List your company

Hallucination Detection on PopQA held-out 50% (test)

85.7AUROCGPT-4o
67.39672.14876.981.652Jun 1, 2026

Evaluation results

MethodLinks
GPT-4oEnsemble Strategy=Weighted Avg.2026.0685.7
GPT-4oEnsemble Strategy=Best J + Best WB2026.0684.6
GPT-4oEnsemble Strategy=Best J + Best BB2026.0684.1
Llama-3-8BEnsemble Strategy=Weighted Avg.2026.0679.9
Llama-3-8BEnsemble Strategy=Best J + Best WB2026.0679.8
Llama-3-8BEnsemble Strategy=Best J + Best BB2026.0679.8
GPT-4oEnsemble Strategy=Best J2026.0678.9
Llama-3-8BEnsemble Strategy=Best J2026.0678.7
Gem.-1.5-FlEnsemble Strategy=Best J + Best BB2026.0675.3
Gem.-1.5-FlEnsemble Strategy=Weighted Avg.2026.0675.1
Gem.-1.5-FlEnsemble Strategy=Best J + Best WB2026.0670.9
Gem.-1.5-FlEnsemble Strategy=Best J2026.0668.1