Vision‑language models (VLMs) can lose accuracy when images are resized, even with minimal changes. The study shows that such small visual configuration changes—like tiling or token arrangement—cause more correctness flips across multiple checkpoints and benchmarks. Interestingly, in many cases the models still read the correct answer but fail to use it, and attention interventions reveal that configuration shifts weaken the use of readable information. By guiding models with field cues and their own transcriptions, the authors correct 97.2% of these errors.
By Dingyang Lin, Yingfeng Luo, Chenglong Wang, Chenwei Zhu, Anxiang Ma, Jingbo Zhu, Tong Xiao
arXiv:2609.24172v1 Announce Type: new
Abstract: Legends are fundamental to chart understanding, as reliable interpretation requires correctly binding legend entries to corresponding visual marks. Whi...
By Xinnuo Zhang, Zhike Tang, Jing Xu, Haoyuan Zhao, Weikai Yang
The paper introduces a causal and temporal evaluation framework for vision‑language models (VLMs) that tracks how visual input, question text, and generated prefixes influence autoregressive decoding. It defines three step‑indexed causal‑drive metrics—Visual Causal Drive (VCD), Question Causal Drive (QCD), and Prefix Causal Drive (PCD)—using a Structural Causal Model and interventions. Experiments on Qwen3‑VL‑8B‑Instruct and other datasets show a shift from early question and visual guidance to increased reliance on generated prefixes, and demonstrate that QCD and PCD reduce recovery error and improve bias detection.
By Shuyao Xiao, Shengling Wang, Haoyu Niu, Ke Chao, Changwei Xu, Xinran Duan, Chaoyong Jiang
Chain-of-Thought (CoT) prompting is widely used as a test-time scaling strategy for Vision-Language Models (VLMs), but it remains unclear what is extended when VLMs generate longer reasoning traces. We ask whether CoT requires continued access to image tokens, or whether it mainly operates over visual information already made available earlier in the forward pass.
The paper investigates how general vision‑language models (VLMs) develop specialized optical character recognition (OCR) capabilities. By applying a causal intervention protocol, the authors identify sparse, stable OCR‑head sets in several VLMs and show that these heads largely overlap with textual retrieval/copy heads found in general VLMs. The study concludes that full‑sequence OCR functions as a dense multimodal copy‑and‑paste mechanism, and that when a VLM is fine‑tuned for OCR, it largely preserves the same head identities while redistributing their functional and causal strengths.
By Yuanxiang Huangfu, Hanmeng Zhong, Linqing Chen, Jeffrey Tiong Jee Hui
Vision-language models (VLMs) are increasingly evaluated on complex image and video understanding tasks, yet conventional metrics primarily assess final-answer quality and reveal little about how diff...
arXiv:2607. 12815v1 Announce Type: new Abstract: Chain-of-Thought (CoT) prompting is widely used as a test-time scaling strategy for Vision-Language Models (VLMs), but it remains unclear what is extended when VLMs generate longer reasoning traces.
By Hiroto Osaka, Shohei Taniguchi, Gouki Minegishi, Kai Yamashita, Masahiro Suzuki, Yutaka Matsuo
arXiv:2606. 19965v1 Announce Type: cross Abstract: Multimodal large language models (MLLMs) are increasingly expected to act on visual information, yet the same scene may require different actions under different task contexts.
By Yihao Wang, Zijian He, Jie Ren, Keze Wang
arXiv:2609.13308v1 Announce Type: cross
Abstract: A companion evaluation found that naming the target part in a manipulation prompt increased action accuracy by 0.32-0.63 across eight vision-language...
By Sarthak Sattigeri
arXiv:2609.26093v1 Announce Type: new
Abstract: Vision-language models can answer spatial relation questions confidently even when the image supports an incompatible relation. We formulate relation-g...
By Feixiang Liu, Qiang Qiu, Qingyang Li, Hui Xu
HALDETECT is a system developed for the English hallucination-detection track of ImageEval 2026, where the task is to identify the single visually grounded statement among three culturally plausible options. The approach treats the problem as a contrastive decision, outputs the answer before an explanation, and bases reasoning on colour/texture, shape/form, and context. The best model fine‑tunes Qwen2.5‑VL‑7B‑Instruct with 4‑bit QLoRA, freezes the vision encoder, and achieves a Contrastive Instability score of 0.035 on the test set, placing third among eight teams.
By Syed Mohaiminul Hoque, Md Sakhawat Hossain
arXiv:2609.16646v1 Announce Type: new
Abstract: When strong multimodal models are widely available, progress requires new scientific methodologies beyond benchmark scores---using models as instrument...
By Zhipeng Zhao, Wenxu Wang, Peishun Liu, Ruichun Tang