MoE-GPS: Guidlines for Prediction Strategy for Dynamic Expert Duplication in MoE Load Balancing
机构 * Princeton University(普林斯顿大学) ; Duke University(杜克大学)
高校专区
机构 * Princeton University(普林斯顿大学) ; Duke University(杜克大学)
机构 * Princeton University(普林斯顿大学) ; UC Berkeley(加州大学伯克利分校) ; Virginia Tech(弗吉尼亚理工学院)
Comments ICLR 2025 Outstanding Paper Runner-Up
机构 * Princeton University(普林斯顿大学)
Journal ref Proceedings of Machine Learning Research 273:238-244, 2025
机构 * Toyota Research Institute (TRI)(丰田研究院) ; Princeton University(普林斯顿大学)
Comments 14 + 5 pages, 10 figures, 4 tables. Accepted to RSS 2025
机构 * NASA Goddard Space Flight Center(美国国家航空航天局戈达德空间飞行中心) ; SETI Institute(搜寻地外文明计划研究所) ; Department of Astronomy, University of Virginia(弗吉尼亚大学天文系) ; Department of Astronomy, San Diego State University(圣地亚哥州立大学天文系) ; Department of Physics, Kavli Institute for Astrophysics and Space Research, M.I.T.(麻省理工学院物理系、Kavli天体物理与空间研究所) ; Department of Computer Science, Princeton University(普林斯顿大学计算机科学系) ; Institute of Plant Physiology and Genetics, Bulgaria(保加利亚植物生理与遗传学研究所)
Comments 40 pages, 39 figures, 4 tables
机构 * PACM, Princeton University(普林斯顿大学) ; Electrical and Computer Engineering Department, Princeton University(普林斯顿大学电子与计算机工程系) ; Courant Institute of Mathematical Sciences, New York University Center for Data Science, New York University(纽约大学应用数学科学学院,纽约大学数据科学中心)
机构 * Department of Computer Science & Princeton Language and Intelligence (PLI), Princeton University(普林斯顿大学计算机科学与普林斯顿语言与智能(PLI)部门)
Comments 30 pages
机构 * School of Physics, Georgia Institute of Technology(佐治亚理工学院物理系) ; Apple Inc.(苹果公司) ; Department of Mechanical Engineering, Stanford University(斯坦福大学机械工程系) ; Department of Physics, Princeton University(普林斯顿大学物理系) ; Department of Physics, Bar-Ilan University(巴伊兰大学物理系) ; Institute for Robotics and Intelligent Machines, Georgia Institute of Technology(佐治亚理工学院机器人与智能机器研究所) ; George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology(佐治亚理工学院乔治·W·伍德鲁夫机械工程学院)
Comments 7 pages, 6 figures, submitted to PRL
机构 * princeton university(普林斯顿大学)
Comments 77 pages, 18 figures; ICML 2025 Main Conference
机构 * University of Maryland(马里兰大学) ; Princeton University(普林斯顿大学) ; University of Sydney(悉尼大学) ; Information Sciences Institute, University of Southern California(南加州大学信息科学研究所)
机构 * Stevens Institute of Technology(史蒂文斯理工学院) ; Princeton University(普林斯顿大学) ; Duke University(杜克大学) ; Google LLC(谷歌公司) ; Lambda Inc.(Lambda公司)
Comments 56 pages, 11 figures
机构 * Princeton University(普林斯顿大学) ; Stanford University(斯坦福大学)
机构 * Princeton University(普林斯顿大学) ; McGill University(麦吉尔大学)
机构 * The Chinese University of Hong Kong(香港中文大学) ; University of Illinois Urbana-Champaign(伊利诺伊大学厄巴娜-香槟分校) ; Princeton University(普林斯顿大学) ; Sun Yat-sen University(中山大学) ; Hong Kong University of Science and Technology(香港科学理工大学)
机构 * The University of Hong Kong(香港大学) ; Tsinghua University(清华大学) ; Duke University(杜克大学) ; University of Rochester(罗切斯特大学) ; The Ohio State University(俄亥俄州立大学) ; Bytedance(字节跳动) ; The University of North Carolina at Chapel Hill(北卡罗来纳大学教堂山分校) ; Apple(苹果公司) ; The Hong Kong Polytechnic University(香港理工大学) ; Princeton University(普林斯顿大学)
Comments The paper is accepted by TMLR
机构 * ETH Zurich(苏黎世联邦理工学院) ; Princeton University(普林斯顿大学)
Comments Published in Transactions on Machine Learning Research (TMLR); Best technical paper at Neurips 2024 SoLaR workshop
机构 * The Hong Kong University of Science and Technology (Guangzhou)(香港科技大学(广州)) ; Princeton University(普林斯顿大学) ; Nanyang Technological University(南洋理工大学) ; Salesforce Research(Salesforce 研究) ; Institute of High Performance Computing (IHPC), Agency for Science, Technology and Research (A*STAR), Singapore(高性能计算研究所,新加坡科技研究局)
机构 * The Hong Kong University of Science and Technology (Guangzhou)(香港科技大学(广州)) ; Princeton University(普林斯顿大学) ; Nanyang Technological University(南洋理工大学) ; Salesforce Research(Salesforce 研究) ; Institute of High Performance Computing (IHPC), Agency for Science, Technology and Research (A*STAR), Singapore(高性能计算研究所,新加坡科技研究局)
机构 * Department of Electrical and Computer Engineering, University of Waterloo(滑铁卢大学电气与计算机工程系) ; PRISMA Lab, Department of Electrical Engineering and Information Technology, University of Naples Federico II(那不勒斯费德里克二世大学电气工程与信息技术系PRISMA实验室) ; H Company(H公司) ; Department of Mechanical and Aerospace Engineering, Princeton University(普林斯顿大学机械与航空航天工程系) ; Department of Science and Methods of Engineering, University of Modena and Reggio Emilia(摩德纳和雷焦艾米利亚大学科学与工程方法系)
机构 * UC Berkeley(加州大学伯克利分校) ; Princeton University(普林斯顿大学)
Comments CVPR 2025
机构 * Princeton University(普林斯顿大学)
Comments Accepted as a CHI Late Breaking Work (2025), cite appropriately
机构 * Department of Operations Research and Financial Engineering, Princeton University(运筹学与金融工程系,普林斯顿大学) ; Department of Electrical and Computer Engineering, Princeton University(电气与计算机工程系,普林斯顿大学)
机构 * Department of Computer Science, Princeton University(普林斯顿大学计算机科学系) ; Princeton Language and Intelligence, Princeton University(普林斯顿语言与智能研究所)
Comments 37 pages, 15 figures, 45 tables
机构 * Shanghai Jiao Tong University(上海交通大学) ; Shanghai AI Laboratory(上海人工智能实验室) ; Wuhan University(武汉大学) ; Tongji University(同济大学) ; Princeton University(普林斯顿大学)
机构 * Princeton University(普林斯顿大学) ; Zoom Video Communications(Zoom视频通讯)
Comments Preprint
Journal ref ICLR 2025
机构 * Computer Science Department, Princeton University(普林斯顿大学计算机科学系)
机构 * Harvard University(哈佛大学) ; Boston University(波士顿大学) ; Cornell University(康奈尔大学) ; Princeton University(普林斯顿大学)
机构 * Virginia Tech(弗吉尼亚理工大学) ; Princeton University(普林斯顿大学) ; University of California, Los Angeles(加州大学洛杉矶分校) ; University of California, Davis(加州大学戴维斯分校) ; University of Pennsylvania(宾夕法尼亚大学)
Comments 15 pages
机构 * Stanford University(斯坦福大学) ; Princeton University(普林斯顿大学) ; Adobe Research(Adobe研究)
Comments Project website: https://ryanpo.com/ssm_wm
机构 * Mercedes-Benz AG(梅赛德斯-奔驰集团) ; Technical University of Munich(慕尼黑技术大学) ; Purdue University(普渡大学) ; Space Robotics Research Group, SnT, University of Luxembourg(卢森堡大学空间机器人研究组) ; Torc Robotics(Torc机器人公司) ; Princeton University(普林斯顿大学)