Concise pytorch implements of MARL algorithms, including MAPPO, MADDPG, MATD3, QMIX and VDN.

Projects from basic algorithms to MARL. Implements MADDPG/MATD3 in Predator-Prey pursuit games with PettingZoo MPE environments.

Multi-agent reinforcement learning framework for Unity environments. Implements MAPPO, MASAC, MATD3, and MADDPG with comprehensive evaluation tools. Features sample-efficient training, competitive analysis, and pre-trained models achieving great performance in Tennis and Soccer environments.