Hallucination Detection on SLAKE Open-Ended
79.94AUCVIHD
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VIHDBackbone=Hulu-4B2026.05 | 79.94 | 81.11 | |
| VIHDBackbone=Lingshu-7B2026.05 | 75.36 | 88.61 | |
| VASEBackbone=Hulu-4B2026.05 | 74.41 | 79.12 | |
| VASEBackbone=Lingshu-7B2026.05 | 72.37 | 85.06 | |
| V-LoopBackbone=Lingshu-7B2026.01 | 67.68 | 84.01 | |
| RadFlagBackbone=InternVL3-8B, Uncertainty-based=true2026.01 | 64.93 | 58.76 | |
| V-LoopBackbone=MedGemma-4B-it2026.01 | 61.87 | 68.6 | |
| SEBackbone=MedGemma-4B-it, Uncertainty-based=true2026.01 | 60.9 | 68.42 | |
| MaxEntBackbone=Hulu-4B2026.05 | 60.45 | 68.22 | |
| AvgEntBackbone=Hulu-4B2026.05 | 60.39 | 68.26 | |
| MaxProbBackbone=Hulu-4B2026.05 | 60.37 | 68.21 | |
| AvgProbBackbone=Hulu-4B2026.05 | 60.3 | 68.24 | |
| V-LoopBackbone=InternVL3-8B2026.01 | 59.97 | 59.77 | |
| MaxEntBackbone=InternVL3-8B2026.01 | 59.53 | 54.51 | |
| RadFlagBackbone=MedGemma-4B-it, Uncertainty-based=true2026.01 | 59.18 | 63.57 | |
| AvgEntBackbone=InternVL3-8B2026.01 | 58.47 | 54.34 | |
| SEBackbone=InternVL3-8B, Uncertainty-based=true2026.01 | 58.24 | 54.03 | |
| RadFlagBackbone=Lingshu-7B2026.05 | 57.07 | 77.14 | |
| MaxEntBackbone=MedGemma-4B-it2026.01 | 56.59 | 60.89 | |
| AvgEntBackbone=MedGemma-4B-it2026.01 | 55.98 | 60.72 | |
| MaxEntBackbone=Lingshu-7B2026.05 | 53.99 | 75.82 | |
| AvgEntBackbone=Lingshu-7B2026.05 | 53.97 | 75.82 | |
| AvgProbBackbone=Lingshu-7B2026.05 | 53.94 | 75.76 | |
| MaxProbBackbone=Lingshu-7B2026.05 | 53.93 | 75.75 | |
| RadFlagBackbone=Lingshu-7B, Uncertainty-based=true2026.01 | 53.29 | 75.05 | |
| MaxEntBackbone=Lingshu-7B2026.01 | 53.16 | 74.39 | |
| AvgEntBackbone=Lingshu-7B2026.01 | 53.13 | 74.38 | |
| SEBackbone=Lingshu-7B2026.05 | 52.72 | 74.13 | |
| SEBackbone=Lingshu-7B, Uncertainty-based=true2026.01 | 50.58 | 73.61 | |
| RadFlagBackbone=Hulu-4B2026.05 | 50 | 59.02 | |
| SEBackbone=Hulu-4B2026.05 | 49.29 | 59.02 | |
| AvgProbBackbone=Lingshu-7B2026.01 | 46.87 | 70.07 | |
| MaxProbBackbone=Lingshu-7B2026.01 | 46.84 | 69.82 | |
| AvgProbBackbone=MedGemma-4B-it2026.01 | 44.02 | 58.77 | |
| MaxProbBackbone=MedGemma-4B-it2026.01 | 43.26 | 56.83 | |
| AvgProbBackbone=InternVL3-8B2026.01 | 41.62 | 46.91 | |
| MaxProbBackbone=InternVL3-8B2026.01 | 40.68 | 45.28 |