视觉大模型 / VLM

M$^\text{4}$World: A Multi-view Multimodal Driving World Model for Interactive Object Manipulation and Minute-long Streaming

M$^\text{4}$World：用于交互式对象操纵和分钟级流的多视图多模态驾驶世界模型

Ke Cheng, Hanqiao Ye, Lei Shi, Yahui Liu, Yunhan Shen, Jingtao Dong, Zhenke Wang, Wenxuan Ao, Weixiang Xu, Kaining Huang, Shuhan Shen

专题命中 文档图表理解 ：VLM(abstract,abstract_cn)；分类 cs.CV

AI总结 针对现有驾驶世界生成方法局限，提出M$^\text{4}$World模型，通过灵活接口与多阶段训练实现对象操纵及长时流稳定，引入后训练与生成模型，并用新管道评估，实验证明其在驾驶模拟中有高质量、可控性与稳定性。

Comments 24 pages, 13 figures

GDP.pdf: Benchmarking Grounded Multimodal Reasoning over Professional PDF Documents

GDP.pdf：针对专业PDF文档的基础多模态推理基准测试

Suhaas Garre, Emily Ritchie, Sushant Mehta, Edwin Chen

机构 * Surge AI 纠错

专题命中 文档图表理解 ：grounding(abstract)；分类 cs.CV

AI总结 该研究针对专业PDF文档构建多模态推理基准测试GDP.pdf，由专业人员编写问题-文档对，通过严格筛选保留问题，有详细评分标准和能力分类。评估七个前沿模型，发现多数错误源于特定模式，公开了完整基准测试。

Comments 9 pages. v2: results updated to July 2026 leaderboard (17 models). Accepted at the 2nd Workshop on Knowledge-Intensive Multimodal Reasoning (KnowledgeMR) at CVPR 2026 (non-archival), under the former title "PDFParse: A Benchmark for Grounded Multimodal Reasoning over Professional PDF Documents". Dataset: https://huggingface.co/datasets/surgeai/GDP.pdf ; Code: https://github.com/surge-ai/gdp-pdf