{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 786, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.2, "learning_rate": 0.00013164556962025315, "loss": 1.6276, "step": 52 }, { "epoch": 0.4, "learning_rate": 0.00019292786421499294, "loss": 1.3899, "step": 104 }, { "epoch": 0.6, "learning_rate": 0.00017821782178217824, "loss": 1.3392, "step": 156 }, { "epoch": 0.79, "learning_rate": 0.00016350777934936352, "loss": 1.3058, "step": 208 }, { "epoch": 0.99, "learning_rate": 0.00014879773691654882, "loss": 1.2995, "step": 260 }, { "epoch": 1.19, "learning_rate": 0.00013408769448373408, "loss": 1.2442, "step": 312 }, { "epoch": 1.39, "learning_rate": 0.00011937765205091938, "loss": 1.2549, "step": 364 }, { "epoch": 1.59, "learning_rate": 0.00010466760961810467, "loss": 1.2237, "step": 416 }, { "epoch": 1.79, "learning_rate": 8.995756718528997e-05, "loss": 1.2321, "step": 468 }, { "epoch": 1.98, "learning_rate": 7.524752475247526e-05, "loss": 1.1997, "step": 520 }, { "epoch": 2.18, "learning_rate": 6.053748231966054e-05, "loss": 1.17, "step": 572 }, { "epoch": 2.38, "learning_rate": 4.582743988684583e-05, "loss": 1.159, "step": 624 }, { "epoch": 2.58, "learning_rate": 3.111739745403111e-05, "loss": 1.1415, "step": 676 }, { "epoch": 2.78, "learning_rate": 1.640735502121641e-05, "loss": 1.1425, "step": 728 }, { "epoch": 2.98, "learning_rate": 1.6973125884016973e-06, "loss": 1.125, "step": 780 } ], "max_steps": 786, "num_train_epochs": 3, "total_flos": 1.957401291789435e+17, "trial_name": null, "trial_params": null }