Constituency Parsing on Penn Treebank WSJ section 22 (dev)
93.5F1 ScoreLSTM+A ensemble
Evaluation results
| Method | Links | |
|---|---|---|
| LSTM+A ensembleTraining Set=high-confidence corpus2014.12 | 93.5 | |
| LSTM+ATraining Set=high-confidence corpus2014.12 | 93.3 | |
| Huang & Harper (2010) ensembleTraining Set=semi-supervised2014.12 | 92.8 | |
| Petrov et al. (2010) ensembleTraining Set=WSJ only2014.12 | 92.5 | |
| McClosky et al. (2006)Training Set=semi-supervised2014.12 | 92.4 | |
| Petrov et al. (2006)Training Set=WSJ only2014.12 | 91.1 | |
| baseline LSTMTraining Set=BerkeleyParser corpus2014.12 | 91 | |
| LSTM+A+D ensembleTraining Set=WSJ only2014.12 | 90.7 | |
| LSTM+A+DTraining Set=WSJ only2014.12 | 88.7 |