These files are not directly used by the WILDS package, and users do not need to look at them to use the package.

This directory contains scripts that were used to preprocess the WILDS datasets from their original forms into the *-wilds forms that we use in our benchmark. The WILDS package automatically downloads the already-processed forms; We archive these scripts here just for reproducibility purposes and for users who are interested in the precise details of the dataset preprocessing.

Some of these scripts have specific requirements beyond what is required for the WILDS package, e.g., specialized software for handling pathology slides.