Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Fine-tune BERT for Extractive Summarization

About

BERT, a pre-trained Transformer model, has achieved ground-breaking performance on multiple NLP tasks. In this paper, we describe BERTSUM, a simple variant of BERT, for extractive summarization. Our system is the state of the art on the CNN/Dailymail dataset, outperforming the previous best-performed system by 1.65 on ROUGE-L. The codes to reproduce our results are available at https://github.com/nlpyang/BertSum

Yang Liu• 2019

Related benchmarks

TaskDatasetResultRank
SummarizationCNN Daily Mail
ROUGE-143.25
67
Extractive SummarizationCNN/Daily Mail (test)
ROUGE-143.25
36
Extractive SummarizationNYT50 (test)
ROUGE-146.66
21
SummarizationCNN/Daily Mail full length (test)
ROUGE-143.25
18
Extractive SummarizationCNN-DM (test)
ROUGE-143.23
18
SummarizationNYT50 limited length (test)
ROUGE-146.66
8
SummarizationCNN/Daily Mail (test)
Relevance58
8
Headline GenerationPANCO (test)
R1 (ROUGE-1)28.09
7
Extractive SummarizationSCILIT (test)
ROUGE-142.53
2
Showing 9 of 9 rows

Other info

Code

Follow for update