List your company

Uncertainty Quantification on Countdown, GSM8K, MATH500, SVAMP Combined

63.7Average ROC-AUCDiSE
52.7855.61558.4561.285Mar 3, 2026

Evaluation results

MethodLinks
DiSEBackbone=LLaDA-Instruct-8B2026.0363.7
DiSEBackbone=LLaDA-1.5-8B2026.0359.2
LLaMA perplexityBackbone=LLaDA-Instruct-8B, Evaluation Model=LLaMA3-Instruct-8B2026.0357.8
MCBackbone=LLaDA-Instruct-8B, Nmc=322026.0357.3
LLaMA perplexityBackbone=LLaDA-1.5-8B, Evaluation Model=LLaMA3-Instruct-8B2026.0357.2
MCBackbone=LLaDA-1.5-8B, Nmc=322026.0356.7
MCBackbone=LLaDA-1.5-8B, Nmc=12026.0353.5
MCBackbone=LLaDA-Instruct-8B, Nmc=12026.0353.2