{ "best_metric": 0.6666666666666666, "best_model_checkpoint": "tiny-bert-sst2-distilled/run-1/checkpoint-160", "epoch": 2.0, "eval_steps": 500, "global_step": 320, "is_hyper_param_search": true, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.0, "grad_norm": 0.6969084143638611, "learning_rate": 2.475738136709632e-05, "loss": 0.064, "step": 160 }, { "epoch": 1.0, "eval_accuracy": 0.6666666666666666, "eval_f1": 0.0, "eval_loss": 0.04275359958410263, "eval_mcc": 0.0, "eval_precision": 0.0, "eval_recall": 0.0, "eval_runtime": 1.9263, "eval_samples_per_second": 661.882, "eval_steps_per_second": 20.765, "step": 160 }, { "epoch": 2.0, "grad_norm": 0.23619167506694794, "learning_rate": 1.8568036025322237e-05, "loss": 0.0425, "step": 320 }, { "epoch": 2.0, "eval_accuracy": 0.6666666666666666, "eval_f1": 0.0, "eval_loss": 0.038055840879678726, "eval_mcc": 0.0, "eval_precision": 0.0, "eval_recall": 0.0, "eval_runtime": 1.8737, "eval_samples_per_second": 680.488, "eval_steps_per_second": 21.349, "step": 320 } ], "logging_steps": 500, "max_steps": 800, "num_input_tokens_seen": 0, "num_train_epochs": 5, "save_steps": 500, "total_flos": 583510875840.0, "train_batch_size": 32, "trial_name": null, "trial_params": { "alpha": 0.031053442748151472, "learning_rate": 3.09467267088704e-05, "num_train_epochs": 5, "temperature": 3 } }