Discrete Compositional Representations as an Abstraction for Goal Conditioned Reinforcement Learning

Open in new window