Fetching the paper…

Enhancing Automated Interpretability with Output-Centric Feature Descriptions · Around