| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 35.0, | |
| "eval_steps": 500, | |
| "global_step": 875, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 6.834115982055664, | |
| "learning_rate": 4.874285714285715e-05, | |
| "loss": 3.1488, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_cer": 0.37662337662337664, | |
| "eval_loss": 0.7865502834320068, | |
| "eval_runtime": 8.8116, | |
| "eval_samples_per_second": 2.27, | |
| "eval_steps_per_second": 0.34, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 5.263820171356201, | |
| "learning_rate": 4.7314285714285714e-05, | |
| "loss": 0.8093, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_cer": 0.4675324675324675, | |
| "eval_loss": 0.8947464823722839, | |
| "eval_runtime": 2.7975, | |
| "eval_samples_per_second": 7.149, | |
| "eval_steps_per_second": 1.072, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 5.471518039703369, | |
| "learning_rate": 4.588571428571429e-05, | |
| "loss": 0.724, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_cer": 0.4155844155844156, | |
| "eval_loss": 0.8557536005973816, | |
| "eval_runtime": 2.7561, | |
| "eval_samples_per_second": 7.257, | |
| "eval_steps_per_second": 1.088, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 6.84343957901001, | |
| "learning_rate": 4.445714285714286e-05, | |
| "loss": 0.7405, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_cer": 0.33766233766233766, | |
| "eval_loss": 0.8277163505554199, | |
| "eval_runtime": 2.8432, | |
| "eval_samples_per_second": 7.034, | |
| "eval_steps_per_second": 1.055, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 4.826574802398682, | |
| "learning_rate": 4.302857142857143e-05, | |
| "loss": 0.7361, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_cer": 0.36363636363636365, | |
| "eval_loss": 0.8111216425895691, | |
| "eval_runtime": 2.8495, | |
| "eval_samples_per_second": 7.019, | |
| "eval_steps_per_second": 1.053, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 4.757009983062744, | |
| "learning_rate": 4.16e-05, | |
| "loss": 0.7061, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_cer": 0.37662337662337664, | |
| "eval_loss": 0.7869578003883362, | |
| "eval_runtime": 2.8778, | |
| "eval_samples_per_second": 6.95, | |
| "eval_steps_per_second": 1.042, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 4.829946517944336, | |
| "learning_rate": 4.017142857142857e-05, | |
| "loss": 0.7159, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_cer": 0.45454545454545453, | |
| "eval_loss": 0.8124594688415527, | |
| "eval_runtime": 2.9455, | |
| "eval_samples_per_second": 6.79, | |
| "eval_steps_per_second": 1.019, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 3.9984848499298096, | |
| "learning_rate": 3.874285714285715e-05, | |
| "loss": 0.707, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_cer": 0.37662337662337664, | |
| "eval_loss": 0.8081725239753723, | |
| "eval_runtime": 2.8597, | |
| "eval_samples_per_second": 6.994, | |
| "eval_steps_per_second": 1.049, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "grad_norm": 3.7987842559814453, | |
| "learning_rate": 3.7314285714285715e-05, | |
| "loss": 0.7073, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_cer": 0.45454545454545453, | |
| "eval_loss": 0.8078359961509705, | |
| "eval_runtime": 2.8752, | |
| "eval_samples_per_second": 6.956, | |
| "eval_steps_per_second": 1.043, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 4.148154258728027, | |
| "learning_rate": 3.588571428571429e-05, | |
| "loss": 0.6956, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_cer": 0.37662337662337664, | |
| "eval_loss": 0.8172693848609924, | |
| "eval_runtime": 2.8623, | |
| "eval_samples_per_second": 6.987, | |
| "eval_steps_per_second": 1.048, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "grad_norm": 3.5226118564605713, | |
| "learning_rate": 3.445714285714286e-05, | |
| "loss": 0.7018, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_cer": 0.37662337662337664, | |
| "eval_loss": 0.8221365809440613, | |
| "eval_runtime": 2.8711, | |
| "eval_samples_per_second": 6.966, | |
| "eval_steps_per_second": 1.045, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "grad_norm": 3.736212730407715, | |
| "learning_rate": 3.302857142857143e-05, | |
| "loss": 0.6901, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_cer": 0.45454545454545453, | |
| "eval_loss": 0.8348967432975769, | |
| "eval_runtime": 2.9677, | |
| "eval_samples_per_second": 6.739, | |
| "eval_steps_per_second": 1.011, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "grad_norm": 3.468735456466675, | |
| "learning_rate": 3.16e-05, | |
| "loss": 0.6862, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_cer": 0.36363636363636365, | |
| "eval_loss": 0.7819182276725769, | |
| "eval_runtime": 2.8624, | |
| "eval_samples_per_second": 6.987, | |
| "eval_steps_per_second": 1.048, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "grad_norm": 4.044895648956299, | |
| "learning_rate": 3.0171428571428572e-05, | |
| "loss": 0.7024, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_cer": 0.36363636363636365, | |
| "eval_loss": 0.7919606566429138, | |
| "eval_runtime": 2.907, | |
| "eval_samples_per_second": 6.88, | |
| "eval_steps_per_second": 1.032, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "grad_norm": 3.781301259994507, | |
| "learning_rate": 2.8742857142857143e-05, | |
| "loss": 0.6858, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_cer": 0.36363636363636365, | |
| "eval_loss": 0.812893807888031, | |
| "eval_runtime": 2.8775, | |
| "eval_samples_per_second": 6.95, | |
| "eval_steps_per_second": 1.043, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "grad_norm": 3.412642240524292, | |
| "learning_rate": 2.7314285714285716e-05, | |
| "loss": 0.6782, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_cer": 0.4025974025974026, | |
| "eval_loss": 0.8198434710502625, | |
| "eval_runtime": 2.89, | |
| "eval_samples_per_second": 6.92, | |
| "eval_steps_per_second": 1.038, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "grad_norm": 2.96364426612854, | |
| "learning_rate": 2.5885714285714286e-05, | |
| "loss": 0.675, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_cer": 0.44155844155844154, | |
| "eval_loss": 0.8002452850341797, | |
| "eval_runtime": 2.9999, | |
| "eval_samples_per_second": 6.667, | |
| "eval_steps_per_second": 1.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "grad_norm": 3.237696647644043, | |
| "learning_rate": 2.445714285714286e-05, | |
| "loss": 0.6749, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_cer": 0.36363636363636365, | |
| "eval_loss": 0.7925994396209717, | |
| "eval_runtime": 2.9067, | |
| "eval_samples_per_second": 6.881, | |
| "eval_steps_per_second": 1.032, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "grad_norm": 3.5985169410705566, | |
| "learning_rate": 2.302857142857143e-05, | |
| "loss": 0.6723, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_cer": 0.4025974025974026, | |
| "eval_loss": 0.8083222508430481, | |
| "eval_runtime": 2.8862, | |
| "eval_samples_per_second": 6.93, | |
| "eval_steps_per_second": 1.039, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "grad_norm": 3.7446236610412598, | |
| "learning_rate": 2.16e-05, | |
| "loss": 0.6736, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_cer": 0.42857142857142855, | |
| "eval_loss": 0.832375705242157, | |
| "eval_runtime": 2.9714, | |
| "eval_samples_per_second": 6.731, | |
| "eval_steps_per_second": 1.01, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "grad_norm": 3.926819324493408, | |
| "learning_rate": 2.0171428571428573e-05, | |
| "loss": 0.6646, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_cer": 0.45454545454545453, | |
| "eval_loss": 0.8589693903923035, | |
| "eval_runtime": 2.9745, | |
| "eval_samples_per_second": 6.724, | |
| "eval_steps_per_second": 1.009, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "grad_norm": 3.9272310733795166, | |
| "learning_rate": 1.8742857142857143e-05, | |
| "loss": 0.6586, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_cer": 0.36363636363636365, | |
| "eval_loss": 0.803028404712677, | |
| "eval_runtime": 2.9162, | |
| "eval_samples_per_second": 6.858, | |
| "eval_steps_per_second": 1.029, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "grad_norm": 3.406860113143921, | |
| "learning_rate": 1.7314285714285717e-05, | |
| "loss": 0.6665, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_cer": 0.4675324675324675, | |
| "eval_loss": 0.8490405082702637, | |
| "eval_runtime": 2.8831, | |
| "eval_samples_per_second": 6.937, | |
| "eval_steps_per_second": 1.041, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "grad_norm": 3.1726770401000977, | |
| "learning_rate": 1.5885714285714287e-05, | |
| "loss": 0.655, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_cer": 0.4675324675324675, | |
| "eval_loss": 0.8455414772033691, | |
| "eval_runtime": 2.9414, | |
| "eval_samples_per_second": 6.799, | |
| "eval_steps_per_second": 1.02, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "grad_norm": 3.506122589111328, | |
| "learning_rate": 1.4457142857142857e-05, | |
| "loss": 0.6641, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_cer": 0.42857142857142855, | |
| "eval_loss": 0.8133634924888611, | |
| "eval_runtime": 3.0374, | |
| "eval_samples_per_second": 6.584, | |
| "eval_steps_per_second": 0.988, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "grad_norm": 4.370969772338867, | |
| "learning_rate": 1.3028571428571429e-05, | |
| "loss": 0.6597, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_cer": 0.4675324675324675, | |
| "eval_loss": 0.8322270512580872, | |
| "eval_runtime": 2.8213, | |
| "eval_samples_per_second": 7.089, | |
| "eval_steps_per_second": 1.063, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "grad_norm": 3.851907253265381, | |
| "learning_rate": 1.16e-05, | |
| "loss": 0.6611, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_cer": 0.45454545454545453, | |
| "eval_loss": 0.8302900195121765, | |
| "eval_runtime": 2.8659, | |
| "eval_samples_per_second": 6.979, | |
| "eval_steps_per_second": 1.047, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "grad_norm": 2.989511013031006, | |
| "learning_rate": 1.0171428571428573e-05, | |
| "loss": 0.6617, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_cer": 0.45454545454545453, | |
| "eval_loss": 0.8202476501464844, | |
| "eval_runtime": 2.9062, | |
| "eval_samples_per_second": 6.882, | |
| "eval_steps_per_second": 1.032, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "grad_norm": 3.977627992630005, | |
| "learning_rate": 8.742857142857143e-06, | |
| "loss": 0.6531, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_cer": 0.45454545454545453, | |
| "eval_loss": 0.8274255394935608, | |
| "eval_runtime": 2.8627, | |
| "eval_samples_per_second": 6.987, | |
| "eval_steps_per_second": 1.048, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "grad_norm": 3.391641855239868, | |
| "learning_rate": 7.314285714285715e-06, | |
| "loss": 0.6491, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_cer": 0.42857142857142855, | |
| "eval_loss": 0.8236328959465027, | |
| "eval_runtime": 2.8678, | |
| "eval_samples_per_second": 6.974, | |
| "eval_steps_per_second": 1.046, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "grad_norm": 3.7944726943969727, | |
| "learning_rate": 5.885714285714286e-06, | |
| "loss": 0.6471, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_cer": 0.4155844155844156, | |
| "eval_loss": 0.8234376907348633, | |
| "eval_runtime": 2.9962, | |
| "eval_samples_per_second": 6.675, | |
| "eval_steps_per_second": 1.001, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "grad_norm": 4.810728073120117, | |
| "learning_rate": 4.457142857142858e-06, | |
| "loss": 0.6446, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_cer": 0.42857142857142855, | |
| "eval_loss": 0.83391934633255, | |
| "eval_runtime": 2.8811, | |
| "eval_samples_per_second": 6.942, | |
| "eval_steps_per_second": 1.041, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "grad_norm": 3.008390188217163, | |
| "learning_rate": 3.028571428571429e-06, | |
| "loss": 0.6439, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_cer": 0.4025974025974026, | |
| "eval_loss": 0.8176433444023132, | |
| "eval_runtime": 2.8474, | |
| "eval_samples_per_second": 7.024, | |
| "eval_steps_per_second": 1.054, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "grad_norm": 4.271653652191162, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 0.6453, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_cer": 0.4025974025974026, | |
| "eval_loss": 0.8156414031982422, | |
| "eval_runtime": 2.8852, | |
| "eval_samples_per_second": 6.932, | |
| "eval_steps_per_second": 1.04, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "grad_norm": 4.2927045822143555, | |
| "learning_rate": 1.7142857142857143e-07, | |
| "loss": 0.6404, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_cer": 0.35064935064935066, | |
| "eval_loss": 0.8252768516540527, | |
| "eval_runtime": 2.8782, | |
| "eval_samples_per_second": 6.949, | |
| "eval_steps_per_second": 1.042, | |
| "step": 875 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 875, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 35, | |
| "save_steps": 500, | |
| "total_flos": 5.237996336971776e+18, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |