data/scripts_hsichen/batch_download.py

from download import download_hf_dataset

repos = [
    'chaiting/pk-2087-L4',
    'chaiting/pk-2354-L5',
    'chaiting/pk-2360-L4',
    'chaiting/pk-2353-L4',
    'chaiting/pk-3976-L5',
    'chaiting/pk-1818-L4',
    'chaiting/pk-2086-L3',
    'chaiting/pk-2089-L6',
    'chaiting/pk-1158-L5',
    'chaiting/pk-1740-L4',
    'chaiting/pk-1819-L5',
    'chaiting/pk-1820-L6',
    'chaiting/pk-867-L4',
    'chaiting/pk-2352-L3',
    'chaiting/pk-2352-L3',
    'chaiting/pk-1817-L3',
    'chaiting/pk-4089-L5',
    'chaiting/pk-1741-L3'
]

for repo in repos:
    basename = repo.split('/')[-1]
    print(f'Downloading {repo}...')
    download_hf_dataset(repo, f'/home/hsichen/LLaMA-Factory/data/chaiting/batch_data/{basename}')
    print(f'Downloaded {repo} successfully.')