Hallucination Detection on VQA-Med Open-Ended 2019 (test)
78.14AUCVASE + V-Loop
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VASE + V-LoopBackbone=InternVL3-8B [20]2026.01 | 78.14 | 61.57 | |
| RadFlag + V-LoopBackbone=InternVL3-8B [20]2026.01 | 76.14 | 58.23 | |
| VASE + V-LoopBackbone=Lingshu-7B [40]2026.01 | 76.01 | 68.6 | |
| SE + V-LoopBackbone=InternVL3-8B [20]2026.01 | 75.6 | 58.99 | |
| VASE + V-LoopBackbone=MedGemma-4B-it [34]2026.01 | 73.51 | 54.67 | |
| RadFlag + V-LoopBackbone=Lingshu-7B [40]2026.01 | 73.42 | 66.43 | |
| SE + V-LoopBackbone=Lingshu-7B [40]2026.01 | 72.51 | 66.33 | |
| VASEBackbone=InternVL3-8B [20]2026.01 | 72.35 | 54.67 | |
| RadFlag + V-LoopBackbone=MedGemma-4B-it [34]2026.01 | 71.11 | 53.89 | |
| VASEBackbone=MedGemma-4B-it [34]2026.01 | 70.92 | 51.46 | |
| SE + V-LoopBackbone=MedGemma-4B-it [34]2026.01 | 69.02 | 51.93 | |
| VASEBackbone=Lingshu-7B [40]2026.01 | 68.02 | 61.23 | |
| RadFlagBackbone=InternVL3-8B [20]2026.01 | 67.69 | 53.58 | |
| SEBackbone=InternVL3-8B [20]2026.01 | 66.18 | 52.19 | |
| RadFlagBackbone=MedGemma-4B-it [34]2026.01 | 64.91 | 36.42 | |
| SEBackbone=MedGemma-4B-it [34]2026.01 | 61.88 | 47.02 | |
| RadFlagBackbone=Lingshu-7B [40]2026.01 | 55.24 | 40.67 | |
| SEBackbone=Lingshu-7B [40]2026.01 | 51.8 | 39.39 |