Skip to content
Better HN
You can predict LLM output sensitivity in closed form | Better HN