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": 5.126477928109984, + "best_model_checkpoint": "./checkpoint-20000", + "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": 4.800000000000001e-07, + "loss": 1.0129, + "step": 25 + }, + { + "epoch": 0.0, + "learning_rate": 9.800000000000001e-07, + "loss": 0.6637, + "step": 50 + }, + { + "epoch": 0.0, + "learning_rate": 1.48e-06, + "loss": 0.2988, + "step": 75 + }, + { + "epoch": 0.01, + "learning_rate": 1.98e-06, + "loss": 0.1983, + "step": 100 + }, + { + "epoch": 0.01, + "learning_rate": 2.4800000000000004e-06, + "loss": 0.1791, + "step": 125 + }, + { + "epoch": 0.01, + "learning_rate": 2.9800000000000003e-06, + "loss": 0.1697, + "step": 150 + }, + { + "epoch": 0.01, + "learning_rate": 3.48e-06, + "loss": 0.1687, + "step": 175 + }, + { + "epoch": 0.01, + "learning_rate": 3.980000000000001e-06, + "loss": 0.1646, + "step": 200 + }, + { + "epoch": 0.01, + "learning_rate": 4.48e-06, + "loss": 0.1726, + "step": 225 + }, + { + "epoch": 0.01, + "learning_rate": 4.980000000000001e-06, + "loss": 0.2303, + "step": 250 + }, + { + "epoch": 0.01, + "learning_rate": 5.480000000000001e-06, + "loss": 0.2275, + "step": 275 + }, + { + "epoch": 0.01, + "learning_rate": 5.98e-06, + "loss": 0.2175, + "step": 300 + }, + { + "epoch": 0.02, + "learning_rate": 6.480000000000001e-06, + "loss": 0.1919, + "step": 325 + }, + { + "epoch": 0.02, + "learning_rate": 6.98e-06, + "loss": 0.2026, + "step": 350 + }, + { + "epoch": 0.02, + "learning_rate": 7.48e-06, + "loss": 0.2155, + "step": 375 + }, + { + "epoch": 0.02, + "learning_rate": 7.980000000000002e-06, + "loss": 0.1923, + "step": 400 + }, + { + "epoch": 0.02, + "learning_rate": 8.48e-06, + "loss": 0.1793, + "step": 425 + }, + { + "epoch": 0.02, + "learning_rate": 8.98e-06, + "loss": 0.1332, + "step": 450 + }, + { + "epoch": 0.02, + "learning_rate": 9.48e-06, + "loss": 0.1336, + "step": 475 + }, + { + "epoch": 1.0, + "learning_rate": 9.980000000000001e-06, + "loss": 0.1497, + "step": 500 + }, + { + "epoch": 1.0, + "learning_rate": 9.987692307692308e-06, + "loss": 0.1546, + "step": 525 + }, + { + "epoch": 1.0, + "learning_rate": 9.974871794871795e-06, + "loss": 0.1912, + "step": 550 + }, + { + "epoch": 1.0, + "learning_rate": 9.962051282051282e-06, + "loss": 0.126, + "step": 575 + }, + { + "epoch": 1.01, + "learning_rate": 9.94923076923077e-06, + "loss": 0.123, + "step": 600 + }, + { + "epoch": 1.01, + "learning_rate": 9.936410256410257e-06, + "loss": 0.1229, + "step": 625 + }, + { + "epoch": 1.01, + "learning_rate": 9.923589743589744e-06, + "loss": 0.1053, + "step": 650 + }, + { + "epoch": 1.01, + "learning_rate": 9.910769230769231e-06, + "loss": 0.1039, + "step": 675 + }, + { + "epoch": 1.01, + "learning_rate": 9.89794871794872e-06, + "loss": 0.0984, + "step": 700 + }, + { + "epoch": 1.01, + "learning_rate": 9.885128205128206e-06, + "loss": 0.1279, + "step": 725 + }, + { + "epoch": 1.01, + "learning_rate": 9.872307692307693e-06, + "loss": 0.1202, + "step": 750 + }, + { + "epoch": 1.01, + "learning_rate": 9.85948717948718e-06, + "loss": 0.1182, + "step": 775 + }, + { + "epoch": 1.02, + "learning_rate": 9.846666666666668e-06, + "loss": 0.1009, + "step": 800 + }, + { + "epoch": 1.02, + "learning_rate": 9.833846153846155e-06, + "loss": 0.0962, + "step": 825 + }, + { + "epoch": 1.02, + "learning_rate": 9.821025641025642e-06, + "loss": 0.097, + "step": 850 + }, + { + "epoch": 1.02, + "learning_rate": 9.808205128205129e-06, + "loss": 0.0947, + "step": 875 + }, + { + "epoch": 1.02, + "learning_rate": 9.795384615384617e-06, + "loss": 0.0835, + "step": 900 + }, + { + "epoch": 1.02, + "learning_rate": 9.782564102564104e-06, + "loss": 0.066, + "step": 925 + }, + { + "epoch": 1.02, + "learning_rate": 9.769743589743591e-06, + "loss": 0.0581, + "step": 950 + }, + { + "epoch": 1.02, + "learning_rate": 9.756923076923078e-06, + "loss": 0.0645, + "step": 975 + }, + { + "epoch": 2.0, + "learning_rate": 9.744102564102565e-06, + "loss": 0.0834, + "step": 1000 + }, + { + "epoch": 2.0, + "eval_loss": 0.1861874759197235, + "eval_runtime": 6775.6214, + "eval_samples_per_second": 2.318, + "eval_steps_per_second": 0.145, + "eval_wer": 6.385162660302675, + "step": 1000 + }, + { + "epoch": 2.0, + "learning_rate": 9.731282051282052e-06, + "loss": 0.1097, + "step": 1025 + }, + { + "epoch": 2.0, + "learning_rate": 9.718461538461538e-06, + "loss": 0.1908, + "step": 1050 + }, + { + "epoch": 2.0, + "learning_rate": 9.705641025641027e-06, + "loss": 0.1416, + "step": 1075 + }, + { + "epoch": 2.01, + "learning_rate": 9.692820512820514e-06, + "loss": 0.1435, + "step": 1100 + }, + { + "epoch": 2.01, + "learning_rate": 9.68e-06, + "loss": 0.1094, + "step": 1125 + }, + { + "epoch": 2.01, + "learning_rate": 9.667179487179487e-06, + "loss": 0.1308, + "step": 1150 + }, + { + "epoch": 2.01, + "learning_rate": 9.654358974358974e-06, + "loss": 0.1309, + "step": 1175 + }, + { + "epoch": 2.01, + "learning_rate": 9.641538461538463e-06, + "loss": 0.1261, + "step": 1200 + }, + { + "epoch": 2.01, + "learning_rate": 9.62871794871795e-06, + "loss": 0.0827, + "step": 1225 + }, + { + "epoch": 2.01, + "learning_rate": 9.615897435897436e-06, + "loss": 0.0854, + "step": 1250 + }, + { + "epoch": 2.01, + "learning_rate": 9.603076923076923e-06, + "loss": 0.0928, + "step": 1275 + }, + { + "epoch": 2.02, + "learning_rate": 9.590256410256412e-06, + "loss": 0.1191, + "step": 1300 + }, + { + "epoch": 2.02, + "learning_rate": 9.577435897435898e-06, + "loss": 0.1493, + "step": 1325 + }, + { + "epoch": 2.02, + "learning_rate": 9.564615384615385e-06, + "loss": 0.1776, + "step": 1350 + }, + { + "epoch": 2.02, + "learning_rate": 9.551794871794872e-06, + "loss": 0.1223, + "step": 1375 + }, + { + "epoch": 2.02, + "learning_rate": 9.53897435897436e-06, + "loss": 0.0958, + "step": 1400 + }, + { + "epoch": 2.02, + "learning_rate": 9.526153846153847e-06, + "loss": 0.1071, + "step": 1425 + }, + { + "epoch": 2.02, + "learning_rate": 9.513333333333334e-06, + "loss": 0.0877, + "step": 1450 + }, + { + "epoch": 3.0, + "learning_rate": 9.500512820512821e-06, + "loss": 0.0842, + "step": 1475 + }, + { + "epoch": 3.0, + "learning_rate": 9.48769230769231e-06, + "loss": 0.1353, + "step": 1500 + }, + { + "epoch": 3.0, + "learning_rate": 9.474871794871796e-06, + "loss": 0.1093, + "step": 1525 + }, + { + "epoch": 3.0, + "learning_rate": 9.462051282051283e-06, + "loss": 0.1079, + "step": 1550 + }, + { + "epoch": 3.01, + "learning_rate": 9.44923076923077e-06, + "loss": 0.1036, + "step": 1575 + }, + { + "epoch": 3.01, + "learning_rate": 9.436410256410257e-06, + "loss": 0.0986, + "step": 1600 + }, + { + "epoch": 3.01, + "learning_rate": 9.423589743589744e-06, + "loss": 0.0913, + "step": 1625 + }, + { + "epoch": 3.01, + "learning_rate": 9.41076923076923e-06, + "loss": 0.0986, + "step": 1650 + }, + { + "epoch": 3.01, + "learning_rate": 9.397948717948719e-06, + "loss": 0.1039, + "step": 1675 + }, + { + "epoch": 3.01, + "learning_rate": 9.385128205128206e-06, + "loss": 0.0924, + "step": 1700 + }, + { + "epoch": 3.01, + "learning_rate": 9.372307692307693e-06, + "loss": 0.0975, + "step": 1725 + }, + { + "epoch": 3.01, + "learning_rate": 9.35948717948718e-06, + "loss": 0.1019, + "step": 1750 + }, + { + "epoch": 3.02, + "learning_rate": 9.346666666666666e-06, + "loss": 0.0976, + "step": 1775 + }, + { + "epoch": 3.02, + "learning_rate": 9.333846153846155e-06, + "loss": 0.0925, + "step": 1800 + }, + { + "epoch": 3.02, + "learning_rate": 9.321025641025642e-06, + "loss": 0.0989, + "step": 1825 + }, + { + "epoch": 3.02, + "learning_rate": 9.308205128205129e-06, + "loss": 0.0993, + "step": 1850 + }, + { + "epoch": 3.02, + "learning_rate": 9.295384615384615e-06, + "loss": 0.1008, + "step": 1875 + }, + { + "epoch": 3.02, + "learning_rate": 9.282564102564104e-06, + "loss": 0.1005, + "step": 1900 + }, + { + "epoch": 3.02, + "learning_rate": 9.26974358974359e-06, + "loss": 0.0941, + "step": 1925 + }, + { + "epoch": 3.02, + "learning_rate": 9.256923076923078e-06, + "loss": 0.0903, + "step": 1950 + }, + { + "epoch": 4.0, + "learning_rate": 9.244102564102564e-06, + "loss": 0.0779, + "step": 1975 + }, + { + "epoch": 4.0, + "learning_rate": 9.231282051282053e-06, + "loss": 0.0871, + "step": 2000 + }, + { + "epoch": 4.0, + "eval_loss": 0.1776791661977768, + "eval_runtime": 5336.9309, + "eval_samples_per_second": 2.943, + "eval_steps_per_second": 0.184, + "eval_wer": 5.917486223111813, + "step": 2000 + }, + { + "epoch": 4.0, + "learning_rate": 9.21846153846154e-06, + "loss": 0.1027, + "step": 2025 + }, + { + "epoch": 4.0, + "learning_rate": 9.205641025641027e-06, + "loss": 0.0979, + "step": 2050 + }, + { + "epoch": 4.01, + "learning_rate": 9.192820512820513e-06, + "loss": 0.0919, + "step": 2075 + }, + { + "epoch": 4.01, + "learning_rate": 9.180000000000002e-06, + "loss": 0.099, + "step": 2100 + }, + { + "epoch": 4.01, + "learning_rate": 9.167179487179489e-06, + "loss": 0.0799, + "step": 2125 + }, + { + "epoch": 4.01, + "learning_rate": 9.154358974358976e-06, + "loss": 0.088, + "step": 2150 + }, + { + "epoch": 4.01, + "learning_rate": 9.141538461538462e-06, + "loss": 0.0911, + "step": 2175 + }, + { + "epoch": 4.01, + "learning_rate": 9.12871794871795e-06, + "loss": 0.0988, + "step": 2200 + }, + { + "epoch": 4.01, + "learning_rate": 9.115897435897436e-06, + "loss": 0.0947, + "step": 2225 + }, + { + "epoch": 4.01, + "learning_rate": 9.103076923076923e-06, + "loss": 0.0913, + "step": 2250 + }, + { + "epoch": 4.02, + "learning_rate": 9.090256410256411e-06, + "loss": 0.0959, + "step": 2275 + }, + { + "epoch": 4.02, + "learning_rate": 9.077435897435898e-06, + "loss": 0.1005, + "step": 2300 + }, + { + "epoch": 4.02, + "learning_rate": 9.064615384615385e-06, + "loss": 0.0893, + "step": 2325 + }, + { + "epoch": 4.02, + "learning_rate": 9.051794871794872e-06, + "loss": 0.1036, + "step": 2350 + }, + { + "epoch": 4.02, + "learning_rate": 9.038974358974359e-06, + "loss": 0.0917, + "step": 2375 + }, + { + "epoch": 4.02, + "learning_rate": 9.026153846153847e-06, + "loss": 0.0901, + "step": 2400 + }, + { + "epoch": 4.02, + "learning_rate": 9.013333333333334e-06, + "loss": 0.083, + "step": 2425 + }, + { + "epoch": 4.02, + "learning_rate": 9.00051282051282e-06, + "loss": 0.0828, + "step": 2450 + }, + { + "epoch": 5.0, + "learning_rate": 8.987692307692308e-06, + "loss": 0.0855, + "step": 2475 + }, + { + "epoch": 5.0, + "learning_rate": 8.974871794871796e-06, + "loss": 0.0883, + "step": 2500 + }, + { + "epoch": 5.0, + "learning_rate": 8.962051282051283e-06, + "loss": 0.1207, + "step": 2525 + }, + { + "epoch": 5.0, + "learning_rate": 8.94923076923077e-06, + "loss": 0.1097, + "step": 2550 + }, + { + "epoch": 5.01, + "learning_rate": 8.936410256410257e-06, + "loss": 0.0855, + "step": 2575 + }, + { + "epoch": 5.01, + "learning_rate": 8.923589743589745e-06, + "loss": 0.0799, + "step": 2600 + }, + { + "epoch": 5.01, + "learning_rate": 8.910769230769232e-06, + "loss": 0.0715, + "step": 2625 + }, + { + "epoch": 5.01, + "learning_rate": 8.897948717948719e-06, + "loss": 0.1086, + "step": 2650 + }, + { + "epoch": 5.01, + "learning_rate": 8.885128205128206e-06, + "loss": 0.0962, + "step": 2675 + }, + { + "epoch": 5.01, + "learning_rate": 8.872307692307694e-06, + "loss": 0.0738, + "step": 2700 + }, + { + "epoch": 5.01, + "learning_rate": 8.859487179487181e-06, + "loss": 0.0893, + "step": 2725 + }, + { + "epoch": 5.01, + "learning_rate": 8.846666666666668e-06, + "loss": 0.0862, + "step": 2750 + }, + { + "epoch": 5.02, + "learning_rate": 8.833846153846155e-06, + "loss": 0.0706, + "step": 2775 + }, + { + "epoch": 5.02, + "learning_rate": 8.821025641025643e-06, + "loss": 0.0977, + "step": 2800 + }, + { + "epoch": 5.02, + "learning_rate": 8.808205128205128e-06, + "loss": 0.226, + "step": 2825 + }, + { + "epoch": 5.02, + "learning_rate": 8.795384615384615e-06, + "loss": 0.109, + "step": 2850 + }, + { + "epoch": 5.02, + "learning_rate": 8.782564102564104e-06, + "loss": 0.1546, + "step": 2875 + }, + { + "epoch": 5.02, + "learning_rate": 8.76974358974359e-06, + "loss": 0.1038, + "step": 2900 + }, + { + "epoch": 5.02, + "learning_rate": 8.756923076923077e-06, + "loss": 0.1091, + "step": 2925 + }, + { + "epoch": 6.0, + "learning_rate": 8.744102564102564e-06, + "loss": 0.1451, + "step": 2950 + }, + { + "epoch": 6.0, + "learning_rate": 8.731282051282051e-06, + "loss": 0.0399, + "step": 2975 + }, + { + "epoch": 6.0, + "learning_rate": 8.71846153846154e-06, + "loss": 0.039, + "step": 3000 + }, + { + "epoch": 6.0, + "eval_loss": 0.17801089584827423, + "eval_runtime": 5418.5565, + "eval_samples_per_second": 2.899, + "eval_steps_per_second": 0.181, + "eval_wer": 5.742348133464206, + "step": 3000 + }, + { + "epoch": 6.0, + "learning_rate": 8.705641025641026e-06, + "loss": 0.0395, + "step": 3025 + }, + { + "epoch": 6.01, + "learning_rate": 8.692820512820513e-06, + "loss": 0.0338, + "step": 3050 + }, + { + "epoch": 6.01, + "learning_rate": 8.68e-06, + "loss": 0.034, + "step": 3075 + }, + { + "epoch": 6.01, + "learning_rate": 8.667179487179488e-06, + "loss": 0.0346, + "step": 3100 + }, + { + "epoch": 6.01, + "learning_rate": 8.654358974358975e-06, + "loss": 0.0359, + "step": 3125 + }, + { + "epoch": 6.01, + "learning_rate": 8.641538461538462e-06, + "loss": 0.0307, + "step": 3150 + }, + { + "epoch": 6.01, + "learning_rate": 8.628717948717949e-06, + "loss": 0.0334, + "step": 3175 + }, + { + "epoch": 6.01, + "learning_rate": 8.615897435897437e-06, + "loss": 0.0329, + "step": 3200 + }, + { + "epoch": 6.01, + "learning_rate": 8.603076923076924e-06, + "loss": 0.0337, + "step": 3225 + }, + { + "epoch": 6.02, + "learning_rate": 8.590256410256411e-06, + "loss": 0.0315, + "step": 3250 + }, + { + "epoch": 6.02, + "learning_rate": 8.577435897435898e-06, + "loss": 0.0319, + "step": 3275 + }, + { + "epoch": 6.02, + "learning_rate": 8.564615384615386e-06, + "loss": 0.036, + "step": 3300 + }, + { + "epoch": 6.02, + "learning_rate": 8.551794871794873e-06, + "loss": 0.0318, + "step": 3325 + }, + { + "epoch": 6.02, + "learning_rate": 8.53897435897436e-06, + "loss": 0.0361, + "step": 3350 + }, + { + "epoch": 6.02, + "learning_rate": 8.526153846153847e-06, + "loss": 0.0332, + "step": 3375 + }, + { + "epoch": 6.02, + "learning_rate": 8.513333333333335e-06, + "loss": 0.0321, + "step": 3400 + }, + { + "epoch": 6.02, + "learning_rate": 8.50051282051282e-06, + "loss": 0.0337, + "step": 3425 + }, + { + "epoch": 7.0, + "learning_rate": 8.487692307692307e-06, + "loss": 0.0464, + "step": 3450 + }, + { + "epoch": 7.0, + "learning_rate": 8.474871794871796e-06, + "loss": 0.0308, + "step": 3475 + }, + { + "epoch": 7.0, + "learning_rate": 8.462051282051283e-06, + "loss": 0.0681, + "step": 3500 + }, + { + "epoch": 7.0, + "learning_rate": 8.44923076923077e-06, + "loss": 0.0575, + "step": 3525 + }, + { + "epoch": 7.01, + "learning_rate": 8.436410256410256e-06, + "loss": 0.0456, + "step": 3550 + }, + { + "epoch": 7.01, + "learning_rate": 8.423589743589745e-06, + "loss": 0.0455, + "step": 3575 + }, + { + "epoch": 7.01, + "learning_rate": 8.410769230769232e-06, + "loss": 0.0453, + "step": 3600 + }, + { + "epoch": 7.01, + "learning_rate": 8.397948717948719e-06, + "loss": 0.0489, + "step": 3625 + }, + { + "epoch": 7.01, + "learning_rate": 8.385128205128205e-06, + "loss": 0.0498, + "step": 3650 + }, + { + "epoch": 7.01, + "learning_rate": 8.372307692307692e-06, + "loss": 0.0281, + "step": 3675 + }, + { + "epoch": 7.01, + "learning_rate": 8.35948717948718e-06, + "loss": 0.027, + "step": 3700 + }, + { + "epoch": 7.01, + "learning_rate": 8.346666666666668e-06, + "loss": 0.0284, + "step": 3725 + }, + { + "epoch": 7.02, + "learning_rate": 8.333846153846154e-06, + "loss": 0.0419, + "step": 3750 + }, + { + "epoch": 7.02, + "learning_rate": 8.321025641025641e-06, + "loss": 0.0557, + "step": 3775 + }, + { + "epoch": 7.02, + "learning_rate": 8.30820512820513e-06, + "loss": 0.0759, + "step": 3800 + }, + { + "epoch": 7.02, + "learning_rate": 8.295384615384616e-06, + "loss": 0.0458, + "step": 3825 + }, + { + "epoch": 7.02, + "learning_rate": 8.282564102564103e-06, + "loss": 0.031, + "step": 3850 + }, + { + "epoch": 7.02, + "learning_rate": 8.26974358974359e-06, + "loss": 0.0375, + "step": 3875 + }, + { + "epoch": 7.02, + "learning_rate": 8.256923076923079e-06, + "loss": 0.0307, + "step": 3900 + }, + { + "epoch": 7.02, + "learning_rate": 8.244102564102565e-06, + "loss": 0.0283, + "step": 3925 + }, + { + "epoch": 8.0, + "learning_rate": 8.231282051282052e-06, + "loss": 0.0138, + "step": 3950 + }, + { + "epoch": 8.0, + "learning_rate": 8.218461538461539e-06, + "loss": 0.0126, + "step": 3975 + }, + { + "epoch": 8.0, + "learning_rate": 8.205641025641028e-06, + "loss": 0.0265, + "step": 4000 + }, + { + "epoch": 8.0, + "eval_loss": 0.21207746863365173, + "eval_runtime": 5376.2988, + "eval_samples_per_second": 2.922, + "eval_steps_per_second": 0.183, + "eval_wer": 5.774424706659738, + "step": 4000 + }, + { + "epoch": 8.0, + "learning_rate": 8.192820512820513e-06, + "loss": 0.016, + "step": 4025 + }, + { + "epoch": 8.01, + "learning_rate": 8.18e-06, + "loss": 0.0155, + "step": 4050 + }, + { + "epoch": 8.01, + "learning_rate": 8.167179487179488e-06, + "loss": 0.0118, + "step": 4075 + }, + { + "epoch": 8.01, + "learning_rate": 8.154358974358975e-06, + "loss": 0.0123, + "step": 4100 + }, + { + "epoch": 8.01, + "learning_rate": 8.141538461538462e-06, + "loss": 0.0151, + "step": 4125 + }, + { + "epoch": 8.01, + "learning_rate": 8.128717948717949e-06, + "loss": 0.0131, + "step": 4150 + }, + { + "epoch": 8.01, + "learning_rate": 8.115897435897437e-06, + "loss": 0.0081, + "step": 4175 + }, + { + "epoch": 8.01, + "learning_rate": 8.103076923076924e-06, + "loss": 0.0069, + "step": 4200 + }, + { + "epoch": 8.01, + "learning_rate": 8.09025641025641e-06, + "loss": 0.0082, + "step": 4225 + }, + { + "epoch": 8.02, + "learning_rate": 8.077435897435898e-06, + "loss": 0.0153, + "step": 4250 + }, + { + "epoch": 8.02, + "learning_rate": 8.064615384615384e-06, + "loss": 0.0177, + "step": 4275 + }, + { + "epoch": 8.02, + "learning_rate": 8.051794871794873e-06, + "loss": 0.0236, + "step": 4300 + }, + { + "epoch": 8.02, + "learning_rate": 8.03897435897436e-06, + "loss": 0.012, + "step": 4325 + }, + { + "epoch": 8.02, + "learning_rate": 8.026153846153847e-06, + "loss": 0.0079, + "step": 4350 + }, + { + "epoch": 8.02, + "learning_rate": 8.013333333333333e-06, + "loss": 0.012, + "step": 4375 + }, + { + "epoch": 8.02, + "learning_rate": 8.000512820512822e-06, + "loss": 0.0076, + "step": 4400 + }, + { + "epoch": 9.0, + "learning_rate": 7.987692307692309e-06, + "loss": 0.0101, + "step": 4425 + }, + { + "epoch": 9.0, + "learning_rate": 7.974871794871796e-06, + "loss": 0.0154, + "step": 4450 + }, + { + "epoch": 9.0, + "learning_rate": 7.962051282051282e-06, + "loss": 0.0142, + "step": 4475 + }, + { + "epoch": 9.0, + "learning_rate": 7.949230769230771e-06, + "loss": 0.0163, + "step": 4500 + }, + { + "epoch": 9.01, + "learning_rate": 7.936410256410258e-06, + "loss": 0.0151, + "step": 4525 + }, + { + "epoch": 9.01, + "learning_rate": 7.923589743589745e-06, + "loss": 0.0111, + "step": 4550 + }, + { + "epoch": 9.01, + "learning_rate": 7.910769230769231e-06, + "loss": 0.0136, + "step": 4575 + }, + { + "epoch": 9.01, + "learning_rate": 7.89794871794872e-06, + "loss": 0.0142, + "step": 4600 + }, + { + "epoch": 9.01, + "learning_rate": 7.885128205128207e-06, + "loss": 0.0109, + "step": 4625 + }, + { + "epoch": 9.01, + "learning_rate": 7.872307692307692e-06, + "loss": 0.0139, + "step": 4650 + }, + { + "epoch": 9.01, + "learning_rate": 7.85948717948718e-06, + "loss": 0.012, + "step": 4675 + }, + { + "epoch": 9.01, + "learning_rate": 7.846666666666667e-06, + "loss": 0.0124, + "step": 4700 + }, + { + "epoch": 9.02, + "learning_rate": 7.833846153846154e-06, + "loss": 0.0127, + "step": 4725 + }, + { + "epoch": 9.02, + "learning_rate": 7.821025641025641e-06, + "loss": 0.0126, + "step": 4750 + }, + { + "epoch": 9.02, + "learning_rate": 7.80820512820513e-06, + "loss": 0.0132, + "step": 4775 + }, + { + "epoch": 9.02, + "learning_rate": 7.795384615384616e-06, + "loss": 0.0132, + "step": 4800 + }, + { + "epoch": 9.02, + "learning_rate": 7.782564102564103e-06, + "loss": 0.0124, + "step": 4825 + }, + { + "epoch": 9.02, + "learning_rate": 7.76974358974359e-06, + "loss": 0.0127, + "step": 4850 + }, + { + "epoch": 9.02, + "learning_rate": 7.756923076923077e-06, + "loss": 0.0106, + "step": 4875 + }, + { + "epoch": 9.02, + "learning_rate": 7.744102564102565e-06, + "loss": 0.0145, + "step": 4900 + }, + { + "epoch": 10.0, + "learning_rate": 7.731282051282052e-06, + "loss": 0.0138, + "step": 4925 + }, + { + "epoch": 10.0, + "learning_rate": 7.718461538461539e-06, + "loss": 0.0074, + "step": 4950 + }, + { + "epoch": 10.0, + "learning_rate": 7.705641025641026e-06, + "loss": 0.007, + "step": 4975 + }, + { + "epoch": 10.0, + "learning_rate": 7.692820512820514e-06, + "loss": 0.0059, + "step": 5000 + }, + { + "epoch": 10.0, + "eval_loss": 0.22191211581230164, + "eval_runtime": 5429.5678, + "eval_samples_per_second": 2.893, + "eval_steps_per_second": 0.181, + "eval_wer": 5.809708937174824, + "step": 5000 + }, + { + "epoch": 10.01, + "learning_rate": 7.680000000000001e-06, + "loss": 0.0047, + "step": 5025 + }, + { + "epoch": 10.01, + "learning_rate": 7.667179487179488e-06, + "loss": 0.0035, + "step": 5050 + }, + { + "epoch": 10.01, + "learning_rate": 7.654358974358975e-06, + "loss": 0.0055, + "step": 5075 + }, + { + "epoch": 10.01, + "learning_rate": 7.641538461538463e-06, + "loss": 0.0068, + "step": 5100 + }, + { + "epoch": 10.01, + "learning_rate": 7.62871794871795e-06, + "loss": 0.0059, + "step": 5125 + }, + { + "epoch": 10.01, + "learning_rate": 7.615897435897437e-06, + "loss": 0.0056, + "step": 5150 + }, + { + "epoch": 10.01, + "learning_rate": 7.6030769230769245e-06, + "loss": 0.0049, + "step": 5175 + }, + { + "epoch": 10.01, + "learning_rate": 7.590256410256411e-06, + "loss": 0.0046, + "step": 5200 + }, + { + "epoch": 10.02, + "learning_rate": 7.577435897435899e-06, + "loss": 0.0056, + "step": 5225 + }, + { + "epoch": 10.02, + "learning_rate": 7.564615384615385e-06, + "loss": 0.0045, + "step": 5250 + }, + { + "epoch": 10.02, + "learning_rate": 7.551794871794872e-06, + "loss": 0.005, + "step": 5275 + }, + { + "epoch": 10.02, + "learning_rate": 7.5389743589743595e-06, + "loss": 0.0047, + "step": 5300 + }, + { + "epoch": 10.02, + "learning_rate": 7.526153846153846e-06, + "loss": 0.0055, + "step": 5325 + }, + { + "epoch": 10.02, + "learning_rate": 7.513333333333334e-06, + "loss": 0.0053, + "step": 5350 + }, + { + "epoch": 10.02, + "learning_rate": 7.500512820512821e-06, + "loss": 0.0035, + "step": 5375 + }, + { + "epoch": 10.02, + "learning_rate": 7.4876923076923085e-06, + "loss": 0.0047, + "step": 5400 + }, + { + "epoch": 11.0, + "learning_rate": 7.474871794871795e-06, + "loss": 0.0048, + "step": 5425 + }, + { + "epoch": 11.0, + "learning_rate": 7.462051282051282e-06, + "loss": 0.0067, + "step": 5450 + }, + { + "epoch": 11.0, + "learning_rate": 7.44923076923077e-06, + "loss": 0.0113, + "step": 5475 + }, + { + "epoch": 11.0, + "learning_rate": 7.436410256410257e-06, + "loss": 0.0065, + "step": 5500 + }, + { + "epoch": 11.01, + "learning_rate": 7.423589743589744e-06, + "loss": 0.0078, + "step": 5525 + }, + { + "epoch": 11.01, + "learning_rate": 7.410769230769231e-06, + "loss": 0.0058, + "step": 5550 + }, + { + "epoch": 11.01, + "learning_rate": 7.397948717948719e-06, + "loss": 0.0064, + "step": 5575 + }, + { + "epoch": 11.01, + "learning_rate": 7.385128205128206e-06, + "loss": 0.0088, + "step": 5600 + }, + { + "epoch": 11.01, + "learning_rate": 7.372307692307693e-06, + "loss": 0.0063, + "step": 5625 + }, + { + "epoch": 11.01, + "learning_rate": 7.35948717948718e-06, + "loss": 0.0052, + "step": 5650 + }, + { + "epoch": 11.01, + "learning_rate": 7.346666666666668e-06, + "loss": 0.0056, + "step": 5675 + }, + { + "epoch": 11.01, + "learning_rate": 7.333846153846155e-06, + "loss": 0.0056, + "step": 5700 + }, + { + "epoch": 11.02, + "learning_rate": 7.321025641025642e-06, + "loss": 0.0077, + "step": 5725 + }, + { + "epoch": 11.02, + "learning_rate": 7.308205128205129e-06, + "loss": 0.0094, + "step": 5750 + }, + { + "epoch": 11.02, + "learning_rate": 7.295384615384617e-06, + "loss": 0.0116, + "step": 5775 + }, + { + "epoch": 11.02, + "learning_rate": 7.2825641025641036e-06, + "loss": 0.0069, + "step": 5800 + }, + { + "epoch": 11.02, + "learning_rate": 7.269743589743591e-06, + "loss": 0.0057, + "step": 5825 + }, + { + "epoch": 11.02, + "learning_rate": 7.256923076923078e-06, + "loss": 0.006, + "step": 5850 + }, + { + "epoch": 11.02, + "learning_rate": 7.244102564102564e-06, + "loss": 0.0067, + "step": 5875 + }, + { + "epoch": 12.0, + "learning_rate": 7.231282051282052e-06, + "loss": 0.0367, + "step": 5900 + }, + { + "epoch": 12.0, + "learning_rate": 7.2184615384615385e-06, + "loss": 0.0965, + "step": 5925 + }, + { + "epoch": 12.0, + "learning_rate": 7.205641025641026e-06, + "loss": 0.0879, + "step": 5950 + }, + { + "epoch": 12.0, + "learning_rate": 7.192820512820513e-06, + "loss": 0.084, + "step": 5975 + }, + { + "epoch": 12.01, + "learning_rate": 7.180000000000001e-06, + "loss": 0.0855, + "step": 6000 + }, + { + "epoch": 12.01, + "eval_loss": 0.18390582501888275, + "eval_runtime": 5436.7897, + "eval_samples_per_second": 2.889, + "eval_steps_per_second": 0.181, + "eval_wer": 5.977790180719413, + "step": 6000 + }, + { + "epoch": 12.01, + "learning_rate": 7.167692307692308e-06, + "loss": 0.0809, + "step": 6025 + }, + { + "epoch": 12.01, + "learning_rate": 7.154871794871795e-06, + "loss": 0.0833, + "step": 6050 + }, + { + "epoch": 12.01, + "learning_rate": 7.142051282051282e-06, + "loss": 0.0931, + "step": 6075 + }, + { + "epoch": 12.01, + "learning_rate": 7.1292307692307695e-06, + "loss": 0.0747, + "step": 6100 + }, + { + "epoch": 12.01, + "learning_rate": 7.116410256410256e-06, + "loss": 0.0834, + "step": 6125 + }, + { + "epoch": 12.01, + "learning_rate": 7.103589743589744e-06, + "loss": 0.0708, + "step": 6150 + }, + { + "epoch": 12.01, + "learning_rate": 7.090769230769231e-06, + "loss": 0.0725, + "step": 6175 + }, + { + "epoch": 12.02, + "learning_rate": 7.0779487179487184e-06, + "loss": 0.0769, + "step": 6200 + }, + { + "epoch": 12.02, + "learning_rate": 7.065128205128205e-06, + "loss": 0.0843, + "step": 6225 + }, + { + "epoch": 12.02, + "learning_rate": 7.052307692307693e-06, + "loss": 0.0866, + "step": 6250 + }, + { + "epoch": 12.02, + "learning_rate": 7.04e-06, + "loss": 0.0767, + "step": 6275 + }, + { + "epoch": 12.02, + "learning_rate": 7.027179487179487e-06, + "loss": 0.0873, + "step": 6300 + }, + { + "epoch": 12.02, + "learning_rate": 7.014358974358975e-06, + "loss": 0.0956, + "step": 6325 + }, + { + "epoch": 12.02, + "learning_rate": 7.001538461538462e-06, + "loss": 0.1365, + "step": 6350 + }, + { + "epoch": 12.02, + "learning_rate": 6.988717948717949e-06, + "loss": 0.081, + "step": 6375 + }, + { + "epoch": 13.0, + "learning_rate": 6.975897435897436e-06, + "loss": 0.0326, + "step": 6400 + }, + { + "epoch": 13.0, + "learning_rate": 6.963076923076924e-06, + "loss": 0.0057, + "step": 6425 + }, + { + "epoch": 13.0, + "learning_rate": 6.950256410256411e-06, + "loss": 0.0084, + "step": 6450 + }, + { + "epoch": 13.0, + "learning_rate": 6.937435897435898e-06, + "loss": 0.0071, + "step": 6475 + }, + { + "epoch": 13.01, + "learning_rate": 6.924615384615385e-06, + "loss": 0.0064, + "step": 6500 + }, + { + "epoch": 13.01, + "learning_rate": 6.911794871794872e-06, + "loss": 0.0052, + "step": 6525 + }, + { + "epoch": 13.01, + "learning_rate": 6.89897435897436e-06, + "loss": 0.0053, + "step": 6550 + }, + { + "epoch": 13.01, + "learning_rate": 6.8861538461538465e-06, + "loss": 0.0052, + "step": 6575 + }, + { + "epoch": 13.01, + "learning_rate": 6.873333333333334e-06, + "loss": 0.0048, + "step": 6600 + }, + { + "epoch": 13.01, + "learning_rate": 6.860512820512821e-06, + "loss": 0.0062, + "step": 6625 + }, + { + "epoch": 13.01, + "learning_rate": 6.847692307692309e-06, + "loss": 0.0031, + "step": 6650 + }, + { + "epoch": 13.01, + "learning_rate": 6.8348717948717955e-06, + "loss": 0.0043, + "step": 6675 + }, + { + "epoch": 13.02, + "learning_rate": 6.822051282051283e-06, + "loss": 0.0054, + "step": 6700 + }, + { + "epoch": 13.02, + "learning_rate": 6.80923076923077e-06, + "loss": 0.0047, + "step": 6725 + }, + { + "epoch": 13.02, + "learning_rate": 6.796410256410258e-06, + "loss": 0.0065, + "step": 6750 + }, + { + "epoch": 13.02, + "learning_rate": 6.7835897435897445e-06, + "loss": 0.0044, + "step": 6775 + }, + { + "epoch": 13.02, + "learning_rate": 6.770769230769232e-06, + "loss": 0.0035, + "step": 6800 + }, + { + "epoch": 13.02, + "learning_rate": 6.757948717948718e-06, + "loss": 0.0048, + "step": 6825 + }, + { + "epoch": 13.02, + "learning_rate": 6.745128205128205e-06, + "loss": 0.0039, + "step": 6850 + }, + { + "epoch": 14.0, + "learning_rate": 6.732307692307693e-06, + "loss": 0.0033, + "step": 6875 + }, + { + "epoch": 14.0, + "learning_rate": 6.7194871794871794e-06, + "loss": 0.004, + "step": 6900 + }, + { + "epoch": 14.0, + "learning_rate": 6.706666666666667e-06, + "loss": 0.0046, + "step": 6925 + }, + { + "epoch": 14.0, + "learning_rate": 6.693846153846154e-06, + "loss": 0.0046, + "step": 6950 + }, + { + "epoch": 14.01, + "learning_rate": 6.681025641025642e-06, + "loss": 0.0037, + "step": 6975 + }, + { + "epoch": 14.01, + "learning_rate": 6.6682051282051284e-06, + "loss": 0.0037, + "step": 7000 + }, + { + "epoch": 14.01, + "eval_loss": 0.22727246582508087, + "eval_runtime": 5481.4532, + "eval_samples_per_second": 2.866, + "eval_steps_per_second": 0.179, + "eval_wer": 5.856540734040301, + "step": 7000 + }, + { + "epoch": 14.01, + "learning_rate": 6.655384615384616e-06, + "loss": 0.0035, + "step": 7025 + }, + { + "epoch": 14.01, + "learning_rate": 6.642564102564103e-06, + "loss": 0.0034, + "step": 7050 + }, + { + "epoch": 14.01, + "learning_rate": 6.629743589743591e-06, + "loss": 0.004, + "step": 7075 + }, + { + "epoch": 14.01, + "learning_rate": 6.616923076923077e-06, + "loss": 0.0038, + "step": 7100 + }, + { + "epoch": 14.01, + "learning_rate": 6.604102564102565e-06, + "loss": 0.0034, + "step": 7125 + }, + { + "epoch": 14.01, + "learning_rate": 6.591282051282052e-06, + "loss": 0.004, + "step": 7150 + }, + { + "epoch": 14.02, + "learning_rate": 6.578461538461539e-06, + "loss": 0.0034, + "step": 7175 + }, + { + "epoch": 14.02, + "learning_rate": 6.565641025641026e-06, + "loss": 0.0044, + "step": 7200 + }, + { + "epoch": 14.02, + "learning_rate": 6.552820512820513e-06, + "loss": 0.0043, + "step": 7225 + }, + { + "epoch": 14.02, + "learning_rate": 6.540000000000001e-06, + "loss": 0.0043, + "step": 7250 + }, + { + "epoch": 14.02, + "learning_rate": 6.527179487179488e-06, + "loss": 0.0039, + "step": 7275 + }, + { + "epoch": 14.02, + "learning_rate": 6.514358974358975e-06, + "loss": 0.004, + "step": 7300 + }, + { + "epoch": 14.02, + "learning_rate": 6.501538461538462e-06, + "loss": 0.0039, + "step": 7325 + }, + { + "epoch": 14.02, + "learning_rate": 6.48871794871795e-06, + "loss": 0.0037, + "step": 7350 + }, + { + "epoch": 15.0, + "learning_rate": 6.475897435897437e-06, + "loss": 0.0204, + "step": 7375 + }, + { + "epoch": 15.0, + "learning_rate": 6.463076923076924e-06, + "loss": 0.0277, + "step": 7400 + }, + { + "epoch": 15.0, + "learning_rate": 6.45025641025641e-06, + "loss": 0.0534, + "step": 7425 + }, + { + "epoch": 15.0, + "learning_rate": 6.437435897435897e-06, + "loss": 0.0376, + "step": 7450 + }, + { + "epoch": 15.01, + "learning_rate": 6.424615384615385e-06, + "loss": 0.0395, + "step": 7475 + }, + { + "epoch": 15.01, + "learning_rate": 6.411794871794872e-06, + "loss": 0.032, + "step": 7500 + }, + { + "epoch": 15.01, + "learning_rate": 6.398974358974359e-06, + "loss": 0.0243, + "step": 7525 + }, + { + "epoch": 15.01, + "learning_rate": 6.386153846153846e-06, + "loss": 0.038, + "step": 7550 + }, + { + "epoch": 15.01, + "learning_rate": 6.373333333333334e-06, + "loss": 0.0395, + "step": 7575 + }, + { + "epoch": 15.01, + "learning_rate": 6.360512820512821e-06, + "loss": 0.0302, + "step": 7600 + }, + { + "epoch": 15.01, + "learning_rate": 6.347692307692308e-06, + "loss": 0.0326, + "step": 7625 + }, + { + "epoch": 15.01, + "learning_rate": 6.334871794871795e-06, + "loss": 0.0358, + "step": 7650 + }, + { + "epoch": 15.02, + "learning_rate": 6.322051282051283e-06, + "loss": 0.0315, + "step": 7675 + }, + { + "epoch": 15.02, + "learning_rate": 6.30923076923077e-06, + "loss": 0.0387, + "step": 7700 + }, + { + "epoch": 15.02, + "learning_rate": 6.296410256410257e-06, + "loss": 0.0685, + "step": 7725 + }, + { + "epoch": 15.02, + "learning_rate": 6.283589743589744e-06, + "loss": 0.0762, + "step": 7750 + }, + { + "epoch": 15.02, + "learning_rate": 6.270769230769231e-06, + "loss": 0.0484, + "step": 7775 + }, + { + "epoch": 15.02, + "learning_rate": 6.257948717948719e-06, + "loss": 0.0619, + "step": 7800 + }, + { + "epoch": 15.02, + "learning_rate": 6.2451282051282055e-06, + "loss": 0.0306, + "step": 7825 + }, + { + "epoch": 15.02, + "learning_rate": 6.232307692307693e-06, + "loss": 0.0755, + "step": 7850 + }, + { + "epoch": 16.0, + "learning_rate": 6.21948717948718e-06, + "loss": 0.0438, + "step": 7875 + }, + { + "epoch": 16.0, + "learning_rate": 6.206666666666668e-06, + "loss": 0.0386, + "step": 7900 + }, + { + "epoch": 16.0, + "learning_rate": 6.1938461538461545e-06, + "loss": 0.0417, + "step": 7925 + }, + { + "epoch": 16.0, + "learning_rate": 6.181025641025642e-06, + "loss": 0.0346, + "step": 7950 + }, + { + "epoch": 16.01, + "learning_rate": 6.168205128205129e-06, + "loss": 0.0372, + "step": 7975 + }, + { + "epoch": 16.01, + "learning_rate": 6.155384615384617e-06, + "loss": 0.0293, + "step": 8000 + }, + { + "epoch": 16.01, + "eval_loss": 0.19646957516670227, + "eval_runtime": 5444.7932, + "eval_samples_per_second": 2.885, + "eval_steps_per_second": 0.18, + "eval_wer": 5.807784342783092, + "step": 8000 + }, + { + "epoch": 16.01, + "learning_rate": 6.142564102564103e-06, + "loss": 0.0322, + "step": 8025 + }, + { + "epoch": 16.01, + "learning_rate": 6.1297435897435895e-06, + "loss": 0.0359, + "step": 8050 + }, + { + "epoch": 16.01, + "learning_rate": 6.116923076923077e-06, + "loss": 0.0359, + "step": 8075 + }, + { + "epoch": 16.01, + "learning_rate": 6.104102564102564e-06, + "loss": 0.0334, + "step": 8100 + }, + { + "epoch": 16.01, + "learning_rate": 6.091282051282052e-06, + "loss": 0.0345, + "step": 8125 + }, + { + "epoch": 16.01, + "learning_rate": 6.0784615384615384e-06, + "loss": 0.0329, + "step": 8150 + }, + { + "epoch": 16.02, + "learning_rate": 6.065641025641026e-06, + "loss": 0.0348, + "step": 8175 + }, + { + "epoch": 16.02, + "learning_rate": 6.052820512820513e-06, + "loss": 0.034, + "step": 8200 + }, + { + "epoch": 16.02, + "learning_rate": 6.040000000000001e-06, + "loss": 0.0361, + "step": 8225 + }, + { + "epoch": 16.02, + "learning_rate": 6.0271794871794874e-06, + "loss": 0.0361, + "step": 8250 + }, + { + "epoch": 16.02, + "learning_rate": 6.014358974358975e-06, + "loss": 0.0352, + "step": 8275 + }, + { + "epoch": 16.02, + "learning_rate": 6.001538461538462e-06, + "loss": 0.0289, + "step": 8300 + }, + { + "epoch": 16.02, + "learning_rate": 5.98871794871795e-06, + "loss": 0.0311, + "step": 8325 + }, + { + "epoch": 17.0, + "learning_rate": 5.9758974358974364e-06, + "loss": 0.0443, + "step": 8350 + }, + { + "epoch": 17.0, + "learning_rate": 5.963076923076923e-06, + "loss": 0.0052, + "step": 8375 + }, + { + "epoch": 17.0, + "learning_rate": 5.950256410256411e-06, + "loss": 0.0036, + "step": 8400 + }, + { + "epoch": 17.0, + "learning_rate": 5.937435897435898e-06, + "loss": 0.0039, + "step": 8425 + }, + { + "epoch": 17.01, + "learning_rate": 5.9246153846153854e-06, + "loss": 0.0029, + "step": 8450 + }, + { + "epoch": 17.01, + "learning_rate": 5.911794871794872e-06, + "loss": 0.0034, + "step": 8475 + }, + { + "epoch": 17.01, + "learning_rate": 5.89897435897436e-06, + "loss": 0.0027, + "step": 8500 + }, + { + "epoch": 17.01, + "learning_rate": 5.886153846153847e-06, + "loss": 0.0028, + "step": 8525 + }, + { + "epoch": 17.01, + "learning_rate": 5.873333333333334e-06, + "loss": 0.0024, + "step": 8550 + }, + { + "epoch": 17.01, + "learning_rate": 5.860512820512821e-06, + "loss": 0.0025, + "step": 8575 + }, + { + "epoch": 17.01, + "learning_rate": 5.847692307692309e-06, + "loss": 0.0032, + "step": 8600 + }, + { + "epoch": 17.01, + "learning_rate": 5.834871794871796e-06, + "loss": 0.0026, + "step": 8625 + }, + { + "epoch": 17.02, + "learning_rate": 5.822051282051282e-06, + "loss": 0.002, + "step": 8650 + }, + { + "epoch": 17.02, + "learning_rate": 5.809230769230769e-06, + "loss": 0.0025, + "step": 8675 + }, + { + "epoch": 17.02, + "learning_rate": 5.796410256410256e-06, + "loss": 0.0025, + "step": 8700 + }, + { + "epoch": 17.02, + "learning_rate": 5.783589743589744e-06, + "loss": 0.0028, + "step": 8725 + }, + { + "epoch": 17.02, + "learning_rate": 5.770769230769231e-06, + "loss": 0.0022, + "step": 8750 + }, + { + "epoch": 17.02, + "learning_rate": 5.757948717948718e-06, + "loss": 0.0023, + "step": 8775 + }, + { + "epoch": 17.02, + "learning_rate": 5.745128205128205e-06, + "loss": 0.0031, + "step": 8800 + }, + { + "epoch": 17.02, + "learning_rate": 5.732307692307693e-06, + "loss": 0.0029, + "step": 8825 + }, + { + "epoch": 18.0, + "learning_rate": 5.71948717948718e-06, + "loss": 0.0669, + "step": 8850 + }, + { + "epoch": 18.0, + "learning_rate": 5.706666666666667e-06, + "loss": 0.103, + "step": 8875 + }, + { + "epoch": 18.0, + "learning_rate": 5.693846153846154e-06, + "loss": 0.0896, + "step": 8900 + }, + { + "epoch": 18.0, + "learning_rate": 5.681025641025642e-06, + "loss": 0.1065, + "step": 8925 + }, + { + "epoch": 18.01, + "learning_rate": 5.668205128205129e-06, + "loss": 0.09, + "step": 8950 + }, + { + "epoch": 18.01, + "learning_rate": 5.6553846153846155e-06, + "loss": 0.0924, + "step": 8975 + }, + { + "epoch": 18.01, + "learning_rate": 5.642564102564103e-06, + "loss": 0.1174, + "step": 9000 + }, + { + "epoch": 18.01, + "eval_loss": 0.19844357669353485, + "eval_runtime": 5473.4057, + "eval_samples_per_second": 2.87, + "eval_steps_per_second": 0.179, + "eval_wer": 5.889258838699744, + "step": 9000 + }, + { + "epoch": 18.01, + "learning_rate": 5.62974358974359e-06, + "loss": 0.1218, + "step": 9025 + }, + { + "epoch": 18.01, + "learning_rate": 5.616923076923078e-06, + "loss": 0.085, + "step": 9050 + }, + { + "epoch": 18.01, + "learning_rate": 5.6041025641025645e-06, + "loss": 0.0917, + "step": 9075 + }, + { + "epoch": 18.01, + "learning_rate": 5.591282051282052e-06, + "loss": 0.1461, + "step": 9100 + }, + { + "epoch": 18.01, + "learning_rate": 5.578461538461539e-06, + "loss": 0.1269, + "step": 9125 + }, + { + "epoch": 18.02, + "learning_rate": 5.565641025641027e-06, + "loss": 0.1152, + "step": 9150 + }, + { + "epoch": 18.02, + "learning_rate": 5.5528205128205135e-06, + "loss": 0.1206, + "step": 9175 + }, + { + "epoch": 18.02, + "learning_rate": 5.540000000000001e-06, + "loss": 0.1251, + "step": 9200 + }, + { + "epoch": 18.02, + "learning_rate": 5.527179487179488e-06, + "loss": 0.1091, + "step": 9225 + }, + { + "epoch": 18.02, + "learning_rate": 5.514358974358974e-06, + "loss": 0.1205, + "step": 9250 + }, + { + "epoch": 18.02, + "learning_rate": 5.501538461538462e-06, + "loss": 0.1343, + "step": 9275 + }, + { + "epoch": 18.02, + "learning_rate": 5.4887179487179485e-06, + "loss": 0.1376, + "step": 9300 + }, + { + "epoch": 18.02, + "learning_rate": 5.475897435897436e-06, + "loss": 0.0926, + "step": 9325 + }, + { + "epoch": 19.0, + "learning_rate": 5.463076923076923e-06, + "loss": 0.0201, + "step": 9350 + }, + { + "epoch": 19.0, + "learning_rate": 5.450256410256411e-06, + "loss": 0.0032, + "step": 9375 + }, + { + "epoch": 19.0, + "learning_rate": 5.4374358974358975e-06, + "loss": 0.003, + "step": 9400 + }, + { + "epoch": 19.0, + "learning_rate": 5.424615384615385e-06, + "loss": 0.0029, + "step": 9425 + }, + { + "epoch": 19.01, + "learning_rate": 5.411794871794872e-06, + "loss": 0.0025, + "step": 9450 + }, + { + "epoch": 19.01, + "learning_rate": 5.39897435897436e-06, + "loss": 0.0022, + "step": 9475 + }, + { + "epoch": 19.01, + "learning_rate": 5.3861538461538464e-06, + "loss": 0.0026, + "step": 9500 + }, + { + "epoch": 19.01, + "learning_rate": 5.373333333333334e-06, + "loss": 0.002, + "step": 9525 + }, + { + "epoch": 19.01, + "learning_rate": 5.360512820512821e-06, + "loss": 0.0022, + "step": 9550 + }, + { + "epoch": 19.01, + "learning_rate": 5.347692307692308e-06, + "loss": 0.0021, + "step": 9575 + }, + { + "epoch": 19.01, + "learning_rate": 5.3348717948717954e-06, + "loss": 0.0016, + "step": 9600 + }, + { + "epoch": 19.01, + "learning_rate": 5.322051282051282e-06, + "loss": 0.0018, + "step": 9625 + }, + { + "epoch": 19.02, + "learning_rate": 5.30923076923077e-06, + "loss": 0.0012, + "step": 9650 + }, + { + "epoch": 19.02, + "learning_rate": 5.296410256410257e-06, + "loss": 0.0022, + "step": 9675 + }, + { + "epoch": 19.02, + "learning_rate": 5.2835897435897444e-06, + "loss": 0.0018, + "step": 9700 + }, + { + "epoch": 19.02, + "learning_rate": 5.270769230769231e-06, + "loss": 0.0016, + "step": 9725 + }, + { + "epoch": 19.02, + "learning_rate": 5.257948717948719e-06, + "loss": 0.0016, + "step": 9750 + }, + { + "epoch": 19.02, + "learning_rate": 5.245128205128206e-06, + "loss": 0.0016, + "step": 9775 + }, + { + "epoch": 19.02, + "learning_rate": 5.2323076923076934e-06, + "loss": 0.0014, + "step": 9800 + }, + { + "epoch": 20.0, + "learning_rate": 5.21948717948718e-06, + "loss": 0.0129, + "step": 9825 + }, + { + "epoch": 20.0, + "learning_rate": 5.206666666666668e-06, + "loss": 0.0553, + "step": 9850 + }, + { + "epoch": 20.0, + "learning_rate": 5.193846153846154e-06, + "loss": 0.0683, + "step": 9875 + }, + { + "epoch": 20.0, + "learning_rate": 5.181025641025641e-06, + "loss": 0.0551, + "step": 9900 + }, + { + "epoch": 20.01, + "learning_rate": 5.168205128205128e-06, + "loss": 0.0387, + "step": 9925 + }, + { + "epoch": 20.01, + "learning_rate": 5.155384615384615e-06, + "loss": 0.0463, + "step": 9950 + }, + { + "epoch": 20.01, + "learning_rate": 5.142564102564103e-06, + "loss": 0.0368, + "step": 9975 + }, + { + "epoch": 20.01, + "learning_rate": 5.12974358974359e-06, + "loss": 0.0355, + "step": 10000 + }, + { + "epoch": 20.01, + "eval_loss": 0.2135934829711914, + "eval_runtime": 5414.6286, + "eval_samples_per_second": 2.901, + "eval_steps_per_second": 0.181, + "eval_wer": 5.866163705998961, + "step": 10000 + }, + { + "epoch": 20.01, + "learning_rate": 5.116923076923077e-06, + "loss": 0.0359, + "step": 10025 + }, + { + "epoch": 20.01, + "learning_rate": 5.104102564102564e-06, + "loss": 0.0527, + "step": 10050 + }, + { + "epoch": 20.01, + "learning_rate": 5.091282051282052e-06, + "loss": 0.0508, + "step": 10075 + }, + { + "epoch": 20.01, + "learning_rate": 5.078461538461539e-06, + "loss": 0.0544, + "step": 10100 + }, + { + "epoch": 20.02, + "learning_rate": 5.065641025641026e-06, + "loss": 0.0423, + "step": 10125 + }, + { + "epoch": 20.02, + "learning_rate": 5.052820512820513e-06, + "loss": 0.0367, + "step": 10150 + }, + { + "epoch": 20.02, + "learning_rate": 5.040512820512821e-06, + "loss": 0.046, + "step": 10175 + }, + { + "epoch": 20.02, + "learning_rate": 5.027692307692308e-06, + "loss": 0.0351, + "step": 10200 + }, + { + "epoch": 20.02, + "learning_rate": 5.014871794871795e-06, + "loss": 0.0322, + "step": 10225 + }, + { + "epoch": 20.02, + "learning_rate": 5.002051282051282e-06, + "loss": 0.0245, + "step": 10250 + }, + { + "epoch": 20.02, + "learning_rate": 4.98923076923077e-06, + "loss": 0.0186, + "step": 10275 + }, + { + "epoch": 20.02, + "learning_rate": 4.9764102564102564e-06, + "loss": 0.0213, + "step": 10300 + }, + { + "epoch": 21.0, + "learning_rate": 4.963589743589744e-06, + "loss": 0.0169, + "step": 10325 + }, + { + "epoch": 21.0, + "learning_rate": 4.950769230769231e-06, + "loss": 0.0197, + "step": 10350 + }, + { + "epoch": 21.0, + "learning_rate": 4.937948717948719e-06, + "loss": 0.0267, + "step": 10375 + }, + { + "epoch": 21.0, + "learning_rate": 4.9251282051282054e-06, + "loss": 0.0095, + "step": 10400 + }, + { + "epoch": 21.01, + "learning_rate": 4.912307692307693e-06, + "loss": 0.0104, + "step": 10425 + }, + { + "epoch": 21.01, + "learning_rate": 4.89948717948718e-06, + "loss": 0.0105, + "step": 10450 + }, + { + "epoch": 21.01, + "learning_rate": 4.886666666666668e-06, + "loss": 0.0089, + "step": 10475 + }, + { + "epoch": 21.01, + "learning_rate": 4.873846153846154e-06, + "loss": 0.0079, + "step": 10500 + }, + { + "epoch": 21.01, + "learning_rate": 4.861025641025641e-06, + "loss": 0.0098, + "step": 10525 + }, + { + "epoch": 21.01, + "learning_rate": 4.848205128205128e-06, + "loss": 0.011, + "step": 10550 + }, + { + "epoch": 21.01, + "learning_rate": 4.835384615384616e-06, + "loss": 0.012, + "step": 10575 + }, + { + "epoch": 21.01, + "learning_rate": 4.8225641025641026e-06, + "loss": 0.0126, + "step": 10600 + }, + { + "epoch": 21.02, + "learning_rate": 4.80974358974359e-06, + "loss": 0.0093, + "step": 10625 + }, + { + "epoch": 21.02, + "learning_rate": 4.796923076923077e-06, + "loss": 0.0078, + "step": 10650 + }, + { + "epoch": 21.02, + "learning_rate": 4.784102564102565e-06, + "loss": 0.0135, + "step": 10675 + }, + { + "epoch": 21.02, + "learning_rate": 4.7712820512820516e-06, + "loss": 0.0073, + "step": 10700 + }, + { + "epoch": 21.02, + "learning_rate": 4.758461538461539e-06, + "loss": 0.007, + "step": 10725 + }, + { + "epoch": 21.02, + "learning_rate": 4.745641025641026e-06, + "loss": 0.0057, + "step": 10750 + }, + { + "epoch": 21.02, + "learning_rate": 4.732820512820514e-06, + "loss": 0.0049, + "step": 10775 + }, + { + "epoch": 21.02, + "learning_rate": 4.7200000000000005e-06, + "loss": 0.0047, + "step": 10800 + }, + { + "epoch": 22.0, + "learning_rate": 4.707179487179487e-06, + "loss": 0.0283, + "step": 10825 + }, + { + "epoch": 22.0, + "learning_rate": 4.694358974358974e-06, + "loss": 0.0247, + "step": 10850 + }, + { + "epoch": 22.0, + "learning_rate": 4.681538461538462e-06, + "loss": 0.0288, + "step": 10875 + }, + { + "epoch": 22.0, + "learning_rate": 4.668717948717949e-06, + "loss": 0.0321, + "step": 10900 + }, + { + "epoch": 22.01, + "learning_rate": 4.655897435897436e-06, + "loss": 0.0276, + "step": 10925 + }, + { + "epoch": 22.01, + "learning_rate": 4.643076923076923e-06, + "loss": 0.0269, + "step": 10950 + }, + { + "epoch": 22.01, + "learning_rate": 4.630256410256411e-06, + "loss": 0.0317, + "step": 10975 + }, + { + "epoch": 22.01, + "learning_rate": 4.617435897435898e-06, + "loss": 0.0279, + "step": 11000 + }, + { + "epoch": 22.01, + "eval_loss": 0.18820956349372864, + "eval_runtime": 5416.5672, + "eval_samples_per_second": 2.9, + "eval_steps_per_second": 0.181, + "eval_wer": 5.496000051322517, + "step": 11000 + }, + { + "epoch": 22.01, + "learning_rate": 4.604615384615385e-06, + "loss": 0.0264, + "step": 11025 + }, + { + "epoch": 22.01, + "learning_rate": 4.591794871794872e-06, + "loss": 0.021, + "step": 11050 + }, + { + "epoch": 22.01, + "learning_rate": 4.57897435897436e-06, + "loss": 0.0222, + "step": 11075 + }, + { + "epoch": 22.01, + "learning_rate": 4.566153846153847e-06, + "loss": 0.0235, + "step": 11100 + }, + { + "epoch": 22.02, + "learning_rate": 4.5533333333333335e-06, + "loss": 0.0227, + "step": 11125 + }, + { + "epoch": 22.02, + "learning_rate": 4.54051282051282e-06, + "loss": 0.03, + "step": 11150 + }, + { + "epoch": 22.02, + "learning_rate": 4.527692307692308e-06, + "loss": 0.0307, + "step": 11175 + }, + { + "epoch": 22.02, + "learning_rate": 4.514871794871795e-06, + "loss": 0.0286, + "step": 11200 + }, + { + "epoch": 22.02, + "learning_rate": 4.5020512820512825e-06, + "loss": 0.0308, + "step": 11225 + }, + { + "epoch": 22.02, + "learning_rate": 4.489230769230769e-06, + "loss": 0.0409, + "step": 11250 + }, + { + "epoch": 22.02, + "learning_rate": 4.476410256410257e-06, + "loss": 0.0392, + "step": 11275 + }, + { + "epoch": 23.0, + "learning_rate": 4.463589743589744e-06, + "loss": 0.0248, + "step": 11300 + }, + { + "epoch": 23.0, + "learning_rate": 4.4507692307692315e-06, + "loss": 0.0099, + "step": 11325 + }, + { + "epoch": 23.0, + "learning_rate": 4.437948717948718e-06, + "loss": 0.0157, + "step": 11350 + }, + { + "epoch": 23.0, + "learning_rate": 4.425128205128206e-06, + "loss": 0.0105, + "step": 11375 + }, + { + "epoch": 23.01, + "learning_rate": 4.412307692307693e-06, + "loss": 0.0135, + "step": 11400 + }, + { + "epoch": 23.01, + "learning_rate": 4.39948717948718e-06, + "loss": 0.0075, + "step": 11425 + }, + { + "epoch": 23.01, + "learning_rate": 4.3866666666666665e-06, + "loss": 0.0071, + "step": 11450 + }, + { + "epoch": 23.01, + "learning_rate": 4.373846153846154e-06, + "loss": 0.0088, + "step": 11475 + }, + { + "epoch": 23.01, + "learning_rate": 4.361025641025641e-06, + "loss": 0.0122, + "step": 11500 + }, + { + "epoch": 23.01, + "learning_rate": 4.348205128205129e-06, + "loss": 0.0095, + "step": 11525 + }, + { + "epoch": 23.01, + "learning_rate": 4.3353846153846154e-06, + "loss": 0.0097, + "step": 11550 + }, + { + "epoch": 23.01, + "learning_rate": 4.322564102564103e-06, + "loss": 0.0104, + "step": 11575 + }, + { + "epoch": 23.02, + "learning_rate": 4.30974358974359e-06, + "loss": 0.0088, + "step": 11600 + }, + { + "epoch": 23.02, + "learning_rate": 4.296923076923078e-06, + "loss": 0.0101, + "step": 11625 + }, + { + "epoch": 23.02, + "learning_rate": 4.2841025641025644e-06, + "loss": 0.0212, + "step": 11650 + }, + { + "epoch": 23.02, + "learning_rate": 4.271282051282052e-06, + "loss": 0.029, + "step": 11675 + }, + { + "epoch": 23.02, + "learning_rate": 4.258461538461539e-06, + "loss": 0.015, + "step": 11700 + }, + { + "epoch": 23.02, + "learning_rate": 4.245641025641026e-06, + "loss": 0.021, + "step": 11725 + }, + { + "epoch": 23.02, + "learning_rate": 4.232820512820513e-06, + "loss": 0.0103, + "step": 11750 + }, + { + "epoch": 23.02, + "learning_rate": 4.22e-06, + "loss": 0.0313, + "step": 11775 + }, + { + "epoch": 24.0, + "learning_rate": 4.207179487179487e-06, + "loss": 0.0215, + "step": 11800 + }, + { + "epoch": 24.0, + "learning_rate": 4.194358974358975e-06, + "loss": 0.0055, + "step": 11825 + }, + { + "epoch": 24.0, + "learning_rate": 4.1815384615384616e-06, + "loss": 0.0038, + "step": 11850 + }, + { + "epoch": 24.0, + "learning_rate": 4.168717948717949e-06, + "loss": 0.0057, + "step": 11875 + }, + { + "epoch": 24.01, + "learning_rate": 4.155897435897436e-06, + "loss": 0.004, + "step": 11900 + }, + { + "epoch": 24.01, + "learning_rate": 4.143076923076924e-06, + "loss": 0.0038, + "step": 11925 + }, + { + "epoch": 24.01, + "learning_rate": 4.1302564102564106e-06, + "loss": 0.0037, + "step": 11950 + }, + { + "epoch": 24.01, + "learning_rate": 4.117435897435898e-06, + "loss": 0.0032, + "step": 11975 + }, + { + "epoch": 24.01, + "learning_rate": 4.104615384615385e-06, + "loss": 0.0043, + "step": 12000 + }, + { + "epoch": 24.01, + "eval_loss": 0.24436695873737335, + "eval_runtime": 5398.0994, + "eval_samples_per_second": 2.91, + "eval_steps_per_second": 0.182, + "eval_wer": 5.335617185344856, + "step": 12000 + }, + { + "epoch": 24.01, + "learning_rate": 4.091794871794872e-06, + "loss": 0.0026, + "step": 12025 + }, + { + "epoch": 24.01, + "learning_rate": 4.078974358974359e-06, + "loss": 0.0018, + "step": 12050 + }, + { + "epoch": 24.01, + "learning_rate": 4.066153846153846e-06, + "loss": 0.0021, + "step": 12075 + }, + { + "epoch": 24.02, + "learning_rate": 4.053333333333333e-06, + "loss": 0.0022, + "step": 12100 + }, + { + "epoch": 24.02, + "learning_rate": 4.040512820512821e-06, + "loss": 0.0033, + "step": 12125 + }, + { + "epoch": 24.02, + "learning_rate": 4.027692307692308e-06, + "loss": 0.0046, + "step": 12150 + }, + { + "epoch": 24.02, + "learning_rate": 4.014871794871795e-06, + "loss": 0.0041, + "step": 12175 + }, + { + "epoch": 24.02, + "learning_rate": 4.002051282051282e-06, + "loss": 0.0025, + "step": 12200 + }, + { + "epoch": 24.02, + "learning_rate": 3.98923076923077e-06, + "loss": 0.0034, + "step": 12225 + }, + { + "epoch": 24.02, + "learning_rate": 3.976410256410257e-06, + "loss": 0.0024, + "step": 12250 + }, + { + "epoch": 24.02, + "learning_rate": 3.963589743589744e-06, + "loss": 0.0023, + "step": 12275 + }, + { + "epoch": 25.0, + "learning_rate": 3.950769230769231e-06, + "loss": 0.0043, + "step": 12300 + }, + { + "epoch": 25.0, + "learning_rate": 3.937948717948718e-06, + "loss": 0.0021, + "step": 12325 + }, + { + "epoch": 25.0, + "learning_rate": 3.925128205128205e-06, + "loss": 0.0041, + "step": 12350 + }, + { + "epoch": 25.0, + "learning_rate": 3.9123076923076925e-06, + "loss": 0.0031, + "step": 12375 + }, + { + "epoch": 25.01, + "learning_rate": 3.899487179487179e-06, + "loss": 0.003, + "step": 12400 + }, + { + "epoch": 25.01, + "learning_rate": 3.886666666666667e-06, + "loss": 0.0024, + "step": 12425 + }, + { + "epoch": 25.01, + "learning_rate": 3.873846153846154e-06, + "loss": 0.0018, + "step": 12450 + }, + { + "epoch": 25.01, + "learning_rate": 3.8610256410256415e-06, + "loss": 0.0026, + "step": 12475 + }, + { + "epoch": 25.01, + "learning_rate": 3.848205128205128e-06, + "loss": 0.0021, + "step": 12500 + }, + { + "epoch": 25.01, + "learning_rate": 3.835384615384616e-06, + "loss": 0.0024, + "step": 12525 + }, + { + "epoch": 25.01, + "learning_rate": 3.822564102564103e-06, + "loss": 0.0026, + "step": 12550 + }, + { + "epoch": 25.02, + "learning_rate": 3.80974358974359e-06, + "loss": 0.0023, + "step": 12575 + }, + { + "epoch": 25.02, + "learning_rate": 3.7969230769230773e-06, + "loss": 0.002, + "step": 12600 + }, + { + "epoch": 25.02, + "learning_rate": 3.784102564102564e-06, + "loss": 0.002, + "step": 12625 + }, + { + "epoch": 25.02, + "learning_rate": 3.7712820512820514e-06, + "loss": 0.0084, + "step": 12650 + }, + { + "epoch": 25.02, + "learning_rate": 3.7584615384615386e-06, + "loss": 0.0046, + "step": 12675 + }, + { + "epoch": 25.02, + "learning_rate": 3.745641025641026e-06, + "loss": 0.0046, + "step": 12700 + }, + { + "epoch": 25.02, + "learning_rate": 3.732820512820513e-06, + "loss": 0.0042, + "step": 12725 + }, + { + "epoch": 25.02, + "learning_rate": 3.7200000000000004e-06, + "loss": 0.0034, + "step": 12750 + }, + { + "epoch": 26.0, + "learning_rate": 3.7071794871794876e-06, + "loss": 0.024, + "step": 12775 + }, + { + "epoch": 26.0, + "learning_rate": 3.694358974358975e-06, + "loss": 0.0312, + "step": 12800 + }, + { + "epoch": 26.0, + "learning_rate": 3.681538461538462e-06, + "loss": 0.0295, + "step": 12825 + }, + { + "epoch": 26.0, + "learning_rate": 3.6687179487179494e-06, + "loss": 0.0337, + "step": 12850 + }, + { + "epoch": 26.01, + "learning_rate": 3.655897435897436e-06, + "loss": 0.0289, + "step": 12875 + }, + { + "epoch": 26.01, + "learning_rate": 3.6430769230769234e-06, + "loss": 0.0355, + "step": 12900 + }, + { + "epoch": 26.01, + "learning_rate": 3.6302564102564103e-06, + "loss": 0.0426, + "step": 12925 + }, + { + "epoch": 26.01, + "learning_rate": 3.6174358974358975e-06, + "loss": 0.0425, + "step": 12950 + }, + { + "epoch": 26.01, + "learning_rate": 3.6046153846153848e-06, + "loss": 0.0311, + "step": 12975 + }, + { + "epoch": 26.01, + "learning_rate": 3.591794871794872e-06, + "loss": 0.0302, + "step": 13000 + }, + { + "epoch": 26.01, + "eval_loss": 0.22225703299045563, + "eval_runtime": 5463.8248, + "eval_samples_per_second": 2.875, + "eval_steps_per_second": 0.18, + "eval_wer": 5.461998883735253, + "step": 13000 + }, + { + "epoch": 26.01, + "learning_rate": 3.5789743589743593e-06, + "loss": 0.0487, + "step": 13025 + }, + { + "epoch": 26.01, + "learning_rate": 3.5661538461538465e-06, + "loss": 0.0445, + "step": 13050 + }, + { + "epoch": 26.02, + "learning_rate": 3.5533333333333338e-06, + "loss": 0.0408, + "step": 13075 + }, + { + "epoch": 26.02, + "learning_rate": 3.540512820512821e-06, + "loss": 0.0428, + "step": 13100 + }, + { + "epoch": 26.02, + "learning_rate": 3.5276923076923083e-06, + "loss": 0.0489, + "step": 13125 + }, + { + "epoch": 26.02, + "learning_rate": 3.5148717948717955e-06, + "loss": 0.0406, + "step": 13150 + }, + { + "epoch": 26.02, + "learning_rate": 3.5020512820512823e-06, + "loss": 0.0464, + "step": 13175 + }, + { + "epoch": 26.02, + "learning_rate": 3.4892307692307696e-06, + "loss": 0.0462, + "step": 13200 + }, + { + "epoch": 26.02, + "learning_rate": 3.4764102564102564e-06, + "loss": 0.0548, + "step": 13225 + }, + { + "epoch": 26.02, + "learning_rate": 3.4635897435897436e-06, + "loss": 0.0363, + "step": 13250 + }, + { + "epoch": 27.0, + "learning_rate": 3.450769230769231e-06, + "loss": 0.0133, + "step": 13275 + }, + { + "epoch": 27.0, + "learning_rate": 3.437948717948718e-06, + "loss": 0.0085, + "step": 13300 + }, + { + "epoch": 27.0, + "learning_rate": 3.4251282051282054e-06, + "loss": 0.0071, + "step": 13325 + }, + { + "epoch": 27.0, + "learning_rate": 3.4123076923076926e-06, + "loss": 0.0072, + "step": 13350 + }, + { + "epoch": 27.01, + "learning_rate": 3.39948717948718e-06, + "loss": 0.0075, + "step": 13375 + }, + { + "epoch": 27.01, + "learning_rate": 3.386666666666667e-06, + "loss": 0.0075, + "step": 13400 + }, + { + "epoch": 27.01, + "learning_rate": 3.3738461538461544e-06, + "loss": 0.008, + "step": 13425 + }, + { + "epoch": 27.01, + "learning_rate": 3.3610256410256416e-06, + "loss": 0.0078, + "step": 13450 + }, + { + "epoch": 27.01, + "learning_rate": 3.3482051282051285e-06, + "loss": 0.005, + "step": 13475 + }, + { + "epoch": 27.01, + "learning_rate": 3.3353846153846157e-06, + "loss": 0.0046, + "step": 13500 + }, + { + "epoch": 27.01, + "learning_rate": 3.322564102564103e-06, + "loss": 0.0045, + "step": 13525 + }, + { + "epoch": 27.01, + "learning_rate": 3.3097435897435898e-06, + "loss": 0.0053, + "step": 13550 + }, + { + "epoch": 27.02, + "learning_rate": 3.296923076923077e-06, + "loss": 0.0054, + "step": 13575 + }, + { + "epoch": 27.02, + "learning_rate": 3.2841025641025643e-06, + "loss": 0.0064, + "step": 13600 + }, + { + "epoch": 27.02, + "learning_rate": 3.2712820512820515e-06, + "loss": 0.0065, + "step": 13625 + }, + { + "epoch": 27.02, + "learning_rate": 3.2584615384615388e-06, + "loss": 0.006, + "step": 13650 + }, + { + "epoch": 27.02, + "learning_rate": 3.245641025641026e-06, + "loss": 0.006, + "step": 13675 + }, + { + "epoch": 27.02, + "learning_rate": 3.2328205128205133e-06, + "loss": 0.0074, + "step": 13700 + }, + { + "epoch": 27.02, + "learning_rate": 3.2200000000000005e-06, + "loss": 0.0131, + "step": 13725 + }, + { + "epoch": 28.0, + "learning_rate": 3.2071794871794878e-06, + "loss": 0.0071, + "step": 13750 + }, + { + "epoch": 28.0, + "learning_rate": 3.194358974358975e-06, + "loss": 0.0012, + "step": 13775 + }, + { + "epoch": 28.0, + "learning_rate": 3.181538461538462e-06, + "loss": 0.0017, + "step": 13800 + }, + { + "epoch": 28.0, + "learning_rate": 3.168717948717949e-06, + "loss": 0.0023, + "step": 13825 + }, + { + "epoch": 28.01, + "learning_rate": 3.155897435897436e-06, + "loss": 0.0016, + "step": 13850 + }, + { + "epoch": 28.01, + "learning_rate": 3.143076923076923e-06, + "loss": 0.0018, + "step": 13875 + }, + { + "epoch": 28.01, + "learning_rate": 3.1302564102564104e-06, + "loss": 0.0013, + "step": 13900 + }, + { + "epoch": 28.01, + "learning_rate": 3.1174358974358976e-06, + "loss": 0.0013, + "step": 13925 + }, + { + "epoch": 28.01, + "learning_rate": 3.104615384615385e-06, + "loss": 0.0014, + "step": 13950 + }, + { + "epoch": 28.01, + "learning_rate": 3.091794871794872e-06, + "loss": 0.0015, + "step": 13975 + }, + { + "epoch": 28.01, + "learning_rate": 3.0789743589743594e-06, + "loss": 0.0011, + "step": 14000 + }, + { + "epoch": 28.01, + "eval_loss": 0.26026812195777893, + "eval_runtime": 5509.8929, + "eval_samples_per_second": 2.851, + "eval_steps_per_second": 0.178, + "eval_wer": 5.560794729177493, + "step": 14000 + }, + { + "epoch": 28.01, + "learning_rate": 3.0661538461538466e-06, + "loss": 0.001, + "step": 14025 + }, + { + "epoch": 28.02, + "learning_rate": 3.053333333333334e-06, + "loss": 0.0011, + "step": 14050 + }, + { + "epoch": 28.02, + "learning_rate": 3.040512820512821e-06, + "loss": 0.0015, + "step": 14075 + }, + { + "epoch": 28.02, + "learning_rate": 3.027692307692308e-06, + "loss": 0.0015, + "step": 14100 + }, + { + "epoch": 28.02, + "learning_rate": 3.014871794871795e-06, + "loss": 0.0022, + "step": 14125 + }, + { + "epoch": 28.02, + "learning_rate": 3.002051282051282e-06, + "loss": 0.0012, + "step": 14150 + }, + { + "epoch": 28.02, + "learning_rate": 2.9892307692307693e-06, + "loss": 0.0008, + "step": 14175 + }, + { + "epoch": 28.02, + "learning_rate": 2.9764102564102565e-06, + "loss": 0.001, + "step": 14200 + }, + { + "epoch": 28.02, + "learning_rate": 2.9635897435897438e-06, + "loss": 0.0012, + "step": 14225 + }, + { + "epoch": 29.0, + "learning_rate": 2.950769230769231e-06, + "loss": 0.0015, + "step": 14250 + }, + { + "epoch": 29.0, + "learning_rate": 2.9379487179487183e-06, + "loss": 0.0004, + "step": 14275 + }, + { + "epoch": 29.0, + "learning_rate": 2.9251282051282055e-06, + "loss": 0.0008, + "step": 14300 + }, + { + "epoch": 29.0, + "learning_rate": 2.9123076923076928e-06, + "loss": 0.0007, + "step": 14325 + }, + { + "epoch": 29.01, + "learning_rate": 2.89948717948718e-06, + "loss": 0.0006, + "step": 14350 + }, + { + "epoch": 29.01, + "learning_rate": 2.8866666666666673e-06, + "loss": 0.0006, + "step": 14375 + }, + { + "epoch": 29.01, + "learning_rate": 2.873846153846154e-06, + "loss": 0.0005, + "step": 14400 + }, + { + "epoch": 29.01, + "learning_rate": 2.8610256410256413e-06, + "loss": 0.0005, + "step": 14425 + }, + { + "epoch": 29.01, + "learning_rate": 2.848205128205128e-06, + "loss": 0.0005, + "step": 14450 + }, + { + "epoch": 29.01, + "learning_rate": 2.8353846153846154e-06, + "loss": 0.0004, + "step": 14475 + }, + { + "epoch": 29.01, + "learning_rate": 2.8225641025641027e-06, + "loss": 0.0004, + "step": 14500 + }, + { + "epoch": 29.01, + "learning_rate": 2.80974358974359e-06, + "loss": 0.0004, + "step": 14525 + }, + { + "epoch": 29.02, + "learning_rate": 2.796923076923077e-06, + "loss": 0.0004, + "step": 14550 + }, + { + "epoch": 29.02, + "learning_rate": 2.7841025641025644e-06, + "loss": 0.0006, + "step": 14575 + }, + { + "epoch": 29.02, + "learning_rate": 2.7712820512820516e-06, + "loss": 0.0007, + "step": 14600 + }, + { + "epoch": 29.02, + "learning_rate": 2.758461538461539e-06, + "loss": 0.0006, + "step": 14625 + }, + { + "epoch": 29.02, + "learning_rate": 2.745641025641026e-06, + "loss": 0.0004, + "step": 14650 + }, + { + "epoch": 29.02, + "learning_rate": 2.7328205128205134e-06, + "loss": 0.0006, + "step": 14675 + }, + { + "epoch": 29.02, + "learning_rate": 2.7200000000000002e-06, + "loss": 0.0005, + "step": 14700 + }, + { + "epoch": 29.02, + "learning_rate": 2.7071794871794875e-06, + "loss": 0.0006, + "step": 14725 + }, + { + "epoch": 30.0, + "learning_rate": 2.6943589743589743e-06, + "loss": 0.0007, + "step": 14750 + }, + { + "epoch": 30.0, + "learning_rate": 2.6815384615384615e-06, + "loss": 0.0008, + "step": 14775 + }, + { + "epoch": 30.0, + "learning_rate": 2.6687179487179488e-06, + "loss": 0.0008, + "step": 14800 + }, + { + "epoch": 30.0, + "learning_rate": 2.655897435897436e-06, + "loss": 0.0008, + "step": 14825 + }, + { + "epoch": 30.01, + "learning_rate": 2.6430769230769233e-06, + "loss": 0.0009, + "step": 14850 + }, + { + "epoch": 30.01, + "learning_rate": 2.6302564102564105e-06, + "loss": 0.0008, + "step": 14875 + }, + { + "epoch": 30.01, + "learning_rate": 2.6174358974358978e-06, + "loss": 0.0015, + "step": 14900 + }, + { + "epoch": 30.01, + "learning_rate": 2.604615384615385e-06, + "loss": 0.0012, + "step": 14925 + }, + { + "epoch": 30.01, + "learning_rate": 2.5917948717948723e-06, + "loss": 0.0008, + "step": 14950 + }, + { + "epoch": 30.01, + "learning_rate": 2.5789743589743595e-06, + "loss": 0.001, + "step": 14975 + }, + { + "epoch": 30.01, + "learning_rate": 2.5661538461538463e-06, + "loss": 0.001, + "step": 15000 + }, + { + "epoch": 30.01, + "eval_loss": 0.24516406655311584, + "eval_runtime": 5555.3108, + "eval_samples_per_second": 2.828, + "eval_steps_per_second": 0.177, + "eval_wer": 5.3086728638606075, + "step": 15000 + }, + { + "epoch": 30.01, + "learning_rate": 2.5533333333333336e-06, + "loss": 0.001, + "step": 15025 + }, + { + "epoch": 30.02, + "learning_rate": 2.5405128205128204e-06, + "loss": 0.001, + "step": 15050 + }, + { + "epoch": 30.02, + "learning_rate": 2.5276923076923077e-06, + "loss": 0.0006, + "step": 15075 + }, + { + "epoch": 30.02, + "learning_rate": 2.514871794871795e-06, + "loss": 0.0007, + "step": 15100 + }, + { + "epoch": 30.02, + "learning_rate": 2.502051282051282e-06, + "loss": 0.0007, + "step": 15125 + }, + { + "epoch": 30.02, + "learning_rate": 2.4892307692307694e-06, + "loss": 0.0011, + "step": 15150 + }, + { + "epoch": 30.02, + "learning_rate": 2.4764102564102567e-06, + "loss": 0.0012, + "step": 15175 + }, + { + "epoch": 30.02, + "learning_rate": 2.463589743589744e-06, + "loss": 0.001, + "step": 15200 + }, + { + "epoch": 31.0, + "learning_rate": 2.450769230769231e-06, + "loss": 0.0009, + "step": 15225 + }, + { + "epoch": 31.0, + "learning_rate": 2.4379487179487184e-06, + "loss": 0.0007, + "step": 15250 + }, + { + "epoch": 31.0, + "learning_rate": 2.4251282051282056e-06, + "loss": 0.0006, + "step": 15275 + }, + { + "epoch": 31.0, + "learning_rate": 2.4123076923076925e-06, + "loss": 0.0004, + "step": 15300 + }, + { + "epoch": 31.01, + "learning_rate": 2.3994871794871797e-06, + "loss": 0.0007, + "step": 15325 + }, + { + "epoch": 31.01, + "learning_rate": 2.386666666666667e-06, + "loss": 0.0004, + "step": 15350 + }, + { + "epoch": 31.01, + "learning_rate": 2.3738461538461542e-06, + "loss": 0.0004, + "step": 15375 + }, + { + "epoch": 31.01, + "learning_rate": 2.3610256410256415e-06, + "loss": 0.0006, + "step": 15400 + }, + { + "epoch": 31.01, + "learning_rate": 2.3482051282051287e-06, + "loss": 0.0005, + "step": 15425 + }, + { + "epoch": 31.01, + "learning_rate": 2.3353846153846155e-06, + "loss": 0.0005, + "step": 15450 + }, + { + "epoch": 31.01, + "learning_rate": 2.3225641025641028e-06, + "loss": 0.0004, + "step": 15475 + }, + { + "epoch": 31.01, + "learning_rate": 2.30974358974359e-06, + "loss": 0.0005, + "step": 15500 + }, + { + "epoch": 31.02, + "learning_rate": 2.2969230769230773e-06, + "loss": 0.0007, + "step": 15525 + }, + { + "epoch": 31.02, + "learning_rate": 2.2841025641025645e-06, + "loss": 0.0006, + "step": 15550 + }, + { + "epoch": 31.02, + "learning_rate": 2.2712820512820518e-06, + "loss": 0.0004, + "step": 15575 + }, + { + "epoch": 31.02, + "learning_rate": 2.2584615384615386e-06, + "loss": 0.0007, + "step": 15600 + }, + { + "epoch": 31.02, + "learning_rate": 2.245641025641026e-06, + "loss": 0.0004, + "step": 15625 + }, + { + "epoch": 31.02, + "learning_rate": 2.232820512820513e-06, + "loss": 0.0004, + "step": 15650 + }, + { + "epoch": 31.02, + "learning_rate": 2.2200000000000003e-06, + "loss": 0.0005, + "step": 15675 + }, + { + "epoch": 31.02, + "learning_rate": 2.2071794871794876e-06, + "loss": 0.0005, + "step": 15700 + }, + { + "epoch": 32.0, + "learning_rate": 2.194358974358975e-06, + "loss": 0.0003, + "step": 15725 + }, + { + "epoch": 32.0, + "learning_rate": 2.1815384615384617e-06, + "loss": 0.0004, + "step": 15750 + }, + { + "epoch": 32.0, + "learning_rate": 2.168717948717949e-06, + "loss": 0.0003, + "step": 15775 + }, + { + "epoch": 32.0, + "learning_rate": 2.155897435897436e-06, + "loss": 0.0003, + "step": 15800 + }, + { + "epoch": 32.01, + "learning_rate": 2.1430769230769234e-06, + "loss": 0.0003, + "step": 15825 + }, + { + "epoch": 32.01, + "learning_rate": 2.1302564102564107e-06, + "loss": 0.0003, + "step": 15850 + }, + { + "epoch": 32.01, + "learning_rate": 2.117435897435898e-06, + "loss": 0.0003, + "step": 15875 + }, + { + "epoch": 32.01, + "learning_rate": 2.1046153846153847e-06, + "loss": 0.0003, + "step": 15900 + }, + { + "epoch": 32.01, + "learning_rate": 2.091794871794872e-06, + "loss": 0.0003, + "step": 15925 + }, + { + "epoch": 32.01, + "learning_rate": 2.0789743589743592e-06, + "loss": 0.0003, + "step": 15950 + }, + { + "epoch": 32.01, + "learning_rate": 2.0661538461538465e-06, + "loss": 0.0003, + "step": 15975 + }, + { + "epoch": 32.01, + "learning_rate": 2.0533333333333337e-06, + "loss": 0.0003, + "step": 16000 + }, + { + "epoch": 32.01, + "eval_loss": 0.2572970688343048, + "eval_runtime": 5398.9628, + "eval_samples_per_second": 2.909, + "eval_steps_per_second": 0.182, + "eval_wer": 5.352297003406532, + "step": 16000 + }, + { + "epoch": 32.02, + "learning_rate": 2.040512820512821e-06, + "loss": 0.0003, + "step": 16025 + }, + { + "epoch": 32.02, + "learning_rate": 2.027692307692308e-06, + "loss": 0.0003, + "step": 16050 + }, + { + "epoch": 32.02, + "learning_rate": 2.014871794871795e-06, + "loss": 0.0003, + "step": 16075 + }, + { + "epoch": 32.02, + "learning_rate": 2.0020512820512823e-06, + "loss": 0.0003, + "step": 16100 + }, + { + "epoch": 32.02, + "learning_rate": 1.9892307692307695e-06, + "loss": 0.0003, + "step": 16125 + }, + { + "epoch": 32.02, + "learning_rate": 1.9764102564102568e-06, + "loss": 0.0003, + "step": 16150 + }, + { + "epoch": 32.02, + "learning_rate": 1.963589743589744e-06, + "loss": 0.0003, + "step": 16175 + }, + { + "epoch": 32.02, + "learning_rate": 1.950769230769231e-06, + "loss": 0.0004, + "step": 16200 + }, + { + "epoch": 33.0, + "learning_rate": 1.937948717948718e-06, + "loss": 0.0009, + "step": 16225 + }, + { + "epoch": 33.0, + "learning_rate": 1.9251282051282054e-06, + "loss": 0.0018, + "step": 16250 + }, + { + "epoch": 33.0, + "learning_rate": 1.9123076923076926e-06, + "loss": 0.002, + "step": 16275 + }, + { + "epoch": 33.0, + "learning_rate": 1.8994871794871796e-06, + "loss": 0.0016, + "step": 16300 + }, + { + "epoch": 33.01, + "learning_rate": 1.8866666666666669e-06, + "loss": 0.0017, + "step": 16325 + }, + { + "epoch": 33.01, + "learning_rate": 1.873846153846154e-06, + "loss": 0.0015, + "step": 16350 + }, + { + "epoch": 33.01, + "learning_rate": 1.8610256410256412e-06, + "loss": 0.0017, + "step": 16375 + }, + { + "epoch": 33.01, + "learning_rate": 1.8482051282051284e-06, + "loss": 0.0015, + "step": 16400 + }, + { + "epoch": 33.01, + "learning_rate": 1.8353846153846155e-06, + "loss": 0.0015, + "step": 16425 + }, + { + "epoch": 33.01, + "learning_rate": 1.8225641025641027e-06, + "loss": 0.0012, + "step": 16450 + }, + { + "epoch": 33.01, + "learning_rate": 1.80974358974359e-06, + "loss": 0.0013, + "step": 16475 + }, + { + "epoch": 33.01, + "learning_rate": 1.796923076923077e-06, + "loss": 0.0015, + "step": 16500 + }, + { + "epoch": 33.02, + "learning_rate": 1.7841025641025642e-06, + "loss": 0.0015, + "step": 16525 + }, + { + "epoch": 33.02, + "learning_rate": 1.7712820512820515e-06, + "loss": 0.0022, + "step": 16550 + }, + { + "epoch": 33.02, + "learning_rate": 1.7584615384615385e-06, + "loss": 0.0017, + "step": 16575 + }, + { + "epoch": 33.02, + "learning_rate": 1.7456410256410258e-06, + "loss": 0.0016, + "step": 16600 + }, + { + "epoch": 33.02, + "learning_rate": 1.732820512820513e-06, + "loss": 0.0022, + "step": 16625 + }, + { + "epoch": 33.02, + "learning_rate": 1.72e-06, + "loss": 0.0023, + "step": 16650 + }, + { + "epoch": 33.02, + "learning_rate": 1.7076923076923077e-06, + "loss": 0.0038, + "step": 16675 + }, + { + "epoch": 34.0, + "learning_rate": 1.694871794871795e-06, + "loss": 0.0014, + "step": 16700 + }, + { + "epoch": 34.0, + "learning_rate": 1.6820512820512822e-06, + "loss": 0.0004, + "step": 16725 + }, + { + "epoch": 34.0, + "learning_rate": 1.6692307692307694e-06, + "loss": 0.0006, + "step": 16750 + }, + { + "epoch": 34.0, + "learning_rate": 1.6564102564102567e-06, + "loss": 0.0006, + "step": 16775 + }, + { + "epoch": 34.01, + "learning_rate": 1.643589743589744e-06, + "loss": 0.0006, + "step": 16800 + }, + { + "epoch": 34.01, + "learning_rate": 1.6307692307692307e-06, + "loss": 0.0005, + "step": 16825 + }, + { + "epoch": 34.01, + "learning_rate": 1.617948717948718e-06, + "loss": 0.0005, + "step": 16850 + }, + { + "epoch": 34.01, + "learning_rate": 1.6051282051282052e-06, + "loss": 0.0006, + "step": 16875 + }, + { + "epoch": 34.01, + "learning_rate": 1.5923076923076925e-06, + "loss": 0.0004, + "step": 16900 + }, + { + "epoch": 34.01, + "learning_rate": 1.5794871794871797e-06, + "loss": 0.0004, + "step": 16925 + }, + { + "epoch": 34.01, + "learning_rate": 1.566666666666667e-06, + "loss": 0.0004, + "step": 16950 + }, + { + "epoch": 34.01, + "learning_rate": 1.5538461538461538e-06, + "loss": 0.0003, + "step": 16975 + }, + { + "epoch": 34.02, + "learning_rate": 1.541025641025641e-06, + "loss": 0.0004, + "step": 17000 + }, + { + "epoch": 34.02, + "eval_loss": 0.2690046727657318, + "eval_runtime": 5420.7174, + "eval_samples_per_second": 2.898, + "eval_steps_per_second": 0.181, + "eval_wer": 5.295200703118484, + "step": 17000 + }, + { + "epoch": 34.02, + "learning_rate": 1.5282051282051283e-06, + "loss": 0.0005, + "step": 17025 + }, + { + "epoch": 34.02, + "learning_rate": 1.5153846153846156e-06, + "loss": 0.0007, + "step": 17050 + }, + { + "epoch": 34.02, + "learning_rate": 1.5025641025641028e-06, + "loss": 0.0004, + "step": 17075 + }, + { + "epoch": 34.02, + "learning_rate": 1.48974358974359e-06, + "loss": 0.0004, + "step": 17100 + }, + { + "epoch": 34.02, + "learning_rate": 1.476923076923077e-06, + "loss": 0.0004, + "step": 17125 + }, + { + "epoch": 34.02, + "learning_rate": 1.4641025641025641e-06, + "loss": 0.0004, + "step": 17150 + }, + { + "epoch": 34.02, + "learning_rate": 1.4512820512820514e-06, + "loss": 0.0003, + "step": 17175 + }, + { + "epoch": 35.0, + "learning_rate": 1.4384615384615386e-06, + "loss": 0.0067, + "step": 17200 + }, + { + "epoch": 35.0, + "learning_rate": 1.4256410256410259e-06, + "loss": 0.01, + "step": 17225 + }, + { + "epoch": 35.0, + "learning_rate": 1.4128205128205131e-06, + "loss": 0.01, + "step": 17250 + }, + { + "epoch": 35.0, + "learning_rate": 1.4000000000000001e-06, + "loss": 0.012, + "step": 17275 + }, + { + "epoch": 35.01, + "learning_rate": 1.3871794871794872e-06, + "loss": 0.0094, + "step": 17300 + }, + { + "epoch": 35.01, + "learning_rate": 1.3743589743589744e-06, + "loss": 0.0086, + "step": 17325 + }, + { + "epoch": 35.01, + "learning_rate": 1.3615384615384617e-06, + "loss": 0.009, + "step": 17350 + }, + { + "epoch": 35.01, + "learning_rate": 1.348717948717949e-06, + "loss": 0.0086, + "step": 17375 + }, + { + "epoch": 35.01, + "learning_rate": 1.3358974358974362e-06, + "loss": 0.0112, + "step": 17400 + }, + { + "epoch": 35.01, + "learning_rate": 1.3230769230769232e-06, + "loss": 0.009, + "step": 17425 + }, + { + "epoch": 35.01, + "learning_rate": 1.3102564102564102e-06, + "loss": 0.0077, + "step": 17450 + }, + { + "epoch": 35.01, + "learning_rate": 1.2974358974358975e-06, + "loss": 0.0095, + "step": 17475 + }, + { + "epoch": 35.02, + "learning_rate": 1.2846153846153847e-06, + "loss": 0.0079, + "step": 17500 + }, + { + "epoch": 35.02, + "learning_rate": 1.271794871794872e-06, + "loss": 0.0098, + "step": 17525 + }, + { + "epoch": 35.02, + "learning_rate": 1.2589743589743592e-06, + "loss": 0.0097, + "step": 17550 + }, + { + "epoch": 35.02, + "learning_rate": 1.2461538461538463e-06, + "loss": 0.0081, + "step": 17575 + }, + { + "epoch": 35.02, + "learning_rate": 1.2333333333333335e-06, + "loss": 0.0091, + "step": 17600 + }, + { + "epoch": 35.02, + "learning_rate": 1.2205128205128206e-06, + "loss": 0.0085, + "step": 17625 + }, + { + "epoch": 35.02, + "learning_rate": 1.2076923076923078e-06, + "loss": 0.0089, + "step": 17650 + }, + { + "epoch": 35.02, + "learning_rate": 1.194871794871795e-06, + "loss": 0.0107, + "step": 17675 + }, + { + "epoch": 36.0, + "learning_rate": 1.182051282051282e-06, + "loss": 0.0019, + "step": 17700 + }, + { + "epoch": 36.0, + "learning_rate": 1.1692307692307693e-06, + "loss": 0.0017, + "step": 17725 + }, + { + "epoch": 36.0, + "learning_rate": 1.1564102564102566e-06, + "loss": 0.0014, + "step": 17750 + }, + { + "epoch": 36.0, + "learning_rate": 1.1435897435897436e-06, + "loss": 0.0016, + "step": 17775 + }, + { + "epoch": 36.01, + "learning_rate": 1.1307692307692309e-06, + "loss": 0.001, + "step": 17800 + }, + { + "epoch": 36.01, + "learning_rate": 1.1179487179487181e-06, + "loss": 0.001, + "step": 17825 + }, + { + "epoch": 36.01, + "learning_rate": 1.1051282051282052e-06, + "loss": 0.0012, + "step": 17850 + }, + { + "epoch": 36.01, + "learning_rate": 1.0923076923076924e-06, + "loss": 0.0015, + "step": 17875 + }, + { + "epoch": 36.01, + "learning_rate": 1.0794871794871797e-06, + "loss": 0.0014, + "step": 17900 + }, + { + "epoch": 36.01, + "learning_rate": 1.066666666666667e-06, + "loss": 0.0012, + "step": 17925 + }, + { + "epoch": 36.01, + "learning_rate": 1.053846153846154e-06, + "loss": 0.0018, + "step": 17950 + }, + { + "epoch": 36.01, + "learning_rate": 1.0410256410256412e-06, + "loss": 0.0011, + "step": 17975 + }, + { + "epoch": 36.02, + "learning_rate": 1.0282051282051284e-06, + "loss": 0.0013, + "step": 18000 + }, + { + "epoch": 36.02, + "eval_loss": 0.23734796047210693, + "eval_runtime": 5420.3987, + "eval_samples_per_second": 2.898, + "eval_steps_per_second": 0.181, + "eval_wer": 5.143799277635572, + "step": 18000 + }, + { + "epoch": 36.02, + "learning_rate": 1.0153846153846155e-06, + "loss": 0.0019, + "step": 18025 + }, + { + "epoch": 36.02, + "learning_rate": 1.0025641025641027e-06, + "loss": 0.0033, + "step": 18050 + }, + { + "epoch": 36.02, + "learning_rate": 9.8974358974359e-07, + "loss": 0.0015, + "step": 18075 + }, + { + "epoch": 36.02, + "learning_rate": 9.76923076923077e-07, + "loss": 0.0026, + "step": 18100 + }, + { + "epoch": 36.02, + "learning_rate": 9.641025641025642e-07, + "loss": 0.0014, + "step": 18125 + }, + { + "epoch": 36.02, + "learning_rate": 9.512820512820514e-07, + "loss": 0.0039, + "step": 18150 + }, + { + "epoch": 37.0, + "learning_rate": 9.384615384615385e-07, + "loss": 0.0037, + "step": 18175 + }, + { + "epoch": 37.0, + "learning_rate": 9.256410256410258e-07, + "loss": 0.0004, + "step": 18200 + }, + { + "epoch": 37.0, + "learning_rate": 9.128205128205129e-07, + "loss": 0.0006, + "step": 18225 + }, + { + "epoch": 37.0, + "learning_rate": 9.000000000000001e-07, + "loss": 0.0006, + "step": 18250 + }, + { + "epoch": 37.01, + "learning_rate": 8.871794871794873e-07, + "loss": 0.0007, + "step": 18275 + }, + { + "epoch": 37.01, + "learning_rate": 8.743589743589745e-07, + "loss": 0.0004, + "step": 18300 + }, + { + "epoch": 37.01, + "learning_rate": 8.615384615384616e-07, + "loss": 0.0004, + "step": 18325 + }, + { + "epoch": 37.01, + "learning_rate": 8.487179487179488e-07, + "loss": 0.0006, + "step": 18350 + }, + { + "epoch": 37.01, + "learning_rate": 8.35897435897436e-07, + "loss": 0.0006, + "step": 18375 + }, + { + "epoch": 37.01, + "learning_rate": 8.230769230769231e-07, + "loss": 0.0005, + "step": 18400 + }, + { + "epoch": 37.01, + "learning_rate": 8.102564102564104e-07, + "loss": 0.0005, + "step": 18425 + }, + { + "epoch": 37.01, + "learning_rate": 7.974358974358975e-07, + "loss": 0.0005, + "step": 18450 + }, + { + "epoch": 37.02, + "learning_rate": 7.846153846153847e-07, + "loss": 0.0005, + "step": 18475 + }, + { + "epoch": 37.02, + "learning_rate": 7.717948717948719e-07, + "loss": 0.0005, + "step": 18500 + }, + { + "epoch": 37.02, + "learning_rate": 7.58974358974359e-07, + "loss": 0.001, + "step": 18525 + }, + { + "epoch": 37.02, + "learning_rate": 7.461538461538462e-07, + "loss": 0.0009, + "step": 18550 + }, + { + "epoch": 37.02, + "learning_rate": 7.333333333333334e-07, + "loss": 0.0007, + "step": 18575 + }, + { + "epoch": 37.02, + "learning_rate": 7.205128205128206e-07, + "loss": 0.001, + "step": 18600 + }, + { + "epoch": 37.02, + "learning_rate": 7.076923076923077e-07, + "loss": 0.0005, + "step": 18625 + }, + { + "epoch": 37.02, + "learning_rate": 6.94871794871795e-07, + "loss": 0.0018, + "step": 18650 + }, + { + "epoch": 38.0, + "learning_rate": 6.820512820512821e-07, + "loss": 0.0007, + "step": 18675 + }, + { + "epoch": 38.0, + "learning_rate": 6.692307692307693e-07, + "loss": 0.0003, + "step": 18700 + }, + { + "epoch": 38.0, + "learning_rate": 6.564102564102565e-07, + "loss": 0.0004, + "step": 18725 + }, + { + "epoch": 38.0, + "learning_rate": 6.435897435897436e-07, + "loss": 0.0005, + "step": 18750 + }, + { + "epoch": 38.01, + "learning_rate": 6.307692307692308e-07, + "loss": 0.0004, + "step": 18775 + }, + { + "epoch": 38.01, + "learning_rate": 6.17948717948718e-07, + "loss": 0.0004, + "step": 18800 + }, + { + "epoch": 38.01, + "learning_rate": 6.051282051282052e-07, + "loss": 0.0004, + "step": 18825 + }, + { + "epoch": 38.01, + "learning_rate": 5.923076923076924e-07, + "loss": 0.0004, + "step": 18850 + }, + { + "epoch": 38.01, + "learning_rate": 5.794871794871796e-07, + "loss": 0.0004, + "step": 18875 + }, + { + "epoch": 38.01, + "learning_rate": 5.666666666666667e-07, + "loss": 0.0006, + "step": 18900 + }, + { + "epoch": 38.01, + "learning_rate": 5.53846153846154e-07, + "loss": 0.0003, + "step": 18925 + }, + { + "epoch": 38.01, + "learning_rate": 5.410256410256411e-07, + "loss": 0.0003, + "step": 18950 + }, + { + "epoch": 38.02, + "learning_rate": 5.282051282051282e-07, + "loss": 0.0003, + "step": 18975 + }, + { + "epoch": 38.02, + "learning_rate": 5.153846153846155e-07, + "loss": 0.0004, + "step": 19000 + }, + { + "epoch": 38.02, + "eval_loss": 0.26179689168930054, + "eval_runtime": 5529.8358, + "eval_samples_per_second": 2.841, + "eval_steps_per_second": 0.178, + "eval_wer": 5.1361009000686435, + "step": 19000 + }, + { + "epoch": 38.02, + "learning_rate": 5.025641025641026e-07, + "loss": 0.0006, + "step": 19025 + }, + { + "epoch": 38.02, + "learning_rate": 4.897435897435898e-07, + "loss": 0.0004, + "step": 19050 + }, + { + "epoch": 38.02, + "learning_rate": 4.76923076923077e-07, + "loss": 0.0003, + "step": 19075 + }, + { + "epoch": 38.02, + "learning_rate": 4.641025641025641e-07, + "loss": 0.0003, + "step": 19100 + }, + { + "epoch": 38.02, + "learning_rate": 4.5128205128205136e-07, + "loss": 0.0003, + "step": 19125 + }, + { + "epoch": 39.0, + "learning_rate": 4.384615384615385e-07, + "loss": 0.0003, + "step": 19150 + }, + { + "epoch": 39.0, + "learning_rate": 4.2564102564102564e-07, + "loss": 0.0002, + "step": 19175 + }, + { + "epoch": 39.0, + "learning_rate": 4.128205128205129e-07, + "loss": 0.0003, + "step": 19200 + }, + { + "epoch": 39.0, + "learning_rate": 4.0000000000000003e-07, + "loss": 0.0004, + "step": 19225 + }, + { + "epoch": 39.01, + "learning_rate": 3.871794871794872e-07, + "loss": 0.0003, + "step": 19250 + }, + { + "epoch": 39.01, + "learning_rate": 3.743589743589744e-07, + "loss": 0.0003, + "step": 19275 + }, + { + "epoch": 39.01, + "learning_rate": 3.6153846153846157e-07, + "loss": 0.0003, + "step": 19300 + }, + { + "epoch": 39.01, + "learning_rate": 3.487179487179487e-07, + "loss": 0.0003, + "step": 19325 + }, + { + "epoch": 39.01, + "learning_rate": 3.3589743589743596e-07, + "loss": 0.0003, + "step": 19350 + }, + { + "epoch": 39.01, + "learning_rate": 3.230769230769231e-07, + "loss": 0.0003, + "step": 19375 + }, + { + "epoch": 39.01, + "learning_rate": 3.102564102564103e-07, + "loss": 0.0003, + "step": 19400 + }, + { + "epoch": 39.01, + "learning_rate": 2.9743589743589744e-07, + "loss": 0.0002, + "step": 19425 + }, + { + "epoch": 39.02, + "learning_rate": 2.8461538461538463e-07, + "loss": 0.0003, + "step": 19450 + }, + { + "epoch": 39.02, + "learning_rate": 2.717948717948718e-07, + "loss": 0.0003, + "step": 19475 + }, + { + "epoch": 39.02, + "learning_rate": 2.5897435897435897e-07, + "loss": 0.0003, + "step": 19500 + }, + { + "epoch": 39.02, + "learning_rate": 2.4615384615384616e-07, + "loss": 0.0004, + "step": 19525 + }, + { + "epoch": 39.02, + "learning_rate": 2.3333333333333336e-07, + "loss": 0.0003, + "step": 19550 + }, + { + "epoch": 39.02, + "learning_rate": 2.2051282051282053e-07, + "loss": 0.0003, + "step": 19575 + }, + { + "epoch": 39.02, + "learning_rate": 2.076923076923077e-07, + "loss": 0.0003, + "step": 19600 + }, + { + "epoch": 39.02, + "learning_rate": 1.948717948717949e-07, + "loss": 0.0003, + "step": 19625 + }, + { + "epoch": 40.0, + "learning_rate": 1.8205128205128206e-07, + "loss": 0.0002, + "step": 19650 + }, + { + "epoch": 40.0, + "learning_rate": 1.6923076923076923e-07, + "loss": 0.0002, + "step": 19675 + }, + { + "epoch": 40.0, + "learning_rate": 1.5641025641025642e-07, + "loss": 0.0003, + "step": 19700 + }, + { + "epoch": 40.0, + "learning_rate": 1.435897435897436e-07, + "loss": 0.0003, + "step": 19725 + }, + { + "epoch": 40.01, + "learning_rate": 1.3076923076923079e-07, + "loss": 0.0003, + "step": 19750 + }, + { + "epoch": 40.01, + "learning_rate": 1.1794871794871797e-07, + "loss": 0.0003, + "step": 19775 + }, + { + "epoch": 40.01, + "learning_rate": 1.0512820512820514e-07, + "loss": 0.0003, + "step": 19800 + }, + { + "epoch": 40.01, + "learning_rate": 9.230769230769232e-08, + "loss": 0.0003, + "step": 19825 + }, + { + "epoch": 40.01, + "learning_rate": 7.94871794871795e-08, + "loss": 0.0003, + "step": 19850 + }, + { + "epoch": 40.01, + "learning_rate": 6.666666666666668e-08, + "loss": 0.0003, + "step": 19875 + }, + { + "epoch": 40.01, + "learning_rate": 5.384615384615385e-08, + "loss": 0.0002, + "step": 19900 + }, + { + "epoch": 40.01, + "learning_rate": 4.102564102564103e-08, + "loss": 0.0002, + "step": 19925 + }, + { + "epoch": 40.02, + "learning_rate": 2.820512820512821e-08, + "loss": 0.0003, + "step": 19950 + }, + { + "epoch": 40.02, + "learning_rate": 1.5384615384615385e-08, + "loss": 0.0003, + "step": 19975 + }, + { + "epoch": 40.02, + "learning_rate": 2.5641025641025645e-09, + "loss": 0.0004, + "step": 20000 + }, + { + "epoch": 40.02, + "eval_loss": 0.26630881428718567, + "eval_runtime": 5538.9041, + "eval_samples_per_second": 2.836, + "eval_steps_per_second": 0.177, + "eval_wer": 5.126477928109984, + "step": 20000 + }, + { + "epoch": 40.02, + "step": 20000, + "total_flos": 2.716979491897344e+21, + "train_loss": 0.033374719058023763, + "train_runtime": 453987.9544, + "train_samples_per_second": 2.819, + "train_steps_per_second": 0.044 + } + ], + "logging_steps": 25, + "max_steps": 20000, + "num_train_epochs": 9223372036854775807, + "save_steps": 1000, + "total_flos": 2.716979491897344e+21, + "trial_name": null, + "trial_params": null +}