I’ve been noticing the swirl patterns in latte art, and I’m curious how they might mirror the way activation functions shape neural nets. What do you think?