INNER CODE UNIT · Python

parse_args

Future-House/BixBench · generate_zeroshot_evals.py:39

def parse_args():
    parser = argparse.ArgumentParser(
        description="BixBench Zero-ShotEvaluation Tool",
        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
    )

    parser.add_argument(
        "--answer-mode",
        choices=["mcq", "openanswer"],
        default="mcq",
        help="Evaluation mode",
    )
    parser.add_argument("--model", default="gpt-4o", help="Model name to use")

    parser.add_argument(
        "--temperature", type=float, default=1.0, help="Model temperature"
    )

View source record →