{ "best_metric": 0.6602564102564102, "best_model_checkpoint": "Output_llama2_70-15-15_new/checkpoint-322", "epoch": 10.0, "eval_steps": 500, "global_step": 460, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.0, "eval_accuracy": 0.4807692307692308, "eval_f1": 0.3121878121878122, "eval_loss": 0.779107391834259, "eval_precision": 0.23113905325443787, "eval_recall": 0.4807692307692308, "eval_runtime": 111.4069, "eval_samples_per_second": 1.4, "eval_steps_per_second": 0.09, "step": 46 }, { "epoch": 2.0, "eval_accuracy": 0.48717948717948717, "eval_f1": 0.3262093156048617, "eval_loss": 0.755558967590332, "eval_precision": 0.7518610421836228, "eval_recall": 0.48717948717948717, "eval_runtime": 111.1784, "eval_samples_per_second": 1.403, "eval_steps_per_second": 0.09, "step": 92 }, { "epoch": 3.0, "eval_accuracy": 0.5064102564102564, "eval_f1": 0.40855426407209083, "eval_loss": 0.7070600390434265, "eval_precision": 0.5707901253675902, "eval_recall": 0.5064102564102564, "eval_runtime": 111.2413, "eval_samples_per_second": 1.402, "eval_steps_per_second": 0.09, "step": 138 }, { "epoch": 4.0, "eval_accuracy": 0.5, "eval_f1": 0.39713058754635955, "eval_loss": 0.7044801115989685, "eval_precision": 0.5548682087143626, "eval_recall": 0.5, "eval_runtime": 111.1012, "eval_samples_per_second": 1.404, "eval_steps_per_second": 0.09, "step": 184 }, { "epoch": 5.0, "eval_accuracy": 0.5705128205128205, "eval_f1": 0.5339812912754868, "eval_loss": 0.6713973879814148, "eval_precision": 0.622663378087679, "eval_recall": 0.5705128205128205, "eval_runtime": 111.3456, "eval_samples_per_second": 1.401, "eval_steps_per_second": 0.09, "step": 230 }, { "epoch": 6.0, "eval_accuracy": 0.4935897435897436, "eval_f1": 0.3931920371648725, "eval_loss": 0.6975941061973572, "eval_precision": 0.530316514471444, "eval_recall": 0.4935897435897436, "eval_runtime": 111.0983, "eval_samples_per_second": 1.404, "eval_steps_per_second": 0.09, "step": 276 }, { "epoch": 7.0, "eval_accuracy": 0.6602564102564102, "eval_f1": 0.6507378243632287, "eval_loss": 0.6453418135643005, "eval_precision": 0.6906434911242604, "eval_recall": 0.6602564102564102, "eval_runtime": 111.1451, "eval_samples_per_second": 1.404, "eval_steps_per_second": 0.09, "step": 322 }, { "epoch": 8.0, "eval_accuracy": 0.5769230769230769, "eval_f1": 0.5353708791208791, "eval_loss": 0.6639875769615173, "eval_precision": 0.6429280397022332, "eval_recall": 0.5769230769230769, "eval_runtime": 111.2239, "eval_samples_per_second": 1.403, "eval_steps_per_second": 0.09, "step": 368 }, { "epoch": 9.0, "eval_accuracy": 0.6153846153846154, "eval_f1": 0.5928187429481648, "eval_loss": 0.645951509475708, "eval_precision": 0.6643282486336569, "eval_recall": 0.6153846153846154, "eval_runtime": 111.2453, "eval_samples_per_second": 1.402, "eval_steps_per_second": 0.09, "step": 414 }, { "epoch": 10.0, "eval_accuracy": 0.6474358974358975, "eval_f1": 0.6375581196222184, "eval_loss": 0.6310403347015381, "eval_precision": 0.6760355029585798, "eval_recall": 0.6474358974358975, "eval_runtime": 111.4617, "eval_samples_per_second": 1.4, "eval_steps_per_second": 0.09, "step": 460 }, { "epoch": 10.0, "step": 460, "total_flos": 2.488087619686564e+17, "train_loss": 0.5811269345490829, "train_runtime": 16870.9489, "train_samples_per_second": 0.433, "train_steps_per_second": 0.027 } ], "logging_steps": 500, "max_steps": 460, "num_input_tokens_seen": 0, "num_train_epochs": 10, "save_steps": 500, "stateful_callbacks": { "EarlyStoppingCallback": { "args": { "early_stopping_patience": 10, "early_stopping_threshold": 0.0 }, "attributes": { "early_stopping_patience_counter": 0 } }, "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2.488087619686564e+17, "train_batch_size": 16, "trial_name": null, "trial_params": null }