{ "best_metric": null, "best_model_checkpoint": null, "epoch": 5.0, "global_step": 7750, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.32, "learning_rate": 9.35483870967742e-06, "loss": 51.2618, "step": 500 }, { "epoch": 0.65, "learning_rate": 8.70967741935484e-06, "loss": 36.8478, "step": 1000 }, { "epoch": 0.97, "learning_rate": 8.064516129032258e-06, "loss": 14.232, "step": 1500 }, { "epoch": 1.29, "learning_rate": 7.4193548387096784e-06, "loss": 0.4812, "step": 2000 }, { "epoch": 1.61, "learning_rate": 6.774193548387097e-06, "loss": 0.3559, "step": 2500 }, { "epoch": 1.94, "learning_rate": 6.129032258064517e-06, "loss": 0.2781, "step": 3000 }, { "epoch": 2.26, "learning_rate": 5.483870967741935e-06, "loss": 0.2514, "step": 3500 }, { "epoch": 2.58, "learning_rate": 4.838709677419355e-06, "loss": 0.2306, "step": 4000 }, { "epoch": 2.9, "learning_rate": 4.193548387096774e-06, "loss": 0.2238, "step": 4500 }, { "epoch": 3.23, "learning_rate": 3.548387096774194e-06, "loss": 0.2172, "step": 5000 }, { "epoch": 3.55, "learning_rate": 2.903225806451613e-06, "loss": 0.2119, "step": 5500 }, { "epoch": 3.87, "learning_rate": 2.2580645161290324e-06, "loss": 0.2101, "step": 6000 }, { "epoch": 4.19, "learning_rate": 1.6129032258064516e-06, "loss": 0.2073, "step": 6500 }, { "epoch": 4.52, "learning_rate": 9.67741935483871e-07, "loss": 0.2082, "step": 7000 }, { "epoch": 4.84, "learning_rate": 3.2258064516129035e-07, "loss": 0.2024, "step": 7500 } ], "max_steps": 7750, "num_train_epochs": 5, "total_flos": 9435041316864000.0, "trial_name": null, "trial_params": null }