| { | |
| "best_metric": 40.73694984646878, | |
| "best_model_checkpoint": "./checkpoint-5000", | |
| "epoch": 4.0576, | |
| "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": 4.2000000000000006e-07, | |
| "loss": 0.4829, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 0.4176, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.42e-06, | |
| "loss": 0.2673, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 0.2489, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.42e-06, | |
| "loss": 0.2574, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.92e-06, | |
| "loss": 0.2479, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.2551, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.243, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.2161, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.2012, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.2304, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.1823, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.2144, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.1998, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.1911, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.2113, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.2114, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.1856, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.2256, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.1621, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.953333333333333e-06, | |
| "loss": 0.2046, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.89777777777778e-06, | |
| "loss": 0.2214, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.842222222222223e-06, | |
| "loss": 0.2874, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.786666666666667e-06, | |
| "loss": 0.2412, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.731111111111113e-06, | |
| "loss": 0.2264, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.675555555555555e-06, | |
| "loss": 0.2179, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.620000000000001e-06, | |
| "loss": 0.2151, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.564444444444445e-06, | |
| "loss": 0.2377, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.508888888888889e-06, | |
| "loss": 0.2338, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.453333333333335e-06, | |
| "loss": 0.2416, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.397777777777779e-06, | |
| "loss": 0.163, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.342222222222223e-06, | |
| "loss": 0.1902, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.286666666666667e-06, | |
| "loss": 0.1751, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.231111111111111e-06, | |
| "loss": 0.288, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.175555555555557e-06, | |
| "loss": 0.2176, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.12e-06, | |
| "loss": 0.2118, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.064444444444447e-06, | |
| "loss": 0.2114, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.008888888888889e-06, | |
| "loss": 0.2073, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.953333333333335e-06, | |
| "loss": 0.2753, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.897777777777779e-06, | |
| "loss": 0.2575, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.22074507176876068, | |
| "eval_runtime": 735.5881, | |
| "eval_samples_per_second": 6.402, | |
| "eval_steps_per_second": 3.202, | |
| "eval_wer": 44.974411463664275, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.842222222222223e-06, | |
| "loss": 0.3282, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.786666666666668e-06, | |
| "loss": 0.1824, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.73111111111111e-06, | |
| "loss": 0.2677, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.675555555555556e-06, | |
| "loss": 0.2362, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.62e-06, | |
| "loss": 0.2008, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.564444444444445e-06, | |
| "loss": 0.2001, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.50888888888889e-06, | |
| "loss": 0.2316, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.453333333333334e-06, | |
| "loss": 0.1498, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.397777777777778e-06, | |
| "loss": 0.1824, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.342222222222222e-06, | |
| "loss": 0.1455, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.286666666666668e-06, | |
| "loss": 0.1347, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.231111111111112e-06, | |
| "loss": 0.1491, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.175555555555556e-06, | |
| "loss": 0.1125, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.120000000000002e-06, | |
| "loss": 0.1255, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.064444444444444e-06, | |
| "loss": 0.1271, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.00888888888889e-06, | |
| "loss": 0.0996, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.953333333333334e-06, | |
| "loss": 0.1004, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.897777777777778e-06, | |
| "loss": 0.1032, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.842222222222224e-06, | |
| "loss": 0.0872, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.786666666666666e-06, | |
| "loss": 0.0899, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.731111111111112e-06, | |
| "loss": 0.0784, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.675555555555556e-06, | |
| "loss": 0.0762, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.620000000000001e-06, | |
| "loss": 0.0655, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.564444444444446e-06, | |
| "loss": 0.0731, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.50888888888889e-06, | |
| "loss": 0.0729, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.453333333333334e-06, | |
| "loss": 0.0476, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.3977777777777786e-06, | |
| "loss": 0.0477, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.342222222222223e-06, | |
| "loss": 0.0535, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.2866666666666675e-06, | |
| "loss": 0.0702, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.231111111111112e-06, | |
| "loss": 0.0674, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.1755555555555556e-06, | |
| "loss": 0.0599, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.1200000000000004e-06, | |
| "loss": 0.0724, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.0644444444444445e-06, | |
| "loss": 0.0665, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.008888888888889e-06, | |
| "loss": 0.0726, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.953333333333334e-06, | |
| "loss": 0.0476, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.897777777777779e-06, | |
| "loss": 0.0971, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.842222222222222e-06, | |
| "loss": 0.0966, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.786666666666667e-06, | |
| "loss": 0.0457, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.731111111111111e-06, | |
| "loss": 0.0555, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.675555555555556e-06, | |
| "loss": 0.0529, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "eval_loss": 0.2098909169435501, | |
| "eval_runtime": 663.2982, | |
| "eval_samples_per_second": 7.099, | |
| "eval_steps_per_second": 3.55, | |
| "eval_wer": 41.31013306038895, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.620000000000001e-06, | |
| "loss": 0.0753, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.564444444444446e-06, | |
| "loss": 0.0592, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.508888888888889e-06, | |
| "loss": 0.0602, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.453333333333334e-06, | |
| "loss": 0.0534, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.397777777777778e-06, | |
| "loss": 0.0543, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.342222222222223e-06, | |
| "loss": 0.0583, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.286666666666668e-06, | |
| "loss": 0.0902, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.231111111111111e-06, | |
| "loss": 0.0735, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.175555555555556e-06, | |
| "loss": 0.07, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.120000000000001e-06, | |
| "loss": 0.0822, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.064444444444445e-06, | |
| "loss": 0.0666, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.00888888888889e-06, | |
| "loss": 0.0579, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.9533333333333345e-06, | |
| "loss": 0.0582, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.897777777777778e-06, | |
| "loss": 0.0566, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.8422222222222226e-06, | |
| "loss": 0.0506, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 5.7866666666666674e-06, | |
| "loss": 0.0397, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 5.7311111111111115e-06, | |
| "loss": 0.0329, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 5.675555555555556e-06, | |
| "loss": 0.045, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 5.620000000000001e-06, | |
| "loss": 0.0394, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 5.5644444444444444e-06, | |
| "loss": 0.0344, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 5.508888888888889e-06, | |
| "loss": 0.0298, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 5.453333333333334e-06, | |
| "loss": 0.0357, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 5.397777777777778e-06, | |
| "loss": 0.0286, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 5.342222222222223e-06, | |
| "loss": 0.0344, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 5.286666666666666e-06, | |
| "loss": 0.0278, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 5.231111111111111e-06, | |
| "loss": 0.0313, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 5.175555555555556e-06, | |
| "loss": 0.0313, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.12e-06, | |
| "loss": 0.0192, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.064444444444445e-06, | |
| "loss": 0.0203, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 5.00888888888889e-06, | |
| "loss": 0.026, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 4.953333333333334e-06, | |
| "loss": 0.0139, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.897777777777778e-06, | |
| "loss": 0.0169, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.842222222222223e-06, | |
| "loss": 0.0158, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.786666666666667e-06, | |
| "loss": 0.0247, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.731111111111112e-06, | |
| "loss": 0.014, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.675555555555556e-06, | |
| "loss": 0.0216, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.620000000000001e-06, | |
| "loss": 0.0146, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.564444444444445e-06, | |
| "loss": 0.0274, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.50888888888889e-06, | |
| "loss": 0.0224, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.453333333333334e-06, | |
| "loss": 0.0212, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "eval_loss": 0.21833518147468567, | |
| "eval_runtime": 659.3954, | |
| "eval_samples_per_second": 7.141, | |
| "eval_steps_per_second": 3.571, | |
| "eval_wer": 41.55578300921187, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.397777777777778e-06, | |
| "loss": 0.0204, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.3422222222222225e-06, | |
| "loss": 0.0192, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.2866666666666666e-06, | |
| "loss": 0.0303, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.2311111111111114e-06, | |
| "loss": 0.016, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.175555555555556e-06, | |
| "loss": 0.0176, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.12e-06, | |
| "loss": 0.0197, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.064444444444444e-06, | |
| "loss": 0.0224, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.008888888888889e-06, | |
| "loss": 0.0155, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.953333333333333e-06, | |
| "loss": 0.0233, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.897777777777778e-06, | |
| "loss": 0.0163, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.842222222222223e-06, | |
| "loss": 0.0274, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.7866666666666667e-06, | |
| "loss": 0.0235, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.7311111111111116e-06, | |
| "loss": 0.0169, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.675555555555556e-06, | |
| "loss": 0.0195, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.62e-06, | |
| "loss": 0.0217, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.564444444444445e-06, | |
| "loss": 0.02, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.508888888888889e-06, | |
| "loss": 0.0205, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 3.4533333333333334e-06, | |
| "loss": 0.0199, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 3.3977777777777783e-06, | |
| "loss": 0.0148, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.3422222222222224e-06, | |
| "loss": 0.0159, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.286666666666667e-06, | |
| "loss": 0.022, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.2311111111111117e-06, | |
| "loss": 0.0173, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.1755555555555557e-06, | |
| "loss": 0.0127, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.12e-06, | |
| "loss": 0.0103, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 3.064444444444445e-06, | |
| "loss": 0.0087, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 3.008888888888889e-06, | |
| "loss": 0.0089, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 2.9533333333333336e-06, | |
| "loss": 0.0098, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 2.8977777777777785e-06, | |
| "loss": 0.0095, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 2.8422222222222225e-06, | |
| "loss": 0.0092, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 2.786666666666667e-06, | |
| "loss": 0.0091, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 2.7311111111111114e-06, | |
| "loss": 0.0078, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 2.675555555555556e-06, | |
| "loss": 0.0069, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 2.6200000000000003e-06, | |
| "loss": 0.0083, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 2.5644444444444444e-06, | |
| "loss": 0.0064, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 2.5088888888888892e-06, | |
| "loss": 0.0096, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 2.4533333333333333e-06, | |
| "loss": 0.0036, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 2.397777777777778e-06, | |
| "loss": 0.0066, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 2.342222222222222e-06, | |
| "loss": 0.0055, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.2866666666666667e-06, | |
| "loss": 0.0033, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.2311111111111115e-06, | |
| "loss": 0.0041, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "eval_loss": 0.21509672701358795, | |
| "eval_runtime": 664.1487, | |
| "eval_samples_per_second": 7.09, | |
| "eval_steps_per_second": 3.546, | |
| "eval_wer": 41.49437052200614, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 2.1755555555555556e-06, | |
| "loss": 0.0053, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 2.12e-06, | |
| "loss": 0.0026, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.064444444444445e-06, | |
| "loss": 0.0036, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.008888888888889e-06, | |
| "loss": 0.0082, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 1.9533333333333334e-06, | |
| "loss": 0.003, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 1.8977777777777779e-06, | |
| "loss": 0.0073, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 1.8422222222222225e-06, | |
| "loss": 0.0049, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 1.7866666666666668e-06, | |
| "loss": 0.0071, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 1.7311111111111112e-06, | |
| "loss": 0.0078, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 1.675555555555556e-06, | |
| "loss": 0.0093, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 1.6200000000000002e-06, | |
| "loss": 0.0064, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 1.5644444444444446e-06, | |
| "loss": 0.0064, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 1.5088888888888889e-06, | |
| "loss": 0.0042, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 1.4533333333333335e-06, | |
| "loss": 0.0058, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 1.397777777777778e-06, | |
| "loss": 0.0055, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 1.3422222222222222e-06, | |
| "loss": 0.0037, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.286666666666667e-06, | |
| "loss": 0.0061, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.2311111111111112e-06, | |
| "loss": 0.0034, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.1755555555555556e-06, | |
| "loss": 0.0055, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.12e-06, | |
| "loss": 0.0046, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 1.0644444444444445e-06, | |
| "loss": 0.0061, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 1.008888888888889e-06, | |
| "loss": 0.0079, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 9.533333333333335e-07, | |
| "loss": 0.0093, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 8.977777777777778e-07, | |
| "loss": 0.0042, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 8.422222222222224e-07, | |
| "loss": 0.0073, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 7.866666666666667e-07, | |
| "loss": 0.0056, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 7.311111111111112e-07, | |
| "loss": 0.003, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 6.755555555555555e-07, | |
| "loss": 0.0042, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 6.200000000000001e-07, | |
| "loss": 0.0026, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.644444444444445e-07, | |
| "loss": 0.0033, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.088888888888889e-07, | |
| "loss": 0.0027, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 4.533333333333334e-07, | |
| "loss": 0.0041, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.9777777777777783e-07, | |
| "loss": 0.0045, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.422222222222223e-07, | |
| "loss": 0.0041, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 2.866666666666667e-07, | |
| "loss": 0.0018, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 2.3111111111111112e-07, | |
| "loss": 0.002, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.7555555555555558e-07, | |
| "loss": 0.0026, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 0.0021, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 6.444444444444445e-08, | |
| "loss": 0.002, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 8.88888888888889e-09, | |
| "loss": 0.003, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "eval_loss": 0.21414214372634888, | |
| "eval_runtime": 675.2414, | |
| "eval_samples_per_second": 6.974, | |
| "eval_steps_per_second": 3.488, | |
| "eval_wer": 40.73694984646878, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "step": 5000, | |
| "total_flos": 1.153648996319232e+19, | |
| "train_loss": 0.08178634889945388, | |
| "train_runtime": 10593.3195, | |
| "train_samples_per_second": 3.776, | |
| "train_steps_per_second": 0.472 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 1.153648996319232e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |