Fetching the paper…

Interpreting and Steering LLMs with Mutual Information-based Explanations on Sparse Autoencoders · Around