MMD-Flagger: Leveraging Maximum Mean Discrepancy to Detect Hallucinations

TL;DR

MMD-Flagger is a novel method that detects hallucinations in large language models by analyzing the distributional differences in outputs generated at different temperature settings, showing promising results in translation and summarization tasks.

Contribution

It introduces a new hallucination detection technique using Maximum Mean Discrepancy, leveraging output distribution shifts across temperature variations.

Findings

Effective in detecting hallucinations in LLM outputs

Competitive performance on translation and summarization datasets

Utilizes distribution shape trajectory for detection

Abstract

Large language models (LLMs) have become pervasive in our everyday life. Yet, a fundamental obstacle prevents their use in many critical applications: their propensity to generate fluent, human-quality content that is not grounded in reality. The detection of such hallucinations is thus of the highest importance. In this work, we propose a new method to flag hallucinated content: MMD-Flagger. It relies on Maximum Mean Discrepancy (MMD), a non-parametric distance between distributions. On a high-level perspective, MMD-Flagger tracks the MMD between the output to inspect and counterparts generated with various temperature parameters. We show empirically that inspecting the shape of this trajectory is sufficient to detect most hallucinations. This novel method is benchmarked on machine translation and summarization datasets, on which it exhibits competitive performance relative to natural competitors.