Medical Question Answering on MedicationQA full n=674 (test)
22F1 ScoreMedBioLM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MedBioLMSetup=FT + RAG2026.05 | 22 | 18.7 | |
| Qwen3-4B MedNLI-ClsSetup=4B + RL2026.05 | 19.1 | 15.3 | |
| Qwen3-4B Likelihood-NLISetup=4B + RL2026.05 | 18.5 | 14.1 | |
| Qwen2.5-7B MedNLI-ClsSetup=7B + RL2026.05 | 17 | 14.4 | |
| GPT-4o (FT)Setup=FT2026.05 | 15.8 | 13.4 | |
| Qwen2.5-7B Likelihood-NLISetup=7B + RL2026.05 | 15 | 13.7 | |
| Qwen2.5-7B GPT-NLISetup=7B + RL2026.05 | 14.9 | 12.7 | |
| Qwen2.5-7B HybridSetup=7B + RL2026.05 | 14 | 12 |