An Intuitive Explanation of Sparse Autoencoders for LLM Interpretability

Manipulating Chess-GPT's World Model

Chess-GPT's Internal World Model