Gemma Scope:helping the safety community shed light on the inner workings of LLM

Announcing a comprehensive, open suite of sparse autoencoders for language model interpretability.

Read more here: External Link