Igor Kuzmin

Set of BERT models with Hierarchical attention pre-trained on conversational data to process multiple utterances at once