| { | |
| "best_metric": 0.11633493479752917, | |
| "best_model_checkpoint": "./hc-impaired-all-v3/checkpoint-1000", | |
| "epoch": 34.36426116838488, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.5479, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.5105, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.3697, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.3095, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.2935, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3e-06, | |
| "loss": 0.3167, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.2239, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.2317, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.2659, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 5e-06, | |
| "loss": 0.2395, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.2413, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6e-06, | |
| "loss": 0.2192, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.1434, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7e-06, | |
| "loss": 0.1786, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.1851, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.1543, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.1652, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 9e-06, | |
| "loss": 0.1407, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.0961, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 1e-05, | |
| "loss": 0.1071, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.1066, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.1302, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.1139, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.083, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.0696, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.0625, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.0779, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.0771, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.0631, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.0449, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.0432, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.0382, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.0409, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.0463, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.0418, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 6.19, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.0227, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.031, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 9e-06, | |
| "loss": 0.0263, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.0221, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 6.87, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.0275, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 6.87, | |
| "eval_loss": 0.22118031978607178, | |
| "eval_runtime": 335.2838, | |
| "eval_samples_per_second": 0.88, | |
| "eval_steps_per_second": 0.11, | |
| "eval_wer": 0.11633493479752917, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.0251, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 7.22, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.0115, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.0171, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 7.56, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.0157, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.0151, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.0177, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.0133, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.0113, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 8.42, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.011, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.0097, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 8.76, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.0112, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.011, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.0116, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.0087, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.0079, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.0073, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 9.79, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.0092, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 9.97, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.007, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.0071, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 10.31, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.0071, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 10.48, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.0058, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 10.65, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.0049, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 10.82, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.0047, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.0056, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 11.17, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.0042, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 11.34, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.0043, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 11.51, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.0043, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 11.68, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.0036, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 11.86, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.0044, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.0032, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 12.2, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.0035, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 12.37, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.0025, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 12.54, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.0032, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 12.71, | |
| "learning_rate": 7e-06, | |
| "loss": 0.003, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 12.89, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.0037, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 13.06, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.0038, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 13.23, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.002, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 13.4, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.0027, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 13.57, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.0036, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 13.75, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.0021, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 13.75, | |
| "eval_loss": 0.305084764957428, | |
| "eval_runtime": 284.7268, | |
| "eval_samples_per_second": 1.036, | |
| "eval_steps_per_second": 0.13, | |
| "eval_wer": 0.11233127430793868, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 13.92, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.0038, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 14.09, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.0029, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 14.26, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.0008, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 14.43, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.001, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 14.6, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.0017, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 14.78, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.0011, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 14.95, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.0015, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 15.12, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.0021, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 15.29, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.0016, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 15.46, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.0015, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 15.64, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.0019, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 15.81, | |
| "learning_rate": 6e-06, | |
| "loss": 0.0014, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 15.98, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.0015, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 16.15, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.0006, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 16.32, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.0011, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 16.49, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.0007, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 16.67, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.0018, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 16.84, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.0006, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.0011, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 17.18, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.0008, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 17.35, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.0006, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 17.53, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.0004, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 17.7, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.0008, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 17.87, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.0006, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.0004, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 18.21, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.0003, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 18.38, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.0004, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 18.56, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.0005, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 18.73, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.0005, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 18.9, | |
| "learning_rate": 5e-06, | |
| "loss": 0.0005, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 19.07, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.0006, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 19.24, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 19.42, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.0003, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 19.59, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.0006, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 19.76, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.0003, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 19.93, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 20.1, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.0002, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 20.27, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.0003, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 20.45, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.0002, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 20.62, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.0004, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 20.62, | |
| "eval_loss": 0.3517112135887146, | |
| "eval_runtime": 284.1633, | |
| "eval_samples_per_second": 1.038, | |
| "eval_steps_per_second": 0.13, | |
| "eval_wer": 0.11130176161061542, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 20.79, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 20.96, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 21.13, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 21.31, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.0003, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 21.48, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.0002, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 21.65, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.0002, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 21.82, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.0004, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 21.99, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.0005, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 22.16, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.0004, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 22.34, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 22.51, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 22.68, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 22.85, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.0002, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 23.02, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.0002, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 23.2, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.0002, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 23.37, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.0002, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 23.54, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.0002, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 23.71, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 23.88, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.0002, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 24.05, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.0002, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 24.23, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 24.4, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.0002, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 24.57, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.0002, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 24.74, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.0001, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 24.91, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.0002, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 25.09, | |
| "learning_rate": 3e-06, | |
| "loss": 0.0001, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 25.26, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 25.43, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 25.6, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.0001, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 25.77, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.0001, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 25.95, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.0001, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 26.12, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 26.29, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.0001, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 26.46, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.0001, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 26.63, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0001, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 26.8, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.0001, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 26.98, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 27.15, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.0001, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 27.32, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 27.49, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0001, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 27.49, | |
| "eval_loss": 0.3759867250919342, | |
| "eval_runtime": 285.2549, | |
| "eval_samples_per_second": 1.034, | |
| "eval_steps_per_second": 0.13, | |
| "eval_wer": 0.11038663921299474, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 27.66, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 27.84, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.0001, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 28.01, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.0001, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 28.18, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.0001, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 28.35, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 28.52, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 28.69, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 0.0001, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 28.87, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 29.04, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 0.0001, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 29.21, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 29.38, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 0.0001, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 29.55, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 0.0001, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 29.73, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.0001, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 29.9, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 30.07, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.0001, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 30.24, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 30.41, | |
| "learning_rate": 1.2777777777777779e-06, | |
| "loss": 0.0001, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 30.58, | |
| "learning_rate": 1.2222222222222223e-06, | |
| "loss": 0.0001, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 30.76, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 0.0001, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 30.93, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0001, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 31.1, | |
| "learning_rate": 1.0555555555555557e-06, | |
| "loss": 0.0001, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 31.27, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.0001, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 31.44, | |
| "learning_rate": 9.444444444444445e-07, | |
| "loss": 0.0001, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 31.62, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 0.0001, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 31.79, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.0001, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 31.96, | |
| "learning_rate": 7.777777777777779e-07, | |
| "loss": 0.0001, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 32.13, | |
| "learning_rate": 7.222222222222222e-07, | |
| "loss": 0.0001, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 32.3, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 0.0001, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 32.47, | |
| "learning_rate": 6.111111111111112e-07, | |
| "loss": 0.0001, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 32.65, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.0001, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 32.82, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.0001, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 32.99, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 0.0001, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 33.16, | |
| "learning_rate": 3.8888888888888895e-07, | |
| "loss": 0.0001, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 33.33, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 0.0001, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 33.51, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.0001, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 33.68, | |
| "learning_rate": 2.2222222222222224e-07, | |
| "loss": 0.0001, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 33.85, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 0.0001, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 34.02, | |
| "learning_rate": 1.1111111111111112e-07, | |
| "loss": 0.0001, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 34.19, | |
| "learning_rate": 5.555555555555556e-08, | |
| "loss": 0.0001, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 34.36, | |
| "learning_rate": 0.0, | |
| "loss": 0.0001, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 34.36, | |
| "eval_loss": 0.38366764783859253, | |
| "eval_runtime": 284.7565, | |
| "eval_samples_per_second": 1.036, | |
| "eval_steps_per_second": 0.13, | |
| "eval_wer": 0.11072981011210249, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 34.36, | |
| "step": 5000, | |
| "total_flos": 2.7110677487026176e+20, | |
| "train_loss": 0.032343216465134175, | |
| "train_runtime": 36155.6286, | |
| "train_samples_per_second": 2.213, | |
| "train_steps_per_second": 0.138 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 35, | |
| "save_steps": 1000, | |
| "total_flos": 2.7110677487026176e+20, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |