{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 342, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.19, "learning_rate": 0.00012571428571428572, "loss": 1.1699, "step": 22 }, { "epoch": 0.39, "learning_rate": 0.0001941368078175896, "loss": 0.9417, "step": 44 }, { "epoch": 0.58, "learning_rate": 0.0001798045602605863, "loss": 0.876, "step": 66 }, { "epoch": 0.77, "learning_rate": 0.00016547231270358308, "loss": 0.8336, "step": 88 }, { "epoch": 0.96, "learning_rate": 0.0001511400651465798, "loss": 0.8132, "step": 110 }, { "epoch": 1.16, "learning_rate": 0.00013680781758957655, "loss": 0.7983, "step": 132 }, { "epoch": 1.35, "learning_rate": 0.00012247557003257332, "loss": 0.7752, "step": 154 }, { "epoch": 1.54, "learning_rate": 0.00010814332247557004, "loss": 0.7826, "step": 176 }, { "epoch": 1.74, "learning_rate": 9.381107491856678e-05, "loss": 0.7814, "step": 198 }, { "epoch": 1.93, "learning_rate": 7.947882736156352e-05, "loss": 0.7853, "step": 220 }, { "epoch": 2.12, "learning_rate": 6.514657980456026e-05, "loss": 0.7673, "step": 242 }, { "epoch": 2.32, "learning_rate": 5.081433224755701e-05, "loss": 0.7548, "step": 264 }, { "epoch": 2.51, "learning_rate": 3.6482084690553745e-05, "loss": 0.7589, "step": 286 }, { "epoch": 2.7, "learning_rate": 2.214983713355049e-05, "loss": 0.7433, "step": 308 }, { "epoch": 2.89, "learning_rate": 7.81758957654723e-06, "loss": 0.7358, "step": 330 } ], "max_steps": 342, "num_train_epochs": 3, "total_flos": 1.1121035283372442e+17, "trial_name": null, "trial_params": null }