| { | |
| "best_metric": 15.594426326712126, | |
| "best_model_checkpoint": "./checkpoint-4000", | |
| "epoch": 37.59176029962547, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.973833272194737e-06, | |
| "loss": 0.469, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.195318418690893e-06, | |
| "loss": 0.2768, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 6.881634451095711e-06, | |
| "loss": 0.2426, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.361221988663844e-06, | |
| "loss": 0.2244, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 7.716963756434345e-06, | |
| "loss": 0.2247, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.019180844200955e-06, | |
| "loss": 0.156, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 8.27351214279797e-06, | |
| "loss": 0.1278, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 8.49307723936858e-06, | |
| "loss": 0.1276, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 8.686247975778677e-06, | |
| "loss": 0.1281, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 8.858694625217149e-06, | |
| "loss": 0.1256, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 9.020362953730323e-06, | |
| "loss": 0.1083, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 9.161852281961698e-06, | |
| "loss": 0.0598, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.29189975311636e-06, | |
| "loss": 0.063, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 9.412218256259678e-06, | |
| "loss": 0.0622, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 9.519831289296397e-06, | |
| "loss": 0.0652, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.624764935335318e-06, | |
| "loss": 0.0661, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 9.719445885591654e-06, | |
| "loss": 0.0342, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 9.812484046603779e-06, | |
| "loss": 0.035, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 9.900435550016748e-06, | |
| "loss": 0.0359, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 9.98382788472848e-06, | |
| "loss": 0.0369, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 9.957777777777779e-06, | |
| "loss": 0.0384, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 9.9e-06, | |
| "loss": 0.0274, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 9.844444444444446e-06, | |
| "loss": 0.0236, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 9.78888888888889e-06, | |
| "loss": 0.0246, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 9.733333333333334e-06, | |
| "loss": 0.0238, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 9.677777777777778e-06, | |
| "loss": 0.0239, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 9.622222222222222e-06, | |
| "loss": 0.0204, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 9.566666666666668e-06, | |
| "loss": 0.0146, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 9.511111111111112e-06, | |
| "loss": 0.0159, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 9.455555555555557e-06, | |
| "loss": 0.0167, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.0154, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 9.342222222222223e-06, | |
| "loss": 0.0149, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 9.286666666666667e-06, | |
| "loss": 0.0107, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 9.231111111111111e-06, | |
| "loss": 0.0107, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 9.175555555555557e-06, | |
| "loss": 0.0108, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 9.12e-06, | |
| "loss": 0.0123, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 9.064444444444447e-06, | |
| "loss": 0.0113, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 9.008888888888889e-06, | |
| "loss": 0.0086, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 8.953333333333335e-06, | |
| "loss": 0.0077, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 7.52, | |
| "learning_rate": 8.897777777777779e-06, | |
| "loss": 0.0076, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 7.52, | |
| "eval_loss": 0.2607421875, | |
| "eval_runtime": 1390.3082, | |
| "eval_samples_per_second": 3.331, | |
| "eval_steps_per_second": 0.416, | |
| "eval_wer": 16.033204862140526, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 7.7, | |
| "learning_rate": 8.842222222222223e-06, | |
| "loss": 0.0069, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 8.786666666666668e-06, | |
| "loss": 0.008, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.0078, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.0066, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 8.46, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.0069, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.0076, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 8.83, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.0073, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.0077, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 9.21, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.0055, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.0044, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 9.58, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.0055, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 9.77, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.0047, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.0051, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 10.15, | |
| "learning_rate": 8.115555555555557e-06, | |
| "loss": 0.0058, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 10.34, | |
| "learning_rate": 8.06e-06, | |
| "loss": 0.0058, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 10.52, | |
| "learning_rate": 8.004444444444445e-06, | |
| "loss": 0.0069, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 10.71, | |
| "learning_rate": 7.948888888888889e-06, | |
| "loss": 0.0076, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 10.9, | |
| "learning_rate": 7.893333333333335e-06, | |
| "loss": 0.0055, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 11.09, | |
| "learning_rate": 7.837777777777779e-06, | |
| "loss": 0.0043, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 11.28, | |
| "learning_rate": 7.782222222222223e-06, | |
| "loss": 0.0032, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 11.46, | |
| "learning_rate": 7.726666666666667e-06, | |
| "loss": 0.0044, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 11.65, | |
| "learning_rate": 7.67111111111111e-06, | |
| "loss": 0.0047, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 11.84, | |
| "learning_rate": 7.6155555555555564e-06, | |
| "loss": 0.0038, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 7.557777777777779e-06, | |
| "loss": 0.0046, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 12.22, | |
| "learning_rate": 7.502222222222223e-06, | |
| "loss": 0.0033, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 12.4, | |
| "learning_rate": 7.446666666666668e-06, | |
| "loss": 0.003, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 12.59, | |
| "learning_rate": 7.3911111111111125e-06, | |
| "loss": 0.0029, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 12.78, | |
| "learning_rate": 7.335555555555556e-06, | |
| "loss": 0.004, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "learning_rate": 7.280000000000001e-06, | |
| "loss": 0.0043, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 13.16, | |
| "learning_rate": 7.224444444444445e-06, | |
| "loss": 0.0027, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 13.34, | |
| "learning_rate": 7.1688888888888895e-06, | |
| "loss": 0.0034, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 13.53, | |
| "learning_rate": 7.113333333333334e-06, | |
| "loss": 0.0025, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 13.72, | |
| "learning_rate": 7.057777777777778e-06, | |
| "loss": 0.0023, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 13.91, | |
| "learning_rate": 7.0022222222222225e-06, | |
| "loss": 0.0029, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 14.1, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.0021, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 14.28, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.0018, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 14.47, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.0019, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 14.66, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.0014, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 14.85, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.0012, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.0013, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "eval_loss": 0.28955078125, | |
| "eval_runtime": 1408.4786, | |
| "eval_samples_per_second": 3.288, | |
| "eval_steps_per_second": 0.411, | |
| "eval_wer": 15.78416839608657, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 15.22, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.001, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 15.41, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.0013, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 15.6, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.0015, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 15.79, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.0017, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 15.97, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.0012, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 16.16, | |
| "learning_rate": 6.331111111111111e-06, | |
| "loss": 0.0015, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 16.35, | |
| "learning_rate": 6.275555555555556e-06, | |
| "loss": 0.0014, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 16.54, | |
| "learning_rate": 6.220000000000001e-06, | |
| "loss": 0.0018, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 16.73, | |
| "learning_rate": 6.1644444444444455e-06, | |
| "loss": 0.0011, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 16.91, | |
| "learning_rate": 6.1088888888888895e-06, | |
| "loss": 0.0013, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 17.1, | |
| "learning_rate": 6.0533333333333335e-06, | |
| "loss": 0.0023, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 17.29, | |
| "learning_rate": 5.9977777777777776e-06, | |
| "loss": 0.0018, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 17.48, | |
| "learning_rate": 5.9422222222222225e-06, | |
| "loss": 0.0024, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 17.67, | |
| "learning_rate": 5.886666666666667e-06, | |
| "loss": 0.0028, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 17.85, | |
| "learning_rate": 5.831111111111112e-06, | |
| "loss": 0.0016, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 5.7733333333333345e-06, | |
| "loss": 0.002, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 18.23, | |
| "learning_rate": 5.7177777777777786e-06, | |
| "loss": 0.002, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 18.42, | |
| "learning_rate": 5.662222222222223e-06, | |
| "loss": 0.0023, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 18.61, | |
| "learning_rate": 5.606666666666667e-06, | |
| "loss": 0.0026, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 18.79, | |
| "learning_rate": 5.5511111111111115e-06, | |
| "loss": 0.0026, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 18.98, | |
| "learning_rate": 5.495555555555556e-06, | |
| "loss": 0.0028, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 19.17, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.0034, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 19.36, | |
| "learning_rate": 5.3844444444444445e-06, | |
| "loss": 0.002, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 19.55, | |
| "learning_rate": 5.328888888888889e-06, | |
| "loss": 0.0022, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 19.73, | |
| "learning_rate": 5.273333333333333e-06, | |
| "loss": 0.0024, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 19.92, | |
| "learning_rate": 5.217777777777778e-06, | |
| "loss": 0.0019, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 20.11, | |
| "learning_rate": 5.1600000000000006e-06, | |
| "loss": 0.0015, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 20.3, | |
| "learning_rate": 5.1044444444444455e-06, | |
| "loss": 0.0014, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 20.49, | |
| "learning_rate": 5.0488888888888895e-06, | |
| "loss": 0.0013, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 20.67, | |
| "learning_rate": 4.9933333333333335e-06, | |
| "loss": 0.0009, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 20.86, | |
| "learning_rate": 4.937777777777778e-06, | |
| "loss": 0.0013, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 21.05, | |
| "learning_rate": 4.8822222222222224e-06, | |
| "loss": 0.0019, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 21.24, | |
| "learning_rate": 4.826666666666667e-06, | |
| "loss": 0.002, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 21.43, | |
| "learning_rate": 4.771111111111111e-06, | |
| "loss": 0.0015, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 21.61, | |
| "learning_rate": 4.715555555555556e-06, | |
| "loss": 0.001, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 21.8, | |
| "learning_rate": 4.66e-06, | |
| "loss": 0.0012, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 21.99, | |
| "learning_rate": 4.604444444444444e-06, | |
| "loss": 0.001, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 22.18, | |
| "learning_rate": 4.546666666666667e-06, | |
| "loss": 0.0009, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 22.37, | |
| "learning_rate": 4.4911111111111115e-06, | |
| "loss": 0.0008, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 22.55, | |
| "learning_rate": 4.4355555555555555e-06, | |
| "loss": 0.0009, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 22.55, | |
| "eval_loss": 0.30419921875, | |
| "eval_runtime": 1375.4073, | |
| "eval_samples_per_second": 3.367, | |
| "eval_steps_per_second": 0.421, | |
| "eval_wer": 16.237770530684852, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 22.74, | |
| "learning_rate": 4.38e-06, | |
| "loss": 0.0014, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 22.93, | |
| "learning_rate": 4.324444444444445e-06, | |
| "loss": 0.0013, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 23.12, | |
| "learning_rate": 4.268888888888889e-06, | |
| "loss": 0.0008, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 23.31, | |
| "learning_rate": 4.213333333333333e-06, | |
| "loss": 0.0007, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 23.49, | |
| "learning_rate": 4.157777777777778e-06, | |
| "loss": 0.0007, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 23.68, | |
| "learning_rate": 4.102222222222222e-06, | |
| "loss": 0.0006, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 23.87, | |
| "learning_rate": 4.046666666666667e-06, | |
| "loss": 0.0005, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 24.06, | |
| "learning_rate": 3.9888888888888895e-06, | |
| "loss": 0.0006, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 24.25, | |
| "learning_rate": 3.9333333333333335e-06, | |
| "loss": 0.0004, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 24.43, | |
| "learning_rate": 3.877777777777778e-06, | |
| "loss": 0.0004, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 24.62, | |
| "learning_rate": 3.8222222222222224e-06, | |
| "loss": 0.0004, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 24.81, | |
| "learning_rate": 3.766666666666667e-06, | |
| "loss": 0.0004, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 3.7111111111111113e-06, | |
| "loss": 0.0006, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 25.19, | |
| "learning_rate": 3.6555555555555562e-06, | |
| "loss": 0.0006, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 25.37, | |
| "learning_rate": 3.6000000000000003e-06, | |
| "loss": 0.0004, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 25.56, | |
| "learning_rate": 3.5444444444444447e-06, | |
| "loss": 0.0005, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 25.75, | |
| "learning_rate": 3.4888888888888896e-06, | |
| "loss": 0.0004, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 25.94, | |
| "learning_rate": 3.4333333333333336e-06, | |
| "loss": 0.0006, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 26.13, | |
| "learning_rate": 3.375555555555556e-06, | |
| "loss": 0.0006, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 26.31, | |
| "learning_rate": 3.3200000000000004e-06, | |
| "loss": 0.0004, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 26.5, | |
| "learning_rate": 3.2644444444444444e-06, | |
| "loss": 0.0004, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 26.69, | |
| "learning_rate": 3.2088888888888893e-06, | |
| "loss": 0.0004, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 26.88, | |
| "learning_rate": 3.1533333333333338e-06, | |
| "loss": 0.0004, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 27.07, | |
| "learning_rate": 3.097777777777778e-06, | |
| "loss": 0.0004, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 27.25, | |
| "learning_rate": 3.0422222222222227e-06, | |
| "loss": 0.0003, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 27.44, | |
| "learning_rate": 2.986666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 27.63, | |
| "learning_rate": 2.931111111111111e-06, | |
| "loss": 0.0003, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 27.82, | |
| "learning_rate": 2.875555555555556e-06, | |
| "loss": 0.0003, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 28.01, | |
| "learning_rate": 2.8177777777777784e-06, | |
| "loss": 0.0003, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 28.19, | |
| "learning_rate": 2.7622222222222224e-06, | |
| "loss": 0.0003, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 28.38, | |
| "learning_rate": 2.706666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 28.57, | |
| "learning_rate": 2.6511111111111113e-06, | |
| "loss": 0.0003, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 28.76, | |
| "learning_rate": 2.5955555555555558e-06, | |
| "loss": 0.0003, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 28.94, | |
| "learning_rate": 2.5400000000000002e-06, | |
| "loss": 0.0003, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 29.13, | |
| "learning_rate": 2.4844444444444447e-06, | |
| "loss": 0.0003, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 29.32, | |
| "learning_rate": 2.428888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 29.51, | |
| "learning_rate": 2.3733333333333336e-06, | |
| "loss": 0.0003, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 29.7, | |
| "learning_rate": 2.317777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 29.88, | |
| "learning_rate": 2.262222222222222e-06, | |
| "loss": 0.0003, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 30.07, | |
| "learning_rate": 2.2044444444444444e-06, | |
| "loss": 0.0003, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 30.07, | |
| "eval_loss": 0.32470703125, | |
| "eval_runtime": 1388.6023, | |
| "eval_samples_per_second": 3.335, | |
| "eval_steps_per_second": 0.417, | |
| "eval_wer": 15.594426326712126, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 30.26, | |
| "learning_rate": 2.148888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 30.45, | |
| "learning_rate": 2.0933333333333338e-06, | |
| "loss": 0.0003, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 30.64, | |
| "learning_rate": 2.037777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 30.82, | |
| "learning_rate": 1.9822222222222223e-06, | |
| "loss": 0.0003, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 31.01, | |
| "learning_rate": 1.926666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 31.2, | |
| "learning_rate": 1.8711111111111114e-06, | |
| "loss": 0.0002, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 31.39, | |
| "learning_rate": 1.8155555555555556e-06, | |
| "loss": 0.0002, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 31.58, | |
| "learning_rate": 1.76e-06, | |
| "loss": 0.0002, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 31.76, | |
| "learning_rate": 1.7044444444444448e-06, | |
| "loss": 0.0002, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 31.95, | |
| "learning_rate": 1.648888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 32.14, | |
| "learning_rate": 1.5911111111111113e-06, | |
| "loss": 0.0002, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 32.33, | |
| "learning_rate": 1.5355555555555558e-06, | |
| "loss": 0.0002, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 32.52, | |
| "learning_rate": 1.48e-06, | |
| "loss": 0.0002, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 32.7, | |
| "learning_rate": 1.4244444444444447e-06, | |
| "loss": 0.0002, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 32.89, | |
| "learning_rate": 1.3688888888888891e-06, | |
| "loss": 0.0002, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 33.08, | |
| "learning_rate": 1.3133333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 33.27, | |
| "learning_rate": 1.2577777777777779e-06, | |
| "loss": 0.0002, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 33.46, | |
| "learning_rate": 1.2022222222222223e-06, | |
| "loss": 0.0002, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 33.64, | |
| "learning_rate": 1.1466666666666668e-06, | |
| "loss": 0.0002, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 33.83, | |
| "learning_rate": 1.0911111111111112e-06, | |
| "loss": 0.0002, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 34.02, | |
| "learning_rate": 1.0333333333333333e-06, | |
| "loss": 0.0002, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 34.21, | |
| "learning_rate": 9.77777777777778e-07, | |
| "loss": 0.0002, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 34.4, | |
| "learning_rate": 9.222222222222222e-07, | |
| "loss": 0.0002, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 34.58, | |
| "learning_rate": 8.666666666666668e-07, | |
| "loss": 0.0002, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 34.77, | |
| "learning_rate": 8.111111111111112e-07, | |
| "loss": 0.0002, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 34.96, | |
| "learning_rate": 7.555555555555556e-07, | |
| "loss": 0.0002, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 35.15, | |
| "learning_rate": 7.000000000000001e-07, | |
| "loss": 0.0002, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 35.34, | |
| "learning_rate": 6.444444444444445e-07, | |
| "loss": 0.0002, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 35.52, | |
| "learning_rate": 5.888888888888889e-07, | |
| "loss": 0.0002, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 35.71, | |
| "learning_rate": 5.333333333333335e-07, | |
| "loss": 0.0002, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 35.9, | |
| "learning_rate": 4.777777777777778e-07, | |
| "loss": 0.0002, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 36.09, | |
| "learning_rate": 4.2000000000000006e-07, | |
| "loss": 0.0002, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 36.28, | |
| "learning_rate": 3.644444444444445e-07, | |
| "loss": 0.0002, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 36.46, | |
| "learning_rate": 3.088888888888889e-07, | |
| "loss": 0.0002, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 36.65, | |
| "learning_rate": 2.533333333333333e-07, | |
| "loss": 0.0002, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 36.84, | |
| "learning_rate": 1.9777777777777778e-07, | |
| "loss": 0.0002, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 37.03, | |
| "learning_rate": 1.4222222222222224e-07, | |
| "loss": 0.0002, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 37.22, | |
| "learning_rate": 8.666666666666668e-08, | |
| "loss": 0.0002, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 37.4, | |
| "learning_rate": 3.1111111111111114e-08, | |
| "loss": 0.0002, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 37.59, | |
| "learning_rate": 0.0, | |
| "loss": 0.0002, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 37.59, | |
| "eval_loss": 0.331298828125, | |
| "eval_runtime": 1391.4178, | |
| "eval_samples_per_second": 3.328, | |
| "eval_steps_per_second": 0.416, | |
| "eval_wer": 15.600355766380078, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 37.59, | |
| "step": 5000, | |
| "total_flos": 6.814572201925149e+20, | |
| "train_loss": 0.01643497463464737, | |
| "train_runtime": 64619.8702, | |
| "train_samples_per_second": 4.952, | |
| "train_steps_per_second": 0.077 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 38, | |
| "total_flos": 6.814572201925149e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |