Blog

Writing

Notes and longer pieces, mostly on machine learning and interpretability.

Summoned by the Question

What mechanistic interpretability says about whether an LLM commits to a secret in 20 Questions.

interpretabilityLLMsmech-interp
Read