{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.710801393728223, "global_step": 3500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.19, "learning_rate": 8.333333333333334e-06, "loss": 2.4875, "step": 250 }, { "epoch": 0.39, "learning_rate": 1.6666666666666667e-05, "loss": 2.2619, "step": 500 }, { "epoch": 0.58, "learning_rate": 2.5e-05, "loss": 2.2339, "step": 750 }, { "epoch": 0.77, "learning_rate": 3.3333333333333335e-05, "loss": 2.2639, "step": 1000 }, { "epoch": 0.97, "learning_rate": 4.166666666666667e-05, "loss": 2.1806, "step": 1250 }, { "epoch": 1.16, "learning_rate": 5e-05, "loss": 2.012, "step": 1500 }, { "epoch": 1.36, "learning_rate": 4.473240623683102e-05, "loss": 1.9184, "step": 1750 }, { "epoch": 1.55, "learning_rate": 3.946481247366203e-05, "loss": 1.925, "step": 2000 }, { "epoch": 1.74, "learning_rate": 3.419721871049305e-05, "loss": 1.8519, "step": 2250 }, { "epoch": 1.94, "learning_rate": 2.8929624947324065e-05, "loss": 1.8232, "step": 2500 }, { "epoch": 2.13, "learning_rate": 2.366203118415508e-05, "loss": 1.5126, "step": 2750 }, { "epoch": 2.32, "learning_rate": 1.8394437420986094e-05, "loss": 1.3707, "step": 3000 }, { "epoch": 2.52, "learning_rate": 1.3126843657817109e-05, "loss": 1.311, "step": 3250 }, { "epoch": 2.71, "learning_rate": 7.859249894648125e-06, "loss": 1.2866, "step": 3500 } ], "max_steps": 3873, "num_train_epochs": 3, "total_flos": 3.956185882637107e+17, "trial_name": null, "trial_params": null }