Grammatical Error Correction on CoNLL-2014 A2
Metric: F0.5 (higher is better)
Metric: F0.5 (higher is better)
| # | Model↕ | F0.5▼ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | VERNet | 63.1 | No | Neural Quality Estimation with Multiple Hypothes... | 2021-05-10 | Code |
| 2 | Bi-LSTM + POS (unrestricted data) | 45.1 | No | Auxiliary Objectives for Neural Error Detection ... | 2017-07-17 | - |
| 3 | Bi-LSTM (unrestricted data) | 44 | No | Compositional Sequence Labeling Models for Error... | 2016-07-20 | - |
| 4 | Ann+PAT+MT | 30.13 | No | Artificial Error Generation with Machine Transla... | 2017-07-17 | - |
| 5 | BiLSTM-JOINT (trained on FCE) | 29.65 | No | Jointly Learning to Label Sentences and Tokens | 2018-11-14 | Code |
| 6 | Bi-LSTM + POS (trained on FCE) | 26.2 | No | Auxiliary Objectives for Neural Error Detection ... | 2017-07-17 | - |
| 7 | Bi-LSTM + LMcost (trained on FCE) | 25.88 | No | Semi-supervised Multitask Learning for Sequence ... | 2017-04-24 | Code |
| 8 | Bi-LSTM (trained on FCE) | 23.9 | No | Compositional Sequence Labeling Models for Error... | 2016-07-20 | - |