| { |
| "best_metric": 23.19300667126754, |
| "best_model_checkpoint": "./output/small/cv9-special-batch8-base/checkpoint-4000", |
| "epoch": 4.84027105517909, |
| "global_step": 5000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.4e-07, |
| "loss": 1.5953, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 9.400000000000001e-07, |
| "loss": 1.5825, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 1.44e-06, |
| "loss": 1.3754, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 1.94e-06, |
| "loss": 1.207, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 2.4400000000000004e-06, |
| "loss": 1.0035, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 2.9400000000000002e-06, |
| "loss": 0.8401, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 3.44e-06, |
| "loss": 0.8021, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 3.94e-06, |
| "loss": 0.7816, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 4.440000000000001e-06, |
| "loss": 0.689, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 4.94e-06, |
| "loss": 0.648, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 5.4400000000000004e-06, |
| "loss": 0.6931, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 5.94e-06, |
| "loss": 0.6811, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 6.440000000000001e-06, |
| "loss": 0.624, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 6.9400000000000005e-06, |
| "loss": 0.6331, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 7.440000000000001e-06, |
| "loss": 0.666, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 7.94e-06, |
| "loss": 0.6223, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 8.44e-06, |
| "loss": 0.6124, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 8.94e-06, |
| "loss": 0.6159, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 9.440000000000001e-06, |
| "loss": 0.5552, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 9.940000000000001e-06, |
| "loss": 0.5695, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 9.951111111111111e-06, |
| "loss": 0.6001, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 9.895555555555557e-06, |
| "loss": 0.5218, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 9.84e-06, |
| "loss": 0.5698, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 9.784444444444445e-06, |
| "loss": 0.6014, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 9.72888888888889e-06, |
| "loss": 0.5341, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 9.673333333333334e-06, |
| "loss": 0.4888, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 9.617777777777778e-06, |
| "loss": 0.5417, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 9.562222222222223e-06, |
| "loss": 0.5562, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 9.506666666666667e-06, |
| "loss": 0.5835, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 9.451111111111112e-06, |
| "loss": 0.5038, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 9.395555555555556e-06, |
| "loss": 0.5457, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 9.340000000000002e-06, |
| "loss": 0.469, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 9.284444444444444e-06, |
| "loss": 0.5071, |
| "step": 825 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 9.22888888888889e-06, |
| "loss": 0.5111, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 9.175555555555557e-06, |
| "loss": 0.4781, |
| "step": 875 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 9.12e-06, |
| "loss": 0.5057, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 9.064444444444447e-06, |
| "loss": 0.5103, |
| "step": 925 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 9.008888888888889e-06, |
| "loss": 0.488, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 8.953333333333335e-06, |
| "loss": 0.5091, |
| "step": 975 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 8.897777777777779e-06, |
| "loss": 0.4648, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.97, |
| "eval_loss": 0.4258050322532654, |
| "eval_runtime": 320.8318, |
| "eval_samples_per_second": 11.289, |
| "eval_steps_per_second": 2.824, |
| "eval_wer": 27.62364849321371, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 8.842222222222223e-06, |
| "loss": 0.5509, |
| "step": 1025 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 8.786666666666668e-06, |
| "loss": 0.4188, |
| "step": 1050 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 8.73111111111111e-06, |
| "loss": 0.351, |
| "step": 1075 |
| }, |
| { |
| "epoch": 1.06, |
| "learning_rate": 8.675555555555556e-06, |
| "loss": 0.3348, |
| "step": 1100 |
| }, |
| { |
| "epoch": 1.09, |
| "learning_rate": 8.62e-06, |
| "loss": 0.3114, |
| "step": 1125 |
| }, |
| { |
| "epoch": 1.11, |
| "learning_rate": 8.564444444444445e-06, |
| "loss": 0.3225, |
| "step": 1150 |
| }, |
| { |
| "epoch": 1.14, |
| "learning_rate": 8.50888888888889e-06, |
| "loss": 0.3408, |
| "step": 1175 |
| }, |
| { |
| "epoch": 1.16, |
| "learning_rate": 8.453333333333334e-06, |
| "loss": 0.3177, |
| "step": 1200 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 8.397777777777778e-06, |
| "loss": 0.3314, |
| "step": 1225 |
| }, |
| { |
| "epoch": 1.21, |
| "learning_rate": 8.342222222222222e-06, |
| "loss": 0.3216, |
| "step": 1250 |
| }, |
| { |
| "epoch": 1.23, |
| "learning_rate": 8.286666666666668e-06, |
| "loss": 0.3521, |
| "step": 1275 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 8.231111111111112e-06, |
| "loss": 0.2922, |
| "step": 1300 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 8.175555555555556e-06, |
| "loss": 0.3127, |
| "step": 1325 |
| }, |
| { |
| "epoch": 1.31, |
| "learning_rate": 8.120000000000002e-06, |
| "loss": 0.3222, |
| "step": 1350 |
| }, |
| { |
| "epoch": 1.33, |
| "learning_rate": 8.064444444444444e-06, |
| "loss": 0.3397, |
| "step": 1375 |
| }, |
| { |
| "epoch": 1.36, |
| "learning_rate": 8.00888888888889e-06, |
| "loss": 0.3617, |
| "step": 1400 |
| }, |
| { |
| "epoch": 1.38, |
| "learning_rate": 7.953333333333334e-06, |
| "loss": 0.3368, |
| "step": 1425 |
| }, |
| { |
| "epoch": 1.4, |
| "learning_rate": 7.897777777777778e-06, |
| "loss": 0.3407, |
| "step": 1450 |
| }, |
| { |
| "epoch": 1.43, |
| "learning_rate": 7.842222222222224e-06, |
| "loss": 0.3215, |
| "step": 1475 |
| }, |
| { |
| "epoch": 1.45, |
| "learning_rate": 7.786666666666666e-06, |
| "loss": 0.3246, |
| "step": 1500 |
| }, |
| { |
| "epoch": 1.48, |
| "learning_rate": 7.731111111111112e-06, |
| "loss": 0.3124, |
| "step": 1525 |
| }, |
| { |
| "epoch": 1.5, |
| "learning_rate": 7.675555555555556e-06, |
| "loss": 0.3023, |
| "step": 1550 |
| }, |
| { |
| "epoch": 1.52, |
| "learning_rate": 7.620000000000001e-06, |
| "loss": 0.3426, |
| "step": 1575 |
| }, |
| { |
| "epoch": 1.55, |
| "learning_rate": 7.564444444444446e-06, |
| "loss": 0.3254, |
| "step": 1600 |
| }, |
| { |
| "epoch": 1.57, |
| "learning_rate": 7.50888888888889e-06, |
| "loss": 0.2945, |
| "step": 1625 |
| }, |
| { |
| "epoch": 1.6, |
| "learning_rate": 7.453333333333334e-06, |
| "loss": 0.3176, |
| "step": 1650 |
| }, |
| { |
| "epoch": 1.62, |
| "learning_rate": 7.3977777777777786e-06, |
| "loss": 0.3185, |
| "step": 1675 |
| }, |
| { |
| "epoch": 1.65, |
| "learning_rate": 7.342222222222223e-06, |
| "loss": 0.3131, |
| "step": 1700 |
| }, |
| { |
| "epoch": 1.67, |
| "learning_rate": 7.2866666666666675e-06, |
| "loss": 0.2917, |
| "step": 1725 |
| }, |
| { |
| "epoch": 1.69, |
| "learning_rate": 7.231111111111112e-06, |
| "loss": 0.2742, |
| "step": 1750 |
| }, |
| { |
| "epoch": 1.72, |
| "learning_rate": 7.1755555555555556e-06, |
| "loss": 0.3514, |
| "step": 1775 |
| }, |
| { |
| "epoch": 1.74, |
| "learning_rate": 7.1200000000000004e-06, |
| "loss": 0.2901, |
| "step": 1800 |
| }, |
| { |
| "epoch": 1.77, |
| "learning_rate": 7.0644444444444445e-06, |
| "loss": 0.2953, |
| "step": 1825 |
| }, |
| { |
| "epoch": 1.79, |
| "learning_rate": 7.008888888888889e-06, |
| "loss": 0.3003, |
| "step": 1850 |
| }, |
| { |
| "epoch": 1.82, |
| "learning_rate": 6.953333333333334e-06, |
| "loss": 0.29, |
| "step": 1875 |
| }, |
| { |
| "epoch": 1.84, |
| "learning_rate": 6.897777777777779e-06, |
| "loss": 0.305, |
| "step": 1900 |
| }, |
| { |
| "epoch": 1.86, |
| "learning_rate": 6.842222222222222e-06, |
| "loss": 0.2921, |
| "step": 1925 |
| }, |
| { |
| "epoch": 1.89, |
| "learning_rate": 6.786666666666667e-06, |
| "loss": 0.2908, |
| "step": 1950 |
| }, |
| { |
| "epoch": 1.91, |
| "learning_rate": 6.731111111111111e-06, |
| "loss": 0.2731, |
| "step": 1975 |
| }, |
| { |
| "epoch": 1.94, |
| "learning_rate": 6.675555555555556e-06, |
| "loss": 0.2992, |
| "step": 2000 |
| }, |
| { |
| "epoch": 1.94, |
| "eval_loss": 0.3830837309360504, |
| "eval_runtime": 315.315, |
| "eval_samples_per_second": 11.487, |
| "eval_steps_per_second": 2.873, |
| "eval_wer": 24.444444444444443, |
| "step": 2000 |
| }, |
| { |
| "epoch": 1.96, |
| "learning_rate": 6.620000000000001e-06, |
| "loss": 0.3173, |
| "step": 2025 |
| }, |
| { |
| "epoch": 1.98, |
| "learning_rate": 6.564444444444446e-06, |
| "loss": 0.3265, |
| "step": 2050 |
| }, |
| { |
| "epoch": 2.01, |
| "learning_rate": 6.508888888888889e-06, |
| "loss": 0.2542, |
| "step": 2075 |
| }, |
| { |
| "epoch": 2.03, |
| "learning_rate": 6.453333333333334e-06, |
| "loss": 0.1756, |
| "step": 2100 |
| }, |
| { |
| "epoch": 2.06, |
| "learning_rate": 6.397777777777778e-06, |
| "loss": 0.1957, |
| "step": 2125 |
| }, |
| { |
| "epoch": 2.08, |
| "learning_rate": 6.342222222222223e-06, |
| "loss": 0.1626, |
| "step": 2150 |
| }, |
| { |
| "epoch": 2.11, |
| "learning_rate": 6.286666666666668e-06, |
| "loss": 0.1738, |
| "step": 2175 |
| }, |
| { |
| "epoch": 2.13, |
| "learning_rate": 6.231111111111111e-06, |
| "loss": 0.1651, |
| "step": 2200 |
| }, |
| { |
| "epoch": 2.15, |
| "learning_rate": 6.175555555555556e-06, |
| "loss": 0.1735, |
| "step": 2225 |
| }, |
| { |
| "epoch": 2.18, |
| "learning_rate": 6.120000000000001e-06, |
| "loss": 0.1803, |
| "step": 2250 |
| }, |
| { |
| "epoch": 2.2, |
| "learning_rate": 6.064444444444445e-06, |
| "loss": 0.1785, |
| "step": 2275 |
| }, |
| { |
| "epoch": 2.23, |
| "learning_rate": 6.00888888888889e-06, |
| "loss": 0.1795, |
| "step": 2300 |
| }, |
| { |
| "epoch": 2.25, |
| "learning_rate": 5.9533333333333345e-06, |
| "loss": 0.1684, |
| "step": 2325 |
| }, |
| { |
| "epoch": 2.27, |
| "learning_rate": 5.897777777777778e-06, |
| "loss": 0.1794, |
| "step": 2350 |
| }, |
| { |
| "epoch": 2.3, |
| "learning_rate": 5.8422222222222226e-06, |
| "loss": 0.1868, |
| "step": 2375 |
| }, |
| { |
| "epoch": 2.32, |
| "learning_rate": 5.7866666666666674e-06, |
| "loss": 0.1912, |
| "step": 2400 |
| }, |
| { |
| "epoch": 2.35, |
| "learning_rate": 5.7311111111111115e-06, |
| "loss": 0.1678, |
| "step": 2425 |
| }, |
| { |
| "epoch": 2.37, |
| "learning_rate": 5.675555555555556e-06, |
| "loss": 0.1553, |
| "step": 2450 |
| }, |
| { |
| "epoch": 2.4, |
| "learning_rate": 5.620000000000001e-06, |
| "loss": 0.1902, |
| "step": 2475 |
| }, |
| { |
| "epoch": 2.42, |
| "learning_rate": 5.5644444444444444e-06, |
| "loss": 0.1891, |
| "step": 2500 |
| }, |
| { |
| "epoch": 2.44, |
| "learning_rate": 5.508888888888889e-06, |
| "loss": 0.1953, |
| "step": 2525 |
| }, |
| { |
| "epoch": 2.47, |
| "learning_rate": 5.453333333333334e-06, |
| "loss": 0.1713, |
| "step": 2550 |
| }, |
| { |
| "epoch": 2.49, |
| "learning_rate": 5.397777777777778e-06, |
| "loss": 0.167, |
| "step": 2575 |
| }, |
| { |
| "epoch": 2.52, |
| "learning_rate": 5.342222222222223e-06, |
| "loss": 0.1553, |
| "step": 2600 |
| }, |
| { |
| "epoch": 2.54, |
| "learning_rate": 5.286666666666666e-06, |
| "loss": 0.1717, |
| "step": 2625 |
| }, |
| { |
| "epoch": 2.57, |
| "learning_rate": 5.231111111111111e-06, |
| "loss": 0.1813, |
| "step": 2650 |
| }, |
| { |
| "epoch": 2.59, |
| "learning_rate": 5.175555555555556e-06, |
| "loss": 0.1988, |
| "step": 2675 |
| }, |
| { |
| "epoch": 2.61, |
| "learning_rate": 5.12e-06, |
| "loss": 0.1738, |
| "step": 2700 |
| }, |
| { |
| "epoch": 2.64, |
| "learning_rate": 5.064444444444445e-06, |
| "loss": 0.1564, |
| "step": 2725 |
| }, |
| { |
| "epoch": 2.66, |
| "learning_rate": 5.00888888888889e-06, |
| "loss": 0.1542, |
| "step": 2750 |
| }, |
| { |
| "epoch": 2.69, |
| "learning_rate": 4.953333333333334e-06, |
| "loss": 0.1663, |
| "step": 2775 |
| }, |
| { |
| "epoch": 2.71, |
| "learning_rate": 4.897777777777778e-06, |
| "loss": 0.1539, |
| "step": 2800 |
| }, |
| { |
| "epoch": 2.73, |
| "learning_rate": 4.842222222222223e-06, |
| "loss": 0.1712, |
| "step": 2825 |
| }, |
| { |
| "epoch": 2.76, |
| "learning_rate": 4.786666666666667e-06, |
| "loss": 0.1796, |
| "step": 2850 |
| }, |
| { |
| "epoch": 2.78, |
| "learning_rate": 4.731111111111112e-06, |
| "loss": 0.1634, |
| "step": 2875 |
| }, |
| { |
| "epoch": 2.81, |
| "learning_rate": 4.675555555555556e-06, |
| "loss": 0.1635, |
| "step": 2900 |
| }, |
| { |
| "epoch": 2.83, |
| "learning_rate": 4.620000000000001e-06, |
| "loss": 0.1886, |
| "step": 2925 |
| }, |
| { |
| "epoch": 2.86, |
| "learning_rate": 4.564444444444445e-06, |
| "loss": 0.191, |
| "step": 2950 |
| }, |
| { |
| "epoch": 2.88, |
| "learning_rate": 4.50888888888889e-06, |
| "loss": 0.1694, |
| "step": 2975 |
| }, |
| { |
| "epoch": 2.9, |
| "learning_rate": 4.453333333333334e-06, |
| "loss": 0.1597, |
| "step": 3000 |
| }, |
| { |
| "epoch": 2.9, |
| "eval_loss": 0.37780889868736267, |
| "eval_runtime": 320.8536, |
| "eval_samples_per_second": 11.289, |
| "eval_steps_per_second": 2.824, |
| "eval_wer": 23.616287094547964, |
| "step": 3000 |
| }, |
| { |
| "epoch": 2.93, |
| "learning_rate": 4.397777777777778e-06, |
| "loss": 0.1755, |
| "step": 3025 |
| }, |
| { |
| "epoch": 2.95, |
| "learning_rate": 4.3422222222222225e-06, |
| "loss": 0.1789, |
| "step": 3050 |
| }, |
| { |
| "epoch": 2.98, |
| "learning_rate": 4.2866666666666666e-06, |
| "loss": 0.2103, |
| "step": 3075 |
| }, |
| { |
| "epoch": 3.0, |
| "learning_rate": 4.2311111111111114e-06, |
| "loss": 0.1666, |
| "step": 3100 |
| }, |
| { |
| "epoch": 3.03, |
| "learning_rate": 4.175555555555556e-06, |
| "loss": 0.0935, |
| "step": 3125 |
| }, |
| { |
| "epoch": 3.05, |
| "learning_rate": 4.12e-06, |
| "loss": 0.1034, |
| "step": 3150 |
| }, |
| { |
| "epoch": 3.07, |
| "learning_rate": 4.064444444444444e-06, |
| "loss": 0.1053, |
| "step": 3175 |
| }, |
| { |
| "epoch": 3.1, |
| "learning_rate": 4.008888888888889e-06, |
| "loss": 0.1156, |
| "step": 3200 |
| }, |
| { |
| "epoch": 3.12, |
| "learning_rate": 3.953333333333333e-06, |
| "loss": 0.1107, |
| "step": 3225 |
| }, |
| { |
| "epoch": 3.15, |
| "learning_rate": 3.897777777777778e-06, |
| "loss": 0.1013, |
| "step": 3250 |
| }, |
| { |
| "epoch": 3.17, |
| "learning_rate": 3.842222222222223e-06, |
| "loss": 0.0944, |
| "step": 3275 |
| }, |
| { |
| "epoch": 3.19, |
| "learning_rate": 3.7866666666666667e-06, |
| "loss": 0.1033, |
| "step": 3300 |
| }, |
| { |
| "epoch": 3.22, |
| "learning_rate": 3.7311111111111116e-06, |
| "loss": 0.1022, |
| "step": 3325 |
| }, |
| { |
| "epoch": 3.24, |
| "learning_rate": 3.675555555555556e-06, |
| "loss": 0.094, |
| "step": 3350 |
| }, |
| { |
| "epoch": 3.27, |
| "learning_rate": 3.62e-06, |
| "loss": 0.1027, |
| "step": 3375 |
| }, |
| { |
| "epoch": 3.29, |
| "learning_rate": 3.564444444444445e-06, |
| "loss": 0.1025, |
| "step": 3400 |
| }, |
| { |
| "epoch": 3.32, |
| "learning_rate": 3.508888888888889e-06, |
| "loss": 0.1028, |
| "step": 3425 |
| }, |
| { |
| "epoch": 3.34, |
| "learning_rate": 3.4533333333333334e-06, |
| "loss": 0.0867, |
| "step": 3450 |
| }, |
| { |
| "epoch": 3.36, |
| "learning_rate": 3.3977777777777783e-06, |
| "loss": 0.1193, |
| "step": 3475 |
| }, |
| { |
| "epoch": 3.39, |
| "learning_rate": 3.3422222222222224e-06, |
| "loss": 0.0888, |
| "step": 3500 |
| }, |
| { |
| "epoch": 3.41, |
| "learning_rate": 3.286666666666667e-06, |
| "loss": 0.1077, |
| "step": 3525 |
| }, |
| { |
| "epoch": 3.44, |
| "learning_rate": 3.2311111111111117e-06, |
| "loss": 0.1018, |
| "step": 3550 |
| }, |
| { |
| "epoch": 3.46, |
| "learning_rate": 3.1755555555555557e-06, |
| "loss": 0.0933, |
| "step": 3575 |
| }, |
| { |
| "epoch": 3.48, |
| "learning_rate": 3.12e-06, |
| "loss": 0.0924, |
| "step": 3600 |
| }, |
| { |
| "epoch": 3.51, |
| "learning_rate": 3.064444444444445e-06, |
| "loss": 0.0903, |
| "step": 3625 |
| }, |
| { |
| "epoch": 3.53, |
| "learning_rate": 3.008888888888889e-06, |
| "loss": 0.1027, |
| "step": 3650 |
| }, |
| { |
| "epoch": 3.56, |
| "learning_rate": 2.9533333333333336e-06, |
| "loss": 0.1039, |
| "step": 3675 |
| }, |
| { |
| "epoch": 3.58, |
| "learning_rate": 2.8977777777777785e-06, |
| "loss": 0.1028, |
| "step": 3700 |
| }, |
| { |
| "epoch": 3.61, |
| "learning_rate": 2.8422222222222225e-06, |
| "loss": 0.1085, |
| "step": 3725 |
| }, |
| { |
| "epoch": 3.63, |
| "learning_rate": 2.786666666666667e-06, |
| "loss": 0.0982, |
| "step": 3750 |
| }, |
| { |
| "epoch": 3.65, |
| "learning_rate": 2.7311111111111114e-06, |
| "loss": 0.1153, |
| "step": 3775 |
| }, |
| { |
| "epoch": 3.68, |
| "learning_rate": 2.675555555555556e-06, |
| "loss": 0.1074, |
| "step": 3800 |
| }, |
| { |
| "epoch": 3.7, |
| "learning_rate": 2.6200000000000003e-06, |
| "loss": 0.1082, |
| "step": 3825 |
| }, |
| { |
| "epoch": 3.73, |
| "learning_rate": 2.5644444444444444e-06, |
| "loss": 0.1221, |
| "step": 3850 |
| }, |
| { |
| "epoch": 3.75, |
| "learning_rate": 2.5088888888888892e-06, |
| "loss": 0.1481, |
| "step": 3875 |
| }, |
| { |
| "epoch": 3.78, |
| "learning_rate": 2.4533333333333333e-06, |
| "loss": 0.1075, |
| "step": 3900 |
| }, |
| { |
| "epoch": 3.8, |
| "learning_rate": 2.397777777777778e-06, |
| "loss": 0.0991, |
| "step": 3925 |
| }, |
| { |
| "epoch": 3.82, |
| "learning_rate": 2.342222222222222e-06, |
| "loss": 0.1001, |
| "step": 3950 |
| }, |
| { |
| "epoch": 3.85, |
| "learning_rate": 2.2866666666666667e-06, |
| "loss": 0.0964, |
| "step": 3975 |
| }, |
| { |
| "epoch": 3.87, |
| "learning_rate": 2.2311111111111115e-06, |
| "loss": 0.1137, |
| "step": 4000 |
| }, |
| { |
| "epoch": 3.87, |
| "eval_loss": 0.37926188111305237, |
| "eval_runtime": 309.1205, |
| "eval_samples_per_second": 11.717, |
| "eval_steps_per_second": 2.931, |
| "eval_wer": 23.19300667126754, |
| "step": 4000 |
| }, |
| { |
| "epoch": 3.9, |
| "learning_rate": 2.1755555555555556e-06, |
| "loss": 0.1026, |
| "step": 4025 |
| }, |
| { |
| "epoch": 3.92, |
| "learning_rate": 2.12e-06, |
| "loss": 0.1151, |
| "step": 4050 |
| }, |
| { |
| "epoch": 3.94, |
| "learning_rate": 2.064444444444445e-06, |
| "loss": 0.0982, |
| "step": 4075 |
| }, |
| { |
| "epoch": 3.97, |
| "learning_rate": 2.008888888888889e-06, |
| "loss": 0.0953, |
| "step": 4100 |
| }, |
| { |
| "epoch": 3.99, |
| "learning_rate": 1.9533333333333334e-06, |
| "loss": 0.098, |
| "step": 4125 |
| }, |
| { |
| "epoch": 4.02, |
| "learning_rate": 1.8977777777777779e-06, |
| "loss": 0.0717, |
| "step": 4150 |
| }, |
| { |
| "epoch": 4.04, |
| "learning_rate": 1.8422222222222225e-06, |
| "loss": 0.0657, |
| "step": 4175 |
| }, |
| { |
| "epoch": 4.07, |
| "learning_rate": 1.7866666666666668e-06, |
| "loss": 0.0724, |
| "step": 4200 |
| }, |
| { |
| "epoch": 4.09, |
| "learning_rate": 1.7311111111111112e-06, |
| "loss": 0.0683, |
| "step": 4225 |
| }, |
| { |
| "epoch": 4.11, |
| "learning_rate": 1.675555555555556e-06, |
| "loss": 0.0703, |
| "step": 4250 |
| }, |
| { |
| "epoch": 4.14, |
| "learning_rate": 1.6200000000000002e-06, |
| "loss": 0.0711, |
| "step": 4275 |
| }, |
| { |
| "epoch": 4.16, |
| "learning_rate": 1.5644444444444446e-06, |
| "loss": 0.069, |
| "step": 4300 |
| }, |
| { |
| "epoch": 4.19, |
| "learning_rate": 1.5088888888888889e-06, |
| "loss": 0.0874, |
| "step": 4325 |
| }, |
| { |
| "epoch": 4.21, |
| "learning_rate": 1.4533333333333335e-06, |
| "loss": 0.0716, |
| "step": 4350 |
| }, |
| { |
| "epoch": 4.24, |
| "learning_rate": 1.397777777777778e-06, |
| "loss": 0.0618, |
| "step": 4375 |
| }, |
| { |
| "epoch": 4.26, |
| "learning_rate": 1.3422222222222222e-06, |
| "loss": 0.0646, |
| "step": 4400 |
| }, |
| { |
| "epoch": 4.28, |
| "learning_rate": 1.286666666666667e-06, |
| "loss": 0.0605, |
| "step": 4425 |
| }, |
| { |
| "epoch": 4.31, |
| "learning_rate": 1.2311111111111112e-06, |
| "loss": 0.0679, |
| "step": 4450 |
| }, |
| { |
| "epoch": 4.33, |
| "learning_rate": 1.1755555555555556e-06, |
| "loss": 0.0687, |
| "step": 4475 |
| }, |
| { |
| "epoch": 4.36, |
| "learning_rate": 1.12e-06, |
| "loss": 0.0675, |
| "step": 4500 |
| }, |
| { |
| "epoch": 4.38, |
| "learning_rate": 1.0644444444444445e-06, |
| "loss": 0.072, |
| "step": 4525 |
| }, |
| { |
| "epoch": 4.4, |
| "learning_rate": 1.008888888888889e-06, |
| "loss": 0.0698, |
| "step": 4550 |
| }, |
| { |
| "epoch": 4.43, |
| "learning_rate": 9.533333333333335e-07, |
| "loss": 0.0692, |
| "step": 4575 |
| }, |
| { |
| "epoch": 4.45, |
| "learning_rate": 8.977777777777778e-07, |
| "loss": 0.0808, |
| "step": 4600 |
| }, |
| { |
| "epoch": 4.48, |
| "learning_rate": 8.422222222222224e-07, |
| "loss": 0.0674, |
| "step": 4625 |
| }, |
| { |
| "epoch": 4.5, |
| "learning_rate": 7.866666666666667e-07, |
| "loss": 0.0773, |
| "step": 4650 |
| }, |
| { |
| "epoch": 4.53, |
| "learning_rate": 7.311111111111112e-07, |
| "loss": 0.0627, |
| "step": 4675 |
| }, |
| { |
| "epoch": 4.55, |
| "learning_rate": 6.755555555555555e-07, |
| "loss": 0.0627, |
| "step": 4700 |
| }, |
| { |
| "epoch": 4.57, |
| "learning_rate": 6.200000000000001e-07, |
| "loss": 0.0686, |
| "step": 4725 |
| }, |
| { |
| "epoch": 4.6, |
| "learning_rate": 5.644444444444445e-07, |
| "loss": 0.0619, |
| "step": 4750 |
| }, |
| { |
| "epoch": 4.62, |
| "learning_rate": 5.088888888888889e-07, |
| "loss": 0.0652, |
| "step": 4775 |
| }, |
| { |
| "epoch": 4.65, |
| "learning_rate": 4.533333333333334e-07, |
| "loss": 0.062, |
| "step": 4800 |
| }, |
| { |
| "epoch": 4.67, |
| "learning_rate": 3.9777777777777783e-07, |
| "loss": 0.0729, |
| "step": 4825 |
| }, |
| { |
| "epoch": 4.7, |
| "learning_rate": 3.422222222222223e-07, |
| "loss": 0.0703, |
| "step": 4850 |
| }, |
| { |
| "epoch": 4.72, |
| "learning_rate": 2.866666666666667e-07, |
| "loss": 0.0738, |
| "step": 4875 |
| }, |
| { |
| "epoch": 4.74, |
| "learning_rate": 2.3111111111111112e-07, |
| "loss": 0.066, |
| "step": 4900 |
| }, |
| { |
| "epoch": 4.77, |
| "learning_rate": 1.7555555555555558e-07, |
| "loss": 0.0702, |
| "step": 4925 |
| }, |
| { |
| "epoch": 4.79, |
| "learning_rate": 1.2000000000000002e-07, |
| "loss": 0.0702, |
| "step": 4950 |
| }, |
| { |
| "epoch": 4.82, |
| "learning_rate": 6.444444444444445e-08, |
| "loss": 0.0772, |
| "step": 4975 |
| }, |
| { |
| "epoch": 4.84, |
| "learning_rate": 8.88888888888889e-09, |
| "loss": 0.0632, |
| "step": 5000 |
| }, |
| { |
| "epoch": 4.84, |
| "eval_loss": 0.38927504420280457, |
| "eval_runtime": 319.9218, |
| "eval_samples_per_second": 11.322, |
| "eval_steps_per_second": 2.832, |
| "eval_wer": 23.501265240395675, |
| "step": 5000 |
| }, |
| { |
| "epoch": 4.84, |
| "step": 5000, |
| "total_flos": 2.59362152644608e+18, |
| "train_loss": 0.2751132179021835, |
| "train_runtime": 3844.695, |
| "train_samples_per_second": 10.404, |
| "train_steps_per_second": 1.3 |
| } |
| ], |
| "max_steps": 5000, |
| "num_train_epochs": 5, |
| "total_flos": 2.59362152644608e+18, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
