{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 402, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.19, "learning_rate": 0.00012682926829268293, "loss": 1.8322, "step": 26 }, { "epoch": 0.39, "learning_rate": 0.00019390581717451522, "loss": 1.5787, "step": 52 }, { "epoch": 0.58, "learning_rate": 0.00017950138504155126, "loss": 1.4229, "step": 78 }, { "epoch": 0.78, "learning_rate": 0.00016509695290858726, "loss": 1.347, "step": 104 }, { "epoch": 0.97, "learning_rate": 0.00015069252077562327, "loss": 1.3379, "step": 130 }, { "epoch": 1.16, "learning_rate": 0.0001362880886426593, "loss": 1.358, "step": 156 }, { "epoch": 1.36, "learning_rate": 0.00012188365650969529, "loss": 1.3522, "step": 182 }, { "epoch": 1.55, "learning_rate": 0.0001074792243767313, "loss": 1.2781, "step": 208 }, { "epoch": 1.75, "learning_rate": 9.307479224376732e-05, "loss": 1.3247, "step": 234 }, { "epoch": 1.94, "learning_rate": 7.867036011080334e-05, "loss": 1.3194, "step": 260 }, { "epoch": 2.13, "learning_rate": 6.426592797783934e-05, "loss": 1.274, "step": 286 }, { "epoch": 2.33, "learning_rate": 4.986149584487535e-05, "loss": 1.3126, "step": 312 }, { "epoch": 2.52, "learning_rate": 3.545706371191136e-05, "loss": 1.3317, "step": 338 }, { "epoch": 2.72, "learning_rate": 2.105263157894737e-05, "loss": 1.2604, "step": 364 }, { "epoch": 2.91, "learning_rate": 6.64819944598338e-06, "loss": 1.1841, "step": 390 } ], "max_steps": 402, "num_train_epochs": 3, "total_flos": 3.324593797123277e+16, "trial_name": null, "trial_params": null }