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

📝

内容提要

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

🏷️

标签

➡️

继续阅读