Instruction Following on Mimic-Instr MIMIC-IV
84.5AccuracyGPT-5
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5Family=Frontier API2026.05 | 84.5 | |
| GPT-4.1Family=Frontier API2026.05 | 65.5 | |
| GPT-4.1-miniFamily=Frontier API2026.05 | 63.7 | |
| MedGemma-27BFamily=Medical / open-source2026.05 | 62.3 | |
| DS-R1-0528Family=Medical / open-source2026.05 | 60.4 | |
| CLR-voyance-8BFamily=8B family, Merge Strategy=DELLA-Linear2026.05 | 58.9 | |
| HuatuoGPT-o1Family=Medical / open-source2026.05 | 55.1 | |
| Qwen3-8B baseFamily=8B family2026.05 | 54.5 | |
| Gemma-3-4BFamily=Medical / open-source2026.05 | 42.7 | |
| MG-Δ+AFamily=4B family, Merge Strategy=DELLA-Linear+AIM2026.05 | 23.2 | |
| CLR-voyance-4BFamily=4B family, Merge Strategy=DELLA-Linear2026.05 | 19.5 | |
| MG-B+AFamily=4B family, Merge Strategy=Breadcrumbs+AIM2026.05 | 19.5 | |
| MG-4B baseFamily=4B family2026.05 | 19.3 | |
| MG-BFamily=4B family, Merge Strategy=Breadcrumbs2026.05 | 19.2 | |
| DS-R1-DistillFamily=Medical / open-source2026.05 | 16.8 |