{ "best_metric": null, "best_model_checkpoint": null, "epoch": 20.0, "global_step": 5240, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.91, "learning_rate": 4.522900763358779e-05, "loss": 2.2998, "step": 500 }, { "epoch": 3.82, "learning_rate": 4.0458015267175576e-05, "loss": 2.0546, "step": 1000 }, { "epoch": 5.73, "learning_rate": 3.568702290076336e-05, "loss": 1.8768, "step": 1500 }, { "epoch": 7.63, "learning_rate": 3.091603053435115e-05, "loss": 1.7311, "step": 2000 }, { "epoch": 9.54, "learning_rate": 2.6145038167938934e-05, "loss": 1.6031, "step": 2500 }, { "epoch": 11.45, "learning_rate": 2.1374045801526718e-05, "loss": 1.495, "step": 3000 }, { "epoch": 13.36, "learning_rate": 1.6603053435114505e-05, "loss": 1.404, "step": 3500 }, { "epoch": 15.27, "learning_rate": 1.1832061068702292e-05, "loss": 1.3284, "step": 4000 }, { "epoch": 17.18, "learning_rate": 7.061068702290078e-06, "loss": 1.2688, "step": 4500 }, { "epoch": 19.08, "learning_rate": 2.2900763358778625e-06, "loss": 1.2258, "step": 5000 }, { "epoch": 20.0, "step": 5240, "total_flos": 1.910567337984e+16, "train_loss": 1.609383287502609, "train_runtime": 6211.091, "train_samples_per_second": 5.886, "train_steps_per_second": 0.844 } ], "max_steps": 5240, "num_train_epochs": 20, "total_flos": 1.910567337984e+16, "trial_name": null, "trial_params": null }