Sergii comments on Automating Mechanistic Interpretability via Program Synthesis