Hallucination Detection on SelfAware Llama outputs (test)
58.7AUROCNon-Cont.
Evaluation Results
| Method | Links | |
|---|---|---|
| Non-Cont.Scorer Category=BB2026.06 | 58.7 | |
| GPT-4o judgeScorer Category=Judge2026.06 | 57.8 | |
| Gemini judgeScorer Category=Judge2026.06 | 57.5 | |
| Sem. Sets Conf.Scorer Category=BB2026.06 | 54.8 | |
| Semantic Neg.Scorer Category=BB2026.06 | 54.2 | |
| Sequence Prob.Scorer Category=WB2026.06 | 50.2 | |
| Exact MatchScorer Category=BB2026.06 | 49.9 | |
| EntailmentScorer Category=BB2026.06 | 49.7 | |
| Mean Token Neg.Scorer Category=WB2026.06 | 49.6 | |
| Prob. MarginScorer Category=WB2026.06 | 49.2 | |
| Sonnet judgeScorer Category=Judge2026.06 | 49.2 | |
| Min Token Prob.Scorer Category=WB2026.06 | 47.2 | |
| Llama judgeScorer Category=Judge2026.06 | 45.8 | |
| P(True)Scorer Category=WB2026.06 | 44.7 | |
| Cosine Sim.Scorer Category=BB2026.06 | 37.5 |