Telling an LLM to take a break doesn’t cause it to take a break—the same amount of computation is always used per token—rather, it resets internal states to “has just taken a break”.Another case of…

Posters: davidad. davidad

3 likes recorded by the source corpus.

Topics: AI & LLMs

View Post: https://x.com/davidad/status/1755888730475487235