Vision Interpretability with Masking Methods

This project is still in progress. I’ve been working with Kannan and Efe since the summer on interpreting vision models via Shapley Values methods.