{ "best_metric": null, "best_model_checkpoint": null, "epoch": 10.703363914373089, "global_step": 7000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.76, "learning_rate": 4.745158002038736e-05, "loss": 0.3992, "step": 500 }, { "epoch": 1.53, "learning_rate": 4.490316004077472e-05, "loss": 0.196, "step": 1000 }, { "epoch": 2.29, "learning_rate": 4.235474006116208e-05, "loss": 0.1557, "step": 1500 }, { "epoch": 3.06, "learning_rate": 3.980632008154944e-05, "loss": 0.1302, "step": 2000 }, { "epoch": 3.82, "learning_rate": 3.72579001019368e-05, "loss": 0.1053, "step": 2500 }, { "epoch": 4.59, "learning_rate": 3.4709480122324164e-05, "loss": 0.0898, "step": 3000 }, { "epoch": 5.35, "learning_rate": 3.2161060142711516e-05, "loss": 0.0762, "step": 3500 }, { "epoch": 6.12, "learning_rate": 2.9612640163098882e-05, "loss": 0.0644, "step": 4000 }, { "epoch": 6.88, "learning_rate": 2.7064220183486238e-05, "loss": 0.0536, "step": 4500 }, { "epoch": 7.65, "learning_rate": 2.45158002038736e-05, "loss": 0.0449, "step": 5000 }, { "epoch": 8.41, "learning_rate": 2.196738022426096e-05, "loss": 0.0391, "step": 5500 }, { "epoch": 9.17, "learning_rate": 1.9418960244648318e-05, "loss": 0.0331, "step": 6000 }, { "epoch": 9.94, "learning_rate": 1.6870540265035677e-05, "loss": 0.0289, "step": 6500 }, { "epoch": 10.7, "learning_rate": 1.4322120285423038e-05, "loss": 0.0248, "step": 7000 } ], "max_steps": 9810, "num_train_epochs": 15, "total_flos": 2.6176216075199188e+16, "trial_name": null, "trial_params": null }