| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.707762557077626, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.2000000000000006e-07, | |
| "loss": 0.8716, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 0.7032, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.42e-06, | |
| "loss": 0.4696, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 0.372, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.42e-06, | |
| "loss": 0.3581, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.92e-06, | |
| "loss": 0.3364, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.3204, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.317, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.3132, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.3021, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.3206, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.319, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.3265, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.3211, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.3143, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.3265, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.2936, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.312, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.3164, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.317, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.953333333333333e-06, | |
| "loss": 0.315, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.89777777777778e-06, | |
| "loss": 0.3249, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 9.842222222222223e-06, | |
| "loss": 0.3308, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.786666666666667e-06, | |
| "loss": 0.3492, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 9.731111111111113e-06, | |
| "loss": 0.3101, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.675555555555555e-06, | |
| "loss": 0.3229, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 9.620000000000001e-06, | |
| "loss": 0.294, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.564444444444445e-06, | |
| "loss": 0.2862, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.508888888888889e-06, | |
| "loss": 0.3107, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.453333333333335e-06, | |
| "loss": 0.2819, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 9.397777777777779e-06, | |
| "loss": 0.3166, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.342222222222223e-06, | |
| "loss": 0.298, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.286666666666667e-06, | |
| "loss": 0.3088, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.231111111111111e-06, | |
| "loss": 0.2821, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.175555555555557e-06, | |
| "loss": 0.2961, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.12e-06, | |
| "loss": 0.169, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.064444444444447e-06, | |
| "loss": 0.1584, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.008888888888889e-06, | |
| "loss": 0.1422, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.953333333333335e-06, | |
| "loss": 0.1615, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.897777777777779e-06, | |
| "loss": 0.1544, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "eval_loss": 0.22604741156101227, | |
| "eval_runtime": 1239.0181, | |
| "eval_samples_per_second": 4.512, | |
| "eval_steps_per_second": 0.282, | |
| "eval_wer": 0.5485307212822796, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 8.842222222222223e-06, | |
| "loss": 0.1593, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 8.786666666666668e-06, | |
| "loss": 0.1527, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 8.73111111111111e-06, | |
| "loss": 0.1534, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 8.675555555555556e-06, | |
| "loss": 0.1723, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 8.62e-06, | |
| "loss": 0.1503, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 8.564444444444445e-06, | |
| "loss": 0.1418, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 8.50888888888889e-06, | |
| "loss": 0.1455, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 8.453333333333334e-06, | |
| "loss": 0.1622, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 8.397777777777778e-06, | |
| "loss": 0.1567, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 8.342222222222222e-06, | |
| "loss": 0.1787, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 8.286666666666668e-06, | |
| "loss": 0.1745, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 8.231111111111112e-06, | |
| "loss": 0.1718, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 8.175555555555556e-06, | |
| "loss": 0.1287, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 8.120000000000002e-06, | |
| "loss": 0.1293, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 8.064444444444444e-06, | |
| "loss": 0.1471, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 8.00888888888889e-06, | |
| "loss": 0.1538, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 7.953333333333334e-06, | |
| "loss": 0.1441, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 7.897777777777778e-06, | |
| "loss": 0.1452, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 7.842222222222224e-06, | |
| "loss": 0.1548, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 7.786666666666666e-06, | |
| "loss": 0.1548, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 7.731111111111112e-06, | |
| "loss": 0.137, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.675555555555556e-06, | |
| "loss": 0.1451, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 7.620000000000001e-06, | |
| "loss": 0.1367, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 7.564444444444446e-06, | |
| "loss": 0.1576, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 7.50888888888889e-06, | |
| "loss": 0.1764, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 7.453333333333334e-06, | |
| "loss": 0.1625, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 7.3977777777777786e-06, | |
| "loss": 0.1481, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 7.342222222222223e-06, | |
| "loss": 0.1523, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 7.2866666666666675e-06, | |
| "loss": 0.1404, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 7.231111111111112e-06, | |
| "loss": 0.1352, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.1755555555555556e-06, | |
| "loss": 0.0618, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.1200000000000004e-06, | |
| "loss": 0.0536, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 7.0644444444444445e-06, | |
| "loss": 0.0685, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 7.008888888888889e-06, | |
| "loss": 0.0755, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 6.953333333333334e-06, | |
| "loss": 0.053, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 6.897777777777779e-06, | |
| "loss": 0.0603, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 6.842222222222222e-06, | |
| "loss": 0.0709, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 6.786666666666667e-06, | |
| "loss": 0.0678, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 6.731111111111111e-06, | |
| "loss": 0.0696, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 6.675555555555556e-06, | |
| "loss": 0.0745, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "eval_loss": 0.2132357656955719, | |
| "eval_runtime": 1250.3673, | |
| "eval_samples_per_second": 4.471, | |
| "eval_steps_per_second": 0.28, | |
| "eval_wer": 0.4967052537845058, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 6.620000000000001e-06, | |
| "loss": 0.0547, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 6.564444444444446e-06, | |
| "loss": 0.06, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 6.508888888888889e-06, | |
| "loss": 0.0576, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 6.453333333333334e-06, | |
| "loss": 0.0797, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 6.397777777777778e-06, | |
| "loss": 0.0564, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 6.342222222222223e-06, | |
| "loss": 0.067, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 6.286666666666668e-06, | |
| "loss": 0.058, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 6.231111111111111e-06, | |
| "loss": 0.0614, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 6.175555555555556e-06, | |
| "loss": 0.0585, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 6.120000000000001e-06, | |
| "loss": 0.0621, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.064444444444445e-06, | |
| "loss": 0.0583, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.00888888888889e-06, | |
| "loss": 0.0609, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.9533333333333345e-06, | |
| "loss": 0.0641, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.897777777777778e-06, | |
| "loss": 0.0563, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 5.8422222222222226e-06, | |
| "loss": 0.0572, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 5.7866666666666674e-06, | |
| "loss": 0.0587, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 5.7311111111111115e-06, | |
| "loss": 0.0596, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 5.675555555555556e-06, | |
| "loss": 0.0605, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 5.620000000000001e-06, | |
| "loss": 0.0646, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 5.5644444444444444e-06, | |
| "loss": 0.0544, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 5.508888888888889e-06, | |
| "loss": 0.0593, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 5.453333333333334e-06, | |
| "loss": 0.0577, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 5.397777777777778e-06, | |
| "loss": 0.0706, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 5.342222222222223e-06, | |
| "loss": 0.056, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 5.286666666666666e-06, | |
| "loss": 0.0621, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 5.231111111111111e-06, | |
| "loss": 0.0297, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.175555555555556e-06, | |
| "loss": 0.0244, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.12e-06, | |
| "loss": 0.0276, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.064444444444445e-06, | |
| "loss": 0.0207, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 5.00888888888889e-06, | |
| "loss": 0.0264, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 4.953333333333334e-06, | |
| "loss": 0.0263, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 4.897777777777778e-06, | |
| "loss": 0.0263, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 4.842222222222223e-06, | |
| "loss": 0.0252, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 4.786666666666667e-06, | |
| "loss": 0.0254, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 4.731111111111112e-06, | |
| "loss": 0.0224, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 4.675555555555556e-06, | |
| "loss": 0.0257, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 4.620000000000001e-06, | |
| "loss": 0.0229, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 4.564444444444445e-06, | |
| "loss": 0.0214, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 4.50888888888889e-06, | |
| "loss": 0.022, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 4.453333333333334e-06, | |
| "loss": 0.0213, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "eval_loss": 0.21137936413288116, | |
| "eval_runtime": 1246.2776, | |
| "eval_samples_per_second": 4.486, | |
| "eval_steps_per_second": 0.281, | |
| "eval_wer": 0.4717720391807658, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 4.397777777777778e-06, | |
| "loss": 0.0261, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 4.3422222222222225e-06, | |
| "loss": 0.0287, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 4.2866666666666666e-06, | |
| "loss": 0.022, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 4.2311111111111114e-06, | |
| "loss": 0.0273, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 4.175555555555556e-06, | |
| "loss": 0.0224, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 4.12e-06, | |
| "loss": 0.0243, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 4.064444444444444e-06, | |
| "loss": 0.0168, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 4.008888888888889e-06, | |
| "loss": 0.0212, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.953333333333333e-06, | |
| "loss": 0.0205, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 3.897777777777778e-06, | |
| "loss": 0.0231, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.842222222222223e-06, | |
| "loss": 0.02, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.7866666666666667e-06, | |
| "loss": 0.0244, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.7311111111111116e-06, | |
| "loss": 0.0195, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 3.675555555555556e-06, | |
| "loss": 0.0211, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 3.62e-06, | |
| "loss": 0.0225, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 3.564444444444445e-06, | |
| "loss": 0.0162, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.508888888888889e-06, | |
| "loss": 0.0253, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 3.4533333333333334e-06, | |
| "loss": 0.0156, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.3977777777777783e-06, | |
| "loss": 0.0219, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.3422222222222224e-06, | |
| "loss": 0.0175, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.286666666666667e-06, | |
| "loss": 0.0133, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 3.2311111111111117e-06, | |
| "loss": 0.0065, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.1755555555555557e-06, | |
| "loss": 0.0087, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.12e-06, | |
| "loss": 0.0068, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.064444444444445e-06, | |
| "loss": 0.009, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.008888888888889e-06, | |
| "loss": 0.0071, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 2.9533333333333336e-06, | |
| "loss": 0.0063, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 2.8977777777777785e-06, | |
| "loss": 0.0082, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 2.8422222222222225e-06, | |
| "loss": 0.0054, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 2.786666666666667e-06, | |
| "loss": 0.0098, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 2.7311111111111114e-06, | |
| "loss": 0.0077, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 2.675555555555556e-06, | |
| "loss": 0.0043, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 2.6200000000000003e-06, | |
| "loss": 0.0064, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 2.5644444444444444e-06, | |
| "loss": 0.0062, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 2.5088888888888892e-06, | |
| "loss": 0.0038, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 2.4533333333333333e-06, | |
| "loss": 0.0121, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 2.397777777777778e-06, | |
| "loss": 0.006, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 2.342222222222222e-06, | |
| "loss": 0.0061, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 2.2866666666666667e-06, | |
| "loss": 0.0081, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 2.2311111111111115e-06, | |
| "loss": 0.0117, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "eval_loss": 0.2196018099784851, | |
| "eval_runtime": 1243.8383, | |
| "eval_samples_per_second": 4.495, | |
| "eval_steps_per_second": 0.281, | |
| "eval_wer": 0.4642920747996438, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 2.1755555555555556e-06, | |
| "loss": 0.0071, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 2.12e-06, | |
| "loss": 0.006, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 2.064444444444445e-06, | |
| "loss": 0.0051, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 2.008888888888889e-06, | |
| "loss": 0.0035, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 1.9533333333333334e-06, | |
| "loss": 0.0052, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 1.8977777777777779e-06, | |
| "loss": 0.0058, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 1.8422222222222225e-06, | |
| "loss": 0.0074, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 1.7866666666666668e-06, | |
| "loss": 0.0075, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 1.7311111111111112e-06, | |
| "loss": 0.0053, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 1.675555555555556e-06, | |
| "loss": 0.0064, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 1.6200000000000002e-06, | |
| "loss": 0.0067, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 1.5644444444444446e-06, | |
| "loss": 0.0044, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 1.5088888888888889e-06, | |
| "loss": 0.0049, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 1.4533333333333335e-06, | |
| "loss": 0.0028, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 1.397777777777778e-06, | |
| "loss": 0.0052, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 1.3422222222222222e-06, | |
| "loss": 0.0027, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 1.286666666666667e-06, | |
| "loss": 0.0021, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 1.2311111111111112e-06, | |
| "loss": 0.002, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 1.1755555555555556e-06, | |
| "loss": 0.0037, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 1.12e-06, | |
| "loss": 0.0014, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 1.0644444444444445e-06, | |
| "loss": 0.002, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 1.008888888888889e-06, | |
| "loss": 0.0012, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 9.533333333333335e-07, | |
| "loss": 0.002, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 8.977777777777778e-07, | |
| "loss": 0.0039, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 8.422222222222224e-07, | |
| "loss": 0.0025, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 7.866666666666667e-07, | |
| "loss": 0.0014, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 7.311111111111112e-07, | |
| "loss": 0.002, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 6.755555555555555e-07, | |
| "loss": 0.0013, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 6.200000000000001e-07, | |
| "loss": 0.0025, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 5.644444444444445e-07, | |
| "loss": 0.0023, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 5.088888888888889e-07, | |
| "loss": 0.0024, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 4.533333333333334e-07, | |
| "loss": 0.0012, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 3.9777777777777783e-07, | |
| "loss": 0.0016, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 3.422222222222223e-07, | |
| "loss": 0.0022, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 2.866666666666667e-07, | |
| "loss": 0.0012, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 2.3111111111111112e-07, | |
| "loss": 0.0011, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 1.7555555555555558e-07, | |
| "loss": 0.0013, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 0.0021, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 6.444444444444445e-08, | |
| "loss": 0.0055, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 8.88888888888889e-09, | |
| "loss": 0.0014, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "eval_loss": 0.22371040284633636, | |
| "eval_runtime": 1249.8343, | |
| "eval_samples_per_second": 4.473, | |
| "eval_steps_per_second": 0.28, | |
| "eval_wer": 0.45734639358860196, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "step": 5000, | |
| "total_flos": 8.163818740396327e+19, | |
| "train_loss": 0.10388498471789062, | |
| "train_runtime": 22145.5379, | |
| "train_samples_per_second": 3.612, | |
| "train_steps_per_second": 0.226 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 6, | |
| "total_flos": 8.163818740396327e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |