Open the Artificial Brain: Sparse Autoencoders for LLM Inspection

Open the Artificial Brain: Sparse Autoencoders for LLM Inspection

A deep dive into LLM visualization and interpretation using sparse autoencoders