Datacard

This repo aims to find and update the missing model cards for Hugging face datasets.

If you find this a worth while pursute, feel free to reach out and let's try to make the Hugging face datasets complete 😉

Setup

# install poetry
git clone --recurse-submodules --remote-submodules git@github.com:Hugging-Face-Supporter/datacards.git
cd datacards
git submodule update

poetry install

Run

poetry shell
python datacards/main.py

WIP

• Look into how to provide multiple answers in model card (ex. Glue dataset)
• Find the datasets that are missing information by parsing the README
• Find ways to know what categories are valid answers
• Create method to filter for missing datasets
• Incorporate the argparse to filter for certiain things
• Toggle between datasets to annotate.
• Save modified files to the README again
• Once done, find ways to create automatic PR to Hugging face datasets
• Incorporate the Huggingface Hub API