0.1058], device='cuda:6'), in_proj_covar=tensor([0.0720, 0.0874, 0.0715, 0.0675, 0.0555, 0.0555, 0.0734, 0.0682], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:14:38,390 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7960, 3.4822, 3.9227, 2.1845, 3.9986, 3.9966, 3.2231, 3.0931], device='cuda:6'), covar=tensor([0.0736, 0.0262, 0.0196, 0.1064, 0.0108, 0.0204, 0.0433, 0.0416], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0111, 0.0100, 0.0140, 0.0084, 0.0131, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 23:14:49,017 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 2.201e+02 2.533e+02 2.997e+02 6.005e+02, threshold=5.066e+02, percent-clipped=2.0 2023-05-01 23:14:58,048 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-05-01 23:14:59,021 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=245747.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 23:15:07,742 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3600, 4.3624, 4.7169, 4.7354, 4.7807, 4.4651, 4.4787, 4.3469], device='cuda:6'), covar=tensor([0.0432, 0.0777, 0.0476, 0.0446, 0.0525, 0.0522, 0.0845, 0.0749], device='cuda:6'), in_proj_covar=tensor([0.0432, 0.0483, 0.0472, 0.0432, 0.0515, 0.0497, 0.0573, 0.0392], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 23:15:08,425 INFO [train.py:904] (6/8) Epoch 25, batch 2150, loss[loss=0.155, simple_loss=0.2405, pruned_loss=0.03478, over 16831.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2543, pruned_loss=0.03974, over 3336047.63 frames. ], batch size: 102, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:15:25,594 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9917, 5.0141, 5.4031, 5.4126, 5.4642, 5.1245, 5.0929, 4.8739], device='cuda:6'), covar=tensor([0.0371, 0.0580, 0.0523, 0.0460, 0.0437, 0.0463, 0.0869, 0.0450], device='cuda:6'), in_proj_covar=tensor([0.0432, 0.0483, 0.0472, 0.0432, 0.0515, 0.0497, 0.0574, 0.0392], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-01 23:16:00,576 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=245792.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:16:15,986 INFO [train.py:904] (6/8) Epoch 25, batch 2200, loss[loss=0.1405, simple_loss=0.2375, pruned_loss=0.02179, over 17068.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2551, pruned_loss=0.04037, over 3336784.10 frames. ], batch size: 50, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:16:34,760 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=245816.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:16:48,014 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7843, 3.8646, 3.0096, 2.3577, 2.4703, 2.4101, 3.9623, 3.3862], device='cuda:6'), covar=tensor([0.2636, 0.0577, 0.1681, 0.3035, 0.2783, 0.2167, 0.0520, 0.1441], device='cuda:6'), in_proj_covar=tensor([0.0331, 0.0274, 0.0311, 0.0321, 0.0303, 0.0271, 0.0302, 0.0348], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 23:17:06,932 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.642e+02 2.208e+02 2.555e+02 2.978e+02 8.197e+02, threshold=5.110e+02, percent-clipped=1.0 2023-05-01 23:17:23,656 INFO [train.py:904] (6/8) Epoch 25, batch 2250, loss[loss=0.1821, simple_loss=0.2628, pruned_loss=0.05069, over 16777.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2562, pruned_loss=0.04074, over 3325450.44 frames. ], batch size: 83, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:17:24,726 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=245853.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:17:25,051 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-01 23:18:16,634 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=245891.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:18:32,205 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=245902.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 23:18:32,949 INFO [train.py:904] (6/8) Epoch 25, batch 2300, loss[loss=0.1619, simple_loss=0.2494, pruned_loss=0.03722, over 16816.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.257, pruned_loss=0.04119, over 3312154.40 frames. ], batch size: 39, lr: 2.71e-03, grad_scale: 2.0 2023-05-01 23:19:24,138 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.666e+02 2.182e+02 2.492e+02 2.872e+02 5.312e+02, threshold=4.984e+02, percent-clipped=2.0 2023-05-01 23:19:42,203 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=245952.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:19:42,915 INFO [train.py:904] (6/8) Epoch 25, batch 2350, loss[loss=0.1548, simple_loss=0.2438, pruned_loss=0.03291, over 15948.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2571, pruned_loss=0.04124, over 3321007.03 frames. ], batch size: 35, lr: 2.71e-03, grad_scale: 2.0 2023-05-01 23:20:39,689 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3312, 5.3643, 5.1635, 4.6487, 5.2233, 2.1699, 4.9934, 5.0868], device='cuda:6'), covar=tensor([0.0096, 0.0075, 0.0226, 0.0383, 0.0106, 0.2581, 0.0131, 0.0198], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0170, 0.0210, 0.0185, 0.0187, 0.0216, 0.0200, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:20:54,783 INFO [train.py:904] (6/8) Epoch 25, batch 2400, loss[loss=0.1564, simple_loss=0.2536, pruned_loss=0.02956, over 17126.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2572, pruned_loss=0.04134, over 3328290.77 frames. ], batch size: 47, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:21:27,948 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2023-05-01 23:21:46,453 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.068e+02 2.493e+02 3.230e+02 1.350e+03, threshold=4.987e+02, percent-clipped=5.0 2023-05-01 23:21:49,265 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=246042.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 23:22:04,193 INFO [train.py:904] (6/8) Epoch 25, batch 2450, loss[loss=0.1625, simple_loss=0.2573, pruned_loss=0.03382, over 16650.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.258, pruned_loss=0.04108, over 3331500.36 frames. ], batch size: 62, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:23:10,964 INFO [train.py:904] (6/8) Epoch 25, batch 2500, loss[loss=0.1678, simple_loss=0.2503, pruned_loss=0.04268, over 16478.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2576, pruned_loss=0.04068, over 3329217.33 frames. ], batch size: 75, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:23:13,024 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2023-05-01 23:23:21,523 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=246111.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:23:28,131 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246116.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:24:01,582 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.119e+02 2.579e+02 3.197e+02 6.008e+02, threshold=5.158e+02, percent-clipped=3.0 2023-05-01 23:24:12,431 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=246148.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:24:18,464 INFO [train.py:904] (6/8) Epoch 25, batch 2550, loss[loss=0.1399, simple_loss=0.2311, pruned_loss=0.02431, over 16834.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2577, pruned_loss=0.04063, over 3328752.07 frames. ], batch size: 42, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:24:26,737 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0918, 5.0244, 4.9493, 4.4147, 4.6296, 4.9913, 4.9279, 4.5838], device='cuda:6'), covar=tensor([0.0641, 0.0656, 0.0334, 0.0427, 0.1033, 0.0535, 0.0335, 0.0835], device='cuda:6'), in_proj_covar=tensor([0.0314, 0.0471, 0.0365, 0.0370, 0.0369, 0.0423, 0.0250, 0.0442], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 23:24:35,187 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246164.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:24:45,509 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=246172.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:25:17,944 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=246196.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 23:25:26,235 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246202.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 23:25:26,988 INFO [train.py:904] (6/8) Epoch 25, batch 2600, loss[loss=0.176, simple_loss=0.2645, pruned_loss=0.04377, over 16521.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2576, pruned_loss=0.04032, over 3325983.29 frames. ], batch size: 75, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:25:36,913 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5118, 4.0382, 4.5260, 2.4274, 4.6527, 4.7545, 3.5011, 3.7616], device='cuda:6'), covar=tensor([0.0550, 0.0241, 0.0158, 0.1094, 0.0075, 0.0179, 0.0406, 0.0368], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0111, 0.0101, 0.0140, 0.0084, 0.0131, 0.0131, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 23:25:52,179 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=246221.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:26:18,784 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.097e+02 2.510e+02 3.006e+02 6.140e+02, threshold=5.021e+02, percent-clipped=2.0 2023-05-01 23:26:29,135 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=246247.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:26:33,158 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246250.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:26:37,198 INFO [train.py:904] (6/8) Epoch 25, batch 2650, loss[loss=0.2, simple_loss=0.2713, pruned_loss=0.06438, over 16860.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2572, pruned_loss=0.03968, over 3333692.88 frames. ], batch size: 116, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:26:42,312 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=246257.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 23:26:43,486 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2148, 5.2199, 4.9445, 4.4321, 5.1083, 1.7226, 4.8407, 4.6852], device='cuda:6'), covar=tensor([0.0107, 0.0088, 0.0239, 0.0388, 0.0098, 0.3201, 0.0133, 0.0284], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0170, 0.0209, 0.0185, 0.0187, 0.0216, 0.0199, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:27:17,669 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=246282.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:27:44,598 INFO [train.py:904] (6/8) Epoch 25, batch 2700, loss[loss=0.1961, simple_loss=0.2687, pruned_loss=0.06178, over 16767.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2571, pruned_loss=0.03932, over 3335469.63 frames. ], batch size: 124, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:28:20,334 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0462, 3.2575, 2.9626, 5.1879, 4.3012, 4.3334, 2.0496, 3.4754], device='cuda:6'), covar=tensor([0.1328, 0.0754, 0.1141, 0.0206, 0.0280, 0.0402, 0.1535, 0.0751], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0179, 0.0198, 0.0198, 0.0206, 0.0219, 0.0206, 0.0197], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 23:28:34,092 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 1.968e+02 2.327e+02 2.625e+02 5.550e+02, threshold=4.654e+02, percent-clipped=1.0 2023-05-01 23:28:36,934 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246342.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 23:28:52,088 INFO [train.py:904] (6/8) Epoch 25, batch 2750, loss[loss=0.1828, simple_loss=0.26, pruned_loss=0.05279, over 16775.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2574, pruned_loss=0.03945, over 3330892.49 frames. ], batch size: 124, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:28:59,229 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7836, 4.2896, 3.0357, 2.3840, 2.7123, 2.6276, 4.6752, 3.4907], device='cuda:6'), covar=tensor([0.3201, 0.0627, 0.2064, 0.3128, 0.3133, 0.2207, 0.0435, 0.1643], device='cuda:6'), in_proj_covar=tensor([0.0332, 0.0274, 0.0311, 0.0321, 0.0303, 0.0272, 0.0302, 0.0349], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-01 23:29:02,731 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-05-01 23:29:13,090 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-05-01 23:29:31,133 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1107, 4.5840, 4.5732, 3.1681, 3.7853, 4.5162, 4.0428, 2.6414], device='cuda:6'), covar=tensor([0.0476, 0.0073, 0.0050, 0.0395, 0.0157, 0.0109, 0.0100, 0.0507], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0088, 0.0089, 0.0136, 0.0101, 0.0112, 0.0098, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 23:29:44,292 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246390.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 23:30:01,769 INFO [train.py:904] (6/8) Epoch 25, batch 2800, loss[loss=0.1831, simple_loss=0.266, pruned_loss=0.05006, over 16418.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2574, pruned_loss=0.03944, over 3331453.19 frames. ], batch size: 75, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:30:54,341 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 1.989e+02 2.330e+02 2.808e+02 5.234e+02, threshold=4.661e+02, percent-clipped=4.0 2023-05-01 23:31:04,338 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246448.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:31:10,742 INFO [train.py:904] (6/8) Epoch 25, batch 2850, loss[loss=0.1709, simple_loss=0.2623, pruned_loss=0.0398, over 17139.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2561, pruned_loss=0.03919, over 3327907.94 frames. ], batch size: 49, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:31:31,939 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=246467.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:32:10,940 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246496.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:32:20,906 INFO [train.py:904] (6/8) Epoch 25, batch 2900, loss[loss=0.1473, simple_loss=0.2261, pruned_loss=0.03426, over 16888.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2552, pruned_loss=0.03972, over 3323550.58 frames. ], batch size: 96, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:33:14,957 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.640e+02 2.258e+02 2.678e+02 3.242e+02 4.768e+02, threshold=5.357e+02, percent-clipped=1.0 2023-05-01 23:33:25,424 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246547.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:33:32,670 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=246552.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 23:33:33,511 INFO [train.py:904] (6/8) Epoch 25, batch 2950, loss[loss=0.1499, simple_loss=0.2424, pruned_loss=0.02875, over 17035.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2542, pruned_loss=0.04002, over 3321318.89 frames. ], batch size: 50, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:34:06,277 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=246577.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:34:32,302 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246595.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:34:42,715 INFO [train.py:904] (6/8) Epoch 25, batch 3000, loss[loss=0.1638, simple_loss=0.2564, pruned_loss=0.03561, over 17067.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2542, pruned_loss=0.0402, over 3326076.88 frames. ], batch size: 55, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:34:42,716 INFO [train.py:929] (6/8) Computing validation loss 2023-05-01 23:34:52,580 INFO [train.py:938] (6/8) Epoch 25, validation: loss=0.1341, simple_loss=0.239, pruned_loss=0.01457, over 944034.00 frames. 2023-05-01 23:34:52,581 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-01 23:35:46,635 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.065e+02 2.466e+02 2.875e+02 5.514e+02, threshold=4.932e+02, percent-clipped=1.0 2023-05-01 23:36:05,624 INFO [train.py:904] (6/8) Epoch 25, batch 3050, loss[loss=0.1811, simple_loss=0.2636, pruned_loss=0.04932, over 16559.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2542, pruned_loss=0.04042, over 3323649.46 frames. ], batch size: 68, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:36:38,364 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=246677.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:37:13,563 INFO [train.py:904] (6/8) Epoch 25, batch 3100, loss[loss=0.1514, simple_loss=0.2545, pruned_loss=0.02417, over 17050.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2548, pruned_loss=0.04058, over 3325378.22 frames. ], batch size: 50, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:37:26,130 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4415, 3.0559, 3.3032, 1.9758, 3.4333, 3.4368, 2.8921, 2.7385], device='cuda:6'), covar=tensor([0.0737, 0.0283, 0.0273, 0.1098, 0.0133, 0.0228, 0.0462, 0.0434], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0112, 0.0101, 0.0141, 0.0084, 0.0132, 0.0131, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 23:38:05,049 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=246738.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:38:08,780 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 2.159e+02 2.563e+02 3.074e+02 4.728e+02, threshold=5.126e+02, percent-clipped=0.0 2023-05-01 23:38:25,017 INFO [train.py:904] (6/8) Epoch 25, batch 3150, loss[loss=0.1723, simple_loss=0.2594, pruned_loss=0.04262, over 16393.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.254, pruned_loss=0.04046, over 3321071.46 frames. ], batch size: 68, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:38:44,296 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246767.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:39:34,228 INFO [train.py:904] (6/8) Epoch 25, batch 3200, loss[loss=0.1654, simple_loss=0.2553, pruned_loss=0.03775, over 15669.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2535, pruned_loss=0.04056, over 3319919.26 frames. ], batch size: 190, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:39:43,994 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9434, 3.0105, 2.8533, 5.1699, 4.2721, 4.5750, 1.7690, 3.3954], device='cuda:6'), covar=tensor([0.1340, 0.0770, 0.1210, 0.0205, 0.0234, 0.0375, 0.1647, 0.0754], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0178, 0.0196, 0.0197, 0.0205, 0.0217, 0.0205, 0.0196], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 23:39:51,749 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246815.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:39:56,005 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4212, 4.4095, 4.3360, 3.8142, 4.4182, 1.6929, 4.1659, 3.9342], device='cuda:6'), covar=tensor([0.0138, 0.0122, 0.0211, 0.0301, 0.0099, 0.2940, 0.0155, 0.0236], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0172, 0.0212, 0.0187, 0.0189, 0.0218, 0.0202, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:40:03,273 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9828, 4.9856, 4.7967, 4.2736, 4.9341, 2.1508, 4.6667, 4.5160], device='cuda:6'), covar=tensor([0.0141, 0.0119, 0.0235, 0.0378, 0.0110, 0.2649, 0.0167, 0.0238], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0172, 0.0212, 0.0187, 0.0189, 0.0218, 0.0202, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:40:27,530 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.068e+02 2.379e+02 2.953e+02 6.702e+02, threshold=4.759e+02, percent-clipped=1.0 2023-05-01 23:40:32,095 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3231, 5.4065, 5.1569, 4.6127, 4.5857, 5.3310, 5.2380, 4.7737], device='cuda:6'), covar=tensor([0.0673, 0.0502, 0.0382, 0.0446, 0.1529, 0.0485, 0.0289, 0.0848], device='cuda:6'), in_proj_covar=tensor([0.0317, 0.0473, 0.0367, 0.0373, 0.0374, 0.0427, 0.0251, 0.0445], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 23:40:42,115 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246852.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 23:40:42,941 INFO [train.py:904] (6/8) Epoch 25, batch 3250, loss[loss=0.1608, simple_loss=0.26, pruned_loss=0.03077, over 17096.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2531, pruned_loss=0.0405, over 3314551.09 frames. ], batch size: 49, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:41:16,981 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246877.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:41:49,852 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246900.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 23:41:53,758 INFO [train.py:904] (6/8) Epoch 25, batch 3300, loss[loss=0.1525, simple_loss=0.2505, pruned_loss=0.02723, over 17140.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2536, pruned_loss=0.04, over 3324303.71 frames. ], batch size: 48, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:42:24,637 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246925.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:42:36,875 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=246934.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:42:46,635 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.089e+02 2.401e+02 2.822e+02 3.775e+02, threshold=4.802e+02, percent-clipped=0.0 2023-05-01 23:43:02,686 INFO [train.py:904] (6/8) Epoch 25, batch 3350, loss[loss=0.1633, simple_loss=0.2482, pruned_loss=0.03922, over 16804.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2543, pruned_loss=0.03926, over 3319248.82 frames. ], batch size: 102, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:44:00,588 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1480, 5.1132, 4.8952, 4.3540, 5.0162, 1.9284, 4.7218, 4.7541], device='cuda:6'), covar=tensor([0.0095, 0.0102, 0.0246, 0.0448, 0.0116, 0.2871, 0.0160, 0.0272], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0172, 0.0212, 0.0188, 0.0190, 0.0218, 0.0202, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:44:01,835 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=246995.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:44:04,472 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.73 vs. limit=5.0 2023-05-01 23:44:13,546 INFO [train.py:904] (6/8) Epoch 25, batch 3400, loss[loss=0.1927, simple_loss=0.2689, pruned_loss=0.0583, over 16184.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2542, pruned_loss=0.03939, over 3324526.73 frames. ], batch size: 165, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:44:56,883 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247033.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:45:08,922 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.556e+02 2.118e+02 2.456e+02 2.904e+02 5.980e+02, threshold=4.912e+02, percent-clipped=3.0 2023-05-01 23:45:26,146 INFO [train.py:904] (6/8) Epoch 25, batch 3450, loss[loss=0.1773, simple_loss=0.2616, pruned_loss=0.04652, over 16329.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2521, pruned_loss=0.0384, over 3334712.92 frames. ], batch size: 165, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:45:53,762 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-01 23:46:35,327 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-05-01 23:46:35,595 INFO [train.py:904] (6/8) Epoch 25, batch 3500, loss[loss=0.1454, simple_loss=0.2351, pruned_loss=0.02792, over 15770.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2511, pruned_loss=0.03829, over 3331709.97 frames. ], batch size: 35, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:46:43,783 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4661, 5.8609, 5.6010, 5.6837, 5.3297, 5.3065, 5.2414, 6.0215], device='cuda:6'), covar=tensor([0.1644, 0.1092, 0.1085, 0.0944, 0.0966, 0.0745, 0.1323, 0.0978], device='cuda:6'), in_proj_covar=tensor([0.0730, 0.0880, 0.0722, 0.0682, 0.0559, 0.0560, 0.0742, 0.0690], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:47:07,357 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2173, 5.2155, 5.1017, 4.5591, 4.7254, 5.1536, 5.0379, 4.7020], device='cuda:6'), covar=tensor([0.0653, 0.0643, 0.0332, 0.0387, 0.1201, 0.0513, 0.0331, 0.0883], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0474, 0.0368, 0.0374, 0.0374, 0.0430, 0.0251, 0.0448], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 23:47:23,728 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 23:47:27,009 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9960, 4.9133, 4.7961, 3.5510, 4.8579, 1.6527, 4.5077, 4.5137], device='cuda:6'), covar=tensor([0.0190, 0.0168, 0.0331, 0.0900, 0.0192, 0.4013, 0.0261, 0.0442], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0172, 0.0213, 0.0188, 0.0191, 0.0219, 0.0202, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:47:30,017 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.124e+02 2.460e+02 2.748e+02 9.578e+02, threshold=4.920e+02, percent-clipped=1.0 2023-05-01 23:47:45,772 INFO [train.py:904] (6/8) Epoch 25, batch 3550, loss[loss=0.1901, simple_loss=0.2617, pruned_loss=0.05926, over 16757.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2503, pruned_loss=0.03798, over 3322975.28 frames. ], batch size: 124, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:48:55,167 INFO [train.py:904] (6/8) Epoch 25, batch 3600, loss[loss=0.1709, simple_loss=0.2494, pruned_loss=0.04619, over 16692.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2498, pruned_loss=0.03767, over 3326921.58 frames. ], batch size: 134, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:49:35,677 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5613, 3.6745, 4.1170, 2.3419, 3.4343, 2.7400, 4.0329, 3.9110], device='cuda:6'), covar=tensor([0.0261, 0.0952, 0.0440, 0.1922, 0.0761, 0.0928, 0.0611, 0.1045], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0170, 0.0171, 0.0156, 0.0148, 0.0133, 0.0147, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-01 23:49:49,107 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.664e+02 2.145e+02 2.569e+02 3.333e+02 7.279e+02, threshold=5.139e+02, percent-clipped=3.0 2023-05-01 23:50:05,473 INFO [train.py:904] (6/8) Epoch 25, batch 3650, loss[loss=0.1631, simple_loss=0.2349, pruned_loss=0.04562, over 16784.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.248, pruned_loss=0.03823, over 3315462.92 frames. ], batch size: 124, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:50:09,637 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2828, 2.3534, 2.4286, 4.0563, 2.3613, 2.6764, 2.4128, 2.5312], device='cuda:6'), covar=tensor([0.1544, 0.3757, 0.3122, 0.0647, 0.3983, 0.2765, 0.3899, 0.3073], device='cuda:6'), in_proj_covar=tensor([0.0417, 0.0465, 0.0384, 0.0339, 0.0445, 0.0534, 0.0438, 0.0546], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:50:59,484 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247290.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:51:17,356 INFO [train.py:904] (6/8) Epoch 25, batch 3700, loss[loss=0.1679, simple_loss=0.2413, pruned_loss=0.04721, over 16905.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2469, pruned_loss=0.03977, over 3297756.03 frames. ], batch size: 109, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:52:01,256 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=247333.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:52:06,618 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4480, 2.5724, 2.4186, 4.2689, 2.3676, 2.8730, 2.5712, 2.6943], device='cuda:6'), covar=tensor([0.1371, 0.3301, 0.2967, 0.0554, 0.3985, 0.2299, 0.3228, 0.3168], device='cuda:6'), in_proj_covar=tensor([0.0418, 0.0467, 0.0385, 0.0339, 0.0446, 0.0535, 0.0439, 0.0547], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:52:12,943 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.806e+02 2.292e+02 2.615e+02 3.085e+02 5.082e+02, threshold=5.229e+02, percent-clipped=0.0 2023-05-01 23:52:18,296 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7007, 1.8469, 2.3406, 2.5531, 2.6556, 2.6809, 1.9522, 2.8170], device='cuda:6'), covar=tensor([0.0200, 0.0543, 0.0345, 0.0319, 0.0349, 0.0341, 0.0573, 0.0203], device='cuda:6'), in_proj_covar=tensor([0.0197, 0.0196, 0.0185, 0.0189, 0.0205, 0.0163, 0.0200, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-01 23:52:23,166 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-05-01 23:52:30,276 INFO [train.py:904] (6/8) Epoch 25, batch 3750, loss[loss=0.1819, simple_loss=0.2692, pruned_loss=0.04731, over 17109.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2482, pruned_loss=0.04144, over 3256744.36 frames. ], batch size: 49, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:53:07,563 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247379.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:53:10,258 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=247381.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:53:42,910 INFO [train.py:904] (6/8) Epoch 25, batch 3800, loss[loss=0.1838, simple_loss=0.2683, pruned_loss=0.04962, over 12334.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2501, pruned_loss=0.04288, over 3250815.28 frames. ], batch size: 246, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:54:16,101 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247426.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:54:36,450 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=247440.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:54:38,773 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.614e+02 2.181e+02 2.707e+02 3.148e+02 5.686e+02, threshold=5.414e+02, percent-clipped=1.0 2023-05-01 23:54:54,709 INFO [train.py:904] (6/8) Epoch 25, batch 3850, loss[loss=0.1673, simple_loss=0.2467, pruned_loss=0.04392, over 16841.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2509, pruned_loss=0.04376, over 3245280.35 frames. ], batch size: 90, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:55:23,011 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2952, 3.3570, 3.5869, 2.4238, 3.2736, 3.6733, 3.3644, 2.0870], device='cuda:6'), covar=tensor([0.0544, 0.0144, 0.0066, 0.0422, 0.0126, 0.0103, 0.0100, 0.0500], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0088, 0.0089, 0.0135, 0.0101, 0.0112, 0.0098, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-01 23:55:43,337 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=247487.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:56:02,881 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247500.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:56:06,077 INFO [train.py:904] (6/8) Epoch 25, batch 3900, loss[loss=0.1562, simple_loss=0.2347, pruned_loss=0.03886, over 16762.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2508, pruned_loss=0.04469, over 3250876.35 frames. ], batch size: 83, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:56:20,899 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.32 vs. limit=5.0 2023-05-01 23:56:39,781 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247525.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:57:00,699 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2312, 4.2717, 4.4303, 4.1474, 4.3184, 4.8150, 4.3572, 4.0051], device='cuda:6'), covar=tensor([0.1797, 0.2318, 0.2370, 0.2345, 0.2531, 0.1209, 0.1681, 0.2765], device='cuda:6'), in_proj_covar=tensor([0.0430, 0.0639, 0.0696, 0.0518, 0.0691, 0.0725, 0.0541, 0.0693], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-01 23:57:03,109 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.630e+02 2.139e+02 2.490e+02 2.933e+02 5.052e+02, threshold=4.980e+02, percent-clipped=0.0 2023-05-01 23:57:13,495 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8647, 2.7263, 2.5514, 4.2340, 3.5091, 4.0127, 1.7633, 2.8146], device='cuda:6'), covar=tensor([0.1303, 0.0706, 0.1157, 0.0211, 0.0224, 0.0510, 0.1492, 0.0970], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0179, 0.0197, 0.0198, 0.0207, 0.0219, 0.0207, 0.0197], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-01 23:57:17,841 INFO [train.py:904] (6/8) Epoch 25, batch 3950, loss[loss=0.1857, simple_loss=0.2624, pruned_loss=0.05448, over 16446.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2506, pruned_loss=0.04492, over 3258807.70 frames. ], batch size: 146, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:57:19,971 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247554.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:57:30,911 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=247561.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:58:06,039 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=247586.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:58:12,183 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=247590.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:58:30,186 INFO [train.py:904] (6/8) Epoch 25, batch 4000, loss[loss=0.1811, simple_loss=0.2581, pruned_loss=0.05202, over 16888.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2504, pruned_loss=0.04523, over 3264612.17 frames. ], batch size: 116, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:58:49,589 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=247615.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:59:16,392 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 23:59:21,889 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=247638.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:59:27,473 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 1.992e+02 2.395e+02 2.922e+02 5.946e+02, threshold=4.790e+02, percent-clipped=1.0 2023-05-01 23:59:44,159 INFO [train.py:904] (6/8) Epoch 25, batch 4050, loss[loss=0.1541, simple_loss=0.2429, pruned_loss=0.03264, over 16491.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2513, pruned_loss=0.04448, over 3265958.22 frames. ], batch size: 68, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:00:00,768 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.47 vs. limit=2.0 2023-05-02 00:00:59,187 INFO [train.py:904] (6/8) Epoch 25, batch 4100, loss[loss=0.193, simple_loss=0.2885, pruned_loss=0.04871, over 16681.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2525, pruned_loss=0.04374, over 3255195.19 frames. ], batch size: 89, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:01:48,423 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247735.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:01:57,927 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.293e+02 2.110e+02 2.475e+02 2.947e+02 8.069e+02, threshold=4.949e+02, percent-clipped=1.0 2023-05-02 00:02:14,951 INFO [train.py:904] (6/8) Epoch 25, batch 4150, loss[loss=0.209, simple_loss=0.3021, pruned_loss=0.05798, over 16221.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2601, pruned_loss=0.04683, over 3197363.79 frames. ], batch size: 165, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:02:38,426 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8714, 1.9765, 2.4508, 2.7724, 2.7673, 3.2920, 2.1652, 3.2282], device='cuda:6'), covar=tensor([0.0239, 0.0554, 0.0363, 0.0356, 0.0343, 0.0183, 0.0572, 0.0162], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0196, 0.0185, 0.0188, 0.0204, 0.0163, 0.0199, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:02:43,739 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 00:03:01,856 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247782.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:03:05,062 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247784.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:03:30,117 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2649, 3.5665, 3.5020, 2.3512, 3.2338, 3.6153, 3.2853, 2.0915], device='cuda:6'), covar=tensor([0.0519, 0.0056, 0.0071, 0.0419, 0.0114, 0.0114, 0.0115, 0.0482], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0087, 0.0088, 0.0135, 0.0101, 0.0111, 0.0098, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 00:03:32,740 INFO [train.py:904] (6/8) Epoch 25, batch 4200, loss[loss=0.2269, simple_loss=0.3159, pruned_loss=0.06895, over 15378.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.2671, pruned_loss=0.04801, over 3192123.55 frames. ], batch size: 190, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:04:06,577 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-02 00:04:30,293 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.155e+02 2.612e+02 3.147e+02 5.112e+02, threshold=5.224e+02, percent-clipped=1.0 2023-05-02 00:04:35,308 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=247845.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:04:46,255 INFO [train.py:904] (6/8) Epoch 25, batch 4250, loss[loss=0.1586, simple_loss=0.2599, pruned_loss=0.02868, over 16912.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2702, pruned_loss=0.04778, over 3172736.36 frames. ], batch size: 96, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:04:52,104 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247856.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:05:16,165 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9967, 2.1828, 2.2940, 2.5725, 1.7453, 3.1369, 1.7814, 2.7149], device='cuda:6'), covar=tensor([0.1179, 0.0826, 0.1176, 0.0185, 0.0127, 0.0441, 0.1628, 0.0806], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0178, 0.0197, 0.0197, 0.0206, 0.0218, 0.0206, 0.0197], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:05:28,797 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247881.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:06:02,534 INFO [train.py:904] (6/8) Epoch 25, batch 4300, loss[loss=0.1875, simple_loss=0.2835, pruned_loss=0.04574, over 16758.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2709, pruned_loss=0.04651, over 3173131.00 frames. ], batch size: 83, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:06:06,118 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6162, 3.7201, 2.8312, 2.3253, 2.6488, 2.5037, 4.0647, 3.3913], device='cuda:6'), covar=tensor([0.2872, 0.0742, 0.1872, 0.2583, 0.2454, 0.2139, 0.0485, 0.1308], device='cuda:6'), in_proj_covar=tensor([0.0330, 0.0273, 0.0310, 0.0319, 0.0303, 0.0269, 0.0301, 0.0345], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 00:06:13,489 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247910.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:07:01,392 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.557e+02 2.163e+02 2.477e+02 3.039e+02 5.024e+02, threshold=4.955e+02, percent-clipped=0.0 2023-05-02 00:07:17,409 INFO [train.py:904] (6/8) Epoch 25, batch 4350, loss[loss=0.1972, simple_loss=0.2835, pruned_loss=0.05547, over 16945.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2739, pruned_loss=0.04739, over 3177029.70 frames. ], batch size: 109, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:07:30,431 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5726, 3.3682, 3.8769, 1.9560, 4.0114, 4.0965, 2.9361, 3.0843], device='cuda:6'), covar=tensor([0.0804, 0.0304, 0.0208, 0.1144, 0.0074, 0.0117, 0.0507, 0.0424], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0111, 0.0100, 0.0139, 0.0085, 0.0130, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:07:46,517 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5308, 3.4029, 3.9084, 1.9286, 4.0473, 4.1296, 2.9794, 3.1017], device='cuda:6'), covar=tensor([0.0891, 0.0321, 0.0231, 0.1267, 0.0083, 0.0126, 0.0517, 0.0440], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0111, 0.0100, 0.0139, 0.0085, 0.0130, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:08:05,109 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247984.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:08:24,070 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247996.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:08:36,576 INFO [train.py:904] (6/8) Epoch 25, batch 4400, loss[loss=0.2091, simple_loss=0.2991, pruned_loss=0.05957, over 16375.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2758, pruned_loss=0.04839, over 3186236.78 frames. ], batch size: 35, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:09:04,137 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3789, 2.5629, 2.9406, 3.3035, 3.1876, 3.9119, 2.4085, 3.7766], device='cuda:6'), covar=tensor([0.0187, 0.0436, 0.0271, 0.0247, 0.0260, 0.0123, 0.0548, 0.0129], device='cuda:6'), in_proj_covar=tensor([0.0195, 0.0195, 0.0184, 0.0188, 0.0203, 0.0161, 0.0199, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:09:24,791 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248035.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:09:27,651 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0388, 4.5176, 3.3700, 2.7381, 3.2204, 2.9621, 4.9247, 4.0005], device='cuda:6'), covar=tensor([0.2573, 0.0514, 0.1559, 0.2228, 0.2175, 0.1781, 0.0335, 0.1014], device='cuda:6'), in_proj_covar=tensor([0.0331, 0.0274, 0.0311, 0.0321, 0.0305, 0.0270, 0.0302, 0.0348], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 00:09:34,452 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 1.997e+02 2.337e+02 2.699e+02 4.708e+02, threshold=4.674e+02, percent-clipped=0.0 2023-05-02 00:09:39,364 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248045.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:09:50,302 INFO [train.py:904] (6/8) Epoch 25, batch 4450, loss[loss=0.2089, simple_loss=0.2903, pruned_loss=0.06374, over 17012.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2793, pruned_loss=0.04976, over 3193788.64 frames. ], batch size: 53, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:09:56,488 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248057.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:10:33,174 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248082.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:10:33,193 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=248082.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:10:35,245 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248083.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:11:04,506 INFO [train.py:904] (6/8) Epoch 25, batch 4500, loss[loss=0.1932, simple_loss=0.2806, pruned_loss=0.05289, over 16690.00 frames. ], tot_loss[loss=0.1906, simple_loss=0.2799, pruned_loss=0.05059, over 3205130.02 frames. ], batch size: 62, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:11:20,259 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1518, 3.8156, 4.3699, 2.2712, 4.6650, 4.7440, 3.3249, 3.7405], device='cuda:6'), covar=tensor([0.0678, 0.0303, 0.0205, 0.1088, 0.0062, 0.0094, 0.0435, 0.0354], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0112, 0.0101, 0.0140, 0.0085, 0.0131, 0.0131, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:11:44,570 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248130.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:12:00,004 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=248140.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:12:01,968 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 1.873e+02 2.071e+02 2.399e+02 5.095e+02, threshold=4.142e+02, percent-clipped=1.0 2023-05-02 00:12:04,315 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248143.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:12:17,509 INFO [train.py:904] (6/8) Epoch 25, batch 4550, loss[loss=0.2039, simple_loss=0.2914, pruned_loss=0.05823, over 16898.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2809, pruned_loss=0.05157, over 3212980.23 frames. ], batch size: 116, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:12:22,838 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248156.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:12:44,071 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.44 vs. limit=2.0 2023-05-02 00:12:51,586 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8902, 4.6532, 4.8078, 5.0644, 5.2174, 4.7236, 5.2726, 5.2773], device='cuda:6'), covar=tensor([0.1687, 0.1347, 0.1837, 0.0803, 0.0689, 0.0951, 0.0671, 0.0662], device='cuda:6'), in_proj_covar=tensor([0.0662, 0.0815, 0.0939, 0.0824, 0.0632, 0.0649, 0.0676, 0.0790], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:13:00,054 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248181.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:13:32,543 INFO [train.py:904] (6/8) Epoch 25, batch 4600, loss[loss=0.184, simple_loss=0.2739, pruned_loss=0.04705, over 16741.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2818, pruned_loss=0.0517, over 3219979.64 frames. ], batch size: 57, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:13:34,234 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248204.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:13:43,442 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248210.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:13:44,619 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0203, 5.2929, 5.0695, 5.1315, 4.8394, 4.6938, 4.6981, 5.3980], device='cuda:6'), covar=tensor([0.1217, 0.0788, 0.0970, 0.0779, 0.0826, 0.0964, 0.1112, 0.0836], device='cuda:6'), in_proj_covar=tensor([0.0703, 0.0847, 0.0699, 0.0657, 0.0539, 0.0543, 0.0714, 0.0667], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:14:11,936 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248229.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:14:29,187 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.305e+02 1.740e+02 1.903e+02 2.276e+02 6.899e+02, threshold=3.806e+02, percent-clipped=0.0 2023-05-02 00:14:46,953 INFO [train.py:904] (6/8) Epoch 25, batch 4650, loss[loss=0.18, simple_loss=0.2667, pruned_loss=0.04665, over 16893.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2812, pruned_loss=0.05214, over 3222595.85 frames. ], batch size: 109, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:14:54,074 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248258.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:15:13,685 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6534, 2.6572, 2.6423, 4.6809, 3.4625, 4.0691, 1.7022, 3.0733], device='cuda:6'), covar=tensor([0.1395, 0.0895, 0.1186, 0.0166, 0.0309, 0.0373, 0.1675, 0.0811], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0178, 0.0197, 0.0195, 0.0205, 0.0217, 0.0206, 0.0196], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:16:01,796 INFO [train.py:904] (6/8) Epoch 25, batch 4700, loss[loss=0.1625, simple_loss=0.2533, pruned_loss=0.03586, over 16434.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2787, pruned_loss=0.05087, over 3227074.69 frames. ], batch size: 75, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:16:36,506 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5173, 5.5310, 5.4090, 4.6244, 5.5134, 2.1848, 5.1932, 5.0594], device='cuda:6'), covar=tensor([0.0091, 0.0100, 0.0156, 0.0457, 0.0081, 0.2613, 0.0117, 0.0221], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0168, 0.0208, 0.0185, 0.0186, 0.0214, 0.0198, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:16:44,371 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=248332.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:16:52,100 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 00:16:56,011 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=248340.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:16:57,978 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 1.897e+02 2.230e+02 2.532e+02 4.236e+02, threshold=4.460e+02, percent-clipped=2.0 2023-05-02 00:17:13,934 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=248352.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:17:14,736 INFO [train.py:904] (6/8) Epoch 25, batch 4750, loss[loss=0.1717, simple_loss=0.2575, pruned_loss=0.04301, over 16817.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2739, pruned_loss=0.04861, over 3237206.57 frames. ], batch size: 39, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:17:21,489 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4103, 1.7885, 2.1175, 2.4134, 2.4467, 2.7555, 1.8399, 2.5287], device='cuda:6'), covar=tensor([0.0243, 0.0582, 0.0351, 0.0380, 0.0386, 0.0237, 0.0604, 0.0202], device='cuda:6'), in_proj_covar=tensor([0.0195, 0.0195, 0.0183, 0.0188, 0.0204, 0.0162, 0.0199, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:17:59,802 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=248383.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:18:13,813 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248393.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:18:29,675 INFO [train.py:904] (6/8) Epoch 25, batch 4800, loss[loss=0.1741, simple_loss=0.273, pruned_loss=0.03754, over 16367.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2707, pruned_loss=0.0466, over 3228795.82 frames. ], batch size: 146, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:19:22,191 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=248438.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:19:25,662 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248440.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:19:27,480 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 1.792e+02 2.090e+02 2.399e+02 3.533e+02, threshold=4.179e+02, percent-clipped=0.0 2023-05-02 00:19:31,550 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248444.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:19:43,834 INFO [train.py:904] (6/8) Epoch 25, batch 4850, loss[loss=0.1868, simple_loss=0.2773, pruned_loss=0.04815, over 17063.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2716, pruned_loss=0.04646, over 3197149.64 frames. ], batch size: 53, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:20:37,518 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248488.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:20:58,429 INFO [train.py:904] (6/8) Epoch 25, batch 4900, loss[loss=0.2038, simple_loss=0.3062, pruned_loss=0.05067, over 15389.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2719, pruned_loss=0.04585, over 3152393.01 frames. ], batch size: 191, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:22:03,445 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 1.924e+02 2.287e+02 2.776e+02 8.047e+02, threshold=4.573e+02, percent-clipped=3.0 2023-05-02 00:22:20,177 INFO [train.py:904] (6/8) Epoch 25, batch 4950, loss[loss=0.1731, simple_loss=0.2597, pruned_loss=0.04327, over 17241.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2712, pruned_loss=0.04505, over 3164126.12 frames. ], batch size: 45, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:22:28,841 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-02 00:23:05,570 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7064, 3.7945, 2.4389, 4.5994, 2.8989, 4.4100, 2.6522, 3.0428], device='cuda:6'), covar=tensor([0.0330, 0.0389, 0.1652, 0.0117, 0.0834, 0.0496, 0.1418, 0.0839], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0180, 0.0195, 0.0169, 0.0177, 0.0220, 0.0203, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:23:31,266 INFO [train.py:904] (6/8) Epoch 25, batch 5000, loss[loss=0.1587, simple_loss=0.2443, pruned_loss=0.03653, over 17053.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2725, pruned_loss=0.04509, over 3173820.77 frames. ], batch size: 53, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:23:48,457 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3545, 5.2853, 5.2416, 4.3608, 5.2845, 1.8647, 4.9690, 4.9652], device='cuda:6'), covar=tensor([0.0091, 0.0090, 0.0150, 0.0529, 0.0109, 0.2699, 0.0124, 0.0224], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0166, 0.0206, 0.0183, 0.0183, 0.0212, 0.0196, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:24:07,380 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8997, 3.9227, 2.6939, 4.8775, 3.0751, 4.7017, 2.7546, 3.1801], device='cuda:6'), covar=tensor([0.0300, 0.0398, 0.1560, 0.0098, 0.0841, 0.0392, 0.1437, 0.0845], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0179, 0.0195, 0.0168, 0.0177, 0.0219, 0.0203, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:24:25,347 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248640.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:24:27,922 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.446e+02 2.126e+02 2.416e+02 2.928e+02 6.487e+02, threshold=4.832e+02, percent-clipped=1.0 2023-05-02 00:24:42,937 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248652.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:24:43,830 INFO [train.py:904] (6/8) Epoch 25, batch 5050, loss[loss=0.1898, simple_loss=0.2786, pruned_loss=0.05053, over 12221.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.273, pruned_loss=0.04491, over 3176740.77 frames. ], batch size: 248, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:25:35,138 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248688.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:25:35,157 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=248688.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:25:39,595 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=248691.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:25:53,426 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248700.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:25:57,288 INFO [train.py:904] (6/8) Epoch 25, batch 5100, loss[loss=0.162, simple_loss=0.2513, pruned_loss=0.03634, over 17092.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2711, pruned_loss=0.04404, over 3182386.60 frames. ], batch size: 49, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:26:15,804 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=248715.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:26:50,104 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248738.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:26:52,083 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=248739.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:26:56,587 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 1.974e+02 2.293e+02 2.738e+02 8.128e+02, threshold=4.585e+02, percent-clipped=1.0 2023-05-02 00:27:12,721 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248752.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:27:13,425 INFO [train.py:904] (6/8) Epoch 25, batch 5150, loss[loss=0.1505, simple_loss=0.2388, pruned_loss=0.03114, over 17210.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2711, pruned_loss=0.04345, over 3186310.02 frames. ], batch size: 45, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:27:47,441 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248776.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:28:03,458 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248786.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:28:28,598 INFO [train.py:904] (6/8) Epoch 25, batch 5200, loss[loss=0.1639, simple_loss=0.2567, pruned_loss=0.03555, over 16758.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2695, pruned_loss=0.04267, over 3196277.93 frames. ], batch size: 83, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:28:51,464 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9186, 5.1848, 5.3669, 5.0754, 5.1893, 5.7168, 5.1528, 4.9039], device='cuda:6'), covar=tensor([0.0933, 0.1799, 0.1894, 0.1987, 0.2355, 0.0913, 0.1612, 0.2422], device='cuda:6'), in_proj_covar=tensor([0.0416, 0.0609, 0.0664, 0.0496, 0.0663, 0.0698, 0.0519, 0.0665], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 00:28:59,142 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-05-02 00:29:25,206 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.065e+02 2.411e+02 2.935e+02 6.100e+02, threshold=4.822e+02, percent-clipped=2.0 2023-05-02 00:29:41,113 INFO [train.py:904] (6/8) Epoch 25, batch 5250, loss[loss=0.1876, simple_loss=0.2735, pruned_loss=0.05092, over 17221.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2669, pruned_loss=0.04224, over 3208785.89 frames. ], batch size: 45, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:29:49,022 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9302, 5.1996, 4.9599, 5.0255, 4.7497, 4.7286, 4.5816, 5.2861], device='cuda:6'), covar=tensor([0.1336, 0.0825, 0.0977, 0.0818, 0.0801, 0.0932, 0.1254, 0.0874], device='cuda:6'), in_proj_covar=tensor([0.0697, 0.0838, 0.0694, 0.0647, 0.0534, 0.0534, 0.0706, 0.0657], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:30:53,681 INFO [train.py:904] (6/8) Epoch 25, batch 5300, loss[loss=0.1831, simple_loss=0.2611, pruned_loss=0.05248, over 12296.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2639, pruned_loss=0.04132, over 3207692.95 frames. ], batch size: 246, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:31:49,345 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3213, 5.3143, 5.1465, 4.7254, 4.7380, 5.2057, 5.1209, 4.9007], device='cuda:6'), covar=tensor([0.0645, 0.0592, 0.0401, 0.0408, 0.1359, 0.0662, 0.0493, 0.0759], device='cuda:6'), in_proj_covar=tensor([0.0304, 0.0458, 0.0356, 0.0361, 0.0361, 0.0417, 0.0243, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:31:51,346 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 1.862e+02 2.129e+02 2.555e+02 4.739e+02, threshold=4.259e+02, percent-clipped=0.0 2023-05-02 00:32:07,997 INFO [train.py:904] (6/8) Epoch 25, batch 5350, loss[loss=0.1807, simple_loss=0.2747, pruned_loss=0.0433, over 16747.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2628, pruned_loss=0.04095, over 3213026.69 frames. ], batch size: 89, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:33:01,129 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248988.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:33:06,105 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-02 00:33:22,803 INFO [train.py:904] (6/8) Epoch 25, batch 5400, loss[loss=0.1914, simple_loss=0.2894, pruned_loss=0.04671, over 16581.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.265, pruned_loss=0.04137, over 3209616.99 frames. ], batch size: 68, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:33:52,065 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9518, 2.0705, 2.5571, 2.8673, 2.7472, 3.4586, 2.3467, 3.3898], device='cuda:6'), covar=tensor([0.0251, 0.0531, 0.0372, 0.0380, 0.0372, 0.0178, 0.0509, 0.0149], device='cuda:6'), in_proj_covar=tensor([0.0195, 0.0195, 0.0182, 0.0188, 0.0202, 0.0161, 0.0200, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:34:12,061 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=249036.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:34:17,275 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=249039.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:34:20,470 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 1.925e+02 2.196e+02 2.578e+02 3.732e+02, threshold=4.392e+02, percent-clipped=0.0 2023-05-02 00:34:29,100 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=249047.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:34:40,330 INFO [train.py:904] (6/8) Epoch 25, batch 5450, loss[loss=0.1789, simple_loss=0.2727, pruned_loss=0.0425, over 16432.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2679, pruned_loss=0.0427, over 3212577.15 frames. ], batch size: 68, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:35:09,494 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=249071.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:35:27,096 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2088, 4.3060, 4.0977, 3.7924, 3.8222, 4.2000, 3.8965, 3.9759], device='cuda:6'), covar=tensor([0.0571, 0.0709, 0.0295, 0.0296, 0.0708, 0.0581, 0.0798, 0.0575], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0457, 0.0355, 0.0360, 0.0359, 0.0416, 0.0242, 0.0428], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:35:29,404 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-02 00:35:34,925 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=249087.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:35:58,372 INFO [train.py:904] (6/8) Epoch 25, batch 5500, loss[loss=0.2186, simple_loss=0.304, pruned_loss=0.06654, over 16851.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2752, pruned_loss=0.04728, over 3173773.44 frames. ], batch size: 116, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:36:09,911 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.33 vs. limit=5.0 2023-05-02 00:37:00,873 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.871e+02 3.023e+02 3.674e+02 4.693e+02 9.094e+02, threshold=7.348e+02, percent-clipped=33.0 2023-05-02 00:37:17,248 INFO [train.py:904] (6/8) Epoch 25, batch 5550, loss[loss=0.2313, simple_loss=0.3123, pruned_loss=0.0752, over 15281.00 frames. ], tot_loss[loss=0.1934, simple_loss=0.2822, pruned_loss=0.05235, over 3136511.92 frames. ], batch size: 190, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:38:41,393 INFO [train.py:904] (6/8) Epoch 25, batch 5600, loss[loss=0.2784, simple_loss=0.3338, pruned_loss=0.1115, over 11144.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2858, pruned_loss=0.0556, over 3099658.73 frames. ], batch size: 247, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:38:48,329 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0203, 2.1726, 2.3335, 3.6339, 2.0784, 2.5275, 2.3067, 2.3534], device='cuda:6'), covar=tensor([0.1447, 0.3472, 0.2790, 0.0614, 0.4165, 0.2341, 0.3420, 0.3221], device='cuda:6'), in_proj_covar=tensor([0.0413, 0.0461, 0.0378, 0.0332, 0.0440, 0.0528, 0.0433, 0.0539], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:38:51,486 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9516, 4.0331, 4.2866, 4.2771, 4.2881, 4.0305, 4.0521, 4.0513], device='cuda:6'), covar=tensor([0.0403, 0.0644, 0.0430, 0.0421, 0.0473, 0.0488, 0.0926, 0.0551], device='cuda:6'), in_proj_covar=tensor([0.0417, 0.0470, 0.0457, 0.0418, 0.0501, 0.0479, 0.0559, 0.0381], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 00:39:02,213 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=249215.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:39:12,807 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-05-02 00:39:47,609 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.932e+02 3.336e+02 4.056e+02 4.975e+02 8.511e+02, threshold=8.112e+02, percent-clipped=2.0 2023-05-02 00:40:04,579 INFO [train.py:904] (6/8) Epoch 25, batch 5650, loss[loss=0.2006, simple_loss=0.2904, pruned_loss=0.05535, over 16734.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2899, pruned_loss=0.05887, over 3086132.12 frames. ], batch size: 89, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:40:14,926 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2009, 5.6768, 5.8990, 5.5936, 5.6568, 6.2138, 5.6713, 5.4521], device='cuda:6'), covar=tensor([0.0921, 0.1837, 0.2544, 0.1940, 0.2268, 0.1003, 0.1786, 0.2677], device='cuda:6'), in_proj_covar=tensor([0.0418, 0.0615, 0.0673, 0.0500, 0.0668, 0.0705, 0.0524, 0.0672], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 00:40:32,505 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6552, 2.5722, 1.9055, 2.6845, 2.0758, 2.7617, 2.1260, 2.3726], device='cuda:6'), covar=tensor([0.0326, 0.0395, 0.1207, 0.0271, 0.0615, 0.0522, 0.1220, 0.0612], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0180, 0.0195, 0.0168, 0.0177, 0.0219, 0.0203, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:40:41,215 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=249276.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:41:22,869 INFO [train.py:904] (6/8) Epoch 25, batch 5700, loss[loss=0.2335, simple_loss=0.3039, pruned_loss=0.08155, over 11323.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2907, pruned_loss=0.06013, over 3067870.38 frames. ], batch size: 248, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:41:39,418 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.59 vs. limit=5.0 2023-05-02 00:42:05,291 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3291, 3.4789, 3.6148, 3.6039, 3.6076, 3.4460, 3.4717, 3.5237], device='cuda:6'), covar=tensor([0.0466, 0.0699, 0.0551, 0.0497, 0.0586, 0.0618, 0.0963, 0.0576], device='cuda:6'), in_proj_covar=tensor([0.0417, 0.0471, 0.0459, 0.0418, 0.0502, 0.0480, 0.0559, 0.0382], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 00:42:25,334 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.887e+02 2.916e+02 3.392e+02 3.948e+02 9.464e+02, threshold=6.785e+02, percent-clipped=1.0 2023-05-02 00:42:35,247 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=249347.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:42:40,301 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-05-02 00:42:43,861 INFO [train.py:904] (6/8) Epoch 25, batch 5750, loss[loss=0.1979, simple_loss=0.2878, pruned_loss=0.05397, over 15341.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2931, pruned_loss=0.06143, over 3031107.85 frames. ], batch size: 191, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:43:13,886 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3224, 3.4430, 3.5875, 3.5782, 3.5877, 3.4246, 3.4638, 3.5045], device='cuda:6'), covar=tensor([0.0471, 0.0835, 0.0499, 0.0458, 0.0563, 0.0614, 0.0887, 0.0616], device='cuda:6'), in_proj_covar=tensor([0.0416, 0.0469, 0.0457, 0.0416, 0.0500, 0.0478, 0.0556, 0.0381], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 00:43:13,908 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=249371.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:43:46,558 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0949, 3.9595, 4.1292, 4.2596, 4.3900, 4.0051, 4.3065, 4.4050], device='cuda:6'), covar=tensor([0.1774, 0.1278, 0.1568, 0.0780, 0.0703, 0.1438, 0.1018, 0.0810], device='cuda:6'), in_proj_covar=tensor([0.0660, 0.0812, 0.0934, 0.0818, 0.0628, 0.0645, 0.0673, 0.0786], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:43:53,993 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=249395.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:44:07,472 INFO [train.py:904] (6/8) Epoch 25, batch 5800, loss[loss=0.2057, simple_loss=0.2846, pruned_loss=0.06341, over 11888.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2937, pruned_loss=0.06119, over 3013989.14 frames. ], batch size: 248, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:44:32,817 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=249419.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:45:09,430 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.796e+02 2.976e+02 3.356e+02 4.382e+02 8.536e+02, threshold=6.712e+02, percent-clipped=3.0 2023-05-02 00:45:26,286 INFO [train.py:904] (6/8) Epoch 25, batch 5850, loss[loss=0.1965, simple_loss=0.2855, pruned_loss=0.05374, over 16856.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2922, pruned_loss=0.06013, over 3029967.46 frames. ], batch size: 116, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:46:36,291 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 00:46:46,941 INFO [train.py:904] (6/8) Epoch 25, batch 5900, loss[loss=0.1924, simple_loss=0.2939, pruned_loss=0.04547, over 16793.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2917, pruned_loss=0.05961, over 3053415.99 frames. ], batch size: 76, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:47:25,023 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9284, 2.0684, 2.2749, 3.3798, 1.9909, 2.3622, 2.2102, 2.2223], device='cuda:6'), covar=tensor([0.1505, 0.3705, 0.2894, 0.0708, 0.4564, 0.2616, 0.3467, 0.3582], device='cuda:6'), in_proj_covar=tensor([0.0410, 0.0459, 0.0376, 0.0330, 0.0439, 0.0525, 0.0429, 0.0535], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:47:40,869 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6383, 1.8223, 2.2513, 2.5477, 2.5511, 2.8785, 2.1093, 2.8256], device='cuda:6'), covar=tensor([0.0216, 0.0590, 0.0345, 0.0334, 0.0359, 0.0209, 0.0527, 0.0159], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0193, 0.0181, 0.0186, 0.0201, 0.0160, 0.0198, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:47:52,231 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 2.624e+02 3.247e+02 4.027e+02 8.250e+02, threshold=6.495e+02, percent-clipped=2.0 2023-05-02 00:48:08,074 INFO [train.py:904] (6/8) Epoch 25, batch 5950, loss[loss=0.2087, simple_loss=0.2995, pruned_loss=0.05894, over 16922.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2926, pruned_loss=0.05817, over 3087269.02 frames. ], batch size: 109, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:48:37,005 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=249571.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:49:09,249 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2927, 3.6021, 3.6849, 2.1051, 3.1745, 2.5434, 3.7027, 3.9683], device='cuda:6'), covar=tensor([0.0285, 0.0782, 0.0673, 0.2301, 0.0893, 0.0996, 0.0715, 0.0933], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0167, 0.0169, 0.0155, 0.0147, 0.0131, 0.0145, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:49:28,143 INFO [train.py:904] (6/8) Epoch 25, batch 6000, loss[loss=0.2061, simple_loss=0.2889, pruned_loss=0.06159, over 16771.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.291, pruned_loss=0.05762, over 3105446.92 frames. ], batch size: 124, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:49:28,143 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 00:49:38,598 INFO [train.py:938] (6/8) Epoch 25, validation: loss=0.1487, simple_loss=0.2613, pruned_loss=0.01811, over 944034.00 frames. 2023-05-02 00:49:38,599 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 00:49:54,510 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-02 00:50:36,538 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.971e+02 2.903e+02 3.391e+02 3.955e+02 5.249e+02, threshold=6.782e+02, percent-clipped=0.0 2023-05-02 00:50:54,725 INFO [train.py:904] (6/8) Epoch 25, batch 6050, loss[loss=0.1808, simple_loss=0.2823, pruned_loss=0.03971, over 16631.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2899, pruned_loss=0.05716, over 3093290.07 frames. ], batch size: 134, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:50:55,825 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2180, 4.3169, 4.5045, 4.2531, 4.3950, 4.8518, 4.3708, 4.0956], device='cuda:6'), covar=tensor([0.1787, 0.2026, 0.2506, 0.2256, 0.2564, 0.1095, 0.1694, 0.2771], device='cuda:6'), in_proj_covar=tensor([0.0419, 0.0617, 0.0678, 0.0504, 0.0670, 0.0707, 0.0525, 0.0675], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 00:51:34,379 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 00:52:12,400 INFO [train.py:904] (6/8) Epoch 25, batch 6100, loss[loss=0.2517, simple_loss=0.3289, pruned_loss=0.08722, over 11842.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2892, pruned_loss=0.05591, over 3110326.86 frames. ], batch size: 246, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:53:02,109 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6893, 3.7566, 2.3911, 4.3866, 2.9065, 4.2951, 2.5902, 3.0535], device='cuda:6'), covar=tensor([0.0295, 0.0421, 0.1654, 0.0194, 0.0847, 0.0563, 0.1439, 0.0817], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0180, 0.0195, 0.0168, 0.0177, 0.0219, 0.0203, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:53:15,556 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.704e+02 2.711e+02 3.338e+02 3.699e+02 9.166e+02, threshold=6.676e+02, percent-clipped=2.0 2023-05-02 00:53:29,950 INFO [train.py:904] (6/8) Epoch 25, batch 6150, loss[loss=0.2234, simple_loss=0.3024, pruned_loss=0.0722, over 11601.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2872, pruned_loss=0.05589, over 3078267.16 frames. ], batch size: 248, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:53:54,744 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.09 vs. limit=5.0 2023-05-02 00:54:51,556 INFO [train.py:904] (6/8) Epoch 25, batch 6200, loss[loss=0.1971, simple_loss=0.285, pruned_loss=0.05461, over 15403.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.2851, pruned_loss=0.05526, over 3085119.27 frames. ], batch size: 191, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:55:10,528 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=249814.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:55:23,462 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6654, 2.4398, 2.3556, 3.5671, 2.4772, 3.7732, 1.4176, 2.7953], device='cuda:6'), covar=tensor([0.1451, 0.0891, 0.1376, 0.0183, 0.0184, 0.0396, 0.1856, 0.0868], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0179, 0.0197, 0.0196, 0.0206, 0.0216, 0.0207, 0.0197], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:55:49,762 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9275, 1.9574, 2.4399, 2.8708, 2.7721, 3.3276, 2.1881, 3.2794], device='cuda:6'), covar=tensor([0.0243, 0.0583, 0.0392, 0.0358, 0.0358, 0.0201, 0.0572, 0.0176], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0195, 0.0183, 0.0188, 0.0203, 0.0161, 0.0200, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:55:55,234 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.970e+02 2.750e+02 3.284e+02 4.072e+02 6.772e+02, threshold=6.569e+02, percent-clipped=1.0 2023-05-02 00:56:10,146 INFO [train.py:904] (6/8) Epoch 25, batch 6250, loss[loss=0.2049, simple_loss=0.294, pruned_loss=0.05789, over 17040.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.285, pruned_loss=0.05541, over 3095468.48 frames. ], batch size: 53, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:56:38,905 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=249871.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:56:44,475 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=249875.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:57:26,264 INFO [train.py:904] (6/8) Epoch 25, batch 6300, loss[loss=0.1975, simple_loss=0.2855, pruned_loss=0.05473, over 16735.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2853, pruned_loss=0.05472, over 3114330.89 frames. ], batch size: 83, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:57:52,593 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=249919.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:58:28,304 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8219, 3.7482, 3.8900, 3.9878, 4.0798, 3.7162, 4.0416, 4.1086], device='cuda:6'), covar=tensor([0.1691, 0.1248, 0.1381, 0.0762, 0.0700, 0.1771, 0.0896, 0.0853], device='cuda:6'), in_proj_covar=tensor([0.0655, 0.0805, 0.0929, 0.0815, 0.0624, 0.0642, 0.0672, 0.0782], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 00:58:29,093 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.822e+02 2.746e+02 3.246e+02 3.923e+02 7.422e+02, threshold=6.492e+02, percent-clipped=1.0 2023-05-02 00:58:37,382 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7143, 2.3536, 2.2909, 3.2705, 2.1800, 3.5169, 1.4927, 2.6875], device='cuda:6'), covar=tensor([0.1458, 0.0878, 0.1330, 0.0200, 0.0161, 0.0394, 0.1903, 0.0896], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0179, 0.0197, 0.0196, 0.0206, 0.0216, 0.0207, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:58:45,042 INFO [train.py:904] (6/8) Epoch 25, batch 6350, loss[loss=0.1847, simple_loss=0.2737, pruned_loss=0.04784, over 16721.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2854, pruned_loss=0.05541, over 3107546.49 frames. ], batch size: 124, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:58:47,496 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1731, 3.2427, 1.8335, 3.4397, 2.3511, 3.4727, 2.0832, 2.6560], device='cuda:6'), covar=tensor([0.0336, 0.0397, 0.1805, 0.0291, 0.0903, 0.0701, 0.1618, 0.0832], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0181, 0.0196, 0.0169, 0.0179, 0.0220, 0.0204, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 00:58:48,594 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=249955.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 00:59:20,600 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=249976.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:00:04,120 INFO [train.py:904] (6/8) Epoch 25, batch 6400, loss[loss=0.1915, simple_loss=0.2784, pruned_loss=0.05228, over 16910.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2858, pruned_loss=0.05663, over 3090613.72 frames. ], batch size: 116, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 01:00:07,127 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9036, 1.3653, 1.7206, 1.7376, 1.8195, 1.8972, 1.5539, 1.8169], device='cuda:6'), covar=tensor([0.0257, 0.0478, 0.0244, 0.0277, 0.0293, 0.0203, 0.0609, 0.0183], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0196, 0.0184, 0.0187, 0.0203, 0.0162, 0.0201, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 01:00:23,789 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=250016.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 01:00:46,764 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0334, 3.3147, 3.4795, 2.0204, 3.0976, 2.2964, 3.5987, 3.6753], device='cuda:6'), covar=tensor([0.0260, 0.0827, 0.0605, 0.2141, 0.0806, 0.1011, 0.0525, 0.0834], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0167, 0.0169, 0.0155, 0.0147, 0.0131, 0.0145, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:00:56,513 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=250037.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:01:05,941 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.084e+02 2.850e+02 3.461e+02 4.247e+02 7.668e+02, threshold=6.921e+02, percent-clipped=1.0 2023-05-02 01:01:20,199 INFO [train.py:904] (6/8) Epoch 25, batch 6450, loss[loss=0.1848, simple_loss=0.2647, pruned_loss=0.05247, over 11682.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2859, pruned_loss=0.05597, over 3085180.76 frames. ], batch size: 247, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 01:01:27,639 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0900, 3.0567, 1.9868, 3.2589, 2.3144, 3.3283, 2.1123, 2.6360], device='cuda:6'), covar=tensor([0.0314, 0.0488, 0.1541, 0.0320, 0.0841, 0.0795, 0.1528, 0.0727], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0181, 0.0196, 0.0169, 0.0178, 0.0220, 0.0204, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:01:33,073 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1038, 4.8463, 4.7655, 5.2368, 5.4713, 4.7988, 5.4044, 5.4294], device='cuda:6'), covar=tensor([0.1793, 0.1453, 0.2561, 0.1045, 0.0853, 0.1254, 0.1063, 0.0943], device='cuda:6'), in_proj_covar=tensor([0.0652, 0.0802, 0.0925, 0.0811, 0.0621, 0.0639, 0.0670, 0.0780], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 01:01:40,195 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.98 vs. limit=5.0 2023-05-02 01:02:13,236 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-02 01:02:36,758 INFO [train.py:904] (6/8) Epoch 25, batch 6500, loss[loss=0.2074, simple_loss=0.2866, pruned_loss=0.06408, over 16651.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2841, pruned_loss=0.05515, over 3104837.63 frames. ], batch size: 62, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 01:03:40,562 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.844e+02 2.643e+02 3.230e+02 3.830e+02 1.038e+03, threshold=6.461e+02, percent-clipped=2.0 2023-05-02 01:03:52,680 INFO [train.py:904] (6/8) Epoch 25, batch 6550, loss[loss=0.2123, simple_loss=0.3099, pruned_loss=0.05741, over 15255.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2869, pruned_loss=0.05615, over 3091998.41 frames. ], batch size: 190, lr: 2.69e-03, grad_scale: 4.0 2023-05-02 01:04:19,063 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=250170.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:05:05,610 INFO [train.py:904] (6/8) Epoch 25, batch 6600, loss[loss=0.2037, simple_loss=0.2837, pruned_loss=0.06181, over 17020.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2888, pruned_loss=0.05656, over 3095982.21 frames. ], batch size: 55, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:06:08,494 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.914e+02 2.823e+02 3.325e+02 3.947e+02 6.947e+02, threshold=6.650e+02, percent-clipped=1.0 2023-05-02 01:06:21,898 INFO [train.py:904] (6/8) Epoch 25, batch 6650, loss[loss=0.1764, simple_loss=0.2628, pruned_loss=0.04503, over 16730.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2891, pruned_loss=0.05786, over 3084492.69 frames. ], batch size: 124, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:07:37,547 INFO [train.py:904] (6/8) Epoch 25, batch 6700, loss[loss=0.2201, simple_loss=0.3103, pruned_loss=0.06495, over 16907.00 frames. ], tot_loss[loss=0.202, simple_loss=0.288, pruned_loss=0.05799, over 3071388.63 frames. ], batch size: 109, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:07:50,663 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=250311.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 01:08:23,194 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=250332.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:08:33,324 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3484, 1.6658, 2.0775, 2.3534, 2.3749, 2.6335, 1.7864, 2.5853], device='cuda:6'), covar=tensor([0.0266, 0.0566, 0.0377, 0.0348, 0.0379, 0.0247, 0.0586, 0.0176], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0194, 0.0182, 0.0185, 0.0200, 0.0160, 0.0198, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 01:08:34,391 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7253, 4.7963, 5.1279, 5.1178, 5.1351, 4.8256, 4.7713, 4.6169], device='cuda:6'), covar=tensor([0.0327, 0.0522, 0.0326, 0.0350, 0.0457, 0.0365, 0.0938, 0.0479], device='cuda:6'), in_proj_covar=tensor([0.0417, 0.0470, 0.0457, 0.0418, 0.0502, 0.0479, 0.0557, 0.0382], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 01:08:41,137 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.782e+02 2.721e+02 3.128e+02 3.707e+02 8.776e+02, threshold=6.256e+02, percent-clipped=1.0 2023-05-02 01:08:54,005 INFO [train.py:904] (6/8) Epoch 25, batch 6750, loss[loss=0.1883, simple_loss=0.2784, pruned_loss=0.04911, over 16203.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2865, pruned_loss=0.05732, over 3090504.47 frames. ], batch size: 165, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:10:10,572 INFO [train.py:904] (6/8) Epoch 25, batch 6800, loss[loss=0.2231, simple_loss=0.2959, pruned_loss=0.07511, over 11735.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2867, pruned_loss=0.05747, over 3081676.58 frames. ], batch size: 248, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:11:16,511 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.894e+02 2.847e+02 3.369e+02 4.013e+02 7.021e+02, threshold=6.738e+02, percent-clipped=2.0 2023-05-02 01:11:26,907 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8944, 2.9262, 2.7153, 5.1297, 3.4632, 4.4365, 1.8573, 3.2109], device='cuda:6'), covar=tensor([0.1538, 0.0904, 0.1349, 0.0199, 0.0446, 0.0446, 0.1769, 0.0874], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0179, 0.0197, 0.0196, 0.0206, 0.0217, 0.0207, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:11:27,491 INFO [train.py:904] (6/8) Epoch 25, batch 6850, loss[loss=0.1854, simple_loss=0.2945, pruned_loss=0.03817, over 16764.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2872, pruned_loss=0.05682, over 3111993.54 frames. ], batch size: 39, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:11:34,057 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2148, 4.3059, 4.1074, 3.7963, 3.7967, 4.2126, 3.9362, 3.9750], device='cuda:6'), covar=tensor([0.0764, 0.1110, 0.0433, 0.0403, 0.0843, 0.0850, 0.1041, 0.0878], device='cuda:6'), in_proj_covar=tensor([0.0298, 0.0450, 0.0348, 0.0354, 0.0354, 0.0407, 0.0239, 0.0423], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 01:11:52,833 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=250470.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:11:52,982 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1626, 1.5882, 1.9825, 2.1378, 2.2252, 2.3896, 1.7520, 2.3643], device='cuda:6'), covar=tensor([0.0238, 0.0533, 0.0286, 0.0338, 0.0339, 0.0241, 0.0553, 0.0172], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0195, 0.0182, 0.0186, 0.0201, 0.0161, 0.0199, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 01:12:43,350 INFO [train.py:904] (6/8) Epoch 25, batch 6900, loss[loss=0.2136, simple_loss=0.308, pruned_loss=0.05955, over 16440.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.289, pruned_loss=0.05638, over 3109592.22 frames. ], batch size: 146, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:13:06,142 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=250518.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:13:47,067 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.918e+02 2.786e+02 3.277e+02 4.153e+02 1.055e+03, threshold=6.554e+02, percent-clipped=5.0 2023-05-02 01:14:00,190 INFO [train.py:904] (6/8) Epoch 25, batch 6950, loss[loss=0.1759, simple_loss=0.27, pruned_loss=0.04091, over 16870.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.291, pruned_loss=0.05816, over 3097268.07 frames. ], batch size: 96, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:15:18,377 INFO [train.py:904] (6/8) Epoch 25, batch 7000, loss[loss=0.1969, simple_loss=0.2953, pruned_loss=0.04921, over 16134.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2912, pruned_loss=0.05757, over 3091144.34 frames. ], batch size: 165, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:15:32,195 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=250611.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 01:16:03,589 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=250632.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:16:18,365 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6042, 3.7373, 2.5194, 2.2613, 2.3551, 2.1851, 3.9593, 3.1307], device='cuda:6'), covar=tensor([0.3403, 0.0769, 0.2441, 0.2852, 0.3135, 0.2578, 0.0615, 0.1571], device='cuda:6'), in_proj_covar=tensor([0.0333, 0.0273, 0.0312, 0.0321, 0.0303, 0.0270, 0.0302, 0.0346], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 01:16:22,012 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.860e+02 2.716e+02 3.192e+02 3.909e+02 7.863e+02, threshold=6.384e+02, percent-clipped=2.0 2023-05-02 01:16:35,089 INFO [train.py:904] (6/8) Epoch 25, batch 7050, loss[loss=0.2177, simple_loss=0.2906, pruned_loss=0.07246, over 11128.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2923, pruned_loss=0.0577, over 3083458.90 frames. ], batch size: 246, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:16:44,699 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=250659.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 01:16:49,655 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=250662.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:17:17,450 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=250680.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:17:43,965 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1382, 2.2694, 2.4366, 3.8388, 2.1110, 2.5084, 2.3721, 2.4265], device='cuda:6'), covar=tensor([0.1555, 0.3774, 0.3019, 0.0644, 0.4620, 0.2836, 0.3677, 0.3638], device='cuda:6'), in_proj_covar=tensor([0.0410, 0.0460, 0.0377, 0.0331, 0.0440, 0.0527, 0.0432, 0.0538], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 01:17:51,094 INFO [train.py:904] (6/8) Epoch 25, batch 7100, loss[loss=0.2149, simple_loss=0.2962, pruned_loss=0.06678, over 16866.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2907, pruned_loss=0.0573, over 3086537.56 frames. ], batch size: 116, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:17:59,093 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9365, 4.7686, 4.9314, 5.1012, 5.3280, 4.6480, 5.2936, 5.3149], device='cuda:6'), covar=tensor([0.1983, 0.1306, 0.1794, 0.0853, 0.0618, 0.1020, 0.0665, 0.0714], device='cuda:6'), in_proj_covar=tensor([0.0651, 0.0800, 0.0924, 0.0810, 0.0622, 0.0642, 0.0673, 0.0782], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 01:17:59,359 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.58 vs. limit=5.0 2023-05-02 01:18:23,460 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=250723.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:18:56,841 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.881e+02 2.742e+02 3.283e+02 4.161e+02 6.949e+02, threshold=6.565e+02, percent-clipped=2.0 2023-05-02 01:19:09,289 INFO [train.py:904] (6/8) Epoch 25, batch 7150, loss[loss=0.1901, simple_loss=0.2817, pruned_loss=0.04921, over 17007.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.289, pruned_loss=0.05714, over 3097448.43 frames. ], batch size: 41, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:20:22,683 INFO [train.py:904] (6/8) Epoch 25, batch 7200, loss[loss=0.1761, simple_loss=0.2636, pruned_loss=0.0443, over 16578.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2868, pruned_loss=0.05601, over 3065817.88 frames. ], batch size: 68, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:20:23,161 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2245, 2.4566, 2.1469, 2.2410, 2.7830, 2.4147, 2.7088, 3.0035], device='cuda:6'), covar=tensor([0.0157, 0.0452, 0.0566, 0.0562, 0.0332, 0.0481, 0.0288, 0.0298], device='cuda:6'), in_proj_covar=tensor([0.0217, 0.0237, 0.0227, 0.0228, 0.0238, 0.0236, 0.0236, 0.0235], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 01:21:08,874 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=250833.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:21:11,312 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 01:21:28,098 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=250845.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:21:28,839 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 2.397e+02 2.821e+02 3.422e+02 6.552e+02, threshold=5.642e+02, percent-clipped=0.0 2023-05-02 01:21:32,793 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8775, 5.2194, 5.4212, 5.1802, 5.2889, 5.7477, 5.2216, 4.9954], device='cuda:6'), covar=tensor([0.1076, 0.1779, 0.2143, 0.1752, 0.2108, 0.0901, 0.1569, 0.2299], device='cuda:6'), in_proj_covar=tensor([0.0421, 0.0617, 0.0678, 0.0506, 0.0671, 0.0704, 0.0527, 0.0676], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 01:21:33,005 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8009, 2.5116, 2.5947, 4.4777, 3.2873, 4.0470, 1.6509, 2.9616], device='cuda:6'), covar=tensor([0.1294, 0.0873, 0.1200, 0.0149, 0.0258, 0.0382, 0.1589, 0.0802], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0182, 0.0200, 0.0198, 0.0209, 0.0219, 0.0209, 0.0201], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:21:41,038 INFO [train.py:904] (6/8) Epoch 25, batch 7250, loss[loss=0.2173, simple_loss=0.2919, pruned_loss=0.07133, over 11200.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.285, pruned_loss=0.05513, over 3051967.77 frames. ], batch size: 248, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:22:20,166 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2789, 3.2956, 2.0217, 3.6079, 2.4648, 3.6131, 2.2244, 2.7218], device='cuda:6'), covar=tensor([0.0297, 0.0418, 0.1667, 0.0252, 0.0857, 0.0712, 0.1461, 0.0754], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0180, 0.0195, 0.0168, 0.0178, 0.0219, 0.0203, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:22:42,110 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=250894.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:22:55,036 INFO [train.py:904] (6/8) Epoch 25, batch 7300, loss[loss=0.216, simple_loss=0.2997, pruned_loss=0.06619, over 15361.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2847, pruned_loss=0.05528, over 3055673.00 frames. ], batch size: 191, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:23:00,836 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=250906.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 01:23:11,177 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.90 vs. limit=5.0 2023-05-02 01:23:59,649 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.919e+02 2.893e+02 3.388e+02 3.918e+02 7.629e+02, threshold=6.776e+02, percent-clipped=7.0 2023-05-02 01:24:09,701 INFO [train.py:904] (6/8) Epoch 25, batch 7350, loss[loss=0.1982, simple_loss=0.2895, pruned_loss=0.05343, over 16755.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2852, pruned_loss=0.05582, over 3054226.02 frames. ], batch size: 83, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:24:10,889 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7438, 2.7007, 2.7871, 2.1084, 2.6708, 2.1605, 2.7269, 2.8874], device='cuda:6'), covar=tensor([0.0271, 0.0810, 0.0511, 0.1808, 0.0817, 0.0909, 0.0570, 0.0806], device='cuda:6'), in_proj_covar=tensor([0.0159, 0.0168, 0.0170, 0.0156, 0.0148, 0.0132, 0.0145, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:24:38,555 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 01:25:03,050 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7447, 1.8164, 1.6364, 1.4979, 1.9192, 1.5548, 1.6382, 1.8672], device='cuda:6'), covar=tensor([0.0185, 0.0275, 0.0390, 0.0330, 0.0210, 0.0263, 0.0173, 0.0213], device='cuda:6'), in_proj_covar=tensor([0.0215, 0.0235, 0.0225, 0.0226, 0.0235, 0.0235, 0.0234, 0.0234], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 01:25:27,062 INFO [train.py:904] (6/8) Epoch 25, batch 7400, loss[loss=0.1994, simple_loss=0.2834, pruned_loss=0.05773, over 17053.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2863, pruned_loss=0.05592, over 3078961.07 frames. ], batch size: 53, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:25:38,912 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4335, 4.6692, 4.4895, 4.4806, 4.2419, 4.1974, 4.2613, 4.7292], device='cuda:6'), covar=tensor([0.1146, 0.0883, 0.0966, 0.0873, 0.0794, 0.1417, 0.1080, 0.0885], device='cuda:6'), in_proj_covar=tensor([0.0688, 0.0833, 0.0687, 0.0641, 0.0528, 0.0534, 0.0698, 0.0649], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 01:25:50,971 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251018.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:26:02,392 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2816, 3.2874, 1.9260, 3.5756, 2.4289, 3.5889, 2.1606, 2.7058], device='cuda:6'), covar=tensor([0.0312, 0.0437, 0.1724, 0.0256, 0.0901, 0.0633, 0.1509, 0.0808], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0181, 0.0197, 0.0169, 0.0179, 0.0220, 0.0205, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:26:35,984 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.828e+02 2.726e+02 3.311e+02 3.902e+02 7.761e+02, threshold=6.622e+02, percent-clipped=1.0 2023-05-02 01:26:46,725 INFO [train.py:904] (6/8) Epoch 25, batch 7450, loss[loss=0.1982, simple_loss=0.2862, pruned_loss=0.0551, over 16738.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2877, pruned_loss=0.05708, over 3078726.54 frames. ], batch size: 76, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:28:06,267 INFO [train.py:904] (6/8) Epoch 25, batch 7500, loss[loss=0.2151, simple_loss=0.3022, pruned_loss=0.06406, over 15306.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.288, pruned_loss=0.05679, over 3065278.21 frames. ], batch size: 190, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:28:38,254 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0873, 3.9445, 4.1373, 4.2408, 4.3859, 3.9639, 4.3363, 4.3919], device='cuda:6'), covar=tensor([0.1776, 0.1284, 0.1447, 0.0798, 0.0641, 0.1498, 0.0911, 0.0783], device='cuda:6'), in_proj_covar=tensor([0.0648, 0.0800, 0.0921, 0.0808, 0.0620, 0.0638, 0.0670, 0.0778], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 01:29:12,775 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.915e+02 2.911e+02 3.474e+02 4.178e+02 7.715e+02, threshold=6.948e+02, percent-clipped=3.0 2023-05-02 01:29:23,360 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8810, 2.8295, 2.8590, 4.8665, 3.7067, 4.2362, 1.7227, 3.2835], device='cuda:6'), covar=tensor([0.1365, 0.0857, 0.1150, 0.0147, 0.0314, 0.0375, 0.1682, 0.0724], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0180, 0.0198, 0.0197, 0.0207, 0.0217, 0.0207, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:29:23,953 INFO [train.py:904] (6/8) Epoch 25, batch 7550, loss[loss=0.1894, simple_loss=0.2707, pruned_loss=0.05403, over 16579.00 frames. ], tot_loss[loss=0.2013, simple_loss=0.2874, pruned_loss=0.05754, over 3058417.05 frames. ], batch size: 57, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:30:19,182 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251189.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:30:32,680 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=251198.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:30:37,923 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251201.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 01:30:39,951 INFO [train.py:904] (6/8) Epoch 25, batch 7600, loss[loss=0.2069, simple_loss=0.2933, pruned_loss=0.06019, over 16247.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2864, pruned_loss=0.05738, over 3067987.84 frames. ], batch size: 165, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:31:43,657 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.097e+02 2.965e+02 3.524e+02 4.371e+02 1.071e+03, threshold=7.047e+02, percent-clipped=4.0 2023-05-02 01:31:53,103 INFO [train.py:904] (6/8) Epoch 25, batch 7650, loss[loss=0.2037, simple_loss=0.2909, pruned_loss=0.05822, over 16779.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2875, pruned_loss=0.05803, over 3074382.81 frames. ], batch size: 124, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:32:03,120 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=251259.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:33:07,046 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=251301.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:33:08,937 INFO [train.py:904] (6/8) Epoch 25, batch 7700, loss[loss=0.1921, simple_loss=0.2832, pruned_loss=0.05051, over 17237.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2872, pruned_loss=0.05804, over 3077953.81 frames. ], batch size: 52, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:33:32,427 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=251318.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:34:14,481 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.002e+02 2.953e+02 3.490e+02 4.215e+02 8.631e+02, threshold=6.980e+02, percent-clipped=3.0 2023-05-02 01:34:25,529 INFO [train.py:904] (6/8) Epoch 25, batch 7750, loss[loss=0.2095, simple_loss=0.2969, pruned_loss=0.06106, over 16326.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.2875, pruned_loss=0.05812, over 3079655.44 frames. ], batch size: 165, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:34:39,758 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5031, 4.0809, 4.0598, 2.8080, 3.7007, 4.1588, 3.6612, 2.3531], device='cuda:6'), covar=tensor([0.0529, 0.0059, 0.0057, 0.0398, 0.0101, 0.0115, 0.0102, 0.0469], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0087, 0.0087, 0.0134, 0.0099, 0.0111, 0.0096, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 01:34:39,787 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=251362.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:34:45,643 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=251366.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:35:40,751 INFO [train.py:904] (6/8) Epoch 25, batch 7800, loss[loss=0.1986, simple_loss=0.2825, pruned_loss=0.05737, over 15364.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2883, pruned_loss=0.05836, over 3093678.66 frames. ], batch size: 190, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:35:42,494 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=251404.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:36:44,894 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-05-02 01:36:45,152 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.753e+02 2.864e+02 3.394e+02 4.001e+02 5.962e+02, threshold=6.789e+02, percent-clipped=0.0 2023-05-02 01:36:55,300 INFO [train.py:904] (6/8) Epoch 25, batch 7850, loss[loss=0.19, simple_loss=0.2751, pruned_loss=0.05247, over 16519.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2894, pruned_loss=0.05855, over 3068566.62 frames. ], batch size: 68, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:37:09,119 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-05-02 01:37:14,501 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=251465.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:37:40,191 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 01:37:50,404 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=251489.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:38:07,468 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=251501.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 01:38:09,238 INFO [train.py:904] (6/8) Epoch 25, batch 7900, loss[loss=0.1859, simple_loss=0.2835, pruned_loss=0.0441, over 16808.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2888, pruned_loss=0.05767, over 3102173.54 frames. ], batch size: 102, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:38:47,677 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6871, 3.4735, 3.9753, 1.9663, 4.1390, 4.1976, 3.1007, 3.0131], device='cuda:6'), covar=tensor([0.0786, 0.0307, 0.0225, 0.1247, 0.0074, 0.0169, 0.0431, 0.0492], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0111, 0.0100, 0.0139, 0.0084, 0.0130, 0.0129, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:38:57,947 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-02 01:39:04,226 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=251537.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:39:17,277 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.008e+02 2.635e+02 3.162e+02 3.936e+02 7.872e+02, threshold=6.324e+02, percent-clipped=1.0 2023-05-02 01:39:22,129 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=251549.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:39:28,551 INFO [train.py:904] (6/8) Epoch 25, batch 7950, loss[loss=0.1848, simple_loss=0.2689, pruned_loss=0.05036, over 16825.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2888, pruned_loss=0.05765, over 3098406.03 frames. ], batch size: 83, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:39:30,134 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251554.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:40:41,095 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7897, 2.7394, 2.8246, 2.1345, 2.7013, 2.1845, 2.7108, 2.9509], device='cuda:6'), covar=tensor([0.0296, 0.0822, 0.0516, 0.1824, 0.0840, 0.0938, 0.0615, 0.0717], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0167, 0.0170, 0.0155, 0.0147, 0.0131, 0.0145, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:40:46,595 INFO [train.py:904] (6/8) Epoch 25, batch 8000, loss[loss=0.2026, simple_loss=0.2883, pruned_loss=0.05845, over 16674.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2897, pruned_loss=0.05876, over 3076600.88 frames. ], batch size: 134, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:41:44,798 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=251641.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:41:51,618 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.813e+02 2.777e+02 3.202e+02 3.994e+02 6.409e+02, threshold=6.404e+02, percent-clipped=2.0 2023-05-02 01:42:01,897 INFO [train.py:904] (6/8) Epoch 25, batch 8050, loss[loss=0.2094, simple_loss=0.2861, pruned_loss=0.06634, over 11598.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2897, pruned_loss=0.05893, over 3067630.16 frames. ], batch size: 248, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:42:09,016 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251657.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:43:16,960 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=251702.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:43:17,637 INFO [train.py:904] (6/8) Epoch 25, batch 8100, loss[loss=0.2045, simple_loss=0.2988, pruned_loss=0.05506, over 15216.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2897, pruned_loss=0.0588, over 3051464.51 frames. ], batch size: 190, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:43:18,532 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.95 vs. limit=5.0 2023-05-02 01:43:27,311 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=251709.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:44:22,958 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.808e+02 3.404e+02 4.395e+02 1.324e+03, threshold=6.808e+02, percent-clipped=5.0 2023-05-02 01:44:33,795 INFO [train.py:904] (6/8) Epoch 25, batch 8150, loss[loss=0.1636, simple_loss=0.2489, pruned_loss=0.03917, over 16628.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2865, pruned_loss=0.05713, over 3073450.73 frames. ], batch size: 57, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:44:41,637 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6964, 2.3857, 2.2290, 3.2004, 1.9517, 3.4965, 1.4965, 2.7187], device='cuda:6'), covar=tensor([0.1473, 0.0880, 0.1382, 0.0234, 0.0148, 0.0415, 0.1870, 0.0866], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0180, 0.0198, 0.0197, 0.0208, 0.0218, 0.0208, 0.0199], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:44:45,086 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251760.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:45:00,574 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=251770.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:45:42,050 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1601, 5.4115, 5.1914, 5.1802, 4.9425, 4.8841, 4.8271, 5.5126], device='cuda:6'), covar=tensor([0.1257, 0.0876, 0.1015, 0.0918, 0.0786, 0.0897, 0.1214, 0.0875], device='cuda:6'), in_proj_covar=tensor([0.0692, 0.0836, 0.0691, 0.0647, 0.0529, 0.0537, 0.0701, 0.0655], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 01:45:50,905 INFO [train.py:904] (6/8) Epoch 25, batch 8200, loss[loss=0.1947, simple_loss=0.2845, pruned_loss=0.05249, over 16847.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2834, pruned_loss=0.05573, over 3094238.13 frames. ], batch size: 116, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:46:59,194 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 2.618e+02 3.043e+02 3.668e+02 8.118e+02, threshold=6.086e+02, percent-clipped=2.0 2023-05-02 01:47:10,669 INFO [train.py:904] (6/8) Epoch 25, batch 8250, loss[loss=0.1813, simple_loss=0.2707, pruned_loss=0.04591, over 16593.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2818, pruned_loss=0.0535, over 3068838.61 frames. ], batch size: 62, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:47:12,496 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=251854.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:47:22,077 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9837, 2.3096, 2.3703, 2.9295, 1.7953, 3.3147, 1.7207, 2.8261], device='cuda:6'), covar=tensor([0.1256, 0.0626, 0.1002, 0.0212, 0.0093, 0.0423, 0.1540, 0.0648], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0181, 0.0199, 0.0197, 0.0208, 0.0218, 0.0209, 0.0199], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:48:21,828 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.89 vs. limit=5.0 2023-05-02 01:48:29,882 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=251902.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:48:30,708 INFO [train.py:904] (6/8) Epoch 25, batch 8300, loss[loss=0.1785, simple_loss=0.279, pruned_loss=0.03897, over 16351.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.2798, pruned_loss=0.051, over 3057019.24 frames. ], batch size: 146, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:49:41,162 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 2.151e+02 2.560e+02 3.132e+02 5.384e+02, threshold=5.121e+02, percent-clipped=0.0 2023-05-02 01:49:52,625 INFO [train.py:904] (6/8) Epoch 25, batch 8350, loss[loss=0.1813, simple_loss=0.2837, pruned_loss=0.03942, over 16645.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2794, pruned_loss=0.04903, over 3054261.46 frames. ], batch size: 134, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:49:59,497 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=251957.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:51:05,924 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251997.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:51:17,313 INFO [train.py:904] (6/8) Epoch 25, batch 8400, loss[loss=0.1578, simple_loss=0.2485, pruned_loss=0.03353, over 12474.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2765, pruned_loss=0.0468, over 3046510.49 frames. ], batch size: 248, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:51:22,059 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=252005.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:52:27,817 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.253e+02 2.668e+02 3.204e+02 6.662e+02, threshold=5.337e+02, percent-clipped=3.0 2023-05-02 01:52:40,175 INFO [train.py:904] (6/8) Epoch 25, batch 8450, loss[loss=0.1684, simple_loss=0.2634, pruned_loss=0.03671, over 16936.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2749, pruned_loss=0.04538, over 3051971.12 frames. ], batch size: 109, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 01:52:51,881 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=252060.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:52:59,910 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=252065.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:53:32,487 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-05-02 01:54:01,927 INFO [train.py:904] (6/8) Epoch 25, batch 8500, loss[loss=0.1555, simple_loss=0.2469, pruned_loss=0.03198, over 15263.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2715, pruned_loss=0.04319, over 3058950.53 frames. ], batch size: 190, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 01:54:09,225 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2023-05-02 01:54:11,578 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=252108.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:55:15,323 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.653e+02 2.299e+02 2.680e+02 3.253e+02 6.153e+02, threshold=5.360e+02, percent-clipped=1.0 2023-05-02 01:55:28,835 INFO [train.py:904] (6/8) Epoch 25, batch 8550, loss[loss=0.1712, simple_loss=0.2543, pruned_loss=0.04402, over 12037.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2695, pruned_loss=0.04232, over 3040619.15 frames. ], batch size: 249, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 01:56:12,697 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6180, 2.6463, 1.8869, 2.7641, 2.0814, 2.8143, 2.1016, 2.3534], device='cuda:6'), covar=tensor([0.0342, 0.0419, 0.1398, 0.0317, 0.0727, 0.0494, 0.1332, 0.0640], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0176, 0.0193, 0.0164, 0.0174, 0.0214, 0.0201, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:56:24,221 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4977, 3.0166, 2.7859, 2.2136, 2.2114, 2.3260, 3.0120, 2.8636], device='cuda:6'), covar=tensor([0.2721, 0.0657, 0.1576, 0.3044, 0.2921, 0.2236, 0.0454, 0.1468], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0267, 0.0305, 0.0315, 0.0298, 0.0266, 0.0295, 0.0338], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 01:57:09,034 INFO [train.py:904] (6/8) Epoch 25, batch 8600, loss[loss=0.1743, simple_loss=0.2737, pruned_loss=0.03745, over 16177.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2695, pruned_loss=0.04171, over 3009200.28 frames. ], batch size: 165, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 01:58:32,388 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0287, 3.0435, 1.8960, 3.3068, 2.2394, 3.2966, 2.1438, 2.6233], device='cuda:6'), covar=tensor([0.0373, 0.0465, 0.1800, 0.0284, 0.1008, 0.0624, 0.1557, 0.0769], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0176, 0.0193, 0.0164, 0.0175, 0.0214, 0.0201, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 01:58:34,356 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.652e+02 2.200e+02 2.678e+02 3.278e+02 9.763e+02, threshold=5.356e+02, percent-clipped=4.0 2023-05-02 01:58:48,811 INFO [train.py:904] (6/8) Epoch 25, batch 8650, loss[loss=0.1515, simple_loss=0.2558, pruned_loss=0.02358, over 16881.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2679, pruned_loss=0.04028, over 3022796.73 frames. ], batch size: 102, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 01:59:45,795 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.57 vs. limit=5.0 2023-05-02 02:00:24,125 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=252297.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:00:33,683 INFO [train.py:904] (6/8) Epoch 25, batch 8700, loss[loss=0.1754, simple_loss=0.2728, pruned_loss=0.03898, over 16388.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2651, pruned_loss=0.03909, over 3025492.30 frames. ], batch size: 146, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:01:53,637 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=252345.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:01:54,399 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.271e+02 2.533e+02 3.010e+02 6.763e+02, threshold=5.067e+02, percent-clipped=2.0 2023-05-02 02:02:09,351 INFO [train.py:904] (6/8) Epoch 25, batch 8750, loss[loss=0.1662, simple_loss=0.2523, pruned_loss=0.04002, over 12266.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2646, pruned_loss=0.03826, over 3045721.85 frames. ], batch size: 248, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:02:41,200 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=252365.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:04:01,512 INFO [train.py:904] (6/8) Epoch 25, batch 8800, loss[loss=0.1772, simple_loss=0.2601, pruned_loss=0.04722, over 12559.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2632, pruned_loss=0.03732, over 3038952.16 frames. ], batch size: 248, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:04:22,162 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=252413.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:04:30,827 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252417.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:05:07,262 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252434.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:05:31,485 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.474e+02 2.215e+02 2.676e+02 3.061e+02 6.379e+02, threshold=5.352e+02, percent-clipped=4.0 2023-05-02 02:05:46,035 INFO [train.py:904] (6/8) Epoch 25, batch 8850, loss[loss=0.1585, simple_loss=0.2494, pruned_loss=0.03378, over 12329.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2663, pruned_loss=0.03703, over 3033545.42 frames. ], batch size: 246, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:05:49,325 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5947, 3.6995, 2.2821, 4.3030, 2.8116, 4.1174, 2.4503, 2.9983], device='cuda:6'), covar=tensor([0.0338, 0.0386, 0.1734, 0.0189, 0.0963, 0.0560, 0.1563, 0.0888], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0174, 0.0191, 0.0162, 0.0173, 0.0211, 0.0200, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 02:06:18,739 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9858, 1.8095, 1.5793, 1.4283, 1.8549, 1.4601, 1.5998, 1.8640], device='cuda:6'), covar=tensor([0.0241, 0.0443, 0.0564, 0.0539, 0.0337, 0.0453, 0.0228, 0.0339], device='cuda:6'), in_proj_covar=tensor([0.0211, 0.0232, 0.0222, 0.0223, 0.0233, 0.0232, 0.0228, 0.0229], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 02:06:29,761 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5281, 3.6434, 2.6800, 2.1678, 2.2807, 2.3871, 3.8998, 3.1432], device='cuda:6'), covar=tensor([0.3291, 0.0734, 0.2168, 0.3099, 0.2981, 0.2278, 0.0477, 0.1405], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0267, 0.0306, 0.0314, 0.0296, 0.0265, 0.0295, 0.0338], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 02:06:39,821 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=252478.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:07:16,725 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=252495.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:07:31,608 INFO [train.py:904] (6/8) Epoch 25, batch 8900, loss[loss=0.1566, simple_loss=0.2549, pruned_loss=0.0292, over 16522.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2659, pruned_loss=0.03615, over 3025657.28 frames. ], batch size: 68, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:07:47,975 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2986, 3.3763, 2.0195, 3.5984, 2.5089, 3.5934, 2.2318, 2.8179], device='cuda:6'), covar=tensor([0.0320, 0.0408, 0.1748, 0.0278, 0.0905, 0.0577, 0.1585, 0.0723], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0174, 0.0192, 0.0162, 0.0174, 0.0211, 0.0200, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 02:09:14,517 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-02 02:09:18,926 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.583e+02 2.220e+02 2.582e+02 3.143e+02 6.266e+02, threshold=5.163e+02, percent-clipped=2.0 2023-05-02 02:09:34,561 INFO [train.py:904] (6/8) Epoch 25, batch 8950, loss[loss=0.1645, simple_loss=0.2578, pruned_loss=0.03562, over 15281.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2657, pruned_loss=0.03639, over 3048983.76 frames. ], batch size: 192, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:09:38,886 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4033, 3.4798, 2.0073, 3.8153, 2.5560, 3.7397, 2.2509, 2.8171], device='cuda:6'), covar=tensor([0.0346, 0.0400, 0.1902, 0.0236, 0.1039, 0.0677, 0.1789, 0.0867], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0173, 0.0191, 0.0161, 0.0173, 0.0210, 0.0198, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 02:10:59,630 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252592.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:11:21,900 INFO [train.py:904] (6/8) Epoch 25, batch 9000, loss[loss=0.1544, simple_loss=0.2527, pruned_loss=0.02809, over 16690.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2627, pruned_loss=0.03534, over 3055931.60 frames. ], batch size: 134, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:11:21,901 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 02:11:31,592 INFO [train.py:938] (6/8) Epoch 25, validation: loss=0.1442, simple_loss=0.248, pruned_loss=0.02014, over 944034.00 frames. 2023-05-02 02:11:31,593 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 02:11:53,598 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252613.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:12:11,137 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4479, 4.2971, 4.5015, 4.6147, 4.7977, 4.3205, 4.8011, 4.8248], device='cuda:6'), covar=tensor([0.2112, 0.1495, 0.1669, 0.1068, 0.0686, 0.1164, 0.0827, 0.0796], device='cuda:6'), in_proj_covar=tensor([0.0621, 0.0763, 0.0879, 0.0776, 0.0592, 0.0613, 0.0643, 0.0745], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 02:13:01,732 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.469e+02 2.125e+02 2.594e+02 3.127e+02 6.449e+02, threshold=5.187e+02, percent-clipped=3.0 2023-05-02 02:13:06,496 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5684, 4.6744, 4.4617, 4.1376, 4.0354, 4.5615, 4.2935, 4.2504], device='cuda:6'), covar=tensor([0.0589, 0.0731, 0.0382, 0.0360, 0.1061, 0.0527, 0.0639, 0.0707], device='cuda:6'), in_proj_covar=tensor([0.0292, 0.0437, 0.0340, 0.0344, 0.0343, 0.0395, 0.0236, 0.0408], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 02:13:14,920 INFO [train.py:904] (6/8) Epoch 25, batch 9050, loss[loss=0.1614, simple_loss=0.258, pruned_loss=0.03239, over 16733.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2636, pruned_loss=0.03588, over 3067303.00 frames. ], batch size: 76, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:13:15,924 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=252653.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:13:47,222 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252668.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:13:56,978 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=252674.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:14:55,837 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2271, 4.2951, 4.1300, 3.8184, 3.8430, 4.2163, 3.9179, 3.9832], device='cuda:6'), covar=tensor([0.0640, 0.0746, 0.0349, 0.0341, 0.0747, 0.0713, 0.0842, 0.0699], device='cuda:6'), in_proj_covar=tensor([0.0291, 0.0436, 0.0339, 0.0343, 0.0341, 0.0393, 0.0234, 0.0407], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 02:14:58,383 INFO [train.py:904] (6/8) Epoch 25, batch 9100, loss[loss=0.1749, simple_loss=0.2686, pruned_loss=0.04064, over 16606.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2631, pruned_loss=0.03656, over 3052661.57 frames. ], batch size: 62, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:16:01,703 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=252729.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:16:41,919 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.125e+02 2.555e+02 2.905e+02 5.383e+02, threshold=5.110e+02, percent-clipped=1.0 2023-05-02 02:16:57,784 INFO [train.py:904] (6/8) Epoch 25, batch 9150, loss[loss=0.1467, simple_loss=0.2478, pruned_loss=0.02282, over 15623.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.263, pruned_loss=0.03612, over 3053800.97 frames. ], batch size: 194, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:17:34,393 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-05-02 02:17:42,773 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=252773.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:18:20,259 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=252790.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:18:43,278 INFO [train.py:904] (6/8) Epoch 25, batch 9200, loss[loss=0.1707, simple_loss=0.2648, pruned_loss=0.03832, over 15380.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.259, pruned_loss=0.03532, over 3047561.11 frames. ], batch size: 190, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:18:56,236 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4904, 4.4587, 4.2781, 3.3756, 4.4036, 1.5548, 4.0881, 4.0087], device='cuda:6'), covar=tensor([0.0163, 0.0157, 0.0269, 0.0498, 0.0165, 0.3281, 0.0197, 0.0366], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0162, 0.0200, 0.0176, 0.0178, 0.0208, 0.0189, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 02:20:05,202 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.337e+02 2.810e+02 3.410e+02 9.767e+02, threshold=5.620e+02, percent-clipped=4.0 2023-05-02 02:20:19,109 INFO [train.py:904] (6/8) Epoch 25, batch 9250, loss[loss=0.159, simple_loss=0.2551, pruned_loss=0.03148, over 16498.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2587, pruned_loss=0.03524, over 3042237.67 frames. ], batch size: 75, lr: 2.67e-03, grad_scale: 16.0 2023-05-02 02:20:28,424 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252857.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:21:21,912 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4042, 3.4946, 2.0888, 3.8931, 2.6105, 3.8176, 2.3735, 2.8650], device='cuda:6'), covar=tensor([0.0323, 0.0417, 0.1713, 0.0226, 0.0931, 0.0636, 0.1499, 0.0779], device='cuda:6'), in_proj_covar=tensor([0.0167, 0.0172, 0.0190, 0.0160, 0.0172, 0.0208, 0.0197, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 02:21:32,639 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8269, 3.1492, 3.5228, 2.1650, 2.9893, 2.2076, 3.4121, 3.3426], device='cuda:6'), covar=tensor([0.0317, 0.0964, 0.0469, 0.2052, 0.0771, 0.1094, 0.0657, 0.1035], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0160, 0.0164, 0.0151, 0.0142, 0.0127, 0.0140, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 02:22:11,611 INFO [train.py:904] (6/8) Epoch 25, batch 9300, loss[loss=0.1619, simple_loss=0.2455, pruned_loss=0.03914, over 12233.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2566, pruned_loss=0.0346, over 3033586.48 frames. ], batch size: 248, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:22:47,960 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=252918.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:23:38,560 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-02 02:23:45,073 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 1.953e+02 2.173e+02 2.787e+02 5.118e+02, threshold=4.346e+02, percent-clipped=0.0 2023-05-02 02:23:47,874 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=252948.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:23:55,314 INFO [train.py:904] (6/8) Epoch 25, batch 9350, loss[loss=0.173, simple_loss=0.2655, pruned_loss=0.04022, over 17003.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2566, pruned_loss=0.03442, over 3046135.25 frames. ], batch size: 109, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:24:30,056 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=252969.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:25:17,154 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252993.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:25:36,665 INFO [train.py:904] (6/8) Epoch 25, batch 9400, loss[loss=0.1621, simple_loss=0.249, pruned_loss=0.03755, over 12444.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2567, pruned_loss=0.03403, over 3044388.76 frames. ], batch size: 248, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:26:10,898 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6858, 1.9104, 2.3230, 2.6753, 2.5322, 3.0963, 2.1650, 3.0388], device='cuda:6'), covar=tensor([0.0256, 0.0617, 0.0439, 0.0385, 0.0414, 0.0206, 0.0550, 0.0166], device='cuda:6'), in_proj_covar=tensor([0.0188, 0.0193, 0.0180, 0.0182, 0.0198, 0.0157, 0.0195, 0.0156], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 02:26:19,765 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253024.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:27:05,784 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.393e+02 2.088e+02 2.613e+02 3.190e+02 8.435e+02, threshold=5.225e+02, percent-clipped=5.0 2023-05-02 02:27:16,649 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-02 02:27:17,025 INFO [train.py:904] (6/8) Epoch 25, batch 9450, loss[loss=0.1612, simple_loss=0.2589, pruned_loss=0.03175, over 16212.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.258, pruned_loss=0.03433, over 3033187.78 frames. ], batch size: 165, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:27:19,819 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253054.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:27:57,712 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253073.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:28:31,198 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253090.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:28:54,659 INFO [train.py:904] (6/8) Epoch 25, batch 9500, loss[loss=0.1609, simple_loss=0.2489, pruned_loss=0.03646, over 12999.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2573, pruned_loss=0.03399, over 3045808.37 frames. ], batch size: 246, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:28:59,435 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8486, 3.1341, 2.9340, 5.1086, 3.7549, 4.5406, 1.8008, 3.3462], device='cuda:6'), covar=tensor([0.1462, 0.0789, 0.1137, 0.0188, 0.0217, 0.0366, 0.1736, 0.0708], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0174, 0.0193, 0.0191, 0.0198, 0.0212, 0.0203, 0.0193], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 02:29:32,065 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2305, 1.5390, 2.0015, 2.1711, 2.2532, 2.4182, 1.6703, 2.3588], device='cuda:6'), covar=tensor([0.0232, 0.0666, 0.0337, 0.0369, 0.0343, 0.0254, 0.0729, 0.0205], device='cuda:6'), in_proj_covar=tensor([0.0189, 0.0194, 0.0181, 0.0183, 0.0199, 0.0158, 0.0196, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 02:29:33,432 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253121.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:29:47,902 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3925, 3.3521, 2.7252, 2.1090, 2.1734, 2.3375, 3.4975, 3.0070], device='cuda:6'), covar=tensor([0.3183, 0.0631, 0.1855, 0.3021, 0.2862, 0.2307, 0.0442, 0.1395], device='cuda:6'), in_proj_covar=tensor([0.0324, 0.0264, 0.0303, 0.0311, 0.0292, 0.0263, 0.0292, 0.0335], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 02:30:06,529 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253138.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:30:26,751 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.108e+02 2.493e+02 2.917e+02 7.333e+02, threshold=4.986e+02, percent-clipped=1.0 2023-05-02 02:30:40,916 INFO [train.py:904] (6/8) Epoch 25, batch 9550, loss[loss=0.1659, simple_loss=0.2556, pruned_loss=0.03812, over 12601.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2568, pruned_loss=0.03392, over 3052438.28 frames. ], batch size: 250, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:31:49,785 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 02:32:22,177 INFO [train.py:904] (6/8) Epoch 25, batch 9600, loss[loss=0.169, simple_loss=0.2654, pruned_loss=0.03626, over 16839.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.258, pruned_loss=0.0348, over 3029224.14 frames. ], batch size: 76, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:32:43,068 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253213.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:33:55,685 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.643e+02 2.388e+02 2.773e+02 3.482e+02 6.416e+02, threshold=5.546e+02, percent-clipped=8.0 2023-05-02 02:33:59,726 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253248.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:34:10,800 INFO [train.py:904] (6/8) Epoch 25, batch 9650, loss[loss=0.1538, simple_loss=0.2553, pruned_loss=0.02616, over 16496.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2603, pruned_loss=0.03525, over 3038636.34 frames. ], batch size: 75, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:34:52,623 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253269.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:35:28,346 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1772, 3.5165, 3.4739, 2.3755, 3.2343, 3.5616, 3.3347, 2.0455], device='cuda:6'), covar=tensor([0.0600, 0.0052, 0.0067, 0.0449, 0.0120, 0.0090, 0.0096, 0.0551], device='cuda:6'), in_proj_covar=tensor([0.0133, 0.0084, 0.0086, 0.0132, 0.0098, 0.0108, 0.0094, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 02:35:46,091 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253296.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:36:00,301 INFO [train.py:904] (6/8) Epoch 25, batch 9700, loss[loss=0.1734, simple_loss=0.2721, pruned_loss=0.03736, over 16776.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2593, pruned_loss=0.03507, over 3030315.94 frames. ], batch size: 83, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:36:15,704 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.27 vs. limit=5.0 2023-05-02 02:36:27,590 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253317.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:36:42,908 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253324.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:36:54,841 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-05-02 02:36:56,157 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=253329.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:37:34,480 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.439e+02 2.159e+02 2.493e+02 2.931e+02 5.551e+02, threshold=4.985e+02, percent-clipped=1.0 2023-05-02 02:37:37,163 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253349.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:37:44,442 INFO [train.py:904] (6/8) Epoch 25, batch 9750, loss[loss=0.1576, simple_loss=0.242, pruned_loss=0.03662, over 12476.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2582, pruned_loss=0.03527, over 3029816.80 frames. ], batch size: 248, lr: 2.67e-03, grad_scale: 4.0 2023-05-02 02:38:21,431 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253372.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:38:38,950 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=253380.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:38:44,585 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4641, 2.5183, 2.1122, 2.3074, 2.8437, 2.5354, 2.9025, 3.0756], device='cuda:6'), covar=tensor([0.0151, 0.0532, 0.0651, 0.0534, 0.0358, 0.0458, 0.0257, 0.0313], device='cuda:6'), in_proj_covar=tensor([0.0212, 0.0236, 0.0225, 0.0225, 0.0234, 0.0234, 0.0228, 0.0231], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 02:39:01,168 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253390.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:39:23,675 INFO [train.py:904] (6/8) Epoch 25, batch 9800, loss[loss=0.1644, simple_loss=0.2683, pruned_loss=0.03019, over 16723.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2585, pruned_loss=0.03458, over 3036700.69 frames. ], batch size: 134, lr: 2.67e-03, grad_scale: 4.0 2023-05-02 02:39:53,081 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=253419.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:40:38,464 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253441.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:40:55,714 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 1.999e+02 2.298e+02 2.745e+02 1.319e+03, threshold=4.597e+02, percent-clipped=1.0 2023-05-02 02:41:06,296 INFO [train.py:904] (6/8) Epoch 25, batch 9850, loss[loss=0.165, simple_loss=0.2608, pruned_loss=0.03457, over 15259.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2598, pruned_loss=0.03442, over 3040101.89 frames. ], batch size: 191, lr: 2.67e-03, grad_scale: 4.0 2023-05-02 02:41:26,278 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5708, 2.7803, 3.1270, 1.9064, 2.7554, 2.0024, 3.1245, 3.0194], device='cuda:6'), covar=tensor([0.0310, 0.0973, 0.0603, 0.2238, 0.0903, 0.1125, 0.0765, 0.0987], device='cuda:6'), in_proj_covar=tensor([0.0153, 0.0160, 0.0165, 0.0151, 0.0142, 0.0128, 0.0140, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 02:42:04,063 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253480.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:42:06,511 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-02 02:43:00,285 INFO [train.py:904] (6/8) Epoch 25, batch 9900, loss[loss=0.161, simple_loss=0.2702, pruned_loss=0.02587, over 16775.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2601, pruned_loss=0.03396, over 3046724.71 frames. ], batch size: 83, lr: 2.67e-03, grad_scale: 4.0 2023-05-02 02:43:25,531 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253513.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:43:41,896 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=253521.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:44:42,785 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8364, 3.8365, 3.9879, 3.7662, 3.9315, 4.3246, 3.9775, 3.6805], device='cuda:6'), covar=tensor([0.2155, 0.2238, 0.2368, 0.2274, 0.2645, 0.1535, 0.1622, 0.2537], device='cuda:6'), in_proj_covar=tensor([0.0395, 0.0581, 0.0645, 0.0478, 0.0632, 0.0670, 0.0504, 0.0637], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 02:44:45,871 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 02:44:46,255 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.406e+02 2.031e+02 2.358e+02 2.986e+02 6.654e+02, threshold=4.715e+02, percent-clipped=4.0 2023-05-02 02:44:53,555 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7455, 2.1689, 1.8807, 1.9977, 2.5230, 2.1715, 2.1865, 2.6280], device='cuda:6'), covar=tensor([0.0221, 0.0514, 0.0573, 0.0525, 0.0278, 0.0420, 0.0265, 0.0273], device='cuda:6'), in_proj_covar=tensor([0.0211, 0.0236, 0.0224, 0.0225, 0.0234, 0.0234, 0.0228, 0.0230], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 02:44:59,712 INFO [train.py:904] (6/8) Epoch 25, batch 9950, loss[loss=0.1706, simple_loss=0.2687, pruned_loss=0.03626, over 16248.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2626, pruned_loss=0.03439, over 3060582.17 frames. ], batch size: 165, lr: 2.67e-03, grad_scale: 4.0 2023-05-02 02:45:18,772 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253561.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:45:34,710 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=253567.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:45:47,191 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3782, 4.3311, 4.6843, 4.6715, 4.6710, 4.4165, 4.3916, 4.3521], device='cuda:6'), covar=tensor([0.0345, 0.0807, 0.0437, 0.0388, 0.0465, 0.0443, 0.0886, 0.0464], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0452, 0.0442, 0.0403, 0.0486, 0.0463, 0.0534, 0.0370], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 02:46:14,379 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253582.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:47:02,499 INFO [train.py:904] (6/8) Epoch 25, batch 10000, loss[loss=0.1645, simple_loss=0.2656, pruned_loss=0.03164, over 16776.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2616, pruned_loss=0.03424, over 3085309.96 frames. ], batch size: 134, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:47:19,372 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4377, 2.0645, 1.8226, 1.7632, 2.3161, 1.9475, 1.8740, 2.3398], device='cuda:6'), covar=tensor([0.0223, 0.0450, 0.0548, 0.0515, 0.0298, 0.0408, 0.0266, 0.0298], device='cuda:6'), in_proj_covar=tensor([0.0212, 0.0236, 0.0225, 0.0225, 0.0235, 0.0235, 0.0228, 0.0230], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 02:47:52,957 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253628.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:48:07,731 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-02 02:48:36,793 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.008e+02 2.354e+02 2.749e+02 5.133e+02, threshold=4.708e+02, percent-clipped=1.0 2023-05-02 02:48:39,830 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253649.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:48:41,822 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=253650.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:48:46,076 INFO [train.py:904] (6/8) Epoch 25, batch 10050, loss[loss=0.1675, simple_loss=0.2671, pruned_loss=0.03395, over 16875.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2614, pruned_loss=0.03411, over 3057667.53 frames. ], batch size: 96, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:49:29,127 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0169, 5.2875, 5.1298, 5.1221, 4.8368, 4.8524, 4.6919, 5.4030], device='cuda:6'), covar=tensor([0.1269, 0.0875, 0.0893, 0.0801, 0.0769, 0.0880, 0.1266, 0.0879], device='cuda:6'), in_proj_covar=tensor([0.0672, 0.0812, 0.0667, 0.0626, 0.0515, 0.0520, 0.0677, 0.0633], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 02:49:48,421 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253685.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:49:53,737 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8461, 5.0550, 4.8892, 4.9172, 4.6266, 4.6404, 4.4243, 5.1443], device='cuda:6'), covar=tensor([0.1148, 0.0868, 0.0894, 0.0807, 0.0755, 0.1017, 0.1308, 0.0858], device='cuda:6'), in_proj_covar=tensor([0.0672, 0.0812, 0.0668, 0.0626, 0.0515, 0.0521, 0.0678, 0.0633], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 02:50:11,423 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253697.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:50:20,895 INFO [train.py:904] (6/8) Epoch 25, batch 10100, loss[loss=0.164, simple_loss=0.2541, pruned_loss=0.03691, over 16635.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2618, pruned_loss=0.03437, over 3049424.09 frames. ], batch size: 134, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:50:28,979 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4727, 3.5205, 3.7083, 3.6845, 3.7045, 3.5445, 3.5810, 3.5845], device='cuda:6'), covar=tensor([0.0537, 0.2078, 0.0859, 0.0700, 0.0754, 0.0832, 0.0768, 0.0788], device='cuda:6'), in_proj_covar=tensor([0.0401, 0.0451, 0.0441, 0.0403, 0.0485, 0.0462, 0.0532, 0.0370], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 02:50:35,651 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253711.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:51:21,587 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253736.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:51:33,769 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.529e+02 2.192e+02 2.668e+02 3.131e+02 7.050e+02, threshold=5.336e+02, percent-clipped=1.0 2023-05-02 02:52:07,291 INFO [train.py:904] (6/8) Epoch 26, batch 0, loss[loss=0.223, simple_loss=0.2882, pruned_loss=0.07887, over 16878.00 frames. ], tot_loss[loss=0.223, simple_loss=0.2882, pruned_loss=0.07887, over 16878.00 frames. ], batch size: 116, lr: 2.61e-03, grad_scale: 8.0 2023-05-02 02:52:07,292 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 02:52:14,675 INFO [train.py:938] (6/8) Epoch 26, validation: loss=0.1437, simple_loss=0.2472, pruned_loss=0.02009, over 944034.00 frames. 2023-05-02 02:52:14,676 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 02:52:44,911 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253775.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:53:23,659 INFO [train.py:904] (6/8) Epoch 26, batch 50, loss[loss=0.1733, simple_loss=0.2706, pruned_loss=0.03804, over 17061.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2659, pruned_loss=0.04701, over 757462.00 frames. ], batch size: 50, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 02:54:28,065 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.564e+02 2.375e+02 2.827e+02 3.414e+02 5.395e+02, threshold=5.654e+02, percent-clipped=1.0 2023-05-02 02:54:30,988 INFO [train.py:904] (6/8) Epoch 26, batch 100, loss[loss=0.1722, simple_loss=0.2513, pruned_loss=0.04652, over 16813.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2615, pruned_loss=0.04422, over 1332232.42 frames. ], batch size: 90, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 02:55:03,744 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253877.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:55:39,318 INFO [train.py:904] (6/8) Epoch 26, batch 150, loss[loss=0.1562, simple_loss=0.2504, pruned_loss=0.03099, over 17206.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2599, pruned_loss=0.04265, over 1779569.38 frames. ], batch size: 44, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 02:56:08,414 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253923.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:56:19,589 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-02 02:56:46,629 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.138e+02 2.556e+02 3.055e+02 5.694e+02, threshold=5.112e+02, percent-clipped=1.0 2023-05-02 02:56:49,068 INFO [train.py:904] (6/8) Epoch 26, batch 200, loss[loss=0.1394, simple_loss=0.2216, pruned_loss=0.02859, over 16787.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.259, pruned_loss=0.0423, over 2129129.37 frames. ], batch size: 42, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 02:57:34,313 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253985.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:58:03,909 INFO [train.py:904] (6/8) Epoch 26, batch 250, loss[loss=0.1905, simple_loss=0.2622, pruned_loss=0.05944, over 16884.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2574, pruned_loss=0.04266, over 2397572.59 frames. ], batch size: 116, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 02:58:07,778 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=254006.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:58:47,184 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=254033.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:58:50,826 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=254036.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:59:10,748 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.700e+02 2.097e+02 2.455e+02 3.037e+02 5.274e+02, threshold=4.910e+02, percent-clipped=1.0 2023-05-02 02:59:14,066 INFO [train.py:904] (6/8) Epoch 26, batch 300, loss[loss=0.1712, simple_loss=0.2505, pruned_loss=0.04599, over 16466.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2548, pruned_loss=0.04154, over 2598242.71 frames. ], batch size: 146, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 02:59:45,468 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=254075.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:59:57,804 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=254084.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:00:23,586 INFO [train.py:904] (6/8) Epoch 26, batch 350, loss[loss=0.1406, simple_loss=0.2312, pruned_loss=0.02503, over 17241.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2533, pruned_loss=0.04049, over 2761064.82 frames. ], batch size: 45, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 03:00:49,177 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3964, 4.2200, 4.4227, 4.5961, 4.7124, 4.2562, 4.5404, 4.6674], device='cuda:6'), covar=tensor([0.1771, 0.1289, 0.1538, 0.0747, 0.0615, 0.1129, 0.2498, 0.0792], device='cuda:6'), in_proj_covar=tensor([0.0643, 0.0787, 0.0908, 0.0800, 0.0612, 0.0633, 0.0668, 0.0771], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:00:52,069 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=254123.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:00:52,367 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1495, 2.1890, 2.3360, 3.7636, 2.2316, 2.5087, 2.2671, 2.3131], device='cuda:6'), covar=tensor([0.1525, 0.3786, 0.3151, 0.0727, 0.3808, 0.2503, 0.3868, 0.3256], device='cuda:6'), in_proj_covar=tensor([0.0411, 0.0461, 0.0380, 0.0331, 0.0442, 0.0525, 0.0432, 0.0538], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:00:58,082 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6207, 4.5004, 4.5587, 4.2219, 4.3017, 4.5897, 4.4020, 4.3397], device='cuda:6'), covar=tensor([0.0576, 0.0804, 0.0309, 0.0310, 0.0766, 0.0465, 0.0482, 0.0651], device='cuda:6'), in_proj_covar=tensor([0.0297, 0.0444, 0.0346, 0.0349, 0.0348, 0.0400, 0.0238, 0.0414], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:01:30,800 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 2.013e+02 2.457e+02 3.049e+02 5.935e+02, threshold=4.915e+02, percent-clipped=1.0 2023-05-02 03:01:33,888 INFO [train.py:904] (6/8) Epoch 26, batch 400, loss[loss=0.1709, simple_loss=0.2504, pruned_loss=0.0457, over 16867.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2533, pruned_loss=0.04111, over 2883187.12 frames. ], batch size: 96, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:02:08,617 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=254177.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:02:44,708 INFO [train.py:904] (6/8) Epoch 26, batch 450, loss[loss=0.155, simple_loss=0.2352, pruned_loss=0.03736, over 15628.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2515, pruned_loss=0.04003, over 2971922.45 frames. ], batch size: 191, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:03:01,494 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6012, 3.6401, 2.3284, 3.8726, 2.9014, 3.8385, 2.5456, 3.0686], device='cuda:6'), covar=tensor([0.0285, 0.0463, 0.1528, 0.0401, 0.0815, 0.0693, 0.1319, 0.0673], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0179, 0.0197, 0.0169, 0.0179, 0.0217, 0.0205, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 03:03:12,813 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=254223.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:03:15,095 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=254225.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:03:50,701 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.473e+02 2.133e+02 2.439e+02 2.881e+02 4.883e+02, threshold=4.879e+02, percent-clipped=0.0 2023-05-02 03:03:52,994 INFO [train.py:904] (6/8) Epoch 26, batch 500, loss[loss=0.1615, simple_loss=0.2374, pruned_loss=0.0428, over 16814.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2503, pruned_loss=0.03981, over 3042135.30 frames. ], batch size: 124, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:04:07,725 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3368, 2.3843, 2.4395, 4.0764, 2.3286, 2.6902, 2.4534, 2.5235], device='cuda:6'), covar=tensor([0.1459, 0.3735, 0.3333, 0.0632, 0.4368, 0.2841, 0.3581, 0.3802], device='cuda:6'), in_proj_covar=tensor([0.0415, 0.0464, 0.0383, 0.0334, 0.0445, 0.0530, 0.0436, 0.0543], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:04:19,050 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=254271.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:05:01,743 INFO [train.py:904] (6/8) Epoch 26, batch 550, loss[loss=0.1683, simple_loss=0.2729, pruned_loss=0.03191, over 17141.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2497, pruned_loss=0.03921, over 3091295.00 frames. ], batch size: 49, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:05:05,465 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=254306.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:05:32,141 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9725, 5.3965, 5.5571, 5.2788, 5.3058, 5.9599, 5.3948, 5.0990], device='cuda:6'), covar=tensor([0.1164, 0.2108, 0.2361, 0.2286, 0.3018, 0.1103, 0.1687, 0.2593], device='cuda:6'), in_proj_covar=tensor([0.0420, 0.0617, 0.0688, 0.0510, 0.0674, 0.0709, 0.0534, 0.0677], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 03:06:08,396 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.696e+02 2.164e+02 2.429e+02 2.784e+02 6.302e+02, threshold=4.858e+02, percent-clipped=1.0 2023-05-02 03:06:09,237 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-05-02 03:06:11,643 INFO [train.py:904] (6/8) Epoch 26, batch 600, loss[loss=0.1599, simple_loss=0.2368, pruned_loss=0.04153, over 15622.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.249, pruned_loss=0.03911, over 3144276.12 frames. ], batch size: 191, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:06:13,074 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=254354.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:07:21,734 INFO [train.py:904] (6/8) Epoch 26, batch 650, loss[loss=0.1644, simple_loss=0.2547, pruned_loss=0.03703, over 16651.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.248, pruned_loss=0.03869, over 3191314.29 frames. ], batch size: 62, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:08:16,596 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4590, 2.3842, 2.2965, 4.2869, 2.3835, 2.7059, 2.4212, 2.4674], device='cuda:6'), covar=tensor([0.1345, 0.3701, 0.3363, 0.0541, 0.4085, 0.2795, 0.3762, 0.3741], device='cuda:6'), in_proj_covar=tensor([0.0416, 0.0465, 0.0384, 0.0334, 0.0445, 0.0531, 0.0436, 0.0544], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:08:28,765 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.176e+02 2.565e+02 3.138e+02 6.256e+02, threshold=5.131e+02, percent-clipped=2.0 2023-05-02 03:08:30,991 INFO [train.py:904] (6/8) Epoch 26, batch 700, loss[loss=0.1635, simple_loss=0.2613, pruned_loss=0.0328, over 17072.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2482, pruned_loss=0.03871, over 3220235.12 frames. ], batch size: 50, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:09:07,477 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3884, 4.6894, 4.5028, 4.5165, 4.2402, 4.2053, 4.2287, 4.7364], device='cuda:6'), covar=tensor([0.1206, 0.0899, 0.0939, 0.0843, 0.0839, 0.1504, 0.1055, 0.0951], device='cuda:6'), in_proj_covar=tensor([0.0715, 0.0864, 0.0710, 0.0667, 0.0550, 0.0554, 0.0723, 0.0672], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:09:41,479 INFO [train.py:904] (6/8) Epoch 26, batch 750, loss[loss=0.1623, simple_loss=0.2441, pruned_loss=0.04021, over 16856.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2478, pruned_loss=0.03829, over 3247288.12 frames. ], batch size: 102, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:10:27,516 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=254536.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:10:46,994 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.386e+02 2.194e+02 2.558e+02 3.127e+02 5.314e+02, threshold=5.115e+02, percent-clipped=1.0 2023-05-02 03:10:50,632 INFO [train.py:904] (6/8) Epoch 26, batch 800, loss[loss=0.1396, simple_loss=0.2401, pruned_loss=0.0195, over 17163.00 frames. ], tot_loss[loss=0.162, simple_loss=0.2476, pruned_loss=0.03816, over 3255389.80 frames. ], batch size: 46, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:11:10,826 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 03:11:20,505 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=254574.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:11:51,481 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=254597.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:11:59,169 INFO [train.py:904] (6/8) Epoch 26, batch 850, loss[loss=0.1644, simple_loss=0.2475, pruned_loss=0.04068, over 16695.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.2466, pruned_loss=0.03756, over 3269919.94 frames. ], batch size: 89, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:12:45,966 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=254635.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:13:08,051 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 2.023e+02 2.316e+02 2.771e+02 4.972e+02, threshold=4.632e+02, percent-clipped=1.0 2023-05-02 03:13:10,212 INFO [train.py:904] (6/8) Epoch 26, batch 900, loss[loss=0.1491, simple_loss=0.2319, pruned_loss=0.03316, over 16024.00 frames. ], tot_loss[loss=0.1604, simple_loss=0.2465, pruned_loss=0.03714, over 3280066.71 frames. ], batch size: 35, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:13:41,840 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2825, 5.3219, 5.7211, 5.6994, 5.7272, 5.3760, 5.3020, 5.1444], device='cuda:6'), covar=tensor([0.0336, 0.0614, 0.0402, 0.0417, 0.0464, 0.0397, 0.0899, 0.0471], device='cuda:6'), in_proj_covar=tensor([0.0428, 0.0481, 0.0469, 0.0429, 0.0515, 0.0493, 0.0568, 0.0393], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 03:14:19,297 INFO [train.py:904] (6/8) Epoch 26, batch 950, loss[loss=0.1474, simple_loss=0.2495, pruned_loss=0.02263, over 17240.00 frames. ], tot_loss[loss=0.1606, simple_loss=0.2465, pruned_loss=0.03739, over 3296887.65 frames. ], batch size: 52, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:14:23,681 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8374, 4.2246, 2.8833, 2.3217, 2.5551, 2.4463, 4.5859, 3.3651], device='cuda:6'), covar=tensor([0.2892, 0.0580, 0.2046, 0.3070, 0.3151, 0.2273, 0.0392, 0.1638], device='cuda:6'), in_proj_covar=tensor([0.0332, 0.0272, 0.0311, 0.0320, 0.0301, 0.0271, 0.0301, 0.0346], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 03:15:07,790 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9117, 2.7458, 2.5540, 4.8407, 3.6240, 4.1733, 1.9084, 2.9572], device='cuda:6'), covar=tensor([0.1470, 0.0980, 0.1501, 0.0288, 0.0316, 0.0534, 0.1691, 0.1004], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0179, 0.0198, 0.0198, 0.0203, 0.0218, 0.0208, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 03:15:14,742 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0421, 3.7601, 4.2362, 2.2957, 4.4170, 4.5682, 3.2245, 3.5394], device='cuda:6'), covar=tensor([0.0741, 0.0321, 0.0277, 0.1153, 0.0106, 0.0190, 0.0507, 0.0462], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0111, 0.0100, 0.0140, 0.0085, 0.0130, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 03:15:24,091 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.276e+02 2.119e+02 2.483e+02 2.968e+02 5.109e+02, threshold=4.965e+02, percent-clipped=3.0 2023-05-02 03:15:27,115 INFO [train.py:904] (6/8) Epoch 26, batch 1000, loss[loss=0.1463, simple_loss=0.2362, pruned_loss=0.02824, over 16996.00 frames. ], tot_loss[loss=0.1609, simple_loss=0.247, pruned_loss=0.03741, over 3305703.18 frames. ], batch size: 41, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:15:31,185 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=254755.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:16:28,379 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=254797.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:16:35,898 INFO [train.py:904] (6/8) Epoch 26, batch 1050, loss[loss=0.163, simple_loss=0.2494, pruned_loss=0.0383, over 16552.00 frames. ], tot_loss[loss=0.1606, simple_loss=0.2462, pruned_loss=0.03754, over 3313158.00 frames. ], batch size: 75, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:16:55,188 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=254816.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:17:42,968 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=254851.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:17:43,591 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.171e+02 2.721e+02 3.139e+02 5.089e+02, threshold=5.442e+02, percent-clipped=1.0 2023-05-02 03:17:44,777 INFO [train.py:904] (6/8) Epoch 26, batch 1100, loss[loss=0.162, simple_loss=0.2533, pruned_loss=0.03537, over 16645.00 frames. ], tot_loss[loss=0.16, simple_loss=0.2453, pruned_loss=0.03731, over 3318376.10 frames. ], batch size: 57, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:17:53,574 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=254858.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:18:36,898 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=254892.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:18:53,028 INFO [train.py:904] (6/8) Epoch 26, batch 1150, loss[loss=0.1465, simple_loss=0.2367, pruned_loss=0.02815, over 17137.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.2447, pruned_loss=0.03716, over 3324918.70 frames. ], batch size: 47, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:19:04,338 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=254912.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:19:29,863 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=254930.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:19:42,423 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0441, 2.2055, 2.8078, 3.0280, 2.8552, 3.5481, 2.3445, 3.5381], device='cuda:6'), covar=tensor([0.0303, 0.0621, 0.0346, 0.0360, 0.0369, 0.0239, 0.0588, 0.0245], device='cuda:6'), in_proj_covar=tensor([0.0195, 0.0198, 0.0186, 0.0188, 0.0205, 0.0163, 0.0201, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:19:59,611 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.804e+02 2.093e+02 2.455e+02 3.027e+02 6.541e+02, threshold=4.911e+02, percent-clipped=2.0 2023-05-02 03:20:00,665 INFO [train.py:904] (6/8) Epoch 26, batch 1200, loss[loss=0.1582, simple_loss=0.2364, pruned_loss=0.03998, over 15567.00 frames. ], tot_loss[loss=0.159, simple_loss=0.2446, pruned_loss=0.03674, over 3329700.57 frames. ], batch size: 190, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:21:10,674 INFO [train.py:904] (6/8) Epoch 26, batch 1250, loss[loss=0.1549, simple_loss=0.2453, pruned_loss=0.03227, over 17117.00 frames. ], tot_loss[loss=0.1591, simple_loss=0.2451, pruned_loss=0.03662, over 3328568.15 frames. ], batch size: 47, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:21:13,219 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4147, 3.4810, 3.6965, 2.6265, 3.3741, 3.8249, 3.4967, 2.1894], device='cuda:6'), covar=tensor([0.0534, 0.0147, 0.0068, 0.0399, 0.0121, 0.0119, 0.0118, 0.0509], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0088, 0.0089, 0.0136, 0.0102, 0.0113, 0.0098, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 03:21:25,562 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5970, 1.9081, 2.3052, 2.4237, 2.5662, 2.6128, 1.9540, 2.7282], device='cuda:6'), covar=tensor([0.0230, 0.0544, 0.0372, 0.0339, 0.0358, 0.0336, 0.0580, 0.0238], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0199, 0.0187, 0.0189, 0.0205, 0.0163, 0.0201, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:22:19,975 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.021e+02 2.372e+02 2.873e+02 4.328e+02, threshold=4.744e+02, percent-clipped=0.0 2023-05-02 03:22:21,123 INFO [train.py:904] (6/8) Epoch 26, batch 1300, loss[loss=0.1632, simple_loss=0.2623, pruned_loss=0.03203, over 16870.00 frames. ], tot_loss[loss=0.1587, simple_loss=0.2445, pruned_loss=0.03649, over 3321501.28 frames. ], batch size: 57, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:23:30,835 INFO [train.py:904] (6/8) Epoch 26, batch 1350, loss[loss=0.1656, simple_loss=0.2484, pruned_loss=0.04136, over 16254.00 frames. ], tot_loss[loss=0.1582, simple_loss=0.2441, pruned_loss=0.03617, over 3306511.11 frames. ], batch size: 165, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:23:42,881 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=255111.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:24:38,900 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.606e+02 2.137e+02 2.466e+02 2.927e+02 7.701e+02, threshold=4.932e+02, percent-clipped=1.0 2023-05-02 03:24:40,852 INFO [train.py:904] (6/8) Epoch 26, batch 1400, loss[loss=0.1532, simple_loss=0.2438, pruned_loss=0.03134, over 17129.00 frames. ], tot_loss[loss=0.1582, simple_loss=0.2446, pruned_loss=0.03588, over 3317381.32 frames. ], batch size: 48, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:24:41,767 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=255153.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:25:23,426 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2063, 2.3919, 2.8669, 3.1886, 3.0202, 3.7311, 2.7517, 3.6568], device='cuda:6'), covar=tensor([0.0287, 0.0554, 0.0449, 0.0374, 0.0392, 0.0210, 0.0469, 0.0217], device='cuda:6'), in_proj_covar=tensor([0.0197, 0.0200, 0.0188, 0.0190, 0.0206, 0.0164, 0.0202, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:25:34,220 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=255192.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:25:48,426 INFO [train.py:904] (6/8) Epoch 26, batch 1450, loss[loss=0.1528, simple_loss=0.231, pruned_loss=0.03733, over 16737.00 frames. ], tot_loss[loss=0.1581, simple_loss=0.2442, pruned_loss=0.03597, over 3320354.86 frames. ], batch size: 124, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:25:54,178 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=255207.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:26:26,862 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=255230.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:26:40,763 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=255240.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:26:56,773 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.573e+02 2.211e+02 2.586e+02 2.983e+02 9.031e+02, threshold=5.172e+02, percent-clipped=4.0 2023-05-02 03:26:57,913 INFO [train.py:904] (6/8) Epoch 26, batch 1500, loss[loss=0.1609, simple_loss=0.2548, pruned_loss=0.0335, over 17055.00 frames. ], tot_loss[loss=0.159, simple_loss=0.2445, pruned_loss=0.0367, over 3313820.61 frames. ], batch size: 50, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:27:31,537 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=255278.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:27:55,972 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7506, 4.6120, 4.6814, 4.3535, 4.3739, 4.6862, 4.5231, 4.4894], device='cuda:6'), covar=tensor([0.0672, 0.1048, 0.0396, 0.0372, 0.1003, 0.0626, 0.0526, 0.0674], device='cuda:6'), in_proj_covar=tensor([0.0314, 0.0468, 0.0365, 0.0370, 0.0368, 0.0424, 0.0251, 0.0439], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 03:28:04,972 INFO [train.py:904] (6/8) Epoch 26, batch 1550, loss[loss=0.1782, simple_loss=0.2684, pruned_loss=0.04401, over 16661.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.2462, pruned_loss=0.03804, over 3311513.23 frames. ], batch size: 62, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:28:35,326 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0373, 3.6037, 4.1262, 2.3260, 4.2288, 4.2872, 3.3032, 3.2822], device='cuda:6'), covar=tensor([0.0677, 0.0299, 0.0209, 0.1135, 0.0104, 0.0231, 0.0423, 0.0452], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0112, 0.0100, 0.0140, 0.0086, 0.0131, 0.0130, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 03:28:40,085 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.74 vs. limit=5.0 2023-05-02 03:29:12,926 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.706e+02 2.260e+02 2.709e+02 3.380e+02 1.198e+03, threshold=5.419e+02, percent-clipped=4.0 2023-05-02 03:29:14,102 INFO [train.py:904] (6/8) Epoch 26, batch 1600, loss[loss=0.1762, simple_loss=0.2695, pruned_loss=0.04145, over 17085.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2479, pruned_loss=0.03873, over 3307706.65 frames. ], batch size: 53, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:30:03,518 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=255388.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:30:12,101 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.92 vs. limit=5.0 2023-05-02 03:30:23,420 INFO [train.py:904] (6/8) Epoch 26, batch 1650, loss[loss=0.1474, simple_loss=0.2368, pruned_loss=0.02905, over 16776.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2498, pruned_loss=0.0393, over 3310002.22 frames. ], batch size: 39, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:30:35,808 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=255411.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:31:10,121 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8811, 5.1836, 5.3699, 5.0745, 5.1539, 5.7662, 5.2362, 4.9057], device='cuda:6'), covar=tensor([0.1310, 0.2039, 0.2787, 0.2222, 0.2868, 0.1108, 0.1850, 0.2635], device='cuda:6'), in_proj_covar=tensor([0.0428, 0.0633, 0.0701, 0.0519, 0.0686, 0.0726, 0.0545, 0.0690], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 03:31:29,037 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=255449.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:31:32,744 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.657e+02 2.212e+02 2.578e+02 3.028e+02 5.571e+02, threshold=5.156e+02, percent-clipped=1.0 2023-05-02 03:31:33,340 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7754, 3.9608, 2.6911, 4.5890, 3.1372, 4.5216, 2.7149, 3.2249], device='cuda:6'), covar=tensor([0.0332, 0.0427, 0.1497, 0.0252, 0.0781, 0.0482, 0.1455, 0.0762], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0183, 0.0201, 0.0175, 0.0181, 0.0223, 0.0208, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 03:31:34,045 INFO [train.py:904] (6/8) Epoch 26, batch 1700, loss[loss=0.1588, simple_loss=0.2576, pruned_loss=0.03002, over 17137.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2511, pruned_loss=0.03965, over 3308565.08 frames. ], batch size: 48, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:31:34,399 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=255453.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:31:43,230 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=255459.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:32:14,004 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7205, 2.7951, 2.3489, 2.6717, 3.0779, 2.7940, 3.3284, 3.2494], device='cuda:6'), covar=tensor([0.0180, 0.0431, 0.0579, 0.0473, 0.0319, 0.0426, 0.0293, 0.0323], device='cuda:6'), in_proj_covar=tensor([0.0231, 0.0249, 0.0236, 0.0237, 0.0248, 0.0247, 0.0246, 0.0246], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:32:40,139 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=255501.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:32:42,741 INFO [train.py:904] (6/8) Epoch 26, batch 1750, loss[loss=0.1718, simple_loss=0.2677, pruned_loss=0.03797, over 17069.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2522, pruned_loss=0.04043, over 3309675.05 frames. ], batch size: 53, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:32:47,895 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=255507.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:32:59,245 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0599, 5.2211, 5.0151, 4.6176, 4.2356, 5.2156, 5.1284, 4.7046], device='cuda:6'), covar=tensor([0.0962, 0.0706, 0.0526, 0.0531, 0.2245, 0.0575, 0.0370, 0.0973], device='cuda:6'), in_proj_covar=tensor([0.0317, 0.0473, 0.0369, 0.0373, 0.0372, 0.0427, 0.0254, 0.0445], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 03:33:33,673 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.41 vs. limit=5.0 2023-05-02 03:33:49,126 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.153e+02 2.652e+02 3.222e+02 7.615e+02, threshold=5.303e+02, percent-clipped=5.0 2023-05-02 03:33:51,387 INFO [train.py:904] (6/8) Epoch 26, batch 1800, loss[loss=0.1586, simple_loss=0.2515, pruned_loss=0.0328, over 17110.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2532, pruned_loss=0.04015, over 3312058.52 frames. ], batch size: 47, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:33:53,447 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 03:33:54,880 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=255555.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:34:35,529 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9339, 2.5478, 2.0328, 2.3372, 2.9021, 2.7027, 2.8971, 3.0235], device='cuda:6'), covar=tensor([0.0246, 0.0457, 0.0631, 0.0502, 0.0290, 0.0362, 0.0254, 0.0302], device='cuda:6'), in_proj_covar=tensor([0.0231, 0.0249, 0.0236, 0.0237, 0.0248, 0.0247, 0.0246, 0.0246], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:34:53,075 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3336, 3.9631, 4.3843, 2.3662, 4.6414, 4.7016, 3.4649, 3.7738], device='cuda:6'), covar=tensor([0.0649, 0.0265, 0.0261, 0.1134, 0.0075, 0.0216, 0.0443, 0.0383], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0113, 0.0100, 0.0141, 0.0086, 0.0132, 0.0131, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 03:34:59,035 INFO [train.py:904] (6/8) Epoch 26, batch 1850, loss[loss=0.2066, simple_loss=0.2975, pruned_loss=0.05786, over 12030.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2546, pruned_loss=0.04009, over 3317298.71 frames. ], batch size: 246, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:35:08,489 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-02 03:35:19,798 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4701, 2.4343, 2.4240, 4.3351, 2.3273, 2.8566, 2.4680, 2.5935], device='cuda:6'), covar=tensor([0.1348, 0.3670, 0.3171, 0.0500, 0.4184, 0.2577, 0.3711, 0.3670], device='cuda:6'), in_proj_covar=tensor([0.0419, 0.0470, 0.0386, 0.0338, 0.0446, 0.0536, 0.0441, 0.0548], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:36:05,021 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.481e+02 2.077e+02 2.353e+02 2.960e+02 6.152e+02, threshold=4.707e+02, percent-clipped=2.0 2023-05-02 03:36:06,158 INFO [train.py:904] (6/8) Epoch 26, batch 1900, loss[loss=0.1668, simple_loss=0.2472, pruned_loss=0.04325, over 16859.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2535, pruned_loss=0.03951, over 3323978.94 frames. ], batch size: 96, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:36:32,141 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7455, 2.4831, 2.3520, 3.9446, 3.1346, 3.9591, 1.4951, 2.9085], device='cuda:6'), covar=tensor([0.1464, 0.0756, 0.1310, 0.0172, 0.0163, 0.0370, 0.1670, 0.0836], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0180, 0.0199, 0.0199, 0.0205, 0.0220, 0.0209, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 03:37:15,826 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9028, 2.8038, 2.7321, 4.2930, 3.5158, 4.1559, 1.7606, 3.1405], device='cuda:6'), covar=tensor([0.1371, 0.0689, 0.1082, 0.0177, 0.0159, 0.0363, 0.1517, 0.0751], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0180, 0.0199, 0.0199, 0.0206, 0.0220, 0.0210, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 03:37:16,452 INFO [train.py:904] (6/8) Epoch 26, batch 1950, loss[loss=0.1723, simple_loss=0.2598, pruned_loss=0.04236, over 16877.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2533, pruned_loss=0.03944, over 3308307.49 frames. ], batch size: 96, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:37:32,840 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3405, 3.6071, 3.9296, 2.3222, 3.2334, 2.5062, 3.7369, 3.7545], device='cuda:6'), covar=tensor([0.0344, 0.0966, 0.0546, 0.2046, 0.0851, 0.1043, 0.0691, 0.1037], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0170, 0.0171, 0.0158, 0.0148, 0.0133, 0.0147, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 03:37:42,738 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-05-02 03:37:55,214 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2298, 5.1473, 5.0965, 4.5541, 4.7227, 5.1444, 5.0538, 4.7272], device='cuda:6'), covar=tensor([0.0628, 0.0559, 0.0356, 0.0430, 0.1213, 0.0509, 0.0343, 0.0869], device='cuda:6'), in_proj_covar=tensor([0.0317, 0.0472, 0.0368, 0.0372, 0.0371, 0.0426, 0.0253, 0.0445], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 03:38:12,956 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=255744.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:38:24,268 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.052e+02 2.408e+02 2.894e+02 5.578e+02, threshold=4.816e+02, percent-clipped=3.0 2023-05-02 03:38:25,464 INFO [train.py:904] (6/8) Epoch 26, batch 2000, loss[loss=0.1448, simple_loss=0.2346, pruned_loss=0.02754, over 16032.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.253, pruned_loss=0.03912, over 3313666.50 frames. ], batch size: 35, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:39:35,298 INFO [train.py:904] (6/8) Epoch 26, batch 2050, loss[loss=0.1788, simple_loss=0.2667, pruned_loss=0.04546, over 17122.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2526, pruned_loss=0.03906, over 3304234.86 frames. ], batch size: 49, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:40:10,802 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6868, 3.6946, 2.3719, 3.9736, 2.9927, 3.9198, 2.4579, 3.0445], device='cuda:6'), covar=tensor([0.0291, 0.0435, 0.1585, 0.0366, 0.0786, 0.0828, 0.1417, 0.0721], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0184, 0.0201, 0.0176, 0.0182, 0.0223, 0.0208, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 03:40:29,674 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1994, 2.2738, 2.4750, 4.0177, 2.2456, 2.6592, 2.3627, 2.4788], device='cuda:6'), covar=tensor([0.1616, 0.3779, 0.2938, 0.0658, 0.4188, 0.2667, 0.4020, 0.3273], device='cuda:6'), in_proj_covar=tensor([0.0419, 0.0470, 0.0385, 0.0338, 0.0446, 0.0536, 0.0440, 0.0548], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:40:44,483 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.146e+02 2.478e+02 2.962e+02 5.196e+02, threshold=4.956e+02, percent-clipped=1.0 2023-05-02 03:40:45,698 INFO [train.py:904] (6/8) Epoch 26, batch 2100, loss[loss=0.1977, simple_loss=0.2827, pruned_loss=0.05639, over 12124.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.253, pruned_loss=0.0391, over 3305610.18 frames. ], batch size: 248, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:41:54,462 INFO [train.py:904] (6/8) Epoch 26, batch 2150, loss[loss=0.1715, simple_loss=0.2689, pruned_loss=0.03706, over 17121.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2543, pruned_loss=0.03962, over 3307115.90 frames. ], batch size: 49, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:42:06,133 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.60 vs. limit=2.0 2023-05-02 03:42:52,697 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6524, 4.5886, 4.5742, 4.2814, 4.3391, 4.6249, 4.3766, 4.3581], device='cuda:6'), covar=tensor([0.0633, 0.0791, 0.0298, 0.0299, 0.0792, 0.0478, 0.0542, 0.0589], device='cuda:6'), in_proj_covar=tensor([0.0319, 0.0474, 0.0369, 0.0374, 0.0372, 0.0427, 0.0254, 0.0446], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 03:43:04,643 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 2.196e+02 2.655e+02 3.354e+02 7.063e+02, threshold=5.310e+02, percent-clipped=7.0 2023-05-02 03:43:05,921 INFO [train.py:904] (6/8) Epoch 26, batch 2200, loss[loss=0.2057, simple_loss=0.2885, pruned_loss=0.06146, over 12166.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2547, pruned_loss=0.04026, over 3302329.10 frames. ], batch size: 246, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:43:10,479 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=255956.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:44:19,344 INFO [train.py:904] (6/8) Epoch 26, batch 2250, loss[loss=0.186, simple_loss=0.2682, pruned_loss=0.05193, over 15457.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2548, pruned_loss=0.04058, over 3309264.67 frames. ], batch size: 191, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:44:40,072 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256017.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:45:09,603 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256038.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:45:17,779 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256044.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:45:27,437 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.334e+02 2.041e+02 2.398e+02 2.796e+02 4.429e+02, threshold=4.795e+02, percent-clipped=0.0 2023-05-02 03:45:29,153 INFO [train.py:904] (6/8) Epoch 26, batch 2300, loss[loss=0.1793, simple_loss=0.2529, pruned_loss=0.05291, over 16665.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2549, pruned_loss=0.04037, over 3315808.79 frames. ], batch size: 134, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:46:09,916 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256081.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:46:14,125 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256084.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:46:24,425 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256092.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:46:32,912 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5621, 2.4902, 2.5303, 4.4470, 2.5430, 2.8528, 2.5523, 2.7366], device='cuda:6'), covar=tensor([0.1365, 0.4072, 0.3278, 0.0527, 0.3923, 0.2791, 0.3860, 0.3638], device='cuda:6'), in_proj_covar=tensor([0.0420, 0.0470, 0.0386, 0.0339, 0.0446, 0.0536, 0.0441, 0.0549], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:46:34,704 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256099.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:46:39,730 INFO [train.py:904] (6/8) Epoch 26, batch 2350, loss[loss=0.1884, simple_loss=0.26, pruned_loss=0.05841, over 16796.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2549, pruned_loss=0.04049, over 3318169.45 frames. ], batch size: 102, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:47:10,199 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5997, 2.4681, 2.5520, 4.4014, 2.4670, 2.8748, 2.5549, 2.7323], device='cuda:6'), covar=tensor([0.1319, 0.3784, 0.3173, 0.0561, 0.4150, 0.2745, 0.3676, 0.3611], device='cuda:6'), in_proj_covar=tensor([0.0421, 0.0471, 0.0386, 0.0339, 0.0447, 0.0537, 0.0442, 0.0549], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:47:19,446 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256132.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:47:32,086 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256142.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:47:37,011 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256145.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:47:42,485 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1483, 2.1198, 2.2594, 3.8573, 2.1824, 2.4434, 2.2392, 2.3015], device='cuda:6'), covar=tensor([0.1686, 0.4137, 0.3267, 0.0697, 0.4309, 0.2847, 0.4251, 0.3518], device='cuda:6'), in_proj_covar=tensor([0.0420, 0.0470, 0.0386, 0.0338, 0.0447, 0.0536, 0.0441, 0.0549], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:47:45,353 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.305e+02 2.073e+02 2.391e+02 2.953e+02 7.650e+02, threshold=4.783e+02, percent-clipped=2.0 2023-05-02 03:47:46,520 INFO [train.py:904] (6/8) Epoch 26, batch 2400, loss[loss=0.1841, simple_loss=0.2822, pruned_loss=0.04295, over 17043.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2554, pruned_loss=0.04049, over 3308718.03 frames. ], batch size: 55, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:47:54,441 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-02 03:48:42,796 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256193.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 03:48:55,676 INFO [train.py:904] (6/8) Epoch 26, batch 2450, loss[loss=0.1441, simple_loss=0.2282, pruned_loss=0.03005, over 16761.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2552, pruned_loss=0.04005, over 3308021.51 frames. ], batch size: 39, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:50:01,725 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.511e+02 2.193e+02 2.630e+02 3.165e+02 5.747e+02, threshold=5.261e+02, percent-clipped=2.0 2023-05-02 03:50:03,664 INFO [train.py:904] (6/8) Epoch 26, batch 2500, loss[loss=0.1736, simple_loss=0.255, pruned_loss=0.04612, over 16787.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2548, pruned_loss=0.04002, over 3315412.02 frames. ], batch size: 83, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:50:48,254 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256285.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:51:13,207 INFO [train.py:904] (6/8) Epoch 26, batch 2550, loss[loss=0.1755, simple_loss=0.2513, pruned_loss=0.04988, over 16711.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2548, pruned_loss=0.03932, over 3322925.38 frames. ], batch size: 134, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:51:26,243 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256312.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:52:14,086 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256346.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:52:22,558 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.118e+02 2.541e+02 2.890e+02 5.508e+02, threshold=5.081e+02, percent-clipped=1.0 2023-05-02 03:52:23,669 INFO [train.py:904] (6/8) Epoch 26, batch 2600, loss[loss=0.1401, simple_loss=0.2263, pruned_loss=0.027, over 16972.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2556, pruned_loss=0.03961, over 3325365.78 frames. ], batch size: 41, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:52:51,412 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256372.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:53:05,476 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1454, 5.7271, 5.9361, 5.6129, 5.7187, 6.2572, 5.7802, 5.5199], device='cuda:6'), covar=tensor([0.0935, 0.1916, 0.2373, 0.1874, 0.2456, 0.0964, 0.1454, 0.2168], device='cuda:6'), in_proj_covar=tensor([0.0434, 0.0638, 0.0705, 0.0522, 0.0692, 0.0727, 0.0547, 0.0697], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 03:53:21,519 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256394.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:53:33,354 INFO [train.py:904] (6/8) Epoch 26, batch 2650, loss[loss=0.1959, simple_loss=0.2746, pruned_loss=0.05861, over 16872.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2558, pruned_loss=0.03928, over 3334934.80 frames. ], batch size: 109, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:54:14,548 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256433.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:54:20,807 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256437.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:54:24,143 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256440.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:54:40,363 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.157e+02 2.407e+02 2.834e+02 4.989e+02, threshold=4.815e+02, percent-clipped=0.0 2023-05-02 03:54:41,407 INFO [train.py:904] (6/8) Epoch 26, batch 2700, loss[loss=0.1699, simple_loss=0.2688, pruned_loss=0.03551, over 17130.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2563, pruned_loss=0.03916, over 3342082.19 frames. ], batch size: 47, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:55:06,492 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256471.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:55:15,342 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7461, 3.9075, 2.9467, 2.3122, 2.6177, 2.6201, 4.1311, 3.4014], device='cuda:6'), covar=tensor([0.2984, 0.0686, 0.2004, 0.3328, 0.2931, 0.2111, 0.0570, 0.1549], device='cuda:6'), in_proj_covar=tensor([0.0333, 0.0276, 0.0312, 0.0324, 0.0305, 0.0273, 0.0303, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 03:55:28,869 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256488.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 03:55:48,616 INFO [train.py:904] (6/8) Epoch 26, batch 2750, loss[loss=0.1874, simple_loss=0.2651, pruned_loss=0.05487, over 16689.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2566, pruned_loss=0.03914, over 3345855.89 frames. ], batch size: 134, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:56:30,084 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256532.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:56:33,736 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2462, 2.3764, 2.4348, 4.0756, 2.2798, 2.6782, 2.4550, 2.5041], device='cuda:6'), covar=tensor([0.1560, 0.3571, 0.3118, 0.0649, 0.4081, 0.2740, 0.3861, 0.3184], device='cuda:6'), in_proj_covar=tensor([0.0421, 0.0471, 0.0386, 0.0339, 0.0447, 0.0538, 0.0442, 0.0551], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:56:56,714 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 2.155e+02 2.522e+02 3.116e+02 8.730e+02, threshold=5.045e+02, percent-clipped=2.0 2023-05-02 03:56:58,711 INFO [train.py:904] (6/8) Epoch 26, batch 2800, loss[loss=0.1277, simple_loss=0.215, pruned_loss=0.02015, over 16846.00 frames. ], tot_loss[loss=0.167, simple_loss=0.256, pruned_loss=0.03896, over 3339905.47 frames. ], batch size: 39, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:57:02,605 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1798, 3.1680, 3.2358, 2.2754, 3.0822, 3.4057, 3.1058, 1.9731], device='cuda:6'), covar=tensor([0.0553, 0.0163, 0.0089, 0.0458, 0.0138, 0.0124, 0.0137, 0.0579], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0088, 0.0089, 0.0136, 0.0101, 0.0113, 0.0098, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 03:58:07,641 INFO [train.py:904] (6/8) Epoch 26, batch 2850, loss[loss=0.2208, simple_loss=0.3038, pruned_loss=0.06889, over 12292.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2549, pruned_loss=0.03861, over 3336465.00 frames. ], batch size: 246, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:58:21,242 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256612.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:59:00,650 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256641.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:59:09,163 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2441, 1.7385, 2.0987, 2.1669, 2.3468, 2.3409, 1.7882, 2.3835], device='cuda:6'), covar=tensor([0.0241, 0.0470, 0.0278, 0.0358, 0.0319, 0.0348, 0.0480, 0.0198], device='cuda:6'), in_proj_covar=tensor([0.0197, 0.0199, 0.0188, 0.0191, 0.0206, 0.0166, 0.0203, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 03:59:15,040 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.116e+02 2.591e+02 2.981e+02 5.904e+02, threshold=5.182e+02, percent-clipped=1.0 2023-05-02 03:59:16,864 INFO [train.py:904] (6/8) Epoch 26, batch 2900, loss[loss=0.1396, simple_loss=0.2248, pruned_loss=0.02719, over 16897.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2543, pruned_loss=0.03909, over 3334524.47 frames. ], batch size: 96, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:59:27,994 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256660.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:59:29,231 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2449, 5.6268, 5.3192, 5.4310, 5.1219, 5.0356, 5.1063, 5.7357], device='cuda:6'), covar=tensor([0.1380, 0.0961, 0.1274, 0.0936, 0.0848, 0.0881, 0.1293, 0.0994], device='cuda:6'), in_proj_covar=tensor([0.0716, 0.0870, 0.0713, 0.0671, 0.0553, 0.0551, 0.0730, 0.0677], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:00:14,009 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256694.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:00:25,265 INFO [train.py:904] (6/8) Epoch 26, batch 2950, loss[loss=0.1572, simple_loss=0.2591, pruned_loss=0.02766, over 17161.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2535, pruned_loss=0.03931, over 3339382.10 frames. ], batch size: 49, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:00:33,225 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9703, 3.1452, 2.8452, 5.1235, 4.2562, 4.4889, 1.8049, 3.3248], device='cuda:6'), covar=tensor([0.1289, 0.0736, 0.1176, 0.0208, 0.0195, 0.0366, 0.1598, 0.0734], device='cuda:6'), in_proj_covar=tensor([0.0170, 0.0180, 0.0199, 0.0200, 0.0206, 0.0219, 0.0209, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 04:01:01,186 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256728.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:01:14,692 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256737.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:01:18,112 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256740.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:01:20,853 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256742.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:01:35,078 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.668e+02 2.243e+02 2.622e+02 3.035e+02 4.998e+02, threshold=5.244e+02, percent-clipped=0.0 2023-05-02 04:01:35,094 INFO [train.py:904] (6/8) Epoch 26, batch 3000, loss[loss=0.1756, simple_loss=0.2579, pruned_loss=0.04668, over 16890.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2545, pruned_loss=0.03985, over 3325448.29 frames. ], batch size: 116, lr: 2.60e-03, grad_scale: 4.0 2023-05-02 04:01:35,094 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 04:01:44,001 INFO [train.py:938] (6/8) Epoch 26, validation: loss=0.1339, simple_loss=0.2392, pruned_loss=0.01435, over 944034.00 frames. 2023-05-02 04:01:44,002 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 04:02:03,762 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6008, 3.6429, 2.2672, 3.9108, 2.9309, 3.8303, 2.3096, 2.9444], device='cuda:6'), covar=tensor([0.0279, 0.0412, 0.1502, 0.0293, 0.0718, 0.0769, 0.1484, 0.0741], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0184, 0.0201, 0.0177, 0.0182, 0.0225, 0.0208, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 04:02:27,413 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256785.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:02:31,514 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256788.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:02:31,602 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256788.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:02:53,042 INFO [train.py:904] (6/8) Epoch 26, batch 3050, loss[loss=0.1671, simple_loss=0.2492, pruned_loss=0.04249, over 16503.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2541, pruned_loss=0.04003, over 3327938.09 frames. ], batch size: 146, lr: 2.60e-03, grad_scale: 4.0 2023-05-02 04:03:19,822 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0646, 5.0360, 4.9269, 4.4715, 4.6105, 5.0241, 4.8251, 4.6933], device='cuda:6'), covar=tensor([0.0593, 0.0545, 0.0301, 0.0390, 0.0989, 0.0448, 0.0424, 0.0703], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0473, 0.0370, 0.0375, 0.0373, 0.0428, 0.0253, 0.0445], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 04:03:27,117 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256827.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:03:27,251 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256827.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 04:03:38,905 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256836.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:04:02,821 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.494e+02 2.111e+02 2.504e+02 3.018e+02 1.623e+03, threshold=5.009e+02, percent-clipped=3.0 2023-05-02 04:04:02,836 INFO [train.py:904] (6/8) Epoch 26, batch 3100, loss[loss=0.1695, simple_loss=0.2455, pruned_loss=0.04673, over 16707.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2535, pruned_loss=0.04004, over 3319167.68 frames. ], batch size: 89, lr: 2.60e-03, grad_scale: 4.0 2023-05-02 04:04:47,245 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256883.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:04:53,641 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256888.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:05:13,356 INFO [train.py:904] (6/8) Epoch 26, batch 3150, loss[loss=0.1749, simple_loss=0.2633, pruned_loss=0.04323, over 16733.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2535, pruned_loss=0.04045, over 3310262.60 frames. ], batch size: 89, lr: 2.60e-03, grad_scale: 4.0 2023-05-02 04:06:05,112 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256941.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:06:09,399 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256944.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 04:06:21,407 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.162e+02 2.494e+02 3.116e+02 1.108e+03, threshold=4.988e+02, percent-clipped=4.0 2023-05-02 04:06:21,429 INFO [train.py:904] (6/8) Epoch 26, batch 3200, loss[loss=0.1463, simple_loss=0.2521, pruned_loss=0.02024, over 17030.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2532, pruned_loss=0.03971, over 3318875.98 frames. ], batch size: 50, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:06:36,898 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2462, 5.1956, 4.9732, 4.4075, 5.0802, 2.0470, 4.7817, 4.8026], device='cuda:6'), covar=tensor([0.0097, 0.0094, 0.0237, 0.0421, 0.0104, 0.2780, 0.0148, 0.0252], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0174, 0.0212, 0.0188, 0.0188, 0.0218, 0.0201, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:06:40,222 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-05-02 04:07:11,961 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256989.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:07:29,905 INFO [train.py:904] (6/8) Epoch 26, batch 3250, loss[loss=0.155, simple_loss=0.2554, pruned_loss=0.02726, over 17295.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.253, pruned_loss=0.0393, over 3329207.87 frames. ], batch size: 52, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:07:43,707 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257013.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:08:03,376 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257028.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:08:05,586 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6232, 2.6381, 2.6792, 4.6371, 2.6235, 3.0065, 2.6687, 2.8064], device='cuda:6'), covar=tensor([0.1298, 0.3484, 0.3078, 0.0528, 0.3822, 0.2368, 0.3446, 0.3272], device='cuda:6'), in_proj_covar=tensor([0.0421, 0.0471, 0.0385, 0.0339, 0.0446, 0.0538, 0.0441, 0.0551], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:08:38,567 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.275e+02 2.633e+02 3.156e+02 7.653e+02, threshold=5.267e+02, percent-clipped=3.0 2023-05-02 04:08:38,583 INFO [train.py:904] (6/8) Epoch 26, batch 3300, loss[loss=0.1625, simple_loss=0.267, pruned_loss=0.02903, over 17275.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2537, pruned_loss=0.03923, over 3332677.68 frames. ], batch size: 52, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:09:07,453 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257074.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:09:09,566 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257076.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:09:48,082 INFO [train.py:904] (6/8) Epoch 26, batch 3350, loss[loss=0.1628, simple_loss=0.2566, pruned_loss=0.03452, over 17299.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2543, pruned_loss=0.0393, over 3327190.76 frames. ], batch size: 52, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:10:20,803 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257127.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:10:27,497 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-02 04:10:46,891 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1783, 3.3345, 3.5277, 2.1451, 2.9844, 2.4033, 3.6634, 3.6608], device='cuda:6'), covar=tensor([0.0275, 0.0902, 0.0643, 0.2034, 0.0855, 0.1047, 0.0537, 0.0844], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0171, 0.0171, 0.0157, 0.0148, 0.0133, 0.0147, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 04:10:56,576 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 2.179e+02 2.494e+02 3.258e+02 7.595e+02, threshold=4.988e+02, percent-clipped=3.0 2023-05-02 04:10:56,591 INFO [train.py:904] (6/8) Epoch 26, batch 3400, loss[loss=0.1676, simple_loss=0.2516, pruned_loss=0.04183, over 16258.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2541, pruned_loss=0.03945, over 3322986.22 frames. ], batch size: 165, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:11:13,568 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257165.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:11:27,560 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257175.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:11:39,283 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257183.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 04:12:06,425 INFO [train.py:904] (6/8) Epoch 26, batch 3450, loss[loss=0.1731, simple_loss=0.2554, pruned_loss=0.04536, over 15528.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2523, pruned_loss=0.03885, over 3314892.87 frames. ], batch size: 191, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:12:22,267 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8570, 1.4510, 1.7391, 1.7312, 1.7520, 1.9912, 1.6281, 1.8578], device='cuda:6'), covar=tensor([0.0267, 0.0463, 0.0244, 0.0317, 0.0306, 0.0226, 0.0467, 0.0179], device='cuda:6'), in_proj_covar=tensor([0.0197, 0.0198, 0.0188, 0.0191, 0.0206, 0.0165, 0.0202, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:12:32,280 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4415, 5.3910, 5.1681, 4.6692, 5.2616, 2.0195, 4.9632, 5.1018], device='cuda:6'), covar=tensor([0.0084, 0.0082, 0.0234, 0.0378, 0.0092, 0.2763, 0.0156, 0.0199], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0174, 0.0212, 0.0188, 0.0189, 0.0219, 0.0202, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:12:34,001 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257222.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:12:39,853 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257226.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:12:57,463 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257239.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 04:12:59,880 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257241.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:13:16,457 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 2.059e+02 2.361e+02 2.793e+02 7.239e+02, threshold=4.722e+02, percent-clipped=1.0 2023-05-02 04:13:16,472 INFO [train.py:904] (6/8) Epoch 26, batch 3500, loss[loss=0.1953, simple_loss=0.2741, pruned_loss=0.05828, over 11936.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2524, pruned_loss=0.03938, over 3311269.44 frames. ], batch size: 246, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:13:35,743 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4479, 2.4045, 2.4648, 4.2233, 2.3589, 2.7944, 2.4794, 2.5327], device='cuda:6'), covar=tensor([0.1319, 0.3697, 0.3090, 0.0543, 0.4117, 0.2616, 0.3603, 0.3722], device='cuda:6'), in_proj_covar=tensor([0.0422, 0.0471, 0.0386, 0.0340, 0.0448, 0.0539, 0.0442, 0.0552], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:13:58,900 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257283.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:14:25,636 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257302.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:14:26,372 INFO [train.py:904] (6/8) Epoch 26, batch 3550, loss[loss=0.1451, simple_loss=0.2294, pruned_loss=0.03039, over 16768.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2511, pruned_loss=0.0392, over 3304882.86 frames. ], batch size: 83, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:15:13,214 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6227, 6.0331, 5.7922, 5.8380, 5.3908, 5.4756, 5.3664, 6.2230], device='cuda:6'), covar=tensor([0.1525, 0.1051, 0.1073, 0.0837, 0.0956, 0.0620, 0.1417, 0.0849], device='cuda:6'), in_proj_covar=tensor([0.0725, 0.0883, 0.0720, 0.0681, 0.0559, 0.0558, 0.0739, 0.0687], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:15:34,858 INFO [train.py:904] (6/8) Epoch 26, batch 3600, loss[loss=0.1562, simple_loss=0.2356, pruned_loss=0.03842, over 16586.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2496, pruned_loss=0.03882, over 3301369.47 frames. ], batch size: 75, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:15:35,974 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.460e+02 2.009e+02 2.326e+02 2.677e+02 5.900e+02, threshold=4.651e+02, percent-clipped=2.0 2023-05-02 04:15:58,537 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257369.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:16:47,824 INFO [train.py:904] (6/8) Epoch 26, batch 3650, loss[loss=0.1717, simple_loss=0.2452, pruned_loss=0.04913, over 16893.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2482, pruned_loss=0.03893, over 3303082.88 frames. ], batch size: 109, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:17:58,064 INFO [train.py:904] (6/8) Epoch 26, batch 3700, loss[loss=0.1451, simple_loss=0.2241, pruned_loss=0.03301, over 16301.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2469, pruned_loss=0.0401, over 3278214.56 frames. ], batch size: 68, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:17:59,893 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.445e+02 2.076e+02 2.563e+02 3.144e+02 6.860e+02, threshold=5.126e+02, percent-clipped=4.0 2023-05-02 04:18:19,099 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257467.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:18:41,362 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257483.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:19:09,129 INFO [train.py:904] (6/8) Epoch 26, batch 3750, loss[loss=0.1694, simple_loss=0.2479, pruned_loss=0.04551, over 16460.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2483, pruned_loss=0.04173, over 3263271.31 frames. ], batch size: 75, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:19:34,001 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257521.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:19:44,860 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257528.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 04:19:49,559 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257531.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:20:01,895 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257539.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 04:20:20,023 INFO [train.py:904] (6/8) Epoch 26, batch 3800, loss[loss=0.1889, simple_loss=0.2621, pruned_loss=0.05786, over 16909.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2497, pruned_loss=0.043, over 3259144.93 frames. ], batch size: 116, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:20:22,164 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.842e+02 2.272e+02 2.465e+02 2.848e+02 4.858e+02, threshold=4.931e+02, percent-clipped=0.0 2023-05-02 04:20:55,728 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257578.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:20:55,834 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257578.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:21:09,654 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257587.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:21:22,647 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257597.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:21:30,542 INFO [train.py:904] (6/8) Epoch 26, batch 3850, loss[loss=0.1612, simple_loss=0.2494, pruned_loss=0.03648, over 15525.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2498, pruned_loss=0.04344, over 3257372.85 frames. ], batch size: 190, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:22:15,681 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6970, 4.8368, 4.9761, 4.7771, 4.8202, 5.4078, 4.9395, 4.5810], device='cuda:6'), covar=tensor([0.1413, 0.1843, 0.2105, 0.1975, 0.2553, 0.0968, 0.1470, 0.2205], device='cuda:6'), in_proj_covar=tensor([0.0433, 0.0638, 0.0704, 0.0522, 0.0692, 0.0728, 0.0546, 0.0698], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 04:22:20,347 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257639.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 04:22:41,228 INFO [train.py:904] (6/8) Epoch 26, batch 3900, loss[loss=0.1613, simple_loss=0.2352, pruned_loss=0.04369, over 16839.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2496, pruned_loss=0.0438, over 3257481.34 frames. ], batch size: 83, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:22:42,469 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.694e+02 2.236e+02 2.511e+02 3.000e+02 4.952e+02, threshold=5.021e+02, percent-clipped=1.0 2023-05-02 04:23:04,676 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257669.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:23:51,421 INFO [train.py:904] (6/8) Epoch 26, batch 3950, loss[loss=0.1539, simple_loss=0.2312, pruned_loss=0.03832, over 16810.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.25, pruned_loss=0.04476, over 3253046.87 frames. ], batch size: 102, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:23:59,700 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3197, 2.4993, 2.4211, 4.2305, 2.3319, 2.8402, 2.5530, 2.6570], device='cuda:6'), covar=tensor([0.1318, 0.3125, 0.2819, 0.0463, 0.3711, 0.2261, 0.3066, 0.2905], device='cuda:6'), in_proj_covar=tensor([0.0421, 0.0471, 0.0385, 0.0339, 0.0446, 0.0538, 0.0442, 0.0551], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:24:12,553 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257717.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:25:02,911 INFO [train.py:904] (6/8) Epoch 26, batch 4000, loss[loss=0.1645, simple_loss=0.2443, pruned_loss=0.04232, over 16671.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2502, pruned_loss=0.04517, over 3260830.06 frames. ], batch size: 89, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:25:03,988 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.101e+02 2.446e+02 2.886e+02 5.630e+02, threshold=4.893e+02, percent-clipped=0.0 2023-05-02 04:25:11,603 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4135, 2.5676, 2.4229, 4.3589, 2.4652, 2.9612, 2.6139, 2.7151], device='cuda:6'), covar=tensor([0.1351, 0.3098, 0.2815, 0.0452, 0.3536, 0.2131, 0.3122, 0.2897], device='cuda:6'), in_proj_covar=tensor([0.0422, 0.0473, 0.0386, 0.0340, 0.0448, 0.0541, 0.0443, 0.0554], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:26:13,222 INFO [train.py:904] (6/8) Epoch 26, batch 4050, loss[loss=0.1763, simple_loss=0.2629, pruned_loss=0.04486, over 16327.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2515, pruned_loss=0.04478, over 3262906.44 frames. ], batch size: 165, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:26:40,087 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257821.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:26:43,250 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257823.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 04:26:52,658 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5868, 5.5596, 5.3924, 4.7793, 5.5767, 2.2315, 5.2847, 4.9970], device='cuda:6'), covar=tensor([0.0061, 0.0054, 0.0165, 0.0330, 0.0054, 0.2768, 0.0110, 0.0240], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0173, 0.0211, 0.0186, 0.0188, 0.0217, 0.0201, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:27:02,829 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1743, 3.2103, 2.0164, 3.4343, 2.4281, 3.5383, 2.2031, 2.6546], device='cuda:6'), covar=tensor([0.0292, 0.0392, 0.1563, 0.0159, 0.0837, 0.0337, 0.1461, 0.0762], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0182, 0.0198, 0.0175, 0.0180, 0.0222, 0.0205, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 04:27:25,364 INFO [train.py:904] (6/8) Epoch 26, batch 4100, loss[loss=0.1897, simple_loss=0.2886, pruned_loss=0.04542, over 16764.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2542, pruned_loss=0.04487, over 3261578.56 frames. ], batch size: 124, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:27:26,539 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.240e+02 1.750e+02 2.045e+02 2.393e+02 4.321e+02, threshold=4.090e+02, percent-clipped=1.0 2023-05-02 04:27:35,119 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257859.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:27:49,182 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257869.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:28:02,720 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257878.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:28:10,499 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257883.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:28:30,715 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257897.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:28:32,576 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2514, 2.9331, 3.1738, 1.7430, 3.3486, 3.4013, 2.7686, 2.6163], device='cuda:6'), covar=tensor([0.0916, 0.0349, 0.0289, 0.1362, 0.0127, 0.0225, 0.0494, 0.0569], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0112, 0.0100, 0.0140, 0.0086, 0.0131, 0.0130, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 04:28:39,711 INFO [train.py:904] (6/8) Epoch 26, batch 4150, loss[loss=0.1982, simple_loss=0.2874, pruned_loss=0.05449, over 16778.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2612, pruned_loss=0.04694, over 3231361.54 frames. ], batch size: 124, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:29:07,216 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257920.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:29:11,786 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4151, 4.6814, 4.5041, 4.5362, 4.2119, 4.1599, 4.1973, 4.7178], device='cuda:6'), covar=tensor([0.1148, 0.0871, 0.0958, 0.0803, 0.0799, 0.1507, 0.1094, 0.0904], device='cuda:6'), in_proj_covar=tensor([0.0714, 0.0868, 0.0708, 0.0670, 0.0551, 0.0551, 0.0730, 0.0678], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:29:15,911 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257926.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:29:28,855 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257934.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 04:29:44,248 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257944.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 04:29:45,355 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257945.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:29:53,544 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257950.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:29:56,804 INFO [train.py:904] (6/8) Epoch 26, batch 4200, loss[loss=0.183, simple_loss=0.2838, pruned_loss=0.04117, over 17126.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2678, pruned_loss=0.04821, over 3208580.18 frames. ], batch size: 48, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:29:58,474 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.624e+02 2.083e+02 2.573e+02 2.968e+02 5.903e+02, threshold=5.147e+02, percent-clipped=5.0 2023-05-02 04:30:51,125 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1298, 2.3001, 2.2672, 3.8430, 2.1784, 2.5666, 2.3025, 2.3565], device='cuda:6'), covar=tensor([0.1449, 0.3812, 0.3214, 0.0580, 0.4355, 0.2619, 0.3932, 0.3660], device='cuda:6'), in_proj_covar=tensor([0.0417, 0.0467, 0.0381, 0.0335, 0.0442, 0.0535, 0.0438, 0.0546], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:31:15,317 INFO [train.py:904] (6/8) Epoch 26, batch 4250, loss[loss=0.179, simple_loss=0.2739, pruned_loss=0.04207, over 15467.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.2705, pruned_loss=0.04731, over 3209163.96 frames. ], batch size: 190, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:31:19,167 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7178, 3.5996, 3.9029, 2.0974, 4.1615, 4.2074, 3.2416, 3.1664], device='cuda:6'), covar=tensor([0.0815, 0.0253, 0.0250, 0.1209, 0.0085, 0.0137, 0.0394, 0.0459], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0112, 0.0100, 0.0139, 0.0086, 0.0130, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 04:31:29,223 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=258011.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:32:01,979 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-02 04:32:29,606 INFO [train.py:904] (6/8) Epoch 26, batch 4300, loss[loss=0.1822, simple_loss=0.2799, pruned_loss=0.0423, over 16266.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2714, pruned_loss=0.04645, over 3207650.66 frames. ], batch size: 165, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:32:31,420 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.134e+02 2.518e+02 2.964e+02 4.474e+02, threshold=5.035e+02, percent-clipped=0.0 2023-05-02 04:32:49,568 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8698, 3.2397, 3.1865, 2.0061, 3.0269, 3.2834, 3.0354, 1.9720], device='cuda:6'), covar=tensor([0.0614, 0.0064, 0.0076, 0.0499, 0.0121, 0.0111, 0.0125, 0.0483], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0088, 0.0089, 0.0136, 0.0101, 0.0113, 0.0098, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 04:33:45,781 INFO [train.py:904] (6/8) Epoch 26, batch 4350, loss[loss=0.1825, simple_loss=0.2742, pruned_loss=0.04542, over 16595.00 frames. ], tot_loss[loss=0.1848, simple_loss=0.2745, pruned_loss=0.04758, over 3178580.19 frames. ], batch size: 57, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:34:15,477 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=258123.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:34:58,989 INFO [train.py:904] (6/8) Epoch 26, batch 4400, loss[loss=0.1955, simple_loss=0.2862, pruned_loss=0.0524, over 15281.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2768, pruned_loss=0.04881, over 3177780.97 frames. ], batch size: 190, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:35:00,099 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.305e+02 2.313e+02 2.752e+02 3.228e+02 5.813e+02, threshold=5.503e+02, percent-clipped=1.0 2023-05-02 04:35:25,259 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=258171.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:35:29,521 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0247, 5.2956, 5.0962, 5.1179, 4.8515, 4.6809, 4.6989, 5.4187], device='cuda:6'), covar=tensor([0.1214, 0.0812, 0.1002, 0.0821, 0.0755, 0.1025, 0.1181, 0.0790], device='cuda:6'), in_proj_covar=tensor([0.0707, 0.0857, 0.0700, 0.0662, 0.0546, 0.0545, 0.0720, 0.0669], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:35:45,094 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4929, 4.4737, 4.3472, 3.5854, 4.4055, 1.6186, 4.1501, 3.7700], device='cuda:6'), covar=tensor([0.0071, 0.0062, 0.0163, 0.0299, 0.0060, 0.3377, 0.0099, 0.0330], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0172, 0.0210, 0.0185, 0.0187, 0.0216, 0.0200, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:36:03,219 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5960, 4.7733, 4.9333, 4.6654, 4.8035, 5.3374, 4.8018, 4.4525], device='cuda:6'), covar=tensor([0.1202, 0.1784, 0.2168, 0.1976, 0.2319, 0.0877, 0.1300, 0.2280], device='cuda:6'), in_proj_covar=tensor([0.0425, 0.0625, 0.0689, 0.0510, 0.0675, 0.0713, 0.0534, 0.0680], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 04:36:11,042 INFO [train.py:904] (6/8) Epoch 26, batch 4450, loss[loss=0.1902, simple_loss=0.2825, pruned_loss=0.04896, over 16379.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2802, pruned_loss=0.05008, over 3193636.51 frames. ], batch size: 35, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:36:19,343 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-02 04:36:28,139 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=258215.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:36:55,325 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=258234.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 04:37:03,959 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=258239.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 04:37:05,348 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4564, 5.4464, 5.1829, 4.5764, 5.4233, 2.0080, 5.0985, 4.6651], device='cuda:6'), covar=tensor([0.0050, 0.0043, 0.0134, 0.0289, 0.0045, 0.2883, 0.0080, 0.0243], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0171, 0.0209, 0.0185, 0.0186, 0.0216, 0.0199, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:37:22,667 INFO [train.py:904] (6/8) Epoch 26, batch 4500, loss[loss=0.1984, simple_loss=0.2825, pruned_loss=0.05721, over 16260.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.281, pruned_loss=0.05075, over 3210348.51 frames. ], batch size: 35, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:37:23,848 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.439e+02 1.900e+02 2.188e+02 2.602e+02 4.588e+02, threshold=4.376e+02, percent-clipped=0.0 2023-05-02 04:37:24,701 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-02 04:37:59,751 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.03 vs. limit=5.0 2023-05-02 04:38:05,642 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=258282.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:38:35,269 INFO [train.py:904] (6/8) Epoch 26, batch 4550, loss[loss=0.2075, simple_loss=0.3017, pruned_loss=0.05663, over 16513.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2814, pruned_loss=0.05151, over 3218695.32 frames. ], batch size: 68, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:38:39,768 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=258306.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:39:48,282 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-02 04:39:48,615 INFO [train.py:904] (6/8) Epoch 26, batch 4600, loss[loss=0.1876, simple_loss=0.2712, pruned_loss=0.05203, over 16610.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.2826, pruned_loss=0.05236, over 3207863.89 frames. ], batch size: 57, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:39:50,255 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.394e+02 1.782e+02 2.088e+02 2.418e+02 3.840e+02, threshold=4.176e+02, percent-clipped=0.0 2023-05-02 04:39:51,513 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3008, 2.5172, 2.4135, 3.9877, 2.3152, 2.8967, 2.5296, 2.5778], device='cuda:6'), covar=tensor([0.1396, 0.3174, 0.2844, 0.0552, 0.3934, 0.2198, 0.3138, 0.3238], device='cuda:6'), in_proj_covar=tensor([0.0419, 0.0469, 0.0382, 0.0336, 0.0445, 0.0537, 0.0440, 0.0548], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:41:03,103 INFO [train.py:904] (6/8) Epoch 26, batch 4650, loss[loss=0.1684, simple_loss=0.2632, pruned_loss=0.03682, over 16881.00 frames. ], tot_loss[loss=0.1934, simple_loss=0.2816, pruned_loss=0.05263, over 3189946.47 frames. ], batch size: 90, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:41:39,277 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0307, 5.4897, 5.7111, 5.3788, 5.5185, 6.0598, 5.4870, 5.1751], device='cuda:6'), covar=tensor([0.0930, 0.1701, 0.2089, 0.1896, 0.2276, 0.0826, 0.1417, 0.2425], device='cuda:6'), in_proj_covar=tensor([0.0424, 0.0624, 0.0685, 0.0509, 0.0674, 0.0714, 0.0532, 0.0679], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 04:42:14,211 INFO [train.py:904] (6/8) Epoch 26, batch 4700, loss[loss=0.1548, simple_loss=0.2516, pruned_loss=0.02897, over 16778.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2784, pruned_loss=0.05097, over 3202001.93 frames. ], batch size: 83, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:42:16,007 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 1.737e+02 2.039e+02 2.431e+02 7.214e+02, threshold=4.078e+02, percent-clipped=1.0 2023-05-02 04:42:17,172 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=258454.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:42:39,373 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6689, 2.6937, 2.2715, 2.6294, 3.0179, 2.6373, 3.0042, 3.1871], device='cuda:6'), covar=tensor([0.0118, 0.0491, 0.0639, 0.0501, 0.0320, 0.0529, 0.0282, 0.0365], device='cuda:6'), in_proj_covar=tensor([0.0230, 0.0245, 0.0234, 0.0235, 0.0245, 0.0245, 0.0244, 0.0245], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:43:06,452 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6658, 4.8530, 4.5641, 4.2431, 3.8621, 4.7728, 4.6228, 4.3711], device='cuda:6'), covar=tensor([0.0808, 0.0628, 0.0458, 0.0453, 0.1887, 0.0556, 0.0509, 0.0728], device='cuda:6'), in_proj_covar=tensor([0.0306, 0.0457, 0.0358, 0.0361, 0.0359, 0.0414, 0.0244, 0.0431], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:43:26,799 INFO [train.py:904] (6/8) Epoch 26, batch 4750, loss[loss=0.168, simple_loss=0.2514, pruned_loss=0.04235, over 16868.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2747, pruned_loss=0.04938, over 3193653.32 frames. ], batch size: 109, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:43:39,675 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=258511.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:43:45,059 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=258515.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:43:45,097 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=258515.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:44:21,162 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=258539.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:44:41,653 INFO [train.py:904] (6/8) Epoch 26, batch 4800, loss[loss=0.1606, simple_loss=0.2614, pruned_loss=0.02995, over 16709.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2714, pruned_loss=0.0476, over 3191909.15 frames. ], batch size: 89, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:44:43,327 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.278e+02 1.748e+02 2.158e+02 2.549e+02 7.061e+02, threshold=4.316e+02, percent-clipped=3.0 2023-05-02 04:44:56,481 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=258563.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:45:11,429 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=258572.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:45:34,406 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=258587.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:45:44,991 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3930, 3.5025, 3.6435, 3.6193, 3.6293, 3.4776, 3.5117, 3.5086], device='cuda:6'), covar=tensor([0.0353, 0.0650, 0.0444, 0.0440, 0.0477, 0.0479, 0.0742, 0.0508], device='cuda:6'), in_proj_covar=tensor([0.0422, 0.0474, 0.0461, 0.0423, 0.0508, 0.0485, 0.0564, 0.0391], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 04:45:57,755 INFO [train.py:904] (6/8) Epoch 26, batch 4850, loss[loss=0.1811, simple_loss=0.2841, pruned_loss=0.03899, over 15394.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2719, pruned_loss=0.04692, over 3171475.85 frames. ], batch size: 190, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:46:03,358 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=258606.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:47:14,836 INFO [train.py:904] (6/8) Epoch 26, batch 4900, loss[loss=0.174, simple_loss=0.2676, pruned_loss=0.04023, over 16428.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2707, pruned_loss=0.0451, over 3182739.97 frames. ], batch size: 146, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:47:16,712 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.370e+02 1.994e+02 2.195e+02 2.536e+02 4.973e+02, threshold=4.389e+02, percent-clipped=2.0 2023-05-02 04:47:17,118 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=258654.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:48:29,883 INFO [train.py:904] (6/8) Epoch 26, batch 4950, loss[loss=0.1836, simple_loss=0.2813, pruned_loss=0.04299, over 16768.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2697, pruned_loss=0.04436, over 3182001.41 frames. ], batch size: 83, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:49:41,062 INFO [train.py:904] (6/8) Epoch 26, batch 5000, loss[loss=0.1896, simple_loss=0.2739, pruned_loss=0.05266, over 16455.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2713, pruned_loss=0.04444, over 3184442.25 frames. ], batch size: 68, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:49:42,188 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.044e+02 2.396e+02 2.809e+02 5.347e+02, threshold=4.792e+02, percent-clipped=3.0 2023-05-02 04:50:54,488 INFO [train.py:904] (6/8) Epoch 26, batch 5050, loss[loss=0.2099, simple_loss=0.3107, pruned_loss=0.05455, over 16915.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2722, pruned_loss=0.04454, over 3200313.22 frames. ], batch size: 109, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:51:05,555 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=258810.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:52:01,359 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9548, 2.7332, 2.8793, 2.0945, 2.6796, 2.1347, 2.7570, 2.8721], device='cuda:6'), covar=tensor([0.0278, 0.0776, 0.0558, 0.1777, 0.0851, 0.0871, 0.0620, 0.0750], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0170, 0.0171, 0.0157, 0.0148, 0.0132, 0.0147, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 04:52:07,189 INFO [train.py:904] (6/8) Epoch 26, batch 5100, loss[loss=0.1496, simple_loss=0.2494, pruned_loss=0.02487, over 16837.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2703, pruned_loss=0.04365, over 3199968.58 frames. ], batch size: 96, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:52:08,929 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 1.999e+02 2.336e+02 2.714e+02 4.372e+02, threshold=4.672e+02, percent-clipped=0.0 2023-05-02 04:52:27,817 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=258867.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:53:21,036 INFO [train.py:904] (6/8) Epoch 26, batch 5150, loss[loss=0.1937, simple_loss=0.2875, pruned_loss=0.04991, over 15449.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2703, pruned_loss=0.04304, over 3193791.83 frames. ], batch size: 190, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:53:43,868 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9565, 2.1208, 2.4640, 2.9110, 2.7888, 3.2695, 2.1474, 3.2811], device='cuda:6'), covar=tensor([0.0219, 0.0556, 0.0364, 0.0338, 0.0348, 0.0194, 0.0611, 0.0176], device='cuda:6'), in_proj_covar=tensor([0.0197, 0.0199, 0.0187, 0.0192, 0.0207, 0.0164, 0.0204, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 04:54:35,102 INFO [train.py:904] (6/8) Epoch 26, batch 5200, loss[loss=0.1467, simple_loss=0.2314, pruned_loss=0.03101, over 17085.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2683, pruned_loss=0.04202, over 3205580.46 frames. ], batch size: 55, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:54:36,751 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 1.846e+02 2.126e+02 2.536e+02 3.966e+02, threshold=4.251e+02, percent-clipped=0.0 2023-05-02 04:54:53,595 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=258965.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:55:48,686 INFO [train.py:904] (6/8) Epoch 26, batch 5250, loss[loss=0.1668, simple_loss=0.25, pruned_loss=0.04178, over 17013.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2661, pruned_loss=0.04207, over 3200143.03 frames. ], batch size: 55, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:56:23,070 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259026.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:56:46,188 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.19 vs. limit=5.0 2023-05-02 04:57:03,176 INFO [train.py:904] (6/8) Epoch 26, batch 5300, loss[loss=0.154, simple_loss=0.245, pruned_loss=0.03144, over 16864.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2623, pruned_loss=0.04041, over 3215099.48 frames. ], batch size: 96, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:57:04,410 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 1.887e+02 2.180e+02 2.580e+02 4.921e+02, threshold=4.360e+02, percent-clipped=4.0 2023-05-02 04:58:18,026 INFO [train.py:904] (6/8) Epoch 26, batch 5350, loss[loss=0.1859, simple_loss=0.2905, pruned_loss=0.04069, over 16776.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2614, pruned_loss=0.03992, over 3215684.71 frames. ], batch size: 83, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:58:27,918 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=259110.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:58:34,291 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.93 vs. limit=5.0 2023-05-02 04:59:10,763 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-02 04:59:31,423 INFO [train.py:904] (6/8) Epoch 26, batch 5400, loss[loss=0.1567, simple_loss=0.2576, pruned_loss=0.02787, over 16701.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2639, pruned_loss=0.04091, over 3207612.00 frames. ], batch size: 89, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:59:32,572 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 1.944e+02 2.298e+02 2.661e+02 4.475e+02, threshold=4.595e+02, percent-clipped=1.0 2023-05-02 04:59:38,449 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=259158.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:59:53,858 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=259167.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:00:48,652 INFO [train.py:904] (6/8) Epoch 26, batch 5450, loss[loss=0.2137, simple_loss=0.3022, pruned_loss=0.06265, over 16380.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2669, pruned_loss=0.0421, over 3219639.50 frames. ], batch size: 146, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 05:01:08,445 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=259215.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:01:25,097 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=259226.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:02:05,203 INFO [train.py:904] (6/8) Epoch 26, batch 5500, loss[loss=0.2149, simple_loss=0.3096, pruned_loss=0.0601, over 16736.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2734, pruned_loss=0.04619, over 3183388.01 frames. ], batch size: 83, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 05:02:06,407 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6012, 3.0683, 3.0660, 1.9539, 2.7487, 2.0366, 3.1625, 3.2972], device='cuda:6'), covar=tensor([0.0270, 0.0738, 0.0626, 0.2133, 0.0907, 0.1065, 0.0696, 0.0837], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0170, 0.0172, 0.0157, 0.0149, 0.0132, 0.0147, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 05:02:07,102 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 2.367e+02 2.972e+02 3.909e+02 6.600e+02, threshold=5.944e+02, percent-clipped=13.0 2023-05-02 05:02:23,193 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=259264.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:02:39,723 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 05:02:58,261 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259287.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:03:22,299 INFO [train.py:904] (6/8) Epoch 26, batch 5550, loss[loss=0.2141, simple_loss=0.3009, pruned_loss=0.06368, over 16225.00 frames. ], tot_loss[loss=0.1916, simple_loss=0.281, pruned_loss=0.05112, over 3161566.23 frames. ], batch size: 165, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 05:03:50,599 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=259321.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:03:57,851 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259325.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 05:04:40,412 INFO [train.py:904] (6/8) Epoch 26, batch 5600, loss[loss=0.2796, simple_loss=0.3372, pruned_loss=0.111, over 11256.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2858, pruned_loss=0.05527, over 3119108.71 frames. ], batch size: 246, lr: 2.58e-03, grad_scale: 16.0 2023-05-02 05:04:41,807 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.957e+02 3.012e+02 3.705e+02 4.647e+02 1.087e+03, threshold=7.410e+02, percent-clipped=8.0 2023-05-02 05:05:00,598 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6139, 2.5613, 1.9133, 2.6373, 2.1447, 2.7543, 2.1490, 2.3695], device='cuda:6'), covar=tensor([0.0332, 0.0377, 0.1155, 0.0302, 0.0607, 0.0521, 0.1157, 0.0592], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0181, 0.0197, 0.0170, 0.0179, 0.0219, 0.0205, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 05:06:04,635 INFO [train.py:904] (6/8) Epoch 26, batch 5650, loss[loss=0.1872, simple_loss=0.2779, pruned_loss=0.04822, over 16470.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2911, pruned_loss=0.05983, over 3068093.71 frames. ], batch size: 68, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:06:22,282 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=259414.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:06:47,953 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0977, 2.4044, 2.6233, 1.9306, 2.7349, 2.8011, 2.4729, 2.3758], device='cuda:6'), covar=tensor([0.0720, 0.0293, 0.0238, 0.0993, 0.0143, 0.0300, 0.0477, 0.0485], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0112, 0.0100, 0.0139, 0.0086, 0.0130, 0.0129, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 05:06:59,079 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4287, 4.4196, 4.3034, 3.5400, 4.3591, 1.6874, 4.1230, 3.8392], device='cuda:6'), covar=tensor([0.0095, 0.0089, 0.0186, 0.0341, 0.0089, 0.3061, 0.0142, 0.0322], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0170, 0.0209, 0.0184, 0.0185, 0.0215, 0.0198, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:07:22,187 INFO [train.py:904] (6/8) Epoch 26, batch 5700, loss[loss=0.2609, simple_loss=0.3278, pruned_loss=0.09706, over 11515.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2933, pruned_loss=0.06185, over 3041415.64 frames. ], batch size: 248, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:07:25,090 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 3.137e+02 3.722e+02 4.714e+02 8.155e+02, threshold=7.444e+02, percent-clipped=2.0 2023-05-02 05:07:48,664 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 05:07:56,933 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259475.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:08:39,968 INFO [train.py:904] (6/8) Epoch 26, batch 5750, loss[loss=0.1971, simple_loss=0.2859, pruned_loss=0.05412, over 16199.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2956, pruned_loss=0.06259, over 3043129.49 frames. ], batch size: 165, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:10:02,530 INFO [train.py:904] (6/8) Epoch 26, batch 5800, loss[loss=0.1979, simple_loss=0.2952, pruned_loss=0.05033, over 16678.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.295, pruned_loss=0.06081, over 3056899.49 frames. ], batch size: 134, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:10:05,674 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.039e+02 2.780e+02 3.397e+02 4.110e+02 5.922e+02, threshold=6.793e+02, percent-clipped=0.0 2023-05-02 05:10:26,987 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0511, 4.0395, 3.9464, 3.1431, 3.9848, 1.8005, 3.7872, 3.4563], device='cuda:6'), covar=tensor([0.0146, 0.0135, 0.0216, 0.0296, 0.0100, 0.3077, 0.0144, 0.0294], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0170, 0.0208, 0.0183, 0.0184, 0.0214, 0.0197, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:10:47,064 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=259582.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:10:50,230 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-05-02 05:11:19,032 INFO [train.py:904] (6/8) Epoch 26, batch 5850, loss[loss=0.203, simple_loss=0.2894, pruned_loss=0.05827, over 17120.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2933, pruned_loss=0.05984, over 3050345.88 frames. ], batch size: 48, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:11:44,962 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=259620.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 05:11:46,908 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=259621.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:12:39,963 INFO [train.py:904] (6/8) Epoch 26, batch 5900, loss[loss=0.2172, simple_loss=0.302, pruned_loss=0.06627, over 16779.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2927, pruned_loss=0.05959, over 3069718.77 frames. ], batch size: 124, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:12:43,692 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.729e+02 2.720e+02 3.451e+02 4.224e+02 7.867e+02, threshold=6.903e+02, percent-clipped=3.0 2023-05-02 05:13:09,229 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=259669.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:13:27,033 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3822, 3.1540, 3.5087, 1.8146, 3.6215, 3.6770, 2.9457, 2.8381], device='cuda:6'), covar=tensor([0.0839, 0.0316, 0.0236, 0.1302, 0.0113, 0.0245, 0.0424, 0.0493], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0111, 0.0100, 0.0139, 0.0086, 0.0130, 0.0129, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 05:14:01,045 INFO [train.py:904] (6/8) Epoch 26, batch 5950, loss[loss=0.2088, simple_loss=0.2975, pruned_loss=0.0601, over 16449.00 frames. ], tot_loss[loss=0.2058, simple_loss=0.2935, pruned_loss=0.05903, over 3070019.16 frames. ], batch size: 146, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:14:06,008 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4269, 3.3831, 3.4446, 3.5169, 3.5651, 3.3112, 3.5346, 3.6081], device='cuda:6'), covar=tensor([0.1248, 0.0911, 0.1029, 0.0615, 0.0691, 0.2589, 0.1171, 0.0882], device='cuda:6'), in_proj_covar=tensor([0.0664, 0.0814, 0.0939, 0.0824, 0.0628, 0.0654, 0.0680, 0.0794], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:14:43,712 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2023-05-02 05:15:12,622 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=259749.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:15:18,049 INFO [train.py:904] (6/8) Epoch 26, batch 6000, loss[loss=0.1724, simple_loss=0.2594, pruned_loss=0.04268, over 16924.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2915, pruned_loss=0.05801, over 3082403.96 frames. ], batch size: 109, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:15:18,050 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 05:15:28,183 INFO [train.py:938] (6/8) Epoch 26, validation: loss=0.1485, simple_loss=0.2607, pruned_loss=0.01818, over 944034.00 frames. 2023-05-02 05:15:28,184 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 05:15:30,552 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.937e+02 2.681e+02 3.578e+02 4.333e+02 8.656e+02, threshold=7.155e+02, percent-clipped=3.0 2023-05-02 05:15:55,081 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=259770.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:16:13,655 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=259782.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:16:25,932 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-02 05:16:46,433 INFO [train.py:904] (6/8) Epoch 26, batch 6050, loss[loss=0.2127, simple_loss=0.2968, pruned_loss=0.06431, over 15420.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2896, pruned_loss=0.057, over 3107837.53 frames. ], batch size: 191, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:16:59,619 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259810.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 05:17:51,090 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259843.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 05:17:57,070 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6752, 1.7191, 1.4562, 1.3777, 1.8220, 1.5214, 1.5680, 1.8770], device='cuda:6'), covar=tensor([0.0261, 0.0412, 0.0560, 0.0484, 0.0304, 0.0373, 0.0228, 0.0305], device='cuda:6'), in_proj_covar=tensor([0.0224, 0.0240, 0.0229, 0.0230, 0.0240, 0.0238, 0.0238, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:18:05,774 INFO [train.py:904] (6/8) Epoch 26, batch 6100, loss[loss=0.1773, simple_loss=0.2729, pruned_loss=0.04082, over 16746.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2895, pruned_loss=0.05607, over 3115842.91 frames. ], batch size: 89, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:18:09,302 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.693e+02 2.757e+02 3.100e+02 3.601e+02 8.920e+02, threshold=6.201e+02, percent-clipped=2.0 2023-05-02 05:18:16,631 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 05:18:27,868 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=259866.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:18:37,239 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0296, 3.0576, 1.8688, 3.2811, 2.3454, 3.3361, 2.1106, 2.5582], device='cuda:6'), covar=tensor([0.0354, 0.0429, 0.1670, 0.0216, 0.0826, 0.0603, 0.1545, 0.0808], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0181, 0.0198, 0.0171, 0.0180, 0.0220, 0.0206, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 05:18:41,958 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6536, 2.6109, 1.8810, 2.7222, 2.1573, 2.8091, 2.1087, 2.3531], device='cuda:6'), covar=tensor([0.0313, 0.0386, 0.1233, 0.0260, 0.0660, 0.0581, 0.1161, 0.0618], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0181, 0.0198, 0.0171, 0.0180, 0.0220, 0.0206, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 05:18:53,003 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=259882.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:19:17,321 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6485, 2.3189, 1.8433, 2.1338, 2.6041, 2.2831, 2.4074, 2.7564], device='cuda:6'), covar=tensor([0.0234, 0.0467, 0.0632, 0.0490, 0.0289, 0.0435, 0.0267, 0.0288], device='cuda:6'), in_proj_covar=tensor([0.0225, 0.0240, 0.0230, 0.0231, 0.0240, 0.0239, 0.0239, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:19:23,804 INFO [train.py:904] (6/8) Epoch 26, batch 6150, loss[loss=0.2189, simple_loss=0.2917, pruned_loss=0.07302, over 11783.00 frames. ], tot_loss[loss=0.2, simple_loss=0.288, pruned_loss=0.05599, over 3104119.75 frames. ], batch size: 248, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:19:49,508 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=259920.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:20:00,760 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259927.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:20:05,846 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=259930.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:20:42,382 INFO [train.py:904] (6/8) Epoch 26, batch 6200, loss[loss=0.1914, simple_loss=0.2884, pruned_loss=0.04719, over 16605.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2863, pruned_loss=0.056, over 3097361.31 frames. ], batch size: 62, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:20:44,647 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.064e+02 2.750e+02 3.295e+02 3.920e+02 8.789e+02, threshold=6.590e+02, percent-clipped=2.0 2023-05-02 05:21:05,648 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=259968.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:21:16,215 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9288, 2.7429, 2.8754, 2.1223, 2.6705, 2.1702, 2.7660, 2.9544], device='cuda:6'), covar=tensor([0.0253, 0.0798, 0.0529, 0.1766, 0.0806, 0.0918, 0.0545, 0.0712], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0169, 0.0171, 0.0156, 0.0148, 0.0132, 0.0146, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 05:21:21,114 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2614, 5.9237, 6.0968, 5.7067, 5.8680, 6.3732, 5.8902, 5.6141], device='cuda:6'), covar=tensor([0.0894, 0.1926, 0.2565, 0.2161, 0.2388, 0.0876, 0.1647, 0.2485], device='cuda:6'), in_proj_covar=tensor([0.0423, 0.0625, 0.0686, 0.0512, 0.0675, 0.0715, 0.0536, 0.0680], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 05:21:37,156 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2186, 2.5270, 2.0616, 2.2830, 2.8589, 2.5622, 2.8111, 3.0479], device='cuda:6'), covar=tensor([0.0188, 0.0507, 0.0619, 0.0527, 0.0291, 0.0426, 0.0247, 0.0312], device='cuda:6'), in_proj_covar=tensor([0.0225, 0.0241, 0.0230, 0.0231, 0.0241, 0.0240, 0.0239, 0.0239], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:22:00,546 INFO [train.py:904] (6/8) Epoch 26, batch 6250, loss[loss=0.2141, simple_loss=0.308, pruned_loss=0.06012, over 16575.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2856, pruned_loss=0.05561, over 3096175.14 frames. ], batch size: 68, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:22:23,634 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8101, 4.7921, 4.6710, 3.9117, 4.7145, 1.9733, 4.4548, 4.2984], device='cuda:6'), covar=tensor([0.0124, 0.0122, 0.0229, 0.0395, 0.0126, 0.2918, 0.0176, 0.0282], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0171, 0.0209, 0.0184, 0.0185, 0.0216, 0.0198, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:22:41,201 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=260030.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:23:15,625 INFO [train.py:904] (6/8) Epoch 26, batch 6300, loss[loss=0.1825, simple_loss=0.2717, pruned_loss=0.04662, over 16491.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2854, pruned_loss=0.0546, over 3117942.11 frames. ], batch size: 68, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:23:19,600 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.731e+02 2.817e+02 3.510e+02 4.108e+02 7.742e+02, threshold=7.020e+02, percent-clipped=4.0 2023-05-02 05:23:36,767 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-02 05:23:44,875 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=260070.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:24:15,048 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9529, 2.7249, 2.6124, 1.9625, 2.5832, 2.7508, 2.5887, 1.9395], device='cuda:6'), covar=tensor([0.0483, 0.0117, 0.0125, 0.0411, 0.0159, 0.0167, 0.0150, 0.0458], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0087, 0.0089, 0.0134, 0.0100, 0.0113, 0.0097, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 05:24:17,954 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=260091.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:24:27,089 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.94 vs. limit=5.0 2023-05-02 05:24:35,601 INFO [train.py:904] (6/8) Epoch 26, batch 6350, loss[loss=0.279, simple_loss=0.3297, pruned_loss=0.1141, over 11225.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2861, pruned_loss=0.05593, over 3095789.85 frames. ], batch size: 246, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:24:39,477 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=260105.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:24:39,595 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4562, 3.4396, 3.4065, 2.6490, 3.3026, 2.1964, 3.1560, 2.6971], device='cuda:6'), covar=tensor([0.0191, 0.0175, 0.0219, 0.0260, 0.0129, 0.2285, 0.0159, 0.0276], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0172, 0.0210, 0.0185, 0.0185, 0.0216, 0.0198, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:24:59,117 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=260118.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:25:23,065 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9105, 2.1682, 2.4682, 3.1353, 2.2449, 2.4190, 2.3542, 2.3232], device='cuda:6'), covar=tensor([0.1533, 0.3583, 0.2491, 0.0752, 0.4200, 0.2390, 0.3358, 0.3157], device='cuda:6'), in_proj_covar=tensor([0.0415, 0.0465, 0.0380, 0.0332, 0.0443, 0.0532, 0.0436, 0.0544], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:25:30,299 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=260138.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:25:52,110 INFO [train.py:904] (6/8) Epoch 26, batch 6400, loss[loss=0.1701, simple_loss=0.2586, pruned_loss=0.04078, over 16533.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2862, pruned_loss=0.05671, over 3092093.54 frames. ], batch size: 68, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:25:54,618 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.816e+02 2.921e+02 3.424e+02 4.125e+02 9.297e+02, threshold=6.848e+02, percent-clipped=3.0 2023-05-02 05:25:56,177 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-02 05:27:08,957 INFO [train.py:904] (6/8) Epoch 26, batch 6450, loss[loss=0.1844, simple_loss=0.2801, pruned_loss=0.04439, over 16874.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2857, pruned_loss=0.05562, over 3089032.82 frames. ], batch size: 83, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:27:35,187 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.12 vs. limit=5.0 2023-05-02 05:27:40,348 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=260222.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:27:43,894 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=260224.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:27:53,550 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-02 05:28:28,517 INFO [train.py:904] (6/8) Epoch 26, batch 6500, loss[loss=0.1913, simple_loss=0.2752, pruned_loss=0.05368, over 16618.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.284, pruned_loss=0.05511, over 3112652.76 frames. ], batch size: 62, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:28:31,534 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.633e+02 2.830e+02 3.427e+02 3.864e+02 5.518e+02, threshold=6.855e+02, percent-clipped=0.0 2023-05-02 05:29:18,541 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.92 vs. limit=5.0 2023-05-02 05:29:20,104 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=260285.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 05:29:49,649 INFO [train.py:904] (6/8) Epoch 26, batch 6550, loss[loss=0.19, simple_loss=0.2881, pruned_loss=0.046, over 15232.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2865, pruned_loss=0.05581, over 3106018.47 frames. ], batch size: 190, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:31:07,557 INFO [train.py:904] (6/8) Epoch 26, batch 6600, loss[loss=0.2262, simple_loss=0.3064, pruned_loss=0.07303, over 16158.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2896, pruned_loss=0.05732, over 3091017.85 frames. ], batch size: 165, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:31:09,946 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.723e+02 2.646e+02 3.224e+02 4.276e+02 9.076e+02, threshold=6.447e+02, percent-clipped=2.0 2023-05-02 05:31:59,515 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=260386.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:32:26,383 INFO [train.py:904] (6/8) Epoch 26, batch 6650, loss[loss=0.175, simple_loss=0.2695, pruned_loss=0.04031, over 16536.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2893, pruned_loss=0.05783, over 3093339.68 frames. ], batch size: 75, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:32:30,384 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=260405.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:33:21,075 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=260438.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:33:43,208 INFO [train.py:904] (6/8) Epoch 26, batch 6700, loss[loss=0.1787, simple_loss=0.2721, pruned_loss=0.04261, over 16909.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.287, pruned_loss=0.05697, over 3098952.38 frames. ], batch size: 90, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:33:43,549 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=260453.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:33:45,912 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.097e+02 2.665e+02 3.203e+02 3.728e+02 7.997e+02, threshold=6.406e+02, percent-clipped=3.0 2023-05-02 05:34:00,860 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6732, 1.7683, 1.5913, 1.4610, 1.9225, 1.5659, 1.5341, 1.8890], device='cuda:6'), covar=tensor([0.0222, 0.0328, 0.0475, 0.0404, 0.0247, 0.0296, 0.0188, 0.0231], device='cuda:6'), in_proj_covar=tensor([0.0224, 0.0240, 0.0230, 0.0231, 0.0241, 0.0240, 0.0239, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:34:35,490 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=260486.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:35:01,131 INFO [train.py:904] (6/8) Epoch 26, batch 6750, loss[loss=0.2267, simple_loss=0.2997, pruned_loss=0.07689, over 11398.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2862, pruned_loss=0.05692, over 3104662.73 frames. ], batch size: 248, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:35:31,738 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=260522.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:36:19,268 INFO [train.py:904] (6/8) Epoch 26, batch 6800, loss[loss=0.1884, simple_loss=0.288, pruned_loss=0.04438, over 16858.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2869, pruned_loss=0.05708, over 3098801.76 frames. ], batch size: 96, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:36:21,473 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.811e+02 2.923e+02 3.478e+02 4.093e+02 6.667e+02, threshold=6.957e+02, percent-clipped=2.0 2023-05-02 05:36:45,311 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=260570.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:37:01,682 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=260580.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:37:24,302 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-05-02 05:37:30,468 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.95 vs. limit=5.0 2023-05-02 05:37:35,979 INFO [train.py:904] (6/8) Epoch 26, batch 6850, loss[loss=0.2212, simple_loss=0.3055, pruned_loss=0.06846, over 15433.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.289, pruned_loss=0.05833, over 3085382.50 frames. ], batch size: 190, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:38:50,099 INFO [train.py:904] (6/8) Epoch 26, batch 6900, loss[loss=0.219, simple_loss=0.305, pruned_loss=0.06652, over 16680.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2915, pruned_loss=0.05813, over 3087868.91 frames. ], batch size: 124, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:38:53,860 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.801e+02 2.543e+02 3.098e+02 3.712e+02 7.299e+02, threshold=6.197e+02, percent-clipped=1.0 2023-05-02 05:39:27,150 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-05-02 05:39:40,138 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=260686.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:40:05,851 INFO [train.py:904] (6/8) Epoch 26, batch 6950, loss[loss=0.1923, simple_loss=0.2798, pruned_loss=0.05236, over 16410.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2922, pruned_loss=0.05927, over 3082023.51 frames. ], batch size: 146, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:40:21,672 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.10 vs. limit=5.0 2023-05-02 05:40:48,836 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=260730.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:40:54,445 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=260734.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:41:21,342 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=260751.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:41:23,880 INFO [train.py:904] (6/8) Epoch 26, batch 7000, loss[loss=0.2069, simple_loss=0.3054, pruned_loss=0.05423, over 16728.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2921, pruned_loss=0.05834, over 3078384.25 frames. ], batch size: 57, lr: 2.58e-03, grad_scale: 2.0 2023-05-02 05:41:29,416 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.929e+02 2.843e+02 3.539e+02 4.501e+02 8.180e+02, threshold=7.079e+02, percent-clipped=4.0 2023-05-02 05:42:24,290 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=260791.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:42:42,012 INFO [train.py:904] (6/8) Epoch 26, batch 7050, loss[loss=0.2177, simple_loss=0.3135, pruned_loss=0.06093, over 16719.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2921, pruned_loss=0.05743, over 3087169.00 frames. ], batch size: 57, lr: 2.58e-03, grad_scale: 2.0 2023-05-02 05:42:56,396 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=260812.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:42:57,750 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 05:43:33,100 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6889, 1.8196, 1.6216, 1.4850, 1.9243, 1.5877, 1.6144, 1.8880], device='cuda:6'), covar=tensor([0.0206, 0.0315, 0.0453, 0.0370, 0.0257, 0.0287, 0.0191, 0.0215], device='cuda:6'), in_proj_covar=tensor([0.0223, 0.0239, 0.0230, 0.0230, 0.0240, 0.0238, 0.0238, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:43:34,880 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1318, 5.1342, 4.9483, 4.2797, 5.0589, 1.7959, 4.7867, 4.6462], device='cuda:6'), covar=tensor([0.0101, 0.0102, 0.0221, 0.0431, 0.0102, 0.3074, 0.0139, 0.0266], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0170, 0.0209, 0.0184, 0.0184, 0.0216, 0.0197, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:43:59,436 INFO [train.py:904] (6/8) Epoch 26, batch 7100, loss[loss=0.1719, simple_loss=0.2693, pruned_loss=0.0372, over 16429.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2899, pruned_loss=0.05668, over 3087599.02 frames. ], batch size: 75, lr: 2.58e-03, grad_scale: 2.0 2023-05-02 05:44:05,368 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.003e+02 2.616e+02 2.972e+02 3.634e+02 7.546e+02, threshold=5.943e+02, percent-clipped=1.0 2023-05-02 05:44:21,999 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 05:44:24,833 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-02 05:44:43,253 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=260880.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 05:44:44,471 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=260881.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:45:17,429 INFO [train.py:904] (6/8) Epoch 26, batch 7150, loss[loss=0.241, simple_loss=0.3043, pruned_loss=0.08889, over 11244.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2888, pruned_loss=0.057, over 3085130.66 frames. ], batch size: 246, lr: 2.58e-03, grad_scale: 2.0 2023-05-02 05:45:18,161 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 05:45:53,512 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=260928.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 05:46:02,244 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4511, 4.6363, 4.7852, 4.5279, 4.6039, 5.1540, 4.5948, 4.3046], device='cuda:6'), covar=tensor([0.1412, 0.1816, 0.2180, 0.2079, 0.2366, 0.0973, 0.1688, 0.2447], device='cuda:6'), in_proj_covar=tensor([0.0424, 0.0631, 0.0692, 0.0514, 0.0681, 0.0721, 0.0541, 0.0688], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 05:46:13,620 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=260942.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:46:29,365 INFO [train.py:904] (6/8) Epoch 26, batch 7200, loss[loss=0.1671, simple_loss=0.266, pruned_loss=0.03412, over 16744.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2865, pruned_loss=0.05579, over 3077419.04 frames. ], batch size: 89, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:46:35,554 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 2.782e+02 3.336e+02 4.472e+02 6.403e+02, threshold=6.673e+02, percent-clipped=2.0 2023-05-02 05:47:18,896 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6182, 1.8009, 1.6526, 1.4338, 1.9159, 1.5398, 1.5481, 1.8953], device='cuda:6'), covar=tensor([0.0182, 0.0291, 0.0416, 0.0363, 0.0223, 0.0270, 0.0155, 0.0206], device='cuda:6'), in_proj_covar=tensor([0.0222, 0.0238, 0.0229, 0.0230, 0.0240, 0.0237, 0.0237, 0.0237], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:47:21,154 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2901, 4.5384, 4.8301, 4.7518, 4.7707, 4.4897, 4.2613, 4.3683], device='cuda:6'), covar=tensor([0.0587, 0.0669, 0.0519, 0.0665, 0.0682, 0.0715, 0.1528, 0.0653], device='cuda:6'), in_proj_covar=tensor([0.0421, 0.0477, 0.0462, 0.0425, 0.0511, 0.0486, 0.0564, 0.0391], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 05:47:53,037 INFO [train.py:904] (6/8) Epoch 26, batch 7250, loss[loss=0.1679, simple_loss=0.2576, pruned_loss=0.03908, over 16384.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2837, pruned_loss=0.05421, over 3094084.29 frames. ], batch size: 146, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:49:09,252 INFO [train.py:904] (6/8) Epoch 26, batch 7300, loss[loss=0.2104, simple_loss=0.3089, pruned_loss=0.05595, over 16876.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2837, pruned_loss=0.05456, over 3073426.73 frames. ], batch size: 102, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:49:15,973 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 2.610e+02 3.123e+02 3.831e+02 8.496e+02, threshold=6.245e+02, percent-clipped=1.0 2023-05-02 05:50:00,936 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=261086.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:50:10,987 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7657, 4.3536, 4.1859, 2.9119, 3.7310, 4.2811, 3.7059, 2.4500], device='cuda:6'), covar=tensor([0.0495, 0.0045, 0.0052, 0.0394, 0.0101, 0.0126, 0.0109, 0.0464], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0087, 0.0088, 0.0133, 0.0100, 0.0112, 0.0097, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-02 05:50:24,125 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-05-02 05:50:26,307 INFO [train.py:904] (6/8) Epoch 26, batch 7350, loss[loss=0.2044, simple_loss=0.2919, pruned_loss=0.05843, over 15232.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2851, pruned_loss=0.05565, over 3068889.36 frames. ], batch size: 190, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:50:33,366 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=261107.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:51:18,023 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=261136.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:51:44,987 INFO [train.py:904] (6/8) Epoch 26, batch 7400, loss[loss=0.2298, simple_loss=0.2963, pruned_loss=0.08162, over 11430.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2864, pruned_loss=0.05619, over 3078409.51 frames. ], batch size: 249, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:51:50,752 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.938e+02 2.670e+02 3.232e+02 3.696e+02 7.635e+02, threshold=6.463e+02, percent-clipped=2.0 2023-05-02 05:52:55,467 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=261197.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:53:04,381 INFO [train.py:904] (6/8) Epoch 26, batch 7450, loss[loss=0.2587, simple_loss=0.3123, pruned_loss=0.1025, over 11105.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2871, pruned_loss=0.05676, over 3085087.79 frames. ], batch size: 246, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:53:15,769 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6296, 2.6553, 2.4837, 4.3613, 3.0667, 4.0031, 1.5482, 2.8556], device='cuda:6'), covar=tensor([0.1484, 0.0882, 0.1303, 0.0221, 0.0240, 0.0441, 0.1738, 0.0876], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0180, 0.0200, 0.0198, 0.0209, 0.0218, 0.0209, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 05:53:46,101 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4068, 3.3706, 2.5070, 2.1527, 2.1692, 2.3042, 3.5064, 3.0168], device='cuda:6'), covar=tensor([0.3246, 0.0768, 0.2115, 0.3078, 0.2872, 0.2235, 0.0559, 0.1589], device='cuda:6'), in_proj_covar=tensor([0.0332, 0.0273, 0.0309, 0.0322, 0.0303, 0.0271, 0.0300, 0.0347], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 05:54:01,684 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=261237.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:54:12,918 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9395, 4.9446, 4.7532, 4.0454, 4.8418, 1.7971, 4.5856, 4.3847], device='cuda:6'), covar=tensor([0.0095, 0.0090, 0.0207, 0.0395, 0.0086, 0.2920, 0.0115, 0.0294], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0169, 0.0207, 0.0182, 0.0183, 0.0214, 0.0195, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:54:27,095 INFO [train.py:904] (6/8) Epoch 26, batch 7500, loss[loss=0.1925, simple_loss=0.2718, pruned_loss=0.05654, over 17111.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2873, pruned_loss=0.05623, over 3071773.34 frames. ], batch size: 47, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:54:33,504 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.912e+02 2.815e+02 3.421e+02 3.855e+02 8.023e+02, threshold=6.841e+02, percent-clipped=2.0 2023-05-02 05:55:45,643 INFO [train.py:904] (6/8) Epoch 26, batch 7550, loss[loss=0.1776, simple_loss=0.2662, pruned_loss=0.04454, over 16771.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2867, pruned_loss=0.05626, over 3082011.82 frames. ], batch size: 83, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:57:01,728 INFO [train.py:904] (6/8) Epoch 26, batch 7600, loss[loss=0.2261, simple_loss=0.2928, pruned_loss=0.07973, over 11282.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2859, pruned_loss=0.05672, over 3069372.36 frames. ], batch size: 246, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 05:57:07,520 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.730e+02 2.718e+02 3.314e+02 3.844e+02 7.163e+02, threshold=6.628e+02, percent-clipped=1.0 2023-05-02 05:57:54,565 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=261386.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:58:00,716 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0478, 4.0583, 3.9730, 3.1822, 4.0166, 1.7713, 3.7974, 3.4729], device='cuda:6'), covar=tensor([0.0170, 0.0134, 0.0199, 0.0315, 0.0102, 0.2992, 0.0151, 0.0300], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0168, 0.0206, 0.0182, 0.0182, 0.0213, 0.0195, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 05:58:20,640 INFO [train.py:904] (6/8) Epoch 26, batch 7650, loss[loss=0.206, simple_loss=0.294, pruned_loss=0.05905, over 16660.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2861, pruned_loss=0.05722, over 3075933.91 frames. ], batch size: 134, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 05:58:26,643 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=261407.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:59:08,102 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=261434.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:59:36,566 INFO [train.py:904] (6/8) Epoch 26, batch 7700, loss[loss=0.2031, simple_loss=0.2843, pruned_loss=0.06097, over 16727.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2864, pruned_loss=0.05799, over 3081842.77 frames. ], batch size: 124, lr: 2.57e-03, grad_scale: 2.0 2023-05-02 05:59:40,111 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=261455.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:59:42,099 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=261456.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:59:46,565 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.047e+02 2.911e+02 3.600e+02 4.177e+02 7.928e+02, threshold=7.200e+02, percent-clipped=2.0 2023-05-02 06:00:28,978 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 06:00:36,508 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=261492.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 06:00:52,204 INFO [train.py:904] (6/8) Epoch 26, batch 7750, loss[loss=0.2181, simple_loss=0.2898, pruned_loss=0.07323, over 11209.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2864, pruned_loss=0.05745, over 3082150.43 frames. ], batch size: 246, lr: 2.57e-03, grad_scale: 2.0 2023-05-02 06:01:14,351 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=261517.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:01:32,543 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9518, 4.1977, 4.0433, 4.0631, 3.7494, 3.8126, 3.8808, 4.1920], device='cuda:6'), covar=tensor([0.1210, 0.0952, 0.1010, 0.0894, 0.0840, 0.1725, 0.1024, 0.1083], device='cuda:6'), in_proj_covar=tensor([0.0698, 0.0837, 0.0690, 0.0647, 0.0534, 0.0536, 0.0708, 0.0657], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 06:01:32,714 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0745, 2.3405, 2.3748, 2.7751, 1.9941, 3.2094, 1.8566, 2.7678], device='cuda:6'), covar=tensor([0.1114, 0.0639, 0.1042, 0.0202, 0.0114, 0.0391, 0.1435, 0.0658], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0180, 0.0199, 0.0198, 0.0208, 0.0218, 0.0208, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 06:01:44,966 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=261537.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:01:57,888 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-02 06:02:09,780 INFO [train.py:904] (6/8) Epoch 26, batch 7800, loss[loss=0.2514, simple_loss=0.3223, pruned_loss=0.09025, over 11498.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2874, pruned_loss=0.05787, over 3076724.94 frames. ], batch size: 246, lr: 2.57e-03, grad_scale: 2.0 2023-05-02 06:02:19,355 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.114e+02 2.737e+02 3.262e+02 3.975e+02 6.664e+02, threshold=6.524e+02, percent-clipped=0.0 2023-05-02 06:02:30,640 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7294, 4.5436, 4.7718, 4.8983, 5.0802, 4.5895, 5.0983, 5.0852], device='cuda:6'), covar=tensor([0.2232, 0.1512, 0.1833, 0.0917, 0.0811, 0.1119, 0.0868, 0.0876], device='cuda:6'), in_proj_covar=tensor([0.0652, 0.0806, 0.0928, 0.0815, 0.0624, 0.0648, 0.0677, 0.0790], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 06:02:32,268 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-02 06:02:36,397 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7783, 3.7802, 3.8794, 3.7124, 3.8889, 4.2213, 3.8717, 3.6248], device='cuda:6'), covar=tensor([0.2141, 0.2286, 0.2870, 0.2614, 0.2691, 0.2089, 0.1892, 0.2774], device='cuda:6'), in_proj_covar=tensor([0.0423, 0.0629, 0.0691, 0.0512, 0.0679, 0.0716, 0.0540, 0.0687], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 06:03:00,429 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=261585.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:03:28,317 INFO [train.py:904] (6/8) Epoch 26, batch 7850, loss[loss=0.1989, simple_loss=0.2857, pruned_loss=0.05601, over 15258.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2878, pruned_loss=0.05724, over 3089963.58 frames. ], batch size: 190, lr: 2.57e-03, grad_scale: 2.0 2023-05-02 06:04:46,515 INFO [train.py:904] (6/8) Epoch 26, batch 7900, loss[loss=0.1908, simple_loss=0.2835, pruned_loss=0.04903, over 17125.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2872, pruned_loss=0.05667, over 3101999.20 frames. ], batch size: 47, lr: 2.57e-03, grad_scale: 2.0 2023-05-02 06:04:55,564 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.825e+02 2.587e+02 3.177e+02 3.708e+02 5.885e+02, threshold=6.353e+02, percent-clipped=0.0 2023-05-02 06:05:56,741 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-05-02 06:06:05,684 INFO [train.py:904] (6/8) Epoch 26, batch 7950, loss[loss=0.2294, simple_loss=0.2958, pruned_loss=0.0815, over 11792.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2872, pruned_loss=0.05695, over 3098394.21 frames. ], batch size: 247, lr: 2.57e-03, grad_scale: 2.0 2023-05-02 06:07:23,908 INFO [train.py:904] (6/8) Epoch 26, batch 8000, loss[loss=0.2091, simple_loss=0.2972, pruned_loss=0.06054, over 16725.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2881, pruned_loss=0.05817, over 3065404.03 frames. ], batch size: 124, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:07:29,115 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6939, 3.8420, 2.9293, 2.2789, 2.5412, 2.4730, 4.1928, 3.4411], device='cuda:6'), covar=tensor([0.2999, 0.0664, 0.1840, 0.2920, 0.2960, 0.2201, 0.0421, 0.1365], device='cuda:6'), in_proj_covar=tensor([0.0334, 0.0274, 0.0310, 0.0323, 0.0305, 0.0273, 0.0302, 0.0349], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 06:07:32,670 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.840e+02 2.545e+02 3.130e+02 3.720e+02 6.505e+02, threshold=6.260e+02, percent-clipped=1.0 2023-05-02 06:07:37,864 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5749, 5.9098, 5.6544, 5.6749, 5.2619, 5.2289, 5.3009, 6.0466], device='cuda:6'), covar=tensor([0.1240, 0.0822, 0.1004, 0.0882, 0.0849, 0.0787, 0.1373, 0.0783], device='cuda:6'), in_proj_covar=tensor([0.0699, 0.0839, 0.0691, 0.0647, 0.0536, 0.0536, 0.0707, 0.0657], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 06:08:24,242 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=261792.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 06:08:40,303 INFO [train.py:904] (6/8) Epoch 26, batch 8050, loss[loss=0.1821, simple_loss=0.2759, pruned_loss=0.0441, over 15270.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2879, pruned_loss=0.05742, over 3071737.73 frames. ], batch size: 190, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:08:54,780 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=261812.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:09:12,986 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=261824.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:09:13,827 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-02 06:09:27,635 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=261833.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:09:37,846 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=261840.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 06:09:57,223 INFO [train.py:904] (6/8) Epoch 26, batch 8100, loss[loss=0.1773, simple_loss=0.2683, pruned_loss=0.04317, over 16432.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2873, pruned_loss=0.05666, over 3081255.99 frames. ], batch size: 75, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:10:06,904 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.718e+02 2.493e+02 2.989e+02 3.536e+02 5.601e+02, threshold=5.979e+02, percent-clipped=0.0 2023-05-02 06:10:46,777 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=261885.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:11:00,945 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=261894.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:11:14,754 INFO [train.py:904] (6/8) Epoch 26, batch 8150, loss[loss=0.1671, simple_loss=0.2601, pruned_loss=0.03704, over 16829.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2847, pruned_loss=0.05557, over 3076288.85 frames. ], batch size: 102, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:12:18,386 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4105, 2.9672, 2.6699, 2.2711, 2.2668, 2.3162, 2.9783, 2.8490], device='cuda:6'), covar=tensor([0.2559, 0.0703, 0.1613, 0.2556, 0.2451, 0.2193, 0.0526, 0.1393], device='cuda:6'), in_proj_covar=tensor([0.0334, 0.0274, 0.0310, 0.0323, 0.0305, 0.0273, 0.0302, 0.0348], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 06:12:24,855 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-05-02 06:12:33,065 INFO [train.py:904] (6/8) Epoch 26, batch 8200, loss[loss=0.2376, simple_loss=0.3024, pruned_loss=0.08645, over 11235.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2829, pruned_loss=0.05536, over 3086034.96 frames. ], batch size: 246, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:12:43,207 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.047e+02 2.684e+02 3.208e+02 4.137e+02 6.714e+02, threshold=6.416e+02, percent-clipped=2.0 2023-05-02 06:13:12,960 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=261977.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:13:58,685 INFO [train.py:904] (6/8) Epoch 26, batch 8250, loss[loss=0.1772, simple_loss=0.2623, pruned_loss=0.04607, over 11951.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2815, pruned_loss=0.0534, over 3038351.13 frames. ], batch size: 247, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:14:57,686 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=262038.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:15:13,029 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9476, 4.2145, 4.0573, 4.0879, 3.7602, 3.8390, 3.8281, 4.2261], device='cuda:6'), covar=tensor([0.1067, 0.0891, 0.0901, 0.0813, 0.0814, 0.1483, 0.0945, 0.0916], device='cuda:6'), in_proj_covar=tensor([0.0690, 0.0829, 0.0683, 0.0638, 0.0529, 0.0530, 0.0697, 0.0649], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 06:15:22,696 INFO [train.py:904] (6/8) Epoch 26, batch 8300, loss[loss=0.1677, simple_loss=0.2509, pruned_loss=0.04224, over 11918.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2784, pruned_loss=0.05036, over 3030917.80 frames. ], batch size: 247, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:15:32,830 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.264e+02 2.806e+02 3.280e+02 5.175e+02, threshold=5.611e+02, percent-clipped=0.0 2023-05-02 06:16:44,223 INFO [train.py:904] (6/8) Epoch 26, batch 8350, loss[loss=0.181, simple_loss=0.2814, pruned_loss=0.04034, over 16410.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.2784, pruned_loss=0.04849, over 3049632.00 frames. ], batch size: 146, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:16:48,362 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0461, 4.0390, 3.9294, 3.1570, 3.9727, 1.9096, 3.7814, 3.5438], device='cuda:6'), covar=tensor([0.0141, 0.0144, 0.0201, 0.0322, 0.0118, 0.2814, 0.0148, 0.0303], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0166, 0.0204, 0.0179, 0.0180, 0.0210, 0.0192, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 06:16:59,149 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=262112.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:17:17,721 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7993, 3.8302, 3.9142, 3.7140, 3.8816, 4.2874, 3.9424, 3.6561], device='cuda:6'), covar=tensor([0.2302, 0.2481, 0.2423, 0.2553, 0.2701, 0.1715, 0.1578, 0.2789], device='cuda:6'), in_proj_covar=tensor([0.0416, 0.0613, 0.0676, 0.0502, 0.0664, 0.0704, 0.0530, 0.0674], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 06:17:47,059 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-02 06:17:58,145 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-02 06:18:01,780 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9517, 2.0760, 2.3799, 3.1887, 2.1431, 2.2747, 2.2746, 2.1937], device='cuda:6'), covar=tensor([0.1344, 0.3758, 0.2867, 0.0758, 0.4719, 0.2893, 0.3733, 0.3825], device='cuda:6'), in_proj_covar=tensor([0.0407, 0.0459, 0.0374, 0.0326, 0.0435, 0.0524, 0.0430, 0.0536], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 06:18:04,036 INFO [train.py:904] (6/8) Epoch 26, batch 8400, loss[loss=0.1853, simple_loss=0.2845, pruned_loss=0.04311, over 15305.00 frames. ], tot_loss[loss=0.1848, simple_loss=0.2762, pruned_loss=0.04667, over 3041307.76 frames. ], batch size: 190, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:18:13,164 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 2.393e+02 2.671e+02 3.015e+02 4.313e+02, threshold=5.342e+02, percent-clipped=0.0 2023-05-02 06:18:15,414 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=262160.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:18:47,987 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=262180.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:19:02,299 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=262189.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:19:24,256 INFO [train.py:904] (6/8) Epoch 26, batch 8450, loss[loss=0.1842, simple_loss=0.2667, pruned_loss=0.05085, over 12364.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2741, pruned_loss=0.04508, over 3040068.94 frames. ], batch size: 247, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:19:40,601 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9696, 3.6406, 4.0020, 2.1471, 4.1442, 4.2066, 3.2782, 3.3074], device='cuda:6'), covar=tensor([0.0600, 0.0256, 0.0204, 0.1097, 0.0078, 0.0177, 0.0361, 0.0384], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0109, 0.0099, 0.0137, 0.0085, 0.0128, 0.0128, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 06:20:14,877 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2956, 4.0891, 4.3446, 4.4724, 4.6148, 4.1507, 4.5697, 4.6199], device='cuda:6'), covar=tensor([0.1925, 0.1450, 0.1563, 0.0815, 0.0630, 0.1335, 0.0823, 0.0914], device='cuda:6'), in_proj_covar=tensor([0.0644, 0.0798, 0.0917, 0.0804, 0.0615, 0.0639, 0.0666, 0.0784], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 06:20:48,803 INFO [train.py:904] (6/8) Epoch 26, batch 8500, loss[loss=0.1544, simple_loss=0.2482, pruned_loss=0.03029, over 16180.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2703, pruned_loss=0.04275, over 3048179.69 frames. ], batch size: 165, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:20:57,922 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.087e+02 2.550e+02 3.186e+02 5.088e+02, threshold=5.100e+02, percent-clipped=0.0 2023-05-02 06:21:26,190 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5499, 3.4791, 2.7360, 2.1405, 2.1879, 2.3284, 3.6280, 3.0882], device='cuda:6'), covar=tensor([0.3062, 0.0690, 0.1967, 0.3377, 0.3247, 0.2461, 0.0454, 0.1537], device='cuda:6'), in_proj_covar=tensor([0.0328, 0.0271, 0.0306, 0.0319, 0.0300, 0.0269, 0.0298, 0.0342], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 06:21:45,373 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-05-02 06:21:51,759 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3669, 4.1535, 4.4167, 4.5500, 4.7326, 4.2971, 4.7050, 4.7207], device='cuda:6'), covar=tensor([0.2029, 0.1518, 0.1642, 0.0855, 0.0645, 0.1192, 0.0718, 0.0958], device='cuda:6'), in_proj_covar=tensor([0.0641, 0.0795, 0.0913, 0.0802, 0.0613, 0.0636, 0.0664, 0.0780], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 06:22:09,457 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 06:22:09,824 INFO [train.py:904] (6/8) Epoch 26, batch 8550, loss[loss=0.1666, simple_loss=0.2664, pruned_loss=0.03345, over 16825.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2682, pruned_loss=0.04144, over 3055092.94 frames. ], batch size: 116, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:23:07,966 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=262333.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:23:47,964 INFO [train.py:904] (6/8) Epoch 26, batch 8600, loss[loss=0.1792, simple_loss=0.2773, pruned_loss=0.04051, over 16362.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2686, pruned_loss=0.04079, over 3041568.49 frames. ], batch size: 146, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:23:59,714 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.293e+02 2.800e+02 3.467e+02 6.859e+02, threshold=5.600e+02, percent-clipped=2.0 2023-05-02 06:24:32,749 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2023-05-02 06:24:57,560 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.00 vs. limit=5.0 2023-05-02 06:25:12,444 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0732, 4.0600, 3.9375, 3.1817, 3.9973, 1.7575, 3.7858, 3.4826], device='cuda:6'), covar=tensor([0.0110, 0.0108, 0.0181, 0.0268, 0.0096, 0.2991, 0.0141, 0.0332], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0166, 0.0203, 0.0178, 0.0179, 0.0210, 0.0192, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 06:25:27,326 INFO [train.py:904] (6/8) Epoch 26, batch 8650, loss[loss=0.1636, simple_loss=0.2506, pruned_loss=0.03825, over 12204.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2667, pruned_loss=0.03927, over 3046382.68 frames. ], batch size: 248, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:27:13,553 INFO [train.py:904] (6/8) Epoch 26, batch 8700, loss[loss=0.1562, simple_loss=0.2463, pruned_loss=0.0331, over 12377.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2641, pruned_loss=0.03796, over 3058812.44 frames. ], batch size: 247, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:27:25,316 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.287e+02 2.080e+02 2.583e+02 3.221e+02 6.009e+02, threshold=5.165e+02, percent-clipped=1.0 2023-05-02 06:28:02,901 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=262480.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:28:04,795 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 06:28:20,416 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=262489.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:28:48,923 INFO [train.py:904] (6/8) Epoch 26, batch 8750, loss[loss=0.1951, simple_loss=0.2918, pruned_loss=0.04916, over 16268.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2643, pruned_loss=0.03746, over 3075185.94 frames. ], batch size: 165, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:29:48,638 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=262528.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:30:09,755 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=262537.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:30:36,103 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6861, 2.5955, 1.8911, 2.8047, 2.0975, 2.8148, 2.1729, 2.4357], device='cuda:6'), covar=tensor([0.0288, 0.0320, 0.1259, 0.0271, 0.0650, 0.0433, 0.1254, 0.0620], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0175, 0.0192, 0.0165, 0.0175, 0.0212, 0.0201, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 06:30:41,539 INFO [train.py:904] (6/8) Epoch 26, batch 8800, loss[loss=0.1869, simple_loss=0.2689, pruned_loss=0.05248, over 12519.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2622, pruned_loss=0.03616, over 3075378.50 frames. ], batch size: 247, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:30:52,412 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.093e+02 2.466e+02 2.812e+02 4.660e+02, threshold=4.932e+02, percent-clipped=0.0 2023-05-02 06:32:25,981 INFO [train.py:904] (6/8) Epoch 26, batch 8850, loss[loss=0.1435, simple_loss=0.2371, pruned_loss=0.02499, over 12258.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2649, pruned_loss=0.036, over 3066482.32 frames. ], batch size: 246, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:33:32,239 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=262633.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:34:13,866 INFO [train.py:904] (6/8) Epoch 26, batch 8900, loss[loss=0.1587, simple_loss=0.2543, pruned_loss=0.03158, over 12907.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2656, pruned_loss=0.03558, over 3061225.66 frames. ], batch size: 250, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:34:26,822 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 2.190e+02 2.649e+02 3.221e+02 8.286e+02, threshold=5.298e+02, percent-clipped=4.0 2023-05-02 06:35:20,125 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=262681.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:36:18,592 INFO [train.py:904] (6/8) Epoch 26, batch 8950, loss[loss=0.1597, simple_loss=0.2543, pruned_loss=0.03258, over 16419.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2648, pruned_loss=0.03534, over 3087095.71 frames. ], batch size: 166, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:36:35,013 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4174, 4.4150, 4.7557, 4.7222, 4.7420, 4.4803, 4.4772, 4.4078], device='cuda:6'), covar=tensor([0.0353, 0.0675, 0.0392, 0.0419, 0.0463, 0.0404, 0.0904, 0.0482], device='cuda:6'), in_proj_covar=tensor([0.0414, 0.0468, 0.0455, 0.0417, 0.0504, 0.0477, 0.0551, 0.0383], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 06:38:00,808 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=262749.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:38:08,094 INFO [train.py:904] (6/8) Epoch 26, batch 9000, loss[loss=0.1404, simple_loss=0.2392, pruned_loss=0.02086, over 16856.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2618, pruned_loss=0.03446, over 3072693.42 frames. ], batch size: 96, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:38:08,094 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 06:38:18,517 INFO [train.py:938] (6/8) Epoch 26, validation: loss=0.1435, simple_loss=0.2475, pruned_loss=0.01976, over 944034.00 frames. 2023-05-02 06:38:18,518 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 06:38:30,742 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.305e+02 2.045e+02 2.519e+02 3.077e+02 5.140e+02, threshold=5.037e+02, percent-clipped=0.0 2023-05-02 06:40:03,316 INFO [train.py:904] (6/8) Epoch 26, batch 9050, loss[loss=0.1687, simple_loss=0.2546, pruned_loss=0.04144, over 16685.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2633, pruned_loss=0.03528, over 3082644.58 frames. ], batch size: 134, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:40:20,424 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=262810.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:41:48,707 INFO [train.py:904] (6/8) Epoch 26, batch 9100, loss[loss=0.1587, simple_loss=0.2517, pruned_loss=0.03282, over 12049.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.263, pruned_loss=0.03599, over 3071684.41 frames. ], batch size: 247, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:42:01,226 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.262e+02 2.654e+02 3.227e+02 6.240e+02, threshold=5.309e+02, percent-clipped=4.0 2023-05-02 06:43:45,864 INFO [train.py:904] (6/8) Epoch 26, batch 9150, loss[loss=0.1689, simple_loss=0.2637, pruned_loss=0.03703, over 16246.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2631, pruned_loss=0.03581, over 3055558.42 frames. ], batch size: 35, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:44:00,919 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7852, 3.1381, 3.4604, 2.1521, 3.0213, 2.2013, 3.3681, 3.2680], device='cuda:6'), covar=tensor([0.0297, 0.0938, 0.0500, 0.2076, 0.0760, 0.1069, 0.0653, 0.0978], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0162, 0.0164, 0.0151, 0.0143, 0.0128, 0.0141, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 06:45:27,072 INFO [train.py:904] (6/8) Epoch 26, batch 9200, loss[loss=0.1745, simple_loss=0.2582, pruned_loss=0.04542, over 16639.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2589, pruned_loss=0.03488, over 3055461.17 frames. ], batch size: 57, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:45:36,599 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.229e+02 2.531e+02 3.050e+02 5.061e+02, threshold=5.062e+02, percent-clipped=0.0 2023-05-02 06:47:01,942 INFO [train.py:904] (6/8) Epoch 26, batch 9250, loss[loss=0.1722, simple_loss=0.2673, pruned_loss=0.03853, over 16741.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2582, pruned_loss=0.0343, over 3063001.18 frames. ], batch size: 124, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:47:11,368 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4364, 3.0743, 2.6987, 2.2740, 2.1559, 2.2558, 2.9776, 2.7887], device='cuda:6'), covar=tensor([0.2667, 0.0637, 0.1709, 0.3005, 0.3201, 0.2426, 0.0427, 0.1594], device='cuda:6'), in_proj_covar=tensor([0.0326, 0.0268, 0.0304, 0.0316, 0.0294, 0.0267, 0.0296, 0.0340], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 06:47:24,603 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1989, 2.5006, 2.6217, 1.9450, 2.8151, 2.8827, 2.5531, 2.5168], device='cuda:6'), covar=tensor([0.0613, 0.0256, 0.0210, 0.0960, 0.0117, 0.0222, 0.0421, 0.0405], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0107, 0.0095, 0.0135, 0.0082, 0.0124, 0.0125, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 06:48:49,716 INFO [train.py:904] (6/8) Epoch 26, batch 9300, loss[loss=0.1443, simple_loss=0.2383, pruned_loss=0.02515, over 16287.00 frames. ], tot_loss[loss=0.1621, simple_loss=0.2564, pruned_loss=0.03386, over 3056148.11 frames. ], batch size: 166, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:49:02,019 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.290e+02 2.028e+02 2.420e+02 3.075e+02 6.199e+02, threshold=4.840e+02, percent-clipped=1.0 2023-05-02 06:50:20,342 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263095.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:50:33,213 INFO [train.py:904] (6/8) Epoch 26, batch 9350, loss[loss=0.1669, simple_loss=0.2606, pruned_loss=0.03663, over 16225.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2567, pruned_loss=0.03421, over 3046893.69 frames. ], batch size: 165, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:50:38,631 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=263105.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:50:51,193 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263111.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:51:48,292 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263140.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:51:52,882 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-02 06:52:14,167 INFO [train.py:904] (6/8) Epoch 26, batch 9400, loss[loss=0.1729, simple_loss=0.2707, pruned_loss=0.03754, over 15314.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2565, pruned_loss=0.03392, over 3044040.47 frames. ], batch size: 191, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:52:19,898 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=263156.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:52:25,026 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.316e+02 2.115e+02 2.425e+02 3.007e+02 4.996e+02, threshold=4.851e+02, percent-clipped=1.0 2023-05-02 06:52:52,638 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=263172.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:53:27,007 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6594, 3.4951, 3.7372, 1.9154, 3.8848, 4.0065, 3.0865, 3.0458], device='cuda:6'), covar=tensor([0.0728, 0.0277, 0.0226, 0.1291, 0.0092, 0.0178, 0.0434, 0.0454], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0107, 0.0096, 0.0135, 0.0082, 0.0124, 0.0126, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 06:53:50,965 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=263201.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:53:53,489 INFO [train.py:904] (6/8) Epoch 26, batch 9450, loss[loss=0.1719, simple_loss=0.2738, pruned_loss=0.03496, over 15469.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2584, pruned_loss=0.03421, over 3044936.42 frames. ], batch size: 191, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:54:14,272 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263214.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:55:19,881 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263246.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:55:33,109 INFO [train.py:904] (6/8) Epoch 26, batch 9500, loss[loss=0.1591, simple_loss=0.2535, pruned_loss=0.03231, over 16390.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2576, pruned_loss=0.03406, over 3043769.35 frames. ], batch size: 146, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:55:47,504 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.223e+02 2.048e+02 2.318e+02 3.052e+02 5.561e+02, threshold=4.636e+02, percent-clipped=2.0 2023-05-02 06:56:19,335 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8883, 3.8620, 3.9866, 3.7769, 3.9600, 4.3410, 3.9909, 3.6853], device='cuda:6'), covar=tensor([0.1902, 0.2426, 0.2508, 0.2338, 0.2502, 0.1491, 0.1630, 0.2773], device='cuda:6'), in_proj_covar=tensor([0.0400, 0.0594, 0.0658, 0.0484, 0.0643, 0.0680, 0.0513, 0.0650], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 06:56:19,526 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=263275.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:57:17,614 INFO [train.py:904] (6/8) Epoch 26, batch 9550, loss[loss=0.176, simple_loss=0.2728, pruned_loss=0.03955, over 16893.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2573, pruned_loss=0.034, over 3047944.68 frames. ], batch size: 116, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:57:27,278 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=263307.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:58:05,311 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0485, 2.1831, 2.2049, 3.5437, 2.1428, 2.4565, 2.2727, 2.3116], device='cuda:6'), covar=tensor([0.1320, 0.3732, 0.3331, 0.0652, 0.4459, 0.2723, 0.3907, 0.3692], device='cuda:6'), in_proj_covar=tensor([0.0406, 0.0457, 0.0375, 0.0323, 0.0435, 0.0521, 0.0430, 0.0533], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 06:58:11,730 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5790, 2.6241, 2.3016, 2.3154, 2.9421, 2.5904, 3.0864, 3.1232], device='cuda:6'), covar=tensor([0.0146, 0.0485, 0.0628, 0.0583, 0.0364, 0.0524, 0.0273, 0.0327], device='cuda:6'), in_proj_covar=tensor([0.0215, 0.0234, 0.0226, 0.0226, 0.0235, 0.0234, 0.0230, 0.0231], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 06:58:53,800 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9854, 3.1646, 3.1977, 1.9784, 2.9387, 3.2241, 3.1150, 1.7944], device='cuda:6'), covar=tensor([0.0601, 0.0072, 0.0075, 0.0535, 0.0145, 0.0114, 0.0088, 0.0603], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0086, 0.0087, 0.0133, 0.0099, 0.0110, 0.0095, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-02 06:58:58,572 INFO [train.py:904] (6/8) Epoch 26, batch 9600, loss[loss=0.1545, simple_loss=0.2439, pruned_loss=0.03254, over 12165.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2588, pruned_loss=0.03477, over 3036078.14 frames. ], batch size: 247, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:59:09,908 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.249e+02 2.699e+02 3.019e+02 5.664e+02, threshold=5.399e+02, percent-clipped=4.0 2023-05-02 07:00:41,944 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 07:00:45,656 INFO [train.py:904] (6/8) Epoch 26, batch 9650, loss[loss=0.1662, simple_loss=0.2608, pruned_loss=0.03577, over 16993.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2603, pruned_loss=0.03495, over 3045869.40 frames. ], batch size: 55, lr: 2.56e-03, grad_scale: 8.0 2023-05-02 07:00:52,234 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=263405.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:02:10,594 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8403, 1.3770, 1.7983, 1.7051, 1.8597, 1.9704, 1.7306, 1.8951], device='cuda:6'), covar=tensor([0.0306, 0.0455, 0.0258, 0.0340, 0.0326, 0.0228, 0.0494, 0.0167], device='cuda:6'), in_proj_covar=tensor([0.0188, 0.0192, 0.0179, 0.0183, 0.0199, 0.0157, 0.0197, 0.0157], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:02:29,079 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=263451.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:02:31,833 INFO [train.py:904] (6/8) Epoch 26, batch 9700, loss[loss=0.1485, simple_loss=0.2388, pruned_loss=0.02905, over 12338.00 frames. ], tot_loss[loss=0.165, simple_loss=0.26, pruned_loss=0.03507, over 3058210.54 frames. ], batch size: 246, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:02:33,069 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=263453.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:02:42,005 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.181e+02 2.600e+02 3.178e+02 8.821e+02, threshold=5.200e+02, percent-clipped=1.0 2023-05-02 07:02:58,816 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=263467.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:03:03,362 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8796, 3.8853, 3.9591, 3.7913, 3.9876, 4.3519, 4.0131, 3.6872], device='cuda:6'), covar=tensor([0.2056, 0.2425, 0.2722, 0.2319, 0.2366, 0.1489, 0.1434, 0.2445], device='cuda:6'), in_proj_covar=tensor([0.0400, 0.0594, 0.0658, 0.0483, 0.0642, 0.0680, 0.0513, 0.0647], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 07:03:22,941 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6830, 2.4270, 2.3195, 3.6616, 2.1072, 3.6402, 1.4842, 2.7416], device='cuda:6'), covar=tensor([0.1654, 0.0856, 0.1306, 0.0188, 0.0099, 0.0355, 0.2068, 0.0834], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0175, 0.0194, 0.0190, 0.0198, 0.0212, 0.0204, 0.0192], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 07:04:02,100 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=263496.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:04:13,671 INFO [train.py:904] (6/8) Epoch 26, batch 9750, loss[loss=0.1761, simple_loss=0.2666, pruned_loss=0.04274, over 16736.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2585, pruned_loss=0.03533, over 3053506.12 frames. ], batch size: 124, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:05:51,431 INFO [train.py:904] (6/8) Epoch 26, batch 9800, loss[loss=0.1952, simple_loss=0.3034, pruned_loss=0.04354, over 16815.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2593, pruned_loss=0.03462, over 3067170.38 frames. ], batch size: 124, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:06:03,264 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 2.162e+02 2.515e+02 2.948e+02 4.356e+02, threshold=5.031e+02, percent-clipped=0.0 2023-05-02 07:06:23,917 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=263570.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:07:33,448 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=263602.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:07:34,241 INFO [train.py:904] (6/8) Epoch 26, batch 9850, loss[loss=0.1589, simple_loss=0.2557, pruned_loss=0.03102, over 16866.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2605, pruned_loss=0.03449, over 3077026.72 frames. ], batch size: 124, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:07:35,355 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9321, 2.1915, 2.3916, 3.1647, 2.2570, 2.3696, 2.3263, 2.2951], device='cuda:6'), covar=tensor([0.1425, 0.3854, 0.2909, 0.0798, 0.4376, 0.2802, 0.3771, 0.3835], device='cuda:6'), in_proj_covar=tensor([0.0408, 0.0459, 0.0377, 0.0326, 0.0437, 0.0522, 0.0431, 0.0534], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:07:53,568 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6128, 4.8129, 4.8974, 4.7039, 4.8203, 5.2899, 4.7631, 4.4469], device='cuda:6'), covar=tensor([0.1154, 0.1747, 0.2011, 0.1884, 0.1944, 0.0798, 0.1441, 0.2238], device='cuda:6'), in_proj_covar=tensor([0.0399, 0.0593, 0.0656, 0.0484, 0.0640, 0.0680, 0.0512, 0.0645], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 07:09:05,833 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4494, 3.3710, 2.7319, 2.1788, 2.2039, 2.3180, 3.5198, 2.9827], device='cuda:6'), covar=tensor([0.3102, 0.0655, 0.1827, 0.3099, 0.2967, 0.2490, 0.0439, 0.1499], device='cuda:6'), in_proj_covar=tensor([0.0326, 0.0268, 0.0305, 0.0317, 0.0293, 0.0267, 0.0296, 0.0339], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 07:09:24,089 INFO [train.py:904] (6/8) Epoch 26, batch 9900, loss[loss=0.1481, simple_loss=0.2549, pruned_loss=0.02066, over 17169.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2603, pruned_loss=0.03409, over 3056687.08 frames. ], batch size: 48, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:09:36,802 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 1.934e+02 2.301e+02 2.936e+02 6.933e+02, threshold=4.602e+02, percent-clipped=3.0 2023-05-02 07:09:42,040 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1616, 2.2699, 2.1767, 3.8612, 2.1470, 2.5628, 2.3084, 2.3638], device='cuda:6'), covar=tensor([0.1338, 0.3891, 0.3340, 0.0553, 0.4502, 0.2760, 0.3834, 0.3846], device='cuda:6'), in_proj_covar=tensor([0.0406, 0.0457, 0.0375, 0.0324, 0.0435, 0.0520, 0.0429, 0.0532], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:10:26,511 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8070, 4.1578, 3.0505, 2.3885, 2.5943, 2.6887, 4.5297, 3.4731], device='cuda:6'), covar=tensor([0.2845, 0.0509, 0.1783, 0.2856, 0.2880, 0.1947, 0.0267, 0.1339], device='cuda:6'), in_proj_covar=tensor([0.0326, 0.0268, 0.0304, 0.0317, 0.0293, 0.0267, 0.0296, 0.0339], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 07:11:20,739 INFO [train.py:904] (6/8) Epoch 26, batch 9950, loss[loss=0.1479, simple_loss=0.2483, pruned_loss=0.02369, over 16529.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2619, pruned_loss=0.03432, over 3032091.84 frames. ], batch size: 68, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:12:44,022 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.17 vs. limit=5.0 2023-05-02 07:13:19,223 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=263751.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:13:21,140 INFO [train.py:904] (6/8) Epoch 26, batch 10000, loss[loss=0.1516, simple_loss=0.2584, pruned_loss=0.02236, over 16923.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2607, pruned_loss=0.03367, over 3047133.18 frames. ], batch size: 102, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:13:34,914 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.519e+02 2.135e+02 2.379e+02 2.761e+02 5.378e+02, threshold=4.758e+02, percent-clipped=3.0 2023-05-02 07:13:48,386 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 07:13:51,380 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=263767.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:13:52,741 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8397, 5.1122, 4.9487, 4.9226, 4.6502, 4.6379, 4.4995, 5.1630], device='cuda:6'), covar=tensor([0.1096, 0.0740, 0.0808, 0.0752, 0.0747, 0.1067, 0.1204, 0.0819], device='cuda:6'), in_proj_covar=tensor([0.0678, 0.0817, 0.0668, 0.0627, 0.0522, 0.0521, 0.0689, 0.0643], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:14:48,924 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=263796.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:14:53,809 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=263799.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:15:00,032 INFO [train.py:904] (6/8) Epoch 26, batch 10050, loss[loss=0.1661, simple_loss=0.2644, pruned_loss=0.03395, over 12167.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2609, pruned_loss=0.03378, over 3048480.99 frames. ], batch size: 248, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:15:23,692 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=263815.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:15:57,715 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263833.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 07:16:17,294 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=263844.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:16:17,805 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 07:16:30,281 INFO [train.py:904] (6/8) Epoch 26, batch 10100, loss[loss=0.1514, simple_loss=0.2443, pruned_loss=0.02927, over 16699.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2615, pruned_loss=0.03389, over 3047093.41 frames. ], batch size: 134, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:16:39,584 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 2.128e+02 2.539e+02 3.188e+02 8.403e+02, threshold=5.079e+02, percent-clipped=3.0 2023-05-02 07:17:02,796 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=263870.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:17:36,334 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=263894.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 07:18:08,995 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=263902.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:18:09,862 INFO [train.py:904] (6/8) Epoch 27, batch 0, loss[loss=0.1859, simple_loss=0.2825, pruned_loss=0.0446, over 17020.00 frames. ], tot_loss[loss=0.1859, simple_loss=0.2825, pruned_loss=0.0446, over 17020.00 frames. ], batch size: 50, lr: 2.51e-03, grad_scale: 16.0 2023-05-02 07:18:09,862 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 07:18:17,113 INFO [train.py:938] (6/8) Epoch 27, validation: loss=0.1434, simple_loss=0.2467, pruned_loss=0.02006, over 944034.00 frames. 2023-05-02 07:18:17,114 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 07:18:38,687 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=263918.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:19:22,226 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=263950.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:19:26,180 INFO [train.py:904] (6/8) Epoch 27, batch 50, loss[loss=0.153, simple_loss=0.235, pruned_loss=0.03548, over 16833.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2677, pruned_loss=0.04601, over 750723.79 frames. ], batch size: 96, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:19:36,256 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263960.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:19:39,887 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.543e+02 2.464e+02 3.043e+02 3.721e+02 7.050e+02, threshold=6.086e+02, percent-clipped=6.0 2023-05-02 07:19:50,788 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4233, 3.3343, 2.6888, 2.1589, 2.2305, 2.2983, 3.4141, 2.9781], device='cuda:6'), covar=tensor([0.2996, 0.0750, 0.1835, 0.2985, 0.2615, 0.2231, 0.0636, 0.1642], device='cuda:6'), in_proj_covar=tensor([0.0326, 0.0268, 0.0306, 0.0317, 0.0293, 0.0268, 0.0297, 0.0340], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 07:20:36,409 INFO [train.py:904] (6/8) Epoch 27, batch 100, loss[loss=0.1331, simple_loss=0.2194, pruned_loss=0.02338, over 16821.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2613, pruned_loss=0.04334, over 1326443.78 frames. ], batch size: 42, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:21:01,025 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=264021.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:21:10,240 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-05-02 07:21:36,111 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0946, 5.7267, 5.8591, 5.5506, 5.7026, 6.2346, 5.7513, 5.4609], device='cuda:6'), covar=tensor([0.0981, 0.2283, 0.3105, 0.2145, 0.2620, 0.0957, 0.1586, 0.2283], device='cuda:6'), in_proj_covar=tensor([0.0407, 0.0602, 0.0669, 0.0493, 0.0652, 0.0693, 0.0522, 0.0657], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 07:21:44,758 INFO [train.py:904] (6/8) Epoch 27, batch 150, loss[loss=0.1495, simple_loss=0.2427, pruned_loss=0.02816, over 17229.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2602, pruned_loss=0.04297, over 1763614.72 frames. ], batch size: 45, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:21:57,549 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.626e+02 2.228e+02 2.541e+02 3.123e+02 6.802e+02, threshold=5.083e+02, percent-clipped=1.0 2023-05-02 07:22:48,855 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3246, 3.0244, 2.5908, 2.2227, 2.2003, 2.1482, 3.0295, 2.7803], device='cuda:6'), covar=tensor([0.2724, 0.0792, 0.1812, 0.2690, 0.2841, 0.2453, 0.0602, 0.1461], device='cuda:6'), in_proj_covar=tensor([0.0328, 0.0270, 0.0308, 0.0319, 0.0295, 0.0270, 0.0299, 0.0343], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 07:22:53,632 INFO [train.py:904] (6/8) Epoch 27, batch 200, loss[loss=0.2271, simple_loss=0.2924, pruned_loss=0.0809, over 16728.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2602, pruned_loss=0.04386, over 2106985.26 frames. ], batch size: 83, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:24:00,835 INFO [train.py:904] (6/8) Epoch 27, batch 250, loss[loss=0.1747, simple_loss=0.2702, pruned_loss=0.03961, over 17083.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.259, pruned_loss=0.04337, over 2383150.15 frames. ], batch size: 47, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:24:14,011 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.137e+02 2.566e+02 3.283e+02 9.180e+02, threshold=5.132e+02, percent-clipped=3.0 2023-05-02 07:24:50,573 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=264189.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 07:24:50,714 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1089, 5.0220, 4.8370, 4.1170, 4.9164, 1.8407, 4.6678, 4.6271], device='cuda:6'), covar=tensor([0.0127, 0.0118, 0.0278, 0.0548, 0.0134, 0.3247, 0.0188, 0.0351], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0167, 0.0203, 0.0176, 0.0180, 0.0212, 0.0193, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:25:10,466 INFO [train.py:904] (6/8) Epoch 27, batch 300, loss[loss=0.1889, simple_loss=0.272, pruned_loss=0.05289, over 12450.00 frames. ], tot_loss[loss=0.171, simple_loss=0.257, pruned_loss=0.04249, over 2593069.47 frames. ], batch size: 247, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:25:42,704 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5753, 3.6585, 4.2224, 2.4440, 3.5115, 2.6541, 4.1044, 3.9205], device='cuda:6'), covar=tensor([0.0262, 0.0953, 0.0497, 0.2017, 0.0742, 0.1015, 0.0557, 0.1061], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0165, 0.0167, 0.0154, 0.0145, 0.0130, 0.0144, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 07:25:53,782 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=264234.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:26:19,420 INFO [train.py:904] (6/8) Epoch 27, batch 350, loss[loss=0.1623, simple_loss=0.2445, pruned_loss=0.04008, over 15974.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2545, pruned_loss=0.04148, over 2750995.14 frames. ], batch size: 35, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:26:34,329 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.070e+02 2.475e+02 2.951e+02 5.112e+02, threshold=4.951e+02, percent-clipped=0.0 2023-05-02 07:26:42,193 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4506, 4.3858, 4.3780, 4.0828, 4.1370, 4.3885, 4.1739, 4.2107], device='cuda:6'), covar=tensor([0.0602, 0.0833, 0.0332, 0.0321, 0.0719, 0.0486, 0.0604, 0.0619], device='cuda:6'), in_proj_covar=tensor([0.0305, 0.0452, 0.0353, 0.0355, 0.0352, 0.0408, 0.0243, 0.0423], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:27:17,920 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=264295.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:27:27,689 INFO [train.py:904] (6/8) Epoch 27, batch 400, loss[loss=0.2037, simple_loss=0.2746, pruned_loss=0.0664, over 16908.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2537, pruned_loss=0.0411, over 2879794.34 frames. ], batch size: 109, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:27:33,126 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.24 vs. limit=5.0 2023-05-02 07:27:45,292 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=264316.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:28:33,916 INFO [train.py:904] (6/8) Epoch 27, batch 450, loss[loss=0.1615, simple_loss=0.2397, pruned_loss=0.04168, over 16785.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2509, pruned_loss=0.03954, over 2981595.95 frames. ], batch size: 124, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:28:48,385 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 2.223e+02 2.597e+02 3.088e+02 6.420e+02, threshold=5.195e+02, percent-clipped=1.0 2023-05-02 07:29:44,911 INFO [train.py:904] (6/8) Epoch 27, batch 500, loss[loss=0.1691, simple_loss=0.2611, pruned_loss=0.03854, over 16809.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2488, pruned_loss=0.0385, over 3056031.36 frames. ], batch size: 102, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:29:59,900 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8451, 3.5415, 3.9544, 2.2064, 4.0861, 4.1004, 3.2630, 3.0754], device='cuda:6'), covar=tensor([0.0778, 0.0308, 0.0213, 0.1189, 0.0120, 0.0217, 0.0410, 0.0490], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0111, 0.0099, 0.0139, 0.0085, 0.0129, 0.0130, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 07:30:51,106 INFO [train.py:904] (6/8) Epoch 27, batch 550, loss[loss=0.1649, simple_loss=0.2511, pruned_loss=0.03936, over 17233.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.249, pruned_loss=0.03869, over 3117760.97 frames. ], batch size: 44, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:31:04,228 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.059e+02 2.347e+02 2.787e+02 4.742e+02, threshold=4.693e+02, percent-clipped=0.0 2023-05-02 07:31:26,712 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=264479.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:31:41,320 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=264489.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 07:31:55,527 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=264500.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 07:31:59,347 INFO [train.py:904] (6/8) Epoch 27, batch 600, loss[loss=0.1633, simple_loss=0.2568, pruned_loss=0.0349, over 16768.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2481, pruned_loss=0.03855, over 3165200.93 frames. ], batch size: 57, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:32:07,742 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8004, 4.3850, 2.8448, 2.2536, 2.8373, 2.4387, 4.6760, 3.5228], device='cuda:6'), covar=tensor([0.3263, 0.0653, 0.2260, 0.3156, 0.2978, 0.2462, 0.0436, 0.1654], device='cuda:6'), in_proj_covar=tensor([0.0334, 0.0275, 0.0314, 0.0325, 0.0302, 0.0275, 0.0304, 0.0351], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 07:32:13,400 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5261, 3.5344, 3.8343, 2.6863, 3.5014, 3.9256, 3.6057, 2.2625], device='cuda:6'), covar=tensor([0.0532, 0.0222, 0.0062, 0.0411, 0.0128, 0.0099, 0.0117, 0.0516], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0088, 0.0088, 0.0134, 0.0100, 0.0112, 0.0096, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 07:32:16,803 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 07:32:46,463 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=264537.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 07:32:50,468 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0563, 4.9916, 4.7327, 3.6299, 4.8397, 1.7120, 4.5008, 4.4053], device='cuda:6'), covar=tensor([0.0158, 0.0131, 0.0337, 0.0793, 0.0184, 0.3934, 0.0241, 0.0498], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0170, 0.0207, 0.0180, 0.0183, 0.0215, 0.0196, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:32:50,505 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=264540.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:33:05,173 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1862, 5.1431, 4.9102, 4.4349, 5.0094, 1.8935, 4.7793, 4.6623], device='cuda:6'), covar=tensor([0.0100, 0.0093, 0.0237, 0.0384, 0.0107, 0.3061, 0.0141, 0.0299], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0170, 0.0206, 0.0179, 0.0183, 0.0215, 0.0196, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:33:08,796 INFO [train.py:904] (6/8) Epoch 27, batch 650, loss[loss=0.1547, simple_loss=0.2407, pruned_loss=0.03431, over 17230.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.2469, pruned_loss=0.03801, over 3192139.44 frames. ], batch size: 45, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:33:18,954 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=264561.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 07:33:20,151 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3010, 2.2652, 2.3903, 3.9939, 2.2700, 2.6011, 2.3695, 2.4315], device='cuda:6'), covar=tensor([0.1553, 0.3965, 0.3232, 0.0721, 0.4203, 0.2839, 0.4143, 0.3400], device='cuda:6'), in_proj_covar=tensor([0.0416, 0.0469, 0.0384, 0.0333, 0.0444, 0.0534, 0.0439, 0.0546], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:33:20,702 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.185e+02 2.561e+02 3.175e+02 5.971e+02, threshold=5.123e+02, percent-clipped=4.0 2023-05-02 07:33:58,957 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=264590.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:34:07,588 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-02 07:34:15,989 INFO [train.py:904] (6/8) Epoch 27, batch 700, loss[loss=0.1484, simple_loss=0.2462, pruned_loss=0.02524, over 17040.00 frames. ], tot_loss[loss=0.1609, simple_loss=0.2466, pruned_loss=0.03759, over 3218827.84 frames. ], batch size: 50, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:34:34,296 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=264616.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:35:22,117 INFO [train.py:904] (6/8) Epoch 27, batch 750, loss[loss=0.172, simple_loss=0.2535, pruned_loss=0.0452, over 16681.00 frames. ], tot_loss[loss=0.1612, simple_loss=0.2473, pruned_loss=0.03753, over 3242405.24 frames. ], batch size: 89, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:35:35,422 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.615e+02 2.184e+02 2.458e+02 2.836e+02 5.785e+02, threshold=4.916e+02, percent-clipped=2.0 2023-05-02 07:35:37,451 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=264664.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:36:27,146 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5652, 3.5451, 3.5102, 2.8166, 3.3526, 2.0972, 3.2239, 2.7691], device='cuda:6'), covar=tensor([0.0184, 0.0154, 0.0183, 0.0240, 0.0118, 0.2492, 0.0157, 0.0316], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0169, 0.0205, 0.0179, 0.0183, 0.0214, 0.0195, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:36:29,179 INFO [train.py:904] (6/8) Epoch 27, batch 800, loss[loss=0.1564, simple_loss=0.2396, pruned_loss=0.03663, over 16753.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.2476, pruned_loss=0.03761, over 3261907.23 frames. ], batch size: 102, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:36:32,207 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.30 vs. limit=5.0 2023-05-02 07:37:08,234 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.07 vs. limit=5.0 2023-05-02 07:37:31,718 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6666, 3.7739, 2.4585, 4.3662, 2.9736, 4.3107, 2.4590, 3.2179], device='cuda:6'), covar=tensor([0.0381, 0.0444, 0.1669, 0.0404, 0.0930, 0.0572, 0.1700, 0.0794], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0182, 0.0197, 0.0173, 0.0181, 0.0221, 0.0207, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 07:37:36,919 INFO [train.py:904] (6/8) Epoch 27, batch 850, loss[loss=0.1682, simple_loss=0.2452, pruned_loss=0.0456, over 16897.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.2471, pruned_loss=0.03749, over 3275622.59 frames. ], batch size: 96, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:37:51,805 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 2.032e+02 2.277e+02 2.694e+02 3.998e+02, threshold=4.553e+02, percent-clipped=0.0 2023-05-02 07:37:53,957 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0422, 5.1533, 4.9564, 4.5039, 4.1465, 5.0793, 5.0789, 4.5646], device='cuda:6'), covar=tensor([0.0908, 0.0683, 0.0596, 0.0609, 0.2343, 0.0669, 0.0391, 0.0987], device='cuda:6'), in_proj_covar=tensor([0.0314, 0.0465, 0.0363, 0.0366, 0.0362, 0.0422, 0.0250, 0.0437], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:38:44,873 INFO [train.py:904] (6/8) Epoch 27, batch 900, loss[loss=0.1694, simple_loss=0.2472, pruned_loss=0.04581, over 16697.00 frames. ], tot_loss[loss=0.1602, simple_loss=0.2466, pruned_loss=0.03695, over 3289225.68 frames. ], batch size: 134, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:38:47,144 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=264804.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:39:19,826 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8656, 2.1886, 2.4222, 3.1431, 2.2145, 2.3350, 2.3637, 2.2994], device='cuda:6'), covar=tensor([0.1546, 0.3563, 0.2983, 0.0851, 0.4229, 0.2657, 0.3373, 0.3494], device='cuda:6'), in_proj_covar=tensor([0.0419, 0.0470, 0.0385, 0.0334, 0.0445, 0.0536, 0.0441, 0.0549], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:39:28,077 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=264835.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:39:51,349 INFO [train.py:904] (6/8) Epoch 27, batch 950, loss[loss=0.1576, simple_loss=0.2429, pruned_loss=0.03614, over 15536.00 frames. ], tot_loss[loss=0.1599, simple_loss=0.2465, pruned_loss=0.03659, over 3288910.05 frames. ], batch size: 190, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:39:53,181 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-02 07:39:55,652 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=264856.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 07:40:04,149 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 2.026e+02 2.339e+02 3.072e+02 8.203e+02, threshold=4.678e+02, percent-clipped=3.0 2023-05-02 07:40:07,737 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=264865.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:40:41,772 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=264890.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:40:57,191 INFO [train.py:904] (6/8) Epoch 27, batch 1000, loss[loss=0.152, simple_loss=0.2345, pruned_loss=0.03479, over 11810.00 frames. ], tot_loss[loss=0.1598, simple_loss=0.2461, pruned_loss=0.03675, over 3300825.69 frames. ], batch size: 246, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:41:47,416 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=264938.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:41:52,338 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2248, 2.3311, 2.4415, 3.9256, 2.3050, 2.6596, 2.3638, 2.4883], device='cuda:6'), covar=tensor([0.1631, 0.3811, 0.3233, 0.0708, 0.4108, 0.2672, 0.4199, 0.3238], device='cuda:6'), in_proj_covar=tensor([0.0421, 0.0473, 0.0387, 0.0337, 0.0448, 0.0539, 0.0443, 0.0552], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:41:52,587 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-02 07:42:07,407 INFO [train.py:904] (6/8) Epoch 27, batch 1050, loss[loss=0.1698, simple_loss=0.2505, pruned_loss=0.0446, over 16498.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.2456, pruned_loss=0.03674, over 3299537.89 frames. ], batch size: 68, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:42:19,713 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.065e+02 2.389e+02 2.901e+02 6.221e+02, threshold=4.777e+02, percent-clipped=3.0 2023-05-02 07:43:09,653 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-05-02 07:43:16,446 INFO [train.py:904] (6/8) Epoch 27, batch 1100, loss[loss=0.1593, simple_loss=0.2417, pruned_loss=0.03847, over 16420.00 frames. ], tot_loss[loss=0.1585, simple_loss=0.2446, pruned_loss=0.03615, over 3311651.48 frames. ], batch size: 75, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:44:12,993 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.89 vs. limit=5.0 2023-05-02 07:44:25,472 INFO [train.py:904] (6/8) Epoch 27, batch 1150, loss[loss=0.1435, simple_loss=0.2415, pruned_loss=0.02275, over 17238.00 frames. ], tot_loss[loss=0.1585, simple_loss=0.2446, pruned_loss=0.03622, over 3319070.04 frames. ], batch size: 45, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:44:39,253 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.259e+02 1.949e+02 2.374e+02 2.990e+02 6.257e+02, threshold=4.748e+02, percent-clipped=3.0 2023-05-02 07:45:34,345 INFO [train.py:904] (6/8) Epoch 27, batch 1200, loss[loss=0.153, simple_loss=0.242, pruned_loss=0.03205, over 17195.00 frames. ], tot_loss[loss=0.1577, simple_loss=0.2442, pruned_loss=0.03565, over 3320546.49 frames. ], batch size: 45, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:46:19,393 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=265135.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:46:29,366 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2428, 3.2616, 2.1903, 3.4707, 2.6410, 3.4541, 2.2450, 2.7513], device='cuda:6'), covar=tensor([0.0335, 0.0436, 0.1513, 0.0375, 0.0800, 0.0725, 0.1478, 0.0729], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0182, 0.0198, 0.0175, 0.0182, 0.0223, 0.0207, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 07:46:42,962 INFO [train.py:904] (6/8) Epoch 27, batch 1250, loss[loss=0.1605, simple_loss=0.2364, pruned_loss=0.04231, over 16689.00 frames. ], tot_loss[loss=0.1584, simple_loss=0.2439, pruned_loss=0.03643, over 3327845.72 frames. ], batch size: 124, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:46:48,391 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=265156.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 07:46:53,529 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=265160.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:46:57,310 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 2.048e+02 2.480e+02 3.047e+02 6.939e+02, threshold=4.960e+02, percent-clipped=4.0 2023-05-02 07:47:25,262 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=265183.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:47:53,022 INFO [train.py:904] (6/8) Epoch 27, batch 1300, loss[loss=0.1453, simple_loss=0.2388, pruned_loss=0.02588, over 17246.00 frames. ], tot_loss[loss=0.158, simple_loss=0.2438, pruned_loss=0.03614, over 3332208.02 frames. ], batch size: 45, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:47:54,318 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=265204.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 07:48:41,199 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2957, 1.6648, 2.0535, 2.1401, 2.2639, 2.3781, 1.7595, 2.3476], device='cuda:6'), covar=tensor([0.0261, 0.0558, 0.0344, 0.0396, 0.0371, 0.0319, 0.0590, 0.0226], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0197, 0.0185, 0.0191, 0.0206, 0.0165, 0.0204, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:49:00,603 INFO [train.py:904] (6/8) Epoch 27, batch 1350, loss[loss=0.1715, simple_loss=0.254, pruned_loss=0.04445, over 16836.00 frames. ], tot_loss[loss=0.1592, simple_loss=0.245, pruned_loss=0.03667, over 3321614.92 frames. ], batch size: 116, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:49:14,443 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.174e+02 2.461e+02 3.019e+02 8.065e+02, threshold=4.923e+02, percent-clipped=2.0 2023-05-02 07:49:35,243 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=265278.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:49:37,371 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2582, 5.6190, 5.4151, 5.4251, 5.0707, 5.0739, 5.0006, 5.7482], device='cuda:6'), covar=tensor([0.1428, 0.0905, 0.0974, 0.0890, 0.0916, 0.0785, 0.1409, 0.0871], device='cuda:6'), in_proj_covar=tensor([0.0722, 0.0867, 0.0711, 0.0670, 0.0554, 0.0549, 0.0736, 0.0681], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:50:07,773 INFO [train.py:904] (6/8) Epoch 27, batch 1400, loss[loss=0.1416, simple_loss=0.2324, pruned_loss=0.02541, over 17197.00 frames. ], tot_loss[loss=0.1592, simple_loss=0.2446, pruned_loss=0.03684, over 3323483.84 frames. ], batch size: 45, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:50:56,846 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=265339.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:51:15,551 INFO [train.py:904] (6/8) Epoch 27, batch 1450, loss[loss=0.1501, simple_loss=0.2488, pruned_loss=0.02567, over 17267.00 frames. ], tot_loss[loss=0.158, simple_loss=0.2435, pruned_loss=0.03619, over 3313443.56 frames. ], batch size: 52, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:51:29,667 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.640e+02 2.111e+02 2.470e+02 3.011e+02 5.399e+02, threshold=4.940e+02, percent-clipped=1.0 2023-05-02 07:51:32,770 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-02 07:52:24,302 INFO [train.py:904] (6/8) Epoch 27, batch 1500, loss[loss=0.1671, simple_loss=0.2447, pruned_loss=0.0448, over 16774.00 frames. ], tot_loss[loss=0.1586, simple_loss=0.2438, pruned_loss=0.03666, over 3316418.92 frames. ], batch size: 124, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:52:41,792 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-05-02 07:52:48,055 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 07:53:36,975 INFO [train.py:904] (6/8) Epoch 27, batch 1550, loss[loss=0.1491, simple_loss=0.2367, pruned_loss=0.03076, over 17193.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.2447, pruned_loss=0.03737, over 3325222.34 frames. ], batch size: 44, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:53:45,998 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=265460.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:53:49,831 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.731e+02 2.243e+02 2.540e+02 3.021e+02 6.885e+02, threshold=5.080e+02, percent-clipped=3.0 2023-05-02 07:54:26,317 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3852, 4.2269, 4.4528, 4.5787, 4.6601, 4.2816, 4.5175, 4.6849], device='cuda:6'), covar=tensor([0.1727, 0.1301, 0.1338, 0.0646, 0.0620, 0.1174, 0.2873, 0.0713], device='cuda:6'), in_proj_covar=tensor([0.0689, 0.0848, 0.0979, 0.0859, 0.0652, 0.0677, 0.0713, 0.0829], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:54:43,792 INFO [train.py:904] (6/8) Epoch 27, batch 1600, loss[loss=0.1457, simple_loss=0.2324, pruned_loss=0.02949, over 16818.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.2461, pruned_loss=0.03777, over 3326886.47 frames. ], batch size: 39, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:54:51,465 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=265508.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:55:21,284 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=265530.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:55:32,003 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0864, 2.3460, 2.7354, 3.0411, 2.9213, 3.5529, 2.5902, 3.5258], device='cuda:6'), covar=tensor([0.0287, 0.0507, 0.0346, 0.0340, 0.0360, 0.0212, 0.0465, 0.0197], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0196, 0.0185, 0.0191, 0.0206, 0.0165, 0.0203, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 07:55:51,090 INFO [train.py:904] (6/8) Epoch 27, batch 1650, loss[loss=0.1722, simple_loss=0.2656, pruned_loss=0.03938, over 17121.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2489, pruned_loss=0.03883, over 3317697.72 frames. ], batch size: 49, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:56:00,142 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-02 07:56:04,414 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.682e+02 2.309e+02 2.667e+02 3.273e+02 6.043e+02, threshold=5.334e+02, percent-clipped=4.0 2023-05-02 07:56:44,628 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=265591.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:57:00,225 INFO [train.py:904] (6/8) Epoch 27, batch 1700, loss[loss=0.1649, simple_loss=0.2459, pruned_loss=0.04191, over 16759.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2509, pruned_loss=0.03922, over 3312016.01 frames. ], batch size: 102, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:57:16,084 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=265614.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:57:42,663 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=265634.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:58:07,548 INFO [train.py:904] (6/8) Epoch 27, batch 1750, loss[loss=0.1728, simple_loss=0.267, pruned_loss=0.03927, over 16748.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2509, pruned_loss=0.03872, over 3318914.03 frames. ], batch size: 57, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:58:22,124 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.193e+02 2.494e+02 2.973e+02 6.444e+02, threshold=4.988e+02, percent-clipped=1.0 2023-05-02 07:58:38,507 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=265675.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 07:58:47,433 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-05-02 07:59:15,654 INFO [train.py:904] (6/8) Epoch 27, batch 1800, loss[loss=0.1438, simple_loss=0.2404, pruned_loss=0.02361, over 17128.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2515, pruned_loss=0.03884, over 3313116.87 frames. ], batch size: 48, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:59:32,432 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=265715.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:00:23,854 INFO [train.py:904] (6/8) Epoch 27, batch 1850, loss[loss=0.1785, simple_loss=0.2573, pruned_loss=0.04982, over 16920.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2531, pruned_loss=0.03895, over 3320952.58 frames. ], batch size: 109, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:00:37,843 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.494e+02 2.122e+02 2.552e+02 2.913e+02 6.429e+02, threshold=5.105e+02, percent-clipped=1.0 2023-05-02 08:00:55,751 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=265776.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:01:33,280 INFO [train.py:904] (6/8) Epoch 27, batch 1900, loss[loss=0.1609, simple_loss=0.2446, pruned_loss=0.0386, over 16471.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.252, pruned_loss=0.0391, over 3307472.83 frames. ], batch size: 146, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:01:33,711 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7055, 4.2808, 4.3136, 3.0254, 3.5815, 4.2999, 3.8431, 2.5448], device='cuda:6'), covar=tensor([0.0573, 0.0086, 0.0057, 0.0396, 0.0150, 0.0099, 0.0108, 0.0515], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0088, 0.0090, 0.0135, 0.0101, 0.0114, 0.0097, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 08:02:10,917 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7005, 3.9164, 2.5068, 4.5455, 3.0630, 4.3777, 2.4840, 3.1544], device='cuda:6'), covar=tensor([0.0334, 0.0403, 0.1659, 0.0254, 0.0856, 0.0604, 0.1663, 0.0823], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0183, 0.0198, 0.0175, 0.0182, 0.0223, 0.0207, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 08:02:26,314 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2023-05-02 08:02:34,452 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4861, 3.5023, 2.2722, 3.7466, 2.8135, 3.6894, 2.3884, 2.8817], device='cuda:6'), covar=tensor([0.0297, 0.0412, 0.1547, 0.0375, 0.0810, 0.0827, 0.1378, 0.0742], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0183, 0.0198, 0.0175, 0.0182, 0.0223, 0.0207, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 08:02:42,098 INFO [train.py:904] (6/8) Epoch 27, batch 1950, loss[loss=0.1807, simple_loss=0.261, pruned_loss=0.0502, over 16699.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2527, pruned_loss=0.03909, over 3304965.56 frames. ], batch size: 134, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:02:54,908 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.085e+02 2.462e+02 3.209e+02 6.071e+02, threshold=4.924e+02, percent-clipped=6.0 2023-05-02 08:03:08,224 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=265872.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:03:27,057 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=265886.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:03:50,713 INFO [train.py:904] (6/8) Epoch 27, batch 2000, loss[loss=0.1417, simple_loss=0.2326, pruned_loss=0.02545, over 17183.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2514, pruned_loss=0.03876, over 3314085.72 frames. ], batch size: 46, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:03:57,410 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7915, 4.3551, 3.1375, 2.2939, 2.7320, 2.6639, 4.6736, 3.5624], device='cuda:6'), covar=tensor([0.3016, 0.0557, 0.1785, 0.3155, 0.2851, 0.2159, 0.0327, 0.1505], device='cuda:6'), in_proj_covar=tensor([0.0333, 0.0275, 0.0311, 0.0325, 0.0303, 0.0274, 0.0304, 0.0350], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 08:04:09,380 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2320, 5.1647, 5.1045, 4.5525, 4.6989, 5.1083, 5.0359, 4.7490], device='cuda:6'), covar=tensor([0.0645, 0.0655, 0.0384, 0.0441, 0.1259, 0.0606, 0.0359, 0.0969], device='cuda:6'), in_proj_covar=tensor([0.0321, 0.0478, 0.0372, 0.0376, 0.0372, 0.0432, 0.0256, 0.0448], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 08:04:09,490 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8557, 2.9396, 2.7251, 4.9931, 3.9024, 4.4370, 1.7094, 3.1834], device='cuda:6'), covar=tensor([0.1398, 0.0833, 0.1268, 0.0201, 0.0230, 0.0378, 0.1672, 0.0783], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0180, 0.0199, 0.0200, 0.0205, 0.0219, 0.0208, 0.0197], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 08:04:10,579 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=265917.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:04:33,629 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=265933.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:04:34,761 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=265934.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:05:01,212 INFO [train.py:904] (6/8) Epoch 27, batch 2050, loss[loss=0.1794, simple_loss=0.2732, pruned_loss=0.04282, over 16560.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2513, pruned_loss=0.03919, over 3318605.53 frames. ], batch size: 62, lr: 2.50e-03, grad_scale: 16.0 2023-05-02 08:05:14,204 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.179e+02 2.495e+02 3.022e+02 6.334e+02, threshold=4.990e+02, percent-clipped=2.0 2023-05-02 08:05:25,389 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=265970.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 08:05:36,740 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=265978.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:05:41,863 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=265982.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:06:04,926 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6628, 3.6068, 4.2106, 2.2883, 3.4034, 2.6476, 4.0619, 3.8642], device='cuda:6'), covar=tensor([0.0250, 0.1085, 0.0465, 0.2102, 0.0805, 0.0980, 0.0549, 0.1117], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0170, 0.0170, 0.0157, 0.0147, 0.0132, 0.0146, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 08:06:14,059 INFO [train.py:904] (6/8) Epoch 27, batch 2100, loss[loss=0.1632, simple_loss=0.2471, pruned_loss=0.03967, over 17237.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2517, pruned_loss=0.03945, over 3321525.71 frames. ], batch size: 45, lr: 2.50e-03, grad_scale: 16.0 2023-05-02 08:06:57,083 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 08:07:06,790 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1723, 2.6699, 2.1754, 2.3853, 2.9617, 2.6992, 3.0286, 3.0767], device='cuda:6'), covar=tensor([0.0231, 0.0421, 0.0570, 0.0503, 0.0290, 0.0398, 0.0285, 0.0323], device='cuda:6'), in_proj_covar=tensor([0.0233, 0.0248, 0.0237, 0.0237, 0.0249, 0.0248, 0.0247, 0.0246], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 08:07:22,091 INFO [train.py:904] (6/8) Epoch 27, batch 2150, loss[loss=0.2026, simple_loss=0.2927, pruned_loss=0.0562, over 16743.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.253, pruned_loss=0.03978, over 3321925.41 frames. ], batch size: 83, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:07:37,030 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.561e+02 2.274e+02 2.676e+02 3.152e+02 5.314e+02, threshold=5.352e+02, percent-clipped=2.0 2023-05-02 08:07:47,118 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266071.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:08:32,965 INFO [train.py:904] (6/8) Epoch 27, batch 2200, loss[loss=0.16, simple_loss=0.2444, pruned_loss=0.03784, over 16781.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2536, pruned_loss=0.03939, over 3313474.40 frames. ], batch size: 83, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:09:34,229 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8393, 4.0412, 3.0574, 2.3817, 2.6087, 2.6424, 4.2166, 3.3815], device='cuda:6'), covar=tensor([0.2897, 0.0634, 0.1875, 0.3145, 0.3159, 0.2190, 0.0538, 0.1796], device='cuda:6'), in_proj_covar=tensor([0.0332, 0.0275, 0.0311, 0.0324, 0.0303, 0.0274, 0.0304, 0.0350], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 08:09:41,315 INFO [train.py:904] (6/8) Epoch 27, batch 2250, loss[loss=0.1585, simple_loss=0.2577, pruned_loss=0.02963, over 16726.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2543, pruned_loss=0.03988, over 3314151.82 frames. ], batch size: 62, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:09:56,594 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.643e+02 2.173e+02 2.503e+02 2.978e+02 4.988e+02, threshold=5.005e+02, percent-clipped=0.0 2023-05-02 08:10:28,077 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266186.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:10:48,231 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6164, 5.9958, 5.7412, 5.8621, 5.4178, 5.4249, 5.3703, 6.1286], device='cuda:6'), covar=tensor([0.1523, 0.1019, 0.1067, 0.0894, 0.0893, 0.0690, 0.1378, 0.0907], device='cuda:6'), in_proj_covar=tensor([0.0722, 0.0872, 0.0712, 0.0672, 0.0554, 0.0550, 0.0737, 0.0684], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 08:10:51,414 INFO [train.py:904] (6/8) Epoch 27, batch 2300, loss[loss=0.1629, simple_loss=0.2617, pruned_loss=0.03203, over 17124.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2544, pruned_loss=0.03988, over 3321148.50 frames. ], batch size: 49, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:11:06,853 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266214.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:11:23,570 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266226.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 08:11:26,475 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266228.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:11:34,360 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266234.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:11:59,649 INFO [train.py:904] (6/8) Epoch 27, batch 2350, loss[loss=0.1745, simple_loss=0.2563, pruned_loss=0.04637, over 16884.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2552, pruned_loss=0.0402, over 3328738.55 frames. ], batch size: 42, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:12:14,159 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.140e+02 2.442e+02 2.999e+02 5.112e+02, threshold=4.884e+02, percent-clipped=1.0 2023-05-02 08:12:22,989 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266270.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 08:12:28,004 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266273.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:12:28,148 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9600, 2.8583, 2.6192, 4.4322, 3.3257, 4.1035, 1.7867, 3.0063], device='cuda:6'), covar=tensor([0.1382, 0.0856, 0.1237, 0.0248, 0.0259, 0.0458, 0.1664, 0.0882], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0180, 0.0199, 0.0200, 0.0206, 0.0219, 0.0208, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 08:12:30,411 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266275.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:12:40,554 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 08:12:46,489 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266287.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 08:13:07,763 INFO [train.py:904] (6/8) Epoch 27, batch 2400, loss[loss=0.1575, simple_loss=0.2567, pruned_loss=0.02913, over 17039.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2555, pruned_loss=0.03983, over 3333858.83 frames. ], batch size: 50, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:13:18,687 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5260, 3.6035, 2.7641, 2.1875, 2.2627, 2.3224, 3.7328, 3.0709], device='cuda:6'), covar=tensor([0.2935, 0.0596, 0.1843, 0.3238, 0.2851, 0.2293, 0.0534, 0.1849], device='cuda:6'), in_proj_covar=tensor([0.0334, 0.0276, 0.0313, 0.0326, 0.0305, 0.0276, 0.0305, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 08:13:28,959 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266318.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:14:07,859 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266346.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:14:16,214 INFO [train.py:904] (6/8) Epoch 27, batch 2450, loss[loss=0.1598, simple_loss=0.2411, pruned_loss=0.03929, over 16611.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2568, pruned_loss=0.04006, over 3326027.98 frames. ], batch size: 76, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:14:18,407 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0821, 5.0948, 4.8359, 4.3212, 5.0011, 1.9201, 4.7054, 4.6909], device='cuda:6'), covar=tensor([0.0109, 0.0105, 0.0255, 0.0405, 0.0102, 0.2872, 0.0161, 0.0255], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0174, 0.0211, 0.0185, 0.0188, 0.0219, 0.0202, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 08:14:24,444 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8758, 4.8599, 4.6801, 4.1375, 4.8257, 1.8824, 4.5500, 4.4096], device='cuda:6'), covar=tensor([0.0157, 0.0147, 0.0246, 0.0387, 0.0131, 0.3014, 0.0192, 0.0276], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0174, 0.0211, 0.0185, 0.0188, 0.0219, 0.0202, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 08:14:31,081 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.203e+02 2.440e+02 2.935e+02 4.080e+02, threshold=4.880e+02, percent-clipped=0.0 2023-05-02 08:14:41,910 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266371.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:15:26,443 INFO [train.py:904] (6/8) Epoch 27, batch 2500, loss[loss=0.1432, simple_loss=0.2368, pruned_loss=0.02484, over 17193.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2564, pruned_loss=0.03939, over 3334718.02 frames. ], batch size: 46, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:15:31,555 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266407.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:15:31,630 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0937, 3.1682, 3.1115, 5.2057, 4.3501, 4.5014, 1.9431, 3.2771], device='cuda:6'), covar=tensor([0.1300, 0.0758, 0.1042, 0.0175, 0.0197, 0.0443, 0.1529, 0.0783], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0180, 0.0199, 0.0200, 0.0206, 0.0219, 0.0208, 0.0197], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 08:15:48,873 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266419.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:16:26,010 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-02 08:16:35,710 INFO [train.py:904] (6/8) Epoch 27, batch 2550, loss[loss=0.1686, simple_loss=0.2646, pruned_loss=0.03627, over 17155.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2563, pruned_loss=0.03946, over 3336522.99 frames. ], batch size: 48, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:16:51,124 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.440e+02 2.045e+02 2.342e+02 2.686e+02 5.745e+02, threshold=4.685e+02, percent-clipped=2.0 2023-05-02 08:16:52,108 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266464.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:17:45,250 INFO [train.py:904] (6/8) Epoch 27, batch 2600, loss[loss=0.1519, simple_loss=0.2509, pruned_loss=0.02644, over 17085.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.256, pruned_loss=0.0389, over 3336002.61 frames. ], batch size: 47, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:18:17,181 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266525.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:18:21,043 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266528.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:18:40,402 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 08:18:55,437 INFO [train.py:904] (6/8) Epoch 27, batch 2650, loss[loss=0.1574, simple_loss=0.2557, pruned_loss=0.02949, over 17108.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2561, pruned_loss=0.03858, over 3344684.28 frames. ], batch size: 47, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:19:11,568 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.008e+02 2.392e+02 2.897e+02 5.210e+02, threshold=4.785e+02, percent-clipped=1.0 2023-05-02 08:19:21,003 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266570.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:19:24,633 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266573.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:19:29,429 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266576.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:19:37,430 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266582.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 08:20:05,798 INFO [train.py:904] (6/8) Epoch 27, batch 2700, loss[loss=0.1509, simple_loss=0.2491, pruned_loss=0.02638, over 17126.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2567, pruned_loss=0.03844, over 3342692.71 frames. ], batch size: 47, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:20:06,393 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4880, 3.5623, 4.0294, 2.2481, 3.2024, 2.5155, 4.0276, 3.7675], device='cuda:6'), covar=tensor([0.0312, 0.0977, 0.0490, 0.2110, 0.0883, 0.1020, 0.0530, 0.1111], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0171, 0.0171, 0.0157, 0.0148, 0.0133, 0.0147, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 08:20:31,908 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266621.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:21:15,403 INFO [train.py:904] (6/8) Epoch 27, batch 2750, loss[loss=0.1712, simple_loss=0.2515, pruned_loss=0.04546, over 16224.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2569, pruned_loss=0.03801, over 3340193.68 frames. ], batch size: 165, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:21:29,197 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.625e+02 2.109e+02 2.416e+02 2.772e+02 5.063e+02, threshold=4.832e+02, percent-clipped=2.0 2023-05-02 08:21:51,262 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266679.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:22:22,656 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266702.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:22:23,525 INFO [train.py:904] (6/8) Epoch 27, batch 2800, loss[loss=0.1745, simple_loss=0.2691, pruned_loss=0.03997, over 17234.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2566, pruned_loss=0.03773, over 3328544.07 frames. ], batch size: 52, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:22:25,180 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266704.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:23:15,466 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266740.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:23:31,365 INFO [train.py:904] (6/8) Epoch 27, batch 2850, loss[loss=0.1788, simple_loss=0.2562, pruned_loss=0.05072, over 16917.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2547, pruned_loss=0.03757, over 3338881.90 frames. ], batch size: 116, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:23:41,900 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 08:23:48,197 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.557e+02 2.101e+02 2.361e+02 2.749e+02 4.923e+02, threshold=4.721e+02, percent-clipped=1.0 2023-05-02 08:23:49,937 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266765.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:24:14,124 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-05-02 08:24:41,243 INFO [train.py:904] (6/8) Epoch 27, batch 2900, loss[loss=0.1944, simple_loss=0.2602, pruned_loss=0.06435, over 16768.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2541, pruned_loss=0.03868, over 3324470.58 frames. ], batch size: 124, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:25:04,248 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266820.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:25:49,404 INFO [train.py:904] (6/8) Epoch 27, batch 2950, loss[loss=0.1426, simple_loss=0.2265, pruned_loss=0.02936, over 16839.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2542, pruned_loss=0.03913, over 3320848.25 frames. ], batch size: 42, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:26:04,411 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.609e+02 2.163e+02 2.599e+02 3.216e+02 6.007e+02, threshold=5.198e+02, percent-clipped=1.0 2023-05-02 08:26:14,736 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266870.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:26:29,698 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266882.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 08:26:58,264 INFO [train.py:904] (6/8) Epoch 27, batch 3000, loss[loss=0.1567, simple_loss=0.2545, pruned_loss=0.02941, over 17125.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2543, pruned_loss=0.03985, over 3311070.76 frames. ], batch size: 49, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:26:58,265 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 08:27:07,044 INFO [train.py:938] (6/8) Epoch 27, validation: loss=0.1336, simple_loss=0.2386, pruned_loss=0.01429, over 944034.00 frames. 2023-05-02 08:27:07,045 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 08:27:27,842 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266918.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:27:30,379 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266920.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:27:45,319 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266930.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 08:27:55,797 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3622, 4.3893, 4.7231, 4.6987, 4.7294, 4.4352, 4.4384, 4.3126], device='cuda:6'), covar=tensor([0.0396, 0.0640, 0.0377, 0.0411, 0.0517, 0.0434, 0.0801, 0.0715], device='cuda:6'), in_proj_covar=tensor([0.0442, 0.0498, 0.0480, 0.0443, 0.0530, 0.0509, 0.0586, 0.0405], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 08:28:09,506 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1225, 3.7527, 4.3566, 2.3112, 4.4876, 4.6983, 3.3490, 3.6015], device='cuda:6'), covar=tensor([0.0738, 0.0282, 0.0256, 0.1176, 0.0088, 0.0161, 0.0439, 0.0395], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0112, 0.0102, 0.0140, 0.0087, 0.0132, 0.0131, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 08:28:16,681 INFO [train.py:904] (6/8) Epoch 27, batch 3050, loss[loss=0.1549, simple_loss=0.2475, pruned_loss=0.03112, over 17058.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2548, pruned_loss=0.04016, over 3308875.52 frames. ], batch size: 50, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:28:30,546 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.662e+02 2.121e+02 2.455e+02 2.792e+02 5.832e+02, threshold=4.910e+02, percent-clipped=2.0 2023-05-02 08:28:55,898 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266981.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:29:24,924 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267002.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:29:25,704 INFO [train.py:904] (6/8) Epoch 27, batch 3100, loss[loss=0.1327, simple_loss=0.2239, pruned_loss=0.02074, over 16822.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2536, pruned_loss=0.0399, over 3308331.00 frames. ], batch size: 42, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:29:45,227 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6250, 4.6607, 4.9913, 4.9674, 5.0193, 4.6817, 4.6617, 4.5560], device='cuda:6'), covar=tensor([0.0426, 0.0775, 0.0489, 0.0478, 0.0583, 0.0590, 0.0990, 0.0596], device='cuda:6'), in_proj_covar=tensor([0.0444, 0.0500, 0.0482, 0.0444, 0.0531, 0.0510, 0.0589, 0.0406], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 08:30:11,860 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=267035.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:30:33,230 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267050.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:30:36,920 INFO [train.py:904] (6/8) Epoch 27, batch 3150, loss[loss=0.1392, simple_loss=0.2323, pruned_loss=0.02302, over 17232.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2529, pruned_loss=0.03956, over 3307960.86 frames. ], batch size: 44, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:30:45,781 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=267060.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:30:50,834 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.616e+02 2.137e+02 2.489e+02 2.954e+02 4.893e+02, threshold=4.979e+02, percent-clipped=1.0 2023-05-02 08:31:06,888 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8377, 2.5407, 2.5770, 4.1040, 3.3344, 4.0490, 1.6360, 2.8213], device='cuda:6'), covar=tensor([0.1465, 0.0838, 0.1236, 0.0197, 0.0204, 0.0414, 0.1706, 0.0967], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0181, 0.0199, 0.0202, 0.0207, 0.0220, 0.0209, 0.0199], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 08:31:10,146 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1696, 5.7651, 5.8613, 5.5060, 5.6447, 6.1935, 5.6424, 5.3353], device='cuda:6'), covar=tensor([0.0884, 0.1801, 0.2693, 0.2056, 0.2543, 0.0957, 0.1623, 0.2553], device='cuda:6'), in_proj_covar=tensor([0.0438, 0.0647, 0.0717, 0.0531, 0.0707, 0.0741, 0.0558, 0.0707], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 08:31:15,475 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=267082.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:31:37,685 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=267098.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:31:43,510 INFO [train.py:904] (6/8) Epoch 27, batch 3200, loss[loss=0.1622, simple_loss=0.2622, pruned_loss=0.03112, over 17137.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2517, pruned_loss=0.03885, over 3316245.53 frames. ], batch size: 48, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:32:08,040 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267120.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:32:39,093 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=267143.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:32:50,799 INFO [train.py:904] (6/8) Epoch 27, batch 3250, loss[loss=0.1769, simple_loss=0.261, pruned_loss=0.04641, over 16838.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2521, pruned_loss=0.0391, over 3316285.88 frames. ], batch size: 102, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:33:01,444 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=267159.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:33:05,714 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-02 08:33:07,327 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.439e+02 2.100e+02 2.558e+02 2.992e+02 5.304e+02, threshold=5.115e+02, percent-clipped=1.0 2023-05-02 08:33:11,482 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8092, 2.9464, 2.7501, 4.7521, 3.6594, 4.1363, 1.8350, 2.9954], device='cuda:6'), covar=tensor([0.1367, 0.0784, 0.1195, 0.0256, 0.0251, 0.0466, 0.1570, 0.0858], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0181, 0.0199, 0.0202, 0.0207, 0.0220, 0.0208, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 08:33:12,826 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267168.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:34:00,077 INFO [train.py:904] (6/8) Epoch 27, batch 3300, loss[loss=0.1839, simple_loss=0.2703, pruned_loss=0.04876, over 16303.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2531, pruned_loss=0.03943, over 3324544.53 frames. ], batch size: 165, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:34:39,330 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-05-02 08:34:40,615 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 08:35:07,427 INFO [train.py:904] (6/8) Epoch 27, batch 3350, loss[loss=0.1514, simple_loss=0.24, pruned_loss=0.03144, over 16801.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2539, pruned_loss=0.03944, over 3315200.84 frames. ], batch size: 42, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:35:08,962 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0999, 4.8146, 5.1184, 5.2989, 5.5578, 4.8550, 5.5022, 5.5335], device='cuda:6'), covar=tensor([0.2117, 0.1508, 0.2062, 0.0863, 0.0572, 0.0984, 0.0539, 0.0658], device='cuda:6'), in_proj_covar=tensor([0.0699, 0.0859, 0.0998, 0.0868, 0.0660, 0.0690, 0.0720, 0.0838], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 08:35:22,678 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 1.912e+02 2.198e+02 2.588e+02 3.700e+02, threshold=4.395e+02, percent-clipped=0.0 2023-05-02 08:35:28,608 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-02 08:35:38,223 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=267276.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:36:14,881 INFO [train.py:904] (6/8) Epoch 27, batch 3400, loss[loss=0.1654, simple_loss=0.2548, pruned_loss=0.03798, over 16738.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2535, pruned_loss=0.03925, over 3309496.25 frames. ], batch size: 57, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:36:39,559 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-02 08:36:54,659 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9445, 2.0402, 2.5202, 2.8818, 2.7679, 3.0341, 2.1550, 3.1790], device='cuda:6'), covar=tensor([0.0245, 0.0566, 0.0396, 0.0326, 0.0368, 0.0306, 0.0618, 0.0207], device='cuda:6'), in_proj_covar=tensor([0.0203, 0.0202, 0.0190, 0.0197, 0.0212, 0.0170, 0.0208, 0.0169], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 08:36:58,125 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267335.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:37:22,626 INFO [train.py:904] (6/8) Epoch 27, batch 3450, loss[loss=0.1737, simple_loss=0.2616, pruned_loss=0.04291, over 16663.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.252, pruned_loss=0.03836, over 3312480.96 frames. ], batch size: 62, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:37:31,843 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267360.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:37:36,673 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 2.041e+02 2.356e+02 2.660e+02 5.077e+02, threshold=4.713e+02, percent-clipped=1.0 2023-05-02 08:37:51,846 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9104, 2.0516, 2.5606, 2.9514, 2.7977, 3.4128, 2.3409, 3.4793], device='cuda:6'), covar=tensor([0.0305, 0.0609, 0.0381, 0.0376, 0.0396, 0.0212, 0.0561, 0.0176], device='cuda:6'), in_proj_covar=tensor([0.0202, 0.0202, 0.0190, 0.0196, 0.0211, 0.0169, 0.0208, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 08:38:03,485 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267383.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:38:30,154 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4090, 5.4147, 5.3403, 4.8010, 4.9156, 5.3633, 5.2583, 4.9708], device='cuda:6'), covar=tensor([0.0657, 0.0504, 0.0302, 0.0370, 0.1219, 0.0480, 0.0293, 0.0782], device='cuda:6'), in_proj_covar=tensor([0.0324, 0.0486, 0.0377, 0.0383, 0.0378, 0.0439, 0.0258, 0.0454], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 08:38:30,914 INFO [train.py:904] (6/8) Epoch 27, batch 3500, loss[loss=0.151, simple_loss=0.2405, pruned_loss=0.03074, over 16943.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.251, pruned_loss=0.03817, over 3324857.47 frames. ], batch size: 41, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:38:39,647 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267408.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:38:39,896 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0390, 3.0757, 2.7964, 2.8607, 3.2859, 2.9981, 3.6018, 3.4942], device='cuda:6'), covar=tensor([0.0156, 0.0415, 0.0509, 0.0474, 0.0316, 0.0408, 0.0280, 0.0293], device='cuda:6'), in_proj_covar=tensor([0.0237, 0.0249, 0.0238, 0.0239, 0.0252, 0.0251, 0.0251, 0.0249], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 08:38:51,365 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-02 08:39:19,830 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=267438.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:39:40,533 INFO [train.py:904] (6/8) Epoch 27, batch 3550, loss[loss=0.1497, simple_loss=0.2444, pruned_loss=0.02756, over 17130.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2499, pruned_loss=0.03791, over 3321744.52 frames. ], batch size: 47, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:39:41,963 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=267454.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:39:48,089 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5867, 3.8591, 4.2333, 2.6335, 3.4215, 2.8262, 3.9511, 4.0401], device='cuda:6'), covar=tensor([0.0380, 0.1044, 0.0443, 0.1942, 0.0832, 0.0943, 0.0793, 0.1127], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0174, 0.0171, 0.0158, 0.0149, 0.0133, 0.0148, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 08:39:53,961 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.045e+02 2.366e+02 2.928e+02 5.405e+02, threshold=4.732e+02, percent-clipped=1.0 2023-05-02 08:40:37,345 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=267494.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:40:48,162 INFO [train.py:904] (6/8) Epoch 27, batch 3600, loss[loss=0.1662, simple_loss=0.2487, pruned_loss=0.04184, over 16471.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2489, pruned_loss=0.03743, over 3326719.15 frames. ], batch size: 146, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:42:00,665 INFO [train.py:904] (6/8) Epoch 27, batch 3650, loss[loss=0.1799, simple_loss=0.246, pruned_loss=0.0569, over 16768.00 frames. ], tot_loss[loss=0.1621, simple_loss=0.2484, pruned_loss=0.03794, over 3316055.15 frames. ], batch size: 83, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:42:03,576 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=267555.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:42:16,686 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.290e+02 2.009e+02 2.392e+02 2.845e+02 4.698e+02, threshold=4.785e+02, percent-clipped=1.0 2023-05-02 08:42:36,602 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267576.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:43:14,275 INFO [train.py:904] (6/8) Epoch 27, batch 3700, loss[loss=0.1539, simple_loss=0.2311, pruned_loss=0.03833, over 16800.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2472, pruned_loss=0.03943, over 3285850.27 frames. ], batch size: 83, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:43:45,316 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267624.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:44:19,779 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0374, 3.0671, 2.7593, 4.4962, 3.7344, 4.2695, 1.7834, 3.1310], device='cuda:6'), covar=tensor([0.1272, 0.0667, 0.1104, 0.0206, 0.0236, 0.0398, 0.1540, 0.0837], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0180, 0.0199, 0.0202, 0.0208, 0.0220, 0.0209, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 08:44:27,015 INFO [train.py:904] (6/8) Epoch 27, batch 3750, loss[loss=0.1799, simple_loss=0.2475, pruned_loss=0.05614, over 16796.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2479, pruned_loss=0.04085, over 3279184.47 frames. ], batch size: 116, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:44:42,769 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.122e+02 2.445e+02 2.969e+02 5.138e+02, threshold=4.890e+02, percent-clipped=1.0 2023-05-02 08:45:40,197 INFO [train.py:904] (6/8) Epoch 27, batch 3800, loss[loss=0.1853, simple_loss=0.257, pruned_loss=0.05686, over 16792.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.249, pruned_loss=0.04228, over 3284819.18 frames. ], batch size: 124, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:45:41,596 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-02 08:46:21,663 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8555, 2.6108, 2.4964, 3.8505, 3.1978, 3.9337, 1.6178, 2.8534], device='cuda:6'), covar=tensor([0.1383, 0.0725, 0.1229, 0.0223, 0.0152, 0.0372, 0.1621, 0.0884], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0180, 0.0198, 0.0202, 0.0207, 0.0219, 0.0208, 0.0197], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 08:46:31,892 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267738.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:46:49,344 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4299, 3.0969, 3.4477, 1.8616, 3.5072, 3.5277, 3.0603, 2.7292], device='cuda:6'), covar=tensor([0.0816, 0.0300, 0.0233, 0.1280, 0.0143, 0.0274, 0.0422, 0.0477], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0112, 0.0102, 0.0139, 0.0087, 0.0132, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 08:46:52,904 INFO [train.py:904] (6/8) Epoch 27, batch 3850, loss[loss=0.1587, simple_loss=0.2386, pruned_loss=0.03947, over 16789.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2496, pruned_loss=0.04305, over 3267205.87 frames. ], batch size: 102, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:46:54,314 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267754.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:47:08,877 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.727e+02 2.090e+02 2.451e+02 2.947e+02 6.738e+02, threshold=4.901e+02, percent-clipped=4.0 2023-05-02 08:47:13,731 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-02 08:47:39,701 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267786.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:47:44,074 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5102, 4.5695, 4.7324, 4.5210, 4.6561, 5.1621, 4.6806, 4.4294], device='cuda:6'), covar=tensor([0.1641, 0.2190, 0.2219, 0.2313, 0.2298, 0.1101, 0.1612, 0.2441], device='cuda:6'), in_proj_covar=tensor([0.0437, 0.0646, 0.0710, 0.0527, 0.0703, 0.0737, 0.0552, 0.0701], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 08:48:02,760 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267802.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:48:03,541 INFO [train.py:904] (6/8) Epoch 27, batch 3900, loss[loss=0.1624, simple_loss=0.2388, pruned_loss=0.04303, over 16854.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2496, pruned_loss=0.04358, over 3259943.61 frames. ], batch size: 96, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:48:10,234 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0409, 3.0995, 3.3302, 2.1569, 2.9253, 2.3200, 3.6164, 3.4881], device='cuda:6'), covar=tensor([0.0251, 0.0960, 0.0641, 0.1978, 0.0863, 0.1058, 0.0458, 0.0831], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0173, 0.0171, 0.0157, 0.0149, 0.0133, 0.0147, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 08:49:11,705 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=267850.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:49:14,858 INFO [train.py:904] (6/8) Epoch 27, batch 3950, loss[loss=0.164, simple_loss=0.24, pruned_loss=0.04397, over 16898.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2495, pruned_loss=0.04423, over 3266679.18 frames. ], batch size: 96, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 08:49:32,326 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.702e+02 2.233e+02 2.530e+02 3.152e+02 5.510e+02, threshold=5.059e+02, percent-clipped=1.0 2023-05-02 08:49:36,825 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0651, 5.4085, 5.2171, 5.1925, 4.9334, 4.8844, 4.8976, 5.5392], device='cuda:6'), covar=tensor([0.1414, 0.0919, 0.0990, 0.0899, 0.0851, 0.0966, 0.1247, 0.0913], device='cuda:6'), in_proj_covar=tensor([0.0735, 0.0889, 0.0723, 0.0685, 0.0565, 0.0560, 0.0752, 0.0699], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 08:50:26,834 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3997, 5.4177, 5.2750, 4.8692, 4.9260, 5.3150, 5.1625, 5.0576], device='cuda:6'), covar=tensor([0.0535, 0.0370, 0.0252, 0.0285, 0.0972, 0.0332, 0.0312, 0.0565], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0482, 0.0374, 0.0380, 0.0376, 0.0436, 0.0256, 0.0451], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 08:50:28,769 INFO [train.py:904] (6/8) Epoch 27, batch 4000, loss[loss=0.1771, simple_loss=0.2563, pruned_loss=0.04898, over 17302.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2493, pruned_loss=0.04439, over 3279697.13 frames. ], batch size: 52, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 08:50:59,104 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-02 08:51:08,751 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6168, 3.7051, 3.4146, 3.1442, 3.2851, 3.5871, 3.3182, 3.4257], device='cuda:6'), covar=tensor([0.0577, 0.0530, 0.0292, 0.0294, 0.0533, 0.0432, 0.1895, 0.0444], device='cuda:6'), in_proj_covar=tensor([0.0321, 0.0481, 0.0373, 0.0380, 0.0375, 0.0435, 0.0256, 0.0450], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 08:51:42,545 INFO [train.py:904] (6/8) Epoch 27, batch 4050, loss[loss=0.167, simple_loss=0.2509, pruned_loss=0.04153, over 17091.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2503, pruned_loss=0.04394, over 3281611.04 frames. ], batch size: 53, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 08:51:58,299 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 1.841e+02 2.143e+02 2.539e+02 4.386e+02, threshold=4.286e+02, percent-clipped=0.0 2023-05-02 08:52:59,882 INFO [train.py:904] (6/8) Epoch 27, batch 4100, loss[loss=0.1992, simple_loss=0.2907, pruned_loss=0.05382, over 15340.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2527, pruned_loss=0.0438, over 3275698.47 frames. ], batch size: 190, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 08:54:16,164 INFO [train.py:904] (6/8) Epoch 27, batch 4150, loss[loss=0.2752, simple_loss=0.3326, pruned_loss=0.1089, over 11270.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2594, pruned_loss=0.04595, over 3242983.67 frames. ], batch size: 248, lr: 2.49e-03, grad_scale: 16.0 2023-05-02 08:54:33,173 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 2.031e+02 2.466e+02 3.151e+02 6.245e+02, threshold=4.932e+02, percent-clipped=8.0 2023-05-02 08:54:57,703 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4376, 3.4023, 3.6606, 2.1167, 3.0557, 2.4273, 3.7675, 3.6905], device='cuda:6'), covar=tensor([0.0249, 0.0952, 0.0595, 0.2189, 0.0954, 0.0992, 0.0657, 0.1081], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0172, 0.0170, 0.0156, 0.0148, 0.0132, 0.0147, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 08:54:59,656 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-02 08:55:21,388 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3386, 4.4156, 4.2198, 3.8934, 3.9384, 4.3308, 3.9829, 4.0909], device='cuda:6'), covar=tensor([0.0527, 0.0468, 0.0255, 0.0277, 0.0675, 0.0411, 0.0772, 0.0551], device='cuda:6'), in_proj_covar=tensor([0.0319, 0.0477, 0.0371, 0.0377, 0.0372, 0.0432, 0.0254, 0.0446], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 08:55:32,119 INFO [train.py:904] (6/8) Epoch 27, batch 4200, loss[loss=0.2003, simple_loss=0.2897, pruned_loss=0.05547, over 16612.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2659, pruned_loss=0.04753, over 3194424.61 frames. ], batch size: 76, lr: 2.49e-03, grad_scale: 16.0 2023-05-02 08:56:41,017 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=268150.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:56:44,662 INFO [train.py:904] (6/8) Epoch 27, batch 4250, loss[loss=0.1525, simple_loss=0.2579, pruned_loss=0.02361, over 16856.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2694, pruned_loss=0.0473, over 3181167.56 frames. ], batch size: 102, lr: 2.49e-03, grad_scale: 16.0 2023-05-02 08:57:00,873 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.686e+02 2.148e+02 2.458e+02 3.113e+02 4.874e+02, threshold=4.916e+02, percent-clipped=0.0 2023-05-02 08:57:50,364 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=268198.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:57:57,375 INFO [train.py:904] (6/8) Epoch 27, batch 4300, loss[loss=0.1953, simple_loss=0.288, pruned_loss=0.05124, over 17227.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2705, pruned_loss=0.04642, over 3180932.73 frames. ], batch size: 44, lr: 2.49e-03, grad_scale: 16.0 2023-05-02 08:58:08,378 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3752, 5.3616, 5.6653, 5.6372, 5.7575, 5.3843, 5.2757, 5.0590], device='cuda:6'), covar=tensor([0.0267, 0.0371, 0.0317, 0.0378, 0.0418, 0.0361, 0.0894, 0.0498], device='cuda:6'), in_proj_covar=tensor([0.0436, 0.0493, 0.0475, 0.0437, 0.0522, 0.0502, 0.0577, 0.0399], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 08:58:10,836 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=268212.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:59:12,216 INFO [train.py:904] (6/8) Epoch 27, batch 4350, loss[loss=0.1935, simple_loss=0.2804, pruned_loss=0.05336, over 16675.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2736, pruned_loss=0.04737, over 3181882.79 frames. ], batch size: 134, lr: 2.49e-03, grad_scale: 16.0 2023-05-02 08:59:27,930 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.498e+02 2.120e+02 2.432e+02 2.957e+02 3.904e+02, threshold=4.863e+02, percent-clipped=0.0 2023-05-02 08:59:43,167 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=268273.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 08:59:45,625 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=268275.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:00:27,002 INFO [train.py:904] (6/8) Epoch 27, batch 4400, loss[loss=0.1799, simple_loss=0.2683, pruned_loss=0.04579, over 16987.00 frames. ], tot_loss[loss=0.1869, simple_loss=0.2763, pruned_loss=0.0488, over 3178992.05 frames. ], batch size: 41, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:00:56,972 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=5.19 vs. limit=5.0 2023-05-02 09:01:14,955 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=268336.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 09:01:37,964 INFO [train.py:904] (6/8) Epoch 27, batch 4450, loss[loss=0.2069, simple_loss=0.2991, pruned_loss=0.05732, over 16463.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2799, pruned_loss=0.05003, over 3201553.15 frames. ], batch size: 35, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:01:55,122 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 1.898e+02 2.329e+02 2.751e+02 5.089e+02, threshold=4.658e+02, percent-clipped=1.0 2023-05-02 09:01:58,034 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-05-02 09:02:50,810 INFO [train.py:904] (6/8) Epoch 27, batch 4500, loss[loss=0.1966, simple_loss=0.2817, pruned_loss=0.05579, over 16361.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2801, pruned_loss=0.05071, over 3198325.13 frames. ], batch size: 35, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:03:22,147 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4434, 3.6680, 3.6451, 2.0965, 3.0316, 2.3127, 3.7242, 3.9035], device='cuda:6'), covar=tensor([0.0232, 0.0764, 0.0648, 0.2296, 0.0949, 0.1085, 0.0615, 0.0830], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0171, 0.0170, 0.0156, 0.0147, 0.0132, 0.0146, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 09:04:03,680 INFO [train.py:904] (6/8) Epoch 27, batch 4550, loss[loss=0.1902, simple_loss=0.2845, pruned_loss=0.0479, over 16877.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2808, pruned_loss=0.05137, over 3214246.93 frames. ], batch size: 96, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:04:20,726 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 1.908e+02 2.230e+02 2.489e+02 1.268e+03, threshold=4.461e+02, percent-clipped=3.0 2023-05-02 09:04:59,868 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2634, 2.9292, 3.2681, 1.8202, 3.3507, 3.3583, 2.7906, 2.6206], device='cuda:6'), covar=tensor([0.0899, 0.0370, 0.0206, 0.1252, 0.0123, 0.0194, 0.0494, 0.0535], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0112, 0.0102, 0.0140, 0.0087, 0.0132, 0.0131, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 09:05:15,490 INFO [train.py:904] (6/8) Epoch 27, batch 4600, loss[loss=0.1812, simple_loss=0.283, pruned_loss=0.03967, over 16880.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2819, pruned_loss=0.05181, over 3208553.57 frames. ], batch size: 96, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:05:23,671 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-02 09:05:55,441 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9598, 4.9736, 5.2860, 5.2557, 5.3359, 4.9983, 4.9580, 4.6500], device='cuda:6'), covar=tensor([0.0275, 0.0414, 0.0294, 0.0352, 0.0365, 0.0339, 0.0832, 0.0490], device='cuda:6'), in_proj_covar=tensor([0.0430, 0.0484, 0.0467, 0.0431, 0.0516, 0.0494, 0.0569, 0.0394], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 09:06:23,483 INFO [train.py:904] (6/8) Epoch 27, batch 4650, loss[loss=0.1668, simple_loss=0.2574, pruned_loss=0.03815, over 16456.00 frames. ], tot_loss[loss=0.1921, simple_loss=0.2807, pruned_loss=0.0518, over 3208078.15 frames. ], batch size: 75, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:06:40,831 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 1.843e+02 2.064e+02 2.500e+02 6.008e+02, threshold=4.128e+02, percent-clipped=2.0 2023-05-02 09:06:45,258 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=268568.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 09:07:32,850 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=268601.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:07:34,884 INFO [train.py:904] (6/8) Epoch 27, batch 4700, loss[loss=0.188, simple_loss=0.2738, pruned_loss=0.05111, over 16904.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2782, pruned_loss=0.05089, over 3198443.43 frames. ], batch size: 109, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:08:15,779 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=268631.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 09:08:17,714 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1894, 5.2156, 5.0524, 4.6376, 4.6513, 5.0982, 4.9883, 4.8103], device='cuda:6'), covar=tensor([0.0601, 0.0553, 0.0307, 0.0324, 0.1154, 0.0646, 0.0327, 0.0661], device='cuda:6'), in_proj_covar=tensor([0.0311, 0.0465, 0.0362, 0.0368, 0.0364, 0.0420, 0.0248, 0.0434], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 09:08:45,864 INFO [train.py:904] (6/8) Epoch 27, batch 4750, loss[loss=0.1481, simple_loss=0.2369, pruned_loss=0.02967, over 16932.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2736, pruned_loss=0.04875, over 3199295.35 frames. ], batch size: 109, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:09:00,233 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=268662.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:09:04,566 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 1.855e+02 2.157e+02 2.478e+02 7.312e+02, threshold=4.313e+02, percent-clipped=3.0 2023-05-02 09:09:31,904 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9303, 5.1911, 5.0065, 5.0261, 4.7353, 4.7109, 4.5744, 5.2834], device='cuda:6'), covar=tensor([0.1193, 0.0823, 0.0865, 0.0767, 0.0796, 0.0918, 0.1243, 0.0904], device='cuda:6'), in_proj_covar=tensor([0.0706, 0.0856, 0.0696, 0.0657, 0.0543, 0.0538, 0.0720, 0.0673], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 09:09:59,302 INFO [train.py:904] (6/8) Epoch 27, batch 4800, loss[loss=0.1554, simple_loss=0.2442, pruned_loss=0.0333, over 17249.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2698, pruned_loss=0.04673, over 3201200.90 frames. ], batch size: 52, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:11:14,169 INFO [train.py:904] (6/8) Epoch 27, batch 4850, loss[loss=0.1691, simple_loss=0.266, pruned_loss=0.0361, over 16939.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2707, pruned_loss=0.04594, over 3188256.46 frames. ], batch size: 109, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:11:31,499 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 1.812e+02 2.075e+02 2.444e+02 6.308e+02, threshold=4.151e+02, percent-clipped=1.0 2023-05-02 09:12:27,702 INFO [train.py:904] (6/8) Epoch 27, batch 4900, loss[loss=0.1632, simple_loss=0.2519, pruned_loss=0.03728, over 12489.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2697, pruned_loss=0.04449, over 3184008.30 frames. ], batch size: 246, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:12:56,960 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2706, 4.0200, 3.9824, 2.4950, 3.5435, 4.0129, 3.5504, 2.2553], device='cuda:6'), covar=tensor([0.0582, 0.0047, 0.0048, 0.0435, 0.0109, 0.0098, 0.0104, 0.0462], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0089, 0.0090, 0.0135, 0.0101, 0.0114, 0.0097, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 09:13:37,966 INFO [train.py:904] (6/8) Epoch 27, batch 4950, loss[loss=0.1792, simple_loss=0.2748, pruned_loss=0.04178, over 16425.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2691, pruned_loss=0.04373, over 3189763.08 frames. ], batch size: 75, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:13:54,431 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 1.970e+02 2.304e+02 2.682e+02 4.795e+02, threshold=4.609e+02, percent-clipped=2.0 2023-05-02 09:13:59,162 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=268868.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 09:14:36,940 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=268894.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:14:50,662 INFO [train.py:904] (6/8) Epoch 27, batch 5000, loss[loss=0.2065, simple_loss=0.2921, pruned_loss=0.06043, over 12454.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2709, pruned_loss=0.04416, over 3186833.08 frames. ], batch size: 247, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:15:09,314 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=268916.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:15:29,377 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=268931.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:16:00,154 INFO [train.py:904] (6/8) Epoch 27, batch 5050, loss[loss=0.1845, simple_loss=0.2747, pruned_loss=0.04718, over 12367.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2717, pruned_loss=0.04414, over 3193656.77 frames. ], batch size: 247, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:16:03,762 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=268955.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:16:05,780 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=268957.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:16:17,236 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 1.970e+02 2.360e+02 2.844e+02 4.584e+02, threshold=4.719e+02, percent-clipped=0.0 2023-05-02 09:16:37,071 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=268979.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:16:54,181 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=268991.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:17:11,192 INFO [train.py:904] (6/8) Epoch 27, batch 5100, loss[loss=0.1703, simple_loss=0.2624, pruned_loss=0.03912, over 16864.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2702, pruned_loss=0.0438, over 3189526.18 frames. ], batch size: 116, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:17:26,911 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9044, 2.1812, 2.2474, 3.4762, 2.0372, 2.4250, 2.2503, 2.2839], device='cuda:6'), covar=tensor([0.1602, 0.3620, 0.3128, 0.0677, 0.4301, 0.2701, 0.3990, 0.3356], device='cuda:6'), in_proj_covar=tensor([0.0421, 0.0474, 0.0384, 0.0337, 0.0446, 0.0541, 0.0444, 0.0555], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 09:17:40,298 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 09:18:12,970 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.27 vs. limit=5.0 2023-05-02 09:18:22,950 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=269052.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:18:23,596 INFO [train.py:904] (6/8) Epoch 27, batch 5150, loss[loss=0.1657, simple_loss=0.2575, pruned_loss=0.03692, over 16570.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2704, pruned_loss=0.04329, over 3170557.03 frames. ], batch size: 62, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:18:41,485 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 1.963e+02 2.152e+02 2.508e+02 3.926e+02, threshold=4.304e+02, percent-clipped=0.0 2023-05-02 09:19:07,817 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5590, 3.4572, 3.9189, 1.9589, 4.0323, 4.0516, 3.1022, 2.9207], device='cuda:6'), covar=tensor([0.0854, 0.0273, 0.0151, 0.1287, 0.0073, 0.0139, 0.0395, 0.0540], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0112, 0.0101, 0.0141, 0.0087, 0.0131, 0.0131, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 09:19:36,079 INFO [train.py:904] (6/8) Epoch 27, batch 5200, loss[loss=0.1722, simple_loss=0.2568, pruned_loss=0.04375, over 16550.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2691, pruned_loss=0.04266, over 3182563.07 frames. ], batch size: 75, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:19:40,738 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5008, 5.7902, 5.5095, 5.6443, 5.2895, 5.2548, 5.1683, 5.9143], device='cuda:6'), covar=tensor([0.1347, 0.0882, 0.1043, 0.0788, 0.0957, 0.0676, 0.1292, 0.0905], device='cuda:6'), in_proj_covar=tensor([0.0710, 0.0862, 0.0703, 0.0662, 0.0547, 0.0543, 0.0726, 0.0677], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 09:20:05,696 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8100, 2.6065, 2.4725, 3.9252, 2.5035, 3.8951, 1.5866, 2.7842], device='cuda:6'), covar=tensor([0.1354, 0.0843, 0.1217, 0.0157, 0.0189, 0.0392, 0.1736, 0.0859], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0179, 0.0199, 0.0200, 0.0206, 0.0217, 0.0208, 0.0197], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 09:20:11,586 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-02 09:20:47,202 INFO [train.py:904] (6/8) Epoch 27, batch 5250, loss[loss=0.1865, simple_loss=0.2726, pruned_loss=0.05017, over 12468.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2667, pruned_loss=0.04245, over 3187016.38 frames. ], batch size: 246, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:21:04,386 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.499e+02 1.922e+02 2.287e+02 2.667e+02 4.356e+02, threshold=4.574e+02, percent-clipped=2.0 2023-05-02 09:22:00,557 INFO [train.py:904] (6/8) Epoch 27, batch 5300, loss[loss=0.1583, simple_loss=0.2516, pruned_loss=0.03255, over 16307.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2632, pruned_loss=0.04124, over 3190161.27 frames. ], batch size: 165, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:22:55,408 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=269241.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:23:08,881 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=269250.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:23:12,961 INFO [train.py:904] (6/8) Epoch 27, batch 5350, loss[loss=0.1709, simple_loss=0.2667, pruned_loss=0.03757, over 16723.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2613, pruned_loss=0.04062, over 3196821.35 frames. ], batch size: 134, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:23:18,871 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=269257.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:23:29,780 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 2.050e+02 2.380e+02 2.740e+02 5.067e+02, threshold=4.761e+02, percent-clipped=1.0 2023-05-02 09:24:22,206 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7879, 3.1033, 3.2978, 2.1124, 2.8538, 2.2301, 3.3316, 3.4285], device='cuda:6'), covar=tensor([0.0331, 0.0832, 0.0624, 0.2071, 0.0935, 0.1030, 0.0621, 0.0910], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0171, 0.0170, 0.0156, 0.0148, 0.0132, 0.0146, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 09:24:24,492 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=269302.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:24:25,136 INFO [train.py:904] (6/8) Epoch 27, batch 5400, loss[loss=0.1874, simple_loss=0.2807, pruned_loss=0.04701, over 16574.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2636, pruned_loss=0.04129, over 3187047.24 frames. ], batch size: 68, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:24:28,455 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=269305.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:24:51,985 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.45 vs. limit=5.0 2023-05-02 09:25:14,088 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4780, 3.3138, 3.8181, 1.8592, 3.9637, 3.9382, 2.9542, 2.8826], device='cuda:6'), covar=tensor([0.0884, 0.0332, 0.0178, 0.1344, 0.0070, 0.0154, 0.0485, 0.0531], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0112, 0.0102, 0.0141, 0.0087, 0.0131, 0.0131, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 09:25:31,283 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=269347.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:25:41,594 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=269352.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:25:42,375 INFO [train.py:904] (6/8) Epoch 27, batch 5450, loss[loss=0.1779, simple_loss=0.2696, pruned_loss=0.04313, over 16852.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2663, pruned_loss=0.04215, over 3188713.85 frames. ], batch size: 116, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:26:01,148 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.011e+02 2.409e+02 3.005e+02 5.920e+02, threshold=4.819e+02, percent-clipped=1.0 2023-05-02 09:27:00,964 INFO [train.py:904] (6/8) Epoch 27, batch 5500, loss[loss=0.1916, simple_loss=0.2831, pruned_loss=0.05007, over 16710.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2733, pruned_loss=0.04636, over 3155613.23 frames. ], batch size: 89, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:27:16,718 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=269413.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:28:00,736 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=269441.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:28:19,421 INFO [train.py:904] (6/8) Epoch 27, batch 5550, loss[loss=0.2208, simple_loss=0.3045, pruned_loss=0.06855, over 15359.00 frames. ], tot_loss[loss=0.1921, simple_loss=0.2807, pruned_loss=0.0518, over 3118904.08 frames. ], batch size: 190, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:28:22,053 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4839, 2.5064, 2.4141, 4.1867, 2.4215, 2.9416, 2.5108, 2.7228], device='cuda:6'), covar=tensor([0.1278, 0.3290, 0.2910, 0.0511, 0.3723, 0.2266, 0.3327, 0.3123], device='cuda:6'), in_proj_covar=tensor([0.0419, 0.0470, 0.0382, 0.0335, 0.0444, 0.0539, 0.0441, 0.0551], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 09:28:26,554 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4218, 3.0085, 2.6307, 2.3906, 2.3836, 2.2010, 2.9677, 2.8747], device='cuda:6'), covar=tensor([0.2313, 0.0692, 0.1686, 0.2459, 0.2356, 0.2479, 0.0579, 0.1356], device='cuda:6'), in_proj_covar=tensor([0.0333, 0.0275, 0.0313, 0.0326, 0.0304, 0.0274, 0.0304, 0.0350], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 09:28:38,497 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.260e+02 3.053e+02 3.491e+02 4.207e+02 9.161e+02, threshold=6.983e+02, percent-clipped=6.0 2023-05-02 09:29:37,475 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=269502.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 09:29:38,093 INFO [train.py:904] (6/8) Epoch 27, batch 5600, loss[loss=0.2877, simple_loss=0.3432, pruned_loss=0.1161, over 11211.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2854, pruned_loss=0.05582, over 3081983.64 frames. ], batch size: 248, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:30:56,853 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=269550.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:31:01,875 INFO [train.py:904] (6/8) Epoch 27, batch 5650, loss[loss=0.207, simple_loss=0.3023, pruned_loss=0.05583, over 16866.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2907, pruned_loss=0.05932, over 3079861.81 frames. ], batch size: 42, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:31:06,704 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.88 vs. limit=5.0 2023-05-02 09:31:20,024 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.426e+02 3.478e+02 4.295e+02 5.160e+02 1.255e+03, threshold=8.591e+02, percent-clipped=5.0 2023-05-02 09:31:39,661 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.57 vs. limit=5.0 2023-05-02 09:32:11,909 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=269597.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:32:13,071 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=269598.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:32:20,977 INFO [train.py:904] (6/8) Epoch 27, batch 5700, loss[loss=0.2657, simple_loss=0.3243, pruned_loss=0.1035, over 11451.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2933, pruned_loss=0.06172, over 3045199.75 frames. ], batch size: 248, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:32:49,177 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.77 vs. limit=5.0 2023-05-02 09:32:54,651 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1024, 4.1759, 4.4277, 4.3994, 4.4310, 4.1658, 4.1297, 4.1658], device='cuda:6'), covar=tensor([0.0381, 0.0636, 0.0408, 0.0440, 0.0454, 0.0491, 0.1019, 0.0575], device='cuda:6'), in_proj_covar=tensor([0.0429, 0.0480, 0.0466, 0.0428, 0.0514, 0.0493, 0.0568, 0.0394], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 09:33:30,901 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3595, 3.0457, 2.6065, 2.3443, 2.3561, 2.1925, 2.9708, 2.8253], device='cuda:6'), covar=tensor([0.2419, 0.0801, 0.1718, 0.2839, 0.2612, 0.2582, 0.0576, 0.1481], device='cuda:6'), in_proj_covar=tensor([0.0334, 0.0275, 0.0313, 0.0327, 0.0305, 0.0274, 0.0305, 0.0350], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 09:33:31,971 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=269647.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:33:41,283 INFO [train.py:904] (6/8) Epoch 27, batch 5750, loss[loss=0.1931, simple_loss=0.2874, pruned_loss=0.04943, over 16919.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2961, pruned_loss=0.06377, over 3007856.45 frames. ], batch size: 90, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:33:59,181 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.119e+02 2.911e+02 3.444e+02 4.109e+02 8.393e+02, threshold=6.889e+02, percent-clipped=0.0 2023-05-02 09:34:50,308 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=269695.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:35:02,093 INFO [train.py:904] (6/8) Epoch 27, batch 5800, loss[loss=0.1854, simple_loss=0.2781, pruned_loss=0.04633, over 16417.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2953, pruned_loss=0.06221, over 3004266.76 frames. ], batch size: 146, lr: 2.49e-03, grad_scale: 4.0 2023-05-02 09:35:11,570 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=269708.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:36:21,258 INFO [train.py:904] (6/8) Epoch 27, batch 5850, loss[loss=0.237, simple_loss=0.3022, pruned_loss=0.08594, over 11471.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2928, pruned_loss=0.06064, over 3003207.13 frames. ], batch size: 248, lr: 2.49e-03, grad_scale: 4.0 2023-05-02 09:36:40,939 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.068e+02 3.085e+02 3.568e+02 4.510e+02 7.349e+02, threshold=7.136e+02, percent-clipped=2.0 2023-05-02 09:36:53,579 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=269773.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:36:57,211 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8915, 2.7179, 2.5941, 1.9534, 2.5842, 2.6741, 2.5701, 1.9941], device='cuda:6'), covar=tensor([0.0494, 0.0097, 0.0095, 0.0416, 0.0144, 0.0157, 0.0134, 0.0416], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0090, 0.0091, 0.0137, 0.0102, 0.0116, 0.0098, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 09:36:57,460 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.24 vs. limit=5.0 2023-05-02 09:37:34,923 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=269797.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 09:37:45,792 INFO [train.py:904] (6/8) Epoch 27, batch 5900, loss[loss=0.1884, simple_loss=0.2717, pruned_loss=0.05259, over 15473.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2923, pruned_loss=0.05991, over 3034597.69 frames. ], batch size: 191, lr: 2.49e-03, grad_scale: 4.0 2023-05-02 09:38:38,180 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=269834.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 09:38:40,739 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=269836.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:39:06,822 INFO [train.py:904] (6/8) Epoch 27, batch 5950, loss[loss=0.2089, simple_loss=0.2981, pruned_loss=0.05986, over 16403.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2925, pruned_loss=0.05804, over 3054746.11 frames. ], batch size: 146, lr: 2.49e-03, grad_scale: 4.0 2023-05-02 09:39:27,598 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.704e+02 3.273e+02 4.077e+02 6.463e+02, threshold=6.547e+02, percent-clipped=0.0 2023-05-02 09:39:33,932 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8169, 4.0735, 3.1010, 2.5364, 2.9248, 2.6116, 4.4831, 3.6796], device='cuda:6'), covar=tensor([0.2943, 0.0711, 0.1834, 0.2714, 0.2516, 0.2062, 0.0444, 0.1313], device='cuda:6'), in_proj_covar=tensor([0.0334, 0.0275, 0.0312, 0.0326, 0.0304, 0.0274, 0.0304, 0.0350], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 09:40:17,653 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=269897.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:40:17,806 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=269897.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:40:24,914 INFO [train.py:904] (6/8) Epoch 27, batch 6000, loss[loss=0.2048, simple_loss=0.2927, pruned_loss=0.05848, over 15390.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2921, pruned_loss=0.05813, over 3053831.62 frames. ], batch size: 191, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:40:24,914 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 09:40:35,105 INFO [train.py:938] (6/8) Epoch 27, validation: loss=0.148, simple_loss=0.2603, pruned_loss=0.01783, over 944034.00 frames. 2023-05-02 09:40:35,106 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 09:41:15,698 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1793, 5.7245, 5.9532, 5.6775, 5.7035, 6.2400, 5.6792, 5.4623], device='cuda:6'), covar=tensor([0.0792, 0.1741, 0.2298, 0.1754, 0.2239, 0.0917, 0.1689, 0.2369], device='cuda:6'), in_proj_covar=tensor([0.0424, 0.0627, 0.0688, 0.0510, 0.0682, 0.0715, 0.0538, 0.0685], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 09:41:37,578 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=269945.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:41:52,113 INFO [train.py:904] (6/8) Epoch 27, batch 6050, loss[loss=0.1878, simple_loss=0.2794, pruned_loss=0.04813, over 16743.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2906, pruned_loss=0.05766, over 3058097.56 frames. ], batch size: 134, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:42:12,243 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.929e+02 2.707e+02 3.068e+02 3.739e+02 6.756e+02, threshold=6.136e+02, percent-clipped=2.0 2023-05-02 09:43:13,692 INFO [train.py:904] (6/8) Epoch 27, batch 6100, loss[loss=0.1848, simple_loss=0.2791, pruned_loss=0.04525, over 16685.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.2901, pruned_loss=0.05683, over 3058725.21 frames. ], batch size: 134, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:43:23,136 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=270008.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:43:42,255 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-02 09:44:33,190 INFO [train.py:904] (6/8) Epoch 27, batch 6150, loss[loss=0.184, simple_loss=0.2684, pruned_loss=0.04981, over 16427.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.288, pruned_loss=0.05647, over 3050927.65 frames. ], batch size: 68, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:44:37,804 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=270056.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:44:53,895 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.916e+02 2.602e+02 3.189e+02 4.066e+02 6.857e+02, threshold=6.378e+02, percent-clipped=4.0 2023-05-02 09:45:02,612 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 09:45:22,288 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0165, 3.8637, 4.0560, 4.1776, 4.2706, 3.9257, 4.2013, 4.2923], device='cuda:6'), covar=tensor([0.1668, 0.1204, 0.1366, 0.0717, 0.0662, 0.1400, 0.0946, 0.0769], device='cuda:6'), in_proj_covar=tensor([0.0666, 0.0816, 0.0948, 0.0828, 0.0630, 0.0660, 0.0687, 0.0801], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 09:45:42,296 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=270097.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 09:45:51,121 INFO [train.py:904] (6/8) Epoch 27, batch 6200, loss[loss=0.1914, simple_loss=0.2735, pruned_loss=0.05462, over 16623.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2862, pruned_loss=0.05587, over 3064953.04 frames. ], batch size: 62, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:46:33,641 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=270129.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 09:46:58,074 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=270145.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:47:09,692 INFO [train.py:904] (6/8) Epoch 27, batch 6250, loss[loss=0.1909, simple_loss=0.2821, pruned_loss=0.04987, over 16385.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.285, pruned_loss=0.05472, over 3083402.05 frames. ], batch size: 146, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:47:29,202 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.849e+02 2.710e+02 3.440e+02 4.218e+02 8.566e+02, threshold=6.880e+02, percent-clipped=4.0 2023-05-02 09:47:29,762 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8658, 1.4021, 1.7194, 1.6857, 1.7946, 1.9147, 1.6480, 1.7781], device='cuda:6'), covar=tensor([0.0279, 0.0435, 0.0235, 0.0344, 0.0285, 0.0216, 0.0480, 0.0149], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0197, 0.0185, 0.0190, 0.0206, 0.0164, 0.0202, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 09:47:58,461 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 09:48:08,660 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=270192.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:48:25,155 INFO [train.py:904] (6/8) Epoch 27, batch 6300, loss[loss=0.201, simple_loss=0.2889, pruned_loss=0.05655, over 16801.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2845, pruned_loss=0.05422, over 3087710.30 frames. ], batch size: 39, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:48:28,593 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.10 vs. limit=5.0 2023-05-02 09:49:45,973 INFO [train.py:904] (6/8) Epoch 27, batch 6350, loss[loss=0.185, simple_loss=0.281, pruned_loss=0.04449, over 16685.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2849, pruned_loss=0.05523, over 3074787.74 frames. ], batch size: 89, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:50:05,608 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.880e+02 2.767e+02 3.274e+02 3.776e+02 6.685e+02, threshold=6.549e+02, percent-clipped=0.0 2023-05-02 09:50:28,873 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=270280.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 09:50:31,771 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9753, 3.7936, 3.9686, 4.1345, 4.2219, 3.8517, 4.1502, 4.2328], device='cuda:6'), covar=tensor([0.1670, 0.1406, 0.1679, 0.0850, 0.0792, 0.1715, 0.1071, 0.0983], device='cuda:6'), in_proj_covar=tensor([0.0669, 0.0818, 0.0950, 0.0830, 0.0631, 0.0663, 0.0690, 0.0805], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 09:51:02,482 INFO [train.py:904] (6/8) Epoch 27, batch 6400, loss[loss=0.1967, simple_loss=0.2831, pruned_loss=0.05518, over 16391.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2859, pruned_loss=0.05662, over 3078132.19 frames. ], batch size: 146, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:51:37,112 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5294, 3.5965, 3.3817, 2.9879, 3.2074, 3.5172, 3.2969, 3.3099], device='cuda:6'), covar=tensor([0.0622, 0.0651, 0.0309, 0.0317, 0.0567, 0.0485, 0.1629, 0.0521], device='cuda:6'), in_proj_covar=tensor([0.0305, 0.0460, 0.0357, 0.0361, 0.0356, 0.0413, 0.0244, 0.0428], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 09:51:39,399 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 09:51:46,039 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1530, 3.5695, 3.5470, 2.3227, 3.3261, 3.6184, 3.3768, 2.0687], device='cuda:6'), covar=tensor([0.0582, 0.0077, 0.0082, 0.0469, 0.0117, 0.0126, 0.0109, 0.0514], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0090, 0.0091, 0.0136, 0.0101, 0.0116, 0.0098, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 09:52:01,760 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=270341.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 09:52:19,098 INFO [train.py:904] (6/8) Epoch 27, batch 6450, loss[loss=0.1809, simple_loss=0.2752, pruned_loss=0.04329, over 16801.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2858, pruned_loss=0.05591, over 3085729.54 frames. ], batch size: 116, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:52:39,046 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.936e+02 2.688e+02 3.236e+02 4.371e+02 9.664e+02, threshold=6.472e+02, percent-clipped=9.0 2023-05-02 09:52:40,620 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-02 09:53:27,559 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=270396.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:53:37,244 INFO [train.py:904] (6/8) Epoch 27, batch 6500, loss[loss=0.2145, simple_loss=0.2909, pruned_loss=0.0691, over 16891.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.2841, pruned_loss=0.05541, over 3096902.13 frames. ], batch size: 109, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:53:49,536 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9117, 2.0272, 2.1652, 3.3680, 1.9984, 2.3082, 2.1739, 2.1757], device='cuda:6'), covar=tensor([0.1630, 0.4052, 0.3265, 0.0733, 0.4727, 0.2849, 0.3985, 0.3726], device='cuda:6'), in_proj_covar=tensor([0.0417, 0.0468, 0.0380, 0.0333, 0.0442, 0.0536, 0.0439, 0.0548], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 09:54:17,006 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=270429.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 09:54:33,581 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=270439.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:54:58,114 INFO [train.py:904] (6/8) Epoch 27, batch 6550, loss[loss=0.2371, simple_loss=0.305, pruned_loss=0.08458, over 11642.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2874, pruned_loss=0.05646, over 3103134.62 frames. ], batch size: 247, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:55:04,722 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=270457.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:55:17,871 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.942e+02 2.679e+02 3.213e+02 3.936e+02 7.742e+02, threshold=6.427e+02, percent-clipped=6.0 2023-05-02 09:55:34,670 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=270477.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:55:58,271 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=270492.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:56:03,799 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-02 09:56:10,634 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=270500.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:56:14,134 INFO [train.py:904] (6/8) Epoch 27, batch 6600, loss[loss=0.2241, simple_loss=0.2987, pruned_loss=0.07471, over 11839.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2894, pruned_loss=0.05677, over 3101173.51 frames. ], batch size: 248, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:57:09,775 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=270540.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:57:30,477 INFO [train.py:904] (6/8) Epoch 27, batch 6650, loss[loss=0.2043, simple_loss=0.2831, pruned_loss=0.06274, over 16727.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2898, pruned_loss=0.05798, over 3086979.04 frames. ], batch size: 39, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:57:50,323 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 2.739e+02 3.273e+02 3.972e+02 7.700e+02, threshold=6.545e+02, percent-clipped=3.0 2023-05-02 09:58:46,095 INFO [train.py:904] (6/8) Epoch 27, batch 6700, loss[loss=0.1745, simple_loss=0.2701, pruned_loss=0.03942, over 16826.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2888, pruned_loss=0.05843, over 3063298.67 frames. ], batch size: 102, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:59:01,573 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=270613.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:59:19,804 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7774, 2.7361, 2.5270, 4.4074, 3.2385, 4.0255, 1.5827, 2.9734], device='cuda:6'), covar=tensor([0.1345, 0.0782, 0.1312, 0.0164, 0.0228, 0.0382, 0.1709, 0.0819], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0180, 0.0201, 0.0202, 0.0208, 0.0219, 0.0210, 0.0199], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 09:59:36,007 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=270636.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 09:59:37,860 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3205, 3.1678, 3.5049, 1.9609, 3.5994, 3.6792, 2.8415, 2.7365], device='cuda:6'), covar=tensor([0.0906, 0.0311, 0.0221, 0.1222, 0.0107, 0.0207, 0.0479, 0.0551], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0111, 0.0101, 0.0139, 0.0086, 0.0131, 0.0130, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 10:00:01,085 INFO [train.py:904] (6/8) Epoch 27, batch 6750, loss[loss=0.177, simple_loss=0.2673, pruned_loss=0.04329, over 16719.00 frames. ], tot_loss[loss=0.2013, simple_loss=0.287, pruned_loss=0.05778, over 3076560.05 frames. ], batch size: 89, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:00:20,166 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.179e+02 2.756e+02 3.273e+02 4.054e+02 1.247e+03, threshold=6.545e+02, percent-clipped=2.0 2023-05-02 10:00:32,938 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=270674.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:01:00,204 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5721, 4.6337, 4.4657, 4.1688, 4.1742, 4.5634, 4.2944, 4.2661], device='cuda:6'), covar=tensor([0.0577, 0.0487, 0.0273, 0.0290, 0.0794, 0.0431, 0.0573, 0.0610], device='cuda:6'), in_proj_covar=tensor([0.0306, 0.0462, 0.0358, 0.0362, 0.0358, 0.0415, 0.0246, 0.0430], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:01:15,210 INFO [train.py:904] (6/8) Epoch 27, batch 6800, loss[loss=0.2051, simple_loss=0.3115, pruned_loss=0.04938, over 16845.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2874, pruned_loss=0.05753, over 3083619.71 frames. ], batch size: 76, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:01:53,345 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=270726.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:02:09,336 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1499, 3.4502, 3.5741, 3.5337, 3.5425, 3.4171, 3.2380, 3.4700], device='cuda:6'), covar=tensor([0.0691, 0.0902, 0.0618, 0.0675, 0.0729, 0.0802, 0.1320, 0.0720], device='cuda:6'), in_proj_covar=tensor([0.0428, 0.0479, 0.0465, 0.0426, 0.0512, 0.0491, 0.0563, 0.0392], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 10:02:33,312 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=270752.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:02:34,169 INFO [train.py:904] (6/8) Epoch 27, batch 6850, loss[loss=0.2496, simple_loss=0.3147, pruned_loss=0.09226, over 11394.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2879, pruned_loss=0.05782, over 3090839.35 frames. ], batch size: 246, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:02:53,223 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.704e+02 3.220e+02 3.695e+02 5.610e+02, threshold=6.440e+02, percent-clipped=0.0 2023-05-02 10:03:23,177 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.61 vs. limit=2.0 2023-05-02 10:03:25,255 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=270787.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:03:37,626 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=270795.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:03:39,565 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0684, 2.4600, 2.5820, 1.9903, 2.6847, 2.7829, 2.4987, 2.4133], device='cuda:6'), covar=tensor([0.0757, 0.0280, 0.0237, 0.0939, 0.0145, 0.0294, 0.0430, 0.0461], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0111, 0.0101, 0.0139, 0.0086, 0.0131, 0.0131, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 10:03:49,486 INFO [train.py:904] (6/8) Epoch 27, batch 6900, loss[loss=0.1959, simple_loss=0.285, pruned_loss=0.05337, over 16719.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2902, pruned_loss=0.05735, over 3101338.43 frames. ], batch size: 124, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:04:00,475 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-02 10:04:57,964 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0119, 3.2537, 3.2060, 2.1069, 3.0136, 3.2334, 3.0654, 2.0213], device='cuda:6'), covar=tensor([0.0556, 0.0071, 0.0083, 0.0466, 0.0121, 0.0124, 0.0117, 0.0487], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0090, 0.0090, 0.0136, 0.0101, 0.0115, 0.0098, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 10:05:07,697 INFO [train.py:904] (6/8) Epoch 27, batch 6950, loss[loss=0.1814, simple_loss=0.2716, pruned_loss=0.04565, over 16891.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2917, pruned_loss=0.05837, over 3103992.28 frames. ], batch size: 116, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:05:28,446 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.875e+02 2.915e+02 3.513e+02 4.432e+02 8.204e+02, threshold=7.026e+02, percent-clipped=5.0 2023-05-02 10:05:30,372 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9648, 2.9179, 2.4547, 2.6680, 3.3208, 2.9084, 3.5024, 3.4535], device='cuda:6'), covar=tensor([0.0115, 0.0458, 0.0569, 0.0486, 0.0278, 0.0422, 0.0256, 0.0281], device='cuda:6'), in_proj_covar=tensor([0.0223, 0.0240, 0.0229, 0.0231, 0.0241, 0.0239, 0.0237, 0.0238], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:06:23,846 INFO [train.py:904] (6/8) Epoch 27, batch 7000, loss[loss=0.1797, simple_loss=0.2875, pruned_loss=0.03595, over 16857.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2913, pruned_loss=0.05735, over 3094754.12 frames. ], batch size: 102, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:06:46,510 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=270917.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:07:15,571 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=270936.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 10:07:40,697 INFO [train.py:904] (6/8) Epoch 27, batch 7050, loss[loss=0.2492, simple_loss=0.3167, pruned_loss=0.09089, over 11142.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2924, pruned_loss=0.05812, over 3059364.39 frames. ], batch size: 246, lr: 2.48e-03, grad_scale: 4.0 2023-05-02 10:08:01,040 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.024e+02 2.738e+02 3.222e+02 3.975e+02 6.307e+02, threshold=6.443e+02, percent-clipped=0.0 2023-05-02 10:08:05,086 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=270969.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:08:18,554 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=270978.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:08:28,177 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=270984.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 10:08:58,235 INFO [train.py:904] (6/8) Epoch 27, batch 7100, loss[loss=0.1951, simple_loss=0.2851, pruned_loss=0.05251, over 16710.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2902, pruned_loss=0.05732, over 3078473.87 frames. ], batch size: 62, lr: 2.48e-03, grad_scale: 4.0 2023-05-02 10:10:15,777 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=271052.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:10:16,497 INFO [train.py:904] (6/8) Epoch 27, batch 7150, loss[loss=0.2245, simple_loss=0.2977, pruned_loss=0.07566, over 11280.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2888, pruned_loss=0.05782, over 3060730.66 frames. ], batch size: 247, lr: 2.48e-03, grad_scale: 4.0 2023-05-02 10:10:24,777 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0093, 3.9068, 4.0632, 4.1803, 4.2830, 3.8971, 4.2110, 4.3011], device='cuda:6'), covar=tensor([0.1668, 0.1198, 0.1318, 0.0660, 0.0577, 0.1617, 0.0876, 0.0741], device='cuda:6'), in_proj_covar=tensor([0.0662, 0.0809, 0.0943, 0.0820, 0.0626, 0.0655, 0.0685, 0.0798], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:10:34,224 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-02 10:10:37,964 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.119e+02 2.616e+02 3.372e+02 3.961e+02 8.125e+02, threshold=6.744e+02, percent-clipped=3.0 2023-05-02 10:10:59,824 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=271082.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:11:18,930 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=271095.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:11:28,070 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=271100.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:11:31,638 INFO [train.py:904] (6/8) Epoch 27, batch 7200, loss[loss=0.1727, simple_loss=0.276, pruned_loss=0.03464, over 16783.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2869, pruned_loss=0.05616, over 3051809.50 frames. ], batch size: 83, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:11:37,682 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4066, 4.5357, 4.6829, 4.4536, 4.5965, 5.0512, 4.4980, 4.2435], device='cuda:6'), covar=tensor([0.1525, 0.1730, 0.2060, 0.1933, 0.2235, 0.0988, 0.1662, 0.2514], device='cuda:6'), in_proj_covar=tensor([0.0425, 0.0632, 0.0693, 0.0514, 0.0684, 0.0723, 0.0543, 0.0691], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 10:11:58,437 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3379, 5.6185, 5.4202, 5.4306, 5.0951, 5.0825, 5.0576, 5.7592], device='cuda:6'), covar=tensor([0.1306, 0.0867, 0.0928, 0.0838, 0.0810, 0.0787, 0.1212, 0.0801], device='cuda:6'), in_proj_covar=tensor([0.0702, 0.0850, 0.0696, 0.0654, 0.0535, 0.0536, 0.0715, 0.0665], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:12:37,999 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=271143.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:12:53,488 INFO [train.py:904] (6/8) Epoch 27, batch 7250, loss[loss=0.1606, simple_loss=0.2518, pruned_loss=0.03475, over 16435.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2844, pruned_loss=0.05452, over 3072956.50 frames. ], batch size: 146, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:13:16,037 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.855e+02 2.666e+02 3.085e+02 3.699e+02 8.603e+02, threshold=6.169e+02, percent-clipped=3.0 2023-05-02 10:13:46,066 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-05-02 10:14:10,772 INFO [train.py:904] (6/8) Epoch 27, batch 7300, loss[loss=0.2118, simple_loss=0.2789, pruned_loss=0.07233, over 11036.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2842, pruned_loss=0.05396, over 3085197.76 frames. ], batch size: 247, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:15:29,743 INFO [train.py:904] (6/8) Epoch 27, batch 7350, loss[loss=0.193, simple_loss=0.2823, pruned_loss=0.05186, over 16252.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2853, pruned_loss=0.05505, over 3080139.98 frames. ], batch size: 165, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:15:42,307 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=271261.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:15:50,537 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.854e+02 2.875e+02 3.303e+02 4.103e+02 7.171e+02, threshold=6.606e+02, percent-clipped=3.0 2023-05-02 10:15:54,731 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=271269.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:16:01,219 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=271273.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:16:40,215 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.64 vs. limit=2.0 2023-05-02 10:16:48,723 INFO [train.py:904] (6/8) Epoch 27, batch 7400, loss[loss=0.1937, simple_loss=0.2894, pruned_loss=0.04902, over 16721.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2866, pruned_loss=0.0556, over 3086061.50 frames. ], batch size: 89, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:16:58,687 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2401, 4.2363, 4.1269, 3.3185, 4.1818, 1.7670, 3.9594, 3.7146], device='cuda:6'), covar=tensor([0.0145, 0.0133, 0.0209, 0.0381, 0.0117, 0.3044, 0.0155, 0.0303], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0170, 0.0209, 0.0183, 0.0184, 0.0214, 0.0196, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:17:11,159 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=271317.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:17:18,913 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7771, 1.4117, 1.7240, 1.6663, 1.7799, 1.8728, 1.6324, 1.7759], device='cuda:6'), covar=tensor([0.0296, 0.0414, 0.0226, 0.0319, 0.0268, 0.0207, 0.0448, 0.0150], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0195, 0.0183, 0.0188, 0.0203, 0.0163, 0.0200, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:17:18,923 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=271322.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:17:28,877 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6271, 4.6801, 4.4844, 4.1720, 4.2209, 4.5940, 4.3554, 4.3046], device='cuda:6'), covar=tensor([0.0590, 0.0597, 0.0308, 0.0313, 0.0800, 0.0507, 0.0535, 0.0677], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0455, 0.0354, 0.0356, 0.0352, 0.0409, 0.0243, 0.0423], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:18:06,473 INFO [train.py:904] (6/8) Epoch 27, batch 7450, loss[loss=0.2102, simple_loss=0.303, pruned_loss=0.0587, over 16302.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.288, pruned_loss=0.05691, over 3081360.83 frames. ], batch size: 165, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:18:30,891 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.940e+02 2.819e+02 3.560e+02 4.384e+02 9.484e+02, threshold=7.119e+02, percent-clipped=1.0 2023-05-02 10:18:32,493 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-05-02 10:18:45,194 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=271375.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:18:55,971 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=271382.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:19:20,763 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5186, 4.5403, 4.8421, 4.8189, 4.8324, 4.5416, 4.5208, 4.4191], device='cuda:6'), covar=tensor([0.0327, 0.0607, 0.0375, 0.0388, 0.0447, 0.0402, 0.0900, 0.0506], device='cuda:6'), in_proj_covar=tensor([0.0426, 0.0477, 0.0464, 0.0424, 0.0510, 0.0488, 0.0562, 0.0391], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 10:19:30,452 INFO [train.py:904] (6/8) Epoch 27, batch 7500, loss[loss=0.1701, simple_loss=0.2698, pruned_loss=0.0352, over 16790.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2877, pruned_loss=0.05607, over 3065235.84 frames. ], batch size: 102, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:19:42,190 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0028, 3.2714, 3.3422, 2.1153, 2.9490, 2.2341, 3.4497, 3.5301], device='cuda:6'), covar=tensor([0.0261, 0.0850, 0.0641, 0.2153, 0.0839, 0.1024, 0.0657, 0.1036], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0171, 0.0170, 0.0156, 0.0148, 0.0133, 0.0146, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 10:20:08,119 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=271426.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:20:14,062 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=271430.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:20:22,990 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=271436.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:20:49,666 INFO [train.py:904] (6/8) Epoch 27, batch 7550, loss[loss=0.202, simple_loss=0.2843, pruned_loss=0.05986, over 16728.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2866, pruned_loss=0.0559, over 3078560.46 frames. ], batch size: 124, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:20:57,850 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6673, 1.8768, 2.2220, 2.6167, 2.5921, 2.9775, 1.8677, 2.9181], device='cuda:6'), covar=tensor([0.0267, 0.0564, 0.0380, 0.0346, 0.0354, 0.0188, 0.0676, 0.0160], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0195, 0.0183, 0.0188, 0.0203, 0.0163, 0.0200, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:21:11,189 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.886e+02 2.650e+02 3.300e+02 4.210e+02 7.541e+02, threshold=6.599e+02, percent-clipped=2.0 2023-05-02 10:21:41,647 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=271487.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:22:05,495 INFO [train.py:904] (6/8) Epoch 27, batch 7600, loss[loss=0.1973, simple_loss=0.2902, pruned_loss=0.05221, over 16773.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.2849, pruned_loss=0.05537, over 3097258.93 frames. ], batch size: 83, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:23:22,937 INFO [train.py:904] (6/8) Epoch 27, batch 7650, loss[loss=0.177, simple_loss=0.2721, pruned_loss=0.04096, over 17132.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2851, pruned_loss=0.05586, over 3092687.88 frames. ], batch size: 49, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:23:30,189 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9306, 3.0446, 3.2571, 1.9509, 2.8592, 2.1628, 3.4090, 3.3494], device='cuda:6'), covar=tensor([0.0304, 0.1001, 0.0679, 0.2431, 0.0971, 0.1141, 0.0708, 0.1132], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0171, 0.0171, 0.0157, 0.0149, 0.0133, 0.0147, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 10:23:31,704 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 10:23:45,488 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.699e+02 2.765e+02 3.286e+02 4.189e+02 6.927e+02, threshold=6.573e+02, percent-clipped=1.0 2023-05-02 10:23:46,638 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6270, 2.4558, 2.3101, 3.9054, 2.5372, 3.9052, 1.4058, 2.7037], device='cuda:6'), covar=tensor([0.1589, 0.0995, 0.1465, 0.0240, 0.0287, 0.0452, 0.2060, 0.0995], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0180, 0.0202, 0.0201, 0.0209, 0.0219, 0.0210, 0.0200], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 10:23:55,688 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=271573.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:24:11,887 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.84 vs. limit=5.0 2023-05-02 10:24:16,592 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.40 vs. limit=5.0 2023-05-02 10:24:33,241 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-05-02 10:24:43,692 INFO [train.py:904] (6/8) Epoch 27, batch 7700, loss[loss=0.1826, simple_loss=0.2756, pruned_loss=0.04477, over 16859.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2853, pruned_loss=0.05612, over 3104098.34 frames. ], batch size: 96, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:25:06,096 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=271617.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:25:09,422 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2246, 4.3000, 4.1212, 3.8295, 3.8650, 4.2223, 3.9165, 3.9898], device='cuda:6'), covar=tensor([0.0614, 0.0684, 0.0308, 0.0302, 0.0752, 0.0543, 0.0813, 0.0630], device='cuda:6'), in_proj_covar=tensor([0.0301, 0.0454, 0.0352, 0.0355, 0.0351, 0.0408, 0.0242, 0.0422], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:25:12,496 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=271621.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:25:12,729 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6739, 2.7877, 2.3259, 2.5543, 3.1078, 2.8100, 3.2432, 3.3321], device='cuda:6'), covar=tensor([0.0126, 0.0442, 0.0558, 0.0457, 0.0301, 0.0400, 0.0248, 0.0253], device='cuda:6'), in_proj_covar=tensor([0.0224, 0.0242, 0.0231, 0.0232, 0.0243, 0.0240, 0.0239, 0.0239], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:25:17,025 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8697, 2.2098, 2.4525, 3.1192, 2.2356, 2.3734, 2.3506, 2.2970], device='cuda:6'), covar=tensor([0.1461, 0.3241, 0.2528, 0.0742, 0.4116, 0.2423, 0.3289, 0.3187], device='cuda:6'), in_proj_covar=tensor([0.0416, 0.0470, 0.0381, 0.0333, 0.0443, 0.0536, 0.0441, 0.0547], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:25:28,124 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1556, 2.0819, 1.7533, 1.8556, 2.3036, 1.9928, 1.9695, 2.4119], device='cuda:6'), covar=tensor([0.0235, 0.0427, 0.0586, 0.0504, 0.0274, 0.0391, 0.0219, 0.0270], device='cuda:6'), in_proj_covar=tensor([0.0225, 0.0242, 0.0231, 0.0233, 0.0244, 0.0240, 0.0240, 0.0240], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:26:02,415 INFO [train.py:904] (6/8) Epoch 27, batch 7750, loss[loss=0.1885, simple_loss=0.2799, pruned_loss=0.04857, over 16477.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2854, pruned_loss=0.05584, over 3114292.23 frames. ], batch size: 68, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:26:11,253 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8316, 1.4802, 1.7450, 1.6688, 1.7982, 1.8884, 1.6707, 1.8393], device='cuda:6'), covar=tensor([0.0278, 0.0372, 0.0223, 0.0329, 0.0280, 0.0200, 0.0443, 0.0157], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0196, 0.0183, 0.0189, 0.0204, 0.0164, 0.0201, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:26:24,500 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.915e+02 2.896e+02 3.348e+02 4.032e+02 8.871e+02, threshold=6.696e+02, percent-clipped=2.0 2023-05-02 10:27:20,098 INFO [train.py:904] (6/8) Epoch 27, batch 7800, loss[loss=0.2188, simple_loss=0.2936, pruned_loss=0.07203, over 11735.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2861, pruned_loss=0.05638, over 3103406.48 frames. ], batch size: 247, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:28:04,771 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=271731.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:28:36,664 INFO [train.py:904] (6/8) Epoch 27, batch 7850, loss[loss=0.207, simple_loss=0.2911, pruned_loss=0.06149, over 16646.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2871, pruned_loss=0.05649, over 3093859.58 frames. ], batch size: 57, lr: 2.48e-03, grad_scale: 4.0 2023-05-02 10:28:57,992 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.815e+02 2.838e+02 3.379e+02 4.179e+02 9.334e+02, threshold=6.757e+02, percent-clipped=5.0 2023-05-02 10:29:12,206 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.89 vs. limit=5.0 2023-05-02 10:29:20,881 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=271782.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:29:50,020 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1614, 3.2045, 1.9644, 3.4347, 2.3974, 3.4536, 2.1441, 2.6623], device='cuda:6'), covar=tensor([0.0345, 0.0446, 0.1776, 0.0331, 0.0889, 0.0858, 0.1599, 0.0842], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0182, 0.0198, 0.0172, 0.0181, 0.0221, 0.0204, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 10:29:52,537 INFO [train.py:904] (6/8) Epoch 27, batch 7900, loss[loss=0.1956, simple_loss=0.2854, pruned_loss=0.05291, over 16970.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2862, pruned_loss=0.05602, over 3094280.64 frames. ], batch size: 109, lr: 2.48e-03, grad_scale: 4.0 2023-05-02 10:31:02,467 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.47 vs. limit=2.0 2023-05-02 10:31:06,591 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5212, 4.5078, 4.3730, 3.6019, 4.4591, 1.7597, 4.1884, 4.0216], device='cuda:6'), covar=tensor([0.0131, 0.0135, 0.0210, 0.0384, 0.0097, 0.3004, 0.0148, 0.0248], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0171, 0.0209, 0.0183, 0.0184, 0.0214, 0.0197, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:31:12,162 INFO [train.py:904] (6/8) Epoch 27, batch 7950, loss[loss=0.1876, simple_loss=0.275, pruned_loss=0.05007, over 16912.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2864, pruned_loss=0.05628, over 3093725.55 frames. ], batch size: 109, lr: 2.48e-03, grad_scale: 4.0 2023-05-02 10:31:15,348 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=271854.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:31:36,730 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.948e+02 2.764e+02 3.233e+02 3.778e+02 5.723e+02, threshold=6.466e+02, percent-clipped=0.0 2023-05-02 10:31:38,867 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.32 vs. limit=5.0 2023-05-02 10:31:54,431 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9689, 3.3350, 3.2358, 2.0837, 3.1569, 3.3804, 3.1540, 1.6175], device='cuda:6'), covar=tensor([0.0666, 0.0101, 0.0117, 0.0547, 0.0141, 0.0170, 0.0147, 0.0735], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0088, 0.0088, 0.0134, 0.0099, 0.0113, 0.0096, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-02 10:32:04,977 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8578, 2.6852, 2.5658, 1.9243, 2.5450, 2.6757, 2.5253, 1.8617], device='cuda:6'), covar=tensor([0.0482, 0.0112, 0.0109, 0.0421, 0.0145, 0.0152, 0.0144, 0.0475], device='cuda:6'), in_proj_covar=tensor([0.0136, 0.0088, 0.0088, 0.0134, 0.0099, 0.0113, 0.0096, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-02 10:32:29,919 INFO [train.py:904] (6/8) Epoch 27, batch 8000, loss[loss=0.1925, simple_loss=0.2812, pruned_loss=0.05197, over 16257.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2869, pruned_loss=0.05661, over 3091571.18 frames. ], batch size: 146, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:32:49,718 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=271915.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:32:52,971 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=271917.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:33:19,524 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=271934.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:33:47,183 INFO [train.py:904] (6/8) Epoch 27, batch 8050, loss[loss=0.2045, simple_loss=0.291, pruned_loss=0.05894, over 15287.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2872, pruned_loss=0.05663, over 3093973.66 frames. ], batch size: 190, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:34:01,654 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-02 10:34:05,653 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=271965.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:34:09,131 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.787e+02 2.577e+02 3.144e+02 3.701e+02 6.468e+02, threshold=6.287e+02, percent-clipped=1.0 2023-05-02 10:34:44,967 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-02 10:34:49,676 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=271995.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:35:04,670 INFO [train.py:904] (6/8) Epoch 27, batch 8100, loss[loss=0.179, simple_loss=0.2753, pruned_loss=0.04137, over 16715.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2867, pruned_loss=0.05601, over 3102269.92 frames. ], batch size: 76, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:35:32,368 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-05-02 10:35:45,689 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272031.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:36:17,759 INFO [train.py:904] (6/8) Epoch 27, batch 8150, loss[loss=0.2274, simple_loss=0.304, pruned_loss=0.07535, over 11850.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2842, pruned_loss=0.05512, over 3100645.82 frames. ], batch size: 247, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:36:39,765 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.743e+02 3.279e+02 3.939e+02 6.192e+02, threshold=6.559e+02, percent-clipped=0.0 2023-05-02 10:36:57,282 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272079.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:37:01,622 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272082.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:37:32,131 INFO [train.py:904] (6/8) Epoch 27, batch 8200, loss[loss=0.2017, simple_loss=0.2754, pruned_loss=0.06399, over 11629.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2819, pruned_loss=0.05475, over 3096283.05 frames. ], batch size: 248, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:38:12,511 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=272127.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:38:16,448 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272130.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:38:53,589 INFO [train.py:904] (6/8) Epoch 27, batch 8250, loss[loss=0.1587, simple_loss=0.2507, pruned_loss=0.03337, over 12018.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2808, pruned_loss=0.05243, over 3062717.04 frames. ], batch size: 247, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:39:09,927 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5664, 3.5330, 3.5190, 2.6338, 3.4190, 2.0053, 3.2737, 2.8908], device='cuda:6'), covar=tensor([0.0174, 0.0164, 0.0199, 0.0267, 0.0136, 0.2713, 0.0160, 0.0320], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0172, 0.0210, 0.0183, 0.0185, 0.0215, 0.0198, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:39:19,036 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.616e+02 2.668e+02 3.172e+02 3.719e+02 6.913e+02, threshold=6.344e+02, percent-clipped=1.0 2023-05-02 10:39:40,209 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=272181.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 10:39:50,760 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=272188.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:40:14,329 INFO [train.py:904] (6/8) Epoch 27, batch 8300, loss[loss=0.1544, simple_loss=0.2419, pruned_loss=0.03344, over 11890.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.2783, pruned_loss=0.04924, over 3070250.80 frames. ], batch size: 246, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:40:26,426 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272210.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:40:46,092 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=272223.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:41:15,192 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=272242.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 10:41:23,004 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=272247.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:41:31,411 INFO [train.py:904] (6/8) Epoch 27, batch 8350, loss[loss=0.2165, simple_loss=0.294, pruned_loss=0.06945, over 11940.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2779, pruned_loss=0.04779, over 3065848.97 frames. ], batch size: 246, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:41:49,246 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2311, 4.2265, 4.5540, 4.5357, 4.5241, 4.3021, 4.2287, 4.3027], device='cuda:6'), covar=tensor([0.0375, 0.0854, 0.0446, 0.0421, 0.0460, 0.0453, 0.0978, 0.0503], device='cuda:6'), in_proj_covar=tensor([0.0428, 0.0482, 0.0466, 0.0428, 0.0513, 0.0491, 0.0564, 0.0393], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 10:41:54,864 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.512e+02 2.144e+02 2.517e+02 3.037e+02 5.148e+02, threshold=5.034e+02, percent-clipped=0.0 2023-05-02 10:42:20,864 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=272284.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:42:30,953 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272290.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:42:50,436 INFO [train.py:904] (6/8) Epoch 27, batch 8400, loss[loss=0.1683, simple_loss=0.2634, pruned_loss=0.03656, over 16237.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2748, pruned_loss=0.04557, over 3072618.73 frames. ], batch size: 165, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:42:59,351 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=272308.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:43:09,896 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8672, 4.8107, 4.5995, 3.8978, 4.6947, 1.7827, 4.4759, 4.4291], device='cuda:6'), covar=tensor([0.0101, 0.0107, 0.0234, 0.0426, 0.0128, 0.2919, 0.0139, 0.0274], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0172, 0.0210, 0.0183, 0.0185, 0.0215, 0.0197, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:44:09,888 INFO [train.py:904] (6/8) Epoch 27, batch 8450, loss[loss=0.161, simple_loss=0.2488, pruned_loss=0.03662, over 12148.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2732, pruned_loss=0.04404, over 3070584.67 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:44:11,701 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4567, 3.3747, 2.7909, 2.2027, 2.1766, 2.4255, 3.5313, 3.0289], device='cuda:6'), covar=tensor([0.3017, 0.0668, 0.1804, 0.3191, 0.2974, 0.2238, 0.0443, 0.1402], device='cuda:6'), in_proj_covar=tensor([0.0329, 0.0270, 0.0307, 0.0321, 0.0300, 0.0271, 0.0299, 0.0343], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 10:44:34,141 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.487e+02 2.148e+02 2.573e+02 3.182e+02 7.232e+02, threshold=5.146e+02, percent-clipped=1.0 2023-05-02 10:44:42,768 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=272373.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:45:32,396 INFO [train.py:904] (6/8) Epoch 27, batch 8500, loss[loss=0.1606, simple_loss=0.252, pruned_loss=0.03462, over 11877.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2699, pruned_loss=0.04211, over 3078050.11 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:46:24,089 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=272434.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:46:57,057 INFO [train.py:904] (6/8) Epoch 27, batch 8550, loss[loss=0.1679, simple_loss=0.2677, pruned_loss=0.03405, over 16662.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.268, pruned_loss=0.04113, over 3058224.35 frames. ], batch size: 134, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:47:25,001 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.247e+02 2.591e+02 3.219e+02 6.393e+02, threshold=5.181e+02, percent-clipped=2.0 2023-05-02 10:47:55,108 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272483.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:48:35,725 INFO [train.py:904] (6/8) Epoch 27, batch 8600, loss[loss=0.1834, simple_loss=0.2904, pruned_loss=0.03823, over 16464.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2683, pruned_loss=0.03993, over 3060033.68 frames. ], batch size: 146, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:48:50,718 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272510.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:49:43,700 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272537.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 10:50:13,940 INFO [train.py:904] (6/8) Epoch 27, batch 8650, loss[loss=0.1746, simple_loss=0.2662, pruned_loss=0.0415, over 12128.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.267, pruned_loss=0.03874, over 3048263.83 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:50:27,063 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272558.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:50:47,045 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 10:50:48,992 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.181e+02 2.564e+02 3.046e+02 5.835e+02, threshold=5.129e+02, percent-clipped=3.0 2023-05-02 10:51:15,238 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272579.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:51:36,685 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272590.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:51:44,324 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2603, 3.4756, 3.4779, 2.4208, 3.1729, 3.5097, 3.3727, 2.0077], device='cuda:6'), covar=tensor([0.0500, 0.0070, 0.0061, 0.0399, 0.0130, 0.0095, 0.0086, 0.0555], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0087, 0.0088, 0.0133, 0.0098, 0.0111, 0.0095, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 10:52:00,241 INFO [train.py:904] (6/8) Epoch 27, batch 8700, loss[loss=0.1696, simple_loss=0.2629, pruned_loss=0.03812, over 16811.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2645, pruned_loss=0.03771, over 3046706.42 frames. ], batch size: 124, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:52:01,242 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272603.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:52:21,573 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=272613.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:53:06,582 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272638.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:53:35,410 INFO [train.py:904] (6/8) Epoch 27, batch 8750, loss[loss=0.1778, simple_loss=0.2814, pruned_loss=0.03707, over 16926.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2631, pruned_loss=0.03686, over 3035715.60 frames. ], batch size: 109, lr: 2.47e-03, grad_scale: 4.0 2023-05-02 10:54:15,879 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.360e+02 2.765e+02 3.360e+02 6.939e+02, threshold=5.529e+02, percent-clipped=4.0 2023-05-02 10:54:28,942 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=272674.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:55:27,074 INFO [train.py:904] (6/8) Epoch 27, batch 8800, loss[loss=0.1717, simple_loss=0.2695, pruned_loss=0.03692, over 16708.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2619, pruned_loss=0.03617, over 3033317.50 frames. ], batch size: 134, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:56:23,109 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272729.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:57:03,530 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5159, 3.4452, 3.5577, 3.6108, 3.6494, 3.3453, 3.6417, 3.6972], device='cuda:6'), covar=tensor([0.1118, 0.0901, 0.0881, 0.0565, 0.0581, 0.2455, 0.0784, 0.0723], device='cuda:6'), in_proj_covar=tensor([0.0635, 0.0778, 0.0904, 0.0789, 0.0606, 0.0632, 0.0661, 0.0772], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 10:57:12,470 INFO [train.py:904] (6/8) Epoch 27, batch 8850, loss[loss=0.1771, simple_loss=0.28, pruned_loss=0.03706, over 16805.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2645, pruned_loss=0.03533, over 3042690.20 frames. ], batch size: 124, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:57:46,549 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.085e+02 2.509e+02 2.907e+02 5.151e+02, threshold=5.018e+02, percent-clipped=0.0 2023-05-02 10:58:17,338 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272783.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:58:57,762 INFO [train.py:904] (6/8) Epoch 27, batch 8900, loss[loss=0.1565, simple_loss=0.2575, pruned_loss=0.02776, over 16928.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2646, pruned_loss=0.03496, over 3037188.03 frames. ], batch size: 96, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:00:01,533 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272831.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:00:20,778 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272837.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 11:00:45,797 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4212, 3.4725, 3.6804, 3.6704, 3.6754, 3.4971, 3.5337, 3.5677], device='cuda:6'), covar=tensor([0.0450, 0.0828, 0.0514, 0.0483, 0.0536, 0.0566, 0.0765, 0.0554], device='cuda:6'), in_proj_covar=tensor([0.0420, 0.0472, 0.0458, 0.0421, 0.0505, 0.0483, 0.0553, 0.0387], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 11:01:00,397 INFO [train.py:904] (6/8) Epoch 27, batch 8950, loss[loss=0.1584, simple_loss=0.255, pruned_loss=0.03091, over 16189.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2641, pruned_loss=0.03506, over 3060549.54 frames. ], batch size: 165, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:01:35,337 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 1.963e+02 2.469e+02 3.081e+02 5.293e+02, threshold=4.938e+02, percent-clipped=1.0 2023-05-02 11:01:57,035 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272879.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:02:12,417 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272885.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 11:02:49,712 INFO [train.py:904] (6/8) Epoch 27, batch 9000, loss[loss=0.1584, simple_loss=0.2396, pruned_loss=0.03858, over 12030.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2611, pruned_loss=0.03419, over 3053169.18 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:02:49,712 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 11:02:57,191 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2138, 2.4458, 2.8655, 3.2553, 3.0045, 3.7085, 2.6356, 3.6894], device='cuda:6'), covar=tensor([0.0268, 0.0528, 0.0384, 0.0314, 0.0361, 0.0188, 0.0505, 0.0188], device='cuda:6'), in_proj_covar=tensor([0.0194, 0.0196, 0.0183, 0.0187, 0.0204, 0.0162, 0.0199, 0.0162], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:02:59,737 INFO [train.py:938] (6/8) Epoch 27, validation: loss=0.1436, simple_loss=0.2474, pruned_loss=0.01989, over 944034.00 frames. 2023-05-02 11:02:59,738 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 11:03:00,803 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272903.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:03:26,180 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4139, 4.6846, 4.5460, 4.5610, 4.2547, 4.1989, 4.2637, 4.7488], device='cuda:6'), covar=tensor([0.1123, 0.0898, 0.0935, 0.0791, 0.0802, 0.1521, 0.1183, 0.0849], device='cuda:6'), in_proj_covar=tensor([0.0691, 0.0837, 0.0686, 0.0643, 0.0527, 0.0529, 0.0699, 0.0654], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:03:51,518 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272927.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:04:42,534 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272951.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:04:44,604 INFO [train.py:904] (6/8) Epoch 27, batch 9050, loss[loss=0.1655, simple_loss=0.2521, pruned_loss=0.03948, over 16692.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2619, pruned_loss=0.03465, over 3059556.82 frames. ], batch size: 134, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:04:56,372 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7339, 2.6330, 2.3922, 4.0952, 2.4065, 3.9250, 1.5865, 2.8925], device='cuda:6'), covar=tensor([0.1389, 0.0812, 0.1309, 0.0173, 0.0117, 0.0461, 0.1679, 0.0789], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0177, 0.0197, 0.0195, 0.0202, 0.0214, 0.0207, 0.0195], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 11:05:18,775 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 2.208e+02 2.579e+02 3.072e+02 5.940e+02, threshold=5.159e+02, percent-clipped=5.0 2023-05-02 11:05:19,931 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272969.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:05:30,416 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6230, 3.7447, 2.8764, 2.2410, 2.4278, 2.4956, 3.9652, 3.2917], device='cuda:6'), covar=tensor([0.2959, 0.0593, 0.1825, 0.3228, 0.2674, 0.2183, 0.0390, 0.1386], device='cuda:6'), in_proj_covar=tensor([0.0327, 0.0268, 0.0306, 0.0319, 0.0296, 0.0270, 0.0297, 0.0341], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 11:06:32,228 INFO [train.py:904] (6/8) Epoch 27, batch 9100, loss[loss=0.1447, simple_loss=0.2396, pruned_loss=0.02489, over 12713.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2614, pruned_loss=0.03488, over 3069467.61 frames. ], batch size: 246, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:07:36,908 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=273029.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:08:32,210 INFO [train.py:904] (6/8) Epoch 27, batch 9150, loss[loss=0.1614, simple_loss=0.2615, pruned_loss=0.03058, over 16703.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2625, pruned_loss=0.03467, over 3078969.66 frames. ], batch size: 134, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:09:08,126 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.260e+02 2.634e+02 3.413e+02 5.711e+02, threshold=5.268e+02, percent-clipped=2.0 2023-05-02 11:09:27,808 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=273077.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:10:18,361 INFO [train.py:904] (6/8) Epoch 27, batch 9200, loss[loss=0.1511, simple_loss=0.2353, pruned_loss=0.03341, over 12110.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2585, pruned_loss=0.03398, over 3083887.28 frames. ], batch size: 247, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:11:53,368 INFO [train.py:904] (6/8) Epoch 27, batch 9250, loss[loss=0.163, simple_loss=0.2606, pruned_loss=0.03273, over 16396.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2584, pruned_loss=0.03397, over 3093119.86 frames. ], batch size: 146, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:12:10,611 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0113, 3.0643, 2.0577, 3.3403, 2.3595, 3.3050, 2.1722, 2.6015], device='cuda:6'), covar=tensor([0.0362, 0.0438, 0.1551, 0.0246, 0.0872, 0.0639, 0.1575, 0.0779], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0175, 0.0191, 0.0164, 0.0175, 0.0211, 0.0199, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 11:12:25,533 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.747e+02 2.225e+02 2.595e+02 3.229e+02 5.341e+02, threshold=5.191e+02, percent-clipped=1.0 2023-05-02 11:13:07,292 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-05-02 11:13:17,904 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3687, 1.6095, 2.0755, 2.3035, 2.3737, 2.5478, 1.6556, 2.4598], device='cuda:6'), covar=tensor([0.0251, 0.0641, 0.0374, 0.0368, 0.0380, 0.0229, 0.0699, 0.0181], device='cuda:6'), in_proj_covar=tensor([0.0191, 0.0193, 0.0181, 0.0184, 0.0200, 0.0159, 0.0196, 0.0159], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:13:43,821 INFO [train.py:904] (6/8) Epoch 27, batch 9300, loss[loss=0.1462, simple_loss=0.2413, pruned_loss=0.02557, over 15416.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.257, pruned_loss=0.0338, over 3090914.98 frames. ], batch size: 192, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:14:02,973 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 11:14:59,138 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6956, 4.7716, 4.6036, 4.2420, 4.2675, 4.6766, 4.5032, 4.3833], device='cuda:6'), covar=tensor([0.0610, 0.0628, 0.0339, 0.0348, 0.0961, 0.0562, 0.0476, 0.0702], device='cuda:6'), in_proj_covar=tensor([0.0299, 0.0449, 0.0349, 0.0351, 0.0346, 0.0405, 0.0241, 0.0419], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:15:28,949 INFO [train.py:904] (6/8) Epoch 27, batch 9350, loss[loss=0.1702, simple_loss=0.2676, pruned_loss=0.0364, over 16774.00 frames. ], tot_loss[loss=0.1621, simple_loss=0.2567, pruned_loss=0.03377, over 3093549.21 frames. ], batch size: 83, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:16:02,980 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.192e+02 2.614e+02 3.426e+02 6.584e+02, threshold=5.229e+02, percent-clipped=2.0 2023-05-02 11:16:03,974 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=273269.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:16:30,857 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.12 vs. limit=5.0 2023-05-02 11:17:04,213 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6955, 4.6209, 4.4514, 3.6946, 4.5641, 1.5249, 4.3873, 4.1916], device='cuda:6'), covar=tensor([0.0107, 0.0135, 0.0240, 0.0428, 0.0131, 0.3308, 0.0145, 0.0340], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0167, 0.0203, 0.0176, 0.0180, 0.0209, 0.0192, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:17:10,294 INFO [train.py:904] (6/8) Epoch 27, batch 9400, loss[loss=0.1615, simple_loss=0.2711, pruned_loss=0.02595, over 16868.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2566, pruned_loss=0.03361, over 3084204.46 frames. ], batch size: 96, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:17:39,667 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=273317.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:17:41,318 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7978, 1.4342, 1.7537, 1.7258, 1.8763, 1.9175, 1.7307, 1.8453], device='cuda:6'), covar=tensor([0.0316, 0.0499, 0.0287, 0.0368, 0.0380, 0.0231, 0.0485, 0.0162], device='cuda:6'), in_proj_covar=tensor([0.0192, 0.0195, 0.0183, 0.0185, 0.0202, 0.0160, 0.0198, 0.0160], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:17:55,072 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3977, 4.6994, 4.5029, 4.5196, 4.2271, 4.2029, 4.1928, 4.7338], device='cuda:6'), covar=tensor([0.1270, 0.0972, 0.0958, 0.0822, 0.0790, 0.1525, 0.1168, 0.0924], device='cuda:6'), in_proj_covar=tensor([0.0690, 0.0837, 0.0684, 0.0643, 0.0526, 0.0530, 0.0699, 0.0654], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:18:39,559 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6128, 3.6775, 3.4883, 3.1880, 3.2935, 3.5745, 3.3695, 3.4083], device='cuda:6'), covar=tensor([0.0556, 0.0544, 0.0299, 0.0272, 0.0487, 0.0468, 0.1285, 0.0513], device='cuda:6'), in_proj_covar=tensor([0.0297, 0.0445, 0.0347, 0.0349, 0.0344, 0.0401, 0.0239, 0.0416], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:18:52,056 INFO [train.py:904] (6/8) Epoch 27, batch 9450, loss[loss=0.1716, simple_loss=0.2688, pruned_loss=0.03723, over 16897.00 frames. ], tot_loss[loss=0.1621, simple_loss=0.2575, pruned_loss=0.03341, over 3063104.49 frames. ], batch size: 116, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:19:21,967 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.197e+02 2.626e+02 3.206e+02 9.861e+02, threshold=5.251e+02, percent-clipped=1.0 2023-05-02 11:19:40,616 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6135, 3.8571, 2.9087, 2.2802, 2.3349, 2.5448, 4.1470, 3.2225], device='cuda:6'), covar=tensor([0.3126, 0.0535, 0.1940, 0.3092, 0.2868, 0.2135, 0.0371, 0.1454], device='cuda:6'), in_proj_covar=tensor([0.0325, 0.0266, 0.0304, 0.0317, 0.0293, 0.0267, 0.0295, 0.0338], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 11:20:32,362 INFO [train.py:904] (6/8) Epoch 27, batch 9500, loss[loss=0.1503, simple_loss=0.2394, pruned_loss=0.03057, over 12791.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.257, pruned_loss=0.03318, over 3061774.82 frames. ], batch size: 250, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:20:49,243 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=273409.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:21:38,364 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=273435.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:22:18,635 INFO [train.py:904] (6/8) Epoch 27, batch 9550, loss[loss=0.1608, simple_loss=0.254, pruned_loss=0.03377, over 12452.00 frames. ], tot_loss[loss=0.161, simple_loss=0.256, pruned_loss=0.03302, over 3064665.78 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:22:51,234 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.160e+02 2.156e+02 2.505e+02 2.980e+02 5.260e+02, threshold=5.009e+02, percent-clipped=1.0 2023-05-02 11:22:54,294 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=273470.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:23:35,379 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2793, 3.5824, 3.6007, 2.5156, 3.2269, 3.6321, 3.4370, 2.1394], device='cuda:6'), covar=tensor([0.0556, 0.0057, 0.0063, 0.0401, 0.0132, 0.0094, 0.0092, 0.0534], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0087, 0.0088, 0.0133, 0.0099, 0.0111, 0.0095, 0.0128], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-02 11:23:46,443 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=273496.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:23:51,497 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0090, 1.8273, 1.6664, 1.5398, 2.0312, 1.6358, 1.5775, 1.9511], device='cuda:6'), covar=tensor([0.0192, 0.0329, 0.0450, 0.0408, 0.0256, 0.0320, 0.0170, 0.0233], device='cuda:6'), in_proj_covar=tensor([0.0215, 0.0235, 0.0224, 0.0226, 0.0236, 0.0233, 0.0230, 0.0230], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:23:57,853 INFO [train.py:904] (6/8) Epoch 27, batch 9600, loss[loss=0.1534, simple_loss=0.2414, pruned_loss=0.03267, over 12118.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2578, pruned_loss=0.03397, over 3055724.64 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:24:29,057 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 11:24:34,275 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-02 11:25:44,212 INFO [train.py:904] (6/8) Epoch 27, batch 9650, loss[loss=0.1726, simple_loss=0.267, pruned_loss=0.03913, over 12741.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2599, pruned_loss=0.03431, over 3054021.20 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:26:23,245 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.167e+02 2.602e+02 2.993e+02 6.217e+02, threshold=5.204e+02, percent-clipped=3.0 2023-05-02 11:26:51,016 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1338, 3.2710, 1.7993, 3.5060, 2.4174, 3.4425, 1.8928, 2.6085], device='cuda:6'), covar=tensor([0.0324, 0.0353, 0.1863, 0.0268, 0.0887, 0.0574, 0.1924, 0.0776], device='cuda:6'), in_proj_covar=tensor([0.0168, 0.0172, 0.0189, 0.0163, 0.0173, 0.0209, 0.0198, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 11:27:19,647 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0102, 4.2725, 4.1226, 4.1623, 3.8236, 3.8480, 3.8826, 4.2771], device='cuda:6'), covar=tensor([0.1118, 0.0939, 0.1004, 0.0764, 0.0789, 0.1754, 0.1021, 0.0946], device='cuda:6'), in_proj_covar=tensor([0.0684, 0.0829, 0.0678, 0.0638, 0.0522, 0.0524, 0.0694, 0.0647], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:27:30,800 INFO [train.py:904] (6/8) Epoch 27, batch 9700, loss[loss=0.1839, simple_loss=0.2763, pruned_loss=0.04577, over 16783.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2593, pruned_loss=0.03417, over 3060406.42 frames. ], batch size: 124, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:27:35,251 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2195, 4.0697, 4.2697, 4.3834, 4.5164, 4.0864, 4.4934, 4.5343], device='cuda:6'), covar=tensor([0.1734, 0.1186, 0.1502, 0.0808, 0.0576, 0.1199, 0.0655, 0.0766], device='cuda:6'), in_proj_covar=tensor([0.0639, 0.0780, 0.0903, 0.0794, 0.0608, 0.0633, 0.0665, 0.0774], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:29:12,471 INFO [train.py:904] (6/8) Epoch 27, batch 9750, loss[loss=0.1587, simple_loss=0.2538, pruned_loss=0.03174, over 15358.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2582, pruned_loss=0.03467, over 3030395.29 frames. ], batch size: 190, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:29:42,115 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 2.115e+02 2.683e+02 3.309e+02 9.068e+02, threshold=5.365e+02, percent-clipped=4.0 2023-05-02 11:30:51,290 INFO [train.py:904] (6/8) Epoch 27, batch 9800, loss[loss=0.177, simple_loss=0.2831, pruned_loss=0.03542, over 16253.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2582, pruned_loss=0.03376, over 3040063.54 frames. ], batch size: 165, lr: 2.47e-03, grad_scale: 4.0 2023-05-02 11:31:16,911 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-05-02 11:32:36,071 INFO [train.py:904] (6/8) Epoch 27, batch 9850, loss[loss=0.1525, simple_loss=0.2538, pruned_loss=0.02561, over 15297.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2594, pruned_loss=0.03333, over 3054474.71 frames. ], batch size: 190, lr: 2.47e-03, grad_scale: 4.0 2023-05-02 11:32:43,912 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-05-02 11:33:00,286 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=273765.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:33:11,082 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.328e+02 2.029e+02 2.351e+02 3.028e+02 6.435e+02, threshold=4.703e+02, percent-clipped=1.0 2023-05-02 11:33:36,824 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2023-05-02 11:34:02,958 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=273791.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:34:25,789 INFO [train.py:904] (6/8) Epoch 27, batch 9900, loss[loss=0.1646, simple_loss=0.2528, pruned_loss=0.03817, over 12596.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2599, pruned_loss=0.03338, over 3047805.09 frames. ], batch size: 247, lr: 2.47e-03, grad_scale: 4.0 2023-05-02 11:35:13,091 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8408, 3.8545, 3.9400, 3.7407, 3.9000, 4.3028, 3.9566, 3.6575], device='cuda:6'), covar=tensor([0.2093, 0.2289, 0.2230, 0.2777, 0.2668, 0.1529, 0.1585, 0.2630], device='cuda:6'), in_proj_covar=tensor([0.0402, 0.0601, 0.0658, 0.0488, 0.0651, 0.0692, 0.0518, 0.0649], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 11:36:11,396 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2924, 4.2923, 4.6230, 4.6066, 4.6106, 4.3534, 4.3181, 4.3345], device='cuda:6'), covar=tensor([0.0418, 0.0719, 0.0513, 0.0478, 0.0507, 0.0524, 0.1013, 0.0531], device='cuda:6'), in_proj_covar=tensor([0.0414, 0.0463, 0.0452, 0.0414, 0.0498, 0.0476, 0.0544, 0.0381], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 11:36:22,573 INFO [train.py:904] (6/8) Epoch 27, batch 9950, loss[loss=0.1514, simple_loss=0.2504, pruned_loss=0.02624, over 16463.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2625, pruned_loss=0.0341, over 3047010.74 frames. ], batch size: 68, lr: 2.47e-03, grad_scale: 4.0 2023-05-02 11:37:03,258 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 1.975e+02 2.538e+02 3.032e+02 5.048e+02, threshold=5.077e+02, percent-clipped=2.0 2023-05-02 11:38:25,114 INFO [train.py:904] (6/8) Epoch 27, batch 10000, loss[loss=0.1718, simple_loss=0.2696, pruned_loss=0.03702, over 16825.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2611, pruned_loss=0.03381, over 3062232.51 frames. ], batch size: 124, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:40:06,628 INFO [train.py:904] (6/8) Epoch 27, batch 10050, loss[loss=0.1693, simple_loss=0.268, pruned_loss=0.03527, over 16671.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2607, pruned_loss=0.03353, over 3057798.20 frames. ], batch size: 134, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:40:24,788 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8439, 3.7994, 3.9556, 3.6984, 3.8686, 4.2856, 3.9673, 3.6887], device='cuda:6'), covar=tensor([0.2135, 0.2564, 0.2739, 0.2755, 0.3025, 0.1685, 0.1764, 0.2933], device='cuda:6'), in_proj_covar=tensor([0.0404, 0.0604, 0.0663, 0.0492, 0.0654, 0.0697, 0.0523, 0.0654], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 11:40:39,173 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.137e+02 2.431e+02 3.038e+02 5.006e+02, threshold=4.862e+02, percent-clipped=0.0 2023-05-02 11:41:39,214 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0041, 3.1770, 3.6834, 2.0410, 3.0180, 2.2930, 3.4145, 3.3553], device='cuda:6'), covar=tensor([0.0291, 0.0904, 0.0480, 0.2271, 0.0819, 0.1036, 0.0709, 0.1108], device='cuda:6'), in_proj_covar=tensor([0.0155, 0.0162, 0.0163, 0.0152, 0.0143, 0.0128, 0.0140, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 11:41:41,059 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=274001.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:41:43,715 INFO [train.py:904] (6/8) Epoch 27, batch 10100, loss[loss=0.1714, simple_loss=0.2649, pruned_loss=0.03894, over 16587.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2614, pruned_loss=0.03395, over 3056316.33 frames. ], batch size: 57, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:42:12,047 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-05-02 11:43:27,165 INFO [train.py:904] (6/8) Epoch 28, batch 0, loss[loss=0.2047, simple_loss=0.2819, pruned_loss=0.06368, over 16321.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2819, pruned_loss=0.06368, over 16321.00 frames. ], batch size: 165, lr: 2.42e-03, grad_scale: 8.0 2023-05-02 11:43:27,165 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 11:43:34,595 INFO [train.py:938] (6/8) Epoch 28, validation: loss=0.1434, simple_loss=0.2464, pruned_loss=0.02022, over 944034.00 frames. 2023-05-02 11:43:34,596 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 11:43:48,089 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=274062.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:43:52,204 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=274065.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:44:01,201 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 2.305e+02 2.657e+02 3.293e+02 5.505e+02, threshold=5.314e+02, percent-clipped=2.0 2023-05-02 11:44:19,278 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5848, 4.5350, 4.8930, 4.8810, 4.9479, 4.6112, 4.5854, 4.4848], device='cuda:6'), covar=tensor([0.0441, 0.1082, 0.0623, 0.0598, 0.0615, 0.0634, 0.1233, 0.0680], device='cuda:6'), in_proj_covar=tensor([0.0414, 0.0464, 0.0454, 0.0416, 0.0499, 0.0477, 0.0544, 0.0382], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 11:44:27,632 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=274091.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:44:43,859 INFO [train.py:904] (6/8) Epoch 28, batch 50, loss[loss=0.1389, simple_loss=0.2226, pruned_loss=0.02766, over 17020.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2622, pruned_loss=0.04483, over 750835.13 frames. ], batch size: 41, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 11:44:45,180 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 11:44:59,068 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=274113.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:45:24,130 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=274131.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:45:34,071 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=274139.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:45:54,004 INFO [train.py:904] (6/8) Epoch 28, batch 100, loss[loss=0.1867, simple_loss=0.2639, pruned_loss=0.05473, over 16869.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2616, pruned_loss=0.04333, over 1311720.07 frames. ], batch size: 109, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 11:46:20,295 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 2.302e+02 2.829e+02 3.561e+02 6.259e+02, threshold=5.657e+02, percent-clipped=1.0 2023-05-02 11:46:23,004 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8970, 5.2591, 5.0169, 4.9605, 4.7411, 4.6294, 4.7224, 5.3289], device='cuda:6'), covar=tensor([0.1376, 0.0978, 0.1168, 0.1020, 0.0908, 0.1199, 0.1359, 0.0980], device='cuda:6'), in_proj_covar=tensor([0.0698, 0.0846, 0.0689, 0.0650, 0.0533, 0.0533, 0.0708, 0.0661], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:46:40,499 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-02 11:46:48,684 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=274192.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:46:52,464 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.59 vs. limit=5.0 2023-05-02 11:47:02,023 INFO [train.py:904] (6/8) Epoch 28, batch 150, loss[loss=0.1536, simple_loss=0.2455, pruned_loss=0.0309, over 17128.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.26, pruned_loss=0.04189, over 1758896.91 frames. ], batch size: 48, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 11:48:08,591 INFO [train.py:904] (6/8) Epoch 28, batch 200, loss[loss=0.1474, simple_loss=0.2346, pruned_loss=0.03013, over 16832.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2599, pruned_loss=0.04124, over 2106964.37 frames. ], batch size: 42, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 11:48:30,322 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=274269.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:48:33,858 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.228e+02 2.594e+02 3.067e+02 5.865e+02, threshold=5.188e+02, percent-clipped=1.0 2023-05-02 11:49:11,536 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3879, 2.3749, 2.4567, 4.3182, 2.3238, 2.7741, 2.4843, 2.5699], device='cuda:6'), covar=tensor([0.1407, 0.3744, 0.3176, 0.0534, 0.4196, 0.2625, 0.3663, 0.3689], device='cuda:6'), in_proj_covar=tensor([0.0415, 0.0468, 0.0382, 0.0331, 0.0442, 0.0532, 0.0440, 0.0544], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:49:16,128 INFO [train.py:904] (6/8) Epoch 28, batch 250, loss[loss=0.1651, simple_loss=0.265, pruned_loss=0.0326, over 17060.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2582, pruned_loss=0.04192, over 2370451.96 frames. ], batch size: 50, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 11:49:16,579 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=274303.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:49:51,990 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=274330.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:50:05,621 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5889, 2.8060, 3.1419, 2.0928, 2.7757, 2.1688, 3.2138, 3.1885], device='cuda:6'), covar=tensor([0.0300, 0.1074, 0.0610, 0.2145, 0.0962, 0.1113, 0.0648, 0.1034], device='cuda:6'), in_proj_covar=tensor([0.0158, 0.0166, 0.0167, 0.0155, 0.0146, 0.0131, 0.0143, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 11:50:19,896 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7824, 4.9538, 5.0915, 4.8632, 4.9087, 5.5267, 5.0345, 4.7007], device='cuda:6'), covar=tensor([0.1350, 0.2174, 0.2803, 0.2132, 0.2551, 0.1015, 0.1706, 0.2440], device='cuda:6'), in_proj_covar=tensor([0.0413, 0.0619, 0.0680, 0.0502, 0.0671, 0.0711, 0.0533, 0.0668], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 11:50:23,017 INFO [train.py:904] (6/8) Epoch 28, batch 300, loss[loss=0.1485, simple_loss=0.2306, pruned_loss=0.03321, over 16845.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2557, pruned_loss=0.04137, over 2587691.73 frames. ], batch size: 96, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:50:30,574 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=274357.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:50:39,364 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=274364.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:50:49,959 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.173e+02 2.525e+02 3.031e+02 5.896e+02, threshold=5.050e+02, percent-clipped=1.0 2023-05-02 11:51:31,008 INFO [train.py:904] (6/8) Epoch 28, batch 350, loss[loss=0.136, simple_loss=0.2285, pruned_loss=0.0218, over 17189.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2532, pruned_loss=0.04037, over 2747640.99 frames. ], batch size: 46, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:52:37,186 INFO [train.py:904] (6/8) Epoch 28, batch 400, loss[loss=0.1508, simple_loss=0.2375, pruned_loss=0.03203, over 15782.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2522, pruned_loss=0.03991, over 2880117.29 frames. ], batch size: 35, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 11:53:03,891 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.265e+02 2.649e+02 3.069e+02 5.270e+02, threshold=5.298e+02, percent-clipped=3.0 2023-05-02 11:53:24,165 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=274487.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:53:44,149 INFO [train.py:904] (6/8) Epoch 28, batch 450, loss[loss=0.1611, simple_loss=0.2587, pruned_loss=0.03173, over 17124.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2516, pruned_loss=0.03966, over 2981716.39 frames. ], batch size: 49, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:54:35,093 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=274540.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:54:52,744 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1376, 1.9362, 2.6921, 3.1536, 2.9371, 3.5508, 2.1164, 3.5125], device='cuda:6'), covar=tensor([0.0281, 0.0747, 0.0375, 0.0344, 0.0371, 0.0230, 0.0813, 0.0196], device='cuda:6'), in_proj_covar=tensor([0.0198, 0.0199, 0.0187, 0.0192, 0.0207, 0.0166, 0.0204, 0.0165], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 11:54:53,414 INFO [train.py:904] (6/8) Epoch 28, batch 500, loss[loss=0.1721, simple_loss=0.2703, pruned_loss=0.03698, over 17144.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2503, pruned_loss=0.03892, over 3060926.74 frames. ], batch size: 48, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:55:21,738 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 2.207e+02 2.697e+02 3.239e+02 1.435e+03, threshold=5.394e+02, percent-clipped=3.0 2023-05-02 11:55:58,925 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=274601.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:56:00,838 INFO [train.py:904] (6/8) Epoch 28, batch 550, loss[loss=0.1494, simple_loss=0.2311, pruned_loss=0.03389, over 15863.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2488, pruned_loss=0.03829, over 3111033.63 frames. ], batch size: 35, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:56:31,523 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=274625.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:57:09,862 INFO [train.py:904] (6/8) Epoch 28, batch 600, loss[loss=0.1621, simple_loss=0.2363, pruned_loss=0.044, over 16732.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.248, pruned_loss=0.03792, over 3160673.45 frames. ], batch size: 124, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:57:15,320 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=274657.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:57:17,513 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=274659.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:57:20,121 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 11:57:36,632 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.130e+02 2.496e+02 2.954e+02 2.616e+03, threshold=4.992e+02, percent-clipped=3.0 2023-05-02 11:57:43,499 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.57 vs. limit=5.0 2023-05-02 11:58:09,115 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6030, 3.6268, 4.2074, 2.5134, 3.3622, 2.5798, 4.1870, 3.8168], device='cuda:6'), covar=tensor([0.0238, 0.1000, 0.0452, 0.1948, 0.0775, 0.1014, 0.0519, 0.1187], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0168, 0.0169, 0.0157, 0.0148, 0.0132, 0.0145, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 11:58:16,066 INFO [train.py:904] (6/8) Epoch 28, batch 650, loss[loss=0.1507, simple_loss=0.2315, pruned_loss=0.03492, over 16887.00 frames. ], tot_loss[loss=0.1602, simple_loss=0.2462, pruned_loss=0.03708, over 3201177.89 frames. ], batch size: 109, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:58:18,639 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=274705.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:58:32,128 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.00 vs. limit=5.0 2023-05-02 11:59:11,201 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-02 11:59:21,413 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-05-02 11:59:25,786 INFO [train.py:904] (6/8) Epoch 28, batch 700, loss[loss=0.195, simple_loss=0.272, pruned_loss=0.05897, over 16701.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.2465, pruned_loss=0.03728, over 3212925.69 frames. ], batch size: 134, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:59:54,342 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.355e+02 2.008e+02 2.338e+02 2.859e+02 5.050e+02, threshold=4.676e+02, percent-clipped=1.0 2023-05-02 12:00:02,920 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2979, 5.2227, 5.1113, 4.5537, 4.7111, 5.1631, 5.1375, 4.7590], device='cuda:6'), covar=tensor([0.0635, 0.0608, 0.0414, 0.0427, 0.1328, 0.0576, 0.0321, 0.0918], device='cuda:6'), in_proj_covar=tensor([0.0309, 0.0463, 0.0361, 0.0362, 0.0358, 0.0417, 0.0248, 0.0432], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 12:00:13,273 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=274787.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:00:15,035 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-05-02 12:00:34,709 INFO [train.py:904] (6/8) Epoch 28, batch 750, loss[loss=0.1798, simple_loss=0.2628, pruned_loss=0.04839, over 16511.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2475, pruned_loss=0.03778, over 3227309.05 frames. ], batch size: 75, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 12:01:14,277 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1521, 5.1434, 4.9138, 4.3730, 5.0048, 1.8351, 4.7751, 4.7834], device='cuda:6'), covar=tensor([0.0103, 0.0099, 0.0248, 0.0432, 0.0127, 0.3007, 0.0147, 0.0259], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0171, 0.0208, 0.0180, 0.0185, 0.0215, 0.0197, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 12:01:19,980 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=274835.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:01:45,403 INFO [train.py:904] (6/8) Epoch 28, batch 800, loss[loss=0.1443, simple_loss=0.2265, pruned_loss=0.03102, over 16835.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.248, pruned_loss=0.03747, over 3245598.78 frames. ], batch size: 42, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:02:15,002 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.216e+02 2.000e+02 2.423e+02 2.829e+02 5.216e+02, threshold=4.846e+02, percent-clipped=1.0 2023-05-02 12:02:43,407 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=274894.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:02:45,596 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-02 12:02:46,867 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=274896.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:02:57,015 INFO [train.py:904] (6/8) Epoch 28, batch 850, loss[loss=0.1436, simple_loss=0.2242, pruned_loss=0.03146, over 16914.00 frames. ], tot_loss[loss=0.1612, simple_loss=0.2478, pruned_loss=0.03732, over 3255729.16 frames. ], batch size: 96, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:03:26,690 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=274925.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:04:05,739 INFO [train.py:904] (6/8) Epoch 28, batch 900, loss[loss=0.1695, simple_loss=0.2596, pruned_loss=0.0397, over 17048.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.2473, pruned_loss=0.03686, over 3278945.69 frames. ], batch size: 55, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:04:09,170 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=274955.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:04:14,434 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=274959.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:04:33,653 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=274973.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:04:34,534 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.476e+02 2.075e+02 2.378e+02 2.870e+02 6.521e+02, threshold=4.756e+02, percent-clipped=4.0 2023-05-02 12:05:14,546 INFO [train.py:904] (6/8) Epoch 28, batch 950, loss[loss=0.1473, simple_loss=0.2427, pruned_loss=0.02592, over 17206.00 frames. ], tot_loss[loss=0.1604, simple_loss=0.2473, pruned_loss=0.03676, over 3296510.50 frames. ], batch size: 46, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:05:20,985 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=275007.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:05:22,299 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0582, 5.0633, 5.4944, 5.4956, 5.4999, 5.1624, 5.1074, 4.9421], device='cuda:6'), covar=tensor([0.0388, 0.0570, 0.0414, 0.0385, 0.0443, 0.0419, 0.0940, 0.0445], device='cuda:6'), in_proj_covar=tensor([0.0440, 0.0494, 0.0478, 0.0439, 0.0525, 0.0503, 0.0576, 0.0402], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 12:06:23,168 INFO [train.py:904] (6/8) Epoch 28, batch 1000, loss[loss=0.1632, simple_loss=0.2586, pruned_loss=0.03391, over 17045.00 frames. ], tot_loss[loss=0.1602, simple_loss=0.2461, pruned_loss=0.03718, over 3297072.49 frames. ], batch size: 55, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:06:52,343 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.067e+02 2.369e+02 2.767e+02 9.232e+02, threshold=4.738e+02, percent-clipped=2.0 2023-05-02 12:07:03,481 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8867, 2.4109, 2.4896, 3.3366, 2.5560, 3.6136, 1.7788, 2.8934], device='cuda:6'), covar=tensor([0.1329, 0.0801, 0.1175, 0.0233, 0.0151, 0.0404, 0.1528, 0.0775], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0180, 0.0200, 0.0201, 0.0205, 0.0219, 0.0209, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 12:07:31,781 INFO [train.py:904] (6/8) Epoch 28, batch 1050, loss[loss=0.1623, simple_loss=0.2454, pruned_loss=0.03957, over 15470.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.245, pruned_loss=0.03723, over 3299284.85 frames. ], batch size: 190, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:08:39,620 INFO [train.py:904] (6/8) Epoch 28, batch 1100, loss[loss=0.1761, simple_loss=0.249, pruned_loss=0.05155, over 16695.00 frames. ], tot_loss[loss=0.1596, simple_loss=0.2448, pruned_loss=0.03724, over 3309065.90 frames. ], batch size: 134, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:08:47,450 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=275158.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:09:07,965 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.388e+02 1.984e+02 2.393e+02 2.792e+02 5.361e+02, threshold=4.787e+02, percent-clipped=2.0 2023-05-02 12:09:39,602 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=275196.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:09:39,968 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2023-05-02 12:09:49,907 INFO [train.py:904] (6/8) Epoch 28, batch 1150, loss[loss=0.1385, simple_loss=0.2183, pruned_loss=0.02939, over 16418.00 frames. ], tot_loss[loss=0.1587, simple_loss=0.2439, pruned_loss=0.0367, over 3307008.91 frames. ], batch size: 75, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:09:51,358 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-02 12:09:58,837 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.60 vs. limit=2.0 2023-05-02 12:10:12,486 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=275219.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:10:22,096 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4041, 5.7339, 5.5556, 5.6069, 5.2208, 5.2338, 5.1502, 5.9047], device='cuda:6'), covar=tensor([0.1483, 0.1189, 0.1048, 0.0955, 0.0918, 0.0746, 0.1340, 0.0959], device='cuda:6'), in_proj_covar=tensor([0.0724, 0.0884, 0.0717, 0.0679, 0.0557, 0.0553, 0.0741, 0.0687], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 12:10:44,954 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=275244.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:10:53,443 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=275250.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:10:57,364 INFO [train.py:904] (6/8) Epoch 28, batch 1200, loss[loss=0.1841, simple_loss=0.2536, pruned_loss=0.05726, over 16477.00 frames. ], tot_loss[loss=0.1583, simple_loss=0.2434, pruned_loss=0.03666, over 3314096.32 frames. ], batch size: 146, lr: 2.42e-03, grad_scale: 4.0 2023-05-02 12:11:20,269 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3220, 3.4614, 3.6274, 2.5585, 3.3327, 3.7421, 3.4657, 2.2071], device='cuda:6'), covar=tensor([0.0524, 0.0229, 0.0077, 0.0418, 0.0150, 0.0120, 0.0113, 0.0533], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0090, 0.0091, 0.0135, 0.0102, 0.0114, 0.0098, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 12:11:21,752 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 12:11:26,948 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.036e+02 2.270e+02 2.715e+02 4.480e+02, threshold=4.540e+02, percent-clipped=0.0 2023-05-02 12:12:06,865 INFO [train.py:904] (6/8) Epoch 28, batch 1250, loss[loss=0.1671, simple_loss=0.2492, pruned_loss=0.04248, over 16668.00 frames. ], tot_loss[loss=0.1576, simple_loss=0.243, pruned_loss=0.03614, over 3308352.28 frames. ], batch size: 62, lr: 2.42e-03, grad_scale: 4.0 2023-05-02 12:12:12,678 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=275307.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:13:09,290 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=275347.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 12:13:17,199 INFO [train.py:904] (6/8) Epoch 28, batch 1300, loss[loss=0.167, simple_loss=0.2487, pruned_loss=0.04268, over 16849.00 frames. ], tot_loss[loss=0.1575, simple_loss=0.2426, pruned_loss=0.03618, over 3309450.15 frames. ], batch size: 116, lr: 2.42e-03, grad_scale: 4.0 2023-05-02 12:13:26,050 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-05-02 12:13:39,150 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=275368.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:13:46,873 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 1.989e+02 2.411e+02 2.818e+02 5.022e+02, threshold=4.822e+02, percent-clipped=1.0 2023-05-02 12:14:02,893 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=275385.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:14:12,683 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0269, 4.9501, 4.8756, 4.4058, 4.5306, 4.9144, 4.7521, 4.5759], device='cuda:6'), covar=tensor([0.0560, 0.0642, 0.0333, 0.0381, 0.1050, 0.0556, 0.0461, 0.0769], device='cuda:6'), in_proj_covar=tensor([0.0317, 0.0475, 0.0369, 0.0372, 0.0367, 0.0426, 0.0254, 0.0443], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 12:14:26,876 INFO [train.py:904] (6/8) Epoch 28, batch 1350, loss[loss=0.1731, simple_loss=0.2521, pruned_loss=0.04706, over 16640.00 frames. ], tot_loss[loss=0.1571, simple_loss=0.2424, pruned_loss=0.03592, over 3312389.19 frames. ], batch size: 134, lr: 2.42e-03, grad_scale: 4.0 2023-05-02 12:14:35,288 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=275408.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 12:14:40,647 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3429, 3.3736, 3.4646, 2.5191, 3.1630, 3.5598, 3.2808, 2.2524], device='cuda:6'), covar=tensor([0.0461, 0.0133, 0.0074, 0.0369, 0.0140, 0.0116, 0.0129, 0.0430], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0090, 0.0091, 0.0135, 0.0101, 0.0114, 0.0098, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 12:15:28,597 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=275446.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:15:37,714 INFO [train.py:904] (6/8) Epoch 28, batch 1400, loss[loss=0.1564, simple_loss=0.2398, pruned_loss=0.03656, over 16772.00 frames. ], tot_loss[loss=0.1572, simple_loss=0.2427, pruned_loss=0.03591, over 3317862.90 frames. ], batch size: 102, lr: 2.42e-03, grad_scale: 4.0 2023-05-02 12:16:00,708 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5178, 5.9893, 5.4520, 5.8906, 5.3734, 5.1491, 5.5390, 6.0451], device='cuda:6'), covar=tensor([0.2498, 0.1403, 0.2621, 0.1596, 0.1877, 0.1343, 0.2420, 0.1927], device='cuda:6'), in_proj_covar=tensor([0.0722, 0.0880, 0.0715, 0.0676, 0.0554, 0.0553, 0.0738, 0.0686], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 12:16:06,676 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.102e+02 2.484e+02 3.057e+02 6.401e+02, threshold=4.968e+02, percent-clipped=1.0 2023-05-02 12:16:07,002 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9804, 5.3194, 5.5028, 5.1779, 5.2848, 5.8939, 5.3036, 5.0108], device='cuda:6'), covar=tensor([0.1202, 0.2204, 0.2621, 0.2199, 0.2744, 0.0993, 0.1830, 0.2477], device='cuda:6'), in_proj_covar=tensor([0.0429, 0.0642, 0.0705, 0.0524, 0.0695, 0.0732, 0.0552, 0.0692], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 12:16:46,092 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1143, 4.2651, 4.3441, 3.2164, 3.5340, 4.3001, 3.9686, 2.5908], device='cuda:6'), covar=tensor([0.0460, 0.0095, 0.0054, 0.0378, 0.0163, 0.0098, 0.0098, 0.0498], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0091, 0.0091, 0.0136, 0.0102, 0.0115, 0.0099, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 12:16:46,828 INFO [train.py:904] (6/8) Epoch 28, batch 1450, loss[loss=0.1608, simple_loss=0.2655, pruned_loss=0.02799, over 17033.00 frames. ], tot_loss[loss=0.157, simple_loss=0.2425, pruned_loss=0.03578, over 3323040.87 frames. ], batch size: 50, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:17:03,348 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=275514.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:17:54,078 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=275550.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:17:57,217 INFO [train.py:904] (6/8) Epoch 28, batch 1500, loss[loss=0.1805, simple_loss=0.2548, pruned_loss=0.05314, over 16427.00 frames. ], tot_loss[loss=0.1577, simple_loss=0.2428, pruned_loss=0.03636, over 3321710.20 frames. ], batch size: 75, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:18:22,142 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0743, 4.1246, 4.4111, 4.4109, 4.4481, 4.1792, 4.1940, 4.1184], device='cuda:6'), covar=tensor([0.0448, 0.0762, 0.0458, 0.0421, 0.0515, 0.0474, 0.0865, 0.0621], device='cuda:6'), in_proj_covar=tensor([0.0438, 0.0492, 0.0475, 0.0437, 0.0522, 0.0501, 0.0574, 0.0401], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 12:18:26,332 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.124e+02 2.433e+02 3.002e+02 9.704e+02, threshold=4.866e+02, percent-clipped=1.0 2023-05-02 12:19:01,053 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=275598.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:19:07,363 INFO [train.py:904] (6/8) Epoch 28, batch 1550, loss[loss=0.1755, simple_loss=0.246, pruned_loss=0.05251, over 16834.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.244, pruned_loss=0.03764, over 3325231.18 frames. ], batch size: 102, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:20:18,406 INFO [train.py:904] (6/8) Epoch 28, batch 1600, loss[loss=0.1521, simple_loss=0.2371, pruned_loss=0.03358, over 16862.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2463, pruned_loss=0.03816, over 3325335.15 frames. ], batch size: 90, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:20:33,167 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=275663.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:20:34,597 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1484, 2.2366, 2.3728, 3.7510, 2.2550, 2.5639, 2.3101, 2.3838], device='cuda:6'), covar=tensor([0.1585, 0.3672, 0.3084, 0.0770, 0.3766, 0.2693, 0.3904, 0.3129], device='cuda:6'), in_proj_covar=tensor([0.0422, 0.0476, 0.0389, 0.0339, 0.0448, 0.0545, 0.0447, 0.0556], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 12:20:48,273 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.650e+02 2.228e+02 2.653e+02 3.158e+02 7.607e+02, threshold=5.306e+02, percent-clipped=1.0 2023-05-02 12:20:54,472 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2552, 5.2305, 4.9868, 4.4940, 5.0980, 1.7658, 4.8340, 4.9036], device='cuda:6'), covar=tensor([0.0132, 0.0115, 0.0262, 0.0476, 0.0118, 0.3183, 0.0177, 0.0258], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0175, 0.0212, 0.0185, 0.0189, 0.0219, 0.0201, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 12:21:04,689 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8046, 3.9125, 2.6274, 4.6047, 3.0989, 4.4940, 2.7932, 3.3168], device='cuda:6'), covar=tensor([0.0341, 0.0394, 0.1657, 0.0289, 0.0843, 0.0591, 0.1454, 0.0770], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0183, 0.0198, 0.0175, 0.0181, 0.0223, 0.0206, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 12:21:28,983 INFO [train.py:904] (6/8) Epoch 28, batch 1650, loss[loss=0.1788, simple_loss=0.2672, pruned_loss=0.04518, over 16531.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2481, pruned_loss=0.03854, over 3313527.83 frames. ], batch size: 68, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:21:29,331 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=275703.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 12:21:58,823 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-02 12:22:21,216 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=275741.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:22:37,917 INFO [train.py:904] (6/8) Epoch 28, batch 1700, loss[loss=0.1426, simple_loss=0.2291, pruned_loss=0.02804, over 17000.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2505, pruned_loss=0.03907, over 3306698.50 frames. ], batch size: 41, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:23:08,710 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.133e+02 2.503e+02 2.990e+02 4.977e+02, threshold=5.005e+02, percent-clipped=0.0 2023-05-02 12:23:26,801 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.57 vs. limit=5.0 2023-05-02 12:23:49,163 INFO [train.py:904] (6/8) Epoch 28, batch 1750, loss[loss=0.1441, simple_loss=0.2334, pruned_loss=0.02739, over 16779.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2514, pruned_loss=0.03918, over 3315152.81 frames. ], batch size: 39, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:24:05,553 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=275814.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:24:20,392 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8163, 2.7370, 2.6594, 1.9167, 2.6737, 2.8046, 2.6348, 1.8803], device='cuda:6'), covar=tensor([0.0490, 0.0131, 0.0103, 0.0429, 0.0156, 0.0132, 0.0128, 0.0466], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0092, 0.0092, 0.0137, 0.0103, 0.0116, 0.0100, 0.0134], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 12:24:29,139 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4490, 5.8423, 5.5848, 5.6777, 5.2394, 5.2059, 5.2300, 5.9701], device='cuda:6'), covar=tensor([0.1624, 0.1034, 0.1199, 0.1013, 0.1041, 0.0849, 0.1425, 0.1047], device='cuda:6'), in_proj_covar=tensor([0.0728, 0.0887, 0.0721, 0.0681, 0.0559, 0.0558, 0.0746, 0.0691], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 12:24:50,735 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4007, 3.4950, 2.2044, 3.6739, 2.8121, 3.6364, 2.3648, 2.8353], device='cuda:6'), covar=tensor([0.0343, 0.0448, 0.1615, 0.0369, 0.0800, 0.0825, 0.1504, 0.0784], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0183, 0.0199, 0.0175, 0.0181, 0.0223, 0.0207, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 12:24:59,129 INFO [train.py:904] (6/8) Epoch 28, batch 1800, loss[loss=0.187, simple_loss=0.2955, pruned_loss=0.03926, over 17269.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2532, pruned_loss=0.03928, over 3308366.75 frames. ], batch size: 52, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:25:08,978 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=275860.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:25:11,868 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=275862.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:25:19,214 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0625, 4.7604, 4.9295, 5.2506, 5.4014, 4.8341, 5.4768, 5.4350], device='cuda:6'), covar=tensor([0.2056, 0.1547, 0.2564, 0.1063, 0.0870, 0.1051, 0.0869, 0.0980], device='cuda:6'), in_proj_covar=tensor([0.0691, 0.0842, 0.0980, 0.0856, 0.0652, 0.0683, 0.0718, 0.0831], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 12:25:30,540 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.316e+02 2.618e+02 3.157e+02 1.143e+03, threshold=5.235e+02, percent-clipped=6.0 2023-05-02 12:25:59,917 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4560, 4.3722, 4.3760, 4.0640, 4.1398, 4.4060, 4.1421, 4.1731], device='cuda:6'), covar=tensor([0.0609, 0.0870, 0.0306, 0.0304, 0.0740, 0.0593, 0.0595, 0.0637], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0478, 0.0372, 0.0374, 0.0370, 0.0429, 0.0255, 0.0446], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 12:26:08,320 INFO [train.py:904] (6/8) Epoch 28, batch 1850, loss[loss=0.1686, simple_loss=0.2555, pruned_loss=0.04091, over 16258.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2542, pruned_loss=0.03962, over 3305214.06 frames. ], batch size: 165, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:26:33,752 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=275921.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:27:16,673 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=275952.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:27:17,449 INFO [train.py:904] (6/8) Epoch 28, batch 1900, loss[loss=0.144, simple_loss=0.2428, pruned_loss=0.02258, over 17140.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2527, pruned_loss=0.03866, over 3303227.13 frames. ], batch size: 47, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:27:32,207 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=275963.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:27:47,687 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.245e+02 2.028e+02 2.382e+02 2.719e+02 5.605e+02, threshold=4.765e+02, percent-clipped=1.0 2023-05-02 12:28:17,383 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-02 12:28:30,501 INFO [train.py:904] (6/8) Epoch 28, batch 1950, loss[loss=0.144, simple_loss=0.2286, pruned_loss=0.02965, over 16827.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2516, pruned_loss=0.03802, over 3315070.10 frames. ], batch size: 39, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:28:30,933 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=276003.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 12:28:40,998 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=276011.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:28:44,271 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276013.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:29:24,704 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=276041.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:29:37,514 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=276051.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 12:29:40,563 INFO [train.py:904] (6/8) Epoch 28, batch 2000, loss[loss=0.155, simple_loss=0.2553, pruned_loss=0.02736, over 17125.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2508, pruned_loss=0.03752, over 3317903.69 frames. ], batch size: 49, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:30:00,525 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4158, 4.4199, 4.7412, 4.7195, 4.7825, 4.4882, 4.4660, 4.3594], device='cuda:6'), covar=tensor([0.0403, 0.0654, 0.0418, 0.0441, 0.0530, 0.0435, 0.0968, 0.0646], device='cuda:6'), in_proj_covar=tensor([0.0442, 0.0497, 0.0480, 0.0442, 0.0527, 0.0507, 0.0582, 0.0405], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 12:30:11,361 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.574e+02 2.104e+02 2.511e+02 2.919e+02 7.567e+02, threshold=5.023e+02, percent-clipped=3.0 2023-05-02 12:30:31,280 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=276089.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:30:44,190 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4725, 5.4436, 5.2695, 4.7957, 5.3744, 2.1989, 5.1506, 5.1938], device='cuda:6'), covar=tensor([0.0108, 0.0079, 0.0203, 0.0347, 0.0092, 0.2686, 0.0123, 0.0189], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0176, 0.0213, 0.0186, 0.0190, 0.0219, 0.0202, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 12:30:50,228 INFO [train.py:904] (6/8) Epoch 28, batch 2050, loss[loss=0.1761, simple_loss=0.2604, pruned_loss=0.04589, over 16262.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2516, pruned_loss=0.03809, over 3296485.36 frames. ], batch size: 165, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:31:25,713 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-05-02 12:31:26,592 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7101, 3.7828, 2.9116, 2.2721, 2.4559, 2.4277, 3.8610, 3.3131], device='cuda:6'), covar=tensor([0.2758, 0.0572, 0.1717, 0.3146, 0.2926, 0.2241, 0.0576, 0.1508], device='cuda:6'), in_proj_covar=tensor([0.0335, 0.0276, 0.0313, 0.0327, 0.0305, 0.0278, 0.0305, 0.0353], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 12:32:00,357 INFO [train.py:904] (6/8) Epoch 28, batch 2100, loss[loss=0.1463, simple_loss=0.2393, pruned_loss=0.02661, over 17059.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.252, pruned_loss=0.03829, over 3311218.77 frames. ], batch size: 50, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:32:30,698 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.546e+02 2.091e+02 2.490e+02 2.967e+02 8.308e+02, threshold=4.979e+02, percent-clipped=2.0 2023-05-02 12:33:09,362 INFO [train.py:904] (6/8) Epoch 28, batch 2150, loss[loss=0.1603, simple_loss=0.2472, pruned_loss=0.03673, over 16479.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2532, pruned_loss=0.03906, over 3307024.65 frames. ], batch size: 68, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:33:27,944 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=276216.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:34:18,066 INFO [train.py:904] (6/8) Epoch 28, batch 2200, loss[loss=0.1678, simple_loss=0.2626, pruned_loss=0.03653, over 17070.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2531, pruned_loss=0.03896, over 3309219.63 frames. ], batch size: 55, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:34:50,518 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.368e+02 2.258e+02 2.667e+02 3.359e+02 8.724e+02, threshold=5.333e+02, percent-clipped=6.0 2023-05-02 12:35:27,983 INFO [train.py:904] (6/8) Epoch 28, batch 2250, loss[loss=0.1776, simple_loss=0.2574, pruned_loss=0.04894, over 16876.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2536, pruned_loss=0.03957, over 3305670.90 frames. ], batch size: 116, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:35:35,416 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=276308.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:35:48,021 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=276317.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:36:37,083 INFO [train.py:904] (6/8) Epoch 28, batch 2300, loss[loss=0.1895, simple_loss=0.2668, pruned_loss=0.05613, over 16769.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2538, pruned_loss=0.03953, over 3308987.81 frames. ], batch size: 124, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:36:55,556 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8430, 2.9772, 2.7487, 4.9379, 3.8483, 4.3531, 1.8534, 3.0203], device='cuda:6'), covar=tensor([0.1473, 0.0837, 0.1274, 0.0267, 0.0320, 0.0460, 0.1675, 0.0906], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0181, 0.0201, 0.0204, 0.0207, 0.0220, 0.0210, 0.0199], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 12:37:08,703 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.645e+02 2.043e+02 2.409e+02 2.878e+02 4.638e+02, threshold=4.818e+02, percent-clipped=0.0 2023-05-02 12:37:12,860 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276378.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:37:46,502 INFO [train.py:904] (6/8) Epoch 28, batch 2350, loss[loss=0.1746, simple_loss=0.2484, pruned_loss=0.0504, over 16921.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2539, pruned_loss=0.03931, over 3320765.99 frames. ], batch size: 109, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:38:09,022 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0887, 3.2512, 3.3015, 2.2683, 3.0850, 3.4230, 3.1178, 2.0461], device='cuda:6'), covar=tensor([0.0551, 0.0133, 0.0080, 0.0442, 0.0150, 0.0100, 0.0129, 0.0514], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0091, 0.0092, 0.0136, 0.0103, 0.0116, 0.0100, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 12:38:54,370 INFO [train.py:904] (6/8) Epoch 28, batch 2400, loss[loss=0.203, simple_loss=0.279, pruned_loss=0.06353, over 16705.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2548, pruned_loss=0.03951, over 3317280.95 frames. ], batch size: 134, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:39:16,520 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7750, 4.9096, 5.0825, 4.8450, 4.8757, 5.5118, 4.9883, 4.6671], device='cuda:6'), covar=tensor([0.1490, 0.2033, 0.2707, 0.2249, 0.2689, 0.1015, 0.1643, 0.2596], device='cuda:6'), in_proj_covar=tensor([0.0434, 0.0650, 0.0716, 0.0531, 0.0701, 0.0740, 0.0556, 0.0702], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 12:39:26,374 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.076e+02 2.394e+02 2.729e+02 1.080e+03, threshold=4.789e+02, percent-clipped=2.0 2023-05-02 12:39:59,703 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-02 12:40:04,310 INFO [train.py:904] (6/8) Epoch 28, batch 2450, loss[loss=0.2093, simple_loss=0.297, pruned_loss=0.06084, over 15905.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2553, pruned_loss=0.03923, over 3316665.88 frames. ], batch size: 35, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:40:23,730 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=276516.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:40:58,044 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7724, 4.7004, 4.6617, 4.3401, 4.2742, 4.7287, 4.5730, 4.4087], device='cuda:6'), covar=tensor([0.0788, 0.1090, 0.0474, 0.0444, 0.1184, 0.0675, 0.0575, 0.0826], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0482, 0.0375, 0.0378, 0.0371, 0.0432, 0.0257, 0.0450], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 12:41:13,986 INFO [train.py:904] (6/8) Epoch 28, batch 2500, loss[loss=0.171, simple_loss=0.25, pruned_loss=0.046, over 16739.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2548, pruned_loss=0.03904, over 3323668.46 frames. ], batch size: 83, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:41:25,129 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5685, 3.0397, 3.4791, 1.8814, 3.5258, 3.5840, 3.0082, 2.7560], device='cuda:6'), covar=tensor([0.0730, 0.0313, 0.0208, 0.1216, 0.0126, 0.0230, 0.0418, 0.0446], device='cuda:6'), in_proj_covar=tensor([0.0147, 0.0110, 0.0102, 0.0139, 0.0086, 0.0131, 0.0129, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 12:41:30,330 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=276564.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:41:45,567 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.004e+02 2.344e+02 2.885e+02 4.375e+02, threshold=4.688e+02, percent-clipped=0.0 2023-05-02 12:42:24,131 INFO [train.py:904] (6/8) Epoch 28, batch 2550, loss[loss=0.1481, simple_loss=0.2412, pruned_loss=0.0275, over 15875.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2548, pruned_loss=0.03898, over 3325213.00 frames. ], batch size: 35, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:42:31,800 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=276608.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:42:34,390 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7521, 2.8345, 3.0586, 2.1286, 2.7159, 2.0992, 3.2814, 3.1917], device='cuda:6'), covar=tensor([0.0283, 0.1110, 0.0732, 0.2111, 0.1042, 0.1197, 0.0648, 0.1006], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0172, 0.0171, 0.0158, 0.0149, 0.0134, 0.0148, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 12:42:36,019 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=276611.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:42:45,774 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6453, 2.5427, 1.9990, 2.7006, 2.0842, 2.8245, 2.1716, 2.3954], device='cuda:6'), covar=tensor([0.0334, 0.0389, 0.1273, 0.0279, 0.0719, 0.0480, 0.1192, 0.0653], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0186, 0.0200, 0.0177, 0.0183, 0.0226, 0.0208, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 12:43:33,172 INFO [train.py:904] (6/8) Epoch 28, batch 2600, loss[loss=0.1456, simple_loss=0.2322, pruned_loss=0.02956, over 17003.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2545, pruned_loss=0.03872, over 3322246.60 frames. ], batch size: 41, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:43:38,040 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=276656.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:43:51,588 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5337, 4.4753, 4.4449, 4.1520, 4.2094, 4.4910, 4.2133, 4.2775], device='cuda:6'), covar=tensor([0.0611, 0.0868, 0.0312, 0.0317, 0.0745, 0.0548, 0.0659, 0.0664], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0482, 0.0376, 0.0378, 0.0372, 0.0433, 0.0257, 0.0451], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 12:43:55,792 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=276668.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:44:01,045 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276672.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:44:02,597 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=276673.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:44:05,671 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.267e+02 2.519e+02 3.133e+02 7.416e+02, threshold=5.037e+02, percent-clipped=2.0 2023-05-02 12:44:11,744 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-05-02 12:44:43,681 INFO [train.py:904] (6/8) Epoch 28, batch 2650, loss[loss=0.1464, simple_loss=0.2398, pruned_loss=0.02655, over 17192.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2555, pruned_loss=0.03885, over 3323782.51 frames. ], batch size: 46, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:45:19,841 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276729.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:45:50,738 INFO [train.py:904] (6/8) Epoch 28, batch 2700, loss[loss=0.1903, simple_loss=0.2752, pruned_loss=0.05271, over 16777.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2555, pruned_loss=0.03864, over 3328069.36 frames. ], batch size: 124, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:46:15,500 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.09 vs. limit=2.0 2023-05-02 12:46:23,523 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.276e+02 2.127e+02 2.476e+02 3.244e+02 1.015e+03, threshold=4.951e+02, percent-clipped=5.0 2023-05-02 12:46:42,560 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=276790.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:47:00,576 INFO [train.py:904] (6/8) Epoch 28, batch 2750, loss[loss=0.1706, simple_loss=0.2683, pruned_loss=0.03641, over 17281.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2556, pruned_loss=0.03814, over 3335121.28 frames. ], batch size: 52, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:48:07,239 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276851.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:48:09,124 INFO [train.py:904] (6/8) Epoch 28, batch 2800, loss[loss=0.1618, simple_loss=0.2571, pruned_loss=0.03324, over 16531.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2557, pruned_loss=0.03798, over 3333996.36 frames. ], batch size: 75, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:48:24,961 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=276863.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:48:41,263 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.121e+02 2.382e+02 2.775e+02 4.197e+02, threshold=4.764e+02, percent-clipped=0.0 2023-05-02 12:49:03,542 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9130, 2.8577, 2.6393, 4.6905, 3.3790, 3.9911, 1.9366, 3.0774], device='cuda:6'), covar=tensor([0.1450, 0.0895, 0.1343, 0.0218, 0.0275, 0.0535, 0.1594, 0.0889], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0180, 0.0200, 0.0204, 0.0206, 0.0219, 0.0209, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 12:49:05,868 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8127, 2.5186, 2.0016, 2.2472, 2.8595, 2.6466, 2.8167, 2.9699], device='cuda:6'), covar=tensor([0.0242, 0.0436, 0.0625, 0.0523, 0.0284, 0.0367, 0.0240, 0.0340], device='cuda:6'), in_proj_covar=tensor([0.0239, 0.0252, 0.0240, 0.0240, 0.0254, 0.0251, 0.0251, 0.0250], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 12:49:20,035 INFO [train.py:904] (6/8) Epoch 28, batch 2850, loss[loss=0.1549, simple_loss=0.2597, pruned_loss=0.02508, over 17118.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2546, pruned_loss=0.03803, over 3327278.00 frames. ], batch size: 48, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:49:50,153 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276924.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:49:57,740 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=276931.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:50:11,770 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0988, 4.1288, 4.4206, 4.3860, 4.4461, 4.1983, 4.2189, 4.1232], device='cuda:6'), covar=tensor([0.0388, 0.0700, 0.0436, 0.0461, 0.0507, 0.0514, 0.0754, 0.0609], device='cuda:6'), in_proj_covar=tensor([0.0449, 0.0509, 0.0487, 0.0452, 0.0538, 0.0516, 0.0595, 0.0412], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 12:50:27,302 INFO [train.py:904] (6/8) Epoch 28, batch 2900, loss[loss=0.132, simple_loss=0.2152, pruned_loss=0.02435, over 16957.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2532, pruned_loss=0.03872, over 3327114.78 frames. ], batch size: 41, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:50:46,672 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=276967.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:50:55,053 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=276973.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:50:56,304 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7393, 3.9234, 2.7635, 4.5042, 3.1315, 4.4063, 2.7729, 3.3466], device='cuda:6'), covar=tensor([0.0380, 0.0427, 0.1427, 0.0378, 0.0819, 0.0577, 0.1433, 0.0727], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0185, 0.0198, 0.0177, 0.0182, 0.0225, 0.0207, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 12:50:58,047 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 2.124e+02 2.391e+02 2.943e+02 6.733e+02, threshold=4.783e+02, percent-clipped=5.0 2023-05-02 12:51:21,985 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276992.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 12:51:36,310 INFO [train.py:904] (6/8) Epoch 28, batch 2950, loss[loss=0.2128, simple_loss=0.2893, pruned_loss=0.06814, over 15337.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.253, pruned_loss=0.03893, over 3328513.95 frames. ], batch size: 190, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:51:48,638 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277011.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:52:01,177 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277021.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:52:05,336 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277024.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:52:45,414 INFO [train.py:904] (6/8) Epoch 28, batch 3000, loss[loss=0.1254, simple_loss=0.2155, pruned_loss=0.0176, over 16776.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2528, pruned_loss=0.03877, over 3336551.59 frames. ], batch size: 39, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:52:45,415 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 12:52:54,788 INFO [train.py:938] (6/8) Epoch 28, validation: loss=0.1335, simple_loss=0.2385, pruned_loss=0.01427, over 944034.00 frames. 2023-05-02 12:52:54,789 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 12:53:21,164 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277072.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:53:25,663 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 2.148e+02 2.585e+02 3.132e+02 1.139e+03, threshold=5.170e+02, percent-clipped=2.0 2023-05-02 12:54:02,140 INFO [train.py:904] (6/8) Epoch 28, batch 3050, loss[loss=0.1849, simple_loss=0.2649, pruned_loss=0.05243, over 16245.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2527, pruned_loss=0.0392, over 3331741.83 frames. ], batch size: 165, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:55:01,327 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277146.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:55:11,783 INFO [train.py:904] (6/8) Epoch 28, batch 3100, loss[loss=0.1401, simple_loss=0.2274, pruned_loss=0.02641, over 16780.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2522, pruned_loss=0.03916, over 3331937.18 frames. ], batch size: 39, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:55:43,285 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.340e+02 2.064e+02 2.462e+02 2.812e+02 4.738e+02, threshold=4.925e+02, percent-clipped=0.0 2023-05-02 12:55:44,303 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4646, 4.5114, 4.4020, 4.1076, 4.0402, 4.5090, 4.2733, 4.1315], device='cuda:6'), covar=tensor([0.0808, 0.0901, 0.0428, 0.0428, 0.1103, 0.0634, 0.0695, 0.0887], device='cuda:6'), in_proj_covar=tensor([0.0324, 0.0485, 0.0379, 0.0381, 0.0375, 0.0436, 0.0260, 0.0455], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 12:55:45,486 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277177.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:56:18,141 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.24 vs. limit=5.0 2023-05-02 12:56:21,091 INFO [train.py:904] (6/8) Epoch 28, batch 3150, loss[loss=0.1708, simple_loss=0.2498, pruned_loss=0.04588, over 15416.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2509, pruned_loss=0.03874, over 3318136.80 frames. ], batch size: 191, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:56:44,545 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277219.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:56:44,665 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277219.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:57:10,827 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277238.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:57:11,994 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6001, 3.5789, 2.7996, 2.2587, 2.3027, 2.3323, 3.7202, 3.1288], device='cuda:6'), covar=tensor([0.2963, 0.0675, 0.1920, 0.3124, 0.3020, 0.2346, 0.0523, 0.1681], device='cuda:6'), in_proj_covar=tensor([0.0337, 0.0279, 0.0315, 0.0330, 0.0309, 0.0280, 0.0308, 0.0358], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 12:57:30,728 INFO [train.py:904] (6/8) Epoch 28, batch 3200, loss[loss=0.1795, simple_loss=0.2529, pruned_loss=0.05308, over 16768.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.251, pruned_loss=0.03868, over 3320554.32 frames. ], batch size: 124, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:57:50,714 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277267.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:58:01,817 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.445e+02 1.963e+02 2.375e+02 2.950e+02 8.294e+02, threshold=4.751e+02, percent-clipped=4.0 2023-05-02 12:58:07,831 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277280.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:58:17,849 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277287.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 12:58:29,296 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2618, 2.8909, 3.1733, 1.8775, 3.2605, 3.2745, 2.7749, 2.5378], device='cuda:6'), covar=tensor([0.0851, 0.0290, 0.0261, 0.1157, 0.0143, 0.0257, 0.0460, 0.0518], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0111, 0.0103, 0.0139, 0.0087, 0.0132, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 12:58:39,066 INFO [train.py:904] (6/8) Epoch 28, batch 3250, loss[loss=0.1538, simple_loss=0.2498, pruned_loss=0.02892, over 17169.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2508, pruned_loss=0.03855, over 3319554.71 frames. ], batch size: 46, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:58:42,402 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277305.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:58:55,899 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277315.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:59:09,102 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277324.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:59:21,682 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0852, 2.1501, 2.2789, 3.8012, 2.1500, 2.4941, 2.2512, 2.3492], device='cuda:6'), covar=tensor([0.1713, 0.4097, 0.3325, 0.0744, 0.4157, 0.2761, 0.4290, 0.3313], device='cuda:6'), in_proj_covar=tensor([0.0427, 0.0480, 0.0391, 0.0342, 0.0449, 0.0551, 0.0451, 0.0561], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 12:59:47,703 INFO [train.py:904] (6/8) Epoch 28, batch 3300, loss[loss=0.1625, simple_loss=0.2533, pruned_loss=0.03588, over 16723.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2521, pruned_loss=0.03917, over 3312469.54 frames. ], batch size: 57, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:00:05,312 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277366.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:00:06,343 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277367.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:00:12,568 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277372.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:00:18,088 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.541e+02 2.125e+02 2.555e+02 3.019e+02 5.327e+02, threshold=5.111e+02, percent-clipped=1.0 2023-05-02 13:00:46,672 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5881, 3.6566, 3.3745, 3.0416, 3.2608, 3.5478, 3.3506, 3.3490], device='cuda:6'), covar=tensor([0.0554, 0.0578, 0.0292, 0.0287, 0.0488, 0.0435, 0.1195, 0.0463], device='cuda:6'), in_proj_covar=tensor([0.0326, 0.0489, 0.0382, 0.0383, 0.0378, 0.0439, 0.0260, 0.0456], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 13:00:55,239 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2021, 3.3245, 3.4936, 2.2204, 2.9989, 2.3436, 3.7137, 3.6354], device='cuda:6'), covar=tensor([0.0247, 0.0913, 0.0626, 0.1990, 0.0883, 0.1041, 0.0534, 0.0836], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0174, 0.0171, 0.0158, 0.0150, 0.0134, 0.0148, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 13:00:55,902 INFO [train.py:904] (6/8) Epoch 28, batch 3350, loss[loss=0.1561, simple_loss=0.2427, pruned_loss=0.03477, over 16788.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2528, pruned_loss=0.03902, over 3311898.35 frames. ], batch size: 83, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:01:55,668 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277446.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:02:03,950 INFO [train.py:904] (6/8) Epoch 28, batch 3400, loss[loss=0.1483, simple_loss=0.2458, pruned_loss=0.02536, over 17224.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2528, pruned_loss=0.03893, over 3318444.87 frames. ], batch size: 52, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:02:34,599 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.063e+02 2.423e+02 2.740e+02 5.592e+02, threshold=4.846e+02, percent-clipped=2.0 2023-05-02 13:03:01,407 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277494.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:03:13,659 INFO [train.py:904] (6/8) Epoch 28, batch 3450, loss[loss=0.1578, simple_loss=0.2408, pruned_loss=0.03742, over 16487.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.251, pruned_loss=0.03797, over 3316829.37 frames. ], batch size: 75, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:03:21,155 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9270, 2.7678, 2.6609, 4.7578, 3.8322, 4.2678, 1.7446, 3.1626], device='cuda:6'), covar=tensor([0.1378, 0.0866, 0.1292, 0.0228, 0.0217, 0.0405, 0.1668, 0.0807], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0181, 0.0201, 0.0206, 0.0208, 0.0220, 0.0210, 0.0200], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 13:03:36,018 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277519.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:03:55,468 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277533.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:04:23,507 INFO [train.py:904] (6/8) Epoch 28, batch 3500, loss[loss=0.1586, simple_loss=0.2589, pruned_loss=0.02911, over 17014.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2503, pruned_loss=0.03738, over 3320188.28 frames. ], batch size: 50, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:04:25,789 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277554.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:04:42,187 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277567.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:04:47,693 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277570.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:04:53,427 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0755, 4.0407, 4.0007, 3.3737, 3.9814, 1.7453, 3.8186, 3.4740], device='cuda:6'), covar=tensor([0.0171, 0.0150, 0.0208, 0.0285, 0.0102, 0.3028, 0.0137, 0.0275], device='cuda:6'), in_proj_covar=tensor([0.0186, 0.0181, 0.0219, 0.0192, 0.0195, 0.0225, 0.0209, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:04:54,555 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277575.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:04:55,301 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.480e+02 2.163e+02 2.511e+02 2.754e+02 5.260e+02, threshold=5.022e+02, percent-clipped=2.0 2023-05-02 13:05:10,674 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277587.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 13:05:32,630 INFO [train.py:904] (6/8) Epoch 28, batch 3550, loss[loss=0.1577, simple_loss=0.2522, pruned_loss=0.03156, over 17019.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2505, pruned_loss=0.03723, over 3323290.09 frames. ], batch size: 50, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:05:35,821 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-05-02 13:05:49,376 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277615.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:06:11,680 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277631.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:06:16,894 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277635.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:06:35,295 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7307, 3.9549, 2.6410, 4.6106, 3.0398, 4.5107, 2.7974, 3.3099], device='cuda:6'), covar=tensor([0.0350, 0.0410, 0.1521, 0.0272, 0.0923, 0.0599, 0.1400, 0.0794], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0186, 0.0199, 0.0179, 0.0183, 0.0227, 0.0207, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 13:06:42,325 INFO [train.py:904] (6/8) Epoch 28, batch 3600, loss[loss=0.1633, simple_loss=0.2476, pruned_loss=0.03951, over 16466.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2493, pruned_loss=0.03671, over 3319418.11 frames. ], batch size: 146, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:06:53,439 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277661.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:07:02,012 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277667.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:07:14,464 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.038e+02 2.278e+02 2.834e+02 5.503e+02, threshold=4.555e+02, percent-clipped=2.0 2023-05-02 13:07:37,024 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277691.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:07:53,647 INFO [train.py:904] (6/8) Epoch 28, batch 3650, loss[loss=0.1547, simple_loss=0.2363, pruned_loss=0.03655, over 16550.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.2476, pruned_loss=0.03704, over 3318681.46 frames. ], batch size: 68, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:08:12,451 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277715.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:08:56,205 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1693, 4.3402, 4.5780, 4.5565, 4.6327, 4.3634, 4.2613, 4.2738], device='cuda:6'), covar=tensor([0.0520, 0.0809, 0.0554, 0.0574, 0.0586, 0.0632, 0.1162, 0.0742], device='cuda:6'), in_proj_covar=tensor([0.0449, 0.0507, 0.0488, 0.0450, 0.0536, 0.0514, 0.0593, 0.0413], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 13:09:06,642 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277752.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 13:09:07,229 INFO [train.py:904] (6/8) Epoch 28, batch 3700, loss[loss=0.1878, simple_loss=0.266, pruned_loss=0.05482, over 16477.00 frames. ], tot_loss[loss=0.1621, simple_loss=0.2467, pruned_loss=0.03873, over 3298292.81 frames. ], batch size: 146, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:09:41,445 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 2.185e+02 2.658e+02 3.107e+02 6.146e+02, threshold=5.316e+02, percent-clipped=3.0 2023-05-02 13:10:22,500 INFO [train.py:904] (6/8) Epoch 28, batch 3750, loss[loss=0.1688, simple_loss=0.251, pruned_loss=0.04328, over 16860.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2474, pruned_loss=0.04029, over 3291069.61 frames. ], batch size: 96, lr: 2.40e-03, grad_scale: 4.0 2023-05-02 13:10:54,229 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-05-02 13:11:06,992 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277833.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:11:35,737 INFO [train.py:904] (6/8) Epoch 28, batch 3800, loss[loss=0.173, simple_loss=0.2582, pruned_loss=0.04384, over 17142.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2491, pruned_loss=0.04189, over 3285485.11 frames. ], batch size: 46, lr: 2.40e-03, grad_scale: 4.0 2023-05-02 13:12:08,784 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277875.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:12:11,083 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.049e+02 2.321e+02 2.888e+02 4.993e+02, threshold=4.641e+02, percent-clipped=0.0 2023-05-02 13:12:17,663 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277881.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:12:48,150 INFO [train.py:904] (6/8) Epoch 28, batch 3850, loss[loss=0.167, simple_loss=0.2495, pruned_loss=0.04219, over 16441.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2494, pruned_loss=0.04254, over 3277265.01 frames. ], batch size: 75, lr: 2.40e-03, grad_scale: 4.0 2023-05-02 13:12:57,560 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9565, 5.3047, 5.4893, 5.2441, 5.3051, 5.8710, 5.3571, 5.0533], device='cuda:6'), covar=tensor([0.1182, 0.1847, 0.2175, 0.2179, 0.2653, 0.1057, 0.1604, 0.2436], device='cuda:6'), in_proj_covar=tensor([0.0437, 0.0655, 0.0721, 0.0534, 0.0709, 0.0743, 0.0560, 0.0708], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 13:12:59,659 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277910.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:12:59,985 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2023-05-02 13:13:12,438 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2875, 2.7486, 2.3185, 2.4516, 3.0524, 2.6951, 3.0604, 3.1979], device='cuda:6'), covar=tensor([0.0211, 0.0427, 0.0594, 0.0496, 0.0286, 0.0412, 0.0285, 0.0286], device='cuda:6'), in_proj_covar=tensor([0.0240, 0.0253, 0.0240, 0.0240, 0.0253, 0.0251, 0.0252, 0.0251], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:13:18,294 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277923.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:13:23,019 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277926.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:14:00,814 INFO [train.py:904] (6/8) Epoch 28, batch 3900, loss[loss=0.1674, simple_loss=0.2416, pruned_loss=0.04659, over 16837.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2485, pruned_loss=0.04275, over 3284722.93 frames. ], batch size: 83, lr: 2.40e-03, grad_scale: 4.0 2023-05-02 13:14:08,084 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2593, 2.3273, 2.3622, 4.0041, 2.3266, 2.6823, 2.3576, 2.4856], device='cuda:6'), covar=tensor([0.1593, 0.3938, 0.3262, 0.0643, 0.4031, 0.2633, 0.4519, 0.3036], device='cuda:6'), in_proj_covar=tensor([0.0426, 0.0480, 0.0390, 0.0342, 0.0448, 0.0550, 0.0451, 0.0561], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:14:13,109 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277961.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:14:27,644 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277971.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:14:36,140 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.618e+02 2.147e+02 2.438e+02 2.765e+02 5.958e+02, threshold=4.876e+02, percent-clipped=2.0 2023-05-02 13:15:16,801 INFO [train.py:904] (6/8) Epoch 28, batch 3950, loss[loss=0.165, simple_loss=0.2466, pruned_loss=0.04165, over 16436.00 frames. ], tot_loss[loss=0.167, simple_loss=0.248, pruned_loss=0.04298, over 3280101.51 frames. ], batch size: 146, lr: 2.40e-03, grad_scale: 4.0 2023-05-02 13:15:25,690 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=278009.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:15:58,214 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=278032.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:16:15,502 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-02 13:16:19,856 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=278047.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 13:16:26,866 INFO [train.py:904] (6/8) Epoch 28, batch 4000, loss[loss=0.1573, simple_loss=0.2445, pruned_loss=0.03504, over 16813.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2489, pruned_loss=0.04355, over 3281587.99 frames. ], batch size: 96, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:17:01,537 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.521e+02 1.968e+02 2.308e+02 2.954e+02 7.046e+02, threshold=4.616e+02, percent-clipped=2.0 2023-05-02 13:17:38,520 INFO [train.py:904] (6/8) Epoch 28, batch 4050, loss[loss=0.1721, simple_loss=0.2561, pruned_loss=0.04404, over 12490.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2489, pruned_loss=0.04262, over 3280104.29 frames. ], batch size: 248, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:18:20,326 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6993, 1.8873, 2.3126, 2.6014, 2.6265, 3.0194, 2.0217, 2.9287], device='cuda:6'), covar=tensor([0.0280, 0.0603, 0.0396, 0.0416, 0.0407, 0.0221, 0.0678, 0.0154], device='cuda:6'), in_proj_covar=tensor([0.0201, 0.0199, 0.0190, 0.0195, 0.0211, 0.0168, 0.0205, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 13:18:52,735 INFO [train.py:904] (6/8) Epoch 28, batch 4100, loss[loss=0.1732, simple_loss=0.2688, pruned_loss=0.03884, over 16891.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2511, pruned_loss=0.0424, over 3286940.91 frames. ], batch size: 116, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:19:25,120 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-05-02 13:19:29,062 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 1.827e+02 2.078e+02 2.388e+02 3.753e+02, threshold=4.156e+02, percent-clipped=0.0 2023-05-02 13:20:09,078 INFO [train.py:904] (6/8) Epoch 28, batch 4150, loss[loss=0.1744, simple_loss=0.2727, pruned_loss=0.03805, over 16738.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2579, pruned_loss=0.04452, over 3240479.38 frames. ], batch size: 89, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:20:19,280 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=278210.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:20:42,448 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=278226.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:20:45,644 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6875, 2.3393, 1.9070, 2.0918, 2.6380, 2.2796, 2.4395, 2.7650], device='cuda:6'), covar=tensor([0.0204, 0.0412, 0.0616, 0.0522, 0.0299, 0.0423, 0.0267, 0.0291], device='cuda:6'), in_proj_covar=tensor([0.0237, 0.0250, 0.0236, 0.0238, 0.0250, 0.0248, 0.0249, 0.0248], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:20:58,876 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3713, 2.5478, 2.4038, 4.2182, 2.4019, 2.8876, 2.5025, 2.6403], device='cuda:6'), covar=tensor([0.1391, 0.3400, 0.2939, 0.0504, 0.3688, 0.2233, 0.3413, 0.3179], device='cuda:6'), in_proj_covar=tensor([0.0424, 0.0477, 0.0387, 0.0339, 0.0445, 0.0548, 0.0448, 0.0558], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:21:22,142 INFO [train.py:904] (6/8) Epoch 28, batch 4200, loss[loss=0.2175, simple_loss=0.2979, pruned_loss=0.06854, over 11276.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2653, pruned_loss=0.04629, over 3216939.31 frames. ], batch size: 246, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:21:30,722 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=278258.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:21:54,355 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=278274.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:21:58,171 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.712e+02 2.281e+02 2.720e+02 3.199e+02 7.364e+02, threshold=5.441e+02, percent-clipped=7.0 2023-05-02 13:22:01,773 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0296, 4.1502, 3.9067, 3.6407, 3.6761, 4.0730, 3.7388, 3.8115], device='cuda:6'), covar=tensor([0.0611, 0.0538, 0.0321, 0.0294, 0.0779, 0.0386, 0.0918, 0.0560], device='cuda:6'), in_proj_covar=tensor([0.0322, 0.0484, 0.0377, 0.0379, 0.0374, 0.0435, 0.0257, 0.0451], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 13:22:37,392 INFO [train.py:904] (6/8) Epoch 28, batch 4250, loss[loss=0.1676, simple_loss=0.2687, pruned_loss=0.0333, over 16227.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2686, pruned_loss=0.04581, over 3214307.94 frames. ], batch size: 165, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:22:52,287 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-05-02 13:22:55,525 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4391, 2.9955, 2.6953, 2.2934, 2.2176, 2.3110, 3.0162, 2.8105], device='cuda:6'), covar=tensor([0.2856, 0.0912, 0.1869, 0.3092, 0.2982, 0.2426, 0.0642, 0.1596], device='cuda:6'), in_proj_covar=tensor([0.0335, 0.0277, 0.0313, 0.0328, 0.0308, 0.0278, 0.0306, 0.0355], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 13:23:07,790 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=278323.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:23:13,432 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=278327.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:23:44,034 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=278347.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 13:23:52,555 INFO [train.py:904] (6/8) Epoch 28, batch 4300, loss[loss=0.1868, simple_loss=0.2858, pruned_loss=0.04384, over 16704.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2694, pruned_loss=0.04496, over 3208549.09 frames. ], batch size: 89, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:24:25,561 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=278374.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:24:29,317 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.019e+02 2.401e+02 2.918e+02 4.618e+02, threshold=4.803e+02, percent-clipped=0.0 2023-05-02 13:24:40,758 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=278384.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:24:56,296 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=278395.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:25:07,418 INFO [train.py:904] (6/8) Epoch 28, batch 4350, loss[loss=0.2035, simple_loss=0.2963, pruned_loss=0.05534, over 16469.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.2729, pruned_loss=0.04608, over 3201751.30 frames. ], batch size: 75, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:25:47,118 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8828, 2.2079, 2.5416, 2.9806, 2.8889, 3.3445, 2.1739, 3.3413], device='cuda:6'), covar=tensor([0.0284, 0.0512, 0.0346, 0.0308, 0.0317, 0.0198, 0.0592, 0.0145], device='cuda:6'), in_proj_covar=tensor([0.0201, 0.0200, 0.0189, 0.0194, 0.0210, 0.0168, 0.0204, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:25:57,099 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5226, 3.9875, 3.8558, 2.4949, 3.6166, 3.9868, 3.6324, 1.9234], device='cuda:6'), covar=tensor([0.0563, 0.0047, 0.0088, 0.0514, 0.0119, 0.0122, 0.0120, 0.0639], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0090, 0.0091, 0.0135, 0.0102, 0.0115, 0.0099, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 13:25:57,134 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=278435.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:26:22,393 INFO [train.py:904] (6/8) Epoch 28, batch 4400, loss[loss=0.1859, simple_loss=0.2792, pruned_loss=0.04628, over 15566.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2747, pruned_loss=0.04726, over 3200500.70 frames. ], batch size: 190, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:26:31,197 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-05-02 13:26:34,054 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0039, 4.8120, 5.0441, 5.1819, 5.3426, 4.7360, 5.3552, 5.3970], device='cuda:6'), covar=tensor([0.1674, 0.1199, 0.1442, 0.0673, 0.0447, 0.0907, 0.0513, 0.0515], device='cuda:6'), in_proj_covar=tensor([0.0686, 0.0839, 0.0971, 0.0854, 0.0649, 0.0679, 0.0711, 0.0823], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:26:58,177 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.247e+02 2.621e+02 3.049e+02 4.749e+02, threshold=5.243e+02, percent-clipped=0.0 2023-05-02 13:27:35,966 INFO [train.py:904] (6/8) Epoch 28, batch 4450, loss[loss=0.2153, simple_loss=0.3071, pruned_loss=0.06178, over 16880.00 frames. ], tot_loss[loss=0.1874, simple_loss=0.2779, pruned_loss=0.04848, over 3201932.27 frames. ], batch size: 116, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:27:48,027 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=278511.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:28:49,071 INFO [train.py:904] (6/8) Epoch 28, batch 4500, loss[loss=0.1751, simple_loss=0.2671, pruned_loss=0.04159, over 16476.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.2781, pruned_loss=0.04923, over 3210398.34 frames. ], batch size: 68, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:28:58,983 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4920, 1.7356, 2.1228, 2.4070, 2.4391, 2.7099, 1.8625, 2.6419], device='cuda:6'), covar=tensor([0.0258, 0.0615, 0.0353, 0.0377, 0.0363, 0.0251, 0.0652, 0.0168], device='cuda:6'), in_proj_covar=tensor([0.0200, 0.0199, 0.0188, 0.0194, 0.0210, 0.0168, 0.0203, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:29:17,601 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=278572.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:29:23,508 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 1.834e+02 2.148e+02 2.381e+02 3.720e+02, threshold=4.296e+02, percent-clipped=0.0 2023-05-02 13:30:01,296 INFO [train.py:904] (6/8) Epoch 28, batch 4550, loss[loss=0.2138, simple_loss=0.3035, pruned_loss=0.06206, over 16428.00 frames. ], tot_loss[loss=0.1899, simple_loss=0.2791, pruned_loss=0.05033, over 3232062.29 frames. ], batch size: 146, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:30:36,983 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=278627.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:31:12,169 INFO [train.py:904] (6/8) Epoch 28, batch 4600, loss[loss=0.187, simple_loss=0.2782, pruned_loss=0.0479, over 16238.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.2797, pruned_loss=0.05062, over 3223786.10 frames. ], batch size: 165, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:31:43,899 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=278675.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:31:46,040 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.292e+02 1.855e+02 2.019e+02 2.344e+02 4.106e+02, threshold=4.037e+02, percent-clipped=0.0 2023-05-02 13:31:49,266 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=278679.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:31:51,615 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-02 13:32:22,022 INFO [train.py:904] (6/8) Epoch 28, batch 4650, loss[loss=0.1826, simple_loss=0.2634, pruned_loss=0.05083, over 17029.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2792, pruned_loss=0.05116, over 3217164.01 frames. ], batch size: 41, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:32:24,853 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2023-05-02 13:33:00,843 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=278730.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:33:06,289 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-05-02 13:33:33,716 INFO [train.py:904] (6/8) Epoch 28, batch 4700, loss[loss=0.1758, simple_loss=0.2644, pruned_loss=0.04361, over 16855.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2771, pruned_loss=0.0504, over 3206554.73 frames. ], batch size: 116, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:34:07,033 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 1.813e+02 1.994e+02 2.322e+02 3.636e+02, threshold=3.989e+02, percent-clipped=0.0 2023-05-02 13:34:45,678 INFO [train.py:904] (6/8) Epoch 28, batch 4750, loss[loss=0.1656, simple_loss=0.2579, pruned_loss=0.03664, over 16548.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2732, pruned_loss=0.04856, over 3202330.56 frames. ], batch size: 75, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:35:37,962 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2495, 5.2428, 5.0632, 4.2645, 5.1659, 1.7803, 4.8637, 4.7277], device='cuda:6'), covar=tensor([0.0100, 0.0121, 0.0186, 0.0552, 0.0112, 0.3057, 0.0150, 0.0272], device='cuda:6'), in_proj_covar=tensor([0.0182, 0.0176, 0.0214, 0.0187, 0.0190, 0.0219, 0.0203, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:35:57,551 INFO [train.py:904] (6/8) Epoch 28, batch 4800, loss[loss=0.1969, simple_loss=0.2906, pruned_loss=0.0516, over 15486.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2695, pruned_loss=0.04633, over 3202177.18 frames. ], batch size: 190, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:36:17,224 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1578, 5.4570, 5.0548, 5.4068, 5.0231, 4.7803, 5.0242, 5.5487], device='cuda:6'), covar=tensor([0.2110, 0.1371, 0.1992, 0.1180, 0.1356, 0.1357, 0.2007, 0.1616], device='cuda:6'), in_proj_covar=tensor([0.0720, 0.0880, 0.0716, 0.0676, 0.0554, 0.0550, 0.0734, 0.0684], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:36:18,859 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=278867.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:36:32,583 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.390e+02 1.887e+02 2.183e+02 2.468e+02 4.688e+02, threshold=4.367e+02, percent-clipped=1.0 2023-05-02 13:37:13,015 INFO [train.py:904] (6/8) Epoch 28, batch 4850, loss[loss=0.157, simple_loss=0.2537, pruned_loss=0.03014, over 16728.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.27, pruned_loss=0.04526, over 3196106.24 frames. ], batch size: 89, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:38:26,573 INFO [train.py:904] (6/8) Epoch 28, batch 4900, loss[loss=0.2007, simple_loss=0.2915, pruned_loss=0.05493, over 15379.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2696, pruned_loss=0.04417, over 3179396.45 frames. ], batch size: 190, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:39:00,724 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 1.895e+02 2.133e+02 2.360e+02 3.975e+02, threshold=4.266e+02, percent-clipped=0.0 2023-05-02 13:39:04,535 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=278979.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:39:21,360 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0962, 5.3518, 5.1539, 5.1828, 4.8992, 4.8666, 4.7003, 5.4585], device='cuda:6'), covar=tensor([0.1344, 0.0888, 0.0935, 0.0866, 0.0806, 0.0897, 0.1238, 0.0797], device='cuda:6'), in_proj_covar=tensor([0.0717, 0.0877, 0.0713, 0.0672, 0.0552, 0.0548, 0.0730, 0.0680], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:39:39,802 INFO [train.py:904] (6/8) Epoch 28, batch 4950, loss[loss=0.1642, simple_loss=0.2613, pruned_loss=0.03354, over 16559.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2694, pruned_loss=0.0437, over 3179528.63 frames. ], batch size: 75, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:40:14,413 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=279027.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:40:19,480 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=279030.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:40:22,141 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-02 13:40:51,497 INFO [train.py:904] (6/8) Epoch 28, batch 5000, loss[loss=0.2021, simple_loss=0.2976, pruned_loss=0.05326, over 16713.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2713, pruned_loss=0.04408, over 3189134.82 frames. ], batch size: 124, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:41:26,569 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.098e+02 2.415e+02 2.843e+02 4.357e+02, threshold=4.831e+02, percent-clipped=1.0 2023-05-02 13:41:28,683 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=279078.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:41:57,118 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4091, 4.1268, 4.1583, 2.5631, 3.5704, 4.1315, 3.5556, 2.2197], device='cuda:6'), covar=tensor([0.0553, 0.0046, 0.0042, 0.0435, 0.0118, 0.0102, 0.0126, 0.0493], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0090, 0.0091, 0.0135, 0.0103, 0.0115, 0.0099, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 13:42:04,191 INFO [train.py:904] (6/8) Epoch 28, batch 5050, loss[loss=0.1648, simple_loss=0.267, pruned_loss=0.0313, over 16890.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2715, pruned_loss=0.04355, over 3199380.12 frames. ], batch size: 96, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:42:37,494 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-02 13:42:45,764 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=279131.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:43:17,182 INFO [train.py:904] (6/8) Epoch 28, batch 5100, loss[loss=0.1803, simple_loss=0.2656, pruned_loss=0.04748, over 16600.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2698, pruned_loss=0.04289, over 3203026.32 frames. ], batch size: 68, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:43:37,532 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=279167.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:43:52,643 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 1.882e+02 2.136e+02 2.468e+02 5.149e+02, threshold=4.272e+02, percent-clipped=1.0 2023-05-02 13:44:09,786 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0130, 5.0623, 5.3597, 5.3586, 5.3673, 5.0795, 5.0050, 4.8232], device='cuda:6'), covar=tensor([0.0251, 0.0504, 0.0361, 0.0337, 0.0351, 0.0296, 0.0816, 0.0400], device='cuda:6'), in_proj_covar=tensor([0.0432, 0.0487, 0.0471, 0.0434, 0.0515, 0.0495, 0.0571, 0.0399], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 13:44:15,321 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=279192.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:44:30,679 INFO [train.py:904] (6/8) Epoch 28, batch 5150, loss[loss=0.1848, simple_loss=0.2877, pruned_loss=0.04091, over 15420.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2691, pruned_loss=0.04194, over 3209637.30 frames. ], batch size: 190, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:44:49,080 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=279215.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:45:42,235 INFO [train.py:904] (6/8) Epoch 28, batch 5200, loss[loss=0.1827, simple_loss=0.2626, pruned_loss=0.05141, over 17025.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2677, pruned_loss=0.04155, over 3206283.22 frames. ], batch size: 55, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:46:17,349 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 1.964e+02 2.228e+02 2.622e+02 8.480e+02, threshold=4.456e+02, percent-clipped=3.0 2023-05-02 13:46:53,552 INFO [train.py:904] (6/8) Epoch 28, batch 5250, loss[loss=0.1607, simple_loss=0.259, pruned_loss=0.03115, over 16861.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2655, pruned_loss=0.04112, over 3198929.50 frames. ], batch size: 102, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:47:03,590 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0120, 3.6234, 3.4811, 2.0592, 3.2310, 3.6393, 3.2630, 1.6607], device='cuda:6'), covar=tensor([0.0769, 0.0098, 0.0128, 0.0624, 0.0179, 0.0242, 0.0287, 0.0799], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0089, 0.0091, 0.0135, 0.0103, 0.0115, 0.0099, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 13:47:16,690 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-02 13:47:57,888 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5705, 4.6353, 4.4543, 4.1077, 4.1307, 4.5608, 4.3274, 4.2737], device='cuda:6'), covar=tensor([0.0645, 0.0509, 0.0326, 0.0359, 0.1007, 0.0531, 0.0546, 0.0652], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0470, 0.0365, 0.0368, 0.0364, 0.0422, 0.0249, 0.0435], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:48:06,860 INFO [train.py:904] (6/8) Epoch 28, batch 5300, loss[loss=0.1455, simple_loss=0.2406, pruned_loss=0.02526, over 16843.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2625, pruned_loss=0.04026, over 3200681.13 frames. ], batch size: 96, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:48:41,219 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.441e+02 1.963e+02 2.333e+02 2.759e+02 6.747e+02, threshold=4.665e+02, percent-clipped=2.0 2023-05-02 13:49:20,977 INFO [train.py:904] (6/8) Epoch 28, batch 5350, loss[loss=0.171, simple_loss=0.2698, pruned_loss=0.03609, over 16676.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2608, pruned_loss=0.03983, over 3193162.74 frames. ], batch size: 134, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:49:42,437 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8630, 2.7787, 2.6355, 1.8666, 2.5490, 2.7433, 2.5642, 1.9076], device='cuda:6'), covar=tensor([0.0511, 0.0086, 0.0086, 0.0414, 0.0150, 0.0129, 0.0145, 0.0453], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0089, 0.0090, 0.0134, 0.0102, 0.0114, 0.0098, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 13:50:32,504 INFO [train.py:904] (6/8) Epoch 28, batch 5400, loss[loss=0.1769, simple_loss=0.2715, pruned_loss=0.04113, over 16661.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.264, pruned_loss=0.04049, over 3186442.45 frames. ], batch size: 62, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:51:08,321 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.450e+02 1.928e+02 2.246e+02 2.506e+02 7.644e+02, threshold=4.493e+02, percent-clipped=1.0 2023-05-02 13:51:24,275 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=279487.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:51:50,347 INFO [train.py:904] (6/8) Epoch 28, batch 5450, loss[loss=0.2345, simple_loss=0.3215, pruned_loss=0.07376, over 15390.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2671, pruned_loss=0.04231, over 3166755.77 frames. ], batch size: 190, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:52:31,319 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=279529.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:52:52,368 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=279542.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:53:09,374 INFO [train.py:904] (6/8) Epoch 28, batch 5500, loss[loss=0.2561, simple_loss=0.3294, pruned_loss=0.09135, over 11686.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2734, pruned_loss=0.04607, over 3146694.32 frames. ], batch size: 246, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:53:18,706 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-02 13:53:47,382 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.670e+02 2.730e+02 3.219e+02 3.961e+02 9.894e+02, threshold=6.439e+02, percent-clipped=12.0 2023-05-02 13:54:09,420 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=279590.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:54:13,911 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7789, 3.9505, 2.7067, 2.4463, 2.7384, 2.5410, 4.0868, 3.4069], device='cuda:6'), covar=tensor([0.3096, 0.0802, 0.2381, 0.3158, 0.2943, 0.2429, 0.0704, 0.1592], device='cuda:6'), in_proj_covar=tensor([0.0335, 0.0275, 0.0312, 0.0327, 0.0305, 0.0276, 0.0305, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 13:54:28,765 INFO [train.py:904] (6/8) Epoch 28, batch 5550, loss[loss=0.2545, simple_loss=0.3194, pruned_loss=0.09481, over 11113.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2794, pruned_loss=0.05032, over 3122570.73 frames. ], batch size: 248, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:54:29,481 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=279603.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:55:48,650 INFO [train.py:904] (6/8) Epoch 28, batch 5600, loss[loss=0.1798, simple_loss=0.2763, pruned_loss=0.04165, over 17119.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2835, pruned_loss=0.05369, over 3107959.98 frames. ], batch size: 48, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:55:50,555 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0340, 4.0967, 3.9225, 3.6714, 3.6758, 4.0370, 3.6863, 3.8667], device='cuda:6'), covar=tensor([0.0607, 0.0694, 0.0294, 0.0309, 0.0719, 0.0552, 0.1220, 0.0561], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0474, 0.0367, 0.0370, 0.0366, 0.0424, 0.0250, 0.0438], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:56:28,890 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.204e+02 3.162e+02 3.859e+02 4.792e+02 1.536e+03, threshold=7.717e+02, percent-clipped=11.0 2023-05-02 13:57:11,930 INFO [train.py:904] (6/8) Epoch 28, batch 5650, loss[loss=0.2845, simple_loss=0.3385, pruned_loss=0.1152, over 10575.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2884, pruned_loss=0.05736, over 3082638.12 frames. ], batch size: 248, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:58:29,312 INFO [train.py:904] (6/8) Epoch 28, batch 5700, loss[loss=0.1934, simple_loss=0.2897, pruned_loss=0.04849, over 16854.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2901, pruned_loss=0.05911, over 3063844.22 frames. ], batch size: 102, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:58:29,840 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0458, 4.0424, 3.9694, 3.1729, 3.9898, 1.7722, 3.7958, 3.4350], device='cuda:6'), covar=tensor([0.0135, 0.0126, 0.0195, 0.0335, 0.0103, 0.3102, 0.0139, 0.0337], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0176, 0.0213, 0.0187, 0.0190, 0.0218, 0.0202, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:59:05,447 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.265e+02 3.251e+02 3.778e+02 4.612e+02 7.364e+02, threshold=7.556e+02, percent-clipped=0.0 2023-05-02 13:59:10,330 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3957, 2.4829, 2.0694, 2.3123, 2.8828, 2.4618, 2.8935, 3.0556], device='cuda:6'), covar=tensor([0.0153, 0.0497, 0.0609, 0.0514, 0.0322, 0.0477, 0.0273, 0.0329], device='cuda:6'), in_proj_covar=tensor([0.0231, 0.0245, 0.0233, 0.0233, 0.0245, 0.0243, 0.0242, 0.0243], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 13:59:22,599 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=279787.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:59:27,077 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.39 vs. limit=5.0 2023-05-02 13:59:46,949 INFO [train.py:904] (6/8) Epoch 28, batch 5750, loss[loss=0.2267, simple_loss=0.2978, pruned_loss=0.07779, over 11252.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2935, pruned_loss=0.06129, over 3029051.13 frames. ], batch size: 246, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 14:00:01,111 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8418, 2.2835, 1.8945, 2.1160, 2.7191, 2.3177, 2.4356, 2.8424], device='cuda:6'), covar=tensor([0.0268, 0.0555, 0.0746, 0.0676, 0.0334, 0.0499, 0.0260, 0.0329], device='cuda:6'), in_proj_covar=tensor([0.0231, 0.0245, 0.0233, 0.0233, 0.0245, 0.0243, 0.0241, 0.0243], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:00:39,834 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=279835.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:00:49,444 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.53 vs. limit=5.0 2023-05-02 14:01:07,318 INFO [train.py:904] (6/8) Epoch 28, batch 5800, loss[loss=0.2107, simple_loss=0.2827, pruned_loss=0.06941, over 12047.00 frames. ], tot_loss[loss=0.2075, simple_loss=0.2935, pruned_loss=0.06076, over 3022225.31 frames. ], batch size: 248, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 14:01:08,293 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-05-02 14:01:27,526 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6091, 2.6072, 1.9222, 2.6780, 2.1376, 2.7928, 2.1731, 2.3858], device='cuda:6'), covar=tensor([0.0340, 0.0360, 0.1243, 0.0315, 0.0647, 0.0464, 0.1155, 0.0590], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0181, 0.0196, 0.0171, 0.0180, 0.0219, 0.0203, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 14:01:46,119 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.097e+02 2.839e+02 3.367e+02 4.183e+02 8.215e+02, threshold=6.734e+02, percent-clipped=2.0 2023-05-02 14:01:57,428 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=279885.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:02:16,986 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=279898.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:02:25,207 INFO [train.py:904] (6/8) Epoch 28, batch 5850, loss[loss=0.2021, simple_loss=0.2746, pruned_loss=0.0648, over 11352.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2911, pruned_loss=0.05881, over 3035481.27 frames. ], batch size: 247, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 14:03:27,953 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6303, 2.6055, 1.9343, 2.7207, 2.1930, 2.8140, 2.2209, 2.4248], device='cuda:6'), covar=tensor([0.0306, 0.0359, 0.1282, 0.0249, 0.0685, 0.0442, 0.1142, 0.0609], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0181, 0.0196, 0.0172, 0.0180, 0.0219, 0.0203, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 14:03:44,211 INFO [train.py:904] (6/8) Epoch 28, batch 5900, loss[loss=0.2078, simple_loss=0.2913, pruned_loss=0.06214, over 15542.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2908, pruned_loss=0.0584, over 3048449.27 frames. ], batch size: 191, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 14:03:57,482 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0419, 5.4369, 5.6137, 5.3028, 5.2291, 5.9557, 5.4142, 5.1505], device='cuda:6'), covar=tensor([0.0966, 0.1883, 0.2549, 0.2017, 0.2783, 0.0836, 0.1608, 0.2518], device='cuda:6'), in_proj_covar=tensor([0.0424, 0.0634, 0.0694, 0.0514, 0.0687, 0.0719, 0.0540, 0.0687], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 14:04:26,171 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.117e+02 2.561e+02 3.002e+02 3.588e+02 6.021e+02, threshold=6.005e+02, percent-clipped=0.0 2023-05-02 14:05:07,641 INFO [train.py:904] (6/8) Epoch 28, batch 5950, loss[loss=0.1817, simple_loss=0.2764, pruned_loss=0.04351, over 16832.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2913, pruned_loss=0.05687, over 3057829.58 frames. ], batch size: 102, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 14:05:48,112 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6841, 4.5570, 4.7446, 4.9002, 5.0617, 4.5216, 5.0709, 5.1019], device='cuda:6'), covar=tensor([0.2127, 0.1269, 0.1675, 0.0782, 0.0605, 0.1106, 0.0652, 0.0647], device='cuda:6'), in_proj_covar=tensor([0.0674, 0.0823, 0.0954, 0.0839, 0.0641, 0.0671, 0.0701, 0.0810], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:06:04,020 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 14:06:20,883 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280051.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:06:23,829 INFO [train.py:904] (6/8) Epoch 28, batch 6000, loss[loss=0.1992, simple_loss=0.2847, pruned_loss=0.05686, over 16391.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2904, pruned_loss=0.05684, over 3043417.48 frames. ], batch size: 146, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 14:06:23,830 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 14:06:34,276 INFO [train.py:938] (6/8) Epoch 28, validation: loss=0.148, simple_loss=0.2602, pruned_loss=0.0179, over 944034.00 frames. 2023-05-02 14:06:34,277 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 14:07:11,168 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.966e+02 2.746e+02 3.426e+02 4.098e+02 7.990e+02, threshold=6.852e+02, percent-clipped=5.0 2023-05-02 14:07:18,429 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.7213, 6.0954, 5.7967, 5.8649, 5.4760, 5.3567, 5.4570, 6.1946], device='cuda:6'), covar=tensor([0.1326, 0.0884, 0.0981, 0.0875, 0.0821, 0.0702, 0.1217, 0.0798], device='cuda:6'), in_proj_covar=tensor([0.0719, 0.0876, 0.0716, 0.0674, 0.0551, 0.0548, 0.0727, 0.0679], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:07:18,777 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2023-05-02 14:07:25,929 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280088.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:07:37,716 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 14:07:51,265 INFO [train.py:904] (6/8) Epoch 28, batch 6050, loss[loss=0.2001, simple_loss=0.2879, pruned_loss=0.05614, over 15363.00 frames. ], tot_loss[loss=0.2013, simple_loss=0.2891, pruned_loss=0.05673, over 3054212.65 frames. ], batch size: 191, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:07:52,339 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9351, 2.3426, 1.9075, 2.1203, 2.7656, 2.4299, 2.5856, 2.9166], device='cuda:6'), covar=tensor([0.0228, 0.0580, 0.0735, 0.0614, 0.0344, 0.0466, 0.0244, 0.0337], device='cuda:6'), in_proj_covar=tensor([0.0228, 0.0242, 0.0231, 0.0231, 0.0243, 0.0240, 0.0239, 0.0241], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:08:06,562 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280112.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:08:20,492 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9754, 2.1511, 2.2246, 3.4726, 2.1078, 2.4610, 2.2591, 2.2811], device='cuda:6'), covar=tensor([0.1649, 0.3741, 0.3206, 0.0721, 0.4307, 0.2632, 0.3763, 0.3454], device='cuda:6'), in_proj_covar=tensor([0.0423, 0.0477, 0.0386, 0.0338, 0.0447, 0.0544, 0.0447, 0.0556], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:08:25,834 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2023-05-02 14:09:02,468 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280149.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:09:08,221 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3605, 4.4098, 4.7358, 4.6862, 4.7137, 4.4246, 4.4220, 4.3583], device='cuda:6'), covar=tensor([0.0350, 0.0658, 0.0391, 0.0439, 0.0482, 0.0417, 0.0951, 0.0516], device='cuda:6'), in_proj_covar=tensor([0.0430, 0.0485, 0.0470, 0.0433, 0.0515, 0.0495, 0.0570, 0.0398], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 14:09:08,867 INFO [train.py:904] (6/8) Epoch 28, batch 6100, loss[loss=0.1784, simple_loss=0.2668, pruned_loss=0.04496, over 16901.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2886, pruned_loss=0.05553, over 3082127.14 frames. ], batch size: 109, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:09:51,400 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.727e+02 2.432e+02 2.900e+02 3.731e+02 6.785e+02, threshold=5.800e+02, percent-clipped=0.0 2023-05-02 14:10:01,928 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=280185.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:10:22,514 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=280198.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:10:29,424 INFO [train.py:904] (6/8) Epoch 28, batch 6150, loss[loss=0.2033, simple_loss=0.2871, pruned_loss=0.05978, over 11712.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2865, pruned_loss=0.05477, over 3077035.48 frames. ], batch size: 246, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:11:17,168 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=280233.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:11:22,992 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-05-02 14:11:36,301 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=280246.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:11:45,495 INFO [train.py:904] (6/8) Epoch 28, batch 6200, loss[loss=0.2018, simple_loss=0.277, pruned_loss=0.06331, over 11702.00 frames. ], tot_loss[loss=0.1963, simple_loss=0.2843, pruned_loss=0.05419, over 3085662.48 frames. ], batch size: 248, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:12:24,022 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.005e+02 2.646e+02 3.232e+02 3.925e+02 8.814e+02, threshold=6.464e+02, percent-clipped=6.0 2023-05-02 14:12:50,636 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3384, 4.4079, 4.2227, 3.9438, 3.9426, 4.3379, 4.0512, 4.0857], device='cuda:6'), covar=tensor([0.0715, 0.0894, 0.0350, 0.0384, 0.0910, 0.0594, 0.0809, 0.0761], device='cuda:6'), in_proj_covar=tensor([0.0314, 0.0473, 0.0366, 0.0369, 0.0364, 0.0423, 0.0250, 0.0437], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:12:53,784 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.94 vs. limit=5.0 2023-05-02 14:13:00,715 INFO [train.py:904] (6/8) Epoch 28, batch 6250, loss[loss=0.1803, simple_loss=0.2736, pruned_loss=0.04349, over 16643.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2839, pruned_loss=0.05367, over 3110220.61 frames. ], batch size: 57, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:13:36,610 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280327.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:14:14,430 INFO [train.py:904] (6/8) Epoch 28, batch 6300, loss[loss=0.2015, simple_loss=0.2822, pruned_loss=0.06038, over 11966.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.2837, pruned_loss=0.05306, over 3110733.95 frames. ], batch size: 247, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:14:20,415 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-05-02 14:14:53,563 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.897e+02 2.662e+02 3.040e+02 3.867e+02 7.586e+02, threshold=6.080e+02, percent-clipped=1.0 2023-05-02 14:15:09,922 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280388.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:15:31,097 INFO [train.py:904] (6/8) Epoch 28, batch 6350, loss[loss=0.2612, simple_loss=0.3197, pruned_loss=0.1013, over 11413.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.285, pruned_loss=0.05465, over 3098176.39 frames. ], batch size: 248, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:15:36,781 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280407.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:16:31,361 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280444.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:16:43,564 INFO [train.py:904] (6/8) Epoch 28, batch 6400, loss[loss=0.1908, simple_loss=0.2757, pruned_loss=0.0529, over 16873.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2853, pruned_loss=0.05584, over 3092285.82 frames. ], batch size: 42, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:16:54,062 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1578, 3.4598, 3.5438, 2.1388, 3.0134, 2.3498, 3.5946, 3.8148], device='cuda:6'), covar=tensor([0.0284, 0.0795, 0.0642, 0.2191, 0.0904, 0.1037, 0.0617, 0.0857], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0172, 0.0172, 0.0158, 0.0149, 0.0133, 0.0147, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 14:17:19,316 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.815e+02 2.958e+02 3.396e+02 3.952e+02 7.468e+02, threshold=6.793e+02, percent-clipped=6.0 2023-05-02 14:17:56,100 INFO [train.py:904] (6/8) Epoch 28, batch 6450, loss[loss=0.1696, simple_loss=0.2613, pruned_loss=0.03893, over 15215.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2853, pruned_loss=0.05532, over 3098445.08 frames. ], batch size: 190, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:18:19,027 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280518.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:18:31,066 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9130, 2.7804, 2.6235, 1.8868, 2.6191, 2.7512, 2.5641, 1.9620], device='cuda:6'), covar=tensor([0.0468, 0.0110, 0.0103, 0.0420, 0.0146, 0.0135, 0.0134, 0.0425], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0089, 0.0090, 0.0134, 0.0102, 0.0115, 0.0097, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 14:18:42,162 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280533.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:19:02,770 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280547.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:19:13,552 INFO [train.py:904] (6/8) Epoch 28, batch 6500, loss[loss=0.1982, simple_loss=0.2764, pruned_loss=0.06006, over 11645.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2834, pruned_loss=0.05476, over 3089482.30 frames. ], batch size: 248, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:19:14,159 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1558, 3.8456, 4.5320, 2.2408, 4.7444, 4.7428, 3.4673, 3.5966], device='cuda:6'), covar=tensor([0.0730, 0.0306, 0.0205, 0.1213, 0.0069, 0.0144, 0.0387, 0.0428], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0111, 0.0103, 0.0140, 0.0087, 0.0131, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 14:19:49,976 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.728e+02 2.616e+02 3.245e+02 4.274e+02 6.880e+02, threshold=6.490e+02, percent-clipped=1.0 2023-05-02 14:19:52,083 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280579.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:20:15,857 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280594.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:20:28,525 INFO [train.py:904] (6/8) Epoch 28, batch 6550, loss[loss=0.2516, simple_loss=0.3199, pruned_loss=0.09163, over 11594.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2861, pruned_loss=0.05557, over 3090258.04 frames. ], batch size: 248, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:20:33,026 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-05-02 14:20:37,507 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280608.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:20:45,844 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280614.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:21:44,262 INFO [train.py:904] (6/8) Epoch 28, batch 6600, loss[loss=0.2114, simple_loss=0.2892, pruned_loss=0.0668, over 11403.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2883, pruned_loss=0.05611, over 3095411.15 frames. ], batch size: 248, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:22:18,724 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280675.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:22:21,695 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.864e+02 2.836e+02 3.315e+02 4.097e+02 1.046e+03, threshold=6.629e+02, percent-clipped=1.0 2023-05-02 14:22:29,514 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280683.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:23:00,603 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280702.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:23:01,181 INFO [train.py:904] (6/8) Epoch 28, batch 6650, loss[loss=0.1885, simple_loss=0.2805, pruned_loss=0.04823, over 16322.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2892, pruned_loss=0.05749, over 3080360.72 frames. ], batch size: 165, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:23:07,196 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=280707.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:24:02,717 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=280744.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:24:15,398 INFO [train.py:904] (6/8) Epoch 28, batch 6700, loss[loss=0.2201, simple_loss=0.2909, pruned_loss=0.07466, over 11599.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2879, pruned_loss=0.05767, over 3074912.43 frames. ], batch size: 248, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:24:18,762 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=280755.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:24:30,607 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280763.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:24:51,655 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.083e+02 2.827e+02 3.469e+02 4.301e+02 6.613e+02, threshold=6.939e+02, percent-clipped=0.0 2023-05-02 14:25:13,161 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=280792.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:25:29,255 INFO [train.py:904] (6/8) Epoch 28, batch 6750, loss[loss=0.184, simple_loss=0.2731, pruned_loss=0.04744, over 16544.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2862, pruned_loss=0.05708, over 3096879.97 frames. ], batch size: 75, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:26:18,410 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-02 14:26:21,628 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9707, 2.1711, 2.1618, 3.5476, 2.0918, 2.5287, 2.2545, 2.3287], device='cuda:6'), covar=tensor([0.1598, 0.3772, 0.3211, 0.0656, 0.4355, 0.2511, 0.3812, 0.3400], device='cuda:6'), in_proj_covar=tensor([0.0420, 0.0473, 0.0384, 0.0336, 0.0445, 0.0542, 0.0445, 0.0553], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:26:41,172 INFO [train.py:904] (6/8) Epoch 28, batch 6800, loss[loss=0.1765, simple_loss=0.2719, pruned_loss=0.04052, over 16795.00 frames. ], tot_loss[loss=0.2, simple_loss=0.286, pruned_loss=0.05697, over 3093464.80 frames. ], batch size: 102, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:26:49,119 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7122, 1.7759, 1.6214, 1.4550, 1.9435, 1.5729, 1.5936, 1.8703], device='cuda:6'), covar=tensor([0.0225, 0.0313, 0.0439, 0.0362, 0.0242, 0.0289, 0.0194, 0.0244], device='cuda:6'), in_proj_covar=tensor([0.0229, 0.0242, 0.0232, 0.0232, 0.0244, 0.0242, 0.0240, 0.0242], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:26:49,493 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 14:27:13,225 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280874.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:27:18,675 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.178e+02 2.733e+02 3.177e+02 3.942e+02 7.279e+02, threshold=6.355e+02, percent-clipped=1.0 2023-05-02 14:27:31,216 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1893, 3.2687, 3.5183, 1.6769, 3.6919, 3.7478, 3.0152, 2.5682], device='cuda:6'), covar=tensor([0.1162, 0.0248, 0.0222, 0.1470, 0.0097, 0.0202, 0.0421, 0.0690], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0111, 0.0103, 0.0139, 0.0086, 0.0131, 0.0129, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 14:27:36,558 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280889.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:27:55,352 INFO [train.py:904] (6/8) Epoch 28, batch 6850, loss[loss=0.1767, simple_loss=0.2881, pruned_loss=0.0327, over 16713.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2868, pruned_loss=0.05719, over 3081078.26 frames. ], batch size: 62, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:27:56,253 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280903.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:29:06,760 INFO [train.py:904] (6/8) Epoch 28, batch 6900, loss[loss=0.2587, simple_loss=0.3219, pruned_loss=0.09778, over 11486.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2888, pruned_loss=0.05634, over 3103874.74 frames. ], batch size: 247, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:29:08,035 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8118, 3.9742, 3.0564, 2.4185, 2.7243, 2.5970, 4.3476, 3.4771], device='cuda:6'), covar=tensor([0.2874, 0.0707, 0.1763, 0.2817, 0.2738, 0.2120, 0.0417, 0.1359], device='cuda:6'), in_proj_covar=tensor([0.0335, 0.0275, 0.0312, 0.0327, 0.0306, 0.0277, 0.0305, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 14:29:22,252 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4206, 5.7519, 5.4762, 5.5638, 5.1821, 5.2017, 5.1436, 5.8504], device='cuda:6'), covar=tensor([0.1328, 0.0929, 0.1131, 0.0889, 0.0841, 0.0681, 0.1224, 0.0874], device='cuda:6'), in_proj_covar=tensor([0.0716, 0.0871, 0.0712, 0.0670, 0.0549, 0.0547, 0.0725, 0.0676], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:29:34,340 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280970.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:29:45,359 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.748e+02 3.135e+02 3.737e+02 6.416e+02, threshold=6.270e+02, percent-clipped=1.0 2023-05-02 14:29:53,922 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=280983.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:30:23,411 INFO [train.py:904] (6/8) Epoch 28, batch 6950, loss[loss=0.1847, simple_loss=0.2766, pruned_loss=0.0464, over 16757.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2904, pruned_loss=0.05765, over 3103257.90 frames. ], batch size: 83, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:31:03,953 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281031.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:31:30,299 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281049.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 14:31:35,801 INFO [train.py:904] (6/8) Epoch 28, batch 7000, loss[loss=0.1989, simple_loss=0.2905, pruned_loss=0.05364, over 16701.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.291, pruned_loss=0.05734, over 3100614.45 frames. ], batch size: 134, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:31:44,077 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=281058.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:32:02,474 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3698, 3.0153, 3.4549, 1.7552, 3.5656, 3.6179, 2.8585, 2.6658], device='cuda:6'), covar=tensor([0.0859, 0.0345, 0.0212, 0.1285, 0.0105, 0.0210, 0.0465, 0.0539], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0112, 0.0104, 0.0140, 0.0087, 0.0132, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 14:32:12,984 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 2.882e+02 3.440e+02 4.031e+02 6.341e+02, threshold=6.881e+02, percent-clipped=1.0 2023-05-02 14:32:24,748 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.73 vs. limit=5.0 2023-05-02 14:32:50,410 INFO [train.py:904] (6/8) Epoch 28, batch 7050, loss[loss=0.199, simple_loss=0.2958, pruned_loss=0.05111, over 16754.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2917, pruned_loss=0.05747, over 3097136.16 frames. ], batch size: 83, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:33:01,796 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=281110.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 14:33:07,185 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4025, 4.2709, 4.4600, 4.6064, 4.7273, 4.2881, 4.7117, 4.7453], device='cuda:6'), covar=tensor([0.1950, 0.1350, 0.1567, 0.0737, 0.0623, 0.1159, 0.0789, 0.0774], device='cuda:6'), in_proj_covar=tensor([0.0664, 0.0810, 0.0938, 0.0824, 0.0631, 0.0658, 0.0692, 0.0797], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:34:04,188 INFO [train.py:904] (6/8) Epoch 28, batch 7100, loss[loss=0.2327, simple_loss=0.2951, pruned_loss=0.08516, over 11465.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2904, pruned_loss=0.05782, over 3069361.85 frames. ], batch size: 247, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:34:29,352 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281169.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:34:37,862 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281174.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:34:43,396 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.809e+02 2.877e+02 3.420e+02 4.394e+02 9.887e+02, threshold=6.841e+02, percent-clipped=1.0 2023-05-02 14:34:56,879 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-02 14:34:59,572 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281189.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:35:20,726 INFO [train.py:904] (6/8) Epoch 28, batch 7150, loss[loss=0.2019, simple_loss=0.2896, pruned_loss=0.05715, over 16395.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2882, pruned_loss=0.0571, over 3082812.74 frames. ], batch size: 146, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:35:21,004 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281203.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:35:47,781 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281222.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:36:00,723 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=281230.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 14:36:09,136 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281237.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:36:29,667 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281251.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:36:31,489 INFO [train.py:904] (6/8) Epoch 28, batch 7200, loss[loss=0.1723, simple_loss=0.2672, pruned_loss=0.03876, over 16741.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2859, pruned_loss=0.05502, over 3079960.61 frames. ], batch size: 134, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:36:39,653 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.05 vs. limit=5.0 2023-05-02 14:36:56,594 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281270.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:37:07,757 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.681e+02 2.572e+02 3.176e+02 3.806e+02 6.231e+02, threshold=6.351e+02, percent-clipped=0.0 2023-05-02 14:37:21,085 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1347, 2.3711, 2.5927, 1.9148, 2.7248, 2.7878, 2.4702, 2.3753], device='cuda:6'), covar=tensor([0.0685, 0.0287, 0.0234, 0.0966, 0.0131, 0.0278, 0.0446, 0.0458], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0111, 0.0103, 0.0139, 0.0086, 0.0131, 0.0129, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 14:37:46,524 INFO [train.py:904] (6/8) Epoch 28, batch 7250, loss[loss=0.2045, simple_loss=0.29, pruned_loss=0.05947, over 16144.00 frames. ], tot_loss[loss=0.1962, simple_loss=0.284, pruned_loss=0.05417, over 3085449.06 frames. ], batch size: 165, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:38:00,864 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6122, 4.8682, 4.6592, 4.6849, 4.4144, 4.3846, 4.3768, 4.9414], device='cuda:6'), covar=tensor([0.1114, 0.0824, 0.0957, 0.0815, 0.0775, 0.1330, 0.1099, 0.0829], device='cuda:6'), in_proj_covar=tensor([0.0714, 0.0867, 0.0709, 0.0667, 0.0547, 0.0546, 0.0721, 0.0674], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:38:09,206 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281318.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:38:59,567 INFO [train.py:904] (6/8) Epoch 28, batch 7300, loss[loss=0.2277, simple_loss=0.2999, pruned_loss=0.07769, over 11569.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2836, pruned_loss=0.05415, over 3082809.09 frames. ], batch size: 246, lr: 2.39e-03, grad_scale: 2.0 2023-05-02 14:39:08,078 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281358.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:39:39,521 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.942e+02 3.043e+02 3.727e+02 4.632e+02 1.394e+03, threshold=7.454e+02, percent-clipped=7.0 2023-05-02 14:40:13,645 INFO [train.py:904] (6/8) Epoch 28, batch 7350, loss[loss=0.1866, simple_loss=0.274, pruned_loss=0.04961, over 17202.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.2848, pruned_loss=0.05511, over 3057248.97 frames. ], batch size: 46, lr: 2.39e-03, grad_scale: 2.0 2023-05-02 14:40:16,308 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=281405.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 14:40:17,486 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281406.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:40:35,052 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2885, 5.2639, 5.0201, 4.2432, 5.1819, 1.8490, 4.9124, 4.6500], device='cuda:6'), covar=tensor([0.0099, 0.0091, 0.0213, 0.0472, 0.0101, 0.3056, 0.0131, 0.0291], device='cuda:6'), in_proj_covar=tensor([0.0180, 0.0174, 0.0213, 0.0186, 0.0188, 0.0218, 0.0200, 0.0180], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:40:50,065 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281426.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:41:28,055 INFO [train.py:904] (6/8) Epoch 28, batch 7400, loss[loss=0.2398, simple_loss=0.3068, pruned_loss=0.0864, over 11113.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2854, pruned_loss=0.05537, over 3061116.63 frames. ], batch size: 247, lr: 2.39e-03, grad_scale: 2.0 2023-05-02 14:41:33,678 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281456.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:42:08,140 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.796e+02 2.701e+02 3.025e+02 3.387e+02 6.725e+02, threshold=6.049e+02, percent-clipped=0.0 2023-05-02 14:42:18,943 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=281487.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:42:42,678 INFO [train.py:904] (6/8) Epoch 28, batch 7450, loss[loss=0.1927, simple_loss=0.2905, pruned_loss=0.04745, over 16869.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2867, pruned_loss=0.05654, over 3069773.07 frames. ], batch size: 102, lr: 2.39e-03, grad_scale: 2.0 2023-05-02 14:43:06,499 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=281517.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:43:17,713 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=281525.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 14:43:58,006 INFO [train.py:904] (6/8) Epoch 28, batch 7500, loss[loss=0.2268, simple_loss=0.2965, pruned_loss=0.07857, over 11369.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2868, pruned_loss=0.05608, over 3061661.05 frames. ], batch size: 247, lr: 2.39e-03, grad_scale: 2.0 2023-05-02 14:44:36,766 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.668e+02 2.813e+02 3.506e+02 4.192e+02 1.275e+03, threshold=7.011e+02, percent-clipped=6.0 2023-05-02 14:45:11,671 INFO [train.py:904] (6/8) Epoch 28, batch 7550, loss[loss=0.251, simple_loss=0.3138, pruned_loss=0.09412, over 11312.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2862, pruned_loss=0.0568, over 3037992.37 frames. ], batch size: 247, lr: 2.39e-03, grad_scale: 2.0 2023-05-02 14:46:25,956 INFO [train.py:904] (6/8) Epoch 28, batch 7600, loss[loss=0.2132, simple_loss=0.2902, pruned_loss=0.06806, over 11695.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2854, pruned_loss=0.05683, over 3044802.83 frames. ], batch size: 247, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:47:04,823 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.856e+02 2.608e+02 3.017e+02 3.545e+02 6.203e+02, threshold=6.033e+02, percent-clipped=0.0 2023-05-02 14:47:22,041 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6468, 4.6285, 4.4811, 3.6899, 4.5689, 1.6943, 4.3381, 4.1180], device='cuda:6'), covar=tensor([0.0112, 0.0097, 0.0207, 0.0361, 0.0098, 0.3143, 0.0135, 0.0291], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0173, 0.0212, 0.0184, 0.0187, 0.0216, 0.0200, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:47:40,214 INFO [train.py:904] (6/8) Epoch 28, batch 7650, loss[loss=0.2211, simple_loss=0.3154, pruned_loss=0.06339, over 16451.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2855, pruned_loss=0.0569, over 3046261.77 frames. ], batch size: 146, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:47:43,632 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281705.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 14:47:45,549 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0479, 4.0412, 3.9869, 3.1446, 3.9987, 1.8822, 3.8286, 3.5285], device='cuda:6'), covar=tensor([0.0174, 0.0134, 0.0200, 0.0308, 0.0109, 0.2940, 0.0153, 0.0327], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0173, 0.0212, 0.0184, 0.0187, 0.0216, 0.0199, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:48:07,233 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-05-02 14:48:19,651 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1436, 2.3240, 2.2382, 3.7870, 2.2479, 2.6214, 2.3225, 2.3957], device='cuda:6'), covar=tensor([0.1496, 0.3498, 0.3176, 0.0612, 0.4305, 0.2427, 0.3619, 0.3425], device='cuda:6'), in_proj_covar=tensor([0.0420, 0.0472, 0.0384, 0.0336, 0.0445, 0.0541, 0.0446, 0.0553], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:48:51,546 INFO [train.py:904] (6/8) Epoch 28, batch 7700, loss[loss=0.2341, simple_loss=0.3026, pruned_loss=0.08286, over 11520.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2861, pruned_loss=0.05729, over 3060146.15 frames. ], batch size: 247, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:48:52,556 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281753.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 14:48:53,722 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9395, 4.7669, 4.9411, 5.1159, 5.2891, 4.7097, 5.3068, 5.2950], device='cuda:6'), covar=tensor([0.1908, 0.1249, 0.1723, 0.0795, 0.0596, 0.0938, 0.0599, 0.0707], device='cuda:6'), in_proj_covar=tensor([0.0663, 0.0811, 0.0938, 0.0825, 0.0630, 0.0657, 0.0692, 0.0798], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:49:31,077 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.035e+02 3.039e+02 3.444e+02 4.107e+02 7.486e+02, threshold=6.887e+02, percent-clipped=3.0 2023-05-02 14:49:35,794 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=281782.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:50:07,525 INFO [train.py:904] (6/8) Epoch 28, batch 7750, loss[loss=0.1984, simple_loss=0.2829, pruned_loss=0.05693, over 17029.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2864, pruned_loss=0.05738, over 3056341.89 frames. ], batch size: 55, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:50:21,324 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=281812.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:50:40,351 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281825.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:51:14,749 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4001, 5.3675, 5.2173, 4.3421, 5.2768, 1.9825, 4.9931, 4.9237], device='cuda:6'), covar=tensor([0.0123, 0.0133, 0.0235, 0.0544, 0.0129, 0.2994, 0.0183, 0.0288], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0173, 0.0212, 0.0184, 0.0187, 0.0216, 0.0199, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:51:16,563 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8188, 4.8990, 4.7051, 4.3395, 4.3450, 4.7595, 4.6411, 4.4592], device='cuda:6'), covar=tensor([0.0778, 0.0803, 0.0378, 0.0397, 0.1188, 0.0767, 0.0492, 0.0816], device='cuda:6'), in_proj_covar=tensor([0.0307, 0.0464, 0.0358, 0.0361, 0.0357, 0.0414, 0.0247, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:51:21,179 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281852.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:51:21,935 INFO [train.py:904] (6/8) Epoch 28, batch 7800, loss[loss=0.1912, simple_loss=0.2787, pruned_loss=0.0518, over 16446.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2874, pruned_loss=0.05809, over 3063832.07 frames. ], batch size: 146, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:51:52,695 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281873.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:52:03,257 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.973e+02 2.715e+02 3.377e+02 4.149e+02 7.003e+02, threshold=6.754e+02, percent-clipped=1.0 2023-05-02 14:52:37,334 INFO [train.py:904] (6/8) Epoch 28, batch 7850, loss[loss=0.215, simple_loss=0.3015, pruned_loss=0.06428, over 15328.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2874, pruned_loss=0.05676, over 3091071.97 frames. ], batch size: 190, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:52:46,235 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3467, 3.9156, 3.8908, 2.5628, 3.6024, 3.9693, 3.5615, 2.3009], device='cuda:6'), covar=tensor([0.0565, 0.0055, 0.0055, 0.0433, 0.0094, 0.0097, 0.0095, 0.0461], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0089, 0.0090, 0.0134, 0.0101, 0.0115, 0.0097, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-02 14:52:47,417 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281910.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 14:52:51,797 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=281913.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:53:38,041 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281945.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:53:48,936 INFO [train.py:904] (6/8) Epoch 28, batch 7900, loss[loss=0.2022, simple_loss=0.291, pruned_loss=0.05668, over 16660.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2865, pruned_loss=0.05597, over 3110060.41 frames. ], batch size: 134, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:54:16,283 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=281971.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 14:54:29,911 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.510e+02 2.970e+02 3.651e+02 5.631e+02, threshold=5.940e+02, percent-clipped=0.0 2023-05-02 14:55:09,783 INFO [train.py:904] (6/8) Epoch 28, batch 7950, loss[loss=0.1922, simple_loss=0.2781, pruned_loss=0.05315, over 16794.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2871, pruned_loss=0.05626, over 3102352.40 frames. ], batch size: 124, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:55:14,738 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282006.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:56:27,142 INFO [train.py:904] (6/8) Epoch 28, batch 8000, loss[loss=0.2134, simple_loss=0.304, pruned_loss=0.06142, over 16390.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2882, pruned_loss=0.05694, over 3094427.30 frames. ], batch size: 146, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:57:07,769 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.806e+02 3.354e+02 4.135e+02 1.069e+03, threshold=6.709e+02, percent-clipped=7.0 2023-05-02 14:57:11,226 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=282082.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:57:24,936 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1256, 2.4002, 2.5894, 1.9489, 2.6984, 2.7846, 2.4577, 2.3666], device='cuda:6'), covar=tensor([0.0743, 0.0291, 0.0242, 0.0981, 0.0147, 0.0318, 0.0458, 0.0453], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0112, 0.0104, 0.0140, 0.0087, 0.0132, 0.0130, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 14:57:33,385 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=282096.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:57:42,454 INFO [train.py:904] (6/8) Epoch 28, batch 8050, loss[loss=0.229, simple_loss=0.3023, pruned_loss=0.07791, over 11693.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2878, pruned_loss=0.05686, over 3077094.08 frames. ], batch size: 247, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:57:56,107 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=282112.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:58:17,577 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7635, 2.5238, 2.4987, 3.7438, 2.4630, 3.7909, 1.4961, 2.8343], device='cuda:6'), covar=tensor([0.1327, 0.0841, 0.1255, 0.0198, 0.0175, 0.0381, 0.1798, 0.0822], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0179, 0.0200, 0.0202, 0.0207, 0.0218, 0.0210, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 14:58:22,174 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=282130.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:58:57,913 INFO [train.py:904] (6/8) Epoch 28, batch 8100, loss[loss=0.2343, simple_loss=0.3036, pruned_loss=0.0825, over 11656.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2875, pruned_loss=0.05633, over 3087528.37 frames. ], batch size: 246, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:59:03,746 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282157.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:59:09,073 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=282160.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:59:15,096 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8921, 2.2145, 2.4420, 3.1294, 2.2565, 2.4055, 2.3748, 2.3181], device='cuda:6'), covar=tensor([0.1572, 0.3513, 0.2655, 0.0799, 0.4111, 0.2420, 0.3432, 0.3290], device='cuda:6'), in_proj_covar=tensor([0.0420, 0.0475, 0.0385, 0.0337, 0.0447, 0.0543, 0.0448, 0.0555], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:59:21,093 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5050, 1.7905, 2.1718, 2.4420, 2.5218, 2.7975, 1.9246, 2.7066], device='cuda:6'), covar=tensor([0.0249, 0.0575, 0.0365, 0.0381, 0.0365, 0.0233, 0.0632, 0.0187], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0198, 0.0185, 0.0191, 0.0208, 0.0164, 0.0202, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 14:59:38,411 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.025e+02 2.655e+02 3.116e+02 4.195e+02 8.434e+02, threshold=6.231e+02, percent-clipped=4.0 2023-05-02 15:00:13,681 INFO [train.py:904] (6/8) Epoch 28, batch 8150, loss[loss=0.2004, simple_loss=0.2717, pruned_loss=0.06451, over 11480.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2853, pruned_loss=0.05557, over 3095101.59 frames. ], batch size: 247, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 15:00:15,372 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6057, 3.7335, 2.3425, 4.2674, 2.8955, 4.2212, 2.5948, 3.0720], device='cuda:6'), covar=tensor([0.0329, 0.0384, 0.1748, 0.0210, 0.0832, 0.0553, 0.1486, 0.0761], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0180, 0.0196, 0.0172, 0.0180, 0.0220, 0.0205, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 15:00:17,543 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-02 15:00:21,792 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=282208.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:01:29,255 INFO [train.py:904] (6/8) Epoch 28, batch 8200, loss[loss=0.1782, simple_loss=0.2664, pruned_loss=0.04499, over 16841.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2825, pruned_loss=0.05513, over 3085666.79 frames. ], batch size: 90, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 15:01:50,076 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=282266.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 15:02:13,182 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.791e+02 2.646e+02 3.224e+02 3.830e+02 8.155e+02, threshold=6.449e+02, percent-clipped=2.0 2023-05-02 15:02:47,664 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=282301.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:02:50,333 INFO [train.py:904] (6/8) Epoch 28, batch 8250, loss[loss=0.1496, simple_loss=0.2515, pruned_loss=0.02386, over 16527.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2808, pruned_loss=0.05232, over 3065928.83 frames. ], batch size: 76, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 15:03:39,309 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9735, 1.7826, 1.5786, 1.3106, 1.8730, 1.4722, 1.5764, 1.9120], device='cuda:6'), covar=tensor([0.0291, 0.0441, 0.0620, 0.0543, 0.0339, 0.0435, 0.0241, 0.0357], device='cuda:6'), in_proj_covar=tensor([0.0225, 0.0239, 0.0231, 0.0231, 0.0240, 0.0239, 0.0237, 0.0239], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:04:07,372 INFO [train.py:904] (6/8) Epoch 28, batch 8300, loss[loss=0.179, simple_loss=0.2717, pruned_loss=0.04313, over 15366.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.2781, pruned_loss=0.04925, over 3072345.63 frames. ], batch size: 190, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 15:04:50,786 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.126e+02 2.473e+02 2.954e+02 5.391e+02, threshold=4.946e+02, percent-clipped=0.0 2023-05-02 15:05:26,438 INFO [train.py:904] (6/8) Epoch 28, batch 8350, loss[loss=0.2062, simple_loss=0.302, pruned_loss=0.05519, over 16777.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.2782, pruned_loss=0.04729, over 3094059.08 frames. ], batch size: 124, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 15:06:43,063 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=282452.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:06:43,824 INFO [train.py:904] (6/8) Epoch 28, batch 8400, loss[loss=0.1679, simple_loss=0.25, pruned_loss=0.04295, over 12168.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2751, pruned_loss=0.04539, over 3079966.83 frames. ], batch size: 248, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 15:07:26,991 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.221e+02 2.637e+02 3.332e+02 6.864e+02, threshold=5.273e+02, percent-clipped=5.0 2023-05-02 15:07:40,619 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=282489.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:08:02,089 INFO [train.py:904] (6/8) Epoch 28, batch 8450, loss[loss=0.1539, simple_loss=0.2534, pruned_loss=0.02725, over 16580.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2731, pruned_loss=0.04384, over 3070098.82 frames. ], batch size: 68, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:08:11,391 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=282508.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:09:17,968 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282550.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:09:21,852 INFO [train.py:904] (6/8) Epoch 28, batch 8500, loss[loss=0.1578, simple_loss=0.2362, pruned_loss=0.03971, over 11961.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2695, pruned_loss=0.04191, over 3064561.88 frames. ], batch size: 247, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:09:28,042 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=282556.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:09:43,300 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=282566.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 15:10:04,696 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2126, 1.7391, 2.0380, 2.2827, 2.4147, 2.5522, 1.8903, 2.5028], device='cuda:6'), covar=tensor([0.0274, 0.0566, 0.0353, 0.0378, 0.0369, 0.0242, 0.0532, 0.0172], device='cuda:6'), in_proj_covar=tensor([0.0195, 0.0197, 0.0184, 0.0189, 0.0206, 0.0163, 0.0200, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:10:07,072 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.128e+02 2.662e+02 3.202e+02 5.740e+02, threshold=5.324e+02, percent-clipped=2.0 2023-05-02 15:10:21,530 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5574, 3.6385, 3.4174, 3.0832, 3.2595, 3.5115, 3.3300, 3.3476], device='cuda:6'), covar=tensor([0.0578, 0.0756, 0.0323, 0.0303, 0.0511, 0.0534, 0.1408, 0.0537], device='cuda:6'), in_proj_covar=tensor([0.0306, 0.0464, 0.0358, 0.0359, 0.0354, 0.0414, 0.0247, 0.0427], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:10:21,910 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 15:10:28,330 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7182, 4.6999, 4.5036, 3.7779, 4.5923, 1.8221, 4.3663, 4.2035], device='cuda:6'), covar=tensor([0.0103, 0.0120, 0.0229, 0.0360, 0.0112, 0.2825, 0.0147, 0.0313], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0171, 0.0209, 0.0181, 0.0184, 0.0213, 0.0196, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:10:42,318 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=282601.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:10:45,533 INFO [train.py:904] (6/8) Epoch 28, batch 8550, loss[loss=0.1562, simple_loss=0.2524, pruned_loss=0.03005, over 17200.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2671, pruned_loss=0.04078, over 3037486.97 frames. ], batch size: 45, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:11:05,500 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=282614.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 15:11:31,444 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-02 15:12:14,466 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=282649.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:12:21,570 INFO [train.py:904] (6/8) Epoch 28, batch 8600, loss[loss=0.1691, simple_loss=0.2604, pruned_loss=0.03885, over 16564.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2674, pruned_loss=0.04003, over 3027779.66 frames. ], batch size: 57, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:12:32,819 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=282658.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:12:53,076 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6793, 4.8616, 4.9663, 4.7703, 4.8621, 5.3818, 4.8577, 4.6072], device='cuda:6'), covar=tensor([0.1176, 0.1743, 0.2515, 0.2006, 0.2543, 0.0907, 0.1703, 0.2443], device='cuda:6'), in_proj_covar=tensor([0.0418, 0.0623, 0.0687, 0.0504, 0.0676, 0.0713, 0.0537, 0.0677], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 15:12:58,385 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4756, 3.4108, 3.5170, 3.5832, 3.6579, 3.3419, 3.6072, 3.6816], device='cuda:6'), covar=tensor([0.1400, 0.0982, 0.1101, 0.0692, 0.0644, 0.2178, 0.0878, 0.0824], device='cuda:6'), in_proj_covar=tensor([0.0654, 0.0802, 0.0925, 0.0815, 0.0621, 0.0648, 0.0681, 0.0789], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:13:12,536 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=282677.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:13:15,367 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2183, 2.5412, 2.6528, 2.0135, 2.7961, 2.8484, 2.6009, 2.5340], device='cuda:6'), covar=tensor([0.0591, 0.0261, 0.0253, 0.0947, 0.0112, 0.0250, 0.0419, 0.0426], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0108, 0.0100, 0.0135, 0.0084, 0.0127, 0.0126, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 15:13:17,326 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.220e+02 2.463e+02 3.072e+02 6.042e+02, threshold=4.927e+02, percent-clipped=1.0 2023-05-02 15:13:50,787 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=282698.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 15:13:58,782 INFO [train.py:904] (6/8) Epoch 28, batch 8650, loss[loss=0.1509, simple_loss=0.2575, pruned_loss=0.02213, over 16724.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2661, pruned_loss=0.03888, over 3036935.94 frames. ], batch size: 76, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:14:28,942 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2579, 2.9970, 3.0893, 1.9598, 3.2907, 3.3311, 2.8631, 2.6977], device='cuda:6'), covar=tensor([0.0754, 0.0277, 0.0253, 0.1106, 0.0093, 0.0190, 0.0410, 0.0459], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0108, 0.0100, 0.0135, 0.0084, 0.0127, 0.0126, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 15:14:36,866 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282719.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:15:14,356 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282738.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:15:24,131 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9362, 2.1649, 2.3817, 3.1939, 2.1841, 2.3835, 2.3525, 2.2905], device='cuda:6'), covar=tensor([0.1562, 0.4080, 0.2969, 0.0781, 0.4717, 0.2904, 0.3885, 0.3986], device='cuda:6'), in_proj_covar=tensor([0.0411, 0.0464, 0.0378, 0.0327, 0.0437, 0.0531, 0.0438, 0.0543], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:15:40,174 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=282752.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:15:40,925 INFO [train.py:904] (6/8) Epoch 28, batch 8700, loss[loss=0.1701, simple_loss=0.2558, pruned_loss=0.04226, over 12418.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2636, pruned_loss=0.03768, over 3039369.67 frames. ], batch size: 250, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:15:53,197 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282759.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 15:16:12,359 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2510, 4.3492, 4.4511, 4.2403, 4.3449, 4.8356, 4.4253, 4.1400], device='cuda:6'), covar=tensor([0.1638, 0.2112, 0.2477, 0.2272, 0.2533, 0.1088, 0.1564, 0.2612], device='cuda:6'), in_proj_covar=tensor([0.0417, 0.0621, 0.0684, 0.0502, 0.0673, 0.0711, 0.0535, 0.0674], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 15:16:29,158 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6430, 2.6768, 1.9106, 2.8306, 2.1720, 2.8247, 2.1869, 2.4219], device='cuda:6'), covar=tensor([0.0334, 0.0354, 0.1254, 0.0298, 0.0666, 0.0428, 0.1249, 0.0602], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0177, 0.0193, 0.0168, 0.0176, 0.0215, 0.0201, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 15:16:29,666 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.234e+02 2.610e+02 3.054e+02 5.139e+02, threshold=5.220e+02, percent-clipped=1.0 2023-05-02 15:17:06,889 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=282800.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:17:12,009 INFO [train.py:904] (6/8) Epoch 28, batch 8750, loss[loss=0.175, simple_loss=0.2765, pruned_loss=0.03678, over 15209.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2632, pruned_loss=0.03687, over 3030416.22 frames. ], batch size: 190, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:17:18,888 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=282805.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:18:47,226 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=282845.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:19:02,828 INFO [train.py:904] (6/8) Epoch 28, batch 8800, loss[loss=0.1749, simple_loss=0.2731, pruned_loss=0.03832, over 16741.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2625, pruned_loss=0.03633, over 3038715.96 frames. ], batch size: 134, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:19:29,222 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282866.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:20:00,725 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.123e+02 2.465e+02 3.098e+02 6.856e+02, threshold=4.929e+02, percent-clipped=4.0 2023-05-02 15:20:47,905 INFO [train.py:904] (6/8) Epoch 28, batch 8850, loss[loss=0.1755, simple_loss=0.2791, pruned_loss=0.03597, over 16629.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2647, pruned_loss=0.03548, over 3043193.00 frames. ], batch size: 62, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:20:55,430 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9658, 1.8052, 1.6011, 1.4370, 1.9695, 1.5517, 1.4999, 1.8862], device='cuda:6'), covar=tensor([0.0211, 0.0388, 0.0543, 0.0424, 0.0292, 0.0353, 0.0180, 0.0268], device='cuda:6'), in_proj_covar=tensor([0.0219, 0.0235, 0.0226, 0.0227, 0.0237, 0.0235, 0.0232, 0.0234], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:22:15,634 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6335, 3.5535, 3.5361, 2.6873, 3.4483, 1.9733, 3.3162, 2.8909], device='cuda:6'), covar=tensor([0.0123, 0.0118, 0.0177, 0.0212, 0.0094, 0.2633, 0.0130, 0.0323], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0169, 0.0207, 0.0180, 0.0183, 0.0213, 0.0196, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:22:35,073 INFO [train.py:904] (6/8) Epoch 28, batch 8900, loss[loss=0.1551, simple_loss=0.2567, pruned_loss=0.02678, over 16740.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2643, pruned_loss=0.03488, over 3025301.38 frames. ], batch size: 83, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:22:57,438 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3225, 3.1285, 3.2670, 1.8345, 3.4738, 3.5149, 2.9053, 2.7813], device='cuda:6'), covar=tensor([0.0794, 0.0294, 0.0272, 0.1227, 0.0093, 0.0203, 0.0435, 0.0472], device='cuda:6'), in_proj_covar=tensor([0.0143, 0.0107, 0.0098, 0.0134, 0.0083, 0.0126, 0.0125, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 15:23:38,028 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.086e+02 2.432e+02 2.867e+02 4.868e+02, threshold=4.864e+02, percent-clipped=0.0 2023-05-02 15:24:39,869 INFO [train.py:904] (6/8) Epoch 28, batch 8950, loss[loss=0.1611, simple_loss=0.2524, pruned_loss=0.03488, over 12819.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2634, pruned_loss=0.03493, over 3050516.89 frames. ], batch size: 250, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:25:04,684 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283014.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:25:45,366 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283033.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:26:27,355 INFO [train.py:904] (6/8) Epoch 28, batch 9000, loss[loss=0.1538, simple_loss=0.243, pruned_loss=0.03226, over 11987.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2602, pruned_loss=0.034, over 3027154.56 frames. ], batch size: 247, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:26:27,356 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 15:26:38,046 INFO [train.py:938] (6/8) Epoch 28, validation: loss=0.1436, simple_loss=0.2472, pruned_loss=0.02006, over 944034.00 frames. 2023-05-02 15:26:38,047 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 15:26:41,526 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283054.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 15:27:10,351 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1365, 2.5141, 2.6033, 1.9143, 2.7634, 2.8097, 2.5118, 2.4378], device='cuda:6'), covar=tensor([0.0643, 0.0287, 0.0245, 0.0997, 0.0126, 0.0277, 0.0483, 0.0454], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0107, 0.0098, 0.0135, 0.0083, 0.0126, 0.0125, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 15:27:36,920 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.008e+02 2.381e+02 2.804e+02 4.985e+02, threshold=4.761e+02, percent-clipped=1.0 2023-05-02 15:28:21,093 INFO [train.py:904] (6/8) Epoch 28, batch 9050, loss[loss=0.1545, simple_loss=0.2453, pruned_loss=0.03187, over 15236.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2613, pruned_loss=0.03458, over 3037599.48 frames. ], batch size: 191, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:28:48,314 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=283115.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:29:46,523 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=283144.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:29:48,308 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283145.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:30:04,160 INFO [train.py:904] (6/8) Epoch 28, batch 9100, loss[loss=0.1805, simple_loss=0.2759, pruned_loss=0.04256, over 16259.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2608, pruned_loss=0.03487, over 3055909.54 frames. ], batch size: 165, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:30:20,702 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283161.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:30:48,425 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6153, 2.1223, 1.8524, 1.9847, 2.4156, 2.0806, 1.9841, 2.5251], device='cuda:6'), covar=tensor([0.0193, 0.0485, 0.0570, 0.0496, 0.0307, 0.0427, 0.0194, 0.0303], device='cuda:6'), in_proj_covar=tensor([0.0221, 0.0238, 0.0229, 0.0229, 0.0239, 0.0237, 0.0233, 0.0236], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:30:58,158 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=283176.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:31:03,952 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.48 vs. limit=2.0 2023-05-02 15:31:08,545 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.518e+02 2.173e+02 2.551e+02 2.852e+02 5.102e+02, threshold=5.103e+02, percent-clipped=2.0 2023-05-02 15:31:10,663 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4552, 1.8147, 2.1493, 2.4736, 2.4860, 2.7895, 1.9905, 2.6740], device='cuda:6'), covar=tensor([0.0333, 0.0605, 0.0408, 0.0374, 0.0389, 0.0226, 0.0605, 0.0165], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0195, 0.0183, 0.0187, 0.0204, 0.0162, 0.0199, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:31:39,558 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283193.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:31:52,706 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0888, 2.4639, 2.5945, 1.9464, 2.7196, 2.8002, 2.5381, 2.4125], device='cuda:6'), covar=tensor([0.0665, 0.0280, 0.0249, 0.0995, 0.0123, 0.0250, 0.0446, 0.0464], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0107, 0.0098, 0.0135, 0.0083, 0.0126, 0.0125, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 15:32:01,019 INFO [train.py:904] (6/8) Epoch 28, batch 9150, loss[loss=0.1828, simple_loss=0.274, pruned_loss=0.04578, over 16160.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2613, pruned_loss=0.0347, over 3053138.07 frames. ], batch size: 165, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:32:02,092 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.08 vs. limit=5.0 2023-05-02 15:32:07,141 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=283205.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:32:54,207 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2401, 3.3376, 2.0379, 3.6366, 2.4451, 3.6239, 2.2211, 2.7292], device='cuda:6'), covar=tensor([0.0381, 0.0424, 0.1794, 0.0255, 0.0918, 0.0514, 0.1666, 0.0828], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0176, 0.0192, 0.0166, 0.0175, 0.0213, 0.0200, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 15:33:44,271 INFO [train.py:904] (6/8) Epoch 28, batch 9200, loss[loss=0.1656, simple_loss=0.2563, pruned_loss=0.03746, over 12181.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2571, pruned_loss=0.03383, over 3044538.91 frames. ], batch size: 249, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:34:34,270 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.118e+02 2.676e+02 3.077e+02 8.618e+02, threshold=5.351e+02, percent-clipped=1.0 2023-05-02 15:34:39,376 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=283283.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:35:20,274 INFO [train.py:904] (6/8) Epoch 28, batch 9250, loss[loss=0.1434, simple_loss=0.2452, pruned_loss=0.02084, over 16889.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2571, pruned_loss=0.03394, over 3043087.05 frames. ], batch size: 96, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:35:44,182 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283314.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:36:28,249 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283333.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:36:55,347 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=283344.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:37:14,584 INFO [train.py:904] (6/8) Epoch 28, batch 9300, loss[loss=0.1447, simple_loss=0.2374, pruned_loss=0.02604, over 16750.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.2551, pruned_loss=0.0332, over 3019385.70 frames. ], batch size: 83, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:37:16,934 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283354.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 15:37:34,863 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283362.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:38:16,507 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.266e+02 1.947e+02 2.341e+02 2.771e+02 4.646e+02, threshold=4.683e+02, percent-clipped=0.0 2023-05-02 15:38:18,082 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283381.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:38:58,952 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283402.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 15:38:59,593 INFO [train.py:904] (6/8) Epoch 28, batch 9350, loss[loss=0.1619, simple_loss=0.2458, pruned_loss=0.03902, over 12190.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.2553, pruned_loss=0.0334, over 3025158.50 frames. ], batch size: 247, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:40:41,027 INFO [train.py:904] (6/8) Epoch 28, batch 9400, loss[loss=0.1486, simple_loss=0.2449, pruned_loss=0.02609, over 16952.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2567, pruned_loss=0.03336, over 3054272.70 frames. ], batch size: 41, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:41:00,053 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283461.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:41:19,444 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283471.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:41:39,542 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.116e+02 2.396e+02 2.984e+02 4.135e+02, threshold=4.791e+02, percent-clipped=0.0 2023-05-02 15:42:20,593 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283500.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:42:24,704 INFO [train.py:904] (6/8) Epoch 28, batch 9450, loss[loss=0.1651, simple_loss=0.2584, pruned_loss=0.03591, over 12408.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2585, pruned_loss=0.03348, over 3059755.06 frames. ], batch size: 247, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:42:36,844 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283509.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:43:49,014 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=283544.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 15:44:06,789 INFO [train.py:904] (6/8) Epoch 28, batch 9500, loss[loss=0.1665, simple_loss=0.2604, pruned_loss=0.03626, over 16151.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2578, pruned_loss=0.03333, over 3084488.58 frames. ], batch size: 165, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:45:03,874 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.031e+02 2.419e+02 2.973e+02 5.266e+02, threshold=4.838e+02, percent-clipped=2.0 2023-05-02 15:45:27,775 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4748, 3.4210, 3.5171, 3.5855, 3.6375, 3.3501, 3.6025, 3.6808], device='cuda:6'), covar=tensor([0.1401, 0.1020, 0.1146, 0.0735, 0.0657, 0.2061, 0.0891, 0.0832], device='cuda:6'), in_proj_covar=tensor([0.0639, 0.0781, 0.0902, 0.0797, 0.0606, 0.0632, 0.0667, 0.0771], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:45:53,231 INFO [train.py:904] (6/8) Epoch 28, batch 9550, loss[loss=0.1612, simple_loss=0.2586, pruned_loss=0.03187, over 12285.00 frames. ], tot_loss[loss=0.162, simple_loss=0.2572, pruned_loss=0.03337, over 3067328.17 frames. ], batch size: 246, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:45:59,190 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=283605.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 15:47:09,791 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283639.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:47:30,656 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1504, 2.2631, 2.2375, 3.9406, 2.1070, 2.6628, 2.3450, 2.4210], device='cuda:6'), covar=tensor([0.1354, 0.3916, 0.3445, 0.0510, 0.4616, 0.2550, 0.3990, 0.3553], device='cuda:6'), in_proj_covar=tensor([0.0409, 0.0463, 0.0379, 0.0325, 0.0438, 0.0529, 0.0437, 0.0541], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:47:34,671 INFO [train.py:904] (6/8) Epoch 28, batch 9600, loss[loss=0.1704, simple_loss=0.2737, pruned_loss=0.03352, over 16862.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2586, pruned_loss=0.03404, over 3044068.20 frames. ], batch size: 102, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:48:29,440 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 2.192e+02 2.621e+02 3.097e+02 5.622e+02, threshold=5.243e+02, percent-clipped=5.0 2023-05-02 15:49:23,006 INFO [train.py:904] (6/8) Epoch 28, batch 9650, loss[loss=0.1664, simple_loss=0.2661, pruned_loss=0.03341, over 16816.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2599, pruned_loss=0.03424, over 3034727.62 frames. ], batch size: 124, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:50:45,925 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0388, 5.2765, 5.0888, 5.0811, 4.8175, 4.8446, 4.6170, 5.3743], device='cuda:6'), covar=tensor([0.1157, 0.0912, 0.1035, 0.0807, 0.0718, 0.0850, 0.1344, 0.0812], device='cuda:6'), in_proj_covar=tensor([0.0692, 0.0838, 0.0683, 0.0647, 0.0528, 0.0530, 0.0696, 0.0652], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:50:59,625 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8252, 2.8246, 2.6705, 4.6825, 3.0027, 4.1708, 1.6465, 3.2338], device='cuda:6'), covar=tensor([0.1380, 0.0786, 0.1149, 0.0125, 0.0120, 0.0351, 0.1673, 0.0656], device='cuda:6'), in_proj_covar=tensor([0.0169, 0.0175, 0.0195, 0.0194, 0.0198, 0.0211, 0.0205, 0.0194], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 15:51:10,203 INFO [train.py:904] (6/8) Epoch 28, batch 9700, loss[loss=0.1814, simple_loss=0.2805, pruned_loss=0.04114, over 16983.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2594, pruned_loss=0.03433, over 3040896.93 frames. ], batch size: 109, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:51:47,283 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283771.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:52:08,987 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.180e+02 2.423e+02 2.886e+02 5.203e+02, threshold=4.846e+02, percent-clipped=0.0 2023-05-02 15:52:48,155 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283800.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:52:52,192 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2363, 3.6055, 3.6652, 2.4420, 3.2571, 3.7084, 3.4676, 2.0747], device='cuda:6'), covar=tensor([0.0569, 0.0070, 0.0063, 0.0443, 0.0150, 0.0097, 0.0100, 0.0558], device='cuda:6'), in_proj_covar=tensor([0.0134, 0.0086, 0.0087, 0.0131, 0.0099, 0.0110, 0.0094, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 15:52:52,808 INFO [train.py:904] (6/8) Epoch 28, batch 9750, loss[loss=0.1561, simple_loss=0.2546, pruned_loss=0.02881, over 16782.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.2585, pruned_loss=0.03449, over 3062711.29 frames. ], batch size: 124, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:53:24,793 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283819.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:54:13,323 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=283842.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:54:24,072 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283848.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:54:31,904 INFO [train.py:904] (6/8) Epoch 28, batch 9800, loss[loss=0.1723, simple_loss=0.2712, pruned_loss=0.03671, over 16633.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2593, pruned_loss=0.03398, over 3079425.45 frames. ], batch size: 57, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:55:23,051 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.061e+02 2.416e+02 3.000e+02 5.773e+02, threshold=4.831e+02, percent-clipped=3.0 2023-05-02 15:56:04,848 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4682, 2.0826, 1.8808, 1.7837, 2.3091, 2.0084, 1.7914, 2.3666], device='cuda:6'), covar=tensor([0.0206, 0.0471, 0.0533, 0.0543, 0.0324, 0.0458, 0.0223, 0.0314], device='cuda:6'), in_proj_covar=tensor([0.0219, 0.0237, 0.0229, 0.0229, 0.0238, 0.0237, 0.0231, 0.0235], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:56:11,010 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283900.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 15:56:15,639 INFO [train.py:904] (6/8) Epoch 28, batch 9850, loss[loss=0.1465, simple_loss=0.2474, pruned_loss=0.0228, over 16179.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2599, pruned_loss=0.0338, over 3065302.05 frames. ], batch size: 165, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:56:17,482 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=283903.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:57:37,460 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283939.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:58:04,203 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4711, 3.4212, 3.5226, 3.5901, 3.6312, 3.3431, 3.6169, 3.6808], device='cuda:6'), covar=tensor([0.1417, 0.1124, 0.1092, 0.0792, 0.0725, 0.2419, 0.0966, 0.0894], device='cuda:6'), in_proj_covar=tensor([0.0641, 0.0784, 0.0903, 0.0800, 0.0608, 0.0635, 0.0669, 0.0774], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 15:58:06,912 INFO [train.py:904] (6/8) Epoch 28, batch 9900, loss[loss=0.1687, simple_loss=0.2692, pruned_loss=0.03414, over 15315.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2601, pruned_loss=0.03336, over 3062534.28 frames. ], batch size: 190, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:58:38,484 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=283966.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:59:13,294 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.328e+02 2.134e+02 2.412e+02 2.887e+02 8.278e+02, threshold=4.823e+02, percent-clipped=3.0 2023-05-02 15:59:30,174 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283987.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:00:07,757 INFO [train.py:904] (6/8) Epoch 28, batch 9950, loss[loss=0.1649, simple_loss=0.2704, pruned_loss=0.02968, over 15391.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2619, pruned_loss=0.03342, over 3066034.17 frames. ], batch size: 192, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 16:01:08,245 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=284027.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:02:08,465 INFO [train.py:904] (6/8) Epoch 28, batch 10000, loss[loss=0.1591, simple_loss=0.2623, pruned_loss=0.02793, over 15352.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2602, pruned_loss=0.03295, over 3065324.63 frames. ], batch size: 191, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 16:03:03,885 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 2.120e+02 2.490e+02 3.242e+02 7.022e+02, threshold=4.980e+02, percent-clipped=2.0 2023-05-02 16:03:27,388 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8436, 1.4084, 1.7188, 1.7654, 1.8757, 1.9419, 1.7015, 1.8995], device='cuda:6'), covar=tensor([0.0290, 0.0461, 0.0261, 0.0320, 0.0347, 0.0257, 0.0479, 0.0175], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0196, 0.0183, 0.0186, 0.0205, 0.0161, 0.0199, 0.0161], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:03:50,397 INFO [train.py:904] (6/8) Epoch 28, batch 10050, loss[loss=0.152, simple_loss=0.252, pruned_loss=0.02598, over 12207.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2597, pruned_loss=0.03246, over 3068028.28 frames. ], batch size: 248, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 16:05:24,422 INFO [train.py:904] (6/8) Epoch 28, batch 10100, loss[loss=0.1424, simple_loss=0.2364, pruned_loss=0.02417, over 15414.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.26, pruned_loss=0.03254, over 3066984.11 frames. ], batch size: 191, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 16:06:20,724 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.509e+02 2.222e+02 2.734e+02 3.195e+02 6.018e+02, threshold=5.468e+02, percent-clipped=8.0 2023-05-02 16:06:39,381 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=284198.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:06:41,999 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=284200.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 16:06:44,465 INFO [train.py:904] (6/8) Epoch 28, batch 10150, loss[loss=0.1561, simple_loss=0.2427, pruned_loss=0.03481, over 12110.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2588, pruned_loss=0.03297, over 3038761.55 frames. ], batch size: 248, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 16:07:10,315 INFO [train.py:904] (6/8) Epoch 29, batch 0, loss[loss=0.2129, simple_loss=0.2881, pruned_loss=0.06886, over 16738.00 frames. ], tot_loss[loss=0.2129, simple_loss=0.2881, pruned_loss=0.06886, over 16738.00 frames. ], batch size: 124, lr: 2.34e-03, grad_scale: 8.0 2023-05-02 16:07:10,315 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 16:07:17,744 INFO [train.py:938] (6/8) Epoch 29, validation: loss=0.1427, simple_loss=0.246, pruned_loss=0.0197, over 944034.00 frames. 2023-05-02 16:07:17,744 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 16:07:53,480 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 16:08:07,747 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8954, 4.5434, 4.5488, 3.2123, 3.7642, 4.5205, 3.9711, 2.6122], device='cuda:6'), covar=tensor([0.0510, 0.0072, 0.0044, 0.0390, 0.0166, 0.0090, 0.0099, 0.0489], device='cuda:6'), in_proj_covar=tensor([0.0135, 0.0087, 0.0088, 0.0132, 0.0100, 0.0111, 0.0095, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-02 16:08:10,907 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4180, 3.5435, 3.9586, 2.2914, 3.2259, 2.5870, 3.7601, 3.8415], device='cuda:6'), covar=tensor([0.0323, 0.1169, 0.0587, 0.2267, 0.0949, 0.1126, 0.0769, 0.1294], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0163, 0.0165, 0.0153, 0.0143, 0.0129, 0.0141, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 16:08:18,231 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=284248.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 16:08:26,889 INFO [train.py:904] (6/8) Epoch 29, batch 50, loss[loss=0.1432, simple_loss=0.2335, pruned_loss=0.02646, over 17022.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2632, pruned_loss=0.04436, over 757925.61 frames. ], batch size: 41, lr: 2.34e-03, grad_scale: 2.0 2023-05-02 16:09:05,863 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.36 vs. limit=5.0 2023-05-02 16:09:08,278 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.709e+02 2.422e+02 3.076e+02 3.842e+02 2.178e+03, threshold=6.152e+02, percent-clipped=5.0 2023-05-02 16:09:25,387 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9584, 3.0858, 2.8400, 5.1413, 4.1729, 4.4228, 1.6773, 3.2916], device='cuda:6'), covar=tensor([0.1443, 0.0786, 0.1251, 0.0275, 0.0254, 0.0496, 0.1750, 0.0802], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0177, 0.0197, 0.0197, 0.0200, 0.0214, 0.0208, 0.0196], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 16:09:37,378 INFO [train.py:904] (6/8) Epoch 29, batch 100, loss[loss=0.1902, simple_loss=0.2691, pruned_loss=0.05569, over 16879.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2631, pruned_loss=0.04346, over 1313397.81 frames. ], batch size: 109, lr: 2.34e-03, grad_scale: 2.0 2023-05-02 16:10:02,627 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=284322.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:10:46,108 INFO [train.py:904] (6/8) Epoch 29, batch 150, loss[loss=0.1763, simple_loss=0.2563, pruned_loss=0.04813, over 16880.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2616, pruned_loss=0.04234, over 1759548.63 frames. ], batch size: 102, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:10:53,677 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.35 vs. limit=5.0 2023-05-02 16:11:06,485 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-05-02 16:11:25,623 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.248e+02 2.614e+02 3.025e+02 1.081e+03, threshold=5.228e+02, percent-clipped=2.0 2023-05-02 16:11:49,575 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6785, 3.7763, 2.3678, 4.1201, 2.9462, 4.0425, 2.4283, 3.0843], device='cuda:6'), covar=tensor([0.0338, 0.0378, 0.1668, 0.0379, 0.0834, 0.0765, 0.1548, 0.0826], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0179, 0.0196, 0.0170, 0.0179, 0.0217, 0.0204, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 16:11:55,141 INFO [train.py:904] (6/8) Epoch 29, batch 200, loss[loss=0.1992, simple_loss=0.2804, pruned_loss=0.05893, over 16305.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2607, pruned_loss=0.04187, over 2112519.76 frames. ], batch size: 165, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:12:51,520 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=284443.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:13:04,661 INFO [train.py:904] (6/8) Epoch 29, batch 250, loss[loss=0.185, simple_loss=0.2664, pruned_loss=0.0518, over 16483.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.258, pruned_loss=0.0411, over 2382048.42 frames. ], batch size: 75, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:13:11,637 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1458, 5.7151, 5.8219, 5.5057, 5.6135, 6.2080, 5.6732, 5.4528], device='cuda:6'), covar=tensor([0.0905, 0.2088, 0.2584, 0.2323, 0.2821, 0.1117, 0.1710, 0.2377], device='cuda:6'), in_proj_covar=tensor([0.0417, 0.0625, 0.0693, 0.0510, 0.0680, 0.0716, 0.0538, 0.0679], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 16:13:47,576 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.177e+02 2.500e+02 3.084e+02 6.085e+02, threshold=5.001e+02, percent-clipped=1.0 2023-05-02 16:13:54,875 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.7895, 6.1497, 5.8078, 5.9487, 5.5060, 5.5654, 5.6646, 6.2960], device='cuda:6'), covar=tensor([0.1411, 0.1068, 0.1478, 0.0938, 0.0931, 0.0644, 0.1188, 0.0983], device='cuda:6'), in_proj_covar=tensor([0.0698, 0.0847, 0.0693, 0.0655, 0.0535, 0.0534, 0.0709, 0.0662], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:14:08,445 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=284498.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:14:16,992 INFO [train.py:904] (6/8) Epoch 29, batch 300, loss[loss=0.1526, simple_loss=0.2351, pruned_loss=0.03507, over 16446.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2555, pruned_loss=0.0402, over 2587476.98 frames. ], batch size: 75, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:14:17,383 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=284504.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:14:26,652 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2128, 5.1690, 4.9669, 4.4882, 4.9564, 1.9088, 4.7731, 4.7678], device='cuda:6'), covar=tensor([0.0121, 0.0115, 0.0253, 0.0411, 0.0134, 0.3044, 0.0172, 0.0289], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0170, 0.0207, 0.0178, 0.0184, 0.0214, 0.0195, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:15:14,306 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=284546.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:15:23,224 INFO [train.py:904] (6/8) Epoch 29, batch 350, loss[loss=0.1752, simple_loss=0.2436, pruned_loss=0.05341, over 16732.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2536, pruned_loss=0.03971, over 2749852.13 frames. ], batch size: 83, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:16:02,935 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.083e+02 2.374e+02 2.884e+02 4.611e+02, threshold=4.748e+02, percent-clipped=0.0 2023-05-02 16:16:31,991 INFO [train.py:904] (6/8) Epoch 29, batch 400, loss[loss=0.1558, simple_loss=0.2524, pruned_loss=0.02956, over 17030.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2513, pruned_loss=0.03945, over 2874389.81 frames. ], batch size: 50, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:16:57,085 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=284622.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:17:36,680 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.23 vs. limit=5.0 2023-05-02 16:17:41,209 INFO [train.py:904] (6/8) Epoch 29, batch 450, loss[loss=0.178, simple_loss=0.2546, pruned_loss=0.05071, over 16878.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2494, pruned_loss=0.03844, over 2977809.36 frames. ], batch size: 96, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:17:45,078 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.98 vs. limit=5.0 2023-05-02 16:17:46,043 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2024, 5.1634, 4.9105, 4.4018, 4.9921, 1.8609, 4.8143, 4.6464], device='cuda:6'), covar=tensor([0.0107, 0.0097, 0.0245, 0.0421, 0.0117, 0.3248, 0.0154, 0.0300], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0172, 0.0210, 0.0180, 0.0186, 0.0216, 0.0198, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:18:03,052 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=284670.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:18:18,521 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 1.978e+02 2.312e+02 2.858e+02 4.933e+02, threshold=4.623e+02, percent-clipped=2.0 2023-05-02 16:18:47,401 INFO [train.py:904] (6/8) Epoch 29, batch 500, loss[loss=0.162, simple_loss=0.243, pruned_loss=0.0405, over 16491.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.248, pruned_loss=0.0379, over 3056726.32 frames. ], batch size: 75, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:18:49,119 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6231, 3.6164, 4.2703, 2.3948, 3.4387, 2.6706, 4.0067, 3.9349], device='cuda:6'), covar=tensor([0.0255, 0.1033, 0.0422, 0.2018, 0.0802, 0.0998, 0.0569, 0.1170], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0168, 0.0169, 0.0156, 0.0147, 0.0132, 0.0145, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 16:19:48,342 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 16:19:56,201 INFO [train.py:904] (6/8) Epoch 29, batch 550, loss[loss=0.1675, simple_loss=0.2466, pruned_loss=0.04421, over 16265.00 frames. ], tot_loss[loss=0.1609, simple_loss=0.2472, pruned_loss=0.03724, over 3103837.00 frames. ], batch size: 165, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:20:19,893 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-05-02 16:20:35,714 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 1.886e+02 2.233e+02 2.634e+02 5.354e+02, threshold=4.465e+02, percent-clipped=1.0 2023-05-02 16:20:41,342 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3484, 3.5101, 3.7032, 2.5708, 3.3533, 3.8299, 3.4590, 2.1172], device='cuda:6'), covar=tensor([0.0549, 0.0188, 0.0068, 0.0430, 0.0141, 0.0102, 0.0126, 0.0549], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0090, 0.0090, 0.0135, 0.0102, 0.0115, 0.0098, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 16:20:57,942 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=284799.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:21:04,186 INFO [train.py:904] (6/8) Epoch 29, batch 600, loss[loss=0.1691, simple_loss=0.2361, pruned_loss=0.051, over 16730.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2468, pruned_loss=0.03792, over 3153838.20 frames. ], batch size: 124, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:21:21,956 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=284816.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:22:12,346 INFO [train.py:904] (6/8) Epoch 29, batch 650, loss[loss=0.1462, simple_loss=0.2293, pruned_loss=0.03157, over 15960.00 frames. ], tot_loss[loss=0.1598, simple_loss=0.2449, pruned_loss=0.03733, over 3196552.58 frames. ], batch size: 35, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:22:46,742 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=284877.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:22:53,843 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.111e+02 2.518e+02 3.065e+02 7.660e+02, threshold=5.037e+02, percent-clipped=4.0 2023-05-02 16:23:22,623 INFO [train.py:904] (6/8) Epoch 29, batch 700, loss[loss=0.1447, simple_loss=0.2251, pruned_loss=0.03219, over 12256.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.2454, pruned_loss=0.03776, over 3221155.62 frames. ], batch size: 247, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:24:05,504 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5222, 2.4235, 2.4237, 4.2580, 2.3907, 2.7881, 2.4824, 2.6486], device='cuda:6'), covar=tensor([0.1341, 0.3886, 0.3351, 0.0589, 0.4180, 0.2777, 0.3887, 0.3734], device='cuda:6'), in_proj_covar=tensor([0.0420, 0.0474, 0.0389, 0.0336, 0.0447, 0.0542, 0.0448, 0.0555], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:24:12,232 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=284939.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:24:30,115 INFO [train.py:904] (6/8) Epoch 29, batch 750, loss[loss=0.1424, simple_loss=0.2246, pruned_loss=0.03017, over 16817.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.2457, pruned_loss=0.0377, over 3246262.26 frames. ], batch size: 39, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:25:13,142 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.091e+02 2.322e+02 2.629e+02 4.322e+02, threshold=4.644e+02, percent-clipped=0.0 2023-05-02 16:25:30,653 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=284996.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:25:37,929 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=285000.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:25:42,932 INFO [train.py:904] (6/8) Epoch 29, batch 800, loss[loss=0.1528, simple_loss=0.2466, pruned_loss=0.02944, over 17117.00 frames. ], tot_loss[loss=0.1596, simple_loss=0.2455, pruned_loss=0.03687, over 3258623.16 frames. ], batch size: 48, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:26:08,909 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4193, 4.6957, 4.5273, 4.5518, 4.2718, 4.2315, 4.2366, 4.7597], device='cuda:6'), covar=tensor([0.1211, 0.0982, 0.0984, 0.0820, 0.0758, 0.1521, 0.1096, 0.0946], device='cuda:6'), in_proj_covar=tensor([0.0720, 0.0869, 0.0712, 0.0673, 0.0549, 0.0549, 0.0730, 0.0681], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:26:34,303 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=285042.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 16:26:49,075 INFO [train.py:904] (6/8) Epoch 29, batch 850, loss[loss=0.1548, simple_loss=0.2387, pruned_loss=0.0355, over 16419.00 frames. ], tot_loss[loss=0.1591, simple_loss=0.2452, pruned_loss=0.03654, over 3265564.60 frames. ], batch size: 146, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:26:53,013 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=285057.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:26:54,231 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9023, 2.9095, 2.6430, 5.0263, 4.0414, 4.3169, 1.6584, 3.0978], device='cuda:6'), covar=tensor([0.1392, 0.0824, 0.1321, 0.0232, 0.0235, 0.0398, 0.1713, 0.0871], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0180, 0.0200, 0.0203, 0.0204, 0.0217, 0.0210, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 16:27:00,102 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1383, 2.6597, 2.2506, 2.4681, 2.9901, 2.7478, 2.9836, 3.0994], device='cuda:6'), covar=tensor([0.0298, 0.0497, 0.0620, 0.0531, 0.0304, 0.0397, 0.0282, 0.0336], device='cuda:6'), in_proj_covar=tensor([0.0232, 0.0246, 0.0236, 0.0236, 0.0247, 0.0246, 0.0241, 0.0245], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:27:09,258 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0938, 4.5714, 4.5757, 3.3733, 3.7670, 4.5149, 3.9788, 2.7216], device='cuda:6'), covar=tensor([0.0477, 0.0075, 0.0047, 0.0368, 0.0169, 0.0106, 0.0116, 0.0490], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0090, 0.0091, 0.0135, 0.0102, 0.0115, 0.0098, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 16:27:31,498 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.142e+02 2.515e+02 3.015e+02 5.883e+02, threshold=5.030e+02, percent-clipped=4.0 2023-05-02 16:27:47,550 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 16:27:49,509 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=285099.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:27:54,993 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=285103.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 16:27:55,637 INFO [train.py:904] (6/8) Epoch 29, batch 900, loss[loss=0.1653, simple_loss=0.2588, pruned_loss=0.03595, over 17126.00 frames. ], tot_loss[loss=0.1591, simple_loss=0.2455, pruned_loss=0.03635, over 3284623.50 frames. ], batch size: 49, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:28:20,437 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4770, 3.3396, 2.7392, 2.2749, 2.1920, 2.3353, 3.4864, 2.9238], device='cuda:6'), covar=tensor([0.3016, 0.0735, 0.1941, 0.2844, 0.2848, 0.2369, 0.0508, 0.1794], device='cuda:6'), in_proj_covar=tensor([0.0338, 0.0274, 0.0314, 0.0328, 0.0304, 0.0279, 0.0305, 0.0353], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 16:28:56,113 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=285147.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:29:05,600 INFO [train.py:904] (6/8) Epoch 29, batch 950, loss[loss=0.1732, simple_loss=0.2617, pruned_loss=0.04237, over 17046.00 frames. ], tot_loss[loss=0.159, simple_loss=0.2457, pruned_loss=0.0362, over 3294655.27 frames. ], batch size: 55, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:29:30,490 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=285172.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:29:47,164 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.394e+02 2.122e+02 2.524e+02 3.029e+02 7.170e+02, threshold=5.047e+02, percent-clipped=3.0 2023-05-02 16:29:53,230 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9111, 2.1326, 2.5919, 2.9779, 2.7146, 3.4433, 2.4072, 3.4440], device='cuda:6'), covar=tensor([0.0320, 0.0588, 0.0401, 0.0399, 0.0449, 0.0241, 0.0541, 0.0184], device='cuda:6'), in_proj_covar=tensor([0.0201, 0.0202, 0.0190, 0.0195, 0.0212, 0.0169, 0.0206, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 16:30:14,198 INFO [train.py:904] (6/8) Epoch 29, batch 1000, loss[loss=0.1434, simple_loss=0.2339, pruned_loss=0.02643, over 17222.00 frames. ], tot_loss[loss=0.1583, simple_loss=0.2447, pruned_loss=0.03591, over 3298740.16 frames. ], batch size: 45, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:30:24,379 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6056, 3.2760, 3.6540, 2.1004, 3.7198, 3.7556, 3.1698, 2.8709], device='cuda:6'), covar=tensor([0.0820, 0.0286, 0.0202, 0.1148, 0.0138, 0.0242, 0.0415, 0.0487], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0111, 0.0102, 0.0140, 0.0087, 0.0132, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 16:31:24,212 INFO [train.py:904] (6/8) Epoch 29, batch 1050, loss[loss=0.1547, simple_loss=0.2354, pruned_loss=0.03706, over 16770.00 frames. ], tot_loss[loss=0.158, simple_loss=0.2445, pruned_loss=0.03577, over 3301424.04 frames. ], batch size: 83, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:32:05,305 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.383e+02 1.938e+02 2.240e+02 2.554e+02 8.528e+02, threshold=4.479e+02, percent-clipped=1.0 2023-05-02 16:32:19,760 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=285295.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:32:31,238 INFO [train.py:904] (6/8) Epoch 29, batch 1100, loss[loss=0.1552, simple_loss=0.2504, pruned_loss=0.02997, over 17277.00 frames. ], tot_loss[loss=0.1582, simple_loss=0.2448, pruned_loss=0.03573, over 3303852.95 frames. ], batch size: 52, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:33:37,435 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=285352.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:33:40,186 INFO [train.py:904] (6/8) Epoch 29, batch 1150, loss[loss=0.1631, simple_loss=0.2561, pruned_loss=0.03502, over 17092.00 frames. ], tot_loss[loss=0.1575, simple_loss=0.2444, pruned_loss=0.03531, over 3314384.34 frames. ], batch size: 53, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:34:22,233 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.496e+02 2.113e+02 2.491e+02 2.895e+02 1.258e+03, threshold=4.983e+02, percent-clipped=2.0 2023-05-02 16:34:40,309 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=285398.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 16:34:47,458 INFO [train.py:904] (6/8) Epoch 29, batch 1200, loss[loss=0.1584, simple_loss=0.2375, pruned_loss=0.03967, over 16525.00 frames. ], tot_loss[loss=0.1569, simple_loss=0.2438, pruned_loss=0.03503, over 3316993.99 frames. ], batch size: 146, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:35:47,428 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1284, 5.6827, 5.8156, 5.4708, 5.5864, 6.1852, 5.6768, 5.4061], device='cuda:6'), covar=tensor([0.0914, 0.2040, 0.2480, 0.2184, 0.2534, 0.0959, 0.1589, 0.2276], device='cuda:6'), in_proj_covar=tensor([0.0432, 0.0645, 0.0717, 0.0524, 0.0703, 0.0737, 0.0555, 0.0701], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 16:35:56,648 INFO [train.py:904] (6/8) Epoch 29, batch 1250, loss[loss=0.1534, simple_loss=0.249, pruned_loss=0.02894, over 17243.00 frames. ], tot_loss[loss=0.1575, simple_loss=0.2441, pruned_loss=0.03552, over 3313305.84 frames. ], batch size: 52, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:36:21,900 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=285472.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:36:36,039 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-05-02 16:36:38,696 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.107e+02 2.455e+02 3.038e+02 4.730e+02, threshold=4.911e+02, percent-clipped=0.0 2023-05-02 16:36:40,504 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8821, 4.4225, 3.1510, 2.4317, 2.7661, 2.7516, 4.8115, 3.6291], device='cuda:6'), covar=tensor([0.3030, 0.0528, 0.1895, 0.3134, 0.2877, 0.2206, 0.0339, 0.1497], device='cuda:6'), in_proj_covar=tensor([0.0339, 0.0275, 0.0315, 0.0329, 0.0306, 0.0281, 0.0306, 0.0355], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 16:37:05,018 INFO [train.py:904] (6/8) Epoch 29, batch 1300, loss[loss=0.1422, simple_loss=0.2331, pruned_loss=0.0257, over 17182.00 frames. ], tot_loss[loss=0.1569, simple_loss=0.2433, pruned_loss=0.03523, over 3306887.88 frames. ], batch size: 46, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:37:28,089 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=285520.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:38:04,225 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2023-05-02 16:38:12,054 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0797, 5.1818, 5.5796, 5.5357, 5.5484, 5.1800, 5.1463, 4.9824], device='cuda:6'), covar=tensor([0.0380, 0.0558, 0.0340, 0.0423, 0.0474, 0.0411, 0.0946, 0.0464], device='cuda:6'), in_proj_covar=tensor([0.0444, 0.0502, 0.0483, 0.0445, 0.0532, 0.0507, 0.0582, 0.0407], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 16:38:13,878 INFO [train.py:904] (6/8) Epoch 29, batch 1350, loss[loss=0.1844, simple_loss=0.2592, pruned_loss=0.05484, over 16895.00 frames. ], tot_loss[loss=0.1569, simple_loss=0.2434, pruned_loss=0.03515, over 3307832.16 frames. ], batch size: 96, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:38:57,506 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7509, 4.8947, 5.0379, 4.8151, 4.8747, 5.4749, 4.9417, 4.6454], device='cuda:6'), covar=tensor([0.1406, 0.2280, 0.2455, 0.2498, 0.2600, 0.1072, 0.1947, 0.2840], device='cuda:6'), in_proj_covar=tensor([0.0431, 0.0643, 0.0716, 0.0524, 0.0701, 0.0734, 0.0554, 0.0699], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 16:38:58,360 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.007e+02 2.357e+02 2.705e+02 4.553e+02, threshold=4.714e+02, percent-clipped=0.0 2023-05-02 16:39:11,759 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=285595.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:39:22,845 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6189, 6.0406, 5.7336, 5.8827, 5.3800, 5.4885, 5.4584, 6.1747], device='cuda:6'), covar=tensor([0.1560, 0.0977, 0.1175, 0.0899, 0.0930, 0.0669, 0.1256, 0.0952], device='cuda:6'), in_proj_covar=tensor([0.0731, 0.0881, 0.0722, 0.0685, 0.0558, 0.0557, 0.0742, 0.0692], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:39:24,823 INFO [train.py:904] (6/8) Epoch 29, batch 1400, loss[loss=0.1412, simple_loss=0.2245, pruned_loss=0.02897, over 11613.00 frames. ], tot_loss[loss=0.1567, simple_loss=0.2431, pruned_loss=0.03511, over 3296939.76 frames. ], batch size: 248, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:39:49,612 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=285622.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:40:12,979 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8347, 5.0632, 5.2096, 4.9812, 5.0488, 5.6629, 5.1083, 4.7394], device='cuda:6'), covar=tensor([0.1373, 0.2014, 0.2451, 0.2100, 0.2577, 0.1054, 0.1809, 0.2701], device='cuda:6'), in_proj_covar=tensor([0.0431, 0.0643, 0.0715, 0.0524, 0.0702, 0.0734, 0.0554, 0.0700], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 16:40:19,822 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=285643.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:40:31,669 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=285652.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:40:33,554 INFO [train.py:904] (6/8) Epoch 29, batch 1450, loss[loss=0.1707, simple_loss=0.2421, pruned_loss=0.04963, over 16845.00 frames. ], tot_loss[loss=0.1569, simple_loss=0.2424, pruned_loss=0.0357, over 3299180.85 frames. ], batch size: 102, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:41:13,991 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=285683.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:41:16,389 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.247e+02 2.721e+02 3.292e+02 6.741e+02, threshold=5.441e+02, percent-clipped=4.0 2023-05-02 16:41:34,326 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=285698.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 16:41:37,507 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=285700.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:41:43,180 INFO [train.py:904] (6/8) Epoch 29, batch 1500, loss[loss=0.1636, simple_loss=0.2446, pruned_loss=0.04127, over 15410.00 frames. ], tot_loss[loss=0.1577, simple_loss=0.2432, pruned_loss=0.03609, over 3296534.80 frames. ], batch size: 190, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:41:44,161 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4928, 5.8872, 5.6420, 5.7178, 5.2606, 5.3834, 5.2713, 6.0420], device='cuda:6'), covar=tensor([0.1631, 0.1059, 0.1163, 0.0904, 0.0943, 0.0725, 0.1360, 0.0982], device='cuda:6'), in_proj_covar=tensor([0.0733, 0.0883, 0.0724, 0.0686, 0.0559, 0.0558, 0.0744, 0.0693], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:41:49,569 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1071, 3.1919, 3.2223, 2.1909, 3.0700, 3.3728, 3.0880, 1.7738], device='cuda:6'), covar=tensor([0.0610, 0.0146, 0.0115, 0.0510, 0.0179, 0.0156, 0.0159, 0.0695], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0091, 0.0092, 0.0136, 0.0103, 0.0117, 0.0099, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 16:42:40,635 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=285746.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 16:42:51,735 INFO [train.py:904] (6/8) Epoch 29, batch 1550, loss[loss=0.1651, simple_loss=0.2406, pruned_loss=0.04481, over 16868.00 frames. ], tot_loss[loss=0.1588, simple_loss=0.2443, pruned_loss=0.03665, over 3300573.40 frames. ], batch size: 96, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:43:07,817 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2398, 2.9107, 3.1458, 1.9275, 3.2327, 3.2750, 2.7756, 2.5663], device='cuda:6'), covar=tensor([0.0874, 0.0313, 0.0263, 0.1097, 0.0145, 0.0251, 0.0474, 0.0515], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0112, 0.0104, 0.0141, 0.0087, 0.0133, 0.0131, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 16:43:07,854 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3103, 2.1096, 1.7653, 1.8058, 2.3527, 2.0814, 2.0629, 2.4385], device='cuda:6'), covar=tensor([0.0290, 0.0461, 0.0604, 0.0549, 0.0289, 0.0405, 0.0278, 0.0349], device='cuda:6'), in_proj_covar=tensor([0.0237, 0.0251, 0.0240, 0.0240, 0.0251, 0.0250, 0.0247, 0.0249], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:43:34,663 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 2.209e+02 2.639e+02 3.034e+02 6.721e+02, threshold=5.278e+02, percent-clipped=2.0 2023-05-02 16:44:00,607 INFO [train.py:904] (6/8) Epoch 29, batch 1600, loss[loss=0.192, simple_loss=0.2808, pruned_loss=0.05159, over 16489.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2465, pruned_loss=0.03749, over 3305175.90 frames. ], batch size: 68, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:44:17,208 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2711, 4.1346, 4.3297, 4.4630, 4.5262, 4.1627, 4.3432, 4.5456], device='cuda:6'), covar=tensor([0.1954, 0.1571, 0.1550, 0.0937, 0.0845, 0.1368, 0.3133, 0.1358], device='cuda:6'), in_proj_covar=tensor([0.0699, 0.0853, 0.0985, 0.0869, 0.0660, 0.0683, 0.0727, 0.0840], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:44:33,731 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2453, 2.3203, 2.5096, 3.9581, 2.3292, 2.6390, 2.3441, 2.4766], device='cuda:6'), covar=tensor([0.1668, 0.3952, 0.3108, 0.0761, 0.4085, 0.2671, 0.4201, 0.3295], device='cuda:6'), in_proj_covar=tensor([0.0424, 0.0478, 0.0391, 0.0339, 0.0449, 0.0548, 0.0451, 0.0562], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:44:53,556 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2023-05-02 16:45:09,516 INFO [train.py:904] (6/8) Epoch 29, batch 1650, loss[loss=0.1578, simple_loss=0.2567, pruned_loss=0.02939, over 17014.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2479, pruned_loss=0.03767, over 3303612.33 frames. ], batch size: 50, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:45:50,990 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.518e+02 2.202e+02 2.447e+02 3.087e+02 4.596e+02, threshold=4.895e+02, percent-clipped=0.0 2023-05-02 16:46:16,879 INFO [train.py:904] (6/8) Epoch 29, batch 1700, loss[loss=0.1806, simple_loss=0.2609, pruned_loss=0.05016, over 16737.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2504, pruned_loss=0.03829, over 3300781.94 frames. ], batch size: 134, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:47:12,947 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2023-05-02 16:47:24,308 INFO [train.py:904] (6/8) Epoch 29, batch 1750, loss[loss=0.1625, simple_loss=0.2613, pruned_loss=0.03183, over 17053.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.251, pruned_loss=0.03815, over 3302199.52 frames. ], batch size: 50, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:47:58,367 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=285978.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:48:07,053 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.447e+02 2.051e+02 2.355e+02 2.951e+02 5.834e+02, threshold=4.710e+02, percent-clipped=3.0 2023-05-02 16:48:36,742 INFO [train.py:904] (6/8) Epoch 29, batch 1800, loss[loss=0.1997, simple_loss=0.2845, pruned_loss=0.05749, over 11961.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2517, pruned_loss=0.03816, over 3308460.07 frames. ], batch size: 246, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:49:43,955 INFO [train.py:904] (6/8) Epoch 29, batch 1850, loss[loss=0.1728, simple_loss=0.2745, pruned_loss=0.03557, over 16661.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2526, pruned_loss=0.03843, over 3308828.38 frames. ], batch size: 57, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:49:54,275 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6698, 3.7464, 4.4615, 2.7289, 3.6054, 3.0002, 4.1851, 4.0656], device='cuda:6'), covar=tensor([0.0233, 0.0950, 0.0384, 0.1822, 0.0749, 0.0879, 0.0503, 0.0998], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0171, 0.0171, 0.0158, 0.0149, 0.0133, 0.0146, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 16:50:28,094 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.081e+02 2.345e+02 2.768e+02 6.474e+02, threshold=4.689e+02, percent-clipped=2.0 2023-05-02 16:50:35,229 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9786, 2.1559, 2.3079, 3.5511, 2.1804, 2.3813, 2.2447, 2.2960], device='cuda:6'), covar=tensor([0.1685, 0.3889, 0.3287, 0.0804, 0.4077, 0.2844, 0.3913, 0.3547], device='cuda:6'), in_proj_covar=tensor([0.0425, 0.0480, 0.0392, 0.0340, 0.0450, 0.0550, 0.0453, 0.0563], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:50:53,098 INFO [train.py:904] (6/8) Epoch 29, batch 1900, loss[loss=0.1881, simple_loss=0.2791, pruned_loss=0.04856, over 12360.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2522, pruned_loss=0.03793, over 3309175.36 frames. ], batch size: 248, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:51:20,346 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286123.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 16:51:29,268 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286129.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:51:55,614 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 16:51:58,698 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-02 16:51:59,693 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-05-02 16:52:04,246 INFO [train.py:904] (6/8) Epoch 29, batch 1950, loss[loss=0.1384, simple_loss=0.232, pruned_loss=0.02244, over 17235.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2529, pruned_loss=0.03806, over 3302900.88 frames. ], batch size: 43, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:52:23,331 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7167, 3.8852, 2.9230, 2.3207, 2.4494, 2.4407, 3.9679, 3.3238], device='cuda:6'), covar=tensor([0.2820, 0.0560, 0.1826, 0.3202, 0.2795, 0.2196, 0.0543, 0.1553], device='cuda:6'), in_proj_covar=tensor([0.0340, 0.0277, 0.0316, 0.0330, 0.0308, 0.0281, 0.0307, 0.0357], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 16:52:45,869 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5862, 2.6898, 2.6674, 4.5151, 2.6439, 3.0090, 2.7211, 2.8412], device='cuda:6'), covar=tensor([0.1445, 0.3627, 0.3153, 0.0586, 0.3827, 0.2682, 0.3721, 0.3650], device='cuda:6'), in_proj_covar=tensor([0.0424, 0.0478, 0.0390, 0.0339, 0.0448, 0.0548, 0.0451, 0.0560], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:52:46,895 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286184.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 16:52:48,710 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.778e+02 2.194e+02 2.537e+02 3.082e+02 2.053e+03, threshold=5.073e+02, percent-clipped=1.0 2023-05-02 16:52:49,815 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2674, 2.1229, 1.8160, 1.8219, 2.3246, 2.0788, 2.0523, 2.4128], device='cuda:6'), covar=tensor([0.0294, 0.0473, 0.0588, 0.0522, 0.0290, 0.0388, 0.0221, 0.0328], device='cuda:6'), in_proj_covar=tensor([0.0239, 0.0253, 0.0240, 0.0241, 0.0253, 0.0251, 0.0249, 0.0251], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 16:52:55,388 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286190.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:53:10,105 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7518, 3.9362, 2.6578, 4.6116, 3.1353, 4.4601, 2.7534, 3.2469], device='cuda:6'), covar=tensor([0.0382, 0.0447, 0.1733, 0.0255, 0.0908, 0.0599, 0.1482, 0.0812], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0184, 0.0200, 0.0178, 0.0183, 0.0225, 0.0209, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 16:53:13,092 INFO [train.py:904] (6/8) Epoch 29, batch 2000, loss[loss=0.1686, simple_loss=0.2611, pruned_loss=0.0381, over 16701.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2522, pruned_loss=0.03771, over 3305126.70 frames. ], batch size: 62, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:54:21,795 INFO [train.py:904] (6/8) Epoch 29, batch 2050, loss[loss=0.1771, simple_loss=0.2505, pruned_loss=0.05183, over 16784.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2523, pruned_loss=0.0377, over 3297674.32 frames. ], batch size: 124, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:54:54,806 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=286278.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:55:04,677 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.125e+02 2.436e+02 2.931e+02 6.185e+02, threshold=4.871e+02, percent-clipped=3.0 2023-05-02 16:55:05,365 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6035, 3.7261, 2.8552, 2.2364, 2.4037, 2.3996, 3.8308, 3.2344], device='cuda:6'), covar=tensor([0.2942, 0.0611, 0.1825, 0.3190, 0.2923, 0.2261, 0.0574, 0.1649], device='cuda:6'), in_proj_covar=tensor([0.0341, 0.0278, 0.0318, 0.0331, 0.0310, 0.0283, 0.0309, 0.0358], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 16:55:18,556 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286295.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:55:30,407 INFO [train.py:904] (6/8) Epoch 29, batch 2100, loss[loss=0.1847, simple_loss=0.2658, pruned_loss=0.05178, over 16879.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.253, pruned_loss=0.03826, over 3305265.83 frames. ], batch size: 116, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:55:54,403 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-02 16:56:00,353 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=286326.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:56:26,231 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286344.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:56:40,455 INFO [train.py:904] (6/8) Epoch 29, batch 2150, loss[loss=0.1504, simple_loss=0.25, pruned_loss=0.02539, over 17107.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2526, pruned_loss=0.03796, over 3321998.65 frames. ], batch size: 48, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:56:44,723 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286356.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:57:24,981 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.220e+02 2.758e+02 3.117e+02 6.077e+02, threshold=5.516e+02, percent-clipped=2.0 2023-05-02 16:57:50,675 INFO [train.py:904] (6/8) Epoch 29, batch 2200, loss[loss=0.1444, simple_loss=0.2396, pruned_loss=0.0246, over 16870.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2526, pruned_loss=0.03822, over 3327354.01 frames. ], batch size: 42, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:57:52,961 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286405.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:58:59,531 INFO [train.py:904] (6/8) Epoch 29, batch 2250, loss[loss=0.1364, simple_loss=0.2211, pruned_loss=0.02589, over 16832.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.253, pruned_loss=0.03842, over 3327211.28 frames. ], batch size: 39, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:59:07,419 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-05-02 16:59:34,557 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=286479.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 16:59:43,710 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=286485.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:59:45,654 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.691e+02 2.269e+02 2.585e+02 3.162e+02 6.397e+02, threshold=5.169e+02, percent-clipped=1.0 2023-05-02 16:59:48,948 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5248, 4.5609, 4.9218, 4.9032, 4.9534, 4.6127, 4.6365, 4.4822], device='cuda:6'), covar=tensor([0.0418, 0.0773, 0.0436, 0.0437, 0.0495, 0.0482, 0.0891, 0.0642], device='cuda:6'), in_proj_covar=tensor([0.0450, 0.0509, 0.0491, 0.0452, 0.0539, 0.0514, 0.0591, 0.0415], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 17:00:08,091 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7534, 4.3222, 2.9462, 2.3669, 2.5734, 2.5146, 4.6940, 3.4663], device='cuda:6'), covar=tensor([0.3283, 0.0554, 0.2126, 0.3223, 0.3374, 0.2432, 0.0381, 0.1645], device='cuda:6'), in_proj_covar=tensor([0.0340, 0.0278, 0.0318, 0.0331, 0.0309, 0.0283, 0.0309, 0.0358], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 17:00:08,670 INFO [train.py:904] (6/8) Epoch 29, batch 2300, loss[loss=0.1536, simple_loss=0.2479, pruned_loss=0.02965, over 17254.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2535, pruned_loss=0.03885, over 3321516.62 frames. ], batch size: 52, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 17:00:12,300 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9269, 2.9343, 3.2908, 2.1207, 2.8571, 2.3039, 3.4196, 3.3517], device='cuda:6'), covar=tensor([0.0260, 0.1103, 0.0631, 0.2079, 0.0919, 0.1043, 0.0583, 0.0942], device='cuda:6'), in_proj_covar=tensor([0.0163, 0.0173, 0.0172, 0.0160, 0.0150, 0.0134, 0.0148, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 17:00:28,084 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5833, 4.6493, 4.9254, 4.9205, 4.9856, 4.6830, 4.6891, 4.5236], device='cuda:6'), covar=tensor([0.0489, 0.0998, 0.0591, 0.0610, 0.0593, 0.0603, 0.0953, 0.0730], device='cuda:6'), in_proj_covar=tensor([0.0448, 0.0506, 0.0490, 0.0450, 0.0537, 0.0513, 0.0589, 0.0413], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 17:00:51,921 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9582, 2.8296, 2.7571, 4.2747, 3.5535, 4.0940, 1.7648, 3.0555], device='cuda:6'), covar=tensor([0.1373, 0.0690, 0.1105, 0.0171, 0.0143, 0.0372, 0.1534, 0.0832], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0181, 0.0201, 0.0206, 0.0207, 0.0221, 0.0211, 0.0200], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 17:01:13,726 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3083, 5.9041, 6.0376, 5.6529, 5.8082, 6.3711, 5.9276, 5.6332], device='cuda:6'), covar=tensor([0.0912, 0.1944, 0.2435, 0.1942, 0.2461, 0.0895, 0.1476, 0.2269], device='cuda:6'), in_proj_covar=tensor([0.0440, 0.0654, 0.0728, 0.0532, 0.0714, 0.0747, 0.0562, 0.0710], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 17:01:17,608 INFO [train.py:904] (6/8) Epoch 29, batch 2350, loss[loss=0.1839, simple_loss=0.2704, pruned_loss=0.04874, over 16119.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2535, pruned_loss=0.03944, over 3320948.88 frames. ], batch size: 164, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 17:01:22,272 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0651, 4.9859, 4.9142, 4.4730, 4.6462, 4.9605, 4.9021, 4.6090], device='cuda:6'), covar=tensor([0.0593, 0.0685, 0.0336, 0.0384, 0.0977, 0.0535, 0.0456, 0.0787], device='cuda:6'), in_proj_covar=tensor([0.0324, 0.0489, 0.0380, 0.0382, 0.0375, 0.0438, 0.0260, 0.0454], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 17:02:03,068 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.083e+02 2.366e+02 2.716e+02 6.133e+02, threshold=4.732e+02, percent-clipped=1.0 2023-05-02 17:02:04,877 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286588.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:02:27,440 INFO [train.py:904] (6/8) Epoch 29, batch 2400, loss[loss=0.1508, simple_loss=0.2438, pruned_loss=0.02886, over 17139.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2541, pruned_loss=0.0391, over 3325464.67 frames. ], batch size: 46, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 17:03:30,962 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286649.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:03:31,977 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2208, 4.2553, 4.5723, 4.5575, 4.6078, 4.3160, 4.3388, 4.2326], device='cuda:6'), covar=tensor([0.0398, 0.0623, 0.0385, 0.0413, 0.0515, 0.0470, 0.0835, 0.0640], device='cuda:6'), in_proj_covar=tensor([0.0449, 0.0509, 0.0491, 0.0452, 0.0539, 0.0515, 0.0592, 0.0415], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 17:03:34,190 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=286651.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:03:37,416 INFO [train.py:904] (6/8) Epoch 29, batch 2450, loss[loss=0.1588, simple_loss=0.2611, pruned_loss=0.0283, over 17227.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2549, pruned_loss=0.03875, over 3326905.56 frames. ], batch size: 52, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 17:04:00,712 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286671.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:04:23,750 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.201e+02 2.590e+02 3.137e+02 8.541e+02, threshold=5.179e+02, percent-clipped=2.0 2023-05-02 17:04:41,150 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=286700.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:04:46,152 INFO [train.py:904] (6/8) Epoch 29, batch 2500, loss[loss=0.1661, simple_loss=0.2372, pruned_loss=0.04752, over 16746.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2552, pruned_loss=0.03838, over 3325415.55 frames. ], batch size: 124, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 17:05:16,508 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.94 vs. limit=5.0 2023-05-02 17:05:25,018 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286732.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:05:38,128 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9192, 2.9030, 2.6179, 2.8440, 3.1746, 3.0408, 3.4405, 3.3890], device='cuda:6'), covar=tensor([0.0172, 0.0518, 0.0559, 0.0522, 0.0359, 0.0420, 0.0291, 0.0319], device='cuda:6'), in_proj_covar=tensor([0.0241, 0.0254, 0.0242, 0.0242, 0.0254, 0.0253, 0.0250, 0.0252], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 17:05:51,647 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-02 17:05:55,678 INFO [train.py:904] (6/8) Epoch 29, batch 2550, loss[loss=0.2228, simple_loss=0.2989, pruned_loss=0.07339, over 12428.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2559, pruned_loss=0.0392, over 3321548.72 frames. ], batch size: 246, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 17:05:58,942 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.54 vs. limit=2.0 2023-05-02 17:06:19,635 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286770.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:06:32,974 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=286779.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:06:41,031 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=286785.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:06:44,659 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.642e+02 2.042e+02 2.418e+02 2.906e+02 6.403e+02, threshold=4.836e+02, percent-clipped=2.0 2023-05-02 17:07:07,734 INFO [train.py:904] (6/8) Epoch 29, batch 2600, loss[loss=0.1683, simple_loss=0.2621, pruned_loss=0.03726, over 16083.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2552, pruned_loss=0.03851, over 3325010.82 frames. ], batch size: 35, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:07:39,059 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=286827.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:07:40,545 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.91 vs. limit=5.0 2023-05-02 17:07:45,108 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286831.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:07:47,350 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=286833.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:08:15,898 INFO [train.py:904] (6/8) Epoch 29, batch 2650, loss[loss=0.1818, simple_loss=0.2753, pruned_loss=0.04419, over 16510.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2559, pruned_loss=0.0383, over 3325225.85 frames. ], batch size: 68, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:08:30,086 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2023-05-02 17:09:00,292 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.057e+02 2.433e+02 2.859e+02 4.818e+02, threshold=4.867e+02, percent-clipped=0.0 2023-05-02 17:09:22,332 INFO [train.py:904] (6/8) Epoch 29, batch 2700, loss[loss=0.1612, simple_loss=0.2442, pruned_loss=0.03906, over 16676.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2555, pruned_loss=0.03767, over 3325254.13 frames. ], batch size: 124, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:09:51,872 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0764, 4.6352, 4.5045, 3.3935, 3.7518, 4.5030, 3.9774, 2.7783], device='cuda:6'), covar=tensor([0.0460, 0.0073, 0.0054, 0.0340, 0.0159, 0.0104, 0.0110, 0.0464], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0091, 0.0093, 0.0137, 0.0104, 0.0117, 0.0100, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 17:10:17,479 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=286944.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:10:26,950 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=286951.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:10:30,031 INFO [train.py:904] (6/8) Epoch 29, batch 2750, loss[loss=0.1692, simple_loss=0.2692, pruned_loss=0.03457, over 16850.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2558, pruned_loss=0.03744, over 3329081.51 frames. ], batch size: 42, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:11:17,992 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 1.930e+02 2.341e+02 2.772e+02 4.818e+02, threshold=4.681e+02, percent-clipped=0.0 2023-05-02 17:11:32,922 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=286999.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:11:35,334 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=287000.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:11:40,740 INFO [train.py:904] (6/8) Epoch 29, batch 2800, loss[loss=0.1894, simple_loss=0.2653, pruned_loss=0.05673, over 16770.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2551, pruned_loss=0.03749, over 3321139.76 frames. ], batch size: 124, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:12:02,221 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2061, 2.4092, 2.4488, 4.0159, 2.3015, 2.7201, 2.4388, 2.5375], device='cuda:6'), covar=tensor([0.1550, 0.3854, 0.3195, 0.0703, 0.4278, 0.2736, 0.3805, 0.3591], device='cuda:6'), in_proj_covar=tensor([0.0427, 0.0482, 0.0393, 0.0342, 0.0450, 0.0553, 0.0455, 0.0565], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 17:12:12,992 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=287027.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:12:42,707 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=287048.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:12:48,471 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-02 17:12:50,053 INFO [train.py:904] (6/8) Epoch 29, batch 2850, loss[loss=0.1682, simple_loss=0.2605, pruned_loss=0.03794, over 17086.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2544, pruned_loss=0.03786, over 3315866.00 frames. ], batch size: 47, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:13:39,986 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.623e+02 2.226e+02 2.685e+02 3.371e+02 7.491e+02, threshold=5.370e+02, percent-clipped=5.0 2023-05-02 17:14:00,571 INFO [train.py:904] (6/8) Epoch 29, batch 2900, loss[loss=0.1584, simple_loss=0.256, pruned_loss=0.03036, over 17125.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2534, pruned_loss=0.03812, over 3326228.23 frames. ], batch size: 49, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:14:31,455 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=287126.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:15:10,985 INFO [train.py:904] (6/8) Epoch 29, batch 2950, loss[loss=0.138, simple_loss=0.222, pruned_loss=0.02703, over 15910.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2532, pruned_loss=0.03846, over 3326738.62 frames. ], batch size: 35, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:15:34,376 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0968, 4.6312, 3.2890, 2.6562, 3.0157, 2.8807, 4.9150, 3.7857], device='cuda:6'), covar=tensor([0.2779, 0.0557, 0.1820, 0.2998, 0.2963, 0.2176, 0.0375, 0.1530], device='cuda:6'), in_proj_covar=tensor([0.0339, 0.0278, 0.0317, 0.0331, 0.0309, 0.0282, 0.0309, 0.0358], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 17:15:59,393 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.092e+02 2.416e+02 3.171e+02 5.496e+02, threshold=4.831e+02, percent-clipped=1.0 2023-05-02 17:16:20,173 INFO [train.py:904] (6/8) Epoch 29, batch 3000, loss[loss=0.1425, simple_loss=0.2334, pruned_loss=0.02577, over 17159.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2532, pruned_loss=0.03882, over 3329654.90 frames. ], batch size: 46, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:16:20,173 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 17:16:28,749 INFO [train.py:938] (6/8) Epoch 29, validation: loss=0.1336, simple_loss=0.2385, pruned_loss=0.01438, over 944034.00 frames. 2023-05-02 17:16:28,750 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 17:16:35,463 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4400, 4.2522, 4.4904, 4.6515, 4.7558, 4.3745, 4.6301, 4.7679], device='cuda:6'), covar=tensor([0.1914, 0.1501, 0.1544, 0.0781, 0.0671, 0.1107, 0.2065, 0.1074], device='cuda:6'), in_proj_covar=tensor([0.0709, 0.0870, 0.1003, 0.0883, 0.0673, 0.0700, 0.0737, 0.0855], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 17:17:12,450 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8017, 4.7838, 4.6375, 4.0571, 4.7388, 1.9344, 4.4683, 4.2691], device='cuda:6'), covar=tensor([0.0198, 0.0140, 0.0235, 0.0378, 0.0146, 0.2832, 0.0160, 0.0294], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0179, 0.0217, 0.0188, 0.0195, 0.0221, 0.0205, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 17:17:12,708 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 17:17:22,343 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.82 vs. limit=5.0 2023-05-02 17:17:25,957 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=287244.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:17:30,519 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8693, 2.9251, 2.6855, 5.1207, 4.1895, 4.3454, 1.7399, 3.2051], device='cuda:6'), covar=tensor([0.1431, 0.0836, 0.1352, 0.0211, 0.0226, 0.0460, 0.1709, 0.0848], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0182, 0.0201, 0.0208, 0.0207, 0.0221, 0.0212, 0.0200], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 17:17:40,005 INFO [train.py:904] (6/8) Epoch 29, batch 3050, loss[loss=0.1717, simple_loss=0.2519, pruned_loss=0.04575, over 16817.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2528, pruned_loss=0.03862, over 3330946.90 frames. ], batch size: 116, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:17:43,540 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6540, 3.7458, 2.8208, 2.2398, 2.3353, 2.3925, 3.8521, 3.1988], device='cuda:6'), covar=tensor([0.2872, 0.0580, 0.1899, 0.3265, 0.2974, 0.2315, 0.0550, 0.1740], device='cuda:6'), in_proj_covar=tensor([0.0337, 0.0276, 0.0315, 0.0330, 0.0308, 0.0281, 0.0307, 0.0357], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 17:17:47,902 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4735, 4.3223, 4.5138, 4.6637, 4.7378, 4.3093, 4.5814, 4.7554], device='cuda:6'), covar=tensor([0.1667, 0.1266, 0.1294, 0.0679, 0.0674, 0.1240, 0.2215, 0.0879], device='cuda:6'), in_proj_covar=tensor([0.0709, 0.0870, 0.1002, 0.0882, 0.0672, 0.0699, 0.0736, 0.0854], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 17:18:29,494 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.177e+02 2.556e+02 3.004e+02 5.914e+02, threshold=5.112e+02, percent-clipped=3.0 2023-05-02 17:18:34,402 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=287292.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:18:49,718 INFO [train.py:904] (6/8) Epoch 29, batch 3100, loss[loss=0.1746, simple_loss=0.2682, pruned_loss=0.04048, over 17068.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2527, pruned_loss=0.03897, over 3327145.83 frames. ], batch size: 55, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:19:22,319 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=287327.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:20:00,321 INFO [train.py:904] (6/8) Epoch 29, batch 3150, loss[loss=0.1686, simple_loss=0.244, pruned_loss=0.04666, over 16867.00 frames. ], tot_loss[loss=0.165, simple_loss=0.252, pruned_loss=0.03903, over 3321796.89 frames. ], batch size: 102, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:20:30,290 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=287375.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:20:49,321 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.202e+02 2.547e+02 3.057e+02 5.331e+02, threshold=5.094e+02, percent-clipped=1.0 2023-05-02 17:21:10,100 INFO [train.py:904] (6/8) Epoch 29, batch 3200, loss[loss=0.1466, simple_loss=0.2322, pruned_loss=0.03047, over 17008.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.2512, pruned_loss=0.03813, over 3329614.47 frames. ], batch size: 41, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:21:20,067 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=287411.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:21:40,297 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=287426.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:22:06,625 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7727, 3.4292, 3.8445, 2.1588, 3.9016, 3.9175, 3.2014, 2.9263], device='cuda:6'), covar=tensor([0.0734, 0.0271, 0.0183, 0.1058, 0.0121, 0.0205, 0.0424, 0.0468], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0112, 0.0103, 0.0139, 0.0087, 0.0133, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 17:22:18,321 INFO [train.py:904] (6/8) Epoch 29, batch 3250, loss[loss=0.1543, simple_loss=0.2432, pruned_loss=0.03265, over 17232.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.251, pruned_loss=0.03829, over 3334151.90 frames. ], batch size: 44, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:22:23,011 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7440, 3.5102, 3.9006, 2.0395, 4.0005, 4.0393, 3.1901, 3.0259], device='cuda:6'), covar=tensor([0.0818, 0.0293, 0.0216, 0.1291, 0.0136, 0.0239, 0.0443, 0.0487], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0112, 0.0103, 0.0139, 0.0087, 0.0133, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 17:22:43,532 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=287472.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:22:45,745 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=287474.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:23:05,921 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.139e+02 2.558e+02 3.030e+02 4.692e+02, threshold=5.116e+02, percent-clipped=0.0 2023-05-02 17:23:26,856 INFO [train.py:904] (6/8) Epoch 29, batch 3300, loss[loss=0.1633, simple_loss=0.2571, pruned_loss=0.03477, over 16745.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2514, pruned_loss=0.0382, over 3325117.42 frames. ], batch size: 83, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:24:11,408 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=287536.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:24:16,905 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9276, 2.9063, 2.5218, 4.5897, 3.6171, 4.1448, 1.7068, 3.0806], device='cuda:6'), covar=tensor([0.1405, 0.0774, 0.1345, 0.0201, 0.0261, 0.0472, 0.1732, 0.0887], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0181, 0.0201, 0.0208, 0.0208, 0.0221, 0.0211, 0.0200], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 17:24:34,315 INFO [train.py:904] (6/8) Epoch 29, batch 3350, loss[loss=0.1616, simple_loss=0.2458, pruned_loss=0.03875, over 16827.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2518, pruned_loss=0.03846, over 3324729.15 frames. ], batch size: 102, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:24:46,522 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=287562.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:25:14,562 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=287583.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:25:21,672 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 2.115e+02 2.472e+02 3.004e+02 5.841e+02, threshold=4.943e+02, percent-clipped=1.0 2023-05-02 17:25:33,268 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1363, 5.0768, 4.8175, 4.1176, 4.9227, 1.7488, 4.6219, 4.5925], device='cuda:6'), covar=tensor([0.0111, 0.0105, 0.0285, 0.0518, 0.0130, 0.3344, 0.0180, 0.0324], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0180, 0.0219, 0.0190, 0.0197, 0.0223, 0.0207, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 17:25:33,304 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=287597.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:25:43,306 INFO [train.py:904] (6/8) Epoch 29, batch 3400, loss[loss=0.1485, simple_loss=0.24, pruned_loss=0.02847, over 17214.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.252, pruned_loss=0.03836, over 3328564.22 frames. ], batch size: 45, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:26:10,412 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=287623.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:26:40,295 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=287644.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:26:53,539 INFO [train.py:904] (6/8) Epoch 29, batch 3450, loss[loss=0.147, simple_loss=0.2291, pruned_loss=0.03247, over 16197.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2505, pruned_loss=0.03767, over 3323063.46 frames. ], batch size: 36, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:27:43,293 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.405e+02 1.986e+02 2.391e+02 2.898e+02 4.066e+02, threshold=4.783e+02, percent-clipped=0.0 2023-05-02 17:28:01,555 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5314, 5.9795, 5.7162, 5.7906, 5.3700, 5.4573, 5.3618, 6.0915], device='cuda:6'), covar=tensor([0.1661, 0.0956, 0.1020, 0.0871, 0.0981, 0.0699, 0.1377, 0.0923], device='cuda:6'), in_proj_covar=tensor([0.0745, 0.0898, 0.0736, 0.0697, 0.0570, 0.0568, 0.0754, 0.0705], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 17:28:04,222 INFO [train.py:904] (6/8) Epoch 29, batch 3500, loss[loss=0.1661, simple_loss=0.2461, pruned_loss=0.04301, over 16832.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2494, pruned_loss=0.03724, over 3332725.76 frames. ], batch size: 116, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:28:26,761 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=287720.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:28:39,632 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7284, 3.7704, 2.4389, 4.0590, 3.1082, 3.9984, 2.6141, 3.1721], device='cuda:6'), covar=tensor([0.0288, 0.0456, 0.1545, 0.0441, 0.0771, 0.0870, 0.1388, 0.0671], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0185, 0.0199, 0.0180, 0.0183, 0.0227, 0.0208, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 17:29:14,561 INFO [train.py:904] (6/8) Epoch 29, batch 3550, loss[loss=0.1477, simple_loss=0.2433, pruned_loss=0.02608, over 17140.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2494, pruned_loss=0.03782, over 3299860.63 frames. ], batch size: 48, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:29:33,153 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=287767.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:29:42,493 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=287773.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:29:53,995 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=287781.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:30:07,185 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 2.118e+02 2.537e+02 3.010e+02 5.172e+02, threshold=5.074e+02, percent-clipped=1.0 2023-05-02 17:30:26,881 INFO [train.py:904] (6/8) Epoch 29, batch 3600, loss[loss=0.1684, simple_loss=0.2429, pruned_loss=0.04693, over 16717.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.2476, pruned_loss=0.03736, over 3305495.40 frames. ], batch size: 124, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:31:11,619 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=287834.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:31:40,143 INFO [train.py:904] (6/8) Epoch 29, batch 3650, loss[loss=0.1627, simple_loss=0.2603, pruned_loss=0.03258, over 16636.00 frames. ], tot_loss[loss=0.161, simple_loss=0.2472, pruned_loss=0.03738, over 3306939.41 frames. ], batch size: 57, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:32:35,499 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.746e+02 2.232e+02 2.562e+02 3.214e+02 7.635e+02, threshold=5.123e+02, percent-clipped=1.0 2023-05-02 17:32:38,219 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=287892.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:32:57,081 INFO [train.py:904] (6/8) Epoch 29, batch 3700, loss[loss=0.1655, simple_loss=0.2394, pruned_loss=0.04579, over 16796.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2459, pruned_loss=0.03896, over 3305511.28 frames. ], batch size: 83, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:33:12,281 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.94 vs. limit=5.0 2023-05-02 17:33:18,789 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=287918.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:33:48,797 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=287939.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:34:09,992 INFO [train.py:904] (6/8) Epoch 29, batch 3750, loss[loss=0.1615, simple_loss=0.2365, pruned_loss=0.0432, over 16770.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2465, pruned_loss=0.04033, over 3283920.17 frames. ], batch size: 102, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:35:05,545 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.573e+02 2.186e+02 2.475e+02 3.056e+02 8.739e+02, threshold=4.949e+02, percent-clipped=4.0 2023-05-02 17:35:29,279 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288003.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:35:30,029 INFO [train.py:904] (6/8) Epoch 29, batch 3800, loss[loss=0.1955, simple_loss=0.2796, pruned_loss=0.05572, over 12180.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2488, pruned_loss=0.04164, over 3271166.58 frames. ], batch size: 246, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:36:40,375 INFO [train.py:904] (6/8) Epoch 29, batch 3850, loss[loss=0.1773, simple_loss=0.254, pruned_loss=0.05027, over 16502.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2489, pruned_loss=0.04214, over 3276212.12 frames. ], batch size: 146, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:36:55,423 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288064.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:36:59,712 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288067.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:37:05,480 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6079, 3.6607, 2.3923, 3.9297, 2.9971, 3.8857, 2.4526, 3.0219], device='cuda:6'), covar=tensor([0.0309, 0.0470, 0.1584, 0.0305, 0.0760, 0.0775, 0.1489, 0.0731], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0185, 0.0199, 0.0180, 0.0183, 0.0227, 0.0208, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 17:37:06,653 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8153, 4.0172, 2.6738, 4.7194, 3.2514, 4.7196, 2.9071, 3.3384], device='cuda:6'), covar=tensor([0.0328, 0.0322, 0.1604, 0.0112, 0.0778, 0.0285, 0.1352, 0.0742], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0185, 0.0199, 0.0180, 0.0183, 0.0227, 0.0208, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 17:37:12,436 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288076.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 17:37:30,811 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.131e+02 2.477e+02 2.774e+02 5.510e+02, threshold=4.954e+02, percent-clipped=1.0 2023-05-02 17:37:50,142 INFO [train.py:904] (6/8) Epoch 29, batch 3900, loss[loss=0.182, simple_loss=0.2527, pruned_loss=0.05565, over 16902.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2485, pruned_loss=0.04254, over 3285554.76 frames. ], batch size: 109, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:37:53,652 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288106.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:38:06,533 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288115.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:38:27,309 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288129.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:38:36,214 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2023-05-02 17:39:00,673 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288152.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 17:39:03,100 INFO [train.py:904] (6/8) Epoch 29, batch 3950, loss[loss=0.1571, simple_loss=0.2481, pruned_loss=0.03309, over 17163.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2486, pruned_loss=0.04323, over 3283830.38 frames. ], batch size: 49, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:39:22,331 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288167.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 17:39:55,368 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.270e+02 2.585e+02 3.079e+02 6.589e+02, threshold=5.170e+02, percent-clipped=2.0 2023-05-02 17:39:58,721 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288192.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:40:16,237 INFO [train.py:904] (6/8) Epoch 29, batch 4000, loss[loss=0.1757, simple_loss=0.2561, pruned_loss=0.04764, over 16624.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2486, pruned_loss=0.04357, over 3282677.85 frames. ], batch size: 134, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:40:30,242 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288213.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 17:40:37,374 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288218.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:41:07,473 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288239.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:41:09,028 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288240.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:41:23,184 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7767, 2.8929, 2.4631, 2.7351, 3.2026, 2.8731, 3.2198, 3.4023], device='cuda:6'), covar=tensor([0.0102, 0.0423, 0.0550, 0.0447, 0.0298, 0.0411, 0.0215, 0.0273], device='cuda:6'), in_proj_covar=tensor([0.0240, 0.0251, 0.0239, 0.0240, 0.0252, 0.0250, 0.0248, 0.0250], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 17:41:25,545 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3233, 5.4445, 5.7302, 5.6851, 5.7807, 5.4469, 5.3827, 5.1167], device='cuda:6'), covar=tensor([0.0306, 0.0442, 0.0361, 0.0437, 0.0536, 0.0368, 0.0949, 0.0461], device='cuda:6'), in_proj_covar=tensor([0.0453, 0.0512, 0.0494, 0.0454, 0.0542, 0.0521, 0.0599, 0.0419], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 17:41:29,240 INFO [train.py:904] (6/8) Epoch 29, batch 4050, loss[loss=0.1783, simple_loss=0.2602, pruned_loss=0.04819, over 16739.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2494, pruned_loss=0.04297, over 3284351.19 frames. ], batch size: 124, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:41:47,213 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288266.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:42:17,241 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288287.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:42:21,065 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 1.858e+02 2.084e+02 2.415e+02 5.689e+02, threshold=4.167e+02, percent-clipped=1.0 2023-05-02 17:42:42,061 INFO [train.py:904] (6/8) Epoch 29, batch 4100, loss[loss=0.187, simple_loss=0.2707, pruned_loss=0.05167, over 16593.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2509, pruned_loss=0.04236, over 3268429.61 frames. ], batch size: 62, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:43:17,878 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288327.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:43:25,749 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-05-02 17:43:45,801 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288345.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:43:59,218 INFO [train.py:904] (6/8) Epoch 29, batch 4150, loss[loss=0.189, simple_loss=0.2821, pruned_loss=0.04792, over 16709.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2578, pruned_loss=0.04477, over 3209130.46 frames. ], batch size: 76, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:44:07,903 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288359.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:44:33,864 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288376.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:44:51,056 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288388.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:44:52,909 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.295e+02 2.546e+02 3.002e+02 5.285e+02, threshold=5.091e+02, percent-clipped=5.0 2023-05-02 17:45:02,562 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288396.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:45:13,928 INFO [train.py:904] (6/8) Epoch 29, batch 4200, loss[loss=0.1924, simple_loss=0.2853, pruned_loss=0.04971, over 16558.00 frames. ], tot_loss[loss=0.179, simple_loss=0.265, pruned_loss=0.04651, over 3176983.18 frames. ], batch size: 68, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:45:18,725 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288406.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:45:45,980 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288424.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:45:53,534 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288429.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:46:30,280 INFO [train.py:904] (6/8) Epoch 29, batch 4250, loss[loss=0.1736, simple_loss=0.2763, pruned_loss=0.03547, over 16492.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2678, pruned_loss=0.04586, over 3170063.25 frames. ], batch size: 68, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:46:36,724 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288457.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:46:43,236 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288462.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 17:47:05,781 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288477.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:47:24,427 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.222e+02 2.492e+02 3.034e+02 5.623e+02, threshold=4.984e+02, percent-clipped=1.0 2023-05-02 17:47:42,867 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-02 17:47:45,985 INFO [train.py:904] (6/8) Epoch 29, batch 4300, loss[loss=0.1894, simple_loss=0.2879, pruned_loss=0.0455, over 16194.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2695, pruned_loss=0.04521, over 3185952.12 frames. ], batch size: 165, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:47:51,166 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288508.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 17:48:59,342 INFO [train.py:904] (6/8) Epoch 29, batch 4350, loss[loss=0.22, simple_loss=0.2984, pruned_loss=0.07083, over 11931.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2726, pruned_loss=0.04637, over 3167163.66 frames. ], batch size: 250, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:49:53,411 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.193e+02 2.447e+02 2.913e+02 5.150e+02, threshold=4.894e+02, percent-clipped=1.0 2023-05-02 17:50:13,839 INFO [train.py:904] (6/8) Epoch 29, batch 4400, loss[loss=0.1926, simple_loss=0.2828, pruned_loss=0.0512, over 16767.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2748, pruned_loss=0.04808, over 3160822.00 frames. ], batch size: 83, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:50:22,414 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288609.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:51:07,899 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0617, 5.5611, 5.8242, 5.4719, 5.5472, 6.1058, 5.6013, 5.2432], device='cuda:6'), covar=tensor([0.0941, 0.1817, 0.1947, 0.1739, 0.2292, 0.0794, 0.1323, 0.2134], device='cuda:6'), in_proj_covar=tensor([0.0435, 0.0646, 0.0713, 0.0524, 0.0701, 0.0733, 0.0553, 0.0697], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 17:51:26,299 INFO [train.py:904] (6/8) Epoch 29, batch 4450, loss[loss=0.2236, simple_loss=0.3136, pruned_loss=0.06681, over 16886.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2791, pruned_loss=0.04954, over 3182792.49 frames. ], batch size: 116, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:51:29,007 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8236, 2.2049, 2.2781, 3.3664, 2.0833, 2.4533, 2.3041, 2.3132], device='cuda:6'), covar=tensor([0.1700, 0.3419, 0.3010, 0.0707, 0.4377, 0.2404, 0.3349, 0.3435], device='cuda:6'), in_proj_covar=tensor([0.0425, 0.0480, 0.0389, 0.0340, 0.0448, 0.0551, 0.0452, 0.0562], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 17:51:34,222 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288659.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:51:49,601 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8815, 4.6198, 4.3739, 3.2138, 3.8748, 4.4683, 3.8684, 2.6700], device='cuda:6'), covar=tensor([0.0495, 0.0026, 0.0057, 0.0368, 0.0108, 0.0084, 0.0096, 0.0436], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0090, 0.0092, 0.0136, 0.0103, 0.0117, 0.0099, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 17:51:49,634 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288670.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:52:08,447 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288683.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:52:18,785 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.398e+02 1.892e+02 2.239e+02 2.604e+02 5.228e+02, threshold=4.478e+02, percent-clipped=1.0 2023-05-02 17:52:34,492 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288701.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:52:37,594 INFO [train.py:904] (6/8) Epoch 29, batch 4500, loss[loss=0.1935, simple_loss=0.2834, pruned_loss=0.05187, over 17147.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2795, pruned_loss=0.05029, over 3197955.24 frames. ], batch size: 47, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:52:42,943 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288707.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:53:43,199 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9912, 3.9257, 4.0686, 4.1648, 4.2481, 3.8792, 4.2173, 4.2947], device='cuda:6'), covar=tensor([0.1496, 0.1058, 0.1146, 0.0596, 0.0511, 0.1501, 0.0795, 0.0617], device='cuda:6'), in_proj_covar=tensor([0.0695, 0.0847, 0.0976, 0.0862, 0.0656, 0.0680, 0.0717, 0.0832], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 17:53:49,129 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288752.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:53:51,862 INFO [train.py:904] (6/8) Epoch 29, batch 4550, loss[loss=0.1974, simple_loss=0.2877, pruned_loss=0.05353, over 17054.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.2798, pruned_loss=0.05096, over 3201382.77 frames. ], batch size: 41, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:54:03,026 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288762.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 17:54:44,107 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 1.747e+02 2.008e+02 2.421e+02 6.234e+02, threshold=4.017e+02, percent-clipped=2.0 2023-05-02 17:54:54,412 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288797.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:55:04,694 INFO [train.py:904] (6/8) Epoch 29, batch 4600, loss[loss=0.1902, simple_loss=0.275, pruned_loss=0.05271, over 16694.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2809, pruned_loss=0.05142, over 3208990.14 frames. ], batch size: 57, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:55:12,175 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288808.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:55:14,479 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288810.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:56:18,389 INFO [train.py:904] (6/8) Epoch 29, batch 4650, loss[loss=0.1775, simple_loss=0.2707, pruned_loss=0.04217, over 15458.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.2796, pruned_loss=0.05115, over 3211518.55 frames. ], batch size: 190, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:56:21,003 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288856.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:56:24,823 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288858.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:56:28,429 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8286, 4.8939, 5.1837, 5.1447, 5.2397, 4.8782, 4.8546, 4.5233], device='cuda:6'), covar=tensor([0.0289, 0.0412, 0.0292, 0.0344, 0.0415, 0.0329, 0.0944, 0.0538], device='cuda:6'), in_proj_covar=tensor([0.0437, 0.0495, 0.0478, 0.0437, 0.0525, 0.0502, 0.0581, 0.0406], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 17:57:00,674 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8633, 5.2118, 5.4069, 5.1653, 5.2332, 5.7939, 5.2716, 4.9507], device='cuda:6'), covar=tensor([0.1109, 0.1753, 0.2194, 0.1894, 0.2292, 0.0795, 0.1417, 0.2356], device='cuda:6'), in_proj_covar=tensor([0.0431, 0.0639, 0.0705, 0.0518, 0.0695, 0.0729, 0.0548, 0.0690], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 17:57:10,668 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.194e+02 1.769e+02 2.025e+02 2.417e+02 4.191e+02, threshold=4.050e+02, percent-clipped=1.0 2023-05-02 17:57:29,733 INFO [train.py:904] (6/8) Epoch 29, batch 4700, loss[loss=0.1602, simple_loss=0.2539, pruned_loss=0.03321, over 16757.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.2766, pruned_loss=0.05007, over 3208910.01 frames. ], batch size: 83, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:58:33,304 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-02 17:58:41,594 INFO [train.py:904] (6/8) Epoch 29, batch 4750, loss[loss=0.1786, simple_loss=0.2624, pruned_loss=0.04742, over 16764.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2718, pruned_loss=0.04756, over 3216743.96 frames. ], batch size: 39, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:58:57,733 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288965.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 17:59:23,546 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288983.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:59:35,274 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.313e+02 1.780e+02 1.987e+02 2.347e+02 4.131e+02, threshold=3.973e+02, percent-clipped=1.0 2023-05-02 17:59:50,949 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=289001.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:59:54,498 INFO [train.py:904] (6/8) Epoch 29, batch 4800, loss[loss=0.1874, simple_loss=0.2898, pruned_loss=0.04247, over 16724.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2682, pruned_loss=0.04528, over 3228391.38 frames. ], batch size: 124, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 18:00:37,065 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289031.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:00:52,584 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3361, 3.4078, 2.0816, 3.7922, 2.6364, 3.7462, 2.1999, 2.7412], device='cuda:6'), covar=tensor([0.0343, 0.0378, 0.1801, 0.0188, 0.0868, 0.0688, 0.1691, 0.0910], device='cuda:6'), in_proj_covar=tensor([0.0179, 0.0183, 0.0199, 0.0177, 0.0182, 0.0225, 0.0207, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 18:01:00,469 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=289046.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 18:01:05,207 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289049.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:01:09,493 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=289052.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:01:12,124 INFO [train.py:904] (6/8) Epoch 29, batch 4850, loss[loss=0.1708, simple_loss=0.2587, pruned_loss=0.0415, over 16862.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2689, pruned_loss=0.04434, over 3219857.83 frames. ], batch size: 42, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 18:02:08,678 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 1.829e+02 2.187e+02 2.631e+02 3.612e+02, threshold=4.373e+02, percent-clipped=0.0 2023-05-02 18:02:22,994 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289100.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:02:28,897 INFO [train.py:904] (6/8) Epoch 29, batch 4900, loss[loss=0.1648, simple_loss=0.2542, pruned_loss=0.03769, over 16750.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2682, pruned_loss=0.04313, over 3207830.05 frames. ], batch size: 89, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 18:02:33,586 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=289107.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 18:03:16,504 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=289136.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:03:36,343 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9815, 2.2743, 2.3177, 3.7080, 2.1675, 2.6150, 2.3106, 2.4161], device='cuda:6'), covar=tensor([0.1738, 0.3761, 0.3134, 0.0660, 0.4204, 0.2583, 0.3925, 0.3259], device='cuda:6'), in_proj_covar=tensor([0.0424, 0.0480, 0.0388, 0.0339, 0.0447, 0.0550, 0.0451, 0.0561], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:03:42,152 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=289153.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:03:43,041 INFO [train.py:904] (6/8) Epoch 29, batch 4950, loss[loss=0.1668, simple_loss=0.2599, pruned_loss=0.03686, over 16401.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2678, pruned_loss=0.04252, over 3207602.00 frames. ], batch size: 75, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 18:04:38,081 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.471e+02 2.040e+02 2.385e+02 2.853e+02 5.286e+02, threshold=4.771e+02, percent-clipped=2.0 2023-05-02 18:04:46,878 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=289197.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:04:56,967 INFO [train.py:904] (6/8) Epoch 29, batch 5000, loss[loss=0.1729, simple_loss=0.2736, pruned_loss=0.03607, over 16781.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2697, pruned_loss=0.04267, over 3218835.20 frames. ], batch size: 83, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 18:05:48,252 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2234, 4.3180, 4.1078, 3.8375, 3.8189, 4.1961, 3.9136, 3.9723], device='cuda:6'), covar=tensor([0.0600, 0.0460, 0.0307, 0.0287, 0.0793, 0.0527, 0.0848, 0.0530], device='cuda:6'), in_proj_covar=tensor([0.0313, 0.0475, 0.0368, 0.0370, 0.0364, 0.0425, 0.0253, 0.0440], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:06:00,747 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3023, 1.6061, 1.9957, 2.2084, 2.3773, 2.5943, 1.7298, 2.5454], device='cuda:6'), covar=tensor([0.0283, 0.0643, 0.0387, 0.0460, 0.0390, 0.0272, 0.0704, 0.0181], device='cuda:6'), in_proj_covar=tensor([0.0200, 0.0201, 0.0189, 0.0196, 0.0213, 0.0169, 0.0206, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 18:06:09,973 INFO [train.py:904] (6/8) Epoch 29, batch 5050, loss[loss=0.1956, simple_loss=0.2836, pruned_loss=0.05386, over 12219.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2704, pruned_loss=0.0427, over 3202769.94 frames. ], batch size: 247, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 18:06:26,922 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=289265.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:07:03,092 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 1.941e+02 2.202e+02 2.540e+02 3.761e+02, threshold=4.404e+02, percent-clipped=0.0 2023-05-02 18:07:22,109 INFO [train.py:904] (6/8) Epoch 29, batch 5100, loss[loss=0.1804, simple_loss=0.2672, pruned_loss=0.04677, over 12074.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2689, pruned_loss=0.0424, over 3202560.11 frames. ], batch size: 246, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:07:34,913 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289313.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:08:06,081 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.5488, 2.3825, 2.4396, 3.7325, 2.5621, 3.7216, 1.4548, 2.8849], device='cuda:6'), covar=tensor([0.1441, 0.0872, 0.1228, 0.0139, 0.0145, 0.0318, 0.1725, 0.0786], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0180, 0.0199, 0.0203, 0.0205, 0.0217, 0.0209, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 18:08:36,361 INFO [train.py:904] (6/8) Epoch 29, batch 5150, loss[loss=0.1666, simple_loss=0.2589, pruned_loss=0.0372, over 16666.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2695, pruned_loss=0.04188, over 3207084.14 frames. ], batch size: 62, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:08:57,717 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-02 18:09:04,568 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0960, 2.3366, 2.2262, 3.8185, 2.1788, 2.6489, 2.3505, 2.4739], device='cuda:6'), covar=tensor([0.1560, 0.3668, 0.3270, 0.0576, 0.4010, 0.2590, 0.3920, 0.3050], device='cuda:6'), in_proj_covar=tensor([0.0424, 0.0480, 0.0389, 0.0340, 0.0448, 0.0551, 0.0452, 0.0561], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:09:29,038 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 1.894e+02 2.237e+02 2.654e+02 4.017e+02, threshold=4.473e+02, percent-clipped=0.0 2023-05-02 18:09:44,622 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1259, 4.1710, 4.4552, 4.4172, 4.4143, 4.2016, 4.1563, 4.1617], device='cuda:6'), covar=tensor([0.0350, 0.0625, 0.0377, 0.0406, 0.0470, 0.0403, 0.0883, 0.0507], device='cuda:6'), in_proj_covar=tensor([0.0437, 0.0494, 0.0479, 0.0438, 0.0525, 0.0502, 0.0582, 0.0406], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 18:09:45,672 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=289402.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 18:09:47,652 INFO [train.py:904] (6/8) Epoch 29, batch 5200, loss[loss=0.1576, simple_loss=0.2502, pruned_loss=0.03252, over 16820.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2678, pruned_loss=0.04118, over 3221043.67 frames. ], batch size: 83, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:10:20,765 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0232, 1.9079, 2.5667, 3.0118, 2.9284, 3.3627, 2.0381, 3.3897], device='cuda:6'), covar=tensor([0.0217, 0.0662, 0.0385, 0.0274, 0.0325, 0.0202, 0.0788, 0.0134], device='cuda:6'), in_proj_covar=tensor([0.0200, 0.0201, 0.0189, 0.0196, 0.0213, 0.0169, 0.0206, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 18:10:27,462 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-02 18:10:59,664 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=289453.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:11:00,510 INFO [train.py:904] (6/8) Epoch 29, batch 5250, loss[loss=0.1689, simple_loss=0.2584, pruned_loss=0.03972, over 16923.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2647, pruned_loss=0.04065, over 3229958.30 frames. ], batch size: 109, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:11:55,405 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 18:11:55,944 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.356e+02 1.894e+02 2.157e+02 2.543e+02 4.571e+02, threshold=4.315e+02, percent-clipped=1.0 2023-05-02 18:11:58,117 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=289492.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:12:10,717 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289501.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:12:15,015 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=289503.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:12:15,675 INFO [train.py:904] (6/8) Epoch 29, batch 5300, loss[loss=0.1435, simple_loss=0.2308, pruned_loss=0.02811, over 16615.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2614, pruned_loss=0.03954, over 3227194.01 frames. ], batch size: 57, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:12:31,680 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4705, 3.7719, 3.9818, 1.9378, 4.3705, 4.2856, 2.9822, 3.0505], device='cuda:6'), covar=tensor([0.1105, 0.0208, 0.0235, 0.1372, 0.0064, 0.0127, 0.0477, 0.0581], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0111, 0.0102, 0.0139, 0.0087, 0.0131, 0.0130, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 18:12:43,395 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.90 vs. limit=5.0 2023-05-02 18:13:26,897 INFO [train.py:904] (6/8) Epoch 29, batch 5350, loss[loss=0.1826, simple_loss=0.2757, pruned_loss=0.04473, over 16800.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2605, pruned_loss=0.0392, over 3228632.06 frames. ], batch size: 76, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:13:42,048 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=289564.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:14:21,403 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.205e+02 1.853e+02 2.169e+02 2.585e+02 3.843e+02, threshold=4.338e+02, percent-clipped=0.0 2023-05-02 18:14:41,092 INFO [train.py:904] (6/8) Epoch 29, batch 5400, loss[loss=0.1848, simple_loss=0.2791, pruned_loss=0.04522, over 16901.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2632, pruned_loss=0.03988, over 3222425.39 frames. ], batch size: 109, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:15:57,431 INFO [train.py:904] (6/8) Epoch 29, batch 5450, loss[loss=0.2157, simple_loss=0.3018, pruned_loss=0.06476, over 16384.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2665, pruned_loss=0.04148, over 3212892.05 frames. ], batch size: 146, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:16:38,887 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8808, 2.7373, 2.8603, 2.2080, 2.7259, 2.2618, 2.7268, 2.9186], device='cuda:6'), covar=tensor([0.0290, 0.0898, 0.0499, 0.1803, 0.0821, 0.0885, 0.0548, 0.0750], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0172, 0.0172, 0.0158, 0.0149, 0.0133, 0.0147, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 18:16:53,635 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.432e+02 2.272e+02 2.915e+02 3.734e+02 7.781e+02, threshold=5.831e+02, percent-clipped=14.0 2023-05-02 18:17:12,127 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=289702.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 18:17:14,644 INFO [train.py:904] (6/8) Epoch 29, batch 5500, loss[loss=0.2224, simple_loss=0.3036, pruned_loss=0.07061, over 16474.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2732, pruned_loss=0.04551, over 3174271.27 frames. ], batch size: 68, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:17:20,430 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-02 18:17:51,871 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5197, 3.5174, 2.6357, 2.2964, 2.4001, 2.3347, 3.7101, 3.3159], device='cuda:6'), covar=tensor([0.2895, 0.0608, 0.1871, 0.2703, 0.2746, 0.2197, 0.0561, 0.1203], device='cuda:6'), in_proj_covar=tensor([0.0334, 0.0274, 0.0312, 0.0327, 0.0305, 0.0277, 0.0305, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 18:18:02,401 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1111, 2.4322, 2.5525, 2.0001, 2.7034, 2.7714, 2.4171, 2.3956], device='cuda:6'), covar=tensor([0.0672, 0.0271, 0.0244, 0.0908, 0.0138, 0.0297, 0.0433, 0.0447], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0111, 0.0102, 0.0138, 0.0087, 0.0132, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 18:18:26,454 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289750.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 18:18:31,706 INFO [train.py:904] (6/8) Epoch 29, batch 5550, loss[loss=0.2454, simple_loss=0.3145, pruned_loss=0.08815, over 11104.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.2803, pruned_loss=0.05094, over 3132468.90 frames. ], batch size: 248, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:19:30,745 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 2.953e+02 3.616e+02 4.343e+02 9.960e+02, threshold=7.231e+02, percent-clipped=12.0 2023-05-02 18:19:34,040 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=289792.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:19:53,194 INFO [train.py:904] (6/8) Epoch 29, batch 5600, loss[loss=0.2021, simple_loss=0.2879, pruned_loss=0.05812, over 16315.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2845, pruned_loss=0.05482, over 3079759.72 frames. ], batch size: 146, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:20:32,163 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1873, 5.2436, 5.0007, 4.2721, 5.1304, 1.9459, 4.9077, 4.8220], device='cuda:6'), covar=tensor([0.0104, 0.0093, 0.0217, 0.0470, 0.0104, 0.2974, 0.0127, 0.0243], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0179, 0.0219, 0.0190, 0.0195, 0.0222, 0.0206, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:20:54,481 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289840.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:21:16,879 INFO [train.py:904] (6/8) Epoch 29, batch 5650, loss[loss=0.2388, simple_loss=0.3158, pruned_loss=0.08086, over 15390.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2892, pruned_loss=0.05852, over 3053158.33 frames. ], batch size: 191, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:21:18,047 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2856, 3.1236, 3.4902, 1.8417, 3.6380, 3.6364, 2.8115, 2.7658], device='cuda:6'), covar=tensor([0.0891, 0.0328, 0.0225, 0.1246, 0.0097, 0.0225, 0.0500, 0.0511], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0111, 0.0103, 0.0138, 0.0087, 0.0131, 0.0129, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 18:21:25,656 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=289859.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:21:32,016 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7098, 2.5305, 2.3375, 3.4458, 2.2113, 3.6309, 1.6533, 2.6278], device='cuda:6'), covar=tensor([0.1454, 0.0877, 0.1402, 0.0237, 0.0202, 0.0448, 0.1823, 0.0971], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0181, 0.0200, 0.0204, 0.0206, 0.0217, 0.0210, 0.0199], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 18:21:35,866 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1772, 3.2720, 2.9499, 5.2839, 4.1186, 4.3381, 2.2165, 3.2732], device='cuda:6'), covar=tensor([0.1233, 0.0732, 0.1191, 0.0132, 0.0357, 0.0439, 0.1440, 0.0856], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0181, 0.0200, 0.0204, 0.0206, 0.0217, 0.0210, 0.0199], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 18:22:15,728 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.920e+02 3.088e+02 3.696e+02 4.617e+02 9.164e+02, threshold=7.392e+02, percent-clipped=2.0 2023-05-02 18:22:35,917 INFO [train.py:904] (6/8) Epoch 29, batch 5700, loss[loss=0.1855, simple_loss=0.2751, pruned_loss=0.04789, over 16350.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2905, pruned_loss=0.05954, over 3063589.13 frames. ], batch size: 146, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:23:02,333 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=289920.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 18:23:22,088 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4583, 5.4726, 5.2065, 4.4340, 5.4389, 2.1232, 5.1485, 5.0411], device='cuda:6'), covar=tensor([0.0080, 0.0085, 0.0200, 0.0471, 0.0084, 0.2905, 0.0112, 0.0225], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0179, 0.0218, 0.0189, 0.0194, 0.0221, 0.0205, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:23:25,078 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9072, 4.1541, 3.9845, 4.0447, 3.7129, 3.7637, 3.8587, 4.1441], device='cuda:6'), covar=tensor([0.1148, 0.0929, 0.1038, 0.0891, 0.0768, 0.1728, 0.0977, 0.0978], device='cuda:6'), in_proj_covar=tensor([0.0717, 0.0860, 0.0709, 0.0668, 0.0549, 0.0548, 0.0722, 0.0677], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:23:55,706 INFO [train.py:904] (6/8) Epoch 29, batch 5750, loss[loss=0.2067, simple_loss=0.2979, pruned_loss=0.05775, over 16177.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2933, pruned_loss=0.06112, over 3048640.35 frames. ], batch size: 165, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:24:22,433 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-02 18:24:41,623 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=289981.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 18:24:47,181 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3860, 4.5380, 4.7152, 4.4359, 4.5352, 5.0350, 4.5751, 4.2548], device='cuda:6'), covar=tensor([0.1591, 0.1883, 0.2618, 0.2129, 0.2376, 0.1085, 0.1694, 0.2632], device='cuda:6'), in_proj_covar=tensor([0.0431, 0.0639, 0.0705, 0.0519, 0.0694, 0.0729, 0.0549, 0.0693], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 18:24:56,847 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.808e+02 2.660e+02 3.372e+02 4.165e+02 9.302e+02, threshold=6.745e+02, percent-clipped=2.0 2023-05-02 18:25:08,762 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=289998.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:25:20,969 INFO [train.py:904] (6/8) Epoch 29, batch 5800, loss[loss=0.2234, simple_loss=0.2933, pruned_loss=0.07679, over 12014.00 frames. ], tot_loss[loss=0.2067, simple_loss=0.2928, pruned_loss=0.06029, over 3028537.85 frames. ], batch size: 247, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:25:52,408 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0397, 3.0361, 1.9051, 3.2198, 2.3644, 3.3068, 2.2060, 2.5632], device='cuda:6'), covar=tensor([0.0338, 0.0416, 0.1688, 0.0329, 0.0838, 0.0578, 0.1467, 0.0793], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0182, 0.0197, 0.0174, 0.0180, 0.0222, 0.0205, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 18:25:56,246 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5165, 3.4741, 3.4524, 2.6496, 3.3109, 2.1249, 3.1303, 2.7554], device='cuda:6'), covar=tensor([0.0199, 0.0167, 0.0222, 0.0250, 0.0134, 0.2495, 0.0154, 0.0292], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0178, 0.0217, 0.0188, 0.0193, 0.0220, 0.0204, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:26:31,309 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5129, 3.4810, 3.4534, 2.7061, 3.3477, 2.1694, 3.1716, 2.8138], device='cuda:6'), covar=tensor([0.0215, 0.0186, 0.0216, 0.0243, 0.0131, 0.2400, 0.0159, 0.0299], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0178, 0.0217, 0.0188, 0.0193, 0.0220, 0.0204, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:26:31,362 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6117, 2.5379, 1.8951, 2.6585, 2.1213, 2.7619, 2.1805, 2.3496], device='cuda:6'), covar=tensor([0.0318, 0.0342, 0.1364, 0.0304, 0.0678, 0.0490, 0.1195, 0.0611], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0182, 0.0197, 0.0174, 0.0181, 0.0222, 0.0205, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 18:26:34,366 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=290050.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:26:39,145 INFO [train.py:904] (6/8) Epoch 29, batch 5850, loss[loss=0.1694, simple_loss=0.2668, pruned_loss=0.03603, over 16876.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2907, pruned_loss=0.05857, over 3029727.32 frames. ], batch size: 96, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:26:48,280 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=290059.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:27:08,490 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0870, 2.2662, 2.2603, 3.5474, 2.1653, 2.5817, 2.3589, 2.3833], device='cuda:6'), covar=tensor([0.1477, 0.3477, 0.3133, 0.0660, 0.4115, 0.2410, 0.3515, 0.3414], device='cuda:6'), in_proj_covar=tensor([0.0421, 0.0475, 0.0385, 0.0336, 0.0444, 0.0545, 0.0447, 0.0556], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:27:39,839 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.636e+02 2.628e+02 3.224e+02 3.685e+02 6.132e+02, threshold=6.447e+02, percent-clipped=0.0 2023-05-02 18:27:53,066 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7310, 4.9882, 4.6093, 4.3166, 3.9520, 4.8409, 4.7237, 4.4272], device='cuda:6'), covar=tensor([0.1354, 0.1364, 0.0678, 0.0850, 0.2115, 0.0885, 0.0929, 0.1312], device='cuda:6'), in_proj_covar=tensor([0.0310, 0.0471, 0.0364, 0.0365, 0.0361, 0.0421, 0.0251, 0.0436], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:28:01,357 INFO [train.py:904] (6/8) Epoch 29, batch 5900, loss[loss=0.1875, simple_loss=0.2822, pruned_loss=0.04639, over 17110.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2902, pruned_loss=0.05839, over 3042235.97 frames. ], batch size: 48, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:28:15,098 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=290111.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:28:32,133 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7061, 4.6698, 4.4909, 3.3973, 4.5846, 1.5918, 4.2715, 4.0567], device='cuda:6'), covar=tensor([0.0184, 0.0192, 0.0267, 0.0627, 0.0170, 0.3861, 0.0251, 0.0429], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0178, 0.0217, 0.0188, 0.0194, 0.0220, 0.0204, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:29:20,949 INFO [train.py:904] (6/8) Epoch 29, batch 5950, loss[loss=0.1736, simple_loss=0.2711, pruned_loss=0.03807, over 16713.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2906, pruned_loss=0.05724, over 3058554.35 frames. ], batch size: 76, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:29:30,115 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=290159.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:29:45,259 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4233, 3.5430, 2.7488, 2.2098, 2.3906, 2.3548, 3.7923, 3.2079], device='cuda:6'), covar=tensor([0.3286, 0.0633, 0.1875, 0.2957, 0.2738, 0.2313, 0.0536, 0.1372], device='cuda:6'), in_proj_covar=tensor([0.0336, 0.0275, 0.0314, 0.0329, 0.0307, 0.0279, 0.0307, 0.0353], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 18:30:15,422 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.46 vs. limit=2.0 2023-05-02 18:30:18,749 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.883e+02 2.594e+02 3.044e+02 3.971e+02 8.061e+02, threshold=6.088e+02, percent-clipped=1.0 2023-05-02 18:30:40,226 INFO [train.py:904] (6/8) Epoch 29, batch 6000, loss[loss=0.2224, simple_loss=0.3058, pruned_loss=0.06954, over 11250.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2894, pruned_loss=0.05629, over 3071470.33 frames. ], batch size: 246, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:30:40,226 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 18:30:50,250 INFO [train.py:938] (6/8) Epoch 29, validation: loss=0.1475, simple_loss=0.2594, pruned_loss=0.01778, over 944034.00 frames. 2023-05-02 18:30:50,251 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 18:30:55,862 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=290207.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:31:32,702 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8841, 3.5768, 4.1429, 2.1245, 4.3120, 4.2803, 3.1159, 3.2657], device='cuda:6'), covar=tensor([0.0784, 0.0321, 0.0193, 0.1286, 0.0079, 0.0182, 0.0460, 0.0504], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0111, 0.0103, 0.0139, 0.0088, 0.0133, 0.0130, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 18:32:11,209 INFO [train.py:904] (6/8) Epoch 29, batch 6050, loss[loss=0.1834, simple_loss=0.2864, pruned_loss=0.04021, over 16693.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2877, pruned_loss=0.05552, over 3077397.96 frames. ], batch size: 89, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:32:42,733 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=290276.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 18:33:04,446 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.589e+02 2.468e+02 2.905e+02 3.476e+02 8.072e+02, threshold=5.810e+02, percent-clipped=1.0 2023-05-02 18:33:26,386 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-05-02 18:33:29,589 INFO [train.py:904] (6/8) Epoch 29, batch 6100, loss[loss=0.1923, simple_loss=0.2834, pruned_loss=0.05058, over 16517.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2873, pruned_loss=0.05459, over 3092565.63 frames. ], batch size: 62, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:33:53,104 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=290318.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:33:53,383 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 18:34:45,756 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5790, 5.8827, 5.5836, 5.6317, 5.2393, 5.2316, 5.2234, 6.0053], device='cuda:6'), covar=tensor([0.1284, 0.0838, 0.1023, 0.0960, 0.0895, 0.0788, 0.1295, 0.0899], device='cuda:6'), in_proj_covar=tensor([0.0716, 0.0859, 0.0707, 0.0669, 0.0548, 0.0548, 0.0721, 0.0676], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:34:46,605 INFO [train.py:904] (6/8) Epoch 29, batch 6150, loss[loss=0.1651, simple_loss=0.2546, pruned_loss=0.03775, over 17229.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2863, pruned_loss=0.05431, over 3112774.76 frames. ], batch size: 52, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:34:47,113 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=290354.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:35:27,457 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=290379.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:35:44,980 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.708e+02 2.649e+02 3.165e+02 3.863e+02 6.871e+02, threshold=6.329e+02, percent-clipped=1.0 2023-05-02 18:36:04,348 INFO [train.py:904] (6/8) Epoch 29, batch 6200, loss[loss=0.1981, simple_loss=0.2722, pruned_loss=0.06204, over 11595.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2845, pruned_loss=0.05411, over 3101236.71 frames. ], batch size: 247, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:36:08,270 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=290406.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:36:19,451 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.59 vs. limit=5.0 2023-05-02 18:37:21,096 INFO [train.py:904] (6/8) Epoch 29, batch 6250, loss[loss=0.1625, simple_loss=0.2655, pruned_loss=0.02978, over 16698.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2843, pruned_loss=0.05358, over 3119865.85 frames. ], batch size: 76, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:38:15,680 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.611e+02 2.626e+02 3.170e+02 3.673e+02 5.508e+02, threshold=6.341e+02, percent-clipped=0.0 2023-05-02 18:38:34,372 INFO [train.py:904] (6/8) Epoch 29, batch 6300, loss[loss=0.1744, simple_loss=0.2678, pruned_loss=0.04049, over 16781.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.2841, pruned_loss=0.05314, over 3117473.88 frames. ], batch size: 102, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:39:52,799 INFO [train.py:904] (6/8) Epoch 29, batch 6350, loss[loss=0.166, simple_loss=0.262, pruned_loss=0.035, over 16815.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.284, pruned_loss=0.0539, over 3113798.30 frames. ], batch size: 102, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:40:28,332 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=290576.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 18:40:50,037 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.932e+02 2.799e+02 3.423e+02 4.274e+02 8.913e+02, threshold=6.847e+02, percent-clipped=6.0 2023-05-02 18:41:09,308 INFO [train.py:904] (6/8) Epoch 29, batch 6400, loss[loss=0.187, simple_loss=0.2766, pruned_loss=0.0487, over 17241.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2853, pruned_loss=0.05563, over 3090846.17 frames. ], batch size: 44, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:41:26,461 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4718, 4.5572, 4.3426, 4.0471, 4.0669, 4.4753, 4.1858, 4.1896], device='cuda:6'), covar=tensor([0.0697, 0.0584, 0.0350, 0.0374, 0.0939, 0.0522, 0.0661, 0.0677], device='cuda:6'), in_proj_covar=tensor([0.0311, 0.0473, 0.0365, 0.0366, 0.0362, 0.0422, 0.0251, 0.0438], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:41:39,269 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=290624.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 18:42:23,290 INFO [train.py:904] (6/8) Epoch 29, batch 6450, loss[loss=0.1813, simple_loss=0.2673, pruned_loss=0.04763, over 16303.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2856, pruned_loss=0.05504, over 3096526.46 frames. ], batch size: 165, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:42:23,867 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=290654.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:42:31,638 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5232, 4.1809, 4.1101, 2.5895, 3.7368, 4.2322, 3.7732, 2.3037], device='cuda:6'), covar=tensor([0.0594, 0.0059, 0.0070, 0.0493, 0.0111, 0.0104, 0.0104, 0.0529], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0091, 0.0092, 0.0137, 0.0104, 0.0117, 0.0100, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 18:42:38,906 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6710, 3.8128, 2.8954, 2.2758, 2.4055, 2.4841, 4.0692, 3.3718], device='cuda:6'), covar=tensor([0.2997, 0.0584, 0.1905, 0.3054, 0.3084, 0.2253, 0.0413, 0.1385], device='cuda:6'), in_proj_covar=tensor([0.0336, 0.0275, 0.0313, 0.0328, 0.0306, 0.0278, 0.0305, 0.0351], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 18:42:53,366 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=290674.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:42:54,902 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7999, 4.8603, 4.6748, 4.3797, 4.3782, 4.7812, 4.5317, 4.5014], device='cuda:6'), covar=tensor([0.0586, 0.0604, 0.0319, 0.0298, 0.0892, 0.0470, 0.0502, 0.0610], device='cuda:6'), in_proj_covar=tensor([0.0311, 0.0473, 0.0365, 0.0366, 0.0362, 0.0422, 0.0250, 0.0438], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:43:19,913 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.946e+02 2.663e+02 2.912e+02 3.666e+02 7.754e+02, threshold=5.824e+02, percent-clipped=2.0 2023-05-02 18:43:37,901 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=290702.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:43:40,038 INFO [train.py:904] (6/8) Epoch 29, batch 6500, loss[loss=0.1969, simple_loss=0.2774, pruned_loss=0.05822, over 16408.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2834, pruned_loss=0.05417, over 3108031.22 frames. ], batch size: 146, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:43:44,264 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=290706.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:44:59,499 INFO [train.py:904] (6/8) Epoch 29, batch 6550, loss[loss=0.2012, simple_loss=0.2872, pruned_loss=0.05759, over 16411.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2859, pruned_loss=0.05515, over 3095189.55 frames. ], batch size: 35, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:44:59,827 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=290754.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:45:58,155 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.498e+02 2.774e+02 3.264e+02 3.867e+02 7.263e+02, threshold=6.527e+02, percent-clipped=1.0 2023-05-02 18:46:19,467 INFO [train.py:904] (6/8) Epoch 29, batch 6600, loss[loss=0.2057, simple_loss=0.2926, pruned_loss=0.05941, over 16440.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.288, pruned_loss=0.05552, over 3082666.92 frames. ], batch size: 68, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:46:22,284 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5872, 4.7846, 4.9988, 4.7538, 4.8583, 5.3585, 4.8359, 4.5584], device='cuda:6'), covar=tensor([0.1327, 0.1880, 0.2222, 0.1894, 0.2251, 0.1028, 0.1777, 0.2522], device='cuda:6'), in_proj_covar=tensor([0.0434, 0.0643, 0.0714, 0.0520, 0.0698, 0.0732, 0.0554, 0.0698], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 18:47:38,810 INFO [train.py:904] (6/8) Epoch 29, batch 6650, loss[loss=0.1786, simple_loss=0.2728, pruned_loss=0.04217, over 16729.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2878, pruned_loss=0.05592, over 3092876.76 frames. ], batch size: 83, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:47:41,314 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=2.14 vs. limit=2.0 2023-05-02 18:47:58,122 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 18:48:35,966 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 2.668e+02 3.227e+02 3.787e+02 6.260e+02, threshold=6.453e+02, percent-clipped=0.0 2023-05-02 18:48:55,746 INFO [train.py:904] (6/8) Epoch 29, batch 6700, loss[loss=0.2458, simple_loss=0.3036, pruned_loss=0.09398, over 11288.00 frames. ], tot_loss[loss=0.1987, simple_loss=0.2861, pruned_loss=0.05564, over 3089341.81 frames. ], batch size: 246, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:49:07,958 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3280, 5.3375, 5.1862, 4.7893, 4.7897, 5.2505, 5.1721, 4.9133], device='cuda:6'), covar=tensor([0.0660, 0.0492, 0.0315, 0.0324, 0.1147, 0.0465, 0.0319, 0.0688], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0473, 0.0365, 0.0366, 0.0362, 0.0422, 0.0251, 0.0439], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 18:50:12,971 INFO [train.py:904] (6/8) Epoch 29, batch 6750, loss[loss=0.1945, simple_loss=0.2742, pruned_loss=0.05744, over 16639.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2841, pruned_loss=0.05526, over 3098560.33 frames. ], batch size: 62, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:50:43,284 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=290974.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:51:10,930 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.985e+02 2.623e+02 3.095e+02 3.882e+02 8.048e+02, threshold=6.190e+02, percent-clipped=2.0 2023-05-02 18:51:28,424 INFO [train.py:904] (6/8) Epoch 29, batch 6800, loss[loss=0.1776, simple_loss=0.2788, pruned_loss=0.0382, over 16765.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2851, pruned_loss=0.05589, over 3107137.61 frames. ], batch size: 102, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:51:57,702 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=291022.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:52:46,127 INFO [train.py:904] (6/8) Epoch 29, batch 6850, loss[loss=0.1954, simple_loss=0.2999, pruned_loss=0.04538, over 16788.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2869, pruned_loss=0.05634, over 3105954.23 frames. ], batch size: 124, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:53:09,389 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=291069.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:53:42,652 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.921e+02 2.621e+02 3.195e+02 3.750e+02 8.276e+02, threshold=6.389e+02, percent-clipped=2.0 2023-05-02 18:54:02,728 INFO [train.py:904] (6/8) Epoch 29, batch 6900, loss[loss=0.2131, simple_loss=0.3022, pruned_loss=0.06202, over 16197.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2888, pruned_loss=0.05588, over 3101254.48 frames. ], batch size: 165, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:54:44,641 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=291130.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:55:01,484 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4932, 4.2054, 4.1752, 2.6946, 3.6784, 4.1702, 3.7059, 2.4442], device='cuda:6'), covar=tensor([0.0611, 0.0054, 0.0062, 0.0471, 0.0119, 0.0139, 0.0104, 0.0488], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0090, 0.0092, 0.0136, 0.0103, 0.0117, 0.0099, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 18:55:20,068 INFO [train.py:904] (6/8) Epoch 29, batch 6950, loss[loss=0.2083, simple_loss=0.2874, pruned_loss=0.06457, over 16784.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2897, pruned_loss=0.05694, over 3100014.90 frames. ], batch size: 124, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:56:18,588 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.816e+02 3.069e+02 3.631e+02 4.410e+02 8.633e+02, threshold=7.261e+02, percent-clipped=4.0 2023-05-02 18:56:36,625 INFO [train.py:904] (6/8) Epoch 29, batch 7000, loss[loss=0.1797, simple_loss=0.2814, pruned_loss=0.03904, over 16185.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2899, pruned_loss=0.05598, over 3108192.24 frames. ], batch size: 165, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:56:58,491 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-02 18:57:38,913 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-05-02 18:57:50,283 INFO [train.py:904] (6/8) Epoch 29, batch 7050, loss[loss=0.2676, simple_loss=0.3281, pruned_loss=0.1036, over 11622.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2907, pruned_loss=0.05583, over 3110822.88 frames. ], batch size: 248, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:58:49,443 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 2.753e+02 3.384e+02 4.060e+02 1.222e+03, threshold=6.768e+02, percent-clipped=4.0 2023-05-02 18:59:07,325 INFO [train.py:904] (6/8) Epoch 29, batch 7100, loss[loss=0.2126, simple_loss=0.2936, pruned_loss=0.0658, over 15506.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2897, pruned_loss=0.05608, over 3100691.62 frames. ], batch size: 191, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 19:00:25,897 INFO [train.py:904] (6/8) Epoch 29, batch 7150, loss[loss=0.2009, simple_loss=0.288, pruned_loss=0.05694, over 16712.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2882, pruned_loss=0.05613, over 3100486.32 frames. ], batch size: 134, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 19:00:47,176 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-02 19:01:19,076 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-02 19:01:23,563 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 2.798e+02 3.396e+02 4.128e+02 9.099e+02, threshold=6.792e+02, percent-clipped=1.0 2023-05-02 19:01:41,833 INFO [train.py:904] (6/8) Epoch 29, batch 7200, loss[loss=0.1757, simple_loss=0.2695, pruned_loss=0.04089, over 16450.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2857, pruned_loss=0.0544, over 3105266.82 frames. ], batch size: 146, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 19:02:06,769 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8520, 5.1024, 4.8929, 4.9107, 4.6460, 4.5556, 4.5599, 5.1662], device='cuda:6'), covar=tensor([0.1144, 0.0778, 0.0882, 0.0832, 0.0781, 0.1166, 0.1099, 0.0765], device='cuda:6'), in_proj_covar=tensor([0.0721, 0.0862, 0.0714, 0.0672, 0.0551, 0.0553, 0.0724, 0.0680], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 19:02:14,451 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=291425.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:03:00,075 INFO [train.py:904] (6/8) Epoch 29, batch 7250, loss[loss=0.1826, simple_loss=0.2679, pruned_loss=0.04863, over 16212.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.2839, pruned_loss=0.05343, over 3102085.46 frames. ], batch size: 165, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:03:55,906 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1388, 3.1574, 1.9691, 3.4523, 2.3969, 3.4865, 2.1080, 2.5619], device='cuda:6'), covar=tensor([0.0349, 0.0462, 0.1778, 0.0245, 0.0997, 0.0626, 0.1601, 0.0853], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0181, 0.0197, 0.0174, 0.0181, 0.0222, 0.0205, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 19:03:58,887 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.627e+02 2.585e+02 3.000e+02 3.629e+02 8.129e+02, threshold=6.000e+02, percent-clipped=2.0 2023-05-02 19:04:06,920 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6658, 2.5874, 1.9813, 2.7033, 2.2187, 2.8108, 2.1593, 2.3710], device='cuda:6'), covar=tensor([0.0304, 0.0357, 0.1196, 0.0278, 0.0630, 0.0513, 0.1170, 0.0548], device='cuda:6'), in_proj_covar=tensor([0.0177, 0.0181, 0.0196, 0.0174, 0.0181, 0.0221, 0.0205, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 19:04:16,116 INFO [train.py:904] (6/8) Epoch 29, batch 7300, loss[loss=0.1972, simple_loss=0.2873, pruned_loss=0.05355, over 16732.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2837, pruned_loss=0.05365, over 3106277.65 frames. ], batch size: 76, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:04:20,672 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8386, 2.7317, 2.8666, 2.2108, 2.7171, 2.1804, 2.6898, 2.9497], device='cuda:6'), covar=tensor([0.0269, 0.0836, 0.0502, 0.1770, 0.0816, 0.0922, 0.0609, 0.0757], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0172, 0.0172, 0.0158, 0.0149, 0.0134, 0.0146, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 19:05:07,532 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.42 vs. limit=2.0 2023-05-02 19:05:32,025 INFO [train.py:904] (6/8) Epoch 29, batch 7350, loss[loss=0.2615, simple_loss=0.3174, pruned_loss=0.1028, over 11381.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2848, pruned_loss=0.05499, over 3070315.24 frames. ], batch size: 250, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:06:05,647 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-05-02 19:06:32,137 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.793e+02 2.867e+02 3.369e+02 3.998e+02 9.840e+02, threshold=6.738e+02, percent-clipped=8.0 2023-05-02 19:06:32,991 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-02 19:06:49,563 INFO [train.py:904] (6/8) Epoch 29, batch 7400, loss[loss=0.2011, simple_loss=0.2921, pruned_loss=0.05501, over 16367.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.286, pruned_loss=0.05577, over 3056069.91 frames. ], batch size: 146, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:06:53,630 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 19:07:11,733 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7507, 2.4684, 2.2336, 3.2819, 2.2401, 3.5512, 1.5574, 2.7066], device='cuda:6'), covar=tensor([0.1502, 0.0852, 0.1438, 0.0242, 0.0185, 0.0401, 0.1933, 0.0887], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0182, 0.0202, 0.0206, 0.0208, 0.0220, 0.0212, 0.0201], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 19:08:07,280 INFO [train.py:904] (6/8) Epoch 29, batch 7450, loss[loss=0.2341, simple_loss=0.2965, pruned_loss=0.08585, over 11498.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2867, pruned_loss=0.05633, over 3057195.68 frames. ], batch size: 247, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:08:29,627 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 19:09:10,905 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.991e+02 2.789e+02 3.286e+02 3.823e+02 6.627e+02, threshold=6.571e+02, percent-clipped=0.0 2023-05-02 19:09:28,108 INFO [train.py:904] (6/8) Epoch 29, batch 7500, loss[loss=0.178, simple_loss=0.2706, pruned_loss=0.04273, over 16832.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2862, pruned_loss=0.05532, over 3057694.60 frames. ], batch size: 90, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:10:01,667 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=291725.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:10:45,759 INFO [train.py:904] (6/8) Epoch 29, batch 7550, loss[loss=0.1929, simple_loss=0.2789, pruned_loss=0.05342, over 16730.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2859, pruned_loss=0.05604, over 3028049.39 frames. ], batch size: 83, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:11:15,423 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=291773.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:11:19,733 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=291776.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:11:26,762 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2275, 5.2595, 5.0568, 4.6685, 4.7518, 5.1381, 5.1094, 4.8190], device='cuda:6'), covar=tensor([0.0634, 0.0499, 0.0332, 0.0357, 0.0942, 0.0529, 0.0339, 0.0727], device='cuda:6'), in_proj_covar=tensor([0.0309, 0.0468, 0.0361, 0.0362, 0.0357, 0.0417, 0.0248, 0.0434], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 19:11:44,713 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.882e+02 2.637e+02 3.162e+02 3.774e+02 1.302e+03, threshold=6.323e+02, percent-clipped=2.0 2023-05-02 19:12:01,405 INFO [train.py:904] (6/8) Epoch 29, batch 7600, loss[loss=0.1696, simple_loss=0.264, pruned_loss=0.03756, over 17013.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2847, pruned_loss=0.05535, over 3065406.93 frames. ], batch size: 53, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:12:19,372 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7880, 1.9601, 2.3991, 2.7466, 2.7097, 3.0960, 2.0213, 3.1152], device='cuda:6'), covar=tensor([0.0260, 0.0610, 0.0408, 0.0363, 0.0378, 0.0252, 0.0707, 0.0171], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0197, 0.0186, 0.0191, 0.0209, 0.0166, 0.0204, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 19:12:52,639 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=291837.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:13:18,969 INFO [train.py:904] (6/8) Epoch 29, batch 7650, loss[loss=0.236, simple_loss=0.305, pruned_loss=0.08347, over 11403.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2859, pruned_loss=0.05619, over 3064649.41 frames. ], batch size: 247, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:14:09,187 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 19:14:20,910 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.976e+02 2.915e+02 3.313e+02 4.275e+02 8.487e+02, threshold=6.627e+02, percent-clipped=5.0 2023-05-02 19:14:36,048 INFO [train.py:904] (6/8) Epoch 29, batch 7700, loss[loss=0.179, simple_loss=0.2705, pruned_loss=0.04377, over 16290.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.285, pruned_loss=0.05593, over 3082322.48 frames. ], batch size: 165, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:14:43,789 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4292, 4.0619, 4.0828, 2.7015, 3.6921, 4.1325, 3.6691, 2.2252], device='cuda:6'), covar=tensor([0.0614, 0.0070, 0.0067, 0.0458, 0.0119, 0.0131, 0.0115, 0.0542], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0091, 0.0092, 0.0136, 0.0103, 0.0117, 0.0099, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 19:15:53,492 INFO [train.py:904] (6/8) Epoch 29, batch 7750, loss[loss=0.2021, simple_loss=0.2838, pruned_loss=0.06022, over 16518.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.285, pruned_loss=0.05593, over 3078175.35 frames. ], batch size: 57, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:16:55,756 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.177e+02 2.797e+02 3.314e+02 3.963e+02 7.468e+02, threshold=6.627e+02, percent-clipped=1.0 2023-05-02 19:17:14,570 INFO [train.py:904] (6/8) Epoch 29, batch 7800, loss[loss=0.1908, simple_loss=0.2829, pruned_loss=0.04931, over 16679.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2858, pruned_loss=0.05655, over 3081904.70 frames. ], batch size: 134, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:17:18,360 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8491, 4.3565, 3.1519, 2.4817, 2.8361, 2.7787, 4.7003, 3.6738], device='cuda:6'), covar=tensor([0.2961, 0.0542, 0.1874, 0.2885, 0.2865, 0.2053, 0.0376, 0.1365], device='cuda:6'), in_proj_covar=tensor([0.0337, 0.0275, 0.0315, 0.0329, 0.0308, 0.0279, 0.0305, 0.0353], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 19:18:25,596 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4694, 3.9687, 3.9544, 2.5879, 3.6285, 4.0255, 3.5771, 2.1347], device='cuda:6'), covar=tensor([0.0600, 0.0080, 0.0079, 0.0488, 0.0132, 0.0120, 0.0125, 0.0569], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0091, 0.0092, 0.0136, 0.0103, 0.0117, 0.0099, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 19:18:30,014 INFO [train.py:904] (6/8) Epoch 29, batch 7850, loss[loss=0.2009, simple_loss=0.2875, pruned_loss=0.05718, over 16176.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2864, pruned_loss=0.056, over 3093791.67 frames. ], batch size: 35, lr: 2.30e-03, grad_scale: 2.0 2023-05-02 19:18:31,063 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=292054.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:18:50,137 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2023-05-02 19:19:30,095 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.904e+02 2.705e+02 3.148e+02 3.782e+02 7.710e+02, threshold=6.297e+02, percent-clipped=1.0 2023-05-02 19:19:43,448 INFO [train.py:904] (6/8) Epoch 29, batch 7900, loss[loss=0.2552, simple_loss=0.3179, pruned_loss=0.09626, over 11699.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2853, pruned_loss=0.05582, over 3071681.26 frames. ], batch size: 248, lr: 2.30e-03, grad_scale: 2.0 2023-05-02 19:19:59,956 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=292115.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:20:26,273 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=292132.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:20:55,167 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-05-02 19:21:01,248 INFO [train.py:904] (6/8) Epoch 29, batch 7950, loss[loss=0.1997, simple_loss=0.2825, pruned_loss=0.05845, over 16877.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2861, pruned_loss=0.05624, over 3075240.28 frames. ], batch size: 42, lr: 2.30e-03, grad_scale: 2.0 2023-05-02 19:22:03,657 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.684e+02 2.577e+02 3.154e+02 3.832e+02 7.252e+02, threshold=6.308e+02, percent-clipped=1.0 2023-05-02 19:22:18,471 INFO [train.py:904] (6/8) Epoch 29, batch 8000, loss[loss=0.1916, simple_loss=0.2851, pruned_loss=0.0491, over 16865.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2866, pruned_loss=0.05722, over 3054142.08 frames. ], batch size: 116, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:23:31,105 INFO [train.py:904] (6/8) Epoch 29, batch 8050, loss[loss=0.2023, simple_loss=0.2841, pruned_loss=0.06026, over 11613.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2869, pruned_loss=0.0572, over 3045681.08 frames. ], batch size: 247, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:24:32,478 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.919e+02 2.613e+02 3.145e+02 3.848e+02 6.977e+02, threshold=6.289e+02, percent-clipped=2.0 2023-05-02 19:24:46,377 INFO [train.py:904] (6/8) Epoch 29, batch 8100, loss[loss=0.1914, simple_loss=0.2742, pruned_loss=0.05426, over 16317.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2862, pruned_loss=0.05651, over 3059694.84 frames. ], batch size: 35, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:25:02,693 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 19:25:03,464 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2306, 5.8827, 6.1177, 5.7521, 5.8335, 6.3609, 5.8389, 5.5510], device='cuda:6'), covar=tensor([0.0907, 0.1670, 0.2218, 0.1846, 0.2093, 0.0828, 0.1536, 0.2338], device='cuda:6'), in_proj_covar=tensor([0.0430, 0.0639, 0.0708, 0.0518, 0.0691, 0.0730, 0.0549, 0.0693], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 19:25:03,642 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3752, 2.3290, 2.9318, 3.2337, 3.1324, 3.7703, 2.4727, 3.7490], device='cuda:6'), covar=tensor([0.0212, 0.0529, 0.0325, 0.0323, 0.0330, 0.0176, 0.0555, 0.0148], device='cuda:6'), in_proj_covar=tensor([0.0196, 0.0197, 0.0187, 0.0192, 0.0208, 0.0166, 0.0204, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 19:25:09,312 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8760, 3.0881, 3.3873, 2.2075, 2.9993, 2.1818, 3.4128, 3.4610], device='cuda:6'), covar=tensor([0.0263, 0.0892, 0.0598, 0.2049, 0.0835, 0.1074, 0.0599, 0.0894], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0171, 0.0171, 0.0157, 0.0148, 0.0134, 0.0145, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 19:26:01,418 INFO [train.py:904] (6/8) Epoch 29, batch 8150, loss[loss=0.2225, simple_loss=0.2897, pruned_loss=0.07764, over 11680.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2833, pruned_loss=0.05513, over 3086891.67 frames. ], batch size: 247, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:27:01,239 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.703e+02 2.736e+02 3.221e+02 3.909e+02 8.294e+02, threshold=6.443e+02, percent-clipped=4.0 2023-05-02 19:27:04,971 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 19:27:15,046 INFO [train.py:904] (6/8) Epoch 29, batch 8200, loss[loss=0.167, simple_loss=0.2627, pruned_loss=0.03562, over 16919.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.281, pruned_loss=0.05511, over 3073122.47 frames. ], batch size: 116, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:27:25,888 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=292410.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:28:00,578 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=292432.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:28:34,795 INFO [train.py:904] (6/8) Epoch 29, batch 8250, loss[loss=0.1666, simple_loss=0.2634, pruned_loss=0.03486, over 16758.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2795, pruned_loss=0.05219, over 3044489.46 frames. ], batch size: 89, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:28:37,897 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4399, 1.7468, 2.1485, 2.3892, 2.4354, 2.6658, 1.9114, 2.6268], device='cuda:6'), covar=tensor([0.0252, 0.0583, 0.0376, 0.0395, 0.0382, 0.0284, 0.0617, 0.0233], device='cuda:6'), in_proj_covar=tensor([0.0195, 0.0196, 0.0185, 0.0191, 0.0207, 0.0166, 0.0203, 0.0167], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 19:29:17,990 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=292480.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:29:40,717 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2624, 4.3692, 4.5088, 4.2871, 4.3524, 4.8503, 4.3981, 4.0899], device='cuda:6'), covar=tensor([0.1742, 0.2100, 0.2311, 0.2027, 0.2525, 0.1115, 0.1649, 0.2597], device='cuda:6'), in_proj_covar=tensor([0.0428, 0.0635, 0.0703, 0.0513, 0.0686, 0.0726, 0.0547, 0.0689], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 19:29:41,435 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.407e+02 2.184e+02 2.675e+02 3.477e+02 6.200e+02, threshold=5.351e+02, percent-clipped=0.0 2023-05-02 19:29:55,807 INFO [train.py:904] (6/8) Epoch 29, batch 8300, loss[loss=0.1721, simple_loss=0.2554, pruned_loss=0.04442, over 12340.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2767, pruned_loss=0.04952, over 3017444.90 frames. ], batch size: 246, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:30:23,626 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9479, 4.9165, 4.6894, 4.0350, 4.7677, 1.9269, 4.5550, 4.4091], device='cuda:6'), covar=tensor([0.0119, 0.0116, 0.0233, 0.0417, 0.0128, 0.2967, 0.0145, 0.0317], device='cuda:6'), in_proj_covar=tensor([0.0182, 0.0176, 0.0215, 0.0185, 0.0190, 0.0219, 0.0202, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 19:31:15,559 INFO [train.py:904] (6/8) Epoch 29, batch 8350, loss[loss=0.16, simple_loss=0.2652, pruned_loss=0.02741, over 16910.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2762, pruned_loss=0.04747, over 3022517.22 frames. ], batch size: 96, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:32:20,503 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.032e+02 2.659e+02 3.191e+02 5.260e+02, threshold=5.318e+02, percent-clipped=0.0 2023-05-02 19:32:36,081 INFO [train.py:904] (6/8) Epoch 29, batch 8400, loss[loss=0.1701, simple_loss=0.2651, pruned_loss=0.03751, over 16364.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2733, pruned_loss=0.04536, over 3007582.29 frames. ], batch size: 146, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:33:14,839 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4340, 3.0541, 2.7919, 2.2948, 2.2402, 2.3824, 3.0357, 2.8453], device='cuda:6'), covar=tensor([0.2691, 0.0646, 0.1626, 0.3070, 0.2798, 0.2330, 0.0491, 0.1677], device='cuda:6'), in_proj_covar=tensor([0.0333, 0.0272, 0.0310, 0.0325, 0.0304, 0.0275, 0.0302, 0.0348], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 19:33:56,324 INFO [train.py:904] (6/8) Epoch 29, batch 8450, loss[loss=0.1635, simple_loss=0.2585, pruned_loss=0.03422, over 17187.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.272, pruned_loss=0.04407, over 3004412.81 frames. ], batch size: 46, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:35:03,479 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.196e+02 2.552e+02 2.926e+02 6.173e+02, threshold=5.103e+02, percent-clipped=1.0 2023-05-02 19:35:19,248 INFO [train.py:904] (6/8) Epoch 29, batch 8500, loss[loss=0.162, simple_loss=0.2575, pruned_loss=0.03325, over 15157.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2684, pruned_loss=0.0417, over 3024388.30 frames. ], batch size: 190, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:35:28,855 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=292710.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:36:42,994 INFO [train.py:904] (6/8) Epoch 29, batch 8550, loss[loss=0.1708, simple_loss=0.2629, pruned_loss=0.03938, over 12269.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2666, pruned_loss=0.04068, over 3026492.43 frames. ], batch size: 248, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:36:46,115 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9424, 4.2307, 4.0701, 4.1206, 3.7663, 3.8448, 3.8530, 4.2350], device='cuda:6'), covar=tensor([0.1224, 0.1021, 0.0967, 0.0817, 0.0840, 0.1716, 0.1019, 0.1023], device='cuda:6'), in_proj_covar=tensor([0.0713, 0.0856, 0.0705, 0.0665, 0.0545, 0.0547, 0.0717, 0.0673], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 19:36:52,495 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=292758.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:36:56,462 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7475, 4.8156, 4.6065, 4.2034, 4.2846, 4.7162, 4.5344, 4.3967], device='cuda:6'), covar=tensor([0.0628, 0.0743, 0.0387, 0.0403, 0.0975, 0.0630, 0.0491, 0.0880], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0468, 0.0360, 0.0361, 0.0356, 0.0415, 0.0249, 0.0431], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 19:38:04,313 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.654e+02 2.162e+02 2.570e+02 3.257e+02 5.024e+02, threshold=5.139e+02, percent-clipped=0.0 2023-05-02 19:38:21,503 INFO [train.py:904] (6/8) Epoch 29, batch 8600, loss[loss=0.1522, simple_loss=0.2522, pruned_loss=0.02614, over 16391.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2674, pruned_loss=0.03979, over 3031231.16 frames. ], batch size: 68, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:40:01,878 INFO [train.py:904] (6/8) Epoch 29, batch 8650, loss[loss=0.169, simple_loss=0.2693, pruned_loss=0.03437, over 16672.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2655, pruned_loss=0.03832, over 3019366.68 frames. ], batch size: 134, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:40:30,534 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-02 19:40:32,243 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 19:41:31,353 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.553e+02 2.024e+02 2.393e+02 2.870e+02 4.192e+02, threshold=4.785e+02, percent-clipped=0.0 2023-05-02 19:41:48,478 INFO [train.py:904] (6/8) Epoch 29, batch 8700, loss[loss=0.1766, simple_loss=0.2816, pruned_loss=0.03585, over 16258.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2634, pruned_loss=0.03726, over 3029552.04 frames. ], batch size: 146, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:43:00,478 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-02 19:43:26,846 INFO [train.py:904] (6/8) Epoch 29, batch 8750, loss[loss=0.197, simple_loss=0.2924, pruned_loss=0.05085, over 16418.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.263, pruned_loss=0.03658, over 3057950.42 frames. ], batch size: 147, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:44:01,888 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=292967.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:44:49,935 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 19:44:58,329 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6064, 4.4350, 4.6408, 4.7735, 4.9483, 4.4861, 4.9624, 4.9643], device='cuda:6'), covar=tensor([0.2039, 0.1392, 0.1735, 0.0947, 0.0570, 0.1039, 0.0650, 0.0720], device='cuda:6'), in_proj_covar=tensor([0.0665, 0.0812, 0.0933, 0.0828, 0.0630, 0.0650, 0.0690, 0.0807], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 19:45:00,944 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.419e+02 2.165e+02 2.592e+02 3.174e+02 6.684e+02, threshold=5.184e+02, percent-clipped=4.0 2023-05-02 19:45:20,129 INFO [train.py:904] (6/8) Epoch 29, batch 8800, loss[loss=0.1754, simple_loss=0.2668, pruned_loss=0.04204, over 16306.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2615, pruned_loss=0.03564, over 3053949.57 frames. ], batch size: 165, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:46:12,291 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=293028.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:47:05,852 INFO [train.py:904] (6/8) Epoch 29, batch 8850, loss[loss=0.1561, simple_loss=0.2475, pruned_loss=0.03236, over 12279.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2633, pruned_loss=0.03539, over 3014435.31 frames. ], batch size: 248, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:48:05,824 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=293081.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:48:35,604 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.279e+02 2.601e+02 3.166e+02 4.771e+02, threshold=5.201e+02, percent-clipped=0.0 2023-05-02 19:48:55,720 INFO [train.py:904] (6/8) Epoch 29, batch 8900, loss[loss=0.1572, simple_loss=0.2477, pruned_loss=0.0333, over 12500.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.264, pruned_loss=0.03473, over 3032129.71 frames. ], batch size: 247, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:49:20,195 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8709, 2.2828, 2.3401, 3.0056, 1.9075, 3.1684, 1.7084, 2.7904], device='cuda:6'), covar=tensor([0.1392, 0.0777, 0.1140, 0.0203, 0.0117, 0.0336, 0.1758, 0.0745], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0181, 0.0200, 0.0201, 0.0205, 0.0216, 0.0210, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 19:49:23,376 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=293117.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:50:34,627 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=293142.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:50:57,011 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-02 19:51:01,958 INFO [train.py:904] (6/8) Epoch 29, batch 8950, loss[loss=0.1603, simple_loss=0.2529, pruned_loss=0.03383, over 16195.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2637, pruned_loss=0.0353, over 3027264.10 frames. ], batch size: 35, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:51:53,129 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=293178.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:52:32,343 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.460e+02 2.109e+02 2.425e+02 3.053e+02 5.475e+02, threshold=4.851e+02, percent-clipped=2.0 2023-05-02 19:52:53,145 INFO [train.py:904] (6/8) Epoch 29, batch 9000, loss[loss=0.1559, simple_loss=0.2443, pruned_loss=0.03379, over 12304.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2611, pruned_loss=0.03442, over 3032192.74 frames. ], batch size: 248, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:52:53,145 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 19:53:02,752 INFO [train.py:938] (6/8) Epoch 29, validation: loss=0.1431, simple_loss=0.2465, pruned_loss=0.01987, over 944034.00 frames. 2023-05-02 19:53:02,753 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 19:53:49,451 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8587, 1.4509, 1.7341, 1.7843, 1.9317, 1.9595, 1.7014, 1.8981], device='cuda:6'), covar=tensor([0.0285, 0.0473, 0.0247, 0.0338, 0.0350, 0.0239, 0.0490, 0.0156], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0194, 0.0183, 0.0187, 0.0205, 0.0163, 0.0200, 0.0164], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 19:54:47,960 INFO [train.py:904] (6/8) Epoch 29, batch 9050, loss[loss=0.1389, simple_loss=0.2353, pruned_loss=0.02127, over 16858.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2612, pruned_loss=0.03449, over 3039387.98 frames. ], batch size: 83, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:56:03,667 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-05-02 19:56:15,025 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.113e+02 2.508e+02 2.955e+02 4.534e+02, threshold=5.015e+02, percent-clipped=0.0 2023-05-02 19:56:34,919 INFO [train.py:904] (6/8) Epoch 29, batch 9100, loss[loss=0.1705, simple_loss=0.2727, pruned_loss=0.03413, over 16849.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2612, pruned_loss=0.03525, over 3038890.96 frames. ], batch size: 96, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:56:36,598 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0124, 3.7257, 4.0761, 1.8313, 4.2696, 4.4450, 3.3737, 3.3956], device='cuda:6'), covar=tensor([0.0737, 0.0289, 0.0284, 0.1485, 0.0101, 0.0142, 0.0388, 0.0454], device='cuda:6'), in_proj_covar=tensor([0.0145, 0.0108, 0.0099, 0.0136, 0.0085, 0.0128, 0.0127, 0.0127], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 19:57:17,597 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=293323.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:57:35,695 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8184, 4.9898, 5.1624, 4.9082, 5.0572, 5.5317, 5.0186, 4.6935], device='cuda:6'), covar=tensor([0.1059, 0.1773, 0.2027, 0.2061, 0.2128, 0.0920, 0.1481, 0.2517], device='cuda:6'), in_proj_covar=tensor([0.0414, 0.0620, 0.0687, 0.0503, 0.0673, 0.0709, 0.0533, 0.0671], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 19:58:34,110 INFO [train.py:904] (6/8) Epoch 29, batch 9150, loss[loss=0.1617, simple_loss=0.2513, pruned_loss=0.03609, over 12155.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2618, pruned_loss=0.03518, over 3045376.01 frames. ], batch size: 246, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:58:51,979 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=293361.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:58:57,364 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=293364.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:00:04,473 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.264e+02 2.108e+02 2.614e+02 3.370e+02 9.014e+02, threshold=5.228e+02, percent-clipped=3.0 2023-05-02 20:00:20,590 INFO [train.py:904] (6/8) Epoch 29, batch 9200, loss[loss=0.1765, simple_loss=0.2777, pruned_loss=0.03763, over 16421.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2575, pruned_loss=0.03416, over 3050812.67 frames. ], batch size: 146, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:00:55,083 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=293422.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:01:00,702 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=293425.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:01:23,542 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=293437.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:01:58,341 INFO [train.py:904] (6/8) Epoch 29, batch 9250, loss[loss=0.177, simple_loss=0.2772, pruned_loss=0.03841, over 15255.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2574, pruned_loss=0.03406, over 3059725.76 frames. ], batch size: 190, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:02:39,069 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=293473.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:03:28,486 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.628e+02 2.175e+02 2.514e+02 3.115e+02 5.451e+02, threshold=5.029e+02, percent-clipped=2.0 2023-05-02 20:03:38,215 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0114, 2.7261, 3.0555, 2.1937, 2.8137, 2.2166, 2.7553, 2.9204], device='cuda:6'), covar=tensor([0.0336, 0.1007, 0.0434, 0.1880, 0.0723, 0.0994, 0.0619, 0.0817], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0165, 0.0166, 0.0153, 0.0144, 0.0130, 0.0142, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:03:49,264 INFO [train.py:904] (6/8) Epoch 29, batch 9300, loss[loss=0.1503, simple_loss=0.2464, pruned_loss=0.02706, over 15395.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.2559, pruned_loss=0.03354, over 3054235.28 frames. ], batch size: 194, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:05:33,314 INFO [train.py:904] (6/8) Epoch 29, batch 9350, loss[loss=0.1572, simple_loss=0.2405, pruned_loss=0.03693, over 12189.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.2556, pruned_loss=0.0333, over 3060888.85 frames. ], batch size: 249, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:06:56,970 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 1.973e+02 2.358e+02 2.873e+02 5.255e+02, threshold=4.716e+02, percent-clipped=1.0 2023-05-02 20:07:01,163 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9615, 4.9834, 5.2890, 5.2693, 5.2859, 5.0741, 4.9341, 4.8409], device='cuda:6'), covar=tensor([0.0357, 0.0607, 0.0382, 0.0397, 0.0435, 0.0330, 0.0861, 0.0375], device='cuda:6'), in_proj_covar=tensor([0.0426, 0.0484, 0.0467, 0.0430, 0.0511, 0.0491, 0.0564, 0.0394], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 20:07:15,889 INFO [train.py:904] (6/8) Epoch 29, batch 9400, loss[loss=0.1725, simple_loss=0.2757, pruned_loss=0.03469, over 16385.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2556, pruned_loss=0.03293, over 3077026.81 frames. ], batch size: 146, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:07:53,878 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=293623.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:08:23,063 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 20:08:55,052 INFO [train.py:904] (6/8) Epoch 29, batch 9450, loss[loss=0.1608, simple_loss=0.2566, pruned_loss=0.03252, over 16336.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.2569, pruned_loss=0.03292, over 3070070.46 frames. ], batch size: 165, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:09:29,881 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=293671.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:10:18,284 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.101e+02 2.359e+02 2.998e+02 5.430e+02, threshold=4.718e+02, percent-clipped=2.0 2023-05-02 20:10:34,114 INFO [train.py:904] (6/8) Epoch 29, batch 9500, loss[loss=0.1407, simple_loss=0.24, pruned_loss=0.02069, over 16852.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.256, pruned_loss=0.03251, over 3063331.42 frames. ], batch size: 102, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:11:02,470 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=293717.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:11:08,985 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=293720.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:11:31,660 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-02 20:11:38,330 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1544, 2.5697, 2.5646, 1.9828, 2.7555, 2.8247, 2.5419, 2.5258], device='cuda:6'), covar=tensor([0.0602, 0.0252, 0.0252, 0.0940, 0.0124, 0.0226, 0.0438, 0.0401], device='cuda:6'), in_proj_covar=tensor([0.0144, 0.0108, 0.0099, 0.0135, 0.0084, 0.0127, 0.0126, 0.0126], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 20:11:41,955 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=293737.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:12:00,068 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0623, 4.1385, 3.9502, 3.6316, 3.7192, 4.0437, 3.7063, 3.8469], device='cuda:6'), covar=tensor([0.0669, 0.0878, 0.0410, 0.0371, 0.0731, 0.0763, 0.1195, 0.0612], device='cuda:6'), in_proj_covar=tensor([0.0302, 0.0455, 0.0353, 0.0353, 0.0348, 0.0405, 0.0243, 0.0420], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 20:12:19,483 INFO [train.py:904] (6/8) Epoch 29, batch 9550, loss[loss=0.1759, simple_loss=0.2787, pruned_loss=0.03657, over 16338.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2568, pruned_loss=0.0329, over 3059914.42 frames. ], batch size: 146, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:12:53,431 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6841, 2.6773, 1.7809, 2.7969, 2.1502, 2.8022, 2.0411, 2.3418], device='cuda:6'), covar=tensor([0.0358, 0.0363, 0.1480, 0.0334, 0.0706, 0.0514, 0.1485, 0.0661], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0175, 0.0191, 0.0167, 0.0176, 0.0213, 0.0201, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:12:54,870 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6941, 2.6663, 1.9115, 2.7968, 2.1684, 2.8261, 2.1753, 2.3812], device='cuda:6'), covar=tensor([0.0359, 0.0415, 0.1434, 0.0330, 0.0757, 0.0562, 0.1349, 0.0694], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0175, 0.0191, 0.0167, 0.0176, 0.0213, 0.0201, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:12:59,206 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=293773.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:13:25,491 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=293785.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:13:25,596 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0130, 3.8639, 4.0694, 4.1721, 4.2679, 3.8627, 4.2550, 4.3204], device='cuda:6'), covar=tensor([0.1739, 0.1162, 0.1334, 0.0720, 0.0595, 0.1638, 0.0741, 0.0701], device='cuda:6'), in_proj_covar=tensor([0.0655, 0.0799, 0.0917, 0.0817, 0.0621, 0.0640, 0.0681, 0.0791], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 20:13:28,634 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.05 vs. limit=5.0 2023-05-02 20:13:43,639 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.327e+02 2.273e+02 2.766e+02 3.522e+02 6.419e+02, threshold=5.532e+02, percent-clipped=5.0 2023-05-02 20:13:59,133 INFO [train.py:904] (6/8) Epoch 29, batch 9600, loss[loss=0.1886, simple_loss=0.2852, pruned_loss=0.04601, over 15505.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2589, pruned_loss=0.03393, over 3061254.79 frames. ], batch size: 191, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:14:32,482 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=293821.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:15:45,742 INFO [train.py:904] (6/8) Epoch 29, batch 9650, loss[loss=0.1641, simple_loss=0.261, pruned_loss=0.03361, over 16837.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2606, pruned_loss=0.03422, over 3061818.55 frames. ], batch size: 116, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:16:38,127 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9213, 2.3553, 2.2461, 3.6051, 1.7444, 3.5583, 1.7063, 2.6379], device='cuda:6'), covar=tensor([0.1330, 0.0977, 0.1433, 0.0211, 0.0124, 0.0565, 0.1685, 0.0992], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0179, 0.0198, 0.0199, 0.0201, 0.0214, 0.0209, 0.0197], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:17:16,474 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.571e+02 2.271e+02 2.644e+02 3.266e+02 5.877e+02, threshold=5.288e+02, percent-clipped=1.0 2023-05-02 20:17:35,789 INFO [train.py:904] (6/8) Epoch 29, batch 9700, loss[loss=0.1575, simple_loss=0.2552, pruned_loss=0.02987, over 16230.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2601, pruned_loss=0.03422, over 3069348.63 frames. ], batch size: 165, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:18:08,963 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8652, 3.2243, 3.4848, 2.0596, 2.9420, 2.2904, 3.3769, 3.4916], device='cuda:6'), covar=tensor([0.0288, 0.0859, 0.0580, 0.2336, 0.0845, 0.1075, 0.0655, 0.0941], device='cuda:6'), in_proj_covar=tensor([0.0157, 0.0166, 0.0167, 0.0154, 0.0145, 0.0130, 0.0142, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:19:17,209 INFO [train.py:904] (6/8) Epoch 29, batch 9750, loss[loss=0.1558, simple_loss=0.2555, pruned_loss=0.02804, over 16411.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.2589, pruned_loss=0.03425, over 3065877.46 frames. ], batch size: 166, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:19:21,692 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.09 vs. limit=2.0 2023-05-02 20:20:17,790 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4835, 3.5744, 3.1237, 5.4586, 4.2563, 4.6425, 2.0713, 3.7507], device='cuda:6'), covar=tensor([0.1114, 0.0640, 0.1061, 0.0123, 0.0228, 0.0344, 0.1518, 0.0570], device='cuda:6'), in_proj_covar=tensor([0.0171, 0.0178, 0.0197, 0.0198, 0.0201, 0.0213, 0.0208, 0.0196], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:20:25,317 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=293987.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:20:38,046 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.578e+02 2.282e+02 2.658e+02 3.259e+02 1.207e+03, threshold=5.315e+02, percent-clipped=3.0 2023-05-02 20:20:56,835 INFO [train.py:904] (6/8) Epoch 29, batch 9800, loss[loss=0.1462, simple_loss=0.2519, pruned_loss=0.02022, over 16812.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2584, pruned_loss=0.03338, over 3077146.82 frames. ], batch size: 83, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:21:22,945 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=294017.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:21:27,499 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=294020.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:21:40,160 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8798, 4.2819, 3.0747, 2.4394, 2.7741, 2.7207, 4.6596, 3.6426], device='cuda:6'), covar=tensor([0.3021, 0.0492, 0.1921, 0.3156, 0.2829, 0.2115, 0.0321, 0.1288], device='cuda:6'), in_proj_covar=tensor([0.0329, 0.0268, 0.0307, 0.0322, 0.0297, 0.0273, 0.0298, 0.0342], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 20:21:58,265 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6384, 3.9489, 3.9429, 2.7750, 3.4566, 3.9805, 3.6489, 2.4131], device='cuda:6'), covar=tensor([0.0498, 0.0050, 0.0053, 0.0393, 0.0126, 0.0080, 0.0090, 0.0480], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0088, 0.0090, 0.0133, 0.0101, 0.0114, 0.0097, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-02 20:22:29,378 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=294048.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:22:39,768 INFO [train.py:904] (6/8) Epoch 29, batch 9850, loss[loss=0.1591, simple_loss=0.2607, pruned_loss=0.0287, over 16863.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2592, pruned_loss=0.03258, over 3087107.04 frames. ], batch size: 96, lr: 2.30e-03, grad_scale: 16.0 2023-05-02 20:23:02,393 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=294065.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:23:09,658 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=294068.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:23:27,298 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-05-02 20:24:14,911 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.054e+02 2.475e+02 2.961e+02 4.915e+02, threshold=4.950e+02, percent-clipped=0.0 2023-05-02 20:24:32,429 INFO [train.py:904] (6/8) Epoch 29, batch 9900, loss[loss=0.1643, simple_loss=0.2488, pruned_loss=0.0399, over 12188.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2597, pruned_loss=0.0327, over 3082150.77 frames. ], batch size: 246, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:25:38,575 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9744, 2.6322, 2.9523, 2.1120, 2.7268, 2.1791, 2.6552, 2.8529], device='cuda:6'), covar=tensor([0.0306, 0.1045, 0.0464, 0.1934, 0.0773, 0.0921, 0.0586, 0.0929], device='cuda:6'), in_proj_covar=tensor([0.0156, 0.0164, 0.0166, 0.0153, 0.0144, 0.0129, 0.0141, 0.0177], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:26:27,773 INFO [train.py:904] (6/8) Epoch 29, batch 9950, loss[loss=0.1509, simple_loss=0.2542, pruned_loss=0.02382, over 16652.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2619, pruned_loss=0.03335, over 3077926.41 frames. ], batch size: 89, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:27:07,138 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=294170.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:27:25,152 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-05-02 20:28:08,137 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.053e+02 2.347e+02 2.830e+02 4.908e+02, threshold=4.695e+02, percent-clipped=0.0 2023-05-02 20:28:27,678 INFO [train.py:904] (6/8) Epoch 29, batch 10000, loss[loss=0.1706, simple_loss=0.2702, pruned_loss=0.03548, over 17029.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2604, pruned_loss=0.03282, over 3091412.91 frames. ], batch size: 109, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:29:21,849 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=294231.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:29:39,421 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 20:29:48,897 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-05-02 20:30:08,074 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-02 20:30:08,528 INFO [train.py:904] (6/8) Epoch 29, batch 10050, loss[loss=0.1696, simple_loss=0.2708, pruned_loss=0.03423, over 15257.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2603, pruned_loss=0.03259, over 3071510.68 frames. ], batch size: 191, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:31:27,887 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.622e+02 2.082e+02 2.498e+02 3.055e+02 5.920e+02, threshold=4.997e+02, percent-clipped=2.0 2023-05-02 20:31:41,186 INFO [train.py:904] (6/8) Epoch 29, batch 10100, loss[loss=0.1613, simple_loss=0.2514, pruned_loss=0.03558, over 16891.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2608, pruned_loss=0.03293, over 3093155.40 frames. ], batch size: 116, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:32:42,415 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=294339.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:32:46,856 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=294343.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:32:51,066 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9246, 5.2325, 5.0588, 5.0164, 4.7506, 4.6833, 4.5863, 5.3036], device='cuda:6'), covar=tensor([0.1221, 0.0845, 0.0934, 0.0796, 0.0812, 0.1092, 0.1266, 0.0873], device='cuda:6'), in_proj_covar=tensor([0.0696, 0.0839, 0.0690, 0.0653, 0.0537, 0.0536, 0.0703, 0.0659], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 20:33:20,746 INFO [train.py:904] (6/8) Epoch 30, batch 0, loss[loss=0.2024, simple_loss=0.2771, pruned_loss=0.06383, over 16840.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2771, pruned_loss=0.06383, over 16840.00 frames. ], batch size: 96, lr: 2.26e-03, grad_scale: 8.0 2023-05-02 20:33:20,747 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 20:33:28,214 INFO [train.py:938] (6/8) Epoch 30, validation: loss=0.1426, simple_loss=0.2458, pruned_loss=0.01974, over 944034.00 frames. 2023-05-02 20:33:28,214 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 20:34:10,663 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7986, 2.7564, 2.5367, 4.2927, 3.3178, 4.0824, 1.6811, 3.0873], device='cuda:6'), covar=tensor([0.1548, 0.0752, 0.1382, 0.0180, 0.0176, 0.0483, 0.1782, 0.0848], device='cuda:6'), in_proj_covar=tensor([0.0172, 0.0178, 0.0198, 0.0199, 0.0200, 0.0214, 0.0209, 0.0196], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:34:28,807 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 2.426e+02 2.887e+02 3.382e+02 7.415e+02, threshold=5.774e+02, percent-clipped=4.0 2023-05-02 20:34:29,671 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=294400.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:34:36,218 INFO [train.py:904] (6/8) Epoch 30, batch 50, loss[loss=0.1907, simple_loss=0.2637, pruned_loss=0.05879, over 16324.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2642, pruned_loss=0.04432, over 747911.07 frames. ], batch size: 145, lr: 2.26e-03, grad_scale: 1.0 2023-05-02 20:35:02,165 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=294423.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:35:22,463 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8815, 2.0638, 2.4202, 2.7260, 2.7968, 2.8211, 2.0765, 3.0385], device='cuda:6'), covar=tensor([0.0250, 0.0569, 0.0427, 0.0377, 0.0385, 0.0330, 0.0662, 0.0231], device='cuda:6'), in_proj_covar=tensor([0.0193, 0.0194, 0.0183, 0.0187, 0.0206, 0.0162, 0.0200, 0.0163], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 20:35:43,885 INFO [train.py:904] (6/8) Epoch 30, batch 100, loss[loss=0.1609, simple_loss=0.2556, pruned_loss=0.03313, over 17082.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2627, pruned_loss=0.04284, over 1311683.74 frames. ], batch size: 55, lr: 2.26e-03, grad_scale: 1.0 2023-05-02 20:35:48,586 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2829, 5.3140, 5.1389, 4.7628, 5.1158, 2.1996, 4.9747, 5.0780], device='cuda:6'), covar=tensor([0.0093, 0.0091, 0.0241, 0.0331, 0.0114, 0.2461, 0.0130, 0.0206], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0170, 0.0208, 0.0178, 0.0186, 0.0215, 0.0196, 0.0175], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 20:36:24,555 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=294484.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:36:32,257 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=294488.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:36:46,207 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.392e+02 2.778e+02 3.445e+02 1.255e+03, threshold=5.557e+02, percent-clipped=5.0 2023-05-02 20:36:51,463 INFO [train.py:904] (6/8) Epoch 30, batch 150, loss[loss=0.1674, simple_loss=0.2604, pruned_loss=0.03726, over 17201.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2604, pruned_loss=0.04147, over 1750463.60 frames. ], batch size: 46, lr: 2.25e-03, grad_scale: 1.0 2023-05-02 20:37:20,775 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=294526.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:37:32,834 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-05-02 20:37:50,455 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8550, 3.9799, 2.5081, 4.6116, 3.0846, 4.5591, 2.6842, 3.4108], device='cuda:6'), covar=tensor([0.0365, 0.0415, 0.1745, 0.0344, 0.0958, 0.0562, 0.1610, 0.0777], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0179, 0.0194, 0.0170, 0.0178, 0.0217, 0.0204, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:37:53,395 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=294549.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:37:59,953 INFO [train.py:904] (6/8) Epoch 30, batch 200, loss[loss=0.1957, simple_loss=0.2708, pruned_loss=0.06029, over 16708.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2601, pruned_loss=0.04166, over 2099079.36 frames. ], batch size: 124, lr: 2.25e-03, grad_scale: 1.0 2023-05-02 20:38:23,406 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=294571.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:38:41,539 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0880, 3.0731, 2.0307, 3.2458, 2.5126, 3.3127, 2.2154, 2.6777], device='cuda:6'), covar=tensor([0.0366, 0.0478, 0.1646, 0.0364, 0.0827, 0.0694, 0.1474, 0.0784], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0179, 0.0194, 0.0171, 0.0179, 0.0217, 0.0204, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:39:01,688 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.189e+02 2.574e+02 3.148e+02 2.009e+03, threshold=5.148e+02, percent-clipped=2.0 2023-05-02 20:39:06,655 INFO [train.py:904] (6/8) Epoch 30, batch 250, loss[loss=0.1652, simple_loss=0.2669, pruned_loss=0.03175, over 17120.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2578, pruned_loss=0.04105, over 2374339.45 frames. ], batch size: 49, lr: 2.25e-03, grad_scale: 1.0 2023-05-02 20:39:45,452 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=294632.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:40:01,623 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=294643.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:40:02,962 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1405, 5.7606, 5.8848, 5.5354, 5.6180, 6.2057, 5.6356, 5.3316], device='cuda:6'), covar=tensor([0.1040, 0.2091, 0.2272, 0.2221, 0.2659, 0.0973, 0.1627, 0.2340], device='cuda:6'), in_proj_covar=tensor([0.0421, 0.0634, 0.0707, 0.0513, 0.0686, 0.0726, 0.0544, 0.0681], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 20:40:10,438 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3986, 5.3515, 5.2128, 4.6835, 4.9026, 5.2878, 5.3239, 4.8810], device='cuda:6'), covar=tensor([0.0635, 0.0555, 0.0408, 0.0405, 0.1171, 0.0590, 0.0254, 0.0846], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0464, 0.0360, 0.0360, 0.0354, 0.0413, 0.0248, 0.0429], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 20:40:15,313 INFO [train.py:904] (6/8) Epoch 30, batch 300, loss[loss=0.1549, simple_loss=0.2506, pruned_loss=0.02955, over 17252.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2547, pruned_loss=0.03923, over 2586232.10 frames. ], batch size: 52, lr: 2.25e-03, grad_scale: 1.0 2023-05-02 20:40:59,144 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9200, 4.4728, 4.4680, 3.1962, 3.7300, 4.4143, 4.0058, 2.6768], device='cuda:6'), covar=tensor([0.0539, 0.0088, 0.0054, 0.0388, 0.0162, 0.0116, 0.0115, 0.0504], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0091, 0.0092, 0.0136, 0.0104, 0.0116, 0.0099, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 20:41:06,778 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=294691.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:41:12,841 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=294695.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:41:19,992 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.370e+02 1.963e+02 2.186e+02 2.662e+02 5.332e+02, threshold=4.373e+02, percent-clipped=1.0 2023-05-02 20:41:25,715 INFO [train.py:904] (6/8) Epoch 30, batch 350, loss[loss=0.1789, simple_loss=0.2564, pruned_loss=0.05071, over 16794.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2528, pruned_loss=0.03883, over 2748155.23 frames. ], batch size: 83, lr: 2.25e-03, grad_scale: 1.0 2023-05-02 20:42:33,013 INFO [train.py:904] (6/8) Epoch 30, batch 400, loss[loss=0.1545, simple_loss=0.2465, pruned_loss=0.03125, over 17074.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2514, pruned_loss=0.03877, over 2878433.14 frames. ], batch size: 50, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:42:55,443 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6253, 3.6014, 4.2921, 2.3023, 3.3903, 2.7307, 3.9810, 3.8575], device='cuda:6'), covar=tensor([0.0215, 0.1067, 0.0445, 0.2121, 0.0767, 0.0969, 0.0583, 0.1145], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0168, 0.0169, 0.0157, 0.0147, 0.0132, 0.0144, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:43:06,963 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=294779.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:43:34,529 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.286e+02 2.158e+02 2.521e+02 2.985e+02 1.602e+03, threshold=5.041e+02, percent-clipped=2.0 2023-05-02 20:43:41,853 INFO [train.py:904] (6/8) Epoch 30, batch 450, loss[loss=0.159, simple_loss=0.2389, pruned_loss=0.03952, over 16762.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2504, pruned_loss=0.03857, over 2978465.85 frames. ], batch size: 134, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:43:50,104 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2611, 3.4952, 3.4672, 2.0374, 2.8549, 2.1105, 3.6631, 3.7568], device='cuda:6'), covar=tensor([0.0308, 0.1057, 0.0863, 0.2711, 0.1178, 0.1479, 0.0739, 0.1285], device='cuda:6'), in_proj_covar=tensor([0.0160, 0.0169, 0.0170, 0.0157, 0.0147, 0.0132, 0.0145, 0.0181], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:44:12,096 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=294826.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:44:16,199 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0322, 3.9234, 4.0978, 4.2116, 4.2639, 3.8454, 4.1034, 4.2788], device='cuda:6'), covar=tensor([0.1634, 0.1182, 0.1221, 0.0669, 0.0614, 0.1654, 0.2902, 0.0749], device='cuda:6'), in_proj_covar=tensor([0.0679, 0.0828, 0.0954, 0.0843, 0.0641, 0.0664, 0.0704, 0.0817], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 20:44:36,121 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=294844.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:44:49,592 INFO [train.py:904] (6/8) Epoch 30, batch 500, loss[loss=0.1673, simple_loss=0.2655, pruned_loss=0.03456, over 16734.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2484, pruned_loss=0.03772, over 3059542.95 frames. ], batch size: 57, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:45:17,742 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=294874.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:45:52,584 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.499e+02 2.033e+02 2.399e+02 2.919e+02 5.551e+02, threshold=4.798e+02, percent-clipped=2.0 2023-05-02 20:45:57,299 INFO [train.py:904] (6/8) Epoch 30, batch 550, loss[loss=0.137, simple_loss=0.225, pruned_loss=0.02447, over 17230.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2476, pruned_loss=0.03746, over 3116848.31 frames. ], batch size: 45, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:46:29,053 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=294927.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:47:00,794 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 20:47:05,461 INFO [train.py:904] (6/8) Epoch 30, batch 600, loss[loss=0.1478, simple_loss=0.2257, pruned_loss=0.03494, over 16809.00 frames. ], tot_loss[loss=0.1606, simple_loss=0.2464, pruned_loss=0.03739, over 3160062.57 frames. ], batch size: 83, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:47:18,804 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9141, 1.9237, 2.4431, 2.7757, 2.7260, 2.6931, 1.9561, 2.9455], device='cuda:6'), covar=tensor([0.0225, 0.0615, 0.0399, 0.0368, 0.0369, 0.0356, 0.0674, 0.0247], device='cuda:6'), in_proj_covar=tensor([0.0199, 0.0200, 0.0188, 0.0193, 0.0212, 0.0167, 0.0206, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 20:47:53,601 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=294989.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:48:01,952 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=294995.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:48:07,356 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.157e+02 2.572e+02 3.014e+02 4.443e+02, threshold=5.143e+02, percent-clipped=0.0 2023-05-02 20:48:13,113 INFO [train.py:904] (6/8) Epoch 30, batch 650, loss[loss=0.1481, simple_loss=0.2428, pruned_loss=0.02673, over 17093.00 frames. ], tot_loss[loss=0.1593, simple_loss=0.2451, pruned_loss=0.0367, over 3186081.21 frames. ], batch size: 49, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:48:41,993 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8481, 5.1716, 4.9780, 4.9575, 4.6675, 4.6593, 4.5710, 5.2653], device='cuda:6'), covar=tensor([0.1241, 0.0900, 0.1012, 0.0953, 0.0885, 0.1108, 0.1404, 0.0922], device='cuda:6'), in_proj_covar=tensor([0.0723, 0.0868, 0.0716, 0.0677, 0.0555, 0.0553, 0.0732, 0.0684], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 20:49:04,985 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295043.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:49:15,941 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295050.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 20:49:20,778 INFO [train.py:904] (6/8) Epoch 30, batch 700, loss[loss=0.1922, simple_loss=0.269, pruned_loss=0.05769, over 16744.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.2456, pruned_loss=0.03669, over 3224166.17 frames. ], batch size: 124, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:49:39,358 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295068.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:49:54,481 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295079.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:50:03,904 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-02 20:50:23,317 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.300e+02 2.027e+02 2.322e+02 2.790e+02 6.383e+02, threshold=4.644e+02, percent-clipped=2.0 2023-05-02 20:50:27,888 INFO [train.py:904] (6/8) Epoch 30, batch 750, loss[loss=0.1377, simple_loss=0.2253, pruned_loss=0.02504, over 17028.00 frames. ], tot_loss[loss=0.1594, simple_loss=0.2461, pruned_loss=0.03632, over 3253958.58 frames. ], batch size: 41, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:50:56,721 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5792, 3.6622, 3.9066, 2.6954, 3.5697, 3.9690, 3.6418, 2.3438], device='cuda:6'), covar=tensor([0.0579, 0.0347, 0.0075, 0.0469, 0.0134, 0.0130, 0.0121, 0.0562], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0092, 0.0093, 0.0137, 0.0104, 0.0117, 0.0100, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 20:51:00,101 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295127.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:51:03,373 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295129.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:51:24,558 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295144.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:51:38,201 INFO [train.py:904] (6/8) Epoch 30, batch 800, loss[loss=0.1545, simple_loss=0.2433, pruned_loss=0.03289, over 16811.00 frames. ], tot_loss[loss=0.1599, simple_loss=0.2466, pruned_loss=0.03663, over 3277023.28 frames. ], batch size: 39, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:51:39,882 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6154, 1.8121, 2.2753, 2.4567, 2.5690, 2.5766, 2.0159, 2.6761], device='cuda:6'), covar=tensor([0.0223, 0.0559, 0.0369, 0.0337, 0.0365, 0.0353, 0.0567, 0.0243], device='cuda:6'), in_proj_covar=tensor([0.0199, 0.0199, 0.0187, 0.0193, 0.0212, 0.0168, 0.0205, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 20:51:49,495 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.38 vs. limit=5.0 2023-05-02 20:51:51,569 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6634, 3.6978, 2.4078, 3.9541, 3.0793, 3.9084, 2.4410, 3.0635], device='cuda:6'), covar=tensor([0.0304, 0.0441, 0.1636, 0.0467, 0.0784, 0.0830, 0.1588, 0.0802], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0183, 0.0197, 0.0175, 0.0181, 0.0222, 0.0207, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:52:27,571 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0646, 3.1220, 3.1858, 2.1655, 3.0195, 3.2771, 3.0538, 2.0099], device='cuda:6'), covar=tensor([0.0602, 0.0134, 0.0087, 0.0479, 0.0154, 0.0143, 0.0122, 0.0531], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0091, 0.0092, 0.0137, 0.0104, 0.0117, 0.0100, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 20:52:29,870 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295192.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:52:42,510 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.377e+02 2.068e+02 2.393e+02 2.794e+02 7.464e+02, threshold=4.786e+02, percent-clipped=2.0 2023-05-02 20:52:45,339 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4381, 3.5951, 4.0697, 2.2576, 3.1951, 2.3202, 3.9506, 3.8939], device='cuda:6'), covar=tensor([0.0260, 0.1022, 0.0514, 0.2150, 0.0879, 0.1148, 0.0554, 0.0992], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0170, 0.0171, 0.0158, 0.0148, 0.0133, 0.0145, 0.0183], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 20:52:48,745 INFO [train.py:904] (6/8) Epoch 30, batch 850, loss[loss=0.1531, simple_loss=0.2368, pruned_loss=0.03469, over 16425.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.2463, pruned_loss=0.03634, over 3293206.19 frames. ], batch size: 75, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:53:18,791 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295225.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:53:21,294 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295227.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:53:56,734 INFO [train.py:904] (6/8) Epoch 30, batch 900, loss[loss=0.1365, simple_loss=0.2206, pruned_loss=0.02615, over 16994.00 frames. ], tot_loss[loss=0.1583, simple_loss=0.245, pruned_loss=0.0358, over 3299835.97 frames. ], batch size: 41, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:54:27,486 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295275.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:54:31,266 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2540, 4.2094, 4.1567, 3.8674, 3.9626, 4.2410, 3.8706, 4.0373], device='cuda:6'), covar=tensor([0.0650, 0.0791, 0.0337, 0.0301, 0.0697, 0.0467, 0.0990, 0.0594], device='cuda:6'), in_proj_covar=tensor([0.0320, 0.0483, 0.0373, 0.0376, 0.0370, 0.0432, 0.0258, 0.0448], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 20:54:34,082 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295279.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:54:41,833 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6904, 4.3627, 4.3905, 2.9918, 3.6114, 4.3590, 3.9020, 2.6979], device='cuda:6'), covar=tensor([0.0574, 0.0079, 0.0055, 0.0434, 0.0164, 0.0107, 0.0110, 0.0483], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0091, 0.0092, 0.0136, 0.0104, 0.0117, 0.0099, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 20:54:43,655 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295286.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:55:02,819 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.063e+02 2.465e+02 2.923e+02 4.814e+02, threshold=4.929e+02, percent-clipped=1.0 2023-05-02 20:55:08,080 INFO [train.py:904] (6/8) Epoch 30, batch 950, loss[loss=0.1759, simple_loss=0.2517, pruned_loss=0.05009, over 12218.00 frames. ], tot_loss[loss=0.159, simple_loss=0.2454, pruned_loss=0.03631, over 3303754.08 frames. ], batch size: 248, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:55:48,100 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3392, 3.3926, 3.7093, 2.3844, 3.1937, 2.5559, 3.8259, 3.7645], device='cuda:6'), covar=tensor([0.0255, 0.1017, 0.0600, 0.2043, 0.0871, 0.0998, 0.0526, 0.1020], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0171, 0.0171, 0.0158, 0.0149, 0.0133, 0.0146, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 20:55:49,182 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295333.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:55:59,238 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295340.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:56:06,375 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295345.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 20:56:17,470 INFO [train.py:904] (6/8) Epoch 30, batch 1000, loss[loss=0.1711, simple_loss=0.266, pruned_loss=0.03816, over 17109.00 frames. ], tot_loss[loss=0.158, simple_loss=0.244, pruned_loss=0.03605, over 3305262.78 frames. ], batch size: 53, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:56:17,938 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8189, 4.0697, 2.2229, 4.6148, 3.0830, 4.4892, 2.3508, 3.2740], device='cuda:6'), covar=tensor([0.0403, 0.0434, 0.2130, 0.0286, 0.0959, 0.0548, 0.2153, 0.0849], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0183, 0.0197, 0.0176, 0.0181, 0.0223, 0.0207, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 20:56:24,313 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295359.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:56:39,487 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-02 20:56:53,390 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-02 20:57:12,869 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295394.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:57:20,047 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.691e+02 2.023e+02 2.384e+02 2.929e+02 5.067e+02, threshold=4.769e+02, percent-clipped=1.0 2023-05-02 20:57:26,454 INFO [train.py:904] (6/8) Epoch 30, batch 1050, loss[loss=0.1493, simple_loss=0.2413, pruned_loss=0.02862, over 17089.00 frames. ], tot_loss[loss=0.1579, simple_loss=0.2437, pruned_loss=0.03606, over 3301690.70 frames. ], batch size: 47, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:57:48,337 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295420.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:57:54,256 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295424.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:58:36,207 INFO [train.py:904] (6/8) Epoch 30, batch 1100, loss[loss=0.1635, simple_loss=0.2512, pruned_loss=0.03787, over 17182.00 frames. ], tot_loss[loss=0.1573, simple_loss=0.2431, pruned_loss=0.03577, over 3312068.71 frames. ], batch size: 44, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:59:32,073 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1087, 4.5711, 4.6064, 3.4142, 3.8353, 4.5213, 4.0603, 2.8786], device='cuda:6'), covar=tensor([0.0490, 0.0062, 0.0044, 0.0351, 0.0146, 0.0107, 0.0094, 0.0460], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0091, 0.0092, 0.0136, 0.0104, 0.0117, 0.0099, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 20:59:33,185 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9383, 5.0398, 5.4001, 5.4111, 5.4378, 5.1043, 5.0121, 4.8803], device='cuda:6'), covar=tensor([0.0407, 0.0598, 0.0539, 0.0465, 0.0495, 0.0511, 0.1119, 0.0511], device='cuda:6'), in_proj_covar=tensor([0.0446, 0.0509, 0.0488, 0.0450, 0.0535, 0.0517, 0.0592, 0.0415], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 20:59:38,543 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.377e+02 2.061e+02 2.413e+02 2.897e+02 1.352e+03, threshold=4.827e+02, percent-clipped=9.0 2023-05-02 20:59:43,305 INFO [train.py:904] (6/8) Epoch 30, batch 1150, loss[loss=0.1781, simple_loss=0.2691, pruned_loss=0.04361, over 16719.00 frames. ], tot_loss[loss=0.1576, simple_loss=0.2436, pruned_loss=0.03578, over 3314185.47 frames. ], batch size: 57, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:00:30,548 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.6989, 3.7541, 2.2906, 4.1249, 3.0465, 4.1063, 2.4564, 3.0864], device='cuda:6'), covar=tensor([0.0350, 0.0480, 0.1919, 0.0458, 0.0837, 0.0670, 0.1661, 0.0849], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0182, 0.0197, 0.0175, 0.0181, 0.0222, 0.0207, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 21:00:41,675 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.02 vs. limit=5.0 2023-05-02 21:00:52,295 INFO [train.py:904] (6/8) Epoch 30, batch 1200, loss[loss=0.1456, simple_loss=0.238, pruned_loss=0.02656, over 16754.00 frames. ], tot_loss[loss=0.1563, simple_loss=0.2426, pruned_loss=0.03497, over 3320481.46 frames. ], batch size: 57, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:00:55,638 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295556.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:00:58,029 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295558.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:01:29,883 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295581.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:01:55,309 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.025e+02 2.431e+02 2.981e+02 5.369e+02, threshold=4.863e+02, percent-clipped=2.0 2023-05-02 21:02:00,804 INFO [train.py:904] (6/8) Epoch 30, batch 1250, loss[loss=0.1578, simple_loss=0.2594, pruned_loss=0.02808, over 17131.00 frames. ], tot_loss[loss=0.1564, simple_loss=0.243, pruned_loss=0.03486, over 3320196.05 frames. ], batch size: 48, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:02:19,221 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295617.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 21:02:21,434 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295619.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 21:02:34,300 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295629.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:02:39,650 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8798, 2.8835, 2.4661, 2.8107, 3.1634, 2.9147, 3.5017, 3.4413], device='cuda:6'), covar=tensor([0.0199, 0.0588, 0.0733, 0.0540, 0.0399, 0.0491, 0.0362, 0.0376], device='cuda:6'), in_proj_covar=tensor([0.0239, 0.0252, 0.0239, 0.0240, 0.0250, 0.0249, 0.0247, 0.0250], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:02:43,115 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.6722, 4.7995, 4.9698, 4.7520, 4.7999, 5.3959, 4.8792, 4.5484], device='cuda:6'), covar=tensor([0.1771, 0.2326, 0.2656, 0.2452, 0.2636, 0.1225, 0.2035, 0.2919], device='cuda:6'), in_proj_covar=tensor([0.0435, 0.0653, 0.0727, 0.0528, 0.0704, 0.0745, 0.0558, 0.0702], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 21:02:43,124 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295635.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:02:46,355 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5623, 2.3445, 1.8515, 2.0977, 2.6669, 2.3884, 2.5541, 2.7372], device='cuda:6'), covar=tensor([0.0323, 0.0481, 0.0663, 0.0592, 0.0288, 0.0437, 0.0252, 0.0362], device='cuda:6'), in_proj_covar=tensor([0.0239, 0.0252, 0.0239, 0.0240, 0.0250, 0.0249, 0.0247, 0.0250], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:02:58,103 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295645.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:03:08,457 INFO [train.py:904] (6/8) Epoch 30, batch 1300, loss[loss=0.1512, simple_loss=0.2399, pruned_loss=0.0312, over 17224.00 frames. ], tot_loss[loss=0.1564, simple_loss=0.2429, pruned_loss=0.03497, over 3326436.03 frames. ], batch size: 43, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:03:52,684 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295685.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:03:57,767 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295689.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:03:59,091 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295690.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:04:02,885 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295693.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:04:11,929 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.150e+02 2.443e+02 2.917e+02 6.484e+02, threshold=4.887e+02, percent-clipped=2.0 2023-05-02 21:04:18,058 INFO [train.py:904] (6/8) Epoch 30, batch 1350, loss[loss=0.1351, simple_loss=0.2224, pruned_loss=0.02389, over 16859.00 frames. ], tot_loss[loss=0.1565, simple_loss=0.2431, pruned_loss=0.03494, over 3332378.33 frames. ], batch size: 42, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:04:33,147 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295715.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:04:38,782 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3568, 5.3008, 5.0780, 4.3912, 5.1419, 2.0629, 4.8568, 4.8404], device='cuda:6'), covar=tensor([0.0108, 0.0099, 0.0248, 0.0490, 0.0128, 0.2893, 0.0172, 0.0296], device='cuda:6'), in_proj_covar=tensor([0.0185, 0.0177, 0.0215, 0.0186, 0.0193, 0.0221, 0.0204, 0.0182], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:04:46,371 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295724.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:05:17,165 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295746.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:05:26,448 INFO [train.py:904] (6/8) Epoch 30, batch 1400, loss[loss=0.1742, simple_loss=0.2498, pruned_loss=0.04931, over 16472.00 frames. ], tot_loss[loss=0.1568, simple_loss=0.2437, pruned_loss=0.0349, over 3332300.37 frames. ], batch size: 146, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:05:52,150 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295772.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:06:07,227 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5213, 4.3790, 4.4423, 4.1077, 4.2275, 4.4683, 4.2512, 4.2482], device='cuda:6'), covar=tensor([0.0665, 0.0978, 0.0348, 0.0342, 0.0766, 0.0520, 0.0553, 0.0692], device='cuda:6'), in_proj_covar=tensor([0.0324, 0.0490, 0.0380, 0.0381, 0.0376, 0.0438, 0.0261, 0.0454], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 21:06:30,028 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.296e+02 2.078e+02 2.399e+02 2.795e+02 5.908e+02, threshold=4.799e+02, percent-clipped=2.0 2023-05-02 21:06:35,136 INFO [train.py:904] (6/8) Epoch 30, batch 1450, loss[loss=0.15, simple_loss=0.2388, pruned_loss=0.03066, over 17204.00 frames. ], tot_loss[loss=0.1564, simple_loss=0.243, pruned_loss=0.03493, over 3326933.27 frames. ], batch size: 44, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:06:57,682 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295820.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:07:44,322 INFO [train.py:904] (6/8) Epoch 30, batch 1500, loss[loss=0.147, simple_loss=0.2394, pruned_loss=0.02728, over 17240.00 frames. ], tot_loss[loss=0.1566, simple_loss=0.2428, pruned_loss=0.03524, over 3321687.00 frames. ], batch size: 45, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:08:11,959 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295873.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:08:22,623 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295881.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:08:22,738 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295881.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:08:47,909 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.156e+02 2.462e+02 3.031e+02 5.873e+02, threshold=4.924e+02, percent-clipped=1.0 2023-05-02 21:08:54,105 INFO [train.py:904] (6/8) Epoch 30, batch 1550, loss[loss=0.158, simple_loss=0.2534, pruned_loss=0.03128, over 17138.00 frames. ], tot_loss[loss=0.1575, simple_loss=0.2436, pruned_loss=0.03573, over 3327726.76 frames. ], batch size: 49, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:09:06,211 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295912.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 21:09:08,349 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295914.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 21:09:14,924 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1104, 5.0842, 4.9906, 4.4822, 4.6639, 5.0109, 4.9139, 4.6410], device='cuda:6'), covar=tensor([0.0608, 0.0537, 0.0358, 0.0396, 0.1128, 0.0478, 0.0416, 0.0822], device='cuda:6'), in_proj_covar=tensor([0.0324, 0.0490, 0.0380, 0.0381, 0.0376, 0.0438, 0.0261, 0.0454], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 21:09:26,395 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-05-02 21:09:29,567 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295929.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:09:36,354 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295934.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 21:09:37,396 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295935.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:10:02,945 INFO [train.py:904] (6/8) Epoch 30, batch 1600, loss[loss=0.1663, simple_loss=0.2673, pruned_loss=0.03267, over 16781.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.2462, pruned_loss=0.03662, over 3331532.23 frames. ], batch size: 57, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:10:42,973 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295983.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:10:45,975 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295985.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:10:51,228 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295989.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:11:05,442 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.556e+02 2.204e+02 2.497e+02 2.999e+02 5.396e+02, threshold=4.994e+02, percent-clipped=1.0 2023-05-02 21:11:14,063 INFO [train.py:904] (6/8) Epoch 30, batch 1650, loss[loss=0.1572, simple_loss=0.2477, pruned_loss=0.03333, over 17216.00 frames. ], tot_loss[loss=0.1612, simple_loss=0.2477, pruned_loss=0.03734, over 3321591.18 frames. ], batch size: 44, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:11:30,748 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296015.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:11:55,131 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.33 vs. limit=5.0 2023-05-02 21:12:00,955 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296037.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:12:06,275 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296041.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:12:16,967 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296049.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:12:20,211 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.00 vs. limit=5.0 2023-05-02 21:12:24,440 INFO [train.py:904] (6/8) Epoch 30, batch 1700, loss[loss=0.195, simple_loss=0.2756, pruned_loss=0.05723, over 16836.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2494, pruned_loss=0.03793, over 3327423.16 frames. ], batch size: 116, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:12:36,865 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296063.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:13:17,412 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-05-02 21:13:26,988 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.214e+02 2.660e+02 3.287e+02 5.248e+02, threshold=5.320e+02, percent-clipped=2.0 2023-05-02 21:13:32,338 INFO [train.py:904] (6/8) Epoch 30, batch 1750, loss[loss=0.1552, simple_loss=0.2531, pruned_loss=0.02861, over 17126.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2508, pruned_loss=0.03784, over 3329731.57 frames. ], batch size: 49, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:13:40,795 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296110.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:13:59,431 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-02 21:14:20,711 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2786, 4.6292, 4.8766, 4.8135, 4.9211, 4.5549, 4.3232, 4.4009], device='cuda:6'), covar=tensor([0.0725, 0.0926, 0.0712, 0.0842, 0.0875, 0.0804, 0.1626, 0.0792], device='cuda:6'), in_proj_covar=tensor([0.0448, 0.0510, 0.0488, 0.0452, 0.0537, 0.0519, 0.0593, 0.0416], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 21:14:38,218 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 21:14:41,099 INFO [train.py:904] (6/8) Epoch 30, batch 1800, loss[loss=0.1648, simple_loss=0.2487, pruned_loss=0.04048, over 16695.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2513, pruned_loss=0.03762, over 3333902.19 frames. ], batch size: 134, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:15:12,948 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296176.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:15:19,804 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0153, 3.6946, 4.1045, 2.0986, 4.2439, 4.3347, 3.2754, 3.3014], device='cuda:6'), covar=tensor([0.0705, 0.0300, 0.0240, 0.1246, 0.0115, 0.0217, 0.0430, 0.0478], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0113, 0.0104, 0.0141, 0.0089, 0.0135, 0.0132, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 21:15:38,122 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2027, 1.6171, 2.0202, 2.0982, 2.2744, 2.3271, 1.8283, 2.3656], device='cuda:6'), covar=tensor([0.0280, 0.0626, 0.0372, 0.0421, 0.0431, 0.0395, 0.0655, 0.0235], device='cuda:6'), in_proj_covar=tensor([0.0202, 0.0202, 0.0190, 0.0196, 0.0214, 0.0171, 0.0207, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 21:15:40,689 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2761, 2.4653, 2.4512, 3.9833, 2.2908, 2.7693, 2.4645, 2.5810], device='cuda:6'), covar=tensor([0.1615, 0.3628, 0.3177, 0.0740, 0.4189, 0.2530, 0.3636, 0.3480], device='cuda:6'), in_proj_covar=tensor([0.0430, 0.0483, 0.0394, 0.0345, 0.0451, 0.0554, 0.0456, 0.0567], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:15:47,287 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.165e+02 2.480e+02 2.953e+02 5.408e+02, threshold=4.959e+02, percent-clipped=1.0 2023-05-02 21:15:51,617 INFO [train.py:904] (6/8) Epoch 30, batch 1850, loss[loss=0.1396, simple_loss=0.2293, pruned_loss=0.02497, over 17232.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2524, pruned_loss=0.03784, over 3327119.82 frames. ], batch size: 45, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:16:03,738 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296212.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 21:16:06,137 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0860, 5.0483, 4.7899, 4.1298, 4.9108, 1.8392, 4.6358, 4.4673], device='cuda:6'), covar=tensor([0.0103, 0.0110, 0.0237, 0.0469, 0.0127, 0.3242, 0.0160, 0.0365], device='cuda:6'), in_proj_covar=tensor([0.0188, 0.0180, 0.0218, 0.0189, 0.0196, 0.0224, 0.0208, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:16:06,141 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296214.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 21:16:28,001 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296229.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 21:17:00,509 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-05-02 21:17:02,020 INFO [train.py:904] (6/8) Epoch 30, batch 1900, loss[loss=0.1912, simple_loss=0.2819, pruned_loss=0.05024, over 17095.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2519, pruned_loss=0.03708, over 3318146.73 frames. ], batch size: 55, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:17:09,979 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296260.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:17:13,526 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296262.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:17:46,290 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296285.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 21:18:06,422 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.062e+02 2.316e+02 2.814e+02 5.874e+02, threshold=4.633e+02, percent-clipped=2.0 2023-05-02 21:18:07,500 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 21:18:10,609 INFO [train.py:904] (6/8) Epoch 30, batch 1950, loss[loss=0.16, simple_loss=0.243, pruned_loss=0.03846, over 16547.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2513, pruned_loss=0.03651, over 3325811.28 frames. ], batch size: 75, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:18:26,966 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296315.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:18:52,565 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296333.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:19:02,381 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296341.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:19:21,068 INFO [train.py:904] (6/8) Epoch 30, batch 2000, loss[loss=0.1595, simple_loss=0.2505, pruned_loss=0.03424, over 12264.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2505, pruned_loss=0.03634, over 3322620.15 frames. ], batch size: 247, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:19:32,096 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1323, 5.0819, 4.8687, 4.2678, 4.9490, 1.8859, 4.7077, 4.5983], device='cuda:6'), covar=tensor([0.0110, 0.0113, 0.0220, 0.0442, 0.0116, 0.3201, 0.0150, 0.0296], device='cuda:6'), in_proj_covar=tensor([0.0188, 0.0180, 0.0218, 0.0188, 0.0196, 0.0223, 0.0207, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:19:51,110 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296376.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:20:09,547 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296389.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:20:25,431 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.330e+02 2.133e+02 2.447e+02 2.850e+02 4.858e+02, threshold=4.894e+02, percent-clipped=1.0 2023-05-02 21:20:30,259 INFO [train.py:904] (6/8) Epoch 30, batch 2050, loss[loss=0.1397, simple_loss=0.2251, pruned_loss=0.02713, over 16757.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.25, pruned_loss=0.03728, over 3307800.68 frames. ], batch size: 39, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:20:32,313 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296405.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:20:49,671 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.14 vs. limit=5.0 2023-05-02 21:21:05,255 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-02 21:21:12,413 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296434.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:21:13,499 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9533, 5.1105, 5.4593, 5.4400, 5.4822, 5.1916, 4.9959, 4.9112], device='cuda:6'), covar=tensor([0.0497, 0.0821, 0.0545, 0.0595, 0.0612, 0.0593, 0.1310, 0.0509], device='cuda:6'), in_proj_covar=tensor([0.0448, 0.0511, 0.0487, 0.0452, 0.0535, 0.0518, 0.0594, 0.0415], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 21:21:13,649 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7848, 2.7552, 2.8502, 5.0953, 4.0208, 4.4354, 1.7038, 3.3083], device='cuda:6'), covar=tensor([0.1393, 0.0866, 0.1174, 0.0205, 0.0207, 0.0396, 0.1673, 0.0752], device='cuda:6'), in_proj_covar=tensor([0.0176, 0.0185, 0.0204, 0.0210, 0.0209, 0.0222, 0.0214, 0.0202], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 21:21:37,726 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1601, 5.1291, 4.9136, 4.3266, 4.9690, 1.9447, 4.7384, 4.7221], device='cuda:6'), covar=tensor([0.0099, 0.0099, 0.0230, 0.0419, 0.0115, 0.3038, 0.0141, 0.0262], device='cuda:6'), in_proj_covar=tensor([0.0187, 0.0179, 0.0217, 0.0187, 0.0195, 0.0222, 0.0207, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:21:39,721 INFO [train.py:904] (6/8) Epoch 30, batch 2100, loss[loss=0.1515, simple_loss=0.2537, pruned_loss=0.02462, over 17111.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2512, pruned_loss=0.0377, over 3302002.70 frames. ], batch size: 49, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:22:09,022 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-02 21:22:09,672 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296476.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:22:33,200 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296492.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:22:36,833 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296495.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:22:39,073 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296497.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:22:46,671 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.431e+02 2.088e+02 2.487e+02 2.838e+02 5.578e+02, threshold=4.974e+02, percent-clipped=1.0 2023-05-02 21:22:48,891 INFO [train.py:904] (6/8) Epoch 30, batch 2150, loss[loss=0.1668, simple_loss=0.2678, pruned_loss=0.03286, over 17253.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.253, pruned_loss=0.03807, over 3298918.45 frames. ], batch size: 52, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:22:49,147 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7937, 3.7748, 3.9140, 3.5815, 3.8024, 4.2442, 3.8784, 3.5592], device='cuda:6'), covar=tensor([0.2121, 0.2486, 0.2520, 0.2665, 0.2726, 0.1977, 0.1736, 0.2692], device='cuda:6'), in_proj_covar=tensor([0.0437, 0.0652, 0.0728, 0.0531, 0.0707, 0.0746, 0.0559, 0.0704], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 21:23:03,828 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296514.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:23:16,236 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296524.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:23:23,441 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296529.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 21:23:25,850 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7276, 3.4683, 3.7843, 2.0381, 3.8334, 3.9079, 3.2055, 3.0007], device='cuda:6'), covar=tensor([0.0765, 0.0263, 0.0215, 0.1220, 0.0155, 0.0241, 0.0433, 0.0465], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0112, 0.0104, 0.0140, 0.0089, 0.0135, 0.0132, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 21:23:56,802 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296553.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:23:57,527 INFO [train.py:904] (6/8) Epoch 30, batch 2200, loss[loss=0.1856, simple_loss=0.2614, pruned_loss=0.05492, over 16765.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2535, pruned_loss=0.03867, over 3305885.34 frames. ], batch size: 102, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:24:04,164 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296558.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:24:04,446 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-02 21:24:27,052 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296575.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:24:30,756 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296577.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:24:31,459 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 21:25:04,056 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.512e+02 2.151e+02 2.512e+02 3.134e+02 5.584e+02, threshold=5.024e+02, percent-clipped=2.0 2023-05-02 21:25:06,331 INFO [train.py:904] (6/8) Epoch 30, batch 2250, loss[loss=0.1545, simple_loss=0.2398, pruned_loss=0.03458, over 16733.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2534, pruned_loss=0.03886, over 3308314.43 frames. ], batch size: 134, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:25:20,103 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296613.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:25:24,182 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4256, 4.6912, 4.5083, 4.5340, 4.2855, 4.2001, 4.2692, 4.7447], device='cuda:6'), covar=tensor([0.1218, 0.0854, 0.1080, 0.0877, 0.0799, 0.1568, 0.1098, 0.0886], device='cuda:6'), in_proj_covar=tensor([0.0742, 0.0894, 0.0735, 0.0697, 0.0572, 0.0566, 0.0751, 0.0704], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:25:25,543 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6970, 3.8482, 2.9491, 2.2727, 2.4812, 2.4540, 4.0103, 3.3319], device='cuda:6'), covar=tensor([0.3013, 0.0631, 0.1889, 0.3423, 0.3052, 0.2381, 0.0540, 0.1592], device='cuda:6'), in_proj_covar=tensor([0.0340, 0.0279, 0.0317, 0.0331, 0.0310, 0.0283, 0.0308, 0.0358], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 21:26:17,731 INFO [train.py:904] (6/8) Epoch 30, batch 2300, loss[loss=0.1758, simple_loss=0.2547, pruned_loss=0.04844, over 16895.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.254, pruned_loss=0.03906, over 3302582.54 frames. ], batch size: 96, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:26:41,376 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296671.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:26:45,094 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296674.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:27:05,599 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296688.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:27:24,079 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.359e+02 2.179e+02 2.509e+02 3.030e+02 4.955e+02, threshold=5.019e+02, percent-clipped=0.0 2023-05-02 21:27:27,335 INFO [train.py:904] (6/8) Epoch 30, batch 2350, loss[loss=0.1549, simple_loss=0.2461, pruned_loss=0.03189, over 15843.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2533, pruned_loss=0.03896, over 3312678.78 frames. ], batch size: 35, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:27:28,827 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296705.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:28:30,543 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296749.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:28:36,423 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296753.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:28:37,278 INFO [train.py:904] (6/8) Epoch 30, batch 2400, loss[loss=0.1536, simple_loss=0.2525, pruned_loss=0.02732, over 17018.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2535, pruned_loss=0.03883, over 3322603.70 frames. ], batch size: 53, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:29:27,151 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296790.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:29:43,967 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.201e+02 2.707e+02 3.099e+02 6.549e+02, threshold=5.413e+02, percent-clipped=3.0 2023-05-02 21:29:46,349 INFO [train.py:904] (6/8) Epoch 30, batch 2450, loss[loss=0.1827, simple_loss=0.2701, pruned_loss=0.04759, over 15429.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.254, pruned_loss=0.03831, over 3321633.48 frames. ], batch size: 190, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:30:22,318 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-05-02 21:30:22,569 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-05-02 21:30:42,764 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-02 21:30:48,328 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296848.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:30:53,562 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.6517, 6.0091, 5.7561, 5.8312, 5.4957, 5.4551, 5.4207, 6.1456], device='cuda:6'), covar=tensor([0.1473, 0.0951, 0.1104, 0.0871, 0.0907, 0.0709, 0.1372, 0.0921], device='cuda:6'), in_proj_covar=tensor([0.0742, 0.0893, 0.0736, 0.0695, 0.0571, 0.0565, 0.0750, 0.0703], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:30:55,476 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296853.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:30:56,382 INFO [train.py:904] (6/8) Epoch 30, batch 2500, loss[loss=0.1589, simple_loss=0.2468, pruned_loss=0.03555, over 15970.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.254, pruned_loss=0.03837, over 3319850.26 frames. ], batch size: 35, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:31:19,659 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296870.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:31:24,748 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296873.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:31:56,955 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8997, 2.1352, 2.2582, 3.3935, 2.1486, 2.3501, 2.2470, 2.2553], device='cuda:6'), covar=tensor([0.1706, 0.3940, 0.3381, 0.0873, 0.4337, 0.2840, 0.4102, 0.3506], device='cuda:6'), in_proj_covar=tensor([0.0432, 0.0485, 0.0396, 0.0347, 0.0453, 0.0557, 0.0458, 0.0569], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:32:04,916 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.483e+02 2.203e+02 2.647e+02 3.147e+02 6.132e+02, threshold=5.294e+02, percent-clipped=2.0 2023-05-02 21:32:07,813 INFO [train.py:904] (6/8) Epoch 30, batch 2550, loss[loss=0.1651, simple_loss=0.256, pruned_loss=0.03712, over 16536.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2544, pruned_loss=0.03836, over 3319311.64 frames. ], batch size: 62, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:32:49,131 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296934.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 21:32:54,172 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296938.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:33:15,497 INFO [train.py:904] (6/8) Epoch 30, batch 2600, loss[loss=0.1564, simple_loss=0.2588, pruned_loss=0.02694, over 17121.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2541, pruned_loss=0.03773, over 3329336.06 frames. ], batch size: 49, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:33:36,661 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296969.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:33:39,828 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296971.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:34:10,134 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3656, 3.6858, 3.8705, 2.5791, 3.4663, 3.9141, 3.5624, 2.2566], device='cuda:6'), covar=tensor([0.0576, 0.0249, 0.0074, 0.0455, 0.0151, 0.0136, 0.0128, 0.0541], device='cuda:6'), in_proj_covar=tensor([0.0140, 0.0092, 0.0093, 0.0136, 0.0105, 0.0117, 0.0100, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 21:34:18,497 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296999.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:34:22,260 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.341e+02 2.007e+02 2.417e+02 2.925e+02 5.072e+02, threshold=4.833e+02, percent-clipped=0.0 2023-05-02 21:34:24,414 INFO [train.py:904] (6/8) Epoch 30, batch 2650, loss[loss=0.1623, simple_loss=0.2602, pruned_loss=0.03215, over 17142.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2549, pruned_loss=0.03774, over 3333502.34 frames. ], batch size: 47, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:34:46,004 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297019.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:35:20,543 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=297044.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:35:34,415 INFO [train.py:904] (6/8) Epoch 30, batch 2700, loss[loss=0.1754, simple_loss=0.2647, pruned_loss=0.04304, over 16801.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2547, pruned_loss=0.03733, over 3323229.68 frames. ], batch size: 42, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:35:43,760 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-02 21:36:24,076 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297090.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:36:42,009 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.120e+02 2.454e+02 2.893e+02 5.534e+02, threshold=4.909e+02, percent-clipped=1.0 2023-05-02 21:36:43,155 INFO [train.py:904] (6/8) Epoch 30, batch 2750, loss[loss=0.1527, simple_loss=0.2414, pruned_loss=0.03198, over 16939.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2541, pruned_loss=0.03678, over 3331240.99 frames. ], batch size: 41, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:36:46,559 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1053, 3.9737, 4.1959, 4.3035, 4.3668, 3.9814, 4.2247, 4.3862], device='cuda:6'), covar=tensor([0.1786, 0.1450, 0.1350, 0.0792, 0.0712, 0.1441, 0.2541, 0.1033], device='cuda:6'), in_proj_covar=tensor([0.0715, 0.0867, 0.1001, 0.0885, 0.0671, 0.0697, 0.0735, 0.0857], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:36:46,976 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2023-05-02 21:37:16,630 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1879, 4.2616, 4.4964, 4.4773, 4.5507, 4.2733, 4.2891, 4.2116], device='cuda:6'), covar=tensor([0.0389, 0.0596, 0.0429, 0.0440, 0.0509, 0.0442, 0.0801, 0.0616], device='cuda:6'), in_proj_covar=tensor([0.0449, 0.0512, 0.0489, 0.0452, 0.0533, 0.0516, 0.0593, 0.0415], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 21:37:30,858 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297138.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:37:44,436 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297148.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:37:50,741 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297153.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:37:51,590 INFO [train.py:904] (6/8) Epoch 30, batch 2800, loss[loss=0.1594, simple_loss=0.2469, pruned_loss=0.03593, over 16477.00 frames. ], tot_loss[loss=0.164, simple_loss=0.254, pruned_loss=0.03702, over 3339714.89 frames. ], batch size: 68, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:38:14,293 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297170.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:38:49,784 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297196.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:38:57,320 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297201.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:39:00,127 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.644e+02 2.070e+02 2.512e+02 3.097e+02 5.652e+02, threshold=5.024e+02, percent-clipped=1.0 2023-05-02 21:39:01,282 INFO [train.py:904] (6/8) Epoch 30, batch 2850, loss[loss=0.1732, simple_loss=0.2719, pruned_loss=0.03731, over 17074.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.254, pruned_loss=0.03694, over 3335838.52 frames. ], batch size: 53, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:39:21,723 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297218.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:39:26,924 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=297222.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:39:37,166 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=297229.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 21:40:10,537 INFO [train.py:904] (6/8) Epoch 30, batch 2900, loss[loss=0.1667, simple_loss=0.2505, pruned_loss=0.04145, over 16414.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2537, pruned_loss=0.03744, over 3324322.11 frames. ], batch size: 146, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:40:27,111 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3442, 3.4338, 3.6025, 2.4366, 3.2832, 3.7462, 3.4605, 2.2091], device='cuda:6'), covar=tensor([0.0564, 0.0168, 0.0077, 0.0454, 0.0155, 0.0106, 0.0112, 0.0544], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0091, 0.0093, 0.0136, 0.0104, 0.0116, 0.0099, 0.0131], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 21:40:30,350 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7799, 2.7270, 2.3941, 2.5162, 3.0124, 2.6687, 3.3749, 3.2064], device='cuda:6'), covar=tensor([0.0195, 0.0514, 0.0604, 0.0555, 0.0352, 0.0500, 0.0251, 0.0363], device='cuda:6'), in_proj_covar=tensor([0.0242, 0.0251, 0.0239, 0.0241, 0.0252, 0.0250, 0.0249, 0.0252], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:40:33,069 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297269.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:40:51,715 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=297283.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:41:08,333 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=297294.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:41:12,774 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8718, 3.6673, 4.0401, 2.1685, 4.0929, 4.1547, 3.3540, 3.1928], device='cuda:6'), covar=tensor([0.0740, 0.0262, 0.0201, 0.1202, 0.0121, 0.0213, 0.0384, 0.0467], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0112, 0.0104, 0.0139, 0.0089, 0.0135, 0.0131, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 21:41:22,344 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.045e+02 2.480e+02 2.886e+02 5.342e+02, threshold=4.960e+02, percent-clipped=3.0 2023-05-02 21:41:22,359 INFO [train.py:904] (6/8) Epoch 30, batch 2950, loss[loss=0.1473, simple_loss=0.2407, pruned_loss=0.02692, over 17214.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2529, pruned_loss=0.03812, over 3316897.54 frames. ], batch size: 45, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:41:41,546 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297317.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:41:43,832 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-05-02 21:42:02,061 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0905, 5.1087, 4.9725, 4.5440, 4.6494, 5.0156, 4.9067, 4.7042], device='cuda:6'), covar=tensor([0.0604, 0.0532, 0.0354, 0.0370, 0.1003, 0.0491, 0.0392, 0.0758], device='cuda:6'), in_proj_covar=tensor([0.0329, 0.0499, 0.0385, 0.0389, 0.0382, 0.0446, 0.0265, 0.0462], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 21:42:19,039 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297344.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:42:33,297 INFO [train.py:904] (6/8) Epoch 30, batch 3000, loss[loss=0.1733, simple_loss=0.2575, pruned_loss=0.04454, over 15558.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2528, pruned_loss=0.03861, over 3312654.73 frames. ], batch size: 190, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:42:33,297 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 21:42:42,092 INFO [train.py:938] (6/8) Epoch 30, validation: loss=0.1331, simple_loss=0.238, pruned_loss=0.01415, over 944034.00 frames. 2023-05-02 21:42:42,093 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 21:43:03,395 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8837, 5.1802, 4.9789, 4.9584, 4.7684, 4.6800, 4.6164, 5.2665], device='cuda:6'), covar=tensor([0.1250, 0.0848, 0.0964, 0.0890, 0.0831, 0.1150, 0.1224, 0.0894], device='cuda:6'), in_proj_covar=tensor([0.0745, 0.0896, 0.0739, 0.0698, 0.0574, 0.0566, 0.0754, 0.0705], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:43:35,254 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297392.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:43:51,743 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.593e+02 2.205e+02 2.625e+02 3.080e+02 5.870e+02, threshold=5.251e+02, percent-clipped=1.0 2023-05-02 21:43:51,758 INFO [train.py:904] (6/8) Epoch 30, batch 3050, loss[loss=0.1728, simple_loss=0.2491, pruned_loss=0.04827, over 16756.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2524, pruned_loss=0.03854, over 3313494.27 frames. ], batch size: 134, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:44:12,758 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8127, 1.9999, 2.4866, 2.7920, 2.7333, 3.2753, 2.2376, 3.2706], device='cuda:6'), covar=tensor([0.0344, 0.0657, 0.0462, 0.0451, 0.0420, 0.0269, 0.0650, 0.0221], device='cuda:6'), in_proj_covar=tensor([0.0206, 0.0205, 0.0194, 0.0199, 0.0218, 0.0174, 0.0210, 0.0174], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 21:44:19,757 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8706, 3.8042, 3.9372, 4.0447, 4.0824, 3.6944, 3.9283, 4.1297], device='cuda:6'), covar=tensor([0.1681, 0.1178, 0.1232, 0.0716, 0.0679, 0.2081, 0.2107, 0.0776], device='cuda:6'), in_proj_covar=tensor([0.0718, 0.0871, 0.1009, 0.0889, 0.0673, 0.0702, 0.0741, 0.0862], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:45:01,424 INFO [train.py:904] (6/8) Epoch 30, batch 3100, loss[loss=0.1463, simple_loss=0.2353, pruned_loss=0.02864, over 17219.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2522, pruned_loss=0.03836, over 3311680.22 frames. ], batch size: 44, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:46:04,651 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.2458, 5.2279, 5.6613, 5.6148, 5.6663, 5.3322, 5.2636, 5.1080], device='cuda:6'), covar=tensor([0.0347, 0.0575, 0.0420, 0.0487, 0.0492, 0.0402, 0.0976, 0.0483], device='cuda:6'), in_proj_covar=tensor([0.0452, 0.0515, 0.0490, 0.0453, 0.0537, 0.0518, 0.0598, 0.0417], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 21:46:07,371 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.135e+02 2.482e+02 3.007e+02 9.801e+02, threshold=4.963e+02, percent-clipped=2.0 2023-05-02 21:46:07,387 INFO [train.py:904] (6/8) Epoch 30, batch 3150, loss[loss=0.1551, simple_loss=0.2493, pruned_loss=0.03041, over 17172.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2515, pruned_loss=0.03783, over 3322890.33 frames. ], batch size: 46, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:46:42,747 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297529.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:47:09,757 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-05-02 21:47:17,203 INFO [train.py:904] (6/8) Epoch 30, batch 3200, loss[loss=0.1659, simple_loss=0.2578, pruned_loss=0.037, over 16491.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2507, pruned_loss=0.03753, over 3322530.81 frames. ], batch size: 68, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:47:30,808 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-02 21:47:49,869 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297577.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:47:51,694 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=297578.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:48:07,152 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8867, 3.9083, 4.1341, 4.1029, 4.1570, 3.9244, 3.9680, 3.9179], device='cuda:6'), covar=tensor([0.0390, 0.0708, 0.0451, 0.0452, 0.0573, 0.0500, 0.0769, 0.0553], device='cuda:6'), in_proj_covar=tensor([0.0453, 0.0517, 0.0492, 0.0455, 0.0538, 0.0519, 0.0600, 0.0419], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:6') 2023-05-02 21:48:13,572 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297594.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:48:27,239 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.469e+02 2.037e+02 2.359e+02 2.785e+02 6.437e+02, threshold=4.718e+02, percent-clipped=1.0 2023-05-02 21:48:27,254 INFO [train.py:904] (6/8) Epoch 30, batch 3250, loss[loss=0.1743, simple_loss=0.2545, pruned_loss=0.04699, over 16717.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2511, pruned_loss=0.03826, over 3324783.24 frames. ], batch size: 83, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:48:28,837 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7644, 4.2050, 4.2879, 2.9719, 3.6507, 4.2719, 3.8115, 2.6293], device='cuda:6'), covar=tensor([0.0527, 0.0099, 0.0061, 0.0429, 0.0138, 0.0109, 0.0103, 0.0483], device='cuda:6'), in_proj_covar=tensor([0.0142, 0.0093, 0.0094, 0.0138, 0.0106, 0.0119, 0.0101, 0.0134], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 21:49:20,240 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297642.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:49:36,043 INFO [train.py:904] (6/8) Epoch 30, batch 3300, loss[loss=0.142, simple_loss=0.2297, pruned_loss=0.02714, over 15852.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2516, pruned_loss=0.03784, over 3324769.90 frames. ], batch size: 35, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:50:46,298 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.242e+02 2.585e+02 3.281e+02 4.783e+02, threshold=5.170e+02, percent-clipped=1.0 2023-05-02 21:50:46,313 INFO [train.py:904] (6/8) Epoch 30, batch 3350, loss[loss=0.1708, simple_loss=0.2528, pruned_loss=0.04441, over 16928.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2526, pruned_loss=0.03823, over 3324394.90 frames. ], batch size: 90, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:51:46,860 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3793, 5.3579, 5.2346, 4.7224, 4.8598, 5.2583, 5.2611, 4.8426], device='cuda:6'), covar=tensor([0.0595, 0.0575, 0.0377, 0.0407, 0.1178, 0.0587, 0.0312, 0.0906], device='cuda:6'), in_proj_covar=tensor([0.0332, 0.0503, 0.0388, 0.0391, 0.0385, 0.0450, 0.0267, 0.0466], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 21:51:56,062 INFO [train.py:904] (6/8) Epoch 30, batch 3400, loss[loss=0.1934, simple_loss=0.2714, pruned_loss=0.05772, over 16908.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2523, pruned_loss=0.03793, over 3318996.97 frames. ], batch size: 116, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:53:07,131 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 1.981e+02 2.351e+02 2.739e+02 4.469e+02, threshold=4.702e+02, percent-clipped=0.0 2023-05-02 21:53:07,146 INFO [train.py:904] (6/8) Epoch 30, batch 3450, loss[loss=0.1601, simple_loss=0.2541, pruned_loss=0.03304, over 17039.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2511, pruned_loss=0.03737, over 3316257.98 frames. ], batch size: 53, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:54:17,196 INFO [train.py:904] (6/8) Epoch 30, batch 3500, loss[loss=0.1655, simple_loss=0.2476, pruned_loss=0.04167, over 16526.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2504, pruned_loss=0.03752, over 3305895.81 frames. ], batch size: 75, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:54:51,993 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297878.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:55:28,082 INFO [train.py:904] (6/8) Epoch 30, batch 3550, loss[loss=0.1611, simple_loss=0.245, pruned_loss=0.03862, over 16693.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2486, pruned_loss=0.037, over 3316262.50 frames. ], batch size: 134, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:55:29,302 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.419e+02 1.998e+02 2.319e+02 2.785e+02 6.034e+02, threshold=4.637e+02, percent-clipped=1.0 2023-05-02 21:55:59,066 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297926.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:56:00,316 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0240, 5.3945, 5.1412, 5.1732, 4.8831, 4.9097, 4.7788, 5.5036], device='cuda:6'), covar=tensor([0.1358, 0.0938, 0.1094, 0.0886, 0.0876, 0.0976, 0.1372, 0.0909], device='cuda:6'), in_proj_covar=tensor([0.0747, 0.0901, 0.0741, 0.0701, 0.0578, 0.0567, 0.0757, 0.0708], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:56:38,331 INFO [train.py:904] (6/8) Epoch 30, batch 3600, loss[loss=0.1655, simple_loss=0.2561, pruned_loss=0.03744, over 16649.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2475, pruned_loss=0.03691, over 3317426.77 frames. ], batch size: 62, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:56:42,425 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=297957.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:57:19,947 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3368, 3.3175, 3.3585, 3.4417, 3.4954, 3.2851, 3.4609, 3.5674], device='cuda:6'), covar=tensor([0.1208, 0.0918, 0.1097, 0.0692, 0.0670, 0.2272, 0.1332, 0.0830], device='cuda:6'), in_proj_covar=tensor([0.0725, 0.0880, 0.1021, 0.0901, 0.0681, 0.0712, 0.0749, 0.0872], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:57:22,655 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.95 vs. limit=5.0 2023-05-02 21:57:24,141 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8570, 4.1947, 4.1937, 3.0027, 3.6882, 4.2226, 3.7313, 2.3172], device='cuda:6'), covar=tensor([0.0496, 0.0139, 0.0071, 0.0410, 0.0144, 0.0128, 0.0121, 0.0567], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0092, 0.0094, 0.0137, 0.0105, 0.0118, 0.0101, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 21:57:53,330 INFO [train.py:904] (6/8) Epoch 30, batch 3650, loss[loss=0.1565, simple_loss=0.2358, pruned_loss=0.03862, over 16844.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.2474, pruned_loss=0.03771, over 3294712.22 frames. ], batch size: 90, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:57:55,114 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.721e+02 2.298e+02 2.680e+02 3.365e+02 6.432e+02, threshold=5.360e+02, percent-clipped=4.0 2023-05-02 21:58:14,262 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=298018.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:58:26,422 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4979, 3.5355, 2.2954, 3.7790, 2.8959, 3.7211, 2.3214, 2.9261], device='cuda:6'), covar=tensor([0.0315, 0.0526, 0.1561, 0.0336, 0.0778, 0.0873, 0.1516, 0.0775], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0186, 0.0199, 0.0180, 0.0184, 0.0227, 0.0208, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 21:58:32,779 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0797, 2.7657, 2.6317, 4.4654, 3.7690, 4.2738, 1.6804, 3.2479], device='cuda:6'), covar=tensor([0.1315, 0.0716, 0.1241, 0.0188, 0.0166, 0.0374, 0.1687, 0.0773], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0183, 0.0202, 0.0209, 0.0208, 0.0220, 0.0212, 0.0200], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 21:58:32,994 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.23 vs. limit=5.0 2023-05-02 21:58:52,329 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9285, 2.1849, 2.3002, 3.4096, 2.1463, 2.3889, 2.2997, 2.2929], device='cuda:6'), covar=tensor([0.1707, 0.3533, 0.3152, 0.0885, 0.4185, 0.2689, 0.3615, 0.3620], device='cuda:6'), in_proj_covar=tensor([0.0432, 0.0484, 0.0394, 0.0347, 0.0452, 0.0556, 0.0457, 0.0568], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 21:59:06,226 INFO [train.py:904] (6/8) Epoch 30, batch 3700, loss[loss=0.2016, simple_loss=0.2759, pruned_loss=0.06363, over 11128.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2459, pruned_loss=0.03875, over 3271163.30 frames. ], batch size: 248, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:59:38,479 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-02 21:59:40,270 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-05-02 21:59:41,479 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-02 22:00:13,353 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-05-02 22:00:17,692 INFO [train.py:904] (6/8) Epoch 30, batch 3750, loss[loss=0.1785, simple_loss=0.2552, pruned_loss=0.05094, over 16717.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2462, pruned_loss=0.03993, over 3249716.87 frames. ], batch size: 124, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:00:19,705 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.710e+02 2.048e+02 2.384e+02 2.845e+02 4.773e+02, threshold=4.768e+02, percent-clipped=0.0 2023-05-02 22:00:53,433 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2842, 3.1223, 3.3936, 1.7985, 3.5414, 3.6019, 3.1124, 2.6432], device='cuda:6'), covar=tensor([0.1128, 0.0277, 0.0260, 0.1374, 0.0150, 0.0225, 0.0424, 0.0648], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0112, 0.0104, 0.0140, 0.0089, 0.0136, 0.0132, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 22:01:30,457 INFO [train.py:904] (6/8) Epoch 30, batch 3800, loss[loss=0.1716, simple_loss=0.2589, pruned_loss=0.04215, over 16371.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2467, pruned_loss=0.04086, over 3253090.40 frames. ], batch size: 35, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:02:15,421 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 22:02:43,880 INFO [train.py:904] (6/8) Epoch 30, batch 3850, loss[loss=0.1433, simple_loss=0.2195, pruned_loss=0.03355, over 16712.00 frames. ], tot_loss[loss=0.165, simple_loss=0.247, pruned_loss=0.04147, over 3249915.55 frames. ], batch size: 83, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:02:44,948 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.556e+02 2.341e+02 2.538e+02 2.759e+02 9.437e+02, threshold=5.075e+02, percent-clipped=1.0 2023-05-02 22:02:48,153 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 22:03:10,377 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5436, 3.3804, 3.7953, 2.2968, 3.1930, 2.4748, 3.9644, 3.8969], device='cuda:6'), covar=tensor([0.0183, 0.0843, 0.0556, 0.2034, 0.0812, 0.0946, 0.0489, 0.1013], device='cuda:6'), in_proj_covar=tensor([0.0164, 0.0174, 0.0172, 0.0159, 0.0150, 0.0134, 0.0147, 0.0188], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 22:03:53,273 INFO [train.py:904] (6/8) Epoch 30, batch 3900, loss[loss=0.1948, simple_loss=0.2779, pruned_loss=0.05585, over 16471.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2469, pruned_loss=0.04181, over 3255274.24 frames. ], batch size: 68, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:05:04,297 INFO [train.py:904] (6/8) Epoch 30, batch 3950, loss[loss=0.1591, simple_loss=0.2384, pruned_loss=0.03993, over 16700.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2471, pruned_loss=0.04236, over 3270266.37 frames. ], batch size: 124, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:05:05,533 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.189e+02 2.590e+02 3.423e+02 7.666e+02, threshold=5.180e+02, percent-clipped=4.0 2023-05-02 22:05:17,508 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=298313.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:06:02,235 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5933, 1.8426, 2.3407, 2.4428, 2.6093, 2.5851, 1.9634, 2.7300], device='cuda:6'), covar=tensor([0.0230, 0.0590, 0.0392, 0.0366, 0.0383, 0.0377, 0.0650, 0.0209], device='cuda:6'), in_proj_covar=tensor([0.0204, 0.0203, 0.0193, 0.0198, 0.0217, 0.0173, 0.0208, 0.0173], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 22:06:15,412 INFO [train.py:904] (6/8) Epoch 30, batch 4000, loss[loss=0.1633, simple_loss=0.2428, pruned_loss=0.0419, over 16875.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.248, pruned_loss=0.04327, over 3271089.84 frames. ], batch size: 109, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:06:41,538 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.5145, 5.5047, 5.2592, 4.5599, 5.4653, 1.9133, 5.1517, 4.8171], device='cuda:6'), covar=tensor([0.0073, 0.0060, 0.0177, 0.0341, 0.0071, 0.3150, 0.0121, 0.0277], device='cuda:6'), in_proj_covar=tensor([0.0189, 0.0180, 0.0220, 0.0191, 0.0198, 0.0223, 0.0209, 0.0187], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:07:04,606 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.9624, 2.6982, 2.7090, 4.8893, 3.7832, 4.2172, 1.7453, 3.1377], device='cuda:6'), covar=tensor([0.1215, 0.0854, 0.1238, 0.0156, 0.0362, 0.0385, 0.1587, 0.0851], device='cuda:6'), in_proj_covar=tensor([0.0174, 0.0183, 0.0202, 0.0209, 0.0208, 0.0220, 0.0212, 0.0201], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 22:07:25,648 INFO [train.py:904] (6/8) Epoch 30, batch 4050, loss[loss=0.1709, simple_loss=0.2619, pruned_loss=0.04, over 15290.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2492, pruned_loss=0.04275, over 3270976.29 frames. ], batch size: 190, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:07:27,603 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.527e+02 1.991e+02 2.231e+02 2.617e+02 4.473e+02, threshold=4.462e+02, percent-clipped=0.0 2023-05-02 22:08:35,791 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7866, 5.0299, 4.8327, 4.8431, 4.5669, 4.5475, 4.4650, 5.1224], device='cuda:6'), covar=tensor([0.1219, 0.0850, 0.1002, 0.0874, 0.0835, 0.1242, 0.1165, 0.0816], device='cuda:6'), in_proj_covar=tensor([0.0747, 0.0901, 0.0739, 0.0701, 0.0577, 0.0570, 0.0759, 0.0708], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:08:37,713 INFO [train.py:904] (6/8) Epoch 30, batch 4100, loss[loss=0.1815, simple_loss=0.276, pruned_loss=0.04352, over 16504.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2507, pruned_loss=0.04238, over 3279161.33 frames. ], batch size: 68, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:09:20,318 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=298482.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:09:39,531 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7642, 3.5460, 4.0344, 2.1188, 4.1800, 4.2539, 3.1435, 3.2390], device='cuda:6'), covar=tensor([0.0828, 0.0292, 0.0225, 0.1219, 0.0091, 0.0157, 0.0475, 0.0461], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0113, 0.0104, 0.0141, 0.0089, 0.0136, 0.0133, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 22:09:53,959 INFO [train.py:904] (6/8) Epoch 30, batch 4150, loss[loss=0.2378, simple_loss=0.3022, pruned_loss=0.08676, over 11322.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2584, pruned_loss=0.04504, over 3234444.80 frames. ], batch size: 248, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:09:56,054 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 2.028e+02 2.326e+02 2.808e+02 4.631e+02, threshold=4.653e+02, percent-clipped=1.0 2023-05-02 22:10:35,456 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=298530.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:10:54,911 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.59 vs. limit=2.0 2023-05-02 22:10:56,363 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=298543.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 22:11:12,253 INFO [train.py:904] (6/8) Epoch 30, batch 4200, loss[loss=0.1976, simple_loss=0.283, pruned_loss=0.05608, over 11341.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2649, pruned_loss=0.04622, over 3223670.37 frames. ], batch size: 248, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:11:31,419 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.90 vs. limit=5.0 2023-05-02 22:12:08,296 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=298591.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:12:26,984 INFO [train.py:904] (6/8) Epoch 30, batch 4250, loss[loss=0.1691, simple_loss=0.2612, pruned_loss=0.0385, over 17017.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2677, pruned_loss=0.04569, over 3209859.11 frames. ], batch size: 55, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:12:28,281 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.166e+02 2.498e+02 2.846e+02 5.761e+02, threshold=4.997e+02, percent-clipped=3.0 2023-05-02 22:12:40,700 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=298613.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:13:04,722 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=298629.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 22:13:14,025 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-05-02 22:13:15,287 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.4497, 5.7240, 5.5000, 5.5618, 5.2213, 5.1191, 5.1370, 5.8579], device='cuda:6'), covar=tensor([0.1122, 0.0780, 0.0973, 0.0835, 0.0846, 0.0765, 0.1158, 0.0762], device='cuda:6'), in_proj_covar=tensor([0.0736, 0.0888, 0.0729, 0.0691, 0.0570, 0.0564, 0.0748, 0.0699], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:13:41,912 INFO [train.py:904] (6/8) Epoch 30, batch 4300, loss[loss=0.1792, simple_loss=0.2686, pruned_loss=0.04488, over 16636.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2689, pruned_loss=0.04469, over 3194812.18 frames. ], batch size: 57, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:13:53,098 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=298661.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:14:35,052 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=298689.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:14:36,346 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=298690.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 22:14:56,150 INFO [train.py:904] (6/8) Epoch 30, batch 4350, loss[loss=0.197, simple_loss=0.2853, pruned_loss=0.05433, over 16692.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2722, pruned_loss=0.04556, over 3192449.17 frames. ], batch size: 62, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:14:57,385 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.491e+02 2.186e+02 2.538e+02 2.866e+02 4.288e+02, threshold=5.076e+02, percent-clipped=0.0 2023-05-02 22:16:04,766 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=298750.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 22:16:09,613 INFO [train.py:904] (6/8) Epoch 30, batch 4400, loss[loss=0.1872, simple_loss=0.2738, pruned_loss=0.05032, over 16615.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2743, pruned_loss=0.04684, over 3185542.36 frames. ], batch size: 134, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:17:21,873 INFO [train.py:904] (6/8) Epoch 30, batch 4450, loss[loss=0.1952, simple_loss=0.2849, pruned_loss=0.05275, over 17200.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2782, pruned_loss=0.04868, over 3206372.17 frames. ], batch size: 44, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:17:23,559 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.049e+02 2.382e+02 3.028e+02 5.771e+02, threshold=4.764e+02, percent-clipped=2.0 2023-05-02 22:18:10,729 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=298838.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 22:18:32,504 INFO [train.py:904] (6/8) Epoch 30, batch 4500, loss[loss=0.201, simple_loss=0.2807, pruned_loss=0.06068, over 16641.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2785, pruned_loss=0.0496, over 3208939.01 frames. ], batch size: 57, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:18:48,995 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1387, 1.9817, 2.8485, 3.0760, 2.9721, 3.5357, 2.2816, 3.5320], device='cuda:6'), covar=tensor([0.0241, 0.0649, 0.0316, 0.0324, 0.0343, 0.0199, 0.0654, 0.0170], device='cuda:6'), in_proj_covar=tensor([0.0203, 0.0202, 0.0191, 0.0198, 0.0216, 0.0172, 0.0208, 0.0172], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 22:19:19,849 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=298886.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:19:31,897 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.59 vs. limit=2.0 2023-05-02 22:19:44,893 INFO [train.py:904] (6/8) Epoch 30, batch 4550, loss[loss=0.2099, simple_loss=0.2992, pruned_loss=0.06027, over 16181.00 frames. ], tot_loss[loss=0.191, simple_loss=0.2802, pruned_loss=0.05091, over 3216947.45 frames. ], batch size: 35, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:19:46,103 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.328e+02 1.782e+02 2.097e+02 2.367e+02 4.731e+02, threshold=4.194e+02, percent-clipped=0.0 2023-05-02 22:20:56,940 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 22:20:57,287 INFO [train.py:904] (6/8) Epoch 30, batch 4600, loss[loss=0.1808, simple_loss=0.2732, pruned_loss=0.04419, over 16728.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.2808, pruned_loss=0.05097, over 3214555.82 frames. ], batch size: 83, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:21:24,191 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3547, 5.5983, 5.3921, 5.4471, 5.1297, 5.0510, 5.1063, 5.7442], device='cuda:6'), covar=tensor([0.1229, 0.0825, 0.1000, 0.0879, 0.0902, 0.0794, 0.1040, 0.0849], device='cuda:6'), in_proj_covar=tensor([0.0730, 0.0881, 0.0724, 0.0686, 0.0566, 0.0560, 0.0739, 0.0693], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:21:43,593 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=298985.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 22:22:09,581 INFO [train.py:904] (6/8) Epoch 30, batch 4650, loss[loss=0.1828, simple_loss=0.2671, pruned_loss=0.0493, over 17225.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2801, pruned_loss=0.05109, over 3221220.65 frames. ], batch size: 45, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:22:10,890 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.373e+02 1.831e+02 2.100e+02 2.427e+02 4.549e+02, threshold=4.199e+02, percent-clipped=1.0 2023-05-02 22:22:46,067 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.9415, 4.7447, 4.7237, 5.1624, 5.2817, 4.7901, 5.3057, 5.3408], device='cuda:6'), covar=tensor([0.1847, 0.1371, 0.2011, 0.0745, 0.0657, 0.1154, 0.0837, 0.0771], device='cuda:6'), in_proj_covar=tensor([0.0696, 0.0844, 0.0977, 0.0862, 0.0654, 0.0681, 0.0715, 0.0832], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:22:50,179 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.4223, 2.6859, 2.2634, 2.3676, 2.9658, 2.5761, 2.9234, 3.1168], device='cuda:6'), covar=tensor([0.0125, 0.0405, 0.0523, 0.0484, 0.0260, 0.0415, 0.0225, 0.0273], device='cuda:6'), in_proj_covar=tensor([0.0236, 0.0247, 0.0236, 0.0238, 0.0247, 0.0245, 0.0246, 0.0248], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:23:09,774 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=299045.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 22:23:22,509 INFO [train.py:904] (6/8) Epoch 30, batch 4700, loss[loss=0.1529, simple_loss=0.2423, pruned_loss=0.03176, over 16720.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2774, pruned_loss=0.05011, over 3223334.62 frames. ], batch size: 89, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:23:32,965 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7592, 2.8789, 2.8863, 5.0367, 3.7794, 4.2426, 1.7706, 3.1356], device='cuda:6'), covar=tensor([0.1338, 0.0809, 0.1136, 0.0128, 0.0298, 0.0349, 0.1655, 0.0819], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0183, 0.0202, 0.0208, 0.0209, 0.0220, 0.0212, 0.0202], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 22:23:33,205 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-02 22:24:02,720 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8830, 4.6886, 4.8926, 5.0827, 5.2446, 4.7535, 5.2523, 5.2731], device='cuda:6'), covar=tensor([0.1913, 0.1356, 0.1810, 0.0796, 0.0534, 0.0876, 0.0600, 0.0628], device='cuda:6'), in_proj_covar=tensor([0.0692, 0.0839, 0.0971, 0.0857, 0.0650, 0.0676, 0.0709, 0.0828], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:24:36,517 INFO [train.py:904] (6/8) Epoch 30, batch 4750, loss[loss=0.1647, simple_loss=0.2517, pruned_loss=0.0388, over 16557.00 frames. ], tot_loss[loss=0.1845, simple_loss=0.273, pruned_loss=0.04798, over 3221145.01 frames. ], batch size: 62, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:24:37,722 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.301e+02 1.904e+02 2.113e+02 2.472e+02 4.285e+02, threshold=4.226e+02, percent-clipped=1.0 2023-05-02 22:25:08,356 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5885, 3.5421, 3.5231, 2.7707, 3.3188, 2.1025, 3.1291, 2.7825], device='cuda:6'), covar=tensor([0.0220, 0.0256, 0.0200, 0.0337, 0.0152, 0.2590, 0.0172, 0.0316], device='cuda:6'), in_proj_covar=tensor([0.0188, 0.0180, 0.0219, 0.0190, 0.0197, 0.0223, 0.0208, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:25:08,379 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=299126.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:25:25,848 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299138.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:25:49,328 INFO [train.py:904] (6/8) Epoch 30, batch 4800, loss[loss=0.1714, simple_loss=0.2698, pruned_loss=0.03655, over 16821.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.2697, pruned_loss=0.04601, over 3220709.24 frames. ], batch size: 83, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:26:37,333 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299186.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:26:37,384 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299186.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:26:39,159 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=299187.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:26:46,014 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=299191.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:27:04,356 INFO [train.py:904] (6/8) Epoch 30, batch 4850, loss[loss=0.1858, simple_loss=0.2863, pruned_loss=0.04261, over 16871.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2705, pruned_loss=0.04495, over 3210908.75 frames. ], batch size: 96, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:27:06,458 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.292e+02 1.894e+02 2.171e+02 2.574e+02 8.129e+02, threshold=4.343e+02, percent-clipped=1.0 2023-05-02 22:27:49,430 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299234.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:28:03,034 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=299243.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:28:16,524 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=299252.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:28:18,992 INFO [train.py:904] (6/8) Epoch 30, batch 4900, loss[loss=0.164, simple_loss=0.2509, pruned_loss=0.0386, over 16228.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2697, pruned_loss=0.04401, over 3180374.44 frames. ], batch size: 35, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:29:05,979 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299285.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 22:29:28,090 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 22:29:33,773 INFO [train.py:904] (6/8) Epoch 30, batch 4950, loss[loss=0.1809, simple_loss=0.2739, pruned_loss=0.04393, over 16463.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2695, pruned_loss=0.04373, over 3165172.67 frames. ], batch size: 146, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:29:34,249 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=299304.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:29:34,846 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.269e+02 1.940e+02 2.221e+02 2.664e+02 5.987e+02, threshold=4.443e+02, percent-clipped=2.0 2023-05-02 22:29:38,506 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.49 vs. limit=5.0 2023-05-02 22:30:02,810 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=299323.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:30:17,899 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299333.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 22:30:36,135 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299345.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:30:47,899 INFO [train.py:904] (6/8) Epoch 30, batch 5000, loss[loss=0.1823, simple_loss=0.2679, pruned_loss=0.04839, over 17000.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2707, pruned_loss=0.0435, over 3180596.16 frames. ], batch size: 53, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:31:01,432 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2224, 3.2449, 2.0125, 3.6112, 2.4219, 3.6014, 2.1593, 2.6516], device='cuda:6'), covar=tensor([0.0379, 0.0457, 0.1746, 0.0167, 0.0944, 0.0549, 0.1658, 0.0931], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0183, 0.0196, 0.0175, 0.0181, 0.0221, 0.0204, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 22:31:33,669 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=299384.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:31:46,452 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299393.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:32:01,716 INFO [train.py:904] (6/8) Epoch 30, batch 5050, loss[loss=0.1639, simple_loss=0.2621, pruned_loss=0.03285, over 16856.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2718, pruned_loss=0.04339, over 3194389.18 frames. ], batch size: 102, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:32:02,876 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.469e+02 2.023e+02 2.400e+02 2.818e+02 3.992e+02, threshold=4.801e+02, percent-clipped=0.0 2023-05-02 22:32:06,431 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5358, 2.2282, 1.9052, 1.9850, 2.5153, 2.2162, 2.0865, 2.6129], device='cuda:6'), covar=tensor([0.0220, 0.0579, 0.0703, 0.0627, 0.0324, 0.0474, 0.0272, 0.0365], device='cuda:6'), in_proj_covar=tensor([0.0236, 0.0246, 0.0236, 0.0237, 0.0247, 0.0245, 0.0245, 0.0248], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:33:13,778 INFO [train.py:904] (6/8) Epoch 30, batch 5100, loss[loss=0.161, simple_loss=0.2479, pruned_loss=0.03701, over 17028.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2698, pruned_loss=0.04256, over 3209493.76 frames. ], batch size: 50, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:33:42,827 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.0154, 4.0918, 3.9203, 3.6292, 3.6069, 3.9981, 3.7059, 3.7824], device='cuda:6'), covar=tensor([0.0643, 0.0505, 0.0374, 0.0342, 0.0841, 0.0540, 0.1081, 0.0622], device='cuda:6'), in_proj_covar=tensor([0.0318, 0.0482, 0.0374, 0.0376, 0.0371, 0.0433, 0.0255, 0.0444], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:33:57,979 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=299482.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:34:14,230 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.2544, 1.6959, 1.9940, 2.2522, 2.3328, 2.5354, 1.7636, 2.5036], device='cuda:6'), covar=tensor([0.0287, 0.0594, 0.0392, 0.0413, 0.0418, 0.0232, 0.0706, 0.0173], device='cuda:6'), in_proj_covar=tensor([0.0200, 0.0200, 0.0188, 0.0195, 0.0213, 0.0169, 0.0206, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 22:34:30,232 INFO [train.py:904] (6/8) Epoch 30, batch 5150, loss[loss=0.164, simple_loss=0.2653, pruned_loss=0.03141, over 16394.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2697, pruned_loss=0.04174, over 3216806.97 frames. ], batch size: 146, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:34:30,736 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0739, 2.4664, 2.5962, 1.9299, 2.7752, 2.8371, 2.5210, 2.4128], device='cuda:6'), covar=tensor([0.0728, 0.0295, 0.0228, 0.1007, 0.0139, 0.0230, 0.0460, 0.0477], device='cuda:6'), in_proj_covar=tensor([0.0149, 0.0112, 0.0103, 0.0139, 0.0088, 0.0133, 0.0132, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 22:34:31,339 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 1.896e+02 2.234e+02 2.667e+02 3.487e+02, threshold=4.469e+02, percent-clipped=0.0 2023-05-02 22:35:33,156 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=299547.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:35:43,063 INFO [train.py:904] (6/8) Epoch 30, batch 5200, loss[loss=0.1787, simple_loss=0.2591, pruned_loss=0.04918, over 16278.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2681, pruned_loss=0.04117, over 3194343.75 frames. ], batch size: 35, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:36:35,232 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5699, 3.6105, 2.7969, 2.2717, 2.3492, 2.5103, 3.7333, 3.1938], device='cuda:6'), covar=tensor([0.2944, 0.0629, 0.1920, 0.2984, 0.2588, 0.2039, 0.0525, 0.1396], device='cuda:6'), in_proj_covar=tensor([0.0336, 0.0275, 0.0313, 0.0328, 0.0306, 0.0279, 0.0304, 0.0352], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 22:36:48,431 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=299599.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:36:55,727 INFO [train.py:904] (6/8) Epoch 30, batch 5250, loss[loss=0.1608, simple_loss=0.2562, pruned_loss=0.03272, over 15396.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2649, pruned_loss=0.04049, over 3216016.57 frames. ], batch size: 190, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:36:56,956 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.459e+02 2.002e+02 2.220e+02 2.747e+02 4.551e+02, threshold=4.439e+02, percent-clipped=1.0 2023-05-02 22:37:00,563 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 22:37:39,217 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5452, 3.6440, 3.4716, 3.1522, 3.2242, 3.5160, 3.2776, 3.3606], device='cuda:6'), covar=tensor([0.0644, 0.0612, 0.0355, 0.0302, 0.0684, 0.0524, 0.1751, 0.0524], device='cuda:6'), in_proj_covar=tensor([0.0319, 0.0483, 0.0375, 0.0376, 0.0371, 0.0435, 0.0256, 0.0445], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:37:56,977 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8511, 3.6231, 4.1763, 2.1770, 4.3645, 4.3856, 3.2335, 3.3401], device='cuda:6'), covar=tensor([0.0761, 0.0309, 0.0178, 0.1213, 0.0072, 0.0124, 0.0410, 0.0484], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0112, 0.0103, 0.0140, 0.0088, 0.0134, 0.0132, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 22:38:06,897 INFO [train.py:904] (6/8) Epoch 30, batch 5300, loss[loss=0.159, simple_loss=0.2492, pruned_loss=0.03435, over 16542.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2612, pruned_loss=0.03943, over 3204898.15 frames. ], batch size: 75, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:38:07,275 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.3759, 5.6568, 5.3980, 5.5030, 5.2395, 5.1928, 5.0290, 5.7832], device='cuda:6'), covar=tensor([0.1270, 0.0804, 0.0950, 0.0758, 0.0666, 0.0725, 0.1240, 0.0733], device='cuda:6'), in_proj_covar=tensor([0.0725, 0.0877, 0.0720, 0.0681, 0.0561, 0.0556, 0.0736, 0.0690], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:38:30,267 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8150, 2.0232, 2.4949, 2.7751, 2.7340, 3.1985, 2.2292, 3.1436], device='cuda:6'), covar=tensor([0.0267, 0.0603, 0.0382, 0.0367, 0.0401, 0.0211, 0.0615, 0.0191], device='cuda:6'), in_proj_covar=tensor([0.0200, 0.0200, 0.0189, 0.0196, 0.0213, 0.0169, 0.0206, 0.0170], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:6') 2023-05-02 22:38:42,831 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=299679.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:39:18,175 INFO [train.py:904] (6/8) Epoch 30, batch 5350, loss[loss=0.1684, simple_loss=0.2659, pruned_loss=0.03545, over 16508.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2596, pruned_loss=0.03875, over 3214782.06 frames. ], batch size: 62, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:39:19,347 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.376e+02 1.953e+02 2.222e+02 2.592e+02 5.137e+02, threshold=4.444e+02, percent-clipped=1.0 2023-05-02 22:40:29,704 INFO [train.py:904] (6/8) Epoch 30, batch 5400, loss[loss=0.1809, simple_loss=0.2844, pruned_loss=0.03866, over 16823.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2619, pruned_loss=0.0392, over 3208374.40 frames. ], batch size: 83, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:40:51,581 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=299769.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:41:10,036 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299782.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:41:17,123 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3758, 4.2421, 4.4144, 4.5656, 4.6904, 4.2867, 4.6824, 4.7184], device='cuda:6'), covar=tensor([0.1713, 0.1228, 0.1608, 0.0770, 0.0562, 0.1167, 0.0719, 0.0701], device='cuda:6'), in_proj_covar=tensor([0.0683, 0.0829, 0.0958, 0.0850, 0.0644, 0.0671, 0.0704, 0.0819], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:41:34,593 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=299798.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 22:41:46,646 INFO [train.py:904] (6/8) Epoch 30, batch 5450, loss[loss=0.1856, simple_loss=0.2711, pruned_loss=0.05, over 11671.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2653, pruned_loss=0.04096, over 3177183.78 frames. ], batch size: 246, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:41:47,796 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 2.110e+02 2.402e+02 2.864e+02 4.092e+02, threshold=4.805e+02, percent-clipped=0.0 2023-05-02 22:42:27,055 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299830.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:42:27,204 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=299830.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:42:27,277 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7361, 2.4533, 2.3200, 3.3382, 2.1157, 3.5447, 1.4893, 2.6902], device='cuda:6'), covar=tensor([0.1392, 0.0828, 0.1326, 0.0211, 0.0161, 0.0399, 0.1896, 0.0873], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0184, 0.0203, 0.0208, 0.0209, 0.0219, 0.0214, 0.0202], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 22:42:52,049 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299847.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:43:02,907 INFO [train.py:904] (6/8) Epoch 30, batch 5500, loss[loss=0.2186, simple_loss=0.2975, pruned_loss=0.06984, over 11850.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2722, pruned_loss=0.04516, over 3137763.77 frames. ], batch size: 247, lr: 2.23e-03, grad_scale: 16.0 2023-05-02 22:43:06,250 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1205, 2.8269, 3.0752, 1.7754, 3.1771, 3.2542, 2.7018, 2.4981], device='cuda:6'), covar=tensor([0.0908, 0.0338, 0.0225, 0.1264, 0.0131, 0.0241, 0.0509, 0.0558], device='cuda:6'), in_proj_covar=tensor([0.0150, 0.0112, 0.0103, 0.0140, 0.0088, 0.0134, 0.0132, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 22:43:11,661 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=299859.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 22:44:09,164 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299895.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:44:14,811 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299899.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:44:21,842 INFO [train.py:904] (6/8) Epoch 30, batch 5550, loss[loss=0.1784, simple_loss=0.2776, pruned_loss=0.03954, over 16641.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2789, pruned_loss=0.04961, over 3121153.89 frames. ], batch size: 89, lr: 2.23e-03, grad_scale: 16.0 2023-05-02 22:44:23,798 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.226e+02 2.904e+02 3.261e+02 4.114e+02 7.155e+02, threshold=6.523e+02, percent-clipped=13.0 2023-05-02 22:45:32,582 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299947.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:45:43,006 INFO [train.py:904] (6/8) Epoch 30, batch 5600, loss[loss=0.1954, simple_loss=0.2798, pruned_loss=0.05545, over 16630.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2818, pruned_loss=0.05235, over 3107017.97 frames. ], batch size: 62, lr: 2.23e-03, grad_scale: 16.0 2023-05-02 22:45:44,417 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-02 22:46:27,206 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299979.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:46:41,930 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7915, 3.8102, 3.9059, 3.6861, 3.8812, 4.2459, 3.8816, 3.6013], device='cuda:6'), covar=tensor([0.2195, 0.2350, 0.2758, 0.2693, 0.2637, 0.1971, 0.1906, 0.2780], device='cuda:6'), in_proj_covar=tensor([0.0430, 0.0635, 0.0707, 0.0515, 0.0687, 0.0725, 0.0545, 0.0686], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 22:47:11,368 INFO [train.py:904] (6/8) Epoch 30, batch 5650, loss[loss=0.2838, simple_loss=0.3422, pruned_loss=0.1127, over 11371.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2867, pruned_loss=0.05571, over 3111711.93 frames. ], batch size: 248, lr: 2.23e-03, grad_scale: 16.0 2023-05-02 22:47:13,271 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.022e+02 3.189e+02 3.868e+02 4.538e+02 6.994e+02, threshold=7.735e+02, percent-clipped=1.0 2023-05-02 22:47:30,037 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0996, 2.4608, 2.5631, 1.9644, 2.7086, 2.7688, 2.4224, 2.3848], device='cuda:6'), covar=tensor([0.0731, 0.0286, 0.0259, 0.0927, 0.0144, 0.0317, 0.0493, 0.0484], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0113, 0.0104, 0.0141, 0.0089, 0.0135, 0.0133, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 22:47:48,229 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=300027.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:48:27,596 INFO [train.py:904] (6/8) Epoch 30, batch 5700, loss[loss=0.1945, simple_loss=0.2907, pruned_loss=0.04916, over 16847.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2887, pruned_loss=0.05776, over 3086359.30 frames. ], batch size: 83, lr: 2.23e-03, grad_scale: 16.0 2023-05-02 22:48:41,921 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-02 22:48:57,374 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=300073.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:49:45,342 INFO [train.py:904] (6/8) Epoch 30, batch 5750, loss[loss=0.2325, simple_loss=0.297, pruned_loss=0.084, over 11577.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2912, pruned_loss=0.05943, over 3063059.58 frames. ], batch size: 246, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:49:49,226 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.972e+02 3.119e+02 3.894e+02 5.149e+02 1.112e+03, threshold=7.788e+02, percent-clipped=3.0 2023-05-02 22:50:18,758 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300125.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:50:35,593 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=300134.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:51:06,732 INFO [train.py:904] (6/8) Epoch 30, batch 5800, loss[loss=0.1692, simple_loss=0.263, pruned_loss=0.03772, over 16699.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.292, pruned_loss=0.05927, over 3041899.38 frames. ], batch size: 89, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:51:07,773 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300154.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 22:52:25,621 INFO [train.py:904] (6/8) Epoch 30, batch 5850, loss[loss=0.1971, simple_loss=0.2775, pruned_loss=0.05836, over 16980.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2895, pruned_loss=0.05753, over 3046843.58 frames. ], batch size: 41, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:52:28,953 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.843e+02 2.638e+02 2.992e+02 3.617e+02 7.830e+02, threshold=5.985e+02, percent-clipped=1.0 2023-05-02 22:52:39,925 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 22:53:30,647 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.0041, 3.1798, 3.1816, 2.1239, 3.0272, 3.2353, 3.0451, 1.9323], device='cuda:6'), covar=tensor([0.0622, 0.0081, 0.0094, 0.0506, 0.0130, 0.0140, 0.0130, 0.0529], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0093, 0.0094, 0.0137, 0.0105, 0.0118, 0.0101, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 22:53:48,213 INFO [train.py:904] (6/8) Epoch 30, batch 5900, loss[loss=0.199, simple_loss=0.2892, pruned_loss=0.05438, over 16751.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2884, pruned_loss=0.05677, over 3062816.21 frames. ], batch size: 124, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:54:06,819 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-02 22:55:08,461 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2211, 4.3077, 4.1433, 3.8417, 3.8244, 4.2271, 3.9086, 3.9906], device='cuda:6'), covar=tensor([0.0653, 0.0759, 0.0301, 0.0320, 0.0780, 0.0480, 0.0962, 0.0603], device='cuda:6'), in_proj_covar=tensor([0.0317, 0.0481, 0.0371, 0.0373, 0.0368, 0.0430, 0.0255, 0.0442], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:55:09,235 INFO [train.py:904] (6/8) Epoch 30, batch 5950, loss[loss=0.2002, simple_loss=0.2931, pruned_loss=0.05362, over 16237.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2894, pruned_loss=0.05548, over 3072130.57 frames. ], batch size: 165, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:55:12,894 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.875e+02 2.782e+02 3.249e+02 4.004e+02 6.648e+02, threshold=6.499e+02, percent-clipped=2.0 2023-05-02 22:56:29,059 INFO [train.py:904] (6/8) Epoch 30, batch 6000, loss[loss=0.1925, simple_loss=0.2819, pruned_loss=0.05159, over 16434.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2879, pruned_loss=0.05496, over 3067372.61 frames. ], batch size: 146, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:56:29,059 INFO [train.py:929] (6/8) Computing validation loss 2023-05-02 22:56:39,791 INFO [train.py:938] (6/8) Epoch 30, validation: loss=0.1471, simple_loss=0.2591, pruned_loss=0.01755, over 944034.00 frames. 2023-05-02 22:56:39,792 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-02 22:57:12,360 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-05-02 22:57:43,566 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2023-05-02 22:57:53,779 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3949, 1.6744, 2.0995, 2.2940, 2.3945, 2.5886, 1.8443, 2.5409], device='cuda:6'), covar=tensor([0.0228, 0.0580, 0.0336, 0.0383, 0.0355, 0.0246, 0.0614, 0.0168], device='cuda:6'), in_proj_covar=tensor([0.0198, 0.0198, 0.0187, 0.0194, 0.0211, 0.0168, 0.0204, 0.0168], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 22:57:57,799 INFO [train.py:904] (6/8) Epoch 30, batch 6050, loss[loss=0.2056, simple_loss=0.3021, pruned_loss=0.05456, over 16495.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2865, pruned_loss=0.0544, over 3078277.87 frames. ], batch size: 68, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:58:01,295 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.745e+02 2.541e+02 2.904e+02 3.584e+02 6.768e+02, threshold=5.808e+02, percent-clipped=1.0 2023-05-02 22:58:29,404 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=300425.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:58:35,952 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300429.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:59:15,657 INFO [train.py:904] (6/8) Epoch 30, batch 6100, loss[loss=0.1889, simple_loss=0.2979, pruned_loss=0.04, over 16890.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.287, pruned_loss=0.05462, over 3067882.26 frames. ], batch size: 96, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:59:17,342 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=300454.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 22:59:47,466 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=300473.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:00:30,284 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-02 23:00:32,833 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=300502.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 23:00:34,965 INFO [train.py:904] (6/8) Epoch 30, batch 6150, loss[loss=0.1759, simple_loss=0.2705, pruned_loss=0.04061, over 16458.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2851, pruned_loss=0.05395, over 3086192.79 frames. ], batch size: 68, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:00:38,226 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.879e+02 2.817e+02 3.397e+02 3.962e+02 7.100e+02, threshold=6.795e+02, percent-clipped=3.0 2023-05-02 23:01:33,858 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3114, 3.7527, 3.7241, 2.3688, 3.4888, 3.8209, 3.5015, 1.9025], device='cuda:6'), covar=tensor([0.0631, 0.0075, 0.0084, 0.0537, 0.0120, 0.0147, 0.0128, 0.0623], device='cuda:6'), in_proj_covar=tensor([0.0141, 0.0093, 0.0094, 0.0137, 0.0105, 0.0118, 0.0101, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 23:01:43,177 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-02 23:01:53,680 INFO [train.py:904] (6/8) Epoch 30, batch 6200, loss[loss=0.1935, simple_loss=0.2834, pruned_loss=0.05182, over 15276.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2838, pruned_loss=0.05376, over 3079772.27 frames. ], batch size: 190, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:03:13,157 INFO [train.py:904] (6/8) Epoch 30, batch 6250, loss[loss=0.2145, simple_loss=0.3027, pruned_loss=0.06316, over 16291.00 frames. ], tot_loss[loss=0.195, simple_loss=0.2833, pruned_loss=0.05333, over 3087205.34 frames. ], batch size: 165, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:03:16,359 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.883e+02 2.845e+02 3.157e+02 4.016e+02 6.897e+02, threshold=6.314e+02, percent-clipped=1.0 2023-05-02 23:03:34,814 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=300617.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:03:39,511 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=2.98 vs. limit=5.0 2023-05-02 23:04:11,994 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4322, 4.5507, 4.7004, 4.4647, 4.5386, 5.0600, 4.5086, 4.2936], device='cuda:6'), covar=tensor([0.1565, 0.1872, 0.2446, 0.2094, 0.2471, 0.1052, 0.1957, 0.2752], device='cuda:6'), in_proj_covar=tensor([0.0435, 0.0643, 0.0719, 0.0524, 0.0697, 0.0735, 0.0553, 0.0697], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 23:04:31,361 INFO [train.py:904] (6/8) Epoch 30, batch 6300, loss[loss=0.236, simple_loss=0.3071, pruned_loss=0.08248, over 11459.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2828, pruned_loss=0.05254, over 3111012.75 frames. ], batch size: 248, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:05:11,483 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=300678.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:05:24,970 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.7155, 4.9704, 4.7644, 4.7609, 4.5156, 4.4780, 4.4876, 5.0663], device='cuda:6'), covar=tensor([0.1266, 0.0890, 0.1008, 0.0937, 0.0865, 0.1438, 0.1153, 0.0825], device='cuda:6'), in_proj_covar=tensor([0.0724, 0.0874, 0.0718, 0.0679, 0.0558, 0.0556, 0.0731, 0.0685], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 23:05:50,438 INFO [train.py:904] (6/8) Epoch 30, batch 6350, loss[loss=0.183, simple_loss=0.2699, pruned_loss=0.04799, over 16922.00 frames. ], tot_loss[loss=0.1953, simple_loss=0.2835, pruned_loss=0.05351, over 3112716.68 frames. ], batch size: 109, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:05:53,979 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.720e+02 2.820e+02 3.153e+02 4.098e+02 7.645e+02, threshold=6.307e+02, percent-clipped=4.0 2023-05-02 23:06:29,151 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=300729.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:07:07,466 INFO [train.py:904] (6/8) Epoch 30, batch 6400, loss[loss=0.1845, simple_loss=0.2714, pruned_loss=0.04882, over 16693.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.2847, pruned_loss=0.05518, over 3098388.88 frames. ], batch size: 76, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:07:31,037 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8596, 3.7617, 3.9022, 4.0242, 4.0920, 3.6956, 4.0592, 4.1228], device='cuda:6'), covar=tensor([0.1710, 0.1156, 0.1409, 0.0737, 0.0724, 0.1900, 0.0970, 0.0795], device='cuda:6'), in_proj_covar=tensor([0.0685, 0.0830, 0.0959, 0.0850, 0.0644, 0.0669, 0.0707, 0.0820], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 23:07:42,793 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=300777.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:07:58,650 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.7705, 3.8390, 2.4789, 4.3120, 2.9476, 4.2688, 2.5984, 3.1142], device='cuda:6'), covar=tensor([0.0296, 0.0379, 0.1705, 0.0316, 0.0797, 0.0678, 0.1431, 0.0816], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0183, 0.0197, 0.0175, 0.0181, 0.0222, 0.0205, 0.0184], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 23:08:19,360 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=300802.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:08:21,347 INFO [train.py:904] (6/8) Epoch 30, batch 6450, loss[loss=0.1839, simple_loss=0.2826, pruned_loss=0.04257, over 17253.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2849, pruned_loss=0.0542, over 3110811.28 frames. ], batch size: 52, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:08:24,280 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.818e+02 2.836e+02 3.339e+02 4.246e+02 8.060e+02, threshold=6.678e+02, percent-clipped=2.0 2023-05-02 23:08:55,660 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9222, 2.7165, 2.8926, 2.1906, 2.7111, 2.2335, 2.7810, 2.9512], device='cuda:6'), covar=tensor([0.0278, 0.0911, 0.0462, 0.1845, 0.0820, 0.0927, 0.0517, 0.0708], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0172, 0.0172, 0.0158, 0.0149, 0.0134, 0.0146, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 23:09:31,623 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2023-05-02 23:09:38,120 INFO [train.py:904] (6/8) Epoch 30, batch 6500, loss[loss=0.1909, simple_loss=0.2836, pruned_loss=0.04912, over 15299.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2833, pruned_loss=0.05394, over 3100962.69 frames. ], batch size: 190, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:09:52,531 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=300863.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 23:10:32,299 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6659, 2.3199, 1.8902, 2.1005, 2.6522, 2.3396, 2.4026, 2.7724], device='cuda:6'), covar=tensor([0.0291, 0.0494, 0.0652, 0.0524, 0.0297, 0.0422, 0.0264, 0.0333], device='cuda:6'), in_proj_covar=tensor([0.0231, 0.0241, 0.0232, 0.0232, 0.0243, 0.0239, 0.0239, 0.0242], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 23:10:58,874 INFO [train.py:904] (6/8) Epoch 30, batch 6550, loss[loss=0.206, simple_loss=0.3082, pruned_loss=0.0519, over 16791.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2857, pruned_loss=0.05455, over 3107302.33 frames. ], batch size: 83, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:11:01,652 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.987e+02 2.626e+02 3.208e+02 3.708e+02 1.015e+03, threshold=6.415e+02, percent-clipped=2.0 2023-05-02 23:11:52,460 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.1233, 5.1851, 5.5151, 5.4676, 5.5208, 5.2075, 5.1349, 4.9230], device='cuda:6'), covar=tensor([0.0341, 0.0574, 0.0407, 0.0451, 0.0481, 0.0439, 0.0852, 0.0495], device='cuda:6'), in_proj_covar=tensor([0.0442, 0.0506, 0.0482, 0.0448, 0.0528, 0.0511, 0.0586, 0.0412], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 23:12:19,420 INFO [train.py:904] (6/8) Epoch 30, batch 6600, loss[loss=0.2021, simple_loss=0.2944, pruned_loss=0.05488, over 16225.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2869, pruned_loss=0.05477, over 3086730.09 frames. ], batch size: 165, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:12:19,864 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5543, 4.6311, 4.9154, 4.8766, 4.9087, 4.6299, 4.6100, 4.4686], device='cuda:6'), covar=tensor([0.0336, 0.0512, 0.0351, 0.0388, 0.0421, 0.0410, 0.0821, 0.0502], device='cuda:6'), in_proj_covar=tensor([0.0442, 0.0505, 0.0482, 0.0447, 0.0528, 0.0511, 0.0586, 0.0411], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 23:12:35,301 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.8753, 5.2554, 5.3894, 5.1219, 5.2180, 5.7552, 5.2294, 4.9639], device='cuda:6'), covar=tensor([0.1088, 0.1776, 0.2311, 0.1825, 0.2220, 0.0938, 0.1702, 0.2439], device='cuda:6'), in_proj_covar=tensor([0.0434, 0.0643, 0.0719, 0.0524, 0.0695, 0.0734, 0.0554, 0.0700], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 23:12:48,891 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300973.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:13:09,741 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5587, 2.2430, 1.9349, 2.0061, 2.5285, 2.2122, 2.2851, 2.6604], device='cuda:6'), covar=tensor([0.0232, 0.0427, 0.0535, 0.0481, 0.0284, 0.0381, 0.0242, 0.0294], device='cuda:6'), in_proj_covar=tensor([0.0231, 0.0241, 0.0232, 0.0232, 0.0243, 0.0239, 0.0239, 0.0242], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 23:13:38,992 INFO [train.py:904] (6/8) Epoch 30, batch 6650, loss[loss=0.249, simple_loss=0.3129, pruned_loss=0.09257, over 11644.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2867, pruned_loss=0.0552, over 3087278.80 frames. ], batch size: 248, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:13:43,932 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 2.674e+02 3.483e+02 3.945e+02 7.489e+02, threshold=6.965e+02, percent-clipped=2.0 2023-05-02 23:14:45,107 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.15 vs. limit=5.0 2023-05-02 23:14:49,188 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.8214, 2.4849, 2.3515, 3.2963, 2.2662, 3.5247, 1.6110, 2.7996], device='cuda:6'), covar=tensor([0.1384, 0.0810, 0.1364, 0.0200, 0.0170, 0.0424, 0.1816, 0.0855], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0183, 0.0202, 0.0208, 0.0208, 0.0220, 0.0212, 0.0202], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 23:14:55,111 INFO [train.py:904] (6/8) Epoch 30, batch 6700, loss[loss=0.1828, simple_loss=0.2622, pruned_loss=0.05172, over 17046.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2858, pruned_loss=0.05562, over 3097101.13 frames. ], batch size: 53, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:15:00,349 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3146, 3.3429, 2.1173, 3.6970, 2.6321, 3.7105, 2.3650, 2.8244], device='cuda:6'), covar=tensor([0.0332, 0.0457, 0.1737, 0.0231, 0.0811, 0.0571, 0.1504, 0.0753], device='cuda:6'), in_proj_covar=tensor([0.0178, 0.0183, 0.0197, 0.0175, 0.0181, 0.0222, 0.0205, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 23:16:08,409 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-05-02 23:16:11,407 INFO [train.py:904] (6/8) Epoch 30, batch 6750, loss[loss=0.1758, simple_loss=0.2676, pruned_loss=0.04196, over 16707.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2845, pruned_loss=0.05541, over 3110777.32 frames. ], batch size: 89, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:16:15,728 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.912e+02 2.866e+02 3.351e+02 4.005e+02 5.751e+02, threshold=6.702e+02, percent-clipped=0.0 2023-05-02 23:17:07,717 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=301140.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:17:29,009 INFO [train.py:904] (6/8) Epoch 30, batch 6800, loss[loss=0.1856, simple_loss=0.2788, pruned_loss=0.04622, over 16371.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2848, pruned_loss=0.05533, over 3125504.41 frames. ], batch size: 68, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:17:35,479 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=301158.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 23:18:43,946 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=301201.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:18:47,466 INFO [train.py:904] (6/8) Epoch 30, batch 6850, loss[loss=0.186, simple_loss=0.281, pruned_loss=0.04548, over 15376.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.286, pruned_loss=0.05612, over 3107893.03 frames. ], batch size: 190, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:18:51,709 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.931e+02 2.806e+02 3.426e+02 4.107e+02 8.215e+02, threshold=6.851e+02, percent-clipped=6.0 2023-05-02 23:19:08,360 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9743, 2.1744, 2.2410, 3.4950, 2.0423, 2.4751, 2.2559, 2.3000], device='cuda:6'), covar=tensor([0.1717, 0.3791, 0.3110, 0.0697, 0.4410, 0.2575, 0.3845, 0.3293], device='cuda:6'), in_proj_covar=tensor([0.0426, 0.0477, 0.0389, 0.0339, 0.0447, 0.0549, 0.0451, 0.0560], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 23:19:36,581 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9206, 3.2233, 3.1413, 1.7176, 2.6958, 1.8971, 3.3567, 3.6101], device='cuda:6'), covar=tensor([0.0262, 0.0908, 0.0852, 0.2755, 0.1184, 0.1418, 0.0694, 0.0836], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0173, 0.0172, 0.0158, 0.0149, 0.0134, 0.0147, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 23:20:04,219 INFO [train.py:904] (6/8) Epoch 30, batch 6900, loss[loss=0.2081, simple_loss=0.2975, pruned_loss=0.05941, over 16348.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2878, pruned_loss=0.05524, over 3123828.11 frames. ], batch size: 165, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:20:35,494 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=301273.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:20:41,760 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6512, 3.3940, 3.9260, 1.9460, 4.0921, 4.1133, 3.1001, 3.0068], device='cuda:6'), covar=tensor([0.0819, 0.0331, 0.0215, 0.1266, 0.0081, 0.0201, 0.0443, 0.0508], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0112, 0.0104, 0.0140, 0.0088, 0.0135, 0.0133, 0.0133], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 23:21:22,700 INFO [train.py:904] (6/8) Epoch 30, batch 6950, loss[loss=0.1849, simple_loss=0.2697, pruned_loss=0.05002, over 17165.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2897, pruned_loss=0.05742, over 3093985.29 frames. ], batch size: 46, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:21:26,935 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.877e+02 3.278e+02 4.034e+02 6.348e+02, threshold=6.555e+02, percent-clipped=0.0 2023-05-02 23:21:50,650 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=301321.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:22:38,929 INFO [train.py:904] (6/8) Epoch 30, batch 7000, loss[loss=0.1891, simple_loss=0.2847, pruned_loss=0.04671, over 17043.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.29, pruned_loss=0.05649, over 3101375.24 frames. ], batch size: 53, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:23:55,569 INFO [train.py:904] (6/8) Epoch 30, batch 7050, loss[loss=0.1902, simple_loss=0.2843, pruned_loss=0.04806, over 16215.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2903, pruned_loss=0.05599, over 3118768.40 frames. ], batch size: 165, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:24:00,750 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.808e+02 2.564e+02 3.175e+02 3.880e+02 7.852e+02, threshold=6.351e+02, percent-clipped=2.0 2023-05-02 23:24:32,130 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-02 23:24:38,967 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=301430.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:25:14,076 INFO [train.py:904] (6/8) Epoch 30, batch 7100, loss[loss=0.1768, simple_loss=0.2617, pruned_loss=0.04594, over 16583.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2887, pruned_loss=0.05596, over 3101048.74 frames. ], batch size: 57, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:25:20,990 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=301458.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:25:25,165 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4082, 3.7097, 3.4289, 1.8291, 2.8478, 2.1398, 3.6166, 3.9813], device='cuda:6'), covar=tensor([0.0241, 0.0802, 0.0787, 0.2766, 0.1205, 0.1316, 0.0703, 0.0986], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0173, 0.0172, 0.0158, 0.0149, 0.0134, 0.0147, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 23:26:14,377 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=301491.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:26:21,593 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=301496.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:26:31,245 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6869, 3.4898, 3.9621, 1.9576, 4.1483, 4.1925, 3.0795, 3.0228], device='cuda:6'), covar=tensor([0.0781, 0.0317, 0.0243, 0.1259, 0.0081, 0.0157, 0.0475, 0.0494], device='cuda:6'), in_proj_covar=tensor([0.0151, 0.0112, 0.0104, 0.0140, 0.0088, 0.0135, 0.0132, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-02 23:26:33,763 INFO [train.py:904] (6/8) Epoch 30, batch 7150, loss[loss=0.1998, simple_loss=0.2905, pruned_loss=0.05459, over 16497.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2871, pruned_loss=0.05608, over 3082568.91 frames. ], batch size: 68, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:26:36,533 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=301506.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:26:37,391 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.858e+02 2.724e+02 3.109e+02 4.036e+02 9.536e+02, threshold=6.218e+02, percent-clipped=1.0 2023-05-02 23:27:49,037 INFO [train.py:904] (6/8) Epoch 30, batch 7200, loss[loss=0.1825, simple_loss=0.2704, pruned_loss=0.04723, over 16734.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2848, pruned_loss=0.05419, over 3083174.58 frames. ], batch size: 134, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:28:00,833 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.8495, 3.9480, 4.1484, 4.1269, 4.1281, 3.9136, 3.9103, 3.9007], device='cuda:6'), covar=tensor([0.0390, 0.0630, 0.0408, 0.0389, 0.0450, 0.0483, 0.0873, 0.0604], device='cuda:6'), in_proj_covar=tensor([0.0443, 0.0506, 0.0483, 0.0448, 0.0528, 0.0513, 0.0588, 0.0412], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 23:29:10,982 INFO [train.py:904] (6/8) Epoch 30, batch 7250, loss[loss=0.1793, simple_loss=0.2567, pruned_loss=0.05093, over 16336.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2824, pruned_loss=0.05316, over 3092345.34 frames. ], batch size: 35, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:29:15,148 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.328e+02 2.684e+02 3.448e+02 5.554e+02, threshold=5.368e+02, percent-clipped=0.0 2023-05-02 23:30:27,540 INFO [train.py:904] (6/8) Epoch 30, batch 7300, loss[loss=0.1934, simple_loss=0.2924, pruned_loss=0.04724, over 16410.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2822, pruned_loss=0.05328, over 3086745.18 frames. ], batch size: 146, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:30:46,120 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4618, 4.6961, 4.4907, 4.4985, 4.2764, 4.1909, 4.2615, 4.7429], device='cuda:6'), covar=tensor([0.1098, 0.0773, 0.1070, 0.0933, 0.0672, 0.1497, 0.1043, 0.0851], device='cuda:6'), in_proj_covar=tensor([0.0722, 0.0872, 0.0719, 0.0681, 0.0556, 0.0557, 0.0732, 0.0685], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 23:31:44,854 INFO [train.py:904] (6/8) Epoch 30, batch 7350, loss[loss=0.1811, simple_loss=0.2696, pruned_loss=0.04632, over 16435.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2833, pruned_loss=0.05385, over 3080688.62 frames. ], batch size: 146, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:31:50,962 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.792e+02 2.739e+02 3.075e+02 3.816e+02 5.602e+02, threshold=6.150e+02, percent-clipped=1.0 2023-05-02 23:31:51,551 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2825, 3.3909, 3.5848, 2.1514, 3.1427, 2.3642, 3.5978, 3.7987], device='cuda:6'), covar=tensor([0.0274, 0.0918, 0.0625, 0.2219, 0.0905, 0.1052, 0.0674, 0.1011], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0173, 0.0172, 0.0159, 0.0150, 0.0135, 0.0147, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 23:32:44,231 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=301743.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:33:00,863 INFO [train.py:904] (6/8) Epoch 30, batch 7400, loss[loss=0.2036, simple_loss=0.2917, pruned_loss=0.05776, over 16650.00 frames. ], tot_loss[loss=0.1963, simple_loss=0.2844, pruned_loss=0.05406, over 3099487.58 frames. ], batch size: 57, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:33:44,389 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9006, 3.1954, 2.9705, 1.7715, 2.6651, 1.8632, 3.2943, 3.5780], device='cuda:6'), covar=tensor([0.0268, 0.0989, 0.0960, 0.2945, 0.1329, 0.1527, 0.0759, 0.0995], device='cuda:6'), in_proj_covar=tensor([0.0162, 0.0173, 0.0172, 0.0159, 0.0150, 0.0135, 0.0147, 0.0186], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 23:33:54,207 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=301786.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:34:10,829 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=301796.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:34:23,455 INFO [train.py:904] (6/8) Epoch 30, batch 7450, loss[loss=0.1959, simple_loss=0.2881, pruned_loss=0.05188, over 16777.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2852, pruned_loss=0.05513, over 3089368.43 frames. ], batch size: 124, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:34:24,252 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=301804.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:34:29,849 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.990e+02 2.877e+02 3.334e+02 4.200e+02 6.879e+02, threshold=6.668e+02, percent-clipped=3.0 2023-05-02 23:35:23,178 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5316, 3.4974, 3.4868, 2.6584, 3.3452, 2.0641, 3.1840, 2.8421], device='cuda:6'), covar=tensor([0.0207, 0.0181, 0.0229, 0.0279, 0.0150, 0.2479, 0.0164, 0.0303], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0175, 0.0214, 0.0186, 0.0191, 0.0218, 0.0202, 0.0179], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 23:35:30,252 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=301844.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:35:45,203 INFO [train.py:904] (6/8) Epoch 30, batch 7500, loss[loss=0.1908, simple_loss=0.278, pruned_loss=0.05179, over 16663.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.286, pruned_loss=0.0552, over 3087361.52 frames. ], batch size: 134, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:36:37,209 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0762, 4.3426, 3.2857, 2.8276, 3.2363, 3.0435, 4.7942, 3.8448], device='cuda:6'), covar=tensor([0.2829, 0.0688, 0.1787, 0.2457, 0.2438, 0.1859, 0.0408, 0.1283], device='cuda:6'), in_proj_covar=tensor([0.0339, 0.0276, 0.0317, 0.0330, 0.0307, 0.0281, 0.0305, 0.0353], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 23:36:57,925 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=301899.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:37:05,267 INFO [train.py:904] (6/8) Epoch 30, batch 7550, loss[loss=0.1911, simple_loss=0.2839, pruned_loss=0.0492, over 16667.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2855, pruned_loss=0.05541, over 3071978.08 frames. ], batch size: 89, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:37:07,765 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=301905.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:37:11,512 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.823e+02 2.767e+02 3.333e+02 4.100e+02 7.380e+02, threshold=6.666e+02, percent-clipped=1.0 2023-05-02 23:38:09,546 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.9487, 4.1686, 4.0095, 4.0467, 3.7519, 3.7831, 3.8175, 4.1551], device='cuda:6'), covar=tensor([0.1087, 0.0870, 0.1028, 0.0920, 0.0769, 0.1804, 0.0995, 0.1075], device='cuda:6'), in_proj_covar=tensor([0.0717, 0.0865, 0.0713, 0.0676, 0.0549, 0.0552, 0.0725, 0.0678], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:6') 2023-05-02 23:38:23,954 INFO [train.py:904] (6/8) Epoch 30, batch 7600, loss[loss=0.1843, simple_loss=0.2701, pruned_loss=0.04928, over 17127.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2842, pruned_loss=0.05501, over 3083604.82 frames. ], batch size: 47, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:38:34,841 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=301960.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 23:38:44,309 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=301966.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:39:14,529 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 23:39:47,150 INFO [train.py:904] (6/8) Epoch 30, batch 7650, loss[loss=0.2177, simple_loss=0.2939, pruned_loss=0.07075, over 11164.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.2844, pruned_loss=0.05561, over 3086162.73 frames. ], batch size: 247, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:39:53,136 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 2.050e+02 2.820e+02 3.410e+02 4.459e+02 8.175e+02, threshold=6.820e+02, percent-clipped=3.0 2023-05-02 23:41:04,693 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([5.0773, 5.0929, 4.8803, 4.1617, 5.0234, 1.7769, 4.7016, 4.4693], device='cuda:6'), covar=tensor([0.0112, 0.0094, 0.0206, 0.0469, 0.0094, 0.3005, 0.0138, 0.0313], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0174, 0.0213, 0.0185, 0.0190, 0.0217, 0.0202, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 23:41:06,720 INFO [train.py:904] (6/8) Epoch 30, batch 7700, loss[loss=0.1907, simple_loss=0.2803, pruned_loss=0.05052, over 16439.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2845, pruned_loss=0.05583, over 3092207.89 frames. ], batch size: 68, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:41:10,202 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5207, 3.5078, 3.4913, 2.6602, 3.3476, 2.1448, 3.1454, 2.7334], device='cuda:6'), covar=tensor([0.0204, 0.0160, 0.0206, 0.0244, 0.0127, 0.2482, 0.0167, 0.0297], device='cuda:6'), in_proj_covar=tensor([0.0184, 0.0174, 0.0213, 0.0185, 0.0190, 0.0217, 0.0202, 0.0178], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 23:41:47,715 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9301, 2.2192, 2.2045, 3.5993, 2.1132, 2.4962, 2.2893, 2.3521], device='cuda:6'), covar=tensor([0.1707, 0.3541, 0.3323, 0.0736, 0.4411, 0.2588, 0.3775, 0.3481], device='cuda:6'), in_proj_covar=tensor([0.0426, 0.0479, 0.0390, 0.0341, 0.0449, 0.0551, 0.0453, 0.0562], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-02 23:41:57,575 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=302086.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:42:18,118 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302099.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:42:26,385 INFO [train.py:904] (6/8) Epoch 30, batch 7750, loss[loss=0.2007, simple_loss=0.2974, pruned_loss=0.05204, over 16736.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.285, pruned_loss=0.05588, over 3089333.69 frames. ], batch size: 39, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:42:33,833 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 2.636e+02 3.091e+02 3.789e+02 8.720e+02, threshold=6.183e+02, percent-clipped=2.0 2023-05-02 23:43:12,802 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=302134.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:43:31,587 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 23:43:41,536 INFO [train.py:904] (6/8) Epoch 30, batch 7800, loss[loss=0.2566, simple_loss=0.3115, pruned_loss=0.1008, over 11306.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2858, pruned_loss=0.05643, over 3081490.69 frames. ], batch size: 246, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:44:03,916 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-05-02 23:44:37,000 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-02 23:44:48,747 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4941, 3.2947, 2.6709, 2.1739, 2.2066, 2.2883, 3.4438, 3.0221], device='cuda:6'), covar=tensor([0.3032, 0.0683, 0.1926, 0.3067, 0.2836, 0.2420, 0.0557, 0.1463], device='cuda:6'), in_proj_covar=tensor([0.0340, 0.0276, 0.0316, 0.0330, 0.0308, 0.0281, 0.0305, 0.0353], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-02 23:44:55,080 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=302200.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:45:00,016 INFO [train.py:904] (6/8) Epoch 30, batch 7850, loss[loss=0.1997, simple_loss=0.2869, pruned_loss=0.05623, over 15247.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2864, pruned_loss=0.05645, over 3057660.69 frames. ], batch size: 191, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:45:06,745 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.702e+02 3.253e+02 3.935e+02 9.747e+02, threshold=6.506e+02, percent-clipped=3.0 2023-05-02 23:46:15,231 INFO [train.py:904] (6/8) Epoch 30, batch 7900, loss[loss=0.2307, simple_loss=0.2992, pruned_loss=0.08106, over 11344.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2858, pruned_loss=0.05627, over 3035752.08 frames. ], batch size: 246, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:46:16,927 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302255.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 23:46:25,379 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302261.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:46:25,499 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=302261.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:47:07,650 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=4.00 vs. limit=5.0 2023-05-02 23:47:22,487 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=302296.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:47:32,970 INFO [train.py:904] (6/8) Epoch 30, batch 7950, loss[loss=0.2229, simple_loss=0.2904, pruned_loss=0.07765, over 11854.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2861, pruned_loss=0.05705, over 3022426.14 frames. ], batch size: 248, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:47:41,024 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.902e+02 2.712e+02 3.186e+02 3.569e+02 8.017e+02, threshold=6.372e+02, percent-clipped=2.0 2023-05-02 23:47:51,432 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1495, 3.4050, 3.4095, 2.2879, 3.2085, 3.4660, 3.1900, 1.9974], device='cuda:6'), covar=tensor([0.0553, 0.0086, 0.0077, 0.0447, 0.0123, 0.0125, 0.0132, 0.0516], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0092, 0.0093, 0.0136, 0.0103, 0.0118, 0.0100, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 23:48:48,634 INFO [train.py:904] (6/8) Epoch 30, batch 8000, loss[loss=0.2158, simple_loss=0.3059, pruned_loss=0.06283, over 15372.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2866, pruned_loss=0.05717, over 3039518.01 frames. ], batch size: 190, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:48:54,515 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=302357.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:49:06,258 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.1616, 3.5220, 3.5153, 2.2870, 3.2656, 3.5496, 3.2739, 1.9762], device='cuda:6'), covar=tensor([0.0592, 0.0081, 0.0078, 0.0466, 0.0124, 0.0128, 0.0124, 0.0530], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0092, 0.0093, 0.0136, 0.0103, 0.0117, 0.0100, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 23:49:57,543 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=302399.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:50:03,889 INFO [train.py:904] (6/8) Epoch 30, batch 8050, loss[loss=0.1861, simple_loss=0.2775, pruned_loss=0.04739, over 16211.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2864, pruned_loss=0.05648, over 3066032.91 frames. ], batch size: 165, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:50:11,661 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.866e+02 2.982e+02 3.316e+02 3.990e+02 1.007e+03, threshold=6.633e+02, percent-clipped=4.0 2023-05-02 23:51:10,642 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=302447.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:51:10,856 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1741, 4.8755, 4.7897, 3.3773, 4.1646, 4.7576, 4.0064, 2.9275], device='cuda:6'), covar=tensor([0.0485, 0.0060, 0.0049, 0.0366, 0.0109, 0.0118, 0.0105, 0.0416], device='cuda:6'), in_proj_covar=tensor([0.0139, 0.0092, 0.0093, 0.0136, 0.0103, 0.0117, 0.0100, 0.0132], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:6') 2023-05-02 23:51:21,030 INFO [train.py:904] (6/8) Epoch 30, batch 8100, loss[loss=0.1913, simple_loss=0.2863, pruned_loss=0.04814, over 16909.00 frames. ], tot_loss[loss=0.1987, simple_loss=0.2858, pruned_loss=0.05579, over 3071941.34 frames. ], batch size: 109, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:51:56,193 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=302477.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:52:35,992 INFO [train.py:904] (6/8) Epoch 30, batch 8150, loss[loss=0.1822, simple_loss=0.2686, pruned_loss=0.04789, over 15349.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.283, pruned_loss=0.05414, over 3099059.33 frames. ], batch size: 190, lr: 2.22e-03, grad_scale: 8.0 2023-05-02 23:52:43,485 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.950e+02 2.749e+02 3.387e+02 4.131e+02 6.245e+02, threshold=6.774e+02, percent-clipped=0.0 2023-05-02 23:53:28,235 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=302538.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 23:53:52,248 INFO [train.py:904] (6/8) Epoch 30, batch 8200, loss[loss=0.2066, simple_loss=0.2809, pruned_loss=0.06617, over 11700.00 frames. ], tot_loss[loss=0.1942, simple_loss=0.281, pruned_loss=0.05372, over 3111371.29 frames. ], batch size: 247, lr: 2.22e-03, grad_scale: 8.0 2023-05-02 23:53:53,895 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=302555.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:53:55,135 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302556.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:54:03,323 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=302561.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:54:22,028 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2489, 4.3347, 4.4267, 4.2080, 4.3999, 4.8436, 4.3715, 4.0439], device='cuda:6'), covar=tensor([0.1623, 0.2104, 0.2425, 0.2214, 0.2482, 0.1025, 0.1729, 0.2628], device='cuda:6'), in_proj_covar=tensor([0.0433, 0.0644, 0.0718, 0.0525, 0.0696, 0.0736, 0.0556, 0.0700], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 23:55:13,043 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=302603.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:55:13,852 INFO [train.py:904] (6/8) Epoch 30, batch 8250, loss[loss=0.1891, simple_loss=0.2921, pruned_loss=0.0431, over 16423.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2796, pruned_loss=0.05106, over 3081116.20 frames. ], batch size: 146, lr: 2.22e-03, grad_scale: 8.0 2023-05-02 23:55:22,071 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.554e+02 2.948e+02 3.684e+02 6.863e+02, threshold=5.896e+02, percent-clipped=1.0 2023-05-02 23:55:23,081 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=302609.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:55:38,305 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-05-02 23:55:57,492 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 23:56:02,644 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=302632.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:56:17,823 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4857, 3.7328, 4.1386, 2.4394, 3.3823, 2.6838, 3.9939, 3.9818], device='cuda:6'), covar=tensor([0.0234, 0.1014, 0.0453, 0.2097, 0.0851, 0.1009, 0.0541, 0.0989], device='cuda:6'), in_proj_covar=tensor([0.0161, 0.0172, 0.0171, 0.0158, 0.0149, 0.0134, 0.0146, 0.0185], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:6') 2023-05-02 23:56:35,852 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302652.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:56:38,717 INFO [train.py:904] (6/8) Epoch 30, batch 8300, loss[loss=0.1645, simple_loss=0.2629, pruned_loss=0.033, over 15299.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2765, pruned_loss=0.04795, over 3074899.85 frames. ], batch size: 191, lr: 2.22e-03, grad_scale: 8.0 2023-05-02 23:57:13,036 INFO [scaling.py:679] (6/8) Whitening: num_groups=1, num_channels=384, metric=3.37 vs. limit=5.0 2023-05-02 23:57:34,002 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.2636, 3.4577, 3.6425, 3.6196, 3.6205, 3.4898, 3.3527, 3.5211], device='cuda:6'), covar=tensor([0.0766, 0.1153, 0.0800, 0.0763, 0.0828, 0.0975, 0.1385, 0.0800], device='cuda:6'), in_proj_covar=tensor([0.0441, 0.0502, 0.0480, 0.0445, 0.0525, 0.0509, 0.0585, 0.0410], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:6') 2023-05-02 23:57:43,564 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=302693.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 23:57:53,885 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5234, 4.6093, 4.7971, 4.5593, 4.6826, 5.1948, 4.7092, 4.4081], device='cuda:6'), covar=tensor([0.1327, 0.2040, 0.2315, 0.2057, 0.2294, 0.0897, 0.1530, 0.2349], device='cuda:6'), in_proj_covar=tensor([0.0429, 0.0638, 0.0711, 0.0520, 0.0690, 0.0729, 0.0551, 0.0693], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-02 23:58:01,685 INFO [train.py:904] (6/8) Epoch 30, batch 8350, loss[loss=0.1754, simple_loss=0.2674, pruned_loss=0.04173, over 11963.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.276, pruned_loss=0.0467, over 3048675.24 frames. ], batch size: 247, lr: 2.22e-03, grad_scale: 8.0 2023-05-02 23:58:09,433 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.283e+02 2.623e+02 3.174e+02 4.916e+02, threshold=5.246e+02, percent-clipped=0.0 2023-05-02 23:58:19,071 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 23:58:41,450 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-05-02 23:59:22,769 INFO [train.py:904] (6/8) Epoch 30, batch 8400, loss[loss=0.176, simple_loss=0.2698, pruned_loss=0.04114, over 16581.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2735, pruned_loss=0.04451, over 3055441.06 frames. ], batch size: 62, lr: 2.22e-03, grad_scale: 8.0 2023-05-02 23:59:37,565 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.5779, 3.6385, 3.3972, 3.1017, 3.2201, 3.5264, 3.3321, 3.3689], device='cuda:6'), covar=tensor([0.0616, 0.0840, 0.0334, 0.0299, 0.0492, 0.0582, 0.1425, 0.0532], device='cuda:6'), in_proj_covar=tensor([0.0312, 0.0474, 0.0364, 0.0365, 0.0359, 0.0421, 0.0252, 0.0434], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:6') 2023-05-03 00:00:29,103 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5186, 3.0657, 2.7641, 2.3138, 2.2528, 2.3460, 3.0418, 2.8646], device='cuda:6'), covar=tensor([0.2728, 0.0656, 0.1701, 0.2987, 0.2960, 0.2495, 0.0477, 0.1604], device='cuda:6'), in_proj_covar=tensor([0.0335, 0.0272, 0.0312, 0.0326, 0.0304, 0.0278, 0.0301, 0.0349], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-03 00:00:44,413 INFO [train.py:904] (6/8) Epoch 30, batch 8450, loss[loss=0.1763, simple_loss=0.2767, pruned_loss=0.03801, over 16335.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2721, pruned_loss=0.043, over 3063628.20 frames. ], batch size: 68, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:00:52,080 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.019e+02 2.324e+02 2.808e+02 4.179e+02, threshold=4.647e+02, percent-clipped=0.0 2023-05-03 00:01:20,262 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.7886, 2.9115, 2.5637, 4.4231, 2.9778, 4.1566, 1.5799, 3.1756], device='cuda:6'), covar=tensor([0.1384, 0.0725, 0.1182, 0.0176, 0.0136, 0.0339, 0.1738, 0.0633], device='cuda:6'), in_proj_covar=tensor([0.0175, 0.0182, 0.0202, 0.0206, 0.0207, 0.0219, 0.0211, 0.0200], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-03 00:01:31,215 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302833.0, num_to_drop=1, layers_to_drop={1} 2023-05-03 00:02:04,572 INFO [train.py:904] (6/8) Epoch 30, batch 8500, loss[loss=0.1622, simple_loss=0.2565, pruned_loss=0.03391, over 16406.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2687, pruned_loss=0.04122, over 3050000.71 frames. ], batch size: 68, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:02:08,549 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=302856.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:02:48,209 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=302881.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:03:26,596 INFO [train.py:904] (6/8) Epoch 30, batch 8550, loss[loss=0.1736, simple_loss=0.2695, pruned_loss=0.03882, over 15394.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2668, pruned_loss=0.04038, over 3057043.37 frames. ], batch size: 191, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:03:28,404 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=302904.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:03:38,833 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.520e+02 2.175e+02 2.493e+02 2.958e+02 5.835e+02, threshold=4.986e+02, percent-clipped=2.0 2023-05-03 00:04:26,472 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-03 00:04:45,023 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=302942.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:04:54,968 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-03 00:05:04,940 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=302952.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:05:08,027 INFO [train.py:904] (6/8) Epoch 30, batch 8600, loss[loss=0.1623, simple_loss=0.2653, pruned_loss=0.02967, over 15349.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2668, pruned_loss=0.03943, over 3050678.55 frames. ], batch size: 192, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:05:25,886 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.3290, 3.1082, 3.3179, 1.8589, 3.4664, 3.5080, 2.8800, 2.7913], device='cuda:6'), covar=tensor([0.0809, 0.0286, 0.0237, 0.1244, 0.0100, 0.0207, 0.0448, 0.0476], device='cuda:6'), in_proj_covar=tensor([0.0148, 0.0110, 0.0102, 0.0137, 0.0086, 0.0132, 0.0130, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-03 00:06:16,088 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302988.0, num_to_drop=1, layers_to_drop={2} 2023-05-03 00:06:39,315 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=303000.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:06:46,606 INFO [train.py:904] (6/8) Epoch 30, batch 8650, loss[loss=0.1682, simple_loss=0.2636, pruned_loss=0.03641, over 15343.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2651, pruned_loss=0.03801, over 3047915.34 frames. ], batch size: 192, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:06:58,831 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.356e+02 2.161e+02 2.643e+02 3.389e+02 5.700e+02, threshold=5.286e+02, percent-clipped=3.0 2023-05-03 00:08:31,269 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303052.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:08:34,579 INFO [train.py:904] (6/8) Epoch 30, batch 8700, loss[loss=0.1591, simple_loss=0.2465, pruned_loss=0.03581, over 12233.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2629, pruned_loss=0.03694, over 3067936.62 frames. ], batch size: 247, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:10:14,424 INFO [train.py:904] (6/8) Epoch 30, batch 8750, loss[loss=0.1601, simple_loss=0.2515, pruned_loss=0.03435, over 12510.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2627, pruned_loss=0.03646, over 3073032.26 frames. ], batch size: 248, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:10:25,189 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 2.109e+02 2.507e+02 3.042e+02 7.094e+02, threshold=5.015e+02, percent-clipped=1.0 2023-05-03 00:10:38,717 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=303113.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:11:22,336 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.4384, 3.6003, 3.6327, 2.5325, 3.3271, 3.6830, 3.4829, 1.6669], device='cuda:6'), covar=tensor([0.0580, 0.0132, 0.0098, 0.0498, 0.0165, 0.0155, 0.0116, 0.0803], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0090, 0.0092, 0.0134, 0.0103, 0.0115, 0.0098, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-03 00:11:23,663 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=303133.0, num_to_drop=1, layers_to_drop={0} 2023-05-03 00:11:39,039 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.0743, 2.2214, 2.1227, 3.5893, 2.1193, 2.5096, 2.3128, 2.2935], device='cuda:6'), covar=tensor([0.1537, 0.3983, 0.3633, 0.0723, 0.4705, 0.2857, 0.3920, 0.4074], device='cuda:6'), in_proj_covar=tensor([0.0419, 0.0472, 0.0385, 0.0332, 0.0441, 0.0541, 0.0445, 0.0552], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-03 00:11:41,508 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-03 00:12:06,331 INFO [train.py:904] (6/8) Epoch 30, batch 8800, loss[loss=0.175, simple_loss=0.2654, pruned_loss=0.04231, over 12301.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2613, pruned_loss=0.03508, over 3086268.95 frames. ], batch size: 248, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:13:03,253 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=303181.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:13:14,966 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6483, 2.2265, 2.1384, 3.5222, 1.9027, 3.4553, 1.5037, 2.6361], device='cuda:6'), covar=tensor([0.1660, 0.1028, 0.1576, 0.0231, 0.0109, 0.0514, 0.2074, 0.0978], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0180, 0.0199, 0.0203, 0.0203, 0.0216, 0.0209, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-03 00:13:50,919 INFO [train.py:904] (6/8) Epoch 30, batch 8850, loss[loss=0.1527, simple_loss=0.2485, pruned_loss=0.0285, over 12366.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2637, pruned_loss=0.03473, over 3066958.19 frames. ], batch size: 246, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:14:00,695 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.249e+02 2.609e+02 3.163e+02 6.486e+02, threshold=5.217e+02, percent-clipped=3.0 2023-05-03 00:15:01,738 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-03 00:15:03,424 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=303237.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:15:29,034 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.6858, 3.9314, 4.0130, 2.9144, 3.4851, 3.9926, 3.6995, 2.3802], device='cuda:6'), covar=tensor([0.0460, 0.0063, 0.0050, 0.0352, 0.0128, 0.0103, 0.0084, 0.0458], device='cuda:6'), in_proj_covar=tensor([0.0138, 0.0090, 0.0091, 0.0134, 0.0102, 0.0115, 0.0098, 0.0130], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-03 00:15:38,681 INFO [train.py:904] (6/8) Epoch 30, batch 8900, loss[loss=0.1766, simple_loss=0.2729, pruned_loss=0.04011, over 16217.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2643, pruned_loss=0.03422, over 3072686.49 frames. ], batch size: 165, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:15:49,493 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303258.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:17:03,849 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=303288.0, num_to_drop=1, layers_to_drop={1} 2023-05-03 00:17:43,147 INFO [train.py:904] (6/8) Epoch 30, batch 8950, loss[loss=0.154, simple_loss=0.2487, pruned_loss=0.02966, over 16992.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2636, pruned_loss=0.03449, over 3095680.08 frames. ], batch size: 116, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:17:53,309 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 2.074e+02 2.389e+02 2.805e+02 4.954e+02, threshold=4.779e+02, percent-clipped=0.0 2023-05-03 00:18:01,467 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.2461, 4.3208, 4.1300, 3.8202, 3.8424, 4.2342, 3.9222, 4.0358], device='cuda:6'), covar=tensor([0.0544, 0.0475, 0.0337, 0.0323, 0.0728, 0.0463, 0.0844, 0.0611], device='cuda:6'), in_proj_covar=tensor([0.0308, 0.0467, 0.0360, 0.0361, 0.0355, 0.0416, 0.0248, 0.0428], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-03 00:18:16,169 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=303319.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:18:17,915 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.8915, 2.1644, 2.4411, 3.1556, 2.2422, 2.3506, 2.3480, 2.2747], device='cuda:6'), covar=tensor([0.1490, 0.3620, 0.3018, 0.0847, 0.4479, 0.2703, 0.3625, 0.3907], device='cuda:6'), in_proj_covar=tensor([0.0421, 0.0474, 0.0387, 0.0334, 0.0442, 0.0543, 0.0448, 0.0553], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-03 00:18:53,962 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=303336.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:19:32,192 INFO [train.py:904] (6/8) Epoch 30, batch 9000, loss[loss=0.1456, simple_loss=0.2384, pruned_loss=0.02642, over 12062.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2604, pruned_loss=0.03326, over 3080037.41 frames. ], batch size: 248, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:19:32,192 INFO [train.py:929] (6/8) Computing validation loss 2023-05-03 00:19:42,082 INFO [train.py:938] (6/8) Epoch 30, validation: loss=0.1431, simple_loss=0.2465, pruned_loss=0.01984, over 944034.00 frames. 2023-05-03 00:19:42,083 INFO [train.py:939] (6/8) Maximum memory allocated so far is 17688MB 2023-05-03 00:21:28,438 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4677, 4.4508, 4.3116, 3.5925, 4.3803, 1.9397, 4.1261, 4.0476], device='cuda:6'), covar=tensor([0.0176, 0.0139, 0.0224, 0.0383, 0.0199, 0.2761, 0.0250, 0.0331], device='cuda:6'), in_proj_covar=tensor([0.0181, 0.0172, 0.0211, 0.0182, 0.0188, 0.0216, 0.0199, 0.0176], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-03 00:21:28,479 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303403.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:21:29,165 INFO [train.py:904] (6/8) Epoch 30, batch 9050, loss[loss=0.1572, simple_loss=0.2447, pruned_loss=0.03482, over 16661.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2614, pruned_loss=0.03387, over 3083810.98 frames. ], batch size: 134, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:21:34,456 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-03 00:21:38,493 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=303408.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:21:39,555 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.159e+02 2.466e+02 3.038e+02 8.070e+02, threshold=4.932e+02, percent-clipped=3.0 2023-05-03 00:23:13,931 INFO [train.py:904] (6/8) Epoch 30, batch 9100, loss[loss=0.1758, simple_loss=0.274, pruned_loss=0.03876, over 15382.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2614, pruned_loss=0.03479, over 3075261.07 frames. ], batch size: 190, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:23:33,670 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=303464.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:24:32,198 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4395, 4.4987, 4.3387, 4.0364, 4.0822, 4.4321, 4.1422, 4.1548], device='cuda:6'), covar=tensor([0.0603, 0.0785, 0.0365, 0.0316, 0.0800, 0.0614, 0.0638, 0.0720], device='cuda:6'), in_proj_covar=tensor([0.0307, 0.0466, 0.0360, 0.0360, 0.0353, 0.0414, 0.0248, 0.0428], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-03 00:25:12,849 INFO [train.py:904] (6/8) Epoch 30, batch 9150, loss[loss=0.1556, simple_loss=0.2571, pruned_loss=0.02703, over 16711.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2621, pruned_loss=0.03455, over 3070210.70 frames. ], batch size: 76, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:25:25,335 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.268e+02 2.639e+02 3.152e+02 6.290e+02, threshold=5.277e+02, percent-clipped=4.0 2023-05-03 00:26:02,165 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.5493, 4.6948, 4.8757, 4.6211, 4.7283, 5.2139, 4.7202, 4.4377], device='cuda:6'), covar=tensor([0.1313, 0.1987, 0.2180, 0.2032, 0.2212, 0.0878, 0.1652, 0.2499], device='cuda:6'), in_proj_covar=tensor([0.0420, 0.0625, 0.0700, 0.0510, 0.0674, 0.0719, 0.0541, 0.0681], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-03 00:26:23,067 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=303537.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:26:54,943 INFO [train.py:904] (6/8) Epoch 30, batch 9200, loss[loss=0.172, simple_loss=0.252, pruned_loss=0.04594, over 12588.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2578, pruned_loss=0.03372, over 3070004.95 frames. ], batch size: 248, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:27:32,027 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303574.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:27:52,353 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=303585.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:28:29,158 INFO [train.py:904] (6/8) Epoch 30, batch 9250, loss[loss=0.1478, simple_loss=0.2462, pruned_loss=0.0247, over 16396.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.257, pruned_loss=0.03322, over 3057078.47 frames. ], batch size: 68, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:28:41,899 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.148e+02 2.583e+02 3.115e+02 7.022e+02, threshold=5.165e+02, percent-clipped=1.0 2023-05-03 00:28:49,566 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=303614.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:29:39,061 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=303635.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:30:19,442 INFO [train.py:904] (6/8) Epoch 30, batch 9300, loss[loss=0.1427, simple_loss=0.2344, pruned_loss=0.02557, over 16504.00 frames. ], tot_loss[loss=0.1606, simple_loss=0.2557, pruned_loss=0.03272, over 3073709.43 frames. ], batch size: 68, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:30:46,229 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.4472, 4.5541, 4.7190, 4.4658, 4.6248, 5.0631, 4.5955, 4.2740], device='cuda:6'), covar=tensor([0.1293, 0.1868, 0.1978, 0.2215, 0.2246, 0.0946, 0.1576, 0.2495], device='cuda:6'), in_proj_covar=tensor([0.0418, 0.0621, 0.0697, 0.0508, 0.0671, 0.0718, 0.0539, 0.0677], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:6') 2023-05-03 00:31:04,341 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([1.6927, 2.6774, 2.5766, 4.2762, 2.5600, 3.9805, 1.6842, 2.9726], device='cuda:6'), covar=tensor([0.1471, 0.0846, 0.1217, 0.0185, 0.0144, 0.0410, 0.1693, 0.0739], device='cuda:6'), in_proj_covar=tensor([0.0173, 0.0180, 0.0199, 0.0202, 0.0202, 0.0215, 0.0210, 0.0198], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:6') 2023-05-03 00:32:05,574 INFO [train.py:904] (6/8) Epoch 30, batch 9350, loss[loss=0.1888, simple_loss=0.2805, pruned_loss=0.04856, over 15310.00 frames. ], tot_loss[loss=0.161, simple_loss=0.2559, pruned_loss=0.03305, over 3061299.10 frames. ], batch size: 191, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:32:13,944 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=303708.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:32:16,692 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.024e+02 2.345e+02 2.902e+02 6.854e+02, threshold=4.690e+02, percent-clipped=2.0 2023-05-03 00:32:44,062 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.7749, 2.2061, 1.9750, 2.0382, 2.4901, 2.1297, 2.1379, 2.6250], device='cuda:6'), covar=tensor([0.0203, 0.0497, 0.0605, 0.0542, 0.0326, 0.0506, 0.0201, 0.0328], device='cuda:6'), in_proj_covar=tensor([0.0227, 0.0241, 0.0231, 0.0232, 0.0242, 0.0239, 0.0235, 0.0239], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-03 00:33:12,242 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.1275, 2.2500, 2.3385, 3.8043, 2.1319, 2.5525, 2.3655, 2.3793], device='cuda:6'), covar=tensor([0.1463, 0.3822, 0.3282, 0.0611, 0.4501, 0.2718, 0.3881, 0.3550], device='cuda:6'), in_proj_covar=tensor([0.0419, 0.0472, 0.0387, 0.0333, 0.0441, 0.0540, 0.0446, 0.0551], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-03 00:33:46,246 INFO [train.py:904] (6/8) Epoch 30, batch 9400, loss[loss=0.1542, simple_loss=0.2659, pruned_loss=0.02123, over 16821.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2557, pruned_loss=0.03288, over 3063542.67 frames. ], batch size: 96, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:33:50,783 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=303756.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:33:56,165 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=303759.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:35:26,329 INFO [train.py:904] (6/8) Epoch 30, batch 9450, loss[loss=0.1642, simple_loss=0.2585, pruned_loss=0.03495, over 16354.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.257, pruned_loss=0.03299, over 3058038.86 frames. ], batch size: 146, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:35:36,995 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.278e+02 2.120e+02 2.504e+02 3.073e+02 7.133e+02, threshold=5.009e+02, percent-clipped=4.0 2023-05-03 00:35:38,528 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303810.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:37:08,345 INFO [train.py:904] (6/8) Epoch 30, batch 9500, loss[loss=0.1476, simple_loss=0.2481, pruned_loss=0.02357, over 16467.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.2562, pruned_loss=0.03246, over 3062160.39 frames. ], batch size: 68, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:37:43,923 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=303871.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:37:59,323 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-03 00:38:36,354 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303896.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:38:54,501 INFO [train.py:904] (6/8) Epoch 30, batch 9550, loss[loss=0.164, simple_loss=0.2567, pruned_loss=0.03565, over 12586.00 frames. ], tot_loss[loss=0.16, simple_loss=0.2554, pruned_loss=0.03227, over 3064152.97 frames. ], batch size: 246, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:39:08,517 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 2.050e+02 2.496e+02 3.159e+02 5.329e+02, threshold=4.992e+02, percent-clipped=3.0 2023-05-03 00:39:18,143 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=303914.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:39:48,222 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.9329, 2.2152, 2.4192, 3.2166, 2.2382, 2.4106, 2.3524, 2.3325], device='cuda:6'), covar=tensor([0.1369, 0.3638, 0.2939, 0.0753, 0.4449, 0.2682, 0.3792, 0.3857], device='cuda:6'), in_proj_covar=tensor([0.0420, 0.0473, 0.0388, 0.0333, 0.0443, 0.0542, 0.0448, 0.0552], device='cuda:6'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-03 00:39:49,715 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=303930.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:40:36,088 INFO [train.py:904] (6/8) Epoch 30, batch 9600, loss[loss=0.1726, simple_loss=0.278, pruned_loss=0.03353, over 16395.00 frames. ], tot_loss[loss=0.1618, simple_loss=0.2574, pruned_loss=0.03308, over 3060254.35 frames. ], batch size: 146, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:40:37,627 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-05-03 00:40:43,416 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=303957.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:40:53,327 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=303962.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:41:16,192 INFO [zipformer.py:625] (6/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303974.0, num_to_drop=1, layers_to_drop={0} 2023-05-03 00:42:31,192 INFO [train.py:904] (6/8) Epoch 30, batch 9650, loss[loss=0.151, simple_loss=0.2447, pruned_loss=0.0287, over 16817.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2592, pruned_loss=0.03341, over 3073592.95 frames. ], batch size: 39, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:42:48,095 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.176e+02 2.625e+02 3.264e+02 9.247e+02, threshold=5.250e+02, percent-clipped=2.0 2023-05-03 00:43:11,347 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([3.3421, 4.7139, 4.7046, 3.5867, 3.9105, 4.6733, 4.1405, 2.7476], device='cuda:6'), covar=tensor([0.0441, 0.0059, 0.0045, 0.0317, 0.0140, 0.0104, 0.0084, 0.0502], device='cuda:6'), in_proj_covar=tensor([0.0137, 0.0089, 0.0091, 0.0133, 0.0102, 0.0113, 0.0096, 0.0129], device='cuda:6'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:6') 2023-05-03 00:43:41,218 INFO [zipformer.py:625] (6/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=304035.0, num_to_drop=1, layers_to_drop={2} 2023-05-03 00:44:20,836 INFO [train.py:904] (6/8) Epoch 30, batch 9700, loss[loss=0.1831, simple_loss=0.2866, pruned_loss=0.03986, over 16799.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2583, pruned_loss=0.03328, over 3062939.38 frames. ], batch size: 124, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:44:30,120 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=304059.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:44:39,544 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.3150, 4.1527, 4.3665, 4.4919, 4.6456, 4.2444, 4.6108, 4.6616], device='cuda:6'), covar=tensor([0.1820, 0.1256, 0.1553, 0.0782, 0.0540, 0.1078, 0.0719, 0.0813], device='cuda:6'), in_proj_covar=tensor([0.0650, 0.0786, 0.0905, 0.0814, 0.0613, 0.0635, 0.0674, 0.0779], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-03 00:45:23,026 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([4.1608, 4.0213, 4.2112, 4.3273, 4.4833, 4.1018, 4.4573, 4.5013], device='cuda:6'), covar=tensor([0.1987, 0.1346, 0.1688, 0.0871, 0.0627, 0.1237, 0.0760, 0.0937], device='cuda:6'), in_proj_covar=tensor([0.0650, 0.0787, 0.0906, 0.0814, 0.0614, 0.0636, 0.0674, 0.0780], device='cuda:6'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:6') 2023-05-03 00:46:04,097 INFO [train.py:904] (6/8) Epoch 30, batch 9750, loss[loss=0.1703, simple_loss=0.277, pruned_loss=0.03182, over 16361.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2577, pruned_loss=0.03373, over 3051638.32 frames. ], batch size: 146, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:46:09,338 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=304107.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:46:17,047 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.422e+02 2.052e+02 2.422e+02 2.861e+02 5.308e+02, threshold=4.844e+02, percent-clipped=2.0 2023-05-03 00:47:41,832 INFO [train.py:904] (6/8) Epoch 30, batch 9800, loss[loss=0.1698, simple_loss=0.2719, pruned_loss=0.03384, over 16857.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2577, pruned_loss=0.03302, over 3068127.97 frames. ], batch size: 116, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:48:05,278 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=304166.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:49:24,990 INFO [train.py:904] (6/8) Epoch 30, batch 9850, loss[loss=0.153, simple_loss=0.2564, pruned_loss=0.02474, over 16789.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2589, pruned_loss=0.03289, over 3073122.05 frames. ], batch size: 83, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:49:39,349 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.408e+02 2.072e+02 2.330e+02 2.794e+02 5.653e+02, threshold=4.660e+02, percent-clipped=2.0 2023-05-03 00:50:19,679 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=304230.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:51:13,334 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=304252.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:51:16,010 INFO [train.py:904] (6/8) Epoch 30, batch 9900, loss[loss=0.145, simple_loss=0.2382, pruned_loss=0.0259, over 12566.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.259, pruned_loss=0.03273, over 3065839.78 frames. ], batch size: 246, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:52:13,324 INFO [zipformer.py:625] (6/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=304278.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:53:14,778 INFO [train.py:904] (6/8) Epoch 30, batch 9950, loss[loss=0.1748, simple_loss=0.2698, pruned_loss=0.0399, over 12439.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2613, pruned_loss=0.03318, over 3076030.75 frames. ], batch size: 248, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:53:31,713 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.067e+02 2.374e+02 2.801e+02 5.953e+02, threshold=4.748e+02, percent-clipped=1.0 2023-05-03 00:54:21,393 INFO [zipformer.py:625] (6/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=304330.0, num_to_drop=1, layers_to_drop={1} 2023-05-03 00:54:55,735 INFO [zipformer.py:1454] (6/8) attn_weights_entropy = tensor([2.5247, 3.5927, 2.8036, 2.1653, 2.2436, 2.4139, 3.7266, 3.0890], device='cuda:6'), covar=tensor([0.3166, 0.0625, 0.1933, 0.3179, 0.2832, 0.2374, 0.0459, 0.1483], device='cuda:6'), in_proj_covar=tensor([0.0331, 0.0269, 0.0309, 0.0323, 0.0297, 0.0276, 0.0297, 0.0345], device='cuda:6'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:6') 2023-05-03 00:55:15,643 INFO [train.py:904] (6/8) Epoch 30, batch 10000, loss[loss=0.1652, simple_loss=0.2592, pruned_loss=0.03563, over 16807.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2604, pruned_loss=0.03287, over 3089871.22 frames. ], batch size: 124, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:56:57,636 INFO [train.py:904] (6/8) Epoch 30, batch 10050, loss[loss=0.1916, simple_loss=0.2786, pruned_loss=0.05231, over 11973.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2603, pruned_loss=0.03292, over 3077631.39 frames. ], batch size: 247, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:57:10,556 INFO [optim.py:368] (6/8) Clipping_scale=2.0, grad-norm quartiles 1.432e+02 2.064e+02 2.395e+02 2.910e+02 6.626e+02, threshold=4.790e+02, percent-clipped=3.0 2023-05-03 00:58:34,282 INFO [train.py:904] (6/8) Epoch 30, batch 10100, loss[loss=0.1485, simple_loss=0.2429, pruned_loss=0.02706, over 16133.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2605, pruned_loss=0.03287, over 3093342.60 frames. ], batch size: 165, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:58:57,146 INFO [zipformer.py:625] (6/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=304466.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:59:26,295 INFO [scaling.py:679] (6/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-05-03 00:59:55,823 INFO [train.py:1169] (6/8) Done!