{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "global_step": 8775, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.02, "learning_rate": 9.962631077216398e-06, "loss": 0.1958, "step": 200 }, { "epoch": 0.05, "learning_rate": 9.886367969494757e-06, "loss": 0.1228, "step": 400 }, { "epoch": 0.07, "learning_rate": 9.810104861773118e-06, "loss": 0.1051, "step": 600 }, { "epoch": 0.09, "learning_rate": 9.733841754051478e-06, "loss": 0.0977, "step": 800 }, { "epoch": 0.11, "learning_rate": 9.657578646329838e-06, "loss": 0.0891, "step": 1000 }, { "epoch": 0.14, "learning_rate": 9.581315538608199e-06, "loss": 0.0827, "step": 1200 }, { "epoch": 0.16, "learning_rate": 9.50505243088656e-06, "loss": 0.0822, "step": 1400 }, { "epoch": 0.18, "learning_rate": 9.42878932316492e-06, "loss": 0.0796, "step": 1600 }, { "epoch": 0.21, "learning_rate": 9.352526215443281e-06, "loss": 0.0777, "step": 1800 }, { "epoch": 0.23, "learning_rate": 9.27626310772164e-06, "loss": 0.0769, "step": 2000 }, { "epoch": 0.25, "learning_rate": 9.200000000000002e-06, "loss": 0.0736, "step": 2200 }, { "epoch": 0.27, "learning_rate": 9.123736892278361e-06, "loss": 0.0721, "step": 2400 }, { "epoch": 0.3, "learning_rate": 9.047473784556723e-06, "loss": 0.0748, "step": 2600 }, { "epoch": 0.32, "learning_rate": 8.971210676835082e-06, "loss": 0.0736, "step": 2800 }, { "epoch": 0.34, "learning_rate": 8.894947569113442e-06, "loss": 0.0714, "step": 3000 }, { "epoch": 0.36, "learning_rate": 8.818684461391803e-06, "loss": 0.0718, "step": 3200 }, { "epoch": 0.39, "learning_rate": 8.742421353670163e-06, "loss": 0.0729, "step": 3400 }, { "epoch": 0.41, "learning_rate": 8.666158245948522e-06, "loss": 0.0711, "step": 3600 }, { "epoch": 0.43, "learning_rate": 8.589895138226883e-06, "loss": 0.0663, "step": 3800 }, { "epoch": 0.46, "learning_rate": 8.513632030505243e-06, "loss": 0.0675, "step": 4000 }, { "epoch": 0.48, "learning_rate": 8.437368922783604e-06, "loss": 0.0689, "step": 4200 }, { "epoch": 0.5, "learning_rate": 8.361105815061964e-06, "loss": 0.0685, "step": 4400 }, { "epoch": 0.52, "learning_rate": 8.284842707340325e-06, "loss": 0.0663, "step": 4600 }, { "epoch": 0.55, "learning_rate": 8.208579599618686e-06, "loss": 0.0651, "step": 4800 }, { "epoch": 0.57, "learning_rate": 8.132316491897046e-06, "loss": 0.0652, "step": 5000 }, { "epoch": 0.59, "learning_rate": 8.056053384175407e-06, "loss": 0.0653, "step": 5200 }, { "epoch": 0.62, "learning_rate": 7.979790276453767e-06, "loss": 0.0656, "step": 5400 }, { "epoch": 0.64, "learning_rate": 7.903527168732126e-06, "loss": 0.0635, "step": 5600 }, { "epoch": 0.66, "learning_rate": 7.827264061010486e-06, "loss": 0.061, "step": 5800 }, { "epoch": 0.68, "learning_rate": 7.751000953288847e-06, "loss": 0.0637, "step": 6000 }, { "epoch": 0.71, "learning_rate": 7.674737845567207e-06, "loss": 0.0621, "step": 6200 }, { "epoch": 0.73, "learning_rate": 7.598856053384176e-06, "loss": 0.0657, "step": 6400 }, { "epoch": 0.75, "learning_rate": 7.522592945662537e-06, "loss": 0.0618, "step": 6600 }, { "epoch": 0.77, "learning_rate": 7.446329837940897e-06, "loss": 0.0617, "step": 6800 }, { "epoch": 0.8, "learning_rate": 7.370066730219257e-06, "loss": 0.0623, "step": 7000 }, { "epoch": 0.82, "learning_rate": 7.293803622497617e-06, "loss": 0.0581, "step": 7200 }, { "epoch": 0.84, "learning_rate": 7.217540514775978e-06, "loss": 0.0612, "step": 7400 }, { "epoch": 0.87, "learning_rate": 7.141277407054338e-06, "loss": 0.0584, "step": 7600 }, { "epoch": 0.89, "learning_rate": 7.065014299332699e-06, "loss": 0.0595, "step": 7800 }, { "epoch": 0.91, "learning_rate": 6.9887511916110584e-06, "loss": 0.0605, "step": 8000 }, { "epoch": 0.93, "learning_rate": 6.912488083889419e-06, "loss": 0.0584, "step": 8200 }, { "epoch": 0.96, "learning_rate": 6.836606291706388e-06, "loss": 0.06, "step": 8400 }, { "epoch": 0.98, "learning_rate": 6.760343183984748e-06, "loss": 0.0589, "step": 8600 }, { "epoch": 1.0, "eval_loss": 0.12254729866981506, "eval_runtime": 5547.0518, "eval_samples_per_second": 6.063, "eval_steps_per_second": 0.189, "eval_wer": 0.06041116269525739, "step": 8775 } ], "max_steps": 26325, "num_train_epochs": 3, "total_flos": 2.1491591961083904e+20, "trial_name": null, "trial_params": null }