{ "best_metric": 14.119648426424725, "best_model_checkpoint": "./checkpoint-9000", "epoch": 48.0112, "eval_steps": 1000, "global_step": 10000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.6000000000000004e-07, "loss": 1.3744, "step": 25 }, { "epoch": 0.01, "learning_rate": 9.600000000000001e-07, "loss": 1.0121, "step": 50 }, { "epoch": 0.01, "learning_rate": 1.46e-06, "loss": 0.7739, "step": 75 }, { "epoch": 0.01, "learning_rate": 1.9600000000000003e-06, "loss": 0.62, "step": 100 }, { "epoch": 0.01, "learning_rate": 2.46e-06, "loss": 0.5381, "step": 125 }, { "epoch": 0.01, "learning_rate": 2.96e-06, "loss": 0.4564, "step": 150 }, { "epoch": 0.02, "learning_rate": 3.46e-06, "loss": 0.4146, "step": 175 }, { "epoch": 0.02, "learning_rate": 3.96e-06, "loss": 0.3752, "step": 200 }, { "epoch": 1.0, "learning_rate": 4.4600000000000005e-06, "loss": 0.3814, "step": 225 }, { "epoch": 1.0, "learning_rate": 4.960000000000001e-06, "loss": 0.3298, "step": 250 }, { "epoch": 1.01, "learning_rate": 5.460000000000001e-06, "loss": 0.3343, "step": 275 }, { "epoch": 1.01, "learning_rate": 5.9600000000000005e-06, "loss": 0.2825, "step": 300 }, { "epoch": 1.01, "learning_rate": 6.460000000000001e-06, "loss": 0.2738, "step": 325 }, { "epoch": 1.01, "learning_rate": 6.96e-06, "loss": 0.2276, "step": 350 }, { "epoch": 1.02, "learning_rate": 7.4600000000000006e-06, "loss": 0.211, "step": 375 }, { "epoch": 1.02, "learning_rate": 7.960000000000002e-06, "loss": 0.1846, "step": 400 }, { "epoch": 2.0, "learning_rate": 8.46e-06, "loss": 0.193, "step": 425 }, { "epoch": 2.0, "learning_rate": 8.96e-06, "loss": 0.1732, "step": 450 }, { "epoch": 2.01, "learning_rate": 9.460000000000001e-06, "loss": 0.1702, "step": 475 }, { "epoch": 2.01, "learning_rate": 9.960000000000001e-06, "loss": 0.1527, "step": 500 }, { "epoch": 2.01, "learning_rate": 9.975789473684211e-06, "loss": 0.1392, "step": 525 }, { "epoch": 2.01, "learning_rate": 9.949473684210526e-06, "loss": 0.1128, "step": 550 }, { "epoch": 2.02, "learning_rate": 9.923157894736844e-06, "loss": 0.1036, "step": 575 }, { "epoch": 2.02, "learning_rate": 9.89684210526316e-06, "loss": 0.0852, "step": 600 }, { "epoch": 3.0, "learning_rate": 9.870526315789474e-06, "loss": 0.0886, "step": 625 }, { "epoch": 3.0, "learning_rate": 9.84421052631579e-06, "loss": 0.0781, "step": 650 }, { "epoch": 3.01, "learning_rate": 9.817894736842106e-06, "loss": 0.0745, "step": 675 }, { "epoch": 3.01, "learning_rate": 9.791578947368422e-06, "loss": 0.067, "step": 700 }, { "epoch": 3.01, "learning_rate": 9.765263157894737e-06, "loss": 0.0599, "step": 725 }, { "epoch": 3.01, "learning_rate": 9.738947368421054e-06, "loss": 0.0514, "step": 750 }, { "epoch": 3.02, "learning_rate": 9.712631578947369e-06, "loss": 0.0454, "step": 775 }, { "epoch": 3.02, "learning_rate": 9.686315789473684e-06, "loss": 0.0386, "step": 800 }, { "epoch": 4.0, "learning_rate": 9.66e-06, "loss": 0.0359, "step": 825 }, { "epoch": 4.0, "learning_rate": 9.633684210526316e-06, "loss": 0.0409, "step": 850 }, { "epoch": 4.01, "learning_rate": 9.607368421052632e-06, "loss": 0.0347, "step": 875 }, { "epoch": 4.01, "learning_rate": 9.581052631578947e-06, "loss": 0.0315, "step": 900 }, { "epoch": 4.01, "learning_rate": 9.554736842105264e-06, "loss": 0.027, "step": 925 }, { "epoch": 4.01, "learning_rate": 9.52842105263158e-06, "loss": 0.0281, "step": 950 }, { "epoch": 4.02, "learning_rate": 9.502105263157896e-06, "loss": 0.0222, "step": 975 }, { "epoch": 4.02, "learning_rate": 9.475789473684212e-06, "loss": 0.0206, "step": 1000 }, { "epoch": 4.02, "eval_loss": 0.29976582527160645, "eval_runtime": 1656.4872, "eval_samples_per_second": 3.979, "eval_steps_per_second": 0.124, "eval_wer": 16.999473449714447, "step": 1000 }, { "epoch": 4.02, "learning_rate": 9.449473684210527e-06, "loss": 0.0199, "step": 1025 }, { "epoch": 5.0, "learning_rate": 9.423157894736842e-06, "loss": 0.0209, "step": 1050 }, { "epoch": 5.0, "learning_rate": 9.396842105263159e-06, "loss": 0.0189, "step": 1075 }, { "epoch": 5.01, "learning_rate": 9.370526315789474e-06, "loss": 0.0192, "step": 1100 }, { "epoch": 5.01, "learning_rate": 9.34421052631579e-06, "loss": 0.015, "step": 1125 }, { "epoch": 5.01, "learning_rate": 9.317894736842105e-06, "loss": 0.0174, "step": 1150 }, { "epoch": 5.01, "learning_rate": 9.291578947368422e-06, "loss": 0.0141, "step": 1175 }, { "epoch": 5.02, "learning_rate": 9.265263157894737e-06, "loss": 0.0142, "step": 1200 }, { "epoch": 5.02, "learning_rate": 9.238947368421052e-06, "loss": 0.0116, "step": 1225 }, { "epoch": 6.0, "learning_rate": 9.21263157894737e-06, "loss": 0.0126, "step": 1250 }, { "epoch": 6.0, "learning_rate": 9.186315789473685e-06, "loss": 0.013, "step": 1275 }, { "epoch": 6.01, "learning_rate": 9.16e-06, "loss": 0.0132, "step": 1300 }, { "epoch": 6.01, "learning_rate": 9.133684210526317e-06, "loss": 0.0113, "step": 1325 }, { "epoch": 6.01, "learning_rate": 9.107368421052632e-06, "loss": 0.0108, "step": 1350 }, { "epoch": 6.01, "learning_rate": 9.081052631578949e-06, "loss": 0.0107, "step": 1375 }, { "epoch": 6.02, "learning_rate": 9.054736842105264e-06, "loss": 0.0086, "step": 1400 }, { "epoch": 6.02, "learning_rate": 9.02842105263158e-06, "loss": 0.0079, "step": 1425 }, { "epoch": 7.0, "learning_rate": 9.002105263157895e-06, "loss": 0.0087, "step": 1450 }, { "epoch": 7.0, "learning_rate": 8.97578947368421e-06, "loss": 0.0073, "step": 1475 }, { "epoch": 7.01, "learning_rate": 8.949473684210527e-06, "loss": 0.0093, "step": 1500 }, { "epoch": 7.01, "learning_rate": 8.923157894736842e-06, "loss": 0.0072, "step": 1525 }, { "epoch": 7.01, "learning_rate": 8.896842105263159e-06, "loss": 0.0081, "step": 1550 }, { "epoch": 7.01, "learning_rate": 8.870526315789474e-06, "loss": 0.0083, "step": 1575 }, { "epoch": 7.02, "learning_rate": 8.84421052631579e-06, "loss": 0.0063, "step": 1600 }, { "epoch": 7.02, "learning_rate": 8.817894736842107e-06, "loss": 0.0074, "step": 1625 }, { "epoch": 8.0, "learning_rate": 8.791578947368422e-06, "loss": 0.0071, "step": 1650 }, { "epoch": 8.0, "learning_rate": 8.765263157894739e-06, "loss": 0.0071, "step": 1675 }, { "epoch": 8.01, "learning_rate": 8.738947368421053e-06, "loss": 0.0064, "step": 1700 }, { "epoch": 8.01, "learning_rate": 8.712631578947368e-06, "loss": 0.006, "step": 1725 }, { "epoch": 8.01, "learning_rate": 8.686315789473685e-06, "loss": 0.005, "step": 1750 }, { "epoch": 8.01, "learning_rate": 8.66e-06, "loss": 0.0055, "step": 1775 }, { "epoch": 8.02, "learning_rate": 8.633684210526317e-06, "loss": 0.0042, "step": 1800 }, { "epoch": 8.02, "learning_rate": 8.607368421052632e-06, "loss": 0.0051, "step": 1825 }, { "epoch": 8.02, "learning_rate": 8.581052631578948e-06, "loss": 0.0048, "step": 1850 }, { "epoch": 9.0, "learning_rate": 8.554736842105263e-06, "loss": 0.006, "step": 1875 }, { "epoch": 9.0, "learning_rate": 8.528421052631578e-06, "loss": 0.0052, "step": 1900 }, { "epoch": 9.01, "learning_rate": 8.502105263157897e-06, "loss": 0.0047, "step": 1925 }, { "epoch": 9.01, "learning_rate": 8.475789473684212e-06, "loss": 0.0044, "step": 1950 }, { "epoch": 9.01, "learning_rate": 8.449473684210527e-06, "loss": 0.0049, "step": 1975 }, { "epoch": 9.01, "learning_rate": 8.423157894736843e-06, "loss": 0.0036, "step": 2000 }, { "epoch": 9.01, "eval_loss": 0.32349300384521484, "eval_runtime": 1644.0636, "eval_samples_per_second": 4.009, "eval_steps_per_second": 0.125, "eval_wer": 15.521082263356151, "step": 2000 }, { "epoch": 9.02, "learning_rate": 8.396842105263158e-06, "loss": 0.0039, "step": 2025 }, { "epoch": 9.02, "learning_rate": 8.370526315789475e-06, "loss": 0.0041, "step": 2050 }, { "epoch": 10.0, "learning_rate": 8.34421052631579e-06, "loss": 0.004, "step": 2075 }, { "epoch": 10.0, "learning_rate": 8.317894736842107e-06, "loss": 0.0033, "step": 2100 }, { "epoch": 10.01, "learning_rate": 8.291578947368422e-06, "loss": 0.0044, "step": 2125 }, { "epoch": 10.01, "learning_rate": 8.265263157894737e-06, "loss": 0.0034, "step": 2150 }, { "epoch": 10.01, "learning_rate": 8.238947368421053e-06, "loss": 0.0044, "step": 2175 }, { "epoch": 10.01, "learning_rate": 8.212631578947368e-06, "loss": 0.0037, "step": 2200 }, { "epoch": 10.02, "learning_rate": 8.186315789473685e-06, "loss": 0.0036, "step": 2225 }, { "epoch": 10.02, "learning_rate": 8.16e-06, "loss": 0.0033, "step": 2250 }, { "epoch": 11.0, "learning_rate": 8.133684210526316e-06, "loss": 0.0033, "step": 2275 }, { "epoch": 11.0, "learning_rate": 8.107368421052633e-06, "loss": 0.0042, "step": 2300 }, { "epoch": 11.01, "learning_rate": 8.081052631578948e-06, "loss": 0.0037, "step": 2325 }, { "epoch": 11.01, "learning_rate": 8.054736842105265e-06, "loss": 0.0042, "step": 2350 }, { "epoch": 11.01, "learning_rate": 8.02842105263158e-06, "loss": 0.0042, "step": 2375 }, { "epoch": 11.01, "learning_rate": 8.002105263157895e-06, "loss": 0.0039, "step": 2400 }, { "epoch": 11.02, "learning_rate": 7.975789473684211e-06, "loss": 0.0036, "step": 2425 }, { "epoch": 11.02, "learning_rate": 7.949473684210526e-06, "loss": 0.0031, "step": 2450 }, { "epoch": 12.0, "learning_rate": 7.923157894736843e-06, "loss": 0.0039, "step": 2475 }, { "epoch": 12.0, "learning_rate": 7.896842105263158e-06, "loss": 0.0035, "step": 2500 }, { "epoch": 12.01, "learning_rate": 7.870526315789475e-06, "loss": 0.0029, "step": 2525 }, { "epoch": 12.01, "learning_rate": 7.84421052631579e-06, "loss": 0.0028, "step": 2550 }, { "epoch": 12.01, "learning_rate": 7.817894736842105e-06, "loss": 0.0032, "step": 2575 }, { "epoch": 12.01, "learning_rate": 7.791578947368423e-06, "loss": 0.0029, "step": 2600 }, { "epoch": 12.02, "learning_rate": 7.765263157894738e-06, "loss": 0.003, "step": 2625 }, { "epoch": 12.02, "learning_rate": 7.738947368421053e-06, "loss": 0.0024, "step": 2650 }, { "epoch": 12.02, "learning_rate": 7.71263157894737e-06, "loss": 0.0028, "step": 2675 }, { "epoch": 13.0, "learning_rate": 7.686315789473685e-06, "loss": 0.0032, "step": 2700 }, { "epoch": 13.0, "learning_rate": 7.660000000000001e-06, "loss": 0.0027, "step": 2725 }, { "epoch": 13.01, "learning_rate": 7.633684210526316e-06, "loss": 0.0031, "step": 2750 }, { "epoch": 13.01, "learning_rate": 7.607368421052632e-06, "loss": 0.0025, "step": 2775 }, { "epoch": 13.01, "learning_rate": 7.581052631578948e-06, "loss": 0.0025, "step": 2800 }, { "epoch": 13.01, "learning_rate": 7.554736842105264e-06, "loss": 0.0021, "step": 2825 }, { "epoch": 13.02, "learning_rate": 7.5284210526315794e-06, "loss": 0.0022, "step": 2850 }, { "epoch": 13.02, "learning_rate": 7.502105263157895e-06, "loss": 0.0017, "step": 2875 }, { "epoch": 14.0, "learning_rate": 7.475789473684211e-06, "loss": 0.0025, "step": 2900 }, { "epoch": 14.0, "learning_rate": 7.449473684210526e-06, "loss": 0.0027, "step": 2925 }, { "epoch": 14.01, "learning_rate": 7.4231578947368436e-06, "loss": 0.0031, "step": 2950 }, { "epoch": 14.01, "learning_rate": 7.3968421052631585e-06, "loss": 0.0021, "step": 2975 }, { "epoch": 14.01, "learning_rate": 7.370526315789474e-06, "loss": 0.0018, "step": 3000 }, { "epoch": 14.01, "eval_loss": 0.3454221785068512, "eval_runtime": 1676.4237, "eval_samples_per_second": 3.932, "eval_steps_per_second": 0.123, "eval_wer": 14.990481590991939, "step": 3000 }, { "epoch": 14.01, "learning_rate": 7.34421052631579e-06, "loss": 0.0024, "step": 3025 }, { "epoch": 14.02, "learning_rate": 7.317894736842106e-06, "loss": 0.003, "step": 3050 }, { "epoch": 14.02, "learning_rate": 7.291578947368422e-06, "loss": 0.0026, "step": 3075 }, { "epoch": 15.0, "learning_rate": 7.265263157894738e-06, "loss": 0.0027, "step": 3100 }, { "epoch": 15.0, "learning_rate": 7.2389473684210534e-06, "loss": 0.0013, "step": 3125 }, { "epoch": 15.01, "learning_rate": 7.212631578947369e-06, "loss": 0.0026, "step": 3150 }, { "epoch": 15.01, "learning_rate": 7.186315789473684e-06, "loss": 0.0023, "step": 3175 }, { "epoch": 15.01, "learning_rate": 7.16e-06, "loss": 0.0023, "step": 3200 }, { "epoch": 15.01, "learning_rate": 7.133684210526316e-06, "loss": 0.0029, "step": 3225 }, { "epoch": 15.02, "learning_rate": 7.107368421052632e-06, "loss": 0.0021, "step": 3250 }, { "epoch": 15.02, "learning_rate": 7.0810526315789475e-06, "loss": 0.0022, "step": 3275 }, { "epoch": 16.0, "learning_rate": 7.054736842105264e-06, "loss": 0.0023, "step": 3300 }, { "epoch": 16.0, "learning_rate": 7.02842105263158e-06, "loss": 0.002, "step": 3325 }, { "epoch": 16.01, "learning_rate": 7.002105263157896e-06, "loss": 0.003, "step": 3350 }, { "epoch": 16.01, "learning_rate": 6.975789473684212e-06, "loss": 0.0018, "step": 3375 }, { "epoch": 16.01, "learning_rate": 6.9494736842105275e-06, "loss": 0.0015, "step": 3400 }, { "epoch": 16.01, "learning_rate": 6.9231578947368424e-06, "loss": 0.0019, "step": 3425 }, { "epoch": 16.02, "learning_rate": 6.896842105263158e-06, "loss": 0.0022, "step": 3450 }, { "epoch": 16.02, "learning_rate": 6.870526315789474e-06, "loss": 0.0016, "step": 3475 }, { "epoch": 16.02, "learning_rate": 6.84421052631579e-06, "loss": 0.0013, "step": 3500 }, { "epoch": 17.0, "learning_rate": 6.817894736842106e-06, "loss": 0.0018, "step": 3525 }, { "epoch": 17.0, "learning_rate": 6.7915789473684215e-06, "loss": 0.001, "step": 3550 }, { "epoch": 17.01, "learning_rate": 6.765263157894737e-06, "loss": 0.0014, "step": 3575 }, { "epoch": 17.01, "learning_rate": 6.738947368421052e-06, "loss": 0.0011, "step": 3600 }, { "epoch": 17.01, "learning_rate": 6.71263157894737e-06, "loss": 0.0016, "step": 3625 }, { "epoch": 17.01, "learning_rate": 6.686315789473685e-06, "loss": 0.0015, "step": 3650 }, { "epoch": 17.02, "learning_rate": 6.660000000000001e-06, "loss": 0.0018, "step": 3675 }, { "epoch": 17.02, "learning_rate": 6.6336842105263164e-06, "loss": 0.0019, "step": 3700 }, { "epoch": 18.0, "learning_rate": 6.607368421052632e-06, "loss": 0.0015, "step": 3725 }, { "epoch": 18.0, "learning_rate": 6.581052631578948e-06, "loss": 0.0015, "step": 3750 }, { "epoch": 18.01, "learning_rate": 6.554736842105264e-06, "loss": 0.0013, "step": 3775 }, { "epoch": 18.01, "learning_rate": 6.52842105263158e-06, "loss": 0.0012, "step": 3800 }, { "epoch": 18.01, "learning_rate": 6.5021052631578955e-06, "loss": 0.0014, "step": 3825 }, { "epoch": 18.01, "learning_rate": 6.4757894736842105e-06, "loss": 0.0013, "step": 3850 }, { "epoch": 18.02, "learning_rate": 6.449473684210526e-06, "loss": 0.0013, "step": 3875 }, { "epoch": 18.02, "learning_rate": 6.423157894736842e-06, "loss": 0.0015, "step": 3900 }, { "epoch": 19.0, "learning_rate": 6.396842105263158e-06, "loss": 0.0013, "step": 3925 }, { "epoch": 19.0, "learning_rate": 6.370526315789474e-06, "loss": 0.0013, "step": 3950 }, { "epoch": 19.01, "learning_rate": 6.3442105263157904e-06, "loss": 0.0013, "step": 3975 }, { "epoch": 19.01, "learning_rate": 6.317894736842106e-06, "loss": 0.0013, "step": 4000 }, { "epoch": 19.01, "eval_loss": 0.3537706434726715, "eval_runtime": 1650.951, "eval_samples_per_second": 3.992, "eval_steps_per_second": 0.125, "eval_wer": 14.943902142654625, "step": 4000 }, { "epoch": 19.01, "learning_rate": 6.291578947368422e-06, "loss": 0.0016, "step": 4025 }, { "epoch": 19.01, "learning_rate": 6.265263157894738e-06, "loss": 0.0014, "step": 4050 }, { "epoch": 19.02, "learning_rate": 6.238947368421054e-06, "loss": 0.0007, "step": 4075 }, { "epoch": 19.02, "learning_rate": 6.212631578947369e-06, "loss": 0.001, "step": 4100 }, { "epoch": 20.0, "learning_rate": 6.1863157894736845e-06, "loss": 0.001, "step": 4125 }, { "epoch": 20.0, "learning_rate": 6.16e-06, "loss": 0.0006, "step": 4150 }, { "epoch": 20.01, "learning_rate": 6.133684210526316e-06, "loss": 0.0013, "step": 4175 }, { "epoch": 20.01, "learning_rate": 6.107368421052632e-06, "loss": 0.0008, "step": 4200 }, { "epoch": 20.01, "learning_rate": 6.081052631578948e-06, "loss": 0.0015, "step": 4225 }, { "epoch": 20.01, "learning_rate": 6.054736842105264e-06, "loss": 0.0006, "step": 4250 }, { "epoch": 20.02, "learning_rate": 6.0284210526315786e-06, "loss": 0.0008, "step": 4275 }, { "epoch": 20.02, "learning_rate": 6.002105263157896e-06, "loss": 0.0011, "step": 4300 }, { "epoch": 20.02, "learning_rate": 5.975789473684212e-06, "loss": 0.0008, "step": 4325 }, { "epoch": 21.0, "learning_rate": 5.949473684210527e-06, "loss": 0.0007, "step": 4350 }, { "epoch": 21.0, "learning_rate": 5.923157894736843e-06, "loss": 0.0008, "step": 4375 }, { "epoch": 21.01, "learning_rate": 5.8968421052631585e-06, "loss": 0.0007, "step": 4400 }, { "epoch": 21.01, "learning_rate": 5.870526315789474e-06, "loss": 0.0006, "step": 4425 }, { "epoch": 21.01, "learning_rate": 5.84421052631579e-06, "loss": 0.0009, "step": 4450 }, { "epoch": 21.01, "learning_rate": 5.817894736842106e-06, "loss": 0.001, "step": 4475 }, { "epoch": 21.02, "learning_rate": 5.791578947368422e-06, "loss": 0.0012, "step": 4500 }, { "epoch": 21.02, "learning_rate": 5.765263157894737e-06, "loss": 0.0008, "step": 4525 }, { "epoch": 22.0, "learning_rate": 5.7389473684210526e-06, "loss": 0.0009, "step": 4550 }, { "epoch": 22.0, "learning_rate": 5.712631578947368e-06, "loss": 0.0009, "step": 4575 }, { "epoch": 22.01, "learning_rate": 5.686315789473684e-06, "loss": 0.0014, "step": 4600 }, { "epoch": 22.01, "learning_rate": 5.66e-06, "loss": 0.0009, "step": 4625 }, { "epoch": 22.01, "learning_rate": 5.633684210526317e-06, "loss": 0.0007, "step": 4650 }, { "epoch": 22.01, "learning_rate": 5.6073684210526325e-06, "loss": 0.0007, "step": 4675 }, { "epoch": 22.02, "learning_rate": 5.581052631578948e-06, "loss": 0.0014, "step": 4700 }, { "epoch": 22.02, "learning_rate": 5.554736842105264e-06, "loss": 0.0008, "step": 4725 }, { "epoch": 23.0, "learning_rate": 5.52842105263158e-06, "loss": 0.0014, "step": 4750 }, { "epoch": 23.0, "learning_rate": 5.502105263157895e-06, "loss": 0.0013, "step": 4775 }, { "epoch": 23.01, "learning_rate": 5.475789473684211e-06, "loss": 0.0013, "step": 4800 }, { "epoch": 23.01, "learning_rate": 5.4494736842105266e-06, "loss": 0.0009, "step": 4825 }, { "epoch": 23.01, "learning_rate": 5.423157894736842e-06, "loss": 0.0014, "step": 4850 }, { "epoch": 23.01, "learning_rate": 5.396842105263158e-06, "loss": 0.0013, "step": 4875 }, { "epoch": 23.02, "learning_rate": 5.370526315789474e-06, "loss": 0.0018, "step": 4900 }, { "epoch": 23.02, "learning_rate": 5.34421052631579e-06, "loss": 0.0021, "step": 4925 }, { "epoch": 24.0, "learning_rate": 5.317894736842105e-06, "loss": 0.0021, "step": 4950 }, { "epoch": 24.0, "learning_rate": 5.291578947368422e-06, "loss": 0.0015, "step": 4975 }, { "epoch": 24.01, "learning_rate": 5.265263157894738e-06, "loss": 0.0013, "step": 5000 }, { "epoch": 24.01, "eval_loss": 0.35868650674819946, "eval_runtime": 1674.0389, "eval_samples_per_second": 3.937, "eval_steps_per_second": 0.123, "eval_wer": 14.856818826197902, "step": 5000 }, { "epoch": 24.01, "learning_rate": 5.238947368421053e-06, "loss": 0.0019, "step": 5025 }, { "epoch": 24.01, "learning_rate": 5.212631578947369e-06, "loss": 0.0013, "step": 5050 }, { "epoch": 24.01, "learning_rate": 5.186315789473685e-06, "loss": 0.0015, "step": 5075 }, { "epoch": 24.02, "learning_rate": 5.1600000000000006e-06, "loss": 0.0024, "step": 5100 }, { "epoch": 24.02, "learning_rate": 5.133684210526316e-06, "loss": 0.002, "step": 5125 }, { "epoch": 24.02, "learning_rate": 5.107368421052632e-06, "loss": 0.0028, "step": 5150 }, { "epoch": 25.0, "learning_rate": 5.081052631578948e-06, "loss": 0.0012, "step": 5175 }, { "epoch": 25.0, "learning_rate": 5.054736842105263e-06, "loss": 0.0015, "step": 5200 }, { "epoch": 25.01, "learning_rate": 5.028421052631579e-06, "loss": 0.0013, "step": 5225 }, { "epoch": 25.01, "learning_rate": 5.002105263157895e-06, "loss": 0.0021, "step": 5250 }, { "epoch": 25.01, "learning_rate": 4.975789473684211e-06, "loss": 0.0013, "step": 5275 }, { "epoch": 25.02, "learning_rate": 4.949473684210527e-06, "loss": 0.0017, "step": 5300 }, { "epoch": 25.02, "learning_rate": 4.923157894736842e-06, "loss": 0.0008, "step": 5325 }, { "epoch": 25.02, "learning_rate": 4.896842105263158e-06, "loss": 0.0014, "step": 5350 }, { "epoch": 26.0, "learning_rate": 4.870526315789474e-06, "loss": 0.001, "step": 5375 }, { "epoch": 26.0, "learning_rate": 4.84421052631579e-06, "loss": 0.0016, "step": 5400 }, { "epoch": 26.01, "learning_rate": 4.817894736842106e-06, "loss": 0.0011, "step": 5425 }, { "epoch": 26.01, "learning_rate": 4.791578947368421e-06, "loss": 0.001, "step": 5450 }, { "epoch": 26.01, "learning_rate": 4.765263157894737e-06, "loss": 0.001, "step": 5475 }, { "epoch": 26.01, "learning_rate": 4.738947368421053e-06, "loss": 0.0013, "step": 5500 }, { "epoch": 26.02, "learning_rate": 4.712631578947369e-06, "loss": 0.0011, "step": 5525 }, { "epoch": 26.02, "learning_rate": 4.6863157894736845e-06, "loss": 0.0004, "step": 5550 }, { "epoch": 27.0, "learning_rate": 4.66e-06, "loss": 0.0008, "step": 5575 }, { "epoch": 27.0, "learning_rate": 4.633684210526316e-06, "loss": 0.0008, "step": 5600 }, { "epoch": 27.01, "learning_rate": 4.607368421052632e-06, "loss": 0.0004, "step": 5625 }, { "epoch": 27.01, "learning_rate": 4.581052631578948e-06, "loss": 0.0005, "step": 5650 }, { "epoch": 27.01, "learning_rate": 4.5547368421052636e-06, "loss": 0.0008, "step": 5675 }, { "epoch": 27.01, "learning_rate": 4.528421052631579e-06, "loss": 0.0004, "step": 5700 }, { "epoch": 27.02, "learning_rate": 4.502105263157895e-06, "loss": 0.0003, "step": 5725 }, { "epoch": 27.02, "learning_rate": 4.475789473684211e-06, "loss": 0.0005, "step": 5750 }, { "epoch": 28.0, "learning_rate": 4.449473684210527e-06, "loss": 0.0002, "step": 5775 }, { "epoch": 28.0, "learning_rate": 4.423157894736843e-06, "loss": 0.0002, "step": 5800 }, { "epoch": 28.01, "learning_rate": 4.3968421052631585e-06, "loss": 0.0002, "step": 5825 }, { "epoch": 28.01, "learning_rate": 4.370526315789474e-06, "loss": 0.0002, "step": 5850 }, { "epoch": 28.01, "learning_rate": 4.344210526315789e-06, "loss": 0.0003, "step": 5875 }, { "epoch": 28.01, "learning_rate": 4.317894736842105e-06, "loss": 0.0002, "step": 5900 }, { "epoch": 28.02, "learning_rate": 4.291578947368422e-06, "loss": 0.0003, "step": 5925 }, { "epoch": 28.02, "learning_rate": 4.2652631578947376e-06, "loss": 0.0002, "step": 5950 }, { "epoch": 29.0, "learning_rate": 4.238947368421053e-06, "loss": 0.0002, "step": 5975 }, { "epoch": 29.0, "learning_rate": 4.212631578947368e-06, "loss": 0.0002, "step": 6000 }, { "epoch": 29.0, "eval_loss": 0.37993738055229187, "eval_runtime": 1654.4478, "eval_samples_per_second": 3.984, "eval_steps_per_second": 0.125, "eval_wer": 14.415326663696382, "step": 6000 }, { "epoch": 29.01, "learning_rate": 4.186315789473684e-06, "loss": 0.0004, "step": 6025 }, { "epoch": 29.01, "learning_rate": 4.16e-06, "loss": 0.0004, "step": 6050 }, { "epoch": 29.01, "learning_rate": 4.133684210526316e-06, "loss": 0.0002, "step": 6075 }, { "epoch": 29.01, "learning_rate": 4.1073684210526325e-06, "loss": 0.0005, "step": 6100 }, { "epoch": 29.02, "learning_rate": 4.0810526315789474e-06, "loss": 0.0005, "step": 6125 }, { "epoch": 29.02, "learning_rate": 4.054736842105263e-06, "loss": 0.0003, "step": 6150 }, { "epoch": 29.02, "learning_rate": 4.028421052631579e-06, "loss": 0.0002, "step": 6175 }, { "epoch": 30.0, "learning_rate": 4.002105263157895e-06, "loss": 0.0004, "step": 6200 }, { "epoch": 30.0, "learning_rate": 3.975789473684211e-06, "loss": 0.0004, "step": 6225 }, { "epoch": 30.01, "learning_rate": 3.9494736842105265e-06, "loss": 0.0002, "step": 6250 }, { "epoch": 30.01, "learning_rate": 3.923157894736842e-06, "loss": 0.0002, "step": 6275 }, { "epoch": 30.01, "learning_rate": 3.896842105263158e-06, "loss": 0.0003, "step": 6300 }, { "epoch": 30.01, "learning_rate": 3.870526315789474e-06, "loss": 0.0002, "step": 6325 }, { "epoch": 30.02, "learning_rate": 3.84421052631579e-06, "loss": 0.0001, "step": 6350 }, { "epoch": 30.02, "learning_rate": 3.817894736842106e-06, "loss": 0.0001, "step": 6375 }, { "epoch": 31.0, "learning_rate": 3.791578947368421e-06, "loss": 0.0006, "step": 6400 }, { "epoch": 31.0, "learning_rate": 3.7652631578947373e-06, "loss": 0.0002, "step": 6425 }, { "epoch": 31.01, "learning_rate": 3.738947368421053e-06, "loss": 0.0009, "step": 6450 }, { "epoch": 31.01, "learning_rate": 3.712631578947369e-06, "loss": 0.0002, "step": 6475 }, { "epoch": 31.01, "learning_rate": 3.6863157894736847e-06, "loss": 0.0002, "step": 6500 }, { "epoch": 31.01, "learning_rate": 3.66e-06, "loss": 0.0002, "step": 6525 }, { "epoch": 31.02, "learning_rate": 3.633684210526316e-06, "loss": 0.0002, "step": 6550 }, { "epoch": 31.02, "learning_rate": 3.6073684210526317e-06, "loss": 0.0001, "step": 6575 }, { "epoch": 32.0, "learning_rate": 3.581052631578948e-06, "loss": 0.0002, "step": 6600 }, { "epoch": 32.0, "learning_rate": 3.554736842105264e-06, "loss": 0.0001, "step": 6625 }, { "epoch": 32.01, "learning_rate": 3.528421052631579e-06, "loss": 0.0001, "step": 6650 }, { "epoch": 32.01, "learning_rate": 3.502105263157895e-06, "loss": 0.0001, "step": 6675 }, { "epoch": 32.01, "learning_rate": 3.475789473684211e-06, "loss": 0.0001, "step": 6700 }, { "epoch": 32.01, "learning_rate": 3.4494736842105262e-06, "loss": 0.0001, "step": 6725 }, { "epoch": 32.02, "learning_rate": 3.423157894736842e-06, "loss": 0.0001, "step": 6750 }, { "epoch": 32.02, "learning_rate": 3.3968421052631583e-06, "loss": 0.0001, "step": 6775 }, { "epoch": 33.0, "learning_rate": 3.370526315789474e-06, "loss": 0.0001, "step": 6800 }, { "epoch": 33.0, "learning_rate": 3.34421052631579e-06, "loss": 0.0001, "step": 6825 }, { "epoch": 33.01, "learning_rate": 3.3178947368421053e-06, "loss": 0.0001, "step": 6850 }, { "epoch": 33.01, "learning_rate": 3.291578947368421e-06, "loss": 0.0001, "step": 6875 }, { "epoch": 33.01, "learning_rate": 3.265263157894737e-06, "loss": 0.0001, "step": 6900 }, { "epoch": 33.01, "learning_rate": 3.238947368421053e-06, "loss": 0.0001, "step": 6925 }, { "epoch": 33.02, "learning_rate": 3.212631578947369e-06, "loss": 0.0001, "step": 6950 }, { "epoch": 33.02, "learning_rate": 3.1863157894736844e-06, "loss": 0.0001, "step": 6975 }, { "epoch": 33.02, "learning_rate": 3.1600000000000002e-06, "loss": 0.0001, "step": 7000 }, { "epoch": 33.02, "eval_loss": 0.393667072057724, "eval_runtime": 1644.8876, "eval_samples_per_second": 4.007, "eval_steps_per_second": 0.125, "eval_wer": 14.206731742881445, "step": 7000 }, { "epoch": 34.0, "learning_rate": 3.133684210526316e-06, "loss": 0.0001, "step": 7025 }, { "epoch": 34.0, "learning_rate": 3.107368421052632e-06, "loss": 0.0001, "step": 7050 }, { "epoch": 34.01, "learning_rate": 3.0810526315789473e-06, "loss": 0.0001, "step": 7075 }, { "epoch": 34.01, "learning_rate": 3.0547368421052635e-06, "loss": 0.0001, "step": 7100 }, { "epoch": 34.01, "learning_rate": 3.0284210526315793e-06, "loss": 0.0001, "step": 7125 }, { "epoch": 34.01, "learning_rate": 3.002105263157895e-06, "loss": 0.0001, "step": 7150 }, { "epoch": 34.02, "learning_rate": 2.975789473684211e-06, "loss": 0.0001, "step": 7175 }, { "epoch": 34.02, "learning_rate": 2.9494736842105264e-06, "loss": 0.0001, "step": 7200 }, { "epoch": 35.0, "learning_rate": 2.923157894736842e-06, "loss": 0.0001, "step": 7225 }, { "epoch": 35.0, "learning_rate": 2.896842105263158e-06, "loss": 0.0001, "step": 7250 }, { "epoch": 35.01, "learning_rate": 2.8705263157894742e-06, "loss": 0.0001, "step": 7275 }, { "epoch": 35.01, "learning_rate": 2.84421052631579e-06, "loss": 0.0001, "step": 7300 }, { "epoch": 35.01, "learning_rate": 2.8178947368421055e-06, "loss": 0.0001, "step": 7325 }, { "epoch": 35.01, "learning_rate": 2.7915789473684213e-06, "loss": 0.0001, "step": 7350 }, { "epoch": 35.02, "learning_rate": 2.765263157894737e-06, "loss": 0.0001, "step": 7375 }, { "epoch": 35.02, "learning_rate": 2.7389473684210525e-06, "loss": 0.0001, "step": 7400 }, { "epoch": 36.0, "learning_rate": 2.7126315789473683e-06, "loss": 0.0001, "step": 7425 }, { "epoch": 36.0, "learning_rate": 2.6863157894736845e-06, "loss": 0.0001, "step": 7450 }, { "epoch": 36.01, "learning_rate": 2.6600000000000004e-06, "loss": 0.0001, "step": 7475 }, { "epoch": 36.01, "learning_rate": 2.633684210526316e-06, "loss": 0.0001, "step": 7500 }, { "epoch": 36.01, "learning_rate": 2.6073684210526316e-06, "loss": 0.0001, "step": 7525 }, { "epoch": 36.01, "learning_rate": 2.5810526315789474e-06, "loss": 0.0001, "step": 7550 }, { "epoch": 36.02, "learning_rate": 2.5547368421052632e-06, "loss": 0.0001, "step": 7575 }, { "epoch": 36.02, "learning_rate": 2.528421052631579e-06, "loss": 0.0001, "step": 7600 }, { "epoch": 37.0, "learning_rate": 2.5021052631578953e-06, "loss": 0.0001, "step": 7625 }, { "epoch": 37.0, "learning_rate": 2.4757894736842107e-06, "loss": 0.0001, "step": 7650 }, { "epoch": 37.01, "learning_rate": 2.4494736842105265e-06, "loss": 0.0001, "step": 7675 }, { "epoch": 37.01, "learning_rate": 2.4231578947368423e-06, "loss": 0.0001, "step": 7700 }, { "epoch": 37.01, "learning_rate": 2.396842105263158e-06, "loss": 0.0001, "step": 7725 }, { "epoch": 37.01, "learning_rate": 2.370526315789474e-06, "loss": 0.0001, "step": 7750 }, { "epoch": 37.02, "learning_rate": 2.3442105263157898e-06, "loss": 0.0001, "step": 7775 }, { "epoch": 37.02, "learning_rate": 2.317894736842105e-06, "loss": 0.0001, "step": 7800 }, { "epoch": 37.02, "learning_rate": 2.2915789473684214e-06, "loss": 0.0001, "step": 7825 }, { "epoch": 38.0, "learning_rate": 2.2652631578947372e-06, "loss": 0.0001, "step": 7850 }, { "epoch": 38.0, "learning_rate": 2.2389473684210526e-06, "loss": 0.0001, "step": 7875 }, { "epoch": 38.01, "learning_rate": 2.2126315789473684e-06, "loss": 0.0001, "step": 7900 }, { "epoch": 38.01, "learning_rate": 2.1863157894736843e-06, "loss": 0.0001, "step": 7925 }, { "epoch": 38.01, "learning_rate": 2.16e-06, "loss": 0.0001, "step": 7950 }, { "epoch": 38.01, "learning_rate": 2.133684210526316e-06, "loss": 0.0001, "step": 7975 }, { "epoch": 38.02, "learning_rate": 2.1073684210526317e-06, "loss": 0.0001, "step": 8000 }, { "epoch": 38.02, "eval_loss": 0.40499618649482727, "eval_runtime": 1658.1493, "eval_samples_per_second": 3.975, "eval_steps_per_second": 0.124, "eval_wer": 14.194580582445623, "step": 8000 }, { "epoch": 38.02, "learning_rate": 2.0810526315789475e-06, "loss": 0.0001, "step": 8025 }, { "epoch": 39.0, "learning_rate": 2.0547368421052633e-06, "loss": 0.0001, "step": 8050 }, { "epoch": 39.0, "learning_rate": 2.028421052631579e-06, "loss": 0.0001, "step": 8075 }, { "epoch": 39.01, "learning_rate": 2.002105263157895e-06, "loss": 0.0001, "step": 8100 }, { "epoch": 39.01, "learning_rate": 1.975789473684211e-06, "loss": 0.0001, "step": 8125 }, { "epoch": 39.01, "learning_rate": 1.949473684210526e-06, "loss": 0.0001, "step": 8150 }, { "epoch": 39.01, "learning_rate": 1.9231578947368424e-06, "loss": 0.0001, "step": 8175 }, { "epoch": 39.02, "learning_rate": 1.896842105263158e-06, "loss": 0.0001, "step": 8200 }, { "epoch": 39.02, "learning_rate": 1.8705263157894737e-06, "loss": 0.0001, "step": 8225 }, { "epoch": 40.0, "learning_rate": 1.8442105263157897e-06, "loss": 0.0001, "step": 8250 }, { "epoch": 40.0, "learning_rate": 1.8178947368421055e-06, "loss": 0.0001, "step": 8275 }, { "epoch": 40.01, "learning_rate": 1.7915789473684211e-06, "loss": 0.0001, "step": 8300 }, { "epoch": 40.01, "learning_rate": 1.7652631578947371e-06, "loss": 0.0001, "step": 8325 }, { "epoch": 40.01, "learning_rate": 1.7389473684210527e-06, "loss": 0.0001, "step": 8350 }, { "epoch": 40.01, "learning_rate": 1.7126315789473686e-06, "loss": 0.0001, "step": 8375 }, { "epoch": 40.02, "learning_rate": 1.6863157894736842e-06, "loss": 0.0001, "step": 8400 }, { "epoch": 40.02, "learning_rate": 1.6600000000000002e-06, "loss": 0.0001, "step": 8425 }, { "epoch": 41.0, "learning_rate": 1.633684210526316e-06, "loss": 0.0001, "step": 8450 }, { "epoch": 41.0, "learning_rate": 1.6073684210526316e-06, "loss": 0.0001, "step": 8475 }, { "epoch": 41.01, "learning_rate": 1.5810526315789477e-06, "loss": 0.0001, "step": 8500 }, { "epoch": 41.01, "learning_rate": 1.5547368421052633e-06, "loss": 0.0001, "step": 8525 }, { "epoch": 41.01, "learning_rate": 1.528421052631579e-06, "loss": 0.0001, "step": 8550 }, { "epoch": 41.01, "learning_rate": 1.5021052631578947e-06, "loss": 0.0001, "step": 8575 }, { "epoch": 41.02, "learning_rate": 1.4757894736842107e-06, "loss": 0.0001, "step": 8600 }, { "epoch": 41.02, "learning_rate": 1.4494736842105263e-06, "loss": 0.0001, "step": 8625 }, { "epoch": 41.02, "learning_rate": 1.4231578947368421e-06, "loss": 0.0001, "step": 8650 }, { "epoch": 42.0, "learning_rate": 1.3968421052631582e-06, "loss": 0.0001, "step": 8675 }, { "epoch": 42.0, "learning_rate": 1.3705263157894738e-06, "loss": 0.0001, "step": 8700 }, { "epoch": 42.01, "learning_rate": 1.3442105263157896e-06, "loss": 0.0001, "step": 8725 }, { "epoch": 42.01, "learning_rate": 1.3178947368421054e-06, "loss": 0.0001, "step": 8750 }, { "epoch": 42.01, "learning_rate": 1.2915789473684212e-06, "loss": 0.0001, "step": 8775 }, { "epoch": 42.01, "learning_rate": 1.2652631578947368e-06, "loss": 0.0001, "step": 8800 }, { "epoch": 42.02, "learning_rate": 1.2389473684210527e-06, "loss": 0.0001, "step": 8825 }, { "epoch": 42.02, "learning_rate": 1.2126315789473685e-06, "loss": 0.0001, "step": 8850 }, { "epoch": 43.0, "learning_rate": 1.1863157894736843e-06, "loss": 0.0001, "step": 8875 }, { "epoch": 43.0, "learning_rate": 1.1600000000000001e-06, "loss": 0.0001, "step": 8900 }, { "epoch": 43.01, "learning_rate": 1.133684210526316e-06, "loss": 0.0001, "step": 8925 }, { "epoch": 43.01, "learning_rate": 1.1073684210526318e-06, "loss": 0.0001, "step": 8950 }, { "epoch": 43.01, "learning_rate": 1.0810526315789474e-06, "loss": 0.0001, "step": 8975 }, { "epoch": 43.01, "learning_rate": 1.0547368421052632e-06, "loss": 0.0001, "step": 9000 }, { "epoch": 43.01, "eval_loss": 0.41194406151771545, "eval_runtime": 1629.9167, "eval_samples_per_second": 4.044, "eval_steps_per_second": 0.126, "eval_wer": 14.119648426424725, "step": 9000 }, { "epoch": 43.02, "learning_rate": 1.028421052631579e-06, "loss": 0.0001, "step": 9025 }, { "epoch": 43.02, "learning_rate": 1.0021052631578948e-06, "loss": 0.0001, "step": 9050 }, { "epoch": 44.0, "learning_rate": 9.757894736842106e-07, "loss": 0.0001, "step": 9075 }, { "epoch": 44.0, "learning_rate": 9.494736842105263e-07, "loss": 0.0001, "step": 9100 }, { "epoch": 44.01, "learning_rate": 9.231578947368422e-07, "loss": 0.0001, "step": 9125 }, { "epoch": 44.01, "learning_rate": 8.968421052631579e-07, "loss": 0.0001, "step": 9150 }, { "epoch": 44.01, "learning_rate": 8.705263157894737e-07, "loss": 0.0001, "step": 9175 }, { "epoch": 44.01, "learning_rate": 8.442105263157896e-07, "loss": 0.0001, "step": 9200 }, { "epoch": 44.02, "learning_rate": 8.178947368421053e-07, "loss": 0.0001, "step": 9225 }, { "epoch": 44.02, "learning_rate": 7.915789473684212e-07, "loss": 0.0001, "step": 9250 }, { "epoch": 45.0, "learning_rate": 7.652631578947369e-07, "loss": 0.0001, "step": 9275 }, { "epoch": 45.0, "learning_rate": 7.389473684210527e-07, "loss": 0.0001, "step": 9300 }, { "epoch": 45.01, "learning_rate": 7.126315789473685e-07, "loss": 0.0001, "step": 9325 }, { "epoch": 45.01, "learning_rate": 6.863157894736842e-07, "loss": 0.0001, "step": 9350 }, { "epoch": 45.01, "learning_rate": 6.6e-07, "loss": 0.0001, "step": 9375 }, { "epoch": 45.01, "learning_rate": 6.336842105263157e-07, "loss": 0.0001, "step": 9400 }, { "epoch": 45.02, "learning_rate": 6.073684210526317e-07, "loss": 0.0001, "step": 9425 }, { "epoch": 45.02, "learning_rate": 5.810526315789474e-07, "loss": 0.0001, "step": 9450 }, { "epoch": 45.02, "learning_rate": 5.547368421052632e-07, "loss": 0.0001, "step": 9475 }, { "epoch": 46.0, "learning_rate": 5.284210526315789e-07, "loss": 0.0001, "step": 9500 }, { "epoch": 46.0, "learning_rate": 5.021052631578948e-07, "loss": 0.0001, "step": 9525 }, { "epoch": 46.01, "learning_rate": 4.757894736842106e-07, "loss": 0.0001, "step": 9550 }, { "epoch": 46.01, "learning_rate": 4.4947368421052637e-07, "loss": 0.0001, "step": 9575 }, { "epoch": 46.01, "learning_rate": 4.2315789473684214e-07, "loss": 0.0001, "step": 9600 }, { "epoch": 46.01, "learning_rate": 3.968421052631579e-07, "loss": 0.0001, "step": 9625 }, { "epoch": 46.02, "learning_rate": 3.7052631578947377e-07, "loss": 0.0001, "step": 9650 }, { "epoch": 46.02, "learning_rate": 3.4421052631578954e-07, "loss": 0.0001, "step": 9675 }, { "epoch": 47.0, "learning_rate": 3.178947368421053e-07, "loss": 0.0001, "step": 9700 }, { "epoch": 47.0, "learning_rate": 2.9157894736842107e-07, "loss": 0.0001, "step": 9725 }, { "epoch": 47.01, "learning_rate": 2.6526315789473684e-07, "loss": 0.0001, "step": 9750 }, { "epoch": 47.01, "learning_rate": 2.3894736842105265e-07, "loss": 0.0001, "step": 9775 }, { "epoch": 47.01, "learning_rate": 2.1263157894736842e-07, "loss": 0.0001, "step": 9800 }, { "epoch": 47.01, "learning_rate": 1.8631578947368424e-07, "loss": 0.0001, "step": 9825 }, { "epoch": 47.02, "learning_rate": 1.6e-07, "loss": 0.0001, "step": 9850 }, { "epoch": 47.02, "learning_rate": 1.3368421052631582e-07, "loss": 0.0001, "step": 9875 }, { "epoch": 48.0, "learning_rate": 1.0736842105263159e-07, "loss": 0.0001, "step": 9900 }, { "epoch": 48.0, "learning_rate": 8.105263157894738e-08, "loss": 0.0001, "step": 9925 }, { "epoch": 48.01, "learning_rate": 5.473684210526316e-08, "loss": 0.0001, "step": 9950 }, { "epoch": 48.01, "learning_rate": 2.842105263157895e-08, "loss": 0.0001, "step": 9975 }, { "epoch": 48.01, "learning_rate": 2.105263157894737e-09, "loss": 0.0001, "step": 10000 }, { "epoch": 48.01, "eval_loss": 0.4149830639362335, "eval_runtime": 1464.85, "eval_samples_per_second": 4.499, "eval_steps_per_second": 0.141, "eval_wer": 14.135849973672485, "step": 10000 }, { "epoch": 48.01, "step": 10000, "total_flos": 6.530891627416781e+20, "train_loss": 0.02586415379194077, "train_runtime": 150418.0511, "train_samples_per_second": 4.255, "train_steps_per_second": 0.066 } ], "logging_steps": 25, "max_steps": 10000, "num_train_epochs": 9223372036854775807, "save_steps": 1000, "total_flos": 6.530891627416781e+20, "trial_name": null, "trial_params": null }