| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "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": 1.0763, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.0817, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.44e-06, | |
| "loss": 0.74, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.94e-06, | |
| "loss": 0.4913, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.3163, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.336, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.3172, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.94e-06, | |
| "loss": 0.327, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.3405, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.2707, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.2802, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.3344, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.2622, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.3155, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.3077, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.2688, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.2683, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.3679, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.4573, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.2967, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 0.3756, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 0.348, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.84e-06, | |
| "loss": 0.3708, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 0.3615, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 0.3703, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 0.3479, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 0.3115, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 0.3658, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 0.3931, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 0.31, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 0.3707, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 0.3718, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 0.4266, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 0.3604, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 0.3439, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 0.3209, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 0.2725, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 0.3711, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 0.3127, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 0.3357, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.3994442820549011, | |
| "eval_runtime": 16638.6565, | |
| "eval_samples_per_second": 0.967, | |
| "eval_steps_per_second": 0.484, | |
| "eval_wer": 16.152331704540064, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.84e-06, | |
| "loss": 0.2788, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 0.33, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.3531, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.3944, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.3192, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.3676, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.3017, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.3797, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.296, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.4296, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.3567, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.329, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.2992, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 0.3141, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 0.3644, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 0.3339, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.3149, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.355, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 0.2989, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.3741, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 0.3216, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.2841, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.3377, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.3056, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.3389, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 0.3007, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.3776, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.3124, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.2804, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.3909, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.3042, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.3236, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.3314, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.2618, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.3153, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.3495, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.3312, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.3386, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.2859, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.3026, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.38022664189338684, | |
| "eval_runtime": 14693.0094, | |
| "eval_samples_per_second": 1.095, | |
| "eval_steps_per_second": 0.548, | |
| "eval_wer": 15.240286541107261, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.2774, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.3064, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 0.3234, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 0.2814, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 0.3351, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.34e-06, | |
| "loss": 0.3656, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.3869, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.3103, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.3658, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.3058, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.3356, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 6.00888888888889e-06, | |
| "loss": 0.316, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.9533333333333345e-06, | |
| "loss": 0.3579, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.897777777777778e-06, | |
| "loss": 0.2479, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.8422222222222226e-06, | |
| "loss": 0.335, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.7866666666666674e-06, | |
| "loss": 0.3542, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.7311111111111115e-06, | |
| "loss": 0.3613, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.675555555555556e-06, | |
| "loss": 0.3314, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.620000000000001e-06, | |
| "loss": 0.4385, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.5644444444444444e-06, | |
| "loss": 0.2933, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.508888888888889e-06, | |
| "loss": 0.3774, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.453333333333334e-06, | |
| "loss": 0.3324, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.397777777777778e-06, | |
| "loss": 0.2947, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.342222222222223e-06, | |
| "loss": 0.2813, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.286666666666666e-06, | |
| "loss": 0.3309, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.231111111111111e-06, | |
| "loss": 0.2638, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.175555555555556e-06, | |
| "loss": 0.2992, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.12e-06, | |
| "loss": 0.3311, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.064444444444445e-06, | |
| "loss": 0.2738, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.00888888888889e-06, | |
| "loss": 0.3203, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.953333333333334e-06, | |
| "loss": 0.3259, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.897777777777778e-06, | |
| "loss": 0.3107, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.842222222222223e-06, | |
| "loss": 0.2592, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.786666666666667e-06, | |
| "loss": 0.2741, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.7333333333333335e-06, | |
| "loss": 0.2967, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.677777777777778e-06, | |
| "loss": 0.3071, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.622222222222222e-06, | |
| "loss": 0.2919, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.566666666666667e-06, | |
| "loss": 0.2978, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.511111111111111e-06, | |
| "loss": 0.2823, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.455555555555555e-06, | |
| "loss": 0.2904, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 0.3389354944229126, | |
| "eval_runtime": 14559.0936, | |
| "eval_samples_per_second": 1.105, | |
| "eval_steps_per_second": 0.553, | |
| "eval_wer": 14.004459414583234, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.3481, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.344444444444445e-06, | |
| "loss": 0.2902, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.288888888888889e-06, | |
| "loss": 0.2798, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.233333333333334e-06, | |
| "loss": 0.2829, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.177777777777778e-06, | |
| "loss": 0.3101, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.122222222222222e-06, | |
| "loss": 0.32, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.066666666666667e-06, | |
| "loss": 0.2372, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.011111111111111e-06, | |
| "loss": 0.2881, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.955555555555556e-06, | |
| "loss": 0.2969, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.3325, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.844444444444445e-06, | |
| "loss": 0.2593, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.7888888888888893e-06, | |
| "loss": 0.2411, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.7333333333333337e-06, | |
| "loss": 0.2914, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.6777777777777778e-06, | |
| "loss": 0.2758, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.6222222222222226e-06, | |
| "loss": 0.3043, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.566666666666667e-06, | |
| "loss": 0.2804, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.511111111111111e-06, | |
| "loss": 0.3264, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.455555555555556e-06, | |
| "loss": 0.288, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.2435, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.3444444444444445e-06, | |
| "loss": 0.2529, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.2888888888888894e-06, | |
| "loss": 0.2523, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.2333333333333334e-06, | |
| "loss": 0.2609, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.177777777777778e-06, | |
| "loss": 0.3121, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.1222222222222228e-06, | |
| "loss": 0.2768, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.066666666666667e-06, | |
| "loss": 0.2618, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.0111111111111113e-06, | |
| "loss": 0.2714, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.955555555555556e-06, | |
| "loss": 0.2898, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.3162, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.8444444444444446e-06, | |
| "loss": 0.2712, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.788888888888889e-06, | |
| "loss": 0.2709, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.7333333333333336e-06, | |
| "loss": 0.2843, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.677777777777778e-06, | |
| "loss": 0.2435, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.6222222222222225e-06, | |
| "loss": 0.2728, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.566666666666667e-06, | |
| "loss": 0.306, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.5111111111111114e-06, | |
| "loss": 0.292, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.455555555555556e-06, | |
| "loss": 0.245, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.2855, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.3444444444444448e-06, | |
| "loss": 0.3038, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.2888888888888892e-06, | |
| "loss": 0.3116, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.2333333333333333e-06, | |
| "loss": 0.2407, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 0.31351420283317566, | |
| "eval_runtime": 14971.3808, | |
| "eval_samples_per_second": 1.075, | |
| "eval_steps_per_second": 0.537, | |
| "eval_wer": 12.794724607429195, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.1777777777777777e-06, | |
| "loss": 0.3365, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.1222222222222226e-06, | |
| "loss": 0.2981, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.0666666666666666e-06, | |
| "loss": 0.317, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.011111111111111e-06, | |
| "loss": 0.3007, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.955555555555556e-06, | |
| "loss": 0.3242, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.3283, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.8444444444444445e-06, | |
| "loss": 0.3226, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.788888888888889e-06, | |
| "loss": 0.2865, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.7333333333333336e-06, | |
| "loss": 0.2912, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.6777777777777779e-06, | |
| "loss": 0.2986, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.6222222222222223e-06, | |
| "loss": 0.3676, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.566666666666667e-06, | |
| "loss": 0.2485, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.5111111111111112e-06, | |
| "loss": 0.2376, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.4555555555555557e-06, | |
| "loss": 0.2854, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.2949, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.3444444444444446e-06, | |
| "loss": 0.2066, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.288888888888889e-06, | |
| "loss": 0.2747, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2333333333333335e-06, | |
| "loss": 0.2533, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1777777777777778e-06, | |
| "loss": 0.2458, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1222222222222222e-06, | |
| "loss": 0.322, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.066666666666667e-06, | |
| "loss": 0.2344, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.0111111111111111e-06, | |
| "loss": 0.2701, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.555555555555556e-07, | |
| "loss": 0.2548, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.2167, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.444444444444445e-07, | |
| "loss": 0.221, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.888888888888889e-07, | |
| "loss": 0.2126, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 7.333333333333334e-07, | |
| "loss": 0.3039, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.777777777777779e-07, | |
| "loss": 0.2154, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.222222222222223e-07, | |
| "loss": 0.2784, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.666666666666667e-07, | |
| "loss": 0.2739, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.111111111111112e-07, | |
| "loss": 0.2935, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.5555555555555563e-07, | |
| "loss": 0.23, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.2517, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.444444444444445e-07, | |
| "loss": 0.2482, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.888888888888889e-07, | |
| "loss": 0.3264, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.3333333333333336e-07, | |
| "loss": 0.2461, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.777777777777778e-07, | |
| "loss": 0.2825, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.2222222222222225e-07, | |
| "loss": 0.2603, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.666666666666668e-08, | |
| "loss": 0.2529, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.1111111111111112e-08, | |
| "loss": 0.2451, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.29616352915763855, | |
| "eval_runtime": 14356.5014, | |
| "eval_samples_per_second": 1.121, | |
| "eval_steps_per_second": 0.56, | |
| "eval_wer": 11.909957777883202, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 5000, | |
| "total_flos": 4.2463420416e+19, | |
| "train_loss": 0.31813771924972534, | |
| "train_runtime": 92330.5211, | |
| "train_samples_per_second": 0.217, | |
| "train_steps_per_second": 0.054 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 4.2463420416e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |