{ "best_metric": 4.678688966300353, "best_model_checkpoint": "./checkpoint-2000", "epoch": 40.73319755600814, "eval_steps": 1000, "global_step": 20000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.05, "learning_rate": 5.000000000000001e-07, "loss": 0.7194, "step": 25 }, { "epoch": 0.1, "learning_rate": 1.0000000000000002e-06, "loss": 0.4266, "step": 50 }, { "epoch": 0.15, "learning_rate": 1.5e-06, "loss": 0.1605, "step": 75 }, { "epoch": 0.2, "learning_rate": 2.0000000000000003e-06, "loss": 0.1402, "step": 100 }, { "epoch": 0.25, "learning_rate": 2.5e-06, "loss": 0.1307, "step": 125 }, { "epoch": 0.31, "learning_rate": 3e-06, "loss": 0.1394, "step": 150 }, { "epoch": 0.36, "learning_rate": 3.5e-06, "loss": 0.1255, "step": 175 }, { "epoch": 0.41, "learning_rate": 4.000000000000001e-06, "loss": 0.1285, "step": 200 }, { "epoch": 0.46, "learning_rate": 4.5e-06, "loss": 0.1278, "step": 225 }, { "epoch": 0.51, "learning_rate": 5e-06, "loss": 0.132, "step": 250 }, { "epoch": 0.56, "learning_rate": 5.500000000000001e-06, "loss": 0.1341, "step": 275 }, { "epoch": 0.61, "learning_rate": 6e-06, "loss": 0.134, "step": 300 }, { "epoch": 0.66, "learning_rate": 6.5000000000000004e-06, "loss": 0.1292, "step": 325 }, { "epoch": 0.71, "learning_rate": 7e-06, "loss": 0.1295, "step": 350 }, { "epoch": 0.76, "learning_rate": 7.500000000000001e-06, "loss": 0.1344, "step": 375 }, { "epoch": 0.81, "learning_rate": 8.000000000000001e-06, "loss": 0.1223, "step": 400 }, { "epoch": 0.87, "learning_rate": 8.5e-06, "loss": 0.1328, "step": 425 }, { "epoch": 0.92, "learning_rate": 9e-06, "loss": 0.1241, "step": 450 }, { "epoch": 0.97, "learning_rate": 9.5e-06, "loss": 0.129, "step": 475 }, { "epoch": 1.02, "learning_rate": 1e-05, "loss": 0.1137, "step": 500 }, { "epoch": 1.07, "learning_rate": 9.987179487179488e-06, "loss": 0.0836, "step": 525 }, { "epoch": 1.12, "learning_rate": 9.974358974358974e-06, "loss": 0.0934, "step": 550 }, { "epoch": 1.17, "learning_rate": 9.961538461538463e-06, "loss": 0.0911, "step": 575 }, { "epoch": 1.22, "learning_rate": 9.94871794871795e-06, "loss": 0.0887, "step": 600 }, { "epoch": 1.27, "learning_rate": 9.935897435897437e-06, "loss": 0.0858, "step": 625 }, { "epoch": 1.32, "learning_rate": 9.923076923076923e-06, "loss": 0.0924, "step": 650 }, { "epoch": 1.37, "learning_rate": 9.910256410256412e-06, "loss": 0.0896, "step": 675 }, { "epoch": 1.43, "learning_rate": 9.897435897435899e-06, "loss": 0.0857, "step": 700 }, { "epoch": 1.48, "learning_rate": 9.884615384615386e-06, "loss": 0.0964, "step": 725 }, { "epoch": 1.53, "learning_rate": 9.871794871794872e-06, "loss": 0.0941, "step": 750 }, { "epoch": 1.58, "learning_rate": 9.858974358974361e-06, "loss": 0.0948, "step": 775 }, { "epoch": 1.63, "learning_rate": 9.846153846153848e-06, "loss": 0.0909, "step": 800 }, { "epoch": 1.68, "learning_rate": 9.833333333333333e-06, "loss": 0.089, "step": 825 }, { "epoch": 1.73, "learning_rate": 9.820512820512821e-06, "loss": 0.091, "step": 850 }, { "epoch": 1.78, "learning_rate": 9.807692307692308e-06, "loss": 0.0986, "step": 875 }, { "epoch": 1.83, "learning_rate": 9.794871794871795e-06, "loss": 0.0905, "step": 900 }, { "epoch": 1.88, "learning_rate": 9.782051282051282e-06, "loss": 0.0962, "step": 925 }, { "epoch": 1.93, "learning_rate": 9.76923076923077e-06, "loss": 0.0918, "step": 950 }, { "epoch": 1.99, "learning_rate": 9.756410256410257e-06, "loss": 0.0967, "step": 975 }, { "epoch": 2.04, "learning_rate": 9.743589743589744e-06, "loss": 0.058, "step": 1000 }, { "epoch": 2.04, "eval_loss": 0.15402629971504211, "eval_runtime": 3271.1965, "eval_samples_per_second": 4.802, "eval_steps_per_second": 0.3, "eval_wer": 4.685104280939459, "step": 1000 }, { "epoch": 2.09, "learning_rate": 9.730769230769231e-06, "loss": 0.0436, "step": 1025 }, { "epoch": 2.14, "learning_rate": 9.71794871794872e-06, "loss": 0.0451, "step": 1050 }, { "epoch": 2.19, "learning_rate": 9.705128205128206e-06, "loss": 0.0454, "step": 1075 }, { "epoch": 2.24, "learning_rate": 9.692307692307693e-06, "loss": 0.0449, "step": 1100 }, { "epoch": 2.29, "learning_rate": 9.67948717948718e-06, "loss": 0.0471, "step": 1125 }, { "epoch": 2.34, "learning_rate": 9.666666666666667e-06, "loss": 0.0428, "step": 1150 }, { "epoch": 2.39, "learning_rate": 9.653846153846155e-06, "loss": 0.0463, "step": 1175 }, { "epoch": 2.44, "learning_rate": 9.641025641025642e-06, "loss": 0.0462, "step": 1200 }, { "epoch": 2.49, "learning_rate": 9.628205128205129e-06, "loss": 0.0471, "step": 1225 }, { "epoch": 2.55, "learning_rate": 9.615384615384616e-06, "loss": 0.0487, "step": 1250 }, { "epoch": 2.6, "learning_rate": 9.602564102564104e-06, "loss": 0.0507, "step": 1275 }, { "epoch": 2.65, "learning_rate": 9.589743589743591e-06, "loss": 0.0484, "step": 1300 }, { "epoch": 2.7, "learning_rate": 9.576923076923078e-06, "loss": 0.05, "step": 1325 }, { "epoch": 2.75, "learning_rate": 9.564102564102565e-06, "loss": 0.0484, "step": 1350 }, { "epoch": 2.8, "learning_rate": 9.551282051282053e-06, "loss": 0.0478, "step": 1375 }, { "epoch": 2.85, "learning_rate": 9.53846153846154e-06, "loss": 0.049, "step": 1400 }, { "epoch": 2.9, "learning_rate": 9.525641025641025e-06, "loss": 0.0483, "step": 1425 }, { "epoch": 2.95, "learning_rate": 9.512820512820514e-06, "loss": 0.0484, "step": 1450 }, { "epoch": 3.0, "learning_rate": 9.5e-06, "loss": 0.0472, "step": 1475 }, { "epoch": 3.05, "learning_rate": 9.487179487179487e-06, "loss": 0.0247, "step": 1500 }, { "epoch": 3.11, "learning_rate": 9.474358974358974e-06, "loss": 0.0214, "step": 1525 }, { "epoch": 3.16, "learning_rate": 9.461538461538463e-06, "loss": 0.0225, "step": 1550 }, { "epoch": 3.21, "learning_rate": 9.44871794871795e-06, "loss": 0.0201, "step": 1575 }, { "epoch": 3.26, "learning_rate": 9.435897435897436e-06, "loss": 0.0221, "step": 1600 }, { "epoch": 3.31, "learning_rate": 9.423076923076923e-06, "loss": 0.0226, "step": 1625 }, { "epoch": 3.36, "learning_rate": 9.410256410256412e-06, "loss": 0.0228, "step": 1650 }, { "epoch": 3.41, "learning_rate": 9.397435897435899e-06, "loss": 0.0241, "step": 1675 }, { "epoch": 3.46, "learning_rate": 9.384615384615385e-06, "loss": 0.0264, "step": 1700 }, { "epoch": 3.51, "learning_rate": 9.371794871794872e-06, "loss": 0.0245, "step": 1725 }, { "epoch": 3.56, "learning_rate": 9.358974358974359e-06, "loss": 0.0247, "step": 1750 }, { "epoch": 3.62, "learning_rate": 9.346153846153847e-06, "loss": 0.0218, "step": 1775 }, { "epoch": 3.67, "learning_rate": 9.333333333333334e-06, "loss": 0.0266, "step": 1800 }, { "epoch": 3.72, "learning_rate": 9.320512820512821e-06, "loss": 0.0247, "step": 1825 }, { "epoch": 3.77, "learning_rate": 9.307692307692308e-06, "loss": 0.0271, "step": 1850 }, { "epoch": 3.82, "learning_rate": 9.294871794871796e-06, "loss": 0.0258, "step": 1875 }, { "epoch": 3.87, "learning_rate": 9.282051282051283e-06, "loss": 0.025, "step": 1900 }, { "epoch": 3.92, "learning_rate": 9.26923076923077e-06, "loss": 0.0286, "step": 1925 }, { "epoch": 3.97, "learning_rate": 9.256410256410257e-06, "loss": 0.0289, "step": 1950 }, { "epoch": 4.02, "learning_rate": 9.243589743589745e-06, "loss": 0.0217, "step": 1975 }, { "epoch": 4.07, "learning_rate": 9.230769230769232e-06, "loss": 0.0124, "step": 2000 }, { "epoch": 4.07, "eval_loss": 0.182881161570549, "eval_runtime": 3187.6234, "eval_samples_per_second": 4.928, "eval_steps_per_second": 0.308, "eval_wer": 4.678688966300353, "step": 2000 }, { "epoch": 4.12, "learning_rate": 9.217948717948717e-06, "loss": 0.0124, "step": 2025 }, { "epoch": 4.18, "learning_rate": 9.205128205128206e-06, "loss": 0.0125, "step": 2050 }, { "epoch": 4.23, "learning_rate": 9.192307692307693e-06, "loss": 0.013, "step": 2075 }, { "epoch": 4.28, "learning_rate": 9.17948717948718e-06, "loss": 0.013, "step": 2100 }, { "epoch": 4.33, "learning_rate": 9.166666666666666e-06, "loss": 0.0134, "step": 2125 }, { "epoch": 4.38, "learning_rate": 9.153846153846155e-06, "loss": 0.0121, "step": 2150 }, { "epoch": 4.43, "learning_rate": 9.141025641025642e-06, "loss": 0.0114, "step": 2175 }, { "epoch": 4.48, "learning_rate": 9.128205128205129e-06, "loss": 0.0134, "step": 2200 }, { "epoch": 4.53, "learning_rate": 9.115384615384615e-06, "loss": 0.0132, "step": 2225 }, { "epoch": 4.58, "learning_rate": 9.102564102564104e-06, "loss": 0.0139, "step": 2250 }, { "epoch": 4.63, "learning_rate": 9.08974358974359e-06, "loss": 0.0129, "step": 2275 }, { "epoch": 4.68, "learning_rate": 9.076923076923078e-06, "loss": 0.0142, "step": 2300 }, { "epoch": 4.74, "learning_rate": 9.064102564102564e-06, "loss": 0.0147, "step": 2325 }, { "epoch": 4.79, "learning_rate": 9.051282051282051e-06, "loss": 0.013, "step": 2350 }, { "epoch": 4.84, "learning_rate": 9.03846153846154e-06, "loss": 0.0134, "step": 2375 }, { "epoch": 4.89, "learning_rate": 9.025641025641027e-06, "loss": 0.0135, "step": 2400 }, { "epoch": 4.94, "learning_rate": 9.012820512820513e-06, "loss": 0.0149, "step": 2425 }, { "epoch": 4.99, "learning_rate": 9e-06, "loss": 0.0159, "step": 2450 }, { "epoch": 5.04, "learning_rate": 8.987179487179489e-06, "loss": 0.0093, "step": 2475 }, { "epoch": 5.09, "learning_rate": 8.974358974358976e-06, "loss": 0.0072, "step": 2500 }, { "epoch": 5.14, "learning_rate": 8.961538461538462e-06, "loss": 0.007, "step": 2525 }, { "epoch": 5.19, "learning_rate": 8.94871794871795e-06, "loss": 0.0068, "step": 2550 }, { "epoch": 5.24, "learning_rate": 8.935897435897438e-06, "loss": 0.0059, "step": 2575 }, { "epoch": 5.3, "learning_rate": 8.923076923076925e-06, "loss": 0.0071, "step": 2600 }, { "epoch": 5.35, "learning_rate": 8.910256410256411e-06, "loss": 0.0079, "step": 2625 }, { "epoch": 5.4, "learning_rate": 8.897435897435898e-06, "loss": 0.0079, "step": 2650 }, { "epoch": 5.45, "learning_rate": 8.884615384615385e-06, "loss": 0.0082, "step": 2675 }, { "epoch": 5.5, "learning_rate": 8.871794871794872e-06, "loss": 0.0076, "step": 2700 }, { "epoch": 5.55, "learning_rate": 8.858974358974359e-06, "loss": 0.0082, "step": 2725 }, { "epoch": 5.6, "learning_rate": 8.846153846153847e-06, "loss": 0.0088, "step": 2750 }, { "epoch": 5.65, "learning_rate": 8.833333333333334e-06, "loss": 0.0083, "step": 2775 }, { "epoch": 5.7, "learning_rate": 8.820512820512821e-06, "loss": 0.0075, "step": 2800 }, { "epoch": 5.75, "learning_rate": 8.807692307692308e-06, "loss": 0.0081, "step": 2825 }, { "epoch": 5.8, "learning_rate": 8.794871794871796e-06, "loss": 0.0088, "step": 2850 }, { "epoch": 5.86, "learning_rate": 8.782051282051283e-06, "loss": 0.0086, "step": 2875 }, { "epoch": 5.91, "learning_rate": 8.76923076923077e-06, "loss": 0.0081, "step": 2900 }, { "epoch": 5.96, "learning_rate": 8.756410256410257e-06, "loss": 0.0084, "step": 2925 }, { "epoch": 6.01, "learning_rate": 8.743589743589743e-06, "loss": 0.0095, "step": 2950 }, { "epoch": 6.06, "learning_rate": 8.730769230769232e-06, "loss": 0.0059, "step": 2975 }, { "epoch": 6.11, "learning_rate": 8.717948717948719e-06, "loss": 0.0052, "step": 3000 }, { "epoch": 6.11, "eval_loss": 0.2190403789281845, "eval_runtime": 3196.1965, "eval_samples_per_second": 4.915, "eval_steps_per_second": 0.307, "eval_wer": 4.8095613849381245, "step": 3000 }, { "epoch": 6.16, "learning_rate": 8.705128205128206e-06, "loss": 0.0051, "step": 3025 }, { "epoch": 6.21, "learning_rate": 8.692307692307692e-06, "loss": 0.0044, "step": 3050 }, { "epoch": 6.26, "learning_rate": 8.679487179487181e-06, "loss": 0.005, "step": 3075 }, { "epoch": 6.31, "learning_rate": 8.666666666666668e-06, "loss": 0.0058, "step": 3100 }, { "epoch": 6.36, "learning_rate": 8.653846153846155e-06, "loss": 0.0049, "step": 3125 }, { "epoch": 6.42, "learning_rate": 8.641025641025641e-06, "loss": 0.005, "step": 3150 }, { "epoch": 6.47, "learning_rate": 8.62820512820513e-06, "loss": 0.0055, "step": 3175 }, { "epoch": 6.52, "learning_rate": 8.615384615384617e-06, "loss": 0.0065, "step": 3200 }, { "epoch": 6.57, "learning_rate": 8.602564102564104e-06, "loss": 0.0062, "step": 3225 }, { "epoch": 6.62, "learning_rate": 8.58974358974359e-06, "loss": 0.0073, "step": 3250 }, { "epoch": 6.67, "learning_rate": 8.576923076923077e-06, "loss": 0.0057, "step": 3275 }, { "epoch": 6.72, "learning_rate": 8.564102564102564e-06, "loss": 0.0065, "step": 3300 }, { "epoch": 6.77, "learning_rate": 8.551282051282051e-06, "loss": 0.0059, "step": 3325 }, { "epoch": 6.82, "learning_rate": 8.53846153846154e-06, "loss": 0.0061, "step": 3350 }, { "epoch": 6.87, "learning_rate": 8.525641025641026e-06, "loss": 0.0066, "step": 3375 }, { "epoch": 6.92, "learning_rate": 8.512820512820513e-06, "loss": 0.0062, "step": 3400 }, { "epoch": 6.98, "learning_rate": 8.5e-06, "loss": 0.0059, "step": 3425 }, { "epoch": 7.03, "learning_rate": 8.487179487179488e-06, "loss": 0.0052, "step": 3450 }, { "epoch": 7.08, "learning_rate": 8.474358974358975e-06, "loss": 0.0039, "step": 3475 }, { "epoch": 7.13, "learning_rate": 8.461538461538462e-06, "loss": 0.0034, "step": 3500 }, { "epoch": 7.18, "learning_rate": 8.448717948717949e-06, "loss": 0.0037, "step": 3525 }, { "epoch": 7.23, "learning_rate": 8.435897435897436e-06, "loss": 0.0046, "step": 3550 }, { "epoch": 7.28, "learning_rate": 8.423076923076924e-06, "loss": 0.0044, "step": 3575 }, { "epoch": 7.33, "learning_rate": 8.410256410256411e-06, "loss": 0.0042, "step": 3600 }, { "epoch": 7.38, "learning_rate": 8.397435897435898e-06, "loss": 0.0043, "step": 3625 }, { "epoch": 7.43, "learning_rate": 8.384615384615385e-06, "loss": 0.0039, "step": 3650 }, { "epoch": 7.48, "learning_rate": 8.371794871794873e-06, "loss": 0.0038, "step": 3675 }, { "epoch": 7.54, "learning_rate": 8.35897435897436e-06, "loss": 0.0036, "step": 3700 }, { "epoch": 7.59, "learning_rate": 8.346153846153847e-06, "loss": 0.0054, "step": 3725 }, { "epoch": 7.64, "learning_rate": 8.333333333333334e-06, "loss": 0.006, "step": 3750 }, { "epoch": 7.69, "learning_rate": 8.320512820512822e-06, "loss": 0.005, "step": 3775 }, { "epoch": 7.74, "learning_rate": 8.307692307692309e-06, "loss": 0.0045, "step": 3800 }, { "epoch": 7.79, "learning_rate": 8.294871794871796e-06, "loss": 0.0052, "step": 3825 }, { "epoch": 7.84, "learning_rate": 8.282051282051283e-06, "loss": 0.0061, "step": 3850 }, { "epoch": 7.89, "learning_rate": 8.26923076923077e-06, "loss": 0.0058, "step": 3875 }, { "epoch": 7.94, "learning_rate": 8.256410256410256e-06, "loss": 0.0045, "step": 3900 }, { "epoch": 7.99, "learning_rate": 8.243589743589743e-06, "loss": 0.0055, "step": 3925 }, { "epoch": 8.04, "learning_rate": 8.230769230769232e-06, "loss": 0.0027, "step": 3950 }, { "epoch": 8.1, "learning_rate": 8.217948717948719e-06, "loss": 0.0025, "step": 3975 }, { "epoch": 8.15, "learning_rate": 8.205128205128205e-06, "loss": 0.0024, "step": 4000 }, { "epoch": 8.15, "eval_loss": 0.2289302498102188, "eval_runtime": 3192.1027, "eval_samples_per_second": 4.921, "eval_steps_per_second": 0.308, "eval_wer": 4.8775637201126525, "step": 4000 }, { "epoch": 8.2, "learning_rate": 8.192307692307692e-06, "loss": 0.0027, "step": 4025 }, { "epoch": 8.25, "learning_rate": 8.17948717948718e-06, "loss": 0.0027, "step": 4050 }, { "epoch": 8.3, "learning_rate": 8.166666666666668e-06, "loss": 0.0036, "step": 4075 }, { "epoch": 8.35, "learning_rate": 8.153846153846154e-06, "loss": 0.004, "step": 4100 }, { "epoch": 8.4, "learning_rate": 8.141025641025641e-06, "loss": 0.0043, "step": 4125 }, { "epoch": 8.45, "learning_rate": 8.12820512820513e-06, "loss": 0.0028, "step": 4150 }, { "epoch": 8.5, "learning_rate": 8.115384615384617e-06, "loss": 0.003, "step": 4175 }, { "epoch": 8.55, "learning_rate": 8.102564102564103e-06, "loss": 0.0041, "step": 4200 }, { "epoch": 8.6, "learning_rate": 8.08974358974359e-06, "loss": 0.0032, "step": 4225 }, { "epoch": 8.66, "learning_rate": 8.076923076923077e-06, "loss": 0.0024, "step": 4250 }, { "epoch": 8.71, "learning_rate": 8.064102564102566e-06, "loss": 0.0029, "step": 4275 }, { "epoch": 8.76, "learning_rate": 8.051282051282052e-06, "loss": 0.0038, "step": 4300 }, { "epoch": 8.81, "learning_rate": 8.03846153846154e-06, "loss": 0.0055, "step": 4325 }, { "epoch": 8.86, "learning_rate": 8.025641025641026e-06, "loss": 0.0056, "step": 4350 }, { "epoch": 8.91, "learning_rate": 8.012820512820515e-06, "loss": 0.0033, "step": 4375 }, { "epoch": 8.96, "learning_rate": 8.000000000000001e-06, "loss": 0.0042, "step": 4400 }, { "epoch": 9.01, "learning_rate": 7.987179487179488e-06, "loss": 0.0038, "step": 4425 }, { "epoch": 9.06, "learning_rate": 7.974358974358975e-06, "loss": 0.0031, "step": 4450 }, { "epoch": 9.11, "learning_rate": 7.961538461538462e-06, "loss": 0.0031, "step": 4475 }, { "epoch": 9.16, "learning_rate": 7.948717948717949e-06, "loss": 0.0029, "step": 4500 }, { "epoch": 9.22, "learning_rate": 7.935897435897435e-06, "loss": 0.0031, "step": 4525 }, { "epoch": 9.27, "learning_rate": 7.923076923076924e-06, "loss": 0.0033, "step": 4550 }, { "epoch": 9.32, "learning_rate": 7.91025641025641e-06, "loss": 0.0029, "step": 4575 }, { "epoch": 9.37, "learning_rate": 7.897435897435898e-06, "loss": 0.0028, "step": 4600 }, { "epoch": 9.42, "learning_rate": 7.884615384615384e-06, "loss": 0.0022, "step": 4625 }, { "epoch": 9.47, "learning_rate": 7.871794871794873e-06, "loss": 0.0046, "step": 4650 }, { "epoch": 9.52, "learning_rate": 7.85897435897436e-06, "loss": 0.0036, "step": 4675 }, { "epoch": 9.57, "learning_rate": 7.846153846153847e-06, "loss": 0.0029, "step": 4700 }, { "epoch": 9.62, "learning_rate": 7.833333333333333e-06, "loss": 0.0031, "step": 4725 }, { "epoch": 9.67, "learning_rate": 7.820512820512822e-06, "loss": 0.0043, "step": 4750 }, { "epoch": 9.73, "learning_rate": 7.807692307692309e-06, "loss": 0.0031, "step": 4775 }, { "epoch": 9.78, "learning_rate": 7.794871794871796e-06, "loss": 0.0037, "step": 4800 }, { "epoch": 9.83, "learning_rate": 7.782051282051282e-06, "loss": 0.0031, "step": 4825 }, { "epoch": 9.88, "learning_rate": 7.76923076923077e-06, "loss": 0.0033, "step": 4850 }, { "epoch": 9.93, "learning_rate": 7.756410256410258e-06, "loss": 0.0043, "step": 4875 }, { "epoch": 9.98, "learning_rate": 7.743589743589745e-06, "loss": 0.0047, "step": 4900 }, { "epoch": 10.03, "learning_rate": 7.730769230769231e-06, "loss": 0.0031, "step": 4925 }, { "epoch": 10.08, "learning_rate": 7.717948717948718e-06, "loss": 0.0018, "step": 4950 }, { "epoch": 10.13, "learning_rate": 7.705128205128207e-06, "loss": 0.003, "step": 4975 }, { "epoch": 10.18, "learning_rate": 7.692307692307694e-06, "loss": 0.0024, "step": 5000 }, { "epoch": 10.18, "eval_loss": 0.2341163605451584, "eval_runtime": 3188.2366, "eval_samples_per_second": 4.927, "eval_steps_per_second": 0.308, "eval_wer": 4.892318943782597, "step": 5000 }, { "epoch": 10.23, "learning_rate": 7.67948717948718e-06, "loss": 0.0026, "step": 5025 }, { "epoch": 10.29, "learning_rate": 7.666666666666667e-06, "loss": 0.0023, "step": 5050 }, { "epoch": 10.34, "learning_rate": 7.653846153846154e-06, "loss": 0.0033, "step": 5075 }, { "epoch": 10.39, "learning_rate": 7.641025641025641e-06, "loss": 0.003, "step": 5100 }, { "epoch": 10.44, "learning_rate": 7.6282051282051286e-06, "loss": 0.0035, "step": 5125 }, { "epoch": 10.49, "learning_rate": 7.615384615384615e-06, "loss": 0.0031, "step": 5150 }, { "epoch": 10.54, "learning_rate": 7.602564102564103e-06, "loss": 0.0041, "step": 5175 }, { "epoch": 10.59, "learning_rate": 7.58974358974359e-06, "loss": 0.0028, "step": 5200 }, { "epoch": 10.64, "learning_rate": 7.5769230769230775e-06, "loss": 0.0027, "step": 5225 }, { "epoch": 10.69, "learning_rate": 7.564102564102564e-06, "loss": 0.0019, "step": 5250 }, { "epoch": 10.74, "learning_rate": 7.551282051282052e-06, "loss": 0.0024, "step": 5275 }, { "epoch": 10.79, "learning_rate": 7.538461538461539e-06, "loss": 0.0021, "step": 5300 }, { "epoch": 10.85, "learning_rate": 7.5256410256410265e-06, "loss": 0.0028, "step": 5325 }, { "epoch": 10.9, "learning_rate": 7.512820512820513e-06, "loss": 0.0029, "step": 5350 }, { "epoch": 10.95, "learning_rate": 7.500000000000001e-06, "loss": 0.0047, "step": 5375 }, { "epoch": 11.0, "learning_rate": 7.487179487179488e-06, "loss": 0.0037, "step": 5400 }, { "epoch": 11.05, "learning_rate": 7.474871794871795e-06, "loss": 0.0033, "step": 5425 }, { "epoch": 11.1, "learning_rate": 7.462051282051282e-06, "loss": 0.0026, "step": 5450 }, { "epoch": 11.15, "learning_rate": 7.44923076923077e-06, "loss": 0.0024, "step": 5475 }, { "epoch": 11.2, "learning_rate": 7.436410256410257e-06, "loss": 0.0025, "step": 5500 }, { "epoch": 11.25, "learning_rate": 7.423589743589744e-06, "loss": 0.0023, "step": 5525 }, { "epoch": 11.3, "learning_rate": 7.410769230769231e-06, "loss": 0.0027, "step": 5550 }, { "epoch": 11.35, "learning_rate": 7.397948717948719e-06, "loss": 0.0027, "step": 5575 }, { "epoch": 11.41, "learning_rate": 7.385128205128206e-06, "loss": 0.0025, "step": 5600 }, { "epoch": 11.46, "learning_rate": 7.372307692307693e-06, "loss": 0.0017, "step": 5625 }, { "epoch": 11.51, "learning_rate": 7.35948717948718e-06, "loss": 0.0031, "step": 5650 }, { "epoch": 11.56, "learning_rate": 7.346666666666668e-06, "loss": 0.0046, "step": 5675 }, { "epoch": 11.61, "learning_rate": 7.333846153846155e-06, "loss": 0.0031, "step": 5700 }, { "epoch": 11.66, "learning_rate": 7.321025641025642e-06, "loss": 0.0029, "step": 5725 }, { "epoch": 11.71, "learning_rate": 7.308205128205129e-06, "loss": 0.0027, "step": 5750 }, { "epoch": 11.76, "learning_rate": 7.295384615384617e-06, "loss": 0.0036, "step": 5775 }, { "epoch": 11.81, "learning_rate": 7.2825641025641036e-06, "loss": 0.003, "step": 5800 }, { "epoch": 11.86, "learning_rate": 7.269743589743591e-06, "loss": 0.0031, "step": 5825 }, { "epoch": 11.91, "learning_rate": 7.256923076923078e-06, "loss": 0.0035, "step": 5850 }, { "epoch": 11.97, "learning_rate": 7.244102564102564e-06, "loss": 0.0034, "step": 5875 }, { "epoch": 12.02, "learning_rate": 7.231282051282052e-06, "loss": 0.0022, "step": 5900 }, { "epoch": 12.07, "learning_rate": 7.2184615384615385e-06, "loss": 0.0021, "step": 5925 }, { "epoch": 12.12, "learning_rate": 7.205641025641026e-06, "loss": 0.0015, "step": 5950 }, { "epoch": 12.17, "learning_rate": 7.192820512820513e-06, "loss": 0.002, "step": 5975 }, { "epoch": 12.22, "learning_rate": 7.180000000000001e-06, "loss": 0.0015, "step": 6000 }, { "epoch": 12.22, "eval_loss": 0.2458568662405014, "eval_runtime": 3184.4333, "eval_samples_per_second": 4.933, "eval_steps_per_second": 0.308, "eval_wer": 4.93401848893679, "step": 6000 }, { "epoch": 12.27, "learning_rate": 7.1671794871794875e-06, "loss": 0.0019, "step": 6025 }, { "epoch": 12.32, "learning_rate": 7.154358974358974e-06, "loss": 0.0022, "step": 6050 }, { "epoch": 12.37, "learning_rate": 7.141538461538462e-06, "loss": 0.0022, "step": 6075 }, { "epoch": 12.42, "learning_rate": 7.128717948717949e-06, "loss": 0.0034, "step": 6100 }, { "epoch": 12.47, "learning_rate": 7.1158974358974365e-06, "loss": 0.0037, "step": 6125 }, { "epoch": 12.53, "learning_rate": 7.103076923076923e-06, "loss": 0.0031, "step": 6150 }, { "epoch": 12.58, "learning_rate": 7.090256410256411e-06, "loss": 0.0024, "step": 6175 }, { "epoch": 12.63, "learning_rate": 7.077435897435898e-06, "loss": 0.0037, "step": 6200 }, { "epoch": 12.68, "learning_rate": 7.0646153846153855e-06, "loss": 0.0028, "step": 6225 }, { "epoch": 12.73, "learning_rate": 7.051794871794872e-06, "loss": 0.0037, "step": 6250 }, { "epoch": 12.78, "learning_rate": 7.03897435897436e-06, "loss": 0.0028, "step": 6275 }, { "epoch": 12.83, "learning_rate": 7.026153846153847e-06, "loss": 0.0034, "step": 6300 }, { "epoch": 12.88, "learning_rate": 7.0133333333333345e-06, "loss": 0.0028, "step": 6325 }, { "epoch": 12.93, "learning_rate": 7.000512820512821e-06, "loss": 0.0032, "step": 6350 }, { "epoch": 12.98, "learning_rate": 6.987692307692309e-06, "loss": 0.0029, "step": 6375 }, { "epoch": 13.03, "learning_rate": 6.974871794871796e-06, "loss": 0.0021, "step": 6400 }, { "epoch": 13.09, "learning_rate": 6.9620512820512835e-06, "loss": 0.0025, "step": 6425 }, { "epoch": 13.14, "learning_rate": 6.94923076923077e-06, "loss": 0.0024, "step": 6450 }, { "epoch": 13.19, "learning_rate": 6.936410256410256e-06, "loss": 0.0022, "step": 6475 }, { "epoch": 13.24, "learning_rate": 6.923589743589744e-06, "loss": 0.0018, "step": 6500 }, { "epoch": 13.29, "learning_rate": 6.910769230769231e-06, "loss": 0.0021, "step": 6525 }, { "epoch": 13.34, "learning_rate": 6.8979487179487185e-06, "loss": 0.0019, "step": 6550 }, { "epoch": 13.39, "learning_rate": 6.885128205128205e-06, "loss": 0.0016, "step": 6575 }, { "epoch": 13.44, "learning_rate": 6.872307692307693e-06, "loss": 0.0023, "step": 6600 }, { "epoch": 13.49, "learning_rate": 6.85948717948718e-06, "loss": 0.0028, "step": 6625 }, { "epoch": 13.54, "learning_rate": 6.846666666666667e-06, "loss": 0.0026, "step": 6650 }, { "epoch": 13.59, "learning_rate": 6.833846153846154e-06, "loss": 0.0024, "step": 6675 }, { "epoch": 13.65, "learning_rate": 6.821025641025641e-06, "loss": 0.0029, "step": 6700 }, { "epoch": 13.7, "learning_rate": 6.808205128205129e-06, "loss": 0.0027, "step": 6725 }, { "epoch": 13.75, "learning_rate": 6.795384615384616e-06, "loss": 0.0028, "step": 6750 }, { "epoch": 13.8, "learning_rate": 6.782564102564103e-06, "loss": 0.0031, "step": 6775 }, { "epoch": 13.85, "learning_rate": 6.76974358974359e-06, "loss": 0.0024, "step": 6800 }, { "epoch": 13.9, "learning_rate": 6.756923076923078e-06, "loss": 0.0021, "step": 6825 }, { "epoch": 13.95, "learning_rate": 6.744102564102565e-06, "loss": 0.002, "step": 6850 }, { "epoch": 14.0, "learning_rate": 6.731282051282052e-06, "loss": 0.0024, "step": 6875 }, { "epoch": 14.05, "learning_rate": 6.718461538461539e-06, "loss": 0.0015, "step": 6900 }, { "epoch": 14.1, "learning_rate": 6.705641025641027e-06, "loss": 0.0022, "step": 6925 }, { "epoch": 14.15, "learning_rate": 6.692820512820514e-06, "loss": 0.0013, "step": 6950 }, { "epoch": 14.21, "learning_rate": 6.680000000000001e-06, "loss": 0.0012, "step": 6975 }, { "epoch": 14.26, "learning_rate": 6.667179487179488e-06, "loss": 0.0021, "step": 7000 }, { "epoch": 14.26, "eval_loss": 0.25577312707901, "eval_runtime": 3186.4396, "eval_samples_per_second": 4.93, "eval_steps_per_second": 0.308, "eval_wer": 4.927603174297683, "step": 7000 }, { "epoch": 14.31, "learning_rate": 6.654358974358976e-06, "loss": 0.0021, "step": 7025 }, { "epoch": 14.36, "learning_rate": 6.641538461538463e-06, "loss": 0.0017, "step": 7050 }, { "epoch": 14.41, "learning_rate": 6.6287179487179486e-06, "loss": 0.0018, "step": 7075 }, { "epoch": 14.46, "learning_rate": 6.615897435897436e-06, "loss": 0.0024, "step": 7100 }, { "epoch": 14.51, "learning_rate": 6.603076923076923e-06, "loss": 0.0023, "step": 7125 }, { "epoch": 14.56, "learning_rate": 6.590256410256411e-06, "loss": 0.0023, "step": 7150 }, { "epoch": 14.61, "learning_rate": 6.5774358974358976e-06, "loss": 0.0026, "step": 7175 }, { "epoch": 14.66, "learning_rate": 6.564615384615385e-06, "loss": 0.0022, "step": 7200 }, { "epoch": 14.71, "learning_rate": 6.551794871794872e-06, "loss": 0.0013, "step": 7225 }, { "epoch": 14.77, "learning_rate": 6.538974358974359e-06, "loss": 0.0014, "step": 7250 }, { "epoch": 14.82, "learning_rate": 6.5261538461538465e-06, "loss": 0.0026, "step": 7275 }, { "epoch": 14.87, "learning_rate": 6.513333333333333e-06, "loss": 0.0021, "step": 7300 }, { "epoch": 14.92, "learning_rate": 6.500512820512821e-06, "loss": 0.0017, "step": 7325 }, { "epoch": 14.97, "learning_rate": 6.487692307692308e-06, "loss": 0.0024, "step": 7350 }, { "epoch": 15.02, "learning_rate": 6.4748717948717955e-06, "loss": 0.003, "step": 7375 }, { "epoch": 15.07, "learning_rate": 6.462051282051282e-06, "loss": 0.0013, "step": 7400 }, { "epoch": 15.12, "learning_rate": 6.44923076923077e-06, "loss": 0.0011, "step": 7425 }, { "epoch": 15.17, "learning_rate": 6.436410256410257e-06, "loss": 0.0015, "step": 7450 }, { "epoch": 15.22, "learning_rate": 6.4235897435897445e-06, "loss": 0.0015, "step": 7475 }, { "epoch": 15.27, "learning_rate": 6.410769230769231e-06, "loss": 0.0013, "step": 7500 }, { "epoch": 15.33, "learning_rate": 6.397948717948719e-06, "loss": 0.0012, "step": 7525 }, { "epoch": 15.38, "learning_rate": 6.385128205128206e-06, "loss": 0.0014, "step": 7550 }, { "epoch": 15.43, "learning_rate": 6.3723076923076935e-06, "loss": 0.0019, "step": 7575 }, { "epoch": 15.48, "learning_rate": 6.35948717948718e-06, "loss": 0.0025, "step": 7600 }, { "epoch": 15.53, "learning_rate": 6.346666666666668e-06, "loss": 0.0021, "step": 7625 }, { "epoch": 15.58, "learning_rate": 6.333846153846155e-06, "loss": 0.0016, "step": 7650 }, { "epoch": 15.63, "learning_rate": 6.3210256410256425e-06, "loss": 0.0017, "step": 7675 }, { "epoch": 15.68, "learning_rate": 6.3082051282051285e-06, "loss": 0.0014, "step": 7700 }, { "epoch": 15.73, "learning_rate": 6.295384615384615e-06, "loss": 0.0018, "step": 7725 }, { "epoch": 15.78, "learning_rate": 6.282564102564103e-06, "loss": 0.0023, "step": 7750 }, { "epoch": 15.84, "learning_rate": 6.26974358974359e-06, "loss": 0.0019, "step": 7775 }, { "epoch": 15.89, "learning_rate": 6.2569230769230775e-06, "loss": 0.0014, "step": 7800 }, { "epoch": 15.94, "learning_rate": 6.244102564102564e-06, "loss": 0.0015, "step": 7825 }, { "epoch": 15.99, "learning_rate": 6.231282051282051e-06, "loss": 0.002, "step": 7850 }, { "epoch": 16.04, "learning_rate": 6.218461538461539e-06, "loss": 0.0016, "step": 7875 }, { "epoch": 16.09, "learning_rate": 6.205641025641026e-06, "loss": 0.002, "step": 7900 }, { "epoch": 16.14, "learning_rate": 6.192820512820513e-06, "loss": 0.001, "step": 7925 }, { "epoch": 16.19, "learning_rate": 6.18e-06, "loss": 0.0013, "step": 7950 }, { "epoch": 16.24, "learning_rate": 6.167179487179488e-06, "loss": 0.0007, "step": 7975 }, { "epoch": 16.29, "learning_rate": 6.154358974358975e-06, "loss": 0.0011, "step": 8000 }, { "epoch": 16.29, "eval_loss": 0.2540310025215149, "eval_runtime": 3185.1352, "eval_samples_per_second": 4.932, "eval_steps_per_second": 0.308, "eval_wer": 5.101458201017469, "step": 8000 }, { "epoch": 16.34, "learning_rate": 6.141538461538462e-06, "loss": 0.001, "step": 8025 }, { "epoch": 16.4, "learning_rate": 6.128717948717949e-06, "loss": 0.0015, "step": 8050 }, { "epoch": 16.45, "learning_rate": 6.115897435897437e-06, "loss": 0.0017, "step": 8075 }, { "epoch": 16.5, "learning_rate": 6.103076923076924e-06, "loss": 0.0022, "step": 8100 }, { "epoch": 16.55, "learning_rate": 6.090256410256411e-06, "loss": 0.0018, "step": 8125 }, { "epoch": 16.6, "learning_rate": 6.077435897435898e-06, "loss": 0.0013, "step": 8150 }, { "epoch": 16.65, "learning_rate": 6.064615384615386e-06, "loss": 0.0011, "step": 8175 }, { "epoch": 16.7, "learning_rate": 6.051794871794873e-06, "loss": 0.0015, "step": 8200 }, { "epoch": 16.75, "learning_rate": 6.03897435897436e-06, "loss": 0.0022, "step": 8225 }, { "epoch": 16.8, "learning_rate": 6.026153846153847e-06, "loss": 0.0022, "step": 8250 }, { "epoch": 16.85, "learning_rate": 6.013333333333335e-06, "loss": 0.0033, "step": 8275 }, { "epoch": 16.9, "learning_rate": 6.000512820512821e-06, "loss": 0.0023, "step": 8300 }, { "epoch": 16.96, "learning_rate": 5.9876923076923076e-06, "loss": 0.002, "step": 8325 }, { "epoch": 17.01, "learning_rate": 5.974871794871795e-06, "loss": 0.0019, "step": 8350 }, { "epoch": 17.06, "learning_rate": 5.962051282051282e-06, "loss": 0.0014, "step": 8375 }, { "epoch": 17.11, "learning_rate": 5.94923076923077e-06, "loss": 0.0011, "step": 8400 }, { "epoch": 17.16, "learning_rate": 5.9364102564102566e-06, "loss": 0.0014, "step": 8425 }, { "epoch": 17.21, "learning_rate": 5.923589743589744e-06, "loss": 0.0019, "step": 8450 }, { "epoch": 17.26, "learning_rate": 5.910769230769231e-06, "loss": 0.002, "step": 8475 }, { "epoch": 17.31, "learning_rate": 5.897948717948718e-06, "loss": 0.0012, "step": 8500 }, { "epoch": 17.36, "learning_rate": 5.8851282051282056e-06, "loss": 0.0009, "step": 8525 }, { "epoch": 17.41, "learning_rate": 5.872307692307692e-06, "loss": 0.0019, "step": 8550 }, { "epoch": 17.46, "learning_rate": 5.85948717948718e-06, "loss": 0.0013, "step": 8575 }, { "epoch": 17.52, "learning_rate": 5.846666666666667e-06, "loss": 0.0012, "step": 8600 }, { "epoch": 17.57, "learning_rate": 5.8338461538461545e-06, "loss": 0.0017, "step": 8625 }, { "epoch": 17.62, "learning_rate": 5.821025641025641e-06, "loss": 0.0009, "step": 8650 }, { "epoch": 17.67, "learning_rate": 5.808205128205129e-06, "loss": 0.0021, "step": 8675 }, { "epoch": 17.72, "learning_rate": 5.795384615384616e-06, "loss": 0.0015, "step": 8700 }, { "epoch": 17.77, "learning_rate": 5.7825641025641035e-06, "loss": 0.0015, "step": 8725 }, { "epoch": 17.82, "learning_rate": 5.76974358974359e-06, "loss": 0.0012, "step": 8750 }, { "epoch": 17.87, "learning_rate": 5.756923076923078e-06, "loss": 0.0015, "step": 8775 }, { "epoch": 17.92, "learning_rate": 5.744102564102565e-06, "loss": 0.0017, "step": 8800 }, { "epoch": 17.97, "learning_rate": 5.7312820512820525e-06, "loss": 0.0017, "step": 8825 }, { "epoch": 18.02, "learning_rate": 5.718461538461539e-06, "loss": 0.0017, "step": 8850 }, { "epoch": 18.08, "learning_rate": 5.705641025641027e-06, "loss": 0.0023, "step": 8875 }, { "epoch": 18.13, "learning_rate": 5.692820512820513e-06, "loss": 0.0014, "step": 8900 }, { "epoch": 18.18, "learning_rate": 5.68e-06, "loss": 0.0016, "step": 8925 }, { "epoch": 18.23, "learning_rate": 5.6671794871794875e-06, "loss": 0.0015, "step": 8950 }, { "epoch": 18.28, "learning_rate": 5.654358974358974e-06, "loss": 0.0009, "step": 8975 }, { "epoch": 18.33, "learning_rate": 5.641538461538462e-06, "loss": 0.0013, "step": 9000 }, { "epoch": 18.33, "eval_loss": 0.26109546422958374, "eval_runtime": 3205.1098, "eval_samples_per_second": 4.901, "eval_steps_per_second": 0.306, "eval_wer": 5.185498822789763, "step": 9000 }, { "epoch": 18.38, "learning_rate": 5.628717948717949e-06, "loss": 0.0013, "step": 9025 }, { "epoch": 18.43, "learning_rate": 5.6158974358974365e-06, "loss": 0.0015, "step": 9050 }, { "epoch": 18.48, "learning_rate": 5.603076923076923e-06, "loss": 0.0014, "step": 9075 }, { "epoch": 18.53, "learning_rate": 5.59025641025641e-06, "loss": 0.001, "step": 9100 }, { "epoch": 18.58, "learning_rate": 5.577435897435898e-06, "loss": 0.0013, "step": 9125 }, { "epoch": 18.64, "learning_rate": 5.564615384615385e-06, "loss": 0.0019, "step": 9150 }, { "epoch": 18.69, "learning_rate": 5.551794871794872e-06, "loss": 0.0017, "step": 9175 }, { "epoch": 18.74, "learning_rate": 5.538974358974359e-06, "loss": 0.0016, "step": 9200 }, { "epoch": 18.79, "learning_rate": 5.526153846153847e-06, "loss": 0.0013, "step": 9225 }, { "epoch": 18.84, "learning_rate": 5.513333333333334e-06, "loss": 0.0017, "step": 9250 }, { "epoch": 18.89, "learning_rate": 5.500512820512821e-06, "loss": 0.0015, "step": 9275 }, { "epoch": 18.94, "learning_rate": 5.487692307692308e-06, "loss": 0.0029, "step": 9300 }, { "epoch": 18.99, "learning_rate": 5.474871794871796e-06, "loss": 0.0031, "step": 9325 }, { "epoch": 19.04, "learning_rate": 5.462051282051283e-06, "loss": 0.0014, "step": 9350 }, { "epoch": 19.09, "learning_rate": 5.44923076923077e-06, "loss": 0.001, "step": 9375 }, { "epoch": 19.14, "learning_rate": 5.436410256410257e-06, "loss": 0.0019, "step": 9400 }, { "epoch": 19.2, "learning_rate": 5.423589743589745e-06, "loss": 0.0014, "step": 9425 }, { "epoch": 19.25, "learning_rate": 5.410769230769232e-06, "loss": 0.0018, "step": 9450 }, { "epoch": 19.3, "learning_rate": 5.397948717948719e-06, "loss": 0.0018, "step": 9475 }, { "epoch": 19.35, "learning_rate": 5.385128205128206e-06, "loss": 0.0007, "step": 9500 }, { "epoch": 19.4, "learning_rate": 5.372307692307692e-06, "loss": 0.0008, "step": 9525 }, { "epoch": 19.45, "learning_rate": 5.35948717948718e-06, "loss": 0.001, "step": 9550 }, { "epoch": 19.5, "learning_rate": 5.346666666666667e-06, "loss": 0.001, "step": 9575 }, { "epoch": 19.55, "learning_rate": 5.333846153846154e-06, "loss": 0.0007, "step": 9600 }, { "epoch": 19.6, "learning_rate": 5.321025641025641e-06, "loss": 0.0016, "step": 9625 }, { "epoch": 19.65, "learning_rate": 5.308205128205129e-06, "loss": 0.0012, "step": 9650 }, { "epoch": 19.7, "learning_rate": 5.2953846153846156e-06, "loss": 0.0009, "step": 9675 }, { "epoch": 19.76, "learning_rate": 5.283076923076923e-06, "loss": 0.0015, "step": 9700 }, { "epoch": 19.81, "learning_rate": 5.27025641025641e-06, "loss": 0.0007, "step": 9725 }, { "epoch": 19.86, "learning_rate": 5.2574358974358975e-06, "loss": 0.0011, "step": 9750 }, { "epoch": 19.91, "learning_rate": 5.244615384615384e-06, "loss": 0.0015, "step": 9775 }, { "epoch": 19.96, "learning_rate": 5.231794871794872e-06, "loss": 0.001, "step": 9800 }, { "epoch": 20.01, "learning_rate": 5.218974358974359e-06, "loss": 0.0024, "step": 9825 }, { "epoch": 20.06, "learning_rate": 5.2061538461538465e-06, "loss": 0.001, "step": 9850 }, { "epoch": 20.11, "learning_rate": 5.193333333333333e-06, "loss": 0.0009, "step": 9875 }, { "epoch": 20.16, "learning_rate": 5.180512820512821e-06, "loss": 0.0008, "step": 9900 }, { "epoch": 20.21, "learning_rate": 5.167692307692308e-06, "loss": 0.0011, "step": 9925 }, { "epoch": 20.26, "learning_rate": 5.1548717948717955e-06, "loss": 0.0009, "step": 9950 }, { "epoch": 20.32, "learning_rate": 5.142051282051282e-06, "loss": 0.0005, "step": 9975 }, { "epoch": 20.37, "learning_rate": 5.12923076923077e-06, "loss": 0.0005, "step": 10000 }, { "epoch": 20.37, "eval_loss": 0.27195730805397034, "eval_runtime": 3197.6224, "eval_samples_per_second": 4.912, "eval_steps_per_second": 0.307, "eval_wer": 4.937867677720254, "step": 10000 }, { "epoch": 20.42, "learning_rate": 5.116410256410257e-06, "loss": 0.0008, "step": 10025 }, { "epoch": 20.47, "learning_rate": 5.1035897435897445e-06, "loss": 0.0007, "step": 10050 }, { "epoch": 20.52, "learning_rate": 5.090769230769231e-06, "loss": 0.0007, "step": 10075 }, { "epoch": 20.57, "learning_rate": 5.077948717948719e-06, "loss": 0.0008, "step": 10100 }, { "epoch": 20.62, "learning_rate": 5.065128205128206e-06, "loss": 0.0015, "step": 10125 }, { "epoch": 20.67, "learning_rate": 5.052307692307693e-06, "loss": 0.0009, "step": 10150 }, { "epoch": 20.72, "learning_rate": 5.03948717948718e-06, "loss": 0.0007, "step": 10175 }, { "epoch": 20.77, "learning_rate": 5.026666666666667e-06, "loss": 0.0005, "step": 10200 }, { "epoch": 20.82, "learning_rate": 5.013846153846155e-06, "loss": 0.0007, "step": 10225 }, { "epoch": 20.88, "learning_rate": 5.001025641025642e-06, "loss": 0.0013, "step": 10250 }, { "epoch": 20.93, "learning_rate": 4.988205128205128e-06, "loss": 0.0007, "step": 10275 }, { "epoch": 20.98, "learning_rate": 4.975384615384616e-06, "loss": 0.0008, "step": 10300 }, { "epoch": 21.03, "learning_rate": 4.962564102564103e-06, "loss": 0.0004, "step": 10325 }, { "epoch": 21.08, "learning_rate": 4.949743589743591e-06, "loss": 0.0004, "step": 10350 }, { "epoch": 21.13, "learning_rate": 4.936923076923077e-06, "loss": 0.0003, "step": 10375 }, { "epoch": 21.18, "learning_rate": 4.924102564102565e-06, "loss": 0.0003, "step": 10400 }, { "epoch": 21.23, "learning_rate": 4.911282051282052e-06, "loss": 0.0003, "step": 10425 }, { "epoch": 21.28, "learning_rate": 4.898461538461539e-06, "loss": 0.0007, "step": 10450 }, { "epoch": 21.33, "learning_rate": 4.8856410256410256e-06, "loss": 0.0009, "step": 10475 }, { "epoch": 21.38, "learning_rate": 4.872820512820513e-06, "loss": 0.0004, "step": 10500 }, { "epoch": 21.44, "learning_rate": 4.86e-06, "loss": 0.0003, "step": 10525 }, { "epoch": 21.49, "learning_rate": 4.847179487179488e-06, "loss": 0.0007, "step": 10550 }, { "epoch": 21.54, "learning_rate": 4.8343589743589746e-06, "loss": 0.0004, "step": 10575 }, { "epoch": 21.59, "learning_rate": 4.821538461538462e-06, "loss": 0.0009, "step": 10600 }, { "epoch": 21.64, "learning_rate": 4.808717948717949e-06, "loss": 0.002, "step": 10625 }, { "epoch": 21.69, "learning_rate": 4.795897435897437e-06, "loss": 0.0013, "step": 10650 }, { "epoch": 21.74, "learning_rate": 4.7830769230769235e-06, "loss": 0.0013, "step": 10675 }, { "epoch": 21.79, "learning_rate": 4.770256410256411e-06, "loss": 0.0015, "step": 10700 }, { "epoch": 21.84, "learning_rate": 4.757435897435898e-06, "loss": 0.0008, "step": 10725 }, { "epoch": 21.89, "learning_rate": 4.744615384615385e-06, "loss": 0.0011, "step": 10750 }, { "epoch": 21.95, "learning_rate": 4.731794871794872e-06, "loss": 0.0009, "step": 10775 }, { "epoch": 22.0, "learning_rate": 4.718974358974359e-06, "loss": 0.0013, "step": 10800 }, { "epoch": 22.05, "learning_rate": 4.706153846153846e-06, "loss": 0.0013, "step": 10825 }, { "epoch": 22.1, "learning_rate": 4.693333333333334e-06, "loss": 0.0007, "step": 10850 }, { "epoch": 22.15, "learning_rate": 4.680512820512821e-06, "loss": 0.001, "step": 10875 }, { "epoch": 22.2, "learning_rate": 4.667692307692308e-06, "loss": 0.0008, "step": 10900 }, { "epoch": 22.25, "learning_rate": 4.654871794871795e-06, "loss": 0.0006, "step": 10925 }, { "epoch": 22.3, "learning_rate": 4.642051282051283e-06, "loss": 0.0007, "step": 10950 }, { "epoch": 22.35, "learning_rate": 4.62923076923077e-06, "loss": 0.0012, "step": 10975 }, { "epoch": 22.4, "learning_rate": 4.616410256410257e-06, "loss": 0.0028, "step": 11000 }, { "epoch": 22.4, "eval_loss": 0.26136788725852966, "eval_runtime": 3186.494, "eval_samples_per_second": 4.93, "eval_steps_per_second": 0.308, "eval_wer": 5.011002264606067, "step": 11000 }, { "epoch": 22.45, "learning_rate": 4.603589743589744e-06, "loss": 0.0008, "step": 11025 }, { "epoch": 22.51, "learning_rate": 4.590769230769232e-06, "loss": 0.0008, "step": 11050 }, { "epoch": 22.56, "learning_rate": 4.577948717948718e-06, "loss": 0.0007, "step": 11075 }, { "epoch": 22.61, "learning_rate": 4.5651282051282055e-06, "loss": 0.001, "step": 11100 }, { "epoch": 22.66, "learning_rate": 4.552307692307692e-06, "loss": 0.001, "step": 11125 }, { "epoch": 22.71, "learning_rate": 4.53948717948718e-06, "loss": 0.0012, "step": 11150 }, { "epoch": 22.76, "learning_rate": 4.526666666666667e-06, "loss": 0.0014, "step": 11175 }, { "epoch": 22.81, "learning_rate": 4.5138461538461545e-06, "loss": 0.0018, "step": 11200 }, { "epoch": 22.86, "learning_rate": 4.501025641025641e-06, "loss": 0.0013, "step": 11225 }, { "epoch": 22.91, "learning_rate": 4.488205128205129e-06, "loss": 0.0014, "step": 11250 }, { "epoch": 22.96, "learning_rate": 4.475384615384616e-06, "loss": 0.0015, "step": 11275 }, { "epoch": 23.01, "learning_rate": 4.4625641025641035e-06, "loss": 0.0018, "step": 11300 }, { "epoch": 23.07, "learning_rate": 4.44974358974359e-06, "loss": 0.0008, "step": 11325 }, { "epoch": 23.12, "learning_rate": 4.436923076923078e-06, "loss": 0.0017, "step": 11350 }, { "epoch": 23.17, "learning_rate": 4.424102564102564e-06, "loss": 0.0017, "step": 11375 }, { "epoch": 23.22, "learning_rate": 4.411282051282052e-06, "loss": 0.0014, "step": 11400 }, { "epoch": 23.27, "learning_rate": 4.3984615384615384e-06, "loss": 0.0014, "step": 11425 }, { "epoch": 23.32, "learning_rate": 4.385641025641026e-06, "loss": 0.0008, "step": 11450 }, { "epoch": 23.37, "learning_rate": 4.372820512820513e-06, "loss": 0.0005, "step": 11475 }, { "epoch": 23.42, "learning_rate": 4.360000000000001e-06, "loss": 0.001, "step": 11500 }, { "epoch": 23.47, "learning_rate": 4.3471794871794874e-06, "loss": 0.0005, "step": 11525 }, { "epoch": 23.52, "learning_rate": 4.334358974358975e-06, "loss": 0.0008, "step": 11550 }, { "epoch": 23.57, "learning_rate": 4.321538461538462e-06, "loss": 0.0007, "step": 11575 }, { "epoch": 23.63, "learning_rate": 4.30871794871795e-06, "loss": 0.0015, "step": 11600 }, { "epoch": 23.68, "learning_rate": 4.2958974358974364e-06, "loss": 0.0014, "step": 11625 }, { "epoch": 23.73, "learning_rate": 4.283076923076924e-06, "loss": 0.001, "step": 11650 }, { "epoch": 23.78, "learning_rate": 4.27025641025641e-06, "loss": 0.001, "step": 11675 }, { "epoch": 23.83, "learning_rate": 4.257435897435898e-06, "loss": 0.0006, "step": 11700 }, { "epoch": 23.88, "learning_rate": 4.2446153846153846e-06, "loss": 0.0007, "step": 11725 }, { "epoch": 23.93, "learning_rate": 4.231794871794872e-06, "loss": 0.0005, "step": 11750 }, { "epoch": 23.98, "learning_rate": 4.218974358974359e-06, "loss": 0.0004, "step": 11775 }, { "epoch": 24.03, "learning_rate": 4.206153846153847e-06, "loss": 0.0005, "step": 11800 }, { "epoch": 24.08, "learning_rate": 4.1933333333333336e-06, "loss": 0.0003, "step": 11825 }, { "epoch": 24.13, "learning_rate": 4.180512820512821e-06, "loss": 0.0005, "step": 11850 }, { "epoch": 24.19, "learning_rate": 4.167692307692308e-06, "loss": 0.0012, "step": 11875 }, { "epoch": 24.24, "learning_rate": 4.154871794871796e-06, "loss": 0.0016, "step": 11900 }, { "epoch": 24.29, "learning_rate": 4.1420512820512826e-06, "loss": 0.0009, "step": 11925 }, { "epoch": 24.34, "learning_rate": 4.12923076923077e-06, "loss": 0.0004, "step": 11950 }, { "epoch": 24.39, "learning_rate": 4.116410256410256e-06, "loss": 0.0006, "step": 11975 }, { "epoch": 24.44, "learning_rate": 4.103589743589744e-06, "loss": 0.0004, "step": 12000 }, { "epoch": 24.44, "eval_loss": 0.26515766978263855, "eval_runtime": 3196.7126, "eval_samples_per_second": 4.914, "eval_steps_per_second": 0.307, "eval_wer": 4.989831726297016, "step": 12000 }, { "epoch": 24.49, "learning_rate": 4.090769230769231e-06, "loss": 0.0005, "step": 12025 }, { "epoch": 24.54, "learning_rate": 4.077948717948718e-06, "loss": 0.0007, "step": 12050 }, { "epoch": 24.59, "learning_rate": 4.065128205128205e-06, "loss": 0.0005, "step": 12075 }, { "epoch": 24.64, "learning_rate": 4.052307692307693e-06, "loss": 0.0007, "step": 12100 }, { "epoch": 24.69, "learning_rate": 4.03948717948718e-06, "loss": 0.0004, "step": 12125 }, { "epoch": 24.75, "learning_rate": 4.026666666666667e-06, "loss": 0.0009, "step": 12150 }, { "epoch": 24.8, "learning_rate": 4.013846153846154e-06, "loss": 0.0006, "step": 12175 }, { "epoch": 24.85, "learning_rate": 4.001025641025642e-06, "loss": 0.0004, "step": 12200 }, { "epoch": 24.9, "learning_rate": 3.988205128205129e-06, "loss": 0.0007, "step": 12225 }, { "epoch": 24.95, "learning_rate": 3.975384615384616e-06, "loss": 0.0006, "step": 12250 }, { "epoch": 25.0, "learning_rate": 3.962564102564102e-06, "loss": 0.0004, "step": 12275 }, { "epoch": 25.05, "learning_rate": 3.94974358974359e-06, "loss": 0.0008, "step": 12300 }, { "epoch": 25.1, "learning_rate": 3.936923076923077e-06, "loss": 0.001, "step": 12325 }, { "epoch": 25.15, "learning_rate": 3.9241025641025645e-06, "loss": 0.0004, "step": 12350 }, { "epoch": 25.2, "learning_rate": 3.911282051282051e-06, "loss": 0.0003, "step": 12375 }, { "epoch": 25.25, "learning_rate": 3.898461538461539e-06, "loss": 0.0004, "step": 12400 }, { "epoch": 25.31, "learning_rate": 3.885641025641026e-06, "loss": 0.0002, "step": 12425 }, { "epoch": 25.36, "learning_rate": 3.8728205128205135e-06, "loss": 0.0006, "step": 12450 }, { "epoch": 25.41, "learning_rate": 3.86e-06, "loss": 0.0005, "step": 12475 }, { "epoch": 25.46, "learning_rate": 3.847179487179488e-06, "loss": 0.0003, "step": 12500 }, { "epoch": 25.51, "learning_rate": 3.834358974358975e-06, "loss": 0.0002, "step": 12525 }, { "epoch": 25.56, "learning_rate": 3.8215384615384625e-06, "loss": 0.0008, "step": 12550 }, { "epoch": 25.61, "learning_rate": 3.808717948717949e-06, "loss": 0.0004, "step": 12575 }, { "epoch": 25.66, "learning_rate": 3.795897435897436e-06, "loss": 0.0003, "step": 12600 }, { "epoch": 25.71, "learning_rate": 3.7830769230769234e-06, "loss": 0.0002, "step": 12625 }, { "epoch": 25.76, "learning_rate": 3.7702564102564106e-06, "loss": 0.0007, "step": 12650 }, { "epoch": 25.81, "learning_rate": 3.7574358974358974e-06, "loss": 0.0007, "step": 12675 }, { "epoch": 25.87, "learning_rate": 3.7446153846153847e-06, "loss": 0.0005, "step": 12700 }, { "epoch": 25.92, "learning_rate": 3.731794871794872e-06, "loss": 0.0003, "step": 12725 }, { "epoch": 25.97, "learning_rate": 3.718974358974359e-06, "loss": 0.0008, "step": 12750 }, { "epoch": 26.02, "learning_rate": 3.7061538461538464e-06, "loss": 0.0006, "step": 12775 }, { "epoch": 26.07, "learning_rate": 3.6933333333333337e-06, "loss": 0.0003, "step": 12800 }, { "epoch": 26.12, "learning_rate": 3.680512820512821e-06, "loss": 0.0002, "step": 12825 }, { "epoch": 26.17, "learning_rate": 3.667692307692308e-06, "loss": 0.0002, "step": 12850 }, { "epoch": 26.22, "learning_rate": 3.6548717948717954e-06, "loss": 0.0004, "step": 12875 }, { "epoch": 26.27, "learning_rate": 3.6420512820512823e-06, "loss": 0.0003, "step": 12900 }, { "epoch": 26.32, "learning_rate": 3.6292307692307695e-06, "loss": 0.0001, "step": 12925 }, { "epoch": 26.37, "learning_rate": 3.6164102564102568e-06, "loss": 0.0002, "step": 12950 }, { "epoch": 26.43, "learning_rate": 3.6035897435897436e-06, "loss": 0.0002, "step": 12975 }, { "epoch": 26.48, "learning_rate": 3.590769230769231e-06, "loss": 0.0004, "step": 13000 }, { "epoch": 26.48, "eval_loss": 0.2849521338939667, "eval_runtime": 3188.1273, "eval_samples_per_second": 4.927, "eval_steps_per_second": 0.308, "eval_wer": 4.977642628482714, "step": 13000 }, { "epoch": 26.53, "learning_rate": 3.577948717948718e-06, "loss": 0.0002, "step": 13025 }, { "epoch": 26.58, "learning_rate": 3.5651282051282053e-06, "loss": 0.0002, "step": 13050 }, { "epoch": 26.63, "learning_rate": 3.5523076923076926e-06, "loss": 0.0002, "step": 13075 }, { "epoch": 26.68, "learning_rate": 3.53948717948718e-06, "loss": 0.0002, "step": 13100 }, { "epoch": 26.73, "learning_rate": 3.526666666666667e-06, "loss": 0.0005, "step": 13125 }, { "epoch": 26.78, "learning_rate": 3.5138461538461543e-06, "loss": 0.0002, "step": 13150 }, { "epoch": 26.83, "learning_rate": 3.5010256410256416e-06, "loss": 0.0002, "step": 13175 }, { "epoch": 26.88, "learning_rate": 3.4882051282051284e-06, "loss": 0.0002, "step": 13200 }, { "epoch": 26.93, "learning_rate": 3.4753846153846156e-06, "loss": 0.0001, "step": 13225 }, { "epoch": 26.99, "learning_rate": 3.462564102564103e-06, "loss": 0.0001, "step": 13250 }, { "epoch": 27.04, "learning_rate": 3.4497435897435897e-06, "loss": 0.0001, "step": 13275 }, { "epoch": 27.09, "learning_rate": 3.436923076923077e-06, "loss": 0.0002, "step": 13300 }, { "epoch": 27.14, "learning_rate": 3.424102564102564e-06, "loss": 0.0001, "step": 13325 }, { "epoch": 27.19, "learning_rate": 3.4112820512820514e-06, "loss": 0.0001, "step": 13350 }, { "epoch": 27.24, "learning_rate": 3.3984615384615387e-06, "loss": 0.0001, "step": 13375 }, { "epoch": 27.29, "learning_rate": 3.385641025641026e-06, "loss": 0.0001, "step": 13400 }, { "epoch": 27.34, "learning_rate": 3.372820512820513e-06, "loss": 0.0001, "step": 13425 }, { "epoch": 27.39, "learning_rate": 3.3600000000000004e-06, "loss": 0.0002, "step": 13450 }, { "epoch": 27.44, "learning_rate": 3.3471794871794877e-06, "loss": 0.0001, "step": 13475 }, { "epoch": 27.49, "learning_rate": 3.3343589743589745e-06, "loss": 0.0004, "step": 13500 }, { "epoch": 27.55, "learning_rate": 3.3215384615384618e-06, "loss": 0.0003, "step": 13525 }, { "epoch": 27.6, "learning_rate": 3.308717948717949e-06, "loss": 0.0003, "step": 13550 }, { "epoch": 27.65, "learning_rate": 3.2958974358974363e-06, "loss": 0.0008, "step": 13575 }, { "epoch": 27.7, "learning_rate": 3.283076923076923e-06, "loss": 0.0004, "step": 13600 }, { "epoch": 27.75, "learning_rate": 3.2702564102564103e-06, "loss": 0.0004, "step": 13625 }, { "epoch": 27.8, "learning_rate": 3.2574358974358976e-06, "loss": 0.0003, "step": 13650 }, { "epoch": 27.85, "learning_rate": 3.244615384615385e-06, "loss": 0.0002, "step": 13675 }, { "epoch": 27.9, "learning_rate": 3.231794871794872e-06, "loss": 0.0003, "step": 13700 }, { "epoch": 27.95, "learning_rate": 3.2189743589743593e-06, "loss": 0.0002, "step": 13725 }, { "epoch": 28.0, "learning_rate": 3.2061538461538466e-06, "loss": 0.0003, "step": 13750 }, { "epoch": 28.05, "learning_rate": 3.193333333333334e-06, "loss": 0.0009, "step": 13775 }, { "epoch": 28.11, "learning_rate": 3.1805128205128206e-06, "loss": 0.0002, "step": 13800 }, { "epoch": 28.16, "learning_rate": 3.167692307692308e-06, "loss": 0.0005, "step": 13825 }, { "epoch": 28.21, "learning_rate": 3.154871794871795e-06, "loss": 0.0004, "step": 13850 }, { "epoch": 28.26, "learning_rate": 3.1420512820512824e-06, "loss": 0.0005, "step": 13875 }, { "epoch": 28.31, "learning_rate": 3.129230769230769e-06, "loss": 0.0004, "step": 13900 }, { "epoch": 28.36, "learning_rate": 3.1164102564102565e-06, "loss": 0.0007, "step": 13925 }, { "epoch": 28.41, "learning_rate": 3.1035897435897437e-06, "loss": 0.0005, "step": 13950 }, { "epoch": 28.46, "learning_rate": 3.090769230769231e-06, "loss": 0.0005, "step": 13975 }, { "epoch": 28.51, "learning_rate": 3.077948717948718e-06, "loss": 0.0006, "step": 14000 }, { "epoch": 28.51, "eval_loss": 0.27358242869377136, "eval_runtime": 3189.207, "eval_samples_per_second": 4.925, "eval_steps_per_second": 0.308, "eval_wer": 4.97315190823534, "step": 14000 }, { "epoch": 28.56, "learning_rate": 3.0651282051282054e-06, "loss": 0.0004, "step": 14025 }, { "epoch": 28.62, "learning_rate": 3.0523076923076927e-06, "loss": 0.0005, "step": 14050 }, { "epoch": 28.67, "learning_rate": 3.03948717948718e-06, "loss": 0.0003, "step": 14075 }, { "epoch": 28.72, "learning_rate": 3.0266666666666668e-06, "loss": 0.0002, "step": 14100 }, { "epoch": 28.77, "learning_rate": 3.013846153846154e-06, "loss": 0.0003, "step": 14125 }, { "epoch": 28.82, "learning_rate": 3.0010256410256413e-06, "loss": 0.0003, "step": 14150 }, { "epoch": 28.87, "learning_rate": 2.9882051282051285e-06, "loss": 0.0002, "step": 14175 }, { "epoch": 28.92, "learning_rate": 2.9753846153846153e-06, "loss": 0.0002, "step": 14200 }, { "epoch": 28.97, "learning_rate": 2.9625641025641026e-06, "loss": 0.0002, "step": 14225 }, { "epoch": 29.02, "learning_rate": 2.94974358974359e-06, "loss": 0.0001, "step": 14250 }, { "epoch": 29.07, "learning_rate": 2.936923076923077e-06, "loss": 0.0001, "step": 14275 }, { "epoch": 29.12, "learning_rate": 2.9241025641025643e-06, "loss": 0.0002, "step": 14300 }, { "epoch": 29.18, "learning_rate": 2.9112820512820516e-06, "loss": 0.0001, "step": 14325 }, { "epoch": 29.23, "learning_rate": 2.898461538461539e-06, "loss": 0.0001, "step": 14350 }, { "epoch": 29.28, "learning_rate": 2.885641025641026e-06, "loss": 0.0001, "step": 14375 }, { "epoch": 29.33, "learning_rate": 2.8728205128205133e-06, "loss": 0.0002, "step": 14400 }, { "epoch": 29.38, "learning_rate": 2.86e-06, "loss": 0.0001, "step": 14425 }, { "epoch": 29.43, "learning_rate": 2.8471794871794874e-06, "loss": 0.0004, "step": 14450 }, { "epoch": 29.48, "learning_rate": 2.8343589743589746e-06, "loss": 0.0007, "step": 14475 }, { "epoch": 29.53, "learning_rate": 2.8215384615384615e-06, "loss": 0.0003, "step": 14500 }, { "epoch": 29.58, "learning_rate": 2.8087179487179487e-06, "loss": 0.0003, "step": 14525 }, { "epoch": 29.63, "learning_rate": 2.796410256410257e-06, "loss": 0.0003, "step": 14550 }, { "epoch": 29.68, "learning_rate": 2.7835897435897434e-06, "loss": 0.0003, "step": 14575 }, { "epoch": 29.74, "learning_rate": 2.7707692307692306e-06, "loss": 0.0002, "step": 14600 }, { "epoch": 29.79, "learning_rate": 2.757948717948718e-06, "loss": 0.0003, "step": 14625 }, { "epoch": 29.84, "learning_rate": 2.745128205128205e-06, "loss": 0.0002, "step": 14650 }, { "epoch": 29.89, "learning_rate": 2.7323076923076924e-06, "loss": 0.0005, "step": 14675 }, { "epoch": 29.94, "learning_rate": 2.7194871794871796e-06, "loss": 0.0005, "step": 14700 }, { "epoch": 29.99, "learning_rate": 2.706666666666667e-06, "loss": 0.0008, "step": 14725 }, { "epoch": 30.04, "learning_rate": 2.693846153846154e-06, "loss": 0.0004, "step": 14750 }, { "epoch": 30.09, "learning_rate": 2.6810256410256414e-06, "loss": 0.0004, "step": 14775 }, { "epoch": 30.14, "learning_rate": 2.6682051282051286e-06, "loss": 0.0003, "step": 14800 }, { "epoch": 30.19, "learning_rate": 2.655384615384616e-06, "loss": 0.0002, "step": 14825 }, { "epoch": 30.24, "learning_rate": 2.642564102564103e-06, "loss": 0.0005, "step": 14850 }, { "epoch": 30.3, "learning_rate": 2.6297435897435904e-06, "loss": 0.0004, "step": 14875 }, { "epoch": 30.35, "learning_rate": 2.6169230769230768e-06, "loss": 0.0004, "step": 14900 }, { "epoch": 30.4, "learning_rate": 2.604102564102564e-06, "loss": 0.0002, "step": 14925 }, { "epoch": 30.45, "learning_rate": 2.5912820512820513e-06, "loss": 0.0003, "step": 14950 }, { "epoch": 30.5, "learning_rate": 2.5784615384615385e-06, "loss": 0.0003, "step": 14975 }, { "epoch": 30.55, "learning_rate": 2.5656410256410257e-06, "loss": 0.0002, "step": 15000 }, { "epoch": 30.55, "eval_loss": 0.29437124729156494, "eval_runtime": 3190.5782, "eval_samples_per_second": 4.923, "eval_steps_per_second": 0.308, "eval_wer": 5.156629906913785, "step": 15000 }, { "epoch": 30.6, "learning_rate": 2.552820512820513e-06, "loss": 0.0005, "step": 15025 }, { "epoch": 30.65, "learning_rate": 2.5400000000000002e-06, "loss": 0.0003, "step": 15050 }, { "epoch": 30.7, "learning_rate": 2.5271794871794875e-06, "loss": 0.0005, "step": 15075 }, { "epoch": 30.75, "learning_rate": 2.5143589743589747e-06, "loss": 0.0004, "step": 15100 }, { "epoch": 30.8, "learning_rate": 2.501538461538462e-06, "loss": 0.0006, "step": 15125 }, { "epoch": 30.86, "learning_rate": 2.488717948717949e-06, "loss": 0.0005, "step": 15150 }, { "epoch": 30.91, "learning_rate": 2.475897435897436e-06, "loss": 0.0008, "step": 15175 }, { "epoch": 30.96, "learning_rate": 2.4630769230769233e-06, "loss": 0.0006, "step": 15200 }, { "epoch": 31.01, "learning_rate": 2.4502564102564106e-06, "loss": 0.0005, "step": 15225 }, { "epoch": 31.06, "learning_rate": 2.437435897435898e-06, "loss": 0.0005, "step": 15250 }, { "epoch": 31.11, "learning_rate": 2.4246153846153846e-06, "loss": 0.0005, "step": 15275 }, { "epoch": 31.16, "learning_rate": 2.411794871794872e-06, "loss": 0.0005, "step": 15300 }, { "epoch": 31.21, "learning_rate": 2.398974358974359e-06, "loss": 0.0006, "step": 15325 }, { "epoch": 31.26, "learning_rate": 2.3861538461538464e-06, "loss": 0.0002, "step": 15350 }, { "epoch": 31.31, "learning_rate": 2.3733333333333336e-06, "loss": 0.0004, "step": 15375 }, { "epoch": 31.36, "learning_rate": 2.3610256410256415e-06, "loss": 0.0005, "step": 15400 }, { "epoch": 31.42, "learning_rate": 2.3482051282051287e-06, "loss": 0.0001, "step": 15425 }, { "epoch": 31.47, "learning_rate": 2.3353846153846155e-06, "loss": 0.0003, "step": 15450 }, { "epoch": 31.52, "learning_rate": 2.3225641025641028e-06, "loss": 0.0004, "step": 15475 }, { "epoch": 31.57, "learning_rate": 2.30974358974359e-06, "loss": 0.0002, "step": 15500 }, { "epoch": 31.62, "learning_rate": 2.2969230769230773e-06, "loss": 0.0001, "step": 15525 }, { "epoch": 31.67, "learning_rate": 2.2841025641025645e-06, "loss": 0.0001, "step": 15550 }, { "epoch": 31.72, "learning_rate": 2.2712820512820518e-06, "loss": 0.0004, "step": 15575 }, { "epoch": 31.77, "learning_rate": 2.2584615384615386e-06, "loss": 0.0003, "step": 15600 }, { "epoch": 31.82, "learning_rate": 2.245641025641026e-06, "loss": 0.0006, "step": 15625 }, { "epoch": 31.87, "learning_rate": 2.232820512820513e-06, "loss": 0.0006, "step": 15650 }, { "epoch": 31.92, "learning_rate": 2.2200000000000003e-06, "loss": 0.0001, "step": 15675 }, { "epoch": 31.98, "learning_rate": 2.2071794871794876e-06, "loss": 0.0001, "step": 15700 }, { "epoch": 32.03, "learning_rate": 2.194358974358975e-06, "loss": 0.0002, "step": 15725 }, { "epoch": 32.08, "learning_rate": 2.1815384615384617e-06, "loss": 0.0003, "step": 15750 }, { "epoch": 32.13, "learning_rate": 2.168717948717949e-06, "loss": 0.0001, "step": 15775 }, { "epoch": 32.18, "learning_rate": 2.155897435897436e-06, "loss": 0.0001, "step": 15800 }, { "epoch": 32.23, "learning_rate": 2.1430769230769234e-06, "loss": 0.0001, "step": 15825 }, { "epoch": 32.28, "learning_rate": 2.1302564102564107e-06, "loss": 0.0001, "step": 15850 }, { "epoch": 32.33, "learning_rate": 2.117435897435898e-06, "loss": 0.0001, "step": 15875 }, { "epoch": 32.38, "learning_rate": 2.1046153846153847e-06, "loss": 0.0001, "step": 15900 }, { "epoch": 32.43, "learning_rate": 2.091794871794872e-06, "loss": 0.0001, "step": 15925 }, { "epoch": 32.48, "learning_rate": 2.0789743589743592e-06, "loss": 0.0002, "step": 15950 }, { "epoch": 32.54, "learning_rate": 2.0661538461538465e-06, "loss": 0.0001, "step": 15975 }, { "epoch": 32.59, "learning_rate": 2.0533333333333337e-06, "loss": 0.0002, "step": 16000 }, { "epoch": 32.59, "eval_loss": 0.2949441373348236, "eval_runtime": 3187.4074, "eval_samples_per_second": 4.928, "eval_steps_per_second": 0.308, "eval_wer": 5.000737761183497, "step": 16000 }, { "epoch": 32.64, "learning_rate": 2.040512820512821e-06, "loss": 0.0001, "step": 16025 }, { "epoch": 32.69, "learning_rate": 2.027692307692308e-06, "loss": 0.0001, "step": 16050 }, { "epoch": 32.74, "learning_rate": 2.014871794871795e-06, "loss": 0.0001, "step": 16075 }, { "epoch": 32.79, "learning_rate": 2.0020512820512823e-06, "loss": 0.0001, "step": 16100 }, { "epoch": 32.84, "learning_rate": 1.9892307692307695e-06, "loss": 0.0001, "step": 16125 }, { "epoch": 32.89, "learning_rate": 1.9764102564102568e-06, "loss": 0.0001, "step": 16150 }, { "epoch": 32.94, "learning_rate": 1.963589743589744e-06, "loss": 0.0001, "step": 16175 }, { "epoch": 32.99, "learning_rate": 1.950769230769231e-06, "loss": 0.0001, "step": 16200 }, { "epoch": 33.04, "learning_rate": 1.937948717948718e-06, "loss": 0.0001, "step": 16225 }, { "epoch": 33.1, "learning_rate": 1.9251282051282054e-06, "loss": 0.0001, "step": 16250 }, { "epoch": 33.15, "learning_rate": 1.9123076923076926e-06, "loss": 0.0001, "step": 16275 }, { "epoch": 33.2, "learning_rate": 1.8994871794871796e-06, "loss": 0.0001, "step": 16300 }, { "epoch": 33.25, "learning_rate": 1.8866666666666669e-06, "loss": 0.0001, "step": 16325 }, { "epoch": 33.3, "learning_rate": 1.873846153846154e-06, "loss": 0.0001, "step": 16350 }, { "epoch": 33.35, "learning_rate": 1.8610256410256412e-06, "loss": 0.0001, "step": 16375 }, { "epoch": 33.4, "learning_rate": 1.8482051282051284e-06, "loss": 0.0001, "step": 16400 }, { "epoch": 33.45, "learning_rate": 1.8353846153846155e-06, "loss": 0.0001, "step": 16425 }, { "epoch": 33.5, "learning_rate": 1.8225641025641027e-06, "loss": 0.0001, "step": 16450 }, { "epoch": 33.55, "learning_rate": 1.80974358974359e-06, "loss": 0.0001, "step": 16475 }, { "epoch": 33.6, "learning_rate": 1.796923076923077e-06, "loss": 0.0001, "step": 16500 }, { "epoch": 33.66, "learning_rate": 1.7841025641025642e-06, "loss": 0.0001, "step": 16525 }, { "epoch": 33.71, "learning_rate": 1.7712820512820515e-06, "loss": 0.0001, "step": 16550 }, { "epoch": 33.76, "learning_rate": 1.7584615384615385e-06, "loss": 0.0001, "step": 16575 }, { "epoch": 33.81, "learning_rate": 1.7456410256410258e-06, "loss": 0.0001, "step": 16600 }, { "epoch": 33.86, "learning_rate": 1.732820512820513e-06, "loss": 0.0001, "step": 16625 }, { "epoch": 33.91, "learning_rate": 1.72e-06, "loss": 0.0001, "step": 16650 }, { "epoch": 33.96, "learning_rate": 1.7071794871794873e-06, "loss": 0.0001, "step": 16675 }, { "epoch": 34.01, "learning_rate": 1.6943589743589745e-06, "loss": 0.0001, "step": 16700 }, { "epoch": 34.06, "learning_rate": 1.6815384615384616e-06, "loss": 0.0004, "step": 16725 }, { "epoch": 34.11, "learning_rate": 1.6687179487179488e-06, "loss": 0.0001, "step": 16750 }, { "epoch": 34.16, "learning_rate": 1.655897435897436e-06, "loss": 0.0001, "step": 16775 }, { "epoch": 34.22, "learning_rate": 1.6430769230769233e-06, "loss": 0.0001, "step": 16800 }, { "epoch": 34.27, "learning_rate": 1.6302564102564104e-06, "loss": 0.0001, "step": 16825 }, { "epoch": 34.32, "learning_rate": 1.6174358974358976e-06, "loss": 0.0001, "step": 16850 }, { "epoch": 34.37, "learning_rate": 1.6046153846153846e-06, "loss": 0.0001, "step": 16875 }, { "epoch": 34.42, "learning_rate": 1.5917948717948719e-06, "loss": 0.0001, "step": 16900 }, { "epoch": 34.47, "learning_rate": 1.5789743589743591e-06, "loss": 0.0001, "step": 16925 }, { "epoch": 34.52, "learning_rate": 1.5661538461538464e-06, "loss": 0.0001, "step": 16950 }, { "epoch": 34.57, "learning_rate": 1.5533333333333334e-06, "loss": 0.0001, "step": 16975 }, { "epoch": 34.62, "learning_rate": 1.5405128205128207e-06, "loss": 0.0001, "step": 17000 }, { "epoch": 34.62, "eval_loss": 0.30941757559776306, "eval_runtime": 3183.3419, "eval_samples_per_second": 4.934, "eval_steps_per_second": 0.308, "eval_wer": 4.955189027245841, "step": 17000 }, { "epoch": 34.67, "learning_rate": 1.5276923076923077e-06, "loss": 0.0001, "step": 17025 }, { "epoch": 34.73, "learning_rate": 1.514871794871795e-06, "loss": 0.0001, "step": 17050 }, { "epoch": 34.78, "learning_rate": 1.5020512820512822e-06, "loss": 0.0001, "step": 17075 }, { "epoch": 34.83, "learning_rate": 1.4892307692307695e-06, "loss": 0.0001, "step": 17100 }, { "epoch": 34.88, "learning_rate": 1.4764102564102565e-06, "loss": 0.0001, "step": 17125 }, { "epoch": 34.93, "learning_rate": 1.4635897435897437e-06, "loss": 0.0001, "step": 17150 }, { "epoch": 34.98, "learning_rate": 1.4507692307692308e-06, "loss": 0.0001, "step": 17175 }, { "epoch": 35.03, "learning_rate": 1.437948717948718e-06, "loss": 0.0001, "step": 17200 }, { "epoch": 35.08, "learning_rate": 1.4251282051282053e-06, "loss": 0.0001, "step": 17225 }, { "epoch": 35.13, "learning_rate": 1.4123076923076925e-06, "loss": 0.0, "step": 17250 }, { "epoch": 35.18, "learning_rate": 1.3994871794871796e-06, "loss": 0.0001, "step": 17275 }, { "epoch": 35.23, "learning_rate": 1.3866666666666668e-06, "loss": 0.0, "step": 17300 }, { "epoch": 35.29, "learning_rate": 1.3738461538461538e-06, "loss": 0.0001, "step": 17325 }, { "epoch": 35.34, "learning_rate": 1.361025641025641e-06, "loss": 0.0001, "step": 17350 }, { "epoch": 35.39, "learning_rate": 1.3482051282051283e-06, "loss": 0.0001, "step": 17375 }, { "epoch": 35.44, "learning_rate": 1.3353846153846156e-06, "loss": 0.0, "step": 17400 }, { "epoch": 35.49, "learning_rate": 1.3225641025641026e-06, "loss": 0.0001, "step": 17425 }, { "epoch": 35.54, "learning_rate": 1.3097435897435899e-06, "loss": 0.0001, "step": 17450 }, { "epoch": 35.59, "learning_rate": 1.296923076923077e-06, "loss": 0.0001, "step": 17475 }, { "epoch": 35.64, "learning_rate": 1.2841025641025641e-06, "loss": 0.0001, "step": 17500 }, { "epoch": 35.69, "learning_rate": 1.2712820512820514e-06, "loss": 0.0001, "step": 17525 }, { "epoch": 35.74, "learning_rate": 1.2584615384615386e-06, "loss": 0.0, "step": 17550 }, { "epoch": 35.79, "learning_rate": 1.2456410256410257e-06, "loss": 0.0001, "step": 17575 }, { "epoch": 35.85, "learning_rate": 1.232820512820513e-06, "loss": 0.0, "step": 17600 }, { "epoch": 35.9, "learning_rate": 1.2200000000000002e-06, "loss": 0.0001, "step": 17625 }, { "epoch": 35.95, "learning_rate": 1.2071794871794872e-06, "loss": 0.0, "step": 17650 }, { "epoch": 36.0, "learning_rate": 1.1943589743589745e-06, "loss": 0.0, "step": 17675 }, { "epoch": 36.05, "learning_rate": 1.1815384615384617e-06, "loss": 0.0, "step": 17700 }, { "epoch": 36.1, "learning_rate": 1.1687179487179487e-06, "loss": 0.0, "step": 17725 }, { "epoch": 36.15, "learning_rate": 1.155897435897436e-06, "loss": 0.0, "step": 17750 }, { "epoch": 36.2, "learning_rate": 1.1430769230769232e-06, "loss": 0.0, "step": 17775 }, { "epoch": 36.25, "learning_rate": 1.1302564102564103e-06, "loss": 0.0, "step": 17800 }, { "epoch": 36.3, "learning_rate": 1.1174358974358975e-06, "loss": 0.0, "step": 17825 }, { "epoch": 36.35, "learning_rate": 1.1046153846153848e-06, "loss": 0.0, "step": 17850 }, { "epoch": 36.41, "learning_rate": 1.0917948717948718e-06, "loss": 0.0, "step": 17875 }, { "epoch": 36.46, "learning_rate": 1.078974358974359e-06, "loss": 0.0, "step": 17900 }, { "epoch": 36.51, "learning_rate": 1.0661538461538463e-06, "loss": 0.0, "step": 17925 }, { "epoch": 36.56, "learning_rate": 1.0533333333333333e-06, "loss": 0.0, "step": 17950 }, { "epoch": 36.61, "learning_rate": 1.0405128205128206e-06, "loss": 0.0, "step": 17975 }, { "epoch": 36.66, "learning_rate": 1.0276923076923078e-06, "loss": 0.0, "step": 18000 }, { "epoch": 36.66, "eval_loss": 0.31845414638519287, "eval_runtime": 3188.8784, "eval_samples_per_second": 4.926, "eval_steps_per_second": 0.308, "eval_wer": 4.962245873348858, "step": 18000 }, { "epoch": 36.71, "learning_rate": 1.0148717948717949e-06, "loss": 0.0, "step": 18025 }, { "epoch": 36.76, "learning_rate": 1.0020512820512821e-06, "loss": 0.0, "step": 18050 }, { "epoch": 36.81, "learning_rate": 9.892307692307694e-07, "loss": 0.0, "step": 18075 }, { "epoch": 36.86, "learning_rate": 9.764102564102564e-07, "loss": 0.0, "step": 18100 }, { "epoch": 36.91, "learning_rate": 9.635897435897437e-07, "loss": 0.0, "step": 18125 }, { "epoch": 36.97, "learning_rate": 9.507692307692308e-07, "loss": 0.0, "step": 18150 }, { "epoch": 37.02, "learning_rate": 9.37948717948718e-07, "loss": 0.0, "step": 18175 }, { "epoch": 37.07, "learning_rate": 9.251282051282053e-07, "loss": 0.0, "step": 18200 }, { "epoch": 37.12, "learning_rate": 9.123076923076923e-07, "loss": 0.0, "step": 18225 }, { "epoch": 37.17, "learning_rate": 8.994871794871796e-07, "loss": 0.0, "step": 18250 }, { "epoch": 37.22, "learning_rate": 8.866666666666668e-07, "loss": 0.0, "step": 18275 }, { "epoch": 37.27, "learning_rate": 8.738461538461539e-07, "loss": 0.0, "step": 18300 }, { "epoch": 37.32, "learning_rate": 8.610256410256411e-07, "loss": 0.0, "step": 18325 }, { "epoch": 37.37, "learning_rate": 8.482051282051284e-07, "loss": 0.0, "step": 18350 }, { "epoch": 37.42, "learning_rate": 8.353846153846154e-07, "loss": 0.0, "step": 18375 }, { "epoch": 37.47, "learning_rate": 8.225641025641026e-07, "loss": 0.0, "step": 18400 }, { "epoch": 37.53, "learning_rate": 8.097435897435899e-07, "loss": 0.0, "step": 18425 }, { "epoch": 37.58, "learning_rate": 7.969230769230769e-07, "loss": 0.0, "step": 18450 }, { "epoch": 37.63, "learning_rate": 7.841025641025642e-07, "loss": 0.0, "step": 18475 }, { "epoch": 37.68, "learning_rate": 7.712820512820514e-07, "loss": 0.0, "step": 18500 }, { "epoch": 37.73, "learning_rate": 7.584615384615385e-07, "loss": 0.0, "step": 18525 }, { "epoch": 37.78, "learning_rate": 7.456410256410257e-07, "loss": 0.0, "step": 18550 }, { "epoch": 37.83, "learning_rate": 7.32820512820513e-07, "loss": 0.0, "step": 18575 }, { "epoch": 37.88, "learning_rate": 7.2e-07, "loss": 0.0, "step": 18600 }, { "epoch": 37.93, "learning_rate": 7.071794871794872e-07, "loss": 0.0, "step": 18625 }, { "epoch": 37.98, "learning_rate": 6.943589743589745e-07, "loss": 0.0, "step": 18650 }, { "epoch": 38.03, "learning_rate": 6.815384615384615e-07, "loss": 0.0, "step": 18675 }, { "epoch": 38.09, "learning_rate": 6.687179487179488e-07, "loss": 0.0, "step": 18700 }, { "epoch": 38.14, "learning_rate": 6.55897435897436e-07, "loss": 0.0, "step": 18725 }, { "epoch": 38.19, "learning_rate": 6.43076923076923e-07, "loss": 0.0, "step": 18750 }, { "epoch": 38.24, "learning_rate": 6.302564102564103e-07, "loss": 0.0, "step": 18775 }, { "epoch": 38.29, "learning_rate": 6.174358974358975e-07, "loss": 0.0, "step": 18800 }, { "epoch": 38.34, "learning_rate": 6.046153846153847e-07, "loss": 0.0, "step": 18825 }, { "epoch": 38.39, "learning_rate": 5.917948717948718e-07, "loss": 0.0, "step": 18850 }, { "epoch": 38.44, "learning_rate": 5.789743589743591e-07, "loss": 0.0, "step": 18875 }, { "epoch": 38.49, "learning_rate": 5.661538461538462e-07, "loss": 0.0, "step": 18900 }, { "epoch": 38.54, "learning_rate": 5.533333333333334e-07, "loss": 0.0, "step": 18925 }, { "epoch": 38.59, "learning_rate": 5.405128205128206e-07, "loss": 0.0, "step": 18950 }, { "epoch": 38.65, "learning_rate": 5.276923076923078e-07, "loss": 0.0, "step": 18975 }, { "epoch": 38.7, "learning_rate": 5.148717948717949e-07, "loss": 0.0, "step": 19000 }, { "epoch": 38.7, "eval_loss": 0.32285359501838684, "eval_runtime": 3211.4026, "eval_samples_per_second": 4.891, "eval_steps_per_second": 0.306, "eval_wer": 4.946207586751092, "step": 19000 }, { "epoch": 38.75, "learning_rate": 5.020512820512821e-07, "loss": 0.0, "step": 19025 }, { "epoch": 38.8, "learning_rate": 4.892307692307693e-07, "loss": 0.0, "step": 19050 }, { "epoch": 38.85, "learning_rate": 4.7641025641025643e-07, "loss": 0.0, "step": 19075 }, { "epoch": 38.9, "learning_rate": 4.635897435897437e-07, "loss": 0.0, "step": 19100 }, { "epoch": 38.95, "learning_rate": 4.507692307692308e-07, "loss": 0.0, "step": 19125 }, { "epoch": 39.0, "learning_rate": 4.3794871794871796e-07, "loss": 0.0, "step": 19150 }, { "epoch": 39.05, "learning_rate": 4.251282051282052e-07, "loss": 0.0, "step": 19175 }, { "epoch": 39.1, "learning_rate": 4.1230769230769235e-07, "loss": 0.0, "step": 19200 }, { "epoch": 39.15, "learning_rate": 3.994871794871795e-07, "loss": 0.0, "step": 19225 }, { "epoch": 39.21, "learning_rate": 3.8666666666666674e-07, "loss": 0.0, "step": 19250 }, { "epoch": 39.26, "learning_rate": 3.738461538461539e-07, "loss": 0.0, "step": 19275 }, { "epoch": 39.31, "learning_rate": 3.61025641025641e-07, "loss": 0.0, "step": 19300 }, { "epoch": 39.36, "learning_rate": 3.4820512820512827e-07, "loss": 0.0, "step": 19325 }, { "epoch": 39.41, "learning_rate": 3.353846153846154e-07, "loss": 0.0, "step": 19350 }, { "epoch": 39.46, "learning_rate": 3.2256410256410256e-07, "loss": 0.0, "step": 19375 }, { "epoch": 39.51, "learning_rate": 3.097435897435898e-07, "loss": 0.0, "step": 19400 }, { "epoch": 39.56, "learning_rate": 2.9692307692307695e-07, "loss": 0.0, "step": 19425 }, { "epoch": 39.61, "learning_rate": 2.8410256410256414e-07, "loss": 0.0, "step": 19450 }, { "epoch": 39.66, "learning_rate": 2.7128205128205134e-07, "loss": 0.0, "step": 19475 }, { "epoch": 39.71, "learning_rate": 2.584615384615385e-07, "loss": 0.0, "step": 19500 }, { "epoch": 39.77, "learning_rate": 2.456410256410257e-07, "loss": 0.0, "step": 19525 }, { "epoch": 39.82, "learning_rate": 2.3282051282051284e-07, "loss": 0.0, "step": 19550 }, { "epoch": 39.87, "learning_rate": 2.2e-07, "loss": 0.0, "step": 19575 }, { "epoch": 39.92, "learning_rate": 2.071794871794872e-07, "loss": 0.0, "step": 19600 }, { "epoch": 39.97, "learning_rate": 1.9435897435897437e-07, "loss": 0.0, "step": 19625 }, { "epoch": 40.02, "learning_rate": 1.8153846153846154e-07, "loss": 0.0, "step": 19650 }, { "epoch": 40.07, "learning_rate": 1.6871794871794874e-07, "loss": 0.0, "step": 19675 }, { "epoch": 40.12, "learning_rate": 1.558974358974359e-07, "loss": 0.0, "step": 19700 }, { "epoch": 40.17, "learning_rate": 1.430769230769231e-07, "loss": 0.0, "step": 19725 }, { "epoch": 40.22, "learning_rate": 1.3025641025641027e-07, "loss": 0.0, "step": 19750 }, { "epoch": 40.27, "learning_rate": 1.1743589743589745e-07, "loss": 0.0, "step": 19775 }, { "epoch": 40.33, "learning_rate": 1.0461538461538462e-07, "loss": 0.0, "step": 19800 }, { "epoch": 40.38, "learning_rate": 9.179487179487179e-08, "loss": 0.0, "step": 19825 }, { "epoch": 40.43, "learning_rate": 7.897435897435898e-08, "loss": 0.0, "step": 19850 }, { "epoch": 40.48, "learning_rate": 6.615384615384615e-08, "loss": 0.0, "step": 19875 }, { "epoch": 40.53, "learning_rate": 5.3333333333333334e-08, "loss": 0.0, "step": 19900 }, { "epoch": 40.58, "learning_rate": 4.0512820512820516e-08, "loss": 0.0, "step": 19925 }, { "epoch": 40.63, "learning_rate": 2.769230769230769e-08, "loss": 0.0, "step": 19950 }, { "epoch": 40.68, "learning_rate": 1.4871794871794873e-08, "loss": 0.0, "step": 19975 }, { "epoch": 40.73, "learning_rate": 2.051282051282051e-09, "loss": 0.0, "step": 20000 }, { "epoch": 40.73, "eval_loss": 0.32449695467948914, "eval_runtime": 3201.0591, "eval_samples_per_second": 4.907, "eval_steps_per_second": 0.307, "eval_wer": 4.9295277686894154, "step": 20000 }, { "epoch": 40.73, "step": 20000, "total_flos": 4.347710632938701e+21, "train_loss": 0.009992324647886562, "train_runtime": 175663.8097, "train_samples_per_second": 7.287, "train_steps_per_second": 0.114 } ], "logging_steps": 25, "max_steps": 20000, "num_input_tokens_seen": 0, "num_train_epochs": 41, "save_steps": 1000, "total_flos": 4.347710632938701e+21, "train_batch_size": 32, "trial_name": null, "trial_params": null }