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": 5.070020005715919, + "best_model_checkpoint": "./checkpoint-19000", + "epoch": 39.0016, + "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": 0.9747, + "step": 25 + }, + { + "epoch": 0.0, + "learning_rate": 9.800000000000001e-07, + "loss": 0.5833, + "step": 50 + }, + { + "epoch": 0.0, + "learning_rate": 1.48e-06, + "loss": 0.2754, + "step": 75 + }, + { + "epoch": 0.01, + "learning_rate": 1.98e-06, + "loss": 0.2088, + "step": 100 + }, + { + "epoch": 0.01, + "learning_rate": 2.4800000000000004e-06, + "loss": 0.1862, + "step": 125 + }, + { + "epoch": 0.01, + "learning_rate": 2.9800000000000003e-06, + "loss": 0.1616, + "step": 150 + }, + { + "epoch": 0.01, + "learning_rate": 3.48e-06, + "loss": 0.1657, + "step": 175 + }, + { + "epoch": 0.01, + "learning_rate": 3.980000000000001e-06, + "loss": 0.1656, + "step": 200 + }, + { + "epoch": 0.01, + "learning_rate": 4.48e-06, + "loss": 0.1571, + "step": 225 + }, + { + "epoch": 0.01, + "learning_rate": 4.980000000000001e-06, + "loss": 0.1595, + "step": 250 + }, + { + "epoch": 0.01, + "learning_rate": 5.480000000000001e-06, + "loss": 0.1492, + "step": 275 + }, + { + "epoch": 0.01, + "learning_rate": 5.98e-06, + "loss": 0.1419, + "step": 300 + }, + { + "epoch": 0.02, + "learning_rate": 6.480000000000001e-06, + "loss": 0.1376, + "step": 325 + }, + { + "epoch": 0.02, + "learning_rate": 6.98e-06, + "loss": 0.1444, + "step": 350 + }, + { + "epoch": 0.02, + "learning_rate": 7.48e-06, + "loss": 0.1402, + "step": 375 + }, + { + "epoch": 0.02, + "learning_rate": 7.980000000000002e-06, + "loss": 0.137, + "step": 400 + }, + { + "epoch": 0.02, + "learning_rate": 8.48e-06, + "loss": 0.1416, + "step": 425 + }, + { + "epoch": 0.02, + "learning_rate": 8.98e-06, + "loss": 0.1373, + "step": 450 + }, + { + "epoch": 0.02, + "learning_rate": 9.48e-06, + "loss": 0.1285, + "step": 475 + }, + { + "epoch": 0.03, + "learning_rate": 9.980000000000001e-06, + "loss": 0.1325, + "step": 500 + }, + { + "epoch": 1.0, + "learning_rate": 9.987692307692308e-06, + "loss": 0.1419, + "step": 525 + }, + { + "epoch": 1.0, + "learning_rate": 9.974871794871795e-06, + "loss": 0.1575, + "step": 550 + }, + { + "epoch": 1.0, + "learning_rate": 9.962051282051282e-06, + "loss": 0.162, + "step": 575 + }, + { + "epoch": 1.0, + "learning_rate": 9.94923076923077e-06, + "loss": 0.1537, + "step": 600 + }, + { + "epoch": 1.01, + "learning_rate": 9.936410256410257e-06, + "loss": 0.1554, + "step": 625 + }, + { + "epoch": 1.01, + "learning_rate": 9.923589743589744e-06, + "loss": 0.1327, + "step": 650 + }, + { + "epoch": 1.01, + "learning_rate": 9.910769230769231e-06, + "loss": 0.1129, + "step": 675 + }, + { + "epoch": 1.01, + "learning_rate": 9.89794871794872e-06, + "loss": 0.1169, + "step": 700 + }, + { + "epoch": 1.01, + "learning_rate": 9.885128205128206e-06, + "loss": 0.1274, + "step": 725 + }, + { + "epoch": 1.01, + "learning_rate": 9.872307692307693e-06, + "loss": 0.1139, + "step": 750 + }, + { + "epoch": 1.01, + "learning_rate": 9.85948717948718e-06, + "loss": 0.1285, + "step": 775 + }, + { + "epoch": 1.01, + "learning_rate": 9.846666666666668e-06, + "loss": 0.1307, + "step": 800 + }, + { + "epoch": 1.02, + "learning_rate": 9.833846153846155e-06, + "loss": 0.1213, + "step": 825 + }, + { + "epoch": 1.02, + "learning_rate": 9.821025641025642e-06, + "loss": 0.1169, + "step": 850 + }, + { + "epoch": 1.02, + "learning_rate": 9.808205128205129e-06, + "loss": 0.1446, + "step": 875 + }, + { + "epoch": 1.02, + "learning_rate": 9.795384615384617e-06, + "loss": 0.109, + "step": 900 + }, + { + "epoch": 1.02, + "learning_rate": 9.782564102564104e-06, + "loss": 0.1017, + "step": 925 + }, + { + "epoch": 1.02, + "learning_rate": 9.769743589743591e-06, + "loss": 0.1015, + "step": 950 + }, + { + "epoch": 1.02, + "learning_rate": 9.756923076923078e-06, + "loss": 0.0992, + "step": 975 + }, + { + "epoch": 1.02, + "learning_rate": 9.744102564102565e-06, + "loss": 0.1059, + "step": 1000 + }, + { + "epoch": 1.02, + "eval_loss": 0.1744348555803299, + "eval_runtime": 5860.9163, + "eval_samples_per_second": 2.795, + "eval_steps_per_second": 0.175, + "eval_wer": 7.634181194627036, + "step": 1000 + }, + { + "epoch": 2.0, + "learning_rate": 9.731282051282052e-06, + "loss": 0.0978, + "step": 1025 + }, + { + "epoch": 2.0, + "learning_rate": 9.718461538461538e-06, + "loss": 0.0887, + "step": 1050 + }, + { + "epoch": 2.0, + "learning_rate": 9.705641025641027e-06, + "loss": 0.0845, + "step": 1075 + }, + { + "epoch": 2.0, + "learning_rate": 9.692820512820514e-06, + "loss": 0.096, + "step": 1100 + }, + { + "epoch": 2.01, + "learning_rate": 9.68e-06, + "loss": 0.1072, + "step": 1125 + }, + { + "epoch": 2.01, + "learning_rate": 9.667179487179487e-06, + "loss": 0.0879, + "step": 1150 + }, + { + "epoch": 2.01, + "learning_rate": 9.654358974358974e-06, + "loss": 0.0915, + "step": 1175 + }, + { + "epoch": 2.01, + "learning_rate": 9.641538461538463e-06, + "loss": 0.1042, + "step": 1200 + }, + { + "epoch": 2.01, + "learning_rate": 9.62871794871795e-06, + "loss": 0.1062, + "step": 1225 + }, + { + "epoch": 2.01, + "learning_rate": 9.615897435897436e-06, + "loss": 0.0983, + "step": 1250 + }, + { + "epoch": 2.01, + "learning_rate": 9.603076923076923e-06, + "loss": 0.0928, + "step": 1275 + }, + { + "epoch": 2.01, + "learning_rate": 9.590256410256412e-06, + "loss": 0.0998, + "step": 1300 + }, + { + "epoch": 2.02, + "learning_rate": 9.577435897435898e-06, + "loss": 0.0985, + "step": 1325 + }, + { + "epoch": 2.02, + "learning_rate": 9.564615384615385e-06, + "loss": 0.0921, + "step": 1350 + }, + { + "epoch": 2.02, + "learning_rate": 9.551794871794872e-06, + "loss": 0.0783, + "step": 1375 + }, + { + "epoch": 2.02, + "learning_rate": 9.53897435897436e-06, + "loss": 0.0593, + "step": 1400 + }, + { + "epoch": 2.02, + "learning_rate": 9.526153846153847e-06, + "loss": 0.055, + "step": 1425 + }, + { + "epoch": 2.02, + "learning_rate": 9.513333333333334e-06, + "loss": 0.0488, + "step": 1450 + }, + { + "epoch": 2.02, + "learning_rate": 9.500512820512821e-06, + "loss": 0.0533, + "step": 1475 + }, + { + "epoch": 2.02, + "learning_rate": 9.48769230769231e-06, + "loss": 0.0431, + "step": 1500 + }, + { + "epoch": 2.03, + "learning_rate": 9.474871794871796e-06, + "loss": 0.0502, + "step": 1525 + }, + { + "epoch": 3.0, + "learning_rate": 9.462051282051283e-06, + "loss": 0.044, + "step": 1550 + }, + { + "epoch": 3.0, + "learning_rate": 9.44923076923077e-06, + "loss": 0.0402, + "step": 1575 + }, + { + "epoch": 3.0, + "learning_rate": 9.436410256410257e-06, + "loss": 0.0359, + "step": 1600 + }, + { + "epoch": 3.0, + "learning_rate": 9.423589743589744e-06, + "loss": 0.0391, + "step": 1625 + }, + { + "epoch": 3.01, + "learning_rate": 9.41076923076923e-06, + "loss": 0.0376, + "step": 1650 + }, + { + "epoch": 3.01, + "learning_rate": 9.397948717948719e-06, + "loss": 0.0352, + "step": 1675 + }, + { + "epoch": 3.01, + "learning_rate": 9.385128205128206e-06, + "loss": 0.0414, + "step": 1700 + }, + { + "epoch": 3.01, + "learning_rate": 9.372307692307693e-06, + "loss": 0.0416, + "step": 1725 + }, + { + "epoch": 3.01, + "learning_rate": 9.35948717948718e-06, + "loss": 0.0395, + "step": 1750 + }, + { + "epoch": 3.01, + "learning_rate": 9.346666666666666e-06, + "loss": 0.0385, + "step": 1775 + }, + { + "epoch": 3.01, + "learning_rate": 9.333846153846155e-06, + "loss": 0.0371, + "step": 1800 + }, + { + "epoch": 3.01, + "learning_rate": 9.321025641025642e-06, + "loss": 0.0414, + "step": 1825 + }, + { + "epoch": 3.02, + "learning_rate": 9.308205128205129e-06, + "loss": 0.0343, + "step": 1850 + }, + { + "epoch": 3.02, + "learning_rate": 9.295384615384615e-06, + "loss": 0.0352, + "step": 1875 + }, + { + "epoch": 3.02, + "learning_rate": 9.282564102564104e-06, + "loss": 0.0241, + "step": 1900 + }, + { + "epoch": 3.02, + "learning_rate": 9.26974358974359e-06, + "loss": 0.017, + "step": 1925 + }, + { + "epoch": 3.02, + "learning_rate": 9.256923076923078e-06, + "loss": 0.0197, + "step": 1950 + }, + { + "epoch": 3.02, + "learning_rate": 9.244102564102564e-06, + "loss": 0.0199, + "step": 1975 + }, + { + "epoch": 3.02, + "learning_rate": 9.231282051282053e-06, + "loss": 0.0159, + "step": 2000 + }, + { + "epoch": 3.02, + "eval_loss": 0.19433525204658508, + "eval_runtime": 5915.6327, + "eval_samples_per_second": 2.769, + "eval_steps_per_second": 0.173, + "eval_wer": 7.384967133466705, + "step": 2000 + }, + { + "epoch": 3.02, + "learning_rate": 9.21846153846154e-06, + "loss": 0.0168, + "step": 2025 + }, + { + "epoch": 4.0, + "learning_rate": 9.205641025641027e-06, + "loss": 0.0241, + "step": 2050 + }, + { + "epoch": 4.0, + "learning_rate": 9.192820512820513e-06, + "loss": 0.0726, + "step": 2075 + }, + { + "epoch": 4.0, + "learning_rate": 9.180000000000002e-06, + "loss": 0.09, + "step": 2100 + }, + { + "epoch": 4.0, + "learning_rate": 9.167179487179489e-06, + "loss": 0.076, + "step": 2125 + }, + { + "epoch": 4.01, + "learning_rate": 9.154358974358976e-06, + "loss": 0.0641, + "step": 2150 + }, + { + "epoch": 4.01, + "learning_rate": 9.141538461538462e-06, + "loss": 0.0607, + "step": 2175 + }, + { + "epoch": 4.01, + "learning_rate": 9.12871794871795e-06, + "loss": 0.0562, + "step": 2200 + }, + { + "epoch": 4.01, + "learning_rate": 9.115897435897436e-06, + "loss": 0.0401, + "step": 2225 + }, + { + "epoch": 4.01, + "learning_rate": 9.103076923076923e-06, + "loss": 0.0663, + "step": 2250 + }, + { + "epoch": 4.01, + "learning_rate": 9.090256410256411e-06, + "loss": 0.0756, + "step": 2275 + }, + { + "epoch": 4.01, + "learning_rate": 9.077435897435898e-06, + "loss": 0.084, + "step": 2300 + }, + { + "epoch": 4.01, + "learning_rate": 9.064615384615385e-06, + "loss": 0.0594, + "step": 2325 + }, + { + "epoch": 4.02, + "learning_rate": 9.051794871794872e-06, + "loss": 0.0625, + "step": 2350 + }, + { + "epoch": 4.02, + "learning_rate": 9.038974358974359e-06, + "loss": 0.0659, + "step": 2375 + }, + { + "epoch": 4.02, + "learning_rate": 9.026153846153847e-06, + "loss": 0.0701, + "step": 2400 + }, + { + "epoch": 4.02, + "learning_rate": 9.013333333333334e-06, + "loss": 0.0712, + "step": 2425 + }, + { + "epoch": 4.02, + "learning_rate": 9.00051282051282e-06, + "loss": 0.0686, + "step": 2450 + }, + { + "epoch": 4.02, + "learning_rate": 8.987692307692308e-06, + "loss": 0.07, + "step": 2475 + }, + { + "epoch": 4.02, + "learning_rate": 8.974871794871796e-06, + "loss": 0.0618, + "step": 2500 + }, + { + "epoch": 4.02, + "learning_rate": 8.962051282051283e-06, + "loss": 0.1045, + "step": 2525 + }, + { + "epoch": 4.03, + "learning_rate": 8.94923076923077e-06, + "loss": 0.1008, + "step": 2550 + }, + { + "epoch": 5.0, + "learning_rate": 8.936410256410257e-06, + "loss": 0.0793, + "step": 2575 + }, + { + "epoch": 5.0, + "learning_rate": 8.923589743589745e-06, + "loss": 0.0912, + "step": 2600 + }, + { + "epoch": 5.0, + "learning_rate": 8.910769230769232e-06, + "loss": 0.0809, + "step": 2625 + }, + { + "epoch": 5.0, + "learning_rate": 8.897948717948719e-06, + "loss": 0.0804, + "step": 2650 + }, + { + "epoch": 5.01, + "learning_rate": 8.885128205128206e-06, + "loss": 0.0709, + "step": 2675 + }, + { + "epoch": 5.01, + "learning_rate": 8.872307692307694e-06, + "loss": 0.0711, + "step": 2700 + }, + { + "epoch": 5.01, + "learning_rate": 8.859487179487181e-06, + "loss": 0.0709, + "step": 2725 + }, + { + "epoch": 5.01, + "learning_rate": 8.846666666666668e-06, + "loss": 0.0736, + "step": 2750 + }, + { + "epoch": 5.01, + "learning_rate": 8.833846153846155e-06, + "loss": 0.0582, + "step": 2775 + }, + { + "epoch": 5.01, + "learning_rate": 8.821025641025643e-06, + "loss": 0.0327, + "step": 2800 + }, + { + "epoch": 5.01, + "learning_rate": 8.808205128205128e-06, + "loss": 0.0283, + "step": 2825 + }, + { + "epoch": 5.01, + "learning_rate": 8.795384615384615e-06, + "loss": 0.0312, + "step": 2850 + }, + { + "epoch": 5.02, + "learning_rate": 8.782564102564104e-06, + "loss": 0.0342, + "step": 2875 + }, + { + "epoch": 5.02, + "learning_rate": 8.76974358974359e-06, + "loss": 0.0336, + "step": 2900 + }, + { + "epoch": 5.02, + "learning_rate": 8.756923076923077e-06, + "loss": 0.0242, + "step": 2925 + }, + { + "epoch": 5.02, + "learning_rate": 8.744102564102564e-06, + "loss": 0.0315, + "step": 2950 + }, + { + "epoch": 5.02, + "learning_rate": 8.731282051282051e-06, + "loss": 0.0439, + "step": 2975 + }, + { + "epoch": 5.02, + "learning_rate": 8.71846153846154e-06, + "loss": 0.0526, + "step": 3000 + }, + { + "epoch": 5.02, + "eval_loss": 0.18990886211395264, + "eval_runtime": 5791.6285, + "eval_samples_per_second": 2.828, + "eval_steps_per_second": 0.177, + "eval_wer": 6.852243498142327, + "step": 3000 + }, + { + "epoch": 5.02, + "learning_rate": 8.705641025641026e-06, + "loss": 0.0626, + "step": 3025 + }, + { + "epoch": 5.02, + "learning_rate": 8.692820512820513e-06, + "loss": 0.0481, + "step": 3050 + }, + { + "epoch": 6.0, + "learning_rate": 8.68e-06, + "loss": 0.0551, + "step": 3075 + }, + { + "epoch": 6.0, + "learning_rate": 8.667179487179488e-06, + "loss": 0.0763, + "step": 3100 + }, + { + "epoch": 6.0, + "learning_rate": 8.654358974358975e-06, + "loss": 0.079, + "step": 3125 + }, + { + "epoch": 6.0, + "learning_rate": 8.641538461538462e-06, + "loss": 0.0752, + "step": 3150 + }, + { + "epoch": 6.01, + "learning_rate": 8.628717948717949e-06, + "loss": 0.0907, + "step": 3175 + }, + { + "epoch": 6.01, + "learning_rate": 8.615897435897437e-06, + "loss": 0.0805, + "step": 3200 + }, + { + "epoch": 6.01, + "learning_rate": 8.603076923076924e-06, + "loss": 0.0719, + "step": 3225 + }, + { + "epoch": 6.01, + "learning_rate": 8.590256410256411e-06, + "loss": 0.0836, + "step": 3250 + }, + { + "epoch": 6.01, + "learning_rate": 8.577435897435898e-06, + "loss": 0.0709, + "step": 3275 + }, + { + "epoch": 6.01, + "learning_rate": 8.564615384615386e-06, + "loss": 0.077, + "step": 3300 + }, + { + "epoch": 6.01, + "learning_rate": 8.551794871794873e-06, + "loss": 0.0647, + "step": 3325 + }, + { + "epoch": 6.01, + "learning_rate": 8.53897435897436e-06, + "loss": 0.0834, + "step": 3350 + }, + { + "epoch": 6.02, + "learning_rate": 8.526153846153847e-06, + "loss": 0.0721, + "step": 3375 + }, + { + "epoch": 6.02, + "learning_rate": 8.513333333333335e-06, + "loss": 0.0627, + "step": 3400 + }, + { + "epoch": 6.02, + "learning_rate": 8.50051282051282e-06, + "loss": 0.0784, + "step": 3425 + }, + { + "epoch": 6.02, + "learning_rate": 8.487692307692307e-06, + "loss": 0.063, + "step": 3450 + }, + { + "epoch": 6.02, + "learning_rate": 8.474871794871796e-06, + "loss": 0.0753, + "step": 3475 + }, + { + "epoch": 6.02, + "learning_rate": 8.462051282051283e-06, + "loss": 0.0792, + "step": 3500 + }, + { + "epoch": 6.02, + "learning_rate": 8.44923076923077e-06, + "loss": 0.077, + "step": 3525 + }, + { + "epoch": 6.02, + "learning_rate": 8.436410256410256e-06, + "loss": 0.0735, + "step": 3550 + }, + { + "epoch": 6.03, + "learning_rate": 8.423589743589745e-06, + "loss": 0.0829, + "step": 3575 + }, + { + "epoch": 7.0, + "learning_rate": 8.410769230769232e-06, + "loss": 0.0426, + "step": 3600 + }, + { + "epoch": 7.0, + "learning_rate": 8.397948717948719e-06, + "loss": 0.0321, + "step": 3625 + }, + { + "epoch": 7.0, + "learning_rate": 8.385128205128205e-06, + "loss": 0.0226, + "step": 3650 + }, + { + "epoch": 7.0, + "learning_rate": 8.372307692307692e-06, + "loss": 0.0386, + "step": 3675 + }, + { + "epoch": 7.01, + "learning_rate": 8.35948717948718e-06, + "loss": 0.0482, + "step": 3700 + }, + { + "epoch": 7.01, + "learning_rate": 8.346666666666668e-06, + "loss": 0.0471, + "step": 3725 + }, + { + "epoch": 7.01, + "learning_rate": 8.333846153846154e-06, + "loss": 0.0467, + "step": 3750 + }, + { + "epoch": 7.01, + "learning_rate": 8.321025641025641e-06, + "loss": 0.0468, + "step": 3775 + }, + { + "epoch": 7.01, + "learning_rate": 8.30820512820513e-06, + "loss": 0.0454, + "step": 3800 + }, + { + "epoch": 7.01, + "learning_rate": 8.295384615384616e-06, + "loss": 0.0421, + "step": 3825 + }, + { + "epoch": 7.01, + "learning_rate": 8.282564102564103e-06, + "loss": 0.0437, + "step": 3850 + }, + { + "epoch": 7.01, + "learning_rate": 8.26974358974359e-06, + "loss": 0.053, + "step": 3875 + }, + { + "epoch": 7.02, + "learning_rate": 8.256923076923079e-06, + "loss": 0.0469, + "step": 3900 + }, + { + "epoch": 7.02, + "learning_rate": 8.244102564102565e-06, + "loss": 0.0514, + "step": 3925 + }, + { + "epoch": 7.02, + "learning_rate": 8.231282051282052e-06, + "loss": 0.0613, + "step": 3950 + }, + { + "epoch": 7.02, + "learning_rate": 8.218461538461539e-06, + "loss": 0.0525, + "step": 3975 + }, + { + "epoch": 7.02, + "learning_rate": 8.205641025641028e-06, + "loss": 0.058, + "step": 4000 + }, + { + "epoch": 7.02, + "eval_loss": 0.17820489406585693, + "eval_runtime": 5806.831, + "eval_samples_per_second": 2.821, + "eval_steps_per_second": 0.176, + "eval_wer": 6.7802229208345235, + "step": 4000 + }, + { + "epoch": 7.02, + "learning_rate": 8.192820512820513e-06, + "loss": 0.0684, + "step": 4025 + }, + { + "epoch": 7.02, + "learning_rate": 8.18e-06, + "loss": 0.0853, + "step": 4050 + }, + { + "epoch": 7.02, + "learning_rate": 8.167179487179488e-06, + "loss": 0.0801, + "step": 4075 + }, + { + "epoch": 8.0, + "learning_rate": 8.154358974358975e-06, + "loss": 0.0636, + "step": 4100 + }, + { + "epoch": 8.0, + "learning_rate": 8.141538461538462e-06, + "loss": 0.0519, + "step": 4125 + }, + { + "epoch": 8.0, + "learning_rate": 8.128717948717949e-06, + "loss": 0.0503, + "step": 4150 + }, + { + "epoch": 8.0, + "learning_rate": 8.115897435897437e-06, + "loss": 0.0508, + "step": 4175 + }, + { + "epoch": 8.01, + "learning_rate": 8.103076923076924e-06, + "loss": 0.0465, + "step": 4200 + }, + { + "epoch": 8.01, + "learning_rate": 8.09025641025641e-06, + "loss": 0.0418, + "step": 4225 + }, + { + "epoch": 8.01, + "learning_rate": 8.077435897435898e-06, + "loss": 0.044, + "step": 4250 + }, + { + "epoch": 8.01, + "learning_rate": 8.064615384615384e-06, + "loss": 0.0408, + "step": 4275 + }, + { + "epoch": 8.01, + "learning_rate": 8.051794871794873e-06, + "loss": 0.0422, + "step": 4300 + }, + { + "epoch": 8.01, + "learning_rate": 8.03897435897436e-06, + "loss": 0.0399, + "step": 4325 + }, + { + "epoch": 8.01, + "learning_rate": 8.026153846153847e-06, + "loss": 0.0416, + "step": 4350 + }, + { + "epoch": 8.01, + "learning_rate": 8.013333333333333e-06, + "loss": 0.038, + "step": 4375 + }, + { + "epoch": 8.02, + "learning_rate": 8.000512820512822e-06, + "loss": 0.0421, + "step": 4400 + }, + { + "epoch": 8.02, + "learning_rate": 7.987692307692309e-06, + "loss": 0.0409, + "step": 4425 + }, + { + "epoch": 8.02, + "learning_rate": 7.974871794871796e-06, + "loss": 0.0396, + "step": 4450 + }, + { + "epoch": 8.02, + "learning_rate": 7.962051282051282e-06, + "loss": 0.0403, + "step": 4475 + }, + { + "epoch": 8.02, + "learning_rate": 7.949230769230771e-06, + "loss": 0.0386, + "step": 4500 + }, + { + "epoch": 8.02, + "learning_rate": 7.936410256410258e-06, + "loss": 0.0386, + "step": 4525 + }, + { + "epoch": 8.02, + "learning_rate": 7.923589743589745e-06, + "loss": 0.0468, + "step": 4550 + }, + { + "epoch": 8.02, + "learning_rate": 7.910769230769231e-06, + "loss": 0.0768, + "step": 4575 + }, + { + "epoch": 8.03, + "learning_rate": 7.89794871794872e-06, + "loss": 0.0666, + "step": 4600 + }, + { + "epoch": 9.0, + "learning_rate": 7.885128205128207e-06, + "loss": 0.0203, + "step": 4625 + }, + { + "epoch": 9.0, + "learning_rate": 7.872307692307692e-06, + "loss": 0.0088, + "step": 4650 + }, + { + "epoch": 9.0, + "learning_rate": 7.85948717948718e-06, + "loss": 0.0089, + "step": 4675 + }, + { + "epoch": 9.0, + "learning_rate": 7.846666666666667e-06, + "loss": 0.0146, + "step": 4700 + }, + { + "epoch": 9.01, + "learning_rate": 7.833846153846154e-06, + "loss": 0.0159, + "step": 4725 + }, + { + "epoch": 9.01, + "learning_rate": 7.821025641025641e-06, + "loss": 0.0172, + "step": 4750 + }, + { + "epoch": 9.01, + "learning_rate": 7.80820512820513e-06, + "loss": 0.017, + "step": 4775 + }, + { + "epoch": 9.01, + "learning_rate": 7.795384615384616e-06, + "loss": 0.0157, + "step": 4800 + }, + { + "epoch": 9.01, + "learning_rate": 7.782564102564103e-06, + "loss": 0.0165, + "step": 4825 + }, + { + "epoch": 9.01, + "learning_rate": 7.76974358974359e-06, + "loss": 0.0149, + "step": 4850 + }, + { + "epoch": 9.01, + "learning_rate": 7.756923076923077e-06, + "loss": 0.0153, + "step": 4875 + }, + { + "epoch": 9.01, + "learning_rate": 7.744102564102565e-06, + "loss": 0.0174, + "step": 4900 + }, + { + "epoch": 9.02, + "learning_rate": 7.731282051282052e-06, + "loss": 0.0162, + "step": 4925 + }, + { + "epoch": 9.02, + "learning_rate": 7.718461538461539e-06, + "loss": 0.0149, + "step": 4950 + }, + { + "epoch": 9.02, + "learning_rate": 7.705641025641026e-06, + "loss": 0.0186, + "step": 4975 + }, + { + "epoch": 9.02, + "learning_rate": 7.692820512820514e-06, + "loss": 0.0161, + "step": 5000 + }, + { + "epoch": 9.02, + "eval_loss": 0.19952422380447388, + "eval_runtime": 5797.4599, + "eval_samples_per_second": 2.825, + "eval_steps_per_second": 0.177, + "eval_wer": 6.633895398685338, + "step": 5000 + }, + { + "epoch": 9.02, + "learning_rate": 7.680000000000001e-06, + "loss": 0.0219, + "step": 5025 + }, + { + "epoch": 9.02, + "learning_rate": 7.667179487179488e-06, + "loss": 0.0224, + "step": 5050 + }, + { + "epoch": 9.02, + "learning_rate": 7.654358974358975e-06, + "loss": 0.0326, + "step": 5075 + }, + { + "epoch": 9.02, + "learning_rate": 7.641538461538463e-06, + "loss": 0.0261, + "step": 5100 + }, + { + "epoch": 10.0, + "learning_rate": 7.62871794871795e-06, + "loss": 0.0249, + "step": 5125 + }, + { + "epoch": 10.0, + "learning_rate": 7.615897435897437e-06, + "loss": 0.046, + "step": 5150 + }, + { + "epoch": 10.0, + "learning_rate": 7.6030769230769245e-06, + "loss": 0.043, + "step": 5175 + }, + { + "epoch": 10.0, + "learning_rate": 7.590256410256411e-06, + "loss": 0.0444, + "step": 5200 + }, + { + "epoch": 10.01, + "learning_rate": 7.577435897435899e-06, + "loss": 0.0423, + "step": 5225 + }, + { + "epoch": 10.01, + "learning_rate": 7.564615384615385e-06, + "loss": 0.0413, + "step": 5250 + }, + { + "epoch": 10.01, + "learning_rate": 7.551794871794872e-06, + "loss": 0.0451, + "step": 5275 + }, + { + "epoch": 10.01, + "learning_rate": 7.5389743589743595e-06, + "loss": 0.0409, + "step": 5300 + }, + { + "epoch": 10.01, + "learning_rate": 7.526153846153846e-06, + "loss": 0.0378, + "step": 5325 + }, + { + "epoch": 10.01, + "learning_rate": 7.513333333333334e-06, + "loss": 0.0411, + "step": 5350 + }, + { + "epoch": 10.01, + "learning_rate": 7.500512820512821e-06, + "loss": 0.0389, + "step": 5375 + }, + { + "epoch": 10.01, + "learning_rate": 7.4876923076923085e-06, + "loss": 0.0439, + "step": 5400 + }, + { + "epoch": 10.02, + "learning_rate": 7.474871794871795e-06, + "loss": 0.0419, + "step": 5425 + }, + { + "epoch": 10.02, + "learning_rate": 7.462051282051282e-06, + "loss": 0.0382, + "step": 5450 + }, + { + "epoch": 10.02, + "learning_rate": 7.44923076923077e-06, + "loss": 0.0479, + "step": 5475 + }, + { + "epoch": 10.02, + "learning_rate": 7.436410256410257e-06, + "loss": 0.0509, + "step": 5500 + }, + { + "epoch": 10.02, + "learning_rate": 7.423589743589744e-06, + "loss": 0.0437, + "step": 5525 + }, + { + "epoch": 10.02, + "learning_rate": 7.410769230769231e-06, + "loss": 0.0438, + "step": 5550 + }, + { + "epoch": 10.02, + "learning_rate": 7.397948717948719e-06, + "loss": 0.0451, + "step": 5575 + }, + { + "epoch": 10.02, + "learning_rate": 7.385128205128206e-06, + "loss": 0.0571, + "step": 5600 + }, + { + "epoch": 10.03, + "learning_rate": 7.372307692307693e-06, + "loss": 0.0748, + "step": 5625 + }, + { + "epoch": 11.0, + "learning_rate": 7.35948717948718e-06, + "loss": 0.064, + "step": 5650 + }, + { + "epoch": 11.0, + "learning_rate": 7.346666666666668e-06, + "loss": 0.0744, + "step": 5675 + }, + { + "epoch": 11.0, + "learning_rate": 7.333846153846155e-06, + "loss": 0.0616, + "step": 5700 + }, + { + "epoch": 11.0, + "learning_rate": 7.321025641025642e-06, + "loss": 0.038, + "step": 5725 + }, + { + "epoch": 11.01, + "learning_rate": 7.308205128205129e-06, + "loss": 0.0689, + "step": 5750 + }, + { + "epoch": 11.01, + "learning_rate": 7.295384615384617e-06, + "loss": 0.0645, + "step": 5775 + }, + { + "epoch": 11.01, + "learning_rate": 7.2825641025641036e-06, + "loss": 0.0511, + "step": 5800 + }, + { + "epoch": 11.01, + "learning_rate": 7.269743589743591e-06, + "loss": 0.0511, + "step": 5825 + }, + { + "epoch": 11.01, + "learning_rate": 7.256923076923078e-06, + "loss": 0.066, + "step": 5850 + }, + { + "epoch": 11.01, + "learning_rate": 7.244102564102564e-06, + "loss": 0.0557, + "step": 5875 + }, + { + "epoch": 11.01, + "learning_rate": 7.231282051282052e-06, + "loss": 0.0566, + "step": 5900 + }, + { + "epoch": 11.01, + "learning_rate": 7.2184615384615385e-06, + "loss": 0.0822, + "step": 5925 + }, + { + "epoch": 11.02, + "learning_rate": 7.205641025641026e-06, + "loss": 0.0594, + "step": 5950 + }, + { + "epoch": 11.02, + "learning_rate": 7.192820512820513e-06, + "loss": 0.0646, + "step": 5975 + }, + { + "epoch": 11.02, + "learning_rate": 7.180000000000001e-06, + "loss": 0.065, + "step": 6000 + }, + { + "epoch": 11.02, + "eval_loss": 0.15629138052463531, + "eval_runtime": 5911.7805, + "eval_samples_per_second": 2.771, + "eval_steps_per_second": 0.173, + "eval_wer": 6.454415547299229, + "step": 6000 + }, + { + "epoch": 11.02, + "learning_rate": 7.1671794871794875e-06, + "loss": 0.0649, + "step": 6025 + }, + { + "epoch": 11.02, + "learning_rate": 7.154358974358974e-06, + "loss": 0.0574, + "step": 6050 + }, + { + "epoch": 11.02, + "learning_rate": 7.141538461538462e-06, + "loss": 0.0494, + "step": 6075 + }, + { + "epoch": 11.02, + "learning_rate": 7.128717948717949e-06, + "loss": 0.0574, + "step": 6100 + }, + { + "epoch": 11.02, + "learning_rate": 7.1158974358974365e-06, + "loss": 0.0651, + "step": 6125 + }, + { + "epoch": 12.0, + "learning_rate": 7.103076923076923e-06, + "loss": 0.0752, + "step": 6150 + }, + { + "epoch": 12.0, + "learning_rate": 7.090256410256411e-06, + "loss": 0.0835, + "step": 6175 + }, + { + "epoch": 12.0, + "learning_rate": 7.077435897435898e-06, + "loss": 0.0831, + "step": 6200 + }, + { + "epoch": 12.0, + "learning_rate": 7.0646153846153855e-06, + "loss": 0.0496, + "step": 6225 + }, + { + "epoch": 12.01, + "learning_rate": 7.052307692307693e-06, + "loss": 0.0406, + "step": 6250 + }, + { + "epoch": 12.01, + "learning_rate": 7.03948717948718e-06, + "loss": 0.0393, + "step": 6275 + }, + { + "epoch": 12.01, + "learning_rate": 7.0266666666666674e-06, + "loss": 0.0502, + "step": 6300 + }, + { + "epoch": 12.01, + "learning_rate": 7.013846153846154e-06, + "loss": 0.042, + "step": 6325 + }, + { + "epoch": 12.01, + "learning_rate": 7.001025641025642e-06, + "loss": 0.0329, + "step": 6350 + }, + { + "epoch": 12.01, + "learning_rate": 6.988205128205129e-06, + "loss": 0.0376, + "step": 6375 + }, + { + "epoch": 12.01, + "learning_rate": 6.9753846153846164e-06, + "loss": 0.0368, + "step": 6400 + }, + { + "epoch": 12.01, + "learning_rate": 6.962564102564103e-06, + "loss": 0.0505, + "step": 6425 + }, + { + "epoch": 12.02, + "learning_rate": 6.949743589743591e-06, + "loss": 0.0497, + "step": 6450 + }, + { + "epoch": 12.02, + "learning_rate": 6.936923076923078e-06, + "loss": 0.0666, + "step": 6475 + }, + { + "epoch": 12.02, + "learning_rate": 6.9241025641025646e-06, + "loss": 0.0344, + "step": 6500 + }, + { + "epoch": 12.02, + "learning_rate": 6.911282051282052e-06, + "loss": 0.0438, + "step": 6525 + }, + { + "epoch": 12.02, + "learning_rate": 6.898461538461539e-06, + "loss": 0.0679, + "step": 6550 + }, + { + "epoch": 12.02, + "learning_rate": 6.885641025641027e-06, + "loss": 0.1006, + "step": 6575 + }, + { + "epoch": 12.02, + "learning_rate": 6.8728205128205136e-06, + "loss": 0.1022, + "step": 6600 + }, + { + "epoch": 12.02, + "learning_rate": 6.860000000000001e-06, + "loss": 0.1058, + "step": 6625 + }, + { + "epoch": 12.03, + "learning_rate": 6.847179487179487e-06, + "loss": 0.1164, + "step": 6650 + }, + { + "epoch": 13.0, + "learning_rate": 6.834358974358974e-06, + "loss": 0.0446, + "step": 6675 + }, + { + "epoch": 13.0, + "learning_rate": 6.821538461538462e-06, + "loss": 0.0379, + "step": 6700 + }, + { + "epoch": 13.0, + "learning_rate": 6.8087179487179485e-06, + "loss": 0.0396, + "step": 6725 + }, + { + "epoch": 13.0, + "learning_rate": 6.795897435897436e-06, + "loss": 0.0355, + "step": 6750 + }, + { + "epoch": 13.01, + "learning_rate": 6.783076923076923e-06, + "loss": 0.0402, + "step": 6775 + }, + { + "epoch": 13.01, + "learning_rate": 6.770256410256411e-06, + "loss": 0.0331, + "step": 6800 + }, + { + "epoch": 13.01, + "learning_rate": 6.7574358974358975e-06, + "loss": 0.0361, + "step": 6825 + }, + { + "epoch": 13.01, + "learning_rate": 6.744615384615385e-06, + "loss": 0.0414, + "step": 6850 + }, + { + "epoch": 13.01, + "learning_rate": 6.731794871794872e-06, + "loss": 0.0387, + "step": 6875 + }, + { + "epoch": 13.01, + "learning_rate": 6.71897435897436e-06, + "loss": 0.035, + "step": 6900 + }, + { + "epoch": 13.01, + "learning_rate": 6.7061538461538465e-06, + "loss": 0.0359, + "step": 6925 + }, + { + "epoch": 13.01, + "learning_rate": 6.693333333333334e-06, + "loss": 0.037, + "step": 6950 + }, + { + "epoch": 13.02, + "learning_rate": 6.680512820512821e-06, + "loss": 0.0568, + "step": 6975 + }, + { + "epoch": 13.02, + "learning_rate": 6.667692307692309e-06, + "loss": 0.082, + "step": 7000 + }, + { + "epoch": 13.02, + "eval_loss": 0.17887301743030548, + "eval_runtime": 5912.8552, + "eval_samples_per_second": 2.77, + "eval_steps_per_second": 0.173, + "eval_wer": 6.030865961703344, + "step": 7000 + }, + { + "epoch": 13.02, + "learning_rate": 6.6548717948717955e-06, + "loss": 0.0698, + "step": 7025 + }, + { + "epoch": 13.02, + "learning_rate": 6.642051282051283e-06, + "loss": 0.0687, + "step": 7050 + }, + { + "epoch": 13.02, + "learning_rate": 6.62923076923077e-06, + "loss": 0.0751, + "step": 7075 + }, + { + "epoch": 13.02, + "learning_rate": 6.616410256410257e-06, + "loss": 0.0739, + "step": 7100 + }, + { + "epoch": 13.02, + "learning_rate": 6.6035897435897445e-06, + "loss": 0.0677, + "step": 7125 + }, + { + "epoch": 13.02, + "learning_rate": 6.590769230769231e-06, + "loss": 0.0714, + "step": 7150 + }, + { + "epoch": 14.0, + "learning_rate": 6.577948717948719e-06, + "loss": 0.0712, + "step": 7175 + }, + { + "epoch": 14.0, + "learning_rate": 6.565128205128206e-06, + "loss": 0.0518, + "step": 7200 + }, + { + "epoch": 14.0, + "learning_rate": 6.5523076923076935e-06, + "loss": 0.0686, + "step": 7225 + }, + { + "epoch": 14.0, + "learning_rate": 6.5394871794871795e-06, + "loss": 0.0468, + "step": 7250 + }, + { + "epoch": 14.01, + "learning_rate": 6.526666666666666e-06, + "loss": 0.0662, + "step": 7275 + }, + { + "epoch": 14.01, + "learning_rate": 6.513846153846154e-06, + "loss": 0.0655, + "step": 7300 + }, + { + "epoch": 14.01, + "learning_rate": 6.501025641025641e-06, + "loss": 0.0611, + "step": 7325 + }, + { + "epoch": 14.01, + "learning_rate": 6.4882051282051285e-06, + "loss": 0.0595, + "step": 7350 + }, + { + "epoch": 14.01, + "learning_rate": 6.475384615384615e-06, + "loss": 0.0626, + "step": 7375 + }, + { + "epoch": 14.01, + "learning_rate": 6.462564102564103e-06, + "loss": 0.0536, + "step": 7400 + }, + { + "epoch": 14.01, + "learning_rate": 6.44974358974359e-06, + "loss": 0.049, + "step": 7425 + }, + { + "epoch": 14.01, + "learning_rate": 6.4369230769230775e-06, + "loss": 0.0724, + "step": 7450 + }, + { + "epoch": 14.02, + "learning_rate": 6.424102564102564e-06, + "loss": 0.0556, + "step": 7475 + }, + { + "epoch": 14.02, + "learning_rate": 6.411282051282052e-06, + "loss": 0.0557, + "step": 7500 + }, + { + "epoch": 14.02, + "learning_rate": 6.398461538461539e-06, + "loss": 0.0512, + "step": 7525 + }, + { + "epoch": 14.02, + "learning_rate": 6.3856410256410264e-06, + "loss": 0.0493, + "step": 7550 + }, + { + "epoch": 14.02, + "learning_rate": 6.372820512820513e-06, + "loss": 0.0655, + "step": 7575 + }, + { + "epoch": 14.02, + "learning_rate": 6.360000000000001e-06, + "loss": 0.0636, + "step": 7600 + }, + { + "epoch": 14.02, + "learning_rate": 6.347179487179488e-06, + "loss": 0.07, + "step": 7625 + }, + { + "epoch": 14.02, + "learning_rate": 6.3343589743589754e-06, + "loss": 0.0625, + "step": 7650 + }, + { + "epoch": 14.03, + "learning_rate": 6.321538461538462e-06, + "loss": 0.064, + "step": 7675 + }, + { + "epoch": 15.0, + "learning_rate": 6.30871794871795e-06, + "loss": 0.0345, + "step": 7700 + }, + { + "epoch": 15.0, + "learning_rate": 6.295897435897437e-06, + "loss": 0.0326, + "step": 7725 + }, + { + "epoch": 15.0, + "learning_rate": 6.283076923076924e-06, + "loss": 0.0402, + "step": 7750 + }, + { + "epoch": 15.0, + "learning_rate": 6.270256410256411e-06, + "loss": 0.0329, + "step": 7775 + }, + { + "epoch": 15.01, + "learning_rate": 6.257435897435898e-06, + "loss": 0.037, + "step": 7800 + }, + { + "epoch": 15.01, + "learning_rate": 6.244615384615386e-06, + "loss": 0.026, + "step": 7825 + }, + { + "epoch": 15.01, + "learning_rate": 6.2317948717948726e-06, + "loss": 0.025, + "step": 7850 + }, + { + "epoch": 15.01, + "learning_rate": 6.2189743589743586e-06, + "loss": 0.0223, + "step": 7875 + }, + { + "epoch": 15.01, + "learning_rate": 6.206153846153846e-06, + "loss": 0.0336, + "step": 7900 + }, + { + "epoch": 15.01, + "learning_rate": 6.193333333333333e-06, + "loss": 0.0302, + "step": 7925 + }, + { + "epoch": 15.01, + "learning_rate": 6.180512820512821e-06, + "loss": 0.0352, + "step": 7950 + }, + { + "epoch": 15.01, + "learning_rate": 6.1676923076923075e-06, + "loss": 0.0391, + "step": 7975 + }, + { + "epoch": 15.02, + "learning_rate": 6.154871794871795e-06, + "loss": 0.0339, + "step": 8000 + }, + { + "epoch": 15.02, + "eval_loss": 0.15094994008541107, + "eval_runtime": 5928.261, + "eval_samples_per_second": 2.763, + "eval_steps_per_second": 0.173, + "eval_wer": 5.75535867390683, + "step": 8000 + }, + { + "epoch": 15.02, + "learning_rate": 6.142051282051282e-06, + "loss": 0.0377, + "step": 8025 + }, + { + "epoch": 15.02, + "learning_rate": 6.12923076923077e-06, + "loss": 0.047, + "step": 8050 + }, + { + "epoch": 15.02, + "learning_rate": 6.1164102564102565e-06, + "loss": 0.0322, + "step": 8075 + }, + { + "epoch": 15.02, + "learning_rate": 6.103589743589744e-06, + "loss": 0.0206, + "step": 8100 + }, + { + "epoch": 15.02, + "learning_rate": 6.090769230769231e-06, + "loss": 0.0284, + "step": 8125 + }, + { + "epoch": 15.02, + "learning_rate": 6.077948717948719e-06, + "loss": 0.0265, + "step": 8150 + }, + { + "epoch": 15.02, + "learning_rate": 6.0651282051282055e-06, + "loss": 0.0322, + "step": 8175 + }, + { + "epoch": 16.0, + "learning_rate": 6.052307692307693e-06, + "loss": 0.0253, + "step": 8200 + }, + { + "epoch": 16.0, + "learning_rate": 6.03948717948718e-06, + "loss": 0.0323, + "step": 8225 + }, + { + "epoch": 16.0, + "learning_rate": 6.026666666666668e-06, + "loss": 0.0201, + "step": 8250 + }, + { + "epoch": 16.0, + "learning_rate": 6.0138461538461545e-06, + "loss": 0.0163, + "step": 8275 + }, + { + "epoch": 16.01, + "learning_rate": 6.001025641025642e-06, + "loss": 0.0205, + "step": 8300 + }, + { + "epoch": 16.01, + "learning_rate": 5.988205128205129e-06, + "loss": 0.0215, + "step": 8325 + }, + { + "epoch": 16.01, + "learning_rate": 5.975384615384616e-06, + "loss": 0.0218, + "step": 8350 + }, + { + "epoch": 16.01, + "learning_rate": 5.9625641025641035e-06, + "loss": 0.026, + "step": 8375 + }, + { + "epoch": 16.01, + "learning_rate": 5.94974358974359e-06, + "loss": 0.0199, + "step": 8400 + }, + { + "epoch": 16.01, + "learning_rate": 5.936923076923078e-06, + "loss": 0.0226, + "step": 8425 + }, + { + "epoch": 16.01, + "learning_rate": 5.924102564102565e-06, + "loss": 0.0362, + "step": 8450 + }, + { + "epoch": 16.01, + "learning_rate": 5.911282051282052e-06, + "loss": 0.0312, + "step": 8475 + }, + { + "epoch": 16.02, + "learning_rate": 5.8984615384615385e-06, + "loss": 0.0292, + "step": 8500 + }, + { + "epoch": 16.02, + "learning_rate": 5.885641025641025e-06, + "loss": 0.0369, + "step": 8525 + }, + { + "epoch": 16.02, + "learning_rate": 5.872820512820513e-06, + "loss": 0.0367, + "step": 8550 + }, + { + "epoch": 16.02, + "learning_rate": 5.86e-06, + "loss": 0.0404, + "step": 8575 + }, + { + "epoch": 16.02, + "learning_rate": 5.8471794871794875e-06, + "loss": 0.0382, + "step": 8600 + }, + { + "epoch": 16.02, + "learning_rate": 5.834358974358974e-06, + "loss": 0.0369, + "step": 8625 + }, + { + "epoch": 16.02, + "learning_rate": 5.821538461538462e-06, + "loss": 0.0417, + "step": 8650 + }, + { + "epoch": 16.02, + "learning_rate": 5.808717948717949e-06, + "loss": 0.0464, + "step": 8675 + }, + { + "epoch": 16.03, + "learning_rate": 5.7958974358974365e-06, + "loss": 0.0633, + "step": 8700 + }, + { + "epoch": 17.0, + "learning_rate": 5.783589743589744e-06, + "loss": 0.0759, + "step": 8725 + }, + { + "epoch": 17.0, + "learning_rate": 5.770769230769231e-06, + "loss": 0.0716, + "step": 8750 + }, + { + "epoch": 17.0, + "learning_rate": 5.757948717948718e-06, + "loss": 0.0338, + "step": 8775 + }, + { + "epoch": 17.0, + "learning_rate": 5.745128205128205e-06, + "loss": 0.0465, + "step": 8800 + }, + { + "epoch": 17.01, + "learning_rate": 5.732307692307693e-06, + "loss": 0.05, + "step": 8825 + }, + { + "epoch": 17.01, + "learning_rate": 5.71948717948718e-06, + "loss": 0.0321, + "step": 8850 + }, + { + "epoch": 17.01, + "learning_rate": 5.706666666666667e-06, + "loss": 0.0212, + "step": 8875 + }, + { + "epoch": 17.01, + "learning_rate": 5.693846153846154e-06, + "loss": 0.0329, + "step": 8900 + }, + { + "epoch": 17.01, + "learning_rate": 5.681025641025642e-06, + "loss": 0.0495, + "step": 8925 + }, + { + "epoch": 17.01, + "learning_rate": 5.668205128205129e-06, + "loss": 0.042, + "step": 8950 + }, + { + "epoch": 17.01, + "learning_rate": 5.6553846153846155e-06, + "loss": 0.0565, + "step": 8975 + }, + { + "epoch": 17.01, + "learning_rate": 5.642564102564103e-06, + "loss": 0.0581, + "step": 9000 + }, + { + "epoch": 17.01, + "eval_loss": 0.15734407305717468, + "eval_runtime": 5915.2914, + "eval_samples_per_second": 2.769, + "eval_steps_per_second": 0.173, + "eval_wer": 6.04458416690483, + "step": 9000 + }, + { + "epoch": 17.02, + "learning_rate": 5.62974358974359e-06, + "loss": 0.043, + "step": 9025 + }, + { + "epoch": 17.02, + "learning_rate": 5.616923076923078e-06, + "loss": 0.0513, + "step": 9050 + }, + { + "epoch": 17.02, + "learning_rate": 5.6041025641025645e-06, + "loss": 0.0523, + "step": 9075 + }, + { + "epoch": 17.02, + "learning_rate": 5.591282051282052e-06, + "loss": 0.0561, + "step": 9100 + }, + { + "epoch": 17.02, + "learning_rate": 5.578461538461539e-06, + "loss": 0.0411, + "step": 9125 + }, + { + "epoch": 17.02, + "learning_rate": 5.565641025641027e-06, + "loss": 0.0467, + "step": 9150 + }, + { + "epoch": 17.02, + "learning_rate": 5.5528205128205135e-06, + "loss": 0.0479, + "step": 9175 + }, + { + "epoch": 17.02, + "learning_rate": 5.540000000000001e-06, + "loss": 0.0559, + "step": 9200 + }, + { + "epoch": 18.0, + "learning_rate": 5.527179487179488e-06, + "loss": 0.0541, + "step": 9225 + }, + { + "epoch": 18.0, + "learning_rate": 5.514358974358974e-06, + "loss": 0.0368, + "step": 9250 + }, + { + "epoch": 18.0, + "learning_rate": 5.501538461538462e-06, + "loss": 0.031, + "step": 9275 + }, + { + "epoch": 18.0, + "learning_rate": 5.4887179487179485e-06, + "loss": 0.0344, + "step": 9300 + }, + { + "epoch": 18.01, + "learning_rate": 5.475897435897436e-06, + "loss": 0.0313, + "step": 9325 + }, + { + "epoch": 18.01, + "learning_rate": 5.463076923076923e-06, + "loss": 0.0276, + "step": 9350 + }, + { + "epoch": 18.01, + "learning_rate": 5.450256410256411e-06, + "loss": 0.0305, + "step": 9375 + }, + { + "epoch": 18.01, + "learning_rate": 5.4374358974358975e-06, + "loss": 0.0313, + "step": 9400 + }, + { + "epoch": 18.01, + "learning_rate": 5.424615384615385e-06, + "loss": 0.0279, + "step": 9425 + }, + { + "epoch": 18.01, + "learning_rate": 5.411794871794872e-06, + "loss": 0.0296, + "step": 9450 + }, + { + "epoch": 18.01, + "learning_rate": 5.39897435897436e-06, + "loss": 0.0288, + "step": 9475 + }, + { + "epoch": 18.01, + "learning_rate": 5.3861538461538464e-06, + "loss": 0.0233, + "step": 9500 + }, + { + "epoch": 18.02, + "learning_rate": 5.373333333333334e-06, + "loss": 0.0261, + "step": 9525 + }, + { + "epoch": 18.02, + "learning_rate": 5.360512820512821e-06, + "loss": 0.0285, + "step": 9550 + }, + { + "epoch": 18.02, + "learning_rate": 5.347692307692308e-06, + "loss": 0.0256, + "step": 9575 + }, + { + "epoch": 18.02, + "learning_rate": 5.3348717948717954e-06, + "loss": 0.0251, + "step": 9600 + }, + { + "epoch": 18.02, + "learning_rate": 5.322051282051282e-06, + "loss": 0.0233, + "step": 9625 + }, + { + "epoch": 18.02, + "learning_rate": 5.30923076923077e-06, + "loss": 0.0218, + "step": 9650 + }, + { + "epoch": 18.02, + "learning_rate": 5.296410256410257e-06, + "loss": 0.0232, + "step": 9675 + }, + { + "epoch": 18.02, + "learning_rate": 5.2835897435897444e-06, + "loss": 0.0229, + "step": 9700 + }, + { + "epoch": 18.03, + "learning_rate": 5.270769230769231e-06, + "loss": 0.0207, + "step": 9725 + }, + { + "epoch": 19.0, + "learning_rate": 5.257948717948719e-06, + "loss": 0.02, + "step": 9750 + }, + { + "epoch": 19.0, + "learning_rate": 5.245128205128206e-06, + "loss": 0.0198, + "step": 9775 + }, + { + "epoch": 19.0, + "learning_rate": 5.2323076923076934e-06, + "loss": 0.019, + "step": 9800 + }, + { + "epoch": 19.0, + "learning_rate": 5.21948717948718e-06, + "loss": 0.0156, + "step": 9825 + }, + { + "epoch": 19.01, + "learning_rate": 5.206666666666668e-06, + "loss": 0.0171, + "step": 9850 + }, + { + "epoch": 19.01, + "learning_rate": 5.193846153846154e-06, + "loss": 0.0175, + "step": 9875 + }, + { + "epoch": 19.01, + "learning_rate": 5.181025641025641e-06, + "loss": 0.0188, + "step": 9900 + }, + { + "epoch": 19.01, + "learning_rate": 5.168205128205128e-06, + "loss": 0.0153, + "step": 9925 + }, + { + "epoch": 19.01, + "learning_rate": 5.155384615384615e-06, + "loss": 0.0163, + "step": 9950 + }, + { + "epoch": 19.01, + "learning_rate": 5.142564102564103e-06, + "loss": 0.0156, + "step": 9975 + }, + { + "epoch": 19.01, + "learning_rate": 5.12974358974359e-06, + "loss": 0.0181, + "step": 10000 + }, + { + "epoch": 19.01, + "eval_loss": 0.1838178187608719, + "eval_runtime": 5865.1115, + "eval_samples_per_second": 2.793, + "eval_steps_per_second": 0.175, + "eval_wer": 5.591311803372393, + "step": 10000 + }, + { + "epoch": 19.01, + "learning_rate": 5.116923076923077e-06, + "loss": 0.0181, + "step": 10025 + }, + { + "epoch": 19.02, + "learning_rate": 5.104102564102564e-06, + "loss": 0.0185, + "step": 10050 + }, + { + "epoch": 19.02, + "learning_rate": 5.091282051282052e-06, + "loss": 0.0211, + "step": 10075 + }, + { + "epoch": 19.02, + "learning_rate": 5.078461538461539e-06, + "loss": 0.0201, + "step": 10100 + }, + { + "epoch": 19.02, + "learning_rate": 5.065641025641026e-06, + "loss": 0.0199, + "step": 10125 + }, + { + "epoch": 19.02, + "learning_rate": 5.052820512820513e-06, + "loss": 0.0186, + "step": 10150 + }, + { + "epoch": 19.02, + "learning_rate": 5.04e-06, + "loss": 0.0195, + "step": 10175 + }, + { + "epoch": 19.02, + "learning_rate": 5.027179487179488e-06, + "loss": 0.0205, + "step": 10200 + }, + { + "epoch": 19.02, + "learning_rate": 5.0143589743589745e-06, + "loss": 0.028, + "step": 10225 + }, + { + "epoch": 20.0, + "learning_rate": 5.001538461538462e-06, + "loss": 0.0268, + "step": 10250 + }, + { + "epoch": 20.0, + "learning_rate": 4.988717948717949e-06, + "loss": 0.0205, + "step": 10275 + }, + { + "epoch": 20.0, + "learning_rate": 4.975897435897437e-06, + "loss": 0.0209, + "step": 10300 + }, + { + "epoch": 20.0, + "learning_rate": 4.9630769230769235e-06, + "loss": 0.017, + "step": 10325 + }, + { + "epoch": 20.01, + "learning_rate": 4.95025641025641e-06, + "loss": 0.0263, + "step": 10350 + }, + { + "epoch": 20.01, + "learning_rate": 4.937435897435898e-06, + "loss": 0.0224, + "step": 10375 + }, + { + "epoch": 20.01, + "learning_rate": 4.924615384615385e-06, + "loss": 0.0208, + "step": 10400 + }, + { + "epoch": 20.01, + "learning_rate": 4.9117948717948725e-06, + "loss": 0.0221, + "step": 10425 + }, + { + "epoch": 20.01, + "learning_rate": 4.898974358974359e-06, + "loss": 0.0196, + "step": 10450 + }, + { + "epoch": 20.01, + "learning_rate": 4.886153846153847e-06, + "loss": 0.0161, + "step": 10475 + }, + { + "epoch": 20.01, + "learning_rate": 4.873333333333334e-06, + "loss": 0.0178, + "step": 10500 + }, + { + "epoch": 20.01, + "learning_rate": 4.860512820512821e-06, + "loss": 0.0245, + "step": 10525 + }, + { + "epoch": 20.02, + "learning_rate": 4.847692307692308e-06, + "loss": 0.0201, + "step": 10550 + }, + { + "epoch": 20.02, + "learning_rate": 4.834871794871795e-06, + "loss": 0.0189, + "step": 10575 + }, + { + "epoch": 20.02, + "learning_rate": 4.822051282051283e-06, + "loss": 0.0163, + "step": 10600 + }, + { + "epoch": 20.02, + "learning_rate": 4.80923076923077e-06, + "loss": 0.0149, + "step": 10625 + }, + { + "epoch": 20.02, + "learning_rate": 4.7964102564102565e-06, + "loss": 0.021, + "step": 10650 + }, + { + "epoch": 20.02, + "learning_rate": 4.783589743589744e-06, + "loss": 0.0216, + "step": 10675 + }, + { + "epoch": 20.02, + "learning_rate": 4.770769230769231e-06, + "loss": 0.0241, + "step": 10700 + }, + { + "epoch": 20.02, + "learning_rate": 4.757948717948719e-06, + "loss": 0.0204, + "step": 10725 + }, + { + "epoch": 20.03, + "learning_rate": 4.7451282051282055e-06, + "loss": 0.0226, + "step": 10750 + }, + { + "epoch": 21.0, + "learning_rate": 4.732307692307693e-06, + "loss": 0.0214, + "step": 10775 + }, + { + "epoch": 21.0, + "learning_rate": 4.71948717948718e-06, + "loss": 0.0273, + "step": 10800 + }, + { + "epoch": 21.0, + "learning_rate": 4.706666666666667e-06, + "loss": 0.0191, + "step": 10825 + }, + { + "epoch": 21.0, + "learning_rate": 4.6938461538461544e-06, + "loss": 0.0133, + "step": 10850 + }, + { + "epoch": 21.01, + "learning_rate": 4.681025641025641e-06, + "loss": 0.0213, + "step": 10875 + }, + { + "epoch": 21.01, + "learning_rate": 4.668205128205129e-06, + "loss": 0.0202, + "step": 10900 + }, + { + "epoch": 21.01, + "learning_rate": 4.655384615384616e-06, + "loss": 0.0158, + "step": 10925 + }, + { + "epoch": 21.01, + "learning_rate": 4.642564102564103e-06, + "loss": 0.0173, + "step": 10950 + }, + { + "epoch": 21.01, + "learning_rate": 4.62974358974359e-06, + "loss": 0.0225, + "step": 10975 + }, + { + "epoch": 21.01, + "learning_rate": 4.616923076923077e-06, + "loss": 0.0188, + "step": 11000 + }, + { + "epoch": 21.01, + "eval_loss": 0.16100507974624634, + "eval_runtime": 5909.3756, + "eval_samples_per_second": 2.772, + "eval_steps_per_second": 0.173, + "eval_wer": 5.480422977993713, + "step": 11000 + }, + { + "epoch": 21.01, + "learning_rate": 4.604102564102565e-06, + "loss": 0.0199, + "step": 11025 + }, + { + "epoch": 21.01, + "learning_rate": 4.591282051282052e-06, + "loss": 0.0211, + "step": 11050 + }, + { + "epoch": 21.02, + "learning_rate": 4.578461538461539e-06, + "loss": 0.0213, + "step": 11075 + }, + { + "epoch": 21.02, + "learning_rate": 4.565641025641026e-06, + "loss": 0.0201, + "step": 11100 + }, + { + "epoch": 21.02, + "learning_rate": 4.552820512820513e-06, + "loss": 0.0201, + "step": 11125 + }, + { + "epoch": 21.02, + "learning_rate": 4.540000000000001e-06, + "loss": 0.0204, + "step": 11150 + }, + { + "epoch": 21.02, + "learning_rate": 4.527179487179487e-06, + "loss": 0.0176, + "step": 11175 + }, + { + "epoch": 21.02, + "learning_rate": 4.514358974358975e-06, + "loss": 0.0164, + "step": 11200 + }, + { + "epoch": 21.02, + "learning_rate": 4.501538461538462e-06, + "loss": 0.0189, + "step": 11225 + }, + { + "epoch": 21.02, + "learning_rate": 4.488717948717949e-06, + "loss": 0.0221, + "step": 11250 + }, + { + "epoch": 22.0, + "learning_rate": 4.475897435897436e-06, + "loss": 0.0209, + "step": 11275 + }, + { + "epoch": 22.0, + "learning_rate": 4.463076923076923e-06, + "loss": 0.0157, + "step": 11300 + }, + { + "epoch": 22.0, + "learning_rate": 4.450256410256411e-06, + "loss": 0.016, + "step": 11325 + }, + { + "epoch": 22.0, + "learning_rate": 4.437435897435898e-06, + "loss": 0.0206, + "step": 11350 + }, + { + "epoch": 22.01, + "learning_rate": 4.424615384615385e-06, + "loss": 0.0193, + "step": 11375 + }, + { + "epoch": 22.01, + "learning_rate": 4.411794871794872e-06, + "loss": 0.0168, + "step": 11400 + }, + { + "epoch": 22.01, + "learning_rate": 4.398974358974359e-06, + "loss": 0.0192, + "step": 11425 + }, + { + "epoch": 22.01, + "learning_rate": 4.386153846153847e-06, + "loss": 0.0213, + "step": 11450 + }, + { + "epoch": 22.01, + "learning_rate": 4.3733333333333335e-06, + "loss": 0.0201, + "step": 11475 + }, + { + "epoch": 22.01, + "learning_rate": 4.360512820512821e-06, + "loss": 0.0184, + "step": 11500 + }, + { + "epoch": 22.01, + "learning_rate": 4.347692307692308e-06, + "loss": 0.0177, + "step": 11525 + }, + { + "epoch": 22.01, + "learning_rate": 4.334871794871795e-06, + "loss": 0.0194, + "step": 11550 + }, + { + "epoch": 22.02, + "learning_rate": 4.3220512820512825e-06, + "loss": 0.0185, + "step": 11575 + }, + { + "epoch": 22.02, + "learning_rate": 4.309230769230769e-06, + "loss": 0.0187, + "step": 11600 + }, + { + "epoch": 22.02, + "learning_rate": 4.296410256410257e-06, + "loss": 0.01, + "step": 11625 + }, + { + "epoch": 22.02, + "learning_rate": 4.283589743589744e-06, + "loss": 0.0041, + "step": 11650 + }, + { + "epoch": 22.02, + "learning_rate": 4.2707692307692315e-06, + "loss": 0.005, + "step": 11675 + }, + { + "epoch": 22.02, + "learning_rate": 4.257948717948718e-06, + "loss": 0.0058, + "step": 11700 + }, + { + "epoch": 22.02, + "learning_rate": 4.245128205128205e-06, + "loss": 0.0055, + "step": 11725 + }, + { + "epoch": 22.02, + "learning_rate": 4.232307692307693e-06, + "loss": 0.0039, + "step": 11750 + }, + { + "epoch": 22.03, + "learning_rate": 4.21948717948718e-06, + "loss": 0.0064, + "step": 11775 + }, + { + "epoch": 23.0, + "learning_rate": 4.206666666666667e-06, + "loss": 0.0183, + "step": 11800 + }, + { + "epoch": 23.0, + "learning_rate": 4.193846153846154e-06, + "loss": 0.0155, + "step": 11825 + }, + { + "epoch": 23.0, + "learning_rate": 4.181025641025641e-06, + "loss": 0.0158, + "step": 11850 + }, + { + "epoch": 23.0, + "learning_rate": 4.168205128205129e-06, + "loss": 0.0143, + "step": 11875 + }, + { + "epoch": 23.01, + "learning_rate": 4.1553846153846155e-06, + "loss": 0.0139, + "step": 11900 + }, + { + "epoch": 23.01, + "learning_rate": 4.142564102564103e-06, + "loss": 0.0135, + "step": 11925 + }, + { + "epoch": 23.01, + "learning_rate": 4.12974358974359e-06, + "loss": 0.0147, + "step": 11950 + }, + { + "epoch": 23.01, + "learning_rate": 4.116923076923078e-06, + "loss": 0.0134, + "step": 11975 + }, + { + "epoch": 23.01, + "learning_rate": 4.1041025641025645e-06, + "loss": 0.0134, + "step": 12000 + }, + { + "epoch": 23.01, + "eval_loss": 0.18212418258190155, + "eval_runtime": 5915.4743, + "eval_samples_per_second": 2.769, + "eval_steps_per_second": 0.173, + "eval_wer": 5.395255787367819, + "step": 12000 + }, + { + "epoch": 23.01, + "learning_rate": 4.091282051282051e-06, + "loss": 0.0124, + "step": 12025 + }, + { + "epoch": 23.01, + "learning_rate": 4.078461538461539e-06, + "loss": 0.0123, + "step": 12050 + }, + { + "epoch": 23.01, + "learning_rate": 4.065641025641026e-06, + "loss": 0.0127, + "step": 12075 + }, + { + "epoch": 23.02, + "learning_rate": 4.0528205128205135e-06, + "loss": 0.0129, + "step": 12100 + }, + { + "epoch": 23.02, + "learning_rate": 4.04e-06, + "loss": 0.0117, + "step": 12125 + }, + { + "epoch": 23.02, + "learning_rate": 4.027179487179487e-06, + "loss": 0.0123, + "step": 12150 + }, + { + "epoch": 23.02, + "learning_rate": 4.014358974358975e-06, + "loss": 0.011, + "step": 12175 + }, + { + "epoch": 23.02, + "learning_rate": 4.001538461538462e-06, + "loss": 0.0115, + "step": 12200 + }, + { + "epoch": 23.02, + "learning_rate": 3.988717948717949e-06, + "loss": 0.0113, + "step": 12225 + }, + { + "epoch": 23.02, + "learning_rate": 3.975897435897436e-06, + "loss": 0.0101, + "step": 12250 + }, + { + "epoch": 23.02, + "learning_rate": 3.963076923076924e-06, + "loss": 0.0106, + "step": 12275 + }, + { + "epoch": 24.0, + "learning_rate": 3.950256410256411e-06, + "loss": 0.0132, + "step": 12300 + }, + { + "epoch": 24.0, + "learning_rate": 3.937435897435897e-06, + "loss": 0.0298, + "step": 12325 + }, + { + "epoch": 24.0, + "learning_rate": 3.924615384615385e-06, + "loss": 0.0261, + "step": 12350 + }, + { + "epoch": 24.0, + "learning_rate": 3.911794871794872e-06, + "loss": 0.0294, + "step": 12375 + }, + { + "epoch": 24.01, + "learning_rate": 3.89897435897436e-06, + "loss": 0.0255, + "step": 12400 + }, + { + "epoch": 24.01, + "learning_rate": 3.886153846153846e-06, + "loss": 0.0259, + "step": 12425 + }, + { + "epoch": 24.01, + "learning_rate": 3.873333333333333e-06, + "loss": 0.028, + "step": 12450 + }, + { + "epoch": 24.01, + "learning_rate": 3.860512820512821e-06, + "loss": 0.026, + "step": 12475 + }, + { + "epoch": 24.01, + "learning_rate": 3.847692307692308e-06, + "loss": 0.0233, + "step": 12500 + }, + { + "epoch": 24.01, + "learning_rate": 3.834871794871795e-06, + "loss": 0.024, + "step": 12525 + }, + { + "epoch": 24.01, + "learning_rate": 3.822051282051282e-06, + "loss": 0.0238, + "step": 12550 + }, + { + "epoch": 24.01, + "learning_rate": 3.8092307692307695e-06, + "loss": 0.0249, + "step": 12575 + }, + { + "epoch": 24.02, + "learning_rate": 3.7964102564102567e-06, + "loss": 0.0204, + "step": 12600 + }, + { + "epoch": 24.02, + "learning_rate": 3.783589743589744e-06, + "loss": 0.0228, + "step": 12625 + }, + { + "epoch": 24.02, + "learning_rate": 3.7707692307692312e-06, + "loss": 0.0253, + "step": 12650 + }, + { + "epoch": 24.02, + "learning_rate": 3.7579487179487185e-06, + "loss": 0.0211, + "step": 12675 + }, + { + "epoch": 24.02, + "learning_rate": 3.7451282051282057e-06, + "loss": 0.0252, + "step": 12700 + }, + { + "epoch": 24.02, + "learning_rate": 3.732307692307693e-06, + "loss": 0.0241, + "step": 12725 + }, + { + "epoch": 24.02, + "learning_rate": 3.7194871794871794e-06, + "loss": 0.0267, + "step": 12750 + }, + { + "epoch": 24.02, + "learning_rate": 3.7066666666666666e-06, + "loss": 0.0226, + "step": 12775 + }, + { + "epoch": 24.03, + "learning_rate": 3.693846153846154e-06, + "loss": 0.0288, + "step": 12800 + }, + { + "epoch": 25.0, + "learning_rate": 3.681025641025641e-06, + "loss": 0.0034, + "step": 12825 + }, + { + "epoch": 25.0, + "learning_rate": 3.6682051282051284e-06, + "loss": 0.0027, + "step": 12850 + }, + { + "epoch": 25.0, + "learning_rate": 3.6553846153846156e-06, + "loss": 0.0028, + "step": 12875 + }, + { + "epoch": 25.0, + "learning_rate": 3.642564102564103e-06, + "loss": 0.0074, + "step": 12900 + }, + { + "epoch": 25.01, + "learning_rate": 3.62974358974359e-06, + "loss": 0.0082, + "step": 12925 + }, + { + "epoch": 25.01, + "learning_rate": 3.6169230769230773e-06, + "loss": 0.0088, + "step": 12950 + }, + { + "epoch": 25.01, + "learning_rate": 3.6041025641025646e-06, + "loss": 0.0084, + "step": 12975 + }, + { + "epoch": 25.01, + "learning_rate": 3.591282051282052e-06, + "loss": 0.008, + "step": 13000 + }, + { + "epoch": 25.01, + "eval_loss": 0.17475968599319458, + "eval_runtime": 6346.4418, + "eval_samples_per_second": 2.581, + "eval_steps_per_second": 0.161, + "eval_wer": 5.380394398399543, + "step": 13000 + }, + { + "epoch": 25.01, + "learning_rate": 3.578461538461539e-06, + "loss": 0.0075, + "step": 13025 + }, + { + "epoch": 25.01, + "learning_rate": 3.5656410256410263e-06, + "loss": 0.0082, + "step": 13050 + }, + { + "epoch": 25.01, + "learning_rate": 3.5533333333333338e-06, + "loss": 0.0098, + "step": 13075 + }, + { + "epoch": 25.02, + "learning_rate": 3.540512820512821e-06, + "loss": 0.0085, + "step": 13100 + }, + { + "epoch": 25.02, + "learning_rate": 3.5276923076923083e-06, + "loss": 0.0073, + "step": 13125 + }, + { + "epoch": 25.02, + "learning_rate": 3.5148717948717955e-06, + "loss": 0.0091, + "step": 13150 + }, + { + "epoch": 25.02, + "learning_rate": 3.5020512820512823e-06, + "loss": 0.008, + "step": 13175 + }, + { + "epoch": 25.02, + "learning_rate": 3.4892307692307696e-06, + "loss": 0.009, + "step": 13200 + }, + { + "epoch": 25.02, + "learning_rate": 3.4764102564102564e-06, + "loss": 0.0095, + "step": 13225 + }, + { + "epoch": 25.02, + "learning_rate": 3.4635897435897436e-06, + "loss": 0.0124, + "step": 13250 + }, + { + "epoch": 25.02, + "learning_rate": 3.450769230769231e-06, + "loss": 0.0163, + "step": 13275 + }, + { + "epoch": 25.02, + "learning_rate": 3.437948717948718e-06, + "loss": 0.0104, + "step": 13300 + }, + { + "epoch": 26.0, + "learning_rate": 3.425641025641026e-06, + "loss": 0.0156, + "step": 13325 + }, + { + "epoch": 26.0, + "learning_rate": 3.416410256410257e-06, + "loss": 5.424, + "step": 13350 + }, + { + "epoch": 26.0, + "learning_rate": 3.404615384615385e-06, + "loss": 7.2571, + "step": 13375 + }, + { + "epoch": 26.0, + "learning_rate": 3.3917948717948722e-06, + "loss": 5.5275, + "step": 13400 + }, + { + "epoch": 26.01, + "learning_rate": 3.378974358974359e-06, + "loss": 0.5044, + "step": 13425 + }, + { + "epoch": 26.01, + "learning_rate": 3.3661538461538463e-06, + "loss": 0.1523, + "step": 13450 + }, + { + "epoch": 26.01, + "learning_rate": 3.3533333333333336e-06, + "loss": 0.0268, + "step": 13475 + }, + { + "epoch": 26.01, + "learning_rate": 3.340512820512821e-06, + "loss": 0.0269, + "step": 13500 + }, + { + "epoch": 26.01, + "learning_rate": 3.327692307692308e-06, + "loss": 0.0112, + "step": 13525 + }, + { + "epoch": 26.01, + "learning_rate": 3.3148717948717953e-06, + "loss": 0.021, + "step": 13550 + }, + { + "epoch": 26.01, + "learning_rate": 3.3020512820512825e-06, + "loss": 0.0188, + "step": 13575 + }, + { + "epoch": 26.01, + "learning_rate": 3.2892307692307694e-06, + "loss": 0.0098, + "step": 13600 + }, + { + "epoch": 26.02, + "learning_rate": 3.2764102564102566e-06, + "loss": 0.0125, + "step": 13625 + }, + { + "epoch": 26.02, + "learning_rate": 3.263589743589744e-06, + "loss": 0.0136, + "step": 13650 + }, + { + "epoch": 26.02, + "learning_rate": 3.250769230769231e-06, + "loss": 0.0164, + "step": 13675 + }, + { + "epoch": 26.02, + "learning_rate": 3.2379487179487184e-06, + "loss": 0.0159, + "step": 13700 + }, + { + "epoch": 26.02, + "learning_rate": 3.225128205128205e-06, + "loss": 0.0162, + "step": 13725 + }, + { + "epoch": 26.02, + "learning_rate": 3.2123076923076924e-06, + "loss": 0.0187, + "step": 13750 + }, + { + "epoch": 26.02, + "learning_rate": 3.1994871794871797e-06, + "loss": 0.0161, + "step": 13775 + }, + { + "epoch": 26.02, + "learning_rate": 3.186666666666667e-06, + "loss": 0.0168, + "step": 13800 + }, + { + "epoch": 27.0, + "learning_rate": 3.173846153846154e-06, + "loss": 0.0209, + "step": 13825 + }, + { + "epoch": 27.0, + "learning_rate": 3.1610256410256414e-06, + "loss": 0.0087, + "step": 13850 + }, + { + "epoch": 27.0, + "learning_rate": 3.1482051282051287e-06, + "loss": 0.0072, + "step": 13875 + }, + { + "epoch": 27.0, + "learning_rate": 3.1353846153846155e-06, + "loss": 0.0067, + "step": 13900 + }, + { + "epoch": 27.01, + "learning_rate": 3.1225641025641027e-06, + "loss": 0.008, + "step": 13925 + }, + { + "epoch": 27.01, + "learning_rate": 3.10974358974359e-06, + "loss": 0.0057, + "step": 13950 + }, + { + "epoch": 27.01, + "learning_rate": 3.0969230769230772e-06, + "loss": 0.0054, + "step": 13975 + }, + { + "epoch": 27.01, + "learning_rate": 3.0841025641025645e-06, + "loss": 0.0071, + "step": 14000 + }, + { + "epoch": 27.01, + "eval_loss": 0.18584036827087402, + "eval_runtime": 5885.4981, + "eval_samples_per_second": 2.783, + "eval_steps_per_second": 0.174, + "eval_wer": 5.470134324092598, + "step": 14000 + }, + { + "epoch": 27.01, + "learning_rate": 3.0712820512820513e-06, + "loss": 0.0065, + "step": 14025 + }, + { + "epoch": 27.01, + "learning_rate": 3.0584615384615386e-06, + "loss": 0.0065, + "step": 14050 + }, + { + "epoch": 27.01, + "learning_rate": 3.045641025641026e-06, + "loss": 0.0064, + "step": 14075 + }, + { + "epoch": 27.01, + "learning_rate": 3.032820512820513e-06, + "loss": 0.0071, + "step": 14100 + }, + { + "epoch": 27.02, + "learning_rate": 3.0200000000000003e-06, + "loss": 0.0075, + "step": 14125 + }, + { + "epoch": 27.02, + "learning_rate": 3.0071794871794876e-06, + "loss": 0.0068, + "step": 14150 + }, + { + "epoch": 27.02, + "learning_rate": 2.994358974358975e-06, + "loss": 0.0046, + "step": 14175 + }, + { + "epoch": 27.02, + "learning_rate": 2.9815384615384616e-06, + "loss": 0.0025, + "step": 14200 + }, + { + "epoch": 27.02, + "learning_rate": 2.968717948717949e-06, + "loss": 0.0028, + "step": 14225 + }, + { + "epoch": 27.02, + "learning_rate": 2.955897435897436e-06, + "loss": 0.0016, + "step": 14250 + }, + { + "epoch": 27.02, + "learning_rate": 2.9430769230769234e-06, + "loss": 0.0024, + "step": 14275 + }, + { + "epoch": 27.02, + "learning_rate": 2.9302564102564106e-06, + "loss": 0.0014, + "step": 14300 + }, + { + "epoch": 27.03, + "learning_rate": 2.917435897435898e-06, + "loss": 0.0015, + "step": 14325 + }, + { + "epoch": 28.0, + "learning_rate": 2.9046153846153847e-06, + "loss": 0.0213, + "step": 14350 + }, + { + "epoch": 28.0, + "learning_rate": 2.891794871794872e-06, + "loss": 0.0359, + "step": 14375 + }, + { + "epoch": 28.0, + "learning_rate": 2.878974358974359e-06, + "loss": 0.0301, + "step": 14400 + }, + { + "epoch": 28.0, + "learning_rate": 2.8661538461538464e-06, + "loss": 0.032, + "step": 14425 + }, + { + "epoch": 28.01, + "learning_rate": 2.8533333333333337e-06, + "loss": 0.0258, + "step": 14450 + }, + { + "epoch": 28.01, + "learning_rate": 2.840512820512821e-06, + "loss": 0.0258, + "step": 14475 + }, + { + "epoch": 28.01, + "learning_rate": 2.8276923076923078e-06, + "loss": 0.0266, + "step": 14500 + }, + { + "epoch": 28.01, + "learning_rate": 2.814871794871795e-06, + "loss": 0.0264, + "step": 14525 + }, + { + "epoch": 28.01, + "learning_rate": 2.8020512820512823e-06, + "loss": 0.0195, + "step": 14550 + }, + { + "epoch": 28.01, + "learning_rate": 2.7892307692307695e-06, + "loss": 0.0061, + "step": 14575 + }, + { + "epoch": 28.01, + "learning_rate": 2.7764102564102567e-06, + "loss": 0.0061, + "step": 14600 + }, + { + "epoch": 28.01, + "learning_rate": 2.763589743589744e-06, + "loss": 0.0076, + "step": 14625 + }, + { + "epoch": 28.02, + "learning_rate": 2.750769230769231e-06, + "loss": 0.0096, + "step": 14650 + }, + { + "epoch": 28.02, + "learning_rate": 2.737948717948718e-06, + "loss": 0.008, + "step": 14675 + }, + { + "epoch": 28.02, + "learning_rate": 2.7251282051282053e-06, + "loss": 0.0073, + "step": 14700 + }, + { + "epoch": 28.02, + "learning_rate": 2.7123076923076926e-06, + "loss": 0.0078, + "step": 14725 + }, + { + "epoch": 28.02, + "learning_rate": 2.69948717948718e-06, + "loss": 0.0148, + "step": 14750 + }, + { + "epoch": 28.02, + "learning_rate": 2.686666666666667e-06, + "loss": 0.0172, + "step": 14775 + }, + { + "epoch": 28.02, + "learning_rate": 2.673846153846154e-06, + "loss": 0.0219, + "step": 14800 + }, + { + "epoch": 28.02, + "learning_rate": 2.661025641025641e-06, + "loss": 0.0157, + "step": 14825 + }, + { + "epoch": 29.0, + "learning_rate": 2.6482051282051284e-06, + "loss": 0.0197, + "step": 14850 + }, + { + "epoch": 29.0, + "learning_rate": 2.6353846153846156e-06, + "loss": 0.033, + "step": 14875 + }, + { + "epoch": 29.0, + "learning_rate": 2.622564102564103e-06, + "loss": 0.0261, + "step": 14900 + }, + { + "epoch": 29.0, + "learning_rate": 2.60974358974359e-06, + "loss": 0.0267, + "step": 14925 + }, + { + "epoch": 29.01, + "learning_rate": 2.596923076923077e-06, + "loss": 0.0397, + "step": 14950 + }, + { + "epoch": 29.01, + "learning_rate": 2.584102564102564e-06, + "loss": 0.0383, + "step": 14975 + }, + { + "epoch": 29.01, + "learning_rate": 2.5712820512820514e-06, + "loss": 0.0371, + "step": 15000 + }, + { + "epoch": 29.01, + "eval_loss": 0.16095739603042603, + "eval_runtime": 5887.3686, + "eval_samples_per_second": 2.782, + "eval_steps_per_second": 0.174, + "eval_wer": 5.659902829379823, + "step": 15000 + }, + { + "epoch": 29.01, + "learning_rate": 2.5584615384615387e-06, + "loss": 0.0405, + "step": 15025 + }, + { + "epoch": 29.01, + "learning_rate": 2.545641025641026e-06, + "loss": 0.0404, + "step": 15050 + }, + { + "epoch": 29.01, + "learning_rate": 2.532820512820513e-06, + "loss": 0.0497, + "step": 15075 + }, + { + "epoch": 29.01, + "learning_rate": 2.52e-06, + "loss": 0.0338, + "step": 15100 + }, + { + "epoch": 29.01, + "learning_rate": 2.5071794871794873e-06, + "loss": 0.0409, + "step": 15125 + }, + { + "epoch": 29.02, + "learning_rate": 2.4943589743589745e-06, + "loss": 0.0384, + "step": 15150 + }, + { + "epoch": 29.02, + "learning_rate": 2.4815384615384618e-06, + "loss": 0.0348, + "step": 15175 + }, + { + "epoch": 29.02, + "learning_rate": 2.468717948717949e-06, + "loss": 0.0348, + "step": 15200 + }, + { + "epoch": 29.02, + "learning_rate": 2.4558974358974363e-06, + "loss": 0.0351, + "step": 15225 + }, + { + "epoch": 29.02, + "learning_rate": 2.4430769230769235e-06, + "loss": 0.0421, + "step": 15250 + }, + { + "epoch": 29.02, + "learning_rate": 2.4302564102564103e-06, + "loss": 0.0456, + "step": 15275 + }, + { + "epoch": 29.02, + "learning_rate": 2.4174358974358976e-06, + "loss": 0.0353, + "step": 15300 + }, + { + "epoch": 29.02, + "learning_rate": 2.404615384615385e-06, + "loss": 0.0357, + "step": 15325 + }, + { + "epoch": 29.03, + "learning_rate": 2.391794871794872e-06, + "loss": 0.0354, + "step": 15350 + }, + { + "epoch": 30.0, + "learning_rate": 2.3789743589743593e-06, + "loss": 0.0233, + "step": 15375 + }, + { + "epoch": 30.0, + "learning_rate": 2.3661538461538466e-06, + "loss": 0.0105, + "step": 15400 + }, + { + "epoch": 30.0, + "learning_rate": 2.3533333333333334e-06, + "loss": 0.0084, + "step": 15425 + }, + { + "epoch": 30.0, + "learning_rate": 2.3405128205128206e-06, + "loss": 0.0075, + "step": 15450 + }, + { + "epoch": 30.01, + "learning_rate": 2.327692307692308e-06, + "loss": 0.0085, + "step": 15475 + }, + { + "epoch": 30.01, + "learning_rate": 2.314871794871795e-06, + "loss": 0.0077, + "step": 15500 + }, + { + "epoch": 30.01, + "learning_rate": 2.3020512820512824e-06, + "loss": 0.0067, + "step": 15525 + }, + { + "epoch": 30.01, + "learning_rate": 2.2892307692307696e-06, + "loss": 0.0077, + "step": 15550 + }, + { + "epoch": 30.01, + "learning_rate": 2.2764102564102565e-06, + "loss": 0.0073, + "step": 15575 + }, + { + "epoch": 30.01, + "learning_rate": 2.2635897435897437e-06, + "loss": 0.0072, + "step": 15600 + }, + { + "epoch": 30.01, + "learning_rate": 2.250769230769231e-06, + "loss": 0.0066, + "step": 15625 + }, + { + "epoch": 30.01, + "learning_rate": 2.237948717948718e-06, + "loss": 0.0066, + "step": 15650 + }, + { + "epoch": 30.02, + "learning_rate": 2.2251282051282054e-06, + "loss": 0.0072, + "step": 15675 + }, + { + "epoch": 30.02, + "learning_rate": 2.2123076923076927e-06, + "loss": 0.0065, + "step": 15700 + }, + { + "epoch": 30.02, + "learning_rate": 2.1994871794871795e-06, + "loss": 0.006, + "step": 15725 + }, + { + "epoch": 30.02, + "learning_rate": 2.1866666666666668e-06, + "loss": 0.0059, + "step": 15750 + }, + { + "epoch": 30.02, + "learning_rate": 2.173846153846154e-06, + "loss": 0.0055, + "step": 15775 + }, + { + "epoch": 30.02, + "learning_rate": 2.1610256410256413e-06, + "loss": 0.0058, + "step": 15800 + }, + { + "epoch": 30.02, + "learning_rate": 2.1482051282051285e-06, + "loss": 0.0051, + "step": 15825 + }, + { + "epoch": 30.02, + "learning_rate": 2.1353846153846158e-06, + "loss": 0.0057, + "step": 15850 + }, + { + "epoch": 31.0, + "learning_rate": 2.1225641025641026e-06, + "loss": 0.0046, + "step": 15875 + }, + { + "epoch": 31.0, + "learning_rate": 2.10974358974359e-06, + "loss": 0.0083, + "step": 15900 + }, + { + "epoch": 31.0, + "learning_rate": 2.096923076923077e-06, + "loss": 0.0086, + "step": 15925 + }, + { + "epoch": 31.0, + "learning_rate": 2.0841025641025643e-06, + "loss": 0.0073, + "step": 15950 + }, + { + "epoch": 31.01, + "learning_rate": 2.0712820512820516e-06, + "loss": 0.0083, + "step": 15975 + }, + { + "epoch": 31.01, + "learning_rate": 2.058461538461539e-06, + "loss": 0.0076, + "step": 16000 + }, + { + "epoch": 31.01, + "eval_loss": 0.1571243405342102, + "eval_runtime": 5828.4576, + "eval_samples_per_second": 2.81, + "eval_steps_per_second": 0.176, + "eval_wer": 5.1654758502429265, + "step": 16000 + }, + { + "epoch": 31.01, + "learning_rate": 2.0456410256410256e-06, + "loss": 0.0078, + "step": 16025 + }, + { + "epoch": 31.01, + "learning_rate": 2.032820512820513e-06, + "loss": 0.0082, + "step": 16050 + }, + { + "epoch": 31.01, + "learning_rate": 2.02e-06, + "loss": 0.006, + "step": 16075 + }, + { + "epoch": 31.01, + "learning_rate": 2.0071794871794874e-06, + "loss": 0.0073, + "step": 16100 + }, + { + "epoch": 31.01, + "learning_rate": 1.9943589743589746e-06, + "loss": 0.0061, + "step": 16125 + }, + { + "epoch": 31.01, + "learning_rate": 1.981538461538462e-06, + "loss": 0.0075, + "step": 16150 + }, + { + "epoch": 31.02, + "learning_rate": 1.9687179487179487e-06, + "loss": 0.007, + "step": 16175 + }, + { + "epoch": 31.02, + "learning_rate": 1.955897435897436e-06, + "loss": 0.0054, + "step": 16200 + }, + { + "epoch": 31.02, + "learning_rate": 1.943076923076923e-06, + "loss": 0.0076, + "step": 16225 + }, + { + "epoch": 31.02, + "learning_rate": 1.9302564102564105e-06, + "loss": 0.006, + "step": 16250 + }, + { + "epoch": 31.02, + "learning_rate": 1.9174358974358977e-06, + "loss": 0.0065, + "step": 16275 + }, + { + "epoch": 31.02, + "learning_rate": 1.9046153846153847e-06, + "loss": 0.0074, + "step": 16300 + }, + { + "epoch": 31.02, + "learning_rate": 1.891794871794872e-06, + "loss": 0.0072, + "step": 16325 + }, + { + "epoch": 31.02, + "learning_rate": 1.8789743589743592e-06, + "loss": 0.0079, + "step": 16350 + }, + { + "epoch": 31.03, + "learning_rate": 1.8661538461538465e-06, + "loss": 0.0083, + "step": 16375 + }, + { + "epoch": 32.0, + "learning_rate": 1.8533333333333333e-06, + "loss": 0.0076, + "step": 16400 + }, + { + "epoch": 32.0, + "learning_rate": 1.8405128205128206e-06, + "loss": 0.0085, + "step": 16425 + }, + { + "epoch": 32.0, + "learning_rate": 1.8276923076923078e-06, + "loss": 0.0076, + "step": 16450 + }, + { + "epoch": 32.0, + "learning_rate": 1.814871794871795e-06, + "loss": 0.0044, + "step": 16475 + }, + { + "epoch": 32.01, + "learning_rate": 1.8020512820512823e-06, + "loss": 0.0074, + "step": 16500 + }, + { + "epoch": 32.01, + "learning_rate": 1.7892307692307695e-06, + "loss": 0.0059, + "step": 16525 + }, + { + "epoch": 32.01, + "learning_rate": 1.7764102564102564e-06, + "loss": 0.0051, + "step": 16550 + }, + { + "epoch": 32.01, + "learning_rate": 1.7635897435897436e-06, + "loss": 0.0061, + "step": 16575 + }, + { + "epoch": 32.01, + "learning_rate": 1.7507692307692309e-06, + "loss": 0.0067, + "step": 16600 + }, + { + "epoch": 32.01, + "learning_rate": 1.7379487179487181e-06, + "loss": 0.0069, + "step": 16625 + }, + { + "epoch": 32.01, + "learning_rate": 1.7251282051282054e-06, + "loss": 0.0047, + "step": 16650 + }, + { + "epoch": 32.01, + "learning_rate": 1.7123076923076926e-06, + "loss": 0.0092, + "step": 16675 + }, + { + "epoch": 32.02, + "learning_rate": 1.6994871794871794e-06, + "loss": 0.0054, + "step": 16700 + }, + { + "epoch": 32.02, + "learning_rate": 1.6866666666666667e-06, + "loss": 0.0082, + "step": 16725 + }, + { + "epoch": 32.02, + "learning_rate": 1.673846153846154e-06, + "loss": 0.0058, + "step": 16750 + }, + { + "epoch": 32.02, + "learning_rate": 1.6610256410256412e-06, + "loss": 0.0064, + "step": 16775 + }, + { + "epoch": 32.02, + "learning_rate": 1.6482051282051284e-06, + "loss": 0.0051, + "step": 16800 + }, + { + "epoch": 32.02, + "learning_rate": 1.6353846153846157e-06, + "loss": 0.0055, + "step": 16825 + }, + { + "epoch": 32.02, + "learning_rate": 1.6225641025641025e-06, + "loss": 0.0061, + "step": 16850 + }, + { + "epoch": 32.02, + "learning_rate": 1.6097435897435897e-06, + "loss": 0.0078, + "step": 16875 + }, + { + "epoch": 33.0, + "learning_rate": 1.596923076923077e-06, + "loss": 0.0112, + "step": 16900 + }, + { + "epoch": 33.0, + "learning_rate": 1.5841025641025642e-06, + "loss": 0.03, + "step": 16925 + }, + { + "epoch": 33.0, + "learning_rate": 1.5712820512820515e-06, + "loss": 0.0246, + "step": 16950 + }, + { + "epoch": 33.0, + "learning_rate": 1.5584615384615387e-06, + "loss": 0.0125, + "step": 16975 + }, + { + "epoch": 33.01, + "learning_rate": 1.5456410256410256e-06, + "loss": 0.0181, + "step": 17000 + }, + { + "epoch": 33.01, + "eval_loss": 0.14488154649734497, + "eval_runtime": 5845.9777, + "eval_samples_per_second": 2.802, + "eval_steps_per_second": 0.175, + "eval_wer": 5.4558445270077165, + "step": 17000 + }, + { + "epoch": 33.01, + "learning_rate": 1.5328205128205128e-06, + "loss": 0.0193, + "step": 17025 + }, + { + "epoch": 33.01, + "learning_rate": 1.52e-06, + "loss": 0.0084, + "step": 17050 + }, + { + "epoch": 33.01, + "learning_rate": 1.5071794871794873e-06, + "loss": 0.0069, + "step": 17075 + }, + { + "epoch": 33.01, + "learning_rate": 1.4943589743589746e-06, + "loss": 0.0153, + "step": 17100 + }, + { + "epoch": 33.01, + "learning_rate": 1.4815384615384618e-06, + "loss": 0.0174, + "step": 17125 + }, + { + "epoch": 33.01, + "learning_rate": 1.4687179487179486e-06, + "loss": 0.0135, + "step": 17150 + }, + { + "epoch": 33.01, + "learning_rate": 1.4558974358974359e-06, + "loss": 0.0173, + "step": 17175 + }, + { + "epoch": 33.02, + "learning_rate": 1.4430769230769231e-06, + "loss": 0.0208, + "step": 17200 + }, + { + "epoch": 33.02, + "learning_rate": 1.4302564102564104e-06, + "loss": 0.0143, + "step": 17225 + }, + { + "epoch": 33.02, + "learning_rate": 1.4174358974358976e-06, + "loss": 0.0205, + "step": 17250 + }, + { + "epoch": 33.02, + "learning_rate": 1.4046153846153849e-06, + "loss": 0.0218, + "step": 17275 + }, + { + "epoch": 33.02, + "learning_rate": 1.3917948717948717e-06, + "loss": 0.0201, + "step": 17300 + }, + { + "epoch": 33.02, + "learning_rate": 1.378974358974359e-06, + "loss": 0.0144, + "step": 17325 + }, + { + "epoch": 33.02, + "learning_rate": 1.3661538461538462e-06, + "loss": 0.016, + "step": 17350 + }, + { + "epoch": 33.02, + "learning_rate": 1.3533333333333334e-06, + "loss": 0.0214, + "step": 17375 + }, + { + "epoch": 33.03, + "learning_rate": 1.3405128205128207e-06, + "loss": 0.0194, + "step": 17400 + }, + { + "epoch": 34.0, + "learning_rate": 1.327692307692308e-06, + "loss": 0.0187, + "step": 17425 + }, + { + "epoch": 34.0, + "learning_rate": 1.3148717948717952e-06, + "loss": 0.0173, + "step": 17450 + }, + { + "epoch": 34.0, + "learning_rate": 1.302051282051282e-06, + "loss": 0.0159, + "step": 17475 + }, + { + "epoch": 34.0, + "learning_rate": 1.2892307692307693e-06, + "loss": 0.0155, + "step": 17500 + }, + { + "epoch": 34.01, + "learning_rate": 1.2764102564102565e-06, + "loss": 0.014, + "step": 17525 + }, + { + "epoch": 34.01, + "learning_rate": 1.2635897435897437e-06, + "loss": 0.0134, + "step": 17550 + }, + { + "epoch": 34.01, + "learning_rate": 1.250769230769231e-06, + "loss": 0.0131, + "step": 17575 + }, + { + "epoch": 34.01, + "learning_rate": 1.237948717948718e-06, + "loss": 0.0131, + "step": 17600 + }, + { + "epoch": 34.01, + "learning_rate": 1.2251282051282053e-06, + "loss": 0.0137, + "step": 17625 + }, + { + "epoch": 34.01, + "learning_rate": 1.2123076923076923e-06, + "loss": 0.0131, + "step": 17650 + }, + { + "epoch": 34.01, + "learning_rate": 1.1994871794871796e-06, + "loss": 0.0142, + "step": 17675 + }, + { + "epoch": 34.01, + "learning_rate": 1.1866666666666668e-06, + "loss": 0.0121, + "step": 17700 + }, + { + "epoch": 34.02, + "learning_rate": 1.1738461538461538e-06, + "loss": 0.0143, + "step": 17725 + }, + { + "epoch": 34.02, + "learning_rate": 1.161025641025641e-06, + "loss": 0.0129, + "step": 17750 + }, + { + "epoch": 34.02, + "learning_rate": 1.1482051282051283e-06, + "loss": 0.0143, + "step": 17775 + }, + { + "epoch": 34.02, + "learning_rate": 1.1353846153846154e-06, + "loss": 0.0131, + "step": 17800 + }, + { + "epoch": 34.02, + "learning_rate": 1.1225641025641026e-06, + "loss": 0.0128, + "step": 17825 + }, + { + "epoch": 34.02, + "learning_rate": 1.1097435897435899e-06, + "loss": 0.0135, + "step": 17850 + }, + { + "epoch": 34.02, + "learning_rate": 1.096923076923077e-06, + "loss": 0.0217, + "step": 17875 + }, + { + "epoch": 34.02, + "learning_rate": 1.0841025641025642e-06, + "loss": 0.027, + "step": 17900 + }, + { + "epoch": 35.0, + "learning_rate": 1.0712820512820514e-06, + "loss": 0.0311, + "step": 17925 + }, + { + "epoch": 35.0, + "learning_rate": 1.0584615384615384e-06, + "loss": 0.0639, + "step": 17950 + }, + { + "epoch": 35.0, + "learning_rate": 1.0456410256410257e-06, + "loss": 0.0593, + "step": 17975 + }, + { + "epoch": 35.0, + "learning_rate": 1.032820512820513e-06, + "loss": 0.0522, + "step": 18000 + }, + { + "epoch": 35.0, + "eval_loss": 0.13397322595119476, + "eval_runtime": 5911.5712, + "eval_samples_per_second": 2.771, + "eval_steps_per_second": 0.173, + "eval_wer": 5.838811088882538, + "step": 18000 + }, + { + "epoch": 35.01, + "learning_rate": 1.02e-06, + "loss": 0.0618, + "step": 18025 + }, + { + "epoch": 35.01, + "learning_rate": 1.0071794871794872e-06, + "loss": 0.0536, + "step": 18050 + }, + { + "epoch": 35.01, + "learning_rate": 9.943589743589745e-07, + "loss": 0.0555, + "step": 18075 + }, + { + "epoch": 35.01, + "learning_rate": 9.815384615384615e-07, + "loss": 0.0487, + "step": 18100 + }, + { + "epoch": 35.01, + "learning_rate": 9.687179487179488e-07, + "loss": 0.0569, + "step": 18125 + }, + { + "epoch": 35.01, + "learning_rate": 9.55897435897436e-07, + "loss": 0.0639, + "step": 18150 + }, + { + "epoch": 35.01, + "learning_rate": 9.430769230769233e-07, + "loss": 0.0584, + "step": 18175 + }, + { + "epoch": 35.01, + "learning_rate": 9.302564102564103e-07, + "loss": 0.0511, + "step": 18200 + }, + { + "epoch": 35.02, + "learning_rate": 9.174358974358975e-07, + "loss": 0.058, + "step": 18225 + }, + { + "epoch": 35.02, + "learning_rate": 9.046153846153848e-07, + "loss": 0.0617, + "step": 18250 + }, + { + "epoch": 35.02, + "learning_rate": 8.917948717948718e-07, + "loss": 0.0626, + "step": 18275 + }, + { + "epoch": 35.02, + "learning_rate": 8.789743589743591e-07, + "loss": 0.0596, + "step": 18300 + }, + { + "epoch": 35.02, + "learning_rate": 8.661538461538463e-07, + "loss": 0.0614, + "step": 18325 + }, + { + "epoch": 35.02, + "learning_rate": 8.533333333333334e-07, + "loss": 0.0547, + "step": 18350 + }, + { + "epoch": 35.02, + "learning_rate": 8.405128205128206e-07, + "loss": 0.0543, + "step": 18375 + }, + { + "epoch": 35.02, + "learning_rate": 8.276923076923078e-07, + "loss": 0.0475, + "step": 18400 + }, + { + "epoch": 35.03, + "learning_rate": 8.148717948717949e-07, + "loss": 0.0573, + "step": 18425 + }, + { + "epoch": 36.0, + "learning_rate": 8.020512820512821e-07, + "loss": 0.0206, + "step": 18450 + }, + { + "epoch": 36.0, + "learning_rate": 7.892307692307694e-07, + "loss": 0.0046, + "step": 18475 + }, + { + "epoch": 36.0, + "learning_rate": 7.764102564102564e-07, + "loss": 0.0036, + "step": 18500 + }, + { + "epoch": 36.0, + "learning_rate": 7.635897435897437e-07, + "loss": 0.0039, + "step": 18525 + }, + { + "epoch": 36.01, + "learning_rate": 7.507692307692309e-07, + "loss": 0.0035, + "step": 18550 + }, + { + "epoch": 36.01, + "learning_rate": 7.37948717948718e-07, + "loss": 0.0041, + "step": 18575 + }, + { + "epoch": 36.01, + "learning_rate": 7.251282051282052e-07, + "loss": 0.0033, + "step": 18600 + }, + { + "epoch": 36.01, + "learning_rate": 7.123076923076924e-07, + "loss": 0.0032, + "step": 18625 + }, + { + "epoch": 36.01, + "learning_rate": 6.994871794871795e-07, + "loss": 0.0028, + "step": 18650 + }, + { + "epoch": 36.01, + "learning_rate": 6.866666666666667e-07, + "loss": 0.0031, + "step": 18675 + }, + { + "epoch": 36.01, + "learning_rate": 6.73846153846154e-07, + "loss": 0.0036, + "step": 18700 + }, + { + "epoch": 36.01, + "learning_rate": 6.61025641025641e-07, + "loss": 0.0035, + "step": 18725 + }, + { + "epoch": 36.02, + "learning_rate": 6.482051282051283e-07, + "loss": 0.0025, + "step": 18750 + }, + { + "epoch": 36.02, + "learning_rate": 6.353846153846155e-07, + "loss": 0.0027, + "step": 18775 + }, + { + "epoch": 36.02, + "learning_rate": 6.225641025641026e-07, + "loss": 0.0034, + "step": 18800 + }, + { + "epoch": 36.02, + "learning_rate": 6.097435897435898e-07, + "loss": 0.0027, + "step": 18825 + }, + { + "epoch": 36.02, + "learning_rate": 5.969230769230769e-07, + "loss": 0.0032, + "step": 18850 + }, + { + "epoch": 36.02, + "learning_rate": 5.841025641025642e-07, + "loss": 0.0033, + "step": 18875 + }, + { + "epoch": 36.02, + "learning_rate": 5.712820512820513e-07, + "loss": 0.0028, + "step": 18900 + }, + { + "epoch": 36.02, + "learning_rate": 5.584615384615385e-07, + "loss": 0.0032, + "step": 18925 + }, + { + "epoch": 37.0, + "learning_rate": 5.456410256410257e-07, + "loss": 0.0084, + "step": 18950 + }, + { + "epoch": 37.0, + "learning_rate": 5.328205128205129e-07, + "loss": 0.0237, + "step": 18975 + }, + { + "epoch": 37.0, + "learning_rate": 5.2e-07, + "loss": 0.0356, + "step": 19000 + }, + { + "epoch": 37.0, + "eval_loss": 0.14577950537204742, + "eval_runtime": 5931.1844, + "eval_samples_per_second": 2.762, + "eval_steps_per_second": 0.173, + "eval_wer": 5.070020005715919, + "step": 19000 + }, + { + "epoch": 37.0, + "learning_rate": 5.071794871794872e-07, + "loss": 0.0248, + "step": 19025 + }, + { + "epoch": 37.01, + "learning_rate": 4.943589743589744e-07, + "loss": 0.0135, + "step": 19050 + }, + { + "epoch": 37.01, + "learning_rate": 4.815384615384615e-07, + "loss": 0.0164, + "step": 19075 + }, + { + "epoch": 37.01, + "learning_rate": 4.687179487179488e-07, + "loss": 0.01, + "step": 19100 + }, + { + "epoch": 37.01, + "learning_rate": 4.558974358974359e-07, + "loss": 0.0133, + "step": 19125 + }, + { + "epoch": 37.01, + "learning_rate": 4.4307692307692306e-07, + "loss": 0.0203, + "step": 19150 + }, + { + "epoch": 37.01, + "learning_rate": 4.302564102564103e-07, + "loss": 0.0123, + "step": 19175 + }, + { + "epoch": 37.01, + "learning_rate": 4.1743589743589745e-07, + "loss": 0.006, + "step": 19200 + }, + { + "epoch": 37.01, + "learning_rate": 4.046153846153846e-07, + "loss": 0.0049, + "step": 19225 + }, + { + "epoch": 37.02, + "learning_rate": 3.9179487179487184e-07, + "loss": 0.0043, + "step": 19250 + }, + { + "epoch": 37.02, + "learning_rate": 3.78974358974359e-07, + "loss": 0.0043, + "step": 19275 + }, + { + "epoch": 37.02, + "learning_rate": 3.661538461538461e-07, + "loss": 0.0038, + "step": 19300 + }, + { + "epoch": 37.02, + "learning_rate": 3.533333333333334e-07, + "loss": 0.0043, + "step": 19325 + }, + { + "epoch": 37.02, + "learning_rate": 3.405128205128205e-07, + "loss": 0.0029, + "step": 19350 + }, + { + "epoch": 37.02, + "learning_rate": 3.276923076923077e-07, + "loss": 0.0038, + "step": 19375 + }, + { + "epoch": 37.02, + "learning_rate": 3.148717948717949e-07, + "loss": 0.0037, + "step": 19400 + }, + { + "epoch": 37.02, + "learning_rate": 3.0205128205128205e-07, + "loss": 0.0039, + "step": 19425 + }, + { + "epoch": 37.03, + "learning_rate": 2.8923076923076924e-07, + "loss": 0.0035, + "step": 19450 + }, + { + "epoch": 38.0, + "learning_rate": 2.7641025641025644e-07, + "loss": 0.0456, + "step": 19475 + }, + { + "epoch": 38.0, + "learning_rate": 2.635897435897436e-07, + "loss": 0.0808, + "step": 19500 + }, + { + "epoch": 38.0, + "learning_rate": 2.507692307692308e-07, + "loss": 0.0534, + "step": 19525 + }, + { + "epoch": 38.0, + "learning_rate": 2.3794871794871797e-07, + "loss": 0.0468, + "step": 19550 + }, + { + "epoch": 38.01, + "learning_rate": 2.2512820512820514e-07, + "loss": 0.0458, + "step": 19575 + }, + { + "epoch": 38.01, + "learning_rate": 2.123076923076923e-07, + "loss": 0.0353, + "step": 19600 + }, + { + "epoch": 38.01, + "learning_rate": 1.994871794871795e-07, + "loss": 0.0416, + "step": 19625 + }, + { + "epoch": 38.01, + "learning_rate": 1.866666666666667e-07, + "loss": 0.0633, + "step": 19650 + }, + { + "epoch": 38.01, + "learning_rate": 1.7384615384615384e-07, + "loss": 0.0496, + "step": 19675 + }, + { + "epoch": 38.01, + "learning_rate": 1.6102564102564103e-07, + "loss": 0.0556, + "step": 19700 + }, + { + "epoch": 38.01, + "learning_rate": 1.482051282051282e-07, + "loss": 0.0993, + "step": 19725 + }, + { + "epoch": 38.01, + "learning_rate": 1.353846153846154e-07, + "loss": 0.063, + "step": 19750 + }, + { + "epoch": 38.02, + "learning_rate": 1.2256410256410257e-07, + "loss": 0.0559, + "step": 19775 + }, + { + "epoch": 38.02, + "learning_rate": 1.0974358974358975e-07, + "loss": 0.0574, + "step": 19800 + }, + { + "epoch": 38.02, + "learning_rate": 9.692307692307693e-08, + "loss": 0.0447, + "step": 19825 + }, + { + "epoch": 38.02, + "learning_rate": 8.410256410256411e-08, + "loss": 0.0289, + "step": 19850 + }, + { + "epoch": 38.02, + "learning_rate": 7.128205128205129e-08, + "loss": 0.0562, + "step": 19875 + }, + { + "epoch": 38.02, + "learning_rate": 5.846153846153847e-08, + "loss": 0.0412, + "step": 19900 + }, + { + "epoch": 38.02, + "learning_rate": 4.5641025641025644e-08, + "loss": 0.0404, + "step": 19925 + }, + { + "epoch": 38.02, + "learning_rate": 3.282051282051282e-08, + "loss": 0.0581, + "step": 19950 + }, + { + "epoch": 39.0, + "learning_rate": 2e-08, + "loss": 0.0417, + "step": 19975 + }, + { + "epoch": 39.0, + "learning_rate": 7.1794871794871805e-09, + "loss": 0.0132, + "step": 20000 + }, + { + "epoch": 39.0, + "eval_loss": 0.13095258176326752, + "eval_runtime": 5907.8092, + "eval_samples_per_second": 2.773, + "eval_steps_per_second": 0.173, + "eval_wer": 5.194055444412689, + "step": 20000 + }, + { + "epoch": 39.0, + "step": 20000, + "total_flos": 2.7169964772655104e+21, + "train_loss": 0.0637142912024632, + "train_runtime": 459101.4744, + "train_samples_per_second": 2.788, + "train_steps_per_second": 0.044 + } + ], + "logging_steps": 25, + "max_steps": 20000, + "num_train_epochs": 9223372036854775807, + "save_steps": 1000, + "total_flos": 2.7169964772655104e+21, + "trial_name": null, + "trial_params": null +}