跳到正文
arXiv cs.CL· Basavaraj R M, Harshal Chandel·· 23 小时前AI 评分42

视觉语言模型的“接地幻觉”:它们会像理想观察者一样权衡检索文本与图像吗?

The Illusion of Grounding: Do Vision-Language Models Weigh Retrieved Text and Images Like an Ideal Observer?

AI 导读

研究构建了 GIB-200K 数据集(20 万条图像-问题样本,源自 COCO 与 Visual Genome),用于测试检索增强视觉语言模型如何融合文本与图像证据。

正文 · 原文

View PDF HTML (experimental)

Abstract:Retrieval-augmented vision-language models (VLMs) answer from two sources of evidence that can each fail: the retrieved text can be wrong or noisy, and the image can be degraded or contradicted. We introduce GIB-200K, 200,000 image-question items built from COCO and Visual Genome, each expanded into controlled stress conditions: hard-negative retrieval noise (1-20 documents), visual ambiguity (Gaussian and motion blur, low-resource crops) and cross-modal conflict with counterfactual flip pairs. Instead of only measuring degradation, we ask how models combine the two modalities. From ideal Bayesian cue combination we derive a testable prediction that yields formal definitions of Knowledge Inertia (text over-weighted relative to the model's own image evidence) and Sensory Capture (the reverse), and we estimate the weights with a measurement-error-corrected regression whose recovery of known weights we verify. A counterfactual flip decomposition separates modality bias from concept priors, and a Grounding Illusion Score measures how much clean accuracy is not grounded in the image. We evaluate 4 open models on one frozen, seeded evaluation split of 2,000 GIB-200K items. All models are accurate on clean inputs (92.5%-94.2%) but adverse accuracy falls to 41.3%-59.9%. The two LLaVA models over-weight the text (Knowledge Inertia, w_T/w_I = 1.229 and 1.322), whereas InternVL3 favours the image (w_T/w_I = 0.680); between 34.8% and 71.4% of the clean answers that are correct are abandoned when the document lies. A training-free consistency check that compares the answer with and without retrieval trades coverage for accuracy under misleading documents, while falling back to the image fails when the image is heavily degraded. All per-instance outputs, code, and an audit that recomputes every reported open-model number are released.
Comments: 10 pages, 5 figures, 11 tables. Code and data: this https URL (archived at this https URL)
Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
Cite as: arXiv:2611.23127 [cs.LG]
  (or arXiv:2611.23127v1 [cs.LG] for this version)
  https://doi.org/10.48550/arXiv.2611.23127

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Basavaraj R M [view email]
[v1] Tue, 17 Nov 2026 17:45:28 UTC (1,297 KB)

来源:arXiv cs.CL · arxiv.org