Pessimism Meets Invariance: Provably Efficient Offline Mean-Field Multi-Agent RL

Open in new window