Fine-tune BERT for Extractive Summarization
2019-03-25arXiv 2019Code Available1· sign in to hype
Yang Liu
Code Available — Be the first to reproduce this paper.
ReproduceCode
- github.com/nlpyang/BertSumOfficialIn paperpytorch★ 0
- github.com/HHousen/TransformerSumpytorch★ 439
- github.com/nakhunchumpolsathien/TR-TPBSnone★ 29
- github.com/vsubramaniam851/typ_embedpytorch★ 0
- github.com/aikawasho/BertSumpytorch★ 0
- github.com/thangarani/bertsumpytorch★ 0
- github.com/lingyu001/nlp_text_summarization_implementationpytorch★ 0
- github.com/raqoon886/KorBertSumpytorch★ 0
- github.com/TidalPaladin/neural-summarizerpytorch★ 0
- github.com/raqoon886/KoBertSumpytorch★ 0
Abstract
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
Tasks
Benchmark Results
| Dataset | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| CNN / Daily Mail | BERTSUM+Transformer | ROUGE-1 | 43.25 | — | Unverified |