Teaching Large Language Models to Express Knowledge Boundary from Their Own Signals

TL;DR

This paper introduces CoKE, a method to teach large language models to recognize and express their knowledge boundaries, reducing hallucinations by admitting ignorance when unsure, thus improving accuracy across domains.

Contribution

The paper proposes CoKE, a novel approach that probes and elicits LLMs' knowledge boundaries, enabling them to better admit ignorance and reduce hallucinations.

Findings

CoKE improves LLMs' ability to express knowledge boundaries.

Significant reduction in hallucinations during question answering.

Enhanced in-domain and out-of-domain performance.

Abstract

Large language models (LLMs) have achieved great success, but their occasional content fabrication, or hallucination, limits their practical application. Hallucination arises because LLMs struggle to admit ignorance due to inadequate training on knowledge boundaries. We call it a limitation of LLMs that they can not accurately express their knowledge boundary, answering questions they know while admitting ignorance to questions they do not know. In this paper, we aim to teach LLMs to recognize and express their knowledge boundary, so they can reduce hallucinations caused by fabricating when they do not know. We propose CoKE, which first probes LLMs' knowledge boundary via internal confidence given a set of questions, and then leverages the probing results to elicit the expression of the knowledge boundary. Extensive experiments show CoKE helps LLMs express knowledge boundaries, answering known questions while declining unknown ones, significantly improving in-domain and out-of-domain performance.