| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 15.0, | |
| "global_step": 1035, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8e-06, | |
| "loss": 11.0428, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.8e-05, | |
| "loss": 11.5863, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.7e-05, | |
| "loss": 10.2859, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.7e-05, | |
| "loss": 7.7802, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.7000000000000004e-05, | |
| "loss": 5.4079, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 5.7e-05, | |
| "loss": 4.1036, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.7e-05, | |
| "loss": 3.9029, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.7e-05, | |
| "loss": 3.4366, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 8.7e-05, | |
| "loss": 3.3931, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 9.7e-05, | |
| "loss": 3.1944, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.000107, | |
| "loss": 3.043, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.00011700000000000001, | |
| "loss": 3.0573, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.000127, | |
| "loss": 2.9868, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 0.00013700000000000002, | |
| "loss": 2.9365, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 0.000147, | |
| "loss": 2.9301, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.000157, | |
| "loss": 2.9331, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 0.00016700000000000002, | |
| "loss": 2.9344, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 0.000177, | |
| "loss": 2.9127, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 0.000187, | |
| "loss": 2.9461, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 0.00019700000000000002, | |
| "loss": 2.9012, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 0.000207, | |
| "loss": 2.8907, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 0.00021700000000000002, | |
| "loss": 2.9403, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 0.00022700000000000002, | |
| "loss": 2.8832, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 0.000237, | |
| "loss": 2.9188, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 0.000247, | |
| "loss": 2.861, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 0.000257, | |
| "loss": 2.8522, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 0.00026700000000000004, | |
| "loss": 2.7281, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 0.000277, | |
| "loss": 2.5301, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 0.000287, | |
| "loss": 2.2375, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 0.000297, | |
| "loss": 1.7317, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 0.000307, | |
| "loss": 1.1891, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 0.000317, | |
| "loss": 0.9012, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 0.00032700000000000003, | |
| "loss": 0.8128, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 0.000337, | |
| "loss": 0.7676, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 0.000347, | |
| "loss": 0.8468, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 0.000357, | |
| "loss": 0.7011, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 0.000367, | |
| "loss": 0.5223, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 0.000377, | |
| "loss": 0.4903, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 0.00038700000000000003, | |
| "loss": 0.4989, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 0.00039700000000000005, | |
| "loss": 0.5719, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 0.00040699999999999997, | |
| "loss": 0.4448, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 0.000417, | |
| "loss": 0.4106, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 6.23, | |
| "learning_rate": 0.000427, | |
| "loss": 0.3305, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 0.000437, | |
| "loss": 0.3317, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 0.000447, | |
| "loss": 0.3787, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 0.00045700000000000005, | |
| "loss": 0.3884, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 0.000467, | |
| "loss": 0.3162, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 0.000477, | |
| "loss": 0.2973, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 0.000487, | |
| "loss": 0.2614, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 0.000497, | |
| "loss": 0.265, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "eval_loss": 0.48176705837249756, | |
| "eval_runtime": 597.1948, | |
| "eval_samples_per_second": 5.683, | |
| "eval_steps_per_second": 0.712, | |
| "eval_wer": 0.44399247261461594, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 0.0004934579439252336, | |
| "loss": 0.2784, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 7.54, | |
| "learning_rate": 0.00048411214953271025, | |
| "loss": 0.2536, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "learning_rate": 0.0004747663551401869, | |
| "loss": 0.2452, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "learning_rate": 0.00046542056074766355, | |
| "loss": 0.2731, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 0.00045607476635514015, | |
| "loss": 0.2524, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "learning_rate": 0.00044672897196261686, | |
| "loss": 0.2179, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 0.0004373831775700935, | |
| "loss": 0.1944, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 0.0004280373831775701, | |
| "loss": 0.1866, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 0.00041869158878504677, | |
| "loss": 0.179, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 8.7, | |
| "learning_rate": 0.0004093457943925234, | |
| "loss": 0.1778, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 0.0004, | |
| "loss": 0.1683, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 0.00039065420560747667, | |
| "loss": 0.1796, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 9.13, | |
| "learning_rate": 0.00038130841121495327, | |
| "loss": 0.17, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 0.0003719626168224299, | |
| "loss": 0.1274, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 9.42, | |
| "learning_rate": 0.0003626168224299066, | |
| "loss": 0.1498, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 9.57, | |
| "learning_rate": 0.0003532710280373832, | |
| "loss": 0.1164, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 9.71, | |
| "learning_rate": 0.00034392523364485983, | |
| "loss": 0.1308, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 9.86, | |
| "learning_rate": 0.0003345794392523365, | |
| "loss": 0.1364, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 0.0003252336448598131, | |
| "loss": 0.124, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "learning_rate": 0.00031588785046728974, | |
| "loss": 0.1068, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 10.29, | |
| "learning_rate": 0.0003065420560747664, | |
| "loss": 0.1209, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 10.43, | |
| "learning_rate": 0.000297196261682243, | |
| "loss": 0.1348, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 10.58, | |
| "learning_rate": 0.00028785046728971965, | |
| "loss": 0.1014, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 10.72, | |
| "learning_rate": 0.0002785046728971963, | |
| "loss": 0.0974, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 10.87, | |
| "learning_rate": 0.0002691588785046729, | |
| "loss": 0.097, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 0.00025981308411214955, | |
| "loss": 0.1105, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 11.16, | |
| "learning_rate": 0.0002504672897196262, | |
| "loss": 0.0849, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 11.3, | |
| "learning_rate": 0.0002411214953271028, | |
| "loss": 0.0777, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 11.45, | |
| "learning_rate": 0.00023177570093457946, | |
| "loss": 0.0741, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 11.59, | |
| "learning_rate": 0.00022242990654205608, | |
| "loss": 0.083, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 11.74, | |
| "learning_rate": 0.0002130841121495327, | |
| "loss": 0.0989, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "learning_rate": 0.00020373831775700934, | |
| "loss": 0.0944, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 0.000194392523364486, | |
| "loss": 0.0713, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 12.17, | |
| "learning_rate": 0.00018504672897196262, | |
| "loss": 0.0823, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 12.32, | |
| "learning_rate": 0.00017570093457943924, | |
| "loss": 0.0632, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 12.46, | |
| "learning_rate": 0.0001663551401869159, | |
| "loss": 0.0671, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 12.61, | |
| "learning_rate": 0.00015700934579439252, | |
| "loss": 0.0722, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 12.75, | |
| "learning_rate": 0.00014766355140186915, | |
| "loss": 0.0657, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 12.9, | |
| "learning_rate": 0.0001383177570093458, | |
| "loss": 0.0749, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 0.00012897196261682243, | |
| "loss": 0.0727, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 13.19, | |
| "learning_rate": 0.00011962616822429907, | |
| "loss": 0.0741, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 13.33, | |
| "learning_rate": 0.00011028037383177571, | |
| "loss": 0.0511, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 13.48, | |
| "learning_rate": 0.00010093457943925234, | |
| "loss": 0.0461, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 13.62, | |
| "learning_rate": 9.158878504672898e-05, | |
| "loss": 0.0517, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 13.77, | |
| "learning_rate": 8.224299065420562e-05, | |
| "loss": 0.0533, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 13.91, | |
| "learning_rate": 7.289719626168224e-05, | |
| "loss": 0.0577, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 14.06, | |
| "learning_rate": 6.355140186915888e-05, | |
| "loss": 0.0571, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 14.2, | |
| "learning_rate": 5.420560747663551e-05, | |
| "loss": 0.048, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 14.35, | |
| "learning_rate": 4.485981308411215e-05, | |
| "loss": 0.0542, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 14.49, | |
| "learning_rate": 3.551401869158879e-05, | |
| "loss": 0.0522, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 14.49, | |
| "eval_loss": 0.39019492268562317, | |
| "eval_runtime": 581.0044, | |
| "eval_samples_per_second": 5.842, | |
| "eval_steps_per_second": 0.731, | |
| "eval_wer": 0.2725472310562171, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 14.64, | |
| "learning_rate": 2.6168224299065423e-05, | |
| "loss": 0.0524, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 14.78, | |
| "learning_rate": 1.6822429906542056e-05, | |
| "loss": 0.0501, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 14.93, | |
| "learning_rate": 7.476635514018692e-06, | |
| "loss": 0.0425, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "step": 1035, | |
| "total_flos": 1.5036330849795359e+19, | |
| "train_loss": 1.3250362157965627, | |
| "train_runtime": 8304.6494, | |
| "train_samples_per_second": 3.963, | |
| "train_steps_per_second": 0.125 | |
| } | |
| ], | |
| "max_steps": 1035, | |
| "num_train_epochs": 15, | |
| "total_flos": 1.5036330849795359e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |