Why does load_dataset make https request when loading a local dataset?

🤗Datasets
1

When calling ds = load_dataset("parquet", data_files=<local_path>, split="train") I get debug logs saying

DEBUG - Starting new HTTPS connection (1): ``s3.amazonaws.com:443
DEBUG - ``https://s3.amazonaws.com:443`` “HEAD /datasets.huggingface.co/datasets/datasets/parquet/parquet.py HTTP/1.1” 404 0

which confuses me, as I thought it just loads a local dataset. Can someone clarify that for me?

1 Like
2

This issue shows how to turn it off, but im still not entirely sure I understand why its needed in the first place.

1 Like