multimodal/tools/instruct_tuning_data/utils.py

import os

MAXCOUNT = 1000
NAMING = "%06d.tar"

def check_sample(sample):
    assert "__key__" in sample
    assert "image_path.txt" in sample
    assert os.path.exists(sample["image_path.txt"])
    assert "dataset.txt" in sample
    assert "data.pyd" in sample