Multi-Agent Imitation Learning: Value is Easy, Regret is Hard | Read Paper on Bytez