 |
ML Newsnew | past | best | rss | submit | about |
| Improving Natural Language Inference with a Pretrained Parser (arxiv.org) | |
| 1 point by sel1 on Sep 19, 2019 | hide | past | pdf | discuss on HN | |
|
In plain words: The system feeds each word's description from a grammar parser that knows how words connect in a sentence into models that judge whether one sentence follows from another. Compared with the same models without grammar info, accuracy rose on all four across three benchmarks. AbstractWe introduce a novel approach to incorporate syntax into natural language inference (NLI) models. Our method uses contextual token-level vector representations from a pretrained dependency parser. Like other contextual embedders, our method is broadly applicable to any neural model. We experiment with four strong NLI models (decomposable attention model, ESIM, BERT, and MT-DNN), and show consistent benefit to accuracy across three NLI benchmarks. | |
| add comment on HN |