def parse_args()

in distilvit/curate_gpt.py [0:0]


def parse_args():
    parser = argparse.ArgumentParser(
        description="Generate alternative text for images in a dataset.",
        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
    )
    parser.add_argument(
        "--sample",
        type=int,
        default=None,
        help="Run on a sample",
    )
    parser.add_argument(
        "--data-dir",
        type=str,
        default="./dummy_data",
        help="Dataset dir",
    )

    parser.add_argument(
        "--dataset",
        type=str,
        default="nlphuji/flickr30k",
        help="Name of the dataset to use",
    )
    parser.add_argument(
        "--target-dataset",
        type=str,
        default="mozilla/flickr30k-transformed-captions-gpt4o",
        help="Name of the target dataset to save to",
    )
    parser.add_argument(
        "--dataset-split",
        type=str,
        default="test",
        help="Name of the dataset split",
    )

    parser.add_argument(
        "--image-column",
        type=str,
        default="image",
        help="Name of the image column in the dataset",
    )
    parser.add_argument(
        "--generated-alt-text-column",
        type=str,
        default="alt_text",
        help="Name of the resulting alt text column",
    )
    parser.add_argument(
        "--image-id-column",
        type=str,
        default="img_id",
        help="Name of the image id column in the dataset",
    )

    return parser.parse_args()