[Adapt] [Seminar] HIBERT: Document Level Pre-training of Hierarchical Bidirectional Transformers for Document Summarization

hyy hongyunyan at sjtu.edu.cn
Tue Jun 4 10:53:02 CST 2019


Dear Adapters,
   In this week, I will give you the talk about the paper “HIBERT: Document Level Pre-training of Hierarchical Bidirectional Transformers for Document Summarization” which is accepted by ACL in 2019. 
In this paper,the core part of a neural extractive summarization model is the hierarchical document encoder. And author proposed a method to pre-train document level hierarchical bidirectional transformer encoders on unlabeled data. The HIBERT model achieves a quite good performance in CNNDM dataset and NYT50 dataset. 


Time: Wed 5:00pm
Venue: SEIEE 3-414

Best regards, 
Sophia
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20190604/c157dc66/attachment.html>


More information about the Adapt mailing list