Don't Throw Away Your Beams: Improving Consistency-based Uncertainties in LLMs via Beam Search

TL;DR

This paper proposes using beam search instead of multinomial sampling for consistency-based uncertainty quantification in large language models, resulting in more reliable estimates and state-of-the-art performance.

Contribution

It introduces a novel beam search-based method for uncertainty quantification, with theoretical analysis and empirical validation showing improved accuracy and reduced variance.

Findings

Beam search improves uncertainty estimation accuracy.

The method achieves state-of-the-art results on six QA datasets.

Theoretical bounds support the effectiveness of beam search over sampling.

Abstract

Consistency-based methods have emerged as an effective approach to uncertainty quantification (UQ) in large language models. These methods typically rely on several generations obtained via multinomial sampling, measuring their agreement level. However, in short-form QA, multinomial sampling is prone to producing duplicates due to peaked distributions, and its stochasticity introduces considerable variance in uncertainty estimates across runs. We introduce a new family of methods that employ beam search to generate candidates for consistency-based UQ, yielding improved performance and reduced variance compared to multinomial sampling. We also provide a theoretical lower bound on the beam set probability mass under which beam search achieves a smaller error than multinomial sampling. We empirically evaluate our approach on six QA datasets and find that its consistent improvements over multinomial sampling lead to state-of-the-art UQ performance.