| { | |
| "best_metric": 35.43550690147549, | |
| "best_model_checkpoint": "./checkpoint-3000", | |
| "epoch": 106.0036, | |
| "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": 2.6504, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 1.9914, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.42e-06, | |
| "loss": 1.0697, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 0.7023, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.42e-06, | |
| "loss": 0.5324, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 2.92e-06, | |
| "loss": 0.45, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.338, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.2858, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.2132, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.1557, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.1146, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.0738, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.0512, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.0324, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.0261, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.0188, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.0163, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.0139, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.0106, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.0106, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 9.953333333333333e-06, | |
| "loss": 0.0105, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 9.89777777777778e-06, | |
| "loss": 0.0101, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 9.842222222222223e-06, | |
| "loss": 0.0072, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 9.786666666666667e-06, | |
| "loss": 0.008, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 9.731111111111113e-06, | |
| "loss": 0.0057, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 9.675555555555555e-06, | |
| "loss": 0.0063, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 9.620000000000001e-06, | |
| "loss": 0.0035, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 9.564444444444445e-06, | |
| "loss": 0.0038, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 9.508888888888889e-06, | |
| "loss": 0.0054, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 9.453333333333335e-06, | |
| "loss": 0.0042, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 9.397777777777779e-06, | |
| "loss": 0.0035, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 9.342222222222223e-06, | |
| "loss": 0.0034, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 9.286666666666667e-06, | |
| "loss": 0.0032, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 9.231111111111111e-06, | |
| "loss": 0.0027, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 9.175555555555557e-06, | |
| "loss": 0.0018, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 9.12e-06, | |
| "loss": 0.0017, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 9.064444444444447e-06, | |
| "loss": 0.0022, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 9.008888888888889e-06, | |
| "loss": 0.0018, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 8.953333333333335e-06, | |
| "loss": 0.0013, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "learning_rate": 8.897777777777779e-06, | |
| "loss": 0.001, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_loss": 0.6506679058074951, | |
| "eval_runtime": 812.1587, | |
| "eval_samples_per_second": 2.791, | |
| "eval_steps_per_second": 0.087, | |
| "eval_wer": 37.32746311280343, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 8.842222222222223e-06, | |
| "loss": 0.001, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "learning_rate": 8.786666666666668e-06, | |
| "loss": 0.0009, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 8.73111111111111e-06, | |
| "loss": 0.0008, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "learning_rate": 8.675555555555556e-06, | |
| "loss": 0.0008, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 23.01, | |
| "learning_rate": 8.62e-06, | |
| "loss": 0.0007, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "learning_rate": 8.564444444444445e-06, | |
| "loss": 0.0007, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "learning_rate": 8.50888888888889e-06, | |
| "loss": 0.0006, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 8.453333333333334e-06, | |
| "loss": 0.0006, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "learning_rate": 8.397777777777778e-06, | |
| "loss": 0.0006, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 26.01, | |
| "learning_rate": 8.342222222222222e-06, | |
| "loss": 0.0006, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "learning_rate": 8.286666666666668e-06, | |
| "loss": 0.0006, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 27.01, | |
| "learning_rate": 8.231111111111112e-06, | |
| "loss": 0.0005, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "learning_rate": 8.175555555555556e-06, | |
| "loss": 0.0005, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 28.01, | |
| "learning_rate": 8.120000000000002e-06, | |
| "loss": 0.0005, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "learning_rate": 8.064444444444444e-06, | |
| "loss": 0.0005, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 29.01, | |
| "learning_rate": 8.00888888888889e-06, | |
| "loss": 0.0005, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "learning_rate": 7.953333333333334e-06, | |
| "loss": 0.0005, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 30.01, | |
| "learning_rate": 7.897777777777778e-06, | |
| "loss": 0.0005, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "learning_rate": 7.842222222222224e-06, | |
| "loss": 0.0004, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 31.01, | |
| "learning_rate": 7.786666666666666e-06, | |
| "loss": 0.0004, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "learning_rate": 7.731111111111112e-06, | |
| "loss": 0.0004, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 32.01, | |
| "learning_rate": 7.675555555555556e-06, | |
| "loss": 0.0004, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "learning_rate": 7.620000000000001e-06, | |
| "loss": 0.0004, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "learning_rate": 7.564444444444446e-06, | |
| "loss": 0.0004, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 34.01, | |
| "learning_rate": 7.50888888888889e-06, | |
| "loss": 0.0004, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "learning_rate": 7.453333333333334e-06, | |
| "loss": 0.0004, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 35.01, | |
| "learning_rate": 7.3977777777777786e-06, | |
| "loss": 0.0004, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "learning_rate": 7.342222222222223e-06, | |
| "loss": 0.0004, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 36.01, | |
| "learning_rate": 7.2866666666666675e-06, | |
| "loss": 0.0004, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "learning_rate": 7.231111111111112e-06, | |
| "loss": 0.0003, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 37.01, | |
| "learning_rate": 7.1755555555555556e-06, | |
| "loss": 0.0003, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "learning_rate": 7.1200000000000004e-06, | |
| "loss": 0.0003, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 38.01, | |
| "learning_rate": 7.0644444444444445e-06, | |
| "loss": 0.0003, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "learning_rate": 7.008888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 39.01, | |
| "learning_rate": 6.953333333333334e-06, | |
| "loss": 0.0003, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 6.897777777777779e-06, | |
| "loss": 0.0003, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 40.01, | |
| "learning_rate": 6.842222222222222e-06, | |
| "loss": 0.0003, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "learning_rate": 6.786666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "learning_rate": 6.731111111111111e-06, | |
| "loss": 0.0003, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 42.01, | |
| "learning_rate": 6.675555555555556e-06, | |
| "loss": 0.0003, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 42.01, | |
| "eval_loss": 0.695440948009491, | |
| "eval_runtime": 776.4908, | |
| "eval_samples_per_second": 2.92, | |
| "eval_steps_per_second": 0.091, | |
| "eval_wer": 36.11375535459305, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "learning_rate": 6.620000000000001e-06, | |
| "loss": 0.0003, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 43.01, | |
| "learning_rate": 6.564444444444446e-06, | |
| "loss": 0.0003, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "learning_rate": 6.508888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 44.01, | |
| "learning_rate": 6.453333333333334e-06, | |
| "loss": 0.0003, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "learning_rate": 6.397777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 45.01, | |
| "learning_rate": 6.342222222222223e-06, | |
| "loss": 0.0003, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "learning_rate": 6.286666666666668e-06, | |
| "loss": 0.0003, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 46.01, | |
| "learning_rate": 6.231111111111111e-06, | |
| "loss": 0.0002, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "learning_rate": 6.175555555555556e-06, | |
| "loss": 0.0002, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 47.01, | |
| "learning_rate": 6.120000000000001e-06, | |
| "loss": 0.0002, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "learning_rate": 6.064444444444445e-06, | |
| "loss": 0.0002, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 48.01, | |
| "learning_rate": 6.00888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "learning_rate": 5.9533333333333345e-06, | |
| "loss": 0.0002, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 49.01, | |
| "learning_rate": 5.897777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 50.01, | |
| "learning_rate": 5.8422222222222226e-06, | |
| "loss": 0.0002, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "learning_rate": 5.7866666666666674e-06, | |
| "loss": 0.0002, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 51.01, | |
| "learning_rate": 5.7311111111111115e-06, | |
| "loss": 0.0002, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "learning_rate": 5.675555555555556e-06, | |
| "loss": 0.0002, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 52.01, | |
| "learning_rate": 5.620000000000001e-06, | |
| "loss": 0.0002, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "learning_rate": 5.5644444444444444e-06, | |
| "loss": 0.0002, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 53.01, | |
| "learning_rate": 5.508888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "learning_rate": 5.453333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 54.01, | |
| "learning_rate": 5.397777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "learning_rate": 5.342222222222223e-06, | |
| "loss": 0.0002, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 55.01, | |
| "learning_rate": 5.286666666666666e-06, | |
| "loss": 0.0002, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "learning_rate": 5.231111111111111e-06, | |
| "loss": 0.0002, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 56.01, | |
| "learning_rate": 5.175555555555556e-06, | |
| "loss": 0.0002, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "learning_rate": 5.12e-06, | |
| "loss": 0.0002, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 57.01, | |
| "learning_rate": 5.064444444444445e-06, | |
| "loss": 0.0002, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "learning_rate": 5.00888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "learning_rate": 4.953333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 59.01, | |
| "learning_rate": 4.897777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 4.842222222222223e-06, | |
| "loss": 0.0002, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 60.01, | |
| "learning_rate": 4.786666666666667e-06, | |
| "loss": 0.0002, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "learning_rate": 4.731111111111112e-06, | |
| "loss": 0.0002, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 61.01, | |
| "learning_rate": 4.675555555555556e-06, | |
| "loss": 0.0002, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "learning_rate": 4.620000000000001e-06, | |
| "loss": 0.0002, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 62.01, | |
| "learning_rate": 4.564444444444445e-06, | |
| "loss": 0.0002, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "learning_rate": 4.50888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 63.01, | |
| "learning_rate": 4.453333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 63.01, | |
| "eval_loss": 0.7222780585289001, | |
| "eval_runtime": 913.8875, | |
| "eval_samples_per_second": 2.481, | |
| "eval_steps_per_second": 0.078, | |
| "eval_wer": 35.43550690147549, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "learning_rate": 4.397777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 64.01, | |
| "learning_rate": 4.3422222222222225e-06, | |
| "loss": 0.0002, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "learning_rate": 4.2866666666666666e-06, | |
| "loss": 0.0002, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 65.01, | |
| "learning_rate": 4.2311111111111114e-06, | |
| "loss": 0.0002, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "learning_rate": 4.175555555555556e-06, | |
| "loss": 0.0002, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "learning_rate": 4.12e-06, | |
| "loss": 0.0002, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 67.01, | |
| "learning_rate": 4.064444444444444e-06, | |
| "loss": 0.0002, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "learning_rate": 4.008888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 68.01, | |
| "learning_rate": 3.953333333333333e-06, | |
| "loss": 0.0001, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "learning_rate": 3.897777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 69.01, | |
| "learning_rate": 3.842222222222223e-06, | |
| "loss": 0.0001, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "learning_rate": 3.7866666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 70.01, | |
| "learning_rate": 3.7311111111111116e-06, | |
| "loss": 0.0001, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "learning_rate": 3.675555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 71.01, | |
| "learning_rate": 3.62e-06, | |
| "loss": 0.0001, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "learning_rate": 3.564444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 72.01, | |
| "learning_rate": 3.508888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "learning_rate": 3.4533333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 73.01, | |
| "learning_rate": 3.3977777777777783e-06, | |
| "loss": 0.0001, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "learning_rate": 3.3422222222222224e-06, | |
| "loss": 0.0001, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 74.01, | |
| "learning_rate": 3.286666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "learning_rate": 3.2311111111111117e-06, | |
| "loss": 0.0001, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "learning_rate": 3.1755555555555557e-06, | |
| "loss": 0.0001, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 76.01, | |
| "learning_rate": 3.12e-06, | |
| "loss": 0.0001, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "learning_rate": 3.064444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 77.01, | |
| "learning_rate": 3.008888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "learning_rate": 2.9533333333333336e-06, | |
| "loss": 0.0001, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 78.01, | |
| "learning_rate": 2.8977777777777785e-06, | |
| "loss": 0.0001, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "learning_rate": 2.8422222222222225e-06, | |
| "loss": 0.0001, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 79.01, | |
| "learning_rate": 2.786666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 2.7311111111111114e-06, | |
| "loss": 0.0001, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 80.01, | |
| "learning_rate": 2.675555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 81.0, | |
| "learning_rate": 2.6200000000000003e-06, | |
| "loss": 0.0001, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 81.01, | |
| "learning_rate": 2.5644444444444444e-06, | |
| "loss": 0.0001, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 82.0, | |
| "learning_rate": 2.5088888888888892e-06, | |
| "loss": 0.0001, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 82.01, | |
| "learning_rate": 2.4533333333333333e-06, | |
| "loss": 0.0001, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 83.0, | |
| "learning_rate": 2.397777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 84.0, | |
| "learning_rate": 2.342222222222222e-06, | |
| "loss": 0.0001, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 84.01, | |
| "learning_rate": 2.2866666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 85.0, | |
| "learning_rate": 2.2311111111111115e-06, | |
| "loss": 0.0001, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 85.0, | |
| "eval_loss": 0.7388416528701782, | |
| "eval_runtime": 749.6129, | |
| "eval_samples_per_second": 3.024, | |
| "eval_steps_per_second": 0.095, | |
| "eval_wer": 35.59019514516897, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 85.01, | |
| "learning_rate": 2.1755555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 86.0, | |
| "learning_rate": 2.12e-06, | |
| "loss": 0.0001, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 86.01, | |
| "learning_rate": 2.064444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 87.0, | |
| "learning_rate": 2.008888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 87.01, | |
| "learning_rate": 1.9533333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 88.0, | |
| "learning_rate": 1.8977777777777779e-06, | |
| "loss": 0.0001, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 88.01, | |
| "learning_rate": 1.8422222222222225e-06, | |
| "loss": 0.0001, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 89.0, | |
| "learning_rate": 1.7866666666666668e-06, | |
| "loss": 0.0001, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 89.01, | |
| "learning_rate": 1.7311111111111112e-06, | |
| "loss": 0.0001, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 90.0, | |
| "learning_rate": 1.675555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 90.01, | |
| "learning_rate": 1.6200000000000002e-06, | |
| "loss": 0.0001, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 91.0, | |
| "learning_rate": 1.5644444444444446e-06, | |
| "loss": 0.0001, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 92.0, | |
| "learning_rate": 1.5088888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 92.01, | |
| "learning_rate": 1.4533333333333335e-06, | |
| "loss": 0.0001, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 93.0, | |
| "learning_rate": 1.397777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 93.01, | |
| "learning_rate": 1.3422222222222222e-06, | |
| "loss": 0.0001, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 94.0, | |
| "learning_rate": 1.286666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 94.01, | |
| "learning_rate": 1.2311111111111112e-06, | |
| "loss": 0.0001, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 95.0, | |
| "learning_rate": 1.1755555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 95.01, | |
| "learning_rate": 1.12e-06, | |
| "loss": 0.0001, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 96.0, | |
| "learning_rate": 1.0644444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 96.01, | |
| "learning_rate": 1.008888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 97.0, | |
| "learning_rate": 9.533333333333335e-07, | |
| "loss": 0.0001, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 97.01, | |
| "learning_rate": 8.977777777777778e-07, | |
| "loss": 0.0001, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 98.0, | |
| "learning_rate": 8.422222222222224e-07, | |
| "loss": 0.0001, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 98.01, | |
| "learning_rate": 7.866666666666667e-07, | |
| "loss": 0.0001, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 99.0, | |
| "learning_rate": 7.311111111111112e-07, | |
| "loss": 0.0001, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 99.01, | |
| "learning_rate": 6.755555555555555e-07, | |
| "loss": 0.0001, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 100.0, | |
| "learning_rate": 6.200000000000001e-07, | |
| "loss": 0.0001, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 101.0, | |
| "learning_rate": 5.644444444444445e-07, | |
| "loss": 0.0001, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 101.01, | |
| "learning_rate": 5.088888888888889e-07, | |
| "loss": 0.0001, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 102.0, | |
| "learning_rate": 4.533333333333334e-07, | |
| "loss": 0.0001, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 102.01, | |
| "learning_rate": 3.9777777777777783e-07, | |
| "loss": 0.0001, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 103.0, | |
| "learning_rate": 3.422222222222223e-07, | |
| "loss": 0.0001, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 103.01, | |
| "learning_rate": 2.866666666666667e-07, | |
| "loss": 0.0001, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 104.0, | |
| "learning_rate": 2.3111111111111112e-07, | |
| "loss": 0.0001, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 104.01, | |
| "learning_rate": 1.7555555555555558e-07, | |
| "loss": 0.0001, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 105.0, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 0.0001, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 105.01, | |
| "learning_rate": 6.444444444444445e-08, | |
| "loss": 0.0001, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 106.0, | |
| "learning_rate": 8.88888888888889e-09, | |
| "loss": 0.0001, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 106.0, | |
| "eval_loss": 0.7465493083000183, | |
| "eval_runtime": 752.5079, | |
| "eval_samples_per_second": 3.013, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 35.673488814850074, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 106.0, | |
| "step": 5000, | |
| "total_flos": 9.231673804996608e+19, | |
| "train_loss": 0.04440673050563782, | |
| "train_runtime": 47208.0079, | |
| "train_samples_per_second": 6.779, | |
| "train_steps_per_second": 0.106 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 9.231673804996608e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |