diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,5010 @@ +{ + "best_metric": 4.98585816777042, + "best_model_checkpoint": "./checkpoint-17000", + "epoch": 100.0, + "eval_steps": 1000, + "global_step": 20000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.12, + "learning_rate": 5.000000000000001e-07, + "loss": 0.8801, + "step": 25 + }, + { + "epoch": 0.25, + "learning_rate": 1.0000000000000002e-06, + "loss": 0.4607, + "step": 50 + }, + { + "epoch": 0.38, + "learning_rate": 1.5e-06, + "loss": 0.2246, + "step": 75 + }, + { + "epoch": 0.5, + "learning_rate": 2.0000000000000003e-06, + "loss": 0.1826, + "step": 100 + }, + { + "epoch": 0.62, + "learning_rate": 2.5e-06, + "loss": 0.1661, + "step": 125 + }, + { + "epoch": 0.75, + "learning_rate": 3e-06, + "loss": 0.1611, + "step": 150 + }, + { + "epoch": 0.88, + "learning_rate": 3.5e-06, + "loss": 0.1477, + "step": 175 + }, + { + "epoch": 1.0, + "learning_rate": 4.000000000000001e-06, + "loss": 0.1354, + "step": 200 + }, + { + "epoch": 1.12, + "learning_rate": 4.5e-06, + "loss": 0.0985, + "step": 225 + }, + { + "epoch": 1.25, + "learning_rate": 5e-06, + "loss": 0.0992, + "step": 250 + }, + { + "epoch": 1.38, + "learning_rate": 5.500000000000001e-06, + "loss": 0.1081, + "step": 275 + }, + { + "epoch": 1.5, + "learning_rate": 6e-06, + "loss": 0.1034, + "step": 300 + }, + { + "epoch": 1.62, + "learning_rate": 6.5000000000000004e-06, + "loss": 0.1013, + "step": 325 + }, + { + "epoch": 1.75, + "learning_rate": 7e-06, + "loss": 0.0967, + "step": 350 + }, + { + "epoch": 1.88, + "learning_rate": 7.500000000000001e-06, + "loss": 0.0991, + "step": 375 + }, + { + "epoch": 2.0, + "learning_rate": 8.000000000000001e-06, + "loss": 0.0962, + "step": 400 + }, + { + "epoch": 2.12, + "learning_rate": 8.5e-06, + "loss": 0.047, + "step": 425 + }, + { + "epoch": 2.25, + "learning_rate": 9e-06, + "loss": 0.0538, + "step": 450 + }, + { + "epoch": 2.38, + "learning_rate": 9.5e-06, + "loss": 0.0563, + "step": 475 + }, + { + "epoch": 2.5, + "learning_rate": 1e-05, + "loss": 0.0563, + "step": 500 + }, + { + "epoch": 2.62, + "learning_rate": 9.987179487179488e-06, + "loss": 0.0552, + "step": 525 + }, + { + "epoch": 2.75, + "learning_rate": 9.974358974358974e-06, + "loss": 0.0651, + "step": 550 + }, + { + "epoch": 2.88, + "learning_rate": 9.961538461538463e-06, + "loss": 0.0607, + "step": 575 + }, + { + "epoch": 3.0, + "learning_rate": 9.94871794871795e-06, + "loss": 0.0625, + "step": 600 + }, + { + "epoch": 3.12, + "learning_rate": 9.935897435897437e-06, + "loss": 0.0313, + "step": 625 + }, + { + "epoch": 3.25, + "learning_rate": 9.923076923076923e-06, + "loss": 0.0292, + "step": 650 + }, + { + "epoch": 3.38, + "learning_rate": 9.910256410256412e-06, + "loss": 0.0302, + "step": 675 + }, + { + "epoch": 3.5, + "learning_rate": 9.897435897435899e-06, + "loss": 0.0315, + "step": 700 + }, + { + "epoch": 3.62, + "learning_rate": 9.884615384615386e-06, + "loss": 0.0306, + "step": 725 + }, + { + "epoch": 3.75, + "learning_rate": 9.871794871794872e-06, + "loss": 0.0302, + "step": 750 + }, + { + "epoch": 3.88, + "learning_rate": 9.858974358974361e-06, + "loss": 0.033, + "step": 775 + }, + { + "epoch": 4.0, + "learning_rate": 9.846153846153848e-06, + "loss": 0.0362, + "step": 800 + }, + { + "epoch": 4.12, + "learning_rate": 9.833333333333333e-06, + "loss": 0.0153, + "step": 825 + }, + { + "epoch": 4.25, + "learning_rate": 9.820512820512821e-06, + "loss": 0.0164, + "step": 850 + }, + { + "epoch": 4.38, + "learning_rate": 9.807692307692308e-06, + "loss": 0.0166, + "step": 875 + }, + { + "epoch": 4.5, + "learning_rate": 9.794871794871795e-06, + "loss": 0.0176, + "step": 900 + }, + { + "epoch": 4.62, + "learning_rate": 9.782051282051282e-06, + "loss": 0.0178, + "step": 925 + }, + { + "epoch": 4.75, + "learning_rate": 9.76923076923077e-06, + "loss": 0.0171, + "step": 950 + }, + { + "epoch": 4.88, + "learning_rate": 9.756410256410257e-06, + "loss": 0.0172, + "step": 975 + }, + { + "epoch": 5.0, + "learning_rate": 9.743589743589744e-06, + "loss": 0.0176, + "step": 1000 + }, + { + "epoch": 5.0, + "eval_loss": 0.15631826221942902, + "eval_runtime": 1350.4292, + "eval_samples_per_second": 4.847, + "eval_steps_per_second": 0.304, + "eval_wer": 5.251448675496689, + "step": 1000 + }, + { + "epoch": 5.12, + "learning_rate": 9.730769230769231e-06, + "loss": 0.0094, + "step": 1025 + }, + { + "epoch": 5.25, + "learning_rate": 9.71794871794872e-06, + "loss": 0.0085, + "step": 1050 + }, + { + "epoch": 5.38, + "learning_rate": 9.705128205128206e-06, + "loss": 0.0078, + "step": 1075 + }, + { + "epoch": 5.5, + "learning_rate": 9.692307692307693e-06, + "loss": 0.0083, + "step": 1100 + }, + { + "epoch": 5.62, + "learning_rate": 9.67948717948718e-06, + "loss": 0.0107, + "step": 1125 + }, + { + "epoch": 5.75, + "learning_rate": 9.666666666666667e-06, + "loss": 0.01, + "step": 1150 + }, + { + "epoch": 5.88, + "learning_rate": 9.653846153846155e-06, + "loss": 0.0094, + "step": 1175 + }, + { + "epoch": 6.0, + "learning_rate": 9.641025641025642e-06, + "loss": 0.0137, + "step": 1200 + }, + { + "epoch": 6.12, + "learning_rate": 9.628205128205129e-06, + "loss": 0.0063, + "step": 1225 + }, + { + "epoch": 6.25, + "learning_rate": 9.615384615384616e-06, + "loss": 0.0047, + "step": 1250 + }, + { + "epoch": 6.38, + "learning_rate": 9.602564102564104e-06, + "loss": 0.0061, + "step": 1275 + }, + { + "epoch": 6.5, + "learning_rate": 9.589743589743591e-06, + "loss": 0.0063, + "step": 1300 + }, + { + "epoch": 6.62, + "learning_rate": 9.576923076923078e-06, + "loss": 0.0057, + "step": 1325 + }, + { + "epoch": 6.75, + "learning_rate": 9.564102564102565e-06, + "loss": 0.0059, + "step": 1350 + }, + { + "epoch": 6.88, + "learning_rate": 9.551282051282053e-06, + "loss": 0.0072, + "step": 1375 + }, + { + "epoch": 7.0, + "learning_rate": 9.53846153846154e-06, + "loss": 0.0067, + "step": 1400 + }, + { + "epoch": 7.12, + "learning_rate": 9.525641025641025e-06, + "loss": 0.0037, + "step": 1425 + }, + { + "epoch": 7.25, + "learning_rate": 9.512820512820514e-06, + "loss": 0.005, + "step": 1450 + }, + { + "epoch": 7.38, + "learning_rate": 9.5e-06, + "loss": 0.0042, + "step": 1475 + }, + { + "epoch": 7.5, + "learning_rate": 9.487179487179487e-06, + "loss": 0.0041, + "step": 1500 + }, + { + "epoch": 7.62, + "learning_rate": 9.474358974358974e-06, + "loss": 0.0054, + "step": 1525 + }, + { + "epoch": 7.75, + "learning_rate": 9.461538461538463e-06, + "loss": 0.0043, + "step": 1550 + }, + { + "epoch": 7.88, + "learning_rate": 9.44871794871795e-06, + "loss": 0.0066, + "step": 1575 + }, + { + "epoch": 8.0, + "learning_rate": 9.435897435897436e-06, + "loss": 0.0047, + "step": 1600 + }, + { + "epoch": 8.12, + "learning_rate": 9.423076923076923e-06, + "loss": 0.0031, + "step": 1625 + }, + { + "epoch": 8.25, + "learning_rate": 9.410256410256412e-06, + "loss": 0.0035, + "step": 1650 + }, + { + "epoch": 8.38, + "learning_rate": 9.397435897435899e-06, + "loss": 0.0034, + "step": 1675 + }, + { + "epoch": 8.5, + "learning_rate": 9.384615384615385e-06, + "loss": 0.0031, + "step": 1700 + }, + { + "epoch": 8.62, + "learning_rate": 9.371794871794872e-06, + "loss": 0.0033, + "step": 1725 + }, + { + "epoch": 8.75, + "learning_rate": 9.358974358974359e-06, + "loss": 0.0046, + "step": 1750 + }, + { + "epoch": 8.88, + "learning_rate": 9.346153846153847e-06, + "loss": 0.0042, + "step": 1775 + }, + { + "epoch": 9.0, + "learning_rate": 9.333333333333334e-06, + "loss": 0.0053, + "step": 1800 + }, + { + "epoch": 9.12, + "learning_rate": 9.320512820512821e-06, + "loss": 0.0039, + "step": 1825 + }, + { + "epoch": 9.25, + "learning_rate": 9.307692307692308e-06, + "loss": 0.0023, + "step": 1850 + }, + { + "epoch": 9.38, + "learning_rate": 9.294871794871796e-06, + "loss": 0.0036, + "step": 1875 + }, + { + "epoch": 9.5, + "learning_rate": 9.282051282051283e-06, + "loss": 0.0046, + "step": 1900 + }, + { + "epoch": 9.62, + "learning_rate": 9.26923076923077e-06, + "loss": 0.0042, + "step": 1925 + }, + { + "epoch": 9.75, + "learning_rate": 9.256410256410257e-06, + "loss": 0.0038, + "step": 1950 + }, + { + "epoch": 9.88, + "learning_rate": 9.243589743589745e-06, + "loss": 0.0042, + "step": 1975 + }, + { + "epoch": 10.0, + "learning_rate": 9.230769230769232e-06, + "loss": 0.004, + "step": 2000 + }, + { + "epoch": 10.0, + "eval_loss": 0.18843267858028412, + "eval_runtime": 1346.8567, + "eval_samples_per_second": 4.86, + "eval_steps_per_second": 0.304, + "eval_wer": 5.565328366445916, + "step": 2000 + }, + { + "epoch": 10.12, + "learning_rate": 9.217948717948717e-06, + "loss": 0.0028, + "step": 2025 + }, + { + "epoch": 10.25, + "learning_rate": 9.205128205128206e-06, + "loss": 0.0018, + "step": 2050 + }, + { + "epoch": 10.38, + "learning_rate": 9.192307692307693e-06, + "loss": 0.0024, + "step": 2075 + }, + { + "epoch": 10.5, + "learning_rate": 9.17948717948718e-06, + "loss": 0.0038, + "step": 2100 + }, + { + "epoch": 10.62, + "learning_rate": 9.166666666666666e-06, + "loss": 0.0029, + "step": 2125 + }, + { + "epoch": 10.75, + "learning_rate": 9.153846153846155e-06, + "loss": 0.003, + "step": 2150 + }, + { + "epoch": 10.88, + "learning_rate": 9.141025641025642e-06, + "loss": 0.0035, + "step": 2175 + }, + { + "epoch": 11.0, + "learning_rate": 9.128205128205129e-06, + "loss": 0.0043, + "step": 2200 + }, + { + "epoch": 11.12, + "learning_rate": 9.115384615384615e-06, + "loss": 0.0022, + "step": 2225 + }, + { + "epoch": 11.25, + "learning_rate": 9.102564102564104e-06, + "loss": 0.0024, + "step": 2250 + }, + { + "epoch": 11.38, + "learning_rate": 9.08974358974359e-06, + "loss": 0.0021, + "step": 2275 + }, + { + "epoch": 11.5, + "learning_rate": 9.076923076923078e-06, + "loss": 0.0029, + "step": 2300 + }, + { + "epoch": 11.62, + "learning_rate": 9.064102564102564e-06, + "loss": 0.0038, + "step": 2325 + }, + { + "epoch": 11.75, + "learning_rate": 9.051282051282051e-06, + "loss": 0.004, + "step": 2350 + }, + { + "epoch": 11.88, + "learning_rate": 9.03846153846154e-06, + "loss": 0.0046, + "step": 2375 + }, + { + "epoch": 12.0, + "learning_rate": 9.025641025641027e-06, + "loss": 0.0038, + "step": 2400 + }, + { + "epoch": 12.12, + "learning_rate": 9.012820512820513e-06, + "loss": 0.0035, + "step": 2425 + }, + { + "epoch": 12.25, + "learning_rate": 9e-06, + "loss": 0.0025, + "step": 2450 + }, + { + "epoch": 12.38, + "learning_rate": 8.987179487179489e-06, + "loss": 0.0026, + "step": 2475 + }, + { + "epoch": 12.5, + "learning_rate": 8.974358974358976e-06, + "loss": 0.003, + "step": 2500 + }, + { + "epoch": 12.62, + "learning_rate": 8.961538461538462e-06, + "loss": 0.0023, + "step": 2525 + }, + { + "epoch": 12.75, + "learning_rate": 8.94871794871795e-06, + "loss": 0.0029, + "step": 2550 + }, + { + "epoch": 12.88, + "learning_rate": 8.935897435897438e-06, + "loss": 0.0041, + "step": 2575 + }, + { + "epoch": 13.0, + "learning_rate": 8.923076923076925e-06, + "loss": 0.0046, + "step": 2600 + }, + { + "epoch": 13.12, + "learning_rate": 8.910256410256411e-06, + "loss": 0.0041, + "step": 2625 + }, + { + "epoch": 13.25, + "learning_rate": 8.897435897435898e-06, + "loss": 0.0031, + "step": 2650 + }, + { + "epoch": 13.38, + "learning_rate": 8.884615384615385e-06, + "loss": 0.0043, + "step": 2675 + }, + { + "epoch": 13.5, + "learning_rate": 8.871794871794872e-06, + "loss": 0.0035, + "step": 2700 + }, + { + "epoch": 13.62, + "learning_rate": 8.858974358974359e-06, + "loss": 0.0027, + "step": 2725 + }, + { + "epoch": 13.75, + "learning_rate": 8.846153846153847e-06, + "loss": 0.0031, + "step": 2750 + }, + { + "epoch": 13.88, + "learning_rate": 8.833333333333334e-06, + "loss": 0.0039, + "step": 2775 + }, + { + "epoch": 14.0, + "learning_rate": 8.820512820512821e-06, + "loss": 0.004, + "step": 2800 + }, + { + "epoch": 14.12, + "learning_rate": 8.807692307692308e-06, + "loss": 0.0018, + "step": 2825 + }, + { + "epoch": 14.25, + "learning_rate": 8.794871794871796e-06, + "loss": 0.0016, + "step": 2850 + }, + { + "epoch": 14.38, + "learning_rate": 8.782051282051283e-06, + "loss": 0.0026, + "step": 2875 + }, + { + "epoch": 14.5, + "learning_rate": 8.76923076923077e-06, + "loss": 0.0024, + "step": 2900 + }, + { + "epoch": 14.62, + "learning_rate": 8.756410256410257e-06, + "loss": 0.0034, + "step": 2925 + }, + { + "epoch": 14.75, + "learning_rate": 8.743589743589743e-06, + "loss": 0.0034, + "step": 2950 + }, + { + "epoch": 14.88, + "learning_rate": 8.730769230769232e-06, + "loss": 0.0035, + "step": 2975 + }, + { + "epoch": 15.0, + "learning_rate": 8.717948717948719e-06, + "loss": 0.0039, + "step": 3000 + }, + { + "epoch": 15.0, + "eval_loss": 0.20515076816082, + "eval_runtime": 1346.2767, + "eval_samples_per_second": 4.862, + "eval_steps_per_second": 0.305, + "eval_wer": 5.537734547461368, + "step": 3000 + }, + { + "epoch": 15.12, + "learning_rate": 8.705128205128206e-06, + "loss": 0.003, + "step": 3025 + }, + { + "epoch": 15.25, + "learning_rate": 8.692307692307692e-06, + "loss": 0.0032, + "step": 3050 + }, + { + "epoch": 15.38, + "learning_rate": 8.679487179487181e-06, + "loss": 0.0043, + "step": 3075 + }, + { + "epoch": 15.5, + "learning_rate": 8.666666666666668e-06, + "loss": 0.0034, + "step": 3100 + }, + { + "epoch": 15.62, + "learning_rate": 8.653846153846155e-06, + "loss": 0.0027, + "step": 3125 + }, + { + "epoch": 15.75, + "learning_rate": 8.641025641025641e-06, + "loss": 0.0021, + "step": 3150 + }, + { + "epoch": 15.88, + "learning_rate": 8.62820512820513e-06, + "loss": 0.0036, + "step": 3175 + }, + { + "epoch": 16.0, + "learning_rate": 8.615384615384617e-06, + "loss": 0.0032, + "step": 3200 + }, + { + "epoch": 16.12, + "learning_rate": 8.602564102564104e-06, + "loss": 0.0023, + "step": 3225 + }, + { + "epoch": 16.25, + "learning_rate": 8.58974358974359e-06, + "loss": 0.0017, + "step": 3250 + }, + { + "epoch": 16.38, + "learning_rate": 8.576923076923077e-06, + "loss": 0.0023, + "step": 3275 + }, + { + "epoch": 16.5, + "learning_rate": 8.564102564102564e-06, + "loss": 0.0018, + "step": 3300 + }, + { + "epoch": 16.62, + "learning_rate": 8.551282051282051e-06, + "loss": 0.002, + "step": 3325 + }, + { + "epoch": 16.75, + "learning_rate": 8.53846153846154e-06, + "loss": 0.0021, + "step": 3350 + }, + { + "epoch": 16.88, + "learning_rate": 8.525641025641026e-06, + "loss": 0.0018, + "step": 3375 + }, + { + "epoch": 17.0, + "learning_rate": 8.512820512820513e-06, + "loss": 0.0019, + "step": 3400 + }, + { + "epoch": 17.12, + "learning_rate": 8.5e-06, + "loss": 0.0012, + "step": 3425 + }, + { + "epoch": 17.25, + "learning_rate": 8.487179487179488e-06, + "loss": 0.0013, + "step": 3450 + }, + { + "epoch": 17.38, + "learning_rate": 8.474358974358975e-06, + "loss": 0.0043, + "step": 3475 + }, + { + "epoch": 17.5, + "learning_rate": 8.461538461538462e-06, + "loss": 0.0017, + "step": 3500 + }, + { + "epoch": 17.62, + "learning_rate": 8.448717948717949e-06, + "loss": 0.0023, + "step": 3525 + }, + { + "epoch": 17.75, + "learning_rate": 8.435897435897436e-06, + "loss": 0.002, + "step": 3550 + }, + { + "epoch": 17.88, + "learning_rate": 8.423076923076924e-06, + "loss": 0.0019, + "step": 3575 + }, + { + "epoch": 18.0, + "learning_rate": 8.410256410256411e-06, + "loss": 0.0012, + "step": 3600 + }, + { + "epoch": 18.12, + "learning_rate": 8.397435897435898e-06, + "loss": 0.0015, + "step": 3625 + }, + { + "epoch": 18.25, + "learning_rate": 8.384615384615385e-06, + "loss": 0.0026, + "step": 3650 + }, + { + "epoch": 18.38, + "learning_rate": 8.371794871794873e-06, + "loss": 0.0022, + "step": 3675 + }, + { + "epoch": 18.5, + "learning_rate": 8.35897435897436e-06, + "loss": 0.0019, + "step": 3700 + }, + { + "epoch": 18.62, + "learning_rate": 8.346153846153847e-06, + "loss": 0.0032, + "step": 3725 + }, + { + "epoch": 18.75, + "learning_rate": 8.333333333333334e-06, + "loss": 0.0057, + "step": 3750 + }, + { + "epoch": 18.88, + "learning_rate": 8.320512820512822e-06, + "loss": 0.0041, + "step": 3775 + }, + { + "epoch": 19.0, + "learning_rate": 8.307692307692309e-06, + "loss": 0.0037, + "step": 3800 + }, + { + "epoch": 19.12, + "learning_rate": 8.294871794871796e-06, + "loss": 0.0024, + "step": 3825 + }, + { + "epoch": 19.25, + "learning_rate": 8.282051282051283e-06, + "loss": 0.0027, + "step": 3850 + }, + { + "epoch": 19.38, + "learning_rate": 8.26923076923077e-06, + "loss": 0.0021, + "step": 3875 + }, + { + "epoch": 19.5, + "learning_rate": 8.256410256410256e-06, + "loss": 0.0029, + "step": 3900 + }, + { + "epoch": 19.62, + "learning_rate": 8.243589743589743e-06, + "loss": 0.002, + "step": 3925 + }, + { + "epoch": 19.75, + "learning_rate": 8.230769230769232e-06, + "loss": 0.0032, + "step": 3950 + }, + { + "epoch": 19.88, + "learning_rate": 8.217948717948719e-06, + "loss": 0.0049, + "step": 3975 + }, + { + "epoch": 20.0, + "learning_rate": 8.205128205128205e-06, + "loss": 0.0033, + "step": 4000 + }, + { + "epoch": 20.0, + "eval_loss": 0.20535099506378174, + "eval_runtime": 1340.274, + "eval_samples_per_second": 4.884, + "eval_steps_per_second": 0.306, + "eval_wer": 5.299737858719647, + "step": 4000 + }, + { + "epoch": 20.12, + "learning_rate": 8.192307692307692e-06, + "loss": 0.0019, + "step": 4025 + }, + { + "epoch": 20.25, + "learning_rate": 8.17948717948718e-06, + "loss": 0.0034, + "step": 4050 + }, + { + "epoch": 20.38, + "learning_rate": 8.166666666666668e-06, + "loss": 0.003, + "step": 4075 + }, + { + "epoch": 20.5, + "learning_rate": 8.153846153846154e-06, + "loss": 0.0029, + "step": 4100 + }, + { + "epoch": 20.62, + "learning_rate": 8.141025641025641e-06, + "loss": 0.0018, + "step": 4125 + }, + { + "epoch": 20.75, + "learning_rate": 8.12820512820513e-06, + "loss": 0.0023, + "step": 4150 + }, + { + "epoch": 20.88, + "learning_rate": 8.115384615384617e-06, + "loss": 0.0018, + "step": 4175 + }, + { + "epoch": 21.0, + "learning_rate": 8.102564102564103e-06, + "loss": 0.0029, + "step": 4200 + }, + { + "epoch": 21.12, + "learning_rate": 8.08974358974359e-06, + "loss": 0.0035, + "step": 4225 + }, + { + "epoch": 21.25, + "learning_rate": 8.076923076923077e-06, + "loss": 0.0018, + "step": 4250 + }, + { + "epoch": 21.38, + "learning_rate": 8.064102564102566e-06, + "loss": 0.0025, + "step": 4275 + }, + { + "epoch": 21.5, + "learning_rate": 8.051282051282052e-06, + "loss": 0.0028, + "step": 4300 + }, + { + "epoch": 21.62, + "learning_rate": 8.03846153846154e-06, + "loss": 0.0021, + "step": 4325 + }, + { + "epoch": 21.75, + "learning_rate": 8.025641025641026e-06, + "loss": 0.0016, + "step": 4350 + }, + { + "epoch": 21.88, + "learning_rate": 8.012820512820515e-06, + "loss": 0.0033, + "step": 4375 + }, + { + "epoch": 22.0, + "learning_rate": 8.000000000000001e-06, + "loss": 0.0032, + "step": 4400 + }, + { + "epoch": 22.12, + "learning_rate": 7.987179487179488e-06, + "loss": 0.0043, + "step": 4425 + }, + { + "epoch": 22.25, + "learning_rate": 7.974358974358975e-06, + "loss": 0.0027, + "step": 4450 + }, + { + "epoch": 22.38, + "learning_rate": 7.961538461538462e-06, + "loss": 0.0029, + "step": 4475 + }, + { + "epoch": 22.5, + "learning_rate": 7.948717948717949e-06, + "loss": 0.0029, + "step": 4500 + }, + { + "epoch": 22.62, + "learning_rate": 7.935897435897435e-06, + "loss": 0.0043, + "step": 4525 + }, + { + "epoch": 22.75, + "learning_rate": 7.923076923076924e-06, + "loss": 0.0029, + "step": 4550 + }, + { + "epoch": 22.88, + "learning_rate": 7.91025641025641e-06, + "loss": 0.0038, + "step": 4575 + }, + { + "epoch": 23.0, + "learning_rate": 7.897435897435898e-06, + "loss": 0.0043, + "step": 4600 + }, + { + "epoch": 23.12, + "learning_rate": 7.884615384615384e-06, + "loss": 0.0034, + "step": 4625 + }, + { + "epoch": 23.25, + "learning_rate": 7.871794871794873e-06, + "loss": 0.0031, + "step": 4650 + }, + { + "epoch": 23.38, + "learning_rate": 7.85897435897436e-06, + "loss": 0.0026, + "step": 4675 + }, + { + "epoch": 23.5, + "learning_rate": 7.846153846153847e-06, + "loss": 0.0026, + "step": 4700 + }, + { + "epoch": 23.62, + "learning_rate": 7.833333333333333e-06, + "loss": 0.003, + "step": 4725 + }, + { + "epoch": 23.75, + "learning_rate": 7.820512820512822e-06, + "loss": 0.0043, + "step": 4750 + }, + { + "epoch": 23.88, + "learning_rate": 7.807692307692309e-06, + "loss": 0.0028, + "step": 4775 + }, + { + "epoch": 24.0, + "learning_rate": 7.794871794871796e-06, + "loss": 0.0033, + "step": 4800 + }, + { + "epoch": 24.12, + "learning_rate": 7.782051282051282e-06, + "loss": 0.002, + "step": 4825 + }, + { + "epoch": 24.25, + "learning_rate": 7.76923076923077e-06, + "loss": 0.0015, + "step": 4850 + }, + { + "epoch": 24.38, + "learning_rate": 7.756410256410258e-06, + "loss": 0.0019, + "step": 4875 + }, + { + "epoch": 24.5, + "learning_rate": 7.743589743589745e-06, + "loss": 0.0011, + "step": 4900 + }, + { + "epoch": 24.62, + "learning_rate": 7.730769230769231e-06, + "loss": 0.0018, + "step": 4925 + }, + { + "epoch": 24.75, + "learning_rate": 7.717948717948718e-06, + "loss": 0.0022, + "step": 4950 + }, + { + "epoch": 24.88, + "learning_rate": 7.705128205128207e-06, + "loss": 0.0013, + "step": 4975 + }, + { + "epoch": 25.0, + "learning_rate": 7.692307692307694e-06, + "loss": 0.0012, + "step": 5000 + }, + { + "epoch": 25.0, + "eval_loss": 0.21148668229579926, + "eval_runtime": 1340.6469, + "eval_samples_per_second": 4.883, + "eval_steps_per_second": 0.306, + "eval_wer": 5.1031318984547465, + "step": 5000 + }, + { + "epoch": 25.12, + "learning_rate": 7.67948717948718e-06, + "loss": 0.0006, + "step": 5025 + }, + { + "epoch": 25.25, + "learning_rate": 7.666666666666667e-06, + "loss": 0.0008, + "step": 5050 + }, + { + "epoch": 25.38, + "learning_rate": 7.653846153846154e-06, + "loss": 0.0006, + "step": 5075 + }, + { + "epoch": 25.5, + "learning_rate": 7.641025641025641e-06, + "loss": 0.0005, + "step": 5100 + }, + { + "epoch": 25.62, + "learning_rate": 7.6282051282051286e-06, + "loss": 0.0003, + "step": 5125 + }, + { + "epoch": 25.75, + "learning_rate": 7.615384615384615e-06, + "loss": 0.0008, + "step": 5150 + }, + { + "epoch": 25.88, + "learning_rate": 7.602564102564103e-06, + "loss": 0.0006, + "step": 5175 + }, + { + "epoch": 26.0, + "learning_rate": 7.58974358974359e-06, + "loss": 0.0007, + "step": 5200 + }, + { + "epoch": 26.12, + "learning_rate": 7.5769230769230775e-06, + "loss": 0.0004, + "step": 5225 + }, + { + "epoch": 26.25, + "learning_rate": 7.564102564102564e-06, + "loss": 0.0006, + "step": 5250 + }, + { + "epoch": 26.38, + "learning_rate": 7.551282051282052e-06, + "loss": 0.0004, + "step": 5275 + }, + { + "epoch": 26.5, + "learning_rate": 7.538461538461539e-06, + "loss": 0.0005, + "step": 5300 + }, + { + "epoch": 26.62, + "learning_rate": 7.5256410256410265e-06, + "loss": 0.0004, + "step": 5325 + }, + { + "epoch": 26.75, + "learning_rate": 7.512820512820513e-06, + "loss": 0.0002, + "step": 5350 + }, + { + "epoch": 26.88, + "learning_rate": 7.500000000000001e-06, + "loss": 0.0003, + "step": 5375 + }, + { + "epoch": 27.0, + "learning_rate": 7.487179487179488e-06, + "loss": 0.0007, + "step": 5400 + }, + { + "epoch": 27.12, + "learning_rate": 7.474358974358975e-06, + "loss": 0.0002, + "step": 5425 + }, + { + "epoch": 27.25, + "learning_rate": 7.461538461538462e-06, + "loss": 0.0002, + "step": 5450 + }, + { + "epoch": 27.38, + "learning_rate": 7.448717948717949e-06, + "loss": 0.0002, + "step": 5475 + }, + { + "epoch": 27.5, + "learning_rate": 7.435897435897437e-06, + "loss": 0.0002, + "step": 5500 + }, + { + "epoch": 27.62, + "learning_rate": 7.423076923076924e-06, + "loss": 0.0002, + "step": 5525 + }, + { + "epoch": 27.75, + "learning_rate": 7.410256410256411e-06, + "loss": 0.0004, + "step": 5550 + }, + { + "epoch": 27.88, + "learning_rate": 7.397435897435898e-06, + "loss": 0.0006, + "step": 5575 + }, + { + "epoch": 28.0, + "learning_rate": 7.384615384615386e-06, + "loss": 0.0004, + "step": 5600 + }, + { + "epoch": 28.12, + "learning_rate": 7.371794871794873e-06, + "loss": 0.0008, + "step": 5625 + }, + { + "epoch": 28.25, + "learning_rate": 7.35897435897436e-06, + "loss": 0.0007, + "step": 5650 + }, + { + "epoch": 28.38, + "learning_rate": 7.346153846153847e-06, + "loss": 0.0003, + "step": 5675 + }, + { + "epoch": 28.5, + "learning_rate": 7.333333333333333e-06, + "loss": 0.0007, + "step": 5700 + }, + { + "epoch": 28.62, + "learning_rate": 7.320512820512821e-06, + "loss": 0.0008, + "step": 5725 + }, + { + "epoch": 28.75, + "learning_rate": 7.307692307692308e-06, + "loss": 0.0004, + "step": 5750 + }, + { + "epoch": 28.88, + "learning_rate": 7.294871794871795e-06, + "loss": 0.0004, + "step": 5775 + }, + { + "epoch": 29.0, + "learning_rate": 7.282051282051282e-06, + "loss": 0.0006, + "step": 5800 + }, + { + "epoch": 29.12, + "learning_rate": 7.26923076923077e-06, + "loss": 0.0004, + "step": 5825 + }, + { + "epoch": 29.25, + "learning_rate": 7.256410256410257e-06, + "loss": 0.0008, + "step": 5850 + }, + { + "epoch": 29.38, + "learning_rate": 7.243589743589744e-06, + "loss": 0.0025, + "step": 5875 + }, + { + "epoch": 29.5, + "learning_rate": 7.230769230769231e-06, + "loss": 0.0011, + "step": 5900 + }, + { + "epoch": 29.62, + "learning_rate": 7.217948717948719e-06, + "loss": 0.0006, + "step": 5925 + }, + { + "epoch": 29.75, + "learning_rate": 7.205128205128206e-06, + "loss": 0.0005, + "step": 5950 + }, + { + "epoch": 29.88, + "learning_rate": 7.192307692307693e-06, + "loss": 0.0013, + "step": 5975 + }, + { + "epoch": 30.0, + "learning_rate": 7.17948717948718e-06, + "loss": 0.001, + "step": 6000 + }, + { + "epoch": 30.0, + "eval_loss": 0.21953216195106506, + "eval_runtime": 1340.2045, + "eval_samples_per_second": 4.884, + "eval_steps_per_second": 0.306, + "eval_wer": 5.239376379690949, + "step": 6000 + }, + { + "epoch": 30.12, + "learning_rate": 7.166666666666667e-06, + "loss": 0.0009, + "step": 6025 + }, + { + "epoch": 30.25, + "learning_rate": 7.153846153846155e-06, + "loss": 0.0009, + "step": 6050 + }, + { + "epoch": 30.38, + "learning_rate": 7.1410256410256414e-06, + "loss": 0.0013, + "step": 6075 + }, + { + "epoch": 30.5, + "learning_rate": 7.128205128205129e-06, + "loss": 0.0017, + "step": 6100 + }, + { + "epoch": 30.62, + "learning_rate": 7.115384615384616e-06, + "loss": 0.0024, + "step": 6125 + }, + { + "epoch": 30.75, + "learning_rate": 7.102564102564104e-06, + "loss": 0.0029, + "step": 6150 + }, + { + "epoch": 30.88, + "learning_rate": 7.0897435897435904e-06, + "loss": 0.0033, + "step": 6175 + }, + { + "epoch": 31.0, + "learning_rate": 7.076923076923078e-06, + "loss": 0.0019, + "step": 6200 + }, + { + "epoch": 31.12, + "learning_rate": 7.064102564102565e-06, + "loss": 0.0019, + "step": 6225 + }, + { + "epoch": 31.25, + "learning_rate": 7.051282051282053e-06, + "loss": 0.0023, + "step": 6250 + }, + { + "epoch": 31.38, + "learning_rate": 7.038461538461539e-06, + "loss": 0.0016, + "step": 6275 + }, + { + "epoch": 31.5, + "learning_rate": 7.025641025641025e-06, + "loss": 0.0032, + "step": 6300 + }, + { + "epoch": 31.62, + "learning_rate": 7.012820512820513e-06, + "loss": 0.0031, + "step": 6325 + }, + { + "epoch": 31.75, + "learning_rate": 7e-06, + "loss": 0.0022, + "step": 6350 + }, + { + "epoch": 31.88, + "learning_rate": 6.9871794871794876e-06, + "loss": 0.0027, + "step": 6375 + }, + { + "epoch": 32.0, + "learning_rate": 6.974358974358974e-06, + "loss": 0.0016, + "step": 6400 + }, + { + "epoch": 32.12, + "learning_rate": 6.961538461538462e-06, + "loss": 0.0023, + "step": 6425 + }, + { + "epoch": 32.25, + "learning_rate": 6.948717948717949e-06, + "loss": 0.0027, + "step": 6450 + }, + { + "epoch": 32.38, + "learning_rate": 6.9358974358974366e-06, + "loss": 0.0023, + "step": 6475 + }, + { + "epoch": 32.5, + "learning_rate": 6.923076923076923e-06, + "loss": 0.0014, + "step": 6500 + }, + { + "epoch": 32.62, + "learning_rate": 6.910256410256411e-06, + "loss": 0.0021, + "step": 6525 + }, + { + "epoch": 32.75, + "learning_rate": 6.897435897435898e-06, + "loss": 0.0033, + "step": 6550 + }, + { + "epoch": 32.88, + "learning_rate": 6.8846153846153855e-06, + "loss": 0.0044, + "step": 6575 + }, + { + "epoch": 33.0, + "learning_rate": 6.871794871794872e-06, + "loss": 0.0033, + "step": 6600 + }, + { + "epoch": 33.12, + "learning_rate": 6.858974358974359e-06, + "loss": 0.002, + "step": 6625 + }, + { + "epoch": 33.25, + "learning_rate": 6.846153846153847e-06, + "loss": 0.0032, + "step": 6650 + }, + { + "epoch": 33.38, + "learning_rate": 6.833333333333334e-06, + "loss": 0.0021, + "step": 6675 + }, + { + "epoch": 33.5, + "learning_rate": 6.820512820512821e-06, + "loss": 0.0029, + "step": 6700 + }, + { + "epoch": 33.62, + "learning_rate": 6.807692307692308e-06, + "loss": 0.0023, + "step": 6725 + }, + { + "epoch": 33.75, + "learning_rate": 6.794871794871796e-06, + "loss": 0.0015, + "step": 6750 + }, + { + "epoch": 33.88, + "learning_rate": 6.782051282051283e-06, + "loss": 0.0024, + "step": 6775 + }, + { + "epoch": 34.0, + "learning_rate": 6.76923076923077e-06, + "loss": 0.0022, + "step": 6800 + }, + { + "epoch": 34.12, + "learning_rate": 6.756410256410257e-06, + "loss": 0.0014, + "step": 6825 + }, + { + "epoch": 34.25, + "learning_rate": 6.743589743589745e-06, + "loss": 0.0017, + "step": 6850 + }, + { + "epoch": 34.38, + "learning_rate": 6.730769230769232e-06, + "loss": 0.0012, + "step": 6875 + }, + { + "epoch": 34.5, + "learning_rate": 6.717948717948718e-06, + "loss": 0.001, + "step": 6900 + }, + { + "epoch": 34.62, + "learning_rate": 6.705128205128205e-06, + "loss": 0.0015, + "step": 6925 + }, + { + "epoch": 34.75, + "learning_rate": 6.692307692307692e-06, + "loss": 0.0017, + "step": 6950 + }, + { + "epoch": 34.88, + "learning_rate": 6.67948717948718e-06, + "loss": 0.0013, + "step": 6975 + }, + { + "epoch": 35.0, + "learning_rate": 6.666666666666667e-06, + "loss": 0.001, + "step": 7000 + }, + { + "epoch": 35.0, + "eval_loss": 0.22571906447410583, + "eval_runtime": 1340.1628, + "eval_samples_per_second": 4.884, + "eval_steps_per_second": 0.306, + "eval_wer": 5.344577814569536, + "step": 7000 + }, + { + "epoch": 35.12, + "learning_rate": 6.653846153846154e-06, + "loss": 0.0003, + "step": 7025 + }, + { + "epoch": 35.25, + "learning_rate": 6.641025641025641e-06, + "loss": 0.0015, + "step": 7050 + }, + { + "epoch": 35.38, + "learning_rate": 6.628205128205129e-06, + "loss": 0.001, + "step": 7075 + }, + { + "epoch": 35.5, + "learning_rate": 6.615384615384616e-06, + "loss": 0.0006, + "step": 7100 + }, + { + "epoch": 35.62, + "learning_rate": 6.602564102564103e-06, + "loss": 0.0012, + "step": 7125 + }, + { + "epoch": 35.75, + "learning_rate": 6.58974358974359e-06, + "loss": 0.0005, + "step": 7150 + }, + { + "epoch": 35.88, + "learning_rate": 6.576923076923078e-06, + "loss": 0.0005, + "step": 7175 + }, + { + "epoch": 36.0, + "learning_rate": 6.564102564102565e-06, + "loss": 0.0007, + "step": 7200 + }, + { + "epoch": 36.12, + "learning_rate": 6.5512820512820515e-06, + "loss": 0.0005, + "step": 7225 + }, + { + "epoch": 36.25, + "learning_rate": 6.538461538461539e-06, + "loss": 0.0004, + "step": 7250 + }, + { + "epoch": 36.38, + "learning_rate": 6.525641025641026e-06, + "loss": 0.0006, + "step": 7275 + }, + { + "epoch": 36.5, + "learning_rate": 6.512820512820514e-06, + "loss": 0.001, + "step": 7300 + }, + { + "epoch": 36.62, + "learning_rate": 6.5000000000000004e-06, + "loss": 0.0009, + "step": 7325 + }, + { + "epoch": 36.75, + "learning_rate": 6.487179487179488e-06, + "loss": 0.0025, + "step": 7350 + }, + { + "epoch": 36.88, + "learning_rate": 6.474358974358975e-06, + "loss": 0.0024, + "step": 7375 + }, + { + "epoch": 37.0, + "learning_rate": 6.461538461538463e-06, + "loss": 0.0024, + "step": 7400 + }, + { + "epoch": 37.12, + "learning_rate": 6.4487179487179494e-06, + "loss": 0.0028, + "step": 7425 + }, + { + "epoch": 37.25, + "learning_rate": 6.435897435897437e-06, + "loss": 0.0019, + "step": 7450 + }, + { + "epoch": 37.38, + "learning_rate": 6.423076923076924e-06, + "loss": 0.0013, + "step": 7475 + }, + { + "epoch": 37.5, + "learning_rate": 6.410256410256412e-06, + "loss": 0.0011, + "step": 7500 + }, + { + "epoch": 37.62, + "learning_rate": 6.397435897435898e-06, + "loss": 0.0011, + "step": 7525 + }, + { + "epoch": 37.75, + "learning_rate": 6.384615384615384e-06, + "loss": 0.0013, + "step": 7550 + }, + { + "epoch": 37.88, + "learning_rate": 6.371794871794872e-06, + "loss": 0.0015, + "step": 7575 + }, + { + "epoch": 38.0, + "learning_rate": 6.358974358974359e-06, + "loss": 0.0012, + "step": 7600 + }, + { + "epoch": 38.12, + "learning_rate": 6.3461538461538466e-06, + "loss": 0.0013, + "step": 7625 + }, + { + "epoch": 38.25, + "learning_rate": 6.333333333333333e-06, + "loss": 0.0005, + "step": 7650 + }, + { + "epoch": 38.38, + "learning_rate": 6.320512820512821e-06, + "loss": 0.0007, + "step": 7675 + }, + { + "epoch": 38.5, + "learning_rate": 6.307692307692308e-06, + "loss": 0.0005, + "step": 7700 + }, + { + "epoch": 38.62, + "learning_rate": 6.2948717948717956e-06, + "loss": 0.0005, + "step": 7725 + }, + { + "epoch": 38.75, + "learning_rate": 6.282051282051282e-06, + "loss": 0.0005, + "step": 7750 + }, + { + "epoch": 38.88, + "learning_rate": 6.26923076923077e-06, + "loss": 0.0008, + "step": 7775 + }, + { + "epoch": 39.0, + "learning_rate": 6.256410256410257e-06, + "loss": 0.0007, + "step": 7800 + }, + { + "epoch": 39.12, + "learning_rate": 6.243589743589744e-06, + "loss": 0.0005, + "step": 7825 + }, + { + "epoch": 39.25, + "learning_rate": 6.230769230769231e-06, + "loss": 0.0002, + "step": 7850 + }, + { + "epoch": 39.38, + "learning_rate": 6.217948717948718e-06, + "loss": 0.0004, + "step": 7875 + }, + { + "epoch": 39.5, + "learning_rate": 6.205128205128206e-06, + "loss": 0.0004, + "step": 7900 + }, + { + "epoch": 39.62, + "learning_rate": 6.192307692307693e-06, + "loss": 0.0007, + "step": 7925 + }, + { + "epoch": 39.75, + "learning_rate": 6.17948717948718e-06, + "loss": 0.0006, + "step": 7950 + }, + { + "epoch": 39.88, + "learning_rate": 6.166666666666667e-06, + "loss": 0.0008, + "step": 7975 + }, + { + "epoch": 40.0, + "learning_rate": 6.153846153846155e-06, + "loss": 0.001, + "step": 8000 + }, + { + "epoch": 40.0, + "eval_loss": 0.21784302592277527, + "eval_runtime": 1340.2704, + "eval_samples_per_second": 4.884, + "eval_steps_per_second": 0.306, + "eval_wer": 5.401490066225166, + "step": 8000 + }, + { + "epoch": 40.12, + "learning_rate": 6.141025641025642e-06, + "loss": 0.0009, + "step": 8025 + }, + { + "epoch": 40.25, + "learning_rate": 6.128205128205129e-06, + "loss": 0.0012, + "step": 8050 + }, + { + "epoch": 40.38, + "learning_rate": 6.115384615384616e-06, + "loss": 0.0004, + "step": 8075 + }, + { + "epoch": 40.5, + "learning_rate": 6.102564102564104e-06, + "loss": 0.0003, + "step": 8100 + }, + { + "epoch": 40.62, + "learning_rate": 6.08974358974359e-06, + "loss": 0.0015, + "step": 8125 + }, + { + "epoch": 40.75, + "learning_rate": 6.076923076923077e-06, + "loss": 0.0007, + "step": 8150 + }, + { + "epoch": 40.88, + "learning_rate": 6.064615384615386e-06, + "loss": 0.0014, + "step": 8175 + }, + { + "epoch": 41.0, + "learning_rate": 6.051794871794873e-06, + "loss": 0.0008, + "step": 8200 + }, + { + "epoch": 41.12, + "learning_rate": 6.03897435897436e-06, + "loss": 0.0019, + "step": 8225 + }, + { + "epoch": 41.25, + "learning_rate": 6.026153846153847e-06, + "loss": 0.0017, + "step": 8250 + }, + { + "epoch": 41.38, + "learning_rate": 6.013333333333335e-06, + "loss": 0.0037, + "step": 8275 + }, + { + "epoch": 41.5, + "learning_rate": 6.000512820512821e-06, + "loss": 0.0013, + "step": 8300 + }, + { + "epoch": 41.62, + "learning_rate": 5.9876923076923076e-06, + "loss": 0.0014, + "step": 8325 + }, + { + "epoch": 41.75, + "learning_rate": 5.974871794871795e-06, + "loss": 0.0012, + "step": 8350 + }, + { + "epoch": 41.88, + "learning_rate": 5.962051282051282e-06, + "loss": 0.0014, + "step": 8375 + }, + { + "epoch": 42.0, + "learning_rate": 5.94923076923077e-06, + "loss": 0.001, + "step": 8400 + }, + { + "epoch": 42.12, + "learning_rate": 5.9364102564102566e-06, + "loss": 0.0024, + "step": 8425 + }, + { + "epoch": 42.25, + "learning_rate": 5.923589743589744e-06, + "loss": 0.001, + "step": 8450 + }, + { + "epoch": 42.38, + "learning_rate": 5.910769230769231e-06, + "loss": 0.0008, + "step": 8475 + }, + { + "epoch": 42.5, + "learning_rate": 5.897948717948718e-06, + "loss": 0.0008, + "step": 8500 + }, + { + "epoch": 42.62, + "learning_rate": 5.8851282051282056e-06, + "loss": 0.0007, + "step": 8525 + }, + { + "epoch": 42.75, + "learning_rate": 5.872307692307692e-06, + "loss": 0.0009, + "step": 8550 + }, + { + "epoch": 42.88, + "learning_rate": 5.85948717948718e-06, + "loss": 0.0011, + "step": 8575 + }, + { + "epoch": 43.0, + "learning_rate": 5.846666666666667e-06, + "loss": 0.0009, + "step": 8600 + }, + { + "epoch": 43.12, + "learning_rate": 5.8338461538461545e-06, + "loss": 0.0012, + "step": 8625 + }, + { + "epoch": 43.25, + "learning_rate": 5.821025641025641e-06, + "loss": 0.001, + "step": 8650 + }, + { + "epoch": 43.38, + "learning_rate": 5.808205128205129e-06, + "loss": 0.0006, + "step": 8675 + }, + { + "epoch": 43.5, + "learning_rate": 5.795384615384616e-06, + "loss": 0.0007, + "step": 8700 + }, + { + "epoch": 43.62, + "learning_rate": 5.7825641025641035e-06, + "loss": 0.0006, + "step": 8725 + }, + { + "epoch": 43.75, + "learning_rate": 5.76974358974359e-06, + "loss": 0.0011, + "step": 8750 + }, + { + "epoch": 43.88, + "learning_rate": 5.756923076923078e-06, + "loss": 0.001, + "step": 8775 + }, + { + "epoch": 44.0, + "learning_rate": 5.744102564102565e-06, + "loss": 0.0009, + "step": 8800 + }, + { + "epoch": 44.12, + "learning_rate": 5.7312820512820525e-06, + "loss": 0.0007, + "step": 8825 + }, + { + "epoch": 44.25, + "learning_rate": 5.718461538461539e-06, + "loss": 0.0009, + "step": 8850 + }, + { + "epoch": 44.38, + "learning_rate": 5.705641025641027e-06, + "loss": 0.0008, + "step": 8875 + }, + { + "epoch": 44.5, + "learning_rate": 5.692820512820513e-06, + "loss": 0.0011, + "step": 8900 + }, + { + "epoch": 44.62, + "learning_rate": 5.68e-06, + "loss": 0.0012, + "step": 8925 + }, + { + "epoch": 44.75, + "learning_rate": 5.6671794871794875e-06, + "loss": 0.0006, + "step": 8950 + }, + { + "epoch": 44.88, + "learning_rate": 5.654358974358974e-06, + "loss": 0.0016, + "step": 8975 + }, + { + "epoch": 45.0, + "learning_rate": 5.641538461538462e-06, + "loss": 0.0008, + "step": 9000 + }, + { + "epoch": 45.0, + "eval_loss": 0.22498832643032074, + "eval_runtime": 1339.9701, + "eval_samples_per_second": 4.885, + "eval_steps_per_second": 0.306, + "eval_wer": 5.470474613686535, + "step": 9000 + }, + { + "epoch": 45.12, + "learning_rate": 5.628717948717949e-06, + "loss": 0.0002, + "step": 9025 + }, + { + "epoch": 45.25, + "learning_rate": 5.6158974358974365e-06, + "loss": 0.0003, + "step": 9050 + }, + { + "epoch": 45.38, + "learning_rate": 5.603076923076923e-06, + "loss": 0.0004, + "step": 9075 + }, + { + "epoch": 45.5, + "learning_rate": 5.59025641025641e-06, + "loss": 0.0004, + "step": 9100 + }, + { + "epoch": 45.62, + "learning_rate": 5.577435897435898e-06, + "loss": 0.0002, + "step": 9125 + }, + { + "epoch": 45.75, + "learning_rate": 5.564615384615385e-06, + "loss": 0.0002, + "step": 9150 + }, + { + "epoch": 45.88, + "learning_rate": 5.551794871794872e-06, + "loss": 0.0003, + "step": 9175 + }, + { + "epoch": 46.0, + "learning_rate": 5.538974358974359e-06, + "loss": 0.0001, + "step": 9200 + }, + { + "epoch": 46.12, + "learning_rate": 5.526153846153847e-06, + "loss": 0.0002, + "step": 9225 + }, + { + "epoch": 46.25, + "learning_rate": 5.513333333333334e-06, + "loss": 0.0005, + "step": 9250 + }, + { + "epoch": 46.38, + "learning_rate": 5.500512820512821e-06, + "loss": 0.0001, + "step": 9275 + }, + { + "epoch": 46.5, + "learning_rate": 5.487692307692308e-06, + "loss": 0.0001, + "step": 9300 + }, + { + "epoch": 46.62, + "learning_rate": 5.474871794871796e-06, + "loss": 0.0002, + "step": 9325 + }, + { + "epoch": 46.75, + "learning_rate": 5.462051282051283e-06, + "loss": 0.0003, + "step": 9350 + }, + { + "epoch": 46.88, + "learning_rate": 5.44923076923077e-06, + "loss": 0.0002, + "step": 9375 + }, + { + "epoch": 47.0, + "learning_rate": 5.436410256410257e-06, + "loss": 0.0001, + "step": 9400 + }, + { + "epoch": 47.12, + "learning_rate": 5.423589743589745e-06, + "loss": 0.0002, + "step": 9425 + }, + { + "epoch": 47.25, + "learning_rate": 5.410769230769232e-06, + "loss": 0.0009, + "step": 9450 + }, + { + "epoch": 47.38, + "learning_rate": 5.397948717948719e-06, + "loss": 0.0002, + "step": 9475 + }, + { + "epoch": 47.5, + "learning_rate": 5.385128205128206e-06, + "loss": 0.0011, + "step": 9500 + }, + { + "epoch": 47.62, + "learning_rate": 5.372307692307692e-06, + "loss": 0.0007, + "step": 9525 + }, + { + "epoch": 47.75, + "learning_rate": 5.35948717948718e-06, + "loss": 0.0012, + "step": 9550 + }, + { + "epoch": 47.88, + "learning_rate": 5.346666666666667e-06, + "loss": 0.0004, + "step": 9575 + }, + { + "epoch": 48.0, + "learning_rate": 5.333846153846154e-06, + "loss": 0.0003, + "step": 9600 + }, + { + "epoch": 48.12, + "learning_rate": 5.321025641025641e-06, + "loss": 0.0003, + "step": 9625 + }, + { + "epoch": 48.25, + "learning_rate": 5.308205128205129e-06, + "loss": 0.0007, + "step": 9650 + }, + { + "epoch": 48.38, + "learning_rate": 5.2953846153846156e-06, + "loss": 0.0003, + "step": 9675 + }, + { + "epoch": 48.5, + "learning_rate": 5.282564102564102e-06, + "loss": 0.0004, + "step": 9700 + }, + { + "epoch": 48.62, + "learning_rate": 5.26974358974359e-06, + "loss": 0.0003, + "step": 9725 + }, + { + "epoch": 48.75, + "learning_rate": 5.256923076923077e-06, + "loss": 0.0005, + "step": 9750 + }, + { + "epoch": 48.88, + "learning_rate": 5.2441025641025646e-06, + "loss": 0.0005, + "step": 9775 + }, + { + "epoch": 49.0, + "learning_rate": 5.231282051282051e-06, + "loss": 0.0006, + "step": 9800 + }, + { + "epoch": 49.12, + "learning_rate": 5.218461538461539e-06, + "loss": 0.0018, + "step": 9825 + }, + { + "epoch": 49.25, + "learning_rate": 5.205641025641026e-06, + "loss": 0.0009, + "step": 9850 + }, + { + "epoch": 49.38, + "learning_rate": 5.1928205128205136e-06, + "loss": 0.0007, + "step": 9875 + }, + { + "epoch": 49.5, + "learning_rate": 5.18e-06, + "loss": 0.0015, + "step": 9900 + }, + { + "epoch": 49.62, + "learning_rate": 5.167179487179488e-06, + "loss": 0.0005, + "step": 9925 + }, + { + "epoch": 49.75, + "learning_rate": 5.154358974358975e-06, + "loss": 0.0009, + "step": 9950 + }, + { + "epoch": 49.88, + "learning_rate": 5.1415384615384625e-06, + "loss": 0.0005, + "step": 9975 + }, + { + "epoch": 50.0, + "learning_rate": 5.128717948717949e-06, + "loss": 0.0008, + "step": 10000 + }, + { + "epoch": 50.0, + "eval_loss": 0.23198513686656952, + "eval_runtime": 1339.9497, + "eval_samples_per_second": 4.885, + "eval_steps_per_second": 0.306, + "eval_wer": 5.294564017660044, + "step": 10000 + }, + { + "epoch": 50.12, + "learning_rate": 5.115897435897437e-06, + "loss": 0.0006, + "step": 10025 + }, + { + "epoch": 50.25, + "learning_rate": 5.103076923076924e-06, + "loss": 0.0011, + "step": 10050 + }, + { + "epoch": 50.38, + "learning_rate": 5.0902564102564115e-06, + "loss": 0.0026, + "step": 10075 + }, + { + "epoch": 50.5, + "learning_rate": 5.077435897435898e-06, + "loss": 0.0008, + "step": 10100 + }, + { + "epoch": 50.62, + "learning_rate": 5.064615384615384e-06, + "loss": 0.0006, + "step": 10125 + }, + { + "epoch": 50.75, + "learning_rate": 5.051794871794872e-06, + "loss": 0.0005, + "step": 10150 + }, + { + "epoch": 50.88, + "learning_rate": 5.038974358974359e-06, + "loss": 0.001, + "step": 10175 + }, + { + "epoch": 51.0, + "learning_rate": 5.0261538461538465e-06, + "loss": 0.0007, + "step": 10200 + }, + { + "epoch": 51.12, + "learning_rate": 5.013333333333333e-06, + "loss": 0.0007, + "step": 10225 + }, + { + "epoch": 51.25, + "learning_rate": 5.000512820512821e-06, + "loss": 0.0009, + "step": 10250 + }, + { + "epoch": 51.38, + "learning_rate": 4.987692307692308e-06, + "loss": 0.0006, + "step": 10275 + }, + { + "epoch": 51.5, + "learning_rate": 4.974871794871795e-06, + "loss": 0.0013, + "step": 10300 + }, + { + "epoch": 51.62, + "learning_rate": 4.962051282051282e-06, + "loss": 0.001, + "step": 10325 + }, + { + "epoch": 51.75, + "learning_rate": 4.949230769230769e-06, + "loss": 0.0015, + "step": 10350 + }, + { + "epoch": 51.88, + "learning_rate": 4.936410256410257e-06, + "loss": 0.0018, + "step": 10375 + }, + { + "epoch": 52.0, + "learning_rate": 4.923589743589744e-06, + "loss": 0.0016, + "step": 10400 + }, + { + "epoch": 52.12, + "learning_rate": 4.910769230769231e-06, + "loss": 0.0018, + "step": 10425 + }, + { + "epoch": 52.25, + "learning_rate": 4.897948717948718e-06, + "loss": 0.0012, + "step": 10450 + }, + { + "epoch": 52.38, + "learning_rate": 4.885128205128206e-06, + "loss": 0.001, + "step": 10475 + }, + { + "epoch": 52.5, + "learning_rate": 4.872307692307693e-06, + "loss": 0.0008, + "step": 10500 + }, + { + "epoch": 52.62, + "learning_rate": 4.86e-06, + "loss": 0.0012, + "step": 10525 + }, + { + "epoch": 52.75, + "learning_rate": 4.847179487179488e-06, + "loss": 0.0021, + "step": 10550 + }, + { + "epoch": 52.88, + "learning_rate": 4.8343589743589746e-06, + "loss": 0.001, + "step": 10575 + }, + { + "epoch": 53.0, + "learning_rate": 4.821538461538462e-06, + "loss": 0.0008, + "step": 10600 + }, + { + "epoch": 53.12, + "learning_rate": 4.808717948717949e-06, + "loss": 0.0007, + "step": 10625 + }, + { + "epoch": 53.25, + "learning_rate": 4.795897435897437e-06, + "loss": 0.0007, + "step": 10650 + }, + { + "epoch": 53.38, + "learning_rate": 4.7830769230769235e-06, + "loss": 0.0007, + "step": 10675 + }, + { + "epoch": 53.5, + "learning_rate": 4.770256410256411e-06, + "loss": 0.0005, + "step": 10700 + }, + { + "epoch": 53.62, + "learning_rate": 4.757435897435898e-06, + "loss": 0.0006, + "step": 10725 + }, + { + "epoch": 53.75, + "learning_rate": 4.744615384615385e-06, + "loss": 0.0015, + "step": 10750 + }, + { + "epoch": 53.88, + "learning_rate": 4.731794871794872e-06, + "loss": 0.0007, + "step": 10775 + }, + { + "epoch": 54.0, + "learning_rate": 4.718974358974359e-06, + "loss": 0.0004, + "step": 10800 + }, + { + "epoch": 54.12, + "learning_rate": 4.706153846153846e-06, + "loss": 0.0006, + "step": 10825 + }, + { + "epoch": 54.25, + "learning_rate": 4.693333333333334e-06, + "loss": 0.0002, + "step": 10850 + }, + { + "epoch": 54.38, + "learning_rate": 4.680512820512821e-06, + "loss": 0.0002, + "step": 10875 + }, + { + "epoch": 54.5, + "learning_rate": 4.667692307692308e-06, + "loss": 0.0005, + "step": 10900 + }, + { + "epoch": 54.62, + "learning_rate": 4.654871794871795e-06, + "loss": 0.0004, + "step": 10925 + }, + { + "epoch": 54.75, + "learning_rate": 4.642051282051283e-06, + "loss": 0.0002, + "step": 10950 + }, + { + "epoch": 54.88, + "learning_rate": 4.62923076923077e-06, + "loss": 0.0003, + "step": 10975 + }, + { + "epoch": 55.0, + "learning_rate": 4.616410256410257e-06, + "loss": 0.0002, + "step": 11000 + }, + { + "epoch": 55.0, + "eval_loss": 0.23683854937553406, + "eval_runtime": 1339.9618, + "eval_samples_per_second": 4.885, + "eval_steps_per_second": 0.306, + "eval_wer": 5.351476269315674, + "step": 11000 + }, + { + "epoch": 55.12, + "learning_rate": 4.603589743589744e-06, + "loss": 0.0001, + "step": 11025 + }, + { + "epoch": 55.25, + "learning_rate": 4.590769230769232e-06, + "loss": 0.0001, + "step": 11050 + }, + { + "epoch": 55.38, + "learning_rate": 4.577948717948718e-06, + "loss": 0.0001, + "step": 11075 + }, + { + "epoch": 55.5, + "learning_rate": 4.5651282051282055e-06, + "loss": 0.0003, + "step": 11100 + }, + { + "epoch": 55.62, + "learning_rate": 4.552307692307692e-06, + "loss": 0.0001, + "step": 11125 + }, + { + "epoch": 55.75, + "learning_rate": 4.53948717948718e-06, + "loss": 0.0001, + "step": 11150 + }, + { + "epoch": 55.88, + "learning_rate": 4.526666666666667e-06, + "loss": 0.0001, + "step": 11175 + }, + { + "epoch": 56.0, + "learning_rate": 4.5138461538461545e-06, + "loss": 0.0001, + "step": 11200 + }, + { + "epoch": 56.12, + "learning_rate": 4.501025641025641e-06, + "loss": 0.0001, + "step": 11225 + }, + { + "epoch": 56.25, + "learning_rate": 4.488205128205129e-06, + "loss": 0.0001, + "step": 11250 + }, + { + "epoch": 56.38, + "learning_rate": 4.475384615384616e-06, + "loss": 0.0001, + "step": 11275 + }, + { + "epoch": 56.5, + "learning_rate": 4.4625641025641035e-06, + "loss": 0.0001, + "step": 11300 + }, + { + "epoch": 56.62, + "learning_rate": 4.44974358974359e-06, + "loss": 0.0001, + "step": 11325 + }, + { + "epoch": 56.75, + "learning_rate": 4.436923076923078e-06, + "loss": 0.0001, + "step": 11350 + }, + { + "epoch": 56.88, + "learning_rate": 4.424102564102564e-06, + "loss": 0.0001, + "step": 11375 + }, + { + "epoch": 57.0, + "learning_rate": 4.411282051282052e-06, + "loss": 0.0001, + "step": 11400 + }, + { + "epoch": 57.12, + "learning_rate": 4.3984615384615384e-06, + "loss": 0.0001, + "step": 11425 + }, + { + "epoch": 57.25, + "learning_rate": 4.385641025641026e-06, + "loss": 0.0001, + "step": 11450 + }, + { + "epoch": 57.38, + "learning_rate": 4.372820512820513e-06, + "loss": 0.0, + "step": 11475 + }, + { + "epoch": 57.5, + "learning_rate": 4.360000000000001e-06, + "loss": 0.0, + "step": 11500 + }, + { + "epoch": 57.62, + "learning_rate": 4.3471794871794874e-06, + "loss": 0.0001, + "step": 11525 + }, + { + "epoch": 57.75, + "learning_rate": 4.334358974358975e-06, + "loss": 0.0001, + "step": 11550 + }, + { + "epoch": 57.88, + "learning_rate": 4.321538461538462e-06, + "loss": 0.0001, + "step": 11575 + }, + { + "epoch": 58.0, + "learning_rate": 4.30871794871795e-06, + "loss": 0.0, + "step": 11600 + }, + { + "epoch": 58.12, + "learning_rate": 4.2958974358974364e-06, + "loss": 0.0, + "step": 11625 + }, + { + "epoch": 58.25, + "learning_rate": 4.283076923076924e-06, + "loss": 0.0, + "step": 11650 + }, + { + "epoch": 58.38, + "learning_rate": 4.27025641025641e-06, + "loss": 0.0, + "step": 11675 + }, + { + "epoch": 58.5, + "learning_rate": 4.257435897435898e-06, + "loss": 0.0, + "step": 11700 + }, + { + "epoch": 58.62, + "learning_rate": 4.2446153846153846e-06, + "loss": 0.0, + "step": 11725 + }, + { + "epoch": 58.75, + "learning_rate": 4.231794871794872e-06, + "loss": 0.0, + "step": 11750 + }, + { + "epoch": 58.88, + "learning_rate": 4.218974358974359e-06, + "loss": 0.0, + "step": 11775 + }, + { + "epoch": 59.0, + "learning_rate": 4.206153846153847e-06, + "loss": 0.0, + "step": 11800 + }, + { + "epoch": 59.12, + "learning_rate": 4.1933333333333336e-06, + "loss": 0.0, + "step": 11825 + }, + { + "epoch": 59.25, + "learning_rate": 4.180512820512821e-06, + "loss": 0.0, + "step": 11850 + }, + { + "epoch": 59.38, + "learning_rate": 4.167692307692308e-06, + "loss": 0.0, + "step": 11875 + }, + { + "epoch": 59.5, + "learning_rate": 4.154871794871796e-06, + "loss": 0.0, + "step": 11900 + }, + { + "epoch": 59.62, + "learning_rate": 4.1420512820512826e-06, + "loss": 0.0, + "step": 11925 + }, + { + "epoch": 59.75, + "learning_rate": 4.12923076923077e-06, + "loss": 0.0, + "step": 11950 + }, + { + "epoch": 59.88, + "learning_rate": 4.116410256410256e-06, + "loss": 0.0, + "step": 11975 + }, + { + "epoch": 60.0, + "learning_rate": 4.103589743589744e-06, + "loss": 0.0, + "step": 12000 + }, + { + "epoch": 60.0, + "eval_loss": 0.2550569176673889, + "eval_runtime": 1349.7143, + "eval_samples_per_second": 4.85, + "eval_steps_per_second": 0.304, + "eval_wer": 5.099682671081678, + "step": 12000 + }, + { + "epoch": 60.12, + "learning_rate": 4.090769230769231e-06, + "loss": 0.0, + "step": 12025 + }, + { + "epoch": 60.25, + "learning_rate": 4.077948717948718e-06, + "loss": 0.0, + "step": 12050 + }, + { + "epoch": 60.38, + "learning_rate": 4.065128205128205e-06, + "loss": 0.0, + "step": 12075 + }, + { + "epoch": 60.5, + "learning_rate": 4.052307692307693e-06, + "loss": 0.0, + "step": 12100 + }, + { + "epoch": 60.62, + "learning_rate": 4.03948717948718e-06, + "loss": 0.0, + "step": 12125 + }, + { + "epoch": 60.75, + "learning_rate": 4.026666666666667e-06, + "loss": 0.0, + "step": 12150 + }, + { + "epoch": 60.88, + "learning_rate": 4.013846153846154e-06, + "loss": 0.0, + "step": 12175 + }, + { + "epoch": 61.0, + "learning_rate": 4.001025641025642e-06, + "loss": 0.0, + "step": 12200 + }, + { + "epoch": 61.12, + "learning_rate": 3.988205128205129e-06, + "loss": 0.0, + "step": 12225 + }, + { + "epoch": 61.25, + "learning_rate": 3.975384615384616e-06, + "loss": 0.0, + "step": 12250 + }, + { + "epoch": 61.38, + "learning_rate": 3.962564102564102e-06, + "loss": 0.0, + "step": 12275 + }, + { + "epoch": 61.5, + "learning_rate": 3.94974358974359e-06, + "loss": 0.0, + "step": 12300 + }, + { + "epoch": 61.62, + "learning_rate": 3.936923076923077e-06, + "loss": 0.0, + "step": 12325 + }, + { + "epoch": 61.75, + "learning_rate": 3.9241025641025645e-06, + "loss": 0.0, + "step": 12350 + }, + { + "epoch": 61.88, + "learning_rate": 3.911282051282051e-06, + "loss": 0.0, + "step": 12375 + }, + { + "epoch": 62.0, + "learning_rate": 3.898461538461539e-06, + "loss": 0.0, + "step": 12400 + }, + { + "epoch": 62.12, + "learning_rate": 3.885641025641026e-06, + "loss": 0.0, + "step": 12425 + }, + { + "epoch": 62.25, + "learning_rate": 3.8728205128205135e-06, + "loss": 0.0, + "step": 12450 + }, + { + "epoch": 62.38, + "learning_rate": 3.86e-06, + "loss": 0.0, + "step": 12475 + }, + { + "epoch": 62.5, + "learning_rate": 3.847179487179488e-06, + "loss": 0.0, + "step": 12500 + }, + { + "epoch": 62.62, + "learning_rate": 3.834358974358975e-06, + "loss": 0.0, + "step": 12525 + }, + { + "epoch": 62.75, + "learning_rate": 3.8215384615384625e-06, + "loss": 0.0, + "step": 12550 + }, + { + "epoch": 62.88, + "learning_rate": 3.808717948717949e-06, + "loss": 0.0, + "step": 12575 + }, + { + "epoch": 63.0, + "learning_rate": 3.795897435897436e-06, + "loss": 0.0, + "step": 12600 + }, + { + "epoch": 63.12, + "learning_rate": 3.7830769230769234e-06, + "loss": 0.0, + "step": 12625 + }, + { + "epoch": 63.25, + "learning_rate": 3.7702564102564106e-06, + "loss": 0.0, + "step": 12650 + }, + { + "epoch": 63.38, + "learning_rate": 3.7574358974358974e-06, + "loss": 0.0, + "step": 12675 + }, + { + "epoch": 63.5, + "learning_rate": 3.7446153846153847e-06, + "loss": 0.0, + "step": 12700 + }, + { + "epoch": 63.62, + "learning_rate": 3.731794871794872e-06, + "loss": 0.0, + "step": 12725 + }, + { + "epoch": 63.75, + "learning_rate": 3.718974358974359e-06, + "loss": 0.0, + "step": 12750 + }, + { + "epoch": 63.88, + "learning_rate": 3.7061538461538464e-06, + "loss": 0.0, + "step": 12775 + }, + { + "epoch": 64.0, + "learning_rate": 3.6933333333333337e-06, + "loss": 0.0, + "step": 12800 + }, + { + "epoch": 64.12, + "learning_rate": 3.680512820512821e-06, + "loss": 0.0, + "step": 12825 + }, + { + "epoch": 64.25, + "learning_rate": 3.667692307692308e-06, + "loss": 0.0, + "step": 12850 + }, + { + "epoch": 64.38, + "learning_rate": 3.6548717948717954e-06, + "loss": 0.0, + "step": 12875 + }, + { + "epoch": 64.5, + "learning_rate": 3.6420512820512823e-06, + "loss": 0.0, + "step": 12900 + }, + { + "epoch": 64.62, + "learning_rate": 3.6292307692307695e-06, + "loss": 0.0, + "step": 12925 + }, + { + "epoch": 64.75, + "learning_rate": 3.6164102564102568e-06, + "loss": 0.0, + "step": 12950 + }, + { + "epoch": 64.88, + "learning_rate": 3.6035897435897436e-06, + "loss": 0.0, + "step": 12975 + }, + { + "epoch": 65.0, + "learning_rate": 3.590769230769231e-06, + "loss": 0.0, + "step": 13000 + }, + { + "epoch": 65.0, + "eval_loss": 0.26338452100753784, + "eval_runtime": 1347.4621, + "eval_samples_per_second": 4.858, + "eval_steps_per_second": 0.304, + "eval_wer": 5.073813465783664, + "step": 13000 + }, + { + "epoch": 65.12, + "learning_rate": 3.577948717948718e-06, + "loss": 0.0, + "step": 13025 + }, + { + "epoch": 65.25, + "learning_rate": 3.5651282051282053e-06, + "loss": 0.0, + "step": 13050 + }, + { + "epoch": 65.38, + "learning_rate": 3.5523076923076926e-06, + "loss": 0.0, + "step": 13075 + }, + { + "epoch": 65.5, + "learning_rate": 3.53948717948718e-06, + "loss": 0.0, + "step": 13100 + }, + { + "epoch": 65.62, + "learning_rate": 3.526666666666667e-06, + "loss": 0.0, + "step": 13125 + }, + { + "epoch": 65.75, + "learning_rate": 3.5138461538461543e-06, + "loss": 0.0, + "step": 13150 + }, + { + "epoch": 65.88, + "learning_rate": 3.5010256410256416e-06, + "loss": 0.0, + "step": 13175 + }, + { + "epoch": 66.0, + "learning_rate": 3.4882051282051284e-06, + "loss": 0.0, + "step": 13200 + }, + { + "epoch": 66.12, + "learning_rate": 3.4753846153846156e-06, + "loss": 0.0, + "step": 13225 + }, + { + "epoch": 66.25, + "learning_rate": 3.462564102564103e-06, + "loss": 0.0, + "step": 13250 + }, + { + "epoch": 66.38, + "learning_rate": 3.4497435897435897e-06, + "loss": 0.0, + "step": 13275 + }, + { + "epoch": 66.5, + "learning_rate": 3.436923076923077e-06, + "loss": 0.0, + "step": 13300 + }, + { + "epoch": 66.62, + "learning_rate": 3.424102564102564e-06, + "loss": 0.0, + "step": 13325 + }, + { + "epoch": 66.75, + "learning_rate": 3.4112820512820514e-06, + "loss": 0.0, + "step": 13350 + }, + { + "epoch": 66.88, + "learning_rate": 3.3984615384615387e-06, + "loss": 0.0, + "step": 13375 + }, + { + "epoch": 67.0, + "learning_rate": 3.385641025641026e-06, + "loss": 0.0, + "step": 13400 + }, + { + "epoch": 67.12, + "learning_rate": 3.372820512820513e-06, + "loss": 0.0, + "step": 13425 + }, + { + "epoch": 67.25, + "learning_rate": 3.3600000000000004e-06, + "loss": 0.0, + "step": 13450 + }, + { + "epoch": 67.38, + "learning_rate": 3.3471794871794877e-06, + "loss": 0.0, + "step": 13475 + }, + { + "epoch": 67.5, + "learning_rate": 3.3343589743589745e-06, + "loss": 0.0, + "step": 13500 + }, + { + "epoch": 67.62, + "learning_rate": 3.3215384615384618e-06, + "loss": 0.0, + "step": 13525 + }, + { + "epoch": 67.75, + "learning_rate": 3.308717948717949e-06, + "loss": 0.0, + "step": 13550 + }, + { + "epoch": 67.88, + "learning_rate": 3.2958974358974363e-06, + "loss": 0.0, + "step": 13575 + }, + { + "epoch": 68.0, + "learning_rate": 3.283076923076923e-06, + "loss": 0.0, + "step": 13600 + }, + { + "epoch": 68.12, + "learning_rate": 3.2702564102564103e-06, + "loss": 0.0, + "step": 13625 + }, + { + "epoch": 68.25, + "learning_rate": 3.2574358974358976e-06, + "loss": 0.0, + "step": 13650 + }, + { + "epoch": 68.38, + "learning_rate": 3.244615384615385e-06, + "loss": 0.0, + "step": 13675 + }, + { + "epoch": 68.5, + "learning_rate": 3.231794871794872e-06, + "loss": 0.0, + "step": 13700 + }, + { + "epoch": 68.62, + "learning_rate": 3.2189743589743593e-06, + "loss": 0.0, + "step": 13725 + }, + { + "epoch": 68.75, + "learning_rate": 3.2061538461538466e-06, + "loss": 0.0, + "step": 13750 + }, + { + "epoch": 68.88, + "learning_rate": 3.193333333333334e-06, + "loss": 0.0, + "step": 13775 + }, + { + "epoch": 69.0, + "learning_rate": 3.1805128205128206e-06, + "loss": 0.0, + "step": 13800 + }, + { + "epoch": 69.12, + "learning_rate": 3.167692307692308e-06, + "loss": 0.0, + "step": 13825 + }, + { + "epoch": 69.25, + "learning_rate": 3.154871794871795e-06, + "loss": 0.0, + "step": 13850 + }, + { + "epoch": 69.38, + "learning_rate": 3.1420512820512824e-06, + "loss": 0.0, + "step": 13875 + }, + { + "epoch": 69.5, + "learning_rate": 3.129230769230769e-06, + "loss": 0.0, + "step": 13900 + }, + { + "epoch": 69.62, + "learning_rate": 3.1164102564102565e-06, + "loss": 0.0, + "step": 13925 + }, + { + "epoch": 69.75, + "learning_rate": 3.1035897435897437e-06, + "loss": 0.0, + "step": 13950 + }, + { + "epoch": 69.88, + "learning_rate": 3.090769230769231e-06, + "loss": 0.0, + "step": 13975 + }, + { + "epoch": 70.0, + "learning_rate": 3.077948717948718e-06, + "loss": 0.0, + "step": 14000 + }, + { + "epoch": 70.0, + "eval_loss": 0.26965048909187317, + "eval_runtime": 1346.2647, + "eval_samples_per_second": 4.862, + "eval_steps_per_second": 0.305, + "eval_wer": 5.035871964679912, + "step": 14000 + }, + { + "epoch": 70.12, + "learning_rate": 3.0651282051282054e-06, + "loss": 0.0, + "step": 14025 + }, + { + "epoch": 70.25, + "learning_rate": 3.0523076923076927e-06, + "loss": 0.0, + "step": 14050 + }, + { + "epoch": 70.38, + "learning_rate": 3.03948717948718e-06, + "loss": 0.0, + "step": 14075 + }, + { + "epoch": 70.5, + "learning_rate": 3.0266666666666668e-06, + "loss": 0.0, + "step": 14100 + }, + { + "epoch": 70.62, + "learning_rate": 3.013846153846154e-06, + "loss": 0.0, + "step": 14125 + }, + { + "epoch": 70.75, + "learning_rate": 3.0010256410256413e-06, + "loss": 0.0, + "step": 14150 + }, + { + "epoch": 70.88, + "learning_rate": 2.9882051282051285e-06, + "loss": 0.0, + "step": 14175 + }, + { + "epoch": 71.0, + "learning_rate": 2.9753846153846153e-06, + "loss": 0.0, + "step": 14200 + }, + { + "epoch": 71.12, + "learning_rate": 2.9625641025641026e-06, + "loss": 0.0, + "step": 14225 + }, + { + "epoch": 71.25, + "learning_rate": 2.94974358974359e-06, + "loss": 0.0, + "step": 14250 + }, + { + "epoch": 71.38, + "learning_rate": 2.936923076923077e-06, + "loss": 0.0, + "step": 14275 + }, + { + "epoch": 71.5, + "learning_rate": 2.9241025641025643e-06, + "loss": 0.0, + "step": 14300 + }, + { + "epoch": 71.62, + "learning_rate": 2.9112820512820516e-06, + "loss": 0.0, + "step": 14325 + }, + { + "epoch": 71.75, + "learning_rate": 2.898461538461539e-06, + "loss": 0.0, + "step": 14350 + }, + { + "epoch": 71.88, + "learning_rate": 2.885641025641026e-06, + "loss": 0.0, + "step": 14375 + }, + { + "epoch": 72.0, + "learning_rate": 2.8728205128205133e-06, + "loss": 0.0, + "step": 14400 + }, + { + "epoch": 72.12, + "learning_rate": 2.86e-06, + "loss": 0.0, + "step": 14425 + }, + { + "epoch": 72.25, + "learning_rate": 2.8471794871794874e-06, + "loss": 0.0, + "step": 14450 + }, + { + "epoch": 72.38, + "learning_rate": 2.8343589743589746e-06, + "loss": 0.0, + "step": 14475 + }, + { + "epoch": 72.5, + "learning_rate": 2.8215384615384615e-06, + "loss": 0.0, + "step": 14500 + }, + { + "epoch": 72.62, + "learning_rate": 2.8087179487179487e-06, + "loss": 0.0, + "step": 14525 + }, + { + "epoch": 72.75, + "learning_rate": 2.795897435897436e-06, + "loss": 0.0, + "step": 14550 + }, + { + "epoch": 72.88, + "learning_rate": 2.783076923076923e-06, + "loss": 0.0, + "step": 14575 + }, + { + "epoch": 73.0, + "learning_rate": 2.7702564102564105e-06, + "loss": 0.0, + "step": 14600 + }, + { + "epoch": 73.12, + "learning_rate": 2.7574358974358977e-06, + "loss": 0.0, + "step": 14625 + }, + { + "epoch": 73.25, + "learning_rate": 2.744615384615385e-06, + "loss": 0.0, + "step": 14650 + }, + { + "epoch": 73.38, + "learning_rate": 2.731794871794872e-06, + "loss": 0.0, + "step": 14675 + }, + { + "epoch": 73.5, + "learning_rate": 2.7189743589743594e-06, + "loss": 0.0, + "step": 14700 + }, + { + "epoch": 73.62, + "learning_rate": 2.7061538461538463e-06, + "loss": 0.0, + "step": 14725 + }, + { + "epoch": 73.75, + "learning_rate": 2.6933333333333335e-06, + "loss": 0.0, + "step": 14750 + }, + { + "epoch": 73.88, + "learning_rate": 2.6805128205128208e-06, + "loss": 0.0, + "step": 14775 + }, + { + "epoch": 74.0, + "learning_rate": 2.6676923076923076e-06, + "loss": 0.0, + "step": 14800 + }, + { + "epoch": 74.12, + "learning_rate": 2.654871794871795e-06, + "loss": 0.0, + "step": 14825 + }, + { + "epoch": 74.25, + "learning_rate": 2.642051282051282e-06, + "loss": 0.0, + "step": 14850 + }, + { + "epoch": 74.38, + "learning_rate": 2.6292307692307693e-06, + "loss": 0.0, + "step": 14875 + }, + { + "epoch": 74.5, + "learning_rate": 2.6164102564102566e-06, + "loss": 0.0, + "step": 14900 + }, + { + "epoch": 74.62, + "learning_rate": 2.603589743589744e-06, + "loss": 0.0, + "step": 14925 + }, + { + "epoch": 74.75, + "learning_rate": 2.590769230769231e-06, + "loss": 0.0, + "step": 14950 + }, + { + "epoch": 74.88, + "learning_rate": 2.5779487179487183e-06, + "loss": 0.0, + "step": 14975 + }, + { + "epoch": 75.0, + "learning_rate": 2.5651282051282056e-06, + "loss": 0.0, + "step": 15000 + }, + { + "epoch": 75.0, + "eval_loss": 0.2752314507961273, + "eval_runtime": 1341.0682, + "eval_samples_per_second": 4.881, + "eval_steps_per_second": 0.306, + "eval_wer": 5.01862582781457, + "step": 15000 + }, + { + "epoch": 75.12, + "learning_rate": 2.5523076923076924e-06, + "loss": 0.0, + "step": 15025 + }, + { + "epoch": 75.25, + "learning_rate": 2.5394871794871797e-06, + "loss": 0.0, + "step": 15050 + }, + { + "epoch": 75.38, + "learning_rate": 2.526666666666667e-06, + "loss": 0.0, + "step": 15075 + }, + { + "epoch": 75.5, + "learning_rate": 2.513846153846154e-06, + "loss": 0.0, + "step": 15100 + }, + { + "epoch": 75.62, + "learning_rate": 2.501025641025641e-06, + "loss": 0.0, + "step": 15125 + }, + { + "epoch": 75.75, + "learning_rate": 2.4882051282051282e-06, + "loss": 0.0, + "step": 15150 + }, + { + "epoch": 75.88, + "learning_rate": 2.4753846153846155e-06, + "loss": 0.0, + "step": 15175 + }, + { + "epoch": 76.0, + "learning_rate": 2.4625641025641027e-06, + "loss": 0.0, + "step": 15200 + }, + { + "epoch": 76.12, + "learning_rate": 2.44974358974359e-06, + "loss": 0.0, + "step": 15225 + }, + { + "epoch": 76.25, + "learning_rate": 2.436923076923077e-06, + "loss": 0.0, + "step": 15250 + }, + { + "epoch": 76.38, + "learning_rate": 2.424102564102564e-06, + "loss": 0.0, + "step": 15275 + }, + { + "epoch": 76.5, + "learning_rate": 2.4112820512820513e-06, + "loss": 0.0, + "step": 15300 + }, + { + "epoch": 76.62, + "learning_rate": 2.3984615384615385e-06, + "loss": 0.0, + "step": 15325 + }, + { + "epoch": 76.75, + "learning_rate": 2.3856410256410258e-06, + "loss": 0.0, + "step": 15350 + }, + { + "epoch": 76.88, + "learning_rate": 2.372820512820513e-06, + "loss": 0.0, + "step": 15375 + }, + { + "epoch": 77.0, + "learning_rate": 2.3600000000000003e-06, + "loss": 0.0, + "step": 15400 + }, + { + "epoch": 77.12, + "learning_rate": 2.347179487179487e-06, + "loss": 0.0, + "step": 15425 + }, + { + "epoch": 77.25, + "learning_rate": 2.3343589743589743e-06, + "loss": 0.0, + "step": 15450 + }, + { + "epoch": 77.38, + "learning_rate": 2.3215384615384616e-06, + "loss": 0.0, + "step": 15475 + }, + { + "epoch": 77.5, + "learning_rate": 2.308717948717949e-06, + "loss": 0.0, + "step": 15500 + }, + { + "epoch": 77.62, + "learning_rate": 2.295897435897436e-06, + "loss": 0.0, + "step": 15525 + }, + { + "epoch": 77.75, + "learning_rate": 2.2830769230769233e-06, + "loss": 0.0, + "step": 15550 + }, + { + "epoch": 77.88, + "learning_rate": 2.27025641025641e-06, + "loss": 0.0, + "step": 15575 + }, + { + "epoch": 78.0, + "learning_rate": 2.2574358974358974e-06, + "loss": 0.0, + "step": 15600 + }, + { + "epoch": 78.12, + "learning_rate": 2.2446153846153847e-06, + "loss": 0.0, + "step": 15625 + }, + { + "epoch": 78.25, + "learning_rate": 2.231794871794872e-06, + "loss": 0.0, + "step": 15650 + }, + { + "epoch": 78.38, + "learning_rate": 2.218974358974359e-06, + "loss": 0.0, + "step": 15675 + }, + { + "epoch": 78.5, + "learning_rate": 2.2061538461538464e-06, + "loss": 0.0, + "step": 15700 + }, + { + "epoch": 78.62, + "learning_rate": 2.1933333333333332e-06, + "loss": 0.0, + "step": 15725 + }, + { + "epoch": 78.75, + "learning_rate": 2.1805128205128205e-06, + "loss": 0.0, + "step": 15750 + }, + { + "epoch": 78.88, + "learning_rate": 2.1676923076923077e-06, + "loss": 0.0, + "step": 15775 + }, + { + "epoch": 79.0, + "learning_rate": 2.154871794871795e-06, + "loss": 0.0, + "step": 15800 + }, + { + "epoch": 79.12, + "learning_rate": 2.1420512820512822e-06, + "loss": 0.0, + "step": 15825 + }, + { + "epoch": 79.25, + "learning_rate": 2.1292307692307695e-06, + "loss": 0.0, + "step": 15850 + }, + { + "epoch": 79.38, + "learning_rate": 2.1164102564102563e-06, + "loss": 0.0, + "step": 15875 + }, + { + "epoch": 79.5, + "learning_rate": 2.1035897435897435e-06, + "loss": 0.0, + "step": 15900 + }, + { + "epoch": 79.62, + "learning_rate": 2.0907692307692308e-06, + "loss": 0.0, + "step": 15925 + }, + { + "epoch": 79.75, + "learning_rate": 2.077948717948718e-06, + "loss": 0.0, + "step": 15950 + }, + { + "epoch": 79.88, + "learning_rate": 2.0651282051282053e-06, + "loss": 0.0, + "step": 15975 + }, + { + "epoch": 80.0, + "learning_rate": 2.0523076923076925e-06, + "loss": 0.0, + "step": 16000 + }, + { + "epoch": 80.0, + "eval_loss": 0.28039103746414185, + "eval_runtime": 1340.8785, + "eval_samples_per_second": 4.882, + "eval_steps_per_second": 0.306, + "eval_wer": 5.00655353200883, + "step": 16000 + }, + { + "epoch": 80.12, + "learning_rate": 2.0394871794871794e-06, + "loss": 0.0, + "step": 16025 + }, + { + "epoch": 80.25, + "learning_rate": 2.0266666666666666e-06, + "loss": 0.0, + "step": 16050 + }, + { + "epoch": 80.38, + "learning_rate": 2.013846153846154e-06, + "loss": 0.0, + "step": 16075 + }, + { + "epoch": 80.5, + "learning_rate": 2.001025641025641e-06, + "loss": 0.0, + "step": 16100 + }, + { + "epoch": 80.62, + "learning_rate": 1.9882051282051283e-06, + "loss": 0.0, + "step": 16125 + }, + { + "epoch": 80.75, + "learning_rate": 1.9753846153846156e-06, + "loss": 0.0, + "step": 16150 + }, + { + "epoch": 80.88, + "learning_rate": 1.9625641025641024e-06, + "loss": 0.0, + "step": 16175 + }, + { + "epoch": 81.0, + "learning_rate": 1.9497435897435897e-06, + "loss": 0.0, + "step": 16200 + }, + { + "epoch": 81.12, + "learning_rate": 1.936923076923077e-06, + "loss": 0.0, + "step": 16225 + }, + { + "epoch": 81.25, + "learning_rate": 1.924102564102564e-06, + "loss": 0.0, + "step": 16250 + }, + { + "epoch": 81.38, + "learning_rate": 1.9112820512820514e-06, + "loss": 0.0, + "step": 16275 + }, + { + "epoch": 81.5, + "learning_rate": 1.8984615384615387e-06, + "loss": 0.0, + "step": 16300 + }, + { + "epoch": 81.62, + "learning_rate": 1.8856410256410257e-06, + "loss": 0.0, + "step": 16325 + }, + { + "epoch": 81.75, + "learning_rate": 1.872820512820513e-06, + "loss": 0.0, + "step": 16350 + }, + { + "epoch": 81.88, + "learning_rate": 1.8600000000000002e-06, + "loss": 0.0, + "step": 16375 + }, + { + "epoch": 82.0, + "learning_rate": 1.8471794871794874e-06, + "loss": 0.0, + "step": 16400 + }, + { + "epoch": 82.12, + "learning_rate": 1.8343589743589747e-06, + "loss": 0.0, + "step": 16425 + }, + { + "epoch": 82.25, + "learning_rate": 1.8215384615384617e-06, + "loss": 0.0, + "step": 16450 + }, + { + "epoch": 82.38, + "learning_rate": 1.8087179487179488e-06, + "loss": 0.0, + "step": 16475 + }, + { + "epoch": 82.5, + "learning_rate": 1.795897435897436e-06, + "loss": 0.0, + "step": 16500 + }, + { + "epoch": 82.62, + "learning_rate": 1.7830769230769233e-06, + "loss": 0.0, + "step": 16525 + }, + { + "epoch": 82.75, + "learning_rate": 1.7702564102564105e-06, + "loss": 0.0, + "step": 16550 + }, + { + "epoch": 82.88, + "learning_rate": 1.7574358974358978e-06, + "loss": 0.0, + "step": 16575 + }, + { + "epoch": 83.0, + "learning_rate": 1.7446153846153848e-06, + "loss": 0.0, + "step": 16600 + }, + { + "epoch": 83.12, + "learning_rate": 1.7317948717948718e-06, + "loss": 0.0, + "step": 16625 + }, + { + "epoch": 83.25, + "learning_rate": 1.718974358974359e-06, + "loss": 0.0, + "step": 16650 + }, + { + "epoch": 83.38, + "learning_rate": 1.7061538461538463e-06, + "loss": 0.0, + "step": 16675 + }, + { + "epoch": 83.5, + "learning_rate": 1.6933333333333336e-06, + "loss": 0.0, + "step": 16700 + }, + { + "epoch": 83.62, + "learning_rate": 1.6805128205128208e-06, + "loss": 0.0, + "step": 16725 + }, + { + "epoch": 83.75, + "learning_rate": 1.6676923076923079e-06, + "loss": 0.0, + "step": 16750 + }, + { + "epoch": 83.88, + "learning_rate": 1.6548717948717949e-06, + "loss": 0.0, + "step": 16775 + }, + { + "epoch": 84.0, + "learning_rate": 1.6420512820512821e-06, + "loss": 0.0, + "step": 16800 + }, + { + "epoch": 84.12, + "learning_rate": 1.6292307692307694e-06, + "loss": 0.0, + "step": 16825 + }, + { + "epoch": 84.25, + "learning_rate": 1.6164102564102566e-06, + "loss": 0.0, + "step": 16850 + }, + { + "epoch": 84.38, + "learning_rate": 1.6035897435897439e-06, + "loss": 0.0, + "step": 16875 + }, + { + "epoch": 84.5, + "learning_rate": 1.590769230769231e-06, + "loss": 0.0, + "step": 16900 + }, + { + "epoch": 84.62, + "learning_rate": 1.577948717948718e-06, + "loss": 0.0, + "step": 16925 + }, + { + "epoch": 84.75, + "learning_rate": 1.5651282051282052e-06, + "loss": 0.0, + "step": 16950 + }, + { + "epoch": 84.88, + "learning_rate": 1.5523076923076924e-06, + "loss": 0.0, + "step": 16975 + }, + { + "epoch": 85.0, + "learning_rate": 1.5394871794871797e-06, + "loss": 0.0, + "step": 17000 + }, + { + "epoch": 85.0, + "eval_loss": 0.2852415442466736, + "eval_runtime": 1342.1289, + "eval_samples_per_second": 4.877, + "eval_steps_per_second": 0.305, + "eval_wer": 4.98585816777042, + "step": 17000 + }, + { + "epoch": 85.12, + "learning_rate": 1.526666666666667e-06, + "loss": 0.0, + "step": 17025 + }, + { + "epoch": 85.25, + "learning_rate": 1.513846153846154e-06, + "loss": 0.0, + "step": 17050 + }, + { + "epoch": 85.38, + "learning_rate": 1.501025641025641e-06, + "loss": 0.0, + "step": 17075 + }, + { + "epoch": 85.5, + "learning_rate": 1.4882051282051283e-06, + "loss": 0.0, + "step": 17100 + }, + { + "epoch": 85.62, + "learning_rate": 1.4753846153846155e-06, + "loss": 0.0, + "step": 17125 + }, + { + "epoch": 85.75, + "learning_rate": 1.4625641025641028e-06, + "loss": 0.0, + "step": 17150 + }, + { + "epoch": 85.88, + "learning_rate": 1.44974358974359e-06, + "loss": 0.0, + "step": 17175 + }, + { + "epoch": 86.0, + "learning_rate": 1.436923076923077e-06, + "loss": 0.0, + "step": 17200 + }, + { + "epoch": 86.12, + "learning_rate": 1.424102564102564e-06, + "loss": 0.0, + "step": 17225 + }, + { + "epoch": 86.25, + "learning_rate": 1.4112820512820513e-06, + "loss": 0.0, + "step": 17250 + }, + { + "epoch": 86.38, + "learning_rate": 1.3984615384615386e-06, + "loss": 0.0, + "step": 17275 + }, + { + "epoch": 86.5, + "learning_rate": 1.3856410256410258e-06, + "loss": 0.0, + "step": 17300 + }, + { + "epoch": 86.62, + "learning_rate": 1.372820512820513e-06, + "loss": 0.0, + "step": 17325 + }, + { + "epoch": 86.75, + "learning_rate": 1.3600000000000001e-06, + "loss": 0.0, + "step": 17350 + }, + { + "epoch": 86.88, + "learning_rate": 1.3471794871794871e-06, + "loss": 0.0, + "step": 17375 + }, + { + "epoch": 87.0, + "learning_rate": 1.3343589743589744e-06, + "loss": 0.0, + "step": 17400 + }, + { + "epoch": 87.12, + "learning_rate": 1.3215384615384616e-06, + "loss": 0.0, + "step": 17425 + }, + { + "epoch": 87.25, + "learning_rate": 1.3087179487179489e-06, + "loss": 0.0, + "step": 17450 + }, + { + "epoch": 87.38, + "learning_rate": 1.2958974358974361e-06, + "loss": 0.0, + "step": 17475 + }, + { + "epoch": 87.5, + "learning_rate": 1.2830769230769232e-06, + "loss": 0.0, + "step": 17500 + }, + { + "epoch": 87.62, + "learning_rate": 1.2702564102564102e-06, + "loss": 0.0, + "step": 17525 + }, + { + "epoch": 87.75, + "learning_rate": 1.2574358974358975e-06, + "loss": 0.0, + "step": 17550 + }, + { + "epoch": 87.88, + "learning_rate": 1.2446153846153847e-06, + "loss": 0.0, + "step": 17575 + }, + { + "epoch": 88.0, + "learning_rate": 1.231794871794872e-06, + "loss": 0.0, + "step": 17600 + }, + { + "epoch": 88.12, + "learning_rate": 1.2189743589743592e-06, + "loss": 0.0, + "step": 17625 + }, + { + "epoch": 88.25, + "learning_rate": 1.2061538461538462e-06, + "loss": 0.0, + "step": 17650 + }, + { + "epoch": 88.38, + "learning_rate": 1.1933333333333335e-06, + "loss": 0.0, + "step": 17675 + }, + { + "epoch": 88.5, + "learning_rate": 1.1805128205128207e-06, + "loss": 0.0, + "step": 17700 + }, + { + "epoch": 88.62, + "learning_rate": 1.1676923076923078e-06, + "loss": 0.0, + "step": 17725 + }, + { + "epoch": 88.75, + "learning_rate": 1.154871794871795e-06, + "loss": 0.0, + "step": 17750 + }, + { + "epoch": 88.88, + "learning_rate": 1.1420512820512823e-06, + "loss": 0.0, + "step": 17775 + }, + { + "epoch": 89.0, + "learning_rate": 1.1292307692307693e-06, + "loss": 0.0, + "step": 17800 + }, + { + "epoch": 89.12, + "learning_rate": 1.1164102564102565e-06, + "loss": 0.0, + "step": 17825 + }, + { + "epoch": 89.25, + "learning_rate": 1.1035897435897438e-06, + "loss": 0.0, + "step": 17850 + }, + { + "epoch": 89.38, + "learning_rate": 1.0907692307692308e-06, + "loss": 0.0, + "step": 17875 + }, + { + "epoch": 89.5, + "learning_rate": 1.077948717948718e-06, + "loss": 0.0, + "step": 17900 + }, + { + "epoch": 89.62, + "learning_rate": 1.0651282051282053e-06, + "loss": 0.0, + "step": 17925 + }, + { + "epoch": 89.75, + "learning_rate": 1.0523076923076924e-06, + "loss": 0.0, + "step": 17950 + }, + { + "epoch": 89.88, + "learning_rate": 1.0394871794871796e-06, + "loss": 0.0, + "step": 17975 + }, + { + "epoch": 90.0, + "learning_rate": 1.0266666666666669e-06, + "loss": 0.0, + "step": 18000 + }, + { + "epoch": 90.0, + "eval_loss": 0.2893517315387726, + "eval_runtime": 1342.0978, + "eval_samples_per_second": 4.877, + "eval_steps_per_second": 0.305, + "eval_wer": 4.989307395143488, + "step": 18000 + }, + { + "epoch": 90.12, + "learning_rate": 1.013846153846154e-06, + "loss": 0.0, + "step": 18025 + }, + { + "epoch": 90.25, + "learning_rate": 1.0010256410256411e-06, + "loss": 0.0, + "step": 18050 + }, + { + "epoch": 90.38, + "learning_rate": 9.882051282051284e-07, + "loss": 0.0, + "step": 18075 + }, + { + "epoch": 90.5, + "learning_rate": 9.753846153846154e-07, + "loss": 0.0, + "step": 18100 + }, + { + "epoch": 90.62, + "learning_rate": 9.625641025641027e-07, + "loss": 0.0, + "step": 18125 + }, + { + "epoch": 90.75, + "learning_rate": 9.497435897435898e-07, + "loss": 0.0, + "step": 18150 + }, + { + "epoch": 90.88, + "learning_rate": 9.36923076923077e-07, + "loss": 0.0, + "step": 18175 + }, + { + "epoch": 91.0, + "learning_rate": 9.241025641025642e-07, + "loss": 0.0, + "step": 18200 + }, + { + "epoch": 91.12, + "learning_rate": 9.112820512820514e-07, + "loss": 0.0, + "step": 18225 + }, + { + "epoch": 91.25, + "learning_rate": 8.984615384615385e-07, + "loss": 0.0, + "step": 18250 + }, + { + "epoch": 91.38, + "learning_rate": 8.856410256410257e-07, + "loss": 0.0, + "step": 18275 + }, + { + "epoch": 91.5, + "learning_rate": 8.728205128205129e-07, + "loss": 0.0, + "step": 18300 + }, + { + "epoch": 91.62, + "learning_rate": 8.6e-07, + "loss": 0.0, + "step": 18325 + }, + { + "epoch": 91.75, + "learning_rate": 8.471794871794873e-07, + "loss": 0.0, + "step": 18350 + }, + { + "epoch": 91.88, + "learning_rate": 8.343589743589744e-07, + "loss": 0.0, + "step": 18375 + }, + { + "epoch": 92.0, + "learning_rate": 8.215384615384617e-07, + "loss": 0.0, + "step": 18400 + }, + { + "epoch": 92.12, + "learning_rate": 8.087179487179488e-07, + "loss": 0.0, + "step": 18425 + }, + { + "epoch": 92.25, + "learning_rate": 7.958974358974359e-07, + "loss": 0.0, + "step": 18450 + }, + { + "epoch": 92.38, + "learning_rate": 7.830769230769232e-07, + "loss": 0.0, + "step": 18475 + }, + { + "epoch": 92.5, + "learning_rate": 7.702564102564103e-07, + "loss": 0.0, + "step": 18500 + }, + { + "epoch": 92.62, + "learning_rate": 7.574358974358975e-07, + "loss": 0.0, + "step": 18525 + }, + { + "epoch": 92.75, + "learning_rate": 7.446153846153847e-07, + "loss": 0.0, + "step": 18550 + }, + { + "epoch": 92.88, + "learning_rate": 7.317948717948719e-07, + "loss": 0.0, + "step": 18575 + }, + { + "epoch": 93.0, + "learning_rate": 7.18974358974359e-07, + "loss": 0.0, + "step": 18600 + }, + { + "epoch": 93.12, + "learning_rate": 7.061538461538463e-07, + "loss": 0.0, + "step": 18625 + }, + { + "epoch": 93.25, + "learning_rate": 6.933333333333334e-07, + "loss": 0.0, + "step": 18650 + }, + { + "epoch": 93.38, + "learning_rate": 6.805128205128205e-07, + "loss": 0.0, + "step": 18675 + }, + { + "epoch": 93.5, + "learning_rate": 6.676923076923078e-07, + "loss": 0.0, + "step": 18700 + }, + { + "epoch": 93.62, + "learning_rate": 6.548717948717949e-07, + "loss": 0.0, + "step": 18725 + }, + { + "epoch": 93.75, + "learning_rate": 6.420512820512821e-07, + "loss": 0.0, + "step": 18750 + }, + { + "epoch": 93.88, + "learning_rate": 6.292307692307693e-07, + "loss": 0.0, + "step": 18775 + }, + { + "epoch": 94.0, + "learning_rate": 6.164102564102565e-07, + "loss": 0.0, + "step": 18800 + }, + { + "epoch": 94.12, + "learning_rate": 6.035897435897436e-07, + "loss": 0.0, + "step": 18825 + }, + { + "epoch": 94.25, + "learning_rate": 5.907692307692309e-07, + "loss": 0.0, + "step": 18850 + }, + { + "epoch": 94.38, + "learning_rate": 5.77948717948718e-07, + "loss": 0.0, + "step": 18875 + }, + { + "epoch": 94.5, + "learning_rate": 5.651282051282051e-07, + "loss": 0.0, + "step": 18900 + }, + { + "epoch": 94.62, + "learning_rate": 5.523076923076924e-07, + "loss": 0.0, + "step": 18925 + }, + { + "epoch": 94.75, + "learning_rate": 5.394871794871795e-07, + "loss": 0.0, + "step": 18950 + }, + { + "epoch": 94.88, + "learning_rate": 5.266666666666667e-07, + "loss": 0.0, + "step": 18975 + }, + { + "epoch": 95.0, + "learning_rate": 5.138461538461539e-07, + "loss": 0.0, + "step": 19000 + }, + { + "epoch": 95.0, + "eval_loss": 0.2926850914955139, + "eval_runtime": 1342.3132, + "eval_samples_per_second": 4.877, + "eval_steps_per_second": 0.305, + "eval_wer": 5.001379690949228, + "step": 19000 + }, + { + "epoch": 95.12, + "learning_rate": 5.010256410256411e-07, + "loss": 0.0, + "step": 19025 + }, + { + "epoch": 95.25, + "learning_rate": 4.882051282051282e-07, + "loss": 0.0, + "step": 19050 + }, + { + "epoch": 95.38, + "learning_rate": 4.753846153846154e-07, + "loss": 0.0, + "step": 19075 + }, + { + "epoch": 95.5, + "learning_rate": 4.6256410256410265e-07, + "loss": 0.0, + "step": 19100 + }, + { + "epoch": 95.62, + "learning_rate": 4.497435897435898e-07, + "loss": 0.0, + "step": 19125 + }, + { + "epoch": 95.75, + "learning_rate": 4.3692307692307693e-07, + "loss": 0.0, + "step": 19150 + }, + { + "epoch": 95.88, + "learning_rate": 4.241025641025642e-07, + "loss": 0.0, + "step": 19175 + }, + { + "epoch": 96.0, + "learning_rate": 4.112820512820513e-07, + "loss": 0.0, + "step": 19200 + }, + { + "epoch": 96.12, + "learning_rate": 3.9846153846153846e-07, + "loss": 0.0, + "step": 19225 + }, + { + "epoch": 96.25, + "learning_rate": 3.856410256410257e-07, + "loss": 0.0, + "step": 19250 + }, + { + "epoch": 96.38, + "learning_rate": 3.7282051282051285e-07, + "loss": 0.0, + "step": 19275 + }, + { + "epoch": 96.5, + "learning_rate": 3.6e-07, + "loss": 0.0, + "step": 19300 + }, + { + "epoch": 96.62, + "learning_rate": 3.4717948717948724e-07, + "loss": 0.0, + "step": 19325 + }, + { + "epoch": 96.75, + "learning_rate": 3.343589743589744e-07, + "loss": 0.0, + "step": 19350 + }, + { + "epoch": 96.88, + "learning_rate": 3.215384615384615e-07, + "loss": 0.0, + "step": 19375 + }, + { + "epoch": 97.0, + "learning_rate": 3.087179487179488e-07, + "loss": 0.0, + "step": 19400 + }, + { + "epoch": 97.12, + "learning_rate": 2.958974358974359e-07, + "loss": 0.0, + "step": 19425 + }, + { + "epoch": 97.25, + "learning_rate": 2.830769230769231e-07, + "loss": 0.0, + "step": 19450 + }, + { + "epoch": 97.38, + "learning_rate": 2.702564102564103e-07, + "loss": 0.0, + "step": 19475 + }, + { + "epoch": 97.5, + "learning_rate": 2.5743589743589745e-07, + "loss": 0.0, + "step": 19500 + }, + { + "epoch": 97.62, + "learning_rate": 2.4461538461538464e-07, + "loss": 0.0, + "step": 19525 + }, + { + "epoch": 97.75, + "learning_rate": 2.3179487179487184e-07, + "loss": 0.0, + "step": 19550 + }, + { + "epoch": 97.88, + "learning_rate": 2.1897435897435898e-07, + "loss": 0.0, + "step": 19575 + }, + { + "epoch": 98.0, + "learning_rate": 2.0615384615384618e-07, + "loss": 0.0, + "step": 19600 + }, + { + "epoch": 98.12, + "learning_rate": 1.9333333333333337e-07, + "loss": 0.0, + "step": 19625 + }, + { + "epoch": 98.25, + "learning_rate": 1.805128205128205e-07, + "loss": 0.0, + "step": 19650 + }, + { + "epoch": 98.38, + "learning_rate": 1.676923076923077e-07, + "loss": 0.0, + "step": 19675 + }, + { + "epoch": 98.5, + "learning_rate": 1.548717948717949e-07, + "loss": 0.0, + "step": 19700 + }, + { + "epoch": 98.62, + "learning_rate": 1.4205128205128207e-07, + "loss": 0.0, + "step": 19725 + }, + { + "epoch": 98.75, + "learning_rate": 1.2923076923076924e-07, + "loss": 0.0, + "step": 19750 + }, + { + "epoch": 98.88, + "learning_rate": 1.1641025641025642e-07, + "loss": 0.0, + "step": 19775 + }, + { + "epoch": 99.0, + "learning_rate": 1.035897435897436e-07, + "loss": 0.0, + "step": 19800 + }, + { + "epoch": 99.12, + "learning_rate": 9.076923076923077e-08, + "loss": 0.0, + "step": 19825 + }, + { + "epoch": 99.25, + "learning_rate": 7.794871794871795e-08, + "loss": 0.0, + "step": 19850 + }, + { + "epoch": 99.38, + "learning_rate": 6.512820512820514e-08, + "loss": 0.0, + "step": 19875 + }, + { + "epoch": 99.5, + "learning_rate": 5.230769230769231e-08, + "loss": 0.0, + "step": 19900 + }, + { + "epoch": 99.62, + "learning_rate": 3.948717948717949e-08, + "loss": 0.0, + "step": 19925 + }, + { + "epoch": 99.75, + "learning_rate": 2.6666666666666667e-08, + "loss": 0.0, + "step": 19950 + }, + { + "epoch": 99.88, + "learning_rate": 1.3846153846153846e-08, + "loss": 0.0, + "step": 19975 + }, + { + "epoch": 100.0, + "learning_rate": 1.0256410256410256e-09, + "loss": 0.0, + "step": 20000 + }, + { + "epoch": 100.0, + "eval_loss": 0.29404520988464355, + "eval_runtime": 1342.1418, + "eval_samples_per_second": 4.877, + "eval_steps_per_second": 0.305, + "eval_wer": 5.008278145695364, + "step": 20000 + }, + { + "epoch": 100.0, + "step": 20000, + "total_flos": 4.346759333412864e+21, + "train_loss": 0.006033917450643639, + "train_runtime": 138849.657, + "train_samples_per_second": 9.219, + "train_steps_per_second": 0.144 + } + ], + "logging_steps": 25, + "max_steps": 20000, + "num_input_tokens_seen": 0, + "num_train_epochs": 100, + "save_steps": 1000, + "total_flos": 4.346759333412864e+21, + "train_batch_size": 32, + "trial_name": null, + "trial_params": null +}