Use the arrow keys to move the red agent from S to G, or press Train RL Solver + Animate to watch a Q-learning agent practice on the current maze and then perform a final learned run.