| { | |
| "best_metric": 8.37737794884072, | |
| "best_model_checkpoint": "./checkpoint-2000", | |
| "epoch": 121.93975903614458, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.113844316514203e-06, | |
| "loss": 0.2509, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.229195710491767e-06, | |
| "loss": 0.1429, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 6.903829450223392e-06, | |
| "loss": 0.0992, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.394062147380189e-06, | |
| "loss": 0.0708, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.756372548783638e-06, | |
| "loss": 0.0552, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 8.051909672656425e-06, | |
| "loss": 0.0337, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 8.310718700403308e-06, | |
| "loss": 0.0294, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 8.517519915806749e-06, | |
| "loss": 0.0227, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 8.707944195594548e-06, | |
| "loss": 0.019, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 8.884648606022435e-06, | |
| "loss": 0.0167, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 9.038013352913754e-06, | |
| "loss": 0.012, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 9.178024397233219e-06, | |
| "loss": 0.0109, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 9.306822072655195e-06, | |
| "loss": 0.0094, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 8.53, | |
| "learning_rate": 9.430661001703139e-06, | |
| "loss": 0.0086, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 9.14, | |
| "learning_rate": 9.541372600623587e-06, | |
| "loss": 0.0094, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "learning_rate": 9.644954810309533e-06, | |
| "loss": 0.0076, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 10.36, | |
| "learning_rate": 9.746043382472281e-06, | |
| "loss": 0.0072, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 10.96, | |
| "learning_rate": 9.837598900861371e-06, | |
| "loss": 0.0065, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 11.58, | |
| "learning_rate": 9.924224332752688e-06, | |
| "loss": 0.0052, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 12.19, | |
| "learning_rate": 9.995555555555556e-06, | |
| "loss": 0.0053, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 12.8, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.0038, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 13.41, | |
| "learning_rate": 9.884444444444445e-06, | |
| "loss": 0.0038, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 9.826666666666667e-06, | |
| "loss": 0.004, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 14.63, | |
| "learning_rate": 9.771111111111113e-06, | |
| "loss": 0.0037, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 15.24, | |
| "learning_rate": 9.715555555555557e-06, | |
| "loss": 0.0047, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 15.84, | |
| "learning_rate": 9.66e-06, | |
| "loss": 0.0047, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 16.46, | |
| "learning_rate": 9.602222222222222e-06, | |
| "loss": 0.0045, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 17.07, | |
| "learning_rate": 9.546666666666668e-06, | |
| "loss": 0.0044, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 17.67, | |
| "learning_rate": 9.491111111111112e-06, | |
| "loss": 0.0031, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 18.29, | |
| "learning_rate": 9.433333333333335e-06, | |
| "loss": 0.0032, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 18.89, | |
| "learning_rate": 9.377777777777779e-06, | |
| "loss": 0.0024, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 19.51, | |
| "learning_rate": 9.322222222222223e-06, | |
| "loss": 0.0023, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 20.12, | |
| "learning_rate": 9.264444444444445e-06, | |
| "loss": 0.0025, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 20.72, | |
| "learning_rate": 9.208888888888889e-06, | |
| "loss": 0.0025, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 21.34, | |
| "learning_rate": 9.153333333333334e-06, | |
| "loss": 0.0025, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 21.94, | |
| "learning_rate": 9.097777777777778e-06, | |
| "loss": 0.0022, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 22.55, | |
| "learning_rate": 9.040000000000002e-06, | |
| "loss": 0.002, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 23.17, | |
| "learning_rate": 8.984444444444446e-06, | |
| "loss": 0.002, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 23.77, | |
| "learning_rate": 8.92888888888889e-06, | |
| "loss": 0.0021, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 24.39, | |
| "learning_rate": 8.871111111111111e-06, | |
| "loss": 0.0022, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 24.39, | |
| "eval_loss": 0.2181396484375, | |
| "eval_runtime": 1286.5853, | |
| "eval_samples_per_second": 1.03, | |
| "eval_steps_per_second": 0.129, | |
| "eval_wer": 8.780685438573874, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 24.99, | |
| "learning_rate": 8.815555555555557e-06, | |
| "loss": 0.003, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 25.6, | |
| "learning_rate": 8.76e-06, | |
| "loss": 0.0027, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 26.22, | |
| "learning_rate": 8.702222222222222e-06, | |
| "loss": 0.0025, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 26.82, | |
| "learning_rate": 8.646666666666668e-06, | |
| "loss": 0.0019, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 27.43, | |
| "learning_rate": 8.591111111111112e-06, | |
| "loss": 0.0021, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 28.05, | |
| "learning_rate": 8.533333333333335e-06, | |
| "loss": 0.0018, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 28.65, | |
| "learning_rate": 8.477777777777778e-06, | |
| "loss": 0.0015, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 29.27, | |
| "learning_rate": 8.422222222222223e-06, | |
| "loss": 0.0012, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 29.87, | |
| "learning_rate": 8.366666666666667e-06, | |
| "loss": 0.0011, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 30.48, | |
| "learning_rate": 8.30888888888889e-06, | |
| "loss": 0.0009, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 31.1, | |
| "learning_rate": 8.253333333333334e-06, | |
| "loss": 0.0011, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 31.7, | |
| "learning_rate": 8.197777777777779e-06, | |
| "loss": 0.0008, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 32.31, | |
| "learning_rate": 8.14e-06, | |
| "loss": 0.0006, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 32.92, | |
| "learning_rate": 8.084444444444444e-06, | |
| "loss": 0.0006, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 33.53, | |
| "learning_rate": 8.02888888888889e-06, | |
| "loss": 0.0006, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 34.14, | |
| "learning_rate": 7.971111111111111e-06, | |
| "loss": 0.001, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 34.75, | |
| "learning_rate": 7.915555555555557e-06, | |
| "loss": 0.0009, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 35.36, | |
| "learning_rate": 7.860000000000001e-06, | |
| "loss": 0.0007, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 35.96, | |
| "learning_rate": 7.804444444444445e-06, | |
| "loss": 0.0006, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 36.58, | |
| "learning_rate": 7.746666666666666e-06, | |
| "loss": 0.0005, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 37.19, | |
| "learning_rate": 7.691111111111112e-06, | |
| "loss": 0.0004, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 37.8, | |
| "learning_rate": 7.635555555555556e-06, | |
| "loss": 0.0004, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 38.41, | |
| "learning_rate": 7.5777777777777785e-06, | |
| "loss": 0.0004, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 39.02, | |
| "learning_rate": 7.5222222222222226e-06, | |
| "loss": 0.0003, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 39.63, | |
| "learning_rate": 7.4666666666666675e-06, | |
| "loss": 0.0003, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 40.24, | |
| "learning_rate": 7.40888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 40.84, | |
| "learning_rate": 7.353333333333334e-06, | |
| "loss": 0.0003, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 41.46, | |
| "learning_rate": 7.297777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 42.07, | |
| "learning_rate": 7.24e-06, | |
| "loss": 0.0003, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 42.67, | |
| "learning_rate": 7.184444444444445e-06, | |
| "loss": 0.0003, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 43.29, | |
| "learning_rate": 7.12888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 43.89, | |
| "learning_rate": 7.073333333333334e-06, | |
| "loss": 0.0003, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 44.51, | |
| "learning_rate": 7.015555555555556e-06, | |
| "loss": 0.0003, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 45.12, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.0003, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 45.72, | |
| "learning_rate": 6.904444444444444e-06, | |
| "loss": 0.0002, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 46.34, | |
| "learning_rate": 6.846666666666667e-06, | |
| "loss": 0.0002, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 46.94, | |
| "learning_rate": 6.7911111111111115e-06, | |
| "loss": 0.0002, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 47.55, | |
| "learning_rate": 6.735555555555556e-06, | |
| "loss": 0.0002, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 48.17, | |
| "learning_rate": 6.677777777777779e-06, | |
| "loss": 0.0002, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 48.77, | |
| "learning_rate": 6.6222222222222236e-06, | |
| "loss": 0.0002, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 48.77, | |
| "eval_loss": 0.25634765625, | |
| "eval_runtime": 1289.0888, | |
| "eval_samples_per_second": 1.028, | |
| "eval_steps_per_second": 0.129, | |
| "eval_wer": 8.37737794884072, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 49.39, | |
| "learning_rate": 6.566666666666667e-06, | |
| "loss": 0.0002, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 49.99, | |
| "learning_rate": 6.511111111111112e-06, | |
| "loss": 0.0002, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 50.6, | |
| "learning_rate": 6.453333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 51.22, | |
| "learning_rate": 6.397777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 51.82, | |
| "learning_rate": 6.342222222222223e-06, | |
| "loss": 0.0002, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 52.43, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.0002, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 53.05, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 53.65, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.0002, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 54.27, | |
| "learning_rate": 6.1155555555555555e-06, | |
| "loss": 0.0002, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 54.87, | |
| "learning_rate": 6.0600000000000004e-06, | |
| "loss": 0.0002, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 55.48, | |
| "learning_rate": 6.004444444444445e-06, | |
| "loss": 0.0002, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 56.1, | |
| "learning_rate": 5.946666666666668e-06, | |
| "loss": 0.0002, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 56.7, | |
| "learning_rate": 5.891111111111112e-06, | |
| "loss": 0.0002, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 57.31, | |
| "learning_rate": 5.8355555555555565e-06, | |
| "loss": 0.0002, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 57.92, | |
| "learning_rate": 5.78e-06, | |
| "loss": 0.0002, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 58.53, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.0002, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 59.14, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.0002, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 59.75, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.0002, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 60.36, | |
| "learning_rate": 5.553333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 60.96, | |
| "learning_rate": 5.497777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 61.58, | |
| "learning_rate": 5.442222222222223e-06, | |
| "loss": 0.0002, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 62.19, | |
| "learning_rate": 5.3844444444444445e-06, | |
| "loss": 0.0002, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 62.8, | |
| "learning_rate": 5.328888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 63.41, | |
| "learning_rate": 5.273333333333333e-06, | |
| "loss": 0.0001, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 64.02, | |
| "learning_rate": 5.215555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 64.63, | |
| "learning_rate": 5.1600000000000006e-06, | |
| "loss": 0.0001, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 65.24, | |
| "learning_rate": 5.1044444444444455e-06, | |
| "loss": 0.0001, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 65.84, | |
| "learning_rate": 5.0488888888888895e-06, | |
| "loss": 0.0001, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 66.46, | |
| "learning_rate": 4.991111111111112e-06, | |
| "loss": 0.0001, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 67.07, | |
| "learning_rate": 4.935555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 67.67, | |
| "learning_rate": 4.880000000000001e-06, | |
| "loss": 0.0001, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 68.29, | |
| "learning_rate": 4.822222222222222e-06, | |
| "loss": 0.0001, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 68.89, | |
| "learning_rate": 4.766666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 69.51, | |
| "learning_rate": 4.711111111111111e-06, | |
| "loss": 0.0001, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 70.12, | |
| "learning_rate": 4.653333333333333e-06, | |
| "loss": 0.0001, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 70.72, | |
| "learning_rate": 4.597777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 71.34, | |
| "learning_rate": 4.542222222222223e-06, | |
| "loss": 0.0001, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 71.94, | |
| "learning_rate": 4.486666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 72.55, | |
| "learning_rate": 4.4288888888888895e-06, | |
| "loss": 0.0001, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 73.17, | |
| "learning_rate": 4.3733333333333335e-06, | |
| "loss": 0.0001, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 73.17, | |
| "eval_loss": 0.275634765625, | |
| "eval_runtime": 1273.9374, | |
| "eval_samples_per_second": 1.04, | |
| "eval_steps_per_second": 0.13, | |
| "eval_wer": 8.451041417285133, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 73.77, | |
| "learning_rate": 4.3177777777777776e-06, | |
| "loss": 0.0001, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 74.39, | |
| "learning_rate": 4.26e-06, | |
| "loss": 0.0001, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 74.99, | |
| "learning_rate": 4.204444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 75.6, | |
| "learning_rate": 4.148888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 76.22, | |
| "learning_rate": 4.091111111111111e-06, | |
| "loss": 0.0001, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 76.82, | |
| "learning_rate": 4.04e-06, | |
| "loss": 0.0001, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 77.43, | |
| "learning_rate": 3.984444444444444e-06, | |
| "loss": 0.0001, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 78.05, | |
| "learning_rate": 3.926666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 78.65, | |
| "learning_rate": 3.8711111111111115e-06, | |
| "loss": 0.0001, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 79.27, | |
| "learning_rate": 3.8155555555555555e-06, | |
| "loss": 0.0001, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 79.87, | |
| "learning_rate": 3.7600000000000004e-06, | |
| "loss": 0.0001, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 80.48, | |
| "learning_rate": 3.7022222222222227e-06, | |
| "loss": 0.0001, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 81.1, | |
| "learning_rate": 3.6466666666666668e-06, | |
| "loss": 0.0001, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 81.7, | |
| "learning_rate": 3.5911111111111112e-06, | |
| "loss": 0.0001, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 82.31, | |
| "learning_rate": 3.5333333333333335e-06, | |
| "loss": 0.0001, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 82.92, | |
| "learning_rate": 3.4777777777777784e-06, | |
| "loss": 0.0001, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 83.53, | |
| "learning_rate": 3.4222222222222224e-06, | |
| "loss": 0.0001, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 84.14, | |
| "learning_rate": 3.3644444444444447e-06, | |
| "loss": 0.0001, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 84.75, | |
| "learning_rate": 3.308888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 85.36, | |
| "learning_rate": 3.2533333333333332e-06, | |
| "loss": 0.0001, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 85.96, | |
| "learning_rate": 3.197777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 86.58, | |
| "learning_rate": 3.1400000000000004e-06, | |
| "loss": 0.0001, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 87.19, | |
| "learning_rate": 3.084444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 87.8, | |
| "learning_rate": 3.028888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 88.41, | |
| "learning_rate": 2.9711111111111112e-06, | |
| "loss": 0.0001, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 89.02, | |
| "learning_rate": 2.915555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 89.63, | |
| "learning_rate": 2.86e-06, | |
| "loss": 0.0001, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 90.24, | |
| "learning_rate": 2.8022222222222225e-06, | |
| "loss": 0.0001, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 90.84, | |
| "learning_rate": 2.746666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 91.46, | |
| "learning_rate": 2.6911111111111114e-06, | |
| "loss": 0.0001, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 92.07, | |
| "learning_rate": 2.6333333333333332e-06, | |
| "loss": 0.0001, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 92.67, | |
| "learning_rate": 2.577777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 93.29, | |
| "learning_rate": 2.5222222222222226e-06, | |
| "loss": 0.0001, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 93.89, | |
| "learning_rate": 2.466666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 94.51, | |
| "learning_rate": 2.4088888888888894e-06, | |
| "loss": 0.0001, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 95.12, | |
| "learning_rate": 2.3533333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 95.72, | |
| "learning_rate": 2.297777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 96.34, | |
| "learning_rate": 2.24e-06, | |
| "loss": 0.0001, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 96.94, | |
| "learning_rate": 2.1844444444444446e-06, | |
| "loss": 0.0001, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 97.55, | |
| "learning_rate": 2.128888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 97.55, | |
| "eval_loss": 0.287109375, | |
| "eval_runtime": 1274.1311, | |
| "eval_samples_per_second": 1.04, | |
| "eval_steps_per_second": 0.13, | |
| "eval_wer": 8.482348391374007, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 98.17, | |
| "learning_rate": 2.0711111111111114e-06, | |
| "loss": 0.0001, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 98.77, | |
| "learning_rate": 2.0155555555555554e-06, | |
| "loss": 0.0001, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 99.39, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 0.0001, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 99.99, | |
| "learning_rate": 1.9044444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 100.6, | |
| "learning_rate": 1.8466666666666668e-06, | |
| "loss": 0.0001, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 101.22, | |
| "learning_rate": 1.7955555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 101.82, | |
| "learning_rate": 1.74e-06, | |
| "loss": 0.0001, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 102.43, | |
| "learning_rate": 1.6822222222222224e-06, | |
| "loss": 0.0001, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 103.05, | |
| "learning_rate": 1.6266666666666666e-06, | |
| "loss": 0.0001, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 103.65, | |
| "learning_rate": 1.5711111111111113e-06, | |
| "loss": 0.0001, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 104.27, | |
| "learning_rate": 1.5133333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 104.87, | |
| "learning_rate": 1.457777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 105.48, | |
| "learning_rate": 1.4022222222222223e-06, | |
| "loss": 0.0001, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 106.1, | |
| "learning_rate": 1.3444444444444446e-06, | |
| "loss": 0.0001, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 106.7, | |
| "learning_rate": 1.288888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 107.31, | |
| "learning_rate": 1.2333333333333335e-06, | |
| "loss": 0.0001, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 107.92, | |
| "learning_rate": 1.1777777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 108.53, | |
| "learning_rate": 1.12e-06, | |
| "loss": 0.0001, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 109.14, | |
| "learning_rate": 1.0644444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 109.75, | |
| "learning_rate": 1.008888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 110.36, | |
| "learning_rate": 9.511111111111111e-07, | |
| "loss": 0.0001, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 110.96, | |
| "learning_rate": 8.955555555555557e-07, | |
| "loss": 0.0001, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 111.58, | |
| "learning_rate": 8.400000000000001e-07, | |
| "loss": 0.0001, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 112.19, | |
| "learning_rate": 7.822222222222223e-07, | |
| "loss": 0.0001, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 112.8, | |
| "learning_rate": 7.266666666666668e-07, | |
| "loss": 0.0001, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 113.41, | |
| "learning_rate": 6.711111111111111e-07, | |
| "loss": 0.0001, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 114.02, | |
| "learning_rate": 6.133333333333333e-07, | |
| "loss": 0.0001, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 114.63, | |
| "learning_rate": 5.577777777777779e-07, | |
| "loss": 0.0001, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 115.24, | |
| "learning_rate": 5.022222222222222e-07, | |
| "loss": 0.0001, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 115.84, | |
| "learning_rate": 4.466666666666667e-07, | |
| "loss": 0.0001, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 116.46, | |
| "learning_rate": 3.8888888888888895e-07, | |
| "loss": 0.0001, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 117.07, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 0.0001, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 117.67, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.0001, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 118.29, | |
| "learning_rate": 2.2e-07, | |
| "loss": 0.0001, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 118.89, | |
| "learning_rate": 1.6444444444444447e-07, | |
| "loss": 0.0001, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 119.51, | |
| "learning_rate": 1.088888888888889e-07, | |
| "loss": 0.0001, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 120.12, | |
| "learning_rate": 5.111111111111112e-08, | |
| "loss": 0.0001, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 120.72, | |
| "learning_rate": 0.0, | |
| "loss": 0.0001, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 121.34, | |
| "learning_rate": 0.0, | |
| "loss": 0.0001, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 121.94, | |
| "learning_rate": 0.0, | |
| "loss": 0.0001, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 121.94, | |
| "eval_loss": 0.291259765625, | |
| "eval_runtime": 1274.9287, | |
| "eval_samples_per_second": 1.039, | |
| "eval_steps_per_second": 0.13, | |
| "eval_wer": 8.473140457818456, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 121.94, | |
| "step": 5000, | |
| "total_flos": 6.860942180681253e+20, | |
| "train_loss": 0.00465274755358696, | |
| "train_runtime": 62864.0618, | |
| "train_samples_per_second": 5.09, | |
| "train_steps_per_second": 0.08 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 122, | |
| "total_flos": 6.860942180681253e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |