huggingface

Search Shortcut cmd + k | ctrl + k

Documentation

huggingface

Downloads n/athis week

GitHub stars 0

Extension repository on GitHub

Extension descriptor (YAML)

Discover, query, profile, and estimate the storage of Hugging Face datasets

Maintainer(s): dentiny

Installing and Loading

INSTALL huggingface FROM community;
LOAD huggingface;

Example

INSTALL huggingface FROM community;
LOAD huggingface;

FROM hf_files(
    'ibm/duorc',
    config = 'ParaphraseRC',
    split = 'train'
);

The huggingface extension works with Parquet datasets hosted on the Hugging Face Hub. It can discover Parquet files, scan records with projection and filter pushdown, profile Parquet metadata, and estimate external-blob storage from a bounded sample.

The extension uses cache_httpfs for cached access to hf:// paths and supports Hugging Face credentials through DuckDB's secret manager. Lance datasets are not currently supported.