Visual Question Answering (Closed-ended) on VQA-RAD closed-ended
82.5AccuracyBiasCareVL
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| BiasCareVL2026.04 | 82.5 | 82.6 | 83 | 82.3 | |
| MEDSIGHTFine-tuning protocol=LoRA, LoRA rank=322026.06 | 82.3 | — | — | — | |
| BioMedGPTYear=20242026.04 | 81.3 | — | — | — | |
| HuatuoGPT-Vision 7BFine-tuning protocol=LoRA, LoRA rank=322026.06 | 76.8 | — | — | — | |
| ExGra-MedFine-tuning protocol=LoRA, LoRA rank=322026.06 | 75.2 | — | — | — | |
| LLaVA-Med 1.5Fine-tuning protocol=LoRA, LoRA rank=322026.06 | 74.4 | — | — | — | |
| HuaTuoGPT-Vision-34BYear=2024, Parameters=34B2026.04 | 68.1 | — | — | — | |
| HuaTuoGPT-Vision-7BYear=2024, Parameters=7B2026.04 | 67.5 | 67.8 | 73.6 | 68.5 | |
| MedPLIB-7BYear=2025, Parameters=7B2026.04 | 66.7 | 66.4 | 68.8 | 67.3 | |
| MIMOYear=20252026.04 | 58.8 | — | — | — | |
| RadFMYear=20252026.04 | 50.6 | — | — | — | |
| DeepSeek-VL2-tinyYear=20252026.04 | 48.8 | 40.6 | 58.1 | 46.6 | |
| DeepSeek-VL2-smallYear=20252026.04 | 48 | 49.8 | 60.7 | 46.9 | |
| Qwen-VL-ChatYear=20232026.04 | 47 | — | — | — | |
| Med-FlamingoYear=20232026.04 | 45.4 | — | — | — | |
| Qwen-VL2.5-3B-InstructYear=2025, Parameters=3B2026.04 | 26.6 | 30.6 | 76.1 | 27.8 |