The code for this is located here: https://github.com/adamkarvonen/chess_llm_interpretability