Hallucination Detection on HaluEval Gemini outputs (test)
0.571AUROCSemantic Neg.
Evaluation Results
| Method | Links | |
|---|---|---|
| Semantic Neg.Scorer Category=BB2026.06 | 0.571 | |
| Sem. Sets Conf.Scorer Category=BB2026.06 | 0.571 | |
| Non-Cont.Scorer Category=BB2026.06 | 0.55 | |
| GPT-4o judgeScorer Category=Judge2026.06 | 0.541 | |
| EntailmentScorer Category=BB2026.06 | 0.54 | |
| Cosine Sim.Scorer Category=BB2026.06 | 0.54 | |
| Exact MatchScorer Category=BB2026.06 | 0.533 | |
| Sequence Prob.Scorer Category=WB2026.06 | 0.53 | |
| P(True)Scorer Category=WB2026.06 | 0.528 | |
| Min Token Prob.Scorer Category=WB2026.06 | 0.525 | |
| Sonnet judgeScorer Category=Judge2026.06 | 0.525 | |
| Mean Token Neg.Scorer Category=WB2026.06 | 0.521 | |
| Prob. MarginScorer Category=WB2026.06 | 0.52 | |
| Gemini judgeScorer Category=Judge2026.06 | 0.514 | |
| Llama judgeScorer Category=Judge2026.06 | 0.506 |