Critique Quality Evaluation on Human Evaluation Overall
66Win RateRCO
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RCOComparison Target=LLaMa-3-70B-Instruct2025.06 | 66 | 18.5 | 15.5 | 50.5 | |
| RCOComparison Target=LLaMa-3-8B-Instruct2025.06 | 49 | 24.5 | 26.5 | 22.5 | |
| RCOComparison Target=LLaMa-3-8B-Instruct2025.06 | 48 | 35 | 17 | 31 | |
| RCOComparison Target=LLaMa-3-8B-Instruct + MSE + DPO2025.06 | 47.5 | 25.5 | 27 | 20.5 | |
| RCOComparison Target=LLaMa-3-70B-Instruct2025.06 | 43 | 33.5 | 23.5 | 19.5 | |
| RCOComparison Target=LLaMa-3-8B-Instruct + MSE + DPO2025.06 | 38 | 32.5 | 29.5 | 8.5 |