{ "best_metric": null, "best_model_checkpoint": null, "epoch": 5.0, "global_step": 470, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.0, "learning_rate": 8e-05, "loss": 3.4965, "step": 94 }, { "epoch": 1.0, "eval_loss": 2.779850721359253, "eval_runtime": 23.1518, "eval_samples_per_second": 8.639, "eval_steps_per_second": 1.08, "step": 94 }, { "epoch": 2.0, "learning_rate": 6e-05, "loss": 3.4526, "step": 188 }, { "epoch": 2.0, "eval_loss": 2.738048791885376, "eval_runtime": 24.2844, "eval_samples_per_second": 8.236, "eval_steps_per_second": 1.029, "step": 188 }, { "epoch": 3.0, "learning_rate": 4e-05, "loss": 3.4012, "step": 282 }, { "epoch": 3.0, "eval_loss": 2.6721441745758057, "eval_runtime": 23.2045, "eval_samples_per_second": 8.619, "eval_steps_per_second": 1.077, "step": 282 }, { "epoch": 4.0, "learning_rate": 2e-05, "loss": 3.2776, "step": 376 }, { "epoch": 4.0, "eval_loss": 2.6650607585906982, "eval_runtime": 24.0682, "eval_samples_per_second": 8.31, "eval_steps_per_second": 1.039, "step": 376 }, { "epoch": 5.0, "learning_rate": 0.0, "loss": 3.2164, "step": 470 }, { "epoch": 5.0, "eval_loss": 2.6554956436157227, "eval_runtime": 22.8547, "eval_samples_per_second": 8.751, "eval_steps_per_second": 1.094, "step": 470 }, { "epoch": 5.0, "step": 470, "total_flos": 1.1450900938752e+17, "train_loss": 3.3688572660405587, "train_runtime": 692.9876, "train_samples_per_second": 5.411, "train_steps_per_second": 0.678 } ], "max_steps": 470, "num_train_epochs": 5, "total_flos": 1.1450900938752e+17, "trial_name": null, "trial_params": null }