|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 24.449877750611247, |
|
"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": 4.800000000000001e-07, |
|
"loss": 0.885, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.600000000000001e-07, |
|
"loss": 0.7798, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.46e-06, |
|
"loss": 0.6352, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.9600000000000003e-06, |
|
"loss": 0.442, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 2.46e-06, |
|
"loss": 0.4772, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 2.96e-06, |
|
"loss": 0.3291, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 3.46e-06, |
|
"loss": 0.3961, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 3.96e-06, |
|
"loss": 0.2712, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 4.4600000000000005e-06, |
|
"loss": 0.3267, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 4.960000000000001e-06, |
|
"loss": 0.2265, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 5.460000000000001e-06, |
|
"loss": 0.2653, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 5.9600000000000005e-06, |
|
"loss": 0.2103, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"learning_rate": 6.460000000000001e-06, |
|
"loss": 0.2456, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"learning_rate": 6.96e-06, |
|
"loss": 0.1962, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 7.4600000000000006e-06, |
|
"loss": 0.2306, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 1.96, |
|
"learning_rate": 7.960000000000002e-06, |
|
"loss": 0.1861, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 8.46e-06, |
|
"loss": 0.1837, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 8.96e-06, |
|
"loss": 0.1389, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 2.32, |
|
"learning_rate": 9.460000000000001e-06, |
|
"loss": 0.1488, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 2.44, |
|
"learning_rate": 9.960000000000001e-06, |
|
"loss": 0.1302, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 2.57, |
|
"learning_rate": 9.94888888888889e-06, |
|
"loss": 0.1473, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 2.69, |
|
"learning_rate": 9.893333333333334e-06, |
|
"loss": 0.13, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 2.81, |
|
"learning_rate": 9.837777777777778e-06, |
|
"loss": 0.1408, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 2.93, |
|
"learning_rate": 9.782222222222222e-06, |
|
"loss": 0.1286, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 3.06, |
|
"learning_rate": 9.726666666666668e-06, |
|
"loss": 0.102, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 3.18, |
|
"learning_rate": 9.671111111111112e-06, |
|
"loss": 0.0804, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 3.3, |
|
"learning_rate": 9.615555555555558e-06, |
|
"loss": 0.0733, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 3.42, |
|
"learning_rate": 9.56e-06, |
|
"loss": 0.0798, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 3.55, |
|
"learning_rate": 9.504444444444446e-06, |
|
"loss": 0.0802, |
|
"step": 725 |
|
}, |
|
{ |
|
"epoch": 3.67, |
|
"learning_rate": 9.44888888888889e-06, |
|
"loss": 0.082, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 3.79, |
|
"learning_rate": 9.393333333333334e-06, |
|
"loss": 0.0776, |
|
"step": 775 |
|
}, |
|
{ |
|
"epoch": 3.91, |
|
"learning_rate": 9.33777777777778e-06, |
|
"loss": 0.0831, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 4.03, |
|
"learning_rate": 9.282222222222222e-06, |
|
"loss": 0.0577, |
|
"step": 825 |
|
}, |
|
{ |
|
"epoch": 4.16, |
|
"learning_rate": 9.226666666666668e-06, |
|
"loss": 0.0493, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 4.28, |
|
"learning_rate": 9.171111111111112e-06, |
|
"loss": 0.0421, |
|
"step": 875 |
|
}, |
|
{ |
|
"epoch": 4.4, |
|
"learning_rate": 9.115555555555556e-06, |
|
"loss": 0.0532, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 4.52, |
|
"learning_rate": 9.060000000000001e-06, |
|
"loss": 0.0416, |
|
"step": 925 |
|
}, |
|
{ |
|
"epoch": 4.65, |
|
"learning_rate": 9.004444444444445e-06, |
|
"loss": 0.049, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 4.77, |
|
"learning_rate": 8.94888888888889e-06, |
|
"loss": 0.042, |
|
"step": 975 |
|
}, |
|
{ |
|
"epoch": 4.89, |
|
"learning_rate": 8.893333333333333e-06, |
|
"loss": 0.0534, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 4.89, |
|
"eval_loss": 0.33749517798423767, |
|
"eval_runtime": 1086.4729, |
|
"eval_samples_per_second": 2.664, |
|
"eval_steps_per_second": 0.167, |
|
"eval_wer": 0.34648268856344705, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 5.01, |
|
"learning_rate": 8.83777777777778e-06, |
|
"loss": 0.0303, |
|
"step": 1025 |
|
}, |
|
{ |
|
"epoch": 5.13, |
|
"learning_rate": 8.782222222222223e-06, |
|
"loss": 0.0324, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 5.26, |
|
"learning_rate": 8.726666666666667e-06, |
|
"loss": 0.021, |
|
"step": 1075 |
|
}, |
|
{ |
|
"epoch": 5.38, |
|
"learning_rate": 8.671111111111113e-06, |
|
"loss": 0.0343, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 5.5, |
|
"learning_rate": 8.615555555555555e-06, |
|
"loss": 0.0211, |
|
"step": 1125 |
|
}, |
|
{ |
|
"epoch": 5.62, |
|
"learning_rate": 8.560000000000001e-06, |
|
"loss": 0.039, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 5.75, |
|
"learning_rate": 8.504444444444445e-06, |
|
"loss": 0.0205, |
|
"step": 1175 |
|
}, |
|
{ |
|
"epoch": 5.87, |
|
"learning_rate": 8.448888888888889e-06, |
|
"loss": 0.0338, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 5.99, |
|
"learning_rate": 8.393333333333335e-06, |
|
"loss": 0.0155, |
|
"step": 1225 |
|
}, |
|
{ |
|
"epoch": 6.11, |
|
"learning_rate": 8.337777777777777e-06, |
|
"loss": 0.0222, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 6.23, |
|
"learning_rate": 8.282222222222223e-06, |
|
"loss": 0.01, |
|
"step": 1275 |
|
}, |
|
{ |
|
"epoch": 6.36, |
|
"learning_rate": 8.226666666666667e-06, |
|
"loss": 0.0241, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 6.48, |
|
"learning_rate": 8.171111111111113e-06, |
|
"loss": 0.009, |
|
"step": 1325 |
|
}, |
|
{ |
|
"epoch": 6.6, |
|
"learning_rate": 8.115555555555557e-06, |
|
"loss": 0.0224, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 6.72, |
|
"learning_rate": 8.06e-06, |
|
"loss": 0.009, |
|
"step": 1375 |
|
}, |
|
{ |
|
"epoch": 6.85, |
|
"learning_rate": 8.004444444444445e-06, |
|
"loss": 0.022, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 6.97, |
|
"learning_rate": 7.948888888888889e-06, |
|
"loss": 0.0095, |
|
"step": 1425 |
|
}, |
|
{ |
|
"epoch": 7.09, |
|
"learning_rate": 7.893333333333335e-06, |
|
"loss": 0.0136, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 7.21, |
|
"learning_rate": 7.837777777777779e-06, |
|
"loss": 0.0062, |
|
"step": 1475 |
|
}, |
|
{ |
|
"epoch": 7.33, |
|
"learning_rate": 7.782222222222223e-06, |
|
"loss": 0.0124, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 7.46, |
|
"learning_rate": 7.726666666666667e-06, |
|
"loss": 0.0062, |
|
"step": 1525 |
|
}, |
|
{ |
|
"epoch": 7.58, |
|
"learning_rate": 7.67111111111111e-06, |
|
"loss": 0.0122, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 7.7, |
|
"learning_rate": 7.6155555555555564e-06, |
|
"loss": 0.0058, |
|
"step": 1575 |
|
}, |
|
{ |
|
"epoch": 7.82, |
|
"learning_rate": 7.5600000000000005e-06, |
|
"loss": 0.0128, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 7.95, |
|
"learning_rate": 7.504444444444445e-06, |
|
"loss": 0.0069, |
|
"step": 1625 |
|
}, |
|
{ |
|
"epoch": 8.07, |
|
"learning_rate": 7.44888888888889e-06, |
|
"loss": 0.0085, |
|
"step": 1650 |
|
}, |
|
{ |
|
"epoch": 8.19, |
|
"learning_rate": 7.393333333333333e-06, |
|
"loss": 0.0048, |
|
"step": 1675 |
|
}, |
|
{ |
|
"epoch": 8.31, |
|
"learning_rate": 7.337777777777778e-06, |
|
"loss": 0.007, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 8.44, |
|
"learning_rate": 7.282222222222222e-06, |
|
"loss": 0.0042, |
|
"step": 1725 |
|
}, |
|
{ |
|
"epoch": 8.56, |
|
"learning_rate": 7.226666666666667e-06, |
|
"loss": 0.007, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 8.68, |
|
"learning_rate": 7.171111111111112e-06, |
|
"loss": 0.0057, |
|
"step": 1775 |
|
}, |
|
{ |
|
"epoch": 8.8, |
|
"learning_rate": 7.115555555555557e-06, |
|
"loss": 0.0083, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 8.92, |
|
"learning_rate": 7.06e-06, |
|
"loss": 0.0061, |
|
"step": 1825 |
|
}, |
|
{ |
|
"epoch": 9.05, |
|
"learning_rate": 7.004444444444445e-06, |
|
"loss": 0.0067, |
|
"step": 1850 |
|
}, |
|
{ |
|
"epoch": 9.17, |
|
"learning_rate": 6.948888888888889e-06, |
|
"loss": 0.0042, |
|
"step": 1875 |
|
}, |
|
{ |
|
"epoch": 9.29, |
|
"learning_rate": 6.893333333333334e-06, |
|
"loss": 0.0035, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 9.41, |
|
"learning_rate": 6.837777777777779e-06, |
|
"loss": 0.0047, |
|
"step": 1925 |
|
}, |
|
{ |
|
"epoch": 9.54, |
|
"learning_rate": 6.782222222222222e-06, |
|
"loss": 0.0035, |
|
"step": 1950 |
|
}, |
|
{ |
|
"epoch": 9.66, |
|
"learning_rate": 6.726666666666667e-06, |
|
"loss": 0.0034, |
|
"step": 1975 |
|
}, |
|
{ |
|
"epoch": 9.78, |
|
"learning_rate": 6.671111111111112e-06, |
|
"loss": 0.0042, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 9.78, |
|
"eval_loss": 0.44431325793266296, |
|
"eval_runtime": 1083.1072, |
|
"eval_samples_per_second": 2.672, |
|
"eval_steps_per_second": 0.167, |
|
"eval_wer": 0.34021840345382204, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 9.9, |
|
"learning_rate": 6.615555555555556e-06, |
|
"loss": 0.0036, |
|
"step": 2025 |
|
}, |
|
{ |
|
"epoch": 10.02, |
|
"learning_rate": 6.560000000000001e-06, |
|
"loss": 0.0033, |
|
"step": 2050 |
|
}, |
|
{ |
|
"epoch": 10.15, |
|
"learning_rate": 6.504444444444446e-06, |
|
"loss": 0.003, |
|
"step": 2075 |
|
}, |
|
{ |
|
"epoch": 10.27, |
|
"learning_rate": 6.448888888888889e-06, |
|
"loss": 0.003, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 10.39, |
|
"learning_rate": 6.393333333333334e-06, |
|
"loss": 0.0034, |
|
"step": 2125 |
|
}, |
|
{ |
|
"epoch": 10.51, |
|
"learning_rate": 6.3377777777777786e-06, |
|
"loss": 0.0023, |
|
"step": 2150 |
|
}, |
|
{ |
|
"epoch": 10.64, |
|
"learning_rate": 6.282222222222223e-06, |
|
"loss": 0.0026, |
|
"step": 2175 |
|
}, |
|
{ |
|
"epoch": 10.76, |
|
"learning_rate": 6.2266666666666675e-06, |
|
"loss": 0.0029, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 10.88, |
|
"learning_rate": 6.171111111111112e-06, |
|
"loss": 0.0037, |
|
"step": 2225 |
|
}, |
|
{ |
|
"epoch": 11.0, |
|
"learning_rate": 6.1155555555555555e-06, |
|
"loss": 0.0014, |
|
"step": 2250 |
|
}, |
|
{ |
|
"epoch": 11.12, |
|
"learning_rate": 6.0600000000000004e-06, |
|
"loss": 0.002, |
|
"step": 2275 |
|
}, |
|
{ |
|
"epoch": 11.25, |
|
"learning_rate": 6.004444444444445e-06, |
|
"loss": 0.0006, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 11.37, |
|
"learning_rate": 5.948888888888889e-06, |
|
"loss": 0.0021, |
|
"step": 2325 |
|
}, |
|
{ |
|
"epoch": 11.49, |
|
"learning_rate": 5.893333333333334e-06, |
|
"loss": 0.0014, |
|
"step": 2350 |
|
}, |
|
{ |
|
"epoch": 11.61, |
|
"learning_rate": 5.837777777777777e-06, |
|
"loss": 0.0017, |
|
"step": 2375 |
|
}, |
|
{ |
|
"epoch": 11.74, |
|
"learning_rate": 5.782222222222222e-06, |
|
"loss": 0.0014, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 11.86, |
|
"learning_rate": 5.726666666666667e-06, |
|
"loss": 0.0019, |
|
"step": 2425 |
|
}, |
|
{ |
|
"epoch": 11.98, |
|
"learning_rate": 5.671111111111112e-06, |
|
"loss": 0.0022, |
|
"step": 2450 |
|
}, |
|
{ |
|
"epoch": 12.1, |
|
"learning_rate": 5.615555555555556e-06, |
|
"loss": 0.0012, |
|
"step": 2475 |
|
}, |
|
{ |
|
"epoch": 12.22, |
|
"learning_rate": 5.560000000000001e-06, |
|
"loss": 0.001, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 12.35, |
|
"learning_rate": 5.504444444444444e-06, |
|
"loss": 0.0012, |
|
"step": 2525 |
|
}, |
|
{ |
|
"epoch": 12.47, |
|
"learning_rate": 5.448888888888889e-06, |
|
"loss": 0.0004, |
|
"step": 2550 |
|
}, |
|
{ |
|
"epoch": 12.59, |
|
"learning_rate": 5.393333333333334e-06, |
|
"loss": 0.0008, |
|
"step": 2575 |
|
}, |
|
{ |
|
"epoch": 12.71, |
|
"learning_rate": 5.337777777777779e-06, |
|
"loss": 0.0003, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 12.84, |
|
"learning_rate": 5.282222222222223e-06, |
|
"loss": 0.0007, |
|
"step": 2625 |
|
}, |
|
{ |
|
"epoch": 12.96, |
|
"learning_rate": 5.226666666666667e-06, |
|
"loss": 0.0005, |
|
"step": 2650 |
|
}, |
|
{ |
|
"epoch": 13.08, |
|
"learning_rate": 5.171111111111111e-06, |
|
"loss": 0.0007, |
|
"step": 2675 |
|
}, |
|
{ |
|
"epoch": 13.2, |
|
"learning_rate": 5.115555555555556e-06, |
|
"loss": 0.0003, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 13.33, |
|
"learning_rate": 5.060000000000001e-06, |
|
"loss": 0.0003, |
|
"step": 2725 |
|
}, |
|
{ |
|
"epoch": 13.45, |
|
"learning_rate": 5.004444444444445e-06, |
|
"loss": 0.0002, |
|
"step": 2750 |
|
}, |
|
{ |
|
"epoch": 13.57, |
|
"learning_rate": 4.94888888888889e-06, |
|
"loss": 0.0003, |
|
"step": 2775 |
|
}, |
|
{ |
|
"epoch": 13.69, |
|
"learning_rate": 4.893333333333334e-06, |
|
"loss": 0.0002, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 13.81, |
|
"learning_rate": 4.837777777777778e-06, |
|
"loss": 0.0003, |
|
"step": 2825 |
|
}, |
|
{ |
|
"epoch": 13.94, |
|
"learning_rate": 4.7822222222222226e-06, |
|
"loss": 0.0002, |
|
"step": 2850 |
|
}, |
|
{ |
|
"epoch": 14.06, |
|
"learning_rate": 4.7266666666666674e-06, |
|
"loss": 0.0008, |
|
"step": 2875 |
|
}, |
|
{ |
|
"epoch": 14.18, |
|
"learning_rate": 4.6711111111111115e-06, |
|
"loss": 0.0002, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 14.3, |
|
"learning_rate": 4.6155555555555555e-06, |
|
"loss": 0.0002, |
|
"step": 2925 |
|
}, |
|
{ |
|
"epoch": 14.43, |
|
"learning_rate": 4.56e-06, |
|
"loss": 0.0002, |
|
"step": 2950 |
|
}, |
|
{ |
|
"epoch": 14.55, |
|
"learning_rate": 4.504444444444444e-06, |
|
"loss": 0.0002, |
|
"step": 2975 |
|
}, |
|
{ |
|
"epoch": 14.67, |
|
"learning_rate": 4.448888888888889e-06, |
|
"loss": 0.0002, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 14.67, |
|
"eval_loss": 0.49726060032844543, |
|
"eval_runtime": 1090.3238, |
|
"eval_samples_per_second": 2.654, |
|
"eval_steps_per_second": 0.166, |
|
"eval_wer": 0.3300601032760518, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 14.79, |
|
"learning_rate": 4.393333333333334e-06, |
|
"loss": 0.0002, |
|
"step": 3025 |
|
}, |
|
{ |
|
"epoch": 14.91, |
|
"learning_rate": 4.337777777777778e-06, |
|
"loss": 0.0002, |
|
"step": 3050 |
|
}, |
|
{ |
|
"epoch": 15.04, |
|
"learning_rate": 4.282222222222222e-06, |
|
"loss": 0.0002, |
|
"step": 3075 |
|
}, |
|
{ |
|
"epoch": 15.16, |
|
"learning_rate": 4.226666666666667e-06, |
|
"loss": 0.0002, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 15.28, |
|
"learning_rate": 4.171111111111111e-06, |
|
"loss": 0.0001, |
|
"step": 3125 |
|
}, |
|
{ |
|
"epoch": 15.4, |
|
"learning_rate": 4.115555555555556e-06, |
|
"loss": 0.0002, |
|
"step": 3150 |
|
}, |
|
{ |
|
"epoch": 15.53, |
|
"learning_rate": 4.060000000000001e-06, |
|
"loss": 0.0001, |
|
"step": 3175 |
|
}, |
|
{ |
|
"epoch": 15.65, |
|
"learning_rate": 4.004444444444445e-06, |
|
"loss": 0.0002, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 15.77, |
|
"learning_rate": 3.948888888888889e-06, |
|
"loss": 0.0001, |
|
"step": 3225 |
|
}, |
|
{ |
|
"epoch": 15.89, |
|
"learning_rate": 3.893333333333333e-06, |
|
"loss": 0.0002, |
|
"step": 3250 |
|
}, |
|
{ |
|
"epoch": 16.01, |
|
"learning_rate": 3.837777777777778e-06, |
|
"loss": 0.0001, |
|
"step": 3275 |
|
}, |
|
{ |
|
"epoch": 16.14, |
|
"learning_rate": 3.782222222222223e-06, |
|
"loss": 0.0002, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 16.26, |
|
"learning_rate": 3.726666666666667e-06, |
|
"loss": 0.0001, |
|
"step": 3325 |
|
}, |
|
{ |
|
"epoch": 16.38, |
|
"learning_rate": 3.6711111111111113e-06, |
|
"loss": 0.0002, |
|
"step": 3350 |
|
}, |
|
{ |
|
"epoch": 16.5, |
|
"learning_rate": 3.615555555555556e-06, |
|
"loss": 0.0001, |
|
"step": 3375 |
|
}, |
|
{ |
|
"epoch": 16.63, |
|
"learning_rate": 3.5600000000000002e-06, |
|
"loss": 0.0002, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 16.75, |
|
"learning_rate": 3.5044444444444447e-06, |
|
"loss": 0.0001, |
|
"step": 3425 |
|
}, |
|
{ |
|
"epoch": 16.87, |
|
"learning_rate": 3.4488888888888896e-06, |
|
"loss": 0.0002, |
|
"step": 3450 |
|
}, |
|
{ |
|
"epoch": 16.99, |
|
"learning_rate": 3.3933333333333336e-06, |
|
"loss": 0.0001, |
|
"step": 3475 |
|
}, |
|
{ |
|
"epoch": 17.11, |
|
"learning_rate": 3.337777777777778e-06, |
|
"loss": 0.0002, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 17.24, |
|
"learning_rate": 3.282222222222223e-06, |
|
"loss": 0.0001, |
|
"step": 3525 |
|
}, |
|
{ |
|
"epoch": 17.36, |
|
"learning_rate": 3.226666666666667e-06, |
|
"loss": 0.0002, |
|
"step": 3550 |
|
}, |
|
{ |
|
"epoch": 17.48, |
|
"learning_rate": 3.1711111111111114e-06, |
|
"loss": 0.0001, |
|
"step": 3575 |
|
}, |
|
{ |
|
"epoch": 17.6, |
|
"learning_rate": 3.1155555555555555e-06, |
|
"loss": 0.0002, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 17.73, |
|
"learning_rate": 3.0600000000000003e-06, |
|
"loss": 0.0001, |
|
"step": 3625 |
|
}, |
|
{ |
|
"epoch": 17.85, |
|
"learning_rate": 3.004444444444445e-06, |
|
"loss": 0.0002, |
|
"step": 3650 |
|
}, |
|
{ |
|
"epoch": 17.97, |
|
"learning_rate": 2.948888888888889e-06, |
|
"loss": 0.0001, |
|
"step": 3675 |
|
}, |
|
{ |
|
"epoch": 18.09, |
|
"learning_rate": 2.8933333333333337e-06, |
|
"loss": 0.0002, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 18.22, |
|
"learning_rate": 2.837777777777778e-06, |
|
"loss": 0.0001, |
|
"step": 3725 |
|
}, |
|
{ |
|
"epoch": 18.34, |
|
"learning_rate": 2.7822222222222222e-06, |
|
"loss": 0.0001, |
|
"step": 3750 |
|
}, |
|
{ |
|
"epoch": 18.46, |
|
"learning_rate": 2.726666666666667e-06, |
|
"loss": 0.0001, |
|
"step": 3775 |
|
}, |
|
{ |
|
"epoch": 18.58, |
|
"learning_rate": 2.6711111111111116e-06, |
|
"loss": 0.0001, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 18.7, |
|
"learning_rate": 2.6155555555555556e-06, |
|
"loss": 0.0001, |
|
"step": 3825 |
|
}, |
|
{ |
|
"epoch": 18.83, |
|
"learning_rate": 2.56e-06, |
|
"loss": 0.0001, |
|
"step": 3850 |
|
}, |
|
{ |
|
"epoch": 18.95, |
|
"learning_rate": 2.504444444444445e-06, |
|
"loss": 0.0001, |
|
"step": 3875 |
|
}, |
|
{ |
|
"epoch": 19.07, |
|
"learning_rate": 2.448888888888889e-06, |
|
"loss": 0.0001, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 19.19, |
|
"learning_rate": 2.3933333333333334e-06, |
|
"loss": 0.0001, |
|
"step": 3925 |
|
}, |
|
{ |
|
"epoch": 19.32, |
|
"learning_rate": 2.337777777777778e-06, |
|
"loss": 0.0001, |
|
"step": 3950 |
|
}, |
|
{ |
|
"epoch": 19.44, |
|
"learning_rate": 2.2822222222222223e-06, |
|
"loss": 0.0001, |
|
"step": 3975 |
|
}, |
|
{ |
|
"epoch": 19.56, |
|
"learning_rate": 2.226666666666667e-06, |
|
"loss": 0.0001, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 19.56, |
|
"eval_loss": 0.5253722071647644, |
|
"eval_runtime": 1086.5288, |
|
"eval_samples_per_second": 2.664, |
|
"eval_steps_per_second": 0.167, |
|
"eval_wer": 0.330906628290866, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 19.68, |
|
"learning_rate": 2.1711111111111113e-06, |
|
"loss": 0.0001, |
|
"step": 4025 |
|
}, |
|
{ |
|
"epoch": 19.8, |
|
"learning_rate": 2.1155555555555557e-06, |
|
"loss": 0.0001, |
|
"step": 4050 |
|
}, |
|
{ |
|
"epoch": 19.93, |
|
"learning_rate": 2.06e-06, |
|
"loss": 0.0001, |
|
"step": 4075 |
|
}, |
|
{ |
|
"epoch": 20.05, |
|
"learning_rate": 2.0044444444444446e-06, |
|
"loss": 0.0001, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 20.17, |
|
"learning_rate": 1.948888888888889e-06, |
|
"loss": 0.0001, |
|
"step": 4125 |
|
}, |
|
{ |
|
"epoch": 20.29, |
|
"learning_rate": 1.8933333333333333e-06, |
|
"loss": 0.0001, |
|
"step": 4150 |
|
}, |
|
{ |
|
"epoch": 20.42, |
|
"learning_rate": 1.837777777777778e-06, |
|
"loss": 0.0001, |
|
"step": 4175 |
|
}, |
|
{ |
|
"epoch": 20.54, |
|
"learning_rate": 1.7822222222222225e-06, |
|
"loss": 0.0001, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 20.66, |
|
"learning_rate": 1.7266666666666667e-06, |
|
"loss": 0.0001, |
|
"step": 4225 |
|
}, |
|
{ |
|
"epoch": 20.78, |
|
"learning_rate": 1.6711111111111112e-06, |
|
"loss": 0.0001, |
|
"step": 4250 |
|
}, |
|
{ |
|
"epoch": 20.9, |
|
"learning_rate": 1.6155555555555559e-06, |
|
"loss": 0.0001, |
|
"step": 4275 |
|
}, |
|
{ |
|
"epoch": 21.03, |
|
"learning_rate": 1.56e-06, |
|
"loss": 0.0001, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 21.15, |
|
"learning_rate": 1.5044444444444446e-06, |
|
"loss": 0.0001, |
|
"step": 4325 |
|
}, |
|
{ |
|
"epoch": 21.27, |
|
"learning_rate": 1.4488888888888892e-06, |
|
"loss": 0.0001, |
|
"step": 4350 |
|
}, |
|
{ |
|
"epoch": 21.39, |
|
"learning_rate": 1.3933333333333335e-06, |
|
"loss": 0.0001, |
|
"step": 4375 |
|
}, |
|
{ |
|
"epoch": 21.52, |
|
"learning_rate": 1.337777777777778e-06, |
|
"loss": 0.0001, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 21.64, |
|
"learning_rate": 1.2822222222222222e-06, |
|
"loss": 0.0001, |
|
"step": 4425 |
|
}, |
|
{ |
|
"epoch": 21.76, |
|
"learning_rate": 1.2266666666666666e-06, |
|
"loss": 0.0001, |
|
"step": 4450 |
|
}, |
|
{ |
|
"epoch": 21.88, |
|
"learning_rate": 1.171111111111111e-06, |
|
"loss": 0.0001, |
|
"step": 4475 |
|
}, |
|
{ |
|
"epoch": 22.0, |
|
"learning_rate": 1.1155555555555558e-06, |
|
"loss": 0.0001, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 22.13, |
|
"learning_rate": 1.06e-06, |
|
"loss": 0.0001, |
|
"step": 4525 |
|
}, |
|
{ |
|
"epoch": 22.25, |
|
"learning_rate": 1.0044444444444445e-06, |
|
"loss": 0.0001, |
|
"step": 4550 |
|
}, |
|
{ |
|
"epoch": 22.37, |
|
"learning_rate": 9.488888888888889e-07, |
|
"loss": 0.0001, |
|
"step": 4575 |
|
}, |
|
{ |
|
"epoch": 22.49, |
|
"learning_rate": 8.933333333333334e-07, |
|
"loss": 0.0001, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 22.62, |
|
"learning_rate": 8.37777777777778e-07, |
|
"loss": 0.0001, |
|
"step": 4625 |
|
}, |
|
{ |
|
"epoch": 22.74, |
|
"learning_rate": 7.822222222222223e-07, |
|
"loss": 0.0001, |
|
"step": 4650 |
|
}, |
|
{ |
|
"epoch": 22.86, |
|
"learning_rate": 7.266666666666668e-07, |
|
"loss": 0.0001, |
|
"step": 4675 |
|
}, |
|
{ |
|
"epoch": 22.98, |
|
"learning_rate": 6.711111111111111e-07, |
|
"loss": 0.0001, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 23.11, |
|
"learning_rate": 6.155555555555556e-07, |
|
"loss": 0.0001, |
|
"step": 4725 |
|
}, |
|
{ |
|
"epoch": 23.23, |
|
"learning_rate": 5.6e-07, |
|
"loss": 0.0001, |
|
"step": 4750 |
|
}, |
|
{ |
|
"epoch": 23.35, |
|
"learning_rate": 5.044444444444445e-07, |
|
"loss": 0.0001, |
|
"step": 4775 |
|
}, |
|
{ |
|
"epoch": 23.47, |
|
"learning_rate": 4.488888888888889e-07, |
|
"loss": 0.0001, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 23.59, |
|
"learning_rate": 3.9333333333333336e-07, |
|
"loss": 0.0001, |
|
"step": 4825 |
|
}, |
|
{ |
|
"epoch": 23.72, |
|
"learning_rate": 3.3777777777777777e-07, |
|
"loss": 0.0001, |
|
"step": 4850 |
|
}, |
|
{ |
|
"epoch": 23.84, |
|
"learning_rate": 2.822222222222222e-07, |
|
"loss": 0.0001, |
|
"step": 4875 |
|
}, |
|
{ |
|
"epoch": 23.96, |
|
"learning_rate": 2.266666666666667e-07, |
|
"loss": 0.0001, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 24.08, |
|
"learning_rate": 1.7111111111111114e-07, |
|
"loss": 0.0001, |
|
"step": 4925 |
|
}, |
|
{ |
|
"epoch": 24.21, |
|
"learning_rate": 1.1555555555555556e-07, |
|
"loss": 0.0001, |
|
"step": 4950 |
|
}, |
|
{ |
|
"epoch": 24.33, |
|
"learning_rate": 6.000000000000001e-08, |
|
"loss": 0.0001, |
|
"step": 4975 |
|
}, |
|
{ |
|
"epoch": 24.45, |
|
"learning_rate": 4.444444444444445e-09, |
|
"loss": 0.0001, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 24.45, |
|
"eval_loss": 0.5355350971221924, |
|
"eval_runtime": 1093.777, |
|
"eval_samples_per_second": 2.646, |
|
"eval_steps_per_second": 0.165, |
|
"eval_wer": 0.330906628290866, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 24.45, |
|
"step": 5000, |
|
"total_flos": 4.614595985276928e+19, |
|
"train_loss": 0.04446965890082065, |
|
"train_runtime": 56869.2369, |
|
"train_samples_per_second": 2.813, |
|
"train_steps_per_second": 0.088 |
|
} |
|
], |
|
"max_steps": 5000, |
|
"num_train_epochs": 25, |
|
"total_flos": 4.614595985276928e+19, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|