光明实验室基础智能研究团队最新突破:Ferret: Federated Full-Parameter Tuning at Scale for Large Language Models。作者:Yao Shu(舒瑶), Wenyang Hu, See-Kiong Ng, Bryan Kian Hsiang Low, Fei Richard Yu。
摘要
概览
结果
总结
原文
Yao Shu, Wenyang Hu, See-Kiong Ng, Bryan Kian Hsiang Low, Fei Richard Yu. Ferret: Federated Full-Parameter Tuning at Scale for Large Language Models. arXiv:2409.06277.
参考文献
[1] Zhen Qin, Daoyuan Chen, Bingchen Qian, Bolin Ding, Yaliang Li, and Shuiguang Deng. Federated full-parameter tuning of billion-sized language models with communication cost under 18 kilobytes. In Proc. ICML, 2024.
[2] Wenzhi Fang, Ziyi Yu, Yuning Jiang, Yuanming Shi, Colin N. Jones, and Yong Zhou. Communication-efficient stochastic zeroth-order optimization for federated learning. IEEE Trans. Signal Process., 70:5058–5073, 2022.
[3] Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, and Blaise Agüera y Arcas. Communication-efficient learning of deep networks from decentralized data. In Proc. AISTATS, 2017.
[4] Yao Shu, Xiaoqiang Lin, Zhongxiang Dai, and Bryan Kian Hsiang Low. Heterogeneous federated zeroth-order optimization using gradient surrogates. In ICML 2024 Workshop on Differentiable Almost Everything: Differentiable Relaxations, Algorithms, Operators, and Simulators, 2024.
素材来源 丨光明实验室基础智能研究团队
编 辑 丨 李沛昱
审 核 丨 郭 锴
Guangming Laboratory
实验室简介
人工智能与数字经济广东省实验室(深圳)(以下简称光明实验室)为广东省政府批准建设的第三批广东省实验室之一。实验室面向世界人工智能与数字经济的前沿理论和未来技术发展趋势,致力于服务国家重大发展战略和重大需求,依托深圳地区产业优势、地缘优势和政策优势,聚集全球科研力量,充分激发科技创新资源的集聚效应。实验室围绕国产AI算力生态建设的任务核心,以多模态AI技术与应用生态建设为关键牵引,通过突破一批关键技术、催生一批原创成果,持续推进科技创新和产业赋能,加快人工智能技术在多元应用供给与全场景渗透,实现科技创新与产业驱动互促双强,不断促进以AI为引擎的新质生产力生成。
人工智能与数字经济广东省实验室(深圳)
官方网站|http://www.gml.ac.cn
招聘联系|hr_office@gml.ac.cn