Generating High-Quality and Informative Conversation Responses with Sequence-to-Sequence Models

   Abstract