Grammatical Error Correction on GMEG-wiki (test)
57.9PrecisionTransformer + BIFI (ours)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Transformer + BIFI (ours)training=unsupervised, critic=LM-Critic2021.09 | 57.9 | 33.6 | 50.6 | |
| Transformer + BIFI with no critictraining=unsupervised, critic=none2021.09 | 53.5 | 27.4 | 44.9 | |
| Transformertraining=unsupervised, data=synthetic paired data as in Awasthi et al. (2019)2021.09 | 52.1 | 26.5 | 43.7 |