| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 24.449877750611247, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.6666, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.4618, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.4005, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.2467, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.2931, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3e-06, | |
| "loss": 0.1967, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.253, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.1654, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.1921, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5e-06, | |
| "loss": 0.1238, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.1626, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6e-06, | |
| "loss": 0.1193, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.1468, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 7e-06, | |
| "loss": 0.1096, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.1499, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.109, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.1019, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 9e-06, | |
| "loss": 0.0609, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.0767, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 1e-05, | |
| "loss": 0.0703, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.0875, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.0741, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.0874, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.0705, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.056, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.0389, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.0449, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.0373, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.0445, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.0419, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.0492, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.0445, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.0344, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.0235, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.0217, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.0249, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.0242, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 9e-06, | |
| "loss": 0.0286, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.0285, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.0282, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "eval_loss": 0.2700173854827881, | |
| "eval_runtime": 1052.0208, | |
| "eval_samples_per_second": 2.751, | |
| "eval_steps_per_second": 0.172, | |
| "eval_wer": 0.2646660458816558, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.0236, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.0164, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.0125, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.0193, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.0132, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.018, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.0156, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.0172, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.0125, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.0123, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 6.23, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.0074, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.0121, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 6.48, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.007, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.0129, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 6.72, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.0072, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.0128, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.0084, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.0108, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.0059, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.0077, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.0053, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.0078, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 7.7, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.0049, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.0088, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 7.95, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.0043, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 8.07, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.0062, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.0047, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.0055, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 8.44, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.0039, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 8.56, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.0066, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 8.68, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.0049, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 8.8, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.0053, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.0035, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 7e-06, | |
| "loss": 0.0044, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 9.17, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.0026, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 9.29, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.0025, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 9.41, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.0024, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 9.54, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.0021, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 9.66, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.0026, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 9.78, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.0025, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 9.78, | |
| "eval_loss": 0.3433513939380646, | |
| "eval_runtime": 1047.6876, | |
| "eval_samples_per_second": 2.762, | |
| "eval_steps_per_second": 0.173, | |
| "eval_wer": 0.25539659696944045, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.0036, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.0041, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 10.15, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.0023, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 10.27, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.0017, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 10.39, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.0017, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 10.51, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.0018, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 10.64, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.0024, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 10.76, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.0021, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 10.88, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.0023, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.0013, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 11.12, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.002, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 11.25, | |
| "learning_rate": 6e-06, | |
| "loss": 0.0007, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 11.37, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.0023, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 11.49, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.0016, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 11.61, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.0014, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 11.74, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.0016, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 11.86, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.0018, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 11.98, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.001, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 12.1, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.0012, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 12.22, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.0007, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 12.35, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.0011, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 12.47, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.0011, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 12.59, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.0012, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 12.71, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.0004, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 12.84, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.001, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 12.96, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.0007, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 13.08, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.0007, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 13.2, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.0008, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 13.33, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.0012, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 13.45, | |
| "learning_rate": 5e-06, | |
| "loss": 0.0004, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 13.57, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.0004, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 13.69, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 13.81, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.0005, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 13.94, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 14.06, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.0003, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 14.18, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.0002, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 14.3, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.0002, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 14.43, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 14.55, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.0001, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 14.67, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.0001, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 14.67, | |
| "eval_loss": 0.36398327350616455, | |
| "eval_runtime": 1041.1039, | |
| "eval_samples_per_second": 2.78, | |
| "eval_steps_per_second": 0.174, | |
| "eval_wer": 0.2471006518242614, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 14.79, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 14.91, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 15.16, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.0002, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 15.28, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 15.4, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.0001, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 15.53, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 15.65, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.0001, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 15.77, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 15.89, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 16.14, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 16.26, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.0001, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 16.38, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.0001, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 16.5, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.0001, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 16.63, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.0002, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 16.75, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.0001, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 16.87, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 16.99, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.0, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 17.11, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.0001, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 17.24, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.0, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 17.36, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.0001, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 17.48, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.0, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 17.6, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.0001, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 17.73, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.0, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 17.85, | |
| "learning_rate": 3e-06, | |
| "loss": 0.0001, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 17.97, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.0, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 18.09, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 18.22, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.0, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 18.34, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.0, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 18.46, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.0001, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 18.58, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 18.7, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.0, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 18.83, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.0001, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 18.95, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 19.07, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.0, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 19.19, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.0, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 19.32, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.0, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 19.44, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.0, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 19.56, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 19.56, | |
| "eval_loss": 0.3901691138744354, | |
| "eval_runtime": 1049.6165, | |
| "eval_samples_per_second": 2.757, | |
| "eval_steps_per_second": 0.172, | |
| "eval_wer": 0.24942859561500041, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 19.68, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.0, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 19.8, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.0, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 19.93, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.0, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 20.05, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.0, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 20.17, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.0, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 20.29, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.0, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 20.42, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 0.0, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 20.54, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 0.0, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 20.66, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 0.0, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 20.78, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.0, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 20.9, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 0.0, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 21.03, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 0.0, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 21.15, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.0, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 21.27, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 0.0, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 21.39, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.0, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 21.52, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.0, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 21.64, | |
| "learning_rate": 1.2777777777777779e-06, | |
| "loss": 0.0, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 21.76, | |
| "learning_rate": 1.2222222222222223e-06, | |
| "loss": 0.0, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 21.88, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 0.0, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 22.13, | |
| "learning_rate": 1.0555555555555557e-06, | |
| "loss": 0.0, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 22.25, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.0, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 22.37, | |
| "learning_rate": 9.444444444444445e-07, | |
| "loss": 0.0, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 22.49, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 0.0, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 22.62, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.0, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 22.74, | |
| "learning_rate": 7.777777777777779e-07, | |
| "loss": 0.0, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 22.86, | |
| "learning_rate": 7.222222222222222e-07, | |
| "loss": 0.0, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 22.98, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 0.0, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 23.11, | |
| "learning_rate": 6.111111111111112e-07, | |
| "loss": 0.0, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 23.23, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.0, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 23.35, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.0, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 23.47, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 0.0, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 23.59, | |
| "learning_rate": 3.8888888888888895e-07, | |
| "loss": 0.0, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 23.72, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 0.0, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 23.84, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.0, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 23.96, | |
| "learning_rate": 2.2222222222222224e-07, | |
| "loss": 0.0, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 24.08, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 0.0, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 24.21, | |
| "learning_rate": 1.1111111111111112e-07, | |
| "loss": 0.0, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 24.33, | |
| "learning_rate": 5.555555555555556e-08, | |
| "loss": 0.0, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 24.45, | |
| "learning_rate": 0.0, | |
| "loss": 0.0, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 24.45, | |
| "eval_loss": 0.3972152769565582, | |
| "eval_runtime": 1049.2951, | |
| "eval_samples_per_second": 2.758, | |
| "eval_steps_per_second": 0.172, | |
| "eval_wer": 0.24870904935240837, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 24.45, | |
| "step": 5000, | |
| "total_flos": 1.6319880713207808e+20, | |
| "train_loss": 0.02697208818818908, | |
| "train_runtime": 40993.3876, | |
| "train_samples_per_second": 3.903, | |
| "train_steps_per_second": 0.122 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_train_epochs": 25, | |
| "save_steps": 1000, | |
| "total_flos": 1.6319880713207808e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |