Medical Question Answering on MedQA, MedMCQA, and PubMedQA (Aggregate 1,000 each)
63.67AccuracyProposed
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ProposedModel=Qwen2.5-7B2026.05 | 63.67 | 100 | 99.97 | 10.26 | |
| CoTModel=Qwen2.5-7B2026.05 | 60.03 | 94.4 | 0.2 | 84.78 | |
| ProposedModel=Phi-4-mini2026.05 | 57.67 | 99.13 | 99.2 | 40.54 | |
| DirectModel=Qwen2.5-7B2026.05 | 56.5 | 91.47 | 0.07 | 92.55 | |
| HypMed-v3Model=Qwen2.5-7B2026.05 | 56.23 | 98.6 | 99.97 | 9.73 | |
| HypMed-v3Model=Phi-4-mini2026.05 | 49.43 | 93.1 | 99.2 | 43.41 | |
| CoTModel=Phi-4-mini2026.05 | 48.17 | 79.23 | 0 | 98.38 | |
| ProposedModel=DeepSeek-R1-32B2026.05 | 38.77 | 72.13 | 44.3 | 2.13 | |
| HypMed-v3Model=DeepSeek-R1-32B2026.05 | 34.13 | 57.93 | 44.3 | 2.66 | |
| DirectModel=Phi-4-mini2026.05 | 30.2 | 54.03 | 0 | 83.29 | |
| DirectModel=DeepSeek-R1-32B2026.05 | 16.9 | 35.57 | 0.1 | 29.86 | |
| ProposedModel=BioMistral-7B2026.05 | 13.07 | 33.3 | 0.4 | 0 | |
| CoTModel=BioMistral-7B2026.05 | 12.47 | 32.03 | 0 | 82.87 | |
| CoTModel=DeepSeek-R1-32B2026.05 | 7.1 | 23.7 | 0.1 | 0 | |
| DirectModel=BioMistral-7B2026.05 | 0.8 | 1.47 | 0 | 13.1 | |
| HypMed-v3Model=BioMistral-7B2026.05 | 0.07 | 0.4 | 0.4 | 0 |