diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,5008 @@ +{ + "best_metric": 4.89488506963824, + "best_model_checkpoint": "./checkpoint-16000", + "epoch": 40.018, + "eval_steps": 1000, + "global_step": 20000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 5.000000000000001e-07, + "loss": 0.8789, + "step": 25 + }, + { + "epoch": 0.0, + "learning_rate": 1.0000000000000002e-06, + "loss": 0.5133, + "step": 50 + }, + { + "epoch": 0.0, + "learning_rate": 1.46e-06, + "loss": 0.3076, + "step": 75 + }, + { + "epoch": 0.01, + "learning_rate": 1.9600000000000003e-06, + "loss": 0.2279, + "step": 100 + }, + { + "epoch": 0.01, + "learning_rate": 2.4400000000000004e-06, + "loss": 0.1953, + "step": 125 + }, + { + "epoch": 0.01, + "learning_rate": 2.92e-06, + "loss": 0.1793, + "step": 150 + }, + { + "epoch": 0.01, + "learning_rate": 3.4200000000000007e-06, + "loss": 0.1733, + "step": 175 + }, + { + "epoch": 0.01, + "learning_rate": 3.920000000000001e-06, + "loss": 0.1657, + "step": 200 + }, + { + "epoch": 0.01, + "learning_rate": 4.42e-06, + "loss": 0.1711, + "step": 225 + }, + { + "epoch": 0.01, + "learning_rate": 4.92e-06, + "loss": 0.2229, + "step": 250 + }, + { + "epoch": 0.01, + "learning_rate": 5.420000000000001e-06, + "loss": 0.2196, + "step": 275 + }, + { + "epoch": 0.01, + "learning_rate": 5.92e-06, + "loss": 0.2124, + "step": 300 + }, + { + "epoch": 0.02, + "learning_rate": 6.42e-06, + "loss": 0.1848, + "step": 325 + }, + { + "epoch": 0.02, + "learning_rate": 6.92e-06, + "loss": 0.1975, + "step": 350 + }, + { + "epoch": 0.02, + "learning_rate": 7.420000000000001e-06, + "loss": 0.2068, + "step": 375 + }, + { + "epoch": 0.02, + "learning_rate": 7.92e-06, + "loss": 0.1862, + "step": 400 + }, + { + "epoch": 0.02, + "learning_rate": 8.42e-06, + "loss": 0.1741, + "step": 425 + }, + { + "epoch": 0.02, + "learning_rate": 8.920000000000001e-06, + "loss": 0.1311, + "step": 450 + }, + { + "epoch": 0.02, + "learning_rate": 9.42e-06, + "loss": 0.13, + "step": 475 + }, + { + "epoch": 1.0, + "learning_rate": 9.920000000000002e-06, + "loss": 0.1473, + "step": 500 + }, + { + "epoch": 1.0, + "learning_rate": 9.98923076923077e-06, + "loss": 0.1572, + "step": 525 + }, + { + "epoch": 1.0, + "learning_rate": 9.976410256410257e-06, + "loss": 0.1913, + "step": 550 + }, + { + "epoch": 1.0, + "learning_rate": 9.963589743589744e-06, + "loss": 0.1376, + "step": 575 + }, + { + "epoch": 1.01, + "learning_rate": 9.950769230769232e-06, + "loss": 0.135, + "step": 600 + }, + { + "epoch": 1.01, + "learning_rate": 9.937948717948719e-06, + "loss": 0.1359, + "step": 625 + }, + { + "epoch": 1.01, + "learning_rate": 9.925128205128206e-06, + "loss": 0.1178, + "step": 650 + }, + { + "epoch": 1.01, + "learning_rate": 9.912307692307693e-06, + "loss": 0.1186, + "step": 675 + }, + { + "epoch": 1.01, + "learning_rate": 9.899487179487181e-06, + "loss": 0.112, + "step": 700 + }, + { + "epoch": 1.01, + "learning_rate": 9.886666666666668e-06, + "loss": 0.1491, + "step": 725 + }, + { + "epoch": 1.01, + "learning_rate": 9.873846153846155e-06, + "loss": 0.1401, + "step": 750 + }, + { + "epoch": 1.01, + "learning_rate": 9.861025641025642e-06, + "loss": 0.137, + "step": 775 + }, + { + "epoch": 1.02, + "learning_rate": 9.84820512820513e-06, + "loss": 0.1145, + "step": 800 + }, + { + "epoch": 1.02, + "learning_rate": 9.835384615384617e-06, + "loss": 0.1179, + "step": 825 + }, + { + "epoch": 1.02, + "learning_rate": 9.822564102564104e-06, + "loss": 0.1171, + "step": 850 + }, + { + "epoch": 1.02, + "learning_rate": 9.80974358974359e-06, + "loss": 0.112, + "step": 875 + }, + { + "epoch": 1.02, + "learning_rate": 9.796923076923077e-06, + "loss": 0.1043, + "step": 900 + }, + { + "epoch": 1.02, + "learning_rate": 9.784102564102564e-06, + "loss": 0.0795, + "step": 925 + }, + { + "epoch": 1.02, + "learning_rate": 9.771282051282051e-06, + "loss": 0.0674, + "step": 950 + }, + { + "epoch": 1.02, + "learning_rate": 9.75846153846154e-06, + "loss": 0.0696, + "step": 975 + }, + { + "epoch": 2.0, + "learning_rate": 9.745641025641026e-06, + "loss": 0.0869, + "step": 1000 + }, + { + "epoch": 2.0, + "eval_loss": 0.17535537481307983, + "eval_runtime": 5551.7218, + "eval_samples_per_second": 2.829, + "eval_steps_per_second": 0.177, + "eval_wer": 6.151645207439199, + "step": 1000 + }, + { + "epoch": 2.0, + "learning_rate": 9.732820512820513e-06, + "loss": 0.1121, + "step": 1025 + }, + { + "epoch": 2.0, + "learning_rate": 9.72e-06, + "loss": 0.1809, + "step": 1050 + }, + { + "epoch": 2.0, + "learning_rate": 9.707179487179487e-06, + "loss": 0.1442, + "step": 1075 + }, + { + "epoch": 2.01, + "learning_rate": 9.694358974358975e-06, + "loss": 0.1406, + "step": 1100 + }, + { + "epoch": 2.01, + "learning_rate": 9.681538461538462e-06, + "loss": 0.1117, + "step": 1125 + }, + { + "epoch": 2.01, + "learning_rate": 9.668717948717949e-06, + "loss": 0.1322, + "step": 1150 + }, + { + "epoch": 2.01, + "learning_rate": 9.655897435897436e-06, + "loss": 0.1308, + "step": 1175 + }, + { + "epoch": 2.01, + "learning_rate": 9.643076923076924e-06, + "loss": 0.1293, + "step": 1200 + }, + { + "epoch": 2.01, + "learning_rate": 9.630256410256411e-06, + "loss": 0.0922, + "step": 1225 + }, + { + "epoch": 2.01, + "learning_rate": 9.617435897435898e-06, + "loss": 0.0934, + "step": 1250 + }, + { + "epoch": 2.01, + "learning_rate": 9.604615384615385e-06, + "loss": 0.0987, + "step": 1275 + }, + { + "epoch": 2.02, + "learning_rate": 9.591794871794873e-06, + "loss": 0.1234, + "step": 1300 + }, + { + "epoch": 2.02, + "learning_rate": 9.57897435897436e-06, + "loss": 0.1491, + "step": 1325 + }, + { + "epoch": 2.02, + "learning_rate": 9.566153846153847e-06, + "loss": 0.1753, + "step": 1350 + }, + { + "epoch": 2.02, + "learning_rate": 9.553333333333334e-06, + "loss": 0.1211, + "step": 1375 + }, + { + "epoch": 2.02, + "learning_rate": 9.540512820512822e-06, + "loss": 0.0999, + "step": 1400 + }, + { + "epoch": 2.02, + "learning_rate": 9.52769230769231e-06, + "loss": 0.111, + "step": 1425 + }, + { + "epoch": 2.02, + "learning_rate": 9.514871794871796e-06, + "loss": 0.0911, + "step": 1450 + }, + { + "epoch": 3.0, + "learning_rate": 9.502051282051283e-06, + "loss": 0.0878, + "step": 1475 + }, + { + "epoch": 3.0, + "learning_rate": 9.48923076923077e-06, + "loss": 0.1315, + "step": 1500 + }, + { + "epoch": 3.0, + "learning_rate": 9.476410256410257e-06, + "loss": 0.1112, + "step": 1525 + }, + { + "epoch": 3.0, + "learning_rate": 9.463589743589743e-06, + "loss": 0.1069, + "step": 1550 + }, + { + "epoch": 3.01, + "learning_rate": 9.450769230769232e-06, + "loss": 0.1044, + "step": 1575 + }, + { + "epoch": 3.01, + "learning_rate": 9.437948717948719e-06, + "loss": 0.1024, + "step": 1600 + }, + { + "epoch": 3.01, + "learning_rate": 9.425128205128206e-06, + "loss": 0.0897, + "step": 1625 + }, + { + "epoch": 3.01, + "learning_rate": 9.412307692307692e-06, + "loss": 0.0987, + "step": 1650 + }, + { + "epoch": 3.01, + "learning_rate": 9.39948717948718e-06, + "loss": 0.1057, + "step": 1675 + }, + { + "epoch": 3.01, + "learning_rate": 9.386666666666668e-06, + "loss": 0.0963, + "step": 1700 + }, + { + "epoch": 3.01, + "learning_rate": 9.373846153846155e-06, + "loss": 0.0997, + "step": 1725 + }, + { + "epoch": 3.01, + "learning_rate": 9.361025641025641e-06, + "loss": 0.1037, + "step": 1750 + }, + { + "epoch": 3.02, + "learning_rate": 9.348205128205128e-06, + "loss": 0.1001, + "step": 1775 + }, + { + "epoch": 3.02, + "learning_rate": 9.335384615384617e-06, + "loss": 0.0931, + "step": 1800 + }, + { + "epoch": 3.02, + "learning_rate": 9.323076923076924e-06, + "loss": 0.1007, + "step": 1825 + }, + { + "epoch": 3.02, + "learning_rate": 9.310256410256411e-06, + "loss": 0.1022, + "step": 1850 + }, + { + "epoch": 3.02, + "learning_rate": 9.297435897435898e-06, + "loss": 0.1005, + "step": 1875 + }, + { + "epoch": 3.02, + "learning_rate": 9.284615384615385e-06, + "loss": 0.1015, + "step": 1900 + }, + { + "epoch": 3.02, + "learning_rate": 9.271794871794873e-06, + "loss": 0.0942, + "step": 1925 + }, + { + "epoch": 3.02, + "learning_rate": 9.25897435897436e-06, + "loss": 0.0934, + "step": 1950 + }, + { + "epoch": 4.0, + "learning_rate": 9.246153846153847e-06, + "loss": 0.0844, + "step": 1975 + }, + { + "epoch": 4.0, + "learning_rate": 9.233333333333334e-06, + "loss": 0.0913, + "step": 2000 + }, + { + "epoch": 4.0, + "eval_loss": 0.16517865657806396, + "eval_runtime": 5467.8276, + "eval_samples_per_second": 2.873, + "eval_steps_per_second": 0.18, + "eval_wer": 5.750046511031134, + "step": 2000 + }, + { + "epoch": 4.0, + "learning_rate": 9.220512820512822e-06, + "loss": 0.102, + "step": 2025 + }, + { + "epoch": 4.0, + "learning_rate": 9.207692307692309e-06, + "loss": 0.095, + "step": 2050 + }, + { + "epoch": 4.01, + "learning_rate": 9.194871794871796e-06, + "loss": 0.0915, + "step": 2075 + }, + { + "epoch": 4.01, + "learning_rate": 9.182051282051283e-06, + "loss": 0.0992, + "step": 2100 + }, + { + "epoch": 4.01, + "learning_rate": 9.169230769230771e-06, + "loss": 0.0808, + "step": 2125 + }, + { + "epoch": 4.01, + "learning_rate": 9.156410256410258e-06, + "loss": 0.0876, + "step": 2150 + }, + { + "epoch": 4.01, + "learning_rate": 9.143589743589745e-06, + "loss": 0.0961, + "step": 2175 + }, + { + "epoch": 4.01, + "learning_rate": 9.130769230769232e-06, + "loss": 0.101, + "step": 2200 + }, + { + "epoch": 4.01, + "learning_rate": 9.117948717948718e-06, + "loss": 0.0943, + "step": 2225 + }, + { + "epoch": 4.01, + "learning_rate": 9.105128205128207e-06, + "loss": 0.0875, + "step": 2250 + }, + { + "epoch": 4.02, + "learning_rate": 9.092307692307692e-06, + "loss": 0.0971, + "step": 2275 + }, + { + "epoch": 4.02, + "learning_rate": 9.07948717948718e-06, + "loss": 0.0972, + "step": 2300 + }, + { + "epoch": 4.02, + "learning_rate": 9.066666666666667e-06, + "loss": 0.0891, + "step": 2325 + }, + { + "epoch": 4.02, + "learning_rate": 9.053846153846154e-06, + "loss": 0.1007, + "step": 2350 + }, + { + "epoch": 4.02, + "learning_rate": 9.041025641025641e-06, + "loss": 0.0942, + "step": 2375 + }, + { + "epoch": 4.02, + "learning_rate": 9.028205128205128e-06, + "loss": 0.091, + "step": 2400 + }, + { + "epoch": 4.02, + "learning_rate": 9.015384615384616e-06, + "loss": 0.0808, + "step": 2425 + }, + { + "epoch": 4.02, + "learning_rate": 9.002564102564103e-06, + "loss": 0.0827, + "step": 2450 + }, + { + "epoch": 5.0, + "learning_rate": 8.98974358974359e-06, + "loss": 0.0853, + "step": 2475 + }, + { + "epoch": 5.0, + "learning_rate": 8.976923076923077e-06, + "loss": 0.0839, + "step": 2500 + }, + { + "epoch": 5.0, + "learning_rate": 8.964102564102565e-06, + "loss": 0.1201, + "step": 2525 + }, + { + "epoch": 5.0, + "learning_rate": 8.951282051282052e-06, + "loss": 0.1055, + "step": 2550 + }, + { + "epoch": 5.01, + "learning_rate": 8.938461538461539e-06, + "loss": 0.0829, + "step": 2575 + }, + { + "epoch": 5.01, + "learning_rate": 8.925641025641026e-06, + "loss": 0.0817, + "step": 2600 + }, + { + "epoch": 5.01, + "learning_rate": 8.912820512820514e-06, + "loss": 0.073, + "step": 2625 + }, + { + "epoch": 5.01, + "learning_rate": 8.900000000000001e-06, + "loss": 0.1087, + "step": 2650 + }, + { + "epoch": 5.01, + "learning_rate": 8.887179487179488e-06, + "loss": 0.0947, + "step": 2675 + }, + { + "epoch": 5.01, + "learning_rate": 8.874358974358975e-06, + "loss": 0.077, + "step": 2700 + }, + { + "epoch": 5.01, + "learning_rate": 8.861538461538463e-06, + "loss": 0.0874, + "step": 2725 + }, + { + "epoch": 5.01, + "learning_rate": 8.84871794871795e-06, + "loss": 0.0842, + "step": 2750 + }, + { + "epoch": 5.02, + "learning_rate": 8.835897435897437e-06, + "loss": 0.0709, + "step": 2775 + }, + { + "epoch": 5.02, + "learning_rate": 8.823076923076924e-06, + "loss": 0.0946, + "step": 2800 + }, + { + "epoch": 5.02, + "learning_rate": 8.81025641025641e-06, + "loss": 0.2053, + "step": 2825 + }, + { + "epoch": 5.02, + "learning_rate": 8.7974358974359e-06, + "loss": 0.1099, + "step": 2850 + }, + { + "epoch": 5.02, + "learning_rate": 8.784615384615386e-06, + "loss": 0.1465, + "step": 2875 + }, + { + "epoch": 5.02, + "learning_rate": 8.771794871794873e-06, + "loss": 0.1039, + "step": 2900 + }, + { + "epoch": 5.02, + "learning_rate": 8.75897435897436e-06, + "loss": 0.1057, + "step": 2925 + }, + { + "epoch": 6.0, + "learning_rate": 8.746153846153846e-06, + "loss": 0.1405, + "step": 2950 + }, + { + "epoch": 6.0, + "learning_rate": 8.733333333333333e-06, + "loss": 0.0535, + "step": 2975 + }, + { + "epoch": 6.0, + "learning_rate": 8.72051282051282e-06, + "loss": 0.051, + "step": 3000 + }, + { + "epoch": 6.0, + "eval_loss": 0.1643298864364624, + "eval_runtime": 5531.8097, + "eval_samples_per_second": 2.84, + "eval_steps_per_second": 0.178, + "eval_wer": 5.77570776958756, + "step": 3000 + }, + { + "epoch": 6.0, + "learning_rate": 8.707692307692309e-06, + "loss": 0.0518, + "step": 3025 + }, + { + "epoch": 6.01, + "learning_rate": 8.694871794871795e-06, + "loss": 0.0435, + "step": 3050 + }, + { + "epoch": 6.01, + "learning_rate": 8.682051282051282e-06, + "loss": 0.0453, + "step": 3075 + }, + { + "epoch": 6.01, + "learning_rate": 8.669230769230769e-06, + "loss": 0.0431, + "step": 3100 + }, + { + "epoch": 6.01, + "learning_rate": 8.656410256410258e-06, + "loss": 0.0474, + "step": 3125 + }, + { + "epoch": 6.01, + "learning_rate": 8.643589743589744e-06, + "loss": 0.0422, + "step": 3150 + }, + { + "epoch": 6.01, + "learning_rate": 8.630769230769231e-06, + "loss": 0.0437, + "step": 3175 + }, + { + "epoch": 6.01, + "learning_rate": 8.617948717948718e-06, + "loss": 0.0443, + "step": 3200 + }, + { + "epoch": 6.01, + "learning_rate": 8.605128205128207e-06, + "loss": 0.0468, + "step": 3225 + }, + { + "epoch": 6.02, + "learning_rate": 8.592307692307693e-06, + "loss": 0.0404, + "step": 3250 + }, + { + "epoch": 6.02, + "learning_rate": 8.57948717948718e-06, + "loss": 0.0415, + "step": 3275 + }, + { + "epoch": 6.02, + "learning_rate": 8.566666666666667e-06, + "loss": 0.0483, + "step": 3300 + }, + { + "epoch": 6.02, + "learning_rate": 8.553846153846156e-06, + "loss": 0.0433, + "step": 3325 + }, + { + "epoch": 6.02, + "learning_rate": 8.541025641025642e-06, + "loss": 0.0506, + "step": 3350 + }, + { + "epoch": 6.02, + "learning_rate": 8.52820512820513e-06, + "loss": 0.0428, + "step": 3375 + }, + { + "epoch": 6.02, + "learning_rate": 8.515384615384616e-06, + "loss": 0.0393, + "step": 3400 + }, + { + "epoch": 6.02, + "learning_rate": 8.502564102564103e-06, + "loss": 0.0412, + "step": 3425 + }, + { + "epoch": 7.0, + "learning_rate": 8.489743589743591e-06, + "loss": 0.0502, + "step": 3450 + }, + { + "epoch": 7.0, + "learning_rate": 8.476923076923078e-06, + "loss": 0.0387, + "step": 3475 + }, + { + "epoch": 7.0, + "learning_rate": 8.464102564102565e-06, + "loss": 0.0835, + "step": 3500 + }, + { + "epoch": 7.0, + "learning_rate": 8.451282051282052e-06, + "loss": 0.0687, + "step": 3525 + }, + { + "epoch": 7.01, + "learning_rate": 8.438461538461539e-06, + "loss": 0.055, + "step": 3550 + }, + { + "epoch": 7.01, + "learning_rate": 8.425641025641026e-06, + "loss": 0.0553, + "step": 3575 + }, + { + "epoch": 7.01, + "learning_rate": 8.412820512820512e-06, + "loss": 0.0553, + "step": 3600 + }, + { + "epoch": 7.01, + "learning_rate": 8.400000000000001e-06, + "loss": 0.06, + "step": 3625 + }, + { + "epoch": 7.01, + "learning_rate": 8.387179487179488e-06, + "loss": 0.0598, + "step": 3650 + }, + { + "epoch": 7.01, + "learning_rate": 8.374358974358975e-06, + "loss": 0.0387, + "step": 3675 + }, + { + "epoch": 7.01, + "learning_rate": 8.361538461538461e-06, + "loss": 0.036, + "step": 3700 + }, + { + "epoch": 7.01, + "learning_rate": 8.34871794871795e-06, + "loss": 0.0375, + "step": 3725 + }, + { + "epoch": 7.02, + "learning_rate": 8.335897435897437e-06, + "loss": 0.0516, + "step": 3750 + }, + { + "epoch": 7.02, + "learning_rate": 8.323076923076924e-06, + "loss": 0.064, + "step": 3775 + }, + { + "epoch": 7.02, + "learning_rate": 8.31025641025641e-06, + "loss": 0.0881, + "step": 3800 + }, + { + "epoch": 7.02, + "learning_rate": 8.297435897435899e-06, + "loss": 0.0538, + "step": 3825 + }, + { + "epoch": 7.02, + "learning_rate": 8.284615384615386e-06, + "loss": 0.0407, + "step": 3850 + }, + { + "epoch": 7.02, + "learning_rate": 8.271794871794873e-06, + "loss": 0.045, + "step": 3875 + }, + { + "epoch": 7.02, + "learning_rate": 8.25897435897436e-06, + "loss": 0.0373, + "step": 3900 + }, + { + "epoch": 7.02, + "learning_rate": 8.246153846153848e-06, + "loss": 0.0342, + "step": 3925 + }, + { + "epoch": 8.0, + "learning_rate": 8.233333333333335e-06, + "loss": 0.0199, + "step": 3950 + }, + { + "epoch": 8.0, + "learning_rate": 8.220512820512822e-06, + "loss": 0.0207, + "step": 3975 + }, + { + "epoch": 8.0, + "learning_rate": 8.207692307692308e-06, + "loss": 0.0391, + "step": 4000 + }, + { + "epoch": 8.0, + "eval_loss": 0.18808600306510925, + "eval_runtime": 5575.6316, + "eval_samples_per_second": 2.817, + "eval_steps_per_second": 0.176, + "eval_wer": 5.658949043155822, + "step": 4000 + }, + { + "epoch": 8.0, + "learning_rate": 8.194871794871795e-06, + "loss": 0.0257, + "step": 4025 + }, + { + "epoch": 8.01, + "learning_rate": 8.182051282051284e-06, + "loss": 0.0241, + "step": 4050 + }, + { + "epoch": 8.01, + "learning_rate": 8.16923076923077e-06, + "loss": 0.0198, + "step": 4075 + }, + { + "epoch": 8.01, + "learning_rate": 8.156410256410257e-06, + "loss": 0.0221, + "step": 4100 + }, + { + "epoch": 8.01, + "learning_rate": 8.143589743589744e-06, + "loss": 0.0228, + "step": 4125 + }, + { + "epoch": 8.01, + "learning_rate": 8.130769230769231e-06, + "loss": 0.019, + "step": 4150 + }, + { + "epoch": 8.01, + "learning_rate": 8.117948717948718e-06, + "loss": 0.0149, + "step": 4175 + }, + { + "epoch": 8.01, + "learning_rate": 8.105128205128205e-06, + "loss": 0.0141, + "step": 4200 + }, + { + "epoch": 8.01, + "learning_rate": 8.092307692307693e-06, + "loss": 0.0162, + "step": 4225 + }, + { + "epoch": 8.02, + "learning_rate": 8.07948717948718e-06, + "loss": 0.0217, + "step": 4250 + }, + { + "epoch": 8.02, + "learning_rate": 8.066666666666667e-06, + "loss": 0.0266, + "step": 4275 + }, + { + "epoch": 8.02, + "learning_rate": 8.053846153846154e-06, + "loss": 0.0345, + "step": 4300 + }, + { + "epoch": 8.02, + "learning_rate": 8.041025641025642e-06, + "loss": 0.0195, + "step": 4325 + }, + { + "epoch": 8.02, + "learning_rate": 8.028205128205129e-06, + "loss": 0.0167, + "step": 4350 + }, + { + "epoch": 8.02, + "learning_rate": 8.015384615384616e-06, + "loss": 0.0173, + "step": 4375 + }, + { + "epoch": 8.02, + "learning_rate": 8.002564102564103e-06, + "loss": 0.0132, + "step": 4400 + }, + { + "epoch": 9.0, + "learning_rate": 7.989743589743591e-06, + "loss": 0.0169, + "step": 4425 + }, + { + "epoch": 9.0, + "learning_rate": 7.976923076923078e-06, + "loss": 0.027, + "step": 4450 + }, + { + "epoch": 9.0, + "learning_rate": 7.964102564102565e-06, + "loss": 0.0238, + "step": 4475 + }, + { + "epoch": 9.0, + "learning_rate": 7.951282051282052e-06, + "loss": 0.0238, + "step": 4500 + }, + { + "epoch": 9.01, + "learning_rate": 7.93846153846154e-06, + "loss": 0.0245, + "step": 4525 + }, + { + "epoch": 9.01, + "learning_rate": 7.925641025641027e-06, + "loss": 0.0191, + "step": 4550 + }, + { + "epoch": 9.01, + "learning_rate": 7.912820512820514e-06, + "loss": 0.0188, + "step": 4575 + }, + { + "epoch": 9.01, + "learning_rate": 7.9e-06, + "loss": 0.0215, + "step": 4600 + }, + { + "epoch": 9.01, + "learning_rate": 7.887179487179487e-06, + "loss": 0.019, + "step": 4625 + }, + { + "epoch": 9.01, + "learning_rate": 7.874358974358976e-06, + "loss": 0.0215, + "step": 4650 + }, + { + "epoch": 9.01, + "learning_rate": 7.861538461538463e-06, + "loss": 0.0197, + "step": 4675 + }, + { + "epoch": 9.01, + "learning_rate": 7.84871794871795e-06, + "loss": 0.0206, + "step": 4700 + }, + { + "epoch": 9.02, + "learning_rate": 7.835897435897436e-06, + "loss": 0.0178, + "step": 4725 + }, + { + "epoch": 9.02, + "learning_rate": 7.823076923076923e-06, + "loss": 0.0201, + "step": 4750 + }, + { + "epoch": 9.02, + "learning_rate": 7.81025641025641e-06, + "loss": 0.0211, + "step": 4775 + }, + { + "epoch": 9.02, + "learning_rate": 7.797435897435897e-06, + "loss": 0.0189, + "step": 4800 + }, + { + "epoch": 9.02, + "learning_rate": 7.784615384615385e-06, + "loss": 0.0217, + "step": 4825 + }, + { + "epoch": 9.02, + "learning_rate": 7.771794871794872e-06, + "loss": 0.0195, + "step": 4850 + }, + { + "epoch": 9.02, + "learning_rate": 7.758974358974359e-06, + "loss": 0.0192, + "step": 4875 + }, + { + "epoch": 9.02, + "learning_rate": 7.746153846153846e-06, + "loss": 0.0203, + "step": 4900 + }, + { + "epoch": 10.0, + "learning_rate": 7.733333333333334e-06, + "loss": 0.0183, + "step": 4925 + }, + { + "epoch": 10.0, + "learning_rate": 7.720512820512821e-06, + "loss": 0.0114, + "step": 4950 + }, + { + "epoch": 10.0, + "learning_rate": 7.707692307692308e-06, + "loss": 0.0112, + "step": 4975 + }, + { + "epoch": 10.0, + "learning_rate": 7.694871794871795e-06, + "loss": 0.0104, + "step": 5000 + }, + { + "epoch": 10.0, + "eval_loss": 0.20255373418331146, + "eval_runtime": 5586.6732, + "eval_samples_per_second": 2.812, + "eval_steps_per_second": 0.176, + "eval_wer": 5.621098686785094, + "step": 5000 + }, + { + "epoch": 10.01, + "learning_rate": 7.682051282051283e-06, + "loss": 0.0096, + "step": 5025 + }, + { + "epoch": 10.01, + "learning_rate": 7.66923076923077e-06, + "loss": 0.0076, + "step": 5050 + }, + { + "epoch": 10.01, + "learning_rate": 7.656410256410257e-06, + "loss": 0.0097, + "step": 5075 + }, + { + "epoch": 10.01, + "learning_rate": 7.643589743589744e-06, + "loss": 0.0091, + "step": 5100 + }, + { + "epoch": 10.01, + "learning_rate": 7.630769230769232e-06, + "loss": 0.0076, + "step": 5125 + }, + { + "epoch": 10.01, + "learning_rate": 7.617948717948719e-06, + "loss": 0.009, + "step": 5150 + }, + { + "epoch": 10.01, + "learning_rate": 7.605128205128206e-06, + "loss": 0.0095, + "step": 5175 + }, + { + "epoch": 10.01, + "learning_rate": 7.592307692307693e-06, + "loss": 0.0098, + "step": 5200 + }, + { + "epoch": 10.02, + "learning_rate": 7.5794871794871805e-06, + "loss": 0.0078, + "step": 5225 + }, + { + "epoch": 10.02, + "learning_rate": 7.566666666666667e-06, + "loss": 0.0089, + "step": 5250 + }, + { + "epoch": 10.02, + "learning_rate": 7.553846153846155e-06, + "loss": 0.0092, + "step": 5275 + }, + { + "epoch": 10.02, + "learning_rate": 7.541025641025642e-06, + "loss": 0.0083, + "step": 5300 + }, + { + "epoch": 10.02, + "learning_rate": 7.528205128205129e-06, + "loss": 0.0092, + "step": 5325 + }, + { + "epoch": 10.02, + "learning_rate": 7.5153846153846155e-06, + "loss": 0.0108, + "step": 5350 + }, + { + "epoch": 10.02, + "learning_rate": 7.502564102564102e-06, + "loss": 0.009, + "step": 5375 + }, + { + "epoch": 10.02, + "learning_rate": 7.48974358974359e-06, + "loss": 0.013, + "step": 5400 + }, + { + "epoch": 11.0, + "learning_rate": 7.476923076923077e-06, + "loss": 0.01, + "step": 5425 + }, + { + "epoch": 11.0, + "learning_rate": 7.4641025641025645e-06, + "loss": 0.0149, + "step": 5450 + }, + { + "epoch": 11.0, + "learning_rate": 7.451282051282051e-06, + "loss": 0.0211, + "step": 5475 + }, + { + "epoch": 11.0, + "learning_rate": 7.438461538461539e-06, + "loss": 0.0146, + "step": 5500 + }, + { + "epoch": 11.01, + "learning_rate": 7.425641025641026e-06, + "loss": 0.0146, + "step": 5525 + }, + { + "epoch": 11.01, + "learning_rate": 7.4128205128205135e-06, + "loss": 0.0099, + "step": 5550 + }, + { + "epoch": 11.01, + "learning_rate": 7.4e-06, + "loss": 0.0132, + "step": 5575 + }, + { + "epoch": 11.01, + "learning_rate": 7.387179487179488e-06, + "loss": 0.0137, + "step": 5600 + }, + { + "epoch": 11.01, + "learning_rate": 7.374358974358975e-06, + "loss": 0.0124, + "step": 5625 + }, + { + "epoch": 11.01, + "learning_rate": 7.3615384615384625e-06, + "loss": 0.009, + "step": 5650 + }, + { + "epoch": 11.01, + "learning_rate": 7.348717948717949e-06, + "loss": 0.0096, + "step": 5675 + }, + { + "epoch": 11.01, + "learning_rate": 7.335897435897437e-06, + "loss": 0.0108, + "step": 5700 + }, + { + "epoch": 11.02, + "learning_rate": 7.323076923076924e-06, + "loss": 0.013, + "step": 5725 + }, + { + "epoch": 11.02, + "learning_rate": 7.3102564102564115e-06, + "loss": 0.0158, + "step": 5750 + }, + { + "epoch": 11.02, + "learning_rate": 7.297435897435898e-06, + "loss": 0.0192, + "step": 5775 + }, + { + "epoch": 11.02, + "learning_rate": 7.284615384615386e-06, + "loss": 0.0118, + "step": 5800 + }, + { + "epoch": 11.02, + "learning_rate": 7.271794871794873e-06, + "loss": 0.0101, + "step": 5825 + }, + { + "epoch": 11.02, + "learning_rate": 7.25897435897436e-06, + "loss": 0.0107, + "step": 5850 + }, + { + "epoch": 11.02, + "learning_rate": 7.246153846153847e-06, + "loss": 0.0094, + "step": 5875 + }, + { + "epoch": 12.0, + "learning_rate": 7.233333333333334e-06, + "loss": 0.0365, + "step": 5900 + }, + { + "epoch": 12.0, + "learning_rate": 7.220512820512822e-06, + "loss": 0.0882, + "step": 5925 + }, + { + "epoch": 12.0, + "learning_rate": 7.207692307692308e-06, + "loss": 0.0847, + "step": 5950 + }, + { + "epoch": 12.0, + "learning_rate": 7.1948717948717946e-06, + "loss": 0.0855, + "step": 5975 + }, + { + "epoch": 12.01, + "learning_rate": 7.182051282051282e-06, + "loss": 0.0806, + "step": 6000 + }, + { + "epoch": 12.01, + "eval_loss": 0.1740662008523941, + "eval_runtime": 5603.9405, + "eval_samples_per_second": 2.803, + "eval_steps_per_second": 0.175, + "eval_wer": 5.7397820076085635, + "step": 6000 + }, + { + "epoch": 12.01, + "learning_rate": 7.169230769230769e-06, + "loss": 0.0761, + "step": 6025 + }, + { + "epoch": 12.01, + "learning_rate": 7.156410256410257e-06, + "loss": 0.0816, + "step": 6050 + }, + { + "epoch": 12.01, + "learning_rate": 7.1435897435897436e-06, + "loss": 0.0872, + "step": 6075 + }, + { + "epoch": 12.01, + "learning_rate": 7.130769230769231e-06, + "loss": 0.0726, + "step": 6100 + }, + { + "epoch": 12.01, + "learning_rate": 7.117948717948718e-06, + "loss": 0.0835, + "step": 6125 + }, + { + "epoch": 12.01, + "learning_rate": 7.105128205128206e-06, + "loss": 0.0687, + "step": 6150 + }, + { + "epoch": 12.01, + "learning_rate": 7.0923076923076926e-06, + "loss": 0.0693, + "step": 6175 + }, + { + "epoch": 12.02, + "learning_rate": 7.07948717948718e-06, + "loss": 0.0724, + "step": 6200 + }, + { + "epoch": 12.02, + "learning_rate": 7.066666666666667e-06, + "loss": 0.0787, + "step": 6225 + }, + { + "epoch": 12.02, + "learning_rate": 7.053846153846155e-06, + "loss": 0.0828, + "step": 6250 + }, + { + "epoch": 12.02, + "learning_rate": 7.0410256410256415e-06, + "loss": 0.0705, + "step": 6275 + }, + { + "epoch": 12.02, + "learning_rate": 7.028205128205129e-06, + "loss": 0.0862, + "step": 6300 + }, + { + "epoch": 12.02, + "learning_rate": 7.015384615384616e-06, + "loss": 0.092, + "step": 6325 + }, + { + "epoch": 12.02, + "learning_rate": 7.002564102564104e-06, + "loss": 0.1257, + "step": 6350 + }, + { + "epoch": 12.02, + "learning_rate": 6.9897435897435905e-06, + "loss": 0.0776, + "step": 6375 + }, + { + "epoch": 13.0, + "learning_rate": 6.976923076923078e-06, + "loss": 0.035, + "step": 6400 + }, + { + "epoch": 13.0, + "learning_rate": 6.964102564102565e-06, + "loss": 0.0101, + "step": 6425 + }, + { + "epoch": 13.0, + "learning_rate": 6.951282051282052e-06, + "loss": 0.0152, + "step": 6450 + }, + { + "epoch": 13.0, + "learning_rate": 6.9384615384615395e-06, + "loss": 0.0099, + "step": 6475 + }, + { + "epoch": 13.01, + "learning_rate": 6.925641025641026e-06, + "loss": 0.0087, + "step": 6500 + }, + { + "epoch": 13.01, + "learning_rate": 6.912820512820514e-06, + "loss": 0.0083, + "step": 6525 + }, + { + "epoch": 13.01, + "learning_rate": 6.9e-06, + "loss": 0.0073, + "step": 6550 + }, + { + "epoch": 13.01, + "learning_rate": 6.887179487179488e-06, + "loss": 0.0085, + "step": 6575 + }, + { + "epoch": 13.01, + "learning_rate": 6.8743589743589745e-06, + "loss": 0.0079, + "step": 6600 + }, + { + "epoch": 13.01, + "learning_rate": 6.861538461538461e-06, + "loss": 0.0056, + "step": 6625 + }, + { + "epoch": 13.01, + "learning_rate": 6.848717948717949e-06, + "loss": 0.0055, + "step": 6650 + }, + { + "epoch": 13.01, + "learning_rate": 6.835897435897436e-06, + "loss": 0.006, + "step": 6675 + }, + { + "epoch": 13.02, + "learning_rate": 6.8230769230769235e-06, + "loss": 0.0073, + "step": 6700 + }, + { + "epoch": 13.02, + "learning_rate": 6.81025641025641e-06, + "loss": 0.0086, + "step": 6725 + }, + { + "epoch": 13.02, + "learning_rate": 6.797435897435898e-06, + "loss": 0.0102, + "step": 6750 + }, + { + "epoch": 13.02, + "learning_rate": 6.784615384615385e-06, + "loss": 0.0075, + "step": 6775 + }, + { + "epoch": 13.02, + "learning_rate": 6.7717948717948725e-06, + "loss": 0.0051, + "step": 6800 + }, + { + "epoch": 13.02, + "learning_rate": 6.758974358974359e-06, + "loss": 0.0069, + "step": 6825 + }, + { + "epoch": 13.02, + "learning_rate": 6.746153846153847e-06, + "loss": 0.0057, + "step": 6850 + }, + { + "epoch": 14.0, + "learning_rate": 6.733333333333334e-06, + "loss": 0.005, + "step": 6875 + }, + { + "epoch": 14.0, + "learning_rate": 6.7205128205128215e-06, + "loss": 0.0078, + "step": 6900 + }, + { + "epoch": 14.0, + "learning_rate": 6.707692307692308e-06, + "loss": 0.008, + "step": 6925 + }, + { + "epoch": 14.0, + "learning_rate": 6.694871794871796e-06, + "loss": 0.009, + "step": 6950 + }, + { + "epoch": 14.01, + "learning_rate": 6.682051282051283e-06, + "loss": 0.0075, + "step": 6975 + }, + { + "epoch": 14.01, + "learning_rate": 6.6692307692307705e-06, + "loss": 0.0077, + "step": 7000 + }, + { + "epoch": 14.01, + "eval_loss": 0.2119489461183548, + "eval_runtime": 5589.6, + "eval_samples_per_second": 2.81, + "eval_steps_per_second": 0.176, + "eval_wer": 5.603777337259506, + "step": 7000 + }, + { + "epoch": 14.01, + "learning_rate": 6.656410256410257e-06, + "loss": 0.0066, + "step": 7025 + }, + { + "epoch": 14.01, + "learning_rate": 6.643589743589744e-06, + "loss": 0.0058, + "step": 7050 + }, + { + "epoch": 14.01, + "learning_rate": 6.630769230769232e-06, + "loss": 0.0073, + "step": 7075 + }, + { + "epoch": 14.01, + "learning_rate": 6.617948717948719e-06, + "loss": 0.0217, + "step": 7100 + }, + { + "epoch": 14.01, + "learning_rate": 6.605128205128206e-06, + "loss": 0.0061, + "step": 7125 + }, + { + "epoch": 14.01, + "learning_rate": 6.592820512820514e-06, + "loss": 0.0062, + "step": 7150 + }, + { + "epoch": 14.02, + "learning_rate": 6.5800000000000005e-06, + "loss": 0.0063, + "step": 7175 + }, + { + "epoch": 14.02, + "learning_rate": 6.567179487179488e-06, + "loss": 0.0064, + "step": 7200 + }, + { + "epoch": 14.02, + "learning_rate": 6.554358974358975e-06, + "loss": 0.0075, + "step": 7225 + }, + { + "epoch": 14.02, + "learning_rate": 6.541538461538463e-06, + "loss": 0.0071, + "step": 7250 + }, + { + "epoch": 14.02, + "learning_rate": 6.5287179487179495e-06, + "loss": 0.0076, + "step": 7275 + }, + { + "epoch": 14.02, + "learning_rate": 6.515897435897437e-06, + "loss": 0.0079, + "step": 7300 + }, + { + "epoch": 14.02, + "learning_rate": 6.503076923076923e-06, + "loss": 0.0071, + "step": 7325 + }, + { + "epoch": 14.02, + "learning_rate": 6.49025641025641e-06, + "loss": 0.007, + "step": 7350 + }, + { + "epoch": 15.0, + "learning_rate": 6.477435897435898e-06, + "loss": 0.026, + "step": 7375 + }, + { + "epoch": 15.0, + "learning_rate": 6.4646153846153845e-06, + "loss": 0.0341, + "step": 7400 + }, + { + "epoch": 15.0, + "learning_rate": 6.451794871794872e-06, + "loss": 0.0605, + "step": 7425 + }, + { + "epoch": 15.0, + "learning_rate": 6.438974358974359e-06, + "loss": 0.0453, + "step": 7450 + }, + { + "epoch": 15.01, + "learning_rate": 6.426153846153847e-06, + "loss": 0.0486, + "step": 7475 + }, + { + "epoch": 15.01, + "learning_rate": 6.4133333333333335e-06, + "loss": 0.0363, + "step": 7500 + }, + { + "epoch": 15.01, + "learning_rate": 6.400512820512821e-06, + "loss": 0.0309, + "step": 7525 + }, + { + "epoch": 15.01, + "learning_rate": 6.387692307692308e-06, + "loss": 0.0463, + "step": 7550 + }, + { + "epoch": 15.01, + "learning_rate": 6.374871794871796e-06, + "loss": 0.0478, + "step": 7575 + }, + { + "epoch": 15.01, + "learning_rate": 6.3620512820512825e-06, + "loss": 0.0361, + "step": 7600 + }, + { + "epoch": 15.01, + "learning_rate": 6.34923076923077e-06, + "loss": 0.0362, + "step": 7625 + }, + { + "epoch": 15.01, + "learning_rate": 6.336410256410257e-06, + "loss": 0.0423, + "step": 7650 + }, + { + "epoch": 15.02, + "learning_rate": 6.323589743589744e-06, + "loss": 0.0355, + "step": 7675 + }, + { + "epoch": 15.02, + "learning_rate": 6.3107692307692315e-06, + "loss": 0.0449, + "step": 7700 + }, + { + "epoch": 15.02, + "learning_rate": 6.297948717948718e-06, + "loss": 0.0793, + "step": 7725 + }, + { + "epoch": 15.02, + "learning_rate": 6.285128205128206e-06, + "loss": 0.0824, + "step": 7750 + }, + { + "epoch": 15.02, + "learning_rate": 6.272307692307693e-06, + "loss": 0.0573, + "step": 7775 + }, + { + "epoch": 15.02, + "learning_rate": 6.2594871794871805e-06, + "loss": 0.0756, + "step": 7800 + }, + { + "epoch": 15.02, + "learning_rate": 6.246666666666667e-06, + "loss": 0.037, + "step": 7825 + }, + { + "epoch": 15.02, + "learning_rate": 6.233846153846155e-06, + "loss": 0.0919, + "step": 7850 + }, + { + "epoch": 16.0, + "learning_rate": 6.221025641025642e-06, + "loss": 0.0552, + "step": 7875 + }, + { + "epoch": 16.0, + "learning_rate": 6.2082051282051294e-06, + "loss": 0.0442, + "step": 7900 + }, + { + "epoch": 16.0, + "learning_rate": 6.195384615384616e-06, + "loss": 0.0458, + "step": 7925 + }, + { + "epoch": 16.0, + "learning_rate": 6.182564102564102e-06, + "loss": 0.0418, + "step": 7950 + }, + { + "epoch": 16.01, + "learning_rate": 6.16974358974359e-06, + "loss": 0.0436, + "step": 7975 + }, + { + "epoch": 16.01, + "learning_rate": 6.156923076923077e-06, + "loss": 0.0357, + "step": 8000 + }, + { + "epoch": 16.01, + "eval_loss": 0.17759384214878082, + "eval_runtime": 5620.562, + "eval_samples_per_second": 2.795, + "eval_steps_per_second": 0.175, + "eval_wer": 5.614683372145987, + "step": 8000 + }, + { + "epoch": 16.01, + "learning_rate": 6.144102564102564e-06, + "loss": 0.0373, + "step": 8025 + }, + { + "epoch": 16.01, + "learning_rate": 6.131282051282051e-06, + "loss": 0.0417, + "step": 8050 + }, + { + "epoch": 16.01, + "learning_rate": 6.118461538461539e-06, + "loss": 0.0454, + "step": 8075 + }, + { + "epoch": 16.01, + "learning_rate": 6.105641025641026e-06, + "loss": 0.0403, + "step": 8100 + }, + { + "epoch": 16.01, + "learning_rate": 6.092820512820513e-06, + "loss": 0.0428, + "step": 8125 + }, + { + "epoch": 16.01, + "learning_rate": 6.08e-06, + "loss": 0.0417, + "step": 8150 + }, + { + "epoch": 16.02, + "learning_rate": 6.067179487179488e-06, + "loss": 0.0424, + "step": 8175 + }, + { + "epoch": 16.02, + "learning_rate": 6.054358974358975e-06, + "loss": 0.0409, + "step": 8200 + }, + { + "epoch": 16.02, + "learning_rate": 6.041538461538462e-06, + "loss": 0.0447, + "step": 8225 + }, + { + "epoch": 16.02, + "learning_rate": 6.028717948717949e-06, + "loss": 0.0464, + "step": 8250 + }, + { + "epoch": 16.02, + "learning_rate": 6.015897435897436e-06, + "loss": 0.043, + "step": 8275 + }, + { + "epoch": 16.02, + "learning_rate": 6.003076923076924e-06, + "loss": 0.0374, + "step": 8300 + }, + { + "epoch": 16.02, + "learning_rate": 5.9902564102564105e-06, + "loss": 0.0367, + "step": 8325 + }, + { + "epoch": 17.0, + "learning_rate": 5.977435897435898e-06, + "loss": 0.0458, + "step": 8350 + }, + { + "epoch": 17.0, + "learning_rate": 5.964615384615385e-06, + "loss": 0.0104, + "step": 8375 + }, + { + "epoch": 17.0, + "learning_rate": 5.951794871794873e-06, + "loss": 0.0072, + "step": 8400 + }, + { + "epoch": 17.0, + "learning_rate": 5.9389743589743595e-06, + "loss": 0.0058, + "step": 8425 + }, + { + "epoch": 17.01, + "learning_rate": 5.926153846153847e-06, + "loss": 0.0062, + "step": 8450 + }, + { + "epoch": 17.01, + "learning_rate": 5.913333333333334e-06, + "loss": 0.0054, + "step": 8475 + }, + { + "epoch": 17.01, + "learning_rate": 5.900512820512822e-06, + "loss": 0.005, + "step": 8500 + }, + { + "epoch": 17.01, + "learning_rate": 5.8876923076923085e-06, + "loss": 0.0061, + "step": 8525 + }, + { + "epoch": 17.01, + "learning_rate": 5.8748717948717945e-06, + "loss": 0.0047, + "step": 8550 + }, + { + "epoch": 17.01, + "learning_rate": 5.862051282051282e-06, + "loss": 0.0047, + "step": 8575 + }, + { + "epoch": 17.01, + "learning_rate": 5.849230769230769e-06, + "loss": 0.0052, + "step": 8600 + }, + { + "epoch": 17.01, + "learning_rate": 5.836410256410257e-06, + "loss": 0.0047, + "step": 8625 + }, + { + "epoch": 17.02, + "learning_rate": 5.8235897435897435e-06, + "loss": 0.0044, + "step": 8650 + }, + { + "epoch": 17.02, + "learning_rate": 5.810769230769231e-06, + "loss": 0.004, + "step": 8675 + }, + { + "epoch": 17.02, + "learning_rate": 5.797948717948718e-06, + "loss": 0.005, + "step": 8700 + }, + { + "epoch": 17.02, + "learning_rate": 5.785128205128206e-06, + "loss": 0.0044, + "step": 8725 + }, + { + "epoch": 17.02, + "learning_rate": 5.7723076923076925e-06, + "loss": 0.0047, + "step": 8750 + }, + { + "epoch": 17.02, + "learning_rate": 5.75948717948718e-06, + "loss": 0.0047, + "step": 8775 + }, + { + "epoch": 17.02, + "learning_rate": 5.746666666666667e-06, + "loss": 0.0043, + "step": 8800 + }, + { + "epoch": 17.02, + "learning_rate": 5.733846153846155e-06, + "loss": 0.004, + "step": 8825 + }, + { + "epoch": 18.0, + "learning_rate": 5.7210256410256415e-06, + "loss": 0.0607, + "step": 8850 + }, + { + "epoch": 18.0, + "learning_rate": 5.708205128205129e-06, + "loss": 0.0939, + "step": 8875 + }, + { + "epoch": 18.0, + "learning_rate": 5.695384615384616e-06, + "loss": 0.0848, + "step": 8900 + }, + { + "epoch": 18.0, + "learning_rate": 5.682564102564103e-06, + "loss": 0.0938, + "step": 8925 + }, + { + "epoch": 18.01, + "learning_rate": 5.6697435897435905e-06, + "loss": 0.0853, + "step": 8950 + }, + { + "epoch": 18.01, + "learning_rate": 5.656923076923077e-06, + "loss": 0.0887, + "step": 8975 + }, + { + "epoch": 18.01, + "learning_rate": 5.644102564102565e-06, + "loss": 0.1087, + "step": 9000 + }, + { + "epoch": 18.01, + "eval_loss": 0.18679018318653107, + "eval_runtime": 5622.427, + "eval_samples_per_second": 2.794, + "eval_steps_per_second": 0.175, + "eval_wer": 5.517170589631569, + "step": 9000 + }, + { + "epoch": 18.01, + "learning_rate": 5.631282051282052e-06, + "loss": 0.1097, + "step": 9025 + }, + { + "epoch": 18.01, + "learning_rate": 5.6184615384615395e-06, + "loss": 0.0773, + "step": 9050 + }, + { + "epoch": 18.01, + "learning_rate": 5.605641025641026e-06, + "loss": 0.0886, + "step": 9075 + }, + { + "epoch": 18.01, + "learning_rate": 5.592820512820514e-06, + "loss": 0.1453, + "step": 9100 + }, + { + "epoch": 18.01, + "learning_rate": 5.580000000000001e-06, + "loss": 0.1193, + "step": 9125 + }, + { + "epoch": 18.02, + "learning_rate": 5.567179487179487e-06, + "loss": 0.1047, + "step": 9150 + }, + { + "epoch": 18.02, + "learning_rate": 5.5543589743589744e-06, + "loss": 0.1135, + "step": 9175 + }, + { + "epoch": 18.02, + "learning_rate": 5.541538461538461e-06, + "loss": 0.1152, + "step": 9200 + }, + { + "epoch": 18.02, + "learning_rate": 5.528717948717949e-06, + "loss": 0.0989, + "step": 9225 + }, + { + "epoch": 18.02, + "learning_rate": 5.515897435897436e-06, + "loss": 0.1154, + "step": 9250 + }, + { + "epoch": 18.02, + "learning_rate": 5.5030769230769234e-06, + "loss": 0.1239, + "step": 9275 + }, + { + "epoch": 18.02, + "learning_rate": 5.49025641025641e-06, + "loss": 0.1244, + "step": 9300 + }, + { + "epoch": 18.02, + "learning_rate": 5.477435897435898e-06, + "loss": 0.0898, + "step": 9325 + }, + { + "epoch": 19.0, + "learning_rate": 5.464615384615385e-06, + "loss": 0.0199, + "step": 9350 + }, + { + "epoch": 19.0, + "learning_rate": 5.451794871794872e-06, + "loss": 0.0053, + "step": 9375 + }, + { + "epoch": 19.0, + "learning_rate": 5.438974358974359e-06, + "loss": 0.0049, + "step": 9400 + }, + { + "epoch": 19.0, + "learning_rate": 5.426153846153847e-06, + "loss": 0.0036, + "step": 9425 + }, + { + "epoch": 19.01, + "learning_rate": 5.413333333333334e-06, + "loss": 0.0041, + "step": 9450 + }, + { + "epoch": 19.01, + "learning_rate": 5.400512820512821e-06, + "loss": 0.0035, + "step": 9475 + }, + { + "epoch": 19.01, + "learning_rate": 5.387692307692308e-06, + "loss": 0.0034, + "step": 9500 + }, + { + "epoch": 19.01, + "learning_rate": 5.374871794871795e-06, + "loss": 0.0033, + "step": 9525 + }, + { + "epoch": 19.01, + "learning_rate": 5.362051282051283e-06, + "loss": 0.0034, + "step": 9550 + }, + { + "epoch": 19.01, + "learning_rate": 5.3492307692307696e-06, + "loss": 0.0031, + "step": 9575 + }, + { + "epoch": 19.01, + "learning_rate": 5.336410256410257e-06, + "loss": 0.0033, + "step": 9600 + }, + { + "epoch": 19.01, + "learning_rate": 5.323589743589744e-06, + "loss": 0.0032, + "step": 9625 + }, + { + "epoch": 19.02, + "learning_rate": 5.310769230769232e-06, + "loss": 0.0027, + "step": 9650 + }, + { + "epoch": 19.02, + "learning_rate": 5.2979487179487185e-06, + "loss": 0.0034, + "step": 9675 + }, + { + "epoch": 19.02, + "learning_rate": 5.285128205128206e-06, + "loss": 0.0029, + "step": 9700 + }, + { + "epoch": 19.02, + "learning_rate": 5.272307692307693e-06, + "loss": 0.0037, + "step": 9725 + }, + { + "epoch": 19.02, + "learning_rate": 5.259487179487181e-06, + "loss": 0.0031, + "step": 9750 + }, + { + "epoch": 19.02, + "learning_rate": 5.246666666666667e-06, + "loss": 0.0028, + "step": 9775 + }, + { + "epoch": 19.02, + "learning_rate": 5.2338461538461535e-06, + "loss": 0.0025, + "step": 9800 + }, + { + "epoch": 20.0, + "learning_rate": 5.221025641025641e-06, + "loss": 0.0133, + "step": 9825 + }, + { + "epoch": 20.0, + "learning_rate": 5.208205128205128e-06, + "loss": 0.0565, + "step": 9850 + }, + { + "epoch": 20.0, + "learning_rate": 5.195384615384616e-06, + "loss": 0.0706, + "step": 9875 + }, + { + "epoch": 20.0, + "learning_rate": 5.1825641025641025e-06, + "loss": 0.0584, + "step": 9900 + }, + { + "epoch": 20.01, + "learning_rate": 5.16974358974359e-06, + "loss": 0.0446, + "step": 9925 + }, + { + "epoch": 20.01, + "learning_rate": 5.156923076923077e-06, + "loss": 0.052, + "step": 9950 + }, + { + "epoch": 20.01, + "learning_rate": 5.144102564102565e-06, + "loss": 0.0432, + "step": 9975 + }, + { + "epoch": 20.01, + "learning_rate": 5.1312820512820515e-06, + "loss": 0.0401, + "step": 10000 + }, + { + "epoch": 20.01, + "eval_loss": 0.20142503082752228, + "eval_runtime": 5472.0584, + "eval_samples_per_second": 2.871, + "eval_steps_per_second": 0.179, + "eval_wer": 5.442752939817933, + "step": 10000 + }, + { + "epoch": 20.01, + "learning_rate": 5.118461538461539e-06, + "loss": 0.0427, + "step": 10025 + }, + { + "epoch": 20.01, + "learning_rate": 5.105641025641026e-06, + "loss": 0.0663, + "step": 10050 + }, + { + "epoch": 20.01, + "learning_rate": 5.092820512820514e-06, + "loss": 0.0643, + "step": 10075 + }, + { + "epoch": 20.01, + "learning_rate": 5.0800000000000005e-06, + "loss": 0.0653, + "step": 10100 + }, + { + "epoch": 20.02, + "learning_rate": 5.067179487179487e-06, + "loss": 0.0505, + "step": 10125 + }, + { + "epoch": 20.02, + "learning_rate": 5.054358974358975e-06, + "loss": 0.0443, + "step": 10150 + }, + { + "epoch": 20.02, + "learning_rate": 5.041538461538462e-06, + "loss": 0.0567, + "step": 10175 + }, + { + "epoch": 20.02, + "learning_rate": 5.029230769230769e-06, + "loss": 0.0432, + "step": 10200 + }, + { + "epoch": 20.02, + "learning_rate": 5.016410256410257e-06, + "loss": 0.0417, + "step": 10225 + }, + { + "epoch": 20.02, + "learning_rate": 5.003589743589744e-06, + "loss": 0.0298, + "step": 10250 + }, + { + "epoch": 20.02, + "learning_rate": 4.990769230769231e-06, + "loss": 0.022, + "step": 10275 + }, + { + "epoch": 20.02, + "learning_rate": 4.977948717948718e-06, + "loss": 0.0248, + "step": 10300 + }, + { + "epoch": 21.0, + "learning_rate": 4.965128205128206e-06, + "loss": 0.0219, + "step": 10325 + }, + { + "epoch": 21.0, + "learning_rate": 4.952307692307693e-06, + "loss": 0.0279, + "step": 10350 + }, + { + "epoch": 21.0, + "learning_rate": 4.93948717948718e-06, + "loss": 0.0334, + "step": 10375 + }, + { + "epoch": 21.0, + "learning_rate": 4.926666666666667e-06, + "loss": 0.0161, + "step": 10400 + }, + { + "epoch": 21.01, + "learning_rate": 4.913846153846154e-06, + "loss": 0.0191, + "step": 10425 + }, + { + "epoch": 21.01, + "learning_rate": 4.901025641025641e-06, + "loss": 0.0188, + "step": 10450 + }, + { + "epoch": 21.01, + "learning_rate": 4.8882051282051285e-06, + "loss": 0.0165, + "step": 10475 + }, + { + "epoch": 21.01, + "learning_rate": 4.875384615384615e-06, + "loss": 0.0142, + "step": 10500 + }, + { + "epoch": 21.01, + "learning_rate": 4.862564102564103e-06, + "loss": 0.0172, + "step": 10525 + }, + { + "epoch": 21.01, + "learning_rate": 4.84974358974359e-06, + "loss": 0.0209, + "step": 10550 + }, + { + "epoch": 21.01, + "learning_rate": 4.8369230769230775e-06, + "loss": 0.0207, + "step": 10575 + }, + { + "epoch": 21.01, + "learning_rate": 4.824102564102564e-06, + "loss": 0.0193, + "step": 10600 + }, + { + "epoch": 21.02, + "learning_rate": 4.811282051282052e-06, + "loss": 0.0142, + "step": 10625 + }, + { + "epoch": 21.02, + "learning_rate": 4.798461538461539e-06, + "loss": 0.0164, + "step": 10650 + }, + { + "epoch": 21.02, + "learning_rate": 4.7856410256410265e-06, + "loss": 0.0185, + "step": 10675 + }, + { + "epoch": 21.02, + "learning_rate": 4.772820512820513e-06, + "loss": 0.0145, + "step": 10700 + }, + { + "epoch": 21.02, + "learning_rate": 4.76e-06, + "loss": 0.0132, + "step": 10725 + }, + { + "epoch": 21.02, + "learning_rate": 4.747179487179487e-06, + "loss": 0.0105, + "step": 10750 + }, + { + "epoch": 21.02, + "learning_rate": 4.734358974358975e-06, + "loss": 0.0095, + "step": 10775 + }, + { + "epoch": 21.02, + "learning_rate": 4.7215384615384615e-06, + "loss": 0.0101, + "step": 10800 + }, + { + "epoch": 22.0, + "learning_rate": 4.708717948717949e-06, + "loss": 0.034, + "step": 10825 + }, + { + "epoch": 22.0, + "learning_rate": 4.695897435897436e-06, + "loss": 0.0307, + "step": 10850 + }, + { + "epoch": 22.0, + "learning_rate": 4.683076923076924e-06, + "loss": 0.0352, + "step": 10875 + }, + { + "epoch": 22.0, + "learning_rate": 4.6702564102564105e-06, + "loss": 0.036, + "step": 10900 + }, + { + "epoch": 22.01, + "learning_rate": 4.657435897435898e-06, + "loss": 0.0307, + "step": 10925 + }, + { + "epoch": 22.01, + "learning_rate": 4.644615384615385e-06, + "loss": 0.0302, + "step": 10950 + }, + { + "epoch": 22.01, + "learning_rate": 4.631794871794873e-06, + "loss": 0.0358, + "step": 10975 + }, + { + "epoch": 22.01, + "learning_rate": 4.6189743589743595e-06, + "loss": 0.0334, + "step": 11000 + }, + { + "epoch": 22.01, + "eval_loss": 0.17510026693344116, + "eval_runtime": 5508.3457, + "eval_samples_per_second": 2.852, + "eval_steps_per_second": 0.178, + "eval_wer": 5.282370073840272, + "step": 11000 + }, + { + "epoch": 22.01, + "learning_rate": 4.606153846153846e-06, + "loss": 0.0341, + "step": 11025 + }, + { + "epoch": 22.01, + "learning_rate": 4.593333333333333e-06, + "loss": 0.0309, + "step": 11050 + }, + { + "epoch": 22.01, + "learning_rate": 4.580512820512821e-06, + "loss": 0.0259, + "step": 11075 + }, + { + "epoch": 22.01, + "learning_rate": 4.567692307692308e-06, + "loss": 0.0297, + "step": 11100 + }, + { + "epoch": 22.02, + "learning_rate": 4.554871794871795e-06, + "loss": 0.0276, + "step": 11125 + }, + { + "epoch": 22.02, + "learning_rate": 4.542051282051282e-06, + "loss": 0.035, + "step": 11150 + }, + { + "epoch": 22.02, + "learning_rate": 4.52923076923077e-06, + "loss": 0.031, + "step": 11175 + }, + { + "epoch": 22.02, + "learning_rate": 4.516410256410257e-06, + "loss": 0.0349, + "step": 11200 + }, + { + "epoch": 22.02, + "learning_rate": 4.503589743589744e-06, + "loss": 0.0367, + "step": 11225 + }, + { + "epoch": 22.02, + "learning_rate": 4.490769230769231e-06, + "loss": 0.0493, + "step": 11250 + }, + { + "epoch": 22.02, + "learning_rate": 4.477948717948719e-06, + "loss": 0.0478, + "step": 11275 + }, + { + "epoch": 23.0, + "learning_rate": 4.465128205128206e-06, + "loss": 0.0319, + "step": 11300 + }, + { + "epoch": 23.0, + "learning_rate": 4.452307692307692e-06, + "loss": 0.0162, + "step": 11325 + }, + { + "epoch": 23.0, + "learning_rate": 4.439487179487179e-06, + "loss": 0.0245, + "step": 11350 + }, + { + "epoch": 23.0, + "learning_rate": 4.426666666666667e-06, + "loss": 0.0189, + "step": 11375 + }, + { + "epoch": 23.01, + "learning_rate": 4.413846153846154e-06, + "loss": 0.0203, + "step": 11400 + }, + { + "epoch": 23.01, + "learning_rate": 4.401025641025641e-06, + "loss": 0.0135, + "step": 11425 + }, + { + "epoch": 23.01, + "learning_rate": 4.388205128205128e-06, + "loss": 0.0121, + "step": 11450 + }, + { + "epoch": 23.01, + "learning_rate": 4.375384615384616e-06, + "loss": 0.0164, + "step": 11475 + }, + { + "epoch": 23.01, + "learning_rate": 4.362564102564103e-06, + "loss": 0.0201, + "step": 11500 + }, + { + "epoch": 23.01, + "learning_rate": 4.34974358974359e-06, + "loss": 0.017, + "step": 11525 + }, + { + "epoch": 23.01, + "learning_rate": 4.336923076923077e-06, + "loss": 0.0152, + "step": 11550 + }, + { + "epoch": 23.01, + "learning_rate": 4.324102564102565e-06, + "loss": 0.0184, + "step": 11575 + }, + { + "epoch": 23.02, + "learning_rate": 4.311282051282052e-06, + "loss": 0.0154, + "step": 11600 + }, + { + "epoch": 23.02, + "learning_rate": 4.2984615384615386e-06, + "loss": 0.0186, + "step": 11625 + }, + { + "epoch": 23.02, + "learning_rate": 4.285641025641025e-06, + "loss": 0.0341, + "step": 11650 + }, + { + "epoch": 23.02, + "learning_rate": 4.272820512820513e-06, + "loss": 0.0433, + "step": 11675 + }, + { + "epoch": 23.02, + "learning_rate": 4.26e-06, + "loss": 0.022, + "step": 11700 + }, + { + "epoch": 23.02, + "learning_rate": 4.2471794871794875e-06, + "loss": 0.0308, + "step": 11725 + }, + { + "epoch": 23.02, + "learning_rate": 4.234358974358974e-06, + "loss": 0.0167, + "step": 11750 + }, + { + "epoch": 23.02, + "learning_rate": 4.221538461538462e-06, + "loss": 0.0457, + "step": 11775 + }, + { + "epoch": 24.0, + "learning_rate": 4.208717948717949e-06, + "loss": 0.0283, + "step": 11800 + }, + { + "epoch": 24.0, + "learning_rate": 4.1958974358974365e-06, + "loss": 0.0086, + "step": 11825 + }, + { + "epoch": 24.0, + "learning_rate": 4.183076923076923e-06, + "loss": 0.0088, + "step": 11850 + }, + { + "epoch": 24.0, + "learning_rate": 4.170256410256411e-06, + "loss": 0.0091, + "step": 11875 + }, + { + "epoch": 24.01, + "learning_rate": 4.157435897435898e-06, + "loss": 0.0057, + "step": 11900 + }, + { + "epoch": 24.01, + "learning_rate": 4.144615384615385e-06, + "loss": 0.0058, + "step": 11925 + }, + { + "epoch": 24.01, + "learning_rate": 4.131794871794872e-06, + "loss": 0.0047, + "step": 11950 + }, + { + "epoch": 24.01, + "learning_rate": 4.118974358974359e-06, + "loss": 0.006, + "step": 11975 + }, + { + "epoch": 24.01, + "learning_rate": 4.106153846153846e-06, + "loss": 0.0071, + "step": 12000 + }, + { + "epoch": 24.01, + "eval_loss": 0.229498490691185, + "eval_runtime": 5519.4715, + "eval_samples_per_second": 2.846, + "eval_steps_per_second": 0.178, + "eval_wer": 5.249010437716918, + "step": 12000 + }, + { + "epoch": 24.01, + "learning_rate": 4.093333333333334e-06, + "loss": 0.0039, + "step": 12025 + }, + { + "epoch": 24.01, + "learning_rate": 4.0805128205128205e-06, + "loss": 0.0044, + "step": 12050 + }, + { + "epoch": 24.01, + "learning_rate": 4.067692307692308e-06, + "loss": 0.0041, + "step": 12075 + }, + { + "epoch": 24.02, + "learning_rate": 4.054871794871795e-06, + "loss": 0.0042, + "step": 12100 + }, + { + "epoch": 24.02, + "learning_rate": 4.042051282051283e-06, + "loss": 0.0055, + "step": 12125 + }, + { + "epoch": 24.02, + "learning_rate": 4.0292307692307695e-06, + "loss": 0.0073, + "step": 12150 + }, + { + "epoch": 24.02, + "learning_rate": 4.016410256410257e-06, + "loss": 0.006, + "step": 12175 + }, + { + "epoch": 24.02, + "learning_rate": 4.003589743589744e-06, + "loss": 0.0047, + "step": 12200 + }, + { + "epoch": 24.02, + "learning_rate": 3.990769230769231e-06, + "loss": 0.0044, + "step": 12225 + }, + { + "epoch": 24.02, + "learning_rate": 3.9779487179487185e-06, + "loss": 0.0046, + "step": 12250 + }, + { + "epoch": 24.02, + "learning_rate": 3.965128205128205e-06, + "loss": 0.0036, + "step": 12275 + }, + { + "epoch": 25.0, + "learning_rate": 3.952307692307692e-06, + "loss": 0.0097, + "step": 12300 + }, + { + "epoch": 25.0, + "learning_rate": 3.93948717948718e-06, + "loss": 0.007, + "step": 12325 + }, + { + "epoch": 25.0, + "learning_rate": 3.926666666666667e-06, + "loss": 0.0073, + "step": 12350 + }, + { + "epoch": 25.0, + "learning_rate": 3.913846153846154e-06, + "loss": 0.0072, + "step": 12375 + }, + { + "epoch": 25.01, + "learning_rate": 3.901025641025641e-06, + "loss": 0.0064, + "step": 12400 + }, + { + "epoch": 25.01, + "learning_rate": 3.888205128205129e-06, + "loss": 0.0052, + "step": 12425 + }, + { + "epoch": 25.01, + "learning_rate": 3.875384615384616e-06, + "loss": 0.0042, + "step": 12450 + }, + { + "epoch": 25.01, + "learning_rate": 3.862564102564103e-06, + "loss": 0.0069, + "step": 12475 + }, + { + "epoch": 25.01, + "learning_rate": 3.84974358974359e-06, + "loss": 0.0063, + "step": 12500 + }, + { + "epoch": 25.01, + "learning_rate": 3.836923076923077e-06, + "loss": 0.0052, + "step": 12525 + }, + { + "epoch": 25.01, + "learning_rate": 3.824102564102565e-06, + "loss": 0.0058, + "step": 12550 + }, + { + "epoch": 25.02, + "learning_rate": 3.8112820512820514e-06, + "loss": 0.0059, + "step": 12575 + }, + { + "epoch": 25.02, + "learning_rate": 3.7984615384615387e-06, + "loss": 0.0051, + "step": 12600 + }, + { + "epoch": 25.02, + "learning_rate": 3.785641025641026e-06, + "loss": 0.0064, + "step": 12625 + }, + { + "epoch": 25.02, + "learning_rate": 3.772820512820513e-06, + "loss": 0.0205, + "step": 12650 + }, + { + "epoch": 25.02, + "learning_rate": 3.7600000000000004e-06, + "loss": 0.0109, + "step": 12675 + }, + { + "epoch": 25.02, + "learning_rate": 3.7471794871794877e-06, + "loss": 0.009, + "step": 12700 + }, + { + "epoch": 25.02, + "learning_rate": 3.734358974358975e-06, + "loss": 0.0081, + "step": 12725 + }, + { + "epoch": 25.02, + "learning_rate": 3.721538461538462e-06, + "loss": 0.0056, + "step": 12750 + }, + { + "epoch": 26.0, + "learning_rate": 3.7087179487179494e-06, + "loss": 0.0367, + "step": 12775 + }, + { + "epoch": 26.0, + "learning_rate": 3.6958974358974362e-06, + "loss": 0.041, + "step": 12800 + }, + { + "epoch": 26.0, + "learning_rate": 3.683076923076923e-06, + "loss": 0.0369, + "step": 12825 + }, + { + "epoch": 26.0, + "learning_rate": 3.6702564102564103e-06, + "loss": 0.0372, + "step": 12850 + }, + { + "epoch": 26.01, + "learning_rate": 3.6574358974358976e-06, + "loss": 0.0359, + "step": 12875 + }, + { + "epoch": 26.01, + "learning_rate": 3.644615384615385e-06, + "loss": 0.044, + "step": 12900 + }, + { + "epoch": 26.01, + "learning_rate": 3.631794871794872e-06, + "loss": 0.0514, + "step": 12925 + }, + { + "epoch": 26.01, + "learning_rate": 3.6189743589743593e-06, + "loss": 0.0517, + "step": 12950 + }, + { + "epoch": 26.01, + "learning_rate": 3.6061538461538466e-06, + "loss": 0.0391, + "step": 12975 + }, + { + "epoch": 26.01, + "learning_rate": 3.593846153846154e-06, + "loss": 0.0374, + "step": 13000 + }, + { + "epoch": 26.01, + "eval_loss": 0.2098151296377182, + "eval_runtime": 6257.1933, + "eval_samples_per_second": 2.51, + "eval_steps_per_second": 0.157, + "eval_wer": 5.2573503467477565, + "step": 13000 + }, + { + "epoch": 26.01, + "learning_rate": 3.5810256410256412e-06, + "loss": 0.0607, + "step": 13025 + }, + { + "epoch": 26.01, + "learning_rate": 3.5682051282051285e-06, + "loss": 0.0533, + "step": 13050 + }, + { + "epoch": 26.02, + "learning_rate": 3.5553846153846157e-06, + "loss": 0.048, + "step": 13075 + }, + { + "epoch": 26.02, + "learning_rate": 3.542564102564103e-06, + "loss": 0.05, + "step": 13100 + }, + { + "epoch": 26.02, + "learning_rate": 3.5297435897435898e-06, + "loss": 0.0534, + "step": 13125 + }, + { + "epoch": 26.02, + "learning_rate": 3.516923076923077e-06, + "loss": 0.0467, + "step": 13150 + }, + { + "epoch": 26.02, + "learning_rate": 3.5041025641025643e-06, + "loss": 0.0552, + "step": 13175 + }, + { + "epoch": 26.02, + "learning_rate": 3.4912820512820515e-06, + "loss": 0.0601, + "step": 13200 + }, + { + "epoch": 26.02, + "learning_rate": 3.4784615384615388e-06, + "loss": 0.0678, + "step": 13225 + }, + { + "epoch": 26.02, + "learning_rate": 3.465641025641026e-06, + "loss": 0.0438, + "step": 13250 + }, + { + "epoch": 27.0, + "learning_rate": 3.4528205128205133e-06, + "loss": 0.0195, + "step": 13275 + }, + { + "epoch": 27.0, + "learning_rate": 3.44e-06, + "loss": 0.0156, + "step": 13300 + }, + { + "epoch": 27.0, + "learning_rate": 3.4271794871794873e-06, + "loss": 0.0141, + "step": 13325 + }, + { + "epoch": 27.0, + "learning_rate": 3.4143589743589746e-06, + "loss": 0.0133, + "step": 13350 + }, + { + "epoch": 27.01, + "learning_rate": 3.401538461538462e-06, + "loss": 0.0153, + "step": 13375 + }, + { + "epoch": 27.01, + "learning_rate": 3.388717948717949e-06, + "loss": 0.0103, + "step": 13400 + }, + { + "epoch": 27.01, + "learning_rate": 3.375897435897436e-06, + "loss": 0.0139, + "step": 13425 + }, + { + "epoch": 27.01, + "learning_rate": 3.363076923076923e-06, + "loss": 0.014, + "step": 13450 + }, + { + "epoch": 27.01, + "learning_rate": 3.3502564102564104e-06, + "loss": 0.0118, + "step": 13475 + }, + { + "epoch": 27.01, + "learning_rate": 3.3374358974358977e-06, + "loss": 0.0111, + "step": 13500 + }, + { + "epoch": 27.01, + "learning_rate": 3.324615384615385e-06, + "loss": 0.0107, + "step": 13525 + }, + { + "epoch": 27.01, + "learning_rate": 3.311794871794872e-06, + "loss": 0.0098, + "step": 13550 + }, + { + "epoch": 27.02, + "learning_rate": 3.2989743589743594e-06, + "loss": 0.0112, + "step": 13575 + }, + { + "epoch": 27.02, + "learning_rate": 3.2861538461538467e-06, + "loss": 0.0124, + "step": 13600 + }, + { + "epoch": 27.02, + "learning_rate": 3.2733333333333335e-06, + "loss": 0.0134, + "step": 13625 + }, + { + "epoch": 27.02, + "learning_rate": 3.2605128205128207e-06, + "loss": 0.0123, + "step": 13650 + }, + { + "epoch": 27.02, + "learning_rate": 3.247692307692308e-06, + "loss": 0.0124, + "step": 13675 + }, + { + "epoch": 27.02, + "learning_rate": 3.2348717948717952e-06, + "loss": 0.0143, + "step": 13700 + }, + { + "epoch": 27.02, + "learning_rate": 3.2220512820512825e-06, + "loss": 0.025, + "step": 13725 + }, + { + "epoch": 28.0, + "learning_rate": 3.2092307692307693e-06, + "loss": 0.0139, + "step": 13750 + }, + { + "epoch": 28.0, + "learning_rate": 3.1964102564102565e-06, + "loss": 0.0028, + "step": 13775 + }, + { + "epoch": 28.0, + "learning_rate": 3.1835897435897438e-06, + "loss": 0.0042, + "step": 13800 + }, + { + "epoch": 28.0, + "learning_rate": 3.170769230769231e-06, + "loss": 0.0047, + "step": 13825 + }, + { + "epoch": 28.01, + "learning_rate": 3.1579487179487183e-06, + "loss": 0.0033, + "step": 13850 + }, + { + "epoch": 28.01, + "learning_rate": 3.1451282051282055e-06, + "loss": 0.0033, + "step": 13875 + }, + { + "epoch": 28.01, + "learning_rate": 3.1323076923076928e-06, + "loss": 0.0028, + "step": 13900 + }, + { + "epoch": 28.01, + "learning_rate": 3.1194871794871796e-06, + "loss": 0.0024, + "step": 13925 + }, + { + "epoch": 28.01, + "learning_rate": 3.106666666666667e-06, + "loss": 0.0032, + "step": 13950 + }, + { + "epoch": 28.01, + "learning_rate": 3.093846153846154e-06, + "loss": 0.0023, + "step": 13975 + }, + { + "epoch": 28.01, + "learning_rate": 3.0810256410256413e-06, + "loss": 0.0023, + "step": 14000 + }, + { + "epoch": 28.01, + "eval_loss": 0.24982614815235138, + "eval_runtime": 6759.6267, + "eval_samples_per_second": 2.324, + "eval_steps_per_second": 0.145, + "eval_wer": 5.041795774873779, + "step": 14000 + }, + { + "epoch": 28.01, + "learning_rate": 3.0682051282051286e-06, + "loss": 0.0022, + "step": 14025 + }, + { + "epoch": 28.02, + "learning_rate": 3.0553846153846154e-06, + "loss": 0.0025, + "step": 14050 + }, + { + "epoch": 28.02, + "learning_rate": 3.0425641025641027e-06, + "loss": 0.0035, + "step": 14075 + }, + { + "epoch": 28.02, + "learning_rate": 3.02974358974359e-06, + "loss": 0.0032, + "step": 14100 + }, + { + "epoch": 28.02, + "learning_rate": 3.016923076923077e-06, + "loss": 0.0039, + "step": 14125 + }, + { + "epoch": 28.02, + "learning_rate": 3.0041025641025644e-06, + "loss": 0.0026, + "step": 14150 + }, + { + "epoch": 28.02, + "learning_rate": 2.9912820512820517e-06, + "loss": 0.0022, + "step": 14175 + }, + { + "epoch": 28.02, + "learning_rate": 2.978461538461539e-06, + "loss": 0.0023, + "step": 14200 + }, + { + "epoch": 28.02, + "learning_rate": 2.9656410256410257e-06, + "loss": 0.0021, + "step": 14225 + }, + { + "epoch": 29.0, + "learning_rate": 2.952820512820513e-06, + "loss": 0.0026, + "step": 14250 + }, + { + "epoch": 29.0, + "learning_rate": 2.9400000000000002e-06, + "loss": 0.0014, + "step": 14275 + }, + { + "epoch": 29.0, + "learning_rate": 2.9271794871794875e-06, + "loss": 0.0023, + "step": 14300 + }, + { + "epoch": 29.0, + "learning_rate": 2.9143589743589747e-06, + "loss": 0.0017, + "step": 14325 + }, + { + "epoch": 29.01, + "learning_rate": 2.9015384615384615e-06, + "loss": 0.0016, + "step": 14350 + }, + { + "epoch": 29.01, + "learning_rate": 2.888717948717949e-06, + "loss": 0.0015, + "step": 14375 + }, + { + "epoch": 29.01, + "learning_rate": 2.875897435897436e-06, + "loss": 0.0012, + "step": 14400 + }, + { + "epoch": 29.01, + "learning_rate": 2.8630769230769233e-06, + "loss": 0.0011, + "step": 14425 + }, + { + "epoch": 29.01, + "learning_rate": 2.8502564102564105e-06, + "loss": 0.0013, + "step": 14450 + }, + { + "epoch": 29.01, + "learning_rate": 2.8374358974358978e-06, + "loss": 0.0009, + "step": 14475 + }, + { + "epoch": 29.01, + "learning_rate": 2.824615384615385e-06, + "loss": 0.0009, + "step": 14500 + }, + { + "epoch": 29.01, + "learning_rate": 2.811794871794872e-06, + "loss": 0.0012, + "step": 14525 + }, + { + "epoch": 29.02, + "learning_rate": 2.798974358974359e-06, + "loss": 0.001, + "step": 14550 + }, + { + "epoch": 29.02, + "learning_rate": 2.7861538461538464e-06, + "loss": 0.0012, + "step": 14575 + }, + { + "epoch": 29.02, + "learning_rate": 2.7733333333333336e-06, + "loss": 0.0015, + "step": 14600 + }, + { + "epoch": 29.02, + "learning_rate": 2.760512820512821e-06, + "loss": 0.0013, + "step": 14625 + }, + { + "epoch": 29.02, + "learning_rate": 2.7476923076923077e-06, + "loss": 0.0014, + "step": 14650 + }, + { + "epoch": 29.02, + "learning_rate": 2.734871794871795e-06, + "loss": 0.0011, + "step": 14675 + }, + { + "epoch": 29.02, + "learning_rate": 2.722051282051282e-06, + "loss": 0.0009, + "step": 14700 + }, + { + "epoch": 29.02, + "learning_rate": 2.7092307692307694e-06, + "loss": 0.0009, + "step": 14725 + }, + { + "epoch": 30.0, + "learning_rate": 2.6964102564102567e-06, + "loss": 0.0016, + "step": 14750 + }, + { + "epoch": 30.0, + "learning_rate": 2.683589743589744e-06, + "loss": 0.0021, + "step": 14775 + }, + { + "epoch": 30.0, + "learning_rate": 2.670769230769231e-06, + "loss": 0.0019, + "step": 14800 + }, + { + "epoch": 30.0, + "learning_rate": 2.657948717948718e-06, + "loss": 0.002, + "step": 14825 + }, + { + "epoch": 30.01, + "learning_rate": 2.6451282051282052e-06, + "loss": 0.0018, + "step": 14850 + }, + { + "epoch": 30.01, + "learning_rate": 2.6323076923076925e-06, + "loss": 0.0017, + "step": 14875 + }, + { + "epoch": 30.01, + "learning_rate": 2.6194871794871797e-06, + "loss": 0.002, + "step": 14900 + }, + { + "epoch": 30.01, + "learning_rate": 2.606666666666667e-06, + "loss": 0.0018, + "step": 14925 + }, + { + "epoch": 30.01, + "learning_rate": 2.593846153846154e-06, + "loss": 0.0016, + "step": 14950 + }, + { + "epoch": 30.01, + "learning_rate": 2.581025641025641e-06, + "loss": 0.0017, + "step": 14975 + }, + { + "epoch": 30.01, + "learning_rate": 2.5682051282051283e-06, + "loss": 0.0025, + "step": 15000 + }, + { + "epoch": 30.01, + "eval_loss": 0.23106679320335388, + "eval_runtime": 6284.6996, + "eval_samples_per_second": 2.499, + "eval_steps_per_second": 0.156, + "eval_wer": 4.938509209184164, + "step": 15000 + }, + { + "epoch": 30.01, + "learning_rate": 2.5553846153846155e-06, + "loss": 0.002, + "step": 15025 + }, + { + "epoch": 30.02, + "learning_rate": 2.542564102564103e-06, + "loss": 0.0018, + "step": 15050 + }, + { + "epoch": 30.02, + "learning_rate": 2.52974358974359e-06, + "loss": 0.0017, + "step": 15075 + }, + { + "epoch": 30.02, + "learning_rate": 2.5169230769230773e-06, + "loss": 0.0016, + "step": 15100 + }, + { + "epoch": 30.02, + "learning_rate": 2.5041025641025645e-06, + "loss": 0.0018, + "step": 15125 + }, + { + "epoch": 30.02, + "learning_rate": 2.4912820512820514e-06, + "loss": 0.0021, + "step": 15150 + }, + { + "epoch": 30.02, + "learning_rate": 2.4784615384615386e-06, + "loss": 0.0022, + "step": 15175 + }, + { + "epoch": 30.02, + "learning_rate": 2.465641025641026e-06, + "loss": 0.0019, + "step": 15200 + }, + { + "epoch": 31.0, + "learning_rate": 2.452820512820513e-06, + "loss": 0.0019, + "step": 15225 + }, + { + "epoch": 31.0, + "learning_rate": 2.4400000000000004e-06, + "loss": 0.0011, + "step": 15250 + }, + { + "epoch": 31.0, + "learning_rate": 2.427179487179487e-06, + "loss": 0.0011, + "step": 15275 + }, + { + "epoch": 31.0, + "learning_rate": 2.4143589743589744e-06, + "loss": 0.001, + "step": 15300 + }, + { + "epoch": 31.01, + "learning_rate": 2.4015384615384617e-06, + "loss": 0.0011, + "step": 15325 + }, + { + "epoch": 31.01, + "learning_rate": 2.388717948717949e-06, + "loss": 0.001, + "step": 15350 + }, + { + "epoch": 31.01, + "learning_rate": 2.375897435897436e-06, + "loss": 0.0017, + "step": 15375 + }, + { + "epoch": 31.01, + "learning_rate": 2.3630769230769234e-06, + "loss": 0.0011, + "step": 15400 + }, + { + "epoch": 31.01, + "learning_rate": 2.3502564102564102e-06, + "loss": 0.0011, + "step": 15425 + }, + { + "epoch": 31.01, + "learning_rate": 2.3374358974358975e-06, + "loss": 0.0009, + "step": 15450 + }, + { + "epoch": 31.01, + "learning_rate": 2.3246153846153847e-06, + "loss": 0.0011, + "step": 15475 + }, + { + "epoch": 31.01, + "learning_rate": 2.311794871794872e-06, + "loss": 0.0012, + "step": 15500 + }, + { + "epoch": 31.02, + "learning_rate": 2.2989743589743592e-06, + "loss": 0.0008, + "step": 15525 + }, + { + "epoch": 31.02, + "learning_rate": 2.2861538461538465e-06, + "loss": 0.0008, + "step": 15550 + }, + { + "epoch": 31.02, + "learning_rate": 2.2733333333333333e-06, + "loss": 0.0009, + "step": 15575 + }, + { + "epoch": 31.02, + "learning_rate": 2.2605128205128206e-06, + "loss": 0.0008, + "step": 15600 + }, + { + "epoch": 31.02, + "learning_rate": 2.247692307692308e-06, + "loss": 0.001, + "step": 15625 + }, + { + "epoch": 31.02, + "learning_rate": 2.234871794871795e-06, + "loss": 0.0008, + "step": 15650 + }, + { + "epoch": 31.02, + "learning_rate": 2.2220512820512823e-06, + "loss": 0.0008, + "step": 15675 + }, + { + "epoch": 31.02, + "learning_rate": 2.2092307692307695e-06, + "loss": 0.0008, + "step": 15700 + }, + { + "epoch": 32.0, + "learning_rate": 2.1964102564102564e-06, + "loss": 0.0007, + "step": 15725 + }, + { + "epoch": 32.0, + "learning_rate": 2.1835897435897436e-06, + "loss": 0.0007, + "step": 15750 + }, + { + "epoch": 32.0, + "learning_rate": 2.170769230769231e-06, + "loss": 0.0007, + "step": 15775 + }, + { + "epoch": 32.0, + "learning_rate": 2.157948717948718e-06, + "loss": 0.0007, + "step": 15800 + }, + { + "epoch": 32.01, + "learning_rate": 2.1451282051282054e-06, + "loss": 0.0007, + "step": 15825 + }, + { + "epoch": 32.01, + "learning_rate": 2.1323076923076926e-06, + "loss": 0.0006, + "step": 15850 + }, + { + "epoch": 32.01, + "learning_rate": 2.1194871794871794e-06, + "loss": 0.0006, + "step": 15875 + }, + { + "epoch": 32.01, + "learning_rate": 2.1066666666666667e-06, + "loss": 0.0006, + "step": 15900 + }, + { + "epoch": 32.01, + "learning_rate": 2.093846153846154e-06, + "loss": 0.0006, + "step": 15925 + }, + { + "epoch": 32.01, + "learning_rate": 2.081025641025641e-06, + "loss": 0.0007, + "step": 15950 + }, + { + "epoch": 32.01, + "learning_rate": 2.0682051282051284e-06, + "loss": 0.0006, + "step": 15975 + }, + { + "epoch": 32.01, + "learning_rate": 2.0553846153846157e-06, + "loss": 0.0006, + "step": 16000 + }, + { + "epoch": 32.01, + "eval_loss": 0.25441616773605347, + "eval_runtime": 5476.5624, + "eval_samples_per_second": 2.868, + "eval_steps_per_second": 0.179, + "eval_wer": 4.89488506963824, + "step": 16000 + }, + { + "epoch": 32.02, + "learning_rate": 2.0425641025641025e-06, + "loss": 0.0006, + "step": 16025 + }, + { + "epoch": 32.02, + "learning_rate": 2.0297435897435897e-06, + "loss": 0.0006, + "step": 16050 + }, + { + "epoch": 32.02, + "learning_rate": 2.016923076923077e-06, + "loss": 0.0006, + "step": 16075 + }, + { + "epoch": 32.02, + "learning_rate": 2.0041025641025642e-06, + "loss": 0.0006, + "step": 16100 + }, + { + "epoch": 32.02, + "learning_rate": 1.9912820512820515e-06, + "loss": 0.0006, + "step": 16125 + }, + { + "epoch": 32.02, + "learning_rate": 1.9784615384615387e-06, + "loss": 0.0006, + "step": 16150 + }, + { + "epoch": 32.02, + "learning_rate": 1.9656410256410256e-06, + "loss": 0.0006, + "step": 16175 + }, + { + "epoch": 32.02, + "learning_rate": 1.952820512820513e-06, + "loss": 0.0006, + "step": 16200 + }, + { + "epoch": 33.0, + "learning_rate": 1.94e-06, + "loss": 0.0033, + "step": 16225 + }, + { + "epoch": 33.0, + "learning_rate": 1.9271794871794873e-06, + "loss": 0.0059, + "step": 16250 + }, + { + "epoch": 33.0, + "learning_rate": 1.9143589743589746e-06, + "loss": 0.0044, + "step": 16275 + }, + { + "epoch": 33.0, + "learning_rate": 1.9015384615384616e-06, + "loss": 0.0051, + "step": 16300 + }, + { + "epoch": 33.01, + "learning_rate": 1.8887179487179488e-06, + "loss": 0.0043, + "step": 16325 + }, + { + "epoch": 33.01, + "learning_rate": 1.875897435897436e-06, + "loss": 0.0045, + "step": 16350 + }, + { + "epoch": 33.01, + "learning_rate": 1.8630769230769233e-06, + "loss": 0.0054, + "step": 16375 + }, + { + "epoch": 33.01, + "learning_rate": 1.8502564102564106e-06, + "loss": 0.0053, + "step": 16400 + }, + { + "epoch": 33.01, + "learning_rate": 1.8374358974358974e-06, + "loss": 0.0047, + "step": 16425 + }, + { + "epoch": 33.01, + "learning_rate": 1.8246153846153847e-06, + "loss": 0.0043, + "step": 16450 + }, + { + "epoch": 33.01, + "learning_rate": 1.811794871794872e-06, + "loss": 0.0034, + "step": 16475 + }, + { + "epoch": 33.01, + "learning_rate": 1.7989743589743592e-06, + "loss": 0.0039, + "step": 16500 + }, + { + "epoch": 33.02, + "learning_rate": 1.7861538461538464e-06, + "loss": 0.0039, + "step": 16525 + }, + { + "epoch": 33.02, + "learning_rate": 1.7733333333333336e-06, + "loss": 0.0044, + "step": 16550 + }, + { + "epoch": 33.02, + "learning_rate": 1.7605128205128205e-06, + "loss": 0.0046, + "step": 16575 + }, + { + "epoch": 33.02, + "learning_rate": 1.7476923076923077e-06, + "loss": 0.0053, + "step": 16600 + }, + { + "epoch": 33.02, + "learning_rate": 1.734871794871795e-06, + "loss": 0.0046, + "step": 16625 + }, + { + "epoch": 33.02, + "learning_rate": 1.7220512820512822e-06, + "loss": 0.0057, + "step": 16650 + }, + { + "epoch": 33.02, + "learning_rate": 1.7092307692307695e-06, + "loss": 0.0096, + "step": 16675 + }, + { + "epoch": 34.0, + "learning_rate": 1.6964102564102567e-06, + "loss": 0.0041, + "step": 16700 + }, + { + "epoch": 34.0, + "learning_rate": 1.683589743589744e-06, + "loss": 0.0011, + "step": 16725 + }, + { + "epoch": 34.0, + "learning_rate": 1.6707692307692308e-06, + "loss": 0.0015, + "step": 16750 + }, + { + "epoch": 34.0, + "learning_rate": 1.657948717948718e-06, + "loss": 0.0018, + "step": 16775 + }, + { + "epoch": 34.01, + "learning_rate": 1.6451282051282053e-06, + "loss": 0.0014, + "step": 16800 + }, + { + "epoch": 34.01, + "learning_rate": 1.6323076923076925e-06, + "loss": 0.0013, + "step": 16825 + }, + { + "epoch": 34.01, + "learning_rate": 1.6194871794871798e-06, + "loss": 0.0011, + "step": 16850 + }, + { + "epoch": 34.01, + "learning_rate": 1.606666666666667e-06, + "loss": 0.0011, + "step": 16875 + }, + { + "epoch": 34.01, + "learning_rate": 1.5938461538461538e-06, + "loss": 0.001, + "step": 16900 + }, + { + "epoch": 34.01, + "learning_rate": 1.581025641025641e-06, + "loss": 0.0012, + "step": 16925 + }, + { + "epoch": 34.01, + "learning_rate": 1.5682051282051283e-06, + "loss": 0.0009, + "step": 16950 + }, + { + "epoch": 34.01, + "learning_rate": 1.5553846153846156e-06, + "loss": 0.0012, + "step": 16975 + }, + { + "epoch": 34.02, + "learning_rate": 1.5425641025641028e-06, + "loss": 0.0009, + "step": 17000 + }, + { + "epoch": 34.02, + "eval_loss": 0.26914408802986145, + "eval_runtime": 5431.6839, + "eval_samples_per_second": 2.892, + "eval_steps_per_second": 0.181, + "eval_wer": 5.124553333718252, + "step": 17000 + }, + { + "epoch": 34.02, + "learning_rate": 1.52974358974359e-06, + "loss": 0.0012, + "step": 17025 + }, + { + "epoch": 34.02, + "learning_rate": 1.516923076923077e-06, + "loss": 0.0015, + "step": 17050 + }, + { + "epoch": 34.02, + "learning_rate": 1.5041025641025642e-06, + "loss": 0.0011, + "step": 17075 + }, + { + "epoch": 34.02, + "learning_rate": 1.4912820512820514e-06, + "loss": 0.001, + "step": 17100 + }, + { + "epoch": 34.02, + "learning_rate": 1.4784615384615387e-06, + "loss": 0.0009, + "step": 17125 + }, + { + "epoch": 34.02, + "learning_rate": 1.465641025641026e-06, + "loss": 0.0008, + "step": 17150 + }, + { + "epoch": 34.02, + "learning_rate": 1.4528205128205132e-06, + "loss": 0.0008, + "step": 17175 + }, + { + "epoch": 35.0, + "learning_rate": 1.44e-06, + "loss": 0.0136, + "step": 17200 + }, + { + "epoch": 35.0, + "learning_rate": 1.4271794871794872e-06, + "loss": 0.0215, + "step": 17225 + }, + { + "epoch": 35.0, + "learning_rate": 1.4143589743589745e-06, + "loss": 0.0184, + "step": 17250 + }, + { + "epoch": 35.0, + "learning_rate": 1.4015384615384617e-06, + "loss": 0.017, + "step": 17275 + }, + { + "epoch": 35.01, + "learning_rate": 1.388717948717949e-06, + "loss": 0.0146, + "step": 17300 + }, + { + "epoch": 35.01, + "learning_rate": 1.3758974358974362e-06, + "loss": 0.0159, + "step": 17325 + }, + { + "epoch": 35.01, + "learning_rate": 1.363076923076923e-06, + "loss": 0.0148, + "step": 17350 + }, + { + "epoch": 35.01, + "learning_rate": 1.3502564102564103e-06, + "loss": 0.0146, + "step": 17375 + }, + { + "epoch": 35.01, + "learning_rate": 1.3374358974358975e-06, + "loss": 0.0186, + "step": 17400 + }, + { + "epoch": 35.01, + "learning_rate": 1.3246153846153848e-06, + "loss": 0.0154, + "step": 17425 + }, + { + "epoch": 35.01, + "learning_rate": 1.311794871794872e-06, + "loss": 0.0152, + "step": 17450 + }, + { + "epoch": 35.01, + "learning_rate": 1.2989743589743593e-06, + "loss": 0.0182, + "step": 17475 + }, + { + "epoch": 35.02, + "learning_rate": 1.286153846153846e-06, + "loss": 0.0176, + "step": 17500 + }, + { + "epoch": 35.02, + "learning_rate": 1.2733333333333334e-06, + "loss": 0.0171, + "step": 17525 + }, + { + "epoch": 35.02, + "learning_rate": 1.2605128205128206e-06, + "loss": 0.0193, + "step": 17550 + }, + { + "epoch": 35.02, + "learning_rate": 1.2476923076923078e-06, + "loss": 0.0182, + "step": 17575 + }, + { + "epoch": 35.02, + "learning_rate": 1.234871794871795e-06, + "loss": 0.0158, + "step": 17600 + }, + { + "epoch": 35.02, + "learning_rate": 1.2220512820512821e-06, + "loss": 0.0158, + "step": 17625 + }, + { + "epoch": 35.02, + "learning_rate": 1.2092307692307694e-06, + "loss": 0.0149, + "step": 17650 + }, + { + "epoch": 35.02, + "learning_rate": 1.1964102564102566e-06, + "loss": 0.0175, + "step": 17675 + }, + { + "epoch": 36.0, + "learning_rate": 1.1835897435897437e-06, + "loss": 0.0053, + "step": 17700 + }, + { + "epoch": 36.0, + "learning_rate": 1.170769230769231e-06, + "loss": 0.0045, + "step": 17725 + }, + { + "epoch": 36.0, + "learning_rate": 1.1579487179487182e-06, + "loss": 0.0044, + "step": 17750 + }, + { + "epoch": 36.0, + "learning_rate": 1.1451282051282052e-06, + "loss": 0.0035, + "step": 17775 + }, + { + "epoch": 36.01, + "learning_rate": 1.1323076923076924e-06, + "loss": 0.0031, + "step": 17800 + }, + { + "epoch": 36.01, + "learning_rate": 1.1194871794871797e-06, + "loss": 0.0031, + "step": 17825 + }, + { + "epoch": 36.01, + "learning_rate": 1.1066666666666667e-06, + "loss": 0.0034, + "step": 17850 + }, + { + "epoch": 36.01, + "learning_rate": 1.093846153846154e-06, + "loss": 0.004, + "step": 17875 + }, + { + "epoch": 36.01, + "learning_rate": 1.0810256410256412e-06, + "loss": 0.0036, + "step": 17900 + }, + { + "epoch": 36.01, + "learning_rate": 1.0682051282051283e-06, + "loss": 0.0031, + "step": 17925 + }, + { + "epoch": 36.01, + "learning_rate": 1.0553846153846155e-06, + "loss": 0.0035, + "step": 17950 + }, + { + "epoch": 36.01, + "learning_rate": 1.0425641025641028e-06, + "loss": 0.0031, + "step": 17975 + }, + { + "epoch": 36.02, + "learning_rate": 1.0297435897435898e-06, + "loss": 0.003, + "step": 18000 + }, + { + "epoch": 36.02, + "eval_loss": 0.22486178576946259, + "eval_runtime": 5489.0585, + "eval_samples_per_second": 2.862, + "eval_steps_per_second": 0.179, + "eval_wer": 5.027682082667744, + "step": 18000 + }, + { + "epoch": 36.02, + "learning_rate": 1.016923076923077e-06, + "loss": 0.0038, + "step": 18025 + }, + { + "epoch": 36.02, + "learning_rate": 1.0041025641025643e-06, + "loss": 0.0091, + "step": 18050 + }, + { + "epoch": 36.02, + "learning_rate": 9.912820512820513e-07, + "loss": 0.0036, + "step": 18075 + }, + { + "epoch": 36.02, + "learning_rate": 9.784615384615386e-07, + "loss": 0.0044, + "step": 18100 + }, + { + "epoch": 36.02, + "learning_rate": 9.656410256410258e-07, + "loss": 0.0035, + "step": 18125 + }, + { + "epoch": 36.02, + "learning_rate": 9.528205128205129e-07, + "loss": 0.0067, + "step": 18150 + }, + { + "epoch": 37.0, + "learning_rate": 9.400000000000001e-07, + "loss": 0.0094, + "step": 18175 + }, + { + "epoch": 37.0, + "learning_rate": 9.271794871794874e-07, + "loss": 0.0016, + "step": 18200 + }, + { + "epoch": 37.0, + "learning_rate": 9.143589743589744e-07, + "loss": 0.0017, + "step": 18225 + }, + { + "epoch": 37.0, + "learning_rate": 9.015384615384616e-07, + "loss": 0.0015, + "step": 18250 + }, + { + "epoch": 37.01, + "learning_rate": 8.887179487179489e-07, + "loss": 0.0016, + "step": 18275 + }, + { + "epoch": 37.01, + "learning_rate": 8.758974358974359e-07, + "loss": 0.0013, + "step": 18300 + }, + { + "epoch": 37.01, + "learning_rate": 8.630769230769232e-07, + "loss": 0.0012, + "step": 18325 + }, + { + "epoch": 37.01, + "learning_rate": 8.502564102564104e-07, + "loss": 0.0018, + "step": 18350 + }, + { + "epoch": 37.01, + "learning_rate": 8.374358974358975e-07, + "loss": 0.0015, + "step": 18375 + }, + { + "epoch": 37.01, + "learning_rate": 8.246153846153847e-07, + "loss": 0.0014, + "step": 18400 + }, + { + "epoch": 37.01, + "learning_rate": 8.11794871794872e-07, + "loss": 0.0013, + "step": 18425 + }, + { + "epoch": 37.01, + "learning_rate": 7.98974358974359e-07, + "loss": 0.0015, + "step": 18450 + }, + { + "epoch": 37.02, + "learning_rate": 7.861538461538462e-07, + "loss": 0.0012, + "step": 18475 + }, + { + "epoch": 37.02, + "learning_rate": 7.733333333333335e-07, + "loss": 0.0014, + "step": 18500 + }, + { + "epoch": 37.02, + "learning_rate": 7.605128205128205e-07, + "loss": 0.0023, + "step": 18525 + }, + { + "epoch": 37.02, + "learning_rate": 7.476923076923078e-07, + "loss": 0.0028, + "step": 18550 + }, + { + "epoch": 37.02, + "learning_rate": 7.34871794871795e-07, + "loss": 0.0015, + "step": 18575 + }, + { + "epoch": 37.02, + "learning_rate": 7.22051282051282e-07, + "loss": 0.0019, + "step": 18600 + }, + { + "epoch": 37.02, + "learning_rate": 7.092307692307693e-07, + "loss": 0.0013, + "step": 18625 + }, + { + "epoch": 37.02, + "learning_rate": 6.964102564102565e-07, + "loss": 0.0036, + "step": 18650 + }, + { + "epoch": 38.0, + "learning_rate": 6.835897435897436e-07, + "loss": 0.0019, + "step": 18675 + }, + { + "epoch": 38.0, + "learning_rate": 6.707692307692308e-07, + "loss": 0.0008, + "step": 18700 + }, + { + "epoch": 38.0, + "learning_rate": 6.579487179487181e-07, + "loss": 0.001, + "step": 18725 + }, + { + "epoch": 38.0, + "learning_rate": 6.451282051282051e-07, + "loss": 0.0009, + "step": 18750 + }, + { + "epoch": 38.01, + "learning_rate": 6.323076923076924e-07, + "loss": 0.0009, + "step": 18775 + }, + { + "epoch": 38.01, + "learning_rate": 6.194871794871796e-07, + "loss": 0.0009, + "step": 18800 + }, + { + "epoch": 38.01, + "learning_rate": 6.066666666666668e-07, + "loss": 0.0008, + "step": 18825 + }, + { + "epoch": 38.01, + "learning_rate": 5.938461538461539e-07, + "loss": 0.0009, + "step": 18850 + }, + { + "epoch": 38.01, + "learning_rate": 5.810256410256411e-07, + "loss": 0.0009, + "step": 18875 + }, + { + "epoch": 38.01, + "learning_rate": 5.682051282051283e-07, + "loss": 0.0007, + "step": 18900 + }, + { + "epoch": 38.01, + "learning_rate": 5.553846153846154e-07, + "loss": 0.0009, + "step": 18925 + }, + { + "epoch": 38.01, + "learning_rate": 5.425641025641027e-07, + "loss": 0.0007, + "step": 18950 + }, + { + "epoch": 38.02, + "learning_rate": 5.297435897435898e-07, + "loss": 0.0008, + "step": 18975 + }, + { + "epoch": 38.02, + "learning_rate": 5.16923076923077e-07, + "loss": 0.0009, + "step": 19000 + }, + { + "epoch": 38.02, + "eval_loss": 0.26029446721076965, + "eval_runtime": 5487.8791, + "eval_samples_per_second": 2.862, + "eval_steps_per_second": 0.179, + "eval_wer": 5.037305054626405, + "step": 19000 + }, + { + "epoch": 38.02, + "learning_rate": 5.041025641025642e-07, + "loss": 0.0011, + "step": 19025 + }, + { + "epoch": 38.02, + "learning_rate": 4.912820512820513e-07, + "loss": 0.0009, + "step": 19050 + }, + { + "epoch": 38.02, + "learning_rate": 4.784615384615385e-07, + "loss": 0.0008, + "step": 19075 + }, + { + "epoch": 38.02, + "learning_rate": 4.656410256410257e-07, + "loss": 0.0007, + "step": 19100 + }, + { + "epoch": 38.02, + "learning_rate": 4.528205128205129e-07, + "loss": 0.0007, + "step": 19125 + }, + { + "epoch": 39.0, + "learning_rate": 4.4e-07, + "loss": 0.0006, + "step": 19150 + }, + { + "epoch": 39.0, + "learning_rate": 4.271794871794872e-07, + "loss": 0.0005, + "step": 19175 + }, + { + "epoch": 39.0, + "learning_rate": 4.143589743589744e-07, + "loss": 0.0007, + "step": 19200 + }, + { + "epoch": 39.0, + "learning_rate": 4.0153846153846155e-07, + "loss": 0.0008, + "step": 19225 + }, + { + "epoch": 39.01, + "learning_rate": 3.8871794871794875e-07, + "loss": 0.0007, + "step": 19250 + }, + { + "epoch": 39.01, + "learning_rate": 3.7589743589743594e-07, + "loss": 0.0007, + "step": 19275 + }, + { + "epoch": 39.01, + "learning_rate": 3.630769230769231e-07, + "loss": 0.0006, + "step": 19300 + }, + { + "epoch": 39.01, + "learning_rate": 3.502564102564103e-07, + "loss": 0.0007, + "step": 19325 + }, + { + "epoch": 39.01, + "learning_rate": 3.374358974358975e-07, + "loss": 0.0007, + "step": 19350 + }, + { + "epoch": 39.01, + "learning_rate": 3.246153846153846e-07, + "loss": 0.0006, + "step": 19375 + }, + { + "epoch": 39.01, + "learning_rate": 3.117948717948718e-07, + "loss": 0.0006, + "step": 19400 + }, + { + "epoch": 39.01, + "learning_rate": 2.98974358974359e-07, + "loss": 0.0008, + "step": 19425 + }, + { + "epoch": 39.02, + "learning_rate": 2.861538461538462e-07, + "loss": 0.0006, + "step": 19450 + }, + { + "epoch": 39.02, + "learning_rate": 2.7333333333333335e-07, + "loss": 0.0007, + "step": 19475 + }, + { + "epoch": 39.02, + "learning_rate": 2.6051282051282054e-07, + "loss": 0.0008, + "step": 19500 + }, + { + "epoch": 39.02, + "learning_rate": 2.4769230769230774e-07, + "loss": 0.0008, + "step": 19525 + }, + { + "epoch": 39.02, + "learning_rate": 2.348717948717949e-07, + "loss": 0.0007, + "step": 19550 + }, + { + "epoch": 39.02, + "learning_rate": 2.2205128205128205e-07, + "loss": 0.0006, + "step": 19575 + }, + { + "epoch": 39.02, + "learning_rate": 2.0923076923076924e-07, + "loss": 0.0006, + "step": 19600 + }, + { + "epoch": 39.02, + "learning_rate": 1.9641025641025644e-07, + "loss": 0.0006, + "step": 19625 + }, + { + "epoch": 40.0, + "learning_rate": 1.8358974358974358e-07, + "loss": 0.0005, + "step": 19650 + }, + { + "epoch": 40.0, + "learning_rate": 1.7076923076923077e-07, + "loss": 0.0005, + "step": 19675 + }, + { + "epoch": 40.0, + "learning_rate": 1.5794871794871797e-07, + "loss": 0.0007, + "step": 19700 + }, + { + "epoch": 40.0, + "learning_rate": 1.4512820512820514e-07, + "loss": 0.0007, + "step": 19725 + }, + { + "epoch": 40.01, + "learning_rate": 1.323076923076923e-07, + "loss": 0.0007, + "step": 19750 + }, + { + "epoch": 40.01, + "learning_rate": 1.194871794871795e-07, + "loss": 0.0007, + "step": 19775 + }, + { + "epoch": 40.01, + "learning_rate": 1.0666666666666667e-07, + "loss": 0.0006, + "step": 19800 + }, + { + "epoch": 40.01, + "learning_rate": 9.384615384615386e-08, + "loss": 0.0006, + "step": 19825 + }, + { + "epoch": 40.01, + "learning_rate": 8.102564102564103e-08, + "loss": 0.0006, + "step": 19850 + }, + { + "epoch": 40.01, + "learning_rate": 6.820512820512821e-08, + "loss": 0.0006, + "step": 19875 + }, + { + "epoch": 40.01, + "learning_rate": 5.538461538461538e-08, + "loss": 0.0005, + "step": 19900 + }, + { + "epoch": 40.01, + "learning_rate": 4.2564102564102564e-08, + "loss": 0.0007, + "step": 19925 + }, + { + "epoch": 40.02, + "learning_rate": 2.9743589743589746e-08, + "loss": 0.0006, + "step": 19950 + }, + { + "epoch": 40.02, + "learning_rate": 1.6923076923076925e-08, + "loss": 0.0007, + "step": 19975 + }, + { + "epoch": 40.02, + "learning_rate": 4.102564102564102e-09, + "loss": 0.0008, + "step": 20000 + }, + { + "epoch": 40.02, + "eval_loss": 0.26569172739982605, + "eval_runtime": 5464.8935, + "eval_samples_per_second": 2.874, + "eval_steps_per_second": 0.18, + "eval_wer": 5.022549830956459, + "step": 20000 + }, + { + "epoch": 40.02, + "step": 20000, + "total_flos": 2.716979491897344e+21, + "train_loss": 0.03653936268617399, + "train_runtime": 463534.7089, + "train_samples_per_second": 2.761, + "train_steps_per_second": 0.043 + } + ], + "logging_steps": 25, + "max_steps": 20000, + "num_train_epochs": 9223372036854775807, + "save_steps": 1000, + "total_flos": 2.716979491897344e+21, + "trial_name": null, + "trial_params": null +}