Hallucination Detection on SLAKE (All)
78.91AUCVIHD
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VIHDBackbone=Hulu-4B2026.05 | 78.91 | 84.96 | |
| VIHDBackbone=Lingshu-7B2026.05 | 72.95 | 86.49 | |
| VASEBackbone=Lingshu-7B2026.05 | 70.56 | 84.52 | |
| VASEBackbone=Hulu-4B2026.05 | 70.31 | 81.27 | |
| RadFlagBackbone=InternVL3-8B, Uncertainty-based=true2026.01 | 62.12 | 59.75 | |
| SEBackbone=InternVL3-8B, Uncertainty-based=true2026.01 | 60.61 | 58.31 | |
| V-LoopBackbone=Lingshu-7B2026.01 | 59.79 | 81.83 | |
| RadFlagBackbone=MedGemma-4B-it, Uncertainty-based=true2026.01 | 59.56 | 58.53 | |
| V-LoopBackbone=MedGemma-4B-it2026.01 | 59.31 | 65.15 | |
| MaxProbBackbone=Hulu-4B2026.05 | 59.26 | 72.62 | |
| MaxEntBackbone=Hulu-4B2026.05 | 59.24 | 72.61 | |
| AvgProbBackbone=Hulu-4B2026.05 | 59.13 | 72.59 | |
| AvgEntBackbone=Hulu-4B2026.05 | 59.03 | 72.56 | |
| MaxEntBackbone=InternVL3-8B2026.01 | 58.81 | 55.93 | |
| AvgEntBackbone=InternVL3-8B2026.01 | 58.09 | 55.58 | |
| MaxEntBackbone=MedGemma-4B-it2026.01 | 57.45 | 57.23 | |
| AvgEntBackbone=MedGemma-4B-it2026.01 | 56.98 | 57.2 | |
| SEBackbone=MedGemma-4B-it, Uncertainty-based=true2026.01 | 56.84 | 59.62 | |
| V-LoopBackbone=InternVL3-8B2026.01 | 56.04 | 57.96 | |
| RadFlagBackbone=Lingshu-7B2026.05 | 55.46 | 78.25 | |
| MaxEntBackbone=Lingshu-7B2026.05 | 53.37 | 77.44 | |
| AvgEntBackbone=Lingshu-7B2026.05 | 53.34 | 77.42 | |
| AvgProbBackbone=Lingshu-7B2026.05 | 53.3 | 77.39 | |
| MaxProbBackbone=Lingshu-7B2026.05 | 53.3 | 77.38 | |
| AvgEntBackbone=Lingshu-7B2026.01 | 53.25 | 75.65 | |
| MaxEntBackbone=Lingshu-7B2026.01 | 53.25 | 75.65 | |
| RadFlagBackbone=Lingshu-7B, Uncertainty-based=true2026.01 | 52.52 | 75.44 | |
| SEBackbone=Lingshu-7B2026.05 | 52.45 | 76.44 | |
| SEBackbone=Lingshu-7B, Uncertainty-based=true2026.01 | 50.89 | 75.61 | |
| RadFlagBackbone=Hulu-4B2026.05 | 50 | 68.12 | |
| SEBackbone=Hulu-4B2026.05 | 49.3 | 66.89 | |
| AvgProbBackbone=Lingshu-7B2026.01 | 46.75 | 69.97 | |
| MaxProbBackbone=Lingshu-7B2026.01 | 46.75 | 70.09 | |
| AvgProbBackbone=MedGemma-4B-it2026.01 | 43.01 | 56.26 | |
| MaxProbBackbone=MedGemma-4B-it2026.01 | 42.65 | 55.02 | |
| AvgProbBackbone=InternVL3-8B2026.01 | 41.24 | 46.19 | |
| MaxProbBackbone=InternVL3-8B2026.01 | 40.38 | 44.39 |