{ "best_metric": 5.504444367585515, "best_model_checkpoint": "./checkpoint-2000", "epoch": 116.61807580174927, "eval_steps": 1000, "global_step": 20000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.15, "learning_rate": 5.000000000000001e-07, "loss": 3.1454, "step": 25 }, { "epoch": 0.29, "learning_rate": 1.0000000000000002e-06, "loss": 1.943, "step": 50 }, { "epoch": 0.44, "learning_rate": 1.5e-06, "loss": 1.1528, "step": 75 }, { "epoch": 0.58, "learning_rate": 2.0000000000000003e-06, "loss": 0.8458, "step": 100 }, { "epoch": 0.73, "learning_rate": 2.5e-06, "loss": 0.6877, "step": 125 }, { "epoch": 0.87, "learning_rate": 3e-06, "loss": 0.5627, "step": 150 }, { "epoch": 1.02, "learning_rate": 3.5e-06, "loss": 0.4879, "step": 175 }, { "epoch": 1.17, "learning_rate": 4.000000000000001e-06, "loss": 0.4146, "step": 200 }, { "epoch": 1.31, "learning_rate": 4.5e-06, "loss": 0.3823, "step": 225 }, { "epoch": 1.46, "learning_rate": 5e-06, "loss": 0.3575, "step": 250 }, { "epoch": 1.6, "learning_rate": 5.500000000000001e-06, "loss": 0.3306, "step": 275 }, { "epoch": 1.75, "learning_rate": 6e-06, "loss": 0.3027, "step": 300 }, { "epoch": 1.9, "learning_rate": 6.5000000000000004e-06, "loss": 0.3015, "step": 325 }, { "epoch": 2.04, "learning_rate": 7e-06, "loss": 0.2553, "step": 350 }, { "epoch": 2.19, "learning_rate": 7.500000000000001e-06, "loss": 0.2229, "step": 375 }, { "epoch": 2.33, "learning_rate": 8.000000000000001e-06, "loss": 0.2095, "step": 400 }, { "epoch": 2.48, "learning_rate": 8.5e-06, "loss": 0.2125, "step": 425 }, { "epoch": 2.62, "learning_rate": 9e-06, "loss": 0.2185, "step": 450 }, { "epoch": 2.77, "learning_rate": 9.5e-06, "loss": 0.2123, "step": 475 }, { "epoch": 2.92, "learning_rate": 1e-05, "loss": 0.2087, "step": 500 }, { "epoch": 3.06, "learning_rate": 9.987179487179488e-06, "loss": 0.1682, "step": 525 }, { "epoch": 3.21, "learning_rate": 9.974358974358974e-06, "loss": 0.1444, "step": 550 }, { "epoch": 3.35, "learning_rate": 9.961538461538463e-06, "loss": 0.1375, "step": 575 }, { "epoch": 3.5, "learning_rate": 9.94871794871795e-06, "loss": 0.1426, "step": 600 }, { "epoch": 3.64, "learning_rate": 9.935897435897437e-06, "loss": 0.1408, "step": 625 }, { "epoch": 3.79, "learning_rate": 9.923076923076923e-06, "loss": 0.1402, "step": 650 }, { "epoch": 3.94, "learning_rate": 9.910256410256412e-06, "loss": 0.1403, "step": 675 }, { "epoch": 4.08, "learning_rate": 9.897435897435899e-06, "loss": 0.1165, "step": 700 }, { "epoch": 4.23, "learning_rate": 9.884615384615386e-06, "loss": 0.094, "step": 725 }, { "epoch": 4.37, "learning_rate": 9.871794871794872e-06, "loss": 0.0946, "step": 750 }, { "epoch": 4.52, "learning_rate": 9.858974358974361e-06, "loss": 0.0956, "step": 775 }, { "epoch": 4.66, "learning_rate": 9.846153846153848e-06, "loss": 0.0938, "step": 800 }, { "epoch": 4.81, "learning_rate": 9.833333333333333e-06, "loss": 0.0916, "step": 825 }, { "epoch": 4.96, "learning_rate": 9.820512820512821e-06, "loss": 0.0963, "step": 850 }, { "epoch": 5.1, "learning_rate": 9.807692307692308e-06, "loss": 0.0736, "step": 875 }, { "epoch": 5.25, "learning_rate": 9.794871794871795e-06, "loss": 0.0675, "step": 900 }, { "epoch": 5.39, "learning_rate": 9.782051282051282e-06, "loss": 0.0651, "step": 925 }, { "epoch": 5.54, "learning_rate": 9.76923076923077e-06, "loss": 0.0682, "step": 950 }, { "epoch": 5.69, "learning_rate": 9.756410256410257e-06, "loss": 0.0696, "step": 975 }, { "epoch": 5.83, "learning_rate": 9.743589743589744e-06, "loss": 0.0659, "step": 1000 }, { "epoch": 5.83, "eval_loss": 0.16631454229354858, "eval_runtime": 1324.5084, "eval_samples_per_second": 4.83, "eval_steps_per_second": 0.302, "eval_wer": 5.959257081589596, "step": 1000 }, { "epoch": 5.98, "learning_rate": 9.730769230769231e-06, "loss": 0.071, "step": 1025 }, { "epoch": 6.12, "learning_rate": 9.71794871794872e-06, "loss": 0.0498, "step": 1050 }, { "epoch": 6.27, "learning_rate": 9.705128205128206e-06, "loss": 0.0451, "step": 1075 }, { "epoch": 6.41, "learning_rate": 9.692307692307693e-06, "loss": 0.0452, "step": 1100 }, { "epoch": 6.56, "learning_rate": 9.67948717948718e-06, "loss": 0.0517, "step": 1125 }, { "epoch": 6.71, "learning_rate": 9.666666666666667e-06, "loss": 0.0492, "step": 1150 }, { "epoch": 6.85, "learning_rate": 9.653846153846155e-06, "loss": 0.0499, "step": 1175 }, { "epoch": 7.0, "learning_rate": 9.641025641025642e-06, "loss": 0.0527, "step": 1200 }, { "epoch": 7.14, "learning_rate": 9.628205128205129e-06, "loss": 0.0344, "step": 1225 }, { "epoch": 7.29, "learning_rate": 9.615384615384616e-06, "loss": 0.0366, "step": 1250 }, { "epoch": 7.43, "learning_rate": 9.602564102564104e-06, "loss": 0.0343, "step": 1275 }, { "epoch": 7.58, "learning_rate": 9.589743589743591e-06, "loss": 0.0379, "step": 1300 }, { "epoch": 7.73, "learning_rate": 9.576923076923078e-06, "loss": 0.0359, "step": 1325 }, { "epoch": 7.87, "learning_rate": 9.564102564102565e-06, "loss": 0.0385, "step": 1350 }, { "epoch": 8.02, "learning_rate": 9.551282051282053e-06, "loss": 0.0387, "step": 1375 }, { "epoch": 8.16, "learning_rate": 9.53846153846154e-06, "loss": 0.0263, "step": 1400 }, { "epoch": 8.31, "learning_rate": 9.525641025641025e-06, "loss": 0.0287, "step": 1425 }, { "epoch": 8.45, "learning_rate": 9.512820512820514e-06, "loss": 0.0259, "step": 1450 }, { "epoch": 8.6, "learning_rate": 9.5e-06, "loss": 0.0284, "step": 1475 }, { "epoch": 8.75, "learning_rate": 9.487179487179487e-06, "loss": 0.0288, "step": 1500 }, { "epoch": 8.89, "learning_rate": 9.474358974358974e-06, "loss": 0.0275, "step": 1525 }, { "epoch": 9.04, "learning_rate": 9.461538461538463e-06, "loss": 0.0262, "step": 1550 }, { "epoch": 9.18, "learning_rate": 9.44871794871795e-06, "loss": 0.0203, "step": 1575 }, { "epoch": 9.33, "learning_rate": 9.435897435897436e-06, "loss": 0.0201, "step": 1600 }, { "epoch": 9.48, "learning_rate": 9.423076923076923e-06, "loss": 0.0218, "step": 1625 }, { "epoch": 9.62, "learning_rate": 9.410256410256412e-06, "loss": 0.0225, "step": 1650 }, { "epoch": 9.77, "learning_rate": 9.397435897435899e-06, "loss": 0.022, "step": 1675 }, { "epoch": 9.91, "learning_rate": 9.384615384615385e-06, "loss": 0.0222, "step": 1700 }, { "epoch": 10.06, "learning_rate": 9.371794871794872e-06, "loss": 0.02, "step": 1725 }, { "epoch": 10.2, "learning_rate": 9.358974358974359e-06, "loss": 0.0176, "step": 1750 }, { "epoch": 10.35, "learning_rate": 9.346153846153847e-06, "loss": 0.0168, "step": 1775 }, { "epoch": 10.5, "learning_rate": 9.333333333333334e-06, "loss": 0.0169, "step": 1800 }, { "epoch": 10.64, "learning_rate": 9.320512820512821e-06, "loss": 0.0149, "step": 1825 }, { "epoch": 10.79, "learning_rate": 9.307692307692308e-06, "loss": 0.0162, "step": 1850 }, { "epoch": 10.93, "learning_rate": 9.294871794871796e-06, "loss": 0.0178, "step": 1875 }, { "epoch": 11.08, "learning_rate": 9.282051282051283e-06, "loss": 0.0145, "step": 1900 }, { "epoch": 11.22, "learning_rate": 9.26923076923077e-06, "loss": 0.0134, "step": 1925 }, { "epoch": 11.37, "learning_rate": 9.256410256410257e-06, "loss": 0.0132, "step": 1950 }, { "epoch": 11.52, "learning_rate": 9.243589743589745e-06, "loss": 0.0152, "step": 1975 }, { "epoch": 11.66, "learning_rate": 9.230769230769232e-06, "loss": 0.0164, "step": 2000 }, { "epoch": 11.66, "eval_loss": 0.1947096288204193, "eval_runtime": 1315.5245, "eval_samples_per_second": 4.863, "eval_steps_per_second": 0.304, "eval_wer": 5.504444367585515, "step": 2000 }, { "epoch": 11.81, "learning_rate": 9.217948717948717e-06, "loss": 0.017, "step": 2025 }, { "epoch": 11.95, "learning_rate": 9.205128205128206e-06, "loss": 0.0159, "step": 2050 }, { "epoch": 12.1, "learning_rate": 9.192307692307693e-06, "loss": 0.0133, "step": 2075 }, { "epoch": 12.24, "learning_rate": 9.17948717948718e-06, "loss": 0.012, "step": 2100 }, { "epoch": 12.39, "learning_rate": 9.166666666666666e-06, "loss": 0.0136, "step": 2125 }, { "epoch": 12.54, "learning_rate": 9.153846153846155e-06, "loss": 0.0127, "step": 2150 }, { "epoch": 12.68, "learning_rate": 9.141025641025642e-06, "loss": 0.0102, "step": 2175 }, { "epoch": 12.83, "learning_rate": 9.128205128205129e-06, "loss": 0.0117, "step": 2200 }, { "epoch": 12.97, "learning_rate": 9.115384615384615e-06, "loss": 0.014, "step": 2225 }, { "epoch": 13.12, "learning_rate": 9.102564102564104e-06, "loss": 0.0109, "step": 2250 }, { "epoch": 13.27, "learning_rate": 9.08974358974359e-06, "loss": 0.0098, "step": 2275 }, { "epoch": 13.41, "learning_rate": 9.076923076923078e-06, "loss": 0.0104, "step": 2300 }, { "epoch": 13.56, "learning_rate": 9.064102564102564e-06, "loss": 0.011, "step": 2325 }, { "epoch": 13.7, "learning_rate": 9.051282051282051e-06, "loss": 0.0116, "step": 2350 }, { "epoch": 13.85, "learning_rate": 9.03846153846154e-06, "loss": 0.0093, "step": 2375 }, { "epoch": 13.99, "learning_rate": 9.025641025641027e-06, "loss": 0.0111, "step": 2400 }, { "epoch": 14.14, "learning_rate": 9.012820512820513e-06, "loss": 0.0079, "step": 2425 }, { "epoch": 14.29, "learning_rate": 9e-06, "loss": 0.0092, "step": 2450 }, { "epoch": 14.43, "learning_rate": 8.987179487179489e-06, "loss": 0.0108, "step": 2475 }, { "epoch": 14.58, "learning_rate": 8.974358974358976e-06, "loss": 0.0096, "step": 2500 }, { "epoch": 14.72, "learning_rate": 8.961538461538462e-06, "loss": 0.0081, "step": 2525 }, { "epoch": 14.87, "learning_rate": 8.94871794871795e-06, "loss": 0.0127, "step": 2550 }, { "epoch": 15.01, "learning_rate": 8.935897435897438e-06, "loss": 0.0105, "step": 2575 }, { "epoch": 15.16, "learning_rate": 8.923076923076925e-06, "loss": 0.0088, "step": 2600 }, { "epoch": 15.31, "learning_rate": 8.910256410256411e-06, "loss": 0.0083, "step": 2625 }, { "epoch": 15.45, "learning_rate": 8.897435897435898e-06, "loss": 0.0086, "step": 2650 }, { "epoch": 15.6, "learning_rate": 8.884615384615385e-06, "loss": 0.0107, "step": 2675 }, { "epoch": 15.74, "learning_rate": 8.871794871794872e-06, "loss": 0.0098, "step": 2700 }, { "epoch": 15.89, "learning_rate": 8.858974358974359e-06, "loss": 0.009, "step": 2725 }, { "epoch": 16.03, "learning_rate": 8.846153846153847e-06, "loss": 0.0087, "step": 2750 }, { "epoch": 16.18, "learning_rate": 8.833333333333334e-06, "loss": 0.0071, "step": 2775 }, { "epoch": 16.33, "learning_rate": 8.820512820512821e-06, "loss": 0.0069, "step": 2800 }, { "epoch": 16.47, "learning_rate": 8.807692307692308e-06, "loss": 0.0082, "step": 2825 }, { "epoch": 16.62, "learning_rate": 8.794871794871796e-06, "loss": 0.0082, "step": 2850 }, { "epoch": 16.76, "learning_rate": 8.782051282051283e-06, "loss": 0.0088, "step": 2875 }, { "epoch": 16.91, "learning_rate": 8.76923076923077e-06, "loss": 0.0079, "step": 2900 }, { "epoch": 17.06, "learning_rate": 8.756410256410257e-06, "loss": 0.0068, "step": 2925 }, { "epoch": 17.2, "learning_rate": 8.743589743589743e-06, "loss": 0.0067, "step": 2950 }, { "epoch": 17.35, "learning_rate": 8.730769230769232e-06, "loss": 0.007, "step": 2975 }, { "epoch": 17.49, "learning_rate": 8.717948717948719e-06, "loss": 0.0069, "step": 3000 }, { "epoch": 17.49, "eval_loss": 0.21648190915584564, "eval_runtime": 1318.9166, "eval_samples_per_second": 4.85, "eval_steps_per_second": 0.303, "eval_wer": 5.750008646629544, "step": 3000 }, { "epoch": 17.64, "learning_rate": 8.705128205128206e-06, "loss": 0.0072, "step": 3025 }, { "epoch": 17.78, "learning_rate": 8.692307692307692e-06, "loss": 0.0073, "step": 3050 }, { "epoch": 17.93, "learning_rate": 8.679487179487181e-06, "loss": 0.0067, "step": 3075 }, { "epoch": 18.08, "learning_rate": 8.666666666666668e-06, "loss": 0.007, "step": 3100 }, { "epoch": 18.22, "learning_rate": 8.653846153846155e-06, "loss": 0.0055, "step": 3125 }, { "epoch": 18.37, "learning_rate": 8.641025641025641e-06, "loss": 0.0058, "step": 3150 }, { "epoch": 18.51, "learning_rate": 8.62820512820513e-06, "loss": 0.0067, "step": 3175 }, { "epoch": 18.66, "learning_rate": 8.615384615384617e-06, "loss": 0.0065, "step": 3200 }, { "epoch": 18.8, "learning_rate": 8.602564102564104e-06, "loss": 0.007, "step": 3225 }, { "epoch": 18.95, "learning_rate": 8.58974358974359e-06, "loss": 0.008, "step": 3250 }, { "epoch": 19.1, "learning_rate": 8.576923076923077e-06, "loss": 0.0059, "step": 3275 }, { "epoch": 19.24, "learning_rate": 8.564102564102564e-06, "loss": 0.0067, "step": 3300 }, { "epoch": 19.39, "learning_rate": 8.551282051282051e-06, "loss": 0.0066, "step": 3325 }, { "epoch": 19.53, "learning_rate": 8.53846153846154e-06, "loss": 0.0069, "step": 3350 }, { "epoch": 19.68, "learning_rate": 8.525641025641026e-06, "loss": 0.007, "step": 3375 }, { "epoch": 19.83, "learning_rate": 8.512820512820513e-06, "loss": 0.0069, "step": 3400 }, { "epoch": 19.97, "learning_rate": 8.5e-06, "loss": 0.0055, "step": 3425 }, { "epoch": 20.12, "learning_rate": 8.487179487179488e-06, "loss": 0.0048, "step": 3450 }, { "epoch": 20.26, "learning_rate": 8.474358974358975e-06, "loss": 0.0043, "step": 3475 }, { "epoch": 20.41, "learning_rate": 8.461538461538462e-06, "loss": 0.0056, "step": 3500 }, { "epoch": 20.55, "learning_rate": 8.448717948717949e-06, "loss": 0.0057, "step": 3525 }, { "epoch": 20.7, "learning_rate": 8.435897435897436e-06, "loss": 0.0056, "step": 3550 }, { "epoch": 20.85, "learning_rate": 8.423076923076924e-06, "loss": 0.0071, "step": 3575 }, { "epoch": 20.99, "learning_rate": 8.410256410256411e-06, "loss": 0.005, "step": 3600 }, { "epoch": 21.14, "learning_rate": 8.397435897435898e-06, "loss": 0.0047, "step": 3625 }, { "epoch": 21.28, "learning_rate": 8.384615384615385e-06, "loss": 0.0049, "step": 3650 }, { "epoch": 21.43, "learning_rate": 8.371794871794873e-06, "loss": 0.0056, "step": 3675 }, { "epoch": 21.57, "learning_rate": 8.35897435897436e-06, "loss": 0.0068, "step": 3700 }, { "epoch": 21.72, "learning_rate": 8.346153846153847e-06, "loss": 0.006, "step": 3725 }, { "epoch": 21.87, "learning_rate": 8.333333333333334e-06, "loss": 0.0061, "step": 3750 }, { "epoch": 22.01, "learning_rate": 8.320512820512822e-06, "loss": 0.0062, "step": 3775 }, { "epoch": 22.16, "learning_rate": 8.307692307692309e-06, "loss": 0.0052, "step": 3800 }, { "epoch": 22.3, "learning_rate": 8.294871794871796e-06, "loss": 0.005, "step": 3825 }, { "epoch": 22.45, "learning_rate": 8.282051282051283e-06, "loss": 0.004, "step": 3850 }, { "epoch": 22.59, "learning_rate": 8.26923076923077e-06, "loss": 0.0049, "step": 3875 }, { "epoch": 22.74, "learning_rate": 8.256410256410256e-06, "loss": 0.0055, "step": 3900 }, { "epoch": 22.89, "learning_rate": 8.243589743589743e-06, "loss": 0.006, "step": 3925 }, { "epoch": 23.03, "learning_rate": 8.230769230769232e-06, "loss": 0.0058, "step": 3950 }, { "epoch": 23.18, "learning_rate": 8.217948717948719e-06, "loss": 0.0048, "step": 3975 }, { "epoch": 23.32, "learning_rate": 8.205128205128205e-06, "loss": 0.0052, "step": 4000 }, { "epoch": 23.32, "eval_loss": 0.22917507588863373, "eval_runtime": 1316.5572, "eval_samples_per_second": 4.859, "eval_steps_per_second": 0.304, "eval_wer": 5.92121191159686, "step": 4000 }, { "epoch": 23.47, "learning_rate": 8.192307692307692e-06, "loss": 0.0047, "step": 4025 }, { "epoch": 23.62, "learning_rate": 8.17948717948718e-06, "loss": 0.0046, "step": 4050 }, { "epoch": 23.76, "learning_rate": 8.166666666666668e-06, "loss": 0.0047, "step": 4075 }, { "epoch": 23.91, "learning_rate": 8.153846153846154e-06, "loss": 0.0052, "step": 4100 }, { "epoch": 24.05, "learning_rate": 8.141025641025641e-06, "loss": 0.0056, "step": 4125 }, { "epoch": 24.2, "learning_rate": 8.12820512820513e-06, "loss": 0.0054, "step": 4150 }, { "epoch": 24.34, "learning_rate": 8.115384615384617e-06, "loss": 0.0048, "step": 4175 }, { "epoch": 24.49, "learning_rate": 8.102564102564103e-06, "loss": 0.0051, "step": 4200 }, { "epoch": 24.64, "learning_rate": 8.08974358974359e-06, "loss": 0.0044, "step": 4225 }, { "epoch": 24.78, "learning_rate": 8.076923076923077e-06, "loss": 0.0051, "step": 4250 }, { "epoch": 24.93, "learning_rate": 8.064102564102566e-06, "loss": 0.0052, "step": 4275 }, { "epoch": 25.07, "learning_rate": 8.051282051282052e-06, "loss": 0.0044, "step": 4300 }, { "epoch": 25.22, "learning_rate": 8.03846153846154e-06, "loss": 0.0045, "step": 4325 }, { "epoch": 25.36, "learning_rate": 8.025641025641026e-06, "loss": 0.004, "step": 4350 }, { "epoch": 25.51, "learning_rate": 8.012820512820515e-06, "loss": 0.0041, "step": 4375 }, { "epoch": 25.66, "learning_rate": 8.000000000000001e-06, "loss": 0.0039, "step": 4400 }, { "epoch": 25.8, "learning_rate": 7.987179487179488e-06, "loss": 0.0051, "step": 4425 }, { "epoch": 25.95, "learning_rate": 7.974358974358975e-06, "loss": 0.0045, "step": 4450 }, { "epoch": 26.09, "learning_rate": 7.961538461538462e-06, "loss": 0.0047, "step": 4475 }, { "epoch": 26.24, "learning_rate": 7.948717948717949e-06, "loss": 0.0037, "step": 4500 }, { "epoch": 26.38, "learning_rate": 7.935897435897435e-06, "loss": 0.0031, "step": 4525 }, { "epoch": 26.53, "learning_rate": 7.923076923076924e-06, "loss": 0.0041, "step": 4550 }, { "epoch": 26.68, "learning_rate": 7.91025641025641e-06, "loss": 0.0031, "step": 4575 }, { "epoch": 26.82, "learning_rate": 7.897435897435898e-06, "loss": 0.0032, "step": 4600 }, { "epoch": 26.97, "learning_rate": 7.884615384615384e-06, "loss": 0.0045, "step": 4625 }, { "epoch": 27.11, "learning_rate": 7.871794871794873e-06, "loss": 0.0029, "step": 4650 }, { "epoch": 27.26, "learning_rate": 7.85897435897436e-06, "loss": 0.003, "step": 4675 }, { "epoch": 27.41, "learning_rate": 7.846153846153847e-06, "loss": 0.0032, "step": 4700 }, { "epoch": 27.55, "learning_rate": 7.833333333333333e-06, "loss": 0.0038, "step": 4725 }, { "epoch": 27.7, "learning_rate": 7.820512820512822e-06, "loss": 0.0037, "step": 4750 }, { "epoch": 27.84, "learning_rate": 7.807692307692309e-06, "loss": 0.0038, "step": 4775 }, { "epoch": 27.99, "learning_rate": 7.794871794871796e-06, "loss": 0.005, "step": 4800 }, { "epoch": 28.13, "learning_rate": 7.782051282051282e-06, "loss": 0.0037, "step": 4825 }, { "epoch": 28.28, "learning_rate": 7.76923076923077e-06, "loss": 0.0047, "step": 4850 }, { "epoch": 28.43, "learning_rate": 7.756410256410258e-06, "loss": 0.0038, "step": 4875 }, { "epoch": 28.57, "learning_rate": 7.743589743589745e-06, "loss": 0.0058, "step": 4900 }, { "epoch": 28.72, "learning_rate": 7.730769230769231e-06, "loss": 0.0046, "step": 4925 }, { "epoch": 28.86, "learning_rate": 7.717948717948718e-06, "loss": 0.0049, "step": 4950 }, { "epoch": 29.01, "learning_rate": 7.705128205128207e-06, "loss": 0.0047, "step": 4975 }, { "epoch": 29.15, "learning_rate": 7.692307692307694e-06, "loss": 0.0032, "step": 5000 }, { "epoch": 29.15, "eval_loss": 0.23203052580356598, "eval_runtime": 1316.573, "eval_samples_per_second": 4.859, "eval_steps_per_second": 0.304, "eval_wer": 5.888354719330405, "step": 5000 }, { "epoch": 29.3, "learning_rate": 7.67948717948718e-06, "loss": 0.003, "step": 5025 }, { "epoch": 29.45, "learning_rate": 7.666666666666667e-06, "loss": 0.0036, "step": 5050 }, { "epoch": 29.59, "learning_rate": 7.653846153846154e-06, "loss": 0.0034, "step": 5075 }, { "epoch": 29.74, "learning_rate": 7.641025641025641e-06, "loss": 0.0032, "step": 5100 }, { "epoch": 29.88, "learning_rate": 7.6282051282051286e-06, "loss": 0.0054, "step": 5125 }, { "epoch": 30.03, "learning_rate": 7.615384615384615e-06, "loss": 0.0055, "step": 5150 }, { "epoch": 30.17, "learning_rate": 7.602564102564103e-06, "loss": 0.0035, "step": 5175 }, { "epoch": 30.32, "learning_rate": 7.58974358974359e-06, "loss": 0.0037, "step": 5200 }, { "epoch": 30.47, "learning_rate": 7.5769230769230775e-06, "loss": 0.0036, "step": 5225 }, { "epoch": 30.61, "learning_rate": 7.564102564102564e-06, "loss": 0.0053, "step": 5250 }, { "epoch": 30.76, "learning_rate": 7.551282051282052e-06, "loss": 0.005, "step": 5275 }, { "epoch": 30.9, "learning_rate": 7.538461538461539e-06, "loss": 0.0045, "step": 5300 }, { "epoch": 31.05, "learning_rate": 7.5256410256410265e-06, "loss": 0.004, "step": 5325 }, { "epoch": 31.2, "learning_rate": 7.512820512820513e-06, "loss": 0.0031, "step": 5350 }, { "epoch": 31.34, "learning_rate": 7.500000000000001e-06, "loss": 0.003, "step": 5375 }, { "epoch": 31.49, "learning_rate": 7.487179487179488e-06, "loss": 0.0041, "step": 5400 }, { "epoch": 31.63, "learning_rate": 7.474358974358975e-06, "loss": 0.0044, "step": 5425 }, { "epoch": 31.78, "learning_rate": 7.461538461538462e-06, "loss": 0.0038, "step": 5450 }, { "epoch": 31.92, "learning_rate": 7.448717948717949e-06, "loss": 0.0038, "step": 5475 }, { "epoch": 32.07, "learning_rate": 7.435897435897437e-06, "loss": 0.0031, "step": 5500 }, { "epoch": 32.22, "learning_rate": 7.423076923076924e-06, "loss": 0.0031, "step": 5525 }, { "epoch": 32.36, "learning_rate": 7.410256410256411e-06, "loss": 0.0036, "step": 5550 }, { "epoch": 32.51, "learning_rate": 7.397435897435898e-06, "loss": 0.0044, "step": 5575 }, { "epoch": 32.65, "learning_rate": 7.384615384615386e-06, "loss": 0.0042, "step": 5600 }, { "epoch": 32.8, "learning_rate": 7.371794871794873e-06, "loss": 0.0038, "step": 5625 }, { "epoch": 32.94, "learning_rate": 7.35897435897436e-06, "loss": 0.0049, "step": 5650 }, { "epoch": 33.09, "learning_rate": 7.346666666666668e-06, "loss": 0.0041, "step": 5675 }, { "epoch": 33.24, "learning_rate": 7.333846153846155e-06, "loss": 0.0037, "step": 5700 }, { "epoch": 33.38, "learning_rate": 7.321025641025642e-06, "loss": 0.004, "step": 5725 }, { "epoch": 33.53, "learning_rate": 7.308205128205129e-06, "loss": 0.0045, "step": 5750 }, { "epoch": 33.67, "learning_rate": 7.295384615384617e-06, "loss": 0.0044, "step": 5775 }, { "epoch": 33.82, "learning_rate": 7.2825641025641036e-06, "loss": 0.003, "step": 5800 }, { "epoch": 33.97, "learning_rate": 7.269743589743591e-06, "loss": 0.0035, "step": 5825 }, { "epoch": 34.11, "learning_rate": 7.256923076923078e-06, "loss": 0.0033, "step": 5850 }, { "epoch": 34.26, "learning_rate": 7.244102564102564e-06, "loss": 0.0035, "step": 5875 }, { "epoch": 34.4, "learning_rate": 7.231282051282052e-06, "loss": 0.0027, "step": 5900 }, { "epoch": 34.55, "learning_rate": 7.2184615384615385e-06, "loss": 0.0043, "step": 5925 }, { "epoch": 34.69, "learning_rate": 7.205641025641026e-06, "loss": 0.0036, "step": 5950 }, { "epoch": 34.84, "learning_rate": 7.192820512820513e-06, "loss": 0.0035, "step": 5975 }, { "epoch": 34.99, "learning_rate": 7.180000000000001e-06, "loss": 0.0037, "step": 6000 }, { "epoch": 34.99, "eval_loss": 0.2434399425983429, "eval_runtime": 1317.8047, "eval_samples_per_second": 4.854, "eval_steps_per_second": 0.304, "eval_wer": 6.064745962024003, "step": 6000 }, { "epoch": 35.13, "learning_rate": 7.1671794871794875e-06, "loss": 0.0029, "step": 6025 }, { "epoch": 35.28, "learning_rate": 7.154358974358974e-06, "loss": 0.0034, "step": 6050 }, { "epoch": 35.42, "learning_rate": 7.141538461538462e-06, "loss": 0.0029, "step": 6075 }, { "epoch": 35.57, "learning_rate": 7.128717948717949e-06, "loss": 0.0024, "step": 6100 }, { "epoch": 35.71, "learning_rate": 7.1158974358974365e-06, "loss": 0.0027, "step": 6125 }, { "epoch": 35.86, "learning_rate": 7.103076923076923e-06, "loss": 0.0032, "step": 6150 }, { "epoch": 36.01, "learning_rate": 7.090256410256411e-06, "loss": 0.003, "step": 6175 }, { "epoch": 36.15, "learning_rate": 7.077435897435898e-06, "loss": 0.002, "step": 6200 }, { "epoch": 36.3, "learning_rate": 7.0646153846153855e-06, "loss": 0.0022, "step": 6225 }, { "epoch": 36.44, "learning_rate": 7.051794871794872e-06, "loss": 0.0016, "step": 6250 }, { "epoch": 36.59, "learning_rate": 7.03897435897436e-06, "loss": 0.0016, "step": 6275 }, { "epoch": 36.73, "learning_rate": 7.026153846153847e-06, "loss": 0.0023, "step": 6300 }, { "epoch": 36.88, "learning_rate": 7.0133333333333345e-06, "loss": 0.0018, "step": 6325 }, { "epoch": 37.03, "learning_rate": 7.000512820512821e-06, "loss": 0.0022, "step": 6350 }, { "epoch": 37.17, "learning_rate": 6.987692307692309e-06, "loss": 0.0022, "step": 6375 }, { "epoch": 37.32, "learning_rate": 6.974871794871796e-06, "loss": 0.0029, "step": 6400 }, { "epoch": 37.46, "learning_rate": 6.9620512820512835e-06, "loss": 0.0035, "step": 6425 }, { "epoch": 37.61, "learning_rate": 6.94923076923077e-06, "loss": 0.0039, "step": 6450 }, { "epoch": 37.76, "learning_rate": 6.936410256410256e-06, "loss": 0.0034, "step": 6475 }, { "epoch": 37.9, "learning_rate": 6.923589743589744e-06, "loss": 0.0035, "step": 6500 }, { "epoch": 38.05, "learning_rate": 6.910769230769231e-06, "loss": 0.0035, "step": 6525 }, { "epoch": 38.19, "learning_rate": 6.8979487179487185e-06, "loss": 0.0024, "step": 6550 }, { "epoch": 38.34, "learning_rate": 6.885128205128205e-06, "loss": 0.0031, "step": 6575 }, { "epoch": 38.48, "learning_rate": 6.872307692307693e-06, "loss": 0.0023, "step": 6600 }, { "epoch": 38.63, "learning_rate": 6.85948717948718e-06, "loss": 0.0017, "step": 6625 }, { "epoch": 38.78, "learning_rate": 6.846666666666667e-06, "loss": 0.0022, "step": 6650 }, { "epoch": 38.92, "learning_rate": 6.833846153846154e-06, "loss": 0.0028, "step": 6675 }, { "epoch": 39.07, "learning_rate": 6.821025641025641e-06, "loss": 0.0028, "step": 6700 }, { "epoch": 39.21, "learning_rate": 6.808205128205129e-06, "loss": 0.0021, "step": 6725 }, { "epoch": 39.36, "learning_rate": 6.795384615384616e-06, "loss": 0.0019, "step": 6750 }, { "epoch": 39.5, "learning_rate": 6.782564102564103e-06, "loss": 0.0021, "step": 6775 }, { "epoch": 39.65, "learning_rate": 6.76974358974359e-06, "loss": 0.0025, "step": 6800 }, { "epoch": 39.8, "learning_rate": 6.756923076923078e-06, "loss": 0.0025, "step": 6825 }, { "epoch": 39.94, "learning_rate": 6.744102564102565e-06, "loss": 0.0027, "step": 6850 }, { "epoch": 40.09, "learning_rate": 6.731282051282052e-06, "loss": 0.003, "step": 6875 }, { "epoch": 40.23, "learning_rate": 6.718461538461539e-06, "loss": 0.002, "step": 6900 }, { "epoch": 40.38, "learning_rate": 6.705641025641027e-06, "loss": 0.003, "step": 6925 }, { "epoch": 40.52, "learning_rate": 6.692820512820514e-06, "loss": 0.0019, "step": 6950 }, { "epoch": 40.67, "learning_rate": 6.680000000000001e-06, "loss": 0.0027, "step": 6975 }, { "epoch": 40.82, "learning_rate": 6.667179487179488e-06, "loss": 0.0022, "step": 7000 }, { "epoch": 40.82, "eval_loss": 0.24650566279888153, "eval_runtime": 1320.5322, "eval_samples_per_second": 4.844, "eval_steps_per_second": 0.303, "eval_wer": 6.111437761560544, "step": 7000 }, { "epoch": 40.96, "learning_rate": 6.654358974358976e-06, "loss": 0.0018, "step": 7025 }, { "epoch": 41.11, "learning_rate": 6.641538461538463e-06, "loss": 0.0029, "step": 7050 }, { "epoch": 41.25, "learning_rate": 6.6287179487179486e-06, "loss": 0.0019, "step": 7075 }, { "epoch": 41.4, "learning_rate": 6.615897435897436e-06, "loss": 0.0016, "step": 7100 }, { "epoch": 41.55, "learning_rate": 6.603076923076923e-06, "loss": 0.002, "step": 7125 }, { "epoch": 41.69, "learning_rate": 6.590256410256411e-06, "loss": 0.0024, "step": 7150 }, { "epoch": 41.84, "learning_rate": 6.5774358974358976e-06, "loss": 0.002, "step": 7175 }, { "epoch": 41.98, "learning_rate": 6.564615384615385e-06, "loss": 0.0026, "step": 7200 }, { "epoch": 42.13, "learning_rate": 6.551794871794872e-06, "loss": 0.0016, "step": 7225 }, { "epoch": 42.27, "learning_rate": 6.538974358974359e-06, "loss": 0.0027, "step": 7250 }, { "epoch": 42.42, "learning_rate": 6.5261538461538465e-06, "loss": 0.0038, "step": 7275 }, { "epoch": 42.57, "learning_rate": 6.513333333333333e-06, "loss": 0.0026, "step": 7300 }, { "epoch": 42.71, "learning_rate": 6.500512820512821e-06, "loss": 0.0036, "step": 7325 }, { "epoch": 42.86, "learning_rate": 6.487692307692308e-06, "loss": 0.003, "step": 7350 }, { "epoch": 43.0, "learning_rate": 6.4748717948717955e-06, "loss": 0.0021, "step": 7375 }, { "epoch": 43.15, "learning_rate": 6.462051282051282e-06, "loss": 0.0021, "step": 7400 }, { "epoch": 43.29, "learning_rate": 6.44923076923077e-06, "loss": 0.0023, "step": 7425 }, { "epoch": 43.44, "learning_rate": 6.436410256410257e-06, "loss": 0.0028, "step": 7450 }, { "epoch": 43.59, "learning_rate": 6.4235897435897445e-06, "loss": 0.0029, "step": 7475 }, { "epoch": 43.73, "learning_rate": 6.410769230769231e-06, "loss": 0.0027, "step": 7500 }, { "epoch": 43.88, "learning_rate": 6.397948717948719e-06, "loss": 0.0023, "step": 7525 }, { "epoch": 44.02, "learning_rate": 6.385128205128206e-06, "loss": 0.003, "step": 7550 }, { "epoch": 44.17, "learning_rate": 6.3723076923076935e-06, "loss": 0.0016, "step": 7575 }, { "epoch": 44.31, "learning_rate": 6.35948717948718e-06, "loss": 0.0032, "step": 7600 }, { "epoch": 44.46, "learning_rate": 6.346666666666668e-06, "loss": 0.0025, "step": 7625 }, { "epoch": 44.61, "learning_rate": 6.333846153846155e-06, "loss": 0.0017, "step": 7650 }, { "epoch": 44.75, "learning_rate": 6.3210256410256425e-06, "loss": 0.002, "step": 7675 }, { "epoch": 44.9, "learning_rate": 6.3082051282051285e-06, "loss": 0.0029, "step": 7700 }, { "epoch": 45.04, "learning_rate": 6.295384615384615e-06, "loss": 0.0026, "step": 7725 }, { "epoch": 45.19, "learning_rate": 6.282564102564103e-06, "loss": 0.0022, "step": 7750 }, { "epoch": 45.34, "learning_rate": 6.26974358974359e-06, "loss": 0.002, "step": 7775 }, { "epoch": 45.48, "learning_rate": 6.2569230769230775e-06, "loss": 0.0024, "step": 7800 }, { "epoch": 45.63, "learning_rate": 6.244102564102564e-06, "loss": 0.0019, "step": 7825 }, { "epoch": 45.77, "learning_rate": 6.231282051282051e-06, "loss": 0.0023, "step": 7850 }, { "epoch": 45.92, "learning_rate": 6.218461538461539e-06, "loss": 0.0026, "step": 7875 }, { "epoch": 46.06, "learning_rate": 6.205641025641026e-06, "loss": 0.0019, "step": 7900 }, { "epoch": 46.21, "learning_rate": 6.192820512820513e-06, "loss": 0.003, "step": 7925 }, { "epoch": 46.36, "learning_rate": 6.18e-06, "loss": 0.003, "step": 7950 }, { "epoch": 46.5, "learning_rate": 6.167179487179488e-06, "loss": 0.0022, "step": 7975 }, { "epoch": 46.65, "learning_rate": 6.154358974358975e-06, "loss": 0.0019, "step": 8000 }, { "epoch": 46.65, "eval_loss": 0.25306838750839233, "eval_runtime": 1315.1452, "eval_samples_per_second": 4.864, "eval_steps_per_second": 0.304, "eval_wer": 5.858956178881472, "step": 8000 }, { "epoch": 46.79, "learning_rate": 6.141538461538462e-06, "loss": 0.0018, "step": 8025 }, { "epoch": 46.94, "learning_rate": 6.128717948717949e-06, "loss": 0.0015, "step": 8050 }, { "epoch": 47.08, "learning_rate": 6.115897435897437e-06, "loss": 0.0013, "step": 8075 }, { "epoch": 47.23, "learning_rate": 6.103076923076924e-06, "loss": 0.0014, "step": 8100 }, { "epoch": 47.38, "learning_rate": 6.090256410256411e-06, "loss": 0.0013, "step": 8125 }, { "epoch": 47.52, "learning_rate": 6.077435897435898e-06, "loss": 0.001, "step": 8150 }, { "epoch": 47.67, "learning_rate": 6.064615384615386e-06, "loss": 0.0011, "step": 8175 }, { "epoch": 47.81, "learning_rate": 6.051794871794873e-06, "loss": 0.002, "step": 8200 }, { "epoch": 47.96, "learning_rate": 6.03897435897436e-06, "loss": 0.0015, "step": 8225 }, { "epoch": 48.1, "learning_rate": 6.026153846153847e-06, "loss": 0.0015, "step": 8250 }, { "epoch": 48.25, "learning_rate": 6.013333333333335e-06, "loss": 0.0012, "step": 8275 }, { "epoch": 48.4, "learning_rate": 6.000512820512821e-06, "loss": 0.0012, "step": 8300 }, { "epoch": 48.54, "learning_rate": 5.9876923076923076e-06, "loss": 0.0014, "step": 8325 }, { "epoch": 48.69, "learning_rate": 5.974871794871795e-06, "loss": 0.0019, "step": 8350 }, { "epoch": 48.83, "learning_rate": 5.962051282051282e-06, "loss": 0.0019, "step": 8375 }, { "epoch": 48.98, "learning_rate": 5.94923076923077e-06, "loss": 0.0014, "step": 8400 }, { "epoch": 49.13, "learning_rate": 5.9364102564102566e-06, "loss": 0.002, "step": 8425 }, { "epoch": 49.27, "learning_rate": 5.923589743589744e-06, "loss": 0.0021, "step": 8450 }, { "epoch": 49.42, "learning_rate": 5.910769230769231e-06, "loss": 0.0021, "step": 8475 }, { "epoch": 49.56, "learning_rate": 5.897948717948718e-06, "loss": 0.0015, "step": 8500 }, { "epoch": 49.71, "learning_rate": 5.8851282051282056e-06, "loss": 0.0013, "step": 8525 }, { "epoch": 49.85, "learning_rate": 5.872307692307692e-06, "loss": 0.0025, "step": 8550 }, { "epoch": 50.0, "learning_rate": 5.85948717948718e-06, "loss": 0.0024, "step": 8575 }, { "epoch": 50.15, "learning_rate": 5.846666666666667e-06, "loss": 0.0023, "step": 8600 }, { "epoch": 50.29, "learning_rate": 5.8338461538461545e-06, "loss": 0.0017, "step": 8625 }, { "epoch": 50.44, "learning_rate": 5.821025641025641e-06, "loss": 0.0021, "step": 8650 }, { "epoch": 50.58, "learning_rate": 5.808205128205129e-06, "loss": 0.0015, "step": 8675 }, { "epoch": 50.73, "learning_rate": 5.795384615384616e-06, "loss": 0.002, "step": 8700 }, { "epoch": 50.87, "learning_rate": 5.7825641025641035e-06, "loss": 0.0026, "step": 8725 }, { "epoch": 51.02, "learning_rate": 5.76974358974359e-06, "loss": 0.0026, "step": 8750 }, { "epoch": 51.17, "learning_rate": 5.756923076923078e-06, "loss": 0.0024, "step": 8775 }, { "epoch": 51.31, "learning_rate": 5.744102564102565e-06, "loss": 0.0025, "step": 8800 }, { "epoch": 51.46, "learning_rate": 5.7312820512820525e-06, "loss": 0.0021, "step": 8825 }, { "epoch": 51.6, "learning_rate": 5.718461538461539e-06, "loss": 0.0015, "step": 8850 }, { "epoch": 51.75, "learning_rate": 5.705641025641027e-06, "loss": 0.0012, "step": 8875 }, { "epoch": 51.9, "learning_rate": 5.692820512820513e-06, "loss": 0.0016, "step": 8900 }, { "epoch": 52.04, "learning_rate": 5.68e-06, "loss": 0.0013, "step": 8925 }, { "epoch": 52.19, "learning_rate": 5.6671794871794875e-06, "loss": 0.0017, "step": 8950 }, { "epoch": 52.33, "learning_rate": 5.654358974358974e-06, "loss": 0.0011, "step": 8975 }, { "epoch": 52.48, "learning_rate": 5.641538461538462e-06, "loss": 0.0009, "step": 9000 }, { "epoch": 52.48, "eval_loss": 0.25673097372055054, "eval_runtime": 1313.2253, "eval_samples_per_second": 4.871, "eval_steps_per_second": 0.305, "eval_wer": 5.845121571611386, "step": 9000 }, { "epoch": 52.62, "learning_rate": 5.628717948717949e-06, "loss": 0.002, "step": 9025 }, { "epoch": 52.77, "learning_rate": 5.6158974358974365e-06, "loss": 0.0012, "step": 9050 }, { "epoch": 52.92, "learning_rate": 5.603076923076923e-06, "loss": 0.0021, "step": 9075 }, { "epoch": 53.06, "learning_rate": 5.59025641025641e-06, "loss": 0.0017, "step": 9100 }, { "epoch": 53.21, "learning_rate": 5.577435897435898e-06, "loss": 0.0011, "step": 9125 }, { "epoch": 53.35, "learning_rate": 5.564615384615385e-06, "loss": 0.0018, "step": 9150 }, { "epoch": 53.5, "learning_rate": 5.551794871794872e-06, "loss": 0.0017, "step": 9175 }, { "epoch": 53.64, "learning_rate": 5.538974358974359e-06, "loss": 0.0012, "step": 9200 }, { "epoch": 53.79, "learning_rate": 5.526153846153847e-06, "loss": 0.0014, "step": 9225 }, { "epoch": 53.94, "learning_rate": 5.513333333333334e-06, "loss": 0.001, "step": 9250 }, { "epoch": 54.08, "learning_rate": 5.500512820512821e-06, "loss": 0.0015, "step": 9275 }, { "epoch": 54.23, "learning_rate": 5.487692307692308e-06, "loss": 0.0015, "step": 9300 }, { "epoch": 54.37, "learning_rate": 5.474871794871796e-06, "loss": 0.0014, "step": 9325 }, { "epoch": 54.52, "learning_rate": 5.462051282051283e-06, "loss": 0.0011, "step": 9350 }, { "epoch": 54.66, "learning_rate": 5.44923076923077e-06, "loss": 0.0013, "step": 9375 }, { "epoch": 54.81, "learning_rate": 5.436410256410257e-06, "loss": 0.0023, "step": 9400 }, { "epoch": 54.96, "learning_rate": 5.423589743589745e-06, "loss": 0.0014, "step": 9425 }, { "epoch": 55.1, "learning_rate": 5.410769230769232e-06, "loss": 0.0017, "step": 9450 }, { "epoch": 55.25, "learning_rate": 5.397948717948719e-06, "loss": 0.0011, "step": 9475 }, { "epoch": 55.39, "learning_rate": 5.385128205128206e-06, "loss": 0.0013, "step": 9500 }, { "epoch": 55.54, "learning_rate": 5.372307692307692e-06, "loss": 0.001, "step": 9525 }, { "epoch": 55.69, "learning_rate": 5.35948717948718e-06, "loss": 0.0014, "step": 9550 }, { "epoch": 55.83, "learning_rate": 5.346666666666667e-06, "loss": 0.0018, "step": 9575 }, { "epoch": 55.98, "learning_rate": 5.333846153846154e-06, "loss": 0.0017, "step": 9600 }, { "epoch": 56.12, "learning_rate": 5.321025641025641e-06, "loss": 0.0015, "step": 9625 }, { "epoch": 56.27, "learning_rate": 5.308205128205129e-06, "loss": 0.0012, "step": 9650 }, { "epoch": 56.41, "learning_rate": 5.2953846153846156e-06, "loss": 0.0019, "step": 9675 }, { "epoch": 56.56, "learning_rate": 5.282564102564102e-06, "loss": 0.002, "step": 9700 }, { "epoch": 56.71, "learning_rate": 5.26974358974359e-06, "loss": 0.001, "step": 9725 }, { "epoch": 56.85, "learning_rate": 5.256923076923077e-06, "loss": 0.0011, "step": 9750 }, { "epoch": 57.0, "learning_rate": 5.2441025641025646e-06, "loss": 0.0008, "step": 9775 }, { "epoch": 57.14, "learning_rate": 5.231282051282051e-06, "loss": 0.0008, "step": 9800 }, { "epoch": 57.29, "learning_rate": 5.218461538461539e-06, "loss": 0.0007, "step": 9825 }, { "epoch": 57.43, "learning_rate": 5.205641025641026e-06, "loss": 0.0023, "step": 9850 }, { "epoch": 57.58, "learning_rate": 5.1928205128205136e-06, "loss": 0.0011, "step": 9875 }, { "epoch": 57.73, "learning_rate": 5.18e-06, "loss": 0.0009, "step": 9900 }, { "epoch": 57.87, "learning_rate": 5.167179487179488e-06, "loss": 0.001, "step": 9925 }, { "epoch": 58.02, "learning_rate": 5.154358974358975e-06, "loss": 0.0007, "step": 9950 }, { "epoch": 58.16, "learning_rate": 5.142051282051282e-06, "loss": 0.0011, "step": 9975 }, { "epoch": 58.31, "learning_rate": 5.12923076923077e-06, "loss": 0.001, "step": 10000 }, { "epoch": 58.31, "eval_loss": 0.2718268930912018, "eval_runtime": 1313.408, "eval_samples_per_second": 4.871, "eval_steps_per_second": 0.305, "eval_wer": 5.767301905717152, "step": 10000 }, { "epoch": 58.45, "learning_rate": 5.116410256410257e-06, "loss": 0.0015, "step": 10025 }, { "epoch": 58.6, "learning_rate": 5.1035897435897445e-06, "loss": 0.0014, "step": 10050 }, { "epoch": 58.75, "learning_rate": 5.090769230769231e-06, "loss": 0.0023, "step": 10075 }, { "epoch": 58.89, "learning_rate": 5.077948717948719e-06, "loss": 0.0016, "step": 10100 }, { "epoch": 59.04, "learning_rate": 5.065128205128206e-06, "loss": 0.0019, "step": 10125 }, { "epoch": 59.18, "learning_rate": 5.052307692307693e-06, "loss": 0.0021, "step": 10150 }, { "epoch": 59.33, "learning_rate": 5.03948717948718e-06, "loss": 0.002, "step": 10175 }, { "epoch": 59.48, "learning_rate": 5.026666666666667e-06, "loss": 0.0023, "step": 10200 }, { "epoch": 59.62, "learning_rate": 5.013846153846155e-06, "loss": 0.0019, "step": 10225 }, { "epoch": 59.77, "learning_rate": 5.001025641025642e-06, "loss": 0.001, "step": 10250 }, { "epoch": 59.91, "learning_rate": 4.988205128205128e-06, "loss": 0.0011, "step": 10275 }, { "epoch": 60.06, "learning_rate": 4.975384615384616e-06, "loss": 0.0015, "step": 10300 }, { "epoch": 60.2, "learning_rate": 4.962564102564103e-06, "loss": 0.0015, "step": 10325 }, { "epoch": 60.35, "learning_rate": 4.949743589743591e-06, "loss": 0.0015, "step": 10350 }, { "epoch": 60.5, "learning_rate": 4.936923076923077e-06, "loss": 0.001, "step": 10375 }, { "epoch": 60.64, "learning_rate": 4.924102564102565e-06, "loss": 0.0009, "step": 10400 }, { "epoch": 60.79, "learning_rate": 4.911282051282052e-06, "loss": 0.0016, "step": 10425 }, { "epoch": 60.93, "learning_rate": 4.898461538461539e-06, "loss": 0.001, "step": 10450 }, { "epoch": 61.08, "learning_rate": 4.8856410256410256e-06, "loss": 0.0008, "step": 10475 }, { "epoch": 61.22, "learning_rate": 4.872820512820513e-06, "loss": 0.0011, "step": 10500 }, { "epoch": 61.37, "learning_rate": 4.86e-06, "loss": 0.0008, "step": 10525 }, { "epoch": 61.52, "learning_rate": 4.847179487179488e-06, "loss": 0.0009, "step": 10550 }, { "epoch": 61.66, "learning_rate": 4.8343589743589746e-06, "loss": 0.0013, "step": 10575 }, { "epoch": 61.81, "learning_rate": 4.821538461538462e-06, "loss": 0.0012, "step": 10600 }, { "epoch": 61.95, "learning_rate": 4.808717948717949e-06, "loss": 0.0012, "step": 10625 }, { "epoch": 62.1, "learning_rate": 4.795897435897437e-06, "loss": 0.0011, "step": 10650 }, { "epoch": 62.24, "learning_rate": 4.7830769230769235e-06, "loss": 0.0008, "step": 10675 }, { "epoch": 62.39, "learning_rate": 4.770256410256411e-06, "loss": 0.0007, "step": 10700 }, { "epoch": 62.54, "learning_rate": 4.757435897435898e-06, "loss": 0.0008, "step": 10725 }, { "epoch": 62.68, "learning_rate": 4.744615384615385e-06, "loss": 0.0015, "step": 10750 }, { "epoch": 62.83, "learning_rate": 4.731794871794872e-06, "loss": 0.0017, "step": 10775 }, { "epoch": 62.97, "learning_rate": 4.718974358974359e-06, "loss": 0.0009, "step": 10800 }, { "epoch": 63.12, "learning_rate": 4.706153846153846e-06, "loss": 0.0022, "step": 10825 }, { "epoch": 63.27, "learning_rate": 4.693333333333334e-06, "loss": 0.0013, "step": 10850 }, { "epoch": 63.41, "learning_rate": 4.680512820512821e-06, "loss": 0.0009, "step": 10875 }, { "epoch": 63.56, "learning_rate": 4.667692307692308e-06, "loss": 0.0011, "step": 10900 }, { "epoch": 63.7, "learning_rate": 4.654871794871795e-06, "loss": 0.0008, "step": 10925 }, { "epoch": 63.85, "learning_rate": 4.642051282051283e-06, "loss": 0.0008, "step": 10950 }, { "epoch": 63.99, "learning_rate": 4.62923076923077e-06, "loss": 0.0006, "step": 10975 }, { "epoch": 64.14, "learning_rate": 4.616410256410257e-06, "loss": 0.0011, "step": 11000 }, { "epoch": 64.14, "eval_loss": 0.26593196392059326, "eval_runtime": 1312.4497, "eval_samples_per_second": 4.874, "eval_steps_per_second": 0.305, "eval_wer": 6.1045204579255, "step": 11000 }, { "epoch": 64.29, "learning_rate": 4.603589743589744e-06, "loss": 0.0009, "step": 11025 }, { "epoch": 64.43, "learning_rate": 4.590769230769232e-06, "loss": 0.001, "step": 11050 }, { "epoch": 64.58, "learning_rate": 4.577948717948718e-06, "loss": 0.001, "step": 11075 }, { "epoch": 64.72, "learning_rate": 4.5651282051282055e-06, "loss": 0.0011, "step": 11100 }, { "epoch": 64.87, "learning_rate": 4.552307692307692e-06, "loss": 0.0012, "step": 11125 }, { "epoch": 65.01, "learning_rate": 4.53948717948718e-06, "loss": 0.001, "step": 11150 }, { "epoch": 65.16, "learning_rate": 4.526666666666667e-06, "loss": 0.0007, "step": 11175 }, { "epoch": 65.31, "learning_rate": 4.5138461538461545e-06, "loss": 0.0006, "step": 11200 }, { "epoch": 65.45, "learning_rate": 4.501025641025641e-06, "loss": 0.0006, "step": 11225 }, { "epoch": 65.6, "learning_rate": 4.488205128205129e-06, "loss": 0.0007, "step": 11250 }, { "epoch": 65.74, "learning_rate": 4.475384615384616e-06, "loss": 0.0011, "step": 11275 }, { "epoch": 65.89, "learning_rate": 4.4625641025641035e-06, "loss": 0.0011, "step": 11300 }, { "epoch": 66.03, "learning_rate": 4.44974358974359e-06, "loss": 0.0008, "step": 11325 }, { "epoch": 66.18, "learning_rate": 4.436923076923078e-06, "loss": 0.0012, "step": 11350 }, { "epoch": 66.33, "learning_rate": 4.424102564102564e-06, "loss": 0.0022, "step": 11375 }, { "epoch": 66.47, "learning_rate": 4.411282051282052e-06, "loss": 0.0014, "step": 11400 }, { "epoch": 66.62, "learning_rate": 4.3984615384615384e-06, "loss": 0.001, "step": 11425 }, { "epoch": 66.76, "learning_rate": 4.385641025641026e-06, "loss": 0.0014, "step": 11450 }, { "epoch": 66.91, "learning_rate": 4.372820512820513e-06, "loss": 0.0014, "step": 11475 }, { "epoch": 67.06, "learning_rate": 4.360000000000001e-06, "loss": 0.0013, "step": 11500 }, { "epoch": 67.2, "learning_rate": 4.3471794871794874e-06, "loss": 0.001, "step": 11525 }, { "epoch": 67.35, "learning_rate": 4.334358974358975e-06, "loss": 0.0008, "step": 11550 }, { "epoch": 67.49, "learning_rate": 4.321538461538462e-06, "loss": 0.0013, "step": 11575 }, { "epoch": 67.64, "learning_rate": 4.30871794871795e-06, "loss": 0.0009, "step": 11600 }, { "epoch": 67.78, "learning_rate": 4.2958974358974364e-06, "loss": 0.0016, "step": 11625 }, { "epoch": 67.93, "learning_rate": 4.283076923076924e-06, "loss": 0.0009, "step": 11650 }, { "epoch": 68.08, "learning_rate": 4.27025641025641e-06, "loss": 0.0015, "step": 11675 }, { "epoch": 68.22, "learning_rate": 4.257435897435898e-06, "loss": 0.0011, "step": 11700 }, { "epoch": 68.37, "learning_rate": 4.2446153846153846e-06, "loss": 0.0009, "step": 11725 }, { "epoch": 68.51, "learning_rate": 4.231794871794872e-06, "loss": 0.001, "step": 11750 }, { "epoch": 68.66, "learning_rate": 4.218974358974359e-06, "loss": 0.001, "step": 11775 }, { "epoch": 68.8, "learning_rate": 4.206153846153847e-06, "loss": 0.0006, "step": 11800 }, { "epoch": 68.95, "learning_rate": 4.1933333333333336e-06, "loss": 0.0006, "step": 11825 }, { "epoch": 69.1, "learning_rate": 4.180512820512821e-06, "loss": 0.001, "step": 11850 }, { "epoch": 69.24, "learning_rate": 4.167692307692308e-06, "loss": 0.0008, "step": 11875 }, { "epoch": 69.39, "learning_rate": 4.154871794871796e-06, "loss": 0.0008, "step": 11900 }, { "epoch": 69.53, "learning_rate": 4.1420512820512826e-06, "loss": 0.0009, "step": 11925 }, { "epoch": 69.68, "learning_rate": 4.12923076923077e-06, "loss": 0.0012, "step": 11950 }, { "epoch": 69.83, "learning_rate": 4.116410256410256e-06, "loss": 0.0009, "step": 11975 }, { "epoch": 69.97, "learning_rate": 4.103589743589744e-06, "loss": 0.0008, "step": 12000 }, { "epoch": 69.97, "eval_loss": 0.2764803469181061, "eval_runtime": 1310.6981, "eval_samples_per_second": 4.881, "eval_steps_per_second": 0.305, "eval_wer": 6.040535399301352, "step": 12000 }, { "epoch": 70.12, "learning_rate": 4.090769230769231e-06, "loss": 0.002, "step": 12025 }, { "epoch": 70.26, "learning_rate": 4.077948717948718e-06, "loss": 0.0011, "step": 12050 }, { "epoch": 70.41, "learning_rate": 4.065128205128205e-06, "loss": 0.0013, "step": 12075 }, { "epoch": 70.55, "learning_rate": 4.052307692307693e-06, "loss": 0.0007, "step": 12100 }, { "epoch": 70.7, "learning_rate": 4.03948717948718e-06, "loss": 0.0006, "step": 12125 }, { "epoch": 70.85, "learning_rate": 4.026666666666667e-06, "loss": 0.0007, "step": 12150 }, { "epoch": 70.99, "learning_rate": 4.013846153846154e-06, "loss": 0.0006, "step": 12175 }, { "epoch": 71.14, "learning_rate": 4.001025641025642e-06, "loss": 0.0008, "step": 12200 }, { "epoch": 71.28, "learning_rate": 3.988205128205129e-06, "loss": 0.0007, "step": 12225 }, { "epoch": 71.43, "learning_rate": 3.975384615384616e-06, "loss": 0.0016, "step": 12250 }, { "epoch": 71.57, "learning_rate": 3.962564102564102e-06, "loss": 0.0007, "step": 12275 }, { "epoch": 71.72, "learning_rate": 3.94974358974359e-06, "loss": 0.0006, "step": 12300 }, { "epoch": 71.87, "learning_rate": 3.936923076923077e-06, "loss": 0.0008, "step": 12325 }, { "epoch": 72.01, "learning_rate": 3.9241025641025645e-06, "loss": 0.0011, "step": 12350 }, { "epoch": 72.16, "learning_rate": 3.911282051282051e-06, "loss": 0.001, "step": 12375 }, { "epoch": 72.3, "learning_rate": 3.898461538461539e-06, "loss": 0.0006, "step": 12400 }, { "epoch": 72.45, "learning_rate": 3.885641025641026e-06, "loss": 0.0005, "step": 12425 }, { "epoch": 72.59, "learning_rate": 3.8728205128205135e-06, "loss": 0.0006, "step": 12450 }, { "epoch": 72.74, "learning_rate": 3.86e-06, "loss": 0.0005, "step": 12475 }, { "epoch": 72.89, "learning_rate": 3.847179487179488e-06, "loss": 0.0008, "step": 12500 }, { "epoch": 73.03, "learning_rate": 3.834871794871795e-06, "loss": 0.0034, "step": 12525 }, { "epoch": 73.18, "learning_rate": 3.822051282051282e-06, "loss": 0.0008, "step": 12550 }, { "epoch": 73.32, "learning_rate": 3.8092307692307695e-06, "loss": 0.0007, "step": 12575 }, { "epoch": 73.47, "learning_rate": 3.7964102564102567e-06, "loss": 0.0005, "step": 12600 }, { "epoch": 73.62, "learning_rate": 3.783589743589744e-06, "loss": 0.0006, "step": 12625 }, { "epoch": 73.76, "learning_rate": 3.7707692307692312e-06, "loss": 0.0006, "step": 12650 }, { "epoch": 73.91, "learning_rate": 3.7579487179487185e-06, "loss": 0.0006, "step": 12675 }, { "epoch": 74.05, "learning_rate": 3.7451282051282057e-06, "loss": 0.0007, "step": 12700 }, { "epoch": 74.2, "learning_rate": 3.732307692307693e-06, "loss": 0.0009, "step": 12725 }, { "epoch": 74.34, "learning_rate": 3.7194871794871794e-06, "loss": 0.0005, "step": 12750 }, { "epoch": 74.49, "learning_rate": 3.7066666666666666e-06, "loss": 0.0004, "step": 12775 }, { "epoch": 74.64, "learning_rate": 3.693846153846154e-06, "loss": 0.0004, "step": 12800 }, { "epoch": 74.78, "learning_rate": 3.681025641025641e-06, "loss": 0.0006, "step": 12825 }, { "epoch": 74.93, "learning_rate": 3.6682051282051284e-06, "loss": 0.0011, "step": 12850 }, { "epoch": 75.07, "learning_rate": 3.6553846153846156e-06, "loss": 0.0015, "step": 12875 }, { "epoch": 75.22, "learning_rate": 3.642564102564103e-06, "loss": 0.0012, "step": 12900 }, { "epoch": 75.36, "learning_rate": 3.62974358974359e-06, "loss": 0.001, "step": 12925 }, { "epoch": 75.51, "learning_rate": 3.6169230769230773e-06, "loss": 0.0006, "step": 12950 }, { "epoch": 75.66, "learning_rate": 3.6041025641025646e-06, "loss": 0.001, "step": 12975 }, { "epoch": 75.8, "learning_rate": 3.591282051282052e-06, "loss": 0.0006, "step": 13000 }, { "epoch": 75.8, "eval_loss": 0.2792523205280304, "eval_runtime": 1311.0637, "eval_samples_per_second": 4.879, "eval_steps_per_second": 0.305, "eval_wer": 6.024971466122506, "step": 13000 }, { "epoch": 75.95, "learning_rate": 3.578461538461539e-06, "loss": 0.0008, "step": 13025 }, { "epoch": 76.09, "learning_rate": 3.5656410256410263e-06, "loss": 0.0005, "step": 13050 }, { "epoch": 76.24, "learning_rate": 3.5528205128205127e-06, "loss": 0.0008, "step": 13075 }, { "epoch": 76.38, "learning_rate": 3.54e-06, "loss": 0.0008, "step": 13100 }, { "epoch": 76.53, "learning_rate": 3.5271794871794872e-06, "loss": 0.0006, "step": 13125 }, { "epoch": 76.68, "learning_rate": 3.5143589743589745e-06, "loss": 0.0006, "step": 13150 }, { "epoch": 76.82, "learning_rate": 3.5015384615384617e-06, "loss": 0.0004, "step": 13175 }, { "epoch": 76.97, "learning_rate": 3.488717948717949e-06, "loss": 0.0005, "step": 13200 }, { "epoch": 77.11, "learning_rate": 3.4758974358974362e-06, "loss": 0.0005, "step": 13225 }, { "epoch": 77.26, "learning_rate": 3.4630769230769235e-06, "loss": 0.0004, "step": 13250 }, { "epoch": 77.41, "learning_rate": 3.4502564102564107e-06, "loss": 0.0004, "step": 13275 }, { "epoch": 77.55, "learning_rate": 3.437435897435898e-06, "loss": 0.0003, "step": 13300 }, { "epoch": 77.7, "learning_rate": 3.4246153846153852e-06, "loss": 0.0005, "step": 13325 }, { "epoch": 77.84, "learning_rate": 3.4117948717948725e-06, "loss": 0.0005, "step": 13350 }, { "epoch": 77.99, "learning_rate": 3.398974358974359e-06, "loss": 0.0004, "step": 13375 }, { "epoch": 78.13, "learning_rate": 3.386153846153846e-06, "loss": 0.0004, "step": 13400 }, { "epoch": 78.28, "learning_rate": 3.3733333333333334e-06, "loss": 0.0003, "step": 13425 }, { "epoch": 78.43, "learning_rate": 3.3605128205128206e-06, "loss": 0.0004, "step": 13450 }, { "epoch": 78.57, "learning_rate": 3.347692307692308e-06, "loss": 0.0007, "step": 13475 }, { "epoch": 78.72, "learning_rate": 3.334871794871795e-06, "loss": 0.0005, "step": 13500 }, { "epoch": 78.86, "learning_rate": 3.3220512820512824e-06, "loss": 0.0003, "step": 13525 }, { "epoch": 79.01, "learning_rate": 3.3092307692307696e-06, "loss": 0.0004, "step": 13550 }, { "epoch": 79.15, "learning_rate": 3.296410256410257e-06, "loss": 0.0004, "step": 13575 }, { "epoch": 79.3, "learning_rate": 3.283589743589744e-06, "loss": 0.0003, "step": 13600 }, { "epoch": 79.45, "learning_rate": 3.2707692307692313e-06, "loss": 0.0004, "step": 13625 }, { "epoch": 79.59, "learning_rate": 3.2579487179487186e-06, "loss": 0.0003, "step": 13650 }, { "epoch": 79.74, "learning_rate": 3.245128205128205e-06, "loss": 0.0003, "step": 13675 }, { "epoch": 79.88, "learning_rate": 3.2323076923076922e-06, "loss": 0.0005, "step": 13700 }, { "epoch": 80.03, "learning_rate": 3.2194871794871795e-06, "loss": 0.0003, "step": 13725 }, { "epoch": 80.17, "learning_rate": 3.2066666666666667e-06, "loss": 0.0005, "step": 13750 }, { "epoch": 80.32, "learning_rate": 3.193846153846154e-06, "loss": 0.0003, "step": 13775 }, { "epoch": 80.47, "learning_rate": 3.1810256410256412e-06, "loss": 0.0005, "step": 13800 }, { "epoch": 80.61, "learning_rate": 3.1682051282051285e-06, "loss": 0.0006, "step": 13825 }, { "epoch": 80.76, "learning_rate": 3.1553846153846157e-06, "loss": 0.0007, "step": 13850 }, { "epoch": 80.9, "learning_rate": 3.142564102564103e-06, "loss": 0.0004, "step": 13875 }, { "epoch": 81.05, "learning_rate": 3.1297435897435902e-06, "loss": 0.0007, "step": 13900 }, { "epoch": 81.2, "learning_rate": 3.1169230769230775e-06, "loss": 0.001, "step": 13925 }, { "epoch": 81.34, "learning_rate": 3.1041025641025647e-06, "loss": 0.0005, "step": 13950 }, { "epoch": 81.49, "learning_rate": 3.091282051282051e-06, "loss": 0.0004, "step": 13975 }, { "epoch": 81.63, "learning_rate": 3.0784615384615384e-06, "loss": 0.0004, "step": 14000 }, { "epoch": 81.63, "eval_loss": 0.28476274013519287, "eval_runtime": 1311.5785, "eval_samples_per_second": 4.877, "eval_steps_per_second": 0.305, "eval_wer": 6.0024902293086155, "step": 14000 }, { "epoch": 81.78, "learning_rate": 3.0656410256410256e-06, "loss": 0.0006, "step": 14025 }, { "epoch": 81.92, "learning_rate": 3.052820512820513e-06, "loss": 0.0004, "step": 14050 }, { "epoch": 82.07, "learning_rate": 3.04e-06, "loss": 0.0004, "step": 14075 }, { "epoch": 82.22, "learning_rate": 3.0271794871794874e-06, "loss": 0.0003, "step": 14100 }, { "epoch": 82.36, "learning_rate": 3.0143589743589746e-06, "loss": 0.0003, "step": 14125 }, { "epoch": 82.51, "learning_rate": 3.001538461538462e-06, "loss": 0.0003, "step": 14150 }, { "epoch": 82.65, "learning_rate": 2.988717948717949e-06, "loss": 0.0004, "step": 14175 }, { "epoch": 82.8, "learning_rate": 2.9758974358974364e-06, "loss": 0.0004, "step": 14200 }, { "epoch": 82.94, "learning_rate": 2.9630769230769236e-06, "loss": 0.0004, "step": 14225 }, { "epoch": 83.09, "learning_rate": 2.950256410256411e-06, "loss": 0.001, "step": 14250 }, { "epoch": 83.24, "learning_rate": 2.9374358974358973e-06, "loss": 0.0006, "step": 14275 }, { "epoch": 83.38, "learning_rate": 2.9246153846153845e-06, "loss": 0.0007, "step": 14300 }, { "epoch": 83.53, "learning_rate": 2.9117948717948717e-06, "loss": 0.0005, "step": 14325 }, { "epoch": 83.67, "learning_rate": 2.898974358974359e-06, "loss": 0.0006, "step": 14350 }, { "epoch": 83.82, "learning_rate": 2.8861538461538462e-06, "loss": 0.0006, "step": 14375 }, { "epoch": 83.97, "learning_rate": 2.8733333333333335e-06, "loss": 0.0006, "step": 14400 }, { "epoch": 84.11, "learning_rate": 2.8605128205128207e-06, "loss": 0.0003, "step": 14425 }, { "epoch": 84.26, "learning_rate": 2.847692307692308e-06, "loss": 0.0009, "step": 14450 }, { "epoch": 84.4, "learning_rate": 2.8348717948717952e-06, "loss": 0.0005, "step": 14475 }, { "epoch": 84.55, "learning_rate": 2.8220512820512825e-06, "loss": 0.0007, "step": 14500 }, { "epoch": 84.69, "learning_rate": 2.8092307692307697e-06, "loss": 0.0004, "step": 14525 }, { "epoch": 84.84, "learning_rate": 2.796410256410257e-06, "loss": 0.0005, "step": 14550 }, { "epoch": 84.99, "learning_rate": 2.7835897435897434e-06, "loss": 0.0005, "step": 14575 }, { "epoch": 85.13, "learning_rate": 2.7707692307692306e-06, "loss": 0.0005, "step": 14600 }, { "epoch": 85.28, "learning_rate": 2.757948717948718e-06, "loss": 0.0005, "step": 14625 }, { "epoch": 85.42, "learning_rate": 2.745128205128205e-06, "loss": 0.0003, "step": 14650 }, { "epoch": 85.57, "learning_rate": 2.7323076923076924e-06, "loss": 0.0003, "step": 14675 }, { "epoch": 85.71, "learning_rate": 2.7194871794871796e-06, "loss": 0.0003, "step": 14700 }, { "epoch": 85.86, "learning_rate": 2.706666666666667e-06, "loss": 0.0005, "step": 14725 }, { "epoch": 86.01, "learning_rate": 2.693846153846154e-06, "loss": 0.0003, "step": 14750 }, { "epoch": 86.15, "learning_rate": 2.6810256410256414e-06, "loss": 0.0004, "step": 14775 }, { "epoch": 86.3, "learning_rate": 2.6682051282051286e-06, "loss": 0.0008, "step": 14800 }, { "epoch": 86.44, "learning_rate": 2.655384615384616e-06, "loss": 0.0007, "step": 14825 }, { "epoch": 86.59, "learning_rate": 2.642564102564103e-06, "loss": 0.0008, "step": 14850 }, { "epoch": 86.73, "learning_rate": 2.6297435897435904e-06, "loss": 0.0005, "step": 14875 }, { "epoch": 86.88, "learning_rate": 2.6169230769230768e-06, "loss": 0.0005, "step": 14900 }, { "epoch": 87.03, "learning_rate": 2.604102564102564e-06, "loss": 0.0005, "step": 14925 }, { "epoch": 87.17, "learning_rate": 2.5912820512820513e-06, "loss": 0.0003, "step": 14950 }, { "epoch": 87.32, "learning_rate": 2.5784615384615385e-06, "loss": 0.0007, "step": 14975 }, { "epoch": 87.46, "learning_rate": 2.5656410256410257e-06, "loss": 0.0005, "step": 15000 }, { "epoch": 87.46, "eval_loss": 0.2789987027645111, "eval_runtime": 1311.0873, "eval_samples_per_second": 4.879, "eval_steps_per_second": 0.305, "eval_wer": 5.9454224743195105, "step": 15000 }, { "epoch": 87.61, "learning_rate": 2.552820512820513e-06, "loss": 0.0005, "step": 15025 }, { "epoch": 87.76, "learning_rate": 2.5400000000000002e-06, "loss": 0.0006, "step": 15050 }, { "epoch": 87.9, "learning_rate": 2.5271794871794875e-06, "loss": 0.0008, "step": 15075 }, { "epoch": 88.05, "learning_rate": 2.5143589743589747e-06, "loss": 0.0006, "step": 15100 }, { "epoch": 88.19, "learning_rate": 2.501538461538462e-06, "loss": 0.0005, "step": 15125 }, { "epoch": 88.34, "learning_rate": 2.488717948717949e-06, "loss": 0.0003, "step": 15150 }, { "epoch": 88.48, "learning_rate": 2.475897435897436e-06, "loss": 0.0003, "step": 15175 }, { "epoch": 88.63, "learning_rate": 2.4630769230769233e-06, "loss": 0.0005, "step": 15200 }, { "epoch": 88.78, "learning_rate": 2.4502564102564106e-06, "loss": 0.0006, "step": 15225 }, { "epoch": 88.92, "learning_rate": 2.437435897435898e-06, "loss": 0.0006, "step": 15250 }, { "epoch": 89.07, "learning_rate": 2.4246153846153846e-06, "loss": 0.0005, "step": 15275 }, { "epoch": 89.21, "learning_rate": 2.411794871794872e-06, "loss": 0.0004, "step": 15300 }, { "epoch": 89.36, "learning_rate": 2.398974358974359e-06, "loss": 0.0006, "step": 15325 }, { "epoch": 89.5, "learning_rate": 2.3861538461538464e-06, "loss": 0.0003, "step": 15350 }, { "epoch": 89.65, "learning_rate": 2.3733333333333336e-06, "loss": 0.0004, "step": 15375 }, { "epoch": 89.8, "learning_rate": 2.360512820512821e-06, "loss": 0.0003, "step": 15400 }, { "epoch": 89.94, "learning_rate": 2.3476923076923077e-06, "loss": 0.0004, "step": 15425 }, { "epoch": 90.09, "learning_rate": 2.334871794871795e-06, "loss": 0.0003, "step": 15450 }, { "epoch": 90.23, "learning_rate": 2.322051282051282e-06, "loss": 0.0008, "step": 15475 }, { "epoch": 90.38, "learning_rate": 2.3092307692307694e-06, "loss": 0.0002, "step": 15500 }, { "epoch": 90.52, "learning_rate": 2.2964102564102567e-06, "loss": 0.0003, "step": 15525 }, { "epoch": 90.67, "learning_rate": 2.283589743589744e-06, "loss": 0.0002, "step": 15550 }, { "epoch": 90.82, "learning_rate": 2.2707692307692308e-06, "loss": 0.0003, "step": 15575 }, { "epoch": 90.96, "learning_rate": 2.257948717948718e-06, "loss": 0.0003, "step": 15600 }, { "epoch": 91.11, "learning_rate": 2.2451282051282053e-06, "loss": 0.0007, "step": 15625 }, { "epoch": 91.25, "learning_rate": 2.2323076923076925e-06, "loss": 0.0003, "step": 15650 }, { "epoch": 91.4, "learning_rate": 2.2194871794871797e-06, "loss": 0.0003, "step": 15675 }, { "epoch": 91.55, "learning_rate": 2.206666666666667e-06, "loss": 0.0003, "step": 15700 }, { "epoch": 91.69, "learning_rate": 2.193846153846154e-06, "loss": 0.0003, "step": 15725 }, { "epoch": 91.84, "learning_rate": 2.181025641025641e-06, "loss": 0.0003, "step": 15750 }, { "epoch": 91.98, "learning_rate": 2.1682051282051283e-06, "loss": 0.0004, "step": 15775 }, { "epoch": 92.13, "learning_rate": 2.1553846153846156e-06, "loss": 0.0003, "step": 15800 }, { "epoch": 92.27, "learning_rate": 2.142564102564103e-06, "loss": 0.0002, "step": 15825 }, { "epoch": 92.42, "learning_rate": 2.12974358974359e-06, "loss": 0.0005, "step": 15850 }, { "epoch": 92.57, "learning_rate": 2.116923076923077e-06, "loss": 0.0004, "step": 15875 }, { "epoch": 92.71, "learning_rate": 2.104102564102564e-06, "loss": 0.0002, "step": 15900 }, { "epoch": 92.86, "learning_rate": 2.0912820512820514e-06, "loss": 0.0002, "step": 15925 }, { "epoch": 93.0, "learning_rate": 2.0784615384615386e-06, "loss": 0.0003, "step": 15950 }, { "epoch": 93.15, "learning_rate": 2.065641025641026e-06, "loss": 0.0002, "step": 15975 }, { "epoch": 93.29, "learning_rate": 2.052820512820513e-06, "loss": 0.0002, "step": 16000 }, { "epoch": 93.29, "eval_loss": 0.2883666753768921, "eval_runtime": 1313.1981, "eval_samples_per_second": 4.871, "eval_steps_per_second": 0.305, "eval_wer": 5.817452357071214, "step": 16000 }, { "epoch": 93.44, "learning_rate": 2.04e-06, "loss": 0.0002, "step": 16025 }, { "epoch": 93.59, "learning_rate": 2.027179487179487e-06, "loss": 0.0003, "step": 16050 }, { "epoch": 93.73, "learning_rate": 2.0143589743589744e-06, "loss": 0.0002, "step": 16075 }, { "epoch": 93.88, "learning_rate": 2.0015384615384617e-06, "loss": 0.0002, "step": 16100 }, { "epoch": 94.02, "learning_rate": 1.988717948717949e-06, "loss": 0.0002, "step": 16125 }, { "epoch": 94.17, "learning_rate": 1.975897435897436e-06, "loss": 0.0002, "step": 16150 }, { "epoch": 94.31, "learning_rate": 1.963076923076923e-06, "loss": 0.0002, "step": 16175 }, { "epoch": 94.46, "learning_rate": 1.9502564102564103e-06, "loss": 0.0003, "step": 16200 }, { "epoch": 94.61, "learning_rate": 1.9374358974358975e-06, "loss": 0.0002, "step": 16225 }, { "epoch": 94.75, "learning_rate": 1.9246153846153848e-06, "loss": 0.0003, "step": 16250 }, { "epoch": 94.9, "learning_rate": 1.911794871794872e-06, "loss": 0.0002, "step": 16275 }, { "epoch": 95.04, "learning_rate": 1.8989743589743593e-06, "loss": 0.0002, "step": 16300 }, { "epoch": 95.19, "learning_rate": 1.8861538461538465e-06, "loss": 0.0002, "step": 16325 }, { "epoch": 95.34, "learning_rate": 1.8733333333333333e-06, "loss": 0.0004, "step": 16350 }, { "epoch": 95.48, "learning_rate": 1.8605128205128206e-06, "loss": 0.0002, "step": 16375 }, { "epoch": 95.63, "learning_rate": 1.8476923076923078e-06, "loss": 0.0002, "step": 16400 }, { "epoch": 95.77, "learning_rate": 1.834871794871795e-06, "loss": 0.0002, "step": 16425 }, { "epoch": 95.92, "learning_rate": 1.8220512820512823e-06, "loss": 0.0002, "step": 16450 }, { "epoch": 96.06, "learning_rate": 1.8092307692307696e-06, "loss": 0.0002, "step": 16475 }, { "epoch": 96.21, "learning_rate": 1.7964102564102564e-06, "loss": 0.0002, "step": 16500 }, { "epoch": 96.36, "learning_rate": 1.7835897435897436e-06, "loss": 0.0002, "step": 16525 }, { "epoch": 96.5, "learning_rate": 1.7707692307692309e-06, "loss": 0.0002, "step": 16550 }, { "epoch": 96.65, "learning_rate": 1.7579487179487181e-06, "loss": 0.0002, "step": 16575 }, { "epoch": 96.79, "learning_rate": 1.7451282051282054e-06, "loss": 0.0002, "step": 16600 }, { "epoch": 96.94, "learning_rate": 1.7323076923076926e-06, "loss": 0.0002, "step": 16625 }, { "epoch": 97.08, "learning_rate": 1.7194871794871795e-06, "loss": 0.0002, "step": 16650 }, { "epoch": 97.23, "learning_rate": 1.7066666666666667e-06, "loss": 0.0002, "step": 16675 }, { "epoch": 97.38, "learning_rate": 1.693846153846154e-06, "loss": 0.0002, "step": 16700 }, { "epoch": 97.52, "learning_rate": 1.6810256410256412e-06, "loss": 0.0001, "step": 16725 }, { "epoch": 97.67, "learning_rate": 1.6682051282051284e-06, "loss": 0.0001, "step": 16750 }, { "epoch": 97.81, "learning_rate": 1.6553846153846157e-06, "loss": 0.0001, "step": 16775 }, { "epoch": 97.96, "learning_rate": 1.6425641025641025e-06, "loss": 0.0002, "step": 16800 }, { "epoch": 98.1, "learning_rate": 1.6297435897435898e-06, "loss": 0.0002, "step": 16825 }, { "epoch": 98.25, "learning_rate": 1.616923076923077e-06, "loss": 0.0004, "step": 16850 }, { "epoch": 98.4, "learning_rate": 1.6041025641025643e-06, "loss": 0.0003, "step": 16875 }, { "epoch": 98.54, "learning_rate": 1.5912820512820515e-06, "loss": 0.0002, "step": 16900 }, { "epoch": 98.69, "learning_rate": 1.5784615384615388e-06, "loss": 0.0003, "step": 16925 }, { "epoch": 98.83, "learning_rate": 1.5656410256410256e-06, "loss": 0.0002, "step": 16950 }, { "epoch": 98.98, "learning_rate": 1.5528205128205128e-06, "loss": 0.0001, "step": 16975 }, { "epoch": 99.13, "learning_rate": 1.54e-06, "loss": 0.0002, "step": 17000 }, { "epoch": 99.13, "eval_loss": 0.29125919938087463, "eval_runtime": 1314.2645, "eval_samples_per_second": 4.867, "eval_steps_per_second": 0.304, "eval_wer": 5.789783142531042, "step": 17000 }, { "epoch": 99.27, "learning_rate": 1.5271794871794873e-06, "loss": 0.0002, "step": 17025 }, { "epoch": 99.42, "learning_rate": 1.5143589743589746e-06, "loss": 0.0001, "step": 17050 }, { "epoch": 99.56, "learning_rate": 1.5015384615384618e-06, "loss": 0.0001, "step": 17075 }, { "epoch": 99.71, "learning_rate": 1.4887179487179486e-06, "loss": 0.0002, "step": 17100 }, { "epoch": 99.85, "learning_rate": 1.475897435897436e-06, "loss": 0.0002, "step": 17125 }, { "epoch": 100.0, "learning_rate": 1.4630769230769231e-06, "loss": 0.0003, "step": 17150 }, { "epoch": 100.15, "learning_rate": 1.4502564102564104e-06, "loss": 0.0001, "step": 17175 }, { "epoch": 100.29, "learning_rate": 1.4374358974358976e-06, "loss": 0.0002, "step": 17200 }, { "epoch": 100.44, "learning_rate": 1.4246153846153849e-06, "loss": 0.0002, "step": 17225 }, { "epoch": 100.58, "learning_rate": 1.4117948717948717e-06, "loss": 0.0002, "step": 17250 }, { "epoch": 100.73, "learning_rate": 1.398974358974359e-06, "loss": 0.0002, "step": 17275 }, { "epoch": 100.87, "learning_rate": 1.3861538461538462e-06, "loss": 0.0001, "step": 17300 }, { "epoch": 101.02, "learning_rate": 1.3733333333333335e-06, "loss": 0.0002, "step": 17325 }, { "epoch": 101.17, "learning_rate": 1.3605128205128207e-06, "loss": 0.0003, "step": 17350 }, { "epoch": 101.31, "learning_rate": 1.347692307692308e-06, "loss": 0.0002, "step": 17375 }, { "epoch": 101.46, "learning_rate": 1.3348717948717948e-06, "loss": 0.0001, "step": 17400 }, { "epoch": 101.6, "learning_rate": 1.322051282051282e-06, "loss": 0.0002, "step": 17425 }, { "epoch": 101.75, "learning_rate": 1.3092307692307693e-06, "loss": 0.0002, "step": 17450 }, { "epoch": 101.9, "learning_rate": 1.2964102564102565e-06, "loss": 0.0001, "step": 17475 }, { "epoch": 102.04, "learning_rate": 1.2835897435897438e-06, "loss": 0.0001, "step": 17500 }, { "epoch": 102.19, "learning_rate": 1.270769230769231e-06, "loss": 0.0001, "step": 17525 }, { "epoch": 102.33, "learning_rate": 1.2579487179487178e-06, "loss": 0.0002, "step": 17550 }, { "epoch": 102.48, "learning_rate": 1.2451282051282053e-06, "loss": 0.0002, "step": 17575 }, { "epoch": 102.62, "learning_rate": 1.2323076923076923e-06, "loss": 0.0002, "step": 17600 }, { "epoch": 102.77, "learning_rate": 1.2194871794871796e-06, "loss": 0.0001, "step": 17625 }, { "epoch": 102.92, "learning_rate": 1.2066666666666668e-06, "loss": 0.0002, "step": 17650 }, { "epoch": 103.06, "learning_rate": 1.1938461538461539e-06, "loss": 0.0002, "step": 17675 }, { "epoch": 103.21, "learning_rate": 1.1810256410256411e-06, "loss": 0.0001, "step": 17700 }, { "epoch": 103.35, "learning_rate": 1.1682051282051284e-06, "loss": 0.0002, "step": 17725 }, { "epoch": 103.5, "learning_rate": 1.1553846153846154e-06, "loss": 0.0001, "step": 17750 }, { "epoch": 103.64, "learning_rate": 1.1425641025641026e-06, "loss": 0.0002, "step": 17775 }, { "epoch": 103.79, "learning_rate": 1.12974358974359e-06, "loss": 0.0003, "step": 17800 }, { "epoch": 103.94, "learning_rate": 1.116923076923077e-06, "loss": 0.0002, "step": 17825 }, { "epoch": 104.08, "learning_rate": 1.1041025641025642e-06, "loss": 0.0001, "step": 17850 }, { "epoch": 104.23, "learning_rate": 1.0912820512820514e-06, "loss": 0.0002, "step": 17875 }, { "epoch": 104.37, "learning_rate": 1.0784615384615385e-06, "loss": 0.0005, "step": 17900 }, { "epoch": 104.52, "learning_rate": 1.0656410256410257e-06, "loss": 0.0004, "step": 17925 }, { "epoch": 104.66, "learning_rate": 1.052820512820513e-06, "loss": 0.0002, "step": 17950 }, { "epoch": 104.81, "learning_rate": 1.04e-06, "loss": 0.0001, "step": 17975 }, { "epoch": 104.96, "learning_rate": 1.0271794871794872e-06, "loss": 0.0001, "step": 18000 }, { "epoch": 104.96, "eval_loss": 0.2900528013706207, "eval_runtime": 1316.0212, "eval_samples_per_second": 4.861, "eval_steps_per_second": 0.304, "eval_wer": 5.725798083906893, "step": 18000 }, { "epoch": 105.1, "learning_rate": 1.0143589743589745e-06, "loss": 0.0002, "step": 18025 }, { "epoch": 105.25, "learning_rate": 1.0015384615384615e-06, "loss": 0.0002, "step": 18050 }, { "epoch": 105.39, "learning_rate": 9.887179487179488e-07, "loss": 0.0001, "step": 18075 }, { "epoch": 105.54, "learning_rate": 9.75897435897436e-07, "loss": 0.0003, "step": 18100 }, { "epoch": 105.69, "learning_rate": 9.63076923076923e-07, "loss": 0.0002, "step": 18125 }, { "epoch": 105.83, "learning_rate": 9.502564102564103e-07, "loss": 0.0001, "step": 18150 }, { "epoch": 105.98, "learning_rate": 9.374358974358976e-07, "loss": 0.0003, "step": 18175 }, { "epoch": 106.12, "learning_rate": 9.246153846153846e-07, "loss": 0.0001, "step": 18200 }, { "epoch": 106.27, "learning_rate": 9.117948717948718e-07, "loss": 0.0002, "step": 18225 }, { "epoch": 106.41, "learning_rate": 8.989743589743591e-07, "loss": 0.0001, "step": 18250 }, { "epoch": 106.56, "learning_rate": 8.861538461538461e-07, "loss": 0.0001, "step": 18275 }, { "epoch": 106.71, "learning_rate": 8.733333333333334e-07, "loss": 0.0004, "step": 18300 }, { "epoch": 106.85, "learning_rate": 8.605128205128206e-07, "loss": 0.0002, "step": 18325 }, { "epoch": 107.0, "learning_rate": 8.476923076923077e-07, "loss": 0.0001, "step": 18350 }, { "epoch": 107.14, "learning_rate": 8.348717948717949e-07, "loss": 0.0002, "step": 18375 }, { "epoch": 107.29, "learning_rate": 8.220512820512822e-07, "loss": 0.0002, "step": 18400 }, { "epoch": 107.43, "learning_rate": 8.092307692307692e-07, "loss": 0.0001, "step": 18425 }, { "epoch": 107.58, "learning_rate": 7.964102564102564e-07, "loss": 0.0001, "step": 18450 }, { "epoch": 107.73, "learning_rate": 7.835897435897437e-07, "loss": 0.0001, "step": 18475 }, { "epoch": 107.87, "learning_rate": 7.707692307692307e-07, "loss": 0.0001, "step": 18500 }, { "epoch": 108.02, "learning_rate": 7.57948717948718e-07, "loss": 0.0001, "step": 18525 }, { "epoch": 108.16, "learning_rate": 7.451282051282052e-07, "loss": 0.0001, "step": 18550 }, { "epoch": 108.31, "learning_rate": 7.323076923076923e-07, "loss": 0.0001, "step": 18575 }, { "epoch": 108.45, "learning_rate": 7.194871794871795e-07, "loss": 0.0001, "step": 18600 }, { "epoch": 108.6, "learning_rate": 7.066666666666667e-07, "loss": 0.0001, "step": 18625 }, { "epoch": 108.75, "learning_rate": 6.938461538461538e-07, "loss": 0.0001, "step": 18650 }, { "epoch": 108.89, "learning_rate": 6.81025641025641e-07, "loss": 0.0001, "step": 18675 }, { "epoch": 109.04, "learning_rate": 6.682051282051283e-07, "loss": 0.0001, "step": 18700 }, { "epoch": 109.18, "learning_rate": 6.553846153846154e-07, "loss": 0.0002, "step": 18725 }, { "epoch": 109.33, "learning_rate": 6.43076923076923e-07, "loss": 0.0001, "step": 18750 }, { "epoch": 109.48, "learning_rate": 6.302564102564103e-07, "loss": 0.0001, "step": 18775 }, { "epoch": 109.62, "learning_rate": 6.174358974358975e-07, "loss": 0.0001, "step": 18800 }, { "epoch": 109.77, "learning_rate": 6.046153846153847e-07, "loss": 0.0001, "step": 18825 }, { "epoch": 109.91, "learning_rate": 5.917948717948718e-07, "loss": 0.0001, "step": 18850 }, { "epoch": 110.06, "learning_rate": 5.789743589743591e-07, "loss": 0.0001, "step": 18875 }, { "epoch": 110.2, "learning_rate": 5.661538461538462e-07, "loss": 0.0001, "step": 18900 }, { "epoch": 110.35, "learning_rate": 5.533333333333334e-07, "loss": 0.0001, "step": 18925 }, { "epoch": 110.5, "learning_rate": 5.405128205128206e-07, "loss": 0.0001, "step": 18950 }, { "epoch": 110.64, "learning_rate": 5.276923076923078e-07, "loss": 0.0001, "step": 18975 }, { "epoch": 110.79, "learning_rate": 5.148717948717949e-07, "loss": 0.0001, "step": 19000 }, { "epoch": 110.79, "eval_loss": 0.2990764379501343, "eval_runtime": 1314.3745, "eval_samples_per_second": 4.867, "eval_steps_per_second": 0.304, "eval_wer": 5.705046173001764, "step": 19000 }, { "epoch": 110.93, "learning_rate": 5.020512820512821e-07, "loss": 0.0001, "step": 19025 }, { "epoch": 111.08, "learning_rate": 4.892307692307693e-07, "loss": 0.0002, "step": 19050 }, { "epoch": 111.22, "learning_rate": 4.7641025641025643e-07, "loss": 0.0002, "step": 19075 }, { "epoch": 111.37, "learning_rate": 4.635897435897437e-07, "loss": 0.0002, "step": 19100 }, { "epoch": 111.52, "learning_rate": 4.507692307692308e-07, "loss": 0.0001, "step": 19125 }, { "epoch": 111.66, "learning_rate": 4.3794871794871796e-07, "loss": 0.0002, "step": 19150 }, { "epoch": 111.81, "learning_rate": 4.251282051282052e-07, "loss": 0.0001, "step": 19175 }, { "epoch": 111.95, "learning_rate": 4.1230769230769235e-07, "loss": 0.0001, "step": 19200 }, { "epoch": 112.1, "learning_rate": 3.994871794871795e-07, "loss": 0.0002, "step": 19225 }, { "epoch": 112.24, "learning_rate": 3.8666666666666674e-07, "loss": 0.0001, "step": 19250 }, { "epoch": 112.39, "learning_rate": 3.738461538461539e-07, "loss": 0.0001, "step": 19275 }, { "epoch": 112.54, "learning_rate": 3.61025641025641e-07, "loss": 0.0002, "step": 19300 }, { "epoch": 112.68, "learning_rate": 3.4820512820512827e-07, "loss": 0.0001, "step": 19325 }, { "epoch": 112.83, "learning_rate": 3.353846153846154e-07, "loss": 0.0001, "step": 19350 }, { "epoch": 112.97, "learning_rate": 3.2256410256410256e-07, "loss": 0.0001, "step": 19375 }, { "epoch": 113.12, "learning_rate": 3.097435897435898e-07, "loss": 0.0001, "step": 19400 }, { "epoch": 113.27, "learning_rate": 2.9692307692307695e-07, "loss": 0.0001, "step": 19425 }, { "epoch": 113.41, "learning_rate": 2.8410256410256414e-07, "loss": 0.0001, "step": 19450 }, { "epoch": 113.56, "learning_rate": 2.7128205128205134e-07, "loss": 0.0002, "step": 19475 }, { "epoch": 113.7, "learning_rate": 2.584615384615385e-07, "loss": 0.0001, "step": 19500 }, { "epoch": 113.85, "learning_rate": 2.456410256410257e-07, "loss": 0.0001, "step": 19525 }, { "epoch": 113.99, "learning_rate": 2.3282051282051284e-07, "loss": 0.0001, "step": 19550 }, { "epoch": 114.14, "learning_rate": 2.2e-07, "loss": 0.0001, "step": 19575 }, { "epoch": 114.29, "learning_rate": 2.071794871794872e-07, "loss": 0.0001, "step": 19600 }, { "epoch": 114.43, "learning_rate": 1.9435897435897437e-07, "loss": 0.0001, "step": 19625 }, { "epoch": 114.58, "learning_rate": 1.8153846153846154e-07, "loss": 0.0001, "step": 19650 }, { "epoch": 114.72, "learning_rate": 1.6871794871794874e-07, "loss": 0.0001, "step": 19675 }, { "epoch": 114.87, "learning_rate": 1.558974358974359e-07, "loss": 0.0001, "step": 19700 }, { "epoch": 115.01, "learning_rate": 1.430769230769231e-07, "loss": 0.0001, "step": 19725 }, { "epoch": 115.16, "learning_rate": 1.3025641025641027e-07, "loss": 0.0001, "step": 19750 }, { "epoch": 115.31, "learning_rate": 1.1743589743589745e-07, "loss": 0.0001, "step": 19775 }, { "epoch": 115.45, "learning_rate": 1.0461538461538462e-07, "loss": 0.0001, "step": 19800 }, { "epoch": 115.6, "learning_rate": 9.179487179487179e-08, "loss": 0.0001, "step": 19825 }, { "epoch": 115.74, "learning_rate": 7.897435897435898e-08, "loss": 0.0001, "step": 19850 }, { "epoch": 115.89, "learning_rate": 6.615384615384615e-08, "loss": 0.0002, "step": 19875 }, { "epoch": 116.03, "learning_rate": 5.3333333333333334e-08, "loss": 0.0001, "step": 19900 }, { "epoch": 116.18, "learning_rate": 4.0512820512820516e-08, "loss": 0.0001, "step": 19925 }, { "epoch": 116.33, "learning_rate": 2.769230769230769e-08, "loss": 0.0001, "step": 19950 }, { "epoch": 116.47, "learning_rate": 1.4871794871794873e-08, "loss": 0.0001, "step": 19975 }, { "epoch": 116.62, "learning_rate": 2.051282051282051e-09, "loss": 0.0001, "step": 20000 }, { "epoch": 116.62, "eval_loss": 0.29805344343185425, "eval_runtime": 1313.5522, "eval_samples_per_second": 4.87, "eval_steps_per_second": 0.305, "eval_wer": 5.701587521184242, "step": 20000 }, { "epoch": 116.62, "step": 20000, "total_flos": 2.7115017106636716e+21, "train_loss": 0.021558468461339363, "train_runtime": 140487.1776, "train_samples_per_second": 9.111, "train_steps_per_second": 0.142 } ], "logging_steps": 25, "max_steps": 20000, "num_input_tokens_seen": 0, "num_train_epochs": 117, "save_steps": 1000, "total_flos": 2.7115017106636716e+21, "train_batch_size": 32, "trial_name": null, "trial_params": null }