Uncategorized Gemma Scope: helping the safety community shed light on the inner workings of language models Google DeepMind Blog July 31, 2024 July 31, 2024 Announcing a comprehensive, open suite of sparse autoencoders for language model interpretability.