train_results.json

{"eval_loss": 0.04157589375972748, "eval_matthews_correlation": 0.9786230864021822, "eval_runtime": 61.069, "eval_samples_per_second": 140.022, "eval_steps_per_second": 17.505}