{ "best_metric": 17.275280898876407, "best_model_checkpoint": "./checkpoint-2000", "epoch": 7.1138, "global_step": 5000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 9.200000000000001e-07, "loss": 1.0761, "step": 50 }, { "epoch": 0.02, "learning_rate": 1.9200000000000003e-06, "loss": 0.5378, "step": 100 }, { "epoch": 0.03, "learning_rate": 2.92e-06, "loss": 0.2491, "step": 150 }, { "epoch": 0.04, "learning_rate": 3.920000000000001e-06, "loss": 0.2272, "step": 200 }, { "epoch": 0.05, "learning_rate": 4.92e-06, "loss": 0.2071, "step": 250 }, { "epoch": 0.06, "learning_rate": 5.92e-06, "loss": 0.1902, "step": 300 }, { "epoch": 0.07, "learning_rate": 6.92e-06, "loss": 0.1945, "step": 350 }, { "epoch": 0.08, "learning_rate": 7.92e-06, "loss": 0.2003, "step": 400 }, { "epoch": 0.09, "learning_rate": 8.920000000000001e-06, "loss": 0.2314, "step": 450 }, { "epoch": 0.1, "learning_rate": 9.920000000000002e-06, "loss": 0.2152, "step": 500 }, { "epoch": 0.11, "learning_rate": 9.89777777777778e-06, "loss": 0.2208, "step": 550 }, { "epoch": 0.12, "learning_rate": 9.786666666666667e-06, "loss": 0.1836, "step": 600 }, { "epoch": 1.0, "learning_rate": 9.675555555555555e-06, "loss": 0.1966, "step": 650 }, { "epoch": 1.01, "learning_rate": 9.564444444444445e-06, "loss": 0.213, "step": 700 }, { "epoch": 1.02, "learning_rate": 9.453333333333335e-06, "loss": 0.1976, "step": 750 }, { "epoch": 1.03, "learning_rate": 9.342222222222223e-06, "loss": 0.1401, "step": 800 }, { "epoch": 1.04, "learning_rate": 9.231111111111111e-06, "loss": 0.1304, "step": 850 }, { "epoch": 1.05, "learning_rate": 9.12e-06, "loss": 0.1214, "step": 900 }, { "epoch": 1.06, "learning_rate": 9.008888888888889e-06, "loss": 0.1042, "step": 950 }, { "epoch": 1.07, "learning_rate": 8.897777777777779e-06, "loss": 0.1044, "step": 1000 }, { "epoch": 1.07, "eval_cer": 4.880950328442613, "eval_loss": 0.2777043581008911, "eval_runtime": 3506.453, "eval_samples_per_second": 2.893, "eval_steps_per_second": 0.181, "eval_wer": 18.404570300637715, "step": 1000 }, { "epoch": 1.08, "learning_rate": 8.786666666666668e-06, "loss": 0.1055, "step": 1050 }, { "epoch": 1.09, "learning_rate": 8.675555555555556e-06, "loss": 0.1061, "step": 1100 }, { "epoch": 1.1, "learning_rate": 8.564444444444445e-06, "loss": 0.0949, "step": 1150 }, { "epoch": 1.11, "learning_rate": 8.453333333333334e-06, "loss": 0.0949, "step": 1200 }, { "epoch": 1.12, "learning_rate": 8.342222222222222e-06, "loss": 0.0825, "step": 1250 }, { "epoch": 2.01, "learning_rate": 8.231111111111112e-06, "loss": 0.0938, "step": 1300 }, { "epoch": 2.02, "learning_rate": 8.120000000000002e-06, "loss": 0.0968, "step": 1350 }, { "epoch": 2.03, "learning_rate": 8.00888888888889e-06, "loss": 0.0798, "step": 1400 }, { "epoch": 2.04, "learning_rate": 7.897777777777778e-06, "loss": 0.0693, "step": 1450 }, { "epoch": 2.05, "learning_rate": 7.786666666666666e-06, "loss": 0.0628, "step": 1500 }, { "epoch": 2.06, "learning_rate": 7.675555555555556e-06, "loss": 0.0568, "step": 1550 }, { "epoch": 2.07, "learning_rate": 7.564444444444446e-06, "loss": 0.0566, "step": 1600 }, { "epoch": 2.08, "learning_rate": 7.453333333333334e-06, "loss": 0.057, "step": 1650 }, { "epoch": 2.09, "learning_rate": 7.342222222222223e-06, "loss": 0.0557, "step": 1700 }, { "epoch": 2.1, "learning_rate": 7.231111111111112e-06, "loss": 0.0507, "step": 1750 }, { "epoch": 2.11, "learning_rate": 7.1200000000000004e-06, "loss": 0.0427, "step": 1800 }, { "epoch": 2.12, "learning_rate": 7.008888888888889e-06, "loss": 0.0463, "step": 1850 }, { "epoch": 3.0, "learning_rate": 6.897777777777779e-06, "loss": 0.0453, "step": 1900 }, { "epoch": 3.01, "learning_rate": 6.786666666666667e-06, "loss": 0.0475, "step": 1950 }, { "epoch": 3.02, "learning_rate": 6.675555555555556e-06, "loss": 0.0469, "step": 2000 }, { "epoch": 3.02, "eval_cer": 4.533542525128877, "eval_loss": 0.27987292408943176, "eval_runtime": 3523.0006, "eval_samples_per_second": 2.879, "eval_steps_per_second": 0.18, "eval_wer": 17.275280898876407, "step": 2000 }, { "epoch": 3.03, "learning_rate": 6.564444444444446e-06, "loss": 0.0393, "step": 2050 }, { "epoch": 3.04, "learning_rate": 6.453333333333334e-06, "loss": 0.0359, "step": 2100 }, { "epoch": 3.05, "learning_rate": 6.342222222222223e-06, "loss": 0.0341, "step": 2150 }, { "epoch": 3.06, "learning_rate": 6.231111111111111e-06, "loss": 0.0285, "step": 2200 }, { "epoch": 3.07, "learning_rate": 6.120000000000001e-06, "loss": 0.034, "step": 2250 }, { "epoch": 3.08, "learning_rate": 6.00888888888889e-06, "loss": 0.0308, "step": 2300 }, { "epoch": 3.09, "learning_rate": 5.897777777777778e-06, "loss": 0.0291, "step": 2350 }, { "epoch": 3.1, "learning_rate": 5.7866666666666674e-06, "loss": 0.0248, "step": 2400 }, { "epoch": 3.11, "learning_rate": 5.675555555555556e-06, "loss": 0.0236, "step": 2450 }, { "epoch": 3.12, "learning_rate": 5.5644444444444444e-06, "loss": 0.0238, "step": 2500 }, { "epoch": 4.0, "learning_rate": 5.453333333333334e-06, "loss": 0.0249, "step": 2550 }, { "epoch": 4.01, "learning_rate": 5.342222222222223e-06, "loss": 0.0251, "step": 2600 }, { "epoch": 4.02, "learning_rate": 5.231111111111111e-06, "loss": 0.0235, "step": 2650 }, { "epoch": 4.03, "learning_rate": 5.12e-06, "loss": 0.0196, "step": 2700 }, { "epoch": 4.04, "learning_rate": 5.00888888888889e-06, "loss": 0.016, "step": 2750 }, { "epoch": 4.05, "learning_rate": 4.897777777777778e-06, "loss": 0.0153, "step": 2800 }, { "epoch": 4.06, "learning_rate": 4.786666666666667e-06, "loss": 0.0147, "step": 2850 }, { "epoch": 4.07, "learning_rate": 4.675555555555556e-06, "loss": 0.0169, "step": 2900 }, { "epoch": 4.08, "learning_rate": 4.564444444444445e-06, "loss": 0.0119, "step": 2950 }, { "epoch": 4.09, "learning_rate": 4.453333333333334e-06, "loss": 0.014, "step": 3000 }, { "epoch": 4.09, "eval_cer": 4.903938437843026, "eval_loss": 0.32019320130348206, "eval_runtime": 3597.0524, "eval_samples_per_second": 2.82, "eval_steps_per_second": 0.176, "eval_wer": 18.080018220467657, "step": 3000 }, { "epoch": 4.1, "learning_rate": 4.3422222222222225e-06, "loss": 0.0096, "step": 3050 }, { "epoch": 4.11, "learning_rate": 4.2311111111111114e-06, "loss": 0.0122, "step": 3100 }, { "epoch": 4.12, "learning_rate": 4.12e-06, "loss": 0.0087, "step": 3150 }, { "epoch": 5.01, "learning_rate": 4.008888888888889e-06, "loss": 0.0112, "step": 3200 }, { "epoch": 5.02, "learning_rate": 3.897777777777778e-06, "loss": 0.011, "step": 3250 }, { "epoch": 5.03, "learning_rate": 3.7866666666666667e-06, "loss": 0.0096, "step": 3300 }, { "epoch": 5.04, "learning_rate": 3.675555555555556e-06, "loss": 0.0081, "step": 3350 }, { "epoch": 5.05, "learning_rate": 3.564444444444445e-06, "loss": 0.0063, "step": 3400 }, { "epoch": 5.06, "learning_rate": 3.4533333333333334e-06, "loss": 0.0081, "step": 3450 }, { "epoch": 5.07, "learning_rate": 3.3422222222222224e-06, "loss": 0.0065, "step": 3500 }, { "epoch": 5.08, "learning_rate": 3.2311111111111117e-06, "loss": 0.0068, "step": 3550 }, { "epoch": 5.09, "learning_rate": 3.12e-06, "loss": 0.0054, "step": 3600 }, { "epoch": 5.1, "learning_rate": 3.008888888888889e-06, "loss": 0.0059, "step": 3650 }, { "epoch": 5.11, "learning_rate": 2.8977777777777785e-06, "loss": 0.0045, "step": 3700 }, { "epoch": 5.12, "learning_rate": 2.786666666666667e-06, "loss": 0.0053, "step": 3750 }, { "epoch": 6.0, "learning_rate": 2.675555555555556e-06, "loss": 0.0045, "step": 3800 }, { "epoch": 6.01, "learning_rate": 2.5644444444444444e-06, "loss": 0.0044, "step": 3850 }, { "epoch": 6.02, "learning_rate": 2.4533333333333333e-06, "loss": 0.005, "step": 3900 }, { "epoch": 6.03, "learning_rate": 2.342222222222222e-06, "loss": 0.0049, "step": 3950 }, { "epoch": 6.04, "learning_rate": 2.2311111111111115e-06, "loss": 0.0039, "step": 4000 }, { "epoch": 6.04, "eval_cer": 5.019166336212594, "eval_loss": 0.33260419964790344, "eval_runtime": 3639.8267, "eval_samples_per_second": 2.787, "eval_steps_per_second": 0.174, "eval_wer": 18.296386273914365, "step": 4000 }, { "epoch": 6.05, "learning_rate": 2.12e-06, "loss": 0.0036, "step": 4050 }, { "epoch": 6.06, "learning_rate": 2.008888888888889e-06, "loss": 0.003, "step": 4100 }, { "epoch": 6.07, "learning_rate": 1.8977777777777779e-06, "loss": 0.0028, "step": 4150 }, { "epoch": 6.08, "learning_rate": 1.7866666666666668e-06, "loss": 0.0029, "step": 4200 }, { "epoch": 6.09, "learning_rate": 1.675555555555556e-06, "loss": 0.0033, "step": 4250 }, { "epoch": 6.1, "learning_rate": 1.5644444444444446e-06, "loss": 0.0026, "step": 4300 }, { "epoch": 6.11, "learning_rate": 1.4533333333333335e-06, "loss": 0.0027, "step": 4350 }, { "epoch": 6.12, "learning_rate": 1.3422222222222222e-06, "loss": 0.0024, "step": 4400 }, { "epoch": 7.0, "learning_rate": 1.2311111111111112e-06, "loss": 0.0029, "step": 4450 }, { "epoch": 7.01, "learning_rate": 1.12e-06, "loss": 0.0029, "step": 4500 }, { "epoch": 7.02, "learning_rate": 1.008888888888889e-06, "loss": 0.0028, "step": 4550 }, { "epoch": 7.03, "learning_rate": 8.977777777777778e-07, "loss": 0.0022, "step": 4600 }, { "epoch": 7.04, "learning_rate": 7.866666666666667e-07, "loss": 0.0023, "step": 4650 }, { "epoch": 7.05, "learning_rate": 6.755555555555555e-07, "loss": 0.0021, "step": 4700 }, { "epoch": 7.06, "learning_rate": 5.644444444444445e-07, "loss": 0.0018, "step": 4750 }, { "epoch": 7.07, "learning_rate": 4.533333333333334e-07, "loss": 0.0018, "step": 4800 }, { "epoch": 7.08, "learning_rate": 3.422222222222223e-07, "loss": 0.0021, "step": 4850 }, { "epoch": 7.09, "learning_rate": 2.3111111111111112e-07, "loss": 0.0021, "step": 4900 }, { "epoch": 7.1, "learning_rate": 1.2000000000000002e-07, "loss": 0.002, "step": 4950 }, { "epoch": 7.11, "learning_rate": 8.88888888888889e-09, "loss": 0.0022, "step": 5000 }, { "epoch": 7.11, "eval_cer": 4.947041142968799, "eval_loss": 0.34528660774230957, "eval_runtime": 3568.6089, "eval_samples_per_second": 2.842, "eval_steps_per_second": 0.178, "eval_wer": 18.03067112055876, "step": 5000 }, { "epoch": 7.11, "step": 5000, "total_flos": 4.616962385559552e+19, "train_loss": 0.07011192187219858, "train_runtime": 43131.9488, "train_samples_per_second": 3.71, "train_steps_per_second": 0.116 } ], "max_steps": 5000, "num_train_epochs": 9223372036854775807, "total_flos": 4.616962385559552e+19, "trial_name": null, "trial_params": null }