x
Topological Data Analysis and Mechanistic Interpretability — LessWrong