Grammatical Error Correction On Falko Merlin
평가 지표
F0.5
평가 결과
이 벤치마크에서 각 모델의 성능 결과
모델 이름 | F0.5 | Paper Title | Repository |
|---|---|---|---|
| mT5-based multimodal MoE | 76.3 | - | - |
| Transformer - synthetic pretrain only | 51.41 | Grammatical Error Correction in Low-Resource Scenarios | |
| Multilayer Convolutional Encoder-Decoder | 43.35 | Using Wikipedia Edits in Low Resource Grammatical Error Correction | - |
| Transformer | 73.71 | Grammatical Error Correction in Low-Resource Scenarios | |
| gT5 xxl | 75.96 | A Simple Recipe for Multilingual Grammatical Error Correction | |
| Llama + 1M BT + gold | 76.75 | To Err Is Human, but Llamas Can Learn It Too |
0 of 6 row(s) selected.