Factual Confidence of LLMs: on Reliability and Robustness of Current Estimators

TL;DR

This paper systematically compares various factual confidence estimators for LLMs, revealing that trained hidden-state probes are most reliable but less accessible, and highlights the instability of LLM confidence across semantically equivalent inputs.

Contribution

It provides a comprehensive empirical comparison of confidence estimators and introduces an experimental framework for fair evaluation of LLM factual confidence methods.

Findings

Hidden-state probes are most reliable confidence estimators.

LLM confidence is often unstable across semantically equivalent inputs.

The study offers a systematic comparison framework for confidence estimators.

Abstract

Large Language Models (LLMs) tend to be unreliable in the factuality of their answers. To address this problem, NLP researchers have proposed a range of techniques to estimate LLM's confidence over facts. However, due to the lack of a systematic comparison, it is not clear how the different methods compare to one another. To fill this gap, we present a survey and empirical comparison of estimators of factual confidence. We define an experimental framework allowing for fair comparison, covering both fact-verification and question answering. Our experiments across a series of LLMs indicate that trained hidden-state probes provide the most reliable confidence estimates, albeit at the expense of requiring access to weights and training data. We also conduct a deeper assessment of factual confidence by measuring the consistency of model behavior under meaning-preserving variations in the input. We find that the confidence of LLMs is often unstable across semantically equivalent inputs, suggesting that there is much room for improvement of the stability of models' parametric knowledge. Our code is available at (https://github.com/amazon-science/factual-confidence-of-llms).