{ "best_metric": null, "best_model_checkpoint": null, "epoch": 10.0, "global_step": 3020, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.66, "learning_rate": 4.1721854304635764e-05, "loss": 1.6291, "step": 500 }, { "epoch": 3.31, "learning_rate": 3.3443708609271526e-05, "loss": 1.2021, "step": 1000 }, { "epoch": 4.97, "learning_rate": 2.5165562913907287e-05, "loss": 1.0667, "step": 1500 }, { "epoch": 6.62, "learning_rate": 1.688741721854305e-05, "loss": 0.9833, "step": 2000 }, { "epoch": 8.28, "learning_rate": 8.609271523178809e-06, "loss": 0.9316, "step": 2500 }, { "epoch": 9.93, "learning_rate": 3.3112582781456954e-07, "loss": 0.9049, "step": 3000 }, { "epoch": 10.0, "step": 3020, "total_flos": 4.16091995209728e+16, "train_loss": 1.118253667465109, "train_runtime": 2440.6659, "train_samples_per_second": 12.345, "train_steps_per_second": 1.237 } ], "max_steps": 3020, "num_train_epochs": 10, "total_flos": 4.16091995209728e+16, "trial_name": null, "trial_params": null }