视觉大模型 / VLM - arXivDaily 专题

2501.17391 2025-10-01 cs.CV cs.AI cs.CL 86%

LFTR: Learning-Free Token Reduction for Multimodal Large Language Models

Zihui Zhao, Yingxin Li, Yang Li

专题命中 VLM训练与架构：multimodal large language model(title,abstract)；LLaVA(abstract)；MLLM(abstract)；分类 cs.CV、cs.AI

2505.07675 2025-10-01 cs.LG cs.AI cs.CV 85%

Simple yet Effective Semi-supervised Knowledge Distillation from Vision-Language Models via Dual-Head Optimization

Seongjae Kang, Dong Bok Lee, Hyungjoon Jang, Sung Ju Hwang

机构 * VUNO Inc.（VUNO公司）； KAIST（韩国科学技术院）

专题命中 VLM训练与架构：vision-language model(title,abstract)；VLM(abstract)；分类 cs.CV、cs.AI、cs.LG

Comments 38 pages, 17 figures, preprint

2509.25568 2025-10-01 cs.CL cs.AI 79%

Probing the Limits of Stylistic Alignment in Vision-Language Models

Asma Farajidizaji, Akash Gupta, Vatsal Raina

机构 * Imperial College London（伦敦帝国学院）； Apta AI

专题命中 VLM训练与架构：vision-language model(title,abstract)；分类 cs.AI

Comments 5 pages, 1 figure, 3 tables

2509.26641 2025-10-01 cs.CV 77%

Query-Kontext: An Unified Multimodal Model for Image Generation and Editing

Yuxin Song, Wenkai Dong, Shizun Wang, Qi Zhang, Song Xue, Tao Yuan, Hu Yang, Haocheng Feng, Hang Zhou, Xinyan Xiao, Jingdong Wang

机构 * Baidu VIS（百度视觉）； National University of Singapore（新加坡国立大学）

专题命中 VLM训练与架构：vision-language model(abstract)；VLM(abstract)；grounding(abstract)；分类 cs.CV

Comments 23 pages, 10 figures

2509.25712 2025-10-01 cs.LG 77%

Expert Merging: Model Merging with Unsupervised Expert Alignment and Importance-Guided Layer Chunking

Dengming Zhang, Xiaowen Ma, Zhenliang Ni, Zhenkai Wu, Han Shu, Xin Jiang, Xinghao Chen

机构 * Zhejiang University（浙江大学）； Huawei Noah’s Ark Lab（华为诺亚实验室）

专题命中 VLM训练与架构：InternVL(abstract)；multimodal large language model(abstract)；MLLM(abstract)；分类 cs.LG

2509.26625 2025-10-01 cs.LG cs.AI cs.CV cs.MM 75%

Learning to See Before Seeing: Demystifying LLM Visual Priors from Language Pre-training

Junlin Han, Shengbang Tong, David Fan, Yufan Ren, Koustuv Sinha, Philip Torr, Filippos Kokkinos

机构 * Meta Superintelligence Labs（Meta 超智能实验室）； University of Oxford（牛津大学）

专题命中 VLM训练与架构：visual reasoning(abstract)；MLLM(abstract)；分类 cs.CV、cs.AI、cs.LG

Comments Project page: https://junlinhan.github.io/projects/lsbs/

2509.22820 2025-10-01 cs.CV cs.AI 73%

MMPB: It's Time for Multi-Modal Personalization

Jaeik Kim, Woojin Kim, Woohyeon Park, Jaeyoung Do

机构 * AIDAS Laboratory（AIDAS实验室）； IPAI ； ECE（电子工程系）； Seoul National University（首尔国立大学）

专题命中 VLM训练与架构：vision-language model(abstract)；VLM(abstract)；分类 cs.CV、cs.AI

Comments Accepted in NeurIPS 2025

2509.26231 2025-10-01 cs.CV 70%

IMG: Calibrating Diffusion Models via Implicit Multimodal Guidance

Jiayi Guo, Chuanhao Yan, Xingqian Xu, Yulin Wang, Kai Wang, Gao Huang, Humphrey Shi

机构 * Tsinghua University（清华大学）

专题命中 VLM训练与架构：multimodal large language model(abstract)；MLLM(abstract)；分类 cs.CV

Comments ICCV 2025

2509.26462 2025-10-01 cs.AI cs.CV cs.LG 67%

Zero-Shot Decentralized Federated Learning

Alessio Masano, Matteo Pennisi, Federica Proietto Salanitri, Concetto Spampinato, Giovanni Bellitto

机构 * University of Catania（卡塔尼亚大学）； PeRCeiVe Lab, University of Catania（PeRCeiVe实验室，卡塔尼亚大学）

专题命中 VLM训练与架构：vision-language model(abstract)；分类 cs.CV、cs.AI、cs.LG

Comments Accepted at International Joint Conference on Neural Networks (IJCNN) 2025. Code available at https://github.com/perceivelab/ZeroDFL

2509.26225 2025-10-01 cs.CV cs.AI 62%

An Experimental Study on Generating Plausible Textual Explanations for Video Summarization

Thomas Eleftheriadis, Evlampios Apostolidis, Vasileios Mezaris

机构 * IEEE CBMI 2025

专题命中 VLM训练与架构：LLaVA(abstract)；分类 cs.CV、cs.AI

Comments IEEE CBMI 2025. This is the authors' accepted version. The final publication is available at https://ieeexplore.ieee.org/

2509.25717 2025-10-01 cs.CV cs.CL cs.LG 62%

Importance Sampling for Multi-Negative Multimodal Direct Preference Optimization

Xintong Li, Chuhan Wang, Junda Wu, Rohan Surana, Tong Yu, Julian McAuley, Jingbo Shang

机构 * University of California, San Diego（加州大学圣迭戈分校）； Adobe Research（Adobe研究）

专题命中 VLM训练与架构：vision-language model(abstract)；分类 cs.CV、cs.LG

Comments Preprint

2509.24200 2025-10-01 cs.CV 57%

UniVid: The Open-Source Unified Video Model

Jiabin Luo, Junhui Lin, Zeyu Zhang, Biao Wu, Meng Fang, Ling Chen, Hao Tang

机构 * Peking University（北京大学）； AI Geeks ； Australian Artificial Intelligence Institute（澳大利亚人工智能研究所）

专题命中 VLM训练与架构：MLLM(abstract)；分类 cs.CV

2509.26276 2025-10-01 cs.CL cs.SD 50%

Optimizing Speech Language Models for Acoustic Consistency

Morteza Rohanian, Michael Krauthammer

专题命中 VLM训练与架构：grounding(abstract)

1. VLM训练与架构 13 篇

LFTR: Learning-Free Token Reduction for Multimodal Large Language Models

Simple yet Effective Semi-supervised Knowledge Distillation from Vision-Language Models via Dual-Head Optimization

Probing the Limits of Stylistic Alignment in Vision-Language Models

Query-Kontext: An Unified Multimodal Model for Image Generation and Editing

Expert Merging: Model Merging with Unsupervised Expert Alignment and Importance-Guided Layer Chunking

Learning to See Before Seeing: Demystifying LLM Visual Priors from Language Pre-training

MMPB: It's Time for Multi-Modal Personalization

IMG: Calibrating Diffusion Models via Implicit Multimodal Guidance

Zero-Shot Decentralized Federated Learning

An Experimental Study on Generating Plausible Textual Explanations for Video Summarization

Importance Sampling for Multi-Negative Multimodal Direct Preference Optimization

UniVid: The Open-Source Unified Video Model

Optimizing Speech Language Models for Acoustic Consistency