{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.848484848484849, "global_step": 4000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.61, "learning_rate": 4.3939393939393944e-05, "loss": 0.1499, "step": 500 }, { "epoch": 1.21, "learning_rate": 3.787878787878788e-05, "loss": 0.0214, "step": 1000 }, { "epoch": 1.82, "learning_rate": 3.181818181818182e-05, "loss": 0.0066, "step": 1500 }, { "epoch": 2.42, "learning_rate": 2.575757575757576e-05, "loss": 0.0032, "step": 2000 }, { "epoch": 3.03, "learning_rate": 1.9696969696969697e-05, "loss": 0.0015, "step": 2500 }, { "epoch": 3.64, "learning_rate": 1.3636363636363637e-05, "loss": 0.0011, "step": 3000 }, { "epoch": 4.24, "learning_rate": 7.5757575757575764e-06, "loss": 0.0012, "step": 3500 }, { "epoch": 4.85, "learning_rate": 1.5151515151515152e-06, "loss": 0.0003, "step": 4000 } ], "max_steps": 4125, "num_train_epochs": 5, "total_flos": 618627535241304.0, "trial_name": null, "trial_params": null }