Believe Your Model: Distribution-Guided Confidence Calibration

TL;DR

This paper introduces DistriVoting and SelfStepConf, novel methods that leverage distributional priors and dynamic confidence adjustments to improve answer selection and confidence calibration in large reasoning models.

Contribution

The paper proposes a new distribution-guided voting method and a dynamic confidence adjustment technique to enhance model reliability and accuracy.

Findings

Significant performance improvements over state-of-the-art methods.

Effective distribution decomposition and filtering enhance answer reliability.

Method generalizes across multiple models and benchmarks.

Abstract

Large Reasoning Models have demonstrated remarkable performance with the advancement of test-time scaling techniques, which enhances prediction accuracy by generating multiple candidate responses and selecting the most reliable answer. While prior work has analyzed that internal model signals like confidence scores can partly indicate response correctness and exhibit a distributional correlation with accuracy, such distributional information has not been fully utilized to guide answer selection. Motivated by this, we propose DistriVoting, which incorporates distributional priors as another signal alongside confidence during voting. Specifically, our method (1) first decomposes the mixed confidence distribution into positive and negative components using Gaussian Mixture Models, (2) then applies a reject filter based on positive/negative samples from them to mitigate overlap between the two distributions. Besides, to further alleviate the overlap from the perspective of distribution itself, we propose SelfStepConf, which uses step-level confidence to dynamically adjust inference process, increasing the separation between the two distributions to improve the reliability of confidences in voting. Experiments across 16 models and 5 benchmarks demonstrate that our method significantly outperforms state-of-the-art approaches.