Fetching the paper…

ConveRT: Efficient and Accurate Conversational Representations from Transformers · Around