This notebook trains an agent to navigate a maze and reach a desired destination. It uses the Gym-MiniGrid's fourRoom-v0 environment as the maze. The agent is trained by using reiforcement learning's vanilla policy gradient (REINFORCE) algorithm.