|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 24.390243902439025, |
|
"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": 4.6000000000000004e-07, |
|
"loss": 0.6767, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.600000000000001e-07, |
|
"loss": 0.5124, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.46e-06, |
|
"loss": 0.348, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.9600000000000003e-06, |
|
"loss": 0.2998, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 2.46e-06, |
|
"loss": 0.2564, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 2.96e-06, |
|
"loss": 0.2307, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 3.46e-06, |
|
"loss": 0.2155, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 3.96e-06, |
|
"loss": 0.2004, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 4.4600000000000005e-06, |
|
"loss": 0.1708, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 4.960000000000001e-06, |
|
"loss": 0.1456, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 5.460000000000001e-06, |
|
"loss": 0.1454, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 5.9600000000000005e-06, |
|
"loss": 0.14, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"learning_rate": 6.460000000000001e-06, |
|
"loss": 0.1267, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"learning_rate": 6.96e-06, |
|
"loss": 0.1302, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 7.4600000000000006e-06, |
|
"loss": 0.1325, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"learning_rate": 7.960000000000002e-06, |
|
"loss": 0.1318, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 8.46e-06, |
|
"loss": 0.0898, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 8.96e-06, |
|
"loss": 0.0725, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 2.32, |
|
"learning_rate": 9.460000000000001e-06, |
|
"loss": 0.0791, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 2.44, |
|
"learning_rate": 9.960000000000001e-06, |
|
"loss": 0.0803, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"learning_rate": 9.94888888888889e-06, |
|
"loss": 0.0819, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 2.68, |
|
"learning_rate": 9.893333333333334e-06, |
|
"loss": 0.0858, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 2.8, |
|
"learning_rate": 9.837777777777778e-06, |
|
"loss": 0.0779, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 2.93, |
|
"learning_rate": 9.782222222222222e-06, |
|
"loss": 0.0776, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 3.05, |
|
"learning_rate": 9.726666666666668e-06, |
|
"loss": 0.0616, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 3.17, |
|
"learning_rate": 9.671111111111112e-06, |
|
"loss": 0.044, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 3.29, |
|
"learning_rate": 9.615555555555558e-06, |
|
"loss": 0.0467, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 3.41, |
|
"learning_rate": 9.56e-06, |
|
"loss": 0.0421, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 3.54, |
|
"learning_rate": 9.504444444444446e-06, |
|
"loss": 0.0424, |
|
"step": 725 |
|
}, |
|
{ |
|
"epoch": 3.66, |
|
"learning_rate": 9.44888888888889e-06, |
|
"loss": 0.0456, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 3.78, |
|
"learning_rate": 9.393333333333334e-06, |
|
"loss": 0.0472, |
|
"step": 775 |
|
}, |
|
{ |
|
"epoch": 3.9, |
|
"learning_rate": 9.33777777777778e-06, |
|
"loss": 0.0441, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 4.02, |
|
"learning_rate": 9.282222222222222e-06, |
|
"loss": 0.0403, |
|
"step": 825 |
|
}, |
|
{ |
|
"epoch": 4.15, |
|
"learning_rate": 9.226666666666668e-06, |
|
"loss": 0.0237, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 4.27, |
|
"learning_rate": 9.171111111111112e-06, |
|
"loss": 0.0269, |
|
"step": 875 |
|
}, |
|
{ |
|
"epoch": 4.39, |
|
"learning_rate": 9.115555555555556e-06, |
|
"loss": 0.0269, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 4.51, |
|
"learning_rate": 9.060000000000001e-06, |
|
"loss": 0.0245, |
|
"step": 925 |
|
}, |
|
{ |
|
"epoch": 4.63, |
|
"learning_rate": 9.004444444444445e-06, |
|
"loss": 0.0263, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 4.76, |
|
"learning_rate": 8.94888888888889e-06, |
|
"loss": 0.0283, |
|
"step": 975 |
|
}, |
|
{ |
|
"epoch": 4.88, |
|
"learning_rate": 8.893333333333333e-06, |
|
"loss": 0.0234, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 4.88, |
|
"eval_loss": 0.27615514397621155, |
|
"eval_runtime": 645.3738, |
|
"eval_samples_per_second": 4.484, |
|
"eval_steps_per_second": 0.141, |
|
"eval_wer": 0.2749513248116482, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"learning_rate": 8.83777777777778e-06, |
|
"loss": 0.0245, |
|
"step": 1025 |
|
}, |
|
{ |
|
"epoch": 5.12, |
|
"learning_rate": 8.782222222222223e-06, |
|
"loss": 0.014, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 5.24, |
|
"learning_rate": 8.726666666666667e-06, |
|
"loss": 0.0152, |
|
"step": 1075 |
|
}, |
|
{ |
|
"epoch": 5.37, |
|
"learning_rate": 8.671111111111113e-06, |
|
"loss": 0.0154, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 5.49, |
|
"learning_rate": 8.615555555555555e-06, |
|
"loss": 0.0169, |
|
"step": 1125 |
|
}, |
|
{ |
|
"epoch": 5.61, |
|
"learning_rate": 8.560000000000001e-06, |
|
"loss": 0.0177, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 5.73, |
|
"learning_rate": 8.504444444444445e-06, |
|
"loss": 0.0157, |
|
"step": 1175 |
|
}, |
|
{ |
|
"epoch": 5.85, |
|
"learning_rate": 8.448888888888889e-06, |
|
"loss": 0.0165, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 5.98, |
|
"learning_rate": 8.393333333333335e-06, |
|
"loss": 0.0166, |
|
"step": 1225 |
|
}, |
|
{ |
|
"epoch": 6.1, |
|
"learning_rate": 8.337777777777777e-06, |
|
"loss": 0.0119, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 6.22, |
|
"learning_rate": 8.282222222222223e-06, |
|
"loss": 0.0094, |
|
"step": 1275 |
|
}, |
|
{ |
|
"epoch": 6.34, |
|
"learning_rate": 8.226666666666667e-06, |
|
"loss": 0.0087, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 6.46, |
|
"learning_rate": 8.171111111111113e-06, |
|
"loss": 0.0117, |
|
"step": 1325 |
|
}, |
|
{ |
|
"epoch": 6.59, |
|
"learning_rate": 8.115555555555557e-06, |
|
"loss": 0.0109, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 6.71, |
|
"learning_rate": 8.06e-06, |
|
"loss": 0.0124, |
|
"step": 1375 |
|
}, |
|
{ |
|
"epoch": 6.83, |
|
"learning_rate": 8.004444444444445e-06, |
|
"loss": 0.0092, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 6.95, |
|
"learning_rate": 7.948888888888889e-06, |
|
"loss": 0.0107, |
|
"step": 1425 |
|
}, |
|
{ |
|
"epoch": 7.07, |
|
"learning_rate": 7.893333333333335e-06, |
|
"loss": 0.0088, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 7.2, |
|
"learning_rate": 7.837777777777779e-06, |
|
"loss": 0.0052, |
|
"step": 1475 |
|
}, |
|
{ |
|
"epoch": 7.32, |
|
"learning_rate": 7.782222222222223e-06, |
|
"loss": 0.0062, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 7.44, |
|
"learning_rate": 7.726666666666667e-06, |
|
"loss": 0.0076, |
|
"step": 1525 |
|
}, |
|
{ |
|
"epoch": 7.56, |
|
"learning_rate": 7.67111111111111e-06, |
|
"loss": 0.0054, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 7.68, |
|
"learning_rate": 7.6155555555555564e-06, |
|
"loss": 0.0059, |
|
"step": 1575 |
|
}, |
|
{ |
|
"epoch": 7.8, |
|
"learning_rate": 7.5600000000000005e-06, |
|
"loss": 0.0066, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 7.93, |
|
"learning_rate": 7.504444444444445e-06, |
|
"loss": 0.0066, |
|
"step": 1625 |
|
}, |
|
{ |
|
"epoch": 8.05, |
|
"learning_rate": 7.44888888888889e-06, |
|
"loss": 0.0081, |
|
"step": 1650 |
|
}, |
|
{ |
|
"epoch": 8.17, |
|
"learning_rate": 7.393333333333333e-06, |
|
"loss": 0.0036, |
|
"step": 1675 |
|
}, |
|
{ |
|
"epoch": 8.29, |
|
"learning_rate": 7.337777777777778e-06, |
|
"loss": 0.0039, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 8.41, |
|
"learning_rate": 7.282222222222222e-06, |
|
"loss": 0.0033, |
|
"step": 1725 |
|
}, |
|
{ |
|
"epoch": 8.54, |
|
"learning_rate": 7.226666666666667e-06, |
|
"loss": 0.0031, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 8.66, |
|
"learning_rate": 7.171111111111112e-06, |
|
"loss": 0.0035, |
|
"step": 1775 |
|
}, |
|
{ |
|
"epoch": 8.78, |
|
"learning_rate": 7.115555555555557e-06, |
|
"loss": 0.0047, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 8.9, |
|
"learning_rate": 7.06e-06, |
|
"loss": 0.0053, |
|
"step": 1825 |
|
}, |
|
{ |
|
"epoch": 9.02, |
|
"learning_rate": 7.004444444444445e-06, |
|
"loss": 0.0043, |
|
"step": 1850 |
|
}, |
|
{ |
|
"epoch": 9.15, |
|
"learning_rate": 6.948888888888889e-06, |
|
"loss": 0.002, |
|
"step": 1875 |
|
}, |
|
{ |
|
"epoch": 9.27, |
|
"learning_rate": 6.893333333333334e-06, |
|
"loss": 0.0015, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 9.39, |
|
"learning_rate": 6.837777777777779e-06, |
|
"loss": 0.0015, |
|
"step": 1925 |
|
}, |
|
{ |
|
"epoch": 9.51, |
|
"learning_rate": 6.782222222222222e-06, |
|
"loss": 0.0027, |
|
"step": 1950 |
|
}, |
|
{ |
|
"epoch": 9.63, |
|
"learning_rate": 6.726666666666667e-06, |
|
"loss": 0.0023, |
|
"step": 1975 |
|
}, |
|
{ |
|
"epoch": 9.76, |
|
"learning_rate": 6.671111111111112e-06, |
|
"loss": 0.003, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 9.76, |
|
"eval_loss": 0.3342243432998657, |
|
"eval_runtime": 576.191, |
|
"eval_samples_per_second": 5.023, |
|
"eval_steps_per_second": 0.158, |
|
"eval_wer": 0.25175653940573944, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 9.88, |
|
"learning_rate": 6.615555555555556e-06, |
|
"loss": 0.0033, |
|
"step": 2025 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"learning_rate": 6.560000000000001e-06, |
|
"loss": 0.0023, |
|
"step": 2050 |
|
}, |
|
{ |
|
"epoch": 10.12, |
|
"learning_rate": 6.504444444444446e-06, |
|
"loss": 0.0018, |
|
"step": 2075 |
|
}, |
|
{ |
|
"epoch": 10.24, |
|
"learning_rate": 6.448888888888889e-06, |
|
"loss": 0.0015, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 10.37, |
|
"learning_rate": 6.393333333333334e-06, |
|
"loss": 0.0015, |
|
"step": 2125 |
|
}, |
|
{ |
|
"epoch": 10.49, |
|
"learning_rate": 6.3377777777777786e-06, |
|
"loss": 0.0015, |
|
"step": 2150 |
|
}, |
|
{ |
|
"epoch": 10.61, |
|
"learning_rate": 6.282222222222223e-06, |
|
"loss": 0.0014, |
|
"step": 2175 |
|
}, |
|
{ |
|
"epoch": 10.73, |
|
"learning_rate": 6.2266666666666675e-06, |
|
"loss": 0.0015, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 10.85, |
|
"learning_rate": 6.171111111111112e-06, |
|
"loss": 0.0021, |
|
"step": 2225 |
|
}, |
|
{ |
|
"epoch": 10.98, |
|
"learning_rate": 6.1155555555555555e-06, |
|
"loss": 0.0026, |
|
"step": 2250 |
|
}, |
|
{ |
|
"epoch": 11.1, |
|
"learning_rate": 6.0600000000000004e-06, |
|
"loss": 0.0022, |
|
"step": 2275 |
|
}, |
|
{ |
|
"epoch": 11.22, |
|
"learning_rate": 6.004444444444445e-06, |
|
"loss": 0.0011, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 11.34, |
|
"learning_rate": 5.948888888888889e-06, |
|
"loss": 0.0016, |
|
"step": 2325 |
|
}, |
|
{ |
|
"epoch": 11.46, |
|
"learning_rate": 5.893333333333334e-06, |
|
"loss": 0.0024, |
|
"step": 2350 |
|
}, |
|
{ |
|
"epoch": 11.59, |
|
"learning_rate": 5.837777777777777e-06, |
|
"loss": 0.0015, |
|
"step": 2375 |
|
}, |
|
{ |
|
"epoch": 11.71, |
|
"learning_rate": 5.782222222222222e-06, |
|
"loss": 0.0014, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 11.83, |
|
"learning_rate": 5.726666666666667e-06, |
|
"loss": 0.0017, |
|
"step": 2425 |
|
}, |
|
{ |
|
"epoch": 11.95, |
|
"learning_rate": 5.671111111111112e-06, |
|
"loss": 0.0019, |
|
"step": 2450 |
|
}, |
|
{ |
|
"epoch": 12.07, |
|
"learning_rate": 5.615555555555556e-06, |
|
"loss": 0.0025, |
|
"step": 2475 |
|
}, |
|
{ |
|
"epoch": 12.2, |
|
"learning_rate": 5.560000000000001e-06, |
|
"loss": 0.0021, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 12.32, |
|
"learning_rate": 5.504444444444444e-06, |
|
"loss": 0.0008, |
|
"step": 2525 |
|
}, |
|
{ |
|
"epoch": 12.44, |
|
"learning_rate": 5.448888888888889e-06, |
|
"loss": 0.0015, |
|
"step": 2550 |
|
}, |
|
{ |
|
"epoch": 12.56, |
|
"learning_rate": 5.393333333333334e-06, |
|
"loss": 0.0007, |
|
"step": 2575 |
|
}, |
|
{ |
|
"epoch": 12.68, |
|
"learning_rate": 5.337777777777779e-06, |
|
"loss": 0.0007, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 12.8, |
|
"learning_rate": 5.282222222222223e-06, |
|
"loss": 0.0007, |
|
"step": 2625 |
|
}, |
|
{ |
|
"epoch": 12.93, |
|
"learning_rate": 5.226666666666667e-06, |
|
"loss": 0.0008, |
|
"step": 2650 |
|
}, |
|
{ |
|
"epoch": 13.05, |
|
"learning_rate": 5.171111111111111e-06, |
|
"loss": 0.0004, |
|
"step": 2675 |
|
}, |
|
{ |
|
"epoch": 13.17, |
|
"learning_rate": 5.115555555555556e-06, |
|
"loss": 0.0003, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 13.29, |
|
"learning_rate": 5.060000000000001e-06, |
|
"loss": 0.0005, |
|
"step": 2725 |
|
}, |
|
{ |
|
"epoch": 13.41, |
|
"learning_rate": 5.004444444444445e-06, |
|
"loss": 0.0004, |
|
"step": 2750 |
|
}, |
|
{ |
|
"epoch": 13.54, |
|
"learning_rate": 4.94888888888889e-06, |
|
"loss": 0.0005, |
|
"step": 2775 |
|
}, |
|
{ |
|
"epoch": 13.66, |
|
"learning_rate": 4.893333333333334e-06, |
|
"loss": 0.0004, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 13.78, |
|
"learning_rate": 4.837777777777778e-06, |
|
"loss": 0.0004, |
|
"step": 2825 |
|
}, |
|
{ |
|
"epoch": 13.9, |
|
"learning_rate": 4.7822222222222226e-06, |
|
"loss": 0.0002, |
|
"step": 2850 |
|
}, |
|
{ |
|
"epoch": 14.02, |
|
"learning_rate": 4.7266666666666674e-06, |
|
"loss": 0.0001, |
|
"step": 2875 |
|
}, |
|
{ |
|
"epoch": 14.15, |
|
"learning_rate": 4.6711111111111115e-06, |
|
"loss": 0.0002, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 14.27, |
|
"learning_rate": 4.6155555555555555e-06, |
|
"loss": 0.0002, |
|
"step": 2925 |
|
}, |
|
{ |
|
"epoch": 14.39, |
|
"learning_rate": 4.56e-06, |
|
"loss": 0.0001, |
|
"step": 2950 |
|
}, |
|
{ |
|
"epoch": 14.51, |
|
"learning_rate": 4.504444444444444e-06, |
|
"loss": 0.0001, |
|
"step": 2975 |
|
}, |
|
{ |
|
"epoch": 14.63, |
|
"learning_rate": 4.448888888888889e-06, |
|
"loss": 0.0001, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 14.63, |
|
"eval_loss": 0.3636488616466522, |
|
"eval_runtime": 584.234, |
|
"eval_samples_per_second": 4.953, |
|
"eval_steps_per_second": 0.156, |
|
"eval_wer": 0.24240243799204267, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 14.76, |
|
"learning_rate": 4.393333333333334e-06, |
|
"loss": 0.0001, |
|
"step": 3025 |
|
}, |
|
{ |
|
"epoch": 14.88, |
|
"learning_rate": 4.337777777777778e-06, |
|
"loss": 0.0001, |
|
"step": 3050 |
|
}, |
|
{ |
|
"epoch": 15.0, |
|
"learning_rate": 4.282222222222222e-06, |
|
"loss": 0.0001, |
|
"step": 3075 |
|
}, |
|
{ |
|
"epoch": 15.12, |
|
"learning_rate": 4.226666666666667e-06, |
|
"loss": 0.0001, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 15.24, |
|
"learning_rate": 4.171111111111111e-06, |
|
"loss": 0.0001, |
|
"step": 3125 |
|
}, |
|
{ |
|
"epoch": 15.37, |
|
"learning_rate": 4.115555555555556e-06, |
|
"loss": 0.0002, |
|
"step": 3150 |
|
}, |
|
{ |
|
"epoch": 15.49, |
|
"learning_rate": 4.060000000000001e-06, |
|
"loss": 0.0002, |
|
"step": 3175 |
|
}, |
|
{ |
|
"epoch": 15.61, |
|
"learning_rate": 4.004444444444445e-06, |
|
"loss": 0.0002, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 15.73, |
|
"learning_rate": 3.948888888888889e-06, |
|
"loss": 0.0001, |
|
"step": 3225 |
|
}, |
|
{ |
|
"epoch": 15.85, |
|
"learning_rate": 3.893333333333333e-06, |
|
"loss": 0.0001, |
|
"step": 3250 |
|
}, |
|
{ |
|
"epoch": 15.98, |
|
"learning_rate": 3.837777777777778e-06, |
|
"loss": 0.0001, |
|
"step": 3275 |
|
}, |
|
{ |
|
"epoch": 16.1, |
|
"learning_rate": 3.782222222222223e-06, |
|
"loss": 0.0001, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 16.22, |
|
"learning_rate": 3.726666666666667e-06, |
|
"loss": 0.0001, |
|
"step": 3325 |
|
}, |
|
{ |
|
"epoch": 16.34, |
|
"learning_rate": 3.6711111111111113e-06, |
|
"loss": 0.0001, |
|
"step": 3350 |
|
}, |
|
{ |
|
"epoch": 16.46, |
|
"learning_rate": 3.615555555555556e-06, |
|
"loss": 0.0001, |
|
"step": 3375 |
|
}, |
|
{ |
|
"epoch": 16.59, |
|
"learning_rate": 3.5600000000000002e-06, |
|
"loss": 0.0001, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 16.71, |
|
"learning_rate": 3.5044444444444447e-06, |
|
"loss": 0.0001, |
|
"step": 3425 |
|
}, |
|
{ |
|
"epoch": 16.83, |
|
"learning_rate": 3.4488888888888896e-06, |
|
"loss": 0.0001, |
|
"step": 3450 |
|
}, |
|
{ |
|
"epoch": 16.95, |
|
"learning_rate": 3.3933333333333336e-06, |
|
"loss": 0.0001, |
|
"step": 3475 |
|
}, |
|
{ |
|
"epoch": 17.07, |
|
"learning_rate": 3.337777777777778e-06, |
|
"loss": 0.0001, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 17.2, |
|
"learning_rate": 3.282222222222223e-06, |
|
"loss": 0.0001, |
|
"step": 3525 |
|
}, |
|
{ |
|
"epoch": 17.32, |
|
"learning_rate": 3.226666666666667e-06, |
|
"loss": 0.0001, |
|
"step": 3550 |
|
}, |
|
{ |
|
"epoch": 17.44, |
|
"learning_rate": 3.1711111111111114e-06, |
|
"loss": 0.0001, |
|
"step": 3575 |
|
}, |
|
{ |
|
"epoch": 17.56, |
|
"learning_rate": 3.1155555555555555e-06, |
|
"loss": 0.0001, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 17.68, |
|
"learning_rate": 3.0600000000000003e-06, |
|
"loss": 0.0001, |
|
"step": 3625 |
|
}, |
|
{ |
|
"epoch": 17.8, |
|
"learning_rate": 3.004444444444445e-06, |
|
"loss": 0.0001, |
|
"step": 3650 |
|
}, |
|
{ |
|
"epoch": 17.93, |
|
"learning_rate": 2.948888888888889e-06, |
|
"loss": 0.0, |
|
"step": 3675 |
|
}, |
|
{ |
|
"epoch": 18.05, |
|
"learning_rate": 2.8933333333333337e-06, |
|
"loss": 0.0001, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 18.17, |
|
"learning_rate": 2.837777777777778e-06, |
|
"loss": 0.0, |
|
"step": 3725 |
|
}, |
|
{ |
|
"epoch": 18.29, |
|
"learning_rate": 2.7822222222222222e-06, |
|
"loss": 0.0, |
|
"step": 3750 |
|
}, |
|
{ |
|
"epoch": 18.41, |
|
"learning_rate": 2.726666666666667e-06, |
|
"loss": 0.0, |
|
"step": 3775 |
|
}, |
|
{ |
|
"epoch": 18.54, |
|
"learning_rate": 2.6711111111111116e-06, |
|
"loss": 0.0, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 18.66, |
|
"learning_rate": 2.6155555555555556e-06, |
|
"loss": 0.0, |
|
"step": 3825 |
|
}, |
|
{ |
|
"epoch": 18.78, |
|
"learning_rate": 2.56e-06, |
|
"loss": 0.0, |
|
"step": 3850 |
|
}, |
|
{ |
|
"epoch": 18.9, |
|
"learning_rate": 2.504444444444445e-06, |
|
"loss": 0.0, |
|
"step": 3875 |
|
}, |
|
{ |
|
"epoch": 19.02, |
|
"learning_rate": 2.448888888888889e-06, |
|
"loss": 0.0, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 19.15, |
|
"learning_rate": 2.3933333333333334e-06, |
|
"loss": 0.0, |
|
"step": 3925 |
|
}, |
|
{ |
|
"epoch": 19.27, |
|
"learning_rate": 2.337777777777778e-06, |
|
"loss": 0.0, |
|
"step": 3950 |
|
}, |
|
{ |
|
"epoch": 19.39, |
|
"learning_rate": 2.2822222222222223e-06, |
|
"loss": 0.0, |
|
"step": 3975 |
|
}, |
|
{ |
|
"epoch": 19.51, |
|
"learning_rate": 2.226666666666667e-06, |
|
"loss": 0.0, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 19.51, |
|
"eval_loss": 0.3877107799053192, |
|
"eval_runtime": 605.8718, |
|
"eval_samples_per_second": 4.777, |
|
"eval_steps_per_second": 0.15, |
|
"eval_wer": 0.24261406924574622, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 19.63, |
|
"learning_rate": 2.1711111111111113e-06, |
|
"loss": 0.0, |
|
"step": 4025 |
|
}, |
|
{ |
|
"epoch": 19.76, |
|
"learning_rate": 2.1155555555555557e-06, |
|
"loss": 0.0, |
|
"step": 4050 |
|
}, |
|
{ |
|
"epoch": 19.88, |
|
"learning_rate": 2.06e-06, |
|
"loss": 0.0, |
|
"step": 4075 |
|
}, |
|
{ |
|
"epoch": 20.0, |
|
"learning_rate": 2.0044444444444446e-06, |
|
"loss": 0.0, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 20.12, |
|
"learning_rate": 1.948888888888889e-06, |
|
"loss": 0.0, |
|
"step": 4125 |
|
}, |
|
{ |
|
"epoch": 20.24, |
|
"learning_rate": 1.8933333333333333e-06, |
|
"loss": 0.0, |
|
"step": 4150 |
|
}, |
|
{ |
|
"epoch": 20.37, |
|
"learning_rate": 1.837777777777778e-06, |
|
"loss": 0.0, |
|
"step": 4175 |
|
}, |
|
{ |
|
"epoch": 20.49, |
|
"learning_rate": 1.7822222222222225e-06, |
|
"loss": 0.0, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 20.61, |
|
"learning_rate": 1.7266666666666667e-06, |
|
"loss": 0.0, |
|
"step": 4225 |
|
}, |
|
{ |
|
"epoch": 20.73, |
|
"learning_rate": 1.6711111111111112e-06, |
|
"loss": 0.0, |
|
"step": 4250 |
|
}, |
|
{ |
|
"epoch": 20.85, |
|
"learning_rate": 1.6155555555555559e-06, |
|
"loss": 0.0, |
|
"step": 4275 |
|
}, |
|
{ |
|
"epoch": 20.98, |
|
"learning_rate": 1.56e-06, |
|
"loss": 0.0, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 21.1, |
|
"learning_rate": 1.5044444444444446e-06, |
|
"loss": 0.0, |
|
"step": 4325 |
|
}, |
|
{ |
|
"epoch": 21.22, |
|
"learning_rate": 1.4488888888888892e-06, |
|
"loss": 0.0, |
|
"step": 4350 |
|
}, |
|
{ |
|
"epoch": 21.34, |
|
"learning_rate": 1.3933333333333335e-06, |
|
"loss": 0.0, |
|
"step": 4375 |
|
}, |
|
{ |
|
"epoch": 21.46, |
|
"learning_rate": 1.337777777777778e-06, |
|
"loss": 0.0, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 21.59, |
|
"learning_rate": 1.2822222222222222e-06, |
|
"loss": 0.0, |
|
"step": 4425 |
|
}, |
|
{ |
|
"epoch": 21.71, |
|
"learning_rate": 1.2266666666666666e-06, |
|
"loss": 0.0, |
|
"step": 4450 |
|
}, |
|
{ |
|
"epoch": 21.83, |
|
"learning_rate": 1.171111111111111e-06, |
|
"loss": 0.0, |
|
"step": 4475 |
|
}, |
|
{ |
|
"epoch": 21.95, |
|
"learning_rate": 1.1155555555555558e-06, |
|
"loss": 0.0, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 22.07, |
|
"learning_rate": 1.06e-06, |
|
"loss": 0.0, |
|
"step": 4525 |
|
}, |
|
{ |
|
"epoch": 22.2, |
|
"learning_rate": 1.0044444444444445e-06, |
|
"loss": 0.0, |
|
"step": 4550 |
|
}, |
|
{ |
|
"epoch": 22.32, |
|
"learning_rate": 9.488888888888889e-07, |
|
"loss": 0.0, |
|
"step": 4575 |
|
}, |
|
{ |
|
"epoch": 22.44, |
|
"learning_rate": 8.933333333333334e-07, |
|
"loss": 0.0, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 22.56, |
|
"learning_rate": 8.37777777777778e-07, |
|
"loss": 0.0, |
|
"step": 4625 |
|
}, |
|
{ |
|
"epoch": 22.68, |
|
"learning_rate": 7.822222222222223e-07, |
|
"loss": 0.0, |
|
"step": 4650 |
|
}, |
|
{ |
|
"epoch": 22.8, |
|
"learning_rate": 7.266666666666668e-07, |
|
"loss": 0.0, |
|
"step": 4675 |
|
}, |
|
{ |
|
"epoch": 22.93, |
|
"learning_rate": 6.711111111111111e-07, |
|
"loss": 0.0, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 23.05, |
|
"learning_rate": 6.155555555555556e-07, |
|
"loss": 0.0, |
|
"step": 4725 |
|
}, |
|
{ |
|
"epoch": 23.17, |
|
"learning_rate": 5.6e-07, |
|
"loss": 0.0, |
|
"step": 4750 |
|
}, |
|
{ |
|
"epoch": 23.29, |
|
"learning_rate": 5.044444444444445e-07, |
|
"loss": 0.0, |
|
"step": 4775 |
|
}, |
|
{ |
|
"epoch": 23.41, |
|
"learning_rate": 4.488888888888889e-07, |
|
"loss": 0.0, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 23.54, |
|
"learning_rate": 3.9333333333333336e-07, |
|
"loss": 0.0, |
|
"step": 4825 |
|
}, |
|
{ |
|
"epoch": 23.66, |
|
"learning_rate": 3.3777777777777777e-07, |
|
"loss": 0.0, |
|
"step": 4850 |
|
}, |
|
{ |
|
"epoch": 23.78, |
|
"learning_rate": 2.822222222222222e-07, |
|
"loss": 0.0, |
|
"step": 4875 |
|
}, |
|
{ |
|
"epoch": 23.9, |
|
"learning_rate": 2.266666666666667e-07, |
|
"loss": 0.0, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 24.02, |
|
"learning_rate": 1.7111111111111114e-07, |
|
"loss": 0.0, |
|
"step": 4925 |
|
}, |
|
{ |
|
"epoch": 24.15, |
|
"learning_rate": 1.1555555555555556e-07, |
|
"loss": 0.0, |
|
"step": 4950 |
|
}, |
|
{ |
|
"epoch": 24.27, |
|
"learning_rate": 6.000000000000001e-08, |
|
"loss": 0.0, |
|
"step": 4975 |
|
}, |
|
{ |
|
"epoch": 24.39, |
|
"learning_rate": 4.444444444444445e-09, |
|
"loss": 0.0, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 24.39, |
|
"eval_loss": 0.3957468569278717, |
|
"eval_runtime": 683.714, |
|
"eval_samples_per_second": 4.233, |
|
"eval_steps_per_second": 0.133, |
|
"eval_wer": 0.24248709049352407, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 24.39, |
|
"step": 5000, |
|
"total_flos": 1.6280689674316415e+20, |
|
"train_loss": 0.027480003798776305, |
|
"train_runtime": 34274.9802, |
|
"train_samples_per_second": 4.668, |
|
"train_steps_per_second": 0.146 |
|
} |
|
], |
|
"max_steps": 5000, |
|
"num_train_epochs": 25, |
|
"total_flos": 1.6280689674316415e+20, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|