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

<div>The framework of the BCTCE (bichannel transformer with context encoding). The left is the utterance encoder. The center is the context encoder which has binary channels. The right is the decoder for response generation.</div>

Complexity

fig1

Figure 1

Figure 1: A Bichannel Transformer with Context Encoding for Document-Driven Conversation Generation in Social Media 