Uncertainty Quantification for LLMs through Minimum Bayes Risk: Bridging Confidence and Consistency

TL;DR

This paper introduces a novel uncertainty quantification method for Large Language Models that combines confidence and consistency measures, leading to more reliable predictions across multiple NLP tasks.

Contribution

It presents a new approach linking uncertainty directly to minimum Bayes risk, improving robustness and efficiency over existing methods.

Findings

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

Revealed key characteristics of LLMs affecting UQ performance.

Proposed a unified framework combining confidence and consistency for better uncertainty estimates.

Abstract

Uncertainty quantification (UQ) methods for Large Language Models (LLMs) encompass a variety of approaches, with two major types being particularly prominent: information-based, which focus on model confidence expressed as token probabilities, and consistency-based, which assess the semantic relationship between multiple outputs generated using repeated sampling. Several recent methods have combined these two approaches to boost UQ performance. However, they sometimes fail to outperform much simpler baseline methods. Our work discusses the fundamental approach to constructing uncertainty measures that directly links uncertainty with the minimum Bayes risks achieved by LLM decoding. Building on these findings, we propose a novel approach to integrating model confidence with output consistency, resulting in a family of efficient and robust UQ methods. Our investigation reveals distinctive characteristics of LLMs as probabilistic models, which help to explain why these UQ methods underperform in certain tasks. Based on these findings, we propose a new way of synthesizing model confidence and output consistency, leading to a family of efficient and robust UQ methods. We evaluate our approach across various tasks such as question answering, abstractive summarization, and machine translation, demonstrating sizable improvements over state-of-the-art UQ approaches.