Grammatical Error Correction on FCGEC (test)
69.44PrecisionAlirector
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| AlirectorModel=BART2024.02 | 69.44 | 36.6 | 58.88 | |
| CopyModel=BART2024.02 | 65.31 | 39.45 | 57.74 | |
| AlirectorModel=Baichuan2-7B2024.02 | 64.49 | 36.22 | 55.78 | |
| Vanilla FTModel=BART2024.02 | 63.85 | 40.16 | 57.11 | |
| CopyModel=Baichuan2-7B2024.02 | 62.14 | 35.47 | 54.01 | |
| Vanilla FTModel=Baichuan2-7B2024.02 | 60.12 | 37.21 | 53.53 | |
| BaichuanParameters=13B, Evaluation Protocol=LoRA Fine-tuning2024.05 | 56.65 | 27.11 | 46.52 | |
| DeGLM-CoGLMParameters=335M+10B, Evaluation Protocol=LoRA Fine-tuning2024.05 | 56.09 | 38.02 | 51.22 | |
| DeCoGLMParameters=335M, Evaluation Protocol=Primary2024.05 | 55.75 | 37.91 | 50.96 | |
| DeGLM-CoGLMParameters=335M+335M, Evaluation Protocol=Primary2024.05 | 52.95 | 39.2 | 49.48 | |
| BaichuanParameters=7B, Evaluation Protocol=LoRA Fine-tuning2024.05 | 51.83 | 24.08 | 42.12 | |
| GECToRParameters=110M, Evaluation Protocol=Primary2024.05 | 46.11 | 34.35 | 43.16 | |
| ChatGLM2Parameters=6B, Evaluation Protocol=LoRA Fine-tuning2024.05 | 44.3 | 17.08 | 33.59 | |
| ChatGLM3Parameters=6B, Evaluation Protocol=LoRA Fine-tuning2024.05 | 41.06 | 19.93 | 33.88 | |
| BARTParameters=400M, Evaluation Protocol=Primary2024.05 | 38.38 | 37.62 | 38.23 | |
| GPT-4 + DeGLMEvaluation Protocol=Zero-shot with DeGLM detection2024.05 | 25.6 | 16.98 | 23.24 | |
| GPT-4Evaluation Protocol=Zero-shot2024.05 | 18.83 | 4.08 | 10.93 |