src/step5_verify_load_splits.py

from datasets import load_from_disk

SPLIT_DIR = "data/splits/comprehensive-car-damage_seed42_test0p2"

def main():
    splits = load_from_disk(SPLIT_DIR)
    print("Loaded keys:", list(splits.keys()))
    print("train:", len(splits["train"]), "val:", len(splits["val"]))

    # sanity: label names
    names = splits["train"].features["label"].names
    print("Label names:", names)

if __name__ == "__main__":
    main()