Towards Principled Unsupervised Multi-Agent Reinforcement Learning | Read Paper on Bytez