---
title: "KnowHal：面向多模态大模型全面幻觉评估的知识驱动基准"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.03782v1"
permalink: "https://feed.astralor.com/items/itm_04e76e8beee3e9fbd4d8868a"
published: "2026-08-04T15:03:29.000Z"
score: 78
channel: "ai"
tags: ["KnowHal","多模态大模型","多模态幻觉","知识幻觉"]
---

# KnowHal：面向多模态大模型全面幻觉评估的知识驱动基准

## AI 点评

> 这篇论文值得关注，因为它把知识幻觉正式拉进了多模态评估的主赛道，不再让知识错误散落在各色小数据集里。对做MLLM评测或想提升模型可信度的团队来说，这1800个配对样本是个很好的起点；对普通研究者，负向问题设计也提醒我们，模型在假前提面前往往过于顺从。

## 速览

幻觉是制约多模态大模型（MLLM）可信度的关键挑战。现有基准多聚焦实体、属性和关系幻觉，知识相关失败常被单独研究，缺乏统一评估框架。为此，KnowHal将知识幻觉显式纳入多模态幻觉评估，覆盖实体、属性、关系、知识四个维度，通过共享图像与实体构建正负配对问题，实现感知错误、知识错误和假前提接受的可控对比。基准包含10个领域50个类别共1800个样本，经半自动管道生成。评测14个代表性MLLM发现，知识维度几乎对所有模型最具挑战性，负向问题导致性能显著下降，反映模型对假前提鲁棒性不足。

## 关键点

- 提出KnowHal基准，首次将知识幻觉与实体、属性、关系幻觉统一到同一评估框架中。
- 采用共享图像和实体的正负配对问题设计，支持感知错误、知识错误与假前提接受的可控对比。
- 覆盖10个领域50个类别共1800个样本，结合LLM辅助、CLIP过滤与人工验证构建。
- 评测14个MLLM发现，知识维度是最大挑战，负向问题显著降低模型表现。

## 全文

Hallucination remains a critical challenge for developing trustworthy Multimodal Large Language Models (MLLMs). While existing benchmarks mainly focus on entity, attribute, and relation hallucinations, knowledge-related failures are often investigated separately, lacking a unified evaluation framework across different hallucination dimensions. To overcome this, we propose \textbf{KnowHal}, a benchmark that explicitly incorporates knowledge hallucination into multimodal hallucination evaluation spanning four dimensions: entity, attribute, relation, and knowledge. KnowHal constructs paired positive and negative questions over shared images and entities, enabling controlled comparisons among perceptual errors, knowledge-related errors, and false-premise acceptance. The benchmark contains 1,800 samples across 10 domains and 50 categories, constructed through a semi-automated pipeline combining LLM assistance, CLIP-based filtering, and human verification. We evaluate 14 representative MLLMs on KnowHal and conduct extensive analyses. Results show that the knowledge dimension consistently presents the greatest challenge for nearly all evaluated models, while most models exhibit substantial performance degradation on negative questions, revealing limited robustness to false premises. By unifying four hallucination dimensions with paired question design, KnowHal addresses an important gap in existing evaluation frameworks and enables a more comprehensive assessment of hallucinations in MLLMs.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_04e76e8beee3e9fbd4d8868a