llm_alphacah.md

in-context RL with AlphaGo-esque self-play

label: experimental, rl, llm, game

Cards Against Humanity type game. Have the LLM play itself and the players. Train the group as an ensemble, taking turns playing and judging responses.

https://chat.openai.com/c/1ea87011-b45c-4823-acdf-430794717e04