List your company

Uncertainty Estimation on Natural Questions (NQ)

0.77AUROCProposed
0.47880.55440.630.7056Apr 30, 2026

Evaluation results

MethodLinks
ProposedBackbone Model=LLaMA-3-8B-Instruct2026.040.770.81
SARBackbone Model=LLaMA-3-8B-Instruct2026.040.750.8
PerplexityBackbone Model=LLaMA-3-8B-Instruct2026.040.670.73
FocusBackbone Model=LLaMA-3-8B-Instruct2026.040.640.69
SemanticBackbone Model=LLaMA-3-8B-Instruct2026.040.630.68
P(True)Backbone Model=LLaMA-3-8B-Instruct2026.040.610.65
EigenBackbone Model=LLaMA-3-8B-Instruct2026.040.570.64
ATRMDBackbone Model=LLaMA-3-8B-Instruct2026.040.540.62
AttentionBackbone Model=LLaMA-3-8B-Instruct2026.040.490.57