Multiverse Computing Interview Question

Could you explain encoder–decoder for Reinforcement Learning?