Offline Multi-Agent Reinforcement Learning Implementations: Solving Overcooked Game with Data-Driven Method