| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 6.0532, | |
| "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": 2.3482, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 2.3433, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.42e-06, | |
| "loss": 1.761, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 1.5988, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 1.4245, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9e-06, | |
| "loss": 1.5665, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 1.4588, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 1.2609, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.4e-06, | |
| "loss": 1.2301, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 1.4295, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 1.415, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.9e-06, | |
| "loss": 1.1804, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 1.3484, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.9e-06, | |
| "loss": 1.3593, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.4e-06, | |
| "loss": 1.179, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.9e-06, | |
| "loss": 1.3824, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 1.1744, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 1.2079, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.4e-06, | |
| "loss": 1.2476, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.9e-06, | |
| "loss": 1.3404, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.955555555555556e-06, | |
| "loss": 1.0947, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.9e-06, | |
| "loss": 1.3787, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.844444444444446e-06, | |
| "loss": 1.0917, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.78888888888889e-06, | |
| "loss": 1.1884, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.733333333333334e-06, | |
| "loss": 1.2814, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.677777777777778e-06, | |
| "loss": 1.1693, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.622222222222222e-06, | |
| "loss": 1.3852, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.566666666666668e-06, | |
| "loss": 1.0322, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.511111111111112e-06, | |
| "loss": 1.1455, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.455555555555557e-06, | |
| "loss": 0.9578, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.4e-06, | |
| "loss": 1.2149, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.344444444444446e-06, | |
| "loss": 0.9243, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.28888888888889e-06, | |
| "loss": 0.7373, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.233333333333334e-06, | |
| "loss": 0.7298, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.17777777777778e-06, | |
| "loss": 0.7556, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.122222222222223e-06, | |
| "loss": 0.6975, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.066666666666667e-06, | |
| "loss": 0.7299, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.011111111111111e-06, | |
| "loss": 0.6555, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.955555555555555e-06, | |
| "loss": 0.6755, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 0.5832, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "eval_cer": 56.474494415937215, | |
| "eval_loss": 1.0634080171585083, | |
| "eval_runtime": 219.1746, | |
| "eval_samples_per_second": 4.499, | |
| "eval_steps_per_second": 2.249, | |
| "eval_wer": 56.30534351145038, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.844444444444445e-06, | |
| "loss": 0.5757, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.788888888888891e-06, | |
| "loss": 0.6065, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.733333333333333e-06, | |
| "loss": 0.5487, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.677777777777779e-06, | |
| "loss": 0.6339, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.622222222222223e-06, | |
| "loss": 0.5477, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.566666666666667e-06, | |
| "loss": 0.597, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.511111111111113e-06, | |
| "loss": 0.5394, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.455555555555555e-06, | |
| "loss": 0.52, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.6005, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.344444444444445e-06, | |
| "loss": 0.4988, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.288888888888889e-06, | |
| "loss": 0.5381, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.233333333333335e-06, | |
| "loss": 0.5005, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.177777777777779e-06, | |
| "loss": 0.3982, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.122222222222223e-06, | |
| "loss": 0.5537, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.066666666666667e-06, | |
| "loss": 0.3985, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.011111111111113e-06, | |
| "loss": 0.3653, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.955555555555557e-06, | |
| "loss": 0.4001, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.4278, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.844444444444446e-06, | |
| "loss": 0.3982, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.788888888888889e-06, | |
| "loss": 0.3967, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.733333333333334e-06, | |
| "loss": 0.4366, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.677777777777778e-06, | |
| "loss": 0.4665, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.622222222222223e-06, | |
| "loss": 0.4292, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 7.566666666666667e-06, | |
| "loss": 0.2518, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.511111111111111e-06, | |
| "loss": 0.1433, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.455555555555556e-06, | |
| "loss": 0.1932, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.1875, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.344444444444445e-06, | |
| "loss": 0.2236, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.28888888888889e-06, | |
| "loss": 0.2445, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.233333333333334e-06, | |
| "loss": 0.1751, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.177777777777778e-06, | |
| "loss": 0.1976, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.122222222222222e-06, | |
| "loss": 0.1307, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.066666666666667e-06, | |
| "loss": 0.1849, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.011111111111112e-06, | |
| "loss": 0.1233, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.955555555555557e-06, | |
| "loss": 0.136, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.1014, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 6.844444444444445e-06, | |
| "loss": 0.1366, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 6.788888888888889e-06, | |
| "loss": 0.2011, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.733333333333334e-06, | |
| "loss": 0.0962, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.677777777777779e-06, | |
| "loss": 0.1467, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "eval_cer": 51.01116812556595, | |
| "eval_loss": 1.040711760520935, | |
| "eval_runtime": 219.0004, | |
| "eval_samples_per_second": 4.502, | |
| "eval_steps_per_second": 2.251, | |
| "eval_wer": 50.961832061068705, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.6222222222222236e-06, | |
| "loss": 0.1376, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.566666666666667e-06, | |
| "loss": 0.1322, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.511111111111112e-06, | |
| "loss": 0.0889, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.455555555555556e-06, | |
| "loss": 0.0725, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.084, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.3444444444444454e-06, | |
| "loss": 0.078, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.28888888888889e-06, | |
| "loss": 0.0666, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.2333333333333335e-06, | |
| "loss": 0.1335, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.177777777777778e-06, | |
| "loss": 0.0919, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.1222222222222224e-06, | |
| "loss": 0.0966, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 6.066666666666667e-06, | |
| "loss": 0.0947, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 6.011111111111112e-06, | |
| "loss": 0.0731, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.955555555555555e-06, | |
| "loss": 0.1585, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.1475, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 5.844444444444445e-06, | |
| "loss": 0.0635, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 5.788888888888889e-06, | |
| "loss": 0.0489, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 5.733333333333334e-06, | |
| "loss": 0.0325, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 5.677777777777779e-06, | |
| "loss": 0.0661, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 5.622222222222222e-06, | |
| "loss": 0.0433, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 5.566666666666667e-06, | |
| "loss": 0.0215, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 5.511111111111112e-06, | |
| "loss": 0.0477, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 5.455555555555556e-06, | |
| "loss": 0.0315, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.0798, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.344444444444446e-06, | |
| "loss": 0.0298, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.288888888888889e-06, | |
| "loss": 0.033, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.233333333333334e-06, | |
| "loss": 0.0151, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.177777777777779e-06, | |
| "loss": 0.0465, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.122222222222223e-06, | |
| "loss": 0.0138, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.0666666666666676e-06, | |
| "loss": 0.0293, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.011111111111111e-06, | |
| "loss": 0.0296, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 4.9555555555555565e-06, | |
| "loss": 0.0336, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.0377, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 4.8444444444444446e-06, | |
| "loss": 0.0345, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 4.7888888888888894e-06, | |
| "loss": 0.0129, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 4.7333333333333335e-06, | |
| "loss": 0.0141, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 4.677777777777778e-06, | |
| "loss": 0.0165, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 4.622222222222222e-06, | |
| "loss": 0.0318, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.566666666666667e-06, | |
| "loss": 0.0161, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.511111111111111e-06, | |
| "loss": 0.028, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 4.455555555555555e-06, | |
| "loss": 0.016, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "eval_cer": 46.5137337760338, | |
| "eval_loss": 1.0225664377212524, | |
| "eval_runtime": 217.5747, | |
| "eval_samples_per_second": 4.532, | |
| "eval_steps_per_second": 2.266, | |
| "eval_wer": 46.44274809160305, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.0509, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 4.344444444444445e-06, | |
| "loss": 0.0254, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 4.288888888888889e-06, | |
| "loss": 0.0075, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 4.233333333333334e-06, | |
| "loss": 0.0303, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 4.177777777777778e-06, | |
| "loss": 0.0127, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 4.122222222222222e-06, | |
| "loss": 0.0184, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 4.066666666666667e-06, | |
| "loss": 0.0311, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 4.011111111111111e-06, | |
| "loss": 0.0012, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 3.955555555555556e-06, | |
| "loss": 0.0106, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.0085, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.844444444444445e-06, | |
| "loss": 0.0016, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.7888888888888893e-06, | |
| "loss": 0.0083, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.7333333333333337e-06, | |
| "loss": 0.0099, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.6777777777777778e-06, | |
| "loss": 0.0017, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.6222222222222226e-06, | |
| "loss": 0.0097, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 3.566666666666667e-06, | |
| "loss": 0.0083, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 3.511111111111111e-06, | |
| "loss": 0.0182, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.455555555555556e-06, | |
| "loss": 0.0041, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.0321, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.3444444444444445e-06, | |
| "loss": 0.0009, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.2888888888888894e-06, | |
| "loss": 0.0008, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.2333333333333334e-06, | |
| "loss": 0.0016, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.177777777777778e-06, | |
| "loss": 0.0368, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.1222222222222228e-06, | |
| "loss": 0.006, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.066666666666667e-06, | |
| "loss": 0.0023, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.0111111111111113e-06, | |
| "loss": 0.001, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 2.955555555555556e-06, | |
| "loss": 0.0012, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.0009, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 2.8444444444444446e-06, | |
| "loss": 0.0009, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 2.788888888888889e-06, | |
| "loss": 0.0005, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 2.7333333333333336e-06, | |
| "loss": 0.0095, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 2.677777777777778e-06, | |
| "loss": 0.0072, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 2.6222222222222225e-06, | |
| "loss": 0.0203, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 2.566666666666667e-06, | |
| "loss": 0.001, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 2.5111111111111114e-06, | |
| "loss": 0.0004, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 2.455555555555556e-06, | |
| "loss": 0.0224, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.0012, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 2.3444444444444448e-06, | |
| "loss": 0.0002, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 2.2888888888888892e-06, | |
| "loss": 0.0003, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 2.2333333333333333e-06, | |
| "loss": 0.0001, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "eval_cer": 45.60821008149713, | |
| "eval_loss": 0.9974298477172852, | |
| "eval_runtime": 214.4422, | |
| "eval_samples_per_second": 4.598, | |
| "eval_steps_per_second": 2.299, | |
| "eval_wer": 45.465648854961835, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 2.1777777777777777e-06, | |
| "loss": 0.0002, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 2.1222222222222226e-06, | |
| "loss": 0.0003, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.0666666666666666e-06, | |
| "loss": 0.0011, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.011111111111111e-06, | |
| "loss": 0.0002, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 1.955555555555556e-06, | |
| "loss": 0.0005, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.0003, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 1.8444444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 1.788888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 1.7333333333333336e-06, | |
| "loss": 0.0018, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 1.6777777777777779e-06, | |
| "loss": 0.0002, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 1.6222222222222223e-06, | |
| "loss": 0.0001, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 1.566666666666667e-06, | |
| "loss": 0.0037, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 1.5111111111111112e-06, | |
| "loss": 0.0002, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 1.4555555555555557e-06, | |
| "loss": 0.0001, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.0001, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 1.3444444444444446e-06, | |
| "loss": 0.0004, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 1.288888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 1.2333333333333335e-06, | |
| "loss": 0.0001, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 1.1777777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 1.1222222222222222e-06, | |
| "loss": 0.0005, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 1.066666666666667e-06, | |
| "loss": 0.0014, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 1.0111111111111111e-06, | |
| "loss": 0.0001, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 9.555555555555556e-07, | |
| "loss": 0.0022, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.0001, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 8.444444444444445e-07, | |
| "loss": 0.0003, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 7.888888888888889e-07, | |
| "loss": 0.0001, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 7.333333333333334e-07, | |
| "loss": 0.0003, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 6.777777777777779e-07, | |
| "loss": 0.0001, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 6.222222222222223e-07, | |
| "loss": 0.0001, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 5.666666666666667e-07, | |
| "loss": 0.0001, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 5.111111111111112e-07, | |
| "loss": 0.0001, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 4.5555555555555563e-07, | |
| "loss": 0.0001, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.0001, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 3.444444444444445e-07, | |
| "loss": 0.0001, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 2.888888888888889e-07, | |
| "loss": 0.0001, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 2.3333333333333336e-07, | |
| "loss": 0.0001, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 1.777777777777778e-07, | |
| "loss": 0.0001, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 1.2222222222222225e-07, | |
| "loss": 0.0001, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 6.666666666666668e-08, | |
| "loss": 0.0001, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 1.1111111111111112e-08, | |
| "loss": 0.0001, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "eval_cer": 45.36673709628735, | |
| "eval_loss": 0.994395911693573, | |
| "eval_runtime": 216.4096, | |
| "eval_samples_per_second": 4.556, | |
| "eval_steps_per_second": 2.278, | |
| "eval_wer": 45.2824427480916, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "step": 5000, | |
| "total_flos": 1.019992544575488e+19, | |
| "train_loss": 0.3284327008752851, | |
| "train_runtime": 3725.671, | |
| "train_samples_per_second": 2.684, | |
| "train_steps_per_second": 1.342 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 1.019992544575488e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |