Constituency Parsing on Cross-domain Test Set (Bio, Dialogue, Forum, Law, Literature, Review)
86.61Accuracy (Bio)Kitaev and Klein (2018)
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Kitaev and Klein (2018)Pre-trained model=BERT, Evaluation setting=Zero-shot, Training source=PTB2021.09 | 86.61 | 86.3 | 86.29 | 92.08 | 86.1 | 83.88 | 86.88 | |
| NFCPre-trained model=BERT, Evaluation setting=Zero-shot, Training source=PTB2021.09 | 86.43 | 89.85 | 88.52 | 95.43 | 90.75 | 88.1 | 89.85 | |
| Liu and Zhang (2017)Pre-trained model=BERT, Evaluation setting=Zero-shot, Training source=PTB2021.09 | 86.33 | 85.56 | 85.42 | 91.5 | 84.84 | 83.53 | 86.2 |