Learning Actionable Representations with Goal-Conditioned Policies | Read Paper on Bytez