{ "best_metric": 5.495679032629054, "best_model_checkpoint": "./checkpoint-1000", "epoch": 70.54673721340389, "eval_steps": 1000, "global_step": 20000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.09, "learning_rate": 4.5999999999999995e-08, "loss": 0.929, "step": 25 }, { "epoch": 0.18, "learning_rate": 9.6e-08, "loss": 0.9271, "step": 50 }, { "epoch": 0.26, "learning_rate": 1.4599999999999998e-07, "loss": 0.876, "step": 75 }, { "epoch": 0.35, "learning_rate": 1.96e-07, "loss": 0.7695, "step": 100 }, { "epoch": 0.44, "learning_rate": 2.46e-07, "loss": 0.5441, "step": 125 }, { "epoch": 0.53, "learning_rate": 2.9599999999999995e-07, "loss": 0.2671, "step": 150 }, { "epoch": 0.62, "learning_rate": 3.4599999999999995e-07, "loss": 0.2277, "step": 175 }, { "epoch": 0.71, "learning_rate": 3.96e-07, "loss": 0.2032, "step": 200 }, { "epoch": 0.79, "learning_rate": 4.46e-07, "loss": 0.1906, "step": 225 }, { "epoch": 0.88, "learning_rate": 4.96e-07, "loss": 0.1852, "step": 250 }, { "epoch": 0.97, "learning_rate": 5.46e-07, "loss": 0.1693, "step": 275 }, { "epoch": 1.06, "learning_rate": 5.96e-07, "loss": 0.1596, "step": 300 }, { "epoch": 1.15, "learning_rate": 6.46e-07, "loss": 0.161, "step": 325 }, { "epoch": 1.23, "learning_rate": 6.959999999999999e-07, "loss": 0.1492, "step": 350 }, { "epoch": 1.32, "learning_rate": 7.459999999999999e-07, "loss": 0.1398, "step": 375 }, { "epoch": 1.41, "learning_rate": 7.96e-07, "loss": 0.1604, "step": 400 }, { "epoch": 1.5, "learning_rate": 8.459999999999999e-07, "loss": 0.1474, "step": 425 }, { "epoch": 1.59, "learning_rate": 8.96e-07, "loss": 0.1462, "step": 450 }, { "epoch": 1.68, "learning_rate": 9.459999999999999e-07, "loss": 0.1506, "step": 475 }, { "epoch": 1.76, "learning_rate": 9.959999999999999e-07, "loss": 0.1425, "step": 500 }, { "epoch": 1.85, "learning_rate": 9.98820512820513e-07, "loss": 0.1434, "step": 525 }, { "epoch": 1.94, "learning_rate": 9.975384615384614e-07, "loss": 0.1416, "step": 550 }, { "epoch": 2.03, "learning_rate": 9.962564102564102e-07, "loss": 0.1295, "step": 575 }, { "epoch": 2.12, "learning_rate": 9.949743589743589e-07, "loss": 0.1054, "step": 600 }, { "epoch": 2.2, "learning_rate": 9.936923076923076e-07, "loss": 0.1112, "step": 625 }, { "epoch": 2.29, "learning_rate": 9.924102564102563e-07, "loss": 0.1118, "step": 650 }, { "epoch": 2.38, "learning_rate": 9.91128205128205e-07, "loss": 0.1069, "step": 675 }, { "epoch": 2.47, "learning_rate": 9.898461538461538e-07, "loss": 0.1022, "step": 700 }, { "epoch": 2.56, "learning_rate": 9.885641025641025e-07, "loss": 0.1111, "step": 725 }, { "epoch": 2.65, "learning_rate": 9.872820512820512e-07, "loss": 0.1099, "step": 750 }, { "epoch": 2.73, "learning_rate": 9.86e-07, "loss": 0.0959, "step": 775 }, { "epoch": 2.82, "learning_rate": 9.847179487179487e-07, "loss": 0.1047, "step": 800 }, { "epoch": 2.91, "learning_rate": 9.834358974358974e-07, "loss": 0.1083, "step": 825 }, { "epoch": 3.0, "learning_rate": 9.821538461538461e-07, "loss": 0.1043, "step": 850 }, { "epoch": 3.09, "learning_rate": 9.808717948717949e-07, "loss": 0.0846, "step": 875 }, { "epoch": 3.17, "learning_rate": 9.795897435897436e-07, "loss": 0.0808, "step": 900 }, { "epoch": 3.26, "learning_rate": 9.783076923076923e-07, "loss": 0.0824, "step": 925 }, { "epoch": 3.35, "learning_rate": 9.77025641025641e-07, "loss": 0.0737, "step": 950 }, { "epoch": 3.44, "learning_rate": 9.757435897435898e-07, "loss": 0.0713, "step": 975 }, { "epoch": 3.53, "learning_rate": 9.744615384615385e-07, "loss": 0.077, "step": 1000 }, { "epoch": 3.53, "eval_loss": 0.16164931654930115, "eval_runtime": 2328.6601, "eval_samples_per_second": 3.896, "eval_steps_per_second": 0.243, "eval_wer": 5.495679032629054, "step": 1000 }, { "epoch": 3.62, "learning_rate": 9.731794871794872e-07, "loss": 0.0749, "step": 1025 }, { "epoch": 3.7, "learning_rate": 9.71897435897436e-07, "loss": 0.0787, "step": 1050 }, { "epoch": 3.79, "learning_rate": 9.706153846153847e-07, "loss": 0.0758, "step": 1075 }, { "epoch": 3.88, "learning_rate": 9.693333333333334e-07, "loss": 0.0805, "step": 1100 }, { "epoch": 3.97, "learning_rate": 9.680512820512821e-07, "loss": 0.0749, "step": 1125 }, { "epoch": 4.06, "learning_rate": 9.667692307692306e-07, "loss": 0.0647, "step": 1150 }, { "epoch": 4.14, "learning_rate": 9.654871794871793e-07, "loss": 0.0532, "step": 1175 }, { "epoch": 4.23, "learning_rate": 9.64205128205128e-07, "loss": 0.054, "step": 1200 }, { "epoch": 4.32, "learning_rate": 9.629230769230768e-07, "loss": 0.0556, "step": 1225 }, { "epoch": 4.41, "learning_rate": 9.616410256410255e-07, "loss": 0.0562, "step": 1250 }, { "epoch": 4.5, "learning_rate": 9.603589743589742e-07, "loss": 0.0559, "step": 1275 }, { "epoch": 4.59, "learning_rate": 9.59076923076923e-07, "loss": 0.0566, "step": 1300 }, { "epoch": 4.67, "learning_rate": 9.577948717948717e-07, "loss": 0.0588, "step": 1325 }, { "epoch": 4.76, "learning_rate": 9.565128205128204e-07, "loss": 0.0555, "step": 1350 }, { "epoch": 4.85, "learning_rate": 9.552307692307691e-07, "loss": 0.0555, "step": 1375 }, { "epoch": 4.94, "learning_rate": 9.539487179487179e-07, "loss": 0.0568, "step": 1400 }, { "epoch": 5.03, "learning_rate": 9.526666666666666e-07, "loss": 0.0518, "step": 1425 }, { "epoch": 5.11, "learning_rate": 9.513846153846153e-07, "loss": 0.0386, "step": 1450 }, { "epoch": 5.2, "learning_rate": 9.50102564102564e-07, "loss": 0.0386, "step": 1475 }, { "epoch": 5.29, "learning_rate": 9.488205128205128e-07, "loss": 0.0369, "step": 1500 }, { "epoch": 5.38, "learning_rate": 9.475384615384615e-07, "loss": 0.0347, "step": 1525 }, { "epoch": 5.47, "learning_rate": 9.462564102564102e-07, "loss": 0.0338, "step": 1550 }, { "epoch": 5.56, "learning_rate": 9.449743589743589e-07, "loss": 0.0417, "step": 1575 }, { "epoch": 5.64, "learning_rate": 9.436923076923077e-07, "loss": 0.0397, "step": 1600 }, { "epoch": 5.73, "learning_rate": 9.424102564102564e-07, "loss": 0.0358, "step": 1625 }, { "epoch": 5.82, "learning_rate": 9.411282051282051e-07, "loss": 0.0368, "step": 1650 }, { "epoch": 5.91, "learning_rate": 9.398461538461538e-07, "loss": 0.0372, "step": 1675 }, { "epoch": 6.0, "learning_rate": 9.385641025641026e-07, "loss": 0.0394, "step": 1700 }, { "epoch": 6.08, "learning_rate": 9.372820512820513e-07, "loss": 0.0209, "step": 1725 }, { "epoch": 6.17, "learning_rate": 9.36e-07, "loss": 0.0216, "step": 1750 }, { "epoch": 6.26, "learning_rate": 9.347179487179486e-07, "loss": 0.0213, "step": 1775 }, { "epoch": 6.35, "learning_rate": 9.334358974358974e-07, "loss": 0.0212, "step": 1800 }, { "epoch": 6.44, "learning_rate": 9.321538461538461e-07, "loss": 0.023, "step": 1825 }, { "epoch": 6.53, "learning_rate": 9.308717948717948e-07, "loss": 0.0213, "step": 1850 }, { "epoch": 6.61, "learning_rate": 9.295897435897435e-07, "loss": 0.0216, "step": 1875 }, { "epoch": 6.7, "learning_rate": 9.283076923076923e-07, "loss": 0.0227, "step": 1900 }, { "epoch": 6.79, "learning_rate": 9.27025641025641e-07, "loss": 0.0243, "step": 1925 }, { "epoch": 6.88, "learning_rate": 9.257435897435897e-07, "loss": 0.0191, "step": 1950 }, { "epoch": 6.97, "learning_rate": 9.244615384615384e-07, "loss": 0.0242, "step": 1975 }, { "epoch": 7.05, "learning_rate": 9.231794871794872e-07, "loss": 0.0155, "step": 2000 }, { "epoch": 7.05, "eval_loss": 0.2548653185367584, "eval_runtime": 2250.6856, "eval_samples_per_second": 4.031, "eval_steps_per_second": 0.252, "eval_wer": 6.195577169519929, "step": 2000 }, { "epoch": 7.14, "learning_rate": 9.218974358974359e-07, "loss": 0.0141, "step": 2025 }, { "epoch": 7.23, "learning_rate": 9.206153846153846e-07, "loss": 0.0121, "step": 2050 }, { "epoch": 7.32, "learning_rate": 9.193333333333333e-07, "loss": 0.0132, "step": 2075 }, { "epoch": 7.41, "learning_rate": 9.180512820512821e-07, "loss": 0.0113, "step": 2100 }, { "epoch": 7.5, "learning_rate": 9.167692307692308e-07, "loss": 0.012, "step": 2125 }, { "epoch": 7.58, "learning_rate": 9.154871794871795e-07, "loss": 0.0115, "step": 2150 }, { "epoch": 7.67, "learning_rate": 9.142051282051282e-07, "loss": 0.0117, "step": 2175 }, { "epoch": 7.76, "learning_rate": 9.12923076923077e-07, "loss": 0.012, "step": 2200 }, { "epoch": 7.85, "learning_rate": 9.116410256410257e-07, "loss": 0.0143, "step": 2225 }, { "epoch": 7.94, "learning_rate": 9.103589743589743e-07, "loss": 0.0118, "step": 2250 }, { "epoch": 8.02, "learning_rate": 9.09076923076923e-07, "loss": 0.0118, "step": 2275 }, { "epoch": 8.11, "learning_rate": 9.077948717948718e-07, "loss": 0.0079, "step": 2300 }, { "epoch": 8.2, "learning_rate": 9.065128205128205e-07, "loss": 0.0079, "step": 2325 }, { "epoch": 8.29, "learning_rate": 9.052307692307692e-07, "loss": 0.0073, "step": 2350 }, { "epoch": 8.38, "learning_rate": 9.039487179487178e-07, "loss": 0.0072, "step": 2375 }, { "epoch": 8.47, "learning_rate": 9.026666666666665e-07, "loss": 0.0074, "step": 2400 }, { "epoch": 8.55, "learning_rate": 9.013846153846153e-07, "loss": 0.0071, "step": 2425 }, { "epoch": 8.64, "learning_rate": 9.00102564102564e-07, "loss": 0.0076, "step": 2450 }, { "epoch": 8.73, "learning_rate": 8.988205128205127e-07, "loss": 0.0087, "step": 2475 }, { "epoch": 8.82, "learning_rate": 8.975384615384614e-07, "loss": 0.0068, "step": 2500 }, { "epoch": 8.91, "learning_rate": 8.962564102564102e-07, "loss": 0.0076, "step": 2525 }, { "epoch": 8.99, "learning_rate": 8.949743589743589e-07, "loss": 0.0083, "step": 2550 }, { "epoch": 9.08, "learning_rate": 8.936923076923076e-07, "loss": 0.0054, "step": 2575 }, { "epoch": 9.17, "learning_rate": 8.924102564102563e-07, "loss": 0.0047, "step": 2600 }, { "epoch": 9.26, "learning_rate": 8.911282051282051e-07, "loss": 0.0048, "step": 2625 }, { "epoch": 9.35, "learning_rate": 8.898461538461538e-07, "loss": 0.0059, "step": 2650 }, { "epoch": 9.44, "learning_rate": 8.885641025641025e-07, "loss": 0.005, "step": 2675 }, { "epoch": 9.52, "learning_rate": 8.872820512820512e-07, "loss": 0.0058, "step": 2700 }, { "epoch": 9.61, "learning_rate": 8.86e-07, "loss": 0.0049, "step": 2725 }, { "epoch": 9.7, "learning_rate": 8.847179487179487e-07, "loss": 0.0053, "step": 2750 }, { "epoch": 9.79, "learning_rate": 8.834358974358974e-07, "loss": 0.005, "step": 2775 }, { "epoch": 9.88, "learning_rate": 8.821538461538461e-07, "loss": 0.0051, "step": 2800 }, { "epoch": 9.96, "learning_rate": 8.808717948717949e-07, "loss": 0.0049, "step": 2825 }, { "epoch": 10.05, "learning_rate": 8.795897435897436e-07, "loss": 0.0049, "step": 2850 }, { "epoch": 10.14, "learning_rate": 8.783076923076923e-07, "loss": 0.0034, "step": 2875 }, { "epoch": 10.23, "learning_rate": 8.77025641025641e-07, "loss": 0.0041, "step": 2900 }, { "epoch": 10.32, "learning_rate": 8.757435897435898e-07, "loss": 0.004, "step": 2925 }, { "epoch": 10.41, "learning_rate": 8.744615384615385e-07, "loss": 0.0037, "step": 2950 }, { "epoch": 10.49, "learning_rate": 8.731794871794872e-07, "loss": 0.004, "step": 2975 }, { "epoch": 10.58, "learning_rate": 8.718974358974358e-07, "loss": 0.0045, "step": 3000 }, { "epoch": 10.58, "eval_loss": 0.31223657727241516, "eval_runtime": 2244.1267, "eval_samples_per_second": 4.043, "eval_steps_per_second": 0.253, "eval_wer": 5.9261328163506715, "step": 3000 }, { "epoch": 10.67, "learning_rate": 8.706153846153846e-07, "loss": 0.0041, "step": 3025 }, { "epoch": 10.76, "learning_rate": 8.693333333333333e-07, "loss": 0.0037, "step": 3050 }, { "epoch": 10.85, "learning_rate": 8.68051282051282e-07, "loss": 0.0041, "step": 3075 }, { "epoch": 10.93, "learning_rate": 8.667692307692307e-07, "loss": 0.0038, "step": 3100 }, { "epoch": 11.02, "learning_rate": 8.654871794871795e-07, "loss": 0.0042, "step": 3125 }, { "epoch": 11.11, "learning_rate": 8.642051282051282e-07, "loss": 0.0033, "step": 3150 }, { "epoch": 11.2, "learning_rate": 8.629230769230769e-07, "loss": 0.0042, "step": 3175 }, { "epoch": 11.29, "learning_rate": 8.616410256410256e-07, "loss": 0.0032, "step": 3200 }, { "epoch": 11.38, "learning_rate": 8.603589743589744e-07, "loss": 0.0033, "step": 3225 }, { "epoch": 11.46, "learning_rate": 8.590769230769231e-07, "loss": 0.0033, "step": 3250 }, { "epoch": 11.55, "learning_rate": 8.577948717948718e-07, "loss": 0.0035, "step": 3275 }, { "epoch": 11.64, "learning_rate": 8.565128205128204e-07, "loss": 0.0034, "step": 3300 }, { "epoch": 11.73, "learning_rate": 8.552307692307691e-07, "loss": 0.0028, "step": 3325 }, { "epoch": 11.82, "learning_rate": 8.539487179487179e-07, "loss": 0.004, "step": 3350 }, { "epoch": 11.9, "learning_rate": 8.526666666666666e-07, "loss": 0.0039, "step": 3375 }, { "epoch": 11.99, "learning_rate": 8.513846153846153e-07, "loss": 0.0032, "step": 3400 }, { "epoch": 12.08, "learning_rate": 8.50102564102564e-07, "loss": 0.0026, "step": 3425 }, { "epoch": 12.17, "learning_rate": 8.488205128205128e-07, "loss": 0.003, "step": 3450 }, { "epoch": 12.26, "learning_rate": 8.475384615384615e-07, "loss": 0.004, "step": 3475 }, { "epoch": 12.35, "learning_rate": 8.462564102564102e-07, "loss": 0.0035, "step": 3500 }, { "epoch": 12.43, "learning_rate": 8.44974358974359e-07, "loss": 0.0027, "step": 3525 }, { "epoch": 12.52, "learning_rate": 8.436923076923077e-07, "loss": 0.0027, "step": 3550 }, { "epoch": 12.61, "learning_rate": 8.424102564102564e-07, "loss": 0.0031, "step": 3575 }, { "epoch": 12.7, "learning_rate": 8.41128205128205e-07, "loss": 0.0024, "step": 3600 }, { "epoch": 12.79, "learning_rate": 8.398461538461537e-07, "loss": 0.0025, "step": 3625 }, { "epoch": 12.87, "learning_rate": 8.385641025641025e-07, "loss": 0.0034, "step": 3650 }, { "epoch": 12.96, "learning_rate": 8.372820512820512e-07, "loss": 0.0025, "step": 3675 }, { "epoch": 13.05, "learning_rate": 8.359999999999999e-07, "loss": 0.0029, "step": 3700 }, { "epoch": 13.14, "learning_rate": 8.347179487179486e-07, "loss": 0.0022, "step": 3725 }, { "epoch": 13.23, "learning_rate": 8.334358974358974e-07, "loss": 0.0029, "step": 3750 }, { "epoch": 13.32, "learning_rate": 8.321538461538461e-07, "loss": 0.0033, "step": 3775 }, { "epoch": 13.4, "learning_rate": 8.308717948717948e-07, "loss": 0.0021, "step": 3800 }, { "epoch": 13.49, "learning_rate": 8.295897435897435e-07, "loss": 0.002, "step": 3825 }, { "epoch": 13.58, "learning_rate": 8.283076923076923e-07, "loss": 0.002, "step": 3850 }, { "epoch": 13.67, "learning_rate": 8.27025641025641e-07, "loss": 0.002, "step": 3875 }, { "epoch": 13.76, "learning_rate": 8.257435897435897e-07, "loss": 0.0023, "step": 3900 }, { "epoch": 13.84, "learning_rate": 8.244615384615384e-07, "loss": 0.0025, "step": 3925 }, { "epoch": 13.93, "learning_rate": 8.231794871794872e-07, "loss": 0.0026, "step": 3950 }, { "epoch": 14.02, "learning_rate": 8.218974358974359e-07, "loss": 0.0027, "step": 3975 }, { "epoch": 14.11, "learning_rate": 8.206153846153846e-07, "loss": 0.0017, "step": 4000 }, { "epoch": 14.11, "eval_loss": 0.3317270576953888, "eval_runtime": 2242.9737, "eval_samples_per_second": 4.045, "eval_steps_per_second": 0.253, "eval_wer": 6.009923438372819, "step": 4000 }, { "epoch": 14.2, "learning_rate": 8.193333333333333e-07, "loss": 0.0034, "step": 4025 }, { "epoch": 14.29, "learning_rate": 8.180512820512821e-07, "loss": 0.0016, "step": 4050 }, { "epoch": 14.37, "learning_rate": 8.167692307692308e-07, "loss": 0.0017, "step": 4075 }, { "epoch": 14.46, "learning_rate": 8.154871794871795e-07, "loss": 0.0017, "step": 4100 }, { "epoch": 14.55, "learning_rate": 8.142051282051282e-07, "loss": 0.0025, "step": 4125 }, { "epoch": 14.64, "learning_rate": 8.12923076923077e-07, "loss": 0.0023, "step": 4150 }, { "epoch": 14.73, "learning_rate": 8.116410256410257e-07, "loss": 0.0017, "step": 4175 }, { "epoch": 14.81, "learning_rate": 8.103589743589743e-07, "loss": 0.0028, "step": 4200 }, { "epoch": 14.9, "learning_rate": 8.09076923076923e-07, "loss": 0.0022, "step": 4225 }, { "epoch": 14.99, "learning_rate": 8.077948717948718e-07, "loss": 0.002, "step": 4250 }, { "epoch": 15.08, "learning_rate": 8.065128205128205e-07, "loss": 0.0018, "step": 4275 }, { "epoch": 15.17, "learning_rate": 8.052307692307692e-07, "loss": 0.0018, "step": 4300 }, { "epoch": 15.26, "learning_rate": 8.039487179487179e-07, "loss": 0.0017, "step": 4325 }, { "epoch": 15.34, "learning_rate": 8.026666666666667e-07, "loss": 0.0016, "step": 4350 }, { "epoch": 15.43, "learning_rate": 8.013846153846153e-07, "loss": 0.0022, "step": 4375 }, { "epoch": 15.52, "learning_rate": 8.00102564102564e-07, "loss": 0.0015, "step": 4400 }, { "epoch": 15.61, "learning_rate": 7.988205128205127e-07, "loss": 0.0023, "step": 4425 }, { "epoch": 15.7, "learning_rate": 7.975384615384614e-07, "loss": 0.0017, "step": 4450 }, { "epoch": 15.78, "learning_rate": 7.962564102564102e-07, "loss": 0.0017, "step": 4475 }, { "epoch": 15.87, "learning_rate": 7.949743589743589e-07, "loss": 0.0023, "step": 4500 }, { "epoch": 15.96, "learning_rate": 7.936923076923076e-07, "loss": 0.0033, "step": 4525 }, { "epoch": 16.05, "learning_rate": 7.924102564102563e-07, "loss": 0.0014, "step": 4550 }, { "epoch": 16.14, "learning_rate": 7.911282051282051e-07, "loss": 0.0019, "step": 4575 }, { "epoch": 16.23, "learning_rate": 7.898461538461538e-07, "loss": 0.0014, "step": 4600 }, { "epoch": 16.31, "learning_rate": 7.885641025641025e-07, "loss": 0.0013, "step": 4625 }, { "epoch": 16.4, "learning_rate": 7.872820512820512e-07, "loss": 0.0023, "step": 4650 }, { "epoch": 16.49, "learning_rate": 7.86e-07, "loss": 0.0017, "step": 4675 }, { "epoch": 16.58, "learning_rate": 7.847179487179487e-07, "loss": 0.0013, "step": 4700 }, { "epoch": 16.67, "learning_rate": 7.834358974358974e-07, "loss": 0.0014, "step": 4725 }, { "epoch": 16.75, "learning_rate": 7.821538461538461e-07, "loss": 0.0017, "step": 4750 }, { "epoch": 16.84, "learning_rate": 7.808717948717949e-07, "loss": 0.0014, "step": 4775 }, { "epoch": 16.93, "learning_rate": 7.795897435897436e-07, "loss": 0.0013, "step": 4800 }, { "epoch": 17.02, "learning_rate": 7.783076923076922e-07, "loss": 0.0013, "step": 4825 }, { "epoch": 17.11, "learning_rate": 7.770256410256409e-07, "loss": 0.0017, "step": 4850 }, { "epoch": 17.2, "learning_rate": 7.757435897435897e-07, "loss": 0.0012, "step": 4875 }, { "epoch": 17.28, "learning_rate": 7.744615384615384e-07, "loss": 0.0013, "step": 4900 }, { "epoch": 17.37, "learning_rate": 7.731794871794871e-07, "loss": 0.0012, "step": 4925 }, { "epoch": 17.46, "learning_rate": 7.718974358974358e-07, "loss": 0.0017, "step": 4950 }, { "epoch": 17.55, "learning_rate": 7.706153846153846e-07, "loss": 0.0014, "step": 4975 }, { "epoch": 17.64, "learning_rate": 7.693333333333333e-07, "loss": 0.0018, "step": 5000 }, { "epoch": 17.64, "eval_loss": 0.3604351580142975, "eval_runtime": 2242.6858, "eval_samples_per_second": 4.045, "eval_steps_per_second": 0.253, "eval_wer": 6.009923438372819, "step": 5000 }, { "epoch": 17.72, "learning_rate": 7.68051282051282e-07, "loss": 0.0017, "step": 5025 }, { "epoch": 17.81, "learning_rate": 7.667692307692307e-07, "loss": 0.0025, "step": 5050 }, { "epoch": 17.9, "learning_rate": 7.654871794871795e-07, "loss": 0.0028, "step": 5075 }, { "epoch": 17.99, "learning_rate": 7.642051282051282e-07, "loss": 0.0031, "step": 5100 }, { "epoch": 18.08, "learning_rate": 7.629230769230769e-07, "loss": 0.0025, "step": 5125 }, { "epoch": 18.17, "learning_rate": 7.616410256410256e-07, "loss": 0.0015, "step": 5150 }, { "epoch": 18.25, "learning_rate": 7.603589743589744e-07, "loss": 0.0017, "step": 5175 }, { "epoch": 18.34, "learning_rate": 7.590769230769231e-07, "loss": 0.0016, "step": 5200 }, { "epoch": 18.43, "learning_rate": 7.577948717948718e-07, "loss": 0.0024, "step": 5225 }, { "epoch": 18.52, "learning_rate": 7.565128205128205e-07, "loss": 0.0017, "step": 5250 }, { "epoch": 18.61, "learning_rate": 7.552307692307693e-07, "loss": 0.0016, "step": 5275 }, { "epoch": 18.69, "learning_rate": 7.54e-07, "loss": 0.0026, "step": 5300 }, { "epoch": 18.78, "learning_rate": 7.527179487179487e-07, "loss": 0.002, "step": 5325 }, { "epoch": 18.87, "learning_rate": 7.514358974358975e-07, "loss": 0.002, "step": 5350 }, { "epoch": 18.96, "learning_rate": 7.501538461538462e-07, "loss": 0.0019, "step": 5375 }, { "epoch": 19.05, "learning_rate": 7.488717948717949e-07, "loss": 0.0024, "step": 5400 }, { "epoch": 19.14, "learning_rate": 7.475897435897436e-07, "loss": 0.0018, "step": 5425 }, { "epoch": 19.22, "learning_rate": 7.463076923076924e-07, "loss": 0.0011, "step": 5450 }, { "epoch": 19.31, "learning_rate": 7.450256410256411e-07, "loss": 0.0012, "step": 5475 }, { "epoch": 19.4, "learning_rate": 7.437435897435898e-07, "loss": 0.001, "step": 5500 }, { "epoch": 19.49, "learning_rate": 7.424615384615385e-07, "loss": 0.0013, "step": 5525 }, { "epoch": 19.58, "learning_rate": 7.411794871794871e-07, "loss": 0.0017, "step": 5550 }, { "epoch": 19.66, "learning_rate": 7.398974358974359e-07, "loss": 0.0027, "step": 5575 }, { "epoch": 19.75, "learning_rate": 7.386153846153845e-07, "loss": 0.0012, "step": 5600 }, { "epoch": 19.84, "learning_rate": 7.373333333333332e-07, "loss": 0.001, "step": 5625 }, { "epoch": 19.93, "learning_rate": 7.360512820512819e-07, "loss": 0.0011, "step": 5650 }, { "epoch": 20.02, "learning_rate": 7.347692307692307e-07, "loss": 0.0012, "step": 5675 }, { "epoch": 20.11, "learning_rate": 7.334871794871794e-07, "loss": 0.001, "step": 5700 }, { "epoch": 20.19, "learning_rate": 7.322051282051281e-07, "loss": 0.0011, "step": 5725 }, { "epoch": 20.28, "learning_rate": 7.309230769230768e-07, "loss": 0.0014, "step": 5750 }, { "epoch": 20.37, "learning_rate": 7.296410256410256e-07, "loss": 0.0013, "step": 5775 }, { "epoch": 20.46, "learning_rate": 7.283589743589743e-07, "loss": 0.001, "step": 5800 }, { "epoch": 20.55, "learning_rate": 7.27076923076923e-07, "loss": 0.0009, "step": 5825 }, { "epoch": 20.63, "learning_rate": 7.257948717948717e-07, "loss": 0.0009, "step": 5850 }, { "epoch": 20.72, "learning_rate": 7.245128205128205e-07, "loss": 0.0009, "step": 5875 }, { "epoch": 20.81, "learning_rate": 7.232307692307692e-07, "loss": 0.0009, "step": 5900 }, { "epoch": 20.9, "learning_rate": 7.219487179487179e-07, "loss": 0.0011, "step": 5925 }, { "epoch": 20.99, "learning_rate": 7.206666666666666e-07, "loss": 0.0008, "step": 5950 }, { "epoch": 21.08, "learning_rate": 7.193846153846154e-07, "loss": 0.0008, "step": 5975 }, { "epoch": 21.16, "learning_rate": 7.181025641025641e-07, "loss": 0.0009, "step": 6000 }, { "epoch": 21.16, "eval_loss": 0.37790602445602417, "eval_runtime": 2245.9507, "eval_samples_per_second": 4.039, "eval_steps_per_second": 0.252, "eval_wer": 6.1791476357900965, "step": 6000 }, { "epoch": 21.25, "learning_rate": 7.168205128205128e-07, "loss": 0.0007, "step": 6025 }, { "epoch": 21.34, "learning_rate": 7.155384615384615e-07, "loss": 0.0021, "step": 6050 }, { "epoch": 21.43, "learning_rate": 7.142564102564103e-07, "loss": 0.0007, "step": 6075 }, { "epoch": 21.52, "learning_rate": 7.12974358974359e-07, "loss": 0.0008, "step": 6100 }, { "epoch": 21.6, "learning_rate": 7.116923076923077e-07, "loss": 0.0007, "step": 6125 }, { "epoch": 21.69, "learning_rate": 7.104102564102564e-07, "loss": 0.0015, "step": 6150 }, { "epoch": 21.78, "learning_rate": 7.091282051282052e-07, "loss": 0.0009, "step": 6175 }, { "epoch": 21.87, "learning_rate": 7.078461538461538e-07, "loss": 0.001, "step": 6200 }, { "epoch": 21.96, "learning_rate": 7.065641025641025e-07, "loss": 0.0009, "step": 6225 }, { "epoch": 22.05, "learning_rate": 7.052820512820512e-07, "loss": 0.001, "step": 6250 }, { "epoch": 22.13, "learning_rate": 7.04e-07, "loss": 0.0007, "step": 6275 }, { "epoch": 22.22, "learning_rate": 7.027179487179487e-07, "loss": 0.0007, "step": 6300 }, { "epoch": 22.31, "learning_rate": 7.014358974358974e-07, "loss": 0.0007, "step": 6325 }, { "epoch": 22.4, "learning_rate": 7.001538461538461e-07, "loss": 0.002, "step": 6350 }, { "epoch": 22.49, "learning_rate": 6.988717948717949e-07, "loss": 0.001, "step": 6375 }, { "epoch": 22.57, "learning_rate": 6.975897435897436e-07, "loss": 0.0008, "step": 6400 }, { "epoch": 22.66, "learning_rate": 6.963076923076923e-07, "loss": 0.0009, "step": 6425 }, { "epoch": 22.75, "learning_rate": 6.95025641025641e-07, "loss": 0.0009, "step": 6450 }, { "epoch": 22.84, "learning_rate": 6.937435897435897e-07, "loss": 0.0016, "step": 6475 }, { "epoch": 22.93, "learning_rate": 6.924615384615385e-07, "loss": 0.0016, "step": 6500 }, { "epoch": 23.02, "learning_rate": 6.911794871794872e-07, "loss": 0.001, "step": 6525 }, { "epoch": 23.1, "learning_rate": 6.898974358974359e-07, "loss": 0.0007, "step": 6550 }, { "epoch": 23.19, "learning_rate": 6.886153846153846e-07, "loss": 0.0007, "step": 6575 }, { "epoch": 23.28, "learning_rate": 6.873333333333334e-07, "loss": 0.0019, "step": 6600 }, { "epoch": 23.37, "learning_rate": 6.86051282051282e-07, "loss": 0.0009, "step": 6625 }, { "epoch": 23.46, "learning_rate": 6.847692307692307e-07, "loss": 0.001, "step": 6650 }, { "epoch": 23.54, "learning_rate": 6.834871794871794e-07, "loss": 0.0012, "step": 6675 }, { "epoch": 23.63, "learning_rate": 6.822051282051282e-07, "loss": 0.0008, "step": 6700 }, { "epoch": 23.72, "learning_rate": 6.809230769230769e-07, "loss": 0.0007, "step": 6725 }, { "epoch": 23.81, "learning_rate": 6.796410256410256e-07, "loss": 0.0014, "step": 6750 }, { "epoch": 23.9, "learning_rate": 6.783589743589743e-07, "loss": 0.001, "step": 6775 }, { "epoch": 23.99, "learning_rate": 6.770769230769231e-07, "loss": 0.0019, "step": 6800 }, { "epoch": 24.07, "learning_rate": 6.757948717948717e-07, "loss": 0.0008, "step": 6825 }, { "epoch": 24.16, "learning_rate": 6.745128205128204e-07, "loss": 0.0008, "step": 6850 }, { "epoch": 24.25, "learning_rate": 6.732307692307691e-07, "loss": 0.0008, "step": 6875 }, { "epoch": 24.34, "learning_rate": 6.719487179487179e-07, "loss": 0.0009, "step": 6900 }, { "epoch": 24.43, "learning_rate": 6.706666666666666e-07, "loss": 0.0011, "step": 6925 }, { "epoch": 24.51, "learning_rate": 6.693846153846153e-07, "loss": 0.0017, "step": 6950 }, { "epoch": 24.6, "learning_rate": 6.68102564102564e-07, "loss": 0.0015, "step": 6975 }, { "epoch": 24.69, "learning_rate": 6.668205128205128e-07, "loss": 0.0012, "step": 7000 }, { "epoch": 24.69, "eval_loss": 0.34700751304626465, "eval_runtime": 2257.836, "eval_samples_per_second": 4.018, "eval_steps_per_second": 0.251, "eval_wer": 6.006637531626852, "step": 7000 }, { "epoch": 24.78, "learning_rate": 6.655384615384615e-07, "loss": 0.0012, "step": 7025 }, { "epoch": 24.87, "learning_rate": 6.642564102564102e-07, "loss": 0.0016, "step": 7050 }, { "epoch": 24.96, "learning_rate": 6.629743589743589e-07, "loss": 0.0011, "step": 7075 }, { "epoch": 25.04, "learning_rate": 6.616923076923077e-07, "loss": 0.0012, "step": 7100 }, { "epoch": 25.13, "learning_rate": 6.604102564102564e-07, "loss": 0.001, "step": 7125 }, { "epoch": 25.22, "learning_rate": 6.591282051282051e-07, "loss": 0.001, "step": 7150 }, { "epoch": 25.31, "learning_rate": 6.578461538461538e-07, "loss": 0.0006, "step": 7175 }, { "epoch": 25.4, "learning_rate": 6.565641025641026e-07, "loss": 0.0008, "step": 7200 }, { "epoch": 25.49, "learning_rate": 6.552820512820513e-07, "loss": 0.0007, "step": 7225 }, { "epoch": 25.57, "learning_rate": 6.54e-07, "loss": 0.0006, "step": 7250 }, { "epoch": 25.66, "learning_rate": 6.527179487179487e-07, "loss": 0.0009, "step": 7275 }, { "epoch": 25.75, "learning_rate": 6.514358974358975e-07, "loss": 0.0006, "step": 7300 }, { "epoch": 25.84, "learning_rate": 6.501538461538462e-07, "loss": 0.0007, "step": 7325 }, { "epoch": 25.93, "learning_rate": 6.488717948717949e-07, "loss": 0.0008, "step": 7350 }, { "epoch": 26.01, "learning_rate": 6.475897435897436e-07, "loss": 0.001, "step": 7375 }, { "epoch": 26.1, "learning_rate": 6.463076923076924e-07, "loss": 0.0007, "step": 7400 }, { "epoch": 26.19, "learning_rate": 6.45025641025641e-07, "loss": 0.0011, "step": 7425 }, { "epoch": 26.28, "learning_rate": 6.437435897435897e-07, "loss": 0.0007, "step": 7450 }, { "epoch": 26.37, "learning_rate": 6.424615384615384e-07, "loss": 0.0011, "step": 7475 }, { "epoch": 26.46, "learning_rate": 6.411794871794871e-07, "loss": 0.0009, "step": 7500 }, { "epoch": 26.54, "learning_rate": 6.398974358974359e-07, "loss": 0.0008, "step": 7525 }, { "epoch": 26.63, "learning_rate": 6.386153846153846e-07, "loss": 0.0008, "step": 7550 }, { "epoch": 26.72, "learning_rate": 6.373333333333333e-07, "loss": 0.0006, "step": 7575 }, { "epoch": 26.81, "learning_rate": 6.36051282051282e-07, "loss": 0.001, "step": 7600 }, { "epoch": 26.9, "learning_rate": 6.347692307692308e-07, "loss": 0.0006, "step": 7625 }, { "epoch": 26.98, "learning_rate": 6.334871794871795e-07, "loss": 0.0011, "step": 7650 }, { "epoch": 27.07, "learning_rate": 6.322051282051281e-07, "loss": 0.0005, "step": 7675 }, { "epoch": 27.16, "learning_rate": 6.309230769230768e-07, "loss": 0.0005, "step": 7700 }, { "epoch": 27.25, "learning_rate": 6.296410256410256e-07, "loss": 0.0006, "step": 7725 }, { "epoch": 27.34, "learning_rate": 6.283589743589743e-07, "loss": 0.0007, "step": 7750 }, { "epoch": 27.43, "learning_rate": 6.27076923076923e-07, "loss": 0.0012, "step": 7775 }, { "epoch": 27.51, "learning_rate": 6.257948717948717e-07, "loss": 0.0016, "step": 7800 }, { "epoch": 27.6, "learning_rate": 6.245128205128205e-07, "loss": 0.0005, "step": 7825 }, { "epoch": 27.69, "learning_rate": 6.232307692307692e-07, "loss": 0.0008, "step": 7850 }, { "epoch": 27.78, "learning_rate": 6.219487179487179e-07, "loss": 0.0007, "step": 7875 }, { "epoch": 27.87, "learning_rate": 6.206666666666666e-07, "loss": 0.0005, "step": 7900 }, { "epoch": 27.95, "learning_rate": 6.193846153846154e-07, "loss": 0.0007, "step": 7925 }, { "epoch": 28.04, "learning_rate": 6.181025641025641e-07, "loss": 0.0006, "step": 7950 }, { "epoch": 28.13, "learning_rate": 6.168205128205128e-07, "loss": 0.0012, "step": 7975 }, { "epoch": 28.22, "learning_rate": 6.155384615384615e-07, "loss": 0.0013, "step": 8000 }, { "epoch": 28.22, "eval_loss": 0.38379231095314026, "eval_runtime": 2248.9853, "eval_samples_per_second": 4.034, "eval_steps_per_second": 0.252, "eval_wer": 6.147931521703414, "step": 8000 }, { "epoch": 28.31, "learning_rate": 6.142564102564102e-07, "loss": 0.0007, "step": 8025 }, { "epoch": 28.4, "learning_rate": 6.129743589743589e-07, "loss": 0.001, "step": 8050 }, { "epoch": 28.48, "learning_rate": 6.116923076923076e-07, "loss": 0.0007, "step": 8075 }, { "epoch": 28.57, "learning_rate": 6.104102564102563e-07, "loss": 0.0006, "step": 8100 }, { "epoch": 28.66, "learning_rate": 6.091282051282051e-07, "loss": 0.0019, "step": 8125 }, { "epoch": 28.75, "learning_rate": 6.078461538461538e-07, "loss": 0.0012, "step": 8150 }, { "epoch": 28.84, "learning_rate": 6.065641025641025e-07, "loss": 0.0009, "step": 8175 }, { "epoch": 28.92, "learning_rate": 6.052820512820512e-07, "loss": 0.0005, "step": 8200 }, { "epoch": 29.01, "learning_rate": 6.04e-07, "loss": 0.0011, "step": 8225 }, { "epoch": 29.1, "learning_rate": 6.027179487179487e-07, "loss": 0.0005, "step": 8250 }, { "epoch": 29.19, "learning_rate": 6.014358974358974e-07, "loss": 0.0006, "step": 8275 }, { "epoch": 29.28, "learning_rate": 6.001538461538461e-07, "loss": 0.0007, "step": 8300 }, { "epoch": 29.37, "learning_rate": 5.988717948717949e-07, "loss": 0.0013, "step": 8325 }, { "epoch": 29.45, "learning_rate": 5.975897435897436e-07, "loss": 0.0017, "step": 8350 }, { "epoch": 29.54, "learning_rate": 5.963076923076923e-07, "loss": 0.0014, "step": 8375 }, { "epoch": 29.63, "learning_rate": 5.95025641025641e-07, "loss": 0.0006, "step": 8400 }, { "epoch": 29.72, "learning_rate": 5.937435897435898e-07, "loss": 0.0013, "step": 8425 }, { "epoch": 29.81, "learning_rate": 5.924615384615385e-07, "loss": 0.0006, "step": 8450 }, { "epoch": 29.89, "learning_rate": 5.911794871794872e-07, "loss": 0.0007, "step": 8475 }, { "epoch": 29.98, "learning_rate": 5.898974358974359e-07, "loss": 0.0007, "step": 8500 }, { "epoch": 30.07, "learning_rate": 5.886153846153847e-07, "loss": 0.0012, "step": 8525 }, { "epoch": 30.16, "learning_rate": 5.873333333333334e-07, "loss": 0.0009, "step": 8550 }, { "epoch": 30.25, "learning_rate": 5.860512820512821e-07, "loss": 0.0004, "step": 8575 }, { "epoch": 30.34, "learning_rate": 5.847692307692308e-07, "loss": 0.0004, "step": 8600 }, { "epoch": 30.42, "learning_rate": 5.834871794871796e-07, "loss": 0.0004, "step": 8625 }, { "epoch": 30.51, "learning_rate": 5.822051282051282e-07, "loss": 0.0005, "step": 8650 }, { "epoch": 30.6, "learning_rate": 5.809230769230769e-07, "loss": 0.0005, "step": 8675 }, { "epoch": 30.69, "learning_rate": 5.796410256410256e-07, "loss": 0.0018, "step": 8700 }, { "epoch": 30.78, "learning_rate": 5.783589743589743e-07, "loss": 0.0005, "step": 8725 }, { "epoch": 30.86, "learning_rate": 5.77076923076923e-07, "loss": 0.0005, "step": 8750 }, { "epoch": 30.95, "learning_rate": 5.757948717948717e-07, "loss": 0.0004, "step": 8775 }, { "epoch": 31.04, "learning_rate": 5.745641025641026e-07, "loss": 0.0007, "step": 8800 }, { "epoch": 31.13, "learning_rate": 5.732820512820512e-07, "loss": 0.0004, "step": 8825 }, { "epoch": 31.22, "learning_rate": 5.719999999999999e-07, "loss": 0.0004, "step": 8850 }, { "epoch": 31.31, "learning_rate": 5.707179487179486e-07, "loss": 0.0005, "step": 8875 }, { "epoch": 31.39, "learning_rate": 5.694358974358973e-07, "loss": 0.0004, "step": 8900 }, { "epoch": 31.48, "learning_rate": 5.681538461538461e-07, "loss": 0.0006, "step": 8925 }, { "epoch": 31.57, "learning_rate": 5.668717948717948e-07, "loss": 0.0004, "step": 8950 }, { "epoch": 31.66, "learning_rate": 5.655897435897435e-07, "loss": 0.0004, "step": 8975 }, { "epoch": 31.75, "learning_rate": 5.643076923076922e-07, "loss": 0.0007, "step": 9000 }, { "epoch": 31.75, "eval_loss": 0.38387608528137207, "eval_runtime": 2254.4584, "eval_samples_per_second": 4.024, "eval_steps_per_second": 0.252, "eval_wer": 6.039496599086518, "step": 9000 }, { "epoch": 31.83, "learning_rate": 5.63025641025641e-07, "loss": 0.0004, "step": 9025 }, { "epoch": 31.92, "learning_rate": 5.617435897435897e-07, "loss": 0.0004, "step": 9050 }, { "epoch": 32.01, "learning_rate": 5.604615384615384e-07, "loss": 0.0004, "step": 9075 }, { "epoch": 32.1, "learning_rate": 5.591794871794871e-07, "loss": 0.0008, "step": 9100 }, { "epoch": 32.19, "learning_rate": 5.578974358974359e-07, "loss": 0.0004, "step": 9125 }, { "epoch": 32.28, "learning_rate": 5.566153846153846e-07, "loss": 0.0005, "step": 9150 }, { "epoch": 32.36, "learning_rate": 5.553333333333333e-07, "loss": 0.0004, "step": 9175 }, { "epoch": 32.45, "learning_rate": 5.54051282051282e-07, "loss": 0.0006, "step": 9200 }, { "epoch": 32.54, "learning_rate": 5.527692307692308e-07, "loss": 0.0003, "step": 9225 }, { "epoch": 32.63, "learning_rate": 5.514871794871795e-07, "loss": 0.0004, "step": 9250 }, { "epoch": 32.72, "learning_rate": 5.502051282051282e-07, "loss": 0.0003, "step": 9275 }, { "epoch": 32.8, "learning_rate": 5.489230769230769e-07, "loss": 0.0004, "step": 9300 }, { "epoch": 32.89, "learning_rate": 5.476410256410257e-07, "loss": 0.0011, "step": 9325 }, { "epoch": 32.98, "learning_rate": 5.463589743589744e-07, "loss": 0.0004, "step": 9350 }, { "epoch": 33.07, "learning_rate": 5.450769230769231e-07, "loss": 0.0003, "step": 9375 }, { "epoch": 33.16, "learning_rate": 5.437948717948718e-07, "loss": 0.0003, "step": 9400 }, { "epoch": 33.25, "learning_rate": 5.425128205128204e-07, "loss": 0.0007, "step": 9425 }, { "epoch": 33.33, "learning_rate": 5.412307692307692e-07, "loss": 0.0004, "step": 9450 }, { "epoch": 33.42, "learning_rate": 5.399487179487179e-07, "loss": 0.0004, "step": 9475 }, { "epoch": 33.51, "learning_rate": 5.386666666666666e-07, "loss": 0.0006, "step": 9500 }, { "epoch": 33.6, "learning_rate": 5.373846153846153e-07, "loss": 0.0003, "step": 9525 }, { "epoch": 33.69, "learning_rate": 5.361025641025641e-07, "loss": 0.0003, "step": 9550 }, { "epoch": 33.77, "learning_rate": 5.348205128205128e-07, "loss": 0.0004, "step": 9575 }, { "epoch": 33.86, "learning_rate": 5.335384615384615e-07, "loss": 0.0003, "step": 9600 }, { "epoch": 33.95, "learning_rate": 5.322564102564102e-07, "loss": 0.0004, "step": 9625 }, { "epoch": 34.04, "learning_rate": 5.30974358974359e-07, "loss": 0.0005, "step": 9650 }, { "epoch": 34.13, "learning_rate": 5.296923076923077e-07, "loss": 0.0003, "step": 9675 }, { "epoch": 34.22, "learning_rate": 5.284102564102564e-07, "loss": 0.0003, "step": 9700 }, { "epoch": 34.3, "learning_rate": 5.271282051282051e-07, "loss": 0.0003, "step": 9725 }, { "epoch": 34.39, "learning_rate": 5.258461538461539e-07, "loss": 0.0003, "step": 9750 }, { "epoch": 34.48, "learning_rate": 5.245641025641026e-07, "loss": 0.0003, "step": 9775 }, { "epoch": 34.57, "learning_rate": 5.232820512820513e-07, "loss": 0.0005, "step": 9800 }, { "epoch": 34.66, "learning_rate": 5.22e-07, "loss": 0.0003, "step": 9825 }, { "epoch": 34.74, "learning_rate": 5.207179487179488e-07, "loss": 0.0003, "step": 9850 }, { "epoch": 34.83, "learning_rate": 5.194358974358975e-07, "loss": 0.0003, "step": 9875 }, { "epoch": 34.92, "learning_rate": 5.181538461538462e-07, "loss": 0.0003, "step": 9900 }, { "epoch": 35.01, "learning_rate": 5.168717948717948e-07, "loss": 0.0003, "step": 9925 }, { "epoch": 35.1, "learning_rate": 5.155897435897436e-07, "loss": 0.0003, "step": 9950 }, { "epoch": 35.19, "learning_rate": 5.143076923076923e-07, "loss": 0.0003, "step": 9975 }, { "epoch": 35.27, "learning_rate": 5.13025641025641e-07, "loss": 0.0003, "step": 10000 }, { "epoch": 35.27, "eval_loss": 0.4090217649936676, "eval_runtime": 2248.3554, "eval_samples_per_second": 4.035, "eval_steps_per_second": 0.252, "eval_wer": 6.205434889757829, "step": 10000 }, { "epoch": 35.36, "learning_rate": 5.117435897435896e-07, "loss": 0.0003, "step": 10025 }, { "epoch": 35.45, "learning_rate": 5.104615384615384e-07, "loss": 0.0003, "step": 10050 }, { "epoch": 35.54, "learning_rate": 5.091794871794871e-07, "loss": 0.0004, "step": 10075 }, { "epoch": 35.63, "learning_rate": 5.078974358974358e-07, "loss": 0.0004, "step": 10100 }, { "epoch": 35.71, "learning_rate": 5.066153846153845e-07, "loss": 0.0006, "step": 10125 }, { "epoch": 35.8, "learning_rate": 5.053333333333333e-07, "loss": 0.0003, "step": 10150 }, { "epoch": 35.89, "learning_rate": 5.04051282051282e-07, "loss": 0.0006, "step": 10175 }, { "epoch": 35.98, "learning_rate": 5.027692307692307e-07, "loss": 0.0007, "step": 10200 }, { "epoch": 36.07, "learning_rate": 5.014871794871794e-07, "loss": 0.0005, "step": 10225 }, { "epoch": 36.16, "learning_rate": 5.002051282051282e-07, "loss": 0.0005, "step": 10250 }, { "epoch": 36.24, "learning_rate": 4.989230769230769e-07, "loss": 0.0008, "step": 10275 }, { "epoch": 36.33, "learning_rate": 4.976410256410256e-07, "loss": 0.0019, "step": 10300 }, { "epoch": 36.42, "learning_rate": 4.963589743589743e-07, "loss": 0.0004, "step": 10325 }, { "epoch": 36.51, "learning_rate": 4.95076923076923e-07, "loss": 0.0005, "step": 10350 }, { "epoch": 36.6, "learning_rate": 4.937948717948718e-07, "loss": 0.0004, "step": 10375 }, { "epoch": 36.68, "learning_rate": 4.925128205128205e-07, "loss": 0.0004, "step": 10400 }, { "epoch": 36.77, "learning_rate": 4.912307692307692e-07, "loss": 0.0003, "step": 10425 }, { "epoch": 36.86, "learning_rate": 4.89948717948718e-07, "loss": 0.0007, "step": 10450 }, { "epoch": 36.95, "learning_rate": 4.886666666666667e-07, "loss": 0.0006, "step": 10475 }, { "epoch": 37.04, "learning_rate": 4.873846153846153e-07, "loss": 0.0003, "step": 10500 }, { "epoch": 37.13, "learning_rate": 4.86102564102564e-07, "loss": 0.0004, "step": 10525 }, { "epoch": 37.21, "learning_rate": 4.848205128205127e-07, "loss": 0.001, "step": 10550 }, { "epoch": 37.3, "learning_rate": 4.835384615384615e-07, "loss": 0.0009, "step": 10575 }, { "epoch": 37.39, "learning_rate": 4.822564102564102e-07, "loss": 0.0007, "step": 10600 }, { "epoch": 37.48, "learning_rate": 4.809743589743589e-07, "loss": 0.0006, "step": 10625 }, { "epoch": 37.57, "learning_rate": 4.796923076923076e-07, "loss": 0.0004, "step": 10650 }, { "epoch": 37.65, "learning_rate": 4.784102564102564e-07, "loss": 0.0003, "step": 10675 }, { "epoch": 37.74, "learning_rate": 4.771282051282051e-07, "loss": 0.0005, "step": 10700 }, { "epoch": 37.83, "learning_rate": 4.758461538461538e-07, "loss": 0.0003, "step": 10725 }, { "epoch": 37.92, "learning_rate": 4.7456410256410254e-07, "loss": 0.0005, "step": 10750 }, { "epoch": 38.01, "learning_rate": 4.7328205128205127e-07, "loss": 0.0003, "step": 10775 }, { "epoch": 38.1, "learning_rate": 4.7199999999999994e-07, "loss": 0.0004, "step": 10800 }, { "epoch": 38.18, "learning_rate": 4.7071794871794866e-07, "loss": 0.0006, "step": 10825 }, { "epoch": 38.27, "learning_rate": 4.694358974358974e-07, "loss": 0.0006, "step": 10850 }, { "epoch": 38.36, "learning_rate": 4.681538461538461e-07, "loss": 0.001, "step": 10875 }, { "epoch": 38.45, "learning_rate": 4.6687179487179484e-07, "loss": 0.0003, "step": 10900 }, { "epoch": 38.54, "learning_rate": 4.6558974358974356e-07, "loss": 0.0003, "step": 10925 }, { "epoch": 38.62, "learning_rate": 4.643076923076923e-07, "loss": 0.0002, "step": 10950 }, { "epoch": 38.71, "learning_rate": 4.63025641025641e-07, "loss": 0.0004, "step": 10975 }, { "epoch": 38.8, "learning_rate": 4.6174358974358974e-07, "loss": 0.0003, "step": 11000 }, { "epoch": 38.8, "eval_loss": 0.405325323343277, "eval_runtime": 2247.2119, "eval_samples_per_second": 4.037, "eval_steps_per_second": 0.252, "eval_wer": 6.2859396050340095, "step": 11000 }, { "epoch": 38.89, "learning_rate": 4.6046153846153846e-07, "loss": 0.0003, "step": 11025 }, { "epoch": 38.98, "learning_rate": 4.591794871794872e-07, "loss": 0.0006, "step": 11050 }, { "epoch": 39.07, "learning_rate": 4.578974358974359e-07, "loss": 0.0003, "step": 11075 }, { "epoch": 39.15, "learning_rate": 4.566153846153846e-07, "loss": 0.0002, "step": 11100 }, { "epoch": 39.24, "learning_rate": 4.553333333333333e-07, "loss": 0.0002, "step": 11125 }, { "epoch": 39.33, "learning_rate": 4.5405128205128203e-07, "loss": 0.0002, "step": 11150 }, { "epoch": 39.42, "learning_rate": 4.5276923076923076e-07, "loss": 0.0002, "step": 11175 }, { "epoch": 39.51, "learning_rate": 4.514871794871795e-07, "loss": 0.0002, "step": 11200 }, { "epoch": 39.59, "learning_rate": 4.502051282051282e-07, "loss": 0.0002, "step": 11225 }, { "epoch": 39.68, "learning_rate": 4.489230769230769e-07, "loss": 0.0002, "step": 11250 }, { "epoch": 39.77, "learning_rate": 4.476410256410256e-07, "loss": 0.0002, "step": 11275 }, { "epoch": 39.86, "learning_rate": 4.4635897435897433e-07, "loss": 0.0002, "step": 11300 }, { "epoch": 39.95, "learning_rate": 4.4507692307692305e-07, "loss": 0.0002, "step": 11325 }, { "epoch": 40.04, "learning_rate": 4.437948717948718e-07, "loss": 0.0002, "step": 11350 }, { "epoch": 40.12, "learning_rate": 4.425128205128205e-07, "loss": 0.0002, "step": 11375 }, { "epoch": 40.21, "learning_rate": 4.412307692307692e-07, "loss": 0.0002, "step": 11400 }, { "epoch": 40.3, "learning_rate": 4.399487179487179e-07, "loss": 0.0002, "step": 11425 }, { "epoch": 40.39, "learning_rate": 4.386666666666666e-07, "loss": 0.0002, "step": 11450 }, { "epoch": 40.48, "learning_rate": 4.3738461538461535e-07, "loss": 0.0002, "step": 11475 }, { "epoch": 40.56, "learning_rate": 4.361025641025641e-07, "loss": 0.0002, "step": 11500 }, { "epoch": 40.65, "learning_rate": 4.348205128205128e-07, "loss": 0.0002, "step": 11525 }, { "epoch": 40.74, "learning_rate": 4.335384615384615e-07, "loss": 0.0002, "step": 11550 }, { "epoch": 40.83, "learning_rate": 4.3225641025641025e-07, "loss": 0.0002, "step": 11575 }, { "epoch": 40.92, "learning_rate": 4.3097435897435897e-07, "loss": 0.0002, "step": 11600 }, { "epoch": 41.01, "learning_rate": 4.296923076923077e-07, "loss": 0.0002, "step": 11625 }, { "epoch": 41.09, "learning_rate": 4.284102564102564e-07, "loss": 0.0002, "step": 11650 }, { "epoch": 41.18, "learning_rate": 4.2712820512820515e-07, "loss": 0.0002, "step": 11675 }, { "epoch": 41.27, "learning_rate": 4.258461538461538e-07, "loss": 0.0002, "step": 11700 }, { "epoch": 41.36, "learning_rate": 4.2456410256410254e-07, "loss": 0.0002, "step": 11725 }, { "epoch": 41.45, "learning_rate": 4.2328205128205127e-07, "loss": 0.0002, "step": 11750 }, { "epoch": 41.53, "learning_rate": 4.2199999999999994e-07, "loss": 0.0002, "step": 11775 }, { "epoch": 41.62, "learning_rate": 4.2071794871794867e-07, "loss": 0.0002, "step": 11800 }, { "epoch": 41.71, "learning_rate": 4.194358974358974e-07, "loss": 0.0002, "step": 11825 }, { "epoch": 41.8, "learning_rate": 4.181538461538461e-07, "loss": 0.0002, "step": 11850 }, { "epoch": 41.89, "learning_rate": 4.1687179487179484e-07, "loss": 0.0002, "step": 11875 }, { "epoch": 41.98, "learning_rate": 4.1558974358974356e-07, "loss": 0.0002, "step": 11900 }, { "epoch": 42.06, "learning_rate": 4.143076923076923e-07, "loss": 0.0002, "step": 11925 }, { "epoch": 42.15, "learning_rate": 4.13025641025641e-07, "loss": 0.0002, "step": 11950 }, { "epoch": 42.24, "learning_rate": 4.1174358974358974e-07, "loss": 0.0002, "step": 11975 }, { "epoch": 42.33, "learning_rate": 4.1046153846153846e-07, "loss": 0.0002, "step": 12000 }, { "epoch": 42.33, "eval_loss": 0.4235469400882721, "eval_runtime": 2255.4094, "eval_samples_per_second": 4.022, "eval_steps_per_second": 0.251, "eval_wer": 6.34672887983439, "step": 12000 }, { "epoch": 42.42, "learning_rate": 4.0917948717948714e-07, "loss": 0.0002, "step": 12025 }, { "epoch": 42.5, "learning_rate": 4.0789743589743586e-07, "loss": 0.0002, "step": 12050 }, { "epoch": 42.59, "learning_rate": 4.066153846153846e-07, "loss": 0.0002, "step": 12075 }, { "epoch": 42.68, "learning_rate": 4.053333333333333e-07, "loss": 0.0002, "step": 12100 }, { "epoch": 42.77, "learning_rate": 4.0405128205128203e-07, "loss": 0.0002, "step": 12125 }, { "epoch": 42.86, "learning_rate": 4.0276923076923076e-07, "loss": 0.0002, "step": 12150 }, { "epoch": 42.95, "learning_rate": 4.014871794871795e-07, "loss": 0.0002, "step": 12175 }, { "epoch": 43.03, "learning_rate": 4.002051282051282e-07, "loss": 0.0002, "step": 12200 }, { "epoch": 43.12, "learning_rate": 3.9892307692307693e-07, "loss": 0.0002, "step": 12225 }, { "epoch": 43.21, "learning_rate": 3.9764102564102566e-07, "loss": 0.0002, "step": 12250 }, { "epoch": 43.3, "learning_rate": 3.963589743589744e-07, "loss": 0.0002, "step": 12275 }, { "epoch": 43.39, "learning_rate": 3.950769230769231e-07, "loss": 0.0002, "step": 12300 }, { "epoch": 43.47, "learning_rate": 3.9379487179487173e-07, "loss": 0.0002, "step": 12325 }, { "epoch": 43.56, "learning_rate": 3.9251282051282045e-07, "loss": 0.0002, "step": 12350 }, { "epoch": 43.65, "learning_rate": 3.912307692307692e-07, "loss": 0.0002, "step": 12375 }, { "epoch": 43.74, "learning_rate": 3.899487179487179e-07, "loss": 0.0002, "step": 12400 }, { "epoch": 43.83, "learning_rate": 3.8866666666666663e-07, "loss": 0.0002, "step": 12425 }, { "epoch": 43.92, "learning_rate": 3.8738461538461535e-07, "loss": 0.0002, "step": 12450 }, { "epoch": 44.0, "learning_rate": 3.861025641025641e-07, "loss": 0.0002, "step": 12475 }, { "epoch": 44.09, "learning_rate": 3.848205128205128e-07, "loss": 0.0002, "step": 12500 }, { "epoch": 44.18, "learning_rate": 3.835384615384615e-07, "loss": 0.0002, "step": 12525 }, { "epoch": 44.27, "learning_rate": 3.8225641025641025e-07, "loss": 0.0002, "step": 12550 }, { "epoch": 44.36, "learning_rate": 3.80974358974359e-07, "loss": 0.0002, "step": 12575 }, { "epoch": 44.44, "learning_rate": 3.796923076923077e-07, "loss": 0.0002, "step": 12600 }, { "epoch": 44.53, "learning_rate": 3.7841025641025637e-07, "loss": 0.0002, "step": 12625 }, { "epoch": 44.62, "learning_rate": 3.771282051282051e-07, "loss": 0.0002, "step": 12650 }, { "epoch": 44.71, "learning_rate": 3.758461538461538e-07, "loss": 0.0002, "step": 12675 }, { "epoch": 44.8, "learning_rate": 3.7456410256410255e-07, "loss": 0.0002, "step": 12700 }, { "epoch": 44.89, "learning_rate": 3.7328205128205127e-07, "loss": 0.0002, "step": 12725 }, { "epoch": 44.97, "learning_rate": 3.72e-07, "loss": 0.0002, "step": 12750 }, { "epoch": 45.06, "learning_rate": 3.707179487179487e-07, "loss": 0.0002, "step": 12775 }, { "epoch": 45.15, "learning_rate": 3.6943589743589745e-07, "loss": 0.0002, "step": 12800 }, { "epoch": 45.24, "learning_rate": 3.6815384615384617e-07, "loss": 0.0002, "step": 12825 }, { "epoch": 45.33, "learning_rate": 3.668717948717949e-07, "loss": 0.0002, "step": 12850 }, { "epoch": 45.41, "learning_rate": 3.6558974358974357e-07, "loss": 0.0002, "step": 12875 }, { "epoch": 45.5, "learning_rate": 3.643076923076923e-07, "loss": 0.0002, "step": 12900 }, { "epoch": 45.59, "learning_rate": 3.6302564102564096e-07, "loss": 0.0002, "step": 12925 }, { "epoch": 45.68, "learning_rate": 3.617435897435897e-07, "loss": 0.0002, "step": 12950 }, { "epoch": 45.77, "learning_rate": 3.604615384615384e-07, "loss": 0.0002, "step": 12975 }, { "epoch": 45.86, "learning_rate": 3.5917948717948714e-07, "loss": 0.0002, "step": 13000 }, { "epoch": 45.86, "eval_loss": 0.4325754940509796, "eval_runtime": 2261.9353, "eval_samples_per_second": 4.011, "eval_steps_per_second": 0.251, "eval_wer": 6.350014786580357, "step": 13000 }, { "epoch": 45.94, "learning_rate": 3.5789743589743586e-07, "loss": 0.0002, "step": 13025 }, { "epoch": 46.03, "learning_rate": 3.566153846153846e-07, "loss": 0.0002, "step": 13050 }, { "epoch": 46.12, "learning_rate": 3.553333333333333e-07, "loss": 0.0002, "step": 13075 }, { "epoch": 46.21, "learning_rate": 3.5405128205128204e-07, "loss": 0.0002, "step": 13100 }, { "epoch": 46.3, "learning_rate": 3.5276923076923076e-07, "loss": 0.0001, "step": 13125 }, { "epoch": 46.38, "learning_rate": 3.514871794871795e-07, "loss": 0.0001, "step": 13150 }, { "epoch": 46.47, "learning_rate": 3.502051282051282e-07, "loss": 0.0001, "step": 13175 }, { "epoch": 46.56, "learning_rate": 3.4892307692307694e-07, "loss": 0.0001, "step": 13200 }, { "epoch": 46.65, "learning_rate": 3.476410256410256e-07, "loss": 0.0001, "step": 13225 }, { "epoch": 46.74, "learning_rate": 3.4635897435897433e-07, "loss": 0.0001, "step": 13250 }, { "epoch": 46.83, "learning_rate": 3.4507692307692306e-07, "loss": 0.0001, "step": 13275 }, { "epoch": 46.91, "learning_rate": 3.437948717948718e-07, "loss": 0.0001, "step": 13300 }, { "epoch": 47.0, "learning_rate": 3.425128205128205e-07, "loss": 0.0001, "step": 13325 }, { "epoch": 47.09, "learning_rate": 3.4123076923076923e-07, "loss": 0.0001, "step": 13350 }, { "epoch": 47.18, "learning_rate": 3.3994871794871796e-07, "loss": 0.0001, "step": 13375 }, { "epoch": 47.27, "learning_rate": 3.386666666666667e-07, "loss": 0.0001, "step": 13400 }, { "epoch": 47.35, "learning_rate": 3.3738461538461535e-07, "loss": 0.0001, "step": 13425 }, { "epoch": 47.44, "learning_rate": 3.361025641025641e-07, "loss": 0.0001, "step": 13450 }, { "epoch": 47.53, "learning_rate": 3.348205128205128e-07, "loss": 0.0001, "step": 13475 }, { "epoch": 47.62, "learning_rate": 3.3353846153846153e-07, "loss": 0.0001, "step": 13500 }, { "epoch": 47.71, "learning_rate": 3.3225641025641025e-07, "loss": 0.0001, "step": 13525 }, { "epoch": 47.8, "learning_rate": 3.309743589743589e-07, "loss": 0.0001, "step": 13550 }, { "epoch": 47.88, "learning_rate": 3.2969230769230765e-07, "loss": 0.0001, "step": 13575 }, { "epoch": 47.97, "learning_rate": 3.284102564102564e-07, "loss": 0.0001, "step": 13600 }, { "epoch": 48.06, "learning_rate": 3.271282051282051e-07, "loss": 0.0001, "step": 13625 }, { "epoch": 48.15, "learning_rate": 3.258461538461538e-07, "loss": 0.0001, "step": 13650 }, { "epoch": 48.24, "learning_rate": 3.2456410256410255e-07, "loss": 0.0001, "step": 13675 }, { "epoch": 48.32, "learning_rate": 3.2328205128205127e-07, "loss": 0.0001, "step": 13700 }, { "epoch": 48.41, "learning_rate": 3.22e-07, "loss": 0.0001, "step": 13725 }, { "epoch": 48.5, "learning_rate": 3.207179487179487e-07, "loss": 0.0001, "step": 13750 }, { "epoch": 48.59, "learning_rate": 3.1943589743589745e-07, "loss": 0.0001, "step": 13775 }, { "epoch": 48.68, "learning_rate": 3.1815384615384617e-07, "loss": 0.0001, "step": 13800 }, { "epoch": 48.77, "learning_rate": 3.168717948717949e-07, "loss": 0.0001, "step": 13825 }, { "epoch": 48.85, "learning_rate": 3.1558974358974357e-07, "loss": 0.0001, "step": 13850 }, { "epoch": 48.94, "learning_rate": 3.143076923076923e-07, "loss": 0.0001, "step": 13875 }, { "epoch": 49.03, "learning_rate": 3.13025641025641e-07, "loss": 0.0001, "step": 13900 }, { "epoch": 49.12, "learning_rate": 3.1174358974358974e-07, "loss": 0.0001, "step": 13925 }, { "epoch": 49.21, "learning_rate": 3.104615384615384e-07, "loss": 0.0001, "step": 13950 }, { "epoch": 49.29, "learning_rate": 3.0917948717948714e-07, "loss": 0.0001, "step": 13975 }, { "epoch": 49.38, "learning_rate": 3.0789743589743586e-07, "loss": 0.0001, "step": 14000 }, { "epoch": 49.38, "eval_loss": 0.44154471158981323, "eval_runtime": 2262.1794, "eval_samples_per_second": 4.01, "eval_steps_per_second": 0.251, "eval_wer": 6.371373180429139, "step": 14000 }, { "epoch": 49.47, "learning_rate": 3.066153846153846e-07, "loss": 0.0001, "step": 14025 }, { "epoch": 49.56, "learning_rate": 3.053333333333333e-07, "loss": 0.0001, "step": 14050 }, { "epoch": 49.65, "learning_rate": 3.0405128205128204e-07, "loss": 0.0001, "step": 14075 }, { "epoch": 49.74, "learning_rate": 3.0276923076923076e-07, "loss": 0.0001, "step": 14100 }, { "epoch": 49.82, "learning_rate": 3.014871794871795e-07, "loss": 0.0001, "step": 14125 }, { "epoch": 49.91, "learning_rate": 3.0020512820512816e-07, "loss": 0.0001, "step": 14150 }, { "epoch": 50.0, "learning_rate": 2.989230769230769e-07, "loss": 0.0001, "step": 14175 }, { "epoch": 50.09, "learning_rate": 2.976410256410256e-07, "loss": 0.0001, "step": 14200 }, { "epoch": 50.18, "learning_rate": 2.9635897435897434e-07, "loss": 0.0001, "step": 14225 }, { "epoch": 50.26, "learning_rate": 2.9507692307692306e-07, "loss": 0.0001, "step": 14250 }, { "epoch": 50.35, "learning_rate": 2.937948717948718e-07, "loss": 0.0001, "step": 14275 }, { "epoch": 50.44, "learning_rate": 2.925128205128205e-07, "loss": 0.0001, "step": 14300 }, { "epoch": 50.53, "learning_rate": 2.9123076923076923e-07, "loss": 0.0001, "step": 14325 }, { "epoch": 50.62, "learning_rate": 2.8994871794871796e-07, "loss": 0.0001, "step": 14350 }, { "epoch": 50.71, "learning_rate": 2.886666666666667e-07, "loss": 0.0001, "step": 14375 }, { "epoch": 50.79, "learning_rate": 2.873846153846154e-07, "loss": 0.0001, "step": 14400 }, { "epoch": 50.88, "learning_rate": 2.8610256410256413e-07, "loss": 0.0001, "step": 14425 }, { "epoch": 50.97, "learning_rate": 2.848205128205128e-07, "loss": 0.0001, "step": 14450 }, { "epoch": 51.06, "learning_rate": 2.8353846153846153e-07, "loss": 0.0001, "step": 14475 }, { "epoch": 51.15, "learning_rate": 2.822564102564102e-07, "loss": 0.0001, "step": 14500 }, { "epoch": 51.23, "learning_rate": 2.8097435897435893e-07, "loss": 0.0001, "step": 14525 }, { "epoch": 51.32, "learning_rate": 2.7969230769230765e-07, "loss": 0.0001, "step": 14550 }, { "epoch": 51.41, "learning_rate": 2.784102564102564e-07, "loss": 0.0001, "step": 14575 }, { "epoch": 51.5, "learning_rate": 2.771282051282051e-07, "loss": 0.0001, "step": 14600 }, { "epoch": 51.59, "learning_rate": 2.758461538461538e-07, "loss": 0.0001, "step": 14625 }, { "epoch": 51.68, "learning_rate": 2.7456410256410255e-07, "loss": 0.0001, "step": 14650 }, { "epoch": 51.76, "learning_rate": 2.732820512820513e-07, "loss": 0.0001, "step": 14675 }, { "epoch": 51.85, "learning_rate": 2.72e-07, "loss": 0.0001, "step": 14700 }, { "epoch": 51.94, "learning_rate": 2.707179487179487e-07, "loss": 0.0001, "step": 14725 }, { "epoch": 52.03, "learning_rate": 2.694358974358974e-07, "loss": 0.0001, "step": 14750 }, { "epoch": 52.12, "learning_rate": 2.681538461538461e-07, "loss": 0.0001, "step": 14775 }, { "epoch": 52.2, "learning_rate": 2.6687179487179485e-07, "loss": 0.0001, "step": 14800 }, { "epoch": 52.29, "learning_rate": 2.6558974358974357e-07, "loss": 0.0001, "step": 14825 }, { "epoch": 52.38, "learning_rate": 2.643076923076923e-07, "loss": 0.0001, "step": 14850 }, { "epoch": 52.47, "learning_rate": 2.63025641025641e-07, "loss": 0.0001, "step": 14875 }, { "epoch": 52.56, "learning_rate": 2.6174358974358975e-07, "loss": 0.0001, "step": 14900 }, { "epoch": 52.65, "learning_rate": 2.6046153846153847e-07, "loss": 0.0001, "step": 14925 }, { "epoch": 52.73, "learning_rate": 2.591794871794872e-07, "loss": 0.0001, "step": 14950 }, { "epoch": 52.82, "learning_rate": 2.578974358974359e-07, "loss": 0.0001, "step": 14975 }, { "epoch": 52.91, "learning_rate": 2.5661538461538464e-07, "loss": 0.0001, "step": 15000 }, { "epoch": 52.91, "eval_loss": 0.45060843229293823, "eval_runtime": 2259.2596, "eval_samples_per_second": 4.015, "eval_steps_per_second": 0.251, "eval_wer": 6.387802714158972, "step": 15000 }, { "epoch": 53.0, "learning_rate": 2.5533333333333337e-07, "loss": 0.0001, "step": 15025 }, { "epoch": 53.09, "learning_rate": 2.54051282051282e-07, "loss": 0.0001, "step": 15050 }, { "epoch": 53.17, "learning_rate": 2.527692307692307e-07, "loss": 0.0001, "step": 15075 }, { "epoch": 53.26, "learning_rate": 2.5148717948717944e-07, "loss": 0.0001, "step": 15100 }, { "epoch": 53.35, "learning_rate": 2.5020512820512816e-07, "loss": 0.0001, "step": 15125 }, { "epoch": 53.44, "learning_rate": 2.489230769230769e-07, "loss": 0.0001, "step": 15150 }, { "epoch": 53.53, "learning_rate": 2.476410256410256e-07, "loss": 0.0001, "step": 15175 }, { "epoch": 53.62, "learning_rate": 2.4635897435897434e-07, "loss": 0.0001, "step": 15200 }, { "epoch": 53.7, "learning_rate": 2.4507692307692306e-07, "loss": 0.0001, "step": 15225 }, { "epoch": 53.79, "learning_rate": 2.437948717948718e-07, "loss": 0.0001, "step": 15250 }, { "epoch": 53.88, "learning_rate": 2.425128205128205e-07, "loss": 0.0001, "step": 15275 }, { "epoch": 53.97, "learning_rate": 2.412307692307692e-07, "loss": 0.0001, "step": 15300 }, { "epoch": 54.06, "learning_rate": 2.399487179487179e-07, "loss": 0.0001, "step": 15325 }, { "epoch": 54.14, "learning_rate": 2.3866666666666663e-07, "loss": 0.0001, "step": 15350 }, { "epoch": 54.23, "learning_rate": 2.3738461538461538e-07, "loss": 0.0001, "step": 15375 }, { "epoch": 54.32, "learning_rate": 2.361025641025641e-07, "loss": 0.0001, "step": 15400 }, { "epoch": 54.41, "learning_rate": 2.3482051282051283e-07, "loss": 0.0001, "step": 15425 }, { "epoch": 54.5, "learning_rate": 2.335384615384615e-07, "loss": 0.0001, "step": 15450 }, { "epoch": 54.59, "learning_rate": 2.3225641025641023e-07, "loss": 0.0001, "step": 15475 }, { "epoch": 54.67, "learning_rate": 2.3097435897435896e-07, "loss": 0.0001, "step": 15500 }, { "epoch": 54.76, "learning_rate": 2.2969230769230768e-07, "loss": 0.0001, "step": 15525 }, { "epoch": 54.85, "learning_rate": 2.284102564102564e-07, "loss": 0.0001, "step": 15550 }, { "epoch": 54.94, "learning_rate": 2.2712820512820513e-07, "loss": 0.0001, "step": 15575 }, { "epoch": 55.03, "learning_rate": 2.2584615384615383e-07, "loss": 0.0001, "step": 15600 }, { "epoch": 55.11, "learning_rate": 2.2456410256410255e-07, "loss": 0.0001, "step": 15625 }, { "epoch": 55.2, "learning_rate": 2.2328205128205128e-07, "loss": 0.0001, "step": 15650 }, { "epoch": 55.29, "learning_rate": 2.22e-07, "loss": 0.0001, "step": 15675 }, { "epoch": 55.38, "learning_rate": 2.2071794871794873e-07, "loss": 0.0001, "step": 15700 }, { "epoch": 55.47, "learning_rate": 2.1943589743589743e-07, "loss": 0.0001, "step": 15725 }, { "epoch": 55.56, "learning_rate": 2.1815384615384612e-07, "loss": 0.0001, "step": 15750 }, { "epoch": 55.64, "learning_rate": 2.1687179487179485e-07, "loss": 0.0001, "step": 15775 }, { "epoch": 55.73, "learning_rate": 2.1558974358974357e-07, "loss": 0.0001, "step": 15800 }, { "epoch": 55.82, "learning_rate": 2.143076923076923e-07, "loss": 0.0001, "step": 15825 }, { "epoch": 55.91, "learning_rate": 2.1302564102564102e-07, "loss": 0.0001, "step": 15850 }, { "epoch": 56.0, "learning_rate": 2.1174358974358975e-07, "loss": 0.0001, "step": 15875 }, { "epoch": 56.08, "learning_rate": 2.1046153846153845e-07, "loss": 0.0001, "step": 15900 }, { "epoch": 56.17, "learning_rate": 2.0917948717948717e-07, "loss": 0.0001, "step": 15925 }, { "epoch": 56.26, "learning_rate": 2.078974358974359e-07, "loss": 0.0001, "step": 15950 }, { "epoch": 56.35, "learning_rate": 2.0661538461538462e-07, "loss": 0.0001, "step": 15975 }, { "epoch": 56.44, "learning_rate": 2.0533333333333332e-07, "loss": 0.0001, "step": 16000 }, { "epoch": 56.44, "eval_loss": 0.4586077332496643, "eval_runtime": 2260.5772, "eval_samples_per_second": 4.013, "eval_steps_per_second": 0.251, "eval_wer": 6.409161108007755, "step": 16000 }, { "epoch": 56.53, "learning_rate": 2.0405128205128204e-07, "loss": 0.0001, "step": 16025 }, { "epoch": 56.61, "learning_rate": 2.0276923076923074e-07, "loss": 0.0001, "step": 16050 }, { "epoch": 56.7, "learning_rate": 2.0148717948717947e-07, "loss": 0.0001, "step": 16075 }, { "epoch": 56.79, "learning_rate": 2.002051282051282e-07, "loss": 0.0001, "step": 16100 }, { "epoch": 56.88, "learning_rate": 1.9892307692307692e-07, "loss": 0.0001, "step": 16125 }, { "epoch": 56.97, "learning_rate": 1.9764102564102564e-07, "loss": 0.0001, "step": 16150 }, { "epoch": 57.05, "learning_rate": 1.9635897435897437e-07, "loss": 0.0001, "step": 16175 }, { "epoch": 57.14, "learning_rate": 1.9507692307692306e-07, "loss": 0.0001, "step": 16200 }, { "epoch": 57.23, "learning_rate": 1.937948717948718e-07, "loss": 0.0001, "step": 16225 }, { "epoch": 57.32, "learning_rate": 1.925128205128205e-07, "loss": 0.0001, "step": 16250 }, { "epoch": 57.41, "learning_rate": 1.912307692307692e-07, "loss": 0.0001, "step": 16275 }, { "epoch": 57.5, "learning_rate": 1.8994871794871794e-07, "loss": 0.0001, "step": 16300 }, { "epoch": 57.58, "learning_rate": 1.8866666666666666e-07, "loss": 0.0001, "step": 16325 }, { "epoch": 57.67, "learning_rate": 1.8738461538461536e-07, "loss": 0.0001, "step": 16350 }, { "epoch": 57.76, "learning_rate": 1.8610256410256409e-07, "loss": 0.0001, "step": 16375 }, { "epoch": 57.85, "learning_rate": 1.848205128205128e-07, "loss": 0.0001, "step": 16400 }, { "epoch": 57.94, "learning_rate": 1.8353846153846153e-07, "loss": 0.0001, "step": 16425 }, { "epoch": 58.02, "learning_rate": 1.8225641025641026e-07, "loss": 0.0001, "step": 16450 }, { "epoch": 58.11, "learning_rate": 1.8097435897435898e-07, "loss": 0.0001, "step": 16475 }, { "epoch": 58.2, "learning_rate": 1.7969230769230768e-07, "loss": 0.0001, "step": 16500 }, { "epoch": 58.29, "learning_rate": 1.7841025641025638e-07, "loss": 0.0001, "step": 16525 }, { "epoch": 58.38, "learning_rate": 1.771282051282051e-07, "loss": 0.0001, "step": 16550 }, { "epoch": 58.47, "learning_rate": 1.7584615384615383e-07, "loss": 0.0001, "step": 16575 }, { "epoch": 58.55, "learning_rate": 1.7456410256410256e-07, "loss": 0.0001, "step": 16600 }, { "epoch": 58.64, "learning_rate": 1.7328205128205128e-07, "loss": 0.0001, "step": 16625 }, { "epoch": 58.73, "learning_rate": 1.7199999999999998e-07, "loss": 0.0001, "step": 16650 }, { "epoch": 58.82, "learning_rate": 1.707179487179487e-07, "loss": 0.0001, "step": 16675 }, { "epoch": 58.91, "learning_rate": 1.6943589743589743e-07, "loss": 0.0001, "step": 16700 }, { "epoch": 58.99, "learning_rate": 1.6815384615384615e-07, "loss": 0.0001, "step": 16725 }, { "epoch": 59.08, "learning_rate": 1.6687179487179488e-07, "loss": 0.0001, "step": 16750 }, { "epoch": 59.17, "learning_rate": 1.655897435897436e-07, "loss": 0.0001, "step": 16775 }, { "epoch": 59.26, "learning_rate": 1.643076923076923e-07, "loss": 0.0001, "step": 16800 }, { "epoch": 59.35, "learning_rate": 1.63025641025641e-07, "loss": 0.0001, "step": 16825 }, { "epoch": 59.44, "learning_rate": 1.6174358974358972e-07, "loss": 0.0001, "step": 16850 }, { "epoch": 59.52, "learning_rate": 1.6046153846153845e-07, "loss": 0.0001, "step": 16875 }, { "epoch": 59.61, "learning_rate": 1.5917948717948717e-07, "loss": 0.0001, "step": 16900 }, { "epoch": 59.7, "learning_rate": 1.578974358974359e-07, "loss": 0.0001, "step": 16925 }, { "epoch": 59.79, "learning_rate": 1.5661538461538462e-07, "loss": 0.0001, "step": 16950 }, { "epoch": 59.88, "learning_rate": 1.5533333333333332e-07, "loss": 0.0001, "step": 16975 }, { "epoch": 59.96, "learning_rate": 1.5405128205128205e-07, "loss": 0.0001, "step": 17000 }, { "epoch": 59.96, "eval_loss": 0.46630626916885376, "eval_runtime": 2261.3865, "eval_samples_per_second": 4.012, "eval_steps_per_second": 0.251, "eval_wer": 6.394374527650905, "step": 17000 }, { "epoch": 60.05, "learning_rate": 1.5276923076923077e-07, "loss": 0.0001, "step": 17025 }, { "epoch": 60.14, "learning_rate": 1.514871794871795e-07, "loss": 0.0001, "step": 17050 }, { "epoch": 60.23, "learning_rate": 1.502051282051282e-07, "loss": 0.0001, "step": 17075 }, { "epoch": 60.32, "learning_rate": 1.4892307692307692e-07, "loss": 0.0001, "step": 17100 }, { "epoch": 60.41, "learning_rate": 1.4764102564102562e-07, "loss": 0.0001, "step": 17125 }, { "epoch": 60.49, "learning_rate": 1.4635897435897434e-07, "loss": 0.0001, "step": 17150 }, { "epoch": 60.58, "learning_rate": 1.4507692307692307e-07, "loss": 0.0001, "step": 17175 }, { "epoch": 60.67, "learning_rate": 1.437948717948718e-07, "loss": 0.0001, "step": 17200 }, { "epoch": 60.76, "learning_rate": 1.4251282051282052e-07, "loss": 0.0001, "step": 17225 }, { "epoch": 60.85, "learning_rate": 1.4123076923076924e-07, "loss": 0.0001, "step": 17250 }, { "epoch": 60.93, "learning_rate": 1.3994871794871794e-07, "loss": 0.0001, "step": 17275 }, { "epoch": 61.02, "learning_rate": 1.3866666666666666e-07, "loss": 0.0001, "step": 17300 }, { "epoch": 61.11, "learning_rate": 1.373846153846154e-07, "loss": 0.0001, "step": 17325 }, { "epoch": 61.2, "learning_rate": 1.361025641025641e-07, "loss": 0.0001, "step": 17350 }, { "epoch": 61.29, "learning_rate": 1.348205128205128e-07, "loss": 0.0001, "step": 17375 }, { "epoch": 61.38, "learning_rate": 1.3353846153846154e-07, "loss": 0.0001, "step": 17400 }, { "epoch": 61.46, "learning_rate": 1.3225641025641024e-07, "loss": 0.0001, "step": 17425 }, { "epoch": 61.55, "learning_rate": 1.3097435897435896e-07, "loss": 0.0001, "step": 17450 }, { "epoch": 61.64, "learning_rate": 1.2969230769230768e-07, "loss": 0.0001, "step": 17475 }, { "epoch": 61.73, "learning_rate": 1.284102564102564e-07, "loss": 0.0001, "step": 17500 }, { "epoch": 61.82, "learning_rate": 1.2712820512820513e-07, "loss": 0.0001, "step": 17525 }, { "epoch": 61.9, "learning_rate": 1.2584615384615386e-07, "loss": 0.0001, "step": 17550 }, { "epoch": 61.99, "learning_rate": 1.2456410256410256e-07, "loss": 0.0001, "step": 17575 }, { "epoch": 62.08, "learning_rate": 1.2328205128205128e-07, "loss": 0.0001, "step": 17600 }, { "epoch": 62.17, "learning_rate": 1.2199999999999998e-07, "loss": 0.0001, "step": 17625 }, { "epoch": 62.26, "learning_rate": 1.207179487179487e-07, "loss": 0.0001, "step": 17650 }, { "epoch": 62.35, "learning_rate": 1.1943589743589743e-07, "loss": 0.0001, "step": 17675 }, { "epoch": 62.43, "learning_rate": 1.1815384615384614e-07, "loss": 0.0001, "step": 17700 }, { "epoch": 62.52, "learning_rate": 1.1687179487179487e-07, "loss": 0.0001, "step": 17725 }, { "epoch": 62.61, "learning_rate": 1.1558974358974359e-07, "loss": 0.0001, "step": 17750 }, { "epoch": 62.7, "learning_rate": 1.143076923076923e-07, "loss": 0.0001, "step": 17775 }, { "epoch": 62.79, "learning_rate": 1.1302564102564103e-07, "loss": 0.0001, "step": 17800 }, { "epoch": 62.87, "learning_rate": 1.1174358974358974e-07, "loss": 0.0001, "step": 17825 }, { "epoch": 62.96, "learning_rate": 1.1046153846153845e-07, "loss": 0.0001, "step": 17850 }, { "epoch": 63.05, "learning_rate": 1.0917948717948718e-07, "loss": 0.0001, "step": 17875 }, { "epoch": 63.14, "learning_rate": 1.078974358974359e-07, "loss": 0.0001, "step": 17900 }, { "epoch": 63.23, "learning_rate": 1.0661538461538461e-07, "loss": 0.0001, "step": 17925 }, { "epoch": 63.32, "learning_rate": 1.0533333333333332e-07, "loss": 0.0001, "step": 17950 }, { "epoch": 63.4, "learning_rate": 1.0405128205128205e-07, "loss": 0.0001, "step": 17975 }, { "epoch": 63.49, "learning_rate": 1.0276923076923076e-07, "loss": 0.0001, "step": 18000 }, { "epoch": 63.49, "eval_loss": 0.47298985719680786, "eval_runtime": 2259.7268, "eval_samples_per_second": 4.015, "eval_steps_per_second": 0.251, "eval_wer": 6.391088620904939, "step": 18000 }, { "epoch": 63.58, "learning_rate": 1.0148717948717948e-07, "loss": 0.0001, "step": 18025 }, { "epoch": 63.67, "learning_rate": 1.0020512820512821e-07, "loss": 0.0001, "step": 18050 }, { "epoch": 63.76, "learning_rate": 9.892307692307692e-08, "loss": 0.0001, "step": 18075 }, { "epoch": 63.84, "learning_rate": 9.764102564102563e-08, "loss": 0.0001, "step": 18100 }, { "epoch": 63.93, "learning_rate": 9.635897435897436e-08, "loss": 0.0001, "step": 18125 }, { "epoch": 64.02, "learning_rate": 9.507692307692307e-08, "loss": 0.0001, "step": 18150 }, { "epoch": 64.11, "learning_rate": 9.37948717948718e-08, "loss": 0.0001, "step": 18175 }, { "epoch": 64.2, "learning_rate": 9.251282051282052e-08, "loss": 0.0001, "step": 18200 }, { "epoch": 64.29, "learning_rate": 9.123076923076922e-08, "loss": 0.0001, "step": 18225 }, { "epoch": 64.37, "learning_rate": 8.994871794871794e-08, "loss": 0.0001, "step": 18250 }, { "epoch": 64.46, "learning_rate": 8.866666666666667e-08, "loss": 0.0001, "step": 18275 }, { "epoch": 64.55, "learning_rate": 8.738461538461538e-08, "loss": 0.0001, "step": 18300 }, { "epoch": 64.64, "learning_rate": 8.61025641025641e-08, "loss": 0.0001, "step": 18325 }, { "epoch": 64.73, "learning_rate": 8.482051282051283e-08, "loss": 0.0001, "step": 18350 }, { "epoch": 64.81, "learning_rate": 8.353846153846153e-08, "loss": 0.0001, "step": 18375 }, { "epoch": 64.9, "learning_rate": 8.225641025641025e-08, "loss": 0.0001, "step": 18400 }, { "epoch": 64.99, "learning_rate": 8.097435897435898e-08, "loss": 0.0001, "step": 18425 }, { "epoch": 65.08, "learning_rate": 7.969230769230769e-08, "loss": 0.0001, "step": 18450 }, { "epoch": 65.17, "learning_rate": 7.841025641025641e-08, "loss": 0.0001, "step": 18475 }, { "epoch": 65.26, "learning_rate": 7.712820512820512e-08, "loss": 0.0001, "step": 18500 }, { "epoch": 65.34, "learning_rate": 7.584615384615384e-08, "loss": 0.0001, "step": 18525 }, { "epoch": 65.43, "learning_rate": 7.456410256410256e-08, "loss": 0.0001, "step": 18550 }, { "epoch": 65.52, "learning_rate": 7.328205128205128e-08, "loss": 0.0001, "step": 18575 }, { "epoch": 65.61, "learning_rate": 7.2e-08, "loss": 0.0001, "step": 18600 }, { "epoch": 65.7, "learning_rate": 7.071794871794871e-08, "loss": 0.0001, "step": 18625 }, { "epoch": 65.78, "learning_rate": 6.943589743589743e-08, "loss": 0.0001, "step": 18650 }, { "epoch": 65.87, "learning_rate": 6.815384615384614e-08, "loss": 0.0001, "step": 18675 }, { "epoch": 65.96, "learning_rate": 6.687179487179487e-08, "loss": 0.0001, "step": 18700 }, { "epoch": 66.05, "learning_rate": 6.55897435897436e-08, "loss": 0.0001, "step": 18725 }, { "epoch": 66.14, "learning_rate": 6.43076923076923e-08, "loss": 0.0001, "step": 18750 }, { "epoch": 66.23, "learning_rate": 6.302564102564102e-08, "loss": 0.0001, "step": 18775 }, { "epoch": 66.31, "learning_rate": 6.174358974358974e-08, "loss": 0.0001, "step": 18800 }, { "epoch": 66.4, "learning_rate": 6.046153846153845e-08, "loss": 0.0001, "step": 18825 }, { "epoch": 66.49, "learning_rate": 5.917948717948718e-08, "loss": 0.0001, "step": 18850 }, { "epoch": 66.58, "learning_rate": 5.7897435897435896e-08, "loss": 0.0001, "step": 18875 }, { "epoch": 66.67, "learning_rate": 5.6615384615384614e-08, "loss": 0.0001, "step": 18900 }, { "epoch": 66.75, "learning_rate": 5.5333333333333326e-08, "loss": 0.0001, "step": 18925 }, { "epoch": 66.84, "learning_rate": 5.405128205128205e-08, "loss": 0.0001, "step": 18950 }, { "epoch": 66.93, "learning_rate": 5.276923076923077e-08, "loss": 0.0001, "step": 18975 }, { "epoch": 67.02, "learning_rate": 5.148717948717948e-08, "loss": 0.0001, "step": 19000 }, { "epoch": 67.02, "eval_loss": 0.47781920433044434, "eval_runtime": 2247.6217, "eval_samples_per_second": 4.036, "eval_steps_per_second": 0.252, "eval_wer": 6.394374527650905, "step": 19000 }, { "epoch": 67.11, "learning_rate": 5.0205128205128205e-08, "loss": 0.0001, "step": 19025 }, { "epoch": 67.2, "learning_rate": 4.8923076923076923e-08, "loss": 0.0001, "step": 19050 }, { "epoch": 67.28, "learning_rate": 4.7641025641025635e-08, "loss": 0.0001, "step": 19075 }, { "epoch": 67.37, "learning_rate": 4.635897435897436e-08, "loss": 0.0001, "step": 19100 }, { "epoch": 67.46, "learning_rate": 4.507692307692308e-08, "loss": 0.0001, "step": 19125 }, { "epoch": 67.55, "learning_rate": 4.379487179487179e-08, "loss": 0.0001, "step": 19150 }, { "epoch": 67.64, "learning_rate": 4.2512820512820514e-08, "loss": 0.0001, "step": 19175 }, { "epoch": 67.72, "learning_rate": 4.1230769230769226e-08, "loss": 0.0001, "step": 19200 }, { "epoch": 67.81, "learning_rate": 3.9948717948717944e-08, "loss": 0.0001, "step": 19225 }, { "epoch": 67.9, "learning_rate": 3.866666666666667e-08, "loss": 0.0001, "step": 19250 }, { "epoch": 67.99, "learning_rate": 3.738461538461538e-08, "loss": 0.0001, "step": 19275 }, { "epoch": 68.08, "learning_rate": 3.61025641025641e-08, "loss": 0.0001, "step": 19300 }, { "epoch": 68.17, "learning_rate": 3.4820512820512823e-08, "loss": 0.0001, "step": 19325 }, { "epoch": 68.25, "learning_rate": 3.3538461538461535e-08, "loss": 0.0001, "step": 19350 }, { "epoch": 68.34, "learning_rate": 3.225641025641025e-08, "loss": 0.0001, "step": 19375 }, { "epoch": 68.43, "learning_rate": 3.097435897435897e-08, "loss": 0.0001, "step": 19400 }, { "epoch": 68.52, "learning_rate": 2.969230769230769e-08, "loss": 0.0001, "step": 19425 }, { "epoch": 68.61, "learning_rate": 2.8410256410256408e-08, "loss": 0.0001, "step": 19450 }, { "epoch": 68.69, "learning_rate": 2.712820512820513e-08, "loss": 0.0001, "step": 19475 }, { "epoch": 68.78, "learning_rate": 2.5846153846153844e-08, "loss": 0.0001, "step": 19500 }, { "epoch": 68.87, "learning_rate": 2.4564102564102562e-08, "loss": 0.0001, "step": 19525 }, { "epoch": 68.96, "learning_rate": 2.3282051282051284e-08, "loss": 0.0001, "step": 19550 }, { "epoch": 69.05, "learning_rate": 2.2e-08, "loss": 0.0001, "step": 19575 }, { "epoch": 69.14, "learning_rate": 2.0717948717948717e-08, "loss": 0.0001, "step": 19600 }, { "epoch": 69.22, "learning_rate": 1.9435897435897435e-08, "loss": 0.0001, "step": 19625 }, { "epoch": 69.31, "learning_rate": 1.8153846153846153e-08, "loss": 0.0001, "step": 19650 }, { "epoch": 69.4, "learning_rate": 1.687179487179487e-08, "loss": 0.0001, "step": 19675 }, { "epoch": 69.49, "learning_rate": 1.558974358974359e-08, "loss": 0.0001, "step": 19700 }, { "epoch": 69.58, "learning_rate": 1.4307692307692308e-08, "loss": 0.0001, "step": 19725 }, { "epoch": 69.66, "learning_rate": 1.3025641025641024e-08, "loss": 0.0001, "step": 19750 }, { "epoch": 69.75, "learning_rate": 1.1743589743589744e-08, "loss": 0.0001, "step": 19775 }, { "epoch": 69.84, "learning_rate": 1.046153846153846e-08, "loss": 0.0001, "step": 19800 }, { "epoch": 69.93, "learning_rate": 9.179487179487179e-09, "loss": 0.0001, "step": 19825 }, { "epoch": 70.02, "learning_rate": 7.897435897435897e-09, "loss": 0.0001, "step": 19850 }, { "epoch": 70.11, "learning_rate": 6.615384615384615e-09, "loss": 0.0001, "step": 19875 }, { "epoch": 70.19, "learning_rate": 5.333333333333333e-09, "loss": 0.0001, "step": 19900 }, { "epoch": 70.28, "learning_rate": 4.051282051282051e-09, "loss": 0.0001, "step": 19925 }, { "epoch": 70.37, "learning_rate": 2.769230769230769e-09, "loss": 0.0001, "step": 19950 }, { "epoch": 70.46, "learning_rate": 1.4871794871794872e-09, "loss": 0.0001, "step": 19975 }, { "epoch": 70.55, "learning_rate": 2.051282051282051e-10, "loss": 0.0001, "step": 20000 }, { "epoch": 70.55, "eval_loss": 0.47987109422683716, "eval_runtime": 2350.7914, "eval_samples_per_second": 3.859, "eval_steps_per_second": 0.241, "eval_wer": 6.399303387769856, "step": 20000 }, { "epoch": 70.55, "step": 20000, "total_flos": 2.717658906624e+21, "train_loss": 0.013972326332342345, "train_runtime": 272612.0522, "train_samples_per_second": 4.695, "train_steps_per_second": 0.073 } ], "logging_steps": 25, "max_steps": 20000, "num_input_tokens_seen": 0, "num_train_epochs": 71, "save_steps": 1000, "total_flos": 2.717658906624e+21, "train_batch_size": 32, "trial_name": null, "trial_params": null }