post-training eval_results.json

{
    "eval_samples": 438,
    "test_accuracy": 0.7328767123287672,
    "test_loss": 0.8578213453292847,
    "test_macro_f1": 0.5900120784858806,
    "test_micro_f1": 0.7328767123287672,
    "test_runtime": 1.8885,
    "test_samples_per_second": 231.936,
    "test_steps_per_second": 2.118
}