| { | |
| "best_metric": 66.04630049931912, | |
| "best_model_checkpoint": "./output/small/yt-special-batch4-base/checkpoint-5000", | |
| "epoch": 0.4294425835265825, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.2000000000000006e-07, | |
| "loss": 2.873, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 2.7105, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.42e-06, | |
| "loss": 2.2533, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 2.1946, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.42e-06, | |
| "loss": 2.1769, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.92e-06, | |
| "loss": 2.031, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 1.6908, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 1.5555, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.42e-06, | |
| "loss": 1.5242, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.92e-06, | |
| "loss": 1.4625, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 1.5447, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.92e-06, | |
| "loss": 1.2566, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.42e-06, | |
| "loss": 1.2499, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.92e-06, | |
| "loss": 1.2861, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 1.2488, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.92e-06, | |
| "loss": 1.272, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.42e-06, | |
| "loss": 1.449, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 1.3319, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.42e-06, | |
| "loss": 1.4091, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 1.1976, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.953333333333333e-06, | |
| "loss": 1.1124, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.89777777777778e-06, | |
| "loss": 1.5198, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.842222222222223e-06, | |
| "loss": 1.3006, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.786666666666667e-06, | |
| "loss": 1.2553, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.731111111111113e-06, | |
| "loss": 1.181, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.675555555555555e-06, | |
| "loss": 1.2383, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.620000000000001e-06, | |
| "loss": 1.3812, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.564444444444445e-06, | |
| "loss": 1.1598, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.508888888888889e-06, | |
| "loss": 1.3842, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.453333333333335e-06, | |
| "loss": 1.4457, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.397777777777779e-06, | |
| "loss": 1.2762, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.342222222222223e-06, | |
| "loss": 1.2001, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.286666666666667e-06, | |
| "loss": 1.0802, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.231111111111111e-06, | |
| "loss": 1.0965, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.175555555555557e-06, | |
| "loss": 1.3009, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.12e-06, | |
| "loss": 1.3254, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.064444444444447e-06, | |
| "loss": 1.2177, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.008888888888889e-06, | |
| "loss": 1.2685, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.953333333333335e-06, | |
| "loss": 1.1188, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.897777777777779e-06, | |
| "loss": 1.4446, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "eval_loss": 1.2313047647476196, | |
| "eval_runtime": 371.4581, | |
| "eval_samples_per_second": 6.598, | |
| "eval_steps_per_second": 3.301, | |
| "eval_wer": 91.59587575384217, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.842222222222223e-06, | |
| "loss": 1.1867, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.786666666666668e-06, | |
| "loss": 1.1923, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.73111111111111e-06, | |
| "loss": 1.3379, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.675555555555556e-06, | |
| "loss": 1.1456, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.62e-06, | |
| "loss": 1.2273, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.564444444444445e-06, | |
| "loss": 1.1864, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.50888888888889e-06, | |
| "loss": 1.3387, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.453333333333334e-06, | |
| "loss": 1.3581, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.397777777777778e-06, | |
| "loss": 1.275, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.342222222222222e-06, | |
| "loss": 1.1069, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.286666666666668e-06, | |
| "loss": 1.3165, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.231111111111112e-06, | |
| "loss": 1.1256, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.175555555555556e-06, | |
| "loss": 1.3302, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.120000000000002e-06, | |
| "loss": 1.1751, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.064444444444444e-06, | |
| "loss": 1.2288, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.00888888888889e-06, | |
| "loss": 0.9965, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.953333333333334e-06, | |
| "loss": 1.1748, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.897777777777778e-06, | |
| "loss": 1.1932, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.842222222222224e-06, | |
| "loss": 1.277, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.786666666666666e-06, | |
| "loss": 1.1456, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.731111111111112e-06, | |
| "loss": 1.1711, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.675555555555556e-06, | |
| "loss": 1.0097, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.620000000000001e-06, | |
| "loss": 1.2093, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.564444444444446e-06, | |
| "loss": 1.0703, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.50888888888889e-06, | |
| "loss": 1.1023, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.453333333333334e-06, | |
| "loss": 1.0636, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.3977777777777786e-06, | |
| "loss": 1.1086, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.342222222222223e-06, | |
| "loss": 1.0284, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.2866666666666675e-06, | |
| "loss": 1.151, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.231111111111112e-06, | |
| "loss": 1.1759, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.1755555555555556e-06, | |
| "loss": 1.2208, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.1200000000000004e-06, | |
| "loss": 1.158, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.0644444444444445e-06, | |
| "loss": 1.236, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.008888888888889e-06, | |
| "loss": 1.0767, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6.953333333333334e-06, | |
| "loss": 1.1283, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6.897777777777779e-06, | |
| "loss": 1.0166, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.842222222222222e-06, | |
| "loss": 1.0964, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.786666666666667e-06, | |
| "loss": 1.0885, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.731111111111111e-06, | |
| "loss": 1.1983, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.675555555555556e-06, | |
| "loss": 1.0599, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "eval_loss": 1.1311960220336914, | |
| "eval_runtime": 573.6162, | |
| "eval_samples_per_second": 4.273, | |
| "eval_steps_per_second": 2.137, | |
| "eval_wer": 106.34200116723949, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.620000000000001e-06, | |
| "loss": 1.1776, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.564444444444446e-06, | |
| "loss": 1.3047, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.508888888888889e-06, | |
| "loss": 1.1054, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.453333333333334e-06, | |
| "loss": 1.0221, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.397777777777778e-06, | |
| "loss": 1.1652, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.342222222222223e-06, | |
| "loss": 1.2054, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.286666666666668e-06, | |
| "loss": 1.1084, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.231111111111111e-06, | |
| "loss": 1.088, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.175555555555556e-06, | |
| "loss": 1.131, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.120000000000001e-06, | |
| "loss": 1.0835, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.064444444444445e-06, | |
| "loss": 1.2146, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.011111111111112e-06, | |
| "loss": 1.3356, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.955555555555555e-06, | |
| "loss": 1.1573, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.9e-06, | |
| "loss": 1.3803, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.844444444444445e-06, | |
| "loss": 1.2143, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.788888888888889e-06, | |
| "loss": 0.9968, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.733333333333334e-06, | |
| "loss": 1.1439, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.677777777777779e-06, | |
| "loss": 1.1418, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.622222222222222e-06, | |
| "loss": 1.1943, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.566666666666667e-06, | |
| "loss": 1.2499, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.511111111111112e-06, | |
| "loss": 1.0368, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.455555555555556e-06, | |
| "loss": 0.9317, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.9876, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.344444444444446e-06, | |
| "loss": 1.2629, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.288888888888889e-06, | |
| "loss": 1.1805, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.233333333333334e-06, | |
| "loss": 0.9991, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.177777777777779e-06, | |
| "loss": 1.2692, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.122222222222223e-06, | |
| "loss": 0.9476, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.0666666666666676e-06, | |
| "loss": 1.1276, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5.011111111111111e-06, | |
| "loss": 1.0929, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.9555555555555565e-06, | |
| "loss": 1.0665, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 1.0465, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.8444444444444446e-06, | |
| "loss": 0.9563, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.7888888888888894e-06, | |
| "loss": 1.2886, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.7333333333333335e-06, | |
| "loss": 1.1789, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.677777777777778e-06, | |
| "loss": 0.9365, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.622222222222222e-06, | |
| "loss": 1.1035, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.566666666666667e-06, | |
| "loss": 1.1096, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.511111111111111e-06, | |
| "loss": 0.8761, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.455555555555555e-06, | |
| "loss": 1.1851, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "eval_loss": 1.080060362815857, | |
| "eval_runtime": 483.6916, | |
| "eval_samples_per_second": 5.067, | |
| "eval_steps_per_second": 2.535, | |
| "eval_wer": 77.31664613189805, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.4e-06, | |
| "loss": 1.2288, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.344444444444445e-06, | |
| "loss": 1.2901, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.288888888888889e-06, | |
| "loss": 0.9715, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.233333333333334e-06, | |
| "loss": 1.0392, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.177777777777778e-06, | |
| "loss": 1.104, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.122222222222222e-06, | |
| "loss": 1.1265, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.066666666666667e-06, | |
| "loss": 1.0865, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.011111111111111e-06, | |
| "loss": 1.0333, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.955555555555556e-06, | |
| "loss": 0.8884, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 1.0726, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.844444444444445e-06, | |
| "loss": 1.0755, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.7888888888888893e-06, | |
| "loss": 1.0754, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.7333333333333337e-06, | |
| "loss": 1.0286, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.6777777777777778e-06, | |
| "loss": 1.0932, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.6222222222222226e-06, | |
| "loss": 0.9988, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.566666666666667e-06, | |
| "loss": 0.997, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.511111111111111e-06, | |
| "loss": 1.1233, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.455555555555556e-06, | |
| "loss": 0.9964, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 1.0635, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.3444444444444445e-06, | |
| "loss": 0.9095, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.2888888888888894e-06, | |
| "loss": 1.1972, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.2333333333333334e-06, | |
| "loss": 1.0363, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.177777777777778e-06, | |
| "loss": 1.0375, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.1222222222222228e-06, | |
| "loss": 0.9243, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.066666666666667e-06, | |
| "loss": 1.0021, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.0111111111111113e-06, | |
| "loss": 1.1643, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.955555555555556e-06, | |
| "loss": 1.0984, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.9e-06, | |
| "loss": 1.1525, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.8444444444444446e-06, | |
| "loss": 0.982, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.788888888888889e-06, | |
| "loss": 0.9339, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.7333333333333336e-06, | |
| "loss": 1.0715, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.677777777777778e-06, | |
| "loss": 1.0495, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.6222222222222225e-06, | |
| "loss": 1.1439, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.566666666666667e-06, | |
| "loss": 1.0615, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.5111111111111114e-06, | |
| "loss": 1.0708, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.455555555555556e-06, | |
| "loss": 1.1386, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 1.1339, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.3444444444444448e-06, | |
| "loss": 0.8847, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.2888888888888892e-06, | |
| "loss": 1.1025, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.2333333333333333e-06, | |
| "loss": 1.0325, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "eval_loss": 1.0379793643951416, | |
| "eval_runtime": 414.138, | |
| "eval_samples_per_second": 5.918, | |
| "eval_steps_per_second": 2.96, | |
| "eval_wer": 71.84358990986317, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.1777777777777777e-06, | |
| "loss": 1.0729, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.1222222222222226e-06, | |
| "loss": 0.9971, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.0666666666666666e-06, | |
| "loss": 0.9761, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.011111111111111e-06, | |
| "loss": 1.2342, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.955555555555556e-06, | |
| "loss": 1.0527, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 1.1445, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.8444444444444445e-06, | |
| "loss": 0.8684, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.788888888888889e-06, | |
| "loss": 1.0673, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.7333333333333336e-06, | |
| "loss": 0.9241, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.6777777777777779e-06, | |
| "loss": 0.9604, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.6222222222222223e-06, | |
| "loss": 0.9731, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.566666666666667e-06, | |
| "loss": 1.1819, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.5111111111111112e-06, | |
| "loss": 1.0006, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4555555555555557e-06, | |
| "loss": 1.1385, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 1.0691, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3444444444444446e-06, | |
| "loss": 1.0082, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.288888888888889e-06, | |
| "loss": 1.1427, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.2333333333333335e-06, | |
| "loss": 0.8985, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.1777777777777778e-06, | |
| "loss": 0.9184, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.1222222222222222e-06, | |
| "loss": 1.0108, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.068888888888889e-06, | |
| "loss": 1.0148, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.0133333333333333e-06, | |
| "loss": 0.8613, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.57777777777778e-07, | |
| "loss": 1.1145, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.022222222222222e-07, | |
| "loss": 1.0819, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.466666666666668e-07, | |
| "loss": 1.0479, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.911111111111111e-07, | |
| "loss": 1.0571, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.355555555555556e-07, | |
| "loss": 1.0167, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.800000000000001e-07, | |
| "loss": 1.1788, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.244444444444445e-07, | |
| "loss": 1.1127, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.68888888888889e-07, | |
| "loss": 1.0466, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.133333333333334e-07, | |
| "loss": 0.9919, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.5777777777777784e-07, | |
| "loss": 0.9679, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.0222222222222224e-07, | |
| "loss": 0.9623, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.466666666666667e-07, | |
| "loss": 1.0163, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.9111111111111116e-07, | |
| "loss": 1.1002, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.3555555555555556e-07, | |
| "loss": 0.9265, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8e-07, | |
| "loss": 1.0592, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2444444444444446e-07, | |
| "loss": 0.9983, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.888888888888889e-08, | |
| "loss": 0.9773, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.3333333333333334e-08, | |
| "loss": 1.008, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "eval_loss": 1.0174591541290283, | |
| "eval_runtime": 307.5669, | |
| "eval_samples_per_second": 7.969, | |
| "eval_steps_per_second": 3.986, | |
| "eval_wer": 66.04630049931912, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "step": 5000, | |
| "total_flos": 1.2971999232e+18, | |
| "train_loss": 1.171582175064087, | |
| "train_runtime": 3966.565, | |
| "train_samples_per_second": 5.042, | |
| "train_steps_per_second": 1.261 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 1, | |
| "total_flos": 1.2971999232e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |