Learning to Route LLMs with Confidence Tokens

TL;DR

This paper introduces Self-REF, a training method that enables large language models to express reliable confidence through confidence tokens, improving their ability to route questions and reject uncertain answers in high-stakes applications.

Contribution

The paper proposes Self-REF, a novel lightweight training strategy that teaches LLMs to generate confidence tokens, enhancing their confidence estimation and downstream routing performance.

Findings

Confidence tokens outperform verbalized confidence in routing tasks.

Self-REF improves the reliability of LLM confidence estimates.

Enhanced confidence estimation leads to better decision-making in high-stakes scenarios.

Abstract

Large language models (LLMs) have demonstrated impressive performance on several tasks and are increasingly deployed in real-world applications. However, especially in high-stakes settings, it becomes vital to know when the output of an LLM may be unreliable. Depending on whether an answer is trustworthy, a system can then choose to route the question to another expert, or otherwise fall back on a safe default behavior. In this work, we study the extent to which LLMs can reliably indicate confidence in their answers, and how this notion of confidence can translate into downstream accuracy gains. We propose Self-Reflection with Error-based Feedback (Self-REF), a lightweight training strategy to teach LLMs to express confidence in whether their answers are correct in a reliable manner. Self-REF introduces confidence tokens into the LLM, from which a confidence score can be extracted. Compared to conventional approaches such as verbalizing confidence and examining token probabilities, we demonstrate empirically that confidence tokens show significant improvements in downstream routing and rejection learning tasks.