Fetching the paper…

HIBERT: Document Level Pre-training of Hierarchical Bidirectional Transformers for Document Summarization · Around