| { | |
| "best_metric": 10.978180052967598, | |
| "best_model_checkpoint": "./checkpoint-10000", | |
| "epoch": 7.0739, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4e-07, | |
| "loss": 0.9795, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 0.5769, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.44e-06, | |
| "loss": 0.6181, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.94e-06, | |
| "loss": 0.6418, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.5181, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.5274, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.5356, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.94e-06, | |
| "loss": 0.4311, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.7091, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.6847, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.524, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.4687, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.3131, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.3863, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.3673, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.3114, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.5311, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.5967, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.5192, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.4347, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.976842105263158e-06, | |
| "loss": 0.6145, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.950526315789475e-06, | |
| "loss": 0.6516, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.92421052631579e-06, | |
| "loss": 0.6159, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.897894736842107e-06, | |
| "loss": 0.6635, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.871578947368422e-06, | |
| "loss": 0.5778, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.845263157894738e-06, | |
| "loss": 0.5212, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.818947368421053e-06, | |
| "loss": 0.6252, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.79263157894737e-06, | |
| "loss": 0.3836, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.766315789473685e-06, | |
| "loss": 0.421, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.74e-06, | |
| "loss": 0.4511, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.713684210526317e-06, | |
| "loss": 0.4606, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.687368421052632e-06, | |
| "loss": 0.4089, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.661052631578948e-06, | |
| "loss": 0.4603, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.634736842105265e-06, | |
| "loss": 0.3939, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.60842105263158e-06, | |
| "loss": 0.3767, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.582105263157897e-06, | |
| "loss": 0.4673, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.555789473684211e-06, | |
| "loss": 0.4722, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.529473684210528e-06, | |
| "loss": 0.6518, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.503157894736843e-06, | |
| "loss": 0.5479, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.476842105263158e-06, | |
| "loss": 0.3998, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_cer": 19.12609606675638, | |
| "eval_loss": 0.2898296117782593, | |
| "eval_runtime": 2637.8663, | |
| "eval_samples_per_second": 4.011, | |
| "eval_steps_per_second": 0.502, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.450526315789475e-06, | |
| "loss": 0.3209, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.42421052631579e-06, | |
| "loss": 0.3491, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.397894736842106e-06, | |
| "loss": 0.3275, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.371578947368421e-06, | |
| "loss": 0.3347, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.345263157894738e-06, | |
| "loss": 0.4085, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.318947368421053e-06, | |
| "loss": 0.6453, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.292631578947368e-06, | |
| "loss": 0.3929, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.266315789473685e-06, | |
| "loss": 0.5635, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.240000000000001e-06, | |
| "loss": 0.6003, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.213684210526316e-06, | |
| "loss": 0.4708, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.187368421052633e-06, | |
| "loss": 0.3895, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.161052631578948e-06, | |
| "loss": 0.4032, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.134736842105265e-06, | |
| "loss": 0.3869, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.10842105263158e-06, | |
| "loss": 0.4624, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.082105263157896e-06, | |
| "loss": 0.4014, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.055789473684211e-06, | |
| "loss": 0.4537, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.029473684210526e-06, | |
| "loss": 0.4391, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.003157894736843e-06, | |
| "loss": 0.3783, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.976842105263158e-06, | |
| "loss": 0.3334, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.950526315789474e-06, | |
| "loss": 0.4029, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.92421052631579e-06, | |
| "loss": 0.2823, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.897894736842106e-06, | |
| "loss": 0.5717, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.871578947368423e-06, | |
| "loss": 0.5025, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.845263157894738e-06, | |
| "loss": 0.2511, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.818947368421054e-06, | |
| "loss": 0.3029, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.79263157894737e-06, | |
| "loss": 0.2048, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.766315789473684e-06, | |
| "loss": 0.2222, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.740000000000001e-06, | |
| "loss": 0.1859, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.713684210526316e-06, | |
| "loss": 0.2013, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.687368421052633e-06, | |
| "loss": 0.2973, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.661052631578948e-06, | |
| "loss": 0.4246, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.634736842105264e-06, | |
| "loss": 0.301, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.60842105263158e-06, | |
| "loss": 0.264, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.582105263157894e-06, | |
| "loss": 0.3699, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.555789473684213e-06, | |
| "loss": 0.4455, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.529473684210528e-06, | |
| "loss": 0.2977, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.503157894736843e-06, | |
| "loss": 0.4167, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.47684210526316e-06, | |
| "loss": 0.3812, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.450526315789474e-06, | |
| "loss": 0.3696, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.42421052631579e-06, | |
| "loss": 0.2414, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "eval_cer": 12.776128941078154, | |
| "eval_loss": 0.28260529041290283, | |
| "eval_runtime": 2599.7667, | |
| "eval_samples_per_second": 4.07, | |
| "eval_steps_per_second": 0.509, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.397894736842106e-06, | |
| "loss": 0.2162, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.371578947368422e-06, | |
| "loss": 0.2105, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.345263157894737e-06, | |
| "loss": 0.2479, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.318947368421052e-06, | |
| "loss": 0.1812, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.292631578947369e-06, | |
| "loss": 0.2211, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.266315789473684e-06, | |
| "loss": 0.2244, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.24e-06, | |
| "loss": 0.1966, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.213684210526316e-06, | |
| "loss": 0.161, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.187368421052632e-06, | |
| "loss": 0.2416, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.161052631578949e-06, | |
| "loss": 0.2791, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.134736842105264e-06, | |
| "loss": 0.3774, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.10842105263158e-06, | |
| "loss": 0.3016, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.082105263157896e-06, | |
| "loss": 0.2055, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.05578947368421e-06, | |
| "loss": 0.1777, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.029473684210527e-06, | |
| "loss": 0.203, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.003157894736842e-06, | |
| "loss": 0.1607, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.976842105263159e-06, | |
| "loss": 0.1528, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.950526315789474e-06, | |
| "loss": 0.2418, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.92421052631579e-06, | |
| "loss": 0.3166, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.897894736842106e-06, | |
| "loss": 0.2895, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.87157894736842e-06, | |
| "loss": 0.3518, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.845263157894739e-06, | |
| "loss": 0.3336, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.818947368421054e-06, | |
| "loss": 0.2864, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.792631578947369e-06, | |
| "loss": 0.2859, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.766315789473685e-06, | |
| "loss": 0.1899, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 7.74e-06, | |
| "loss": 0.2016, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 7.713684210526317e-06, | |
| "loss": 0.2545, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.687368421052632e-06, | |
| "loss": 0.215, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.661052631578949e-06, | |
| "loss": 0.3004, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.634736842105264e-06, | |
| "loss": 0.2364, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.6084210526315795e-06, | |
| "loss": 0.2095, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.582105263157895e-06, | |
| "loss": 0.2106, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.555789473684211e-06, | |
| "loss": 0.2222, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.529473684210526e-06, | |
| "loss": 0.1603, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.503157894736842e-06, | |
| "loss": 0.3861, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.476842105263159e-06, | |
| "loss": 0.2711, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.4505263157894744e-06, | |
| "loss": 0.1412, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.42421052631579e-06, | |
| "loss": 0.2065, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.397894736842106e-06, | |
| "loss": 0.11, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.371578947368422e-06, | |
| "loss": 0.1197, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "eval_cer": 12.432023890722068, | |
| "eval_loss": 0.29520827531814575, | |
| "eval_runtime": 2631.2042, | |
| "eval_samples_per_second": 4.021, | |
| "eval_steps_per_second": 0.503, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.345263157894738e-06, | |
| "loss": 0.0958, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.3189473684210535e-06, | |
| "loss": 0.0993, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.292631578947369e-06, | |
| "loss": 0.1724, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.266315789473684e-06, | |
| "loss": 0.2281, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.24e-06, | |
| "loss": 0.1545, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.213684210526316e-06, | |
| "loss": 0.1804, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.187368421052632e-06, | |
| "loss": 0.2058, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.161052631578948e-06, | |
| "loss": 0.234, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.134736842105264e-06, | |
| "loss": 0.2734, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.10842105263158e-06, | |
| "loss": 0.2279, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.082105263157896e-06, | |
| "loss": 0.2153, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.055789473684212e-06, | |
| "loss": 0.1732, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.0294736842105275e-06, | |
| "loss": 0.1928, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.0031578947368425e-06, | |
| "loss": 0.0833, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 6.976842105263158e-06, | |
| "loss": 0.1013, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.950526315789474e-06, | |
| "loss": 0.1463, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.92421052631579e-06, | |
| "loss": 0.1042, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.897894736842106e-06, | |
| "loss": 0.0706, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.871578947368422e-06, | |
| "loss": 0.1119, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.845263157894737e-06, | |
| "loss": 0.0848, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.818947368421052e-06, | |
| "loss": 0.0679, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.792631578947368e-06, | |
| "loss": 0.1489, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.766315789473685e-06, | |
| "loss": 0.1184, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.740000000000001e-06, | |
| "loss": 0.2467, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.7136842105263165e-06, | |
| "loss": 0.2125, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.687368421052632e-06, | |
| "loss": 0.1153, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.661052631578948e-06, | |
| "loss": 0.0608, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.634736842105264e-06, | |
| "loss": 0.0644, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.60842105263158e-06, | |
| "loss": 0.0739, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.582105263157896e-06, | |
| "loss": 0.0783, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.5557894736842106e-06, | |
| "loss": 0.1506, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.529473684210526e-06, | |
| "loss": 0.2, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.503157894736842e-06, | |
| "loss": 0.1605, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.476842105263158e-06, | |
| "loss": 0.1771, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.450526315789474e-06, | |
| "loss": 0.1563, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.42421052631579e-06, | |
| "loss": 0.1633, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.397894736842106e-06, | |
| "loss": 0.1662, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.371578947368422e-06, | |
| "loss": 0.1108, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 6.345263157894738e-06, | |
| "loss": 0.1496, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 6.318947368421054e-06, | |
| "loss": 0.2034, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_cer": 13.197043154460156, | |
| "eval_loss": 0.29621946811676025, | |
| "eval_runtime": 2528.8287, | |
| "eval_samples_per_second": 4.184, | |
| "eval_steps_per_second": 0.523, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.292631578947369e-06, | |
| "loss": 0.0928, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.266315789473685e-06, | |
| "loss": 0.1862, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.24e-06, | |
| "loss": 0.1285, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.213684210526316e-06, | |
| "loss": 0.1005, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.187368421052632e-06, | |
| "loss": 0.1318, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.161052631578948e-06, | |
| "loss": 0.0988, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.134736842105264e-06, | |
| "loss": 0.1077, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.108421052631579e-06, | |
| "loss": 0.2411, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 6.0821052631578945e-06, | |
| "loss": 0.1673, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 6.055789473684212e-06, | |
| "loss": 0.078, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 6.029473684210527e-06, | |
| "loss": 0.0786, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 6.003157894736843e-06, | |
| "loss": 0.039, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 5.976842105263159e-06, | |
| "loss": 0.0716, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 5.950526315789474e-06, | |
| "loss": 0.0485, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 5.92421052631579e-06, | |
| "loss": 0.0567, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 5.897894736842106e-06, | |
| "loss": 0.0962, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.871578947368422e-06, | |
| "loss": 0.1454, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.845263157894737e-06, | |
| "loss": 0.0905, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.818947368421053e-06, | |
| "loss": 0.0866, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.7926315789473685e-06, | |
| "loss": 0.1452, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.766315789473684e-06, | |
| "loss": 0.1202, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.74e-06, | |
| "loss": 0.1344, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.713684210526316e-06, | |
| "loss": 0.1256, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.687368421052633e-06, | |
| "loss": 0.1132, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.661052631578948e-06, | |
| "loss": 0.0773, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.634736842105264e-06, | |
| "loss": 0.0898, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.60842105263158e-06, | |
| "loss": 0.0349, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.582105263157895e-06, | |
| "loss": 0.0427, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.555789473684211e-06, | |
| "loss": 0.0561, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.529473684210527e-06, | |
| "loss": 0.0591, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.5031578947368425e-06, | |
| "loss": 0.0277, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.476842105263158e-06, | |
| "loss": 0.0691, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.450526315789474e-06, | |
| "loss": 0.0339, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.42421052631579e-06, | |
| "loss": 0.0395, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.397894736842105e-06, | |
| "loss": 0.0626, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.371578947368421e-06, | |
| "loss": 0.106, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.345263157894738e-06, | |
| "loss": 0.1073, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.318947368421053e-06, | |
| "loss": 0.0855, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.292631578947369e-06, | |
| "loss": 0.0722, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.266315789473685e-06, | |
| "loss": 0.0344, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "eval_cer": 11.512157354323742, | |
| "eval_loss": 0.30385497212409973, | |
| "eval_runtime": 2372.663, | |
| "eval_samples_per_second": 4.46, | |
| "eval_steps_per_second": 0.558, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.240000000000001e-06, | |
| "loss": 0.0272, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.2136842105263165e-06, | |
| "loss": 0.0178, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.187368421052632e-06, | |
| "loss": 0.0405, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.161052631578948e-06, | |
| "loss": 0.0838, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.134736842105263e-06, | |
| "loss": 0.0736, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.108421052631579e-06, | |
| "loss": 0.1135, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.082105263157895e-06, | |
| "loss": 0.0937, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.0557894736842105e-06, | |
| "loss": 0.1089, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.029473684210526e-06, | |
| "loss": 0.1071, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.003157894736842e-06, | |
| "loss": 0.0752, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 4.976842105263158e-06, | |
| "loss": 0.0653, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 4.950526315789474e-06, | |
| "loss": 0.0596, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 4.92421052631579e-06, | |
| "loss": 0.0797, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 4.897894736842106e-06, | |
| "loss": 0.0512, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 4.871578947368421e-06, | |
| "loss": 0.0841, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 4.845263157894737e-06, | |
| "loss": 0.0714, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 4.818947368421053e-06, | |
| "loss": 0.0561, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 4.792631578947369e-06, | |
| "loss": 0.0613, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 4.7663157894736845e-06, | |
| "loss": 0.0581, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 4.74e-06, | |
| "loss": 0.0596, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 4.713684210526316e-06, | |
| "loss": 0.1357, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 4.687368421052632e-06, | |
| "loss": 0.0703, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 4.661052631578948e-06, | |
| "loss": 0.0497, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 4.634736842105264e-06, | |
| "loss": 0.0399, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 4.6084210526315794e-06, | |
| "loss": 0.0292, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.582105263157895e-06, | |
| "loss": 0.0246, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.55578947368421e-06, | |
| "loss": 0.015, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.529473684210527e-06, | |
| "loss": 0.0178, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.503157894736843e-06, | |
| "loss": 0.0653, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.4768421052631585e-06, | |
| "loss": 0.0653, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.450526315789474e-06, | |
| "loss": 0.041, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.424210526315789e-06, | |
| "loss": 0.0417, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.397894736842105e-06, | |
| "loss": 0.0754, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.371578947368421e-06, | |
| "loss": 0.0528, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.345263157894738e-06, | |
| "loss": 0.064, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.3189473684210535e-06, | |
| "loss": 0.057, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.2926315789473684e-06, | |
| "loss": 0.0577, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.266315789473684e-06, | |
| "loss": 0.0398, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.24e-06, | |
| "loss": 0.0362, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.213684210526316e-06, | |
| "loss": 0.0226, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "eval_cer": 11.354852188446674, | |
| "eval_loss": 0.30825433135032654, | |
| "eval_runtime": 2371.9841, | |
| "eval_samples_per_second": 4.461, | |
| "eval_steps_per_second": 0.558, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.1873684210526325e-06, | |
| "loss": 0.0224, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.1610526315789475e-06, | |
| "loss": 0.0172, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.134736842105263e-06, | |
| "loss": 0.0292, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.108421052631579e-06, | |
| "loss": 0.0129, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.082105263157895e-06, | |
| "loss": 0.0269, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.055789473684211e-06, | |
| "loss": 0.0172, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.029473684210527e-06, | |
| "loss": 0.0133, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.0031578947368424e-06, | |
| "loss": 0.0265, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.976842105263158e-06, | |
| "loss": 0.0373, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.950526315789474e-06, | |
| "loss": 0.0578, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.92421052631579e-06, | |
| "loss": 0.0388, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.897894736842106e-06, | |
| "loss": 0.0242, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.8715789473684215e-06, | |
| "loss": 0.0154, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.845263157894737e-06, | |
| "loss": 0.0101, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.818947368421053e-06, | |
| "loss": 0.0088, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.792631578947369e-06, | |
| "loss": 0.0192, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.766315789473685e-06, | |
| "loss": 0.0383, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.74e-06, | |
| "loss": 0.0537, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.713684210526316e-06, | |
| "loss": 0.0566, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.687368421052632e-06, | |
| "loss": 0.0443, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.6610526315789472e-06, | |
| "loss": 0.0434, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.6347368421052635e-06, | |
| "loss": 0.0471, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.6084210526315793e-06, | |
| "loss": 0.0223, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.582105263157895e-06, | |
| "loss": 0.023, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.555789473684211e-06, | |
| "loss": 0.0321, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.5294736842105263e-06, | |
| "loss": 0.05, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.503157894736842e-06, | |
| "loss": 0.03, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.476842105263158e-06, | |
| "loss": 0.0499, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.450526315789474e-06, | |
| "loss": 0.034, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.425263157894737e-06, | |
| "loss": 0.022, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.398947368421053e-06, | |
| "loss": 0.028, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.3726315789473683e-06, | |
| "loss": 0.0269, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.3463157894736846e-06, | |
| "loss": 0.0185, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.3200000000000004e-06, | |
| "loss": 0.0549, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.2936842105263162e-06, | |
| "loss": 0.0451, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.267368421052632e-06, | |
| "loss": 0.0228, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.2410526315789474e-06, | |
| "loss": 0.0255, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.2147368421052633e-06, | |
| "loss": 0.015, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.188421052631579e-06, | |
| "loss": 0.0143, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.1621052631578953e-06, | |
| "loss": 0.0097, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "eval_cer": 11.44395081755673, | |
| "eval_loss": 0.3186756670475006, | |
| "eval_runtime": 2351.9601, | |
| "eval_samples_per_second": 4.499, | |
| "eval_steps_per_second": 0.563, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.135789473684211e-06, | |
| "loss": 0.0113, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.1094736842105265e-06, | |
| "loss": 0.0331, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.0831578947368423e-06, | |
| "loss": 0.0353, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.056842105263158e-06, | |
| "loss": 0.0233, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.0305263157894736e-06, | |
| "loss": 0.0127, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.0042105263157894e-06, | |
| "loss": 0.0345, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 2.9778947368421056e-06, | |
| "loss": 0.0318, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 2.9515789473684214e-06, | |
| "loss": 0.0281, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 2.9252631578947373e-06, | |
| "loss": 0.0216, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 2.8989473684210526e-06, | |
| "loss": 0.0246, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 2.8726315789473685e-06, | |
| "loss": 0.0171, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 2.8463157894736843e-06, | |
| "loss": 0.0163, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 2.82e-06, | |
| "loss": 0.0027, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 2.7936842105263163e-06, | |
| "loss": 0.007, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 2.7673684210526317e-06, | |
| "loss": 0.0132, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 2.7410526315789476e-06, | |
| "loss": 0.0109, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 2.7147368421052634e-06, | |
| "loss": 0.003, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 2.688421052631579e-06, | |
| "loss": 0.0146, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.6621052631578946e-06, | |
| "loss": 0.0085, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.635789473684211e-06, | |
| "loss": 0.0037, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.6094736842105267e-06, | |
| "loss": 0.0142, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.5831578947368425e-06, | |
| "loss": 0.0197, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 2.5568421052631583e-06, | |
| "loss": 0.0235, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 2.5305263157894737e-06, | |
| "loss": 0.0173, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 2.5042105263157895e-06, | |
| "loss": 0.0125, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 2.4778947368421053e-06, | |
| "loss": 0.005, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.451578947368421e-06, | |
| "loss": 0.0029, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.425263157894737e-06, | |
| "loss": 0.0063, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.3989473684210528e-06, | |
| "loss": 0.0135, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.3726315789473686e-06, | |
| "loss": 0.022, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 2.3463157894736844e-06, | |
| "loss": 0.0199, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 2.3200000000000002e-06, | |
| "loss": 0.0315, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 2.293684210526316e-06, | |
| "loss": 0.0208, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 2.267368421052632e-06, | |
| "loss": 0.0185, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 2.2410526315789473e-06, | |
| "loss": 0.0142, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 2.2147368421052635e-06, | |
| "loss": 0.013, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 2.188421052631579e-06, | |
| "loss": 0.0111, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 2.1621052631578947e-06, | |
| "loss": 0.019, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 2.135789473684211e-06, | |
| "loss": 0.0159, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 2.1094736842105264e-06, | |
| "loss": 0.0121, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "eval_cer": 11.22581279456314, | |
| "eval_loss": 0.3172953724861145, | |
| "eval_runtime": 2351.6739, | |
| "eval_samples_per_second": 4.499, | |
| "eval_steps_per_second": 0.563, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 2.083157894736842e-06, | |
| "loss": 0.0215, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 2.056842105263158e-06, | |
| "loss": 0.0139, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 2.030526315789474e-06, | |
| "loss": 0.0159, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.0042105263157896e-06, | |
| "loss": 0.0122, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 1.9778947368421055e-06, | |
| "loss": 0.0144, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 1.9515789473684213e-06, | |
| "loss": 0.0136, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 1.925263157894737e-06, | |
| "loss": 0.0396, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 1.8989473684210527e-06, | |
| "loss": 0.0207, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 1.8726315789473687e-06, | |
| "loss": 0.0124, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 1.8463157894736843e-06, | |
| "loss": 0.0095, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 1.8200000000000002e-06, | |
| "loss": 0.0041, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 1.7936842105263158e-06, | |
| "loss": 0.006, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 1.7673684210526318e-06, | |
| "loss": 0.0054, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 1.7410526315789474e-06, | |
| "loss": 0.0068, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 1.7147368421052632e-06, | |
| "loss": 0.0184, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 1.6884210526315792e-06, | |
| "loss": 0.0157, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 1.6621052631578948e-06, | |
| "loss": 0.0049, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 1.6357894736842107e-06, | |
| "loss": 0.0068, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 1.6094736842105265e-06, | |
| "loss": 0.0127, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 1.5831578947368423e-06, | |
| "loss": 0.0127, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 1.556842105263158e-06, | |
| "loss": 0.0094, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 1.5305263157894737e-06, | |
| "loss": 0.0103, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 1.5042105263157898e-06, | |
| "loss": 0.0072, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 1.4778947368421054e-06, | |
| "loss": 0.0091, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 1.4515789473684212e-06, | |
| "loss": 0.0033, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 1.425263157894737e-06, | |
| "loss": 0.0026, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 1.3989473684210528e-06, | |
| "loss": 0.0037, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 1.3726315789473684e-06, | |
| "loss": 0.0031, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 1.3463157894736842e-06, | |
| "loss": 0.0034, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 1.32e-06, | |
| "loss": 0.0028, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 1.2936842105263159e-06, | |
| "loss": 0.0045, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 1.2673684210526315e-06, | |
| "loss": 0.003, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 1.2410526315789475e-06, | |
| "loss": 0.002, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 1.2147368421052633e-06, | |
| "loss": 0.0036, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 1.188421052631579e-06, | |
| "loss": 0.01, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.1621052631578948e-06, | |
| "loss": 0.0123, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.1357894736842106e-06, | |
| "loss": 0.0058, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.1094736842105264e-06, | |
| "loss": 0.0056, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.0831578947368422e-06, | |
| "loss": 0.0026, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.0568421052631578e-06, | |
| "loss": 0.0015, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "eval_cer": 11.141015478582533, | |
| "eval_loss": 0.3219420909881592, | |
| "eval_runtime": 2360.5215, | |
| "eval_samples_per_second": 4.482, | |
| "eval_steps_per_second": 0.56, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.0305263157894739e-06, | |
| "loss": 0.0022, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.0042105263157897e-06, | |
| "loss": 0.0055, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 9.778947368421053e-07, | |
| "loss": 0.0083, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 9.515789473684212e-07, | |
| "loss": 0.0128, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 9.252631578947368e-07, | |
| "loss": 0.0122, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 8.989473684210527e-07, | |
| "loss": 0.0082, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 8.726315789473686e-07, | |
| "loss": 0.0101, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 8.463157894736843e-07, | |
| "loss": 0.0034, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 8.200000000000001e-07, | |
| "loss": 0.0061, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 7.936842105263158e-07, | |
| "loss": 0.0044, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 7.673684210526316e-07, | |
| "loss": 0.0051, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 7.410526315789475e-07, | |
| "loss": 0.0078, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 7.147368421052632e-07, | |
| "loss": 0.0041, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 6.884210526315791e-07, | |
| "loss": 0.0108, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 6.621052631578948e-07, | |
| "loss": 0.0048, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 6.357894736842106e-07, | |
| "loss": 0.0053, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 6.094736842105263e-07, | |
| "loss": 0.0071, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 5.831578947368421e-07, | |
| "loss": 0.0039, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 5.56842105263158e-07, | |
| "loss": 0.0087, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 5.305263157894737e-07, | |
| "loss": 0.0161, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 5.042105263157895e-07, | |
| "loss": 0.0092, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 4.778947368421053e-07, | |
| "loss": 0.0064, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 4.5157894736842107e-07, | |
| "loss": 0.0042, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 4.2526315789473684e-07, | |
| "loss": 0.0008, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.9894736842105266e-07, | |
| "loss": 0.0028, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.726315789473685e-07, | |
| "loss": 0.002, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.4631578947368424e-07, | |
| "loss": 0.0023, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.2e-07, | |
| "loss": 0.0072, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 2.936842105263158e-07, | |
| "loss": 0.0038, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 2.6736842105263164e-07, | |
| "loss": 0.0023, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 2.410526315789474e-07, | |
| "loss": 0.0036, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 2.1473684210526317e-07, | |
| "loss": 0.0036, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 1.8842105263157897e-07, | |
| "loss": 0.0048, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 1.6210526315789476e-07, | |
| "loss": 0.0045, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 1.3578947368421055e-07, | |
| "loss": 0.0047, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 1.0947368421052632e-07, | |
| "loss": 0.0039, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 8.315789473684211e-08, | |
| "loss": 0.0028, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 5.68421052631579e-08, | |
| "loss": 0.0016, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 3.0526315789473686e-08, | |
| "loss": 0.0015, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 4.210526315789474e-09, | |
| "loss": 0.0019, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "eval_cer": 10.978180052967598, | |
| "eval_loss": 0.32260650396347046, | |
| "eval_runtime": 2355.0099, | |
| "eval_samples_per_second": 4.493, | |
| "eval_steps_per_second": 0.562, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "step": 10000, | |
| "total_flos": 1.6325391979708416e+20, | |
| "train_loss": 0.14908174896221607, | |
| "train_runtime": 62798.3056, | |
| "train_samples_per_second": 2.548, | |
| "train_steps_per_second": 0.159 | |
| } | |
| ], | |
| "max_steps": 10000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 1.6325391979708416e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |