Named Entity Recognition on CoNLL Cased (test)
92.8F1 ScoreDevlin et al. 2019
Evaluation Results
| Method | Links | |
|---|---|---|
| Devlin et al. 2019Train Case=Cased2019.12 | 92.8 | |
| Clark et al. 2018Train Case=Cased2019.12 | 92.6 | |
| Chiu and Nichols 2016Train Case=Cased2019.12 | 91.6 | |
| BiLSTM-CRF+BERT uncased + truecaserTrain Case=Uncased, Embeddings=BERT uncased, Truecaser=included2019.12 | 91.2 | |
| BiLSTM-CRF+BERT uncasedTrain Case=Uncased, Embeddings=BERT uncased2019.12 | 91 | |
| Lample et al. 2016Train Case=Cased2019.12 | 90.9 | |
| Data augmentation (Mayhew et al. 2019)Train Case=Cased+Uncased2019.12 | 90.4 | |
| GloVe + Gold case vectorsTrain Case=Uncased, Embeddings=GloVe + Gold case vectors2019.12 | 90.4 | |
| BiLSTM-CRF+GloVe uncased + truecaserTrain Case=Cased, Embeddings=GloVe uncased, Truecaser=included2019.12 | 90.3 | |
| BiLSTM-CRF+GloVe uncasedTrain Case=Cased, Embeddings=GloVe uncased2019.12 | 90.2 | |
| BiLSTM-CRF+GloVe uncased + truecaserTrain Case=Uncased, Embeddings=GloVe uncased, Truecaser=included2019.12 | 88.3 | |
| BiLSTM-CRF+GloVe uncasedTrain Case=Uncased, Embeddings=GloVe uncased2019.12 | 87.3 |