diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,5008 @@ +{ + "best_metric": 4.671620462989425, + "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": 4.4e-07, + "loss": 0.8428, + "step": 25 + }, + { + "epoch": 0.0, + "learning_rate": 9.400000000000001e-07, + "loss": 0.5317, + "step": 50 + }, + { + "epoch": 0.0, + "learning_rate": 1.44e-06, + "loss": 0.2909, + "step": 75 + }, + { + "epoch": 0.01, + "learning_rate": 1.94e-06, + "loss": 0.229, + "step": 100 + }, + { + "epoch": 0.01, + "learning_rate": 2.4400000000000004e-06, + "loss": 0.1949, + "step": 125 + }, + { + "epoch": 0.01, + "learning_rate": 2.9400000000000002e-06, + "loss": 0.1658, + "step": 150 + }, + { + "epoch": 0.01, + "learning_rate": 3.44e-06, + "loss": 0.1663, + "step": 175 + }, + { + "epoch": 0.01, + "learning_rate": 3.94e-06, + "loss": 0.1654, + "step": 200 + }, + { + "epoch": 0.01, + "learning_rate": 4.440000000000001e-06, + "loss": 0.1557, + "step": 225 + }, + { + "epoch": 0.01, + "learning_rate": 4.94e-06, + "loss": 0.1573, + "step": 250 + }, + { + "epoch": 0.01, + "learning_rate": 5.4400000000000004e-06, + "loss": 0.1488, + "step": 275 + }, + { + "epoch": 0.01, + "learning_rate": 5.94e-06, + "loss": 0.1451, + "step": 300 + }, + { + "epoch": 0.02, + "learning_rate": 6.440000000000001e-06, + "loss": 0.1386, + "step": 325 + }, + { + "epoch": 0.02, + "learning_rate": 6.9400000000000005e-06, + "loss": 0.1418, + "step": 350 + }, + { + "epoch": 0.02, + "learning_rate": 7.440000000000001e-06, + "loss": 0.1389, + "step": 375 + }, + { + "epoch": 0.02, + "learning_rate": 7.94e-06, + "loss": 0.1342, + "step": 400 + }, + { + "epoch": 0.02, + "learning_rate": 8.42e-06, + "loss": 0.1425, + "step": 425 + }, + { + "epoch": 0.02, + "learning_rate": 8.920000000000001e-06, + "loss": 0.1325, + "step": 450 + }, + { + "epoch": 0.02, + "learning_rate": 9.42e-06, + "loss": 0.1304, + "step": 475 + }, + { + "epoch": 0.03, + "learning_rate": 9.920000000000002e-06, + "loss": 0.1312, + "step": 500 + }, + { + "epoch": 1.0, + "learning_rate": 9.98923076923077e-06, + "loss": 0.143, + "step": 525 + }, + { + "epoch": 1.0, + "learning_rate": 9.976410256410257e-06, + "loss": 0.156, + "step": 550 + }, + { + "epoch": 1.0, + "learning_rate": 9.963589743589744e-06, + "loss": 0.16, + "step": 575 + }, + { + "epoch": 1.0, + "learning_rate": 9.950769230769232e-06, + "loss": 0.1545, + "step": 600 + }, + { + "epoch": 1.01, + "learning_rate": 9.937948717948719e-06, + "loss": 0.155, + "step": 625 + }, + { + "epoch": 1.01, + "learning_rate": 9.925128205128206e-06, + "loss": 0.1329, + "step": 650 + }, + { + "epoch": 1.01, + "learning_rate": 9.912307692307693e-06, + "loss": 0.1168, + "step": 675 + }, + { + "epoch": 1.01, + "learning_rate": 9.899487179487181e-06, + "loss": 0.1219, + "step": 700 + }, + { + "epoch": 1.01, + "learning_rate": 9.886666666666668e-06, + "loss": 0.1281, + "step": 725 + }, + { + "epoch": 1.01, + "learning_rate": 9.873846153846155e-06, + "loss": 0.1222, + "step": 750 + }, + { + "epoch": 1.01, + "learning_rate": 9.861025641025642e-06, + "loss": 0.1305, + "step": 775 + }, + { + "epoch": 1.01, + "learning_rate": 9.84820512820513e-06, + "loss": 0.1328, + "step": 800 + }, + { + "epoch": 1.02, + "learning_rate": 9.835384615384617e-06, + "loss": 0.1255, + "step": 825 + }, + { + "epoch": 1.02, + "learning_rate": 9.822564102564104e-06, + "loss": 0.1198, + "step": 850 + }, + { + "epoch": 1.02, + "learning_rate": 9.80974358974359e-06, + "loss": 0.1459, + "step": 875 + }, + { + "epoch": 1.02, + "learning_rate": 9.796923076923077e-06, + "loss": 0.115, + "step": 900 + }, + { + "epoch": 1.02, + "learning_rate": 9.784102564102564e-06, + "loss": 0.1083, + "step": 925 + }, + { + "epoch": 1.02, + "learning_rate": 9.771282051282051e-06, + "loss": 0.108, + "step": 950 + }, + { + "epoch": 1.02, + "learning_rate": 9.75846153846154e-06, + "loss": 0.1051, + "step": 975 + }, + { + "epoch": 1.02, + "learning_rate": 9.745641025641026e-06, + "loss": 0.1072, + "step": 1000 + }, + { + "epoch": 1.02, + "eval_loss": 0.16367217898368835, + "eval_runtime": 5885.0411, + "eval_samples_per_second": 2.783, + "eval_steps_per_second": 0.174, + "eval_wer": 7.032866533295228, + "step": 1000 + }, + { + "epoch": 2.0, + "learning_rate": 9.732820512820513e-06, + "loss": 0.1019, + "step": 1025 + }, + { + "epoch": 2.0, + "learning_rate": 9.72e-06, + "loss": 0.0958, + "step": 1050 + }, + { + "epoch": 2.0, + "learning_rate": 9.707179487179487e-06, + "loss": 0.0899, + "step": 1075 + }, + { + "epoch": 2.0, + "learning_rate": 9.694358974358975e-06, + "loss": 0.1002, + "step": 1100 + }, + { + "epoch": 2.01, + "learning_rate": 9.681538461538462e-06, + "loss": 0.1065, + "step": 1125 + }, + { + "epoch": 2.01, + "learning_rate": 9.668717948717949e-06, + "loss": 0.0946, + "step": 1150 + }, + { + "epoch": 2.01, + "learning_rate": 9.655897435897436e-06, + "loss": 0.0952, + "step": 1175 + }, + { + "epoch": 2.01, + "learning_rate": 9.643076923076924e-06, + "loss": 0.1057, + "step": 1200 + }, + { + "epoch": 2.01, + "learning_rate": 9.630256410256411e-06, + "loss": 0.1123, + "step": 1225 + }, + { + "epoch": 2.01, + "learning_rate": 9.617435897435898e-06, + "loss": 0.1048, + "step": 1250 + }, + { + "epoch": 2.01, + "learning_rate": 9.604615384615385e-06, + "loss": 0.0945, + "step": 1275 + }, + { + "epoch": 2.01, + "learning_rate": 9.591794871794873e-06, + "loss": 0.1004, + "step": 1300 + }, + { + "epoch": 2.02, + "learning_rate": 9.57897435897436e-06, + "loss": 0.1019, + "step": 1325 + }, + { + "epoch": 2.02, + "learning_rate": 9.566153846153847e-06, + "loss": 0.0929, + "step": 1350 + }, + { + "epoch": 2.02, + "learning_rate": 9.553333333333334e-06, + "loss": 0.0826, + "step": 1375 + }, + { + "epoch": 2.02, + "learning_rate": 9.540512820512822e-06, + "loss": 0.0632, + "step": 1400 + }, + { + "epoch": 2.02, + "learning_rate": 9.52769230769231e-06, + "loss": 0.0609, + "step": 1425 + }, + { + "epoch": 2.02, + "learning_rate": 9.514871794871796e-06, + "loss": 0.0592, + "step": 1450 + }, + { + "epoch": 2.02, + "learning_rate": 9.502051282051283e-06, + "loss": 0.0589, + "step": 1475 + }, + { + "epoch": 2.02, + "learning_rate": 9.48923076923077e-06, + "loss": 0.0498, + "step": 1500 + }, + { + "epoch": 2.03, + "learning_rate": 9.476410256410257e-06, + "loss": 0.056, + "step": 1525 + }, + { + "epoch": 3.0, + "learning_rate": 9.463589743589743e-06, + "loss": 0.0527, + "step": 1550 + }, + { + "epoch": 3.0, + "learning_rate": 9.450769230769232e-06, + "loss": 0.0482, + "step": 1575 + }, + { + "epoch": 3.0, + "learning_rate": 9.437948717948719e-06, + "loss": 0.0465, + "step": 1600 + }, + { + "epoch": 3.0, + "learning_rate": 9.425128205128206e-06, + "loss": 0.0485, + "step": 1625 + }, + { + "epoch": 3.01, + "learning_rate": 9.412307692307692e-06, + "loss": 0.0493, + "step": 1650 + }, + { + "epoch": 3.01, + "learning_rate": 9.39948717948718e-06, + "loss": 0.0441, + "step": 1675 + }, + { + "epoch": 3.01, + "learning_rate": 9.386666666666668e-06, + "loss": 0.049, + "step": 1700 + }, + { + "epoch": 3.01, + "learning_rate": 9.373846153846155e-06, + "loss": 0.0529, + "step": 1725 + }, + { + "epoch": 3.01, + "learning_rate": 9.361025641025641e-06, + "loss": 0.0489, + "step": 1750 + }, + { + "epoch": 3.01, + "learning_rate": 9.348205128205128e-06, + "loss": 0.0487, + "step": 1775 + }, + { + "epoch": 3.01, + "learning_rate": 9.335384615384617e-06, + "loss": 0.0464, + "step": 1800 + }, + { + "epoch": 3.01, + "learning_rate": 9.322564102564104e-06, + "loss": 0.0535, + "step": 1825 + }, + { + "epoch": 3.02, + "learning_rate": 9.30974358974359e-06, + "loss": 0.0455, + "step": 1850 + }, + { + "epoch": 3.02, + "learning_rate": 9.296923076923077e-06, + "loss": 0.0437, + "step": 1875 + }, + { + "epoch": 3.02, + "learning_rate": 9.284102564102566e-06, + "loss": 0.0291, + "step": 1900 + }, + { + "epoch": 3.02, + "learning_rate": 9.271282051282053e-06, + "loss": 0.0259, + "step": 1925 + }, + { + "epoch": 3.02, + "learning_rate": 9.25846153846154e-06, + "loss": 0.0265, + "step": 1950 + }, + { + "epoch": 3.02, + "learning_rate": 9.245641025641026e-06, + "loss": 0.0261, + "step": 1975 + }, + { + "epoch": 3.02, + "learning_rate": 9.232820512820515e-06, + "loss": 0.0239, + "step": 2000 + }, + { + "epoch": 3.02, + "eval_loss": 0.1783781796693802, + "eval_runtime": 5894.2183, + "eval_samples_per_second": 2.779, + "eval_steps_per_second": 0.174, + "eval_wer": 7.027722206344669, + "step": 2000 + }, + { + "epoch": 3.02, + "learning_rate": 9.220000000000002e-06, + "loss": 0.0222, + "step": 2025 + }, + { + "epoch": 4.0, + "learning_rate": 9.207179487179488e-06, + "loss": 0.0301, + "step": 2050 + }, + { + "epoch": 4.0, + "learning_rate": 9.194358974358975e-06, + "loss": 0.0712, + "step": 2075 + }, + { + "epoch": 4.0, + "learning_rate": 9.181538461538464e-06, + "loss": 0.0902, + "step": 2100 + }, + { + "epoch": 4.0, + "learning_rate": 9.168717948717949e-06, + "loss": 0.0748, + "step": 2125 + }, + { + "epoch": 4.01, + "learning_rate": 9.155897435897436e-06, + "loss": 0.0688, + "step": 2150 + }, + { + "epoch": 4.01, + "learning_rate": 9.143076923076924e-06, + "loss": 0.0626, + "step": 2175 + }, + { + "epoch": 4.01, + "learning_rate": 9.130256410256411e-06, + "loss": 0.0574, + "step": 2200 + }, + { + "epoch": 4.01, + "learning_rate": 9.117435897435898e-06, + "loss": 0.045, + "step": 2225 + }, + { + "epoch": 4.01, + "learning_rate": 9.104615384615385e-06, + "loss": 0.0657, + "step": 2250 + }, + { + "epoch": 4.01, + "learning_rate": 9.091794871794873e-06, + "loss": 0.0792, + "step": 2275 + }, + { + "epoch": 4.01, + "learning_rate": 9.07897435897436e-06, + "loss": 0.085, + "step": 2300 + }, + { + "epoch": 4.01, + "learning_rate": 9.066153846153847e-06, + "loss": 0.0612, + "step": 2325 + }, + { + "epoch": 4.02, + "learning_rate": 9.053333333333334e-06, + "loss": 0.0618, + "step": 2350 + }, + { + "epoch": 4.02, + "learning_rate": 9.04051282051282e-06, + "loss": 0.0693, + "step": 2375 + }, + { + "epoch": 4.02, + "learning_rate": 9.027692307692309e-06, + "loss": 0.0714, + "step": 2400 + }, + { + "epoch": 4.02, + "learning_rate": 9.014871794871796e-06, + "loss": 0.0704, + "step": 2425 + }, + { + "epoch": 4.02, + "learning_rate": 9.002051282051283e-06, + "loss": 0.0661, + "step": 2450 + }, + { + "epoch": 4.02, + "learning_rate": 8.98923076923077e-06, + "loss": 0.0705, + "step": 2475 + }, + { + "epoch": 4.02, + "learning_rate": 8.976410256410258e-06, + "loss": 0.0633, + "step": 2500 + }, + { + "epoch": 4.02, + "learning_rate": 8.963589743589745e-06, + "loss": 0.1, + "step": 2525 + }, + { + "epoch": 4.03, + "learning_rate": 8.950769230769232e-06, + "loss": 0.0937, + "step": 2550 + }, + { + "epoch": 5.0, + "learning_rate": 8.937948717948718e-06, + "loss": 0.0816, + "step": 2575 + }, + { + "epoch": 5.0, + "learning_rate": 8.925128205128207e-06, + "loss": 0.0865, + "step": 2600 + }, + { + "epoch": 5.0, + "learning_rate": 8.912307692307694e-06, + "loss": 0.0786, + "step": 2625 + }, + { + "epoch": 5.0, + "learning_rate": 8.89948717948718e-06, + "loss": 0.0823, + "step": 2650 + }, + { + "epoch": 5.01, + "learning_rate": 8.886666666666667e-06, + "loss": 0.0693, + "step": 2675 + }, + { + "epoch": 5.01, + "learning_rate": 8.873846153846156e-06, + "loss": 0.0678, + "step": 2700 + }, + { + "epoch": 5.01, + "learning_rate": 8.861025641025641e-06, + "loss": 0.0724, + "step": 2725 + }, + { + "epoch": 5.01, + "learning_rate": 8.848205128205128e-06, + "loss": 0.07, + "step": 2750 + }, + { + "epoch": 5.01, + "learning_rate": 8.835384615384616e-06, + "loss": 0.0583, + "step": 2775 + }, + { + "epoch": 5.01, + "learning_rate": 8.822564102564103e-06, + "loss": 0.0358, + "step": 2800 + }, + { + "epoch": 5.01, + "learning_rate": 8.80974358974359e-06, + "loss": 0.0303, + "step": 2825 + }, + { + "epoch": 5.01, + "learning_rate": 8.796923076923077e-06, + "loss": 0.0315, + "step": 2850 + }, + { + "epoch": 5.02, + "learning_rate": 8.784102564102565e-06, + "loss": 0.0369, + "step": 2875 + }, + { + "epoch": 5.02, + "learning_rate": 8.771282051282052e-06, + "loss": 0.0332, + "step": 2900 + }, + { + "epoch": 5.02, + "learning_rate": 8.758461538461539e-06, + "loss": 0.0282, + "step": 2925 + }, + { + "epoch": 5.02, + "learning_rate": 8.745641025641026e-06, + "loss": 0.0336, + "step": 2950 + }, + { + "epoch": 5.02, + "learning_rate": 8.732820512820513e-06, + "loss": 0.0417, + "step": 2975 + }, + { + "epoch": 5.02, + "learning_rate": 8.720000000000001e-06, + "loss": 0.0507, + "step": 3000 + }, + { + "epoch": 5.02, + "eval_loss": 0.17540615797042847, + "eval_runtime": 5878.7582, + "eval_samples_per_second": 2.786, + "eval_steps_per_second": 0.174, + "eval_wer": 6.577307802229209, + "step": 3000 + }, + { + "epoch": 5.02, + "learning_rate": 8.707179487179488e-06, + "loss": 0.0606, + "step": 3025 + }, + { + "epoch": 5.02, + "learning_rate": 8.694358974358975e-06, + "loss": 0.0487, + "step": 3050 + }, + { + "epoch": 6.0, + "learning_rate": 8.681538461538462e-06, + "loss": 0.0567, + "step": 3075 + }, + { + "epoch": 6.0, + "learning_rate": 8.66871794871795e-06, + "loss": 0.0717, + "step": 3100 + }, + { + "epoch": 6.0, + "learning_rate": 8.655897435897437e-06, + "loss": 0.0773, + "step": 3125 + }, + { + "epoch": 6.0, + "learning_rate": 8.643076923076924e-06, + "loss": 0.0758, + "step": 3150 + }, + { + "epoch": 6.01, + "learning_rate": 8.63025641025641e-06, + "loss": 0.0837, + "step": 3175 + }, + { + "epoch": 6.01, + "learning_rate": 8.6174358974359e-06, + "loss": 0.0751, + "step": 3200 + }, + { + "epoch": 6.01, + "learning_rate": 8.604615384615386e-06, + "loss": 0.0716, + "step": 3225 + }, + { + "epoch": 6.01, + "learning_rate": 8.591794871794873e-06, + "loss": 0.0838, + "step": 3250 + }, + { + "epoch": 6.01, + "learning_rate": 8.57897435897436e-06, + "loss": 0.0684, + "step": 3275 + }, + { + "epoch": 6.01, + "learning_rate": 8.566153846153848e-06, + "loss": 0.0743, + "step": 3300 + }, + { + "epoch": 6.01, + "learning_rate": 8.553333333333333e-06, + "loss": 0.066, + "step": 3325 + }, + { + "epoch": 6.01, + "learning_rate": 8.54051282051282e-06, + "loss": 0.0828, + "step": 3350 + }, + { + "epoch": 6.02, + "learning_rate": 8.527692307692309e-06, + "loss": 0.0714, + "step": 3375 + }, + { + "epoch": 6.02, + "learning_rate": 8.514871794871795e-06, + "loss": 0.0651, + "step": 3400 + }, + { + "epoch": 6.02, + "learning_rate": 8.502051282051282e-06, + "loss": 0.0761, + "step": 3425 + }, + { + "epoch": 6.02, + "learning_rate": 8.489230769230769e-06, + "loss": 0.0637, + "step": 3450 + }, + { + "epoch": 6.02, + "learning_rate": 8.476410256410258e-06, + "loss": 0.0734, + "step": 3475 + }, + { + "epoch": 6.02, + "learning_rate": 8.463589743589744e-06, + "loss": 0.0744, + "step": 3500 + }, + { + "epoch": 6.02, + "learning_rate": 8.450769230769231e-06, + "loss": 0.0745, + "step": 3525 + }, + { + "epoch": 6.02, + "learning_rate": 8.437948717948718e-06, + "loss": 0.0734, + "step": 3550 + }, + { + "epoch": 6.03, + "learning_rate": 8.425128205128205e-06, + "loss": 0.0818, + "step": 3575 + }, + { + "epoch": 7.0, + "learning_rate": 8.412307692307693e-06, + "loss": 0.0445, + "step": 3600 + }, + { + "epoch": 7.0, + "learning_rate": 8.39948717948718e-06, + "loss": 0.0314, + "step": 3625 + }, + { + "epoch": 7.0, + "learning_rate": 8.386666666666667e-06, + "loss": 0.0219, + "step": 3650 + }, + { + "epoch": 7.0, + "learning_rate": 8.373846153846154e-06, + "loss": 0.0388, + "step": 3675 + }, + { + "epoch": 7.01, + "learning_rate": 8.361025641025642e-06, + "loss": 0.0468, + "step": 3700 + }, + { + "epoch": 7.01, + "learning_rate": 8.34820512820513e-06, + "loss": 0.0471, + "step": 3725 + }, + { + "epoch": 7.01, + "learning_rate": 8.335384615384616e-06, + "loss": 0.0432, + "step": 3750 + }, + { + "epoch": 7.01, + "learning_rate": 8.322564102564103e-06, + "loss": 0.0449, + "step": 3775 + }, + { + "epoch": 7.01, + "learning_rate": 8.309743589743591e-06, + "loss": 0.0427, + "step": 3800 + }, + { + "epoch": 7.01, + "learning_rate": 8.296923076923078e-06, + "loss": 0.0435, + "step": 3825 + }, + { + "epoch": 7.01, + "learning_rate": 8.284102564102565e-06, + "loss": 0.0451, + "step": 3850 + }, + { + "epoch": 7.01, + "learning_rate": 8.271282051282052e-06, + "loss": 0.0523, + "step": 3875 + }, + { + "epoch": 7.02, + "learning_rate": 8.25846153846154e-06, + "loss": 0.0454, + "step": 3900 + }, + { + "epoch": 7.02, + "learning_rate": 8.245641025641027e-06, + "loss": 0.0494, + "step": 3925 + }, + { + "epoch": 7.02, + "learning_rate": 8.232820512820512e-06, + "loss": 0.0589, + "step": 3950 + }, + { + "epoch": 7.02, + "learning_rate": 8.220000000000001e-06, + "loss": 0.0524, + "step": 3975 + }, + { + "epoch": 7.02, + "learning_rate": 8.207179487179488e-06, + "loss": 0.0571, + "step": 4000 + }, + { + "epoch": 7.02, + "eval_loss": 0.16203409433364868, + "eval_runtime": 6048.831, + "eval_samples_per_second": 2.708, + "eval_steps_per_second": 0.169, + "eval_wer": 6.5047156330380105, + "step": 4000 + }, + { + "epoch": 7.02, + "learning_rate": 8.194358974358975e-06, + "loss": 0.0639, + "step": 4025 + }, + { + "epoch": 7.02, + "learning_rate": 8.181538461538461e-06, + "loss": 0.0804, + "step": 4050 + }, + { + "epoch": 7.02, + "learning_rate": 8.16871794871795e-06, + "loss": 0.08, + "step": 4075 + }, + { + "epoch": 8.0, + "learning_rate": 8.155897435897437e-06, + "loss": 0.0647, + "step": 4100 + }, + { + "epoch": 8.0, + "learning_rate": 8.143076923076924e-06, + "loss": 0.0503, + "step": 4125 + }, + { + "epoch": 8.0, + "learning_rate": 8.13025641025641e-06, + "loss": 0.0508, + "step": 4150 + }, + { + "epoch": 8.0, + "learning_rate": 8.117435897435897e-06, + "loss": 0.0493, + "step": 4175 + }, + { + "epoch": 8.01, + "learning_rate": 8.104615384615386e-06, + "loss": 0.0461, + "step": 4200 + }, + { + "epoch": 8.01, + "learning_rate": 8.091794871794873e-06, + "loss": 0.0417, + "step": 4225 + }, + { + "epoch": 8.01, + "learning_rate": 8.07897435897436e-06, + "loss": 0.0412, + "step": 4250 + }, + { + "epoch": 8.01, + "learning_rate": 8.066153846153846e-06, + "loss": 0.0401, + "step": 4275 + }, + { + "epoch": 8.01, + "learning_rate": 8.053333333333335e-06, + "loss": 0.0406, + "step": 4300 + }, + { + "epoch": 8.01, + "learning_rate": 8.040512820512822e-06, + "loss": 0.0414, + "step": 4325 + }, + { + "epoch": 8.01, + "learning_rate": 8.027692307692308e-06, + "loss": 0.0423, + "step": 4350 + }, + { + "epoch": 8.01, + "learning_rate": 8.014871794871795e-06, + "loss": 0.039, + "step": 4375 + }, + { + "epoch": 8.02, + "learning_rate": 8.002051282051284e-06, + "loss": 0.0378, + "step": 4400 + }, + { + "epoch": 8.02, + "learning_rate": 7.98923076923077e-06, + "loss": 0.0401, + "step": 4425 + }, + { + "epoch": 8.02, + "learning_rate": 7.976410256410257e-06, + "loss": 0.0395, + "step": 4450 + }, + { + "epoch": 8.02, + "learning_rate": 7.963589743589744e-06, + "loss": 0.0377, + "step": 4475 + }, + { + "epoch": 8.02, + "learning_rate": 7.950769230769233e-06, + "loss": 0.0384, + "step": 4500 + }, + { + "epoch": 8.02, + "learning_rate": 7.93794871794872e-06, + "loss": 0.0367, + "step": 4525 + }, + { + "epoch": 8.02, + "learning_rate": 7.925128205128205e-06, + "loss": 0.0437, + "step": 4550 + }, + { + "epoch": 8.02, + "learning_rate": 7.912307692307693e-06, + "loss": 0.0694, + "step": 4575 + }, + { + "epoch": 8.03, + "learning_rate": 7.89948717948718e-06, + "loss": 0.0667, + "step": 4600 + }, + { + "epoch": 9.0, + "learning_rate": 7.886666666666667e-06, + "loss": 0.0237, + "step": 4625 + }, + { + "epoch": 9.0, + "learning_rate": 7.873846153846154e-06, + "loss": 0.0143, + "step": 4650 + }, + { + "epoch": 9.0, + "learning_rate": 7.861025641025642e-06, + "loss": 0.0117, + "step": 4675 + }, + { + "epoch": 9.0, + "learning_rate": 7.848205128205129e-06, + "loss": 0.018, + "step": 4700 + }, + { + "epoch": 9.01, + "learning_rate": 7.835384615384616e-06, + "loss": 0.0192, + "step": 4725 + }, + { + "epoch": 9.01, + "learning_rate": 7.822564102564103e-06, + "loss": 0.0193, + "step": 4750 + }, + { + "epoch": 9.01, + "learning_rate": 7.81025641025641e-06, + "loss": 0.0191, + "step": 4775 + }, + { + "epoch": 9.01, + "learning_rate": 7.797435897435897e-06, + "loss": 0.0185, + "step": 4800 + }, + { + "epoch": 9.01, + "learning_rate": 7.784615384615385e-06, + "loss": 0.0203, + "step": 4825 + }, + { + "epoch": 9.01, + "learning_rate": 7.771794871794872e-06, + "loss": 0.0203, + "step": 4850 + }, + { + "epoch": 9.01, + "learning_rate": 7.758974358974359e-06, + "loss": 0.019, + "step": 4875 + }, + { + "epoch": 9.01, + "learning_rate": 7.746153846153846e-06, + "loss": 0.022, + "step": 4900 + }, + { + "epoch": 9.02, + "learning_rate": 7.733333333333334e-06, + "loss": 0.0186, + "step": 4925 + }, + { + "epoch": 9.02, + "learning_rate": 7.720512820512821e-06, + "loss": 0.0195, + "step": 4950 + }, + { + "epoch": 9.02, + "learning_rate": 7.707692307692308e-06, + "loss": 0.0219, + "step": 4975 + }, + { + "epoch": 9.02, + "learning_rate": 7.694871794871795e-06, + "loss": 0.0193, + "step": 5000 + }, + { + "epoch": 9.02, + "eval_loss": 0.18205508589744568, + "eval_runtime": 5871.3706, + "eval_samples_per_second": 2.79, + "eval_steps_per_second": 0.174, + "eval_wer": 6.488711060302944, + "step": 5000 + }, + { + "epoch": 9.02, + "learning_rate": 7.682051282051283e-06, + "loss": 0.0237, + "step": 5025 + }, + { + "epoch": 9.02, + "learning_rate": 7.66923076923077e-06, + "loss": 0.0248, + "step": 5050 + }, + { + "epoch": 9.02, + "learning_rate": 7.656410256410257e-06, + "loss": 0.0372, + "step": 5075 + }, + { + "epoch": 9.02, + "learning_rate": 7.643589743589744e-06, + "loss": 0.0309, + "step": 5100 + }, + { + "epoch": 10.0, + "learning_rate": 7.630769230769232e-06, + "loss": 0.0297, + "step": 5125 + }, + { + "epoch": 10.0, + "learning_rate": 7.617948717948719e-06, + "loss": 0.0462, + "step": 5150 + }, + { + "epoch": 10.0, + "learning_rate": 7.605128205128206e-06, + "loss": 0.0404, + "step": 5175 + }, + { + "epoch": 10.0, + "learning_rate": 7.592307692307693e-06, + "loss": 0.0422, + "step": 5200 + }, + { + "epoch": 10.01, + "learning_rate": 7.5794871794871805e-06, + "loss": 0.0431, + "step": 5225 + }, + { + "epoch": 10.01, + "learning_rate": 7.566666666666667e-06, + "loss": 0.0391, + "step": 5250 + }, + { + "epoch": 10.01, + "learning_rate": 7.553846153846155e-06, + "loss": 0.0403, + "step": 5275 + }, + { + "epoch": 10.01, + "learning_rate": 7.541025641025642e-06, + "loss": 0.0382, + "step": 5300 + }, + { + "epoch": 10.01, + "learning_rate": 7.528205128205129e-06, + "loss": 0.0353, + "step": 5325 + }, + { + "epoch": 10.01, + "learning_rate": 7.5153846153846155e-06, + "loss": 0.0392, + "step": 5350 + }, + { + "epoch": 10.01, + "learning_rate": 7.503076923076924e-06, + "loss": 0.0387, + "step": 5375 + }, + { + "epoch": 10.01, + "learning_rate": 7.4902564102564114e-06, + "loss": 0.0441, + "step": 5400 + }, + { + "epoch": 10.02, + "learning_rate": 7.477435897435898e-06, + "loss": 0.0436, + "step": 5425 + }, + { + "epoch": 10.02, + "learning_rate": 7.464615384615386e-06, + "loss": 0.0387, + "step": 5450 + }, + { + "epoch": 10.02, + "learning_rate": 7.451794871794873e-06, + "loss": 0.0478, + "step": 5475 + }, + { + "epoch": 10.02, + "learning_rate": 7.438974358974359e-06, + "loss": 0.0476, + "step": 5500 + }, + { + "epoch": 10.02, + "learning_rate": 7.426153846153846e-06, + "loss": 0.0428, + "step": 5525 + }, + { + "epoch": 10.02, + "learning_rate": 7.413333333333333e-06, + "loss": 0.0433, + "step": 5550 + }, + { + "epoch": 10.02, + "learning_rate": 7.400512820512821e-06, + "loss": 0.0439, + "step": 5575 + }, + { + "epoch": 10.02, + "learning_rate": 7.387692307692308e-06, + "loss": 0.057, + "step": 5600 + }, + { + "epoch": 10.03, + "learning_rate": 7.374871794871795e-06, + "loss": 0.0682, + "step": 5625 + }, + { + "epoch": 11.0, + "learning_rate": 7.362564102564104e-06, + "loss": 0.0614, + "step": 5650 + }, + { + "epoch": 11.0, + "learning_rate": 7.34974358974359e-06, + "loss": 0.0712, + "step": 5675 + }, + { + "epoch": 11.0, + "learning_rate": 7.336923076923077e-06, + "loss": 0.0649, + "step": 5700 + }, + { + "epoch": 11.0, + "learning_rate": 7.324102564102564e-06, + "loss": 0.0368, + "step": 5725 + }, + { + "epoch": 11.01, + "learning_rate": 7.311282051282052e-06, + "loss": 0.0642, + "step": 5750 + }, + { + "epoch": 11.01, + "learning_rate": 7.298461538461539e-06, + "loss": 0.0603, + "step": 5775 + }, + { + "epoch": 11.01, + "learning_rate": 7.285641025641026e-06, + "loss": 0.0507, + "step": 5800 + }, + { + "epoch": 11.01, + "learning_rate": 7.272820512820513e-06, + "loss": 0.0489, + "step": 5825 + }, + { + "epoch": 11.01, + "learning_rate": 7.260000000000001e-06, + "loss": 0.0637, + "step": 5850 + }, + { + "epoch": 11.01, + "learning_rate": 7.247179487179488e-06, + "loss": 0.0532, + "step": 5875 + }, + { + "epoch": 11.01, + "learning_rate": 7.2343589743589744e-06, + "loss": 0.053, + "step": 5900 + }, + { + "epoch": 11.01, + "learning_rate": 7.221538461538462e-06, + "loss": 0.0775, + "step": 5925 + }, + { + "epoch": 11.02, + "learning_rate": 7.208717948717949e-06, + "loss": 0.0589, + "step": 5950 + }, + { + "epoch": 11.02, + "learning_rate": 7.195897435897437e-06, + "loss": 0.0617, + "step": 5975 + }, + { + "epoch": 11.02, + "learning_rate": 7.1830769230769234e-06, + "loss": 0.0625, + "step": 6000 + }, + { + "epoch": 11.02, + "eval_loss": 0.14429950714111328, + "eval_runtime": 5943.2838, + "eval_samples_per_second": 2.756, + "eval_steps_per_second": 0.172, + "eval_wer": 6.758502429265505, + "step": 6000 + }, + { + "epoch": 11.02, + "learning_rate": 7.170256410256411e-06, + "loss": 0.062, + "step": 6025 + }, + { + "epoch": 11.02, + "learning_rate": 7.157435897435898e-06, + "loss": 0.0548, + "step": 6050 + }, + { + "epoch": 11.02, + "learning_rate": 7.144615384615386e-06, + "loss": 0.0492, + "step": 6075 + }, + { + "epoch": 11.02, + "learning_rate": 7.1317948717948724e-06, + "loss": 0.0529, + "step": 6100 + }, + { + "epoch": 11.02, + "learning_rate": 7.11897435897436e-06, + "loss": 0.0624, + "step": 6125 + }, + { + "epoch": 12.0, + "learning_rate": 7.106153846153847e-06, + "loss": 0.0752, + "step": 6150 + }, + { + "epoch": 12.0, + "learning_rate": 7.093333333333335e-06, + "loss": 0.0786, + "step": 6175 + }, + { + "epoch": 12.0, + "learning_rate": 7.080512820512821e-06, + "loss": 0.079, + "step": 6200 + }, + { + "epoch": 12.0, + "learning_rate": 7.067692307692309e-06, + "loss": 0.0495, + "step": 6225 + }, + { + "epoch": 12.01, + "learning_rate": 7.054871794871796e-06, + "loss": 0.0411, + "step": 6250 + }, + { + "epoch": 12.01, + "learning_rate": 7.042051282051282e-06, + "loss": 0.0394, + "step": 6275 + }, + { + "epoch": 12.01, + "learning_rate": 7.0292307692307696e-06, + "loss": 0.0489, + "step": 6300 + }, + { + "epoch": 12.01, + "learning_rate": 7.016410256410256e-06, + "loss": 0.0408, + "step": 6325 + }, + { + "epoch": 12.01, + "learning_rate": 7.003589743589744e-06, + "loss": 0.0334, + "step": 6350 + }, + { + "epoch": 12.01, + "learning_rate": 6.990769230769231e-06, + "loss": 0.034, + "step": 6375 + }, + { + "epoch": 12.01, + "learning_rate": 6.9779487179487186e-06, + "loss": 0.0346, + "step": 6400 + }, + { + "epoch": 12.01, + "learning_rate": 6.965128205128205e-06, + "loss": 0.0483, + "step": 6425 + }, + { + "epoch": 12.02, + "learning_rate": 6.952307692307693e-06, + "loss": 0.0479, + "step": 6450 + }, + { + "epoch": 12.02, + "learning_rate": 6.93948717948718e-06, + "loss": 0.063, + "step": 6475 + }, + { + "epoch": 12.02, + "learning_rate": 6.926666666666667e-06, + "loss": 0.035, + "step": 6500 + }, + { + "epoch": 12.02, + "learning_rate": 6.913846153846154e-06, + "loss": 0.0428, + "step": 6525 + }, + { + "epoch": 12.02, + "learning_rate": 6.901025641025641e-06, + "loss": 0.0664, + "step": 6550 + }, + { + "epoch": 12.02, + "learning_rate": 6.888205128205129e-06, + "loss": 0.0944, + "step": 6575 + }, + { + "epoch": 12.02, + "learning_rate": 6.875384615384616e-06, + "loss": 0.097, + "step": 6600 + }, + { + "epoch": 12.02, + "learning_rate": 6.862564102564103e-06, + "loss": 0.0963, + "step": 6625 + }, + { + "epoch": 12.03, + "learning_rate": 6.84974358974359e-06, + "loss": 0.106, + "step": 6650 + }, + { + "epoch": 13.0, + "learning_rate": 6.836923076923078e-06, + "loss": 0.0443, + "step": 6675 + }, + { + "epoch": 13.0, + "learning_rate": 6.824102564102565e-06, + "loss": 0.0373, + "step": 6700 + }, + { + "epoch": 13.0, + "learning_rate": 6.811282051282052e-06, + "loss": 0.0404, + "step": 6725 + }, + { + "epoch": 13.0, + "learning_rate": 6.798461538461539e-06, + "loss": 0.0328, + "step": 6750 + }, + { + "epoch": 13.01, + "learning_rate": 6.785641025641027e-06, + "loss": 0.0381, + "step": 6775 + }, + { + "epoch": 13.01, + "learning_rate": 6.772820512820514e-06, + "loss": 0.0323, + "step": 6800 + }, + { + "epoch": 13.01, + "learning_rate": 6.760000000000001e-06, + "loss": 0.0339, + "step": 6825 + }, + { + "epoch": 13.01, + "learning_rate": 6.747179487179488e-06, + "loss": 0.0412, + "step": 6850 + }, + { + "epoch": 13.01, + "learning_rate": 6.734358974358976e-06, + "loss": 0.0383, + "step": 6875 + }, + { + "epoch": 13.01, + "learning_rate": 6.721538461538462e-06, + "loss": 0.0336, + "step": 6900 + }, + { + "epoch": 13.01, + "learning_rate": 6.708717948717949e-06, + "loss": 0.0329, + "step": 6925 + }, + { + "epoch": 13.01, + "learning_rate": 6.695897435897436e-06, + "loss": 0.0346, + "step": 6950 + }, + { + "epoch": 13.02, + "learning_rate": 6.683076923076923e-06, + "loss": 0.0526, + "step": 6975 + }, + { + "epoch": 13.02, + "learning_rate": 6.670256410256411e-06, + "loss": 0.0752, + "step": 7000 + }, + { + "epoch": 13.02, + "eval_loss": 0.16529309749603271, + "eval_runtime": 5901.988, + "eval_samples_per_second": 2.775, + "eval_steps_per_second": 0.174, + "eval_wer": 5.909688482423549, + "step": 7000 + }, + { + "epoch": 13.02, + "learning_rate": 6.657435897435898e-06, + "loss": 0.0655, + "step": 7025 + }, + { + "epoch": 13.02, + "learning_rate": 6.644615384615385e-06, + "loss": 0.0655, + "step": 7050 + }, + { + "epoch": 13.02, + "learning_rate": 6.631794871794872e-06, + "loss": 0.0694, + "step": 7075 + }, + { + "epoch": 13.02, + "learning_rate": 6.618974358974359e-06, + "loss": 0.0682, + "step": 7100 + }, + { + "epoch": 13.02, + "learning_rate": 6.606153846153847e-06, + "loss": 0.0625, + "step": 7125 + }, + { + "epoch": 13.02, + "learning_rate": 6.5933333333333335e-06, + "loss": 0.0663, + "step": 7150 + }, + { + "epoch": 14.0, + "learning_rate": 6.580512820512821e-06, + "loss": 0.0681, + "step": 7175 + }, + { + "epoch": 14.0, + "learning_rate": 6.567692307692308e-06, + "loss": 0.0487, + "step": 7200 + }, + { + "epoch": 14.0, + "learning_rate": 6.554871794871796e-06, + "loss": 0.0634, + "step": 7225 + }, + { + "epoch": 14.0, + "learning_rate": 6.5420512820512824e-06, + "loss": 0.0465, + "step": 7250 + }, + { + "epoch": 14.01, + "learning_rate": 6.52923076923077e-06, + "loss": 0.0618, + "step": 7275 + }, + { + "epoch": 14.01, + "learning_rate": 6.516410256410257e-06, + "loss": 0.063, + "step": 7300 + }, + { + "epoch": 14.01, + "learning_rate": 6.503589743589745e-06, + "loss": 0.0586, + "step": 7325 + }, + { + "epoch": 14.01, + "learning_rate": 6.4907692307692314e-06, + "loss": 0.0593, + "step": 7350 + }, + { + "epoch": 14.01, + "learning_rate": 6.477948717948719e-06, + "loss": 0.0604, + "step": 7375 + }, + { + "epoch": 14.01, + "learning_rate": 6.465128205128206e-06, + "loss": 0.0528, + "step": 7400 + }, + { + "epoch": 14.01, + "learning_rate": 6.452307692307694e-06, + "loss": 0.0464, + "step": 7425 + }, + { + "epoch": 14.01, + "learning_rate": 6.4394871794871804e-06, + "loss": 0.0685, + "step": 7450 + }, + { + "epoch": 14.02, + "learning_rate": 6.426666666666668e-06, + "loss": 0.0533, + "step": 7475 + }, + { + "epoch": 14.02, + "learning_rate": 6.413846153846154e-06, + "loss": 0.0536, + "step": 7500 + }, + { + "epoch": 14.02, + "learning_rate": 6.401025641025641e-06, + "loss": 0.0485, + "step": 7525 + }, + { + "epoch": 14.02, + "learning_rate": 6.3882051282051286e-06, + "loss": 0.0462, + "step": 7550 + }, + { + "epoch": 14.02, + "learning_rate": 6.375384615384615e-06, + "loss": 0.0593, + "step": 7575 + }, + { + "epoch": 14.02, + "learning_rate": 6.362564102564103e-06, + "loss": 0.0594, + "step": 7600 + }, + { + "epoch": 14.02, + "learning_rate": 6.34974358974359e-06, + "loss": 0.0656, + "step": 7625 + }, + { + "epoch": 14.02, + "learning_rate": 6.3369230769230776e-06, + "loss": 0.0602, + "step": 7650 + }, + { + "epoch": 14.03, + "learning_rate": 6.324102564102564e-06, + "loss": 0.0611, + "step": 7675 + }, + { + "epoch": 15.0, + "learning_rate": 6.311282051282051e-06, + "loss": 0.0374, + "step": 7700 + }, + { + "epoch": 15.0, + "learning_rate": 6.298461538461539e-06, + "loss": 0.0352, + "step": 7725 + }, + { + "epoch": 15.0, + "learning_rate": 6.285641025641026e-06, + "loss": 0.0441, + "step": 7750 + }, + { + "epoch": 15.0, + "learning_rate": 6.272820512820513e-06, + "loss": 0.0364, + "step": 7775 + }, + { + "epoch": 15.01, + "learning_rate": 6.26e-06, + "loss": 0.0403, + "step": 7800 + }, + { + "epoch": 15.01, + "learning_rate": 6.247179487179488e-06, + "loss": 0.0294, + "step": 7825 + }, + { + "epoch": 15.01, + "learning_rate": 6.234358974358975e-06, + "loss": 0.0293, + "step": 7850 + }, + { + "epoch": 15.01, + "learning_rate": 6.221538461538462e-06, + "loss": 0.0273, + "step": 7875 + }, + { + "epoch": 15.01, + "learning_rate": 6.208717948717949e-06, + "loss": 0.0373, + "step": 7900 + }, + { + "epoch": 15.01, + "learning_rate": 6.195897435897437e-06, + "loss": 0.0333, + "step": 7925 + }, + { + "epoch": 15.01, + "learning_rate": 6.183076923076924e-06, + "loss": 0.0383, + "step": 7950 + }, + { + "epoch": 15.01, + "learning_rate": 6.170256410256411e-06, + "loss": 0.0429, + "step": 7975 + }, + { + "epoch": 15.02, + "learning_rate": 6.157435897435898e-06, + "loss": 0.0359, + "step": 8000 + }, + { + "epoch": 15.02, + "eval_loss": 0.14058415591716766, + "eval_runtime": 5922.6947, + "eval_samples_per_second": 2.766, + "eval_steps_per_second": 0.173, + "eval_wer": 5.875964561303229, + "step": 8000 + }, + { + "epoch": 15.02, + "learning_rate": 6.144615384615386e-06, + "loss": 0.0413, + "step": 8025 + }, + { + "epoch": 15.02, + "learning_rate": 6.131794871794873e-06, + "loss": 0.0522, + "step": 8050 + }, + { + "epoch": 15.02, + "learning_rate": 6.11897435897436e-06, + "loss": 0.0359, + "step": 8075 + }, + { + "epoch": 15.02, + "learning_rate": 6.106153846153846e-06, + "loss": 0.0242, + "step": 8100 + }, + { + "epoch": 15.02, + "learning_rate": 6.093333333333333e-06, + "loss": 0.0332, + "step": 8125 + }, + { + "epoch": 15.02, + "learning_rate": 6.080512820512821e-06, + "loss": 0.0303, + "step": 8150 + }, + { + "epoch": 15.02, + "learning_rate": 6.067692307692308e-06, + "loss": 0.0363, + "step": 8175 + }, + { + "epoch": 16.0, + "learning_rate": 6.054871794871795e-06, + "loss": 0.0288, + "step": 8200 + }, + { + "epoch": 16.0, + "learning_rate": 6.042051282051282e-06, + "loss": 0.0355, + "step": 8225 + }, + { + "epoch": 16.0, + "learning_rate": 6.02923076923077e-06, + "loss": 0.0224, + "step": 8250 + }, + { + "epoch": 16.0, + "learning_rate": 6.016410256410257e-06, + "loss": 0.0199, + "step": 8275 + }, + { + "epoch": 16.01, + "learning_rate": 6.0035897435897435e-06, + "loss": 0.024, + "step": 8300 + }, + { + "epoch": 16.01, + "learning_rate": 5.990769230769231e-06, + "loss": 0.0244, + "step": 8325 + }, + { + "epoch": 16.01, + "learning_rate": 5.977948717948718e-06, + "loss": 0.0238, + "step": 8350 + }, + { + "epoch": 16.01, + "learning_rate": 5.965128205128206e-06, + "loss": 0.0272, + "step": 8375 + }, + { + "epoch": 16.01, + "learning_rate": 5.9523076923076925e-06, + "loss": 0.0238, + "step": 8400 + }, + { + "epoch": 16.01, + "learning_rate": 5.93948717948718e-06, + "loss": 0.0256, + "step": 8425 + }, + { + "epoch": 16.01, + "learning_rate": 5.926666666666667e-06, + "loss": 0.0417, + "step": 8450 + }, + { + "epoch": 16.01, + "learning_rate": 5.913846153846155e-06, + "loss": 0.0342, + "step": 8475 + }, + { + "epoch": 16.02, + "learning_rate": 5.9010256410256415e-06, + "loss": 0.0301, + "step": 8500 + }, + { + "epoch": 16.02, + "learning_rate": 5.888205128205129e-06, + "loss": 0.0354, + "step": 8525 + }, + { + "epoch": 16.02, + "learning_rate": 5.875384615384616e-06, + "loss": 0.0365, + "step": 8550 + }, + { + "epoch": 16.02, + "learning_rate": 5.862564102564104e-06, + "loss": 0.0377, + "step": 8575 + }, + { + "epoch": 16.02, + "learning_rate": 5.8497435897435904e-06, + "loss": 0.0373, + "step": 8600 + }, + { + "epoch": 16.02, + "learning_rate": 5.836923076923078e-06, + "loss": 0.0361, + "step": 8625 + }, + { + "epoch": 16.02, + "learning_rate": 5.824102564102565e-06, + "loss": 0.0413, + "step": 8650 + }, + { + "epoch": 16.02, + "learning_rate": 5.811282051282053e-06, + "loss": 0.0425, + "step": 8675 + }, + { + "epoch": 16.03, + "learning_rate": 5.7984615384615394e-06, + "loss": 0.057, + "step": 8700 + }, + { + "epoch": 17.0, + "learning_rate": 5.785641025641025e-06, + "loss": 0.0711, + "step": 8725 + }, + { + "epoch": 17.0, + "learning_rate": 5.772820512820513e-06, + "loss": 0.0688, + "step": 8750 + }, + { + "epoch": 17.0, + "learning_rate": 5.76e-06, + "loss": 0.0355, + "step": 8775 + }, + { + "epoch": 17.0, + "learning_rate": 5.747179487179488e-06, + "loss": 0.0439, + "step": 8800 + }, + { + "epoch": 17.01, + "learning_rate": 5.734358974358974e-06, + "loss": 0.0464, + "step": 8825 + }, + { + "epoch": 17.01, + "learning_rate": 5.721538461538462e-06, + "loss": 0.0306, + "step": 8850 + }, + { + "epoch": 17.01, + "learning_rate": 5.708717948717949e-06, + "loss": 0.0192, + "step": 8875 + }, + { + "epoch": 17.01, + "learning_rate": 5.6958974358974366e-06, + "loss": 0.0319, + "step": 8900 + }, + { + "epoch": 17.01, + "learning_rate": 5.683076923076923e-06, + "loss": 0.0464, + "step": 8925 + }, + { + "epoch": 17.01, + "learning_rate": 5.67025641025641e-06, + "loss": 0.0429, + "step": 8950 + }, + { + "epoch": 17.01, + "learning_rate": 5.657435897435898e-06, + "loss": 0.0506, + "step": 8975 + }, + { + "epoch": 17.01, + "learning_rate": 5.644615384615385e-06, + "loss": 0.0565, + "step": 9000 + }, + { + "epoch": 17.01, + "eval_loss": 0.14963489770889282, + "eval_runtime": 5831.5589, + "eval_samples_per_second": 2.809, + "eval_steps_per_second": 0.176, + "eval_wer": 5.967990854529866, + "step": 9000 + }, + { + "epoch": 17.02, + "learning_rate": 5.631794871794872e-06, + "loss": 0.0427, + "step": 9025 + }, + { + "epoch": 17.02, + "learning_rate": 5.618974358974359e-06, + "loss": 0.0473, + "step": 9050 + }, + { + "epoch": 17.02, + "learning_rate": 5.606153846153847e-06, + "loss": 0.0498, + "step": 9075 + }, + { + "epoch": 17.02, + "learning_rate": 5.593333333333334e-06, + "loss": 0.0513, + "step": 9100 + }, + { + "epoch": 17.02, + "learning_rate": 5.580512820512821e-06, + "loss": 0.0402, + "step": 9125 + }, + { + "epoch": 17.02, + "learning_rate": 5.567692307692308e-06, + "loss": 0.0438, + "step": 9150 + }, + { + "epoch": 17.02, + "learning_rate": 5.554871794871796e-06, + "loss": 0.0437, + "step": 9175 + }, + { + "epoch": 17.02, + "learning_rate": 5.542051282051283e-06, + "loss": 0.0524, + "step": 9200 + }, + { + "epoch": 18.0, + "learning_rate": 5.52923076923077e-06, + "loss": 0.0499, + "step": 9225 + }, + { + "epoch": 18.0, + "learning_rate": 5.516410256410257e-06, + "loss": 0.0367, + "step": 9250 + }, + { + "epoch": 18.0, + "learning_rate": 5.503589743589745e-06, + "loss": 0.0301, + "step": 9275 + }, + { + "epoch": 18.0, + "learning_rate": 5.490769230769232e-06, + "loss": 0.0332, + "step": 9300 + }, + { + "epoch": 18.01, + "learning_rate": 5.477948717948718e-06, + "loss": 0.0308, + "step": 9325 + }, + { + "epoch": 18.01, + "learning_rate": 5.465128205128205e-06, + "loss": 0.0279, + "step": 9350 + }, + { + "epoch": 18.01, + "learning_rate": 5.452307692307692e-06, + "loss": 0.0302, + "step": 9375 + }, + { + "epoch": 18.01, + "learning_rate": 5.43948717948718e-06, + "loss": 0.0318, + "step": 9400 + }, + { + "epoch": 18.01, + "learning_rate": 5.426666666666667e-06, + "loss": 0.0272, + "step": 9425 + }, + { + "epoch": 18.01, + "learning_rate": 5.413846153846154e-06, + "loss": 0.0308, + "step": 9450 + }, + { + "epoch": 18.01, + "learning_rate": 5.401025641025641e-06, + "loss": 0.0301, + "step": 9475 + }, + { + "epoch": 18.01, + "learning_rate": 5.388205128205129e-06, + "loss": 0.0251, + "step": 9500 + }, + { + "epoch": 18.02, + "learning_rate": 5.375384615384616e-06, + "loss": 0.0265, + "step": 9525 + }, + { + "epoch": 18.02, + "learning_rate": 5.3625641025641025e-06, + "loss": 0.028, + "step": 9550 + }, + { + "epoch": 18.02, + "learning_rate": 5.34974358974359e-06, + "loss": 0.0265, + "step": 9575 + }, + { + "epoch": 18.02, + "learning_rate": 5.336923076923077e-06, + "loss": 0.0267, + "step": 9600 + }, + { + "epoch": 18.02, + "learning_rate": 5.324102564102565e-06, + "loss": 0.0252, + "step": 9625 + }, + { + "epoch": 18.02, + "learning_rate": 5.3112820512820515e-06, + "loss": 0.0225, + "step": 9650 + }, + { + "epoch": 18.02, + "learning_rate": 5.298461538461539e-06, + "loss": 0.0228, + "step": 9675 + }, + { + "epoch": 18.02, + "learning_rate": 5.285641025641026e-06, + "loss": 0.0228, + "step": 9700 + }, + { + "epoch": 18.03, + "learning_rate": 5.272820512820514e-06, + "loss": 0.0214, + "step": 9725 + }, + { + "epoch": 19.0, + "learning_rate": 5.2600000000000005e-06, + "loss": 0.022, + "step": 9750 + }, + { + "epoch": 19.0, + "learning_rate": 5.247179487179488e-06, + "loss": 0.0212, + "step": 9775 + }, + { + "epoch": 19.0, + "learning_rate": 5.234358974358975e-06, + "loss": 0.0201, + "step": 9800 + }, + { + "epoch": 19.0, + "learning_rate": 5.221538461538463e-06, + "loss": 0.0188, + "step": 9825 + }, + { + "epoch": 19.01, + "learning_rate": 5.2087179487179495e-06, + "loss": 0.0199, + "step": 9850 + }, + { + "epoch": 19.01, + "learning_rate": 5.195897435897437e-06, + "loss": 0.0192, + "step": 9875 + }, + { + "epoch": 19.01, + "learning_rate": 5.183076923076924e-06, + "loss": 0.0201, + "step": 9900 + }, + { + "epoch": 19.01, + "learning_rate": 5.170256410256412e-06, + "loss": 0.0175, + "step": 9925 + }, + { + "epoch": 19.01, + "learning_rate": 5.157435897435898e-06, + "loss": 0.0191, + "step": 9950 + }, + { + "epoch": 19.01, + "learning_rate": 5.144615384615384e-06, + "loss": 0.0186, + "step": 9975 + }, + { + "epoch": 19.01, + "learning_rate": 5.131794871794872e-06, + "loss": 0.0196, + "step": 10000 + }, + { + "epoch": 19.01, + "eval_loss": 0.17882370948791504, + "eval_runtime": 5918.0937, + "eval_samples_per_second": 2.768, + "eval_steps_per_second": 0.173, + "eval_wer": 5.27464989997142, + "step": 10000 + }, + { + "epoch": 19.01, + "learning_rate": 5.118974358974359e-06, + "loss": 0.0216, + "step": 10025 + }, + { + "epoch": 19.02, + "learning_rate": 5.106153846153847e-06, + "loss": 0.0212, + "step": 10050 + }, + { + "epoch": 19.02, + "learning_rate": 5.093333333333333e-06, + "loss": 0.0233, + "step": 10075 + }, + { + "epoch": 19.02, + "learning_rate": 5.080512820512821e-06, + "loss": 0.021, + "step": 10100 + }, + { + "epoch": 19.02, + "learning_rate": 5.067692307692308e-06, + "loss": 0.0219, + "step": 10125 + }, + { + "epoch": 19.02, + "learning_rate": 5.054871794871795e-06, + "loss": 0.0204, + "step": 10150 + }, + { + "epoch": 19.02, + "learning_rate": 5.042051282051282e-06, + "loss": 0.0232, + "step": 10175 + }, + { + "epoch": 19.02, + "learning_rate": 5.029230769230769e-06, + "loss": 0.0243, + "step": 10200 + }, + { + "epoch": 19.02, + "learning_rate": 5.016410256410257e-06, + "loss": 0.0314, + "step": 10225 + }, + { + "epoch": 20.0, + "learning_rate": 5.003589743589744e-06, + "loss": 0.0312, + "step": 10250 + }, + { + "epoch": 20.0, + "learning_rate": 4.990769230769231e-06, + "loss": 0.0251, + "step": 10275 + }, + { + "epoch": 20.0, + "learning_rate": 4.977948717948718e-06, + "loss": 0.025, + "step": 10300 + }, + { + "epoch": 20.0, + "learning_rate": 4.965128205128206e-06, + "loss": 0.0204, + "step": 10325 + }, + { + "epoch": 20.01, + "learning_rate": 4.952307692307693e-06, + "loss": 0.028, + "step": 10350 + }, + { + "epoch": 20.01, + "learning_rate": 4.93948717948718e-06, + "loss": 0.025, + "step": 10375 + }, + { + "epoch": 20.01, + "learning_rate": 4.926666666666667e-06, + "loss": 0.0237, + "step": 10400 + }, + { + "epoch": 20.01, + "learning_rate": 4.913846153846154e-06, + "loss": 0.025, + "step": 10425 + }, + { + "epoch": 20.01, + "learning_rate": 4.901025641025641e-06, + "loss": 0.0237, + "step": 10450 + }, + { + "epoch": 20.01, + "learning_rate": 4.8882051282051285e-06, + "loss": 0.0184, + "step": 10475 + }, + { + "epoch": 20.01, + "learning_rate": 4.875384615384615e-06, + "loss": 0.0207, + "step": 10500 + }, + { + "epoch": 20.01, + "learning_rate": 4.862564102564103e-06, + "loss": 0.0269, + "step": 10525 + }, + { + "epoch": 20.02, + "learning_rate": 4.84974358974359e-06, + "loss": 0.0221, + "step": 10550 + }, + { + "epoch": 20.02, + "learning_rate": 4.8369230769230775e-06, + "loss": 0.0244, + "step": 10575 + }, + { + "epoch": 20.02, + "learning_rate": 4.824102564102564e-06, + "loss": 0.0198, + "step": 10600 + }, + { + "epoch": 20.02, + "learning_rate": 4.811282051282052e-06, + "loss": 0.0195, + "step": 10625 + }, + { + "epoch": 20.02, + "learning_rate": 4.798461538461539e-06, + "loss": 0.024, + "step": 10650 + }, + { + "epoch": 20.02, + "learning_rate": 4.7856410256410265e-06, + "loss": 0.0264, + "step": 10675 + }, + { + "epoch": 20.02, + "learning_rate": 4.772820512820513e-06, + "loss": 0.0286, + "step": 10700 + }, + { + "epoch": 20.02, + "learning_rate": 4.76e-06, + "loss": 0.0228, + "step": 10725 + }, + { + "epoch": 20.03, + "learning_rate": 4.747179487179487e-06, + "loss": 0.0258, + "step": 10750 + }, + { + "epoch": 21.0, + "learning_rate": 4.734358974358975e-06, + "loss": 0.0255, + "step": 10775 + }, + { + "epoch": 21.0, + "learning_rate": 4.7215384615384615e-06, + "loss": 0.0291, + "step": 10800 + }, + { + "epoch": 21.0, + "learning_rate": 4.708717948717949e-06, + "loss": 0.0221, + "step": 10825 + }, + { + "epoch": 21.0, + "learning_rate": 4.695897435897436e-06, + "loss": 0.0155, + "step": 10850 + }, + { + "epoch": 21.01, + "learning_rate": 4.683076923076924e-06, + "loss": 0.0252, + "step": 10875 + }, + { + "epoch": 21.01, + "learning_rate": 4.6702564102564105e-06, + "loss": 0.0226, + "step": 10900 + }, + { + "epoch": 21.01, + "learning_rate": 4.657435897435898e-06, + "loss": 0.0182, + "step": 10925 + }, + { + "epoch": 21.01, + "learning_rate": 4.644615384615385e-06, + "loss": 0.0196, + "step": 10950 + }, + { + "epoch": 21.01, + "learning_rate": 4.631794871794873e-06, + "loss": 0.0251, + "step": 10975 + }, + { + "epoch": 21.01, + "learning_rate": 4.6189743589743595e-06, + "loss": 0.0215, + "step": 11000 + }, + { + "epoch": 21.01, + "eval_loss": 0.1538822501897812, + "eval_runtime": 5905.7414, + "eval_samples_per_second": 2.774, + "eval_steps_per_second": 0.173, + "eval_wer": 5.389539868533866, + "step": 11000 + }, + { + "epoch": 21.01, + "learning_rate": 4.606153846153846e-06, + "loss": 0.0223, + "step": 11025 + }, + { + "epoch": 21.01, + "learning_rate": 4.593333333333333e-06, + "loss": 0.0242, + "step": 11050 + }, + { + "epoch": 21.02, + "learning_rate": 4.580512820512821e-06, + "loss": 0.0252, + "step": 11075 + }, + { + "epoch": 21.02, + "learning_rate": 4.567692307692308e-06, + "loss": 0.0233, + "step": 11100 + }, + { + "epoch": 21.02, + "learning_rate": 4.554871794871795e-06, + "loss": 0.0238, + "step": 11125 + }, + { + "epoch": 21.02, + "learning_rate": 4.542051282051282e-06, + "loss": 0.0231, + "step": 11150 + }, + { + "epoch": 21.02, + "learning_rate": 4.52923076923077e-06, + "loss": 0.0224, + "step": 11175 + }, + { + "epoch": 21.02, + "learning_rate": 4.516410256410257e-06, + "loss": 0.0189, + "step": 11200 + }, + { + "epoch": 21.02, + "learning_rate": 4.503589743589744e-06, + "loss": 0.0222, + "step": 11225 + }, + { + "epoch": 21.02, + "learning_rate": 4.490769230769231e-06, + "loss": 0.0248, + "step": 11250 + }, + { + "epoch": 22.0, + "learning_rate": 4.477948717948719e-06, + "loss": 0.0244, + "step": 11275 + }, + { + "epoch": 22.0, + "learning_rate": 4.465128205128206e-06, + "loss": 0.0193, + "step": 11300 + }, + { + "epoch": 22.0, + "learning_rate": 4.452307692307692e-06, + "loss": 0.0199, + "step": 11325 + }, + { + "epoch": 22.0, + "learning_rate": 4.439487179487179e-06, + "loss": 0.0237, + "step": 11350 + }, + { + "epoch": 22.01, + "learning_rate": 4.426666666666667e-06, + "loss": 0.0233, + "step": 11375 + }, + { + "epoch": 22.01, + "learning_rate": 4.413846153846154e-06, + "loss": 0.0207, + "step": 11400 + }, + { + "epoch": 22.01, + "learning_rate": 4.401025641025641e-06, + "loss": 0.0247, + "step": 11425 + }, + { + "epoch": 22.01, + "learning_rate": 4.388205128205128e-06, + "loss": 0.0271, + "step": 11450 + }, + { + "epoch": 22.01, + "learning_rate": 4.375384615384616e-06, + "loss": 0.0242, + "step": 11475 + }, + { + "epoch": 22.01, + "learning_rate": 4.362564102564103e-06, + "loss": 0.0235, + "step": 11500 + }, + { + "epoch": 22.01, + "learning_rate": 4.34974358974359e-06, + "loss": 0.0226, + "step": 11525 + }, + { + "epoch": 22.01, + "learning_rate": 4.336923076923077e-06, + "loss": 0.0253, + "step": 11550 + }, + { + "epoch": 22.02, + "learning_rate": 4.324102564102565e-06, + "loss": 0.0228, + "step": 11575 + }, + { + "epoch": 22.02, + "learning_rate": 4.311282051282052e-06, + "loss": 0.0228, + "step": 11600 + }, + { + "epoch": 22.02, + "learning_rate": 4.2984615384615386e-06, + "loss": 0.0123, + "step": 11625 + }, + { + "epoch": 22.02, + "learning_rate": 4.285641025641025e-06, + "loss": 0.0067, + "step": 11650 + }, + { + "epoch": 22.02, + "learning_rate": 4.272820512820513e-06, + "loss": 0.0073, + "step": 11675 + }, + { + "epoch": 22.02, + "learning_rate": 4.26e-06, + "loss": 0.0082, + "step": 11700 + }, + { + "epoch": 22.02, + "learning_rate": 4.2471794871794875e-06, + "loss": 0.0077, + "step": 11725 + }, + { + "epoch": 22.02, + "learning_rate": 4.234358974358974e-06, + "loss": 0.0059, + "step": 11750 + }, + { + "epoch": 22.03, + "learning_rate": 4.221538461538462e-06, + "loss": 0.0083, + "step": 11775 + }, + { + "epoch": 23.0, + "learning_rate": 4.208717948717949e-06, + "loss": 0.0216, + "step": 11800 + }, + { + "epoch": 23.0, + "learning_rate": 4.1958974358974365e-06, + "loss": 0.0189, + "step": 11825 + }, + { + "epoch": 23.0, + "learning_rate": 4.183076923076923e-06, + "loss": 0.0194, + "step": 11850 + }, + { + "epoch": 23.0, + "learning_rate": 4.170256410256411e-06, + "loss": 0.0175, + "step": 11875 + }, + { + "epoch": 23.01, + "learning_rate": 4.157435897435898e-06, + "loss": 0.0181, + "step": 11900 + }, + { + "epoch": 23.01, + "learning_rate": 4.144615384615385e-06, + "loss": 0.0168, + "step": 11925 + }, + { + "epoch": 23.01, + "learning_rate": 4.131794871794872e-06, + "loss": 0.0187, + "step": 11950 + }, + { + "epoch": 23.01, + "learning_rate": 4.118974358974359e-06, + "loss": 0.0167, + "step": 11975 + }, + { + "epoch": 23.01, + "learning_rate": 4.106153846153846e-06, + "loss": 0.0178, + "step": 12000 + }, + { + "epoch": 23.01, + "eval_loss": 0.1800205558538437, + "eval_runtime": 5915.5164, + "eval_samples_per_second": 2.769, + "eval_steps_per_second": 0.173, + "eval_wer": 5.376393255215776, + "step": 12000 + }, + { + "epoch": 23.01, + "learning_rate": 4.093333333333334e-06, + "loss": 0.0165, + "step": 12025 + }, + { + "epoch": 23.01, + "learning_rate": 4.0805128205128205e-06, + "loss": 0.0156, + "step": 12050 + }, + { + "epoch": 23.01, + "learning_rate": 4.067692307692308e-06, + "loss": 0.0166, + "step": 12075 + }, + { + "epoch": 23.02, + "learning_rate": 4.054871794871795e-06, + "loss": 0.0165, + "step": 12100 + }, + { + "epoch": 23.02, + "learning_rate": 4.042051282051283e-06, + "loss": 0.0151, + "step": 12125 + }, + { + "epoch": 23.02, + "learning_rate": 4.0292307692307695e-06, + "loss": 0.0166, + "step": 12150 + }, + { + "epoch": 23.02, + "learning_rate": 4.016410256410257e-06, + "loss": 0.0141, + "step": 12175 + }, + { + "epoch": 23.02, + "learning_rate": 4.003589743589744e-06, + "loss": 0.0149, + "step": 12200 + }, + { + "epoch": 23.02, + "learning_rate": 3.990769230769231e-06, + "loss": 0.013, + "step": 12225 + }, + { + "epoch": 23.02, + "learning_rate": 3.9779487179487185e-06, + "loss": 0.0137, + "step": 12250 + }, + { + "epoch": 23.02, + "learning_rate": 3.965128205128205e-06, + "loss": 0.0132, + "step": 12275 + }, + { + "epoch": 24.0, + "learning_rate": 3.952307692307692e-06, + "loss": 0.0167, + "step": 12300 + }, + { + "epoch": 24.0, + "learning_rate": 3.93948717948718e-06, + "loss": 0.0332, + "step": 12325 + }, + { + "epoch": 24.0, + "learning_rate": 3.926666666666667e-06, + "loss": 0.0304, + "step": 12350 + }, + { + "epoch": 24.0, + "learning_rate": 3.913846153846154e-06, + "loss": 0.0312, + "step": 12375 + }, + { + "epoch": 24.01, + "learning_rate": 3.901025641025641e-06, + "loss": 0.0287, + "step": 12400 + }, + { + "epoch": 24.01, + "learning_rate": 3.888205128205129e-06, + "loss": 0.0293, + "step": 12425 + }, + { + "epoch": 24.01, + "learning_rate": 3.875384615384616e-06, + "loss": 0.0317, + "step": 12450 + }, + { + "epoch": 24.01, + "learning_rate": 3.862564102564103e-06, + "loss": 0.0315, + "step": 12475 + }, + { + "epoch": 24.01, + "learning_rate": 3.84974358974359e-06, + "loss": 0.0278, + "step": 12500 + }, + { + "epoch": 24.01, + "learning_rate": 3.836923076923077e-06, + "loss": 0.0269, + "step": 12525 + }, + { + "epoch": 24.01, + "learning_rate": 3.824102564102565e-06, + "loss": 0.0289, + "step": 12550 + }, + { + "epoch": 24.01, + "learning_rate": 3.8112820512820514e-06, + "loss": 0.0273, + "step": 12575 + }, + { + "epoch": 24.02, + "learning_rate": 3.7984615384615387e-06, + "loss": 0.025, + "step": 12600 + }, + { + "epoch": 24.02, + "learning_rate": 3.785641025641026e-06, + "loss": 0.0286, + "step": 12625 + }, + { + "epoch": 24.02, + "learning_rate": 3.772820512820513e-06, + "loss": 0.0287, + "step": 12650 + }, + { + "epoch": 24.02, + "learning_rate": 3.7600000000000004e-06, + "loss": 0.0247, + "step": 12675 + }, + { + "epoch": 24.02, + "learning_rate": 3.7471794871794877e-06, + "loss": 0.028, + "step": 12700 + }, + { + "epoch": 24.02, + "learning_rate": 3.734358974358975e-06, + "loss": 0.0265, + "step": 12725 + }, + { + "epoch": 24.02, + "learning_rate": 3.721538461538462e-06, + "loss": 0.0303, + "step": 12750 + }, + { + "epoch": 24.02, + "learning_rate": 3.7087179487179494e-06, + "loss": 0.0265, + "step": 12775 + }, + { + "epoch": 24.03, + "learning_rate": 3.6958974358974362e-06, + "loss": 0.032, + "step": 12800 + }, + { + "epoch": 25.0, + "learning_rate": 3.683076923076923e-06, + "loss": 0.0053, + "step": 12825 + }, + { + "epoch": 25.0, + "learning_rate": 3.6702564102564103e-06, + "loss": 0.005, + "step": 12850 + }, + { + "epoch": 25.0, + "learning_rate": 3.6574358974358976e-06, + "loss": 0.005, + "step": 12875 + }, + { + "epoch": 25.0, + "learning_rate": 3.644615384615385e-06, + "loss": 0.0118, + "step": 12900 + }, + { + "epoch": 25.01, + "learning_rate": 3.631794871794872e-06, + "loss": 0.0117, + "step": 12925 + }, + { + "epoch": 25.01, + "learning_rate": 3.6189743589743593e-06, + "loss": 0.0118, + "step": 12950 + }, + { + "epoch": 25.01, + "learning_rate": 3.6061538461538466e-06, + "loss": 0.0118, + "step": 12975 + }, + { + "epoch": 25.01, + "learning_rate": 3.593333333333334e-06, + "loss": 0.0114, + "step": 13000 + }, + { + "epoch": 25.01, + "eval_loss": 0.17090679705142975, + "eval_runtime": 5865.554, + "eval_samples_per_second": 2.793, + "eval_steps_per_second": 0.175, + "eval_wer": 5.207773649614175, + "step": 13000 + }, + { + "epoch": 25.01, + "learning_rate": 3.580512820512821e-06, + "loss": 0.0109, + "step": 13025 + }, + { + "epoch": 25.01, + "learning_rate": 3.5676923076923083e-06, + "loss": 0.0124, + "step": 13050 + }, + { + "epoch": 25.01, + "learning_rate": 3.5548717948717955e-06, + "loss": 0.0118, + "step": 13075 + }, + { + "epoch": 25.02, + "learning_rate": 3.5420512820512824e-06, + "loss": 0.0121, + "step": 13100 + }, + { + "epoch": 25.02, + "learning_rate": 3.5292307692307696e-06, + "loss": 0.0101, + "step": 13125 + }, + { + "epoch": 25.02, + "learning_rate": 3.5164102564102564e-06, + "loss": 0.0124, + "step": 13150 + }, + { + "epoch": 25.02, + "learning_rate": 3.5035897435897437e-06, + "loss": 0.0108, + "step": 13175 + }, + { + "epoch": 25.02, + "learning_rate": 3.490769230769231e-06, + "loss": 0.0133, + "step": 13200 + }, + { + "epoch": 25.02, + "learning_rate": 3.477948717948718e-06, + "loss": 0.0134, + "step": 13225 + }, + { + "epoch": 25.02, + "learning_rate": 3.4651282051282054e-06, + "loss": 0.017, + "step": 13250 + }, + { + "epoch": 25.02, + "learning_rate": 3.4523076923076927e-06, + "loss": 0.023, + "step": 13275 + }, + { + "epoch": 25.02, + "learning_rate": 3.43948717948718e-06, + "loss": 0.0157, + "step": 13300 + }, + { + "epoch": 26.0, + "learning_rate": 3.426666666666667e-06, + "loss": 0.0157, + "step": 13325 + }, + { + "epoch": 26.0, + "learning_rate": 3.4138461538461544e-06, + "loss": 0.0172, + "step": 13350 + }, + { + "epoch": 26.0, + "learning_rate": 3.4010256410256417e-06, + "loss": 0.0091, + "step": 13375 + }, + { + "epoch": 26.0, + "learning_rate": 3.3882051282051285e-06, + "loss": 0.0097, + "step": 13400 + }, + { + "epoch": 26.01, + "learning_rate": 3.3753846153846157e-06, + "loss": 0.0128, + "step": 13425 + }, + { + "epoch": 26.01, + "learning_rate": 3.3625641025641026e-06, + "loss": 0.0123, + "step": 13450 + }, + { + "epoch": 26.01, + "learning_rate": 3.34974358974359e-06, + "loss": 0.0115, + "step": 13475 + }, + { + "epoch": 26.01, + "learning_rate": 3.336923076923077e-06, + "loss": 0.0129, + "step": 13500 + }, + { + "epoch": 26.01, + "learning_rate": 3.3241025641025643e-06, + "loss": 0.0118, + "step": 13525 + }, + { + "epoch": 26.01, + "learning_rate": 3.3112820512820516e-06, + "loss": 0.0138, + "step": 13550 + }, + { + "epoch": 26.01, + "learning_rate": 3.298461538461539e-06, + "loss": 0.018, + "step": 13575 + }, + { + "epoch": 26.01, + "learning_rate": 3.285641025641026e-06, + "loss": 0.0127, + "step": 13600 + }, + { + "epoch": 26.02, + "learning_rate": 3.2728205128205133e-06, + "loss": 0.0154, + "step": 13625 + }, + { + "epoch": 26.02, + "learning_rate": 3.2600000000000006e-06, + "loss": 0.0157, + "step": 13650 + }, + { + "epoch": 26.02, + "learning_rate": 3.247179487179488e-06, + "loss": 0.0195, + "step": 13675 + }, + { + "epoch": 26.02, + "learning_rate": 3.2348717948717952e-06, + "loss": 0.0179, + "step": 13700 + }, + { + "epoch": 26.02, + "learning_rate": 3.2220512820512825e-06, + "loss": 0.0191, + "step": 13725 + }, + { + "epoch": 26.02, + "learning_rate": 3.2092307692307693e-06, + "loss": 0.0215, + "step": 13750 + }, + { + "epoch": 26.02, + "learning_rate": 3.1964102564102565e-06, + "loss": 0.0186, + "step": 13775 + }, + { + "epoch": 26.02, + "learning_rate": 3.1835897435897438e-06, + "loss": 0.0207, + "step": 13800 + }, + { + "epoch": 27.0, + "learning_rate": 3.170769230769231e-06, + "loss": 0.0221, + "step": 13825 + }, + { + "epoch": 27.0, + "learning_rate": 3.1579487179487183e-06, + "loss": 0.0126, + "step": 13850 + }, + { + "epoch": 27.0, + "learning_rate": 3.1451282051282055e-06, + "loss": 0.0122, + "step": 13875 + }, + { + "epoch": 27.0, + "learning_rate": 3.1323076923076928e-06, + "loss": 0.0125, + "step": 13900 + }, + { + "epoch": 27.01, + "learning_rate": 3.1194871794871796e-06, + "loss": 0.0125, + "step": 13925 + }, + { + "epoch": 27.01, + "learning_rate": 3.106666666666667e-06, + "loss": 0.0096, + "step": 13950 + }, + { + "epoch": 27.01, + "learning_rate": 3.093846153846154e-06, + "loss": 0.0092, + "step": 13975 + }, + { + "epoch": 27.01, + "learning_rate": 3.0810256410256413e-06, + "loss": 0.0123, + "step": 14000 + }, + { + "epoch": 27.01, + "eval_loss": 0.18274465203285217, + "eval_runtime": 5946.1315, + "eval_samples_per_second": 2.755, + "eval_steps_per_second": 0.172, + "eval_wer": 5.200342955130037, + "step": 14000 + }, + { + "epoch": 27.01, + "learning_rate": 3.0682051282051286e-06, + "loss": 0.012, + "step": 14025 + }, + { + "epoch": 27.01, + "learning_rate": 3.0553846153846154e-06, + "loss": 0.0107, + "step": 14050 + }, + { + "epoch": 27.01, + "learning_rate": 3.0425641025641027e-06, + "loss": 0.0106, + "step": 14075 + }, + { + "epoch": 27.01, + "learning_rate": 3.02974358974359e-06, + "loss": 0.0104, + "step": 14100 + }, + { + "epoch": 27.02, + "learning_rate": 3.016923076923077e-06, + "loss": 0.0116, + "step": 14125 + }, + { + "epoch": 27.02, + "learning_rate": 3.0041025641025644e-06, + "loss": 0.0098, + "step": 14150 + }, + { + "epoch": 27.02, + "learning_rate": 2.9912820512820517e-06, + "loss": 0.0076, + "step": 14175 + }, + { + "epoch": 27.02, + "learning_rate": 2.978461538461539e-06, + "loss": 0.0039, + "step": 14200 + }, + { + "epoch": 27.02, + "learning_rate": 2.9656410256410257e-06, + "loss": 0.0029, + "step": 14225 + }, + { + "epoch": 27.02, + "learning_rate": 2.952820512820513e-06, + "loss": 0.0034, + "step": 14250 + }, + { + "epoch": 27.02, + "learning_rate": 2.9400000000000002e-06, + "loss": 0.0038, + "step": 14275 + }, + { + "epoch": 27.02, + "learning_rate": 2.9271794871794875e-06, + "loss": 0.0022, + "step": 14300 + }, + { + "epoch": 27.03, + "learning_rate": 2.9143589743589747e-06, + "loss": 0.0027, + "step": 14325 + }, + { + "epoch": 28.0, + "learning_rate": 2.9015384615384615e-06, + "loss": 0.0215, + "step": 14350 + }, + { + "epoch": 28.0, + "learning_rate": 2.888717948717949e-06, + "loss": 0.0378, + "step": 14375 + }, + { + "epoch": 28.0, + "learning_rate": 2.875897435897436e-06, + "loss": 0.0323, + "step": 14400 + }, + { + "epoch": 28.0, + "learning_rate": 2.8630769230769233e-06, + "loss": 0.0358, + "step": 14425 + }, + { + "epoch": 28.01, + "learning_rate": 2.8502564102564105e-06, + "loss": 0.0286, + "step": 14450 + }, + { + "epoch": 28.01, + "learning_rate": 2.8374358974358978e-06, + "loss": 0.0288, + "step": 14475 + }, + { + "epoch": 28.01, + "learning_rate": 2.824615384615385e-06, + "loss": 0.0295, + "step": 14500 + }, + { + "epoch": 28.01, + "learning_rate": 2.811794871794872e-06, + "loss": 0.0294, + "step": 14525 + }, + { + "epoch": 28.01, + "learning_rate": 2.798974358974359e-06, + "loss": 0.0207, + "step": 14550 + }, + { + "epoch": 28.01, + "learning_rate": 2.7861538461538464e-06, + "loss": 0.0086, + "step": 14575 + }, + { + "epoch": 28.01, + "learning_rate": 2.7733333333333336e-06, + "loss": 0.0077, + "step": 14600 + }, + { + "epoch": 28.01, + "learning_rate": 2.760512820512821e-06, + "loss": 0.0085, + "step": 14625 + }, + { + "epoch": 28.02, + "learning_rate": 2.7476923076923077e-06, + "loss": 0.0118, + "step": 14650 + }, + { + "epoch": 28.02, + "learning_rate": 2.734871794871795e-06, + "loss": 0.0099, + "step": 14675 + }, + { + "epoch": 28.02, + "learning_rate": 2.722051282051282e-06, + "loss": 0.009, + "step": 14700 + }, + { + "epoch": 28.02, + "learning_rate": 2.7092307692307694e-06, + "loss": 0.0104, + "step": 14725 + }, + { + "epoch": 28.02, + "learning_rate": 2.6964102564102567e-06, + "loss": 0.0161, + "step": 14750 + }, + { + "epoch": 28.02, + "learning_rate": 2.683589743589744e-06, + "loss": 0.0189, + "step": 14775 + }, + { + "epoch": 28.02, + "learning_rate": 2.670769230769231e-06, + "loss": 0.0241, + "step": 14800 + }, + { + "epoch": 28.02, + "learning_rate": 2.657948717948718e-06, + "loss": 0.0196, + "step": 14825 + }, + { + "epoch": 29.0, + "learning_rate": 2.6451282051282052e-06, + "loss": 0.0221, + "step": 14850 + }, + { + "epoch": 29.0, + "learning_rate": 2.6323076923076925e-06, + "loss": 0.032, + "step": 14875 + }, + { + "epoch": 29.0, + "learning_rate": 2.6194871794871797e-06, + "loss": 0.0248, + "step": 14900 + }, + { + "epoch": 29.0, + "learning_rate": 2.606666666666667e-06, + "loss": 0.0244, + "step": 14925 + }, + { + "epoch": 29.01, + "learning_rate": 2.593846153846154e-06, + "loss": 0.0348, + "step": 14950 + }, + { + "epoch": 29.01, + "learning_rate": 2.581025641025641e-06, + "loss": 0.0365, + "step": 14975 + }, + { + "epoch": 29.01, + "learning_rate": 2.5682051282051283e-06, + "loss": 0.0337, + "step": 15000 + }, + { + "epoch": 29.01, + "eval_loss": 0.15532256662845612, + "eval_runtime": 6199.8735, + "eval_samples_per_second": 2.642, + "eval_steps_per_second": 0.165, + "eval_wer": 5.3655330094312665, + "step": 15000 + }, + { + "epoch": 29.01, + "learning_rate": 2.5553846153846155e-06, + "loss": 0.0376, + "step": 15025 + }, + { + "epoch": 29.01, + "learning_rate": 2.542564102564103e-06, + "loss": 0.0373, + "step": 15050 + }, + { + "epoch": 29.01, + "learning_rate": 2.52974358974359e-06, + "loss": 0.0446, + "step": 15075 + }, + { + "epoch": 29.01, + "learning_rate": 2.5169230769230773e-06, + "loss": 0.0335, + "step": 15100 + }, + { + "epoch": 29.01, + "learning_rate": 2.5041025641025645e-06, + "loss": 0.0398, + "step": 15125 + }, + { + "epoch": 29.02, + "learning_rate": 2.4912820512820514e-06, + "loss": 0.0346, + "step": 15150 + }, + { + "epoch": 29.02, + "learning_rate": 2.4784615384615386e-06, + "loss": 0.0316, + "step": 15175 + }, + { + "epoch": 29.02, + "learning_rate": 2.465641025641026e-06, + "loss": 0.0329, + "step": 15200 + }, + { + "epoch": 29.02, + "learning_rate": 2.452820512820513e-06, + "loss": 0.0326, + "step": 15225 + }, + { + "epoch": 29.02, + "learning_rate": 2.4400000000000004e-06, + "loss": 0.0413, + "step": 15250 + }, + { + "epoch": 29.02, + "learning_rate": 2.427179487179487e-06, + "loss": 0.0427, + "step": 15275 + }, + { + "epoch": 29.02, + "learning_rate": 2.4143589743589744e-06, + "loss": 0.0339, + "step": 15300 + }, + { + "epoch": 29.02, + "learning_rate": 2.4015384615384617e-06, + "loss": 0.0337, + "step": 15325 + }, + { + "epoch": 29.03, + "learning_rate": 2.388717948717949e-06, + "loss": 0.034, + "step": 15350 + }, + { + "epoch": 30.0, + "learning_rate": 2.375897435897436e-06, + "loss": 0.024, + "step": 15375 + }, + { + "epoch": 30.0, + "learning_rate": 2.3630769230769234e-06, + "loss": 0.0149, + "step": 15400 + }, + { + "epoch": 30.0, + "learning_rate": 2.3502564102564102e-06, + "loss": 0.0116, + "step": 15425 + }, + { + "epoch": 30.0, + "learning_rate": 2.3374358974358975e-06, + "loss": 0.0112, + "step": 15450 + }, + { + "epoch": 30.01, + "learning_rate": 2.3246153846153847e-06, + "loss": 0.0137, + "step": 15475 + }, + { + "epoch": 30.01, + "learning_rate": 2.311794871794872e-06, + "loss": 0.0113, + "step": 15500 + }, + { + "epoch": 30.01, + "learning_rate": 2.2989743589743592e-06, + "loss": 0.011, + "step": 15525 + }, + { + "epoch": 30.01, + "learning_rate": 2.2861538461538465e-06, + "loss": 0.0113, + "step": 15550 + }, + { + "epoch": 30.01, + "learning_rate": 2.2733333333333333e-06, + "loss": 0.0101, + "step": 15575 + }, + { + "epoch": 30.01, + "learning_rate": 2.2605128205128206e-06, + "loss": 0.0108, + "step": 15600 + }, + { + "epoch": 30.01, + "learning_rate": 2.247692307692308e-06, + "loss": 0.0103, + "step": 15625 + }, + { + "epoch": 30.01, + "learning_rate": 2.234871794871795e-06, + "loss": 0.0104, + "step": 15650 + }, + { + "epoch": 30.02, + "learning_rate": 2.2220512820512823e-06, + "loss": 0.011, + "step": 15675 + }, + { + "epoch": 30.02, + "learning_rate": 2.2092307692307695e-06, + "loss": 0.0096, + "step": 15700 + }, + { + "epoch": 30.02, + "learning_rate": 2.1964102564102564e-06, + "loss": 0.0101, + "step": 15725 + }, + { + "epoch": 30.02, + "learning_rate": 2.1835897435897436e-06, + "loss": 0.0086, + "step": 15750 + }, + { + "epoch": 30.02, + "learning_rate": 2.170769230769231e-06, + "loss": 0.0088, + "step": 15775 + }, + { + "epoch": 30.02, + "learning_rate": 2.157948717948718e-06, + "loss": 0.0087, + "step": 15800 + }, + { + "epoch": 30.02, + "learning_rate": 2.1451282051282054e-06, + "loss": 0.0079, + "step": 15825 + }, + { + "epoch": 30.02, + "learning_rate": 2.1323076923076926e-06, + "loss": 0.009, + "step": 15850 + }, + { + "epoch": 31.0, + "learning_rate": 2.1194871794871794e-06, + "loss": 0.0072, + "step": 15875 + }, + { + "epoch": 31.0, + "learning_rate": 2.1066666666666667e-06, + "loss": 0.0103, + "step": 15900 + }, + { + "epoch": 31.0, + "learning_rate": 2.093846153846154e-06, + "loss": 0.0135, + "step": 15925 + }, + { + "epoch": 31.0, + "learning_rate": 2.081025641025641e-06, + "loss": 0.0111, + "step": 15950 + }, + { + "epoch": 31.01, + "learning_rate": 2.0682051282051284e-06, + "loss": 0.0112, + "step": 15975 + }, + { + "epoch": 31.01, + "learning_rate": 2.0553846153846157e-06, + "loss": 0.0108, + "step": 16000 + }, + { + "epoch": 31.01, + "eval_loss": 0.14759808778762817, + "eval_runtime": 6075.5393, + "eval_samples_per_second": 2.696, + "eval_steps_per_second": 0.169, + "eval_wer": 4.915118605315804, + "step": 16000 + }, + { + "epoch": 31.01, + "learning_rate": 2.0425641025641025e-06, + "loss": 0.0104, + "step": 16025 + }, + { + "epoch": 31.01, + "learning_rate": 2.0297435897435897e-06, + "loss": 0.0115, + "step": 16050 + }, + { + "epoch": 31.01, + "learning_rate": 2.016923076923077e-06, + "loss": 0.0105, + "step": 16075 + }, + { + "epoch": 31.01, + "learning_rate": 2.0041025641025642e-06, + "loss": 0.0113, + "step": 16100 + }, + { + "epoch": 31.01, + "learning_rate": 1.9912820512820515e-06, + "loss": 0.0099, + "step": 16125 + }, + { + "epoch": 31.01, + "learning_rate": 1.9784615384615387e-06, + "loss": 0.0115, + "step": 16150 + }, + { + "epoch": 31.02, + "learning_rate": 1.9656410256410256e-06, + "loss": 0.0099, + "step": 16175 + }, + { + "epoch": 31.02, + "learning_rate": 1.952820512820513e-06, + "loss": 0.0093, + "step": 16200 + }, + { + "epoch": 31.02, + "learning_rate": 1.94e-06, + "loss": 0.0105, + "step": 16225 + }, + { + "epoch": 31.02, + "learning_rate": 1.9271794871794873e-06, + "loss": 0.009, + "step": 16250 + }, + { + "epoch": 31.02, + "learning_rate": 1.9143589743589746e-06, + "loss": 0.0103, + "step": 16275 + }, + { + "epoch": 31.02, + "learning_rate": 1.9015384615384616e-06, + "loss": 0.0111, + "step": 16300 + }, + { + "epoch": 31.02, + "learning_rate": 1.8887179487179488e-06, + "loss": 0.0108, + "step": 16325 + }, + { + "epoch": 31.02, + "learning_rate": 1.875897435897436e-06, + "loss": 0.0101, + "step": 16350 + }, + { + "epoch": 31.03, + "learning_rate": 1.8630769230769233e-06, + "loss": 0.0123, + "step": 16375 + }, + { + "epoch": 32.0, + "learning_rate": 1.8502564102564106e-06, + "loss": 0.0091, + "step": 16400 + }, + { + "epoch": 32.0, + "learning_rate": 1.8374358974358974e-06, + "loss": 0.01, + "step": 16425 + }, + { + "epoch": 32.0, + "learning_rate": 1.8246153846153847e-06, + "loss": 0.0114, + "step": 16450 + }, + { + "epoch": 32.0, + "learning_rate": 1.811794871794872e-06, + "loss": 0.0064, + "step": 16475 + }, + { + "epoch": 32.01, + "learning_rate": 1.7989743589743592e-06, + "loss": 0.0102, + "step": 16500 + }, + { + "epoch": 32.01, + "learning_rate": 1.7861538461538464e-06, + "loss": 0.0097, + "step": 16525 + }, + { + "epoch": 32.01, + "learning_rate": 1.7733333333333336e-06, + "loss": 0.0071, + "step": 16550 + }, + { + "epoch": 32.01, + "learning_rate": 1.7605128205128205e-06, + "loss": 0.0083, + "step": 16575 + }, + { + "epoch": 32.01, + "learning_rate": 1.7476923076923077e-06, + "loss": 0.0096, + "step": 16600 + }, + { + "epoch": 32.01, + "learning_rate": 1.734871794871795e-06, + "loss": 0.0085, + "step": 16625 + }, + { + "epoch": 32.01, + "learning_rate": 1.7220512820512822e-06, + "loss": 0.0071, + "step": 16650 + }, + { + "epoch": 32.01, + "learning_rate": 1.7092307692307695e-06, + "loss": 0.0107, + "step": 16675 + }, + { + "epoch": 32.02, + "learning_rate": 1.6964102564102567e-06, + "loss": 0.0086, + "step": 16700 + }, + { + "epoch": 32.02, + "learning_rate": 1.683589743589744e-06, + "loss": 0.0106, + "step": 16725 + }, + { + "epoch": 32.02, + "learning_rate": 1.6707692307692308e-06, + "loss": 0.0097, + "step": 16750 + }, + { + "epoch": 32.02, + "learning_rate": 1.657948717948718e-06, + "loss": 0.01, + "step": 16775 + }, + { + "epoch": 32.02, + "learning_rate": 1.6451282051282053e-06, + "loss": 0.0084, + "step": 16800 + }, + { + "epoch": 32.02, + "learning_rate": 1.6323076923076925e-06, + "loss": 0.0087, + "step": 16825 + }, + { + "epoch": 32.02, + "learning_rate": 1.6194871794871798e-06, + "loss": 0.0085, + "step": 16850 + }, + { + "epoch": 32.02, + "learning_rate": 1.606666666666667e-06, + "loss": 0.0097, + "step": 16875 + }, + { + "epoch": 33.0, + "learning_rate": 1.5938461538461538e-06, + "loss": 0.0132, + "step": 16900 + }, + { + "epoch": 33.0, + "learning_rate": 1.581025641025641e-06, + "loss": 0.031, + "step": 16925 + }, + { + "epoch": 33.0, + "learning_rate": 1.5682051282051283e-06, + "loss": 0.0266, + "step": 16950 + }, + { + "epoch": 33.0, + "learning_rate": 1.5553846153846156e-06, + "loss": 0.0148, + "step": 16975 + }, + { + "epoch": 33.01, + "learning_rate": 1.5425641025641028e-06, + "loss": 0.0194, + "step": 17000 + }, + { + "epoch": 33.01, + "eval_loss": 0.13962695002555847, + "eval_runtime": 5922.6519, + "eval_samples_per_second": 2.766, + "eval_steps_per_second": 0.173, + "eval_wer": 4.847670763075165, + "step": 17000 + }, + { + "epoch": 33.01, + "learning_rate": 1.52974358974359e-06, + "loss": 0.0213, + "step": 17025 + }, + { + "epoch": 33.01, + "learning_rate": 1.516923076923077e-06, + "loss": 0.0101, + "step": 17050 + }, + { + "epoch": 33.01, + "learning_rate": 1.5041025641025642e-06, + "loss": 0.0073, + "step": 17075 + }, + { + "epoch": 33.01, + "learning_rate": 1.4912820512820514e-06, + "loss": 0.0162, + "step": 17100 + }, + { + "epoch": 33.01, + "learning_rate": 1.4784615384615387e-06, + "loss": 0.0206, + "step": 17125 + }, + { + "epoch": 33.01, + "learning_rate": 1.465641025641026e-06, + "loss": 0.0162, + "step": 17150 + }, + { + "epoch": 33.01, + "learning_rate": 1.4528205128205132e-06, + "loss": 0.0202, + "step": 17175 + }, + { + "epoch": 33.02, + "learning_rate": 1.44e-06, + "loss": 0.0227, + "step": 17200 + }, + { + "epoch": 33.02, + "learning_rate": 1.4271794871794872e-06, + "loss": 0.0172, + "step": 17225 + }, + { + "epoch": 33.02, + "learning_rate": 1.4143589743589745e-06, + "loss": 0.023, + "step": 17250 + }, + { + "epoch": 33.02, + "learning_rate": 1.4015384615384617e-06, + "loss": 0.0228, + "step": 17275 + }, + { + "epoch": 33.02, + "learning_rate": 1.388717948717949e-06, + "loss": 0.022, + "step": 17300 + }, + { + "epoch": 33.02, + "learning_rate": 1.3758974358974362e-06, + "loss": 0.0176, + "step": 17325 + }, + { + "epoch": 33.02, + "learning_rate": 1.363076923076923e-06, + "loss": 0.0172, + "step": 17350 + }, + { + "epoch": 33.02, + "learning_rate": 1.3502564102564103e-06, + "loss": 0.0235, + "step": 17375 + }, + { + "epoch": 33.03, + "learning_rate": 1.3374358974358975e-06, + "loss": 0.0217, + "step": 17400 + }, + { + "epoch": 34.0, + "learning_rate": 1.3246153846153848e-06, + "loss": 0.0217, + "step": 17425 + }, + { + "epoch": 34.0, + "learning_rate": 1.311794871794872e-06, + "loss": 0.021, + "step": 17450 + }, + { + "epoch": 34.0, + "learning_rate": 1.2989743589743593e-06, + "loss": 0.0175, + "step": 17475 + }, + { + "epoch": 34.0, + "learning_rate": 1.286153846153846e-06, + "loss": 0.017, + "step": 17500 + }, + { + "epoch": 34.01, + "learning_rate": 1.2733333333333334e-06, + "loss": 0.0162, + "step": 17525 + }, + { + "epoch": 34.01, + "learning_rate": 1.2605128205128206e-06, + "loss": 0.0167, + "step": 17550 + }, + { + "epoch": 34.01, + "learning_rate": 1.2476923076923078e-06, + "loss": 0.0157, + "step": 17575 + }, + { + "epoch": 34.01, + "learning_rate": 1.2353846153846155e-06, + "loss": 0.016, + "step": 17600 + }, + { + "epoch": 34.01, + "learning_rate": 1.2225641025641025e-06, + "loss": 0.0164, + "step": 17625 + }, + { + "epoch": 34.01, + "learning_rate": 1.2097435897435898e-06, + "loss": 0.0158, + "step": 17650 + }, + { + "epoch": 34.01, + "learning_rate": 1.196923076923077e-06, + "loss": 0.0172, + "step": 17675 + }, + { + "epoch": 34.01, + "learning_rate": 1.184102564102564e-06, + "loss": 0.0139, + "step": 17700 + }, + { + "epoch": 34.02, + "learning_rate": 1.1712820512820513e-06, + "loss": 0.0167, + "step": 17725 + }, + { + "epoch": 34.02, + "learning_rate": 1.1584615384615385e-06, + "loss": 0.0158, + "step": 17750 + }, + { + "epoch": 34.02, + "learning_rate": 1.1456410256410256e-06, + "loss": 0.0161, + "step": 17775 + }, + { + "epoch": 34.02, + "learning_rate": 1.1328205128205128e-06, + "loss": 0.0146, + "step": 17800 + }, + { + "epoch": 34.02, + "learning_rate": 1.12e-06, + "loss": 0.0148, + "step": 17825 + }, + { + "epoch": 34.02, + "learning_rate": 1.1071794871794873e-06, + "loss": 0.0152, + "step": 17850 + }, + { + "epoch": 34.02, + "learning_rate": 1.0943589743589744e-06, + "loss": 0.0233, + "step": 17875 + }, + { + "epoch": 34.02, + "learning_rate": 1.0815384615384616e-06, + "loss": 0.0312, + "step": 17900 + }, + { + "epoch": 35.0, + "learning_rate": 1.0687179487179489e-06, + "loss": 0.0318, + "step": 17925 + }, + { + "epoch": 35.0, + "learning_rate": 1.0558974358974359e-06, + "loss": 0.062, + "step": 17950 + }, + { + "epoch": 35.0, + "learning_rate": 1.0430769230769231e-06, + "loss": 0.0541, + "step": 17975 + }, + { + "epoch": 35.0, + "learning_rate": 1.0302564102564104e-06, + "loss": 0.0472, + "step": 18000 + }, + { + "epoch": 35.0, + "eval_loss": 0.12018702924251556, + "eval_runtime": 6083.5102, + "eval_samples_per_second": 2.693, + "eval_steps_per_second": 0.168, + "eval_wer": 4.8716776221777645, + "step": 18000 + }, + { + "epoch": 35.01, + "learning_rate": 1.0174358974358974e-06, + "loss": 0.0534, + "step": 18025 + }, + { + "epoch": 35.01, + "learning_rate": 1.0046153846153847e-06, + "loss": 0.0493, + "step": 18050 + }, + { + "epoch": 35.01, + "learning_rate": 9.91794871794872e-07, + "loss": 0.0495, + "step": 18075 + }, + { + "epoch": 35.01, + "learning_rate": 9.78974358974359e-07, + "loss": 0.0429, + "step": 18100 + }, + { + "epoch": 35.01, + "learning_rate": 9.661538461538462e-07, + "loss": 0.0501, + "step": 18125 + }, + { + "epoch": 35.01, + "learning_rate": 9.533333333333335e-07, + "loss": 0.0566, + "step": 18150 + }, + { + "epoch": 35.01, + "learning_rate": 9.405128205128206e-07, + "loss": 0.0519, + "step": 18175 + }, + { + "epoch": 35.01, + "learning_rate": 9.276923076923077e-07, + "loss": 0.0455, + "step": 18200 + }, + { + "epoch": 35.02, + "learning_rate": 9.14871794871795e-07, + "loss": 0.053, + "step": 18225 + }, + { + "epoch": 35.02, + "learning_rate": 9.020512820512821e-07, + "loss": 0.0572, + "step": 18250 + }, + { + "epoch": 35.02, + "learning_rate": 8.892307692307693e-07, + "loss": 0.0556, + "step": 18275 + }, + { + "epoch": 35.02, + "learning_rate": 8.764102564102565e-07, + "loss": 0.054, + "step": 18300 + }, + { + "epoch": 35.02, + "learning_rate": 8.635897435897437e-07, + "loss": 0.0542, + "step": 18325 + }, + { + "epoch": 35.02, + "learning_rate": 8.507692307692308e-07, + "loss": 0.0505, + "step": 18350 + }, + { + "epoch": 35.02, + "learning_rate": 8.37948717948718e-07, + "loss": 0.0504, + "step": 18375 + }, + { + "epoch": 35.02, + "learning_rate": 8.251282051282052e-07, + "loss": 0.0431, + "step": 18400 + }, + { + "epoch": 35.03, + "learning_rate": 8.123076923076923e-07, + "loss": 0.0522, + "step": 18425 + }, + { + "epoch": 36.0, + "learning_rate": 7.994871794871796e-07, + "loss": 0.0208, + "step": 18450 + }, + { + "epoch": 36.0, + "learning_rate": 7.866666666666667e-07, + "loss": 0.0096, + "step": 18475 + }, + { + "epoch": 36.0, + "learning_rate": 7.738461538461539e-07, + "loss": 0.0078, + "step": 18500 + }, + { + "epoch": 36.0, + "learning_rate": 7.610256410256411e-07, + "loss": 0.0076, + "step": 18525 + }, + { + "epoch": 36.01, + "learning_rate": 7.482051282051283e-07, + "loss": 0.0052, + "step": 18550 + }, + { + "epoch": 36.01, + "learning_rate": 7.353846153846154e-07, + "loss": 0.0068, + "step": 18575 + }, + { + "epoch": 36.01, + "learning_rate": 7.225641025641026e-07, + "loss": 0.0059, + "step": 18600 + }, + { + "epoch": 36.01, + "learning_rate": 7.097435897435898e-07, + "loss": 0.0059, + "step": 18625 + }, + { + "epoch": 36.01, + "learning_rate": 6.969230769230769e-07, + "loss": 0.0053, + "step": 18650 + }, + { + "epoch": 36.01, + "learning_rate": 6.841025641025642e-07, + "loss": 0.0053, + "step": 18675 + }, + { + "epoch": 36.01, + "learning_rate": 6.712820512820513e-07, + "loss": 0.0068, + "step": 18700 + }, + { + "epoch": 36.01, + "learning_rate": 6.584615384615385e-07, + "loss": 0.0052, + "step": 18725 + }, + { + "epoch": 36.02, + "learning_rate": 6.456410256410257e-07, + "loss": 0.0048, + "step": 18750 + }, + { + "epoch": 36.02, + "learning_rate": 6.328205128205129e-07, + "loss": 0.0059, + "step": 18775 + }, + { + "epoch": 36.02, + "learning_rate": 6.200000000000001e-07, + "loss": 0.0047, + "step": 18800 + }, + { + "epoch": 36.02, + "learning_rate": 6.071794871794872e-07, + "loss": 0.0047, + "step": 18825 + }, + { + "epoch": 36.02, + "learning_rate": 5.943589743589744e-07, + "loss": 0.0058, + "step": 18850 + }, + { + "epoch": 36.02, + "learning_rate": 5.815384615384616e-07, + "loss": 0.0059, + "step": 18875 + }, + { + "epoch": 36.02, + "learning_rate": 5.687179487179488e-07, + "loss": 0.005, + "step": 18900 + }, + { + "epoch": 36.02, + "learning_rate": 5.558974358974359e-07, + "loss": 0.0059, + "step": 18925 + }, + { + "epoch": 37.0, + "learning_rate": 5.430769230769232e-07, + "loss": 0.0101, + "step": 18950 + }, + { + "epoch": 37.0, + "learning_rate": 5.302564102564103e-07, + "loss": 0.0257, + "step": 18975 + }, + { + "epoch": 37.0, + "learning_rate": 5.174358974358974e-07, + "loss": 0.0401, + "step": 19000 + }, + { + "epoch": 37.0, + "eval_loss": 0.14943751692771912, + "eval_runtime": 5977.2624, + "eval_samples_per_second": 2.74, + "eval_steps_per_second": 0.171, + "eval_wer": 4.671620462989425, + "step": 19000 + }, + { + "epoch": 37.0, + "learning_rate": 5.046153846153847e-07, + "loss": 0.0268, + "step": 19025 + }, + { + "epoch": 37.01, + "learning_rate": 4.917948717948718e-07, + "loss": 0.0138, + "step": 19050 + }, + { + "epoch": 37.01, + "learning_rate": 4.78974358974359e-07, + "loss": 0.0184, + "step": 19075 + }, + { + "epoch": 37.01, + "learning_rate": 4.661538461538462e-07, + "loss": 0.0114, + "step": 19100 + }, + { + "epoch": 37.01, + "learning_rate": 4.533333333333334e-07, + "loss": 0.0128, + "step": 19125 + }, + { + "epoch": 37.01, + "learning_rate": 4.4051282051282056e-07, + "loss": 0.022, + "step": 19150 + }, + { + "epoch": 37.01, + "learning_rate": 4.276923076923077e-07, + "loss": 0.0146, + "step": 19175 + }, + { + "epoch": 37.01, + "learning_rate": 4.1487179487179495e-07, + "loss": 0.0105, + "step": 19200 + }, + { + "epoch": 37.01, + "learning_rate": 4.020512820512821e-07, + "loss": 0.008, + "step": 19225 + }, + { + "epoch": 37.02, + "learning_rate": 3.8923076923076924e-07, + "loss": 0.009, + "step": 19250 + }, + { + "epoch": 37.02, + "learning_rate": 3.764102564102565e-07, + "loss": 0.0081, + "step": 19275 + }, + { + "epoch": 37.02, + "learning_rate": 3.6358974358974363e-07, + "loss": 0.0077, + "step": 19300 + }, + { + "epoch": 37.02, + "learning_rate": 3.5076923076923077e-07, + "loss": 0.008, + "step": 19325 + }, + { + "epoch": 37.02, + "learning_rate": 3.37948717948718e-07, + "loss": 0.0066, + "step": 19350 + }, + { + "epoch": 37.02, + "learning_rate": 3.2512820512820516e-07, + "loss": 0.0073, + "step": 19375 + }, + { + "epoch": 37.02, + "learning_rate": 3.1230769230769236e-07, + "loss": 0.0072, + "step": 19400 + }, + { + "epoch": 37.02, + "learning_rate": 2.994871794871795e-07, + "loss": 0.007, + "step": 19425 + }, + { + "epoch": 37.03, + "learning_rate": 2.866666666666667e-07, + "loss": 0.0071, + "step": 19450 + }, + { + "epoch": 38.0, + "learning_rate": 2.738461538461539e-07, + "loss": 0.0404, + "step": 19475 + }, + { + "epoch": 38.0, + "learning_rate": 2.6102564102564103e-07, + "loss": 0.0755, + "step": 19500 + }, + { + "epoch": 38.0, + "learning_rate": 2.482051282051282e-07, + "loss": 0.052, + "step": 19525 + }, + { + "epoch": 38.0, + "learning_rate": 2.3538461538461542e-07, + "loss": 0.0443, + "step": 19550 + }, + { + "epoch": 38.01, + "learning_rate": 2.2256410256410256e-07, + "loss": 0.044, + "step": 19575 + }, + { + "epoch": 38.01, + "learning_rate": 2.0974358974358976e-07, + "loss": 0.0333, + "step": 19600 + }, + { + "epoch": 38.01, + "learning_rate": 1.9692307692307695e-07, + "loss": 0.039, + "step": 19625 + }, + { + "epoch": 38.01, + "learning_rate": 1.841025641025641e-07, + "loss": 0.0591, + "step": 19650 + }, + { + "epoch": 38.01, + "learning_rate": 1.712820512820513e-07, + "loss": 0.047, + "step": 19675 + }, + { + "epoch": 38.01, + "learning_rate": 1.5846153846153848e-07, + "loss": 0.0527, + "step": 19700 + }, + { + "epoch": 38.01, + "learning_rate": 1.4564102564102565e-07, + "loss": 0.0927, + "step": 19725 + }, + { + "epoch": 38.01, + "learning_rate": 1.3333333333333336e-07, + "loss": 0.0573, + "step": 19750 + }, + { + "epoch": 38.02, + "learning_rate": 1.2051282051282053e-07, + "loss": 0.0523, + "step": 19775 + }, + { + "epoch": 38.02, + "learning_rate": 1.076923076923077e-07, + "loss": 0.0516, + "step": 19800 + }, + { + "epoch": 38.02, + "learning_rate": 9.487179487179488e-08, + "loss": 0.043, + "step": 19825 + }, + { + "epoch": 38.02, + "learning_rate": 8.205128205128206e-08, + "loss": 0.0266, + "step": 19850 + }, + { + "epoch": 38.02, + "learning_rate": 6.923076923076924e-08, + "loss": 0.0516, + "step": 19875 + }, + { + "epoch": 38.02, + "learning_rate": 5.641025641025642e-08, + "loss": 0.0371, + "step": 19900 + }, + { + "epoch": 38.02, + "learning_rate": 4.358974358974359e-08, + "loss": 0.0361, + "step": 19925 + }, + { + "epoch": 38.02, + "learning_rate": 3.076923076923077e-08, + "loss": 0.0535, + "step": 19950 + }, + { + "epoch": 39.0, + "learning_rate": 1.794871794871795e-08, + "loss": 0.0376, + "step": 19975 + }, + { + "epoch": 39.0, + "learning_rate": 5.128205128205129e-09, + "loss": 0.0127, + "step": 20000 + }, + { + "epoch": 39.0, + "eval_loss": 0.11867784708738327, + "eval_runtime": 5992.6904, + "eval_samples_per_second": 2.733, + "eval_steps_per_second": 0.171, + "eval_wer": 4.727636467562161, + "step": 20000 + }, + { + "epoch": 39.0, + "step": 20000, + "total_flos": 2.7169964772655104e+21, + "train_loss": 0.041109147554636, + "train_runtime": 463477.2931, + "train_samples_per_second": 2.762, + "train_steps_per_second": 0.043 + } + ], + "logging_steps": 25, + "max_steps": 20000, + "num_train_epochs": 9223372036854775807, + "save_steps": 1000, + "total_flos": 2.7169964772655104e+21, + "trial_name": null, + "trial_params": null +}