arXivDaily arXiv每日学术速递 周一至周五更新

高校专区

University of Chinese Academy of Sciences(中国科学院大学)

2026-01-30 至 2026-01-30 共收录 10
2601.21947 2026-01-30 cs.AI

ToolWeaver: Weaving Collaborative Semantics for Scalable Tool Use in Large Language Models

ToolWeaver: 为大语言模型中的可扩展工具使用编织协作语义

Bowen Fang, Wen Ye, Yunyue Su, Jinghao Zhang, Qiang Liu, Yesheng Liu, Xin Sun, Shu Wu, Jiabing Yang, Baole Wei, Liang Wang

机构 * New Laboratory of Pattern Recognition (NLPR), Institute of Automation, Chinese Academy of Sciences (CASIA)(模式识别新实验室(NLPR),自动化研究所,中国科学院(CASIA)) School of Artificial Intelligence, University of Chinese Academy of Sciences(人工智能学院,中国科学院大学) Zhongguancun Academy(中关村学院) Zhongguancun Institute of Artificial Intelligence(中关村人工智能研究所)

AI总结 ToolWeaver通过编码工具为层次序列,解决大语言模型中工具使用中的语义和可扩展性问题,提升工具协作学习的效率与效果。

Comments 10pages, 12 figures, Accepted to ICLR 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2601.21751 2026-01-30 cs.CV

Dynamic Topology Awareness: Breaking the Granularity Rigidity in Vision-Language Navigation

动态拓扑感知:突破视觉语言导航中的粒度刚性

Jiankun Peng, Jianyuan Guo, Ying Xu, Yue Liu, Jiashuang Yan, Xuanwei Ye, Houhua Li, Xiaoming Wang

机构 * Aerospace Information Research Institute, Chinese Academy of Sciences(中国科学院航空航天信息研究所) School of Electronic, Electrical and Communication Engineering, University of Chinese Academy of Sciences(中国科学院大学电子电气与通信工程学院) Department of Computer Science, City University of Hong Kong(香港城市大学计算机科学系)

AI总结 DGNav通过动态拓扑导航框架,解决视觉语言导航中的粒度刚性问题,提升导航效率与安全性。

详情

展开后加载摘要…

URL PDF HTML 收藏
2601.21716 2026-01-30 cs.CV cs.AI

DreamActor-M2: Universal Character Image Animation via Spatiotemporal In-Context Learning

DreamActor-M2: 通过时空上下文学习实现通用角色图像动画

Mingshuang Luo, Shuang Liang, Zhengkun Rong, Yuxuan Luo, Tianshu Hu, Ruibing Hou, Hong Chang, Yong Li, Yuan Zhang, Mingyuan Gao

机构 * Key Lab of Intell. Info. Process., ICT, CAS(智能信息处理重点实验室) University of Chinese Academy of Sciences(中国科学院大学) Southeast University(东南大学)

AI总结 DreamActor-M2通过时空上下文学习实现通用角色图像动画,解决运动注入策略和姿态先验依赖问题,提升跨领域泛化能力。

详情

展开后加载摘要…

URL PDF HTML 收藏
2601.21341 2026-01-30 cs.CV

Dynamical Adapter Fusion: Constructing A Global Adapter for Pre-Trained Model-based Class-Incremental Learning

动态适配器融合:为基于预训练模型的类增量学习构建一个全局适配器

Ruiqi Liu, Boyu Diao, Zijia An, Zhulin An, Fei Wang, Yongjun Xu

机构 * Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China(中国科学院计算技术研究所) University of Chinese Academy of Sciences, Beijing, China(中国科学院大学)

AI总结 本文提出动态适配器融合方法,通过整合任务特定参数、全局适配器参数和初始化参数,解决类增量学习中的灾难性遗忘问题,实现状态最先进的性能。

详情

展开后加载摘要…

URL PDF HTML 收藏
2601.21238 2026-01-30 cs.CV cs.AI

PTQ4ARVG: Post-Training Quantization for AutoRegressive Visual Generation Models

PTQ4ARVG:后训练量化用于自回归视觉生成模型

Xuewen Liu, Zhikai Li, Jing Zhang, Mengjuan Chen, Qingyi Gu

机构 * Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所) School of Artificial Intelligence, University of Chinese Academy of Sciences(中国科学院大学人工智能学院)

AI总结 PTQ4ARVG提出一种无需训练的后训练量化框架,解决ARVG模型在通道、令牌和样本层面的量化难题,实现8位和6位高效量化。

Comments ICLR 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2601.17865 2026-01-30 cs.CL

D-Models and E-Models: Diversity-Stability Trade-offs in the Sampling Behavior of Large Language Models

D模型和E模型:大语言模型采样行为中的多样性-稳定性权衡

Jia Gu, Liang Pang, Huawei Shen, Xueqi Cheng

机构 * State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences(人工智能安全国家重点实验室,计算技术研究所,中国科学院) University of Chinese Academy of Sciences(中国科学院大学)

AI总结 研究揭示了大语言模型中D模型与E模型在采样行为上的多样性-稳定性权衡,为任务应用中的模型选择提供了指导。

Comments 12 pages, 10 figures. Accepted by WWW'26

详情

展开后加载摘要…

URL PDF HTML 收藏
2509.03803 2026-01-30 cs.CV

Causality-guided Prompt Learning for Vision-language Models via Visual Granulation

基于视觉粒化的因果引导提示学习用于视觉语言模型

Mengyu Gao, Qiulei Dong

机构 * State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences(多模态人工智能系统国家重点实验室,自动化研究所,中国科学院) School of Artificial Intelligence, University of Chinese Academy of Sciences(人工智能学院,中国科学院大学)

AI总结 本文提出CaPL方法,通过视觉粒化和因果推理提升细粒度识别性能,实验表明其在细粒度数据集上表现优异。

Comments Updated version

详情

展开后加载摘要…

URL PDF HTML 收藏
2506.20966 2026-01-30 cs.RO cs.AI

Parallels Between VLA Model Post-Training and Human Motor Learning: Progress, Challenges, and Trends

VLA模型后训练与人类运动学习的类比:进展、挑战与趋势

Tian-Yu Xiang, Ao-Qun Jin, Xiao-Hu Zhou, Mei-Jiang Gui, Xiao-Liang Xie, Shi-Qi Liu, Shuang-Yi Wang, Sheng-Bin Duan, Fu-Chao Xie, Wen-Kai Wang, Si-Cheng Wang, Ling-Yun Li, Tian Tu, Zeng-Guang Hou

机构 * State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences(多模态人工智能系统国家重点实验室,自动化研究所,中国科学院) School of Artificial Intelligence, University of Chinese Academy of Sciences(中国科学院大学人工智能学院) The Grainger College of Engineering, University of Illinois Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校格拉inger工程学院) CAS Center for Excellence in Brain Science and Intelligence Technology(中国科学院脑科学与智能技术卓越创新中心) Joint Laboratory of Intelligence Science and Technology, Institute of Systems Engineering, Macau University of Science and Technology(澳门科技大学系统工程学院智能科学与技术联合实验室)

AI总结 本文从人类运动学习角度综述了VLA模型后训练的进展、挑战与趋势,提出四类后训练方法并探讨了其在机器人操作中的应用与未来方向。

详情

展开后加载摘要…

URL PDF HTML 收藏
2410.00564 2026-01-30 cs.LG cs.AI

Scaling Offline Model-Based RL via Jointly-Optimized World-Action Model Pretraining

通过联合优化的世界-动作模型扩展离线模型基于的强化学习

Jie Cheng, Ruixi Qiao, Yingwei Ma, Binhua Li, Gang Xiong, Qinghai Miao, Yongbin Li, Yisheng Lv

机构 * State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences(多模态人工智能系统国家重点实验室,自动化研究所,中国科学院) School of Artificial Intelligence, University of Chinese Academy of Sciences(中国科学院大学人工智能学院) Alibaba Group(阿里巴巴集团)

AI总结 JOWA通过联合优化的世界-动作模型扩展离线RL,实现高效泛化和高性能

Comments Accepted by ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2309.16117 2026-01-30 cs.LG cs.AI

Low-redundancy Distillation for Continual Learning

低冗余蒸馏用于持续学习

RuiQi Liu, Boyu Diao, Libo Huang, Zijia An, Hangda Liu, Zhulin An, Yongjun Xu

机构 * Institute of Computing Technology(计算技术研究所) University of Chinese Academy of Sciences(中国科学院大学)

AI总结 LoRD通过消除持续学习中的冗余来提升模型性能和训练效率,实现高准确率与低计算开销的平衡。

Comments Accepted by Pattern Recognition

详情

展开后加载摘要…

URL PDF HTML 收藏