Writing

Notes on mechanistic interpretability, AI safety, and research. Includes original posts and annotated reposts.