A Context-Aware Dual-Metric Framework for Confidence Estimation in Large Language Models

TL;DR

This paper introduces CRUX, a novel framework for confidence estimation in large language models that considers context relevance through two new metrics, improving trustworthiness especially in safety-critical applications.

Contribution

CRUX is the first framework to integrate context faithfulness and consistency for confidence estimation in LLMs, using contrastive sampling and global answer consistency metrics.

Findings

CRUX achieves the highest AUROC on multiple benchmark datasets.

It effectively captures data uncertainty and model confidence.

Outperforms existing confidence estimation methods.

Abstract

Accurate confidence estimation is essential for trustworthy large language models (LLMs) systems, as it empowers the user to determine when to trust outputs and enables reliable deployment in safety-critical applications. Current confidence estimation methods for LLMs neglect the relevance between responses and contextual information, a crucial factor in output quality evaluation, particularly in scenarios where background knowledge is provided. To bridge this gap, we propose CRUX (Context-aware entropy Reduction and Unified consistency eXamination), the first framework that integrates context faithfulness and consistency for confidence estimation via two novel metrics. First, contextual entropy reduction represents data uncertainty with the information gain through contrastive sampling with and without context. Second, unified consistency examination captures potential model uncertainty through the global consistency of the generated answers with and without context. Experiments across three benchmark datasets (CoQA, SQuAD, QuAC) and two domain-specific datasets (BioASQ, EduQG) demonstrate CRUX's effectiveness, achieving the highest AUROC than existing baselines.