HD-NDEs: Neural Differential Equations for Hallucination Detection in LLMs

TL;DR

This paper introduces HD-NDEs, a neural differential equations-based method to detect hallucinations in large language models by modeling their dynamics in latent space, significantly improving detection accuracy.

Contribution

The paper presents a novel neural differential equations approach for hallucination detection in LLMs, capturing full model dynamics for more reliable truth assessment.

Findings

Over 14% improvement in AUC-ROC on True-False dataset

Effective across five datasets and six LLMs

Outperforms existing classification-based methods

Abstract

In recent years, large language models (LLMs) have made remarkable advancements, yet hallucination, where models produce inaccurate or non-factual statements, remains a significant challenge for real-world deployment. Although current classification-based methods, such as SAPLMA, are highly efficient in mitigating hallucinations, they struggle when non-factual information arises in the early or mid-sequence of outputs, reducing their reliability. To address these issues, we propose Hallucination Detection-Neural Differential Equations (HD-NDEs), a novel method that systematically assesses the truthfulness of statements by capturing the full dynamics of LLMs within their latent space. Our approaches apply neural differential equations (Neural DEs) to model the dynamic system in the latent space of LLMs. Then, the sequence in the latent space is mapped to the classification space for truth assessment. The extensive experiments across five datasets and six widely used LLMs demonstrate the effectiveness of HD-NDEs, especially, achieving over 14% improvement in AUC-ROC on the True-False dataset compared to state-of-the-art techniques.