Scientific Paper Feedback Generation on Human Evaluation Feedback Level
72.3Validity RateGemini-3-flash
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Gemini-3-flashn (per model)=652026.04 | 72.3 | 56.9 | 4.42 | 3.4 | |
| GOODPOINT-DPOn (per model)=622026.04 | 58.1 | 40.3 | 3.5 | 2.77 | |
| Qwen3-8Bn (per model)=652026.04 | 41.5 | 32.3 | 2.89 | 2.25 |