Text Generation on HealthBench
44.83RatioEvoRubrics
Evaluation results
| Method | Links | |
|---|---|---|
| EvoRubricsModel=Qwen3-8B2026.06 | 44.83 | |
| Qwen3-8BMethod=Vanilla2026.06 | 43.56 | |
| Qwen3-8BMethod=SFT2026.06 | 41.92 | |
| Deepseek-V4-FlashMethod=Prompt2026.06 | 37.58 | |
| Qwen3-32BMethod=Prompt2026.06 | 31.4 | |
| EvoRubricsModel=Qwen3-4B2026.06 | 27.87 | |
| Qwen3-4BMethod=Vanilla2026.06 | 26.33 | |
| Qwen3-4BMethod=SFT2026.06 | 26.1 |