| { | |
| "best_metric": 73.01634232878185, | |
| "best_model_checkpoint": "./checkpoint-5000", | |
| "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": 4.4e-07, | |
| "loss": 8.1423, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 7.8269, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.44e-06, | |
| "loss": 7.4222, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.94e-06, | |
| "loss": 6.5485, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 5.8875, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 5.7146, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.44e-06, | |
| "loss": 4.936, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.94e-06, | |
| "loss": 4.5072, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 4.5945, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.94e-06, | |
| "loss": 4.1959, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 3.983, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 5.94e-06, | |
| "loss": 4.0814, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 3.8272, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 3.5149, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 3.6732, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.94e-06, | |
| "loss": 3.5016, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.44e-06, | |
| "loss": 3.3331, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.94e-06, | |
| "loss": 3.1667, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 3.1868, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 2.8215, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 2.8644, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 2.9609, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.84e-06, | |
| "loss": 2.6745, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 2.4092, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 2.5981, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 2.2719, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 2.1846, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 2.4153, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 2.1667, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 2.0301, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 2.0543, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 1.9007, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 1.775, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 1.9376, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 1.837, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 1.7839, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 1.6248, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 1.6341, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 1.4855, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 1.5547, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "eval_loss": 1.924423098564148, | |
| "eval_runtime": 1185.7279, | |
| "eval_samples_per_second": 4.275, | |
| "eval_steps_per_second": 0.067, | |
| "eval_wer": 113.44479088270079, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 8.84e-06, | |
| "loss": 1.6874, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 1.5095, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 1.3423, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 1.4235, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 1.3073, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 1.2523, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 1.4573, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 1.3365, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 1.1963, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 8.34e-06, | |
| "loss": 1.2298, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 1.1486, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 1.0733, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 1.2373, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 1.1736, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 1.1319, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 1.0365, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 1.0208, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.9332, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 1.038, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 1.1189, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 1.0068, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.8809, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.9296, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.848, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.8336, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 1.0223, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.9194, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.8019, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.8496, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.7617, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.7217, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.8814, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.8341, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.7786, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.7169, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.7032, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.6348, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.7406, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.7945, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.7244, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "eval_loss": 1.4592993259429932, | |
| "eval_runtime": 1048.4388, | |
| "eval_samples_per_second": 4.835, | |
| "eval_steps_per_second": 0.076, | |
| "eval_wer": 81.01279432319105, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.6017, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.6666, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 0.5832, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 0.5903, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 0.7552, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 6.34e-06, | |
| "loss": 0.6376, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.5746, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.6044, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.5465, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.5101, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.6415, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 6.006666666666667e-06, | |
| "loss": 0.5965, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 5.951111111111112e-06, | |
| "loss": 0.5793, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 5.895555555555557e-06, | |
| "loss": 0.5146, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 5.84e-06, | |
| "loss": 0.5071, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 5.784444444444445e-06, | |
| "loss": 0.4509, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 5.72888888888889e-06, | |
| "loss": 0.5399, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 5.673333333333334e-06, | |
| "loss": 0.592, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 5.617777777777779e-06, | |
| "loss": 0.5237, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 5.562222222222222e-06, | |
| "loss": 0.4457, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 5.506666666666667e-06, | |
| "loss": 0.4826, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 5.451111111111112e-06, | |
| "loss": 0.4252, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 5.3955555555555565e-06, | |
| "loss": 0.435, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 5.3400000000000005e-06, | |
| "loss": 0.5675, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 5.2844444444444454e-06, | |
| "loss": 0.4779, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 5.228888888888889e-06, | |
| "loss": 0.4128, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 5.1733333333333335e-06, | |
| "loss": 0.4529, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 5.117777777777778e-06, | |
| "loss": 0.3921, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 5.062222222222222e-06, | |
| "loss": 0.3782, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 5.006666666666667e-06, | |
| "loss": 0.4842, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 4.951111111111111e-06, | |
| "loss": 0.4412, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 4.895555555555556e-06, | |
| "loss": 0.423, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 4.84e-06, | |
| "loss": 0.3937, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 4.784444444444445e-06, | |
| "loss": 0.3722, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 4.728888888888889e-06, | |
| "loss": 0.3247, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 4.673333333333333e-06, | |
| "loss": 0.415, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 4.617777777777778e-06, | |
| "loss": 0.4442, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 4.562222222222222e-06, | |
| "loss": 0.3891, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 4.506666666666667e-06, | |
| "loss": 0.3393, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 4.451111111111112e-06, | |
| "loss": 0.3583, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "eval_loss": 1.4019403457641602, | |
| "eval_runtime": 1037.1427, | |
| "eval_samples_per_second": 4.887, | |
| "eval_steps_per_second": 0.077, | |
| "eval_wer": 74.34146865928395, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 4.395555555555556e-06, | |
| "loss": 0.3164, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 4.34e-06, | |
| "loss": 0.3259, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 4.284444444444445e-06, | |
| "loss": 0.4335, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 4.228888888888889e-06, | |
| "loss": 0.3616, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 4.173333333333334e-06, | |
| "loss": 0.3108, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 4.117777777777779e-06, | |
| "loss": 0.3438, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 4.062222222222223e-06, | |
| "loss": 0.2942, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 4.006666666666667e-06, | |
| "loss": 0.2834, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 3.951111111111112e-06, | |
| "loss": 0.3776, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 3.895555555555556e-06, | |
| "loss": 0.3448, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 0.3145, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 20.03, | |
| "learning_rate": 3.784444444444445e-06, | |
| "loss": 0.3069, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "learning_rate": 3.728888888888889e-06, | |
| "loss": 0.2872, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 3.673333333333334e-06, | |
| "loss": 0.2423, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 3.617777777777778e-06, | |
| "loss": 0.3258, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 21.02, | |
| "learning_rate": 3.5622222222222224e-06, | |
| "loss": 0.3419, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 21.02, | |
| "learning_rate": 3.5066666666666673e-06, | |
| "loss": 0.3062, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 21.03, | |
| "learning_rate": 3.4511111111111113e-06, | |
| "loss": 0.2604, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 21.03, | |
| "learning_rate": 3.3955555555555558e-06, | |
| "loss": 0.2776, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "learning_rate": 3.3400000000000006e-06, | |
| "loss": 0.24, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 3.2844444444444447e-06, | |
| "loss": 0.2569, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 3.228888888888889e-06, | |
| "loss": 0.3486, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 22.02, | |
| "learning_rate": 3.173333333333334e-06, | |
| "loss": 0.2795, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 22.02, | |
| "learning_rate": 3.117777777777778e-06, | |
| "loss": 0.2426, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 22.03, | |
| "learning_rate": 3.0622222222222225e-06, | |
| "loss": 0.2677, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "learning_rate": 3.0066666666666674e-06, | |
| "loss": 0.2328, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 23.01, | |
| "learning_rate": 2.9511111111111114e-06, | |
| "loss": 0.2195, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 23.01, | |
| "learning_rate": 2.895555555555556e-06, | |
| "loss": 0.3037, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 23.02, | |
| "learning_rate": 2.84e-06, | |
| "loss": 0.2776, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 23.02, | |
| "learning_rate": 2.784444444444445e-06, | |
| "loss": 0.2467, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 23.03, | |
| "learning_rate": 2.7288888888888893e-06, | |
| "loss": 0.2432, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "learning_rate": 2.6733333333333333e-06, | |
| "loss": 0.2286, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "learning_rate": 2.617777777777778e-06, | |
| "loss": 0.1972, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "learning_rate": 2.5622222222222226e-06, | |
| "loss": 0.2645, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 24.02, | |
| "learning_rate": 2.5066666666666667e-06, | |
| "loss": 0.2765, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 24.02, | |
| "learning_rate": 2.451111111111111e-06, | |
| "loss": 0.2413, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 24.03, | |
| "learning_rate": 2.3955555555555556e-06, | |
| "loss": 0.2153, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 24.03, | |
| "learning_rate": 2.3400000000000005e-06, | |
| "loss": 0.2231, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 2.2844444444444445e-06, | |
| "loss": 0.1911, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 25.01, | |
| "learning_rate": 2.228888888888889e-06, | |
| "loss": 0.2157, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 25.01, | |
| "eval_loss": 1.4249294996261597, | |
| "eval_runtime": 947.0942, | |
| "eval_samples_per_second": 5.352, | |
| "eval_steps_per_second": 0.084, | |
| "eval_wer": 73.89528007741103, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 25.02, | |
| "learning_rate": 2.1733333333333334e-06, | |
| "loss": 0.2837, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 25.02, | |
| "learning_rate": 2.117777777777778e-06, | |
| "loss": 0.2335, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 25.02, | |
| "learning_rate": 2.0622222222222223e-06, | |
| "loss": 0.1945, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 25.03, | |
| "learning_rate": 2.006666666666667e-06, | |
| "loss": 0.2243, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "learning_rate": 1.9511111111111113e-06, | |
| "loss": 0.1902, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 26.01, | |
| "learning_rate": 1.8955555555555557e-06, | |
| "loss": 0.181, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 26.01, | |
| "learning_rate": 1.8400000000000002e-06, | |
| "loss": 0.2623, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 1.7844444444444444e-06, | |
| "loss": 0.2248, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 1.728888888888889e-06, | |
| "loss": 0.2033, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 26.03, | |
| "learning_rate": 1.6733333333333335e-06, | |
| "loss": 0.2094, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "learning_rate": 1.6177777777777778e-06, | |
| "loss": 0.1873, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 27.01, | |
| "learning_rate": 1.5622222222222225e-06, | |
| "loss": 0.1661, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 27.01, | |
| "learning_rate": 1.506666666666667e-06, | |
| "loss": 0.2289, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 27.02, | |
| "learning_rate": 1.4511111111111112e-06, | |
| "loss": 0.2331, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 27.02, | |
| "learning_rate": 1.3955555555555556e-06, | |
| "loss": 0.2085, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 27.03, | |
| "learning_rate": 1.34e-06, | |
| "loss": 0.1849, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "learning_rate": 1.2844444444444445e-06, | |
| "loss": 0.19, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 28.01, | |
| "learning_rate": 1.228888888888889e-06, | |
| "loss": 0.1642, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 28.01, | |
| "learning_rate": 1.1733333333333335e-06, | |
| "loss": 0.1884, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 28.02, | |
| "learning_rate": 1.117777777777778e-06, | |
| "loss": 0.2501, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 28.02, | |
| "learning_rate": 1.0622222222222222e-06, | |
| "loss": 0.2036, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 28.03, | |
| "learning_rate": 1.0066666666666668e-06, | |
| "loss": 0.1695, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 28.03, | |
| "learning_rate": 9.511111111111111e-07, | |
| "loss": 0.1955, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "learning_rate": 8.955555555555557e-07, | |
| "loss": 0.166, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 29.01, | |
| "learning_rate": 8.400000000000001e-07, | |
| "loss": 0.1614, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 29.01, | |
| "learning_rate": 7.844444444444445e-07, | |
| "loss": 0.2334, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 29.02, | |
| "learning_rate": 7.28888888888889e-07, | |
| "loss": 0.2081, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 29.02, | |
| "learning_rate": 6.733333333333334e-07, | |
| "loss": 0.178, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 29.03, | |
| "learning_rate": 6.177777777777778e-07, | |
| "loss": 0.1898, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "learning_rate": 5.622222222222223e-07, | |
| "loss": 0.1705, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 30.01, | |
| "learning_rate": 5.066666666666667e-07, | |
| "loss": 0.1512, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 30.01, | |
| "learning_rate": 4.511111111111111e-07, | |
| "loss": 0.2121, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 30.02, | |
| "learning_rate": 3.9555555555555557e-07, | |
| "loss": 0.2107, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 30.02, | |
| "learning_rate": 3.4000000000000003e-07, | |
| "loss": 0.1924, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 30.03, | |
| "learning_rate": 2.844444444444445e-07, | |
| "loss": 0.17, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "learning_rate": 2.2888888888888892e-07, | |
| "loss": 0.1754, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 31.01, | |
| "learning_rate": 1.7333333333333335e-07, | |
| "loss": 0.1506, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 31.01, | |
| "learning_rate": 1.1777777777777778e-07, | |
| "loss": 0.184, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 31.02, | |
| "learning_rate": 6.222222222222223e-08, | |
| "loss": 0.2325, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 31.02, | |
| "learning_rate": 6.666666666666667e-09, | |
| "loss": 0.1897, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 31.02, | |
| "eval_loss": 1.4344117641448975, | |
| "eval_runtime": 1007.6428, | |
| "eval_samples_per_second": 5.031, | |
| "eval_steps_per_second": 0.079, | |
| "eval_wer": 73.01634232878185, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 31.02, | |
| "step": 5000, | |
| "total_flos": 5.693284122624e+18, | |
| "train_loss": 1.0857838482379913, | |
| "train_runtime": 51440.7533, | |
| "train_samples_per_second": 12.441, | |
| "train_steps_per_second": 0.097 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 5.693284122624e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |