{ "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": 3.5369, "step": 94 }, { "epoch": 1.0, "eval_loss": 2.7679553031921387, "eval_runtime": 21.844, "eval_samples_per_second": 9.156, "eval_steps_per_second": 1.144, "step": 94 }, { "epoch": 2.0, "learning_rate": 6e-05, "loss": 3.4721, "step": 188 }, { "epoch": 2.0, "eval_loss": 2.7605249881744385, "eval_runtime": 20.3138, "eval_samples_per_second": 9.846, "eval_steps_per_second": 1.231, "step": 188 }, { "epoch": 3.0, "learning_rate": 4e-05, "loss": 3.4243, "step": 282 }, { "epoch": 3.0, "eval_loss": 2.6854496002197266, "eval_runtime": 20.7618, "eval_samples_per_second": 9.633, "eval_steps_per_second": 1.204, "step": 282 }, { "epoch": 4.0, "learning_rate": 2e-05, "loss": 3.3027, "step": 376 }, { "epoch": 4.0, "eval_loss": 2.6660890579223633, "eval_runtime": 20.6653, "eval_samples_per_second": 9.678, "eval_steps_per_second": 1.21, "step": 376 }, { "epoch": 5.0, "learning_rate": 0.0, "loss": 3.2875, "step": 470 }, { "epoch": 5.0, "eval_loss": 2.6691887378692627, "eval_runtime": 20.1586, "eval_samples_per_second": 9.921, "eval_steps_per_second": 1.24, "step": 470 }, { "epoch": 5.0, "step": 470, "total_flos": 1.1450900938752e+17, "train_loss": 3.4047226276803526, "train_runtime": 644.0168, "train_samples_per_second": 5.823, "train_steps_per_second": 0.73 } ], "max_steps": 470, "num_train_epochs": 5, "total_flos": 1.1450900938752e+17, "trial_name": null, "trial_params": null }