Hallucination Detection on TriviaQA (out-of-distribution)
80.44AUCHalluSAE
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| HalluSAECategory=Ours2026.04 | 80.44 | 76.2 | 75.6 | |
| HaloScopeCategory=Supervised2026.04 | 78.94 | 72.21 | 72.83 | |
| Semantic EntropyCategory=Uncertainty2026.04 | 71.85 | 69.21 | 71.88 | |
| LN-EntropyCategory=Uncertainty2026.04 | 70.94 | 65.75 | 66.5 | |
| Lexical SimilarityCategory=Consistency2026.04 | 65.4 | 64.25 | 53.5 | |
| SAPLMACategory=Supervised2026.04 | 64.33 | 59.5 | 55.12 | |
| SelfCheckGPTCategory=Consistency2026.04 | 63.53 | 61.51 | 57.12 | |
| EigenScoreCategory=Internal State2026.04 | 61.51 | 60.25 | 66.5 | |
| MHADCategory=Internal State2026.04 | 61.01 | 60.75 | 76.51 |