| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 14.996463098325867, | |
| "eval_steps": 2000, | |
| "global_step": 31800, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00014879999999999998, | |
| "loss": 14.5956, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0002985, | |
| "loss": 3.7568, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0002951785714285714, | |
| "loss": 1.8728, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.00029030844155844153, | |
| "loss": 1.5566, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "eval_cer": 0.2632472070390033, | |
| "eval_loss": 1.0225796699523926, | |
| "eval_runtime": 2668.5503, | |
| "eval_samples_per_second": 6.338, | |
| "eval_steps_per_second": 0.396, | |
| "eval_wer": 0.618374838459038, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00028543831168831166, | |
| "loss": 1.3765, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.0002805681818181818, | |
| "loss": 1.2884, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.0002757077922077922, | |
| "loss": 1.2363, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.0002708376623376623, | |
| "loss": 1.179, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "eval_cer": 0.20008986332921824, | |
| "eval_loss": 0.7681829333305359, | |
| "eval_runtime": 2604.8981, | |
| "eval_samples_per_second": 6.493, | |
| "eval_steps_per_second": 0.406, | |
| "eval_wer": 0.49895741661315, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.00026596753246753245, | |
| "loss": 1.1289, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 0.0002610974025974026, | |
| "loss": 1.0784, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 0.000256237012987013, | |
| "loss": 1.0664, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 0.0002513668831168831, | |
| "loss": 1.0432, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "eval_cer": 0.17490413851167133, | |
| "eval_loss": 0.6633431315422058, | |
| "eval_runtime": 2540.6608, | |
| "eval_samples_per_second": 6.657, | |
| "eval_steps_per_second": 0.416, | |
| "eval_wer": 0.45160906690768976, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 0.00024649675324675324, | |
| "loss": 1.0052, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 0.00024162662337662335, | |
| "loss": 0.9657, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 0.0002367564935064935, | |
| "loss": 0.9627, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 0.00023189610389610387, | |
| "loss": 0.9413, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "eval_cer": 0.16238652573537432, | |
| "eval_loss": 0.615895688533783, | |
| "eval_runtime": 2544.7169, | |
| "eval_samples_per_second": 6.647, | |
| "eval_steps_per_second": 0.416, | |
| "eval_wer": 0.4259210807982304, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 0.000227025974025974, | |
| "loss": 0.9393, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 0.00022215584415584413, | |
| "loss": 0.8932, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 0.00021728571428571427, | |
| "loss": 0.8898, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 0.00021242532467532466, | |
| "loss": 0.8765, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "eval_cer": 0.15377316925278206, | |
| "eval_loss": 0.5792471766471863, | |
| "eval_runtime": 2541.286, | |
| "eval_samples_per_second": 6.656, | |
| "eval_steps_per_second": 0.416, | |
| "eval_wer": 0.40610010386191914, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 0.0002075551948051948, | |
| "loss": 0.874, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 0.00020268506493506492, | |
| "loss": 0.8393, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 0.0001978246753246753, | |
| "loss": 0.8254, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 0.00019295454545454545, | |
| "loss": 0.8248, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "eval_cer": 0.14464288051230823, | |
| "eval_loss": 0.5455637574195862, | |
| "eval_runtime": 2603.5132, | |
| "eval_samples_per_second": 6.497, | |
| "eval_steps_per_second": 0.406, | |
| "eval_wer": 0.3876665953111497, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 0.00018808441558441555, | |
| "loss": 0.815, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 0.0001832142857142857, | |
| "loss": 0.7914, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 0.00017834415584415584, | |
| "loss": 0.7721, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 0.00017347402597402595, | |
| "loss": 0.7714, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "eval_cer": 0.1396851293639334, | |
| "eval_loss": 0.5316255688667297, | |
| "eval_runtime": 2697.2289, | |
| "eval_samples_per_second": 6.271, | |
| "eval_steps_per_second": 0.392, | |
| "eval_wer": 0.37104868824774634, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 6.84, | |
| "learning_rate": 0.00016861363636363634, | |
| "loss": 0.7775, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 0.00016375324675324673, | |
| "loss": 0.7576, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 0.0001588831168831169, | |
| "loss": 0.74, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 0.000154012987012987, | |
| "loss": 0.7388, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "eval_cer": 0.13560201887129913, | |
| "eval_loss": 0.5172015428543091, | |
| "eval_runtime": 2608.9857, | |
| "eval_samples_per_second": 6.483, | |
| "eval_steps_per_second": 0.406, | |
| "eval_wer": 0.3657208096472659, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 0.00014914285714285713, | |
| "loss": 0.7315, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 0.00014427272727272726, | |
| "loss": 0.725, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "learning_rate": 0.0001394025974025974, | |
| "loss": 0.6876, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 0.00013453246753246753, | |
| "loss": 0.6912, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "eval_cer": 0.12909128980051213, | |
| "eval_loss": 0.4891507625579834, | |
| "eval_runtime": 2622.6389, | |
| "eval_samples_per_second": 6.449, | |
| "eval_steps_per_second": 0.403, | |
| "eval_wer": 0.35079164981883626, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 0.00012966233766233766, | |
| "loss": 0.6912, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 0.00012480194805194805, | |
| "loss": 0.6847, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 0.00011993181818181817, | |
| "loss": 0.6578, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "learning_rate": 0.00011506168831168829, | |
| "loss": 0.6549, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "eval_cer": 0.12405937959404459, | |
| "eval_loss": 0.4693571627140045, | |
| "eval_runtime": 2599.1484, | |
| "eval_samples_per_second": 6.508, | |
| "eval_steps_per_second": 0.407, | |
| "eval_wer": 0.33972361629760006, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 9.67, | |
| "learning_rate": 0.00011019155844155843, | |
| "loss": 0.646, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "learning_rate": 0.00010533116883116881, | |
| "loss": 0.6467, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "learning_rate": 0.00010046103896103896, | |
| "loss": 0.6287, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 10.37, | |
| "learning_rate": 9.560064935064933e-05, | |
| "loss": 0.614, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 10.37, | |
| "eval_cer": 0.12051065054943137, | |
| "eval_loss": 0.461481511592865, | |
| "eval_runtime": 2605.4751, | |
| "eval_samples_per_second": 6.492, | |
| "eval_steps_per_second": 0.406, | |
| "eval_wer": 0.33093895931942696, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 10.61, | |
| "learning_rate": 9.073051948051948e-05, | |
| "loss": 0.6153, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 10.85, | |
| "learning_rate": 8.58603896103896e-05, | |
| "loss": 0.6157, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 11.08, | |
| "learning_rate": 8.099025974025973e-05, | |
| "loss": 0.6054, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 11.32, | |
| "learning_rate": 7.612987012987012e-05, | |
| "loss": 0.5901, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 11.32, | |
| "eval_cer": 0.1176620702591641, | |
| "eval_loss": 0.4488585889339447, | |
| "eval_runtime": 2619.3349, | |
| "eval_samples_per_second": 6.457, | |
| "eval_steps_per_second": 0.404, | |
| "eval_wer": 0.32152003107929183, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "learning_rate": 7.125974025974026e-05, | |
| "loss": 0.5808, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 11.79, | |
| "learning_rate": 6.638961038961039e-05, | |
| "loss": 0.5845, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 6.151948051948051e-05, | |
| "loss": 0.5777, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 12.26, | |
| "learning_rate": 5.664935064935064e-05, | |
| "loss": 0.555, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 12.26, | |
| "eval_cer": 0.11478513503492019, | |
| "eval_loss": 0.4419253170490265, | |
| "eval_runtime": 2591.0294, | |
| "eval_samples_per_second": 6.528, | |
| "eval_steps_per_second": 0.408, | |
| "eval_wer": 0.31628729316810567, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 5.178896103896104e-05, | |
| "loss": 0.5641, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 12.73, | |
| "learning_rate": 4.691883116883116e-05, | |
| "loss": 0.5552, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "learning_rate": 4.2048701298701296e-05, | |
| "loss": 0.5523, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 13.2, | |
| "learning_rate": 3.717857142857143e-05, | |
| "loss": 0.5377, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 13.2, | |
| "eval_cer": 0.11216557536523336, | |
| "eval_loss": 0.4320293366909027, | |
| "eval_runtime": 2509.0221, | |
| "eval_samples_per_second": 6.741, | |
| "eval_steps_per_second": 0.422, | |
| "eval_wer": 0.3102855013517906, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 13.44, | |
| "learning_rate": 3.231818181818181e-05, | |
| "loss": 0.5329, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 13.68, | |
| "learning_rate": 2.7448051948051945e-05, | |
| "loss": 0.533, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 13.91, | |
| "learning_rate": 2.2577922077922077e-05, | |
| "loss": 0.5289, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "learning_rate": 1.7707792207792207e-05, | |
| "loss": 0.5253, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "eval_cer": 0.11016764309426488, | |
| "eval_loss": 0.4250529706478119, | |
| "eval_runtime": 2522.5601, | |
| "eval_samples_per_second": 6.705, | |
| "eval_steps_per_second": 0.419, | |
| "eval_wer": 0.3051637609114478, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 14.38, | |
| "learning_rate": 1.2837662337662336e-05, | |
| "loss": 0.5196, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 14.62, | |
| "learning_rate": 7.967532467532467e-06, | |
| "loss": 0.5163, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 14.85, | |
| "learning_rate": 3.1071428571428566e-06, | |
| "loss": 0.5197, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "step": 31800, | |
| "total_flos": 2.1355952953477825e+20, | |
| "train_loss": 1.0558512441767087, | |
| "train_runtime": 76407.9259, | |
| "train_samples_per_second": 26.642, | |
| "train_steps_per_second": 0.416 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 31800, | |
| "num_train_epochs": 15, | |
| "save_steps": 2000, | |
| "total_flos": 2.1355952953477825e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |