List your company

Hallucination Detection on TriviaQA Gemini outputs (test)

0.67AUROCMin Token Prob.
0.532720.568360.6040.63964Jun 1, 2026

Evaluation results

MethodLinks
Min Token Prob.Scorer Category=WB2026.060.67
EntailmentScorer Category=BB2026.060.655
Exact MatchScorer Category=BB2026.060.634
Non-Cont.Scorer Category=BB2026.060.63
Sonnet judgeScorer Category=Judge2026.060.627
Sequence Prob.Scorer Category=WB2026.060.622
Semantic Neg.Scorer Category=BB2026.060.619
Sem. Sets Conf.Scorer Category=BB2026.060.618
Cosine Sim.Scorer Category=BB2026.060.604
Prob. MarginScorer Category=WB2026.060.603
P(True)Scorer Category=WB2026.060.588
Mean Token Neg.Scorer Category=WB2026.060.585
GPT-4o judgeScorer Category=Judge2026.060.572
Llama judgeScorer Category=Judge2026.060.568
Gemini judgeScorer Category=Judge2026.060.538