alt Hacker NewsI mean, that LLM idea _sounds_ ridiculous, but similar ideas have worked really well in machine learning for games like Chess and AI.
I mean, that LLM idea _sounds_ ridiculous, but similar ideas have worked really well in machine learning for games like Chess and AI.
This is assuming that the knowledge space being aimed at is discoverable solely by exploration of 1v1 games. Maths and maybe some of the sciences could be set up like this if you were very clever about it, but not much else.