FREAK: A Fine-grained Hallucination Evaluation Benchmark for Advanced MLLMs

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Yin, Zhihan, Liang, Jianxin, Wang, Yueqian, Yao, Yifeng, Zhang, Huishuai, Zhao, Dongyan
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917354677469184
author Yin, Zhihan
Liang, Jianxin
Wang, Yueqian
Yao, Yifeng
Zhang, Huishuai
Zhao, Dongyan
author_facet Yin, Zhihan
Liang, Jianxin
Wang, Yueqian
Yao, Yifeng
Zhang, Huishuai
Zhao, Dongyan
contents Multimodal Large Language Models (MLLMs) suffer from hallucinations. Existing hallucination evaluation benchmarks are often limited by over-simplified tasks leading to saturated metrics, or insufficient diversity that fails to adequately assess the hallucination extent in state-of-the-art multimodal models. To address this gap, we propose FREAK, a comprehensive multimodal benchmark designed for fine-grained hallucination assessment in MLLMs. Through high-quality photorealistic images featuring fine-grained counter-commonsense edits, FREAK innovatively evaluates hallucination phenomena in detailed visual perception of MLLMs. Extensive experiments on FREAK show severe hallucination issues in SOTA models regarding detailed visual perception. To enable deeper investigation, we curate a controlled subset to indirectly evaluate the model's ability to perceive target detailed information. Through systematic evaluation of prevailing Chain-of-Thought (CoT) prompting techniques within this task, we reveal critical insights regarding hallucination patterns and model reasoning processes.
format Preprint
id arxiv_https___arxiv_org_abs_2603_19765
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle FREAK: A Fine-grained Hallucination Evaluation Benchmark for Advanced MLLMs
Yin, Zhihan
Liang, Jianxin
Wang, Yueqian
Yao, Yifeng
Zhang, Huishuai
Zhao, Dongyan
Computer Vision and Pattern Recognition
Multimodal Large Language Models (MLLMs) suffer from hallucinations. Existing hallucination evaluation benchmarks are often limited by over-simplified tasks leading to saturated metrics, or insufficient diversity that fails to adequately assess the hallucination extent in state-of-the-art multimodal models. To address this gap, we propose FREAK, a comprehensive multimodal benchmark designed for fine-grained hallucination assessment in MLLMs. Through high-quality photorealistic images featuring fine-grained counter-commonsense edits, FREAK innovatively evaluates hallucination phenomena in detailed visual perception of MLLMs. Extensive experiments on FREAK show severe hallucination issues in SOTA models regarding detailed visual perception. To enable deeper investigation, we curate a controlled subset to indirectly evaluate the model's ability to perceive target detailed information. Through systematic evaluation of prevailing Chain-of-Thought (CoT) prompting techniques within this task, we reveal critical insights regarding hallucination patterns and model reasoning processes.
title FREAK: A Fine-grained Hallucination Evaluation Benchmark for Advanced MLLMs
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.19765