A Bichannel Transformer with Context Encoding for Document-Driven Conversation Generation in Social Media

Figure 2

Training time of various models.