Table 5 Results on flat NER datasets.
From: A diffusion enhanced CRF and BiLSTM framework for accurate entity recognition
Model | CoNLL2003 | ||
|---|---|---|---|
Pr.(%) | Rec.(%) | F1(%) | |
Devlin et al. (2019) | - | - | 92.8 |
Yu et al. (2020) | 92.74 | 92.15 | 92.35 |
Li et al. (2020b) | 92.47 | 92.95 | 92.70 |
Sequence Labeling BERT | 91.93 | 91.54 | 91.73 |
Sequence Labeling BART | 89.60 | 91.63 | 90.60 |
BERT-MRC | 90.61 | 91.55 | 91.08 |
PromptNER [BERT-large] | 92.48 | 92.33 | 92.41 |
Shen et al. (2023) | 92.83 | 92.50 | 92.65 |
Ours | 92.54 | 93.10 | 92.79 |
