Quantum-Train-Based Distributed Multi-Agent Reinforcement Learning