| { | |
| "best_metric": 8.998885586924219, | |
| "best_model_checkpoint": "./checkpoint-5000", | |
| "epoch": 1.0, | |
| "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": 4.4e-07, | |
| "loss": 0.5955, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 0.421, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.44e-06, | |
| "loss": 0.2827, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 0.1982, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.42e-06, | |
| "loss": 0.1954, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.92e-06, | |
| "loss": 0.1979, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.1864, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.1927, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.2034, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.2536, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.2312, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.3123, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.2504, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.2816, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.2424, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.1967, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.188, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.1838, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.2108, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.1665, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.953333333333333e-06, | |
| "loss": 0.1591, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.89777777777778e-06, | |
| "loss": 0.1597, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.842222222222223e-06, | |
| "loss": 0.1741, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.786666666666667e-06, | |
| "loss": 0.1803, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.731111111111113e-06, | |
| "loss": 0.1544, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.675555555555555e-06, | |
| "loss": 0.1465, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.620000000000001e-06, | |
| "loss": 0.1319, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.564444444444445e-06, | |
| "loss": 0.1435, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.508888888888889e-06, | |
| "loss": 0.1237, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.453333333333335e-06, | |
| "loss": 0.1337, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.397777777777779e-06, | |
| "loss": 0.121, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.342222222222223e-06, | |
| "loss": 0.1191, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.286666666666667e-06, | |
| "loss": 0.1319, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.231111111111111e-06, | |
| "loss": 0.1101, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.175555555555557e-06, | |
| "loss": 0.1121, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.12e-06, | |
| "loss": 0.1232, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.064444444444447e-06, | |
| "loss": 0.1204, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.008888888888889e-06, | |
| "loss": 0.1069, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.953333333333335e-06, | |
| "loss": 0.1224, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.897777777777779e-06, | |
| "loss": 0.0987, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.19660557806491852, | |
| "eval_runtime": 2156.5521, | |
| "eval_samples_per_second": 0.786, | |
| "eval_steps_per_second": 0.098, | |
| "eval_wer": 13.651560178306092, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.842222222222223e-06, | |
| "loss": 0.1184, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.786666666666668e-06, | |
| "loss": 0.1071, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.73111111111111e-06, | |
| "loss": 0.0923, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.675555555555556e-06, | |
| "loss": 0.1384, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.62e-06, | |
| "loss": 0.1082, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.564444444444445e-06, | |
| "loss": 0.1162, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.50888888888889e-06, | |
| "loss": 0.0938, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.453333333333334e-06, | |
| "loss": 0.1061, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.397777777777778e-06, | |
| "loss": 0.1042, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.342222222222222e-06, | |
| "loss": 0.0926, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.286666666666668e-06, | |
| "loss": 0.0926, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.231111111111112e-06, | |
| "loss": 0.0671, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.175555555555556e-06, | |
| "loss": 0.0894, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.120000000000002e-06, | |
| "loss": 0.0872, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.064444444444444e-06, | |
| "loss": 0.0981, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.00888888888889e-06, | |
| "loss": 0.0886, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.953333333333334e-06, | |
| "loss": 0.0735, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.897777777777778e-06, | |
| "loss": 0.1254, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.842222222222224e-06, | |
| "loss": 0.1351, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.786666666666666e-06, | |
| "loss": 0.1256, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.731111111111112e-06, | |
| "loss": 0.1623, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.675555555555556e-06, | |
| "loss": 0.1293, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.620000000000001e-06, | |
| "loss": 0.0962, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.564444444444446e-06, | |
| "loss": 0.1324, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.50888888888889e-06, | |
| "loss": 0.1344, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.453333333333334e-06, | |
| "loss": 0.0704, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.3977777777777786e-06, | |
| "loss": 0.0813, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.342222222222223e-06, | |
| "loss": 0.0701, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.28888888888889e-06, | |
| "loss": 0.0612, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.235555555555556e-06, | |
| "loss": 0.0747, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.180000000000001e-06, | |
| "loss": 0.0674, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.124444444444445e-06, | |
| "loss": 0.0577, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.06888888888889e-06, | |
| "loss": 0.0597, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.0133333333333345e-06, | |
| "loss": 0.0661, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.9577777777777785e-06, | |
| "loss": 0.0594, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.902222222222223e-06, | |
| "loss": 0.0527, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.846666666666667e-06, | |
| "loss": 0.0674, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.7911111111111115e-06, | |
| "loss": 0.0735, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.735555555555556e-06, | |
| "loss": 0.0643, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.680000000000001e-06, | |
| "loss": 0.0772, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.18119178712368011, | |
| "eval_runtime": 2204.2882, | |
| "eval_samples_per_second": 0.769, | |
| "eval_steps_per_second": 0.096, | |
| "eval_wer": 12.277117384843981, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.6244444444444445e-06, | |
| "loss": 0.0592, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.568888888888889e-06, | |
| "loss": 0.0557, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.513333333333333e-06, | |
| "loss": 0.062, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.457777777777778e-06, | |
| "loss": 0.0518, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.402222222222223e-06, | |
| "loss": 0.0451, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.346666666666668e-06, | |
| "loss": 0.0645, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.291111111111111e-06, | |
| "loss": 0.057, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.235555555555556e-06, | |
| "loss": 0.0651, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.18e-06, | |
| "loss": 0.0516, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.124444444444445e-06, | |
| "loss": 0.0532, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 6.06888888888889e-06, | |
| "loss": 0.0546, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 6.013333333333335e-06, | |
| "loss": 0.0464, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.957777777777778e-06, | |
| "loss": 0.0509, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.902222222222223e-06, | |
| "loss": 0.0588, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.846666666666667e-06, | |
| "loss": 0.0796, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.791111111111112e-06, | |
| "loss": 0.0563, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.735555555555557e-06, | |
| "loss": 0.0687, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.68e-06, | |
| "loss": 0.0454, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.624444444444445e-06, | |
| "loss": 0.0453, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.56888888888889e-06, | |
| "loss": 0.0501, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.513333333333334e-06, | |
| "loss": 0.0486, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.4577777777777785e-06, | |
| "loss": 0.0535, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.402222222222223e-06, | |
| "loss": 0.0413, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.346666666666667e-06, | |
| "loss": 0.0448, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.2911111111111115e-06, | |
| "loss": 0.0669, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.235555555555556e-06, | |
| "loss": 0.0789, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.18e-06, | |
| "loss": 0.0686, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.124444444444445e-06, | |
| "loss": 0.089, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.06888888888889e-06, | |
| "loss": 0.0563, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.013333333333333e-06, | |
| "loss": 0.0774, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.957777777777778e-06, | |
| "loss": 0.103, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.902222222222222e-06, | |
| "loss": 0.0887, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.846666666666667e-06, | |
| "loss": 0.1001, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.791111111111111e-06, | |
| "loss": 0.0777, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.735555555555556e-06, | |
| "loss": 0.0508, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.680000000000001e-06, | |
| "loss": 0.0509, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.624444444444445e-06, | |
| "loss": 0.0425, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.568888888888889e-06, | |
| "loss": 0.0445, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.513333333333333e-06, | |
| "loss": 0.0541, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.457777777777778e-06, | |
| "loss": 0.0398, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 0.17341840267181396, | |
| "eval_runtime": 2240.6936, | |
| "eval_samples_per_second": 0.757, | |
| "eval_steps_per_second": 0.095, | |
| "eval_wer": 11.311292719167906, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.402222222222223e-06, | |
| "loss": 0.0544, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.346666666666667e-06, | |
| "loss": 0.0347, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.291111111111112e-06, | |
| "loss": 0.0457, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.235555555555556e-06, | |
| "loss": 0.0353, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.18e-06, | |
| "loss": 0.0519, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.124444444444445e-06, | |
| "loss": 0.0391, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.0688888888888896e-06, | |
| "loss": 0.0385, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.013333333333334e-06, | |
| "loss": 0.0364, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.9577777777777785e-06, | |
| "loss": 0.0505, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.9022222222222225e-06, | |
| "loss": 0.0343, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.8466666666666665e-06, | |
| "loss": 0.0383, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.7911111111111114e-06, | |
| "loss": 0.032, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.7355555555555555e-06, | |
| "loss": 0.0277, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.6800000000000003e-06, | |
| "loss": 0.0416, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.624444444444445e-06, | |
| "loss": 0.0298, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.568888888888889e-06, | |
| "loss": 0.0272, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.5133333333333337e-06, | |
| "loss": 0.0387, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.457777777777778e-06, | |
| "loss": 0.0325, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.4022222222222222e-06, | |
| "loss": 0.0417, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.346666666666667e-06, | |
| "loss": 0.0416, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.2911111111111116e-06, | |
| "loss": 0.0237, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.2355555555555556e-06, | |
| "loss": 0.03, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.1800000000000005e-06, | |
| "loss": 0.0341, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.124444444444445e-06, | |
| "loss": 0.0301, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.068888888888889e-06, | |
| "loss": 0.0349, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.013333333333334e-06, | |
| "loss": 0.0293, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.957777777777778e-06, | |
| "loss": 0.0379, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.9022222222222223e-06, | |
| "loss": 0.0338, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.8466666666666672e-06, | |
| "loss": 0.0317, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.7911111111111113e-06, | |
| "loss": 0.0367, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.7355555555555557e-06, | |
| "loss": 0.028, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.68e-06, | |
| "loss": 0.0255, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.6244444444444446e-06, | |
| "loss": 0.024, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.568888888888889e-06, | |
| "loss": 0.0491, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.5133333333333336e-06, | |
| "loss": 0.0617, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.457777777777778e-06, | |
| "loss": 0.0642, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.4022222222222225e-06, | |
| "loss": 0.0726, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.346666666666667e-06, | |
| "loss": 0.0703, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.2911111111111114e-06, | |
| "loss": 0.0625, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.235555555555556e-06, | |
| "loss": 0.0775, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 0.16992205381393433, | |
| "eval_runtime": 2146.6353, | |
| "eval_samples_per_second": 0.79, | |
| "eval_steps_per_second": 0.099, | |
| "eval_wer": 9.797548291233284, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.1800000000000003e-06, | |
| "loss": 0.0666, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.1244444444444443e-06, | |
| "loss": 0.0509, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.0688888888888892e-06, | |
| "loss": 0.0531, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.0133333333333337e-06, | |
| "loss": 0.0396, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.9577777777777777e-06, | |
| "loss": 0.0445, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.9022222222222222e-06, | |
| "loss": 0.0393, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.8466666666666668e-06, | |
| "loss": 0.0411, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.7911111111111113e-06, | |
| "loss": 0.0507, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.7355555555555555e-06, | |
| "loss": 0.035, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.6800000000000002e-06, | |
| "loss": 0.0411, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.6244444444444447e-06, | |
| "loss": 0.0346, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.568888888888889e-06, | |
| "loss": 0.0358, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.5133333333333334e-06, | |
| "loss": 0.0404, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.457777777777778e-06, | |
| "loss": 0.0369, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.4022222222222223e-06, | |
| "loss": 0.0511, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.3466666666666668e-06, | |
| "loss": 0.0368, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2911111111111112e-06, | |
| "loss": 0.0293, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2355555555555557e-06, | |
| "loss": 0.028, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1800000000000001e-06, | |
| "loss": 0.0179, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1244444444444446e-06, | |
| "loss": 0.0337, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.068888888888889e-06, | |
| "loss": 0.0219, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.0133333333333333e-06, | |
| "loss": 0.0337, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.57777777777778e-07, | |
| "loss": 0.021, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.022222222222222e-07, | |
| "loss": 0.0197, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.466666666666668e-07, | |
| "loss": 0.0198, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.911111111111111e-07, | |
| "loss": 0.0172, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 7.355555555555556e-07, | |
| "loss": 0.0226, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.800000000000001e-07, | |
| "loss": 0.0199, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.244444444444445e-07, | |
| "loss": 0.0235, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.68888888888889e-07, | |
| "loss": 0.0197, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.133333333333334e-07, | |
| "loss": 0.022, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.5777777777777784e-07, | |
| "loss": 0.0211, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.0222222222222224e-07, | |
| "loss": 0.0276, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.466666666666667e-07, | |
| "loss": 0.022, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.9111111111111116e-07, | |
| "loss": 0.0297, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.3555555555555556e-07, | |
| "loss": 0.0165, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.8e-07, | |
| "loss": 0.0199, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.2444444444444446e-07, | |
| "loss": 0.0181, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.888888888888889e-08, | |
| "loss": 0.0202, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.3333333333333334e-08, | |
| "loss": 0.0314, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.18952082097530365, | |
| "eval_runtime": 2132.1866, | |
| "eval_samples_per_second": 0.795, | |
| "eval_steps_per_second": 0.099, | |
| "eval_wer": 8.998885586924219, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 5000, | |
| "total_flos": 8.4926840832e+19, | |
| "train_loss": 0.08331226392388344, | |
| "train_runtime": 25405.394, | |
| "train_samples_per_second": 1.574, | |
| "train_steps_per_second": 0.197 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 8.4926840832e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |