[论文] HALP: Detecting Hallucinations in Vision-Language Models without Gener...

论文概要

研究领域: NLP 作者: Sai Akhil Kogilathota, Sripadha Vallabha E G, Vineeth N Balasubramanian 发布时间: 2026-03-06 arXiv: 2603.05465

中文摘要

视觉语言模型（VLM）在理解和生成图像内容描述方面展现出令人印象深刻的能力。然而，它们容易产生幻觉——生成与视觉内容不符的描述。现有的幻觉检测方法通常需要生成输出令牌然后进行验证，计算成本高昂。本文提出了 HALP（通过潜在投影检测幻觉），一种无需生成单个令牌即可检测 VLM 幻觉的方法。HALP 利用模型的内部表示来识别模型可能产生幻觉的情况，从而在推理时实现高效且有效的幻觉检测。

原文摘要

Vision-Language Models (VLMs) have shown impressive capabilities in understanding and generating content about images. However, they are prone to hallucinations - generating descriptions that are not grounded in the visual content. Existing methods for detecting hallucinations typically require generating output tokens and then verifying them, which is computationally expensive. In this work, we propose HALP (Hallucination Detection via Latent Projection), a method that can detect hallucinations in VLMs without generating a single token. HALP leverages the internal representations of the model to identify when the model is likely to hallucinate, enabling efficient and effective hallucination detection at inference time.

--- *自动采集于 2026-03-07*

#论文 #arXiv #NLP #小凯