Abstract
End-to-end models for goal-orientated dialogue are challenging to train, because linguistic and strategic aspects are entangled in latent state vec tors. We introduce an approach to learning representations of messages in dialogues by maximizing the likelihood of subsequent sentences and actions, which decouples the semantics of the dialogue utterance from its linguistic realization. We then use these latent sentence representations for hierarchical language generation, planning and reinforcement learning. Experiments show that our approach increases the endtask reward achieved by the model, improves the effectiveness of long-term planning using rollouts, and allows self-play reinforcement learning to improve decision making without diverging from human language. Our hierarchical latentvariable model outperforms previous work both linguistically and strategically.