| { | |
| "best_metric": 19.368282518096073, | |
| "best_model_checkpoint": "./checkpoint-4000", | |
| "epoch": 6.1252, | |
| "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": 4.800000000000001e-07, | |
| "loss": 1.4296, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.800000000000001e-07, | |
| "loss": 1.1127, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.48e-06, | |
| "loss": 0.5985, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.98e-06, | |
| "loss": 0.3874, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.4800000000000004e-06, | |
| "loss": 0.2839, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9800000000000003e-06, | |
| "loss": 0.2291, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.48e-06, | |
| "loss": 0.1973, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.980000000000001e-06, | |
| "loss": 0.1512, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.48e-06, | |
| "loss": 0.1399, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.980000000000001e-06, | |
| "loss": 0.1351, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.480000000000001e-06, | |
| "loss": 0.1304, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.98e-06, | |
| "loss": 0.1156, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.480000000000001e-06, | |
| "loss": 0.1333, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.98e-06, | |
| "loss": 0.0772, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.48e-06, | |
| "loss": 0.095, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.980000000000002e-06, | |
| "loss": 0.0779, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.48e-06, | |
| "loss": 0.0813, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.98e-06, | |
| "loss": 0.0933, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.48e-06, | |
| "loss": 0.0805, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.980000000000001e-06, | |
| "loss": 0.0665, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.946666666666667e-06, | |
| "loss": 0.0787, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.891111111111113e-06, | |
| "loss": 0.0872, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.835555555555556e-06, | |
| "loss": 0.0749, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.780000000000001e-06, | |
| "loss": 0.0721, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.724444444444445e-06, | |
| "loss": 0.0908, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.66888888888889e-06, | |
| "loss": 0.0845, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.613333333333335e-06, | |
| "loss": 0.0622, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.557777777777777e-06, | |
| "loss": 0.0559, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.502222222222223e-06, | |
| "loss": 0.0556, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.446666666666667e-06, | |
| "loss": 0.0966, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.391111111111111e-06, | |
| "loss": 0.0812, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.335555555555557e-06, | |
| "loss": 0.0492, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.280000000000001e-06, | |
| "loss": 0.0381, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.224444444444445e-06, | |
| "loss": 0.0428, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.168888888888889e-06, | |
| "loss": 0.0405, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.113333333333335e-06, | |
| "loss": 0.0474, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.057777777777779e-06, | |
| "loss": 0.041, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.002222222222223e-06, | |
| "loss": 0.0394, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.946666666666669e-06, | |
| "loss": 0.0437, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.891111111111111e-06, | |
| "loss": 0.0471, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "eval_cer": 8.981987991994664, | |
| "eval_loss": 0.5961337685585022, | |
| "eval_runtime": 892.0726, | |
| "eval_samples_per_second": 1.387, | |
| "eval_steps_per_second": 0.044, | |
| "eval_wer": 20.289537179205965, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.835555555555557e-06, | |
| "loss": 0.0379, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.78e-06, | |
| "loss": 0.0464, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.724444444444445e-06, | |
| "loss": 0.025, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.66888888888889e-06, | |
| "loss": 0.0324, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.613333333333333e-06, | |
| "loss": 0.025, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.557777777777778e-06, | |
| "loss": 0.0304, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.502222222222223e-06, | |
| "loss": 0.0356, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.446666666666668e-06, | |
| "loss": 0.031, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.391111111111112e-06, | |
| "loss": 0.0182, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.335555555555556e-06, | |
| "loss": 0.0344, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.28e-06, | |
| "loss": 0.0329, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.224444444444444e-06, | |
| "loss": 0.032, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.16888888888889e-06, | |
| "loss": 0.0192, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.113333333333334e-06, | |
| "loss": 0.0359, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.057777777777778e-06, | |
| "loss": 0.0386, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.002222222222222e-06, | |
| "loss": 0.0201, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.946666666666666e-06, | |
| "loss": 0.0213, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.891111111111112e-06, | |
| "loss": 0.0195, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 7.835555555555556e-06, | |
| "loss": 0.0423, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.78e-06, | |
| "loss": 0.042, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.724444444444446e-06, | |
| "loss": 0.0225, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.66888888888889e-06, | |
| "loss": 0.0139, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.613333333333334e-06, | |
| "loss": 0.0132, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.557777777777779e-06, | |
| "loss": 0.017, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.502222222222223e-06, | |
| "loss": 0.016, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.446666666666668e-06, | |
| "loss": 0.0206, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.3911111111111125e-06, | |
| "loss": 0.0157, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.335555555555556e-06, | |
| "loss": 0.019, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.280000000000001e-06, | |
| "loss": 0.0208, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.224444444444445e-06, | |
| "loss": 0.0148, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.1688888888888895e-06, | |
| "loss": 0.0231, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.113333333333334e-06, | |
| "loss": 0.0121, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.057777777777778e-06, | |
| "loss": 0.0121, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 7.0022222222222225e-06, | |
| "loss": 0.0115, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.946666666666667e-06, | |
| "loss": 0.0136, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.891111111111111e-06, | |
| "loss": 0.0146, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.835555555555556e-06, | |
| "loss": 0.0148, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.780000000000001e-06, | |
| "loss": 0.0093, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.724444444444444e-06, | |
| "loss": 0.0154, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.668888888888889e-06, | |
| "loss": 0.0194, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "eval_cer": 9.710473649099399, | |
| "eval_loss": 1.0999209880828857, | |
| "eval_runtime": 990.357, | |
| "eval_samples_per_second": 1.249, | |
| "eval_steps_per_second": 0.039, | |
| "eval_wer": 22.614608466769027, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.613333333333334e-06, | |
| "loss": 0.0166, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.557777777777778e-06, | |
| "loss": 0.0068, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.502222222222223e-06, | |
| "loss": 0.0149, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.446666666666668e-06, | |
| "loss": 0.0184, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.391111111111111e-06, | |
| "loss": 0.0097, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 6.335555555555556e-06, | |
| "loss": 0.0089, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 6.280000000000001e-06, | |
| "loss": 0.0097, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 6.224444444444445e-06, | |
| "loss": 0.0176, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.16888888888889e-06, | |
| "loss": 0.0239, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.113333333333333e-06, | |
| "loss": 0.0129, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.057777777777778e-06, | |
| "loss": 0.006, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.002222222222223e-06, | |
| "loss": 0.0041, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 5.946666666666668e-06, | |
| "loss": 0.0076, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 5.891111111111112e-06, | |
| "loss": 0.0049, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 5.8355555555555565e-06, | |
| "loss": 0.0085, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 5.78e-06, | |
| "loss": 0.0082, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.724444444444445e-06, | |
| "loss": 0.008, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.6688888888888895e-06, | |
| "loss": 0.0086, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.613333333333334e-06, | |
| "loss": 0.0081, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.557777777777778e-06, | |
| "loss": 0.0089, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.5022222222222224e-06, | |
| "loss": 0.0077, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.4466666666666665e-06, | |
| "loss": 0.0044, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.391111111111111e-06, | |
| "loss": 0.0056, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.335555555555556e-06, | |
| "loss": 0.0068, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.28e-06, | |
| "loss": 0.0058, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.224444444444445e-06, | |
| "loss": 0.0082, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.168888888888889e-06, | |
| "loss": 0.0038, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.113333333333333e-06, | |
| "loss": 0.0066, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.057777777777778e-06, | |
| "loss": 0.0093, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.002222222222223e-06, | |
| "loss": 0.0074, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.946666666666667e-06, | |
| "loss": 0.0033, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.891111111111111e-06, | |
| "loss": 0.0071, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 4.835555555555556e-06, | |
| "loss": 0.0076, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 4.78e-06, | |
| "loss": 0.0045, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 4.724444444444445e-06, | |
| "loss": 0.0058, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 4.66888888888889e-06, | |
| "loss": 0.0026, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 4.613333333333334e-06, | |
| "loss": 0.0065, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 4.557777777777778e-06, | |
| "loss": 0.0097, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 4.502222222222223e-06, | |
| "loss": 0.006, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 4.446666666666667e-06, | |
| "loss": 0.002, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "eval_cer": 8.837891927951969, | |
| "eval_loss": 0.7289467453956604, | |
| "eval_runtime": 919.6749, | |
| "eval_samples_per_second": 1.345, | |
| "eval_steps_per_second": 0.042, | |
| "eval_wer": 20.201798640052644, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 4.391111111111112e-06, | |
| "loss": 0.0018, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 4.3355555555555565e-06, | |
| "loss": 0.0034, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 4.2800000000000005e-06, | |
| "loss": 0.0022, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.2244444444444446e-06, | |
| "loss": 0.0038, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.168888888888889e-06, | |
| "loss": 0.0019, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.1133333333333335e-06, | |
| "loss": 0.0043, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.057777777777778e-06, | |
| "loss": 0.0033, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.002222222222222e-06, | |
| "loss": 0.0024, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.946666666666667e-06, | |
| "loss": 0.0038, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.891111111111111e-06, | |
| "loss": 0.0026, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.835555555555555e-06, | |
| "loss": 0.0012, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.7800000000000002e-06, | |
| "loss": 0.0019, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.724444444444445e-06, | |
| "loss": 0.0026, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.668888888888889e-06, | |
| "loss": 0.0023, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.6133333333333336e-06, | |
| "loss": 0.0031, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.5577777777777785e-06, | |
| "loss": 0.0012, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.5022222222222225e-06, | |
| "loss": 0.0036, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.446666666666667e-06, | |
| "loss": 0.0038, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.391111111111111e-06, | |
| "loss": 0.0024, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.335555555555556e-06, | |
| "loss": 0.0013, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.2800000000000004e-06, | |
| "loss": 0.0022, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.2244444444444444e-06, | |
| "loss": 0.002, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.1688888888888893e-06, | |
| "loss": 0.0029, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.1133333333333337e-06, | |
| "loss": 0.0028, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.0577777777777778e-06, | |
| "loss": 0.0014, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.0022222222222227e-06, | |
| "loss": 0.0021, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 2.946666666666667e-06, | |
| "loss": 0.0042, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 2.891111111111111e-06, | |
| "loss": 0.0021, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 2.835555555555556e-06, | |
| "loss": 0.0014, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 2.7800000000000005e-06, | |
| "loss": 0.0008, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.7244444444444445e-06, | |
| "loss": 0.0008, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.6688888888888894e-06, | |
| "loss": 0.0006, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 2.6133333333333334e-06, | |
| "loss": 0.0012, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 2.557777777777778e-06, | |
| "loss": 0.0017, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 2.5022222222222224e-06, | |
| "loss": 0.0019, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 2.446666666666667e-06, | |
| "loss": 0.0012, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 2.3911111111111113e-06, | |
| "loss": 0.0009, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 2.3355555555555557e-06, | |
| "loss": 0.0011, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 2.28e-06, | |
| "loss": 0.0009, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 2.2244444444444447e-06, | |
| "loss": 0.0006, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "eval_cer": 8.437625083388927, | |
| "eval_loss": 0.7791085243225098, | |
| "eval_runtime": 930.9266, | |
| "eval_samples_per_second": 1.329, | |
| "eval_steps_per_second": 0.042, | |
| "eval_wer": 19.368282518096073, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 2.168888888888889e-06, | |
| "loss": 0.0006, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 2.1133333333333336e-06, | |
| "loss": 0.0009, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.057777777777778e-06, | |
| "loss": 0.0008, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.0022222222222225e-06, | |
| "loss": 0.0013, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 1.9466666666666665e-06, | |
| "loss": 0.0004, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 1.8911111111111114e-06, | |
| "loss": 0.0005, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 1.8355555555555557e-06, | |
| "loss": 0.0016, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 1.7800000000000001e-06, | |
| "loss": 0.0009, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 1.7244444444444448e-06, | |
| "loss": 0.0004, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 1.668888888888889e-06, | |
| "loss": 0.0006, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 1.6133333333333335e-06, | |
| "loss": 0.0007, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 1.5577777777777777e-06, | |
| "loss": 0.0013, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 1.5022222222222224e-06, | |
| "loss": 0.0009, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 1.4466666666666669e-06, | |
| "loss": 0.0005, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 1.3911111111111111e-06, | |
| "loss": 0.0003, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 1.3355555555555558e-06, | |
| "loss": 0.0011, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 1.28e-06, | |
| "loss": 0.0005, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 1.2244444444444445e-06, | |
| "loss": 0.0004, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 1.168888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 1.1133333333333334e-06, | |
| "loss": 0.0003, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 1.0577777777777779e-06, | |
| "loss": 0.0003, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 1.0022222222222223e-06, | |
| "loss": 0.0003, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 9.466666666666667e-07, | |
| "loss": 0.0005, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 8.911111111111112e-07, | |
| "loss": 0.0003, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 8.355555555555556e-07, | |
| "loss": 0.0003, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 7.8e-07, | |
| "loss": 0.0004, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 7.244444444444446e-07, | |
| "loss": 0.0003, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 6.68888888888889e-07, | |
| "loss": 0.0003, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 6.133333333333333e-07, | |
| "loss": 0.0003, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 5.577777777777779e-07, | |
| "loss": 0.0003, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 5.022222222222222e-07, | |
| "loss": 0.0002, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 4.466666666666667e-07, | |
| "loss": 0.0002, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 3.9111111111111115e-07, | |
| "loss": 0.0004, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 3.3555555555555556e-07, | |
| "loss": 0.0003, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 2.8e-07, | |
| "loss": 0.0002, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 2.2444444444444445e-07, | |
| "loss": 0.0004, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.6888888888888888e-07, | |
| "loss": 0.0003, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 1.1333333333333336e-07, | |
| "loss": 0.0003, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 5.777777777777778e-08, | |
| "loss": 0.0003, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 2.2222222222222225e-09, | |
| "loss": 0.0003, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "eval_cer": 8.728485657104738, | |
| "eval_loss": 0.7427007555961609, | |
| "eval_runtime": 932.7395, | |
| "eval_samples_per_second": 1.326, | |
| "eval_steps_per_second": 0.042, | |
| "eval_wer": 19.390217152884404, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "step": 5000, | |
| "total_flos": 3.2648946870583296e+20, | |
| "train_loss": 0.04111601705942303, | |
| "train_runtime": 74950.7687, | |
| "train_samples_per_second": 4.269, | |
| "train_steps_per_second": 0.067 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 3.2648946870583296e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |