| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 33.78378378378378, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.4e-07, | |
| "loss": 1.0761, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 0.9171, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.44e-06, | |
| "loss": 0.9707, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.94e-06, | |
| "loss": 1.0289, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.9489, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.9441, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.44e-06, | |
| "loss": 1.5917, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.94e-06, | |
| "loss": 1.6768, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 1.5563, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.94e-06, | |
| "loss": 1.6267, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 1.8075, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 5.94e-06, | |
| "loss": 1.804, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 2.3545, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 3.6551, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 4.6441, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 7.92e-06, | |
| "loss": 5.3999, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 8.42e-06, | |
| "loss": 6.0272, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 5.7311, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 9.42e-06, | |
| "loss": 6.0189, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 7.2512, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 9.953333333333333e-06, | |
| "loss": 6.7626, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 9.89777777777778e-06, | |
| "loss": 5.912, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 9.842222222222223e-06, | |
| "loss": 7.2678, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 9.786666666666667e-06, | |
| "loss": 7.7191, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 9.731111111111113e-06, | |
| "loss": 5.3031, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 9.675555555555555e-06, | |
| "loss": 7.4638, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 9.620000000000001e-06, | |
| "loss": 6.5911, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 9.564444444444445e-06, | |
| "loss": 5.5009, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 9.508888888888889e-06, | |
| "loss": 7.2869, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 9.455555555555557e-06, | |
| "loss": 6.3539, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "learning_rate": 9.402222222222222e-06, | |
| "loss": 4.408, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 9.346666666666666e-06, | |
| "loss": 6.2703, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 9.291111111111112e-06, | |
| "loss": 6.6461, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 9.235555555555556e-06, | |
| "loss": 5.4521, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 9.180000000000002e-06, | |
| "loss": 6.3987, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 9.124444444444444e-06, | |
| "loss": 6.1241, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 9.06888888888889e-06, | |
| "loss": 4.0049, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 9.013333333333334e-06, | |
| "loss": 5.8297, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 8.957777777777778e-06, | |
| "loss": 5.7196, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 8.902222222222224e-06, | |
| "loss": 4.9319, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "eval_loss": 10.077404975891113, | |
| "eval_runtime": 246.976, | |
| "eval_samples_per_second": 9.973, | |
| "eval_steps_per_second": 0.624, | |
| "eval_wer": 73.98924171438487, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 8.846666666666668e-06, | |
| "loss": 5.367, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 8.791111111111112e-06, | |
| "loss": 5.9988, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "learning_rate": 8.735555555555556e-06, | |
| "loss": 3.7539, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 7.43, | |
| "learning_rate": 8.68e-06, | |
| "loss": 4.6308, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 8.624444444444446e-06, | |
| "loss": 5.3367, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 8.56888888888889e-06, | |
| "loss": 4.8021, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 8.513333333333335e-06, | |
| "loss": 4.3516, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 8.457777777777778e-06, | |
| "loss": 5.1473, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 8.28, | |
| "learning_rate": 8.402222222222223e-06, | |
| "loss": 4.1365, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 8.45, | |
| "learning_rate": 8.346666666666668e-06, | |
| "loss": 3.7336, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 8.291111111111112e-06, | |
| "loss": 4.9819, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 8.78, | |
| "learning_rate": 8.235555555555557e-06, | |
| "loss": 4.1889, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 8.95, | |
| "learning_rate": 8.18e-06, | |
| "loss": 3.6101, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 9.12, | |
| "learning_rate": 8.124444444444445e-06, | |
| "loss": 4.9865, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 9.29, | |
| "learning_rate": 8.06888888888889e-06, | |
| "loss": 4.0579, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 9.46, | |
| "learning_rate": 8.013333333333333e-06, | |
| "loss": 3.2807, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 9.63, | |
| "learning_rate": 7.957777777777779e-06, | |
| "loss": 5.2015, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 7.902222222222223e-06, | |
| "loss": 4.2603, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 9.97, | |
| "learning_rate": 7.846666666666667e-06, | |
| "loss": 2.8056, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "learning_rate": 7.791111111111111e-06, | |
| "loss": 4.4941, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 10.3, | |
| "learning_rate": 7.735555555555557e-06, | |
| "loss": 3.6729, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 10.47, | |
| "learning_rate": 7.680000000000001e-06, | |
| "loss": 2.5204, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 10.64, | |
| "learning_rate": 7.624444444444445e-06, | |
| "loss": 4.6931, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 10.81, | |
| "learning_rate": 7.56888888888889e-06, | |
| "loss": 4.064, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 10.98, | |
| "learning_rate": 7.513333333333334e-06, | |
| "loss": 2.5234, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 11.15, | |
| "learning_rate": 7.457777777777778e-06, | |
| "loss": 3.3534, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 11.32, | |
| "learning_rate": 7.402222222222223e-06, | |
| "loss": 3.4619, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 11.49, | |
| "learning_rate": 7.346666666666668e-06, | |
| "loss": 2.6103, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 11.66, | |
| "learning_rate": 7.291111111111112e-06, | |
| "loss": 3.8041, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 11.82, | |
| "learning_rate": 7.235555555555556e-06, | |
| "loss": 3.5229, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 11.99, | |
| "learning_rate": 7.180000000000001e-06, | |
| "loss": 2.5819, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 12.16, | |
| "learning_rate": 7.124444444444445e-06, | |
| "loss": 3.2352, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 12.33, | |
| "learning_rate": 7.06888888888889e-06, | |
| "loss": 3.1571, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 7.0133333333333345e-06, | |
| "loss": 2.2571, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 12.67, | |
| "learning_rate": 6.9577777777777785e-06, | |
| "loss": 3.1133, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 12.84, | |
| "learning_rate": 6.902222222222223e-06, | |
| "loss": 3.0995, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 6.846666666666667e-06, | |
| "loss": 2.1959, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 13.18, | |
| "learning_rate": 6.7911111111111115e-06, | |
| "loss": 2.8142, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 13.34, | |
| "learning_rate": 6.735555555555556e-06, | |
| "loss": 3.1267, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 13.51, | |
| "learning_rate": 6.680000000000001e-06, | |
| "loss": 2.6116, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 13.51, | |
| "eval_loss": 11.408943176269531, | |
| "eval_runtime": 241.1367, | |
| "eval_samples_per_second": 10.214, | |
| "eval_steps_per_second": 0.639, | |
| "eval_wer": 67.04841228526809, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 13.68, | |
| "learning_rate": 6.6244444444444445e-06, | |
| "loss": 2.517, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 13.85, | |
| "learning_rate": 6.568888888888889e-06, | |
| "loss": 3.0989, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 6.513333333333333e-06, | |
| "loss": 2.0728, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 14.19, | |
| "learning_rate": 6.457777777777778e-06, | |
| "loss": 2.0813, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 14.36, | |
| "learning_rate": 6.402222222222223e-06, | |
| "loss": 3.0881, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 14.53, | |
| "learning_rate": 6.346666666666668e-06, | |
| "loss": 2.3762, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 14.7, | |
| "learning_rate": 6.291111111111111e-06, | |
| "loss": 2.2093, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 14.86, | |
| "learning_rate": 6.235555555555556e-06, | |
| "loss": 2.8965, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 6.18e-06, | |
| "loss": 2.3463, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 15.2, | |
| "learning_rate": 6.124444444444445e-06, | |
| "loss": 1.6227, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 15.37, | |
| "learning_rate": 6.06888888888889e-06, | |
| "loss": 2.7423, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 15.54, | |
| "learning_rate": 6.013333333333335e-06, | |
| "loss": 2.522, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 15.71, | |
| "learning_rate": 5.957777777777778e-06, | |
| "loss": 1.6232, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 15.88, | |
| "learning_rate": 5.902222222222223e-06, | |
| "loss": 2.6979, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 16.05, | |
| "learning_rate": 5.846666666666667e-06, | |
| "loss": 2.0168, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 16.22, | |
| "learning_rate": 5.791111111111112e-06, | |
| "loss": 1.3373, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 16.39, | |
| "learning_rate": 5.735555555555557e-06, | |
| "loss": 2.2285, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 16.55, | |
| "learning_rate": 5.68e-06, | |
| "loss": 2.3166, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 16.72, | |
| "learning_rate": 5.624444444444445e-06, | |
| "loss": 1.387, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 16.89, | |
| "learning_rate": 5.56888888888889e-06, | |
| "loss": 2.4226, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 17.06, | |
| "learning_rate": 5.513333333333334e-06, | |
| "loss": 2.0713, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 17.23, | |
| "learning_rate": 5.4577777777777785e-06, | |
| "loss": 1.2785, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 17.4, | |
| "learning_rate": 5.402222222222223e-06, | |
| "loss": 1.8165, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 17.57, | |
| "learning_rate": 5.346666666666667e-06, | |
| "loss": 1.9608, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 17.74, | |
| "learning_rate": 5.2911111111111115e-06, | |
| "loss": 1.5858, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 17.91, | |
| "learning_rate": 5.235555555555556e-06, | |
| "loss": 2.024, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 18.07, | |
| "learning_rate": 5.18e-06, | |
| "loss": 1.8938, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 18.24, | |
| "learning_rate": 5.124444444444445e-06, | |
| "loss": 1.1712, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 18.41, | |
| "learning_rate": 5.06888888888889e-06, | |
| "loss": 1.8954, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 18.58, | |
| "learning_rate": 5.013333333333333e-06, | |
| "loss": 1.7302, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 18.75, | |
| "learning_rate": 4.957777777777778e-06, | |
| "loss": 1.3895, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 18.92, | |
| "learning_rate": 4.902222222222222e-06, | |
| "loss": 1.4083, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 19.09, | |
| "learning_rate": 4.846666666666667e-06, | |
| "loss": 1.7037, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 19.26, | |
| "learning_rate": 4.791111111111111e-06, | |
| "loss": 1.0114, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 19.43, | |
| "learning_rate": 4.735555555555556e-06, | |
| "loss": 1.4758, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 19.59, | |
| "learning_rate": 4.680000000000001e-06, | |
| "loss": 1.7263, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 19.76, | |
| "learning_rate": 4.624444444444445e-06, | |
| "loss": 1.311, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 19.93, | |
| "learning_rate": 4.568888888888889e-06, | |
| "loss": 1.1316, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 20.1, | |
| "learning_rate": 4.513333333333333e-06, | |
| "loss": 1.5401, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 20.27, | |
| "learning_rate": 4.457777777777778e-06, | |
| "loss": 0.9607, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 20.27, | |
| "eval_loss": 11.82657241821289, | |
| "eval_runtime": 243.7624, | |
| "eval_samples_per_second": 10.104, | |
| "eval_steps_per_second": 0.632, | |
| "eval_wer": 60.944820406038524, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 20.44, | |
| "learning_rate": 4.402222222222223e-06, | |
| "loss": 1.2625, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 20.61, | |
| "learning_rate": 4.346666666666667e-06, | |
| "loss": 1.4382, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 20.78, | |
| "learning_rate": 4.291111111111112e-06, | |
| "loss": 1.1822, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 20.95, | |
| "learning_rate": 4.235555555555556e-06, | |
| "loss": 0.683, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 21.11, | |
| "learning_rate": 4.18e-06, | |
| "loss": 1.3987, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 21.28, | |
| "learning_rate": 4.124444444444445e-06, | |
| "loss": 1.0554, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 21.45, | |
| "learning_rate": 4.0688888888888896e-06, | |
| "loss": 0.9125, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 21.62, | |
| "learning_rate": 4.013333333333334e-06, | |
| "loss": 1.2349, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 21.79, | |
| "learning_rate": 3.9577777777777785e-06, | |
| "loss": 1.4298, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 21.96, | |
| "learning_rate": 3.9022222222222225e-06, | |
| "loss": 0.8018, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 22.13, | |
| "learning_rate": 3.8466666666666665e-06, | |
| "loss": 1.2586, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 22.3, | |
| "learning_rate": 3.7911111111111114e-06, | |
| "loss": 0.9665, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 22.47, | |
| "learning_rate": 3.7355555555555555e-06, | |
| "loss": 0.4815, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 22.64, | |
| "learning_rate": 3.6800000000000003e-06, | |
| "loss": 1.1611, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 22.8, | |
| "learning_rate": 3.624444444444445e-06, | |
| "loss": 0.9511, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 22.97, | |
| "learning_rate": 3.568888888888889e-06, | |
| "loss": 0.4956, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 23.14, | |
| "learning_rate": 3.5133333333333337e-06, | |
| "loss": 0.7079, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 23.31, | |
| "learning_rate": 3.457777777777778e-06, | |
| "loss": 0.7881, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 23.48, | |
| "learning_rate": 3.4022222222222222e-06, | |
| "loss": 0.3986, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 23.65, | |
| "learning_rate": 3.346666666666667e-06, | |
| "loss": 0.8256, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 23.82, | |
| "learning_rate": 3.2911111111111116e-06, | |
| "loss": 0.6971, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 23.99, | |
| "learning_rate": 3.2355555555555556e-06, | |
| "loss": 0.5003, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 24.16, | |
| "learning_rate": 3.1800000000000005e-06, | |
| "loss": 0.7882, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 24.32, | |
| "learning_rate": 3.124444444444445e-06, | |
| "loss": 0.5888, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 24.49, | |
| "learning_rate": 3.068888888888889e-06, | |
| "loss": 0.6318, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 24.66, | |
| "learning_rate": 3.013333333333334e-06, | |
| "loss": 0.8246, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 24.83, | |
| "learning_rate": 2.957777777777778e-06, | |
| "loss": 0.7554, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 2.9022222222222223e-06, | |
| "loss": 0.5072, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 25.17, | |
| "learning_rate": 2.8466666666666672e-06, | |
| "loss": 0.6132, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 25.34, | |
| "learning_rate": 2.7911111111111113e-06, | |
| "loss": 0.661, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 25.51, | |
| "learning_rate": 2.7355555555555557e-06, | |
| "loss": 0.4763, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 25.68, | |
| "learning_rate": 2.68e-06, | |
| "loss": 0.5485, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 25.84, | |
| "learning_rate": 2.6244444444444446e-06, | |
| "loss": 0.5781, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 26.01, | |
| "learning_rate": 2.568888888888889e-06, | |
| "loss": 0.4007, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 26.18, | |
| "learning_rate": 2.5133333333333336e-06, | |
| "loss": 0.3914, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 26.35, | |
| "learning_rate": 2.457777777777778e-06, | |
| "loss": 0.4562, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 26.52, | |
| "learning_rate": 2.4022222222222225e-06, | |
| "loss": 0.3649, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 26.69, | |
| "learning_rate": 2.346666666666667e-06, | |
| "loss": 0.3235, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 26.86, | |
| "learning_rate": 2.2911111111111114e-06, | |
| "loss": 0.5641, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 27.03, | |
| "learning_rate": 2.235555555555556e-06, | |
| "loss": 0.3464, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 27.03, | |
| "eval_loss": 9.949957847595215, | |
| "eval_runtime": 246.2029, | |
| "eval_samples_per_second": 10.004, | |
| "eval_steps_per_second": 0.626, | |
| "eval_wer": 52.121290994273814, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 27.2, | |
| "learning_rate": 2.1800000000000003e-06, | |
| "loss": 0.237, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 27.36, | |
| "learning_rate": 2.1244444444444443e-06, | |
| "loss": 0.2766, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 27.53, | |
| "learning_rate": 2.0688888888888892e-06, | |
| "loss": 0.2813, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 27.7, | |
| "learning_rate": 2.0133333333333337e-06, | |
| "loss": 0.1791, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 27.87, | |
| "learning_rate": 1.9577777777777777e-06, | |
| "loss": 0.2774, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 28.04, | |
| "learning_rate": 1.9022222222222222e-06, | |
| "loss": 0.2763, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 28.21, | |
| "learning_rate": 1.8466666666666668e-06, | |
| "loss": 0.1649, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 28.38, | |
| "learning_rate": 1.7911111111111113e-06, | |
| "loss": 0.3234, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 28.55, | |
| "learning_rate": 1.7355555555555555e-06, | |
| "loss": 0.3068, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 28.72, | |
| "learning_rate": 1.6800000000000002e-06, | |
| "loss": 0.2139, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 28.89, | |
| "learning_rate": 1.6244444444444447e-06, | |
| "loss": 0.3518, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 29.05, | |
| "learning_rate": 1.568888888888889e-06, | |
| "loss": 0.321, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 29.22, | |
| "learning_rate": 1.5133333333333334e-06, | |
| "loss": 0.0825, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 29.39, | |
| "learning_rate": 1.457777777777778e-06, | |
| "loss": 0.1898, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 29.56, | |
| "learning_rate": 1.4022222222222223e-06, | |
| "loss": 0.219, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 29.73, | |
| "learning_rate": 1.3466666666666668e-06, | |
| "loss": 0.1058, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 29.9, | |
| "learning_rate": 1.2911111111111112e-06, | |
| "loss": 0.1666, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 30.07, | |
| "learning_rate": 1.2355555555555557e-06, | |
| "loss": 0.1786, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 30.24, | |
| "learning_rate": 1.1800000000000001e-06, | |
| "loss": 0.0728, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 30.41, | |
| "learning_rate": 1.1244444444444446e-06, | |
| "loss": 0.1202, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 30.57, | |
| "learning_rate": 1.068888888888889e-06, | |
| "loss": 0.1826, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 30.74, | |
| "learning_rate": 1.0133333333333333e-06, | |
| "loss": 0.1013, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 30.91, | |
| "learning_rate": 9.57777777777778e-07, | |
| "loss": 0.1335, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 31.08, | |
| "learning_rate": 9.022222222222222e-07, | |
| "loss": 0.062, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 31.25, | |
| "learning_rate": 8.466666666666668e-07, | |
| "loss": 0.0308, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 31.42, | |
| "learning_rate": 7.911111111111111e-07, | |
| "loss": 0.0679, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 31.59, | |
| "learning_rate": 7.355555555555556e-07, | |
| "loss": 0.0583, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 31.76, | |
| "learning_rate": 6.800000000000001e-07, | |
| "loss": 0.044, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 31.93, | |
| "learning_rate": 6.244444444444445e-07, | |
| "loss": 0.047, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 32.09, | |
| "learning_rate": 5.68888888888889e-07, | |
| "loss": 0.0376, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 32.26, | |
| "learning_rate": 5.133333333333334e-07, | |
| "loss": 0.0268, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 32.43, | |
| "learning_rate": 4.5777777777777784e-07, | |
| "loss": 0.0392, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 32.6, | |
| "learning_rate": 4.0222222222222224e-07, | |
| "loss": 0.0264, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 32.77, | |
| "learning_rate": 3.466666666666667e-07, | |
| "loss": 0.0248, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 32.94, | |
| "learning_rate": 2.9111111111111116e-07, | |
| "loss": 0.0182, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 33.11, | |
| "learning_rate": 2.3555555555555556e-07, | |
| "loss": 0.0149, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 33.28, | |
| "learning_rate": 1.8e-07, | |
| "loss": 0.0076, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 33.45, | |
| "learning_rate": 1.2444444444444446e-07, | |
| "loss": 0.0084, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 33.61, | |
| "learning_rate": 6.888888888888889e-08, | |
| "loss": 0.0159, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 33.78, | |
| "learning_rate": 1.3333333333333334e-08, | |
| "loss": 0.0122, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 33.78, | |
| "eval_loss": 9.061223030090332, | |
| "eval_runtime": 246.5625, | |
| "eval_samples_per_second": 9.989, | |
| "eval_steps_per_second": 0.625, | |
| "eval_wer": 44.67291341315287, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 33.78, | |
| "step": 5000, | |
| "total_flos": 4.598319781272512e+19, | |
| "train_loss": 2.2148335320025683, | |
| "train_runtime": 12302.5688, | |
| "train_samples_per_second": 13.005, | |
| "train_steps_per_second": 0.406 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 34, | |
| "total_flos": 4.598319781272512e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |