Medical Domain Question Answering on HealthBench In-domain, Seen
22.97ScoreEvoRubrics
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| EvoRubricsModel=Qwen3-8B, Category=Ours2026.06 | 22.97 | 0.4227 | |
| EvoRubricsModel=Qwen3-4B, Category=Ours2026.06 | 20.8 | 0.3664 | |
| GoldenRubricsModel=Qwen3-8B, Category=RL w/ Rubric2026.06 | 20.47 | 0.3833 | |
| OnlineRubricsModel=Qwen3-8B, Category=RL w/ Rubric2026.06 | 20.23 | 0.3774 | |
| SFTModel=Qwen3-8B, Category=Supervised2026.06 | 18.52 | 0.3427 | |
| PromptModel=Deepseek-V4-Flash, Category=Vanilla2026.06 | 18.04 | 0.3303 | |
| PromptModel=Qwen3-32B, Category=Vanilla2026.06 | 17.17 | 0.3207 | |
| GoldenRubricsModel=Qwen3-4B, Category=RL w/ Rubric2026.06 | 15.82 | 0.2775 | |
| PromptModel=Qwen3-8B, Category=Vanilla2026.06 | 14.4 | 0.2592 | |
| SFTModel=Qwen3-4B, Category=Supervised2026.06 | 13.04 | 0.2375 | |
| PromptModel=Qwen3-4B, Category=Vanilla2026.06 | 12.62 | 0.2248 | |
| RuscaRLModel=Qwen3-4B, Category=RL w/ Rubric2026.06 | 8.85 | 0.1889 | |
| RuscaRLModel=Qwen3-8B, Category=RL w/ Rubric2026.06 | 8.24 | 0.1423 | |
| OnlineRubricsModel=Qwen3-4B, Category=RL w/ Rubric2026.06 | 5.92 | 0.0804 |