dataset_loader.py

from huggingface_hub import Repository
from huggingface_hub import login
import os

def load_dataset():
    login(token = os.environ['HUB_TOKEN'])

    repo = Repository(
        local_dir="mystartup_data",
        repo_type="dataset",
        clone_from="gheng/mystartup_data",
        token=True
    )
    repo.git_pull()