{ "best_metric": null, "best_model_checkpoint": null, "epoch": 15.0, "global_step": 5190, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.45, "learning_rate": 9.036608863198459e-05, "loss": 4.3892, "step": 500 }, { "epoch": 2.89, "learning_rate": 8.073217726396918e-05, "loss": 2.2906, "step": 1000 }, { "epoch": 4.34, "learning_rate": 7.109826589595376e-05, "loss": 1.8127, "step": 1500 }, { "epoch": 5.78, "learning_rate": 6.146435452793834e-05, "loss": 1.6294, "step": 2000 }, { "epoch": 7.23, "learning_rate": 5.1830443159922926e-05, "loss": 1.54, "step": 2500 }, { "epoch": 8.67, "learning_rate": 4.2196531791907514e-05, "loss": 1.4846, "step": 3000 }, { "epoch": 10.12, "learning_rate": 3.25626204238921e-05, "loss": 1.4445, "step": 3500 }, { "epoch": 11.56, "learning_rate": 2.2928709055876687e-05, "loss": 1.4234, "step": 4000 }, { "epoch": 13.01, "learning_rate": 1.329479768786127e-05, "loss": 1.4035, "step": 4500 }, { "epoch": 14.45, "learning_rate": 3.6608863198458577e-06, "loss": 1.3972, "step": 5000 }, { "epoch": 15.0, "step": 5190, "total_flos": 2.6052822149208474e+17, "train_loss": 1.8643955473265896, "train_runtime": 1757.3318, "train_samples_per_second": 755.987, "train_steps_per_second": 2.953 } ], "max_steps": 5190, "num_train_epochs": 15, "total_flos": 2.6052822149208474e+17, "trial_name": null, "trial_params": null }