Multiple-choice VQA on SKINCON
75.2AccuracyInstFree
Evaluation Results
| Method | Links | |
|---|---|---|
| InstFreeresponse shuffling=true2026.03 | 75.2 | |
| InstFree2026.03 | 69.8 | |
| LLaMA-3.2-11B-Visionfine-tuning=true2026.03 | 59.4 | |
| PubMedVision-7Bfine-tuning=false2026.03 | 47 | |
| Qwen2.5-VL-3Bfine-tuning=true2026.03 | 42 | |
| LLaMA-3.2-11B-Visionfine-tuning=false2026.03 | 39.9 | |
| Qwen2.5-VL-3Bfine-tuning=false2026.03 | 37.7 | |
| MedGemma-4Bfine-tuning=false2026.03 | 14.3 | |
| MedGemma-4Bfine-tuning=true2026.03 | 8.9 | |
| BLIP-2fine-tuning=false2026.03 | 8.7 |