Reinforcement Learning with Prototypical Representations | Read Paper on Bytez