I've just finished creating a Magic the Gathering rules engine, and now I'm currently training an LLM agent to play games against itself through reinforcement learning.
How did you do it?