Commit
•
8ed4a53
1
Parent(s):
50ba432
Upload folder using huggingface_hub
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- .gitattributes +15 -0
- finetuning_concatenated_config.json +54 -0
- run_parler_tts_training.py +1763 -0
- wandb/debug-cli.sanchit.log +0 -0
- wandb/debug-internal.log +0 -0
- wandb/debug.log +35 -0
- wandb/run-20240513_204652-m0g0ap7d/files/conda-environment.yaml +248 -0
- wandb/run-20240513_204652-m0g0ap7d/files/config.yaml +86 -0
- wandb/run-20240513_204652-m0g0ap7d/files/output.log +180 -0
- wandb/run-20240513_204652-m0g0ap7d/files/requirements.txt +225 -0
- wandb/run-20240513_204652-m0g0ap7d/files/wandb-metadata.json +706 -0
- wandb/run-20240513_204652-m0g0ap7d/files/wandb-summary.json +1 -0
- wandb/run-20240513_204652-m0g0ap7d/logs/debug-internal.log +455 -0
- wandb/run-20240513_204652-m0g0ap7d/logs/debug.log +29 -0
- wandb/run-20240513_204652-m0g0ap7d/run-m0g0ap7d.wandb +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/conda-environment.yaml +248 -0
- wandb/run-20240513_205249-qaoje1x9/files/config.yaml +88 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_193b029d494fd24e7cfa.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_345bfb6a72849809d361.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_3c9adbd9374e0fb5ce3d.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_3cd94e4824cc6c8fb09c.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_3ed6544e58dd861a5d9e.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_9da1fed11be9d614d9ec.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_ec838b0233dbe87d33f3.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_0db946e177a69cbe11f5.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_32c9af8d48e757598000.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_341c52fd92336c009f67.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_43ed5d3749c912acb591.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_75818e76e9e077f058be.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_d24330f3382b9e6ea7ea.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_ec7dcb5421538131ede7.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_2794dcaf322bd12e2814.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_2ef5b33e2eaf98dca4a6.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_4ca836a112634417b82e.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_837a3499e3f93538b643.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_b3650df61e399b05257d.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_d33ccbefe990db0dce2b.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_e8ca5038019cad3cde86.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_4899e0da4615e883ad13.wav +3 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_492597073098578f0605.wav +3 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_67d9409a306e3614ec3f.wav +3 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_7c47ba927ac118ffaacc.wav +3 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_89ad32d31f3e70178cc1.wav +3 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_cd644667186ae0518a3c.wav +3 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_f7405ef7b645b3265477.wav +3 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_508_00e9064c0bdbd6b9428d.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_508_11adac906cb7e2ef30c6.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_508_5619d97860f92fc1a62d.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_508_bcb03b95f0470920bdc6.wav +0 -0
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_508_e48b4ff2b12d5ffdb11c.wav +0 -0
.gitattributes
CHANGED
@@ -33,3 +33,18 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_4899e0da4615e883ad13.wav filter=lfs diff=lfs merge=lfs -text
|
37 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_492597073098578f0605.wav filter=lfs diff=lfs merge=lfs -text
|
38 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_67d9409a306e3614ec3f.wav filter=lfs diff=lfs merge=lfs -text
|
39 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_7c47ba927ac118ffaacc.wav filter=lfs diff=lfs merge=lfs -text
|
40 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_89ad32d31f3e70178cc1.wav filter=lfs diff=lfs merge=lfs -text
|
41 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_cd644667186ae0518a3c.wav filter=lfs diff=lfs merge=lfs -text
|
42 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_f7405ef7b645b3265477.wav filter=lfs diff=lfs merge=lfs -text
|
43 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_09db3d4c95c46a3bae02.wav filter=lfs diff=lfs merge=lfs -text
|
44 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_3349ee3cb71e3d789f00.wav filter=lfs diff=lfs merge=lfs -text
|
45 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_63964662df5bc7731b7a.wav filter=lfs diff=lfs merge=lfs -text
|
46 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_7ab4dededd2d3a052def.wav filter=lfs diff=lfs merge=lfs -text
|
47 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_b8be5a72ea66619af007.wav filter=lfs diff=lfs merge=lfs -text
|
48 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_d1c21fd700287f9f3109.wav filter=lfs diff=lfs merge=lfs -text
|
49 |
+
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_e8feadb7228a5b136ce8.wav filter=lfs diff=lfs merge=lfs -text
|
50 |
+
wandb/run-20240513_205249-qaoje1x9/run-qaoje1x9.wandb filter=lfs diff=lfs merge=lfs -text
|
finetuning_concatenated_config.json
ADDED
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"model_name_or_path": "parler-tts/parler_tts_mini_v0.1",
|
3 |
+
"feature_extractor_name": "parler-tts/dac_44khZ_8kbps",
|
4 |
+
"description_tokenizer_name": "parler-tts/parler_tts_mini_v0.1",
|
5 |
+
"prompt_tokenizer_name": "parler-tts/parler_tts_mini_v0.1",
|
6 |
+
"report_to": ["wandb"],
|
7 |
+
"overwrite_output_dir": true,
|
8 |
+
"train_dataset_name": "sanchit-gandhi/expresso-concatenated-half-normal+reach-vb/jenny_tts_dataset+sanchit-gandhi/libritts_r_test+sanchit-gandhi/libritts_r_test",
|
9 |
+
"train_metadata_dataset_name": "sanchit-gandhi/expresso-concatenated-half-normal-tags-mistral+ylacombe/jenny-tts-10k-tagged+parler-tts/libritts_r_tags_tagged_10k_generated+parler-tts/libritts_r_tags_tagged_10k_generated",
|
10 |
+
"train_dataset_config_name": "read+default+clean+other",
|
11 |
+
"train_split_name": "train+train[:20%]+test.clean+test.other",
|
12 |
+
"eval_dataset_name": "sanchit-gandhi/expresso-concatenated-half-normal+reach-vb/jenny_tts_dataset+sanchit-gandhi/libritts_r_test+sanchit-gandhi/libritts_r_test",
|
13 |
+
"eval_metadata_dataset_name": "sanchit-gandhi/expresso-concatenated-half-normal-tags-mistral+ylacombe/jenny-tts-10k-tagged+parler-tts/libritts_r_tags_tagged_10k_generated+parler-tts/libritts_r_tags_tagged_10k_generated",
|
14 |
+
"eval_dataset_config_name": "read+default+clean+other",
|
15 |
+
"eval_split_name": "train+train[:20%]+test.clean+test.other",
|
16 |
+
"max_eval_samples": 8,
|
17 |
+
"per_device_eval_batch_size": 16,
|
18 |
+
"target_audio_column_name": "audio",
|
19 |
+
"description_column_name": "text_description",
|
20 |
+
"prompt_column_name": "text",
|
21 |
+
"max_duration_in_seconds": 30.0,
|
22 |
+
"min_duration_in_seconds": 2.0,
|
23 |
+
"max_text_length": 400,
|
24 |
+
"preprocessing_num_workers": 2,
|
25 |
+
"do_train": true,
|
26 |
+
"num_train_epochs": 8,
|
27 |
+
"max_steps": -1,
|
28 |
+
"gradient_accumulation_steps": 8,
|
29 |
+
"gradient_checkpointing": true,
|
30 |
+
"per_device_train_batch_size": 16,
|
31 |
+
"learning_rate": 0.00008,
|
32 |
+
"adam_beta1": 0.9,
|
33 |
+
"adam_beta2": 0.99,
|
34 |
+
"weight_decay": 0.01,
|
35 |
+
"lr_scheduler_type": "cosine",
|
36 |
+
"warmup_steps": 250,
|
37 |
+
"logging_steps": 5,
|
38 |
+
"freeze_text_encoder": true,
|
39 |
+
"audio_encoder_per_device_batch_size": 4,
|
40 |
+
"dtype": "bfloat16",
|
41 |
+
"seed": 456,
|
42 |
+
"output_dir": "../output_dir_training_constant_concat/",
|
43 |
+
"temporary_save_to_disk": "../audio_code_tmp_constant_concat/",
|
44 |
+
"save_to_disk": "../tmp_dataset_audio_constant_concat/",
|
45 |
+
"dataloader_num_workers": 4,
|
46 |
+
"do_eval": true,
|
47 |
+
"predict_with_generate": true,
|
48 |
+
"include_inputs_for_metrics": true,
|
49 |
+
"save_strategy": "epoch",
|
50 |
+
"evaluation_strategy": "epoch",
|
51 |
+
"save_total_limit": 5,
|
52 |
+
"group_by_length": true
|
53 |
+
}
|
54 |
+
|
run_parler_tts_training.py
ADDED
@@ -0,0 +1,1763 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
#!/usr/bin/env python
|
2 |
+
# coding=utf-8
|
3 |
+
# Copyright 2024 The HuggingFace Inc. team. All rights reserved.
|
4 |
+
#
|
5 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
6 |
+
# you may not use this file except in compliance with the License.
|
7 |
+
# You may obtain a copy of the License at
|
8 |
+
#
|
9 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
10 |
+
#
|
11 |
+
# Unless required by applicable law or agreed to in writing, software
|
12 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
13 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
14 |
+
# See the License for the specific language governing permissions and
|
15 |
+
# limitations under the License.
|
16 |
+
|
17 |
+
""" Train Parler-TTS using 🤗 Accelerate"""
|
18 |
+
|
19 |
+
import logging
|
20 |
+
import os
|
21 |
+
import re
|
22 |
+
import shutil
|
23 |
+
import sys
|
24 |
+
import time
|
25 |
+
from dataclasses import dataclass, field
|
26 |
+
from datetime import timedelta
|
27 |
+
from pathlib import Path
|
28 |
+
from typing import Dict, List, Optional, Set, Union
|
29 |
+
|
30 |
+
import datasets
|
31 |
+
import evaluate
|
32 |
+
import numpy as np
|
33 |
+
import torch
|
34 |
+
import transformers
|
35 |
+
from accelerate import Accelerator
|
36 |
+
from accelerate.utils import AutocastKwargs, InitProcessGroupKwargs, TorchDynamoPlugin, set_seed
|
37 |
+
from accelerate.utils.memory import release_memory
|
38 |
+
from datasets import Dataset, DatasetDict, IterableDataset, concatenate_datasets, interleave_datasets, load_dataset
|
39 |
+
from huggingface_hub import Repository, create_repo
|
40 |
+
from multiprocess import set_start_method
|
41 |
+
from torch.utils.data import DataLoader
|
42 |
+
from tqdm import tqdm
|
43 |
+
from transformers import (
|
44 |
+
AutoFeatureExtractor,
|
45 |
+
AutoModel,
|
46 |
+
AutoProcessor,
|
47 |
+
AutoTokenizer,
|
48 |
+
HfArgumentParser,
|
49 |
+
Seq2SeqTrainingArguments,
|
50 |
+
pipeline,
|
51 |
+
)
|
52 |
+
from transformers.optimization import get_scheduler
|
53 |
+
from transformers.trainer_pt_utils import LengthGroupedSampler
|
54 |
+
from transformers.utils import send_example_telemetry
|
55 |
+
from wandb import Audio
|
56 |
+
|
57 |
+
from parler_tts import (
|
58 |
+
ParlerTTSConfig,
|
59 |
+
ParlerTTSForConditionalGeneration,
|
60 |
+
build_delay_pattern_mask,
|
61 |
+
)
|
62 |
+
|
63 |
+
|
64 |
+
logger = logging.getLogger(__name__)
|
65 |
+
|
66 |
+
|
67 |
+
def list_field(default=None, metadata=None):
|
68 |
+
return field(default_factory=lambda: default, metadata=metadata)
|
69 |
+
|
70 |
+
|
71 |
+
_RE_CHECKPOINT = re.compile(r"^checkpoint-(\d+)-epoch-(\d+)$")
|
72 |
+
|
73 |
+
|
74 |
+
def get_last_checkpoint(folder):
|
75 |
+
content = os.listdir(folder)
|
76 |
+
checkpoints = [
|
77 |
+
path
|
78 |
+
for path in content
|
79 |
+
if _RE_CHECKPOINT.search(path) is not None and os.path.isdir(os.path.join(folder, path))
|
80 |
+
]
|
81 |
+
if len(checkpoints) == 0:
|
82 |
+
return
|
83 |
+
return os.path.join(folder, max(checkpoints, key=lambda x: int(_RE_CHECKPOINT.search(x).groups()[0])))
|
84 |
+
|
85 |
+
|
86 |
+
def sorted_checkpoints(output_dir=None, checkpoint_prefix="checkpoint") -> List[str]:
|
87 |
+
"""Helper function to sort saved checkpoints from oldest to newest."""
|
88 |
+
ordering_and_checkpoint_path = []
|
89 |
+
|
90 |
+
glob_checkpoints = [str(x) for x in Path(output_dir).glob(f"{checkpoint_prefix}-*") if os.path.isdir(x)]
|
91 |
+
|
92 |
+
for path in glob_checkpoints:
|
93 |
+
regex_match = re.match(f".*{checkpoint_prefix}-([0-9]+)", path)
|
94 |
+
if regex_match is not None and regex_match.groups() is not None:
|
95 |
+
ordering_and_checkpoint_path.append((int(regex_match.groups()[0]), path))
|
96 |
+
|
97 |
+
checkpoints_sorted = sorted(ordering_and_checkpoint_path)
|
98 |
+
checkpoints_sorted = [checkpoint[1] for checkpoint in checkpoints_sorted]
|
99 |
+
return checkpoints_sorted
|
100 |
+
|
101 |
+
|
102 |
+
def rotate_checkpoints(save_total_limit=None, output_dir=None, checkpoint_prefix="checkpoint") -> None:
|
103 |
+
"""Helper function to delete old checkpoints."""
|
104 |
+
if save_total_limit is None or save_total_limit <= 0:
|
105 |
+
return
|
106 |
+
# Check if we should delete older checkpoint(s)
|
107 |
+
checkpoints_sorted = sorted_checkpoints(output_dir=output_dir, checkpoint_prefix=checkpoint_prefix)
|
108 |
+
if len(checkpoints_sorted) <= save_total_limit:
|
109 |
+
return
|
110 |
+
|
111 |
+
number_of_checkpoints_to_delete = max(0, len(checkpoints_sorted) - save_total_limit)
|
112 |
+
checkpoints_to_be_deleted = checkpoints_sorted[:number_of_checkpoints_to_delete]
|
113 |
+
for checkpoint in checkpoints_to_be_deleted:
|
114 |
+
logger.info(f"Deleting older checkpoint [{checkpoint}] due to args.save_total_limit")
|
115 |
+
shutil.rmtree(checkpoint, ignore_errors=True)
|
116 |
+
|
117 |
+
|
118 |
+
def log_metric(
|
119 |
+
accelerator,
|
120 |
+
metrics: Dict,
|
121 |
+
train_time: float,
|
122 |
+
step: int,
|
123 |
+
epoch: int,
|
124 |
+
learning_rate: float = None,
|
125 |
+
prefix: str = "train",
|
126 |
+
):
|
127 |
+
"""Helper function to log all training/evaluation metrics with the correct prefixes and styling."""
|
128 |
+
log_metrics = {}
|
129 |
+
for k, v in metrics.items():
|
130 |
+
log_metrics[f"{prefix}/{k}"] = v
|
131 |
+
log_metrics[f"{prefix}/time"] = train_time
|
132 |
+
log_metrics[f"{prefix}/epoch"] = epoch
|
133 |
+
if learning_rate is not None:
|
134 |
+
log_metrics[f"{prefix}/learning_rate"] = learning_rate
|
135 |
+
accelerator.log(log_metrics, step=step)
|
136 |
+
|
137 |
+
|
138 |
+
def log_pred(
|
139 |
+
accelerator,
|
140 |
+
pred_descriptions: List[str],
|
141 |
+
pred_prompts: List[str],
|
142 |
+
transcriptions: List[str],
|
143 |
+
audios: List[torch.Tensor],
|
144 |
+
sampling_rate: int,
|
145 |
+
step: int,
|
146 |
+
prefix: str = "eval",
|
147 |
+
num_lines: int = 200000,
|
148 |
+
):
|
149 |
+
"""Helper function to log target/predicted transcriptions to weights and biases (wandb)."""
|
150 |
+
if accelerator.is_main_process:
|
151 |
+
wandb_tracker = accelerator.get_tracker("wandb")
|
152 |
+
# pretty name for current step: step 50000 -> step 50k
|
153 |
+
cur_step_pretty = f"{int(step // 1000)}k" if step > 1000 else step
|
154 |
+
prefix_pretty = prefix.replace("/", "-")
|
155 |
+
|
156 |
+
# convert str data to a wandb compatible format
|
157 |
+
str_data = [[pred_descriptions[i], pred_prompts[i], transcriptions[i]] for i in range(len(pred_descriptions))]
|
158 |
+
# log as a table with the appropriate headers
|
159 |
+
wandb_tracker.log_table(
|
160 |
+
table_name=f"predictions/{prefix_pretty}-step-{cur_step_pretty}",
|
161 |
+
columns=["Target descriptions", "Target prompts", "Predicted transcriptions"],
|
162 |
+
data=str_data[:num_lines],
|
163 |
+
step=step,
|
164 |
+
commit=False,
|
165 |
+
)
|
166 |
+
|
167 |
+
# wandb can only loads 100 audios per step
|
168 |
+
wandb_tracker.log(
|
169 |
+
{
|
170 |
+
f"Speech samples/{prefix}": [
|
171 |
+
Audio(
|
172 |
+
audio,
|
173 |
+
caption=f"{pred_prompts[i]} --- DESCRIPTION: {pred_descriptions[i]}",
|
174 |
+
sample_rate=sampling_rate,
|
175 |
+
)
|
176 |
+
for (i, audio) in enumerate(audios[: min(len(audios), 100)])
|
177 |
+
]
|
178 |
+
},
|
179 |
+
step=step,
|
180 |
+
)
|
181 |
+
|
182 |
+
|
183 |
+
@dataclass
|
184 |
+
class ModelArguments:
|
185 |
+
"""
|
186 |
+
Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
|
187 |
+
"""
|
188 |
+
|
189 |
+
model_name_or_path: str = field(
|
190 |
+
metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"}
|
191 |
+
)
|
192 |
+
config_name: Optional[str] = field(
|
193 |
+
default=None, metadata={"help": "Pretrained config name or path if not the same as model_name"}
|
194 |
+
)
|
195 |
+
feature_extractor_name: Optional[str] = field(
|
196 |
+
default=None, metadata={"help": "Pretrained feature extractor name or path if not the same as model_name"}
|
197 |
+
)
|
198 |
+
description_tokenizer_name: Optional[str] = field(
|
199 |
+
default=None, metadata={"help": "Pretrained description tokenizer name or path if not the same as model_name"}
|
200 |
+
)
|
201 |
+
prompt_tokenizer_name: Optional[str] = field(
|
202 |
+
default=None,
|
203 |
+
metadata={"help": "Pretrained prompt tokenizer name or path if not the same as description_tokenizer_name"},
|
204 |
+
)
|
205 |
+
cache_dir: Optional[str] = field(
|
206 |
+
default=None,
|
207 |
+
metadata={"help": "Where to store the pretrained models downloaded from huggingface.co"},
|
208 |
+
)
|
209 |
+
use_fast_tokenizer: bool = field(
|
210 |
+
default=True,
|
211 |
+
metadata={"help": "Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."},
|
212 |
+
)
|
213 |
+
model_revision: str = field(
|
214 |
+
default="main",
|
215 |
+
metadata={"help": "The specific model version to use (can be a branch name, tag name or commit id)."},
|
216 |
+
)
|
217 |
+
pad_token_id: int = field(
|
218 |
+
default=None,
|
219 |
+
metadata={"help": "If specified, change the model pad token id."},
|
220 |
+
)
|
221 |
+
decoder_start_token_id: int = field(
|
222 |
+
default=None,
|
223 |
+
metadata={"help": "If specified, change the model decoder start token id."},
|
224 |
+
)
|
225 |
+
freeze_text_encoder: bool = field(
|
226 |
+
default=False,
|
227 |
+
metadata={"help": "Whether to freeze the text encoder."},
|
228 |
+
)
|
229 |
+
do_sample: bool = field(
|
230 |
+
default=True,
|
231 |
+
metadata={"help": "Whether to do sampling or greedy decoding."},
|
232 |
+
)
|
233 |
+
temperature: float = field(
|
234 |
+
default=1.0,
|
235 |
+
metadata={"help": "Temperature if sampling."},
|
236 |
+
)
|
237 |
+
max_length: int = field(
|
238 |
+
default=2580,
|
239 |
+
metadata={"help": "Generation max length."},
|
240 |
+
)
|
241 |
+
bandwidth: float = field(
|
242 |
+
default=6,
|
243 |
+
metadata={"help": "Audio encoder bandwidth."},
|
244 |
+
)
|
245 |
+
asr_model_name_or_path: str = field(
|
246 |
+
default="distil-whisper/distil-large-v2",
|
247 |
+
metadata={
|
248 |
+
"help": "Used to compute WER during evaluation. Path to pretrained model or model identifier from huggingface.co/models"
|
249 |
+
},
|
250 |
+
)
|
251 |
+
clap_model_name_or_path: str = field(
|
252 |
+
default="laion/larger_clap_music_and_speech",
|
253 |
+
metadata={
|
254 |
+
"help": "Used to compute audio similarity during evaluation. Path to pretrained model or model identifier from huggingface.co/models"
|
255 |
+
},
|
256 |
+
)
|
257 |
+
|
258 |
+
|
259 |
+
@dataclass
|
260 |
+
class DataTrainingArguments:
|
261 |
+
"""
|
262 |
+
Arguments pertaining to what data we are going to input our model for training and eval.
|
263 |
+
|
264 |
+
Using `HfArgumentParser` we can turn this class
|
265 |
+
into argparse arguments to be able to specify them on
|
266 |
+
the command line.
|
267 |
+
"""
|
268 |
+
|
269 |
+
train_dataset_name: str = field(
|
270 |
+
default=None,
|
271 |
+
metadata={
|
272 |
+
"help": "The name of the training dataset to use (via the datasets library). Load and combine "
|
273 |
+
"multiple datasets by separating dataset ids by a '+' symbol. For example, to load and combine "
|
274 |
+
" librispeech and common voice, set `train_dataset_name='librispeech_asr+common_voice'`."
|
275 |
+
},
|
276 |
+
)
|
277 |
+
train_dataset_config_name: Optional[str] = field(
|
278 |
+
default=None,
|
279 |
+
metadata={
|
280 |
+
"help": "The configuration name of the training dataset to use (via the datasets library). Load and combine "
|
281 |
+
"multiple datasets by separating dataset configs by a '+' symbol."
|
282 |
+
},
|
283 |
+
)
|
284 |
+
train_split_name: str = field(
|
285 |
+
default="train",
|
286 |
+
metadata={
|
287 |
+
"help": ("The name of the training data set split to use (via the datasets library). Defaults to 'train'")
|
288 |
+
},
|
289 |
+
)
|
290 |
+
train_dataset_samples: str = field(
|
291 |
+
default=None,
|
292 |
+
metadata={
|
293 |
+
"help": "Number of samples in the training data. Load and combine "
|
294 |
+
"multiple datasets by separating dataset samples by a '+' symbol."
|
295 |
+
},
|
296 |
+
)
|
297 |
+
train_metadata_dataset_name: str = field(
|
298 |
+
default=None,
|
299 |
+
metadata={
|
300 |
+
"help": "The name of the metadata training dataset to use (via the datasets library). Load and combine "
|
301 |
+
"multiple datasets by separating dataset ids by a '+' symbol. For example, to load and combine "
|
302 |
+
" librispeech and common voice, set `train_dataset_name='librispeech_asr+common_voice'`."
|
303 |
+
},
|
304 |
+
)
|
305 |
+
eval_dataset_name: str = field(
|
306 |
+
default=None,
|
307 |
+
metadata={
|
308 |
+
"help": "The name of the evaluation dataset to use (via the datasets library). Defaults to the training dataset name if unspecified."
|
309 |
+
},
|
310 |
+
)
|
311 |
+
eval_dataset_config_name: Optional[str] = field(
|
312 |
+
default=None,
|
313 |
+
metadata={
|
314 |
+
"help": "The configuration name of the evaluation dataset to use (via the datasets library). Defaults to the training dataset config name if unspecified"
|
315 |
+
},
|
316 |
+
)
|
317 |
+
eval_split_name: str = field(
|
318 |
+
default="test",
|
319 |
+
metadata={
|
320 |
+
"help": "The name of the evaluation data set split to use (via the datasets library). Defaults to 'test'"
|
321 |
+
},
|
322 |
+
)
|
323 |
+
eval_metadata_dataset_name: str = field(
|
324 |
+
default=None,
|
325 |
+
metadata={
|
326 |
+
"help": "The name of the metadata training dataset to use (via the datasets library). Load and combine "
|
327 |
+
"multiple datasets by separating dataset ids by a '+' symbol. For example, to load and combine "
|
328 |
+
" librispeech and common voice, set `train_dataset_name='librispeech_asr+common_voice'`."
|
329 |
+
},
|
330 |
+
)
|
331 |
+
target_audio_column_name: str = field(
|
332 |
+
default="audio",
|
333 |
+
metadata={"help": "The name of the dataset column containing the target audio data. Defaults to 'audio'"},
|
334 |
+
)
|
335 |
+
description_column_name: str = field(
|
336 |
+
default=None,
|
337 |
+
metadata={"help": "The name of the dataset column containing the description text data. Defaults to 'None'."},
|
338 |
+
)
|
339 |
+
prompt_column_name: str = field(
|
340 |
+
default=None,
|
341 |
+
metadata={"help": "The name of the dataset column containing the prompt text data. Defaults to 'None'."},
|
342 |
+
)
|
343 |
+
overwrite_cache: bool = field(
|
344 |
+
default=False, metadata={"help": "Overwrite the cached preprocessed datasets or not."}
|
345 |
+
)
|
346 |
+
preprocessing_num_workers: Optional[int] = field(
|
347 |
+
default=None,
|
348 |
+
metadata={"help": "The number of processes to use for the preprocessing."},
|
349 |
+
)
|
350 |
+
max_train_samples: Optional[int] = field(
|
351 |
+
default=None,
|
352 |
+
metadata={
|
353 |
+
"help": (
|
354 |
+
"For debugging purposes or quicker training, truncate the number of training examples to this "
|
355 |
+
"value if set."
|
356 |
+
)
|
357 |
+
},
|
358 |
+
)
|
359 |
+
max_eval_samples: Optional[int] = field(
|
360 |
+
default=None,
|
361 |
+
metadata={
|
362 |
+
"help": (
|
363 |
+
"For debugging purposes or quicker training, truncate the number of validation examples to this "
|
364 |
+
"value if set."
|
365 |
+
)
|
366 |
+
},
|
367 |
+
)
|
368 |
+
max_duration_in_seconds: float = field(
|
369 |
+
default=35.0,
|
370 |
+
metadata={
|
371 |
+
"help": (
|
372 |
+
"Filter audio files that are longer than `max_duration_in_seconds` seconds to 'max_duration_in_seconds`."
|
373 |
+
"Also, used to set maximum audio length if `pad_to_max_length=True`."
|
374 |
+
)
|
375 |
+
},
|
376 |
+
)
|
377 |
+
min_duration_in_seconds: float = field(
|
378 |
+
default=0.0, metadata={"help": "Filter audio files that are shorter than `min_duration_in_seconds` seconds"}
|
379 |
+
)
|
380 |
+
max_text_length: int = field(
|
381 |
+
default=500, metadata={"help": "If set, max description lengths in number of characters."}
|
382 |
+
)
|
383 |
+
max_prompt_token_length: int = field(
|
384 |
+
default=None,
|
385 |
+
metadata={
|
386 |
+
"help": (
|
387 |
+
"If set, filter samples with prompts that are longer than `max_prompt_token_length` tokens."
|
388 |
+
"Also, used to set maximum prompt token length if `pad_to_max_length=True`."
|
389 |
+
)
|
390 |
+
},
|
391 |
+
)
|
392 |
+
max_description_token_length: int = field(
|
393 |
+
default=None,
|
394 |
+
metadata={
|
395 |
+
"help": (
|
396 |
+
"If set, filter samples with descriptions that are longer than `max_description_token_length` tokens."
|
397 |
+
"Also, used to set maximum desription token length if `pad_to_max_length=True`."
|
398 |
+
)
|
399 |
+
},
|
400 |
+
)
|
401 |
+
pad_to_max_length: bool = field(
|
402 |
+
default=False,
|
403 |
+
metadata={
|
404 |
+
"help": (
|
405 |
+
"If `True`, pad audio, prompt and description to a maximum length set with respectively "
|
406 |
+
"`max_duration_in_seconds`, `max_prompt_token_length`, `max_description_token_length`."
|
407 |
+
)
|
408 |
+
},
|
409 |
+
)
|
410 |
+
preprocessing_only: bool = field(
|
411 |
+
default=False,
|
412 |
+
metadata={
|
413 |
+
"help": (
|
414 |
+
"Whether to only do data preprocessing and skip training. This is especially useful when data"
|
415 |
+
" preprocessing errors out in distributed training due to timeout. In this case, one should run the"
|
416 |
+
" preprocessing in a non-distributed setup with `preprocessing_only=True` so that the cached datasets"
|
417 |
+
" can consequently be loaded in distributed training."
|
418 |
+
" In this training script, `save_to_disk` must be set to the path in which the dataset should be saved. "
|
419 |
+
)
|
420 |
+
},
|
421 |
+
)
|
422 |
+
token: str = field(
|
423 |
+
default=None,
|
424 |
+
metadata={
|
425 |
+
"help": (
|
426 |
+
"The token to use as HTTP bearer authorization for remote files. If not specified, will use the token "
|
427 |
+
"generated when running `huggingface-cli login` (stored in `~/.huggingface`)."
|
428 |
+
)
|
429 |
+
},
|
430 |
+
)
|
431 |
+
use_auth_token: bool = field(
|
432 |
+
default=None,
|
433 |
+
metadata={
|
434 |
+
"help": "The `use_auth_token` argument is deprecated and will be removed in v4.34. Please use `token` instead."
|
435 |
+
},
|
436 |
+
)
|
437 |
+
trust_remote_code: bool = field(
|
438 |
+
default=False,
|
439 |
+
metadata={
|
440 |
+
"help": (
|
441 |
+
"Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
|
442 |
+
"should only be set to `True` for repositories you trust and in which you have read the code, as it will "
|
443 |
+
"execute code present on the Hub on your local machine."
|
444 |
+
)
|
445 |
+
},
|
446 |
+
)
|
447 |
+
add_audio_samples_to_wandb: bool = field(
|
448 |
+
default=False,
|
449 |
+
metadata={"help": "If set and if `wandb` in args.report_to, will add generated audio samples to wandb logs."},
|
450 |
+
)
|
451 |
+
id_column_name: str = field(default=None, metadata={"help": "id column name."})
|
452 |
+
wandb_project: str = field(
|
453 |
+
default="parler-speech",
|
454 |
+
metadata={"help": "The name of the wandb project."},
|
455 |
+
)
|
456 |
+
save_to_disk: str = field(
|
457 |
+
default=None,
|
458 |
+
metadata={
|
459 |
+
"help": "If set, will save the dataset to this path if this is an empyt folder. If not empty, will load the datasets from it."
|
460 |
+
},
|
461 |
+
)
|
462 |
+
temporary_save_to_disk: str = field(default=None, metadata={"help": "Temporarily save audio labels here."})
|
463 |
+
pad_to_multiple_of: Optional[int] = field(
|
464 |
+
default=2,
|
465 |
+
metadata={"help": ("Pad to multiple of for tokenizers.")},
|
466 |
+
)
|
467 |
+
|
468 |
+
|
469 |
+
@dataclass
|
470 |
+
class ParlerTTSTrainingArguments(Seq2SeqTrainingArguments):
|
471 |
+
dtype: Optional[str] = field(
|
472 |
+
default="float32",
|
473 |
+
metadata={
|
474 |
+
"help": (
|
475 |
+
"The data type (dtype) in which to run training. One of `float32` (full-precision), "
|
476 |
+
"`float16` or `bfloat16` (both half-precision)."
|
477 |
+
)
|
478 |
+
},
|
479 |
+
)
|
480 |
+
audio_encoder_per_device_batch_size: int = field(
|
481 |
+
default=8,
|
482 |
+
metadata={"help": ("Specify the batch size of the audio encoding pre-processing steps.")},
|
483 |
+
)
|
484 |
+
|
485 |
+
|
486 |
+
@dataclass
|
487 |
+
class DataCollatorEncodecWithPadding:
|
488 |
+
"""
|
489 |
+
Data collator that will dynamically pad the inputs received to the longest sequence in the batch or
|
490 |
+
to `max_length` if `max_length` is set and `padding=max_length`.
|
491 |
+
"""
|
492 |
+
|
493 |
+
feature_extractor: AutoFeatureExtractor
|
494 |
+
audio_column_name: str
|
495 |
+
feature_extractor_input_name: Optional[str] = "input_values"
|
496 |
+
max_length: Optional[int] = None
|
497 |
+
padding: Optional[str] = "longest"
|
498 |
+
|
499 |
+
def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
|
500 |
+
# split inputs and labels since they have to be of different lengths and need
|
501 |
+
# different padding methods
|
502 |
+
audios = [feature[self.audio_column_name]["array"] for feature in features]
|
503 |
+
len_audio = [len(audio) for audio in audios]
|
504 |
+
|
505 |
+
batch = self.feature_extractor(audios, return_tensors="pt", padding=self.padding, max_length=self.max_length, sampling_rate=self.feature_extractor.sampling_rate)
|
506 |
+
batch["len_audio"] = torch.tensor(len_audio).unsqueeze(1)
|
507 |
+
return batch
|
508 |
+
|
509 |
+
|
510 |
+
@dataclass
|
511 |
+
class DataCollatorParlerTTSWithPadding:
|
512 |
+
"""
|
513 |
+
Data collator that will dynamically pad the inputs received.
|
514 |
+
Args:
|
515 |
+
prompt_tokenizer (:class:`~transformers.AutoTokenizer`)
|
516 |
+
The prompt_tokenizer used for proccessing the data.
|
517 |
+
description_tokenizer (:class:`~transformers.AutoTokenizer`)
|
518 |
+
The description_tokenizer used for proccessing the data.
|
519 |
+
padding (:obj:`bool`, :obj:`str` or :class:`~transformers.tokenization_utils_base.PaddingStrategy`, `optional`, defaults to :obj:`True`):
|
520 |
+
Select a strategy to pad the returned sequences (according to the model's padding side and padding index)
|
521 |
+
among:
|
522 |
+
* :obj:`True` or :obj:`'longest'`: Pad to the longest sequence in the batch (or no padding if only a single
|
523 |
+
sequence if provided).
|
524 |
+
* :obj:`'max_length'`: Pad to a maximum length specified with the argument :obj:`max_length` or to the
|
525 |
+
maximum acceptable input length for the model if that argument is not provided.
|
526 |
+
* :obj:`False` or :obj:`'do_not_pad'` (default): No padding (i.e., can output a batch with sequences of
|
527 |
+
different lengths).
|
528 |
+
pad_to_multiple_of (:obj:`int`, `optional`):
|
529 |
+
If set will pad the sequence to a multiple of the provided value.
|
530 |
+
This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability >=
|
531 |
+
7.5 (Volta).
|
532 |
+
"""
|
533 |
+
|
534 |
+
prompt_tokenizer: AutoTokenizer
|
535 |
+
description_tokenizer: AutoTokenizer
|
536 |
+
padding: Union[bool, str] = "longest"
|
537 |
+
pad_to_multiple_of: Optional[int] = None
|
538 |
+
prompt_max_length: Optional[int] = None
|
539 |
+
description_max_length: Optional[int] = None
|
540 |
+
audio_max_length: Optional[int] = None
|
541 |
+
|
542 |
+
def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
|
543 |
+
# split inputs and labels since they have to be of different lengths and need
|
544 |
+
# different padding methods
|
545 |
+
|
546 |
+
labels = [torch.tensor(feature["labels"]).transpose(0, 1) for feature in features]
|
547 |
+
# (bsz, seq_len, num_codebooks)
|
548 |
+
labels = torch.nn.utils.rnn.pad_sequence(labels, batch_first=True, padding_value=-100)
|
549 |
+
if self.audio_max_length is not None and self.padding == "max_length":
|
550 |
+
labels = torch.nn.functional.pad(labels, pad=(0, 0, 0, max(self.audio_max_length - labels.shape[1], 0)))
|
551 |
+
|
552 |
+
input_ids = [{"input_ids": feature["input_ids"]} for feature in features]
|
553 |
+
|
554 |
+
input_ids = self.description_tokenizer.pad(
|
555 |
+
input_ids,
|
556 |
+
return_tensors="pt",
|
557 |
+
padding=self.padding,
|
558 |
+
pad_to_multiple_of=self.pad_to_multiple_of,
|
559 |
+
max_length=self.description_max_length,
|
560 |
+
)
|
561 |
+
|
562 |
+
batch = {"labels": labels, **input_ids}
|
563 |
+
|
564 |
+
if self.audio_max_length is not None and self.padding == "max_length":
|
565 |
+
# if we do torch.compile, we need to also specify the attention_mask
|
566 |
+
decoder_attention_mask = torch.ones(labels.shape[:2], dtype=input_ids["attention_mask"].dtype)
|
567 |
+
batch["decoder_attention_mask"] = decoder_attention_mask
|
568 |
+
|
569 |
+
prompt_input_ids = [{"input_ids": feature["prompt_input_ids"]} for feature in features]
|
570 |
+
prompt_input_ids = self.prompt_tokenizer.pad(
|
571 |
+
prompt_input_ids,
|
572 |
+
return_tensors="pt",
|
573 |
+
padding=self.padding,
|
574 |
+
pad_to_multiple_of=self.pad_to_multiple_of,
|
575 |
+
max_length=self.prompt_max_length,
|
576 |
+
)
|
577 |
+
|
578 |
+
batch["prompt_input_ids"] = prompt_input_ids["input_ids"]
|
579 |
+
if "attention_mask" in prompt_input_ids:
|
580 |
+
batch["prompt_attention_mask"] = prompt_input_ids["attention_mask"]
|
581 |
+
|
582 |
+
return batch
|
583 |
+
|
584 |
+
|
585 |
+
def convert_dataset_str_to_list(
|
586 |
+
dataset_names,
|
587 |
+
dataset_config_names,
|
588 |
+
metadata_dataset_names=None,
|
589 |
+
splits=None,
|
590 |
+
dataset_samples=None,
|
591 |
+
default_split="train",
|
592 |
+
):
|
593 |
+
if isinstance(dataset_names, str):
|
594 |
+
dataset_names = dataset_names.split("+")
|
595 |
+
dataset_config_names = dataset_config_names.split("+")
|
596 |
+
splits = splits.split("+") if splits is not None else None
|
597 |
+
dataset_samples = dataset_samples.split("+") if dataset_samples is not None else None
|
598 |
+
metadata_dataset_names = metadata_dataset_names.split("+") if metadata_dataset_names is not None else None
|
599 |
+
|
600 |
+
# basic checks to ensure we've got the right number of datasets/configs/splits/columns/probs
|
601 |
+
if len(dataset_names) != len(dataset_config_names):
|
602 |
+
raise ValueError(
|
603 |
+
f"Ensure one config is passed for each dataset, got {len(dataset_names)} datasets and"
|
604 |
+
f" {len(dataset_config_names)} configs."
|
605 |
+
)
|
606 |
+
|
607 |
+
if splits is not None and len(splits) != len(dataset_names):
|
608 |
+
raise ValueError(
|
609 |
+
f"Ensure one split is passed for each dataset, got {len(dataset_names)} datasets and {len(splits)} splits."
|
610 |
+
)
|
611 |
+
|
612 |
+
if metadata_dataset_names is not None and len(metadata_dataset_names) != len(dataset_names):
|
613 |
+
raise ValueError(
|
614 |
+
f"Ensure one metadata dataset is passed for each dataset, got {len(dataset_names)} datasets and {len(metadata_dataset_names)} metadata datasets."
|
615 |
+
)
|
616 |
+
|
617 |
+
if dataset_samples is not None:
|
618 |
+
if len(dataset_samples) != len(dataset_names):
|
619 |
+
raise ValueError(
|
620 |
+
f"Ensure one sample is passed for each dataset, got {len(dataset_names)} datasets and "
|
621 |
+
f"{len(dataset_samples)} samples."
|
622 |
+
)
|
623 |
+
dataset_samples = [float(ds_sample) for ds_sample in dataset_samples]
|
624 |
+
else:
|
625 |
+
dataset_samples = [None] * len(dataset_names)
|
626 |
+
|
627 |
+
splits = splits if splits is not None else [default_split for _ in range(len(dataset_names))]
|
628 |
+
|
629 |
+
dataset_names_dict = []
|
630 |
+
for i, ds_name in enumerate(dataset_names):
|
631 |
+
dataset_names_dict.append(
|
632 |
+
{
|
633 |
+
"name": ds_name,
|
634 |
+
"config": dataset_config_names[i],
|
635 |
+
"split": splits[i],
|
636 |
+
"metadata_dataset_name": metadata_dataset_names[i],
|
637 |
+
"samples": dataset_samples[i],
|
638 |
+
}
|
639 |
+
)
|
640 |
+
return dataset_names_dict
|
641 |
+
|
642 |
+
|
643 |
+
def load_multiple_datasets(
|
644 |
+
accelerator: Accelerator,
|
645 |
+
dataset_names: Union[List, str],
|
646 |
+
dataset_config_names: Union[List, str],
|
647 |
+
metadata_dataset_names: Optional[str] = None,
|
648 |
+
splits: Optional[Union[List, str]] = None,
|
649 |
+
label_column_names: Optional[List] = None,
|
650 |
+
stopping_strategy: Optional[str] = "first_exhausted",
|
651 |
+
dataset_samples: Optional[Union[List, np.array]] = None,
|
652 |
+
streaming: Optional[bool] = False,
|
653 |
+
seed: Optional[int] = None,
|
654 |
+
id_column_name: Optional[str] = None,
|
655 |
+
columns_to_keep: Optional[Set[str]] = None,
|
656 |
+
prompt_column_name: Optional[str] = None,
|
657 |
+
sampling_rate: Optional[int] = None,
|
658 |
+
audio_column_name: Optional[str] = None,
|
659 |
+
**kwargs,
|
660 |
+
) -> Union[Dataset, IterableDataset]:
|
661 |
+
dataset_names_dict = convert_dataset_str_to_list(
|
662 |
+
dataset_names, dataset_config_names, metadata_dataset_names, splits, label_column_names, dataset_samples
|
663 |
+
)
|
664 |
+
|
665 |
+
if dataset_samples is not None:
|
666 |
+
dataset_samples = [ds_dict["samples"] for ds_dict in dataset_names_dict]
|
667 |
+
probabilities = np.array(dataset_samples) / np.sum(dataset_samples)
|
668 |
+
else:
|
669 |
+
probabilities = None
|
670 |
+
|
671 |
+
all_datasets = []
|
672 |
+
# iterate over the datasets we want to interleave
|
673 |
+
for dataset_dict in tqdm(dataset_names_dict, desc="Combining datasets..."):
|
674 |
+
with accelerator.main_process_first():
|
675 |
+
dataset = load_dataset(
|
676 |
+
dataset_dict["name"],
|
677 |
+
dataset_dict["config"],
|
678 |
+
split=dataset_dict["split"],
|
679 |
+
streaming=streaming,
|
680 |
+
**kwargs,
|
681 |
+
)
|
682 |
+
dataset_features = dataset.features.keys()
|
683 |
+
|
684 |
+
if sampling_rate is not None and audio_column_name is not None:
|
685 |
+
# resample target audio
|
686 |
+
dataset = dataset.cast_column(audio_column_name, datasets.features.Audio(sampling_rate=sampling_rate))
|
687 |
+
|
688 |
+
metadata_dataset_name = dataset_dict["metadata_dataset_name"]
|
689 |
+
if metadata_dataset_name is not None:
|
690 |
+
logger.info(
|
691 |
+
f'Merging {dataset_dict["name"]} - {dataset_dict["split"]} with {metadata_dataset_name} - {dataset_dict["split"]}'
|
692 |
+
)
|
693 |
+
metadata_dataset = load_dataset(
|
694 |
+
metadata_dataset_name,
|
695 |
+
dataset_dict["config"],
|
696 |
+
split=dataset_dict["split"],
|
697 |
+
streaming=streaming,
|
698 |
+
**kwargs,
|
699 |
+
)
|
700 |
+
|
701 |
+
# TODO(YL): I forgot to create unique ids for MLS english.
|
702 |
+
# To iterate faster, I bypass the original id check and do another one. - Done once because assuming it won't change next time
|
703 |
+
# if dataset_dict["name"] == "parler-tts/mls_eng_10k":
|
704 |
+
# def concat_ids(book_id, speaker_id, begin_time):
|
705 |
+
# return {"id": f"{book_id}_{speaker_id}_{str(begin_time).replace('.', '_')}"}
|
706 |
+
# dataset = dataset.map(concat_ids, input_columns=["book_id", "speaker_id", "begin_time"], num_proc=24)
|
707 |
+
# metadata_dataset = metadata_dataset.map(concat_ids, input_columns=["book_id", "speaker_id", "begin_time"], num_proc=24)
|
708 |
+
# metadata_dataset = metadata_dataset.rename_column(id_column_name, f"metadata_{id_column_name}")
|
709 |
+
|
710 |
+
if dataset_dict["name"] != "parler-tts/mls_eng_10k":
|
711 |
+
if id_column_name is not None and id_column_name not in dataset.column_names:
|
712 |
+
raise ValueError(
|
713 |
+
f"id_column_name={id_column_name} but has not been found in the dataset columns"
|
714 |
+
f"- one of {', '.join(list(dataset.column_names))}."
|
715 |
+
)
|
716 |
+
if id_column_name is not None and id_column_name not in metadata_dataset.column_names:
|
717 |
+
raise ValueError(
|
718 |
+
f"id_column_name={id_column_name} but has not been found in the metadata dataset columns"
|
719 |
+
f"- one of {', '.join(list(metadata_dataset.column_names))}."
|
720 |
+
)
|
721 |
+
elif id_column_name is not None:
|
722 |
+
metadata_dataset = metadata_dataset.rename_column(id_column_name, f"metadata_{id_column_name}")
|
723 |
+
|
724 |
+
metadata_columns_to_remove = set(metadata_dataset.column_names).intersection(set(dataset.column_names))
|
725 |
+
|
726 |
+
if prompt_column_name is not None:
|
727 |
+
# We might have applied some transformations to the prompts (e.g punctuation restoration)
|
728 |
+
# so we make sure to remove it from the original dataset
|
729 |
+
if prompt_column_name in dataset.column_names:
|
730 |
+
logger.info(
|
731 |
+
f"REMOVE {prompt_column_name} from dataset {dataset_dict['name']} - dataset_dict['split']"
|
732 |
+
)
|
733 |
+
dataset.remove_columns(prompt_column_name)
|
734 |
+
|
735 |
+
metadata_columns_to_remove = set(metadata_dataset.column_names).intersection(set(dataset.column_names))
|
736 |
+
metadata_dataset = metadata_dataset.remove_columns(metadata_columns_to_remove)
|
737 |
+
|
738 |
+
dataset = concatenate_datasets([dataset, metadata_dataset], axis=1)
|
739 |
+
|
740 |
+
if id_column_name is not None and dataset_dict["name"] != "parler-tts/mls_eng_10k":
|
741 |
+
if (
|
742 |
+
len(
|
743 |
+
dataset.filter(
|
744 |
+
lambda id1, id2: id1 != id2,
|
745 |
+
input_columns=[id_column_name, f"metadata_{id_column_name}"],
|
746 |
+
)
|
747 |
+
)
|
748 |
+
!= 0
|
749 |
+
):
|
750 |
+
raise ValueError(
|
751 |
+
f"Concatenate didn't work. Some ids don't correspond on dataset {dataset_dict['name']}"
|
752 |
+
)
|
753 |
+
|
754 |
+
dataset_features = dataset.features.keys()
|
755 |
+
|
756 |
+
if columns_to_keep is not None:
|
757 |
+
dataset = dataset.remove_columns(set(dataset_features - columns_to_keep))
|
758 |
+
all_datasets.append(dataset)
|
759 |
+
|
760 |
+
if len(all_datasets) == 1:
|
761 |
+
# we have a single dataset so just return it as is
|
762 |
+
return all_datasets[0]
|
763 |
+
|
764 |
+
if streaming:
|
765 |
+
interleaved_dataset = interleave_datasets(
|
766 |
+
all_datasets,
|
767 |
+
stopping_strategy=stopping_strategy,
|
768 |
+
probabilities=probabilities,
|
769 |
+
seed=seed,
|
770 |
+
)
|
771 |
+
else:
|
772 |
+
with accelerator.main_process_first():
|
773 |
+
interleaved_dataset = concatenate_datasets(all_datasets)
|
774 |
+
|
775 |
+
return interleaved_dataset
|
776 |
+
|
777 |
+
|
778 |
+
def main():
|
779 |
+
# See all possible arguments in src/transformers/training_args.py
|
780 |
+
# or by passing the --help flag to this script.
|
781 |
+
# We now keep distinct sets of args, for a cleaner separation of concerns.
|
782 |
+
|
783 |
+
parser = HfArgumentParser((ModelArguments, DataTrainingArguments, ParlerTTSTrainingArguments))
|
784 |
+
if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
|
785 |
+
# If we pass only one argument to the script and it's the path to a json file,
|
786 |
+
# let's parse it to get our arguments.
|
787 |
+
model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
|
788 |
+
else:
|
789 |
+
model_args, data_args, training_args = parser.parse_args_into_dataclasses()
|
790 |
+
|
791 |
+
# Sending telemetry. Tracking the example usage helps us better allocate resources to maintain them. The
|
792 |
+
# information sent is the one passed as arguments along with your Python/PyTorch versions.
|
793 |
+
send_example_telemetry("run_parler_tts", model_args, data_args)
|
794 |
+
|
795 |
+
if training_args.dtype == "float16":
|
796 |
+
mixed_precision = "fp16"
|
797 |
+
elif training_args.dtype == "bfloat16":
|
798 |
+
mixed_precision = "bf16"
|
799 |
+
else:
|
800 |
+
mixed_precision = "no"
|
801 |
+
|
802 |
+
if data_args.pad_to_max_length and (
|
803 |
+
data_args.max_duration_in_seconds is None
|
804 |
+
or data_args.max_prompt_token_length is None
|
805 |
+
or data_args.max_description_token_length is None
|
806 |
+
):
|
807 |
+
raise ValueError(
|
808 |
+
"`pad_to_max_length` is `True` but one of the following parameters has not been set: `max_duration_in_seconds`, `max_prompt_token_length`, `max_description_token_length`"
|
809 |
+
)
|
810 |
+
|
811 |
+
padding = "max_length" if data_args.pad_to_max_length else "longest"
|
812 |
+
|
813 |
+
####### A. Preparation
|
814 |
+
kwargs_handlers = [InitProcessGroupKwargs(timeout=timedelta(minutes=60))]
|
815 |
+
if training_args.torch_compile:
|
816 |
+
# TODO(YL): add more compile modes?
|
817 |
+
kwargs_handlers.append(TorchDynamoPlugin(backend="inductor", mode="default")) # reduce-overhead
|
818 |
+
|
819 |
+
accelerator = Accelerator(
|
820 |
+
gradient_accumulation_steps=training_args.gradient_accumulation_steps,
|
821 |
+
mixed_precision=mixed_precision,
|
822 |
+
log_with=training_args.report_to,
|
823 |
+
project_dir=training_args.output_dir,
|
824 |
+
kwargs_handlers=kwargs_handlers,
|
825 |
+
)
|
826 |
+
|
827 |
+
accelerator.init_trackers(
|
828 |
+
project_name=data_args.wandb_project,
|
829 |
+
config={
|
830 |
+
"learning_rate": training_args.learning_rate,
|
831 |
+
"model_name_or_path": model_args.model_name_or_path,
|
832 |
+
"num_train_epochs": training_args.num_train_epochs,
|
833 |
+
"gradient_accumulation_steps": training_args.gradient_accumulation_steps,
|
834 |
+
"per_device_train_batch_size": training_args.per_device_train_batch_size,
|
835 |
+
"global_batch_size": training_args.per_device_train_batch_size * accelerator.num_processes,
|
836 |
+
"mixed_precision": mixed_precision,
|
837 |
+
"lr_scheduler_type": training_args.lr_scheduler_type,
|
838 |
+
"warmup_steps": training_args.warmup_steps,
|
839 |
+
"freeze_text_encoder": model_args.freeze_text_encoder,
|
840 |
+
"max_duration_in_seconds": data_args.max_duration_in_seconds,
|
841 |
+
"weight_decay": training_args.weight_decay,
|
842 |
+
"adam_beta1": training_args.adam_beta1,
|
843 |
+
"adam_beta2": training_args.adam_beta2,
|
844 |
+
"temperature": model_args.temperature,
|
845 |
+
},
|
846 |
+
)
|
847 |
+
|
848 |
+
# Detecting last checkpoint and eventually continue from last checkpoint
|
849 |
+
last_checkpoint = None
|
850 |
+
if os.path.isdir(training_args.output_dir) and training_args.do_train and not training_args.overwrite_output_dir:
|
851 |
+
last_checkpoint = get_last_checkpoint(training_args.output_dir)
|
852 |
+
if last_checkpoint is None and len(os.listdir(training_args.output_dir)) > 0:
|
853 |
+
raise ValueError(
|
854 |
+
f"Output directory ({training_args.output_dir}) already exists and is not empty. "
|
855 |
+
"Use --overwrite_output_dir to overcome."
|
856 |
+
)
|
857 |
+
elif last_checkpoint is not None and training_args.resume_from_checkpoint is None:
|
858 |
+
logger.info(
|
859 |
+
f"Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change "
|
860 |
+
"the `--output_dir` or add `--overwrite_output_dir` to train from scratch."
|
861 |
+
)
|
862 |
+
|
863 |
+
# Setup logging
|
864 |
+
logging.basicConfig(
|
865 |
+
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
|
866 |
+
datefmt="%m/%d/%Y %H:%M:%S",
|
867 |
+
handlers=[logging.StreamHandler(sys.stdout)],
|
868 |
+
)
|
869 |
+
logger.setLevel(logging.INFO if accelerator.is_main_process else logging.WARN)
|
870 |
+
|
871 |
+
# Log a small summary on each proces
|
872 |
+
logger.warning(
|
873 |
+
f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}, "
|
874 |
+
f"distributed training: {training_args.parallel_mode.value == 'distributed'}, 16-bits training: {training_args.fp16}"
|
875 |
+
)
|
876 |
+
|
877 |
+
# Set the verbosity to info of the Transformers logger (on main process only)
|
878 |
+
if accelerator.is_local_main_process:
|
879 |
+
datasets.utils.logging.set_verbosity_warning()
|
880 |
+
transformers.utils.logging.set_verbosity_info()
|
881 |
+
else:
|
882 |
+
datasets.utils.logging.set_verbosity_error()
|
883 |
+
transformers.utils.logging.set_verbosity_error()
|
884 |
+
|
885 |
+
logger.info("Training/evaluation parameters %s", training_args)
|
886 |
+
|
887 |
+
# Set seed before initializing model.
|
888 |
+
set_seed(training_args.seed)
|
889 |
+
num_workers = data_args.preprocessing_num_workers
|
890 |
+
|
891 |
+
# 1. First, lett's instantiate the feature extractor, tokenizers and model
|
892 |
+
# Note for distributed training, the .from_pretrained methods guarantee that only
|
893 |
+
# one local process can concurrently download model & vocab.
|
894 |
+
|
895 |
+
# load feature extractor
|
896 |
+
feature_extractor = AutoFeatureExtractor.from_pretrained(
|
897 |
+
model_args.feature_extractor_name or model_args.model_name_or_path,
|
898 |
+
cache_dir=model_args.cache_dir,
|
899 |
+
token=data_args.token,
|
900 |
+
trust_remote_code=data_args.trust_remote_code,
|
901 |
+
)
|
902 |
+
sampling_rate = feature_extractor.sampling_rate
|
903 |
+
|
904 |
+
# load prompt tokenizer
|
905 |
+
prompt_tokenizer = AutoTokenizer.from_pretrained(
|
906 |
+
model_args.prompt_tokenizer_name or model_args.description_tokenizer_name or model_args.model_name_or_path,
|
907 |
+
cache_dir=model_args.cache_dir,
|
908 |
+
token=data_args.token,
|
909 |
+
trust_remote_code=data_args.trust_remote_code,
|
910 |
+
use_fast=model_args.use_fast_tokenizer,
|
911 |
+
padding_side="left", # prompt has to be padded on the left bc it's preprend to codebooks hidden states
|
912 |
+
)
|
913 |
+
|
914 |
+
# load description tokenizer
|
915 |
+
description_tokenizer = AutoTokenizer.from_pretrained(
|
916 |
+
model_args.description_tokenizer_name or model_args.model_name_or_path,
|
917 |
+
cache_dir=model_args.cache_dir,
|
918 |
+
token=data_args.token,
|
919 |
+
trust_remote_code=data_args.trust_remote_code,
|
920 |
+
use_fast=model_args.use_fast_tokenizer,
|
921 |
+
)
|
922 |
+
|
923 |
+
if model_args.use_fast_tokenizer:
|
924 |
+
logger.warning(
|
925 |
+
"Disabling fast tokenizer warning: https://github.com/huggingface/transformers/blob/main/src/transformers/tokenization_utils_base.py#L3231-L3235"
|
926 |
+
)
|
927 |
+
prompt_tokenizer.deprecation_warnings["Asking-to-pad-a-fast-tokenizer"] = True
|
928 |
+
description_tokenizer.deprecation_warnings["Asking-to-pad-a-fast-tokenizer"] = True
|
929 |
+
|
930 |
+
# 2. Now, let's load the dataset
|
931 |
+
|
932 |
+
if data_args.save_to_disk is not None:
|
933 |
+
os.makedirs(data_args.save_to_disk, exist_ok=True)
|
934 |
+
|
935 |
+
# assume that the dataset has been saved to `save_to_disk` if the latter is not empty
|
936 |
+
dataset_was_precomputed = len(os.listdir(data_args.save_to_disk)) > 0
|
937 |
+
if dataset_was_precomputed:
|
938 |
+
vectorized_datasets = datasets.load_from_disk(data_args.save_to_disk)
|
939 |
+
else:
|
940 |
+
raw_datasets = DatasetDict()
|
941 |
+
|
942 |
+
columns_to_keep = {
|
943 |
+
"target_audio_column_name": data_args.target_audio_column_name,
|
944 |
+
"prompt_column_name": data_args.prompt_column_name,
|
945 |
+
}
|
946 |
+
if data_args.description_column_name is not None:
|
947 |
+
columns_to_keep["description_column_name"] = data_args.description_column_name
|
948 |
+
|
949 |
+
if training_args.do_train:
|
950 |
+
raw_datasets["train"] = load_multiple_datasets(
|
951 |
+
accelerator,
|
952 |
+
data_args.train_dataset_name,
|
953 |
+
data_args.train_dataset_config_name,
|
954 |
+
metadata_dataset_names=data_args.train_metadata_dataset_name,
|
955 |
+
splits=data_args.train_split_name,
|
956 |
+
dataset_samples=data_args.train_dataset_samples,
|
957 |
+
seed=training_args.seed,
|
958 |
+
cache_dir=model_args.cache_dir,
|
959 |
+
num_proc=data_args.preprocessing_num_workers,
|
960 |
+
id_column_name=data_args.id_column_name,
|
961 |
+
columns_to_keep=columns_to_keep.values(),
|
962 |
+
prompt_column_name=data_args.prompt_column_name,
|
963 |
+
audio_column_name=data_args.target_audio_column_name,
|
964 |
+
sampling_rate=sampling_rate,
|
965 |
+
# streaming=data_args.streaming, TODO(SG): optionally enable streaming mode
|
966 |
+
)
|
967 |
+
|
968 |
+
for key in columns_to_keep:
|
969 |
+
if columns_to_keep[key] not in raw_datasets["train"].column_names:
|
970 |
+
raise ValueError(
|
971 |
+
f"--{key} '{columns_to_keep[key]}' not found in dataset '{data_args.train_dataset_name}'."
|
972 |
+
f" Make sure to set `--{key}` to the correct audio column - one of"
|
973 |
+
f" {', '.join(raw_datasets['train'].column_names)}."
|
974 |
+
)
|
975 |
+
|
976 |
+
if data_args.max_train_samples is not None:
|
977 |
+
raw_datasets["train"] = raw_datasets["train"].select(range(data_args.max_train_samples))
|
978 |
+
|
979 |
+
if training_args.do_eval:
|
980 |
+
raw_datasets["eval"] = load_multiple_datasets(
|
981 |
+
accelerator,
|
982 |
+
data_args.eval_dataset_name if data_args.eval_dataset_name else data_args.train_dataset_name,
|
983 |
+
data_args.eval_dataset_config_name
|
984 |
+
if data_args.eval_dataset_config_name
|
985 |
+
else data_args.train_dataset_config_name,
|
986 |
+
metadata_dataset_names=data_args.eval_metadata_dataset_name,
|
987 |
+
splits=data_args.eval_split_name,
|
988 |
+
cache_dir=model_args.cache_dir,
|
989 |
+
num_proc=data_args.preprocessing_num_workers,
|
990 |
+
id_column_name=data_args.id_column_name,
|
991 |
+
columns_to_keep=columns_to_keep.values(),
|
992 |
+
prompt_column_name=data_args.prompt_column_name,
|
993 |
+
audio_column_name=data_args.target_audio_column_name,
|
994 |
+
sampling_rate=sampling_rate,
|
995 |
+
# streaming=data_args.streaming, TODO(SG): optionally enable streaming mode
|
996 |
+
)
|
997 |
+
|
998 |
+
if data_args.max_eval_samples is not None:
|
999 |
+
raw_datasets["eval"] = (
|
1000 |
+
raw_datasets["eval"].shuffle(seed=training_args.seed).select(range(data_args.max_eval_samples))
|
1001 |
+
)
|
1002 |
+
|
1003 |
+
# 3. Next, let's load the config.
|
1004 |
+
config = ParlerTTSConfig.from_pretrained(
|
1005 |
+
model_args.model_name_or_path,
|
1006 |
+
cache_dir=model_args.cache_dir,
|
1007 |
+
token=data_args.token,
|
1008 |
+
trust_remote_code=data_args.trust_remote_code,
|
1009 |
+
)
|
1010 |
+
|
1011 |
+
# update pad token id and decoder_start_token_id
|
1012 |
+
config.update(
|
1013 |
+
{
|
1014 |
+
"pad_token_id": model_args.pad_token_id if model_args.pad_token_id is not None else config.pad_token_id,
|
1015 |
+
"decoder_start_token_id": (
|
1016 |
+
model_args.decoder_start_token_id
|
1017 |
+
if model_args.decoder_start_token_id is not None
|
1018 |
+
else config.decoder_start_token_id
|
1019 |
+
),
|
1020 |
+
}
|
1021 |
+
)
|
1022 |
+
|
1023 |
+
# create model
|
1024 |
+
model = ParlerTTSForConditionalGeneration.from_pretrained(
|
1025 |
+
model_args.model_name_or_path,
|
1026 |
+
cache_dir=model_args.cache_dir,
|
1027 |
+
config=config,
|
1028 |
+
token=data_args.token,
|
1029 |
+
trust_remote_code=data_args.trust_remote_code,
|
1030 |
+
)
|
1031 |
+
|
1032 |
+
# enable gradient checkpointing if necessary
|
1033 |
+
if training_args.gradient_checkpointing:
|
1034 |
+
model.gradient_checkpointing_enable()
|
1035 |
+
|
1036 |
+
# 4. Now we preprocess the datasets including loading the audio, resampling and normalization
|
1037 |
+
# Thankfully, `datasets` takes care of automatically loading and resampling the audio,
|
1038 |
+
# so that we just need to set the correct target sampling rate and normalize the input
|
1039 |
+
# via the `feature_extractor`
|
1040 |
+
|
1041 |
+
# derive max & min input length for sample rate & max duration
|
1042 |
+
sampling_rate = feature_extractor.sampling_rate
|
1043 |
+
max_target_length = data_args.max_duration_in_seconds * sampling_rate
|
1044 |
+
min_target_length = data_args.min_duration_in_seconds * sampling_rate
|
1045 |
+
target_audio_column_name = data_args.target_audio_column_name
|
1046 |
+
description_column_name = data_args.description_column_name
|
1047 |
+
prompt_column_name = data_args.prompt_column_name
|
1048 |
+
feature_extractor_input_name = feature_extractor.model_input_names[0]
|
1049 |
+
audio_encoder_pad_token_id = config.decoder.pad_token_id
|
1050 |
+
audio_encoder_eos_token_id = config.decoder.eos_token_id
|
1051 |
+
audio_encoder_bos_token_id = model.generation_config.decoder_start_token_id
|
1052 |
+
max_length = model.generation_config.max_length
|
1053 |
+
num_codebooks = model.decoder.config.num_codebooks
|
1054 |
+
bandwidth = model_args.bandwidth
|
1055 |
+
|
1056 |
+
# Freeze Encoders
|
1057 |
+
model.freeze_encoders(model_args.freeze_text_encoder)
|
1058 |
+
|
1059 |
+
# Test all gather - used for warmout and avoiding timeout
|
1060 |
+
test_tensor = torch.tensor([accelerator.process_index], device=accelerator.device)
|
1061 |
+
gathered_tensor = accelerator.gather(test_tensor)
|
1062 |
+
print("gathered_tensor", gathered_tensor)
|
1063 |
+
accelerator.wait_for_everyone()
|
1064 |
+
|
1065 |
+
if not dataset_was_precomputed:
|
1066 |
+
# Filter on text length
|
1067 |
+
if description_column_name is not None and data_args.max_text_length is not None:
|
1068 |
+
with accelerator.main_process_first():
|
1069 |
+
# filter description that is shorter than max_text_length
|
1070 |
+
raw_datasets = raw_datasets.filter(
|
1071 |
+
lambda x: len(x) < data_args.max_text_length,
|
1072 |
+
num_proc=num_workers,
|
1073 |
+
input_columns=[description_column_name],
|
1074 |
+
)
|
1075 |
+
|
1076 |
+
# Preprocessing the dataset.
|
1077 |
+
# We need to tokenize the texts.
|
1078 |
+
def pass_through_processors(description, prompt):
|
1079 |
+
batch = {}
|
1080 |
+
|
1081 |
+
batch["input_ids"] = description_tokenizer(description.strip())["input_ids"]
|
1082 |
+
batch["prompt_input_ids"] = prompt_tokenizer(prompt.strip())["input_ids"]
|
1083 |
+
|
1084 |
+
return batch
|
1085 |
+
|
1086 |
+
with accelerator.main_process_first():
|
1087 |
+
# this is a trick to avoid to rewrite the entire audio column which takes ages
|
1088 |
+
vectorized_datasets = raw_datasets.map(
|
1089 |
+
pass_through_processors,
|
1090 |
+
remove_columns=next(iter(raw_datasets.values())).column_names,
|
1091 |
+
input_columns=[description_column_name, prompt_column_name],
|
1092 |
+
num_proc=num_workers,
|
1093 |
+
desc="preprocess datasets",
|
1094 |
+
)
|
1095 |
+
|
1096 |
+
# We use Accelerate to perform distributed inference
|
1097 |
+
# T5 doesn't support fp16
|
1098 |
+
autocast_kwargs = AutocastKwargs(enabled=(mixed_precision != "fp16"))
|
1099 |
+
|
1100 |
+
# Now we encode the audio labels with encodec.
|
1101 |
+
####### B. Encode audio
|
1102 |
+
|
1103 |
+
logger.info("*** Encode target audio with encodec ***")
|
1104 |
+
|
1105 |
+
# no need to prepare audio_decoder because used for inference without mixed precision
|
1106 |
+
# see: https://huggingface.co/docs/accelerate/main/en/package_reference/accelerator#accelerate.Accelerator.prepare
|
1107 |
+
if training_args.torch_compile:
|
1108 |
+
audio_decoder = accelerator.prepare_model(model.audio_encoder, evaluation_mode=True)
|
1109 |
+
else:
|
1110 |
+
audio_decoder = model.audio_encoder
|
1111 |
+
|
1112 |
+
encoder_data_collator = DataCollatorEncodecWithPadding(
|
1113 |
+
feature_extractor,
|
1114 |
+
audio_column_name=target_audio_column_name,
|
1115 |
+
feature_extractor_input_name=feature_extractor_input_name,
|
1116 |
+
max_length=max_target_length,
|
1117 |
+
padding=padding,
|
1118 |
+
)
|
1119 |
+
|
1120 |
+
def apply_audio_decoder(batch):
|
1121 |
+
len_audio = batch.pop("len_audio")
|
1122 |
+
audio_decoder.to(batch["input_values"].device).eval()
|
1123 |
+
with torch.no_grad():
|
1124 |
+
labels = audio_decoder.encode(**batch, bandwidth=bandwidth)["audio_codes"]
|
1125 |
+
output = {}
|
1126 |
+
output["len_audio"] = len_audio
|
1127 |
+
# (1, bsz, codebooks, seq_len) -> (bsz, seq_len, codebooks)
|
1128 |
+
output["labels"] = labels.squeeze(0).transpose(1, 2)
|
1129 |
+
output["ratio"] = torch.ones_like(len_audio) * labels.shape[-1] / len_audio.max()
|
1130 |
+
return output
|
1131 |
+
|
1132 |
+
for split in vectorized_datasets:
|
1133 |
+
data_loader = DataLoader(
|
1134 |
+
raw_datasets[split],
|
1135 |
+
batch_size=training_args.audio_encoder_per_device_batch_size,
|
1136 |
+
collate_fn=encoder_data_collator,
|
1137 |
+
num_workers=training_args.dataloader_num_workers,
|
1138 |
+
pin_memory=True,
|
1139 |
+
)
|
1140 |
+
data_loader = accelerator.prepare(data_loader)
|
1141 |
+
|
1142 |
+
all_generated_labels = []
|
1143 |
+
all_lens = []
|
1144 |
+
for batch in tqdm(data_loader, disable=not accelerator.is_local_main_process):
|
1145 |
+
generate_labels = apply_audio_decoder(batch)
|
1146 |
+
generate_labels = accelerator.pad_across_processes(generate_labels, dim=1, pad_index=0)
|
1147 |
+
generate_labels = accelerator.gather_for_metrics(generate_labels)
|
1148 |
+
|
1149 |
+
if accelerator.is_main_process:
|
1150 |
+
lab = generate_labels["labels"].cpu().transpose(1, 2).to(torch.int16)
|
1151 |
+
rat = generate_labels["ratio"].cpu().squeeze()
|
1152 |
+
lens = generate_labels["len_audio"].cpu().squeeze()
|
1153 |
+
lab = [l[:, : int(ratio * length)] for (l, ratio, length) in zip(lab, rat, lens)]
|
1154 |
+
|
1155 |
+
all_generated_labels.extend(lab)
|
1156 |
+
all_lens.extend(lens)
|
1157 |
+
|
1158 |
+
# (1, codebooks, seq_len) where seq_len=1
|
1159 |
+
bos_labels = torch.ones((1, num_codebooks, 1)) * audio_encoder_bos_token_id
|
1160 |
+
|
1161 |
+
if accelerator.is_main_process:
|
1162 |
+
tmp_labels = Dataset.from_dict({"labels": all_generated_labels, "target_length": all_lens})
|
1163 |
+
tmp_labels.save_to_disk(
|
1164 |
+
os.path.join(data_args.temporary_save_to_disk, split),
|
1165 |
+
num_proc=1 if split == "eval" else data_args.preprocessing_num_workers,
|
1166 |
+
)
|
1167 |
+
accelerator.wait_for_everyone()
|
1168 |
+
del all_generated_labels
|
1169 |
+
|
1170 |
+
tmp_labels = datasets.load_from_disk(os.path.join(data_args.temporary_save_to_disk, split))
|
1171 |
+
with accelerator.main_process_first():
|
1172 |
+
vectorized_datasets[split] = concatenate_datasets([vectorized_datasets[split], tmp_labels], axis=1)
|
1173 |
+
|
1174 |
+
def postprocess_dataset(labels):
|
1175 |
+
# (1, codebooks, seq_len)
|
1176 |
+
labels = torch.tensor(labels).unsqueeze(0)
|
1177 |
+
# add bos
|
1178 |
+
labels = torch.cat([bos_labels, labels], dim=-1)
|
1179 |
+
|
1180 |
+
labels, delay_pattern_mask = build_delay_pattern_mask(
|
1181 |
+
labels,
|
1182 |
+
bos_token_id=audio_encoder_bos_token_id,
|
1183 |
+
pad_token_id=audio_encoder_eos_token_id,
|
1184 |
+
max_length=labels.shape[-1] + num_codebooks,
|
1185 |
+
num_codebooks=num_codebooks,
|
1186 |
+
)
|
1187 |
+
|
1188 |
+
# the first ids of the delay pattern mask are precisely labels, we use the rest of the labels mask
|
1189 |
+
# to take care of EOS
|
1190 |
+
# we want labels to look like this:
|
1191 |
+
# - [B, a, b, E, E, E, E]
|
1192 |
+
# - [B, B, c, d, E, E, E]
|
1193 |
+
# - [B, B, B, e, f, E, E]
|
1194 |
+
# - [B, B, B, B, g, h, E]
|
1195 |
+
labels = torch.where(delay_pattern_mask == -1, audio_encoder_eos_token_id, delay_pattern_mask)
|
1196 |
+
|
1197 |
+
# the first timestamp is associated to a row full of BOS, let's get rid of it
|
1198 |
+
# we also remove the last timestampts (full of PAD)
|
1199 |
+
output = {"labels": labels[:, 1:]}
|
1200 |
+
return output
|
1201 |
+
|
1202 |
+
with accelerator.main_process_first():
|
1203 |
+
vectorized_datasets[split] = vectorized_datasets[split].map(
|
1204 |
+
postprocess_dataset,
|
1205 |
+
num_proc=data_args.preprocessing_num_workers, # this one is resource consuming if many processor.
|
1206 |
+
input_columns=["labels"],
|
1207 |
+
desc="Postprocessing labeling",
|
1208 |
+
)
|
1209 |
+
|
1210 |
+
accelerator.free_memory()
|
1211 |
+
del generate_labels, all_lens
|
1212 |
+
|
1213 |
+
with accelerator.main_process_first():
|
1214 |
+
# NOTE: filtering is done at the end because in the `datasets` library, caching audio files is done after most operations
|
1215 |
+
# caching audio files is time and disk-space consuming, so we want to avoid it at all costs, especially for large (>1Kh) audio datasets.
|
1216 |
+
# That's also why we avoid to concat the processed datasets (vectorized_datasets) with the audio column present in raw_datasets.
|
1217 |
+
|
1218 |
+
def is_audio_in_length_range(length):
|
1219 |
+
return length > min_target_length and length < max_target_length
|
1220 |
+
|
1221 |
+
# filter data that is shorter than min_target_length
|
1222 |
+
vectorized_datasets = vectorized_datasets.filter(
|
1223 |
+
is_audio_in_length_range,
|
1224 |
+
num_proc=num_workers,
|
1225 |
+
input_columns=["target_length"],
|
1226 |
+
)
|
1227 |
+
|
1228 |
+
if description_column_name is not None and data_args.max_description_token_length is not None:
|
1229 |
+
with accelerator.main_process_first():
|
1230 |
+
# filter description that is shorter than max_text_length
|
1231 |
+
vectorized_datasets = vectorized_datasets.filter(
|
1232 |
+
lambda x: len(x) < data_args.max_description_token_length,
|
1233 |
+
num_proc=num_workers,
|
1234 |
+
input_columns=["input_ids"],
|
1235 |
+
)
|
1236 |
+
|
1237 |
+
if data_args.max_prompt_token_length is not None:
|
1238 |
+
with accelerator.main_process_first():
|
1239 |
+
# filter description that is shorter than max_text_length
|
1240 |
+
vectorized_datasets = vectorized_datasets.filter(
|
1241 |
+
lambda x: len(x) < data_args.max_prompt_token_length,
|
1242 |
+
num_proc=num_workers,
|
1243 |
+
input_columns=["prompt_input_ids"],
|
1244 |
+
)
|
1245 |
+
|
1246 |
+
if data_args.save_to_disk is not None and not dataset_was_precomputed:
|
1247 |
+
if accelerator.is_main_process:
|
1248 |
+
vectorized_datasets.save_to_disk(
|
1249 |
+
data_args.save_to_disk,
|
1250 |
+
num_proc=min(data_args.preprocessing_num_workers, len(vectorized_datasets["eval"]) - 1),
|
1251 |
+
)
|
1252 |
+
logger.info(f"Dataset saved at {data_args.save_to_disk}")
|
1253 |
+
|
1254 |
+
audio_max_length = None
|
1255 |
+
if training_args.torch_compile:
|
1256 |
+
audio_max_length = max(vectorized_datasets["train"]["target_length"])
|
1257 |
+
with accelerator.main_process_first():
|
1258 |
+
max_sample = vectorized_datasets["train"].filter(
|
1259 |
+
lambda x: x == audio_max_length,
|
1260 |
+
num_proc=num_workers,
|
1261 |
+
input_columns=["target_length"],
|
1262 |
+
)
|
1263 |
+
audio_max_length = torch.tensor(max_sample[0]["labels"]).shape[1]
|
1264 |
+
|
1265 |
+
# for large datasets it is advised to run the preprocessing on a
|
1266 |
+
# single machine first with ``args.preprocessing_only`` since there will mostly likely
|
1267 |
+
# be a timeout when running the script in distributed mode.
|
1268 |
+
# In a second step ``args.preprocessing_only`` can then be set to `False` to load the
|
1269 |
+
# cached dataset
|
1270 |
+
if data_args.preprocessing_only and data_args.save_to_disk is None:
|
1271 |
+
raise ValueError(
|
1272 |
+
"`preprocessing_only=True` but `save_to_disk` is not set. The latter should indicates where to save the dataset locally."
|
1273 |
+
)
|
1274 |
+
elif data_args.preprocessing_only:
|
1275 |
+
logger.info(f"Data preprocessing finished. Files save at {data_args.save_to_disk}")
|
1276 |
+
return
|
1277 |
+
|
1278 |
+
# 6. Next, we can prepare the training.
|
1279 |
+
|
1280 |
+
# Let's use word CLAP similary and WER metrics as our evaluation metrics,
|
1281 |
+
|
1282 |
+
# Define evaluation metrics during training, *i.e.* CLAP similarity
|
1283 |
+
clap = AutoModel.from_pretrained(model_args.clap_model_name_or_path)
|
1284 |
+
clap_processor = AutoProcessor.from_pretrained(model_args.clap_model_name_or_path)
|
1285 |
+
metric = evaluate.load("wer")
|
1286 |
+
|
1287 |
+
def clap_similarity(texts, audios, device):
|
1288 |
+
clap_inputs = clap_processor(text=texts, audios=audios, padding=True, return_tensors="pt").to(device)
|
1289 |
+
clap.to(device)
|
1290 |
+
with torch.no_grad():
|
1291 |
+
text_features = clap.get_text_features(
|
1292 |
+
clap_inputs["input_ids"], attention_mask=clap_inputs.get("attention_mask", None)
|
1293 |
+
)
|
1294 |
+
audio_features = clap.get_audio_features(clap_inputs["input_features"])
|
1295 |
+
|
1296 |
+
cosine_sim = torch.nn.functional.cosine_similarity(audio_features, text_features, dim=1, eps=1e-8)
|
1297 |
+
|
1298 |
+
clap.to("cpu")
|
1299 |
+
clap_inputs.to("cpu")
|
1300 |
+
return cosine_sim.mean().to("cpu")
|
1301 |
+
|
1302 |
+
def wer(prompts, audios, device):
|
1303 |
+
asr_pipeline = pipeline(model=model_args.asr_model_name_or_path, device=device)
|
1304 |
+
transcriptions = asr_pipeline(
|
1305 |
+
[{"raw": audio, "sampling_rate": sampling_rate} for audio in audios],
|
1306 |
+
batch_size=int(training_args.per_device_eval_batch_size),
|
1307 |
+
)
|
1308 |
+
|
1309 |
+
word_error = 100 * metric.compute(
|
1310 |
+
predictions=[t["text"].lower() for t in transcriptions], references=[t.lower() for t in prompts]
|
1311 |
+
)
|
1312 |
+
|
1313 |
+
return word_error, [t["text"] for t in transcriptions]
|
1314 |
+
|
1315 |
+
eval_methods = {"clap": clap_similarity, "wer": wer}
|
1316 |
+
|
1317 |
+
def compute_metrics(audios, descriptions, prompts, device="cpu"):
|
1318 |
+
input_ids = descriptions
|
1319 |
+
texts = description_tokenizer.batch_decode(input_ids, skip_special_tokens=True)
|
1320 |
+
prompts = prompt_tokenizer.batch_decode(prompts, skip_special_tokens=True)
|
1321 |
+
audios = [a.cpu().numpy() for a in audios]
|
1322 |
+
results = {"clap": eval_methods["clap"](texts, audios, device)}
|
1323 |
+
word_error, transcriptions = eval_methods["wer"](prompts, audios, device)
|
1324 |
+
results["wer"] = word_error
|
1325 |
+
|
1326 |
+
return results, texts, prompts, audios, transcriptions
|
1327 |
+
|
1328 |
+
# Define Training Schedule
|
1329 |
+
# Store some constants
|
1330 |
+
per_device_train_batch_size = int(training_args.per_device_train_batch_size)
|
1331 |
+
train_batch_size = per_device_train_batch_size * accelerator.num_processes
|
1332 |
+
gradient_accumulation_steps = int(training_args.gradient_accumulation_steps)
|
1333 |
+
per_device_eval_batch_size = int(training_args.per_device_eval_batch_size)
|
1334 |
+
|
1335 |
+
if training_args.max_steps < 0:
|
1336 |
+
num_epochs = int(training_args.num_train_epochs)
|
1337 |
+
steps_per_epoch = len(vectorized_datasets["train"]) // (train_batch_size * gradient_accumulation_steps)
|
1338 |
+
total_train_steps = steps_per_epoch * num_epochs
|
1339 |
+
elif training_args.max_steps > 0:
|
1340 |
+
logger.info("max_steps is given, it will override any value given in num_train_epochs")
|
1341 |
+
total_train_steps = int(training_args.max_steps)
|
1342 |
+
# Setting a very large number of epochs so we go as many times as necessary over the iterator.
|
1343 |
+
num_epochs = sys.maxsize
|
1344 |
+
steps_per_epoch = total_train_steps
|
1345 |
+
|
1346 |
+
if training_args.evaluation_strategy == "epoch":
|
1347 |
+
eval_steps = steps_per_epoch
|
1348 |
+
elif training_args.eval_steps is None:
|
1349 |
+
logger.info(f"eval_steps is not set, evaluating at the end of each epoch")
|
1350 |
+
eval_steps = steps_per_epoch
|
1351 |
+
else:
|
1352 |
+
eval_steps = training_args.eval_steps
|
1353 |
+
|
1354 |
+
if training_args.save_strategy == "epoch":
|
1355 |
+
save_steps = steps_per_epoch
|
1356 |
+
elif training_args.save_strategy == "steps":
|
1357 |
+
save_steps = training_args.save_steps
|
1358 |
+
else:
|
1359 |
+
save_steps = sys.maxsize
|
1360 |
+
|
1361 |
+
# T5 doesn't support fp16
|
1362 |
+
autocast_kwargs = AutocastKwargs(enabled=(mixed_precision != "fp16"))
|
1363 |
+
|
1364 |
+
# Define optimizer, LR scheduler, collator
|
1365 |
+
optimizer = torch.optim.AdamW(
|
1366 |
+
params=model.parameters(),
|
1367 |
+
lr=training_args.learning_rate,
|
1368 |
+
betas=(training_args.adam_beta1, training_args.adam_beta2),
|
1369 |
+
eps=training_args.adam_epsilon,
|
1370 |
+
weight_decay=training_args.weight_decay,
|
1371 |
+
)
|
1372 |
+
|
1373 |
+
# LR scheduler gets stepped by `num_processes` each time -> account for this in warmup / total steps
|
1374 |
+
lr_scheduler = get_scheduler(
|
1375 |
+
name=training_args.lr_scheduler_type,
|
1376 |
+
optimizer=optimizer,
|
1377 |
+
num_warmup_steps=training_args.get_warmup_steps(total_train_steps) * accelerator.num_processes,
|
1378 |
+
num_training_steps=total_train_steps * accelerator.num_processes,
|
1379 |
+
)
|
1380 |
+
|
1381 |
+
# Instantiate custom data collator
|
1382 |
+
data_collator = DataCollatorParlerTTSWithPadding(
|
1383 |
+
prompt_tokenizer=prompt_tokenizer,
|
1384 |
+
description_tokenizer=description_tokenizer,
|
1385 |
+
pad_to_multiple_of=data_args.pad_to_multiple_of,
|
1386 |
+
padding=padding,
|
1387 |
+
prompt_max_length=data_args.max_prompt_token_length,
|
1388 |
+
description_max_length=data_args.max_description_token_length,
|
1389 |
+
audio_max_length=audio_max_length,
|
1390 |
+
)
|
1391 |
+
|
1392 |
+
# Prepare everything with accelerate
|
1393 |
+
model, optimizer, lr_scheduler = accelerator.prepare(model, optimizer, lr_scheduler)
|
1394 |
+
|
1395 |
+
logger.info("***** Running training *****")
|
1396 |
+
logger.info(f" Num examples = {total_train_steps * train_batch_size * gradient_accumulation_steps}")
|
1397 |
+
logger.info(" Instantaneous batch size per device =" f" {per_device_train_batch_size}")
|
1398 |
+
logger.info(" Gradient accumulation steps =" f" {gradient_accumulation_steps}")
|
1399 |
+
logger.info(
|
1400 |
+
f" Total train batch size (w. parallel & distributed) = {train_batch_size * gradient_accumulation_steps}"
|
1401 |
+
)
|
1402 |
+
logger.info(f" Total optimization steps = {total_train_steps}")
|
1403 |
+
|
1404 |
+
# ======================== Training ================================
|
1405 |
+
train_time = 0
|
1406 |
+
train_start = time.time()
|
1407 |
+
steps_trained_progress_bar = tqdm(
|
1408 |
+
range(total_train_steps), desc="Train steps ... ", position=0, disable=not accelerator.is_local_main_process
|
1409 |
+
)
|
1410 |
+
continue_training = True
|
1411 |
+
epochs_trained = 0
|
1412 |
+
cur_step = 0
|
1413 |
+
|
1414 |
+
checkpoint = None
|
1415 |
+
if training_args.resume_from_checkpoint is not None:
|
1416 |
+
checkpoint = training_args.resume_from_checkpoint
|
1417 |
+
elif last_checkpoint is not None:
|
1418 |
+
checkpoint = last_checkpoint
|
1419 |
+
|
1420 |
+
if accelerator.is_main_process:
|
1421 |
+
if training_args.push_to_hub:
|
1422 |
+
# Retrieve of infer repo_name
|
1423 |
+
repo_name = training_args.hub_model_id
|
1424 |
+
if repo_name is None:
|
1425 |
+
repo_name = Path(training_args.output_dir).absolute().name
|
1426 |
+
# Create repo and retrieve repo_id
|
1427 |
+
repo_id = create_repo(repo_name, exist_ok=True, token=training_args.hub_token).repo_id
|
1428 |
+
# Clone repo locally
|
1429 |
+
repo = Repository(training_args.output_dir, clone_from=repo_id, token=training_args.hub_token)
|
1430 |
+
|
1431 |
+
with open(os.path.join(training_args.output_dir, ".gitignore"), "w+") as gitignore:
|
1432 |
+
if "wandb" not in gitignore:
|
1433 |
+
gitignore.write("wandb\n")
|
1434 |
+
elif training_args.output_dir is not None:
|
1435 |
+
os.makedirs(training_args.output_dir, exist_ok=True)
|
1436 |
+
accelerator.wait_for_everyone()
|
1437 |
+
|
1438 |
+
# Now save everything to be able to create a single processor later
|
1439 |
+
# make sure all processes wait until data is saved
|
1440 |
+
with accelerator.main_process_first():
|
1441 |
+
# only the main process saves them
|
1442 |
+
if accelerator.is_main_process:
|
1443 |
+
# save feature extractor, tokenizer and config
|
1444 |
+
if (
|
1445 |
+
model_args.prompt_tokenizer_name is None
|
1446 |
+
and model_args.description_tokenizer_name
|
1447 |
+
or (model_args.prompt_tokenizer_name == model_args.description_tokenizer_name)
|
1448 |
+
):
|
1449 |
+
prompt_tokenizer.save_pretrained(training_args.output_dir)
|
1450 |
+
else:
|
1451 |
+
logger.warning(
|
1452 |
+
"Prompt tokenizer ('{model_args.prompt_tokenizer_name}') and description tokenizer ('{model_args.description_tokenizer_name}') are not the same. Saving only the prompt tokenizer."
|
1453 |
+
)
|
1454 |
+
prompt_tokenizer.save_pretrained(training_args.output_dir)
|
1455 |
+
|
1456 |
+
feature_extractor.save_pretrained(training_args.output_dir)
|
1457 |
+
config.save_pretrained(training_args.output_dir)
|
1458 |
+
|
1459 |
+
if checkpoint is not None:
|
1460 |
+
accelerator.load_state(checkpoint)
|
1461 |
+
# Find num steps and epoch from saved state string pattern
|
1462 |
+
pattern = r"checkpoint-(\d+)-epoch-(\d+)"
|
1463 |
+
match = re.search(pattern, checkpoint)
|
1464 |
+
cur_step = int(match.group(1))
|
1465 |
+
epochs_trained = int(match.group(2))
|
1466 |
+
|
1467 |
+
logger.info(" Continuing training from checkpoint, will skip to saved global_step")
|
1468 |
+
logger.info(f" Continuing training from epoch {epochs_trained}")
|
1469 |
+
logger.info(f" Continuing training from global step {cur_step}")
|
1470 |
+
|
1471 |
+
steps_trained_progress_bar.update(cur_step)
|
1472 |
+
|
1473 |
+
for epoch in range(0, epochs_trained):
|
1474 |
+
vectorized_datasets["train"] = vectorized_datasets["train"].shuffle(training_args.seed)
|
1475 |
+
|
1476 |
+
if training_args.max_steps < 0:
|
1477 |
+
# we know exactly the number of steps per epoch, so can skip through the required number of batches
|
1478 |
+
resume_step = (cur_step - epochs_trained * steps_per_epoch) * gradient_accumulation_steps
|
1479 |
+
else:
|
1480 |
+
# Currently we don't know how many steps we've taken in the current epoch
|
1481 |
+
# So we just shuffle the dataset one extra time and start from a fresh epoch
|
1482 |
+
# This is "good enough" for our purposes but not fully correct
|
1483 |
+
resume_step = None
|
1484 |
+
vectorized_datasets["train"] = vectorized_datasets["train"].shuffle(training_args.seed)
|
1485 |
+
else:
|
1486 |
+
resume_step = None
|
1487 |
+
|
1488 |
+
gen_kwargs = {
|
1489 |
+
"do_sample": model_args.do_sample,
|
1490 |
+
"temperature": model_args.temperature,
|
1491 |
+
"max_length": model_args.max_length,
|
1492 |
+
}
|
1493 |
+
|
1494 |
+
# Define gradient update step fn
|
1495 |
+
def train_step(
|
1496 |
+
batch,
|
1497 |
+
accelerator,
|
1498 |
+
autocast_kwargs,
|
1499 |
+
):
|
1500 |
+
model.train()
|
1501 |
+
|
1502 |
+
if mixed_precision == "fp16":
|
1503 |
+
# fp16 doesn't work with T5-like models
|
1504 |
+
with accelerator.autocast(autocast_handler=autocast_kwargs):
|
1505 |
+
if training_args.parallel_mode.value != "distributed":
|
1506 |
+
encoder_outputs = model.text_encoder(
|
1507 |
+
input_ids=batch.get("input_ids"), attention_mask=batch.get("attention_mask", None)
|
1508 |
+
)
|
1509 |
+
else:
|
1510 |
+
encoder_outputs = model.module.text_encoder(
|
1511 |
+
input_ids=batch.get("input_ids"), attention_mask=batch.get("attention_mask", None)
|
1512 |
+
)
|
1513 |
+
batch["encoder_outputs"] = encoder_outputs
|
1514 |
+
|
1515 |
+
outputs = model(**batch)
|
1516 |
+
# CE (data) loss
|
1517 |
+
ce_loss = outputs.loss
|
1518 |
+
|
1519 |
+
metrics = {"loss": ce_loss}
|
1520 |
+
return ce_loss, metrics
|
1521 |
+
|
1522 |
+
# Define eval fn
|
1523 |
+
def eval_step(
|
1524 |
+
batch,
|
1525 |
+
accelerator,
|
1526 |
+
autocast_kwargs,
|
1527 |
+
):
|
1528 |
+
eval_model = model if not training_args.torch_compile else model._orig_mod
|
1529 |
+
eval_model.eval()
|
1530 |
+
|
1531 |
+
if mixed_precision == "fp16":
|
1532 |
+
# fp16 doesn't work with T5-like models
|
1533 |
+
with accelerator.autocast(autocast_handler=autocast_kwargs):
|
1534 |
+
with torch.no_grad():
|
1535 |
+
if training_args.parallel_mode.value != "distributed" or training_args.torch_compile:
|
1536 |
+
encoder_outputs = eval_model.text_encoder(
|
1537 |
+
input_ids=batch.get("input_ids"), attention_mask=batch.get("attention_mask", None)
|
1538 |
+
)
|
1539 |
+
else:
|
1540 |
+
encoder_outputs = eval_model.module.text_encoder(
|
1541 |
+
input_ids=batch.get("input_ids"), attention_mask=batch.get("attention_mask", None)
|
1542 |
+
)
|
1543 |
+
batch["encoder_outputs"] = encoder_outputs
|
1544 |
+
|
1545 |
+
with torch.no_grad():
|
1546 |
+
outputs = eval_model(**batch)
|
1547 |
+
# CE (data) loss
|
1548 |
+
ce_loss = outputs.loss
|
1549 |
+
metrics = {"loss": ce_loss}
|
1550 |
+
return metrics
|
1551 |
+
|
1552 |
+
def generate_step(batch):
|
1553 |
+
batch.pop("decoder_attention_mask", None)
|
1554 |
+
eval_model = accelerator.unwrap_model(model, keep_fp32_wrapper=mixed_precision != "fp16").eval()
|
1555 |
+
if training_args.torch_compile:
|
1556 |
+
eval_model = model._orig_mod
|
1557 |
+
|
1558 |
+
output_audios = eval_model.generate(**batch, **gen_kwargs)
|
1559 |
+
output_audios = accelerator.pad_across_processes(output_audios, dim=1, pad_index=0)
|
1560 |
+
return output_audios
|
1561 |
+
|
1562 |
+
for epoch in range(epochs_trained, num_epochs):
|
1563 |
+
vectorized_datasets["train"] = vectorized_datasets["train"].shuffle(training_args.seed)
|
1564 |
+
sampler = None
|
1565 |
+
if training_args.group_by_length:
|
1566 |
+
sampler = LengthGroupedSampler(train_batch_size, lengths=vectorized_datasets["train"]["target_length"])
|
1567 |
+
train_dataloader = DataLoader(
|
1568 |
+
vectorized_datasets["train"],
|
1569 |
+
collate_fn=data_collator,
|
1570 |
+
batch_size=per_device_train_batch_size,
|
1571 |
+
sampler=sampler,
|
1572 |
+
num_workers=training_args.dataloader_num_workers,
|
1573 |
+
pin_memory=training_args.dataloader_pin_memory,
|
1574 |
+
)
|
1575 |
+
train_dataloader = accelerator.prepare(train_dataloader)
|
1576 |
+
if hasattr(train_dataloader, "dataset") and isinstance(train_dataloader.dataset, IterableDataset):
|
1577 |
+
train_dataloader.dataset.set_epoch(epoch)
|
1578 |
+
|
1579 |
+
if resume_step is not None:
|
1580 |
+
# Skip the first N batches in the dataloader when resuming from a checkpoint
|
1581 |
+
train_dataloader = accelerator.skip_first_batches(train_dataloader, resume_step)
|
1582 |
+
resume_step = None
|
1583 |
+
|
1584 |
+
for batch in train_dataloader:
|
1585 |
+
with accelerator.accumulate(model):
|
1586 |
+
loss, train_metric = train_step(batch, accelerator, autocast_kwargs)
|
1587 |
+
accelerator.backward(loss)
|
1588 |
+
if accelerator.sync_gradients:
|
1589 |
+
accelerator.clip_grad_norm_(model.parameters(), training_args.max_grad_norm)
|
1590 |
+
optimizer.step()
|
1591 |
+
lr_scheduler.step()
|
1592 |
+
optimizer.zero_grad()
|
1593 |
+
|
1594 |
+
# Check if the accelerator has performed an optimization step behind the scenes
|
1595 |
+
if accelerator.sync_gradients:
|
1596 |
+
steps_trained_progress_bar.update(1)
|
1597 |
+
cur_step += 1
|
1598 |
+
|
1599 |
+
if cur_step % training_args.logging_steps == 0:
|
1600 |
+
steps_trained_progress_bar.write(
|
1601 |
+
f"Step... ({cur_step} / {total_train_steps} | Loss:"
|
1602 |
+
f" {train_metric['loss']}, Learning Rate:"
|
1603 |
+
f" {lr_scheduler.get_last_lr()[0]})"
|
1604 |
+
)
|
1605 |
+
log_metric(
|
1606 |
+
accelerator,
|
1607 |
+
metrics=train_metric,
|
1608 |
+
learning_rate=lr_scheduler.get_last_lr()[0],
|
1609 |
+
train_time=train_time + time.time() - train_start,
|
1610 |
+
step=cur_step,
|
1611 |
+
epoch=epoch + (cur_step - epoch * steps_per_epoch) / steps_per_epoch,
|
1612 |
+
prefix="train",
|
1613 |
+
)
|
1614 |
+
|
1615 |
+
# save checkpoint and weights after each save_steps and at the end of training
|
1616 |
+
if (cur_step % save_steps == 0) or cur_step == total_train_steps:
|
1617 |
+
intermediate_dir = os.path.join(training_args.output_dir, f"checkpoint-{cur_step}-epoch-{epoch}")
|
1618 |
+
# safe_serialization=False to avoid shared tensors saving issue (TODO(YL): it's a temporary fix)
|
1619 |
+
# https://github.com/huggingface/transformers/issues/27293#issuecomment-1872560074
|
1620 |
+
accelerator.save_state(output_dir=intermediate_dir, safe_serialization=False)
|
1621 |
+
accelerator.wait_for_everyone()
|
1622 |
+
if accelerator.is_main_process:
|
1623 |
+
rotate_checkpoints(training_args.save_total_limit, output_dir=training_args.output_dir)
|
1624 |
+
|
1625 |
+
if cur_step == total_train_steps:
|
1626 |
+
# un-wrap student model for save
|
1627 |
+
unwrapped_model = accelerator.unwrap_model(model)
|
1628 |
+
unwrapped_model.save_pretrained(training_args.output_dir)
|
1629 |
+
|
1630 |
+
if training_args.push_to_hub:
|
1631 |
+
repo.push_to_hub(
|
1632 |
+
commit_message=f"Saving train state of step {cur_step}",
|
1633 |
+
blocking=False,
|
1634 |
+
)
|
1635 |
+
|
1636 |
+
if training_args.do_eval and (cur_step % eval_steps == 0 or cur_step == total_train_steps):
|
1637 |
+
train_time += time.time() - train_start
|
1638 |
+
# ======================== Evaluating ==============================
|
1639 |
+
eval_metrics = []
|
1640 |
+
eval_preds = []
|
1641 |
+
eval_descriptions = []
|
1642 |
+
eval_prompts = []
|
1643 |
+
eval_start = time.time()
|
1644 |
+
|
1645 |
+
# release training input batch
|
1646 |
+
batch = release_memory(batch)
|
1647 |
+
|
1648 |
+
validation_dataloader = DataLoader(
|
1649 |
+
vectorized_datasets["eval"],
|
1650 |
+
collate_fn=data_collator,
|
1651 |
+
batch_size=per_device_eval_batch_size,
|
1652 |
+
drop_last=False,
|
1653 |
+
num_workers=training_args.dataloader_pin_memory,
|
1654 |
+
pin_memory=training_args.dataloader_pin_memory,
|
1655 |
+
)
|
1656 |
+
validation_dataloader = accelerator.prepare(validation_dataloader)
|
1657 |
+
|
1658 |
+
for batch in tqdm(
|
1659 |
+
validation_dataloader,
|
1660 |
+
desc="Evaluating - Inference ...",
|
1661 |
+
position=2,
|
1662 |
+
disable=not accelerator.is_local_main_process,
|
1663 |
+
):
|
1664 |
+
# Model forward
|
1665 |
+
eval_metric = eval_step(batch, accelerator, autocast_kwargs)
|
1666 |
+
eval_metric = accelerator.gather_for_metrics(eval_metric)
|
1667 |
+
eval_metrics.append(eval_metric)
|
1668 |
+
|
1669 |
+
if training_args.predict_with_generate:
|
1670 |
+
validation_dataloader = DataLoader(
|
1671 |
+
vectorized_datasets["eval"],
|
1672 |
+
collate_fn=data_collator,
|
1673 |
+
batch_size=per_device_eval_batch_size,
|
1674 |
+
drop_last=False,
|
1675 |
+
num_workers=training_args.dataloader_pin_memory,
|
1676 |
+
pin_memory=training_args.dataloader_pin_memory,
|
1677 |
+
)
|
1678 |
+
validation_dataloader = accelerator.prepare(validation_dataloader)
|
1679 |
+
# generation
|
1680 |
+
for batch in tqdm(
|
1681 |
+
validation_dataloader,
|
1682 |
+
desc="Evaluating - Generation ...",
|
1683 |
+
position=2,
|
1684 |
+
disable=not accelerator.is_local_main_process,
|
1685 |
+
):
|
1686 |
+
generated_audios = generate_step(batch)
|
1687 |
+
# Gather all predictions and targets
|
1688 |
+
generated_audios, input_ids, prompts = accelerator.pad_across_processes(
|
1689 |
+
(generated_audios, batch["input_ids"], batch["prompt_input_ids"]), dim=1, pad_index=0
|
1690 |
+
)
|
1691 |
+
generated_audios, input_ids, prompts = accelerator.gather_for_metrics(
|
1692 |
+
(generated_audios, input_ids, prompts)
|
1693 |
+
)
|
1694 |
+
eval_preds.extend(generated_audios.to("cpu"))
|
1695 |
+
eval_descriptions.extend(input_ids.to("cpu"))
|
1696 |
+
eval_prompts.extend(prompts.to("cpu"))
|
1697 |
+
|
1698 |
+
eval_time = time.time() - eval_start
|
1699 |
+
# normalize eval metrics
|
1700 |
+
eval_metrics = {
|
1701 |
+
key: torch.mean(torch.cat([d[key].unsqueeze(0) for d in eval_metrics]))
|
1702 |
+
for key in eval_metrics[0]
|
1703 |
+
}
|
1704 |
+
|
1705 |
+
# compute metrics
|
1706 |
+
metrics_desc = ""
|
1707 |
+
if training_args.predict_with_generate:
|
1708 |
+
metric_values, pred_descriptions, pred_prompts, audios, transcriptions = compute_metrics(
|
1709 |
+
eval_preds, eval_descriptions, eval_prompts, accelerator.device
|
1710 |
+
)
|
1711 |
+
eval_metrics.update(metric_values)
|
1712 |
+
metrics_desc = " ".join([f"Eval {key}: {value} |" for key, value in metric_values.items()])
|
1713 |
+
if "wandb" in training_args.report_to:
|
1714 |
+
log_pred(
|
1715 |
+
accelerator,
|
1716 |
+
pred_descriptions,
|
1717 |
+
pred_prompts,
|
1718 |
+
transcriptions,
|
1719 |
+
audios,
|
1720 |
+
sampling_rate=sampling_rate,
|
1721 |
+
step=cur_step,
|
1722 |
+
prefix="eval",
|
1723 |
+
)
|
1724 |
+
|
1725 |
+
# Print metrics and update progress bar
|
1726 |
+
steps_trained_progress_bar.write(
|
1727 |
+
f"Eval results for step ({cur_step} / {total_train_steps} | Eval Loss: {eval_metrics['loss']} |"
|
1728 |
+
f" {metrics_desc})"
|
1729 |
+
)
|
1730 |
+
|
1731 |
+
log_metric(
|
1732 |
+
accelerator,
|
1733 |
+
metrics=eval_metrics,
|
1734 |
+
train_time=eval_time,
|
1735 |
+
step=cur_step,
|
1736 |
+
epoch=epoch + (cur_step - epoch * steps_per_epoch) / steps_per_epoch,
|
1737 |
+
prefix="eval",
|
1738 |
+
)
|
1739 |
+
|
1740 |
+
# release eval batch and relax metrics
|
1741 |
+
eval_metrics = []
|
1742 |
+
eval_preds = []
|
1743 |
+
eval_descriptions = []
|
1744 |
+
eval_prompts = []
|
1745 |
+
batch = release_memory(batch)
|
1746 |
+
|
1747 |
+
# flush the train metrics
|
1748 |
+
train_start = time.time()
|
1749 |
+
|
1750 |
+
# break condition
|
1751 |
+
if cur_step == total_train_steps:
|
1752 |
+
continue_training = False
|
1753 |
+
break
|
1754 |
+
|
1755 |
+
if not continue_training:
|
1756 |
+
break
|
1757 |
+
|
1758 |
+
accelerator.end_training()
|
1759 |
+
|
1760 |
+
|
1761 |
+
if __name__ == "__main__":
|
1762 |
+
set_start_method("spawn")
|
1763 |
+
main()
|
wandb/debug-cli.sanchit.log
ADDED
File without changes
|
wandb/debug-internal.log
ADDED
The diff for this file is too large to render.
See raw diff
|
|
wandb/debug.log
ADDED
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_setup.py:_flush():76] Current SDK version is 0.17.0
|
2 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_setup.py:_flush():76] Configure stats pid to 1257680
|
3 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_setup.py:_flush():76] Loading settings from /home/sanchit/.config/wandb/settings
|
4 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_setup.py:_flush():76] Loading settings from /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/settings
|
5 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_setup.py:_flush():76] Loading settings from environment variables: {}
|
6 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_setup.py:_flush():76] Applying setup settings: {'_disable_service': False}
|
7 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_setup.py:_flush():76] Inferring run settings from compute environment: {'program_relpath': 'run_parler_tts_training.py', 'program_abspath': '/raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/run_parler_tts_training.py', 'program': '/raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/run_parler_tts_training.py'}
|
8 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_setup.py:_flush():76] Applying login settings: {}
|
9 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_init.py:_log_setup():520] Logging user logs to /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_205249-qaoje1x9/logs/debug.log
|
10 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_init.py:_log_setup():521] Logging internal logs to /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_205249-qaoje1x9/logs/debug-internal.log
|
11 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_init.py:init():560] calling init triggers
|
12 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_init.py:init():567] wandb.init called with sweep_config: {}
|
13 |
+
config: {}
|
14 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_init.py:init():610] starting backend
|
15 |
+
2024-05-13 20:52:49,015 INFO MainThread:1257680 [wandb_init.py:init():614] setting up manager
|
16 |
+
2024-05-13 20:52:49,019 INFO MainThread:1257680 [backend.py:_multiprocessing_setup():105] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
|
17 |
+
2024-05-13 20:52:49,019 INFO MainThread:1257680 [wandb_init.py:init():622] backend started and connected
|
18 |
+
2024-05-13 20:52:49,021 INFO MainThread:1257680 [wandb_init.py:init():711] updated telemetry
|
19 |
+
2024-05-13 20:52:49,024 INFO MainThread:1257680 [wandb_init.py:init():744] communicating run to backend with 90.0 second timeout
|
20 |
+
2024-05-13 20:52:49,413 INFO MainThread:1257680 [wandb_run.py:_on_init():2396] communicating current version
|
21 |
+
2024-05-13 20:52:49,474 INFO MainThread:1257680 [wandb_run.py:_on_init():2405] got version response
|
22 |
+
2024-05-13 20:52:49,475 INFO MainThread:1257680 [wandb_init.py:init():795] starting run threads in backend
|
23 |
+
2024-05-13 20:52:52,103 INFO MainThread:1257680 [wandb_run.py:_console_start():2374] atexit reg
|
24 |
+
2024-05-13 20:52:52,103 INFO MainThread:1257680 [wandb_run.py:_redirect():2229] redirect: wrap_raw
|
25 |
+
2024-05-13 20:52:52,103 INFO MainThread:1257680 [wandb_run.py:_redirect():2294] Wrapping output streams.
|
26 |
+
2024-05-13 20:52:52,103 INFO MainThread:1257680 [wandb_run.py:_redirect():2319] Redirects installed.
|
27 |
+
2024-05-13 20:52:52,104 INFO MainThread:1257680 [wandb_init.py:init():838] run started, returning control to user process
|
28 |
+
2024-05-13 20:52:52,104 INFO MainThread:1257680 [wandb_run.py:_config_callback():1376] config_cb None None {'learning_rate': 8e-05, 'model_name_or_path': 'parler-tts/parler_tts_mini_v0.1', 'num_train_epochs': 8, 'gradient_accumulation_steps': 8, 'per_device_train_batch_size': 16, 'global_batch_size': 16, 'mixed_precision': 'bf16', 'lr_scheduler_type': 'SchedulerType.COSINE', 'warmup_steps': 250, 'freeze_text_encoder': True, 'max_duration_in_seconds': 30.0, 'weight_decay': 0.01, 'adam_beta1': 0.9, 'adam_beta2': 0.99, 'temperature': 1.0}
|
29 |
+
2024-05-13 23:01:32,489 INFO MainThread:1257680 [wandb_run.py:_finish():2103] finishing run sanchit-gandhi/parler-speech/qaoje1x9
|
30 |
+
2024-05-13 23:01:32,489 INFO MainThread:1257680 [wandb_run.py:_atexit_cleanup():2343] got exitcode: 0
|
31 |
+
2024-05-13 23:01:32,489 INFO MainThread:1257680 [wandb_run.py:_restore():2326] restore
|
32 |
+
2024-05-13 23:01:32,489 INFO MainThread:1257680 [wandb_run.py:_restore():2332] restore done
|
33 |
+
2024-05-13 23:01:46,253 INFO MainThread:1257680 [wandb_run.py:_footer_history_summary_info():3994] rendering history
|
34 |
+
2024-05-13 23:01:46,254 INFO MainThread:1257680 [wandb_run.py:_footer_history_summary_info():4026] rendering summary
|
35 |
+
2024-05-13 23:01:46,256 INFO MainThread:1257680 [wandb_run.py:_footer_sync_info():3953] logging synced files
|
wandb/run-20240513_204652-m0g0ap7d/files/conda-environment.yaml
ADDED
@@ -0,0 +1,248 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
name: venv
|
2 |
+
channels:
|
3 |
+
- defaults
|
4 |
+
dependencies:
|
5 |
+
- _libgcc_mutex=0.1=main
|
6 |
+
- _openmp_mutex=5.1=1_gnu
|
7 |
+
- bzip2=1.0.8=h5eee18b_6
|
8 |
+
- ca-certificates=2024.3.11=h06a4308_0
|
9 |
+
- ld_impl_linux-64=2.38=h1181459_1
|
10 |
+
- libffi=3.4.4=h6a678d5_1
|
11 |
+
- libgcc-ng=11.2.0=h1234567_1
|
12 |
+
- libgomp=11.2.0=h1234567_1
|
13 |
+
- libstdcxx-ng=11.2.0=h1234567_1
|
14 |
+
- libuuid=1.41.5=h5eee18b_0
|
15 |
+
- ncurses=6.4=h6a678d5_0
|
16 |
+
- openssl=3.0.13=h7f8727e_1
|
17 |
+
- pip=24.0=py311h06a4308_0
|
18 |
+
- python=3.11.9=h955ad1f_0
|
19 |
+
- readline=8.2=h5eee18b_0
|
20 |
+
- setuptools=69.5.1=py311h06a4308_0
|
21 |
+
- sqlite=3.45.3=h5eee18b_0
|
22 |
+
- tk=8.6.14=h39e8969_0
|
23 |
+
- wheel=0.43.0=py311h06a4308_0
|
24 |
+
- xz=5.4.6=h5eee18b_1
|
25 |
+
- zlib=1.2.13=h5eee18b_1
|
26 |
+
- pip:
|
27 |
+
- absl-py==2.1.0
|
28 |
+
- accelerate==0.30.0
|
29 |
+
- aiohttp==3.9.5
|
30 |
+
- aiosignal==1.3.1
|
31 |
+
- aniso8601==9.0.1
|
32 |
+
- annotated-types==0.6.0
|
33 |
+
- anyio==4.3.0
|
34 |
+
- argbind==0.3.7
|
35 |
+
- argon2-cffi==23.1.0
|
36 |
+
- argon2-cffi-bindings==21.2.0
|
37 |
+
- arrow==1.3.0
|
38 |
+
- asttokens==2.4.1
|
39 |
+
- async-lru==2.0.4
|
40 |
+
- attrs==23.2.0
|
41 |
+
- audioread==3.0.1
|
42 |
+
- babel==2.15.0
|
43 |
+
- beautifulsoup4==4.12.3
|
44 |
+
- bidict==0.23.1
|
45 |
+
- bitsandbytes==0.43.1
|
46 |
+
- bleach==6.1.0
|
47 |
+
- certifi==2024.2.2
|
48 |
+
- cffi==1.16.0
|
49 |
+
- charset-normalizer==3.3.2
|
50 |
+
- click==8.1.7
|
51 |
+
- coloredlogs==14.0
|
52 |
+
- comm==0.2.2
|
53 |
+
- contourpy==1.2.1
|
54 |
+
- cycler==0.12.1
|
55 |
+
- datasets==2.19.1
|
56 |
+
- debugpy==1.8.1
|
57 |
+
- decorator==5.1.1
|
58 |
+
- defusedxml==0.7.1
|
59 |
+
- descript-audio-codec==1.0.0
|
60 |
+
- descript-audiotools==0.7.2
|
61 |
+
- dill==0.3.8
|
62 |
+
- dnspython==2.3.0
|
63 |
+
- docker-pycreds==0.4.0
|
64 |
+
- docstring-parser==0.16
|
65 |
+
- editdistance==0.8.1
|
66 |
+
- einops==0.8.0
|
67 |
+
- et-xmlfile==1.1.0
|
68 |
+
- evaluate==0.4.2
|
69 |
+
- eventlet==0.36.1
|
70 |
+
- executing==2.0.1
|
71 |
+
- fastjsonschema==2.19.1
|
72 |
+
- ffmpy==0.3.2
|
73 |
+
- filelock==3.14.0
|
74 |
+
- fire==0.6.0
|
75 |
+
- flask==2.2.5
|
76 |
+
- flask-cors==4.0.1
|
77 |
+
- flask-restful==0.3.10
|
78 |
+
- flask-socketio==5.3.6
|
79 |
+
- flask-talisman==1.1.0
|
80 |
+
- flatten-dict==0.4.2
|
81 |
+
- fonttools==4.51.0
|
82 |
+
- fqdn==1.5.1
|
83 |
+
- frozenlist==1.4.1
|
84 |
+
- fsspec==2024.3.1
|
85 |
+
- future==1.0.0
|
86 |
+
- g2p==2.0.0
|
87 |
+
- gitdb==4.0.11
|
88 |
+
- gitpython==3.1.43
|
89 |
+
- greenlet==3.0.3
|
90 |
+
- grpcio==1.63.0
|
91 |
+
- h11==0.14.0
|
92 |
+
- httpcore==1.0.5
|
93 |
+
- httpx==0.27.0
|
94 |
+
- huggingface-hub==0.23.0
|
95 |
+
- humanfriendly==10.0
|
96 |
+
- idna==3.7
|
97 |
+
- importlib-resources==6.4.0
|
98 |
+
- ipdb==0.13.13
|
99 |
+
- ipykernel==6.29.4
|
100 |
+
- ipython==8.24.0
|
101 |
+
- isoduration==20.11.0
|
102 |
+
- itsdangerous==2.2.0
|
103 |
+
- jedi==0.19.1
|
104 |
+
- jinja2==3.1.4
|
105 |
+
- jiwer==3.0.4
|
106 |
+
- joblib==1.4.2
|
107 |
+
- json5==0.9.25
|
108 |
+
- jsonpointer==2.4
|
109 |
+
- jsonschema==4.22.0
|
110 |
+
- jsonschema-specifications==2023.12.1
|
111 |
+
- julius==0.2.7
|
112 |
+
- jupyter-client==8.6.1
|
113 |
+
- jupyter-core==5.7.2
|
114 |
+
- jupyter-events==0.10.0
|
115 |
+
- jupyter-lsp==2.2.5
|
116 |
+
- jupyter-server==2.14.0
|
117 |
+
- jupyter-server-terminals==0.5.3
|
118 |
+
- jupyterlab==4.2.0
|
119 |
+
- jupyterlab-pygments==0.3.0
|
120 |
+
- jupyterlab-server==2.27.1
|
121 |
+
- kiwisolver==1.4.5
|
122 |
+
- lazy-loader==0.4
|
123 |
+
- librosa==0.10.2
|
124 |
+
- llvmlite==0.42.0
|
125 |
+
- markdown==3.6
|
126 |
+
- markdown-it-py==3.0.0
|
127 |
+
- markdown2==2.4.13
|
128 |
+
- markupsafe==2.1.5
|
129 |
+
- matplotlib==3.8.4
|
130 |
+
- matplotlib-inline==0.1.7
|
131 |
+
- mdurl==0.1.2
|
132 |
+
- mistune==3.0.2
|
133 |
+
- mpmath==1.3.0
|
134 |
+
- msgpack==1.0.8
|
135 |
+
- multidict==6.0.5
|
136 |
+
- multiprocess==0.70.16
|
137 |
+
- munkres==1.1.4
|
138 |
+
- nbclient==0.10.0
|
139 |
+
- nbconvert==7.16.4
|
140 |
+
- nbformat==5.10.4
|
141 |
+
- nest-asyncio==1.6.0
|
142 |
+
- networkx==3.3
|
143 |
+
- notebook-shim==0.2.4
|
144 |
+
- numba==0.59.1
|
145 |
+
- numpy==1.26.4
|
146 |
+
- nvidia-cublas-cu12==12.1.3.1
|
147 |
+
- nvidia-cuda-cupti-cu12==12.1.105
|
148 |
+
- nvidia-cuda-nvrtc-cu12==12.1.105
|
149 |
+
- nvidia-cuda-runtime-cu12==12.1.105
|
150 |
+
- nvidia-cudnn-cu12==8.9.2.26
|
151 |
+
- nvidia-cufft-cu12==11.0.2.54
|
152 |
+
- nvidia-curand-cu12==10.3.2.106
|
153 |
+
- nvidia-cusolver-cu12==11.4.5.107
|
154 |
+
- nvidia-cusparse-cu12==12.1.0.106
|
155 |
+
- nvidia-nccl-cu12==2.20.5
|
156 |
+
- nvidia-nvjitlink-cu12==12.4.127
|
157 |
+
- nvidia-nvtx-cu12==12.1.105
|
158 |
+
- openpyxl==3.1.2
|
159 |
+
- overrides==7.7.0
|
160 |
+
- packaging==24.0
|
161 |
+
- pandas==2.2.2
|
162 |
+
- pandocfilters==1.5.1
|
163 |
+
- panphon==0.20.0
|
164 |
+
- parler-tts==0.1
|
165 |
+
- parso==0.8.4
|
166 |
+
- pexpect==4.9.0
|
167 |
+
- pillow==10.3.0
|
168 |
+
- platformdirs==4.2.1
|
169 |
+
- pooch==1.8.1
|
170 |
+
- prometheus-client==0.20.0
|
171 |
+
- prompt-toolkit==3.0.43
|
172 |
+
- protobuf==3.19.6
|
173 |
+
- psutil==5.9.8
|
174 |
+
- ptyprocess==0.7.0
|
175 |
+
- pure-eval==0.2.2
|
176 |
+
- pyarrow==16.0.0
|
177 |
+
- pyarrow-hotfix==0.6
|
178 |
+
- pycparser==2.22
|
179 |
+
- pydantic==2.7.1
|
180 |
+
- pydantic-core==2.18.2
|
181 |
+
- pygments==2.18.0
|
182 |
+
- pyloudnorm==0.1.1
|
183 |
+
- pyparsing==3.1.2
|
184 |
+
- pystoi==0.4.1
|
185 |
+
- python-dateutil==2.9.0.post0
|
186 |
+
- python-engineio==4.9.0
|
187 |
+
- python-json-logger==2.0.7
|
188 |
+
- python-socketio==5.11.2
|
189 |
+
- pytz==2024.1
|
190 |
+
- pyyaml==6.0.1
|
191 |
+
- pyzmq==26.0.3
|
192 |
+
- randomname==0.2.1
|
193 |
+
- rapidfuzz==3.9.0
|
194 |
+
- referencing==0.35.1
|
195 |
+
- regex==2024.4.28
|
196 |
+
- requests==2.31.0
|
197 |
+
- rfc3339-validator==0.1.4
|
198 |
+
- rfc3986-validator==0.1.1
|
199 |
+
- rich==13.7.1
|
200 |
+
- rpds-py==0.18.1
|
201 |
+
- safetensors==0.4.3
|
202 |
+
- scikit-learn==1.4.2
|
203 |
+
- scipy==1.13.0
|
204 |
+
- send2trash==1.8.3
|
205 |
+
- sentencepiece==0.2.0
|
206 |
+
- sentry-sdk==2.1.1
|
207 |
+
- setproctitle==1.3.3
|
208 |
+
- simple-websocket==1.0.0
|
209 |
+
- six==1.16.0
|
210 |
+
- smmap==5.0.1
|
211 |
+
- sniffio==1.3.1
|
212 |
+
- soundfile==0.12.1
|
213 |
+
- soupsieve==2.5
|
214 |
+
- soxr==0.3.7
|
215 |
+
- stack-data==0.6.3
|
216 |
+
- sympy==1.12
|
217 |
+
- tensorboard==2.16.2
|
218 |
+
- tensorboard-data-server==0.7.2
|
219 |
+
- termcolor==2.4.0
|
220 |
+
- terminado==0.18.1
|
221 |
+
- text-unidecode==1.3
|
222 |
+
- threadpoolctl==3.5.0
|
223 |
+
- tinycss2==1.3.0
|
224 |
+
- tokenizers==0.19.1
|
225 |
+
- torch==2.3.0
|
226 |
+
- torch-stoi==0.2.1
|
227 |
+
- torchaudio==2.3.0
|
228 |
+
- tornado==6.4
|
229 |
+
- tqdm==4.66.4
|
230 |
+
- traitlets==5.14.3
|
231 |
+
- transformers==4.41.0.dev0
|
232 |
+
- triton==2.3.0
|
233 |
+
- types-python-dateutil==2.9.0.20240316
|
234 |
+
- typing-extensions==4.11.0
|
235 |
+
- tzdata==2024.1
|
236 |
+
- unicodecsv==0.14.1
|
237 |
+
- uri-template==1.3.0
|
238 |
+
- urllib3==2.2.1
|
239 |
+
- wandb==0.17.0
|
240 |
+
- wcwidth==0.2.13
|
241 |
+
- webcolors==1.13
|
242 |
+
- webencodings==0.5.1
|
243 |
+
- websocket-client==1.8.0
|
244 |
+
- werkzeug==3.0.3
|
245 |
+
- wsproto==1.2.0
|
246 |
+
- xxhash==3.4.1
|
247 |
+
- yarl==1.9.4
|
248 |
+
prefix: /home/sanchit/miniconda3/envs/venv
|
wandb/run-20240513_204652-m0g0ap7d/files/config.yaml
ADDED
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
wandb_version: 1
|
2 |
+
|
3 |
+
_wandb:
|
4 |
+
desc: null
|
5 |
+
value:
|
6 |
+
python_version: 3.11.9
|
7 |
+
cli_version: 0.17.0
|
8 |
+
framework: huggingface
|
9 |
+
huggingface_version: 4.41.0.dev0
|
10 |
+
is_jupyter_run: false
|
11 |
+
is_kaggle_kernel: false
|
12 |
+
start_time: 1715626012
|
13 |
+
t:
|
14 |
+
1:
|
15 |
+
- 1
|
16 |
+
- 5
|
17 |
+
- 11
|
18 |
+
- 49
|
19 |
+
- 51
|
20 |
+
- 53
|
21 |
+
- 55
|
22 |
+
- 71
|
23 |
+
- 100
|
24 |
+
2:
|
25 |
+
- 1
|
26 |
+
- 5
|
27 |
+
- 11
|
28 |
+
- 49
|
29 |
+
- 51
|
30 |
+
- 53
|
31 |
+
- 55
|
32 |
+
- 71
|
33 |
+
- 100
|
34 |
+
3:
|
35 |
+
- 23
|
36 |
+
4: 3.11.9
|
37 |
+
5: 0.17.0
|
38 |
+
6: 4.41.0.dev0
|
39 |
+
8:
|
40 |
+
- 5
|
41 |
+
13: linux-x86_64
|
42 |
+
learning_rate:
|
43 |
+
desc: null
|
44 |
+
value: 8.0e-05
|
45 |
+
model_name_or_path:
|
46 |
+
desc: null
|
47 |
+
value: parler-tts/parler_tts_mini_v0.1
|
48 |
+
num_train_epochs:
|
49 |
+
desc: null
|
50 |
+
value: 8
|
51 |
+
gradient_accumulation_steps:
|
52 |
+
desc: null
|
53 |
+
value: 8
|
54 |
+
per_device_train_batch_size:
|
55 |
+
desc: null
|
56 |
+
value: 16
|
57 |
+
global_batch_size:
|
58 |
+
desc: null
|
59 |
+
value: 16
|
60 |
+
mixed_precision:
|
61 |
+
desc: null
|
62 |
+
value: bf16
|
63 |
+
lr_scheduler_type:
|
64 |
+
desc: null
|
65 |
+
value: SchedulerType.COSINE
|
66 |
+
warmup_steps:
|
67 |
+
desc: null
|
68 |
+
value: 250
|
69 |
+
freeze_text_encoder:
|
70 |
+
desc: null
|
71 |
+
value: true
|
72 |
+
max_duration_in_seconds:
|
73 |
+
desc: null
|
74 |
+
value: 30.0
|
75 |
+
weight_decay:
|
76 |
+
desc: null
|
77 |
+
value: 0.01
|
78 |
+
adam_beta1:
|
79 |
+
desc: null
|
80 |
+
value: 0.9
|
81 |
+
adam_beta2:
|
82 |
+
desc: null
|
83 |
+
value: 0.99
|
84 |
+
temperature:
|
85 |
+
desc: null
|
86 |
+
value: 1.0
|
wandb/run-20240513_204652-m0g0ap7d/files/output.log
ADDED
@@ -0,0 +1,180 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
05/13/2024 20:46:55 - WARNING - __main__ - Process rank: 0, device: cuda:0, n_gpu: 1, distributed training: False, 16-bits training: False
|
2 |
+
05/13/2024 20:46:55 - INFO - __main__ - Training/evaluation parameters ParlerTTSTrainingArguments(
|
3 |
+
_n_gpu=1,
|
4 |
+
accelerator_config={'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None},
|
5 |
+
adafactor=False,
|
6 |
+
adam_beta1=0.9,
|
7 |
+
adam_beta2=0.99,
|
8 |
+
adam_epsilon=1e-08,
|
9 |
+
audio_encoder_per_device_batch_size=4,
|
10 |
+
auto_find_batch_size=False,
|
11 |
+
batch_eval_metrics=False,
|
12 |
+
bf16=False,
|
13 |
+
bf16_full_eval=False,
|
14 |
+
data_seed=None,
|
15 |
+
dataloader_drop_last=False,
|
16 |
+
dataloader_num_workers=4,
|
17 |
+
dataloader_persistent_workers=False,
|
18 |
+
dataloader_pin_memory=True,
|
19 |
+
dataloader_prefetch_factor=None,
|
20 |
+
ddp_backend=None,
|
21 |
+
ddp_broadcast_buffers=None,
|
22 |
+
ddp_bucket_cap_mb=None,
|
23 |
+
ddp_find_unused_parameters=None,
|
24 |
+
ddp_timeout=1800,
|
25 |
+
debug=[],
|
26 |
+
deepspeed=None,
|
27 |
+
disable_tqdm=False,
|
28 |
+
dispatch_batches=None,
|
29 |
+
do_eval=True,
|
30 |
+
do_predict=False,
|
31 |
+
do_train=True,
|
32 |
+
dtype=bfloat16,
|
33 |
+
eval_accumulation_steps=None,
|
34 |
+
eval_delay=0,
|
35 |
+
eval_do_concat_batches=True,
|
36 |
+
eval_steps=None,
|
37 |
+
eval_strategy=IntervalStrategy.EPOCH,
|
38 |
+
evaluation_strategy=epoch,
|
39 |
+
fp16=False,
|
40 |
+
fp16_backend=auto,
|
41 |
+
fp16_full_eval=False,
|
42 |
+
fp16_opt_level=O1,
|
43 |
+
fsdp=[],
|
44 |
+
fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False},
|
45 |
+
fsdp_min_num_params=0,
|
46 |
+
fsdp_transformer_layer_cls_to_wrap=None,
|
47 |
+
full_determinism=False,
|
48 |
+
generation_config=None,
|
49 |
+
generation_max_length=None,
|
50 |
+
generation_num_beams=None,
|
51 |
+
gradient_accumulation_steps=8,
|
52 |
+
gradient_checkpointing=True,
|
53 |
+
gradient_checkpointing_kwargs=None,
|
54 |
+
greater_is_better=None,
|
55 |
+
group_by_length=True,
|
56 |
+
half_precision_backend=auto,
|
57 |
+
hub_always_push=False,
|
58 |
+
hub_model_id=None,
|
59 |
+
hub_private_repo=False,
|
60 |
+
hub_strategy=HubStrategy.EVERY_SAVE,
|
61 |
+
hub_token=<HUB_TOKEN>,
|
62 |
+
ignore_data_skip=False,
|
63 |
+
include_inputs_for_metrics=True,
|
64 |
+
include_num_input_tokens_seen=False,
|
65 |
+
include_tokens_per_second=False,
|
66 |
+
jit_mode_eval=False,
|
67 |
+
label_names=None,
|
68 |
+
label_smoothing_factor=0.0,
|
69 |
+
learning_rate=8e-05,
|
70 |
+
length_column_name=length,
|
71 |
+
load_best_model_at_end=False,
|
72 |
+
local_rank=0,
|
73 |
+
log_level=passive,
|
74 |
+
log_level_replica=warning,
|
75 |
+
log_on_each_node=True,
|
76 |
+
logging_dir=../output_dir_training_constant_concat/runs/May13_20-46-51_hf-dgx-01,
|
77 |
+
logging_first_step=False,
|
78 |
+
logging_nan_inf_filter=True,
|
79 |
+
logging_steps=5,
|
80 |
+
logging_strategy=IntervalStrategy.STEPS,
|
81 |
+
lr_scheduler_kwargs={},
|
82 |
+
lr_scheduler_type=SchedulerType.COSINE,
|
83 |
+
max_grad_norm=1.0,
|
84 |
+
max_steps=-1,
|
85 |
+
metric_for_best_model=None,
|
86 |
+
mp_parameters=,
|
87 |
+
neftune_noise_alpha=None,
|
88 |
+
no_cuda=False,
|
89 |
+
num_train_epochs=8,
|
90 |
+
optim=OptimizerNames.ADAMW_TORCH,
|
91 |
+
optim_args=None,
|
92 |
+
optim_target_modules=None,
|
93 |
+
output_dir=../output_dir_training_constant_concat/,
|
94 |
+
overwrite_output_dir=True,
|
95 |
+
past_index=-1,
|
96 |
+
per_device_eval_batch_size=16,
|
97 |
+
per_device_train_batch_size=16,
|
98 |
+
predict_with_generate=True,
|
99 |
+
prediction_loss_only=False,
|
100 |
+
push_to_hub=False,
|
101 |
+
push_to_hub_model_id=None,
|
102 |
+
push_to_hub_organization=None,
|
103 |
+
push_to_hub_token=<PUSH_TO_HUB_TOKEN>,
|
104 |
+
ray_scope=last,
|
105 |
+
remove_unused_columns=True,
|
106 |
+
report_to=['wandb'],
|
107 |
+
restore_callback_states_from_checkpoint=False,
|
108 |
+
resume_from_checkpoint=None,
|
109 |
+
run_name=../output_dir_training_constant_concat/,
|
110 |
+
save_on_each_node=False,
|
111 |
+
save_only_model=False,
|
112 |
+
save_safetensors=True,
|
113 |
+
save_steps=500,
|
114 |
+
save_strategy=IntervalStrategy.EPOCH,
|
115 |
+
save_total_limit=5,
|
116 |
+
seed=456,
|
117 |
+
skip_memory_metrics=True,
|
118 |
+
sortish_sampler=False,
|
119 |
+
split_batches=None,
|
120 |
+
tf32=None,
|
121 |
+
torch_compile=False,
|
122 |
+
torch_compile_backend=None,
|
123 |
+
torch_compile_mode=None,
|
124 |
+
torchdynamo=None,
|
125 |
+
tpu_metrics_debug=False,
|
126 |
+
tpu_num_cores=None,
|
127 |
+
use_cpu=False,
|
128 |
+
use_ipex=False,
|
129 |
+
use_legacy_prediction_loop=False,
|
130 |
+
use_mps_device=False,
|
131 |
+
warmup_ratio=0.0,
|
132 |
+
warmup_steps=250,
|
133 |
+
weight_decay=0.01,
|
134 |
+
)
|
135 |
+
05/13/2024 20:46:57 - WARNING - __main__ - Disabling fast tokenizer warning: https://github.com/huggingface/transformers/blob/main/src/transformers/tokenization_utils_base.py#L3231-L3235
|
136 |
+
loading configuration file preprocessor_config.json from cache at /raid/.cache/huggingface/models--parler-tts--dac_44khZ_8kbps/snapshots/db52bea859d9411e0beb44a3ea923a8731ee4197/preprocessor_config.json
|
137 |
+
Feature extractor EncodecFeatureExtractor {
|
138 |
+
"chunk_length_s": null,
|
139 |
+
"feature_extractor_type": "EncodecFeatureExtractor",
|
140 |
+
"feature_size": 1,
|
141 |
+
"overlap": null,
|
142 |
+
"padding_side": "right",
|
143 |
+
"padding_value": 0.0,
|
144 |
+
"return_attention_mask": true,
|
145 |
+
"sampling_rate": 44100
|
146 |
+
}
|
147 |
+
loading file spiece.model from cache at /raid/.cache/huggingface/models--parler-tts--parler_tts_mini_v0.1/snapshots/e02fd18e77d38b49a85c7a9a85189a64b8472544/spiece.model
|
148 |
+
loading file tokenizer.json from cache at /raid/.cache/huggingface/models--parler-tts--parler_tts_mini_v0.1/snapshots/e02fd18e77d38b49a85c7a9a85189a64b8472544/tokenizer.json
|
149 |
+
loading file added_tokens.json from cache at None
|
150 |
+
loading file special_tokens_map.json from cache at /raid/.cache/huggingface/models--parler-tts--parler_tts_mini_v0.1/snapshots/e02fd18e77d38b49a85c7a9a85189a64b8472544/special_tokens_map.json
|
151 |
+
loading file tokenizer_config.json from cache at /raid/.cache/huggingface/models--parler-tts--parler_tts_mini_v0.1/snapshots/e02fd18e77d38b49a85c7a9a85189a64b8472544/tokenizer_config.json
|
152 |
+
You set `add_prefix_space`. The tokenizer needs to be converted from the slow tokenizers
|
153 |
+
loading file spiece.model from cache at /raid/.cache/huggingface/models--parler-tts--parler_tts_mini_v0.1/snapshots/e02fd18e77d38b49a85c7a9a85189a64b8472544/spiece.model
|
154 |
+
loading file tokenizer.json from cache at /raid/.cache/huggingface/models--parler-tts--parler_tts_mini_v0.1/snapshots/e02fd18e77d38b49a85c7a9a85189a64b8472544/tokenizer.json
|
155 |
+
loading file added_tokens.json from cache at None
|
156 |
+
loading file special_tokens_map.json from cache at /raid/.cache/huggingface/models--parler-tts--parler_tts_mini_v0.1/snapshots/e02fd18e77d38b49a85c7a9a85189a64b8472544/special_tokens_map.json
|
157 |
+
loading file tokenizer_config.json from cache at /raid/.cache/huggingface/models--parler-tts--parler_tts_mini_v0.1/snapshots/e02fd18e77d38b49a85c7a9a85189a64b8472544/tokenizer_config.json
|
158 |
+
Combining datasets...: 0%| | 0/4 [00:00<?, ?it/s]
|
159 |
+
Combining datasets...: 0%| | 0/4 [03:35<?, ?it/s]
|
160 |
+
Traceback (most recent call last):
|
161 |
+
File "/raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/run_parler_tts_training.py", line 1763, in <module>
|
162 |
+
main()
|
163 |
+
File "/raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/run_parler_tts_training.py", line 950, in main
|
164 |
+
raw_datasets["train"] = load_multiple_datasets(
|
165 |
+
^^^^^^^^^^^^^^^^^^^^^^^
|
166 |
+
File "/raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/run_parler_tts_training.py", line 693, in load_multiple_datasets
|
167 |
+
metadata_dataset = load_dataset(
|
168 |
+
^^^^^^^^^^^^^
|
169 |
+
File "/home/sanchit/miniconda3/envs/venv/lib/python3.11/site-packages/datasets/load.py", line 2587, in load_dataset
|
170 |
+
builder_instance = load_dataset_builder(
|
171 |
+
^^^^^^^^^^^^^^^^^^^^^
|
172 |
+
File "/home/sanchit/miniconda3/envs/venv/lib/python3.11/site-packages/datasets/load.py", line 2296, in load_dataset_builder
|
173 |
+
builder_instance: DatasetBuilder = builder_cls(
|
174 |
+
^^^^^^^^^^^^
|
175 |
+
File "/home/sanchit/miniconda3/envs/venv/lib/python3.11/site-packages/datasets/builder.py", line 374, in __init__
|
176 |
+
self.config, self.config_id = self._create_builder_config(
|
177 |
+
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
178 |
+
File "/home/sanchit/miniconda3/envs/venv/lib/python3.11/site-packages/datasets/builder.py", line 599, in _create_builder_config
|
179 |
+
raise ValueError(
|
180 |
+
ValueError: BuilderConfig 'read' not found. Available: ['default']
|
wandb/run-20240513_204652-m0g0ap7d/files/requirements.txt
ADDED
@@ -0,0 +1,225 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
Babel==2.15.0
|
2 |
+
Flask-Cors==4.0.1
|
3 |
+
Flask-RESTful==0.3.10
|
4 |
+
Flask-SocketIO==5.3.6
|
5 |
+
Flask==2.2.5
|
6 |
+
GitPython==3.1.43
|
7 |
+
Jinja2==3.1.4
|
8 |
+
Markdown==3.6
|
9 |
+
MarkupSafe==2.1.5
|
10 |
+
PyYAML==6.0.1
|
11 |
+
Pygments==2.18.0
|
12 |
+
Send2Trash==1.8.3
|
13 |
+
Werkzeug==3.0.3
|
14 |
+
absl-py==2.1.0
|
15 |
+
accelerate==0.30.0
|
16 |
+
aiohttp==3.9.5
|
17 |
+
aiosignal==1.3.1
|
18 |
+
aniso8601==9.0.1
|
19 |
+
annotated-types==0.6.0
|
20 |
+
anyio==4.3.0
|
21 |
+
argbind==0.3.7
|
22 |
+
argon2-cffi-bindings==21.2.0
|
23 |
+
argon2-cffi==23.1.0
|
24 |
+
arrow==1.3.0
|
25 |
+
asttokens==2.4.1
|
26 |
+
async-lru==2.0.4
|
27 |
+
attrs==23.2.0
|
28 |
+
audioread==3.0.1
|
29 |
+
beautifulsoup4==4.12.3
|
30 |
+
bidict==0.23.1
|
31 |
+
bitsandbytes==0.43.1
|
32 |
+
bleach==6.1.0
|
33 |
+
certifi==2024.2.2
|
34 |
+
cffi==1.16.0
|
35 |
+
charset-normalizer==3.3.2
|
36 |
+
click==8.1.7
|
37 |
+
coloredlogs==14.0
|
38 |
+
comm==0.2.2
|
39 |
+
contourpy==1.2.1
|
40 |
+
cycler==0.12.1
|
41 |
+
datasets==2.19.1
|
42 |
+
debugpy==1.8.1
|
43 |
+
decorator==5.1.1
|
44 |
+
defusedxml==0.7.1
|
45 |
+
descript-audio-codec==1.0.0
|
46 |
+
descript-audiotools==0.7.2
|
47 |
+
dill==0.3.8
|
48 |
+
dnspython==2.3.0
|
49 |
+
docker-pycreds==0.4.0
|
50 |
+
docstring_parser==0.16
|
51 |
+
editdistance==0.8.1
|
52 |
+
einops==0.8.0
|
53 |
+
et-xmlfile==1.1.0
|
54 |
+
evaluate==0.4.2
|
55 |
+
eventlet==0.36.1
|
56 |
+
executing==2.0.1
|
57 |
+
fastjsonschema==2.19.1
|
58 |
+
ffmpy==0.3.2
|
59 |
+
filelock==3.14.0
|
60 |
+
fire==0.6.0
|
61 |
+
flask-talisman==1.1.0
|
62 |
+
flatten-dict==0.4.2
|
63 |
+
fonttools==4.51.0
|
64 |
+
fqdn==1.5.1
|
65 |
+
frozenlist==1.4.1
|
66 |
+
fsspec==2024.3.1
|
67 |
+
future==1.0.0
|
68 |
+
g2p==2.0.0
|
69 |
+
gitdb==4.0.11
|
70 |
+
greenlet==3.0.3
|
71 |
+
grpcio==1.63.0
|
72 |
+
h11==0.14.0
|
73 |
+
httpcore==1.0.5
|
74 |
+
httpx==0.27.0
|
75 |
+
huggingface-hub==0.23.0
|
76 |
+
humanfriendly==10.0
|
77 |
+
idna==3.7
|
78 |
+
importlib_resources==6.4.0
|
79 |
+
ipdb==0.13.13
|
80 |
+
ipykernel==6.29.4
|
81 |
+
ipython==8.24.0
|
82 |
+
isoduration==20.11.0
|
83 |
+
itsdangerous==2.2.0
|
84 |
+
jedi==0.19.1
|
85 |
+
jiwer==3.0.4
|
86 |
+
joblib==1.4.2
|
87 |
+
json5==0.9.25
|
88 |
+
jsonpointer==2.4
|
89 |
+
jsonschema-specifications==2023.12.1
|
90 |
+
jsonschema==4.22.0
|
91 |
+
julius==0.2.7
|
92 |
+
jupyter-events==0.10.0
|
93 |
+
jupyter-lsp==2.2.5
|
94 |
+
jupyter_client==8.6.1
|
95 |
+
jupyter_core==5.7.2
|
96 |
+
jupyter_server==2.14.0
|
97 |
+
jupyter_server_terminals==0.5.3
|
98 |
+
jupyterlab==4.2.0
|
99 |
+
jupyterlab_pygments==0.3.0
|
100 |
+
jupyterlab_server==2.27.1
|
101 |
+
kiwisolver==1.4.5
|
102 |
+
lazy_loader==0.4
|
103 |
+
librosa==0.10.2
|
104 |
+
llvmlite==0.42.0
|
105 |
+
markdown-it-py==3.0.0
|
106 |
+
markdown2==2.4.13
|
107 |
+
matplotlib-inline==0.1.7
|
108 |
+
matplotlib==3.8.4
|
109 |
+
mdurl==0.1.2
|
110 |
+
mistune==3.0.2
|
111 |
+
mpmath==1.3.0
|
112 |
+
msgpack==1.0.8
|
113 |
+
multidict==6.0.5
|
114 |
+
multiprocess==0.70.16
|
115 |
+
munkres==1.1.4
|
116 |
+
nbclient==0.10.0
|
117 |
+
nbconvert==7.16.4
|
118 |
+
nbformat==5.10.4
|
119 |
+
nest-asyncio==1.6.0
|
120 |
+
networkx==3.3
|
121 |
+
notebook_shim==0.2.4
|
122 |
+
numba==0.59.1
|
123 |
+
numpy==1.26.4
|
124 |
+
nvidia-cublas-cu12==12.1.3.1
|
125 |
+
nvidia-cuda-cupti-cu12==12.1.105
|
126 |
+
nvidia-cuda-nvrtc-cu12==12.1.105
|
127 |
+
nvidia-cuda-runtime-cu12==12.1.105
|
128 |
+
nvidia-cudnn-cu12==8.9.2.26
|
129 |
+
nvidia-cufft-cu12==11.0.2.54
|
130 |
+
nvidia-curand-cu12==10.3.2.106
|
131 |
+
nvidia-cusolver-cu12==11.4.5.107
|
132 |
+
nvidia-cusparse-cu12==12.1.0.106
|
133 |
+
nvidia-nccl-cu12==2.20.5
|
134 |
+
nvidia-nvjitlink-cu12==12.4.127
|
135 |
+
nvidia-nvtx-cu12==12.1.105
|
136 |
+
openpyxl==3.1.2
|
137 |
+
overrides==7.7.0
|
138 |
+
packaging==24.0
|
139 |
+
pandas==2.2.2
|
140 |
+
pandocfilters==1.5.1
|
141 |
+
panphon==0.20.0
|
142 |
+
parler_tts==0.1
|
143 |
+
parso==0.8.4
|
144 |
+
pexpect==4.9.0
|
145 |
+
pillow==10.3.0
|
146 |
+
pip==24.0
|
147 |
+
platformdirs==4.2.1
|
148 |
+
pooch==1.8.1
|
149 |
+
prometheus_client==0.20.0
|
150 |
+
prompt-toolkit==3.0.43
|
151 |
+
protobuf==3.19.6
|
152 |
+
psutil==5.9.8
|
153 |
+
ptyprocess==0.7.0
|
154 |
+
pure-eval==0.2.2
|
155 |
+
pyarrow-hotfix==0.6
|
156 |
+
pyarrow==16.0.0
|
157 |
+
pycparser==2.22
|
158 |
+
pydantic==2.7.1
|
159 |
+
pydantic_core==2.18.2
|
160 |
+
pyloudnorm==0.1.1
|
161 |
+
pyparsing==3.1.2
|
162 |
+
pystoi==0.4.1
|
163 |
+
python-dateutil==2.9.0.post0
|
164 |
+
python-engineio==4.9.0
|
165 |
+
python-json-logger==2.0.7
|
166 |
+
python-socketio==5.11.2
|
167 |
+
pytz==2024.1
|
168 |
+
pyzmq==26.0.3
|
169 |
+
randomname==0.2.1
|
170 |
+
rapidfuzz==3.9.0
|
171 |
+
referencing==0.35.1
|
172 |
+
regex==2024.4.28
|
173 |
+
requests==2.31.0
|
174 |
+
rfc3339-validator==0.1.4
|
175 |
+
rfc3986-validator==0.1.1
|
176 |
+
rich==13.7.1
|
177 |
+
rpds-py==0.18.1
|
178 |
+
safetensors==0.4.3
|
179 |
+
scikit-learn==1.4.2
|
180 |
+
scipy==1.13.0
|
181 |
+
sentencepiece==0.2.0
|
182 |
+
sentry-sdk==2.1.1
|
183 |
+
setproctitle==1.3.3
|
184 |
+
setuptools==69.5.1
|
185 |
+
simple-websocket==1.0.0
|
186 |
+
six==1.16.0
|
187 |
+
smmap==5.0.1
|
188 |
+
sniffio==1.3.1
|
189 |
+
soundfile==0.12.1
|
190 |
+
soupsieve==2.5
|
191 |
+
soxr==0.3.7
|
192 |
+
stack-data==0.6.3
|
193 |
+
sympy==1.12
|
194 |
+
tensorboard-data-server==0.7.2
|
195 |
+
tensorboard==2.16.2
|
196 |
+
termcolor==2.4.0
|
197 |
+
terminado==0.18.1
|
198 |
+
text-unidecode==1.3
|
199 |
+
threadpoolctl==3.5.0
|
200 |
+
tinycss2==1.3.0
|
201 |
+
tokenizers==0.19.1
|
202 |
+
torch-stoi==0.2.1
|
203 |
+
torch==2.3.0
|
204 |
+
torchaudio==2.3.0
|
205 |
+
tornado==6.4
|
206 |
+
tqdm==4.66.4
|
207 |
+
traitlets==5.14.3
|
208 |
+
transformers==4.41.0.dev0
|
209 |
+
transformers==4.41.0.dev0
|
210 |
+
triton==2.3.0
|
211 |
+
types-python-dateutil==2.9.0.20240316
|
212 |
+
typing_extensions==4.11.0
|
213 |
+
tzdata==2024.1
|
214 |
+
unicodecsv==0.14.1
|
215 |
+
uri-template==1.3.0
|
216 |
+
urllib3==2.2.1
|
217 |
+
wandb==0.17.0
|
218 |
+
wcwidth==0.2.13
|
219 |
+
webcolors==1.13
|
220 |
+
webencodings==0.5.1
|
221 |
+
websocket-client==1.8.0
|
222 |
+
wheel==0.43.0
|
223 |
+
wsproto==1.2.0
|
224 |
+
xxhash==3.4.1
|
225 |
+
yarl==1.9.4
|
wandb/run-20240513_204652-m0g0ap7d/files/wandb-metadata.json
ADDED
@@ -0,0 +1,706 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"os": "Linux-5.4.0-166-generic-x86_64-with-glibc2.31",
|
3 |
+
"python": "3.11.9",
|
4 |
+
"heartbeatAt": "2024-05-13T18:46:53.365083",
|
5 |
+
"startedAt": "2024-05-13T18:46:52.816759",
|
6 |
+
"docker": null,
|
7 |
+
"cuda": null,
|
8 |
+
"args": [
|
9 |
+
"finetuning_concatenated_config.json"
|
10 |
+
],
|
11 |
+
"state": "running",
|
12 |
+
"program": "/raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/run_parler_tts_training.py",
|
13 |
+
"codePathLocal": "run_parler_tts_training.py",
|
14 |
+
"codePath": "run_parler_tts_training.py",
|
15 |
+
"git": {
|
16 |
+
"remote": "https://huggingface.co/sanchit-gandhi/parler-tts-mini-v0.1-expresso-concatenated-combined",
|
17 |
+
"commit": "50ba4323d7b8bb052629aa1b88283b9df081a821"
|
18 |
+
},
|
19 |
+
"email": "[email protected]",
|
20 |
+
"root": "/raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined",
|
21 |
+
"host": "hf-dgx-01",
|
22 |
+
"username": "sanchit",
|
23 |
+
"executable": "/home/sanchit/miniconda3/envs/venv/bin/python",
|
24 |
+
"cpu_count": 64,
|
25 |
+
"cpu_count_logical": 128,
|
26 |
+
"cpu_freq": {
|
27 |
+
"current": 2257.736234375,
|
28 |
+
"min": 1500.0,
|
29 |
+
"max": 2250.0
|
30 |
+
},
|
31 |
+
"cpu_freq_per_core": [
|
32 |
+
{
|
33 |
+
"current": 1795.281,
|
34 |
+
"min": 1500.0,
|
35 |
+
"max": 2250.0
|
36 |
+
},
|
37 |
+
{
|
38 |
+
"current": 1794.292,
|
39 |
+
"min": 1500.0,
|
40 |
+
"max": 2250.0
|
41 |
+
},
|
42 |
+
{
|
43 |
+
"current": 1795.78,
|
44 |
+
"min": 1500.0,
|
45 |
+
"max": 2250.0
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"current": 1792.55,
|
49 |
+
"min": 1500.0,
|
50 |
+
"max": 2250.0
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"current": 1742.094,
|
54 |
+
"min": 1500.0,
|
55 |
+
"max": 2250.0
|
56 |
+
},
|
57 |
+
{
|
58 |
+
"current": 3026.54,
|
59 |
+
"min": 1500.0,
|
60 |
+
"max": 2250.0
|
61 |
+
},
|
62 |
+
{
|
63 |
+
"current": 1786.214,
|
64 |
+
"min": 1500.0,
|
65 |
+
"max": 2250.0
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"current": 1742.547,
|
69 |
+
"min": 1500.0,
|
70 |
+
"max": 2250.0
|
71 |
+
},
|
72 |
+
{
|
73 |
+
"current": 1728.916,
|
74 |
+
"min": 1500.0,
|
75 |
+
"max": 2250.0
|
76 |
+
},
|
77 |
+
{
|
78 |
+
"current": 1734.023,
|
79 |
+
"min": 1500.0,
|
80 |
+
"max": 2250.0
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"current": 3195.219,
|
84 |
+
"min": 1500.0,
|
85 |
+
"max": 2250.0
|
86 |
+
},
|
87 |
+
{
|
88 |
+
"current": 1733.722,
|
89 |
+
"min": 1500.0,
|
90 |
+
"max": 2250.0
|
91 |
+
},
|
92 |
+
{
|
93 |
+
"current": 3341.911,
|
94 |
+
"min": 1500.0,
|
95 |
+
"max": 2250.0
|
96 |
+
},
|
97 |
+
{
|
98 |
+
"current": 3325.608,
|
99 |
+
"min": 1500.0,
|
100 |
+
"max": 2250.0
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"current": 3233.258,
|
104 |
+
"min": 1500.0,
|
105 |
+
"max": 2250.0
|
106 |
+
},
|
107 |
+
{
|
108 |
+
"current": 1669.847,
|
109 |
+
"min": 1500.0,
|
110 |
+
"max": 2250.0
|
111 |
+
},
|
112 |
+
{
|
113 |
+
"current": 1791.948,
|
114 |
+
"min": 1500.0,
|
115 |
+
"max": 2250.0
|
116 |
+
},
|
117 |
+
{
|
118 |
+
"current": 1796.794,
|
119 |
+
"min": 1500.0,
|
120 |
+
"max": 2250.0
|
121 |
+
},
|
122 |
+
{
|
123 |
+
"current": 1791.49,
|
124 |
+
"min": 1500.0,
|
125 |
+
"max": 2250.0
|
126 |
+
},
|
127 |
+
{
|
128 |
+
"current": 1793.945,
|
129 |
+
"min": 1500.0,
|
130 |
+
"max": 2250.0
|
131 |
+
},
|
132 |
+
{
|
133 |
+
"current": 3342.943,
|
134 |
+
"min": 1500.0,
|
135 |
+
"max": 2250.0
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"current": 1669.791,
|
139 |
+
"min": 1500.0,
|
140 |
+
"max": 2250.0
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"current": 1669.593,
|
144 |
+
"min": 1500.0,
|
145 |
+
"max": 2250.0
|
146 |
+
},
|
147 |
+
{
|
148 |
+
"current": 1694.312,
|
149 |
+
"min": 1500.0,
|
150 |
+
"max": 2250.0
|
151 |
+
},
|
152 |
+
{
|
153 |
+
"current": 1873.727,
|
154 |
+
"min": 1500.0,
|
155 |
+
"max": 2250.0
|
156 |
+
},
|
157 |
+
{
|
158 |
+
"current": 1724.813,
|
159 |
+
"min": 1500.0,
|
160 |
+
"max": 2250.0
|
161 |
+
},
|
162 |
+
{
|
163 |
+
"current": 2354.471,
|
164 |
+
"min": 1500.0,
|
165 |
+
"max": 2250.0
|
166 |
+
},
|
167 |
+
{
|
168 |
+
"current": 1718.662,
|
169 |
+
"min": 1500.0,
|
170 |
+
"max": 2250.0
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"current": 1670.588,
|
174 |
+
"min": 1500.0,
|
175 |
+
"max": 2250.0
|
176 |
+
},
|
177 |
+
{
|
178 |
+
"current": 1665.577,
|
179 |
+
"min": 1500.0,
|
180 |
+
"max": 2250.0
|
181 |
+
},
|
182 |
+
{
|
183 |
+
"current": 1616.671,
|
184 |
+
"min": 1500.0,
|
185 |
+
"max": 2250.0
|
186 |
+
},
|
187 |
+
{
|
188 |
+
"current": 2080.81,
|
189 |
+
"min": 1500.0,
|
190 |
+
"max": 2250.0
|
191 |
+
},
|
192 |
+
{
|
193 |
+
"current": 1670.666,
|
194 |
+
"min": 1500.0,
|
195 |
+
"max": 2250.0
|
196 |
+
},
|
197 |
+
{
|
198 |
+
"current": 1652.559,
|
199 |
+
"min": 1500.0,
|
200 |
+
"max": 2250.0
|
201 |
+
},
|
202 |
+
{
|
203 |
+
"current": 3323.654,
|
204 |
+
"min": 1500.0,
|
205 |
+
"max": 2250.0
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"current": 1671.311,
|
209 |
+
"min": 1500.0,
|
210 |
+
"max": 2250.0
|
211 |
+
},
|
212 |
+
{
|
213 |
+
"current": 1726.286,
|
214 |
+
"min": 1500.0,
|
215 |
+
"max": 2250.0
|
216 |
+
},
|
217 |
+
{
|
218 |
+
"current": 1670.365,
|
219 |
+
"min": 1500.0,
|
220 |
+
"max": 2250.0
|
221 |
+
},
|
222 |
+
{
|
223 |
+
"current": 3320.57,
|
224 |
+
"min": 1500.0,
|
225 |
+
"max": 2250.0
|
226 |
+
},
|
227 |
+
{
|
228 |
+
"current": 1669.941,
|
229 |
+
"min": 1500.0,
|
230 |
+
"max": 2250.0
|
231 |
+
},
|
232 |
+
{
|
233 |
+
"current": 1791.021,
|
234 |
+
"min": 1500.0,
|
235 |
+
"max": 2250.0
|
236 |
+
},
|
237 |
+
{
|
238 |
+
"current": 1796.246,
|
239 |
+
"min": 1500.0,
|
240 |
+
"max": 2250.0
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"current": 1793.946,
|
244 |
+
"min": 1500.0,
|
245 |
+
"max": 2250.0
|
246 |
+
},
|
247 |
+
{
|
248 |
+
"current": 1794.848,
|
249 |
+
"min": 1500.0,
|
250 |
+
"max": 2250.0
|
251 |
+
},
|
252 |
+
{
|
253 |
+
"current": 3339.327,
|
254 |
+
"min": 1500.0,
|
255 |
+
"max": 2250.0
|
256 |
+
},
|
257 |
+
{
|
258 |
+
"current": 3344.315,
|
259 |
+
"min": 1500.0,
|
260 |
+
"max": 2250.0
|
261 |
+
},
|
262 |
+
{
|
263 |
+
"current": 3338.901,
|
264 |
+
"min": 1500.0,
|
265 |
+
"max": 2250.0
|
266 |
+
},
|
267 |
+
{
|
268 |
+
"current": 1668.541,
|
269 |
+
"min": 1500.0,
|
270 |
+
"max": 2250.0
|
271 |
+
},
|
272 |
+
{
|
273 |
+
"current": 1794.526,
|
274 |
+
"min": 1500.0,
|
275 |
+
"max": 2250.0
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"current": 1792.886,
|
279 |
+
"min": 1500.0,
|
280 |
+
"max": 2250.0
|
281 |
+
},
|
282 |
+
{
|
283 |
+
"current": 1796.844,
|
284 |
+
"min": 1500.0,
|
285 |
+
"max": 2250.0
|
286 |
+
},
|
287 |
+
{
|
288 |
+
"current": 1793.81,
|
289 |
+
"min": 1500.0,
|
290 |
+
"max": 2250.0
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"current": 1724.861,
|
294 |
+
"min": 1500.0,
|
295 |
+
"max": 2250.0
|
296 |
+
},
|
297 |
+
{
|
298 |
+
"current": 2294.458,
|
299 |
+
"min": 1500.0,
|
300 |
+
"max": 2250.0
|
301 |
+
},
|
302 |
+
{
|
303 |
+
"current": 1720.835,
|
304 |
+
"min": 1500.0,
|
305 |
+
"max": 2250.0
|
306 |
+
},
|
307 |
+
{
|
308 |
+
"current": 1720.155,
|
309 |
+
"min": 1500.0,
|
310 |
+
"max": 2250.0
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"current": 1668.96,
|
314 |
+
"min": 1500.0,
|
315 |
+
"max": 2250.0
|
316 |
+
},
|
317 |
+
{
|
318 |
+
"current": 1976.5,
|
319 |
+
"min": 1500.0,
|
320 |
+
"max": 2250.0
|
321 |
+
},
|
322 |
+
{
|
323 |
+
"current": 2241.578,
|
324 |
+
"min": 1500.0,
|
325 |
+
"max": 2250.0
|
326 |
+
},
|
327 |
+
{
|
328 |
+
"current": 1671.964,
|
329 |
+
"min": 1500.0,
|
330 |
+
"max": 2250.0
|
331 |
+
},
|
332 |
+
{
|
333 |
+
"current": 3319.623,
|
334 |
+
"min": 1500.0,
|
335 |
+
"max": 2250.0
|
336 |
+
},
|
337 |
+
{
|
338 |
+
"current": 1670.777,
|
339 |
+
"min": 1500.0,
|
340 |
+
"max": 2250.0
|
341 |
+
},
|
342 |
+
{
|
343 |
+
"current": 1670.389,
|
344 |
+
"min": 1500.0,
|
345 |
+
"max": 2250.0
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"current": 1669.629,
|
349 |
+
"min": 1500.0,
|
350 |
+
"max": 2250.0
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"current": 1794.19,
|
354 |
+
"min": 1500.0,
|
355 |
+
"max": 2250.0
|
356 |
+
},
|
357 |
+
{
|
358 |
+
"current": 1794.138,
|
359 |
+
"min": 1500.0,
|
360 |
+
"max": 2250.0
|
361 |
+
},
|
362 |
+
{
|
363 |
+
"current": 1796.317,
|
364 |
+
"min": 1500.0,
|
365 |
+
"max": 2250.0
|
366 |
+
},
|
367 |
+
{
|
368 |
+
"current": 1792.821,
|
369 |
+
"min": 1500.0,
|
370 |
+
"max": 2250.0
|
371 |
+
},
|
372 |
+
{
|
373 |
+
"current": 1794.716,
|
374 |
+
"min": 1500.0,
|
375 |
+
"max": 2250.0
|
376 |
+
},
|
377 |
+
{
|
378 |
+
"current": 1793.624,
|
379 |
+
"min": 1500.0,
|
380 |
+
"max": 2250.0
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"current": 1796.346,
|
384 |
+
"min": 1500.0,
|
385 |
+
"max": 2250.0
|
386 |
+
},
|
387 |
+
{
|
388 |
+
"current": 1793.897,
|
389 |
+
"min": 1500.0,
|
390 |
+
"max": 2250.0
|
391 |
+
},
|
392 |
+
{
|
393 |
+
"current": 1735.424,
|
394 |
+
"min": 1500.0,
|
395 |
+
"max": 2250.0
|
396 |
+
},
|
397 |
+
{
|
398 |
+
"current": 1738.64,
|
399 |
+
"min": 1500.0,
|
400 |
+
"max": 2250.0
|
401 |
+
},
|
402 |
+
{
|
403 |
+
"current": 1979.998,
|
404 |
+
"min": 1500.0,
|
405 |
+
"max": 2250.0
|
406 |
+
},
|
407 |
+
{
|
408 |
+
"current": 1737.286,
|
409 |
+
"min": 1500.0,
|
410 |
+
"max": 2250.0
|
411 |
+
},
|
412 |
+
{
|
413 |
+
"current": 3313.748,
|
414 |
+
"min": 1500.0,
|
415 |
+
"max": 2250.0
|
416 |
+
},
|
417 |
+
{
|
418 |
+
"current": 3337.223,
|
419 |
+
"min": 1500.0,
|
420 |
+
"max": 2250.0
|
421 |
+
},
|
422 |
+
{
|
423 |
+
"current": 1671.416,
|
424 |
+
"min": 1500.0,
|
425 |
+
"max": 2250.0
|
426 |
+
},
|
427 |
+
{
|
428 |
+
"current": 1670.005,
|
429 |
+
"min": 1500.0,
|
430 |
+
"max": 2250.0
|
431 |
+
},
|
432 |
+
{
|
433 |
+
"current": 1794.276,
|
434 |
+
"min": 1500.0,
|
435 |
+
"max": 2250.0
|
436 |
+
},
|
437 |
+
{
|
438 |
+
"current": 1738.22,
|
439 |
+
"min": 1500.0,
|
440 |
+
"max": 2250.0
|
441 |
+
},
|
442 |
+
{
|
443 |
+
"current": 1742.737,
|
444 |
+
"min": 1500.0,
|
445 |
+
"max": 2250.0
|
446 |
+
},
|
447 |
+
{
|
448 |
+
"current": 1770.535,
|
449 |
+
"min": 1500.0,
|
450 |
+
"max": 2250.0
|
451 |
+
},
|
452 |
+
{
|
453 |
+
"current": 3320.252,
|
454 |
+
"min": 1500.0,
|
455 |
+
"max": 2250.0
|
456 |
+
},
|
457 |
+
{
|
458 |
+
"current": 1671.037,
|
459 |
+
"min": 1500.0,
|
460 |
+
"max": 2250.0
|
461 |
+
},
|
462 |
+
{
|
463 |
+
"current": 1669.549,
|
464 |
+
"min": 1500.0,
|
465 |
+
"max": 2250.0
|
466 |
+
},
|
467 |
+
{
|
468 |
+
"current": 1670.948,
|
469 |
+
"min": 1500.0,
|
470 |
+
"max": 2250.0
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"current": 2843.391,
|
474 |
+
"min": 1500.0,
|
475 |
+
"max": 2250.0
|
476 |
+
},
|
477 |
+
{
|
478 |
+
"current": 2348.589,
|
479 |
+
"min": 1500.0,
|
480 |
+
"max": 2250.0
|
481 |
+
},
|
482 |
+
{
|
483 |
+
"current": 3287.915,
|
484 |
+
"min": 1500.0,
|
485 |
+
"max": 2250.0
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"current": 2340.192,
|
489 |
+
"min": 1500.0,
|
490 |
+
"max": 2250.0
|
491 |
+
},
|
492 |
+
{
|
493 |
+
"current": 2426.358,
|
494 |
+
"min": 1500.0,
|
495 |
+
"max": 2250.0
|
496 |
+
},
|
497 |
+
{
|
498 |
+
"current": 2415.833,
|
499 |
+
"min": 1500.0,
|
500 |
+
"max": 2250.0
|
501 |
+
},
|
502 |
+
{
|
503 |
+
"current": 2419.416,
|
504 |
+
"min": 1500.0,
|
505 |
+
"max": 2250.0
|
506 |
+
},
|
507 |
+
{
|
508 |
+
"current": 2277.433,
|
509 |
+
"min": 1500.0,
|
510 |
+
"max": 2250.0
|
511 |
+
},
|
512 |
+
{
|
513 |
+
"current": 2365.562,
|
514 |
+
"min": 1500.0,
|
515 |
+
"max": 2250.0
|
516 |
+
},
|
517 |
+
{
|
518 |
+
"current": 2400.6,
|
519 |
+
"min": 1500.0,
|
520 |
+
"max": 2250.0
|
521 |
+
},
|
522 |
+
{
|
523 |
+
"current": 2075.143,
|
524 |
+
"min": 1500.0,
|
525 |
+
"max": 2250.0
|
526 |
+
},
|
527 |
+
{
|
528 |
+
"current": 2382.295,
|
529 |
+
"min": 1500.0,
|
530 |
+
"max": 2250.0
|
531 |
+
},
|
532 |
+
{
|
533 |
+
"current": 3066.339,
|
534 |
+
"min": 1500.0,
|
535 |
+
"max": 2250.0
|
536 |
+
},
|
537 |
+
{
|
538 |
+
"current": 2466.631,
|
539 |
+
"min": 1500.0,
|
540 |
+
"max": 2250.0
|
541 |
+
},
|
542 |
+
{
|
543 |
+
"current": 3100.81,
|
544 |
+
"min": 1500.0,
|
545 |
+
"max": 2250.0
|
546 |
+
},
|
547 |
+
{
|
548 |
+
"current": 2421.93,
|
549 |
+
"min": 1500.0,
|
550 |
+
"max": 2250.0
|
551 |
+
},
|
552 |
+
{
|
553 |
+
"current": 3233.829,
|
554 |
+
"min": 1500.0,
|
555 |
+
"max": 2250.0
|
556 |
+
},
|
557 |
+
{
|
558 |
+
"current": 2234.583,
|
559 |
+
"min": 1500.0,
|
560 |
+
"max": 2250.0
|
561 |
+
},
|
562 |
+
{
|
563 |
+
"current": 2452.089,
|
564 |
+
"min": 1500.0,
|
565 |
+
"max": 2250.0
|
566 |
+
},
|
567 |
+
{
|
568 |
+
"current": 2975.985,
|
569 |
+
"min": 1500.0,
|
570 |
+
"max": 2250.0
|
571 |
+
},
|
572 |
+
{
|
573 |
+
"current": 3301.512,
|
574 |
+
"min": 1500.0,
|
575 |
+
"max": 2250.0
|
576 |
+
},
|
577 |
+
{
|
578 |
+
"current": 3336.905,
|
579 |
+
"min": 1500.0,
|
580 |
+
"max": 2250.0
|
581 |
+
},
|
582 |
+
{
|
583 |
+
"current": 2984.87,
|
584 |
+
"min": 1500.0,
|
585 |
+
"max": 2250.0
|
586 |
+
},
|
587 |
+
{
|
588 |
+
"current": 2384.306,
|
589 |
+
"min": 1500.0,
|
590 |
+
"max": 2250.0
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"current": 2965.197,
|
594 |
+
"min": 1500.0,
|
595 |
+
"max": 2250.0
|
596 |
+
},
|
597 |
+
{
|
598 |
+
"current": 1929.067,
|
599 |
+
"min": 1500.0,
|
600 |
+
"max": 2250.0
|
601 |
+
},
|
602 |
+
{
|
603 |
+
"current": 1986.731,
|
604 |
+
"min": 1500.0,
|
605 |
+
"max": 2250.0
|
606 |
+
},
|
607 |
+
{
|
608 |
+
"current": 1999.412,
|
609 |
+
"min": 1500.0,
|
610 |
+
"max": 2250.0
|
611 |
+
},
|
612 |
+
{
|
613 |
+
"current": 2477.541,
|
614 |
+
"min": 1500.0,
|
615 |
+
"max": 2250.0
|
616 |
+
},
|
617 |
+
{
|
618 |
+
"current": 3111.851,
|
619 |
+
"min": 1500.0,
|
620 |
+
"max": 2250.0
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"current": 2009.907,
|
624 |
+
"min": 1500.0,
|
625 |
+
"max": 2250.0
|
626 |
+
},
|
627 |
+
{
|
628 |
+
"current": 1993.784,
|
629 |
+
"min": 1500.0,
|
630 |
+
"max": 2250.0
|
631 |
+
},
|
632 |
+
{
|
633 |
+
"current": 2144.459,
|
634 |
+
"min": 1500.0,
|
635 |
+
"max": 2250.0
|
636 |
+
},
|
637 |
+
{
|
638 |
+
"current": 3337.426,
|
639 |
+
"min": 1500.0,
|
640 |
+
"max": 2250.0
|
641 |
+
},
|
642 |
+
{
|
643 |
+
"current": 3320.114,
|
644 |
+
"min": 1500.0,
|
645 |
+
"max": 2250.0
|
646 |
+
},
|
647 |
+
{
|
648 |
+
"current": 2169.719,
|
649 |
+
"min": 1500.0,
|
650 |
+
"max": 2250.0
|
651 |
+
},
|
652 |
+
{
|
653 |
+
"current": 3308.644,
|
654 |
+
"min": 1500.0,
|
655 |
+
"max": 2250.0
|
656 |
+
},
|
657 |
+
{
|
658 |
+
"current": 2111.633,
|
659 |
+
"min": 1500.0,
|
660 |
+
"max": 2250.0
|
661 |
+
},
|
662 |
+
{
|
663 |
+
"current": 2123.71,
|
664 |
+
"min": 1500.0,
|
665 |
+
"max": 2250.0
|
666 |
+
},
|
667 |
+
{
|
668 |
+
"current": 2153.49,
|
669 |
+
"min": 1500.0,
|
670 |
+
"max": 2250.0
|
671 |
+
}
|
672 |
+
],
|
673 |
+
"disk": {
|
674 |
+
"/": {
|
675 |
+
"total": 1757.8785285949707,
|
676 |
+
"used": 1663.5005989074707
|
677 |
+
}
|
678 |
+
},
|
679 |
+
"gpu": "NVIDIA A100-SXM4-80GB",
|
680 |
+
"gpu_count": 5,
|
681 |
+
"gpu_devices": [
|
682 |
+
{
|
683 |
+
"name": "NVIDIA A100-SXM4-80GB",
|
684 |
+
"memory_total": 85899345920
|
685 |
+
},
|
686 |
+
{
|
687 |
+
"name": "NVIDIA A100-SXM4-80GB",
|
688 |
+
"memory_total": 85899345920
|
689 |
+
},
|
690 |
+
{
|
691 |
+
"name": "NVIDIA A100-SXM4-80GB",
|
692 |
+
"memory_total": 85899345920
|
693 |
+
},
|
694 |
+
{
|
695 |
+
"name": "NVIDIA DGX Display",
|
696 |
+
"memory_total": 4294967296
|
697 |
+
},
|
698 |
+
{
|
699 |
+
"name": "NVIDIA A100-SXM4-80GB",
|
700 |
+
"memory_total": 85899345920
|
701 |
+
}
|
702 |
+
],
|
703 |
+
"memory": {
|
704 |
+
"total": 503.5396919250488
|
705 |
+
}
|
706 |
+
}
|
wandb/run-20240513_204652-m0g0ap7d/files/wandb-summary.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"_wandb": {"runtime": 219}}
|
wandb/run-20240513_204652-m0g0ap7d/logs/debug-internal.log
ADDED
@@ -0,0 +1,455 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
2024-05-13 20:46:52,823 INFO StreamThr :1244775 [internal.py:wandb_internal():85] W&B internal server running at pid: 1244775, started at: 2024-05-13 20:46:52.823043
|
2 |
+
2024-05-13 20:46:52,825 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status
|
3 |
+
2024-05-13 20:46:52,826 INFO WriterThread:1244775 [datastore.py:open_for_write():87] open: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/run-m0g0ap7d.wandb
|
4 |
+
2024-05-13 20:46:52,828 DEBUG SenderThread:1244775 [sender.py:send():378] send: header
|
5 |
+
2024-05-13 20:46:52,829 DEBUG SenderThread:1244775 [sender.py:send():378] send: run
|
6 |
+
2024-05-13 20:46:53,223 INFO SenderThread:1244775 [dir_watcher.py:__init__():211] watching files in: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files
|
7 |
+
2024-05-13 20:46:53,223 INFO SenderThread:1244775 [sender.py:_start_run_threads():1123] run started: m0g0ap7d with start time 1715626012.822173
|
8 |
+
2024-05-13 20:46:53,229 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: check_version
|
9 |
+
2024-05-13 20:46:53,229 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: check_version
|
10 |
+
2024-05-13 20:46:53,293 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: run_start
|
11 |
+
2024-05-13 20:46:53,323 DEBUG HandlerThread:1244775 [system_info.py:__init__():26] System info init
|
12 |
+
2024-05-13 20:46:53,323 DEBUG HandlerThread:1244775 [system_info.py:__init__():41] System info init done
|
13 |
+
2024-05-13 20:46:53,323 INFO HandlerThread:1244775 [system_monitor.py:start():194] Starting system monitor
|
14 |
+
2024-05-13 20:46:53,323 INFO SystemMonitor:1244775 [system_monitor.py:_start():158] Starting system asset monitoring threads
|
15 |
+
2024-05-13 20:46:53,324 INFO HandlerThread:1244775 [system_monitor.py:probe():214] Collecting system info
|
16 |
+
2024-05-13 20:46:53,324 INFO SystemMonitor:1244775 [interfaces.py:start():188] Started cpu monitoring
|
17 |
+
2024-05-13 20:46:53,325 INFO SystemMonitor:1244775 [interfaces.py:start():188] Started disk monitoring
|
18 |
+
2024-05-13 20:46:53,325 INFO SystemMonitor:1244775 [interfaces.py:start():188] Started gpu monitoring
|
19 |
+
2024-05-13 20:46:53,327 INFO SystemMonitor:1244775 [interfaces.py:start():188] Started memory monitoring
|
20 |
+
2024-05-13 20:46:53,329 INFO SystemMonitor:1244775 [interfaces.py:start():188] Started network monitoring
|
21 |
+
2024-05-13 20:46:53,365 DEBUG HandlerThread:1244775 [system_info.py:probe():150] Probing system
|
22 |
+
2024-05-13 20:46:53,366 DEBUG HandlerThread:1244775 [system_info.py:_probe_git():135] Probing git
|
23 |
+
2024-05-13 20:46:53,372 DEBUG HandlerThread:1244775 [system_info.py:_probe_git():143] Probing git done
|
24 |
+
2024-05-13 20:46:53,372 DEBUG HandlerThread:1244775 [system_info.py:probe():198] Probing system done
|
25 |
+
2024-05-13 20:46:53,372 DEBUG HandlerThread:1244775 [system_monitor.py:probe():223] {'os': 'Linux-5.4.0-166-generic-x86_64-with-glibc2.31', 'python': '3.11.9', 'heartbeatAt': '2024-05-13T18:46:53.365083', 'startedAt': '2024-05-13T18:46:52.816759', 'docker': None, 'cuda': None, 'args': ('finetuning_concatenated_config.json',), 'state': 'running', 'program': '/raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/run_parler_tts_training.py', 'codePathLocal': 'run_parler_tts_training.py', 'codePath': 'run_parler_tts_training.py', 'git': {'remote': 'https://huggingface.co/sanchit-gandhi/parler-tts-mini-v0.1-expresso-concatenated-combined', 'commit': '50ba4323d7b8bb052629aa1b88283b9df081a821'}, 'email': '[email protected]', 'root': '/raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined', 'host': 'hf-dgx-01', 'username': 'sanchit', 'executable': '/home/sanchit/miniconda3/envs/venv/bin/python', 'cpu_count': 64, 'cpu_count_logical': 128, 'cpu_freq': {'current': 2257.736234375, 'min': 1500.0, 'max': 2250.0}, 'cpu_freq_per_core': [{'current': 1795.281, 'min': 1500.0, 'max': 2250.0}, {'current': 1794.292, 'min': 1500.0, 'max': 2250.0}, {'current': 1795.78, 'min': 1500.0, 'max': 2250.0}, {'current': 1792.55, 'min': 1500.0, 'max': 2250.0}, {'current': 1742.094, 'min': 1500.0, 'max': 2250.0}, {'current': 3026.54, 'min': 1500.0, 'max': 2250.0}, {'current': 1786.214, 'min': 1500.0, 'max': 2250.0}, {'current': 1742.547, 'min': 1500.0, 'max': 2250.0}, {'current': 1728.916, 'min': 1500.0, 'max': 2250.0}, {'current': 1734.023, 'min': 1500.0, 'max': 2250.0}, {'current': 3195.219, 'min': 1500.0, 'max': 2250.0}, {'current': 1733.722, 'min': 1500.0, 'max': 2250.0}, {'current': 3341.911, 'min': 1500.0, 'max': 2250.0}, {'current': 3325.608, 'min': 1500.0, 'max': 2250.0}, {'current': 3233.258, 'min': 1500.0, 'max': 2250.0}, {'current': 1669.847, 'min': 1500.0, 'max': 2250.0}, {'current': 1791.948, 'min': 1500.0, 'max': 2250.0}, {'current': 1796.794, 'min': 1500.0, 'max': 2250.0}, {'current': 1791.49, 'min': 1500.0, 'max': 2250.0}, {'current': 1793.945, 'min': 1500.0, 'max': 2250.0}, {'current': 3342.943, 'min': 1500.0, 'max': 2250.0}, {'current': 1669.791, 'min': 1500.0, 'max': 2250.0}, {'current': 1669.593, 'min': 1500.0, 'max': 2250.0}, {'current': 1694.312, 'min': 1500.0, 'max': 2250.0}, {'current': 1873.727, 'min': 1500.0, 'max': 2250.0}, {'current': 1724.813, 'min': 1500.0, 'max': 2250.0}, {'current': 2354.471, 'min': 1500.0, 'max': 2250.0}, {'current': 1718.662, 'min': 1500.0, 'max': 2250.0}, {'current': 1670.588, 'min': 1500.0, 'max': 2250.0}, {'current': 1665.577, 'min': 1500.0, 'max': 2250.0}, {'current': 1616.671, 'min': 1500.0, 'max': 2250.0}, {'current': 2080.81, 'min': 1500.0, 'max': 2250.0}, {'current': 1670.666, 'min': 1500.0, 'max': 2250.0}, {'current': 1652.559, 'min': 1500.0, 'max': 2250.0}, {'current': 3323.654, 'min': 1500.0, 'max': 2250.0}, {'current': 1671.311, 'min': 1500.0, 'max': 2250.0}, {'current': 1726.286, 'min': 1500.0, 'max': 2250.0}, {'current': 1670.365, 'min': 1500.0, 'max': 2250.0}, {'current': 3320.57, 'min': 1500.0, 'max': 2250.0}, {'current': 1669.941, 'min': 1500.0, 'max': 2250.0}, {'current': 1791.021, 'min': 1500.0, 'max': 2250.0}, {'current': 1796.246, 'min': 1500.0, 'max': 2250.0}, {'current': 1793.946, 'min': 1500.0, 'max': 2250.0}, {'current': 1794.848, 'min': 1500.0, 'max': 2250.0}, {'current': 3339.327, 'min': 1500.0, 'max': 2250.0}, {'current': 3344.315, 'min': 1500.0, 'max': 2250.0}, {'current': 3338.901, 'min': 1500.0, 'max': 2250.0}, {'current': 1668.541, 'min': 1500.0, 'max': 2250.0}, {'current': 1794.526, 'min': 1500.0, 'max': 2250.0}, {'current': 1792.886, 'min': 1500.0, 'max': 2250.0}, {'current': 1796.844, 'min': 1500.0, 'max': 2250.0}, {'current': 1793.81, 'min': 1500.0, 'max': 2250.0}, {'current': 1724.861, 'min': 1500.0, 'max': 2250.0}, {'current': 2294.458, 'min': 1500.0, 'max': 2250.0}, {'current': 1720.835, 'min': 1500.0, 'max': 2250.0}, {'current': 1720.155, 'min': 1500.0, 'max': 2250.0}, {'current': 1668.96, 'min': 1500.0, 'max': 2250.0}, {'current': 1976.5, 'min': 1500.0, 'max': 2250.0}, {'current': 2241.578, 'min': 1500.0, 'max': 2250.0}, {'current': 1671.964, 'min': 1500.0, 'max': 2250.0}, {'current': 3319.623, 'min': 1500.0, 'max': 2250.0}, {'current': 1670.777, 'min': 1500.0, 'max': 2250.0}, {'current': 1670.389, 'min': 1500.0, 'max': 2250.0}, {'current': 1669.629, 'min': 1500.0, 'max': 2250.0}, {'current': 1794.19, 'min': 1500.0, 'max': 2250.0}, {'current': 1794.138, 'min': 1500.0, 'max': 2250.0}, {'current': 1796.317, 'min': 1500.0, 'max': 2250.0}, {'current': 1792.821, 'min': 1500.0, 'max': 2250.0}, {'current': 1794.716, 'min': 1500.0, 'max': 2250.0}, {'current': 1793.624, 'min': 1500.0, 'max': 2250.0}, {'current': 1796.346, 'min': 1500.0, 'max': 2250.0}, {'current': 1793.897, 'min': 1500.0, 'max': 2250.0}, {'current': 1735.424, 'min': 1500.0, 'max': 2250.0}, {'current': 1738.64, 'min': 1500.0, 'max': 2250.0}, {'current': 1979.998, 'min': 1500.0, 'max': 2250.0}, {'current': 1737.286, 'min': 1500.0, 'max': 2250.0}, {'current': 3313.748, 'min': 1500.0, 'max': 2250.0}, {'current': 3337.223, 'min': 1500.0, 'max': 2250.0}, {'current': 1671.416, 'min': 1500.0, 'max': 2250.0}, {'current': 1670.005, 'min': 1500.0, 'max': 2250.0}, {'current': 1794.276, 'min': 1500.0, 'max': 2250.0}, {'current': 1738.22, 'min': 1500.0, 'max': 2250.0}, {'current': 1742.737, 'min': 1500.0, 'max': 2250.0}, {'current': 1770.535, 'min': 1500.0, 'max': 2250.0}, {'current': 3320.252, 'min': 1500.0, 'max': 2250.0}, {'current': 1671.037, 'min': 1500.0, 'max': 2250.0}, {'current': 1669.549, 'min': 1500.0, 'max': 2250.0}, {'current': 1670.948, 'min': 1500.0, 'max': 2250.0}, {'current': 2843.391, 'min': 1500.0, 'max': 2250.0}, {'current': 2348.589, 'min': 1500.0, 'max': 2250.0}, {'current': 3287.915, 'min': 1500.0, 'max': 2250.0}, {'current': 2340.192, 'min': 1500.0, 'max': 2250.0}, {'current': 2426.358, 'min': 1500.0, 'max': 2250.0}, {'current': 2415.833, 'min': 1500.0, 'max': 2250.0}, {'current': 2419.416, 'min': 1500.0, 'max': 2250.0}, {'current': 2277.433, 'min': 1500.0, 'max': 2250.0}, {'current': 2365.562, 'min': 1500.0, 'max': 2250.0}, {'current': 2400.6, 'min': 1500.0, 'max': 2250.0}, {'current': 2075.143, 'min': 1500.0, 'max': 2250.0}, {'current': 2382.295, 'min': 1500.0, 'max': 2250.0}, {'current': 3066.339, 'min': 1500.0, 'max': 2250.0}, {'current': 2466.631, 'min': 1500.0, 'max': 2250.0}, {'current': 3100.81, 'min': 1500.0, 'max': 2250.0}, {'current': 2421.93, 'min': 1500.0, 'max': 2250.0}, {'current': 3233.829, 'min': 1500.0, 'max': 2250.0}, {'current': 2234.583, 'min': 1500.0, 'max': 2250.0}, {'current': 2452.089, 'min': 1500.0, 'max': 2250.0}, {'current': 2975.985, 'min': 1500.0, 'max': 2250.0}, {'current': 3301.512, 'min': 1500.0, 'max': 2250.0}, {'current': 3336.905, 'min': 1500.0, 'max': 2250.0}, {'current': 2984.87, 'min': 1500.0, 'max': 2250.0}, {'current': 2384.306, 'min': 1500.0, 'max': 2250.0}, {'current': 2965.197, 'min': 1500.0, 'max': 2250.0}, {'current': 1929.067, 'min': 1500.0, 'max': 2250.0}, {'current': 1986.731, 'min': 1500.0, 'max': 2250.0}, {'current': 1999.412, 'min': 1500.0, 'max': 2250.0}, {'current': 2477.541, 'min': 1500.0, 'max': 2250.0}, {'current': 3111.851, 'min': 1500.0, 'max': 2250.0}, {'current': 2009.907, 'min': 1500.0, 'max': 2250.0}, {'current': 1993.784, 'min': 1500.0, 'max': 2250.0}, {'current': 2144.459, 'min': 1500.0, 'max': 2250.0}, {'current': 3337.426, 'min': 1500.0, 'max': 2250.0}, {'current': 3320.114, 'min': 1500.0, 'max': 2250.0}, {'current': 2169.719, 'min': 1500.0, 'max': 2250.0}, {'current': 3308.644, 'min': 1500.0, 'max': 2250.0}, {'current': 2111.633, 'min': 1500.0, 'max': 2250.0}, {'current': 2123.71, 'min': 1500.0, 'max': 2250.0}, {'current': 2153.49, 'min': 1500.0, 'max': 2250.0}], 'disk': {'/': {'total': 1757.8785285949707, 'used': 1663.5005989074707}}, 'gpu': 'NVIDIA A100-SXM4-80GB', 'gpu_count': 5, 'gpu_devices': [{'name': 'NVIDIA A100-SXM4-80GB', 'memory_total': 85899345920}, {'name': 'NVIDIA A100-SXM4-80GB', 'memory_total': 85899345920}, {'name': 'NVIDIA A100-SXM4-80GB', 'memory_total': 85899345920}, {'name': 'NVIDIA DGX Display', 'memory_total': 4294967296}, {'name': 'NVIDIA A100-SXM4-80GB', 'memory_total': 85899345920}], 'memory': {'total': 503.5396919250488}}
|
26 |
+
2024-05-13 20:46:53,372 INFO HandlerThread:1244775 [system_monitor.py:probe():224] Finished collecting system info
|
27 |
+
2024-05-13 20:46:53,372 INFO HandlerThread:1244775 [system_monitor.py:probe():227] Publishing system info
|
28 |
+
2024-05-13 20:46:53,372 DEBUG HandlerThread:1244775 [system_info.py:_save_conda():207] Saving list of conda packages installed into the current environment
|
29 |
+
2024-05-13 20:46:53,387 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
30 |
+
2024-05-13 20:46:53,400 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
31 |
+
2024-05-13 20:46:54,224 INFO Thread-12 :1244775 [dir_watcher.py:_on_file_created():271] file/dir created: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/conda-environment.yaml
|
32 |
+
2024-05-13 20:46:55,418 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
33 |
+
2024-05-13 20:46:55,429 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
34 |
+
2024-05-13 20:46:55,741 DEBUG HandlerThread:1244775 [system_info.py:_save_conda():222] Saving conda packages done
|
35 |
+
2024-05-13 20:46:55,742 INFO HandlerThread:1244775 [system_monitor.py:probe():229] Finished publishing system info
|
36 |
+
2024-05-13 20:46:55,750 DEBUG SenderThread:1244775 [sender.py:send():378] send: files
|
37 |
+
2024-05-13 20:46:55,750 INFO SenderThread:1244775 [sender.py:_save_file():1389] saving file wandb-metadata.json with policy now
|
38 |
+
2024-05-13 20:46:55,863 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: python_packages
|
39 |
+
2024-05-13 20:46:55,863 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
40 |
+
2024-05-13 20:46:55,863 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: python_packages
|
41 |
+
2024-05-13 20:46:55,865 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
42 |
+
2024-05-13 20:46:56,093 DEBUG SenderThread:1244775 [sender.py:send():378] send: telemetry
|
43 |
+
2024-05-13 20:46:56,093 DEBUG SenderThread:1244775 [sender.py:send():378] send: config
|
44 |
+
2024-05-13 20:46:56,224 INFO Thread-12 :1244775 [dir_watcher.py:_on_file_modified():288] file/dir modified: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/conda-environment.yaml
|
45 |
+
2024-05-13 20:46:56,224 INFO Thread-12 :1244775 [dir_watcher.py:_on_file_created():271] file/dir created: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/wandb-metadata.json
|
46 |
+
2024-05-13 20:46:56,224 INFO Thread-12 :1244775 [dir_watcher.py:_on_file_created():271] file/dir created: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/requirements.txt
|
47 |
+
2024-05-13 20:46:56,224 INFO Thread-12 :1244775 [dir_watcher.py:_on_file_created():271] file/dir created: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/output.log
|
48 |
+
2024-05-13 20:46:56,261 INFO wandb-upload_0:1244775 [upload_job.py:push():130] Uploaded file /tmp/tmpewnm1an9wandb/l0duo91p-wandb-metadata.json
|
49 |
+
2024-05-13 20:46:58,225 INFO Thread-12 :1244775 [dir_watcher.py:_on_file_modified():288] file/dir modified: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/output.log
|
50 |
+
2024-05-13 20:46:58,331 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
51 |
+
2024-05-13 20:46:58,343 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
52 |
+
2024-05-13 20:46:58,386 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
53 |
+
2024-05-13 20:47:00,362 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
54 |
+
2024-05-13 20:47:00,375 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
55 |
+
2024-05-13 20:47:03,387 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
56 |
+
2024-05-13 20:47:03,431 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
57 |
+
2024-05-13 20:47:03,443 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
58 |
+
2024-05-13 20:47:05,466 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
59 |
+
2024-05-13 20:47:05,478 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
60 |
+
2024-05-13 20:47:08,388 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
61 |
+
2024-05-13 20:47:08,679 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
62 |
+
2024-05-13 20:47:08,692 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
63 |
+
2024-05-13 20:47:10,713 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
64 |
+
2024-05-13 20:47:10,724 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
65 |
+
2024-05-13 20:47:10,863 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
66 |
+
2024-05-13 20:47:10,863 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
67 |
+
2024-05-13 20:47:12,746 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
68 |
+
2024-05-13 20:47:12,757 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
69 |
+
2024-05-13 20:47:14,094 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
70 |
+
2024-05-13 20:47:15,627 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
71 |
+
2024-05-13 20:47:15,637 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
72 |
+
2024-05-13 20:47:17,658 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
73 |
+
2024-05-13 20:47:17,668 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
74 |
+
2024-05-13 20:47:19,096 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
75 |
+
2024-05-13 20:47:20,779 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
76 |
+
2024-05-13 20:47:20,799 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
77 |
+
2024-05-13 20:47:22,817 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
78 |
+
2024-05-13 20:47:22,830 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
79 |
+
2024-05-13 20:47:24,099 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
80 |
+
2024-05-13 20:47:24,858 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
81 |
+
2024-05-13 20:47:24,870 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
82 |
+
2024-05-13 20:47:25,233 INFO Thread-12 :1244775 [dir_watcher.py:_on_file_modified():288] file/dir modified: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/config.yaml
|
83 |
+
2024-05-13 20:47:25,863 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
84 |
+
2024-05-13 20:47:25,863 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
85 |
+
2024-05-13 20:47:27,736 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
86 |
+
2024-05-13 20:47:27,747 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
87 |
+
2024-05-13 20:47:30,092 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
88 |
+
2024-05-13 20:47:31,414 ERROR gpu :1244775 [interfaces.py:aggregate():159] Failed to serialize metric: division by zero
|
89 |
+
2024-05-13 20:47:31,434 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
90 |
+
2024-05-13 20:47:31,462 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
91 |
+
2024-05-13 20:47:34,494 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
92 |
+
2024-05-13 20:47:34,518 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
93 |
+
2024-05-13 20:47:35,093 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
94 |
+
2024-05-13 20:47:36,569 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
95 |
+
2024-05-13 20:47:36,599 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
96 |
+
2024-05-13 20:47:38,635 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
97 |
+
2024-05-13 20:47:38,658 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
98 |
+
2024-05-13 20:47:40,093 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
99 |
+
2024-05-13 20:47:40,863 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
100 |
+
2024-05-13 20:47:40,864 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
101 |
+
2024-05-13 20:47:41,560 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
102 |
+
2024-05-13 20:47:41,584 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
103 |
+
2024-05-13 20:47:43,631 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
104 |
+
2024-05-13 20:47:43,652 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
105 |
+
2024-05-13 20:47:46,084 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
106 |
+
2024-05-13 20:47:46,570 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
107 |
+
2024-05-13 20:47:46,604 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
108 |
+
2024-05-13 20:47:48,647 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
109 |
+
2024-05-13 20:47:48,659 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
110 |
+
2024-05-13 20:47:51,084 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
111 |
+
2024-05-13 20:47:51,664 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
112 |
+
2024-05-13 20:47:51,686 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
113 |
+
2024-05-13 20:47:53,329 DEBUG SystemMonitor:1244775 [system_monitor.py:_start():172] Starting system metrics aggregation loop
|
114 |
+
2024-05-13 20:47:53,333 DEBUG SenderThread:1244775 [sender.py:send():378] send: stats
|
115 |
+
2024-05-13 20:47:53,709 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
116 |
+
2024-05-13 20:47:53,724 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
117 |
+
2024-05-13 20:47:55,863 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
118 |
+
2024-05-13 20:47:55,864 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
119 |
+
2024-05-13 20:47:56,593 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
120 |
+
2024-05-13 20:47:56,607 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
121 |
+
2024-05-13 20:47:57,080 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
122 |
+
2024-05-13 20:47:58,627 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
123 |
+
2024-05-13 20:47:58,641 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
124 |
+
2024-05-13 20:48:01,653 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
125 |
+
2024-05-13 20:48:01,664 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
126 |
+
2024-05-13 20:48:02,081 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
127 |
+
2024-05-13 20:48:03,684 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
128 |
+
2024-05-13 20:48:03,696 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
129 |
+
2024-05-13 20:48:06,665 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
130 |
+
2024-05-13 20:48:06,679 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
131 |
+
2024-05-13 20:48:07,082 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
132 |
+
2024-05-13 20:48:10,344 ERROR gpu :1244775 [interfaces.py:aggregate():159] Failed to serialize metric: division by zero
|
133 |
+
2024-05-13 20:48:10,366 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
134 |
+
2024-05-13 20:48:10,380 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
135 |
+
2024-05-13 20:48:10,863 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
136 |
+
2024-05-13 20:48:10,864 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
137 |
+
2024-05-13 20:48:13,048 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
138 |
+
2024-05-13 20:48:13,506 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
139 |
+
2024-05-13 20:48:13,529 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
140 |
+
2024-05-13 20:48:15,558 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
141 |
+
2024-05-13 20:48:15,586 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
142 |
+
2024-05-13 20:48:18,050 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
143 |
+
2024-05-13 20:48:18,552 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
144 |
+
2024-05-13 20:48:18,572 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
145 |
+
2024-05-13 20:48:20,626 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
146 |
+
2024-05-13 20:48:20,644 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
147 |
+
2024-05-13 20:48:23,050 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
148 |
+
2024-05-13 20:48:23,336 DEBUG SenderThread:1244775 [sender.py:send():378] send: stats
|
149 |
+
2024-05-13 20:48:23,683 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
150 |
+
2024-05-13 20:48:23,707 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
151 |
+
2024-05-13 20:48:25,750 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
152 |
+
2024-05-13 20:48:25,769 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
153 |
+
2024-05-13 20:48:25,863 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
154 |
+
2024-05-13 20:48:25,864 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
155 |
+
2024-05-13 20:48:28,681 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
156 |
+
2024-05-13 20:48:28,701 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
157 |
+
2024-05-13 20:48:29,005 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
158 |
+
2024-05-13 20:48:30,725 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
159 |
+
2024-05-13 20:48:30,747 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
160 |
+
2024-05-13 20:48:33,782 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
161 |
+
2024-05-13 20:48:33,801 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
162 |
+
2024-05-13 20:48:34,006 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
163 |
+
2024-05-13 20:48:35,835 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
164 |
+
2024-05-13 20:48:35,858 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
165 |
+
2024-05-13 20:48:38,877 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
166 |
+
2024-05-13 20:48:38,904 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
167 |
+
2024-05-13 20:48:39,007 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
168 |
+
2024-05-13 20:48:40,863 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
169 |
+
2024-05-13 20:48:40,864 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
170 |
+
2024-05-13 20:48:40,932 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
171 |
+
2024-05-13 20:48:40,946 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
172 |
+
2024-05-13 20:48:42,969 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
173 |
+
2024-05-13 20:48:42,980 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
174 |
+
2024-05-13 20:48:44,036 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
175 |
+
2024-05-13 20:48:45,801 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
176 |
+
2024-05-13 20:48:45,831 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
177 |
+
2024-05-13 20:48:49,037 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
178 |
+
2024-05-13 20:48:49,512 ERROR gpu :1244775 [interfaces.py:aggregate():159] Failed to serialize metric: division by zero
|
179 |
+
2024-05-13 20:48:49,550 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
180 |
+
2024-05-13 20:48:49,567 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
181 |
+
2024-05-13 20:48:52,479 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
182 |
+
2024-05-13 20:48:52,494 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
183 |
+
2024-05-13 20:48:53,338 DEBUG SenderThread:1244775 [sender.py:send():378] send: stats
|
184 |
+
2024-05-13 20:48:54,339 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
185 |
+
2024-05-13 20:48:54,518 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
186 |
+
2024-05-13 20:48:54,530 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
187 |
+
2024-05-13 20:48:55,864 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
188 |
+
2024-05-13 20:48:55,864 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
189 |
+
2024-05-13 20:48:57,576 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
190 |
+
2024-05-13 20:48:57,589 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
191 |
+
2024-05-13 20:48:59,615 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
192 |
+
2024-05-13 20:48:59,628 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
193 |
+
2024-05-13 20:49:00,007 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
194 |
+
2024-05-13 20:49:02,716 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
195 |
+
2024-05-13 20:49:02,730 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
196 |
+
2024-05-13 20:49:04,748 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
197 |
+
2024-05-13 20:49:04,763 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
198 |
+
2024-05-13 20:49:05,008 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
199 |
+
2024-05-13 20:49:07,719 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
200 |
+
2024-05-13 20:49:07,741 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
201 |
+
2024-05-13 20:49:09,766 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
202 |
+
2024-05-13 20:49:09,776 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
203 |
+
2024-05-13 20:49:10,009 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
204 |
+
2024-05-13 20:49:10,864 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
205 |
+
2024-05-13 20:49:10,864 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
206 |
+
2024-05-13 20:49:12,861 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
207 |
+
2024-05-13 20:49:12,875 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
208 |
+
2024-05-13 20:49:14,916 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
209 |
+
2024-05-13 20:49:14,934 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
210 |
+
2024-05-13 20:49:15,085 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
211 |
+
2024-05-13 20:49:17,938 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
212 |
+
2024-05-13 20:49:17,957 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
213 |
+
2024-05-13 20:49:20,002 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
214 |
+
2024-05-13 20:49:20,012 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
215 |
+
2024-05-13 20:49:20,086 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
216 |
+
2024-05-13 20:49:22,621 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
217 |
+
2024-05-13 20:49:22,641 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
218 |
+
2024-05-13 20:49:23,339 DEBUG SenderThread:1244775 [sender.py:send():378] send: stats
|
219 |
+
2024-05-13 20:49:24,675 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
220 |
+
2024-05-13 20:49:24,686 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
221 |
+
2024-05-13 20:49:25,341 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
222 |
+
2024-05-13 20:49:25,864 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
223 |
+
2024-05-13 20:49:25,865 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
224 |
+
2024-05-13 20:49:28,370 ERROR gpu :1244775 [interfaces.py:aggregate():159] Failed to serialize metric: division by zero
|
225 |
+
2024-05-13 20:49:28,612 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
226 |
+
2024-05-13 20:49:28,633 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
227 |
+
2024-05-13 20:49:30,656 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
228 |
+
2024-05-13 20:49:30,678 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
229 |
+
2024-05-13 20:49:31,038 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
230 |
+
2024-05-13 20:49:33,121 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
231 |
+
2024-05-13 20:49:33,150 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
232 |
+
2024-05-13 20:49:35,177 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
233 |
+
2024-05-13 20:49:35,217 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
234 |
+
2024-05-13 20:49:36,039 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
235 |
+
2024-05-13 20:49:37,932 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
236 |
+
2024-05-13 20:49:37,967 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
237 |
+
2024-05-13 20:49:40,002 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
238 |
+
2024-05-13 20:49:40,032 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
239 |
+
2024-05-13 20:49:40,864 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
240 |
+
2024-05-13 20:49:40,864 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
241 |
+
2024-05-13 20:49:42,037 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
242 |
+
2024-05-13 20:49:42,295 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
243 |
+
2024-05-13 20:49:42,620 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
244 |
+
2024-05-13 20:49:44,641 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
245 |
+
2024-05-13 20:49:44,688 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
246 |
+
2024-05-13 20:49:47,037 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
247 |
+
2024-05-13 20:49:47,242 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
248 |
+
2024-05-13 20:49:47,286 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
249 |
+
2024-05-13 20:49:49,336 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
250 |
+
2024-05-13 20:49:49,362 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
251 |
+
2024-05-13 20:49:51,898 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
252 |
+
2024-05-13 20:49:51,927 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
253 |
+
2024-05-13 20:49:52,038 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
254 |
+
2024-05-13 20:49:53,343 DEBUG SenderThread:1244775 [sender.py:send():378] send: stats
|
255 |
+
2024-05-13 20:49:54,378 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
256 |
+
2024-05-13 20:49:54,396 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
257 |
+
2024-05-13 20:49:55,864 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
258 |
+
2024-05-13 20:49:55,864 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
259 |
+
2024-05-13 20:49:56,417 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
260 |
+
2024-05-13 20:49:56,465 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
261 |
+
2024-05-13 20:49:58,028 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
262 |
+
2024-05-13 20:49:58,997 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
263 |
+
2024-05-13 20:49:59,011 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
264 |
+
2024-05-13 20:50:01,057 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
265 |
+
2024-05-13 20:50:01,097 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
266 |
+
2024-05-13 20:50:03,029 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
267 |
+
2024-05-13 20:50:05,200 ERROR gpu :1244775 [interfaces.py:aggregate():159] Failed to serialize metric: division by zero
|
268 |
+
2024-05-13 20:50:05,253 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
269 |
+
2024-05-13 20:50:05,272 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
270 |
+
2024-05-13 20:50:07,311 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
271 |
+
2024-05-13 20:50:07,330 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
272 |
+
2024-05-13 20:50:08,029 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
273 |
+
2024-05-13 20:50:09,887 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
274 |
+
2024-05-13 20:50:09,918 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
275 |
+
2024-05-13 20:50:10,864 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
276 |
+
2024-05-13 20:50:10,864 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
277 |
+
2024-05-13 20:50:11,934 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
278 |
+
2024-05-13 20:50:11,942 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
279 |
+
2024-05-13 20:50:13,047 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
280 |
+
2024-05-13 20:50:14,680 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
281 |
+
2024-05-13 20:50:14,697 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
282 |
+
2024-05-13 20:50:16,719 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
283 |
+
2024-05-13 20:50:16,738 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
284 |
+
2024-05-13 20:50:18,047 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
285 |
+
2024-05-13 20:50:19,232 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
286 |
+
2024-05-13 20:50:19,254 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
287 |
+
2024-05-13 20:50:21,272 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
288 |
+
2024-05-13 20:50:21,282 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
289 |
+
2024-05-13 20:50:23,048 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
290 |
+
2024-05-13 20:50:23,346 DEBUG SenderThread:1244775 [sender.py:send():378] send: stats
|
291 |
+
2024-05-13 20:50:23,802 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
292 |
+
2024-05-13 20:50:23,817 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
293 |
+
2024-05-13 20:50:25,839 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
294 |
+
2024-05-13 20:50:25,850 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
295 |
+
2024-05-13 20:50:25,864 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: stop_status
|
296 |
+
2024-05-13 20:50:25,865 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: stop_status
|
297 |
+
2024-05-13 20:50:28,061 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
298 |
+
2024-05-13 20:50:28,256 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
299 |
+
2024-05-13 20:50:28,265 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
300 |
+
2024-05-13 20:50:30,284 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
301 |
+
2024-05-13 20:50:30,296 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
302 |
+
2024-05-13 20:50:32,293 INFO Thread-12 :1244775 [dir_watcher.py:_on_file_modified():288] file/dir modified: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/output.log
|
303 |
+
2024-05-13 20:50:32,787 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
304 |
+
2024-05-13 20:50:32,796 ERROR gpu :1244775 [interfaces.py:monitor():142] Failed to sample metric: Not Supported
|
305 |
+
2024-05-13 20:50:33,152 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
306 |
+
2024-05-13 20:50:33,158 DEBUG SenderThread:1244775 [sender.py:send():378] send: exit
|
307 |
+
2024-05-13 20:50:33,158 INFO SenderThread:1244775 [sender.py:send_exit():585] handling exit code: 1
|
308 |
+
2024-05-13 20:50:33,159 INFO SenderThread:1244775 [sender.py:send_exit():587] handling runtime: 219
|
309 |
+
2024-05-13 20:50:33,159 INFO SenderThread:1244775 [sender.py:_save_file():1389] saving file wandb-summary.json with policy end
|
310 |
+
2024-05-13 20:50:33,159 INFO SenderThread:1244775 [sender.py:send_exit():593] send defer
|
311 |
+
2024-05-13 20:50:33,159 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
312 |
+
2024-05-13 20:50:33,159 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 0
|
313 |
+
2024-05-13 20:50:33,159 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
314 |
+
2024-05-13 20:50:33,160 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 0
|
315 |
+
2024-05-13 20:50:33,160 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 1
|
316 |
+
2024-05-13 20:50:33,160 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
317 |
+
2024-05-13 20:50:33,160 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 1
|
318 |
+
2024-05-13 20:50:33,160 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
319 |
+
2024-05-13 20:50:33,160 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 1
|
320 |
+
2024-05-13 20:50:33,160 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 2
|
321 |
+
2024-05-13 20:50:33,160 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
322 |
+
2024-05-13 20:50:33,160 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 2
|
323 |
+
2024-05-13 20:50:33,160 INFO HandlerThread:1244775 [system_monitor.py:finish():203] Stopping system monitor
|
324 |
+
2024-05-13 20:50:33,161 DEBUG SystemMonitor:1244775 [system_monitor.py:_start():179] Finished system metrics aggregation loop
|
325 |
+
2024-05-13 20:50:33,161 DEBUG SystemMonitor:1244775 [system_monitor.py:_start():183] Publishing last batch of metrics
|
326 |
+
2024-05-13 20:50:33,161 INFO HandlerThread:1244775 [interfaces.py:finish():200] Joined cpu monitor
|
327 |
+
2024-05-13 20:50:33,164 INFO HandlerThread:1244775 [interfaces.py:finish():200] Joined disk monitor
|
328 |
+
2024-05-13 20:50:33,293 INFO Thread-12 :1244775 [dir_watcher.py:_on_file_created():271] file/dir created: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/wandb-summary.json
|
329 |
+
2024-05-13 20:50:34,293 INFO Thread-12 :1244775 [dir_watcher.py:_on_file_modified():288] file/dir modified: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/output.log
|
330 |
+
2024-05-13 20:50:34,797 ERROR gpu :1244775 [interfaces.py:aggregate():159] Failed to serialize metric: division by zero
|
331 |
+
2024-05-13 20:50:34,797 INFO HandlerThread:1244775 [interfaces.py:finish():200] Joined gpu monitor
|
332 |
+
2024-05-13 20:50:34,797 INFO HandlerThread:1244775 [interfaces.py:finish():200] Joined memory monitor
|
333 |
+
2024-05-13 20:50:34,797 INFO HandlerThread:1244775 [interfaces.py:finish():200] Joined network monitor
|
334 |
+
2024-05-13 20:50:34,798 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: poll_exit
|
335 |
+
2024-05-13 20:50:34,799 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
336 |
+
2024-05-13 20:50:34,799 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 2
|
337 |
+
2024-05-13 20:50:34,799 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 3
|
338 |
+
2024-05-13 20:50:34,800 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
339 |
+
2024-05-13 20:50:34,800 DEBUG SenderThread:1244775 [sender.py:send():378] send: stats
|
340 |
+
2024-05-13 20:50:34,800 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 3
|
341 |
+
2024-05-13 20:50:34,800 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: poll_exit
|
342 |
+
2024-05-13 20:50:34,801 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
343 |
+
2024-05-13 20:50:34,801 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 3
|
344 |
+
2024-05-13 20:50:34,801 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 4
|
345 |
+
2024-05-13 20:50:34,801 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
346 |
+
2024-05-13 20:50:34,801 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 4
|
347 |
+
2024-05-13 20:50:34,801 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
348 |
+
2024-05-13 20:50:34,801 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 4
|
349 |
+
2024-05-13 20:50:34,801 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 5
|
350 |
+
2024-05-13 20:50:34,802 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
351 |
+
2024-05-13 20:50:34,802 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 5
|
352 |
+
2024-05-13 20:50:34,802 DEBUG SenderThread:1244775 [sender.py:send():378] send: summary
|
353 |
+
2024-05-13 20:50:34,802 INFO SenderThread:1244775 [sender.py:_save_file():1389] saving file wandb-summary.json with policy end
|
354 |
+
2024-05-13 20:50:34,802 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
355 |
+
2024-05-13 20:50:34,802 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 5
|
356 |
+
2024-05-13 20:50:34,802 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 6
|
357 |
+
2024-05-13 20:50:34,802 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
358 |
+
2024-05-13 20:50:34,802 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 6
|
359 |
+
2024-05-13 20:50:34,803 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
360 |
+
2024-05-13 20:50:34,803 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 6
|
361 |
+
2024-05-13 20:50:34,803 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 7
|
362 |
+
2024-05-13 20:50:34,803 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: status_report
|
363 |
+
2024-05-13 20:50:34,803 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
364 |
+
2024-05-13 20:50:34,803 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 7
|
365 |
+
2024-05-13 20:50:34,803 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
366 |
+
2024-05-13 20:50:34,803 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 7
|
367 |
+
2024-05-13 20:50:35,159 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: poll_exit
|
368 |
+
2024-05-13 20:50:35,294 INFO Thread-12 :1244775 [dir_watcher.py:_on_file_modified():288] file/dir modified: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/wandb-summary.json
|
369 |
+
2024-05-13 20:50:38,152 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 8
|
370 |
+
2024-05-13 20:50:38,152 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: poll_exit
|
371 |
+
2024-05-13 20:50:38,152 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
372 |
+
2024-05-13 20:50:38,153 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 8
|
373 |
+
2024-05-13 20:50:38,153 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
374 |
+
2024-05-13 20:50:38,153 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 8
|
375 |
+
2024-05-13 20:50:38,153 INFO SenderThread:1244775 [job_builder.py:build():432] Attempting to build job artifact
|
376 |
+
2024-05-13 20:50:38,153 INFO SenderThread:1244775 [job_builder.py:_get_source_type():565] is repo sourced job
|
377 |
+
2024-05-13 20:50:38,160 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: poll_exit
|
378 |
+
2024-05-13 20:50:38,179 INFO SenderThread:1244775 [job_builder.py:build():541] adding wandb-job metadata file
|
379 |
+
2024-05-13 20:50:38,181 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 9
|
380 |
+
2024-05-13 20:50:38,181 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: poll_exit
|
381 |
+
2024-05-13 20:50:38,181 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
382 |
+
2024-05-13 20:50:38,182 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 9
|
383 |
+
2024-05-13 20:50:38,182 DEBUG SenderThread:1244775 [sender.py:send():378] send: artifact
|
384 |
+
2024-05-13 20:50:38,294 INFO Thread-12 :1244775 [dir_watcher.py:_on_file_modified():288] file/dir modified: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/output.log
|
385 |
+
2024-05-13 20:50:39,160 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: poll_exit
|
386 |
+
2024-05-13 20:50:39,209 INFO wandb-upload_0:1244775 [upload_job.py:push():88] Uploaded file /home/sanchit/.local/share/wandb/artifacts/staging/tmp34vs1_ku
|
387 |
+
2024-05-13 20:50:39,238 INFO wandb-upload_1:1244775 [upload_job.py:push():88] Uploaded file /tmp/tmp_ne7l6g3/wandb-job.json
|
388 |
+
2024-05-13 20:50:40,085 INFO SenderThread:1244775 [sender.py:send_artifact():1467] sent artifact job-https___huggingface.co_sanchit-gandhi_parler-tts-mini-v0.1-expresso-concatenated-combined_run_parler_tts_training.py - {'id': 'QXJ0aWZhY3Q6ODM0NzI5NzMx', 'state': 'PENDING', 'artifactSequence': {'id': 'QXJ0aWZhY3RDb2xsZWN0aW9uOjE3NDIzMTI3NQ==', 'latestArtifact': None}}
|
389 |
+
2024-05-13 20:50:40,086 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
390 |
+
2024-05-13 20:50:40,086 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 9
|
391 |
+
2024-05-13 20:50:40,086 INFO SenderThread:1244775 [dir_watcher.py:finish():358] shutting down directory watcher
|
392 |
+
2024-05-13 20:50:40,295 INFO SenderThread:1244775 [dir_watcher.py:finish():388] scan: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files
|
393 |
+
2024-05-13 20:50:40,295 INFO SenderThread:1244775 [dir_watcher.py:finish():402] scan save: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/conda-environment.yaml conda-environment.yaml
|
394 |
+
2024-05-13 20:50:40,295 INFO SenderThread:1244775 [dir_watcher.py:finish():402] scan save: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/wandb-summary.json wandb-summary.json
|
395 |
+
2024-05-13 20:50:40,295 INFO SenderThread:1244775 [dir_watcher.py:finish():402] scan save: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/output.log output.log
|
396 |
+
2024-05-13 20:50:40,297 INFO SenderThread:1244775 [dir_watcher.py:finish():402] scan save: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/requirements.txt requirements.txt
|
397 |
+
2024-05-13 20:50:40,300 INFO SenderThread:1244775 [dir_watcher.py:finish():402] scan save: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/config.yaml config.yaml
|
398 |
+
2024-05-13 20:50:40,301 INFO SenderThread:1244775 [dir_watcher.py:finish():402] scan save: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/wandb-metadata.json wandb-metadata.json
|
399 |
+
2024-05-13 20:50:40,301 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 10
|
400 |
+
2024-05-13 20:50:40,301 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: poll_exit
|
401 |
+
2024-05-13 20:50:40,301 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
402 |
+
2024-05-13 20:50:40,302 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 10
|
403 |
+
2024-05-13 20:50:40,306 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
404 |
+
2024-05-13 20:50:40,306 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 10
|
405 |
+
2024-05-13 20:50:40,306 INFO SenderThread:1244775 [file_pusher.py:finish():169] shutting down file pusher
|
406 |
+
2024-05-13 20:50:40,632 INFO wandb-upload_0:1244775 [upload_job.py:push():130] Uploaded file /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/conda-environment.yaml
|
407 |
+
2024-05-13 20:50:40,639 INFO wandb-upload_1:1244775 [upload_job.py:push():130] Uploaded file /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/output.log
|
408 |
+
2024-05-13 20:50:40,758 INFO wandb-upload_3:1244775 [upload_job.py:push():130] Uploaded file /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/requirements.txt
|
409 |
+
2024-05-13 20:50:40,773 INFO wandb-upload_2:1244775 [upload_job.py:push():130] Uploaded file /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/wandb-summary.json
|
410 |
+
2024-05-13 20:50:40,918 INFO wandb-upload_4:1244775 [upload_job.py:push():130] Uploaded file /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/files/config.yaml
|
411 |
+
2024-05-13 20:50:41,118 INFO Thread-11 (_thread_body):1244775 [sender.py:transition_state():613] send defer: 11
|
412 |
+
2024-05-13 20:50:41,119 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
413 |
+
2024-05-13 20:50:41,119 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 11
|
414 |
+
2024-05-13 20:50:41,119 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
415 |
+
2024-05-13 20:50:41,119 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 11
|
416 |
+
2024-05-13 20:50:41,120 INFO SenderThread:1244775 [file_pusher.py:join():175] waiting for file pusher
|
417 |
+
2024-05-13 20:50:41,120 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 12
|
418 |
+
2024-05-13 20:50:41,120 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
419 |
+
2024-05-13 20:50:41,120 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 12
|
420 |
+
2024-05-13 20:50:41,120 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
421 |
+
2024-05-13 20:50:41,120 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 12
|
422 |
+
2024-05-13 20:50:41,120 INFO SenderThread:1244775 [file_stream.py:finish():601] file stream finish called
|
423 |
+
2024-05-13 20:50:41,161 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: poll_exit
|
424 |
+
2024-05-13 20:50:41,510 INFO SenderThread:1244775 [file_stream.py:finish():605] file stream finish is done
|
425 |
+
2024-05-13 20:50:41,510 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 13
|
426 |
+
2024-05-13 20:50:41,510 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: poll_exit
|
427 |
+
2024-05-13 20:50:41,510 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
428 |
+
2024-05-13 20:50:41,510 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 13
|
429 |
+
2024-05-13 20:50:41,510 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
430 |
+
2024-05-13 20:50:41,510 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 13
|
431 |
+
2024-05-13 20:50:41,510 INFO SenderThread:1244775 [sender.py:transition_state():613] send defer: 14
|
432 |
+
2024-05-13 20:50:41,511 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: defer
|
433 |
+
2024-05-13 20:50:41,511 INFO HandlerThread:1244775 [handler.py:handle_request_defer():184] handle defer: 14
|
434 |
+
2024-05-13 20:50:41,511 DEBUG SenderThread:1244775 [sender.py:send():378] send: final
|
435 |
+
2024-05-13 20:50:41,511 DEBUG SenderThread:1244775 [sender.py:send():378] send: footer
|
436 |
+
2024-05-13 20:50:41,511 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: defer
|
437 |
+
2024-05-13 20:50:41,511 INFO SenderThread:1244775 [sender.py:send_request_defer():609] handle sender defer: 14
|
438 |
+
2024-05-13 20:50:41,512 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: poll_exit
|
439 |
+
2024-05-13 20:50:41,512 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: poll_exit
|
440 |
+
2024-05-13 20:50:41,512 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: poll_exit
|
441 |
+
2024-05-13 20:50:41,512 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: poll_exit
|
442 |
+
2024-05-13 20:50:41,513 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: server_info
|
443 |
+
2024-05-13 20:50:41,513 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: get_summary
|
444 |
+
2024-05-13 20:50:41,513 DEBUG SenderThread:1244775 [sender.py:send_request():405] send_request: server_info
|
445 |
+
2024-05-13 20:50:41,515 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: sampled_history
|
446 |
+
2024-05-13 20:50:41,515 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: internal_messages
|
447 |
+
2024-05-13 20:50:41,651 INFO MainThread:1244775 [wandb_run.py:_footer_history_summary_info():3994] rendering history
|
448 |
+
2024-05-13 20:50:41,651 INFO MainThread:1244775 [wandb_run.py:_footer_history_summary_info():4026] rendering summary
|
449 |
+
2024-05-13 20:50:41,651 INFO MainThread:1244775 [wandb_run.py:_footer_sync_info():3953] logging synced files
|
450 |
+
2024-05-13 20:50:41,651 DEBUG HandlerThread:1244775 [handler.py:handle_request():158] handle_request: shutdown
|
451 |
+
2024-05-13 20:50:41,651 INFO HandlerThread:1244775 [handler.py:finish():882] shutting down handler
|
452 |
+
2024-05-13 20:50:42,513 INFO WriterThread:1244775 [datastore.py:close():296] close: /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/run-m0g0ap7d.wandb
|
453 |
+
2024-05-13 20:50:42,651 INFO SenderThread:1244775 [sender.py:finish():1545] shutting down sender
|
454 |
+
2024-05-13 20:50:42,651 INFO SenderThread:1244775 [file_pusher.py:finish():169] shutting down file pusher
|
455 |
+
2024-05-13 20:50:42,651 INFO SenderThread:1244775 [file_pusher.py:join():175] waiting for file pusher
|
wandb/run-20240513_204652-m0g0ap7d/logs/debug.log
ADDED
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
2024-05-13 20:46:52,817 INFO MainThread:1244577 [wandb_setup.py:_flush():76] Current SDK version is 0.17.0
|
2 |
+
2024-05-13 20:46:52,817 INFO MainThread:1244577 [wandb_setup.py:_flush():76] Configure stats pid to 1244577
|
3 |
+
2024-05-13 20:46:52,817 INFO MainThread:1244577 [wandb_setup.py:_flush():76] Loading settings from /home/sanchit/.config/wandb/settings
|
4 |
+
2024-05-13 20:46:52,817 INFO MainThread:1244577 [wandb_setup.py:_flush():76] Loading settings from /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/settings
|
5 |
+
2024-05-13 20:46:52,817 INFO MainThread:1244577 [wandb_setup.py:_flush():76] Loading settings from environment variables: {}
|
6 |
+
2024-05-13 20:46:52,817 INFO MainThread:1244577 [wandb_setup.py:_flush():76] Applying setup settings: {'_disable_service': False}
|
7 |
+
2024-05-13 20:46:52,817 INFO MainThread:1244577 [wandb_setup.py:_flush():76] Inferring run settings from compute environment: {'program_relpath': 'run_parler_tts_training.py', 'program_abspath': '/raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/run_parler_tts_training.py', 'program': '/raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/run_parler_tts_training.py'}
|
8 |
+
2024-05-13 20:46:52,817 INFO MainThread:1244577 [wandb_setup.py:_flush():76] Applying login settings: {}
|
9 |
+
2024-05-13 20:46:52,817 INFO MainThread:1244577 [wandb_init.py:_log_setup():520] Logging user logs to /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/logs/debug.log
|
10 |
+
2024-05-13 20:46:52,818 INFO MainThread:1244577 [wandb_init.py:_log_setup():521] Logging internal logs to /raid/sanchit/parler-tts-mini-v0.1-expresso-concatenated-combined/wandb/run-20240513_204652-m0g0ap7d/logs/debug-internal.log
|
11 |
+
2024-05-13 20:46:52,818 INFO MainThread:1244577 [wandb_init.py:init():560] calling init triggers
|
12 |
+
2024-05-13 20:46:52,818 INFO MainThread:1244577 [wandb_init.py:init():567] wandb.init called with sweep_config: {}
|
13 |
+
config: {}
|
14 |
+
2024-05-13 20:46:52,818 INFO MainThread:1244577 [wandb_init.py:init():610] starting backend
|
15 |
+
2024-05-13 20:46:52,818 INFO MainThread:1244577 [wandb_init.py:init():614] setting up manager
|
16 |
+
2024-05-13 20:46:52,821 INFO MainThread:1244577 [backend.py:_multiprocessing_setup():105] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
|
17 |
+
2024-05-13 20:46:52,822 INFO MainThread:1244577 [wandb_init.py:init():622] backend started and connected
|
18 |
+
2024-05-13 20:46:52,823 INFO MainThread:1244577 [wandb_init.py:init():711] updated telemetry
|
19 |
+
2024-05-13 20:46:52,826 INFO MainThread:1244577 [wandb_init.py:init():744] communicating run to backend with 90.0 second timeout
|
20 |
+
2024-05-13 20:46:53,228 INFO MainThread:1244577 [wandb_run.py:_on_init():2396] communicating current version
|
21 |
+
2024-05-13 20:46:53,287 INFO MainThread:1244577 [wandb_run.py:_on_init():2405] got version response
|
22 |
+
2024-05-13 20:46:53,287 INFO MainThread:1244577 [wandb_init.py:init():795] starting run threads in backend
|
23 |
+
2024-05-13 20:46:55,863 INFO MainThread:1244577 [wandb_run.py:_console_start():2374] atexit reg
|
24 |
+
2024-05-13 20:46:55,863 INFO MainThread:1244577 [wandb_run.py:_redirect():2229] redirect: wrap_raw
|
25 |
+
2024-05-13 20:46:55,863 INFO MainThread:1244577 [wandb_run.py:_redirect():2294] Wrapping output streams.
|
26 |
+
2024-05-13 20:46:55,863 INFO MainThread:1244577 [wandb_run.py:_redirect():2319] Redirects installed.
|
27 |
+
2024-05-13 20:46:55,864 INFO MainThread:1244577 [wandb_init.py:init():838] run started, returning control to user process
|
28 |
+
2024-05-13 20:46:55,864 INFO MainThread:1244577 [wandb_run.py:_config_callback():1376] config_cb None None {'learning_rate': 8e-05, 'model_name_or_path': 'parler-tts/parler_tts_mini_v0.1', 'num_train_epochs': 8, 'gradient_accumulation_steps': 8, 'per_device_train_batch_size': 16, 'global_batch_size': 16, 'mixed_precision': 'bf16', 'lr_scheduler_type': 'SchedulerType.COSINE', 'warmup_steps': 250, 'freeze_text_encoder': True, 'max_duration_in_seconds': 30.0, 'weight_decay': 0.01, 'adam_beta1': 0.9, 'adam_beta2': 0.99, 'temperature': 1.0}
|
29 |
+
2024-05-13 20:50:42,653 WARNING MsgRouterThr:1244577 [router.py:message_loop():77] message_loop has been closed
|
wandb/run-20240513_204652-m0g0ap7d/run-m0g0ap7d.wandb
ADDED
Binary file (45 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/conda-environment.yaml
ADDED
@@ -0,0 +1,248 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
name: venv
|
2 |
+
channels:
|
3 |
+
- defaults
|
4 |
+
dependencies:
|
5 |
+
- _libgcc_mutex=0.1=main
|
6 |
+
- _openmp_mutex=5.1=1_gnu
|
7 |
+
- bzip2=1.0.8=h5eee18b_6
|
8 |
+
- ca-certificates=2024.3.11=h06a4308_0
|
9 |
+
- ld_impl_linux-64=2.38=h1181459_1
|
10 |
+
- libffi=3.4.4=h6a678d5_1
|
11 |
+
- libgcc-ng=11.2.0=h1234567_1
|
12 |
+
- libgomp=11.2.0=h1234567_1
|
13 |
+
- libstdcxx-ng=11.2.0=h1234567_1
|
14 |
+
- libuuid=1.41.5=h5eee18b_0
|
15 |
+
- ncurses=6.4=h6a678d5_0
|
16 |
+
- openssl=3.0.13=h7f8727e_1
|
17 |
+
- pip=24.0=py311h06a4308_0
|
18 |
+
- python=3.11.9=h955ad1f_0
|
19 |
+
- readline=8.2=h5eee18b_0
|
20 |
+
- setuptools=69.5.1=py311h06a4308_0
|
21 |
+
- sqlite=3.45.3=h5eee18b_0
|
22 |
+
- tk=8.6.14=h39e8969_0
|
23 |
+
- wheel=0.43.0=py311h06a4308_0
|
24 |
+
- xz=5.4.6=h5eee18b_1
|
25 |
+
- zlib=1.2.13=h5eee18b_1
|
26 |
+
- pip:
|
27 |
+
- absl-py==2.1.0
|
28 |
+
- accelerate==0.30.0
|
29 |
+
- aiohttp==3.9.5
|
30 |
+
- aiosignal==1.3.1
|
31 |
+
- aniso8601==9.0.1
|
32 |
+
- annotated-types==0.6.0
|
33 |
+
- anyio==4.3.0
|
34 |
+
- argbind==0.3.7
|
35 |
+
- argon2-cffi==23.1.0
|
36 |
+
- argon2-cffi-bindings==21.2.0
|
37 |
+
- arrow==1.3.0
|
38 |
+
- asttokens==2.4.1
|
39 |
+
- async-lru==2.0.4
|
40 |
+
- attrs==23.2.0
|
41 |
+
- audioread==3.0.1
|
42 |
+
- babel==2.15.0
|
43 |
+
- beautifulsoup4==4.12.3
|
44 |
+
- bidict==0.23.1
|
45 |
+
- bitsandbytes==0.43.1
|
46 |
+
- bleach==6.1.0
|
47 |
+
- certifi==2024.2.2
|
48 |
+
- cffi==1.16.0
|
49 |
+
- charset-normalizer==3.3.2
|
50 |
+
- click==8.1.7
|
51 |
+
- coloredlogs==14.0
|
52 |
+
- comm==0.2.2
|
53 |
+
- contourpy==1.2.1
|
54 |
+
- cycler==0.12.1
|
55 |
+
- datasets==2.19.1
|
56 |
+
- debugpy==1.8.1
|
57 |
+
- decorator==5.1.1
|
58 |
+
- defusedxml==0.7.1
|
59 |
+
- descript-audio-codec==1.0.0
|
60 |
+
- descript-audiotools==0.7.2
|
61 |
+
- dill==0.3.8
|
62 |
+
- dnspython==2.3.0
|
63 |
+
- docker-pycreds==0.4.0
|
64 |
+
- docstring-parser==0.16
|
65 |
+
- editdistance==0.8.1
|
66 |
+
- einops==0.8.0
|
67 |
+
- et-xmlfile==1.1.0
|
68 |
+
- evaluate==0.4.2
|
69 |
+
- eventlet==0.36.1
|
70 |
+
- executing==2.0.1
|
71 |
+
- fastjsonschema==2.19.1
|
72 |
+
- ffmpy==0.3.2
|
73 |
+
- filelock==3.14.0
|
74 |
+
- fire==0.6.0
|
75 |
+
- flask==2.2.5
|
76 |
+
- flask-cors==4.0.1
|
77 |
+
- flask-restful==0.3.10
|
78 |
+
- flask-socketio==5.3.6
|
79 |
+
- flask-talisman==1.1.0
|
80 |
+
- flatten-dict==0.4.2
|
81 |
+
- fonttools==4.51.0
|
82 |
+
- fqdn==1.5.1
|
83 |
+
- frozenlist==1.4.1
|
84 |
+
- fsspec==2024.3.1
|
85 |
+
- future==1.0.0
|
86 |
+
- g2p==2.0.0
|
87 |
+
- gitdb==4.0.11
|
88 |
+
- gitpython==3.1.43
|
89 |
+
- greenlet==3.0.3
|
90 |
+
- grpcio==1.63.0
|
91 |
+
- h11==0.14.0
|
92 |
+
- httpcore==1.0.5
|
93 |
+
- httpx==0.27.0
|
94 |
+
- huggingface-hub==0.23.0
|
95 |
+
- humanfriendly==10.0
|
96 |
+
- idna==3.7
|
97 |
+
- importlib-resources==6.4.0
|
98 |
+
- ipdb==0.13.13
|
99 |
+
- ipykernel==6.29.4
|
100 |
+
- ipython==8.24.0
|
101 |
+
- isoduration==20.11.0
|
102 |
+
- itsdangerous==2.2.0
|
103 |
+
- jedi==0.19.1
|
104 |
+
- jinja2==3.1.4
|
105 |
+
- jiwer==3.0.4
|
106 |
+
- joblib==1.4.2
|
107 |
+
- json5==0.9.25
|
108 |
+
- jsonpointer==2.4
|
109 |
+
- jsonschema==4.22.0
|
110 |
+
- jsonschema-specifications==2023.12.1
|
111 |
+
- julius==0.2.7
|
112 |
+
- jupyter-client==8.6.1
|
113 |
+
- jupyter-core==5.7.2
|
114 |
+
- jupyter-events==0.10.0
|
115 |
+
- jupyter-lsp==2.2.5
|
116 |
+
- jupyter-server==2.14.0
|
117 |
+
- jupyter-server-terminals==0.5.3
|
118 |
+
- jupyterlab==4.2.0
|
119 |
+
- jupyterlab-pygments==0.3.0
|
120 |
+
- jupyterlab-server==2.27.1
|
121 |
+
- kiwisolver==1.4.5
|
122 |
+
- lazy-loader==0.4
|
123 |
+
- librosa==0.10.2
|
124 |
+
- llvmlite==0.42.0
|
125 |
+
- markdown==3.6
|
126 |
+
- markdown-it-py==3.0.0
|
127 |
+
- markdown2==2.4.13
|
128 |
+
- markupsafe==2.1.5
|
129 |
+
- matplotlib==3.8.4
|
130 |
+
- matplotlib-inline==0.1.7
|
131 |
+
- mdurl==0.1.2
|
132 |
+
- mistune==3.0.2
|
133 |
+
- mpmath==1.3.0
|
134 |
+
- msgpack==1.0.8
|
135 |
+
- multidict==6.0.5
|
136 |
+
- multiprocess==0.70.16
|
137 |
+
- munkres==1.1.4
|
138 |
+
- nbclient==0.10.0
|
139 |
+
- nbconvert==7.16.4
|
140 |
+
- nbformat==5.10.4
|
141 |
+
- nest-asyncio==1.6.0
|
142 |
+
- networkx==3.3
|
143 |
+
- notebook-shim==0.2.4
|
144 |
+
- numba==0.59.1
|
145 |
+
- numpy==1.26.4
|
146 |
+
- nvidia-cublas-cu12==12.1.3.1
|
147 |
+
- nvidia-cuda-cupti-cu12==12.1.105
|
148 |
+
- nvidia-cuda-nvrtc-cu12==12.1.105
|
149 |
+
- nvidia-cuda-runtime-cu12==12.1.105
|
150 |
+
- nvidia-cudnn-cu12==8.9.2.26
|
151 |
+
- nvidia-cufft-cu12==11.0.2.54
|
152 |
+
- nvidia-curand-cu12==10.3.2.106
|
153 |
+
- nvidia-cusolver-cu12==11.4.5.107
|
154 |
+
- nvidia-cusparse-cu12==12.1.0.106
|
155 |
+
- nvidia-nccl-cu12==2.20.5
|
156 |
+
- nvidia-nvjitlink-cu12==12.4.127
|
157 |
+
- nvidia-nvtx-cu12==12.1.105
|
158 |
+
- openpyxl==3.1.2
|
159 |
+
- overrides==7.7.0
|
160 |
+
- packaging==24.0
|
161 |
+
- pandas==2.2.2
|
162 |
+
- pandocfilters==1.5.1
|
163 |
+
- panphon==0.20.0
|
164 |
+
- parler-tts==0.1
|
165 |
+
- parso==0.8.4
|
166 |
+
- pexpect==4.9.0
|
167 |
+
- pillow==10.3.0
|
168 |
+
- platformdirs==4.2.1
|
169 |
+
- pooch==1.8.1
|
170 |
+
- prometheus-client==0.20.0
|
171 |
+
- prompt-toolkit==3.0.43
|
172 |
+
- protobuf==3.19.6
|
173 |
+
- psutil==5.9.8
|
174 |
+
- ptyprocess==0.7.0
|
175 |
+
- pure-eval==0.2.2
|
176 |
+
- pyarrow==16.0.0
|
177 |
+
- pyarrow-hotfix==0.6
|
178 |
+
- pycparser==2.22
|
179 |
+
- pydantic==2.7.1
|
180 |
+
- pydantic-core==2.18.2
|
181 |
+
- pygments==2.18.0
|
182 |
+
- pyloudnorm==0.1.1
|
183 |
+
- pyparsing==3.1.2
|
184 |
+
- pystoi==0.4.1
|
185 |
+
- python-dateutil==2.9.0.post0
|
186 |
+
- python-engineio==4.9.0
|
187 |
+
- python-json-logger==2.0.7
|
188 |
+
- python-socketio==5.11.2
|
189 |
+
- pytz==2024.1
|
190 |
+
- pyyaml==6.0.1
|
191 |
+
- pyzmq==26.0.3
|
192 |
+
- randomname==0.2.1
|
193 |
+
- rapidfuzz==3.9.0
|
194 |
+
- referencing==0.35.1
|
195 |
+
- regex==2024.4.28
|
196 |
+
- requests==2.31.0
|
197 |
+
- rfc3339-validator==0.1.4
|
198 |
+
- rfc3986-validator==0.1.1
|
199 |
+
- rich==13.7.1
|
200 |
+
- rpds-py==0.18.1
|
201 |
+
- safetensors==0.4.3
|
202 |
+
- scikit-learn==1.4.2
|
203 |
+
- scipy==1.13.0
|
204 |
+
- send2trash==1.8.3
|
205 |
+
- sentencepiece==0.2.0
|
206 |
+
- sentry-sdk==2.1.1
|
207 |
+
- setproctitle==1.3.3
|
208 |
+
- simple-websocket==1.0.0
|
209 |
+
- six==1.16.0
|
210 |
+
- smmap==5.0.1
|
211 |
+
- sniffio==1.3.1
|
212 |
+
- soundfile==0.12.1
|
213 |
+
- soupsieve==2.5
|
214 |
+
- soxr==0.3.7
|
215 |
+
- stack-data==0.6.3
|
216 |
+
- sympy==1.12
|
217 |
+
- tensorboard==2.16.2
|
218 |
+
- tensorboard-data-server==0.7.2
|
219 |
+
- termcolor==2.4.0
|
220 |
+
- terminado==0.18.1
|
221 |
+
- text-unidecode==1.3
|
222 |
+
- threadpoolctl==3.5.0
|
223 |
+
- tinycss2==1.3.0
|
224 |
+
- tokenizers==0.19.1
|
225 |
+
- torch==2.3.0
|
226 |
+
- torch-stoi==0.2.1
|
227 |
+
- torchaudio==2.3.0
|
228 |
+
- tornado==6.4
|
229 |
+
- tqdm==4.66.4
|
230 |
+
- traitlets==5.14.3
|
231 |
+
- transformers==4.41.0.dev0
|
232 |
+
- triton==2.3.0
|
233 |
+
- types-python-dateutil==2.9.0.20240316
|
234 |
+
- typing-extensions==4.11.0
|
235 |
+
- tzdata==2024.1
|
236 |
+
- unicodecsv==0.14.1
|
237 |
+
- uri-template==1.3.0
|
238 |
+
- urllib3==2.2.1
|
239 |
+
- wandb==0.17.0
|
240 |
+
- wcwidth==0.2.13
|
241 |
+
- webcolors==1.13
|
242 |
+
- webencodings==0.5.1
|
243 |
+
- websocket-client==1.8.0
|
244 |
+
- werkzeug==3.0.3
|
245 |
+
- wsproto==1.2.0
|
246 |
+
- xxhash==3.4.1
|
247 |
+
- yarl==1.9.4
|
248 |
+
prefix: /home/sanchit/miniconda3/envs/venv
|
wandb/run-20240513_205249-qaoje1x9/files/config.yaml
ADDED
@@ -0,0 +1,88 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
wandb_version: 1
|
2 |
+
|
3 |
+
_wandb:
|
4 |
+
desc: null
|
5 |
+
value:
|
6 |
+
python_version: 3.11.9
|
7 |
+
cli_version: 0.17.0
|
8 |
+
framework: huggingface
|
9 |
+
huggingface_version: 4.41.0.dev0
|
10 |
+
is_jupyter_run: false
|
11 |
+
is_kaggle_kernel: false
|
12 |
+
start_time: 1715626369
|
13 |
+
t:
|
14 |
+
1:
|
15 |
+
- 1
|
16 |
+
- 5
|
17 |
+
- 11
|
18 |
+
- 49
|
19 |
+
- 51
|
20 |
+
- 53
|
21 |
+
- 55
|
22 |
+
- 71
|
23 |
+
- 100
|
24 |
+
2:
|
25 |
+
- 1
|
26 |
+
- 5
|
27 |
+
- 11
|
28 |
+
- 49
|
29 |
+
- 51
|
30 |
+
- 53
|
31 |
+
- 55
|
32 |
+
- 71
|
33 |
+
- 100
|
34 |
+
3:
|
35 |
+
- 2
|
36 |
+
- 23
|
37 |
+
- 61
|
38 |
+
4: 3.11.9
|
39 |
+
5: 0.17.0
|
40 |
+
6: 4.41.0.dev0
|
41 |
+
8:
|
42 |
+
- 5
|
43 |
+
13: linux-x86_64
|
44 |
+
learning_rate:
|
45 |
+
desc: null
|
46 |
+
value: 8.0e-05
|
47 |
+
model_name_or_path:
|
48 |
+
desc: null
|
49 |
+
value: parler-tts/parler_tts_mini_v0.1
|
50 |
+
num_train_epochs:
|
51 |
+
desc: null
|
52 |
+
value: 8
|
53 |
+
gradient_accumulation_steps:
|
54 |
+
desc: null
|
55 |
+
value: 8
|
56 |
+
per_device_train_batch_size:
|
57 |
+
desc: null
|
58 |
+
value: 16
|
59 |
+
global_batch_size:
|
60 |
+
desc: null
|
61 |
+
value: 16
|
62 |
+
mixed_precision:
|
63 |
+
desc: null
|
64 |
+
value: bf16
|
65 |
+
lr_scheduler_type:
|
66 |
+
desc: null
|
67 |
+
value: SchedulerType.COSINE
|
68 |
+
warmup_steps:
|
69 |
+
desc: null
|
70 |
+
value: 250
|
71 |
+
freeze_text_encoder:
|
72 |
+
desc: null
|
73 |
+
value: true
|
74 |
+
max_duration_in_seconds:
|
75 |
+
desc: null
|
76 |
+
value: 30.0
|
77 |
+
weight_decay:
|
78 |
+
desc: null
|
79 |
+
value: 0.01
|
80 |
+
adam_beta1:
|
81 |
+
desc: null
|
82 |
+
value: 0.9
|
83 |
+
adam_beta2:
|
84 |
+
desc: null
|
85 |
+
value: 0.99
|
86 |
+
temperature:
|
87 |
+
desc: null
|
88 |
+
value: 1.0
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_193b029d494fd24e7cfa.wav
ADDED
Binary file (962 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_345bfb6a72849809d361.wav
ADDED
Binary file (962 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_3c9adbd9374e0fb5ce3d.wav
ADDED
Binary file (962 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_3cd94e4824cc6c8fb09c.wav
ADDED
Binary file (962 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_3ed6544e58dd861a5d9e.wav
ADDED
Binary file (962 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_9da1fed11be9d614d9ec.wav
ADDED
Binary file (962 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_1016_ec838b0233dbe87d33f3.wav
ADDED
Binary file (962 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_0db946e177a69cbe11f5.wav
ADDED
Binary file (842 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_32c9af8d48e757598000.wav
ADDED
Binary file (842 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_341c52fd92336c009f67.wav
ADDED
Binary file (842 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_43ed5d3749c912acb591.wav
ADDED
Binary file (842 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_75818e76e9e077f058be.wav
ADDED
Binary file (842 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_d24330f3382b9e6ea7ea.wav
ADDED
Binary file (842 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_127_ec7dcb5421538131ede7.wav
ADDED
Binary file (842 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_2794dcaf322bd12e2814.wav
ADDED
Binary file (810 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_2ef5b33e2eaf98dca4a6.wav
ADDED
Binary file (810 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_4ca836a112634417b82e.wav
ADDED
Binary file (810 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_837a3499e3f93538b643.wav
ADDED
Binary file (810 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_b3650df61e399b05257d.wav
ADDED
Binary file (810 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_d33ccbefe990db0dce2b.wav
ADDED
Binary file (810 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_254_e8ca5038019cad3cde86.wav
ADDED
Binary file (810 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_4899e0da4615e883ad13.wav
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4899e0da4615e883ad137c4ebecee7d753f2512e69b2610ee121046fafa228a2
|
3 |
+
size 1282092
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_492597073098578f0605.wav
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:492597073098578f0605b080944b218a8c03fd49715aa37e5b6b9f2bbdc975fe
|
3 |
+
size 1282092
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_67d9409a306e3614ec3f.wav
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:67d9409a306e3614ec3f84ea766046f8674e3cd4a8b13d1231dd6a356164f6d3
|
3 |
+
size 1282092
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_7c47ba927ac118ffaacc.wav
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7c47ba927ac118ffaacca70183d37077fd7a6d6fc2c649e51b7fb5137609cd8e
|
3 |
+
size 1282092
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_89ad32d31f3e70178cc1.wav
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:89ad32d31f3e70178cc19b1fb1c8ceda0377b104575def497602c02ba8bd1196
|
3 |
+
size 1282092
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_cd644667186ae0518a3c.wav
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cd644667186ae0518a3c08763c4aabf1e1f4d5754f3025b1e40779e8f068d489
|
3 |
+
size 1282092
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_381_f7405ef7b645b3265477.wav
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f7405ef7b645b3265477b3e4e65bf1313dbc4445631e516872a054a694bb5e4a
|
3 |
+
size 1282092
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_508_00e9064c0bdbd6b9428d.wav
ADDED
Binary file (996 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_508_11adac906cb7e2ef30c6.wav
ADDED
Binary file (996 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_508_5619d97860f92fc1a62d.wav
ADDED
Binary file (996 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_508_bcb03b95f0470920bdc6.wav
ADDED
Binary file (996 kB). View file
|
|
wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech samples/eval_508_e48b4ff2b12d5ffdb11c.wav
ADDED
Binary file (996 kB). View file
|
|