Hallucination Detection on MIMIC-VQA
90.89PR AUCCoEV
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CoEVModel Type=General-purpose, Base Model=Qwen2-3B2026.06 | 90.89 | 72.65 | |
| CoEVModel Type=Medical-tuned, Base Model=Lingshu-7B2026.06 | 85.1 | 70.6 | |
| RadFlagModel Type=Medical-tuned, Base Model=Lingshu-7B2026.06 | 84.1 | 61.37 | |
| CoEVModel Type=General-purpose, Base Model=InternVL3-2B2026.06 | 81.31 | 76.95 | |
| RadFlagModel Type=General-purpose, Base Model=Qwen2-3B2026.06 | 79.91 | 61.08 | |
| RadFlagModel Type=Medical-tuned, Base Model=LLaVA-Med2026.06 | 79.54 | 54.11 | |
| CoEVModel Type=Medical-tuned, Base Model=LLaVA-Med2026.06 | 72.06 | 76.12 | |
| VASEModel Type=General-purpose, Base Model=Qwen2-3B2026.06 | 71.01 | 64.34 | |
| RadFlagModel Type=General-purpose, Base Model=InternVL3-2B2026.06 | 70.54 | 62.71 | |
| SEModel Type=Medical-tuned, Base Model=Lingshu-7B2026.06 | 66.69 | 61.79 | |
| VASEModel Type=Medical-tuned, Base Model=Lingshu-7B2026.06 | 65.29 | 65.51 | |
| VASEModel Type=General-purpose, Base Model=InternVL3-2B2026.06 | 62.79 | 68.06 | |
| VASEModel Type=Medical-tuned, Base Model=LLaVA-Med2026.06 | 62.57 | 71.08 | |
| SEModel Type=Medical-tuned, Base Model=LLaVA-Med2026.06 | 62.4 | 62 | |
| SEModel Type=General-purpose, Base Model=Qwen2-3B2026.06 | 61.66 | 60.18 | |
| SEModel Type=General-purpose, Base Model=InternVL3-2B2026.06 | 60.93 | 64.1 |