This project allows you to train AI agents on custom-built multiplayer environments, through self-play reinforcement learning.