| { | |
| "best_metric": 44.19148478658209, | |
| "best_model_checkpoint": "./checkpoint-1000", | |
| "epoch": 31.0204, | |
| "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": 4.2000000000000006e-07, | |
| "loss": 4.4125, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 3.8695, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.42e-06, | |
| "loss": 2.9975, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 2.2517, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.42e-06, | |
| "loss": 1.4513, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.92e-06, | |
| "loss": 1.0241, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.9404, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.8316, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.9362, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.8711, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.8359, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.729, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.6871, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.6619, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.7008, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.7207, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.7039, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.5865, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.5384, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.5595, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.953333333333333e-06, | |
| "loss": 0.5347, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.89777777777778e-06, | |
| "loss": 0.6092, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.842222222222223e-06, | |
| "loss": 0.5707, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 9.786666666666667e-06, | |
| "loss": 0.4879, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 9.731111111111113e-06, | |
| "loss": 0.4475, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 9.675555555555555e-06, | |
| "loss": 0.4395, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 9.620000000000001e-06, | |
| "loss": 0.4044, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 9.564444444444445e-06, | |
| "loss": 0.4884, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 9.508888888888889e-06, | |
| "loss": 0.4477, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 9.453333333333335e-06, | |
| "loss": 0.4298, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 9.397777777777779e-06, | |
| "loss": 0.3643, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 9.342222222222223e-06, | |
| "loss": 0.3446, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 9.286666666666667e-06, | |
| "loss": 0.3292, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 9.231111111111111e-06, | |
| "loss": 0.3715, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 9.175555555555557e-06, | |
| "loss": 0.3683, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 9.12e-06, | |
| "loss": 0.3747, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 9.064444444444447e-06, | |
| "loss": 0.292, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 9.008888888888889e-06, | |
| "loss": 0.272, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 8.953333333333335e-06, | |
| "loss": 0.276, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 8.897777777777779e-06, | |
| "loss": 0.2768, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "eval_loss": 0.6929294466972351, | |
| "eval_runtime": 737.447, | |
| "eval_samples_per_second": 6.874, | |
| "eval_steps_per_second": 0.216, | |
| "eval_wer": 44.19148478658209, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 8.842222222222223e-06, | |
| "loss": 0.3181, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 8.786666666666668e-06, | |
| "loss": 0.2944, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 8.73111111111111e-06, | |
| "loss": 0.2488, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 8.675555555555556e-06, | |
| "loss": 0.2283, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 8.62e-06, | |
| "loss": 0.2268, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 8.564444444444445e-06, | |
| "loss": 0.202, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 8.50888888888889e-06, | |
| "loss": 0.2612, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 8.453333333333334e-06, | |
| "loss": 0.2341, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 8.397777777777778e-06, | |
| "loss": 0.2217, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 8.342222222222222e-06, | |
| "loss": 0.1883, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 8.286666666666668e-06, | |
| "loss": 0.1827, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 8.231111111111112e-06, | |
| "loss": 0.1628, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 8.175555555555556e-06, | |
| "loss": 0.1977, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 8.120000000000002e-06, | |
| "loss": 0.187, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 8.064444444444444e-06, | |
| "loss": 0.192, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 8.00888888888889e-06, | |
| "loss": 0.1551, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 7.953333333333334e-06, | |
| "loss": 0.1368, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 7.897777777777778e-06, | |
| "loss": 0.1383, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 7.842222222222224e-06, | |
| "loss": 0.1432, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 7.786666666666666e-06, | |
| "loss": 0.1628, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 7.731111111111112e-06, | |
| "loss": 0.1561, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 7.675555555555556e-06, | |
| "loss": 0.1251, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 7.620000000000001e-06, | |
| "loss": 0.112, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 7.564444444444446e-06, | |
| "loss": 0.1132, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 7.50888888888889e-06, | |
| "loss": 0.1012, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 7.453333333333334e-06, | |
| "loss": 0.1329, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 7.3977777777777786e-06, | |
| "loss": 0.1212, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 7.342222222222223e-06, | |
| "loss": 0.1081, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 7.2866666666666675e-06, | |
| "loss": 0.0952, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 7.231111111111112e-06, | |
| "loss": 0.0878, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 7.1755555555555556e-06, | |
| "loss": 0.0797, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 7.1200000000000004e-06, | |
| "loss": 0.1004, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 7.0644444444444445e-06, | |
| "loss": 0.0966, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 7.008888888888889e-06, | |
| "loss": 0.0917, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 6.953333333333334e-06, | |
| "loss": 0.0746, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 6.897777777777779e-06, | |
| "loss": 0.0677, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 6.842222222222222e-06, | |
| "loss": 0.0651, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 6.786666666666667e-06, | |
| "loss": 0.0714, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 6.731111111111111e-06, | |
| "loss": 0.0797, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 6.675555555555556e-06, | |
| "loss": 0.0748, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "eval_loss": 0.7672464847564697, | |
| "eval_runtime": 887.6918, | |
| "eval_samples_per_second": 5.71, | |
| "eval_steps_per_second": 0.179, | |
| "eval_wer": 44.992473927534675, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 6.620000000000001e-06, | |
| "loss": 0.059, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 6.564444444444446e-06, | |
| "loss": 0.0549, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 6.508888888888889e-06, | |
| "loss": 0.0543, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 6.453333333333334e-06, | |
| "loss": 0.0477, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 6.397777777777778e-06, | |
| "loss": 0.0658, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 6.342222222222223e-06, | |
| "loss": 0.0554, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 6.286666666666668e-06, | |
| "loss": 0.05, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 6.231111111111111e-06, | |
| "loss": 0.0452, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 6.175555555555556e-06, | |
| "loss": 0.0422, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 6.120000000000001e-06, | |
| "loss": 0.0379, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 6.064444444444445e-06, | |
| "loss": 0.0468, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 6.00888888888889e-06, | |
| "loss": 0.0455, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 5.9533333333333345e-06, | |
| "loss": 0.0431, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 5.897777777777778e-06, | |
| "loss": 0.0349, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 5.8422222222222226e-06, | |
| "loss": 0.0325, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 5.7866666666666674e-06, | |
| "loss": 0.0307, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 5.7311111111111115e-06, | |
| "loss": 0.0333, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 5.675555555555556e-06, | |
| "loss": 0.0379, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 5.620000000000001e-06, | |
| "loss": 0.034, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 5.5644444444444444e-06, | |
| "loss": 0.027, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 5.508888888888889e-06, | |
| "loss": 0.026, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 5.453333333333334e-06, | |
| "loss": 0.0258, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 5.397777777777778e-06, | |
| "loss": 0.0232, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 5.342222222222223e-06, | |
| "loss": 0.0312, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 5.286666666666666e-06, | |
| "loss": 0.0262, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 5.231111111111111e-06, | |
| "loss": 0.0228, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 5.175555555555556e-06, | |
| "loss": 0.0215, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 5.12e-06, | |
| "loss": 0.0203, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 5.064444444444445e-06, | |
| "loss": 0.018, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 5.00888888888889e-06, | |
| "loss": 0.023, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 4.953333333333334e-06, | |
| "loss": 0.0208, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 4.897777777777778e-06, | |
| "loss": 0.02, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 4.842222222222223e-06, | |
| "loss": 0.0175, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 4.786666666666667e-06, | |
| "loss": 0.0166, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 4.731111111111112e-06, | |
| "loss": 0.0147, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 4.675555555555556e-06, | |
| "loss": 0.017, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 4.620000000000001e-06, | |
| "loss": 0.0186, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 4.564444444444445e-06, | |
| "loss": 0.0167, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 4.50888888888889e-06, | |
| "loss": 0.0142, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 4.453333333333334e-06, | |
| "loss": 0.0143, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "eval_loss": 0.8664703369140625, | |
| "eval_runtime": 681.4862, | |
| "eval_samples_per_second": 7.438, | |
| "eval_steps_per_second": 0.233, | |
| "eval_wer": 45.00860122567466, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 4.397777777777778e-06, | |
| "loss": 0.0136, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 4.3422222222222225e-06, | |
| "loss": 0.0128, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 4.2866666666666666e-06, | |
| "loss": 0.0165, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 4.2311111111111114e-06, | |
| "loss": 0.0139, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 4.175555555555556e-06, | |
| "loss": 0.0124, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 4.12e-06, | |
| "loss": 0.0131, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 4.064444444444444e-06, | |
| "loss": 0.0117, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 4.008888888888889e-06, | |
| "loss": 0.011, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 3.953333333333333e-06, | |
| "loss": 0.0134, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 3.897777777777778e-06, | |
| "loss": 0.0125, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 3.842222222222223e-06, | |
| "loss": 0.0118, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 20.03, | |
| "learning_rate": 3.7866666666666667e-06, | |
| "loss": 0.0114, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "learning_rate": 3.7311111111111116e-06, | |
| "loss": 0.0108, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 3.675555555555556e-06, | |
| "loss": 0.0094, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 3.62e-06, | |
| "loss": 0.0114, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 21.02, | |
| "learning_rate": 3.564444444444445e-06, | |
| "loss": 0.0115, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 21.02, | |
| "learning_rate": 3.508888888888889e-06, | |
| "loss": 0.011, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 21.03, | |
| "learning_rate": 3.4533333333333334e-06, | |
| "loss": 0.0096, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 21.03, | |
| "learning_rate": 3.3977777777777783e-06, | |
| "loss": 0.01, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "learning_rate": 3.3422222222222224e-06, | |
| "loss": 0.0092, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 3.286666666666667e-06, | |
| "loss": 0.0091, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 3.2311111111111117e-06, | |
| "loss": 0.0113, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 22.02, | |
| "learning_rate": 3.1755555555555557e-06, | |
| "loss": 0.0096, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 22.02, | |
| "learning_rate": 3.12e-06, | |
| "loss": 0.0089, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 22.03, | |
| "learning_rate": 3.064444444444445e-06, | |
| "loss": 0.0095, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "learning_rate": 3.008888888888889e-06, | |
| "loss": 0.0086, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 23.01, | |
| "learning_rate": 2.9533333333333336e-06, | |
| "loss": 0.0079, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 23.01, | |
| "learning_rate": 2.8977777777777785e-06, | |
| "loss": 0.0098, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 23.02, | |
| "learning_rate": 2.8422222222222225e-06, | |
| "loss": 0.0091, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 23.02, | |
| "learning_rate": 2.786666666666667e-06, | |
| "loss": 0.0087, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 23.03, | |
| "learning_rate": 2.7311111111111114e-06, | |
| "loss": 0.0087, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "learning_rate": 2.675555555555556e-06, | |
| "loss": 0.0081, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "learning_rate": 2.6200000000000003e-06, | |
| "loss": 0.0073, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "learning_rate": 2.5644444444444444e-06, | |
| "loss": 0.0087, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 24.02, | |
| "learning_rate": 2.5088888888888892e-06, | |
| "loss": 0.0087, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 24.02, | |
| "learning_rate": 2.4533333333333333e-06, | |
| "loss": 0.0082, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 24.03, | |
| "learning_rate": 2.397777777777778e-06, | |
| "loss": 0.0077, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 24.03, | |
| "learning_rate": 2.342222222222222e-06, | |
| "loss": 0.0078, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 2.2866666666666667e-06, | |
| "loss": 0.0071, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 25.01, | |
| "learning_rate": 2.2311111111111115e-06, | |
| "loss": 0.0074, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 25.01, | |
| "eval_loss": 0.9178622961044312, | |
| "eval_runtime": 647.6309, | |
| "eval_samples_per_second": 7.827, | |
| "eval_steps_per_second": 0.246, | |
| "eval_wer": 45.81765401569724, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 25.02, | |
| "learning_rate": 2.1755555555555556e-06, | |
| "loss": 0.0088, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 25.02, | |
| "learning_rate": 2.12e-06, | |
| "loss": 0.0076, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 25.02, | |
| "learning_rate": 2.064444444444445e-06, | |
| "loss": 0.007, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 25.03, | |
| "learning_rate": 2.008888888888889e-06, | |
| "loss": 0.0077, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "learning_rate": 1.9533333333333334e-06, | |
| "loss": 0.0069, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 26.01, | |
| "learning_rate": 1.8977777777777779e-06, | |
| "loss": 0.0065, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 26.01, | |
| "learning_rate": 1.8422222222222225e-06, | |
| "loss": 0.0081, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 1.7866666666666668e-06, | |
| "loss": 0.0073, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 1.7311111111111112e-06, | |
| "loss": 0.007, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 26.03, | |
| "learning_rate": 1.675555555555556e-06, | |
| "loss": 0.0072, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "learning_rate": 1.6200000000000002e-06, | |
| "loss": 0.0067, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 27.01, | |
| "learning_rate": 1.5644444444444446e-06, | |
| "loss": 0.0062, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 27.01, | |
| "learning_rate": 1.5088888888888889e-06, | |
| "loss": 0.0074, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 27.02, | |
| "learning_rate": 1.4533333333333335e-06, | |
| "loss": 0.0071, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 27.02, | |
| "learning_rate": 1.397777777777778e-06, | |
| "loss": 0.007, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 27.03, | |
| "learning_rate": 1.3422222222222222e-06, | |
| "loss": 0.0066, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "learning_rate": 1.286666666666667e-06, | |
| "loss": 0.0066, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 28.01, | |
| "learning_rate": 1.2311111111111112e-06, | |
| "loss": 0.0061, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 28.01, | |
| "learning_rate": 1.1755555555555556e-06, | |
| "loss": 0.0063, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 28.02, | |
| "learning_rate": 1.12e-06, | |
| "loss": 0.0076, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 28.02, | |
| "learning_rate": 1.0644444444444445e-06, | |
| "loss": 0.0066, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 28.03, | |
| "learning_rate": 1.008888888888889e-06, | |
| "loss": 0.0061, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 28.03, | |
| "learning_rate": 9.533333333333335e-07, | |
| "loss": 0.0068, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "learning_rate": 8.977777777777778e-07, | |
| "loss": 0.006, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 29.01, | |
| "learning_rate": 8.422222222222224e-07, | |
| "loss": 0.0058, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 29.01, | |
| "learning_rate": 7.866666666666667e-07, | |
| "loss": 0.0073, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 29.02, | |
| "learning_rate": 7.311111111111112e-07, | |
| "loss": 0.0064, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 29.02, | |
| "learning_rate": 6.755555555555555e-07, | |
| "loss": 0.0062, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 29.03, | |
| "learning_rate": 6.200000000000001e-07, | |
| "loss": 0.0066, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "learning_rate": 5.644444444444445e-07, | |
| "loss": 0.0061, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 30.01, | |
| "learning_rate": 5.088888888888889e-07, | |
| "loss": 0.0055, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 30.01, | |
| "learning_rate": 4.533333333333334e-07, | |
| "loss": 0.0069, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 30.02, | |
| "learning_rate": 3.9777777777777783e-07, | |
| "loss": 0.0065, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 30.02, | |
| "learning_rate": 3.422222222222223e-07, | |
| "loss": 0.0064, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 30.03, | |
| "learning_rate": 2.866666666666667e-07, | |
| "loss": 0.0061, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "learning_rate": 2.3111111111111112e-07, | |
| "loss": 0.0061, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 31.01, | |
| "learning_rate": 1.7555555555555558e-07, | |
| "loss": 0.0056, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 31.01, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 0.0061, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 31.02, | |
| "learning_rate": 6.444444444444445e-08, | |
| "loss": 0.0071, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 31.02, | |
| "learning_rate": 8.88888888888889e-09, | |
| "loss": 0.0062, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 31.02, | |
| "eval_loss": 0.9326218366622925, | |
| "eval_runtime": 629.2584, | |
| "eval_samples_per_second": 8.056, | |
| "eval_steps_per_second": 0.253, | |
| "eval_wer": 45.78271153639394, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 31.02, | |
| "step": 5000, | |
| "total_flos": 7.87193775489024e+18, | |
| "train_loss": 0.2122017632007599, | |
| "train_runtime": 29527.9232, | |
| "train_samples_per_second": 10.837, | |
| "train_steps_per_second": 0.169 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 7.87193775489024e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |