Yunjian Xu

For multi-agent offline safe reinforcement learning (MOSRL), we propose the first algorithm MOSDT, and the first dataset and benchmark MOSDB.