The study investigates how unified vision‑language models (VLMs) can simultaneously support visual understanding and generation. Using controlled benchmarks (SmartWatch and modified CelebA) that pair VQA, captioning, and text‑to‑image tasks, the authors evaluate several LLM‑based architectures built on SigLIP and VQ‑VAE visual spaces. Results show that mixed training can improve both understanding and generation, but the gains depend on how well the visual input and output spaces are aligned; misaligned or distorted visual spaces can weaken or reverse these benefits. The paper also demonstrates that balancing data across tasks and controlling attribute frequencies can help recover underrepresented visual concepts, and that the transfer is driven more by the base language model’s learned relationships than by visual adapters.
By Jihai Zhang, Tianle Li, Linjie Li, Zhengyuan Yang, Yu Cheng
arXiv:2606. 15631v1 Announce Type: cross Abstract: Extending a vision-language-action (VLA) policy to a new task typically requires task-specific teleoperated demonstrations and per-task fine-tuning, making adaptation costly in both data collection and compute.
By Jeongeun Park, Juhan Park, Taekyung Kim, Sungjoon Choi, Dongyoon Han, Sangdoo Yun
TwinICL is a procedurally generated benchmark that pairs matched text and image versions of tasks to enable controlled comparison of in‑context learning (ICL) across modalities. Experiments on six open‑weight models and 38 tasks show that multimodal ICL consistently underperforms text‑only ICL, with varying gaps by task family. Interventions targeting visual access, task framing, and reasoning can recover strong multimodal performance on a diagnostic subset, yet a modality gap remains even when explicit task instructions are provided, highlighting the dual role of demonstrations as context and evidence.
By Zihan Xue, Po-Yi Lu, Serhii Honcharenko, Zih-Ching Chen, Hsuan-Tien Lin, Nanyun Peng, I-Hung Hsu, Kuan-Hao Huang
arXiv:2602. 00344v2 Announce Type: replace-cross Abstract: While Retrieval-Augmented Generation (RAG) is one of the dominant paradigms for enhancing Large Vision-Language Models (LVLMs) on knowledge-based VQA tasks, recent work attributes RAG failures to insufficient attention towards the retrieved context, proposing to reduce the attention allocated to image tokens.
By Beidi Zhao, Wenlong Deng, Xinting Liao, Yushu Li, Nazim Shaikh, Yao Nie, Xiaoxiao Li
arXiv:2606. 04434v1 Announce Type: cross Abstract: Multimodal In-Context Learning (ICL) has emerged as a practical inference paradigm for Multimodal Large Language Models, where a small set of interleaved image-text In-Context Demonstrations (ICDs) conditions the model to solve new tasks.
By Niloufar Alipour Talemi, Hossein Kashiani, Fatemeh Afghah
arXiv:2610.02117v1 Announce Type: cross
Abstract: On-policy self-distillation has recently emerged as an effective approach for improving language-model reasoning by supervising students with a froze...
By Sophia Sirko-Galouchenko, Monika Wysoczanska, Andrei Bursuc, Nicolas Thome, Spyros Gidaris
arXiv:2607. 07179v1 Announce Type: cross Abstract: Document Visual Question Answering (DocVQA) presents a complex multimodal challenge, requiring models to exploit visual, textual, and layout information from documents.
By Miguel Lopez-Duran, Elena Marrero, Julian Fierrez, Marta Robledo-Moreno, Ruben Vera-Rodriguez, Daniel DeAlcala, Aythami Morales, Ruben Tolosana, Oscar Delgado, Alvaro Ortigosa, Javier Ortega-Garcia
arXiv:2604.09531v2 Announce Type: replace-cross
Abstract: Vision-language models (VLMs) still struggle with visual perception tasks such as spatial understanding and viewpoint recognition, largely be...
By Guanyu Zhou, Yida Yin, Wenhao Chai, Shengbang Tong, Xingyu Fu, Zhuang Liu
arXiv:2606. 02735v1 Announce Type: cross Abstract: Generalization remains a central bottleneck for vision-language-action (VLA) models: under distractors, appearance shifts, and semantically similar tasks, the policy must often infer local execution details from coarse instructions while also deciding which parts of the image matter for control.
By Yueh-Hua Wu, Tatsuya Matsushima, Kei Ota
Hidden‑Shot introduces an implicit prompt mechanism that extracts task‑specific visual information and merges it with in‑task processing to boost one‑shot performance on new low‑level vision tasks. The method injects this prompt cost‑effectively while minimally altering the base generalist model’s architecture. A data‑driven evaluation framework, C/U assessment, is proposed to systematically test generalization across conventional and unconventional tasks, and experiments on seven and ten datasets show Hidden‑Shot outperforming state‑of‑the‑art models.
By Shao-Jun Xia, Xianzheng Ma, Zichong Meng
The paper investigates why multimodal large language models (MLLMs) struggle with vision‑centric tasks when visual evidence conflicts with pretrained language knowledge. Using image reconstruction and a new WhatIfVis benchmark, the authors show that MLLMs preserve coarse‑grained visual attributes but fail to consistently use them, and that supervised fine‑tuning and activation patching can improve controllability of visual context sensitivity. The study demonstrates that the main bottleneck lies in the models’ inability to reliably regulate their reliance on visual evidence rather than in visual perception itself.
By Jiaang Li, Chengzu Li, Zhaochong An, Yifei Yuan, Xi Liu, Serge Belongie, V\'esteinn Sn{\ae}bjarnarson
arXiv:2609.37250v1 Announce Type: cross
Abstract: World-action models (WAMs) couple future visual-state prediction with action generation. By adapting video generators or image-editing models pretrai...
By Yang Zhang, Jiangyuan Zhao, Chenyou Fan, Jiayu Hu, Xiu Yuan, Chenjia Bai, Xiu Li