Hallucination Detection on LLaMa 1 (test)
0.894AUROCEigenTrack
Evaluation Results
| Method | Links | |
|---|---|---|
| EigenTrackModel Scale=7B2026.01 | 0.894 | |
| LapEigvalsModel Scale=7B2026.01 | 0.871 | |
| EigenTrackModel Scale=3B2026.01 | 0.861 | |
| HaloScopeModel Scale=7B2026.01 | 0.861 | |
| EigenTrackModel Scale=1B2026.01 | 0.842 | |
| INSIDEModel Scale=3B2026.01 | 0.831 | |
| HaloScopeModel Scale=3B2026.01 | 0.827 | |
| HaloScopeModel Scale=1B2026.01 | 0.82 | |
| LapEigvalsModel Scale=3B2026.01 | 0.819 | |
| INSIDEModel Scale=7B2026.01 | 0.81 | |
| SelfCheckGPTModel Scale=7B2026.01 | 0.809 | |
| SelfCheckGPTModel Scale=3B2026.01 | 0.804 | |
| LapEigvalsModel Scale=1B2026.01 | 0.785 | |
| INSIDEModel Scale=1B2026.01 | 0.753 | |
| SelfCheckGPTModel Scale=1B2026.01 | 0.739 |