Cover Image

Youtu-VL: Unleashing Visual Potential via Unified Vision-Language Supervision

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Wei, Zhixiang, Li, Yi, Kan, Zhehan, Jiang, Xinghua, Long, Zuwei, Liu, Shifeng, Shen, Hongze, Liu, Wei, Tan, Xiaoyu, Lin, Haojia, Zhu, Yubo, Li, Qianyu, Yin, Di, Cao, Haoyu, Gu, Weibo, Li, Xin, Liu, Yinsong, Jiang, Deqiang, Sun, Xing, Wu, Yunsheng, Tang, Mingkong, Liu, Shuangyin, Tang, Lexiang, Lin, Haodong, Lu, Junru, Qin, Jiarui, Qiao, Lingfeng, Qiao, Ruizhi, Ke, Bo, He, Jianfeng, Li, Ke, Li, Yangning, Shen, Yunhang, Zhang, Mengdan, Chen, Peixian, Yin, Kun, Liu, Bing, Wu, Yunfei, Chen, Huang, Cai, Zhongpeng, Li, Xiaotian
Format:	Preprint
Published:	2026
Subjects:	Computer Vision and Pattern Recognition
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

_version_	1866911402910810112
author	Wei, Zhixiang Li, Yi Kan, Zhehan Jiang, Xinghua Long, Zuwei Liu, Shifeng Shen, Hongze Liu, Wei Tan, Xiaoyu Lin, Haojia Zhu, Yubo Li, Qianyu Yin, Di Cao, Haoyu Gu, Weibo Li, Xin Liu, Yinsong Jiang, Deqiang Sun, Xing Wu, Yunsheng Tang, Mingkong Liu, Shuangyin Tang, Lexiang Lin, Haodong Lu, Junru Qin, Jiarui Qiao, Lingfeng Qiao, Ruizhi Ke, Bo He, Jianfeng Li, Ke Li, Yangning Shen, Yunhang Zhang, Mengdan Chen, Peixian Yin, Kun Liu, Bing Wu, Yunfei Chen, Huang Cai, Zhongpeng Li, Xiaotian
author_facet	Wei, Zhixiang Li, Yi Kan, Zhehan Jiang, Xinghua Long, Zuwei Liu, Shifeng Shen, Hongze Liu, Wei Tan, Xiaoyu Lin, Haojia Zhu, Yubo Li, Qianyu Yin, Di Cao, Haoyu Gu, Weibo Li, Xin Liu, Yinsong Jiang, Deqiang Sun, Xing Wu, Yunsheng Tang, Mingkong Liu, Shuangyin Tang, Lexiang Lin, Haodong Lu, Junru Qin, Jiarui Qiao, Lingfeng Qiao, Ruizhi Ke, Bo He, Jianfeng Li, Ke Li, Yangning Shen, Yunhang Zhang, Mengdan Chen, Peixian Yin, Kun Liu, Bing Wu, Yunfei Chen, Huang Cai, Zhongpeng Li, Xiaotian
contents	Despite the significant advancements represented by Vision-Language Models (VLMs), current architectures often exhibit limitations in retaining fine-grained visual information, leading to coarse-grained multimodal comprehension. We attribute this deficiency to a suboptimal training paradigm inherent in prevailing VLMs, which exhibits a text-dominant optimization bias by conceptualizing visual signals merely as passive conditional inputs rather than supervisory targets. To mitigate this, we introduce Youtu-VL, a framework leveraging the Vision-Language Unified Autoregressive Supervision (VLUAS) paradigm, which fundamentally shifts the optimization objective from ``vision-as-input'' to ``vision-as-target.'' By integrating visual tokens directly into the prediction stream, Youtu-VL applies unified autoregressive supervision to both visual details and linguistic content. Furthermore, we extend this paradigm to encompass vision-centric tasks, enabling a standard VLM to perform vision-centric tasks without task-specific additions. Extensive empirical evaluations demonstrate that Youtu-VL achieves competitive performance on both general multimodal tasks and vision-centric tasks, establishing a robust foundation for the development of comprehensive generalist visual agents.
format	Preprint
id	arxiv_https___arxiv_org_abs_2601_19798
institution	arXiv
publishDate	2026
record_format	arxiv
spellingShingle	Youtu-VL: Unleashing Visual Potential via Unified Vision-Language Supervision Wei, Zhixiang Li, Yi Kan, Zhehan Jiang, Xinghua Long, Zuwei Liu, Shifeng Shen, Hongze Liu, Wei Tan, Xiaoyu Lin, Haojia Zhu, Yubo Li, Qianyu Yin, Di Cao, Haoyu Gu, Weibo Li, Xin Liu, Yinsong Jiang, Deqiang Sun, Xing Wu, Yunsheng Tang, Mingkong Liu, Shuangyin Tang, Lexiang Lin, Haodong Lu, Junru Qin, Jiarui Qiao, Lingfeng Qiao, Ruizhi Ke, Bo He, Jianfeng Li, Ke Li, Yangning Shen, Yunhang Zhang, Mengdan Chen, Peixian Yin, Kun Liu, Bing Wu, Yunfei Chen, Huang Cai, Zhongpeng Li, Xiaotian Computer Vision and Pattern Recognition Despite the significant advancements represented by Vision-Language Models (VLMs), current architectures often exhibit limitations in retaining fine-grained visual information, leading to coarse-grained multimodal comprehension. We attribute this deficiency to a suboptimal training paradigm inherent in prevailing VLMs, which exhibits a text-dominant optimization bias by conceptualizing visual signals merely as passive conditional inputs rather than supervisory targets. To mitigate this, we introduce Youtu-VL, a framework leveraging the Vision-Language Unified Autoregressive Supervision (VLUAS) paradigm, which fundamentally shifts the optimization objective from ``vision-as-input'' to ``vision-as-target.'' By integrating visual tokens directly into the prediction stream, Youtu-VL applies unified autoregressive supervision to both visual details and linguistic content. Furthermore, we extend this paradigm to encompass vision-centric tasks, enabling a standard VLM to perform vision-centric tasks without task-specific additions. Extensive empirical evaluations demonstrate that Youtu-VL achieves competitive performance on both general multimodal tasks and vision-centric tasks, establishing a robust foundation for the development of comprehensive generalist visual agents.
title	Youtu-VL: Unleashing Visual Potential via Unified Vision-Language Supervision
topic	Computer Vision and Pattern Recognition
url	https://arxiv.org/abs/2601.19798