Graph Representation Learning Augmented Model Manipulation on Federated Fine-Tuning of LLMs
图表示学习增强的模型操纵在联邦微调大型语言模型中的应用
机构 * Centre for neXt Communications (CXC), Department of Engineering, University of Cambridge(下一代通讯中心(CXC),工程系,剑桥大学) ; Interdisciplinary Centre for Security, Reliability and Trust (SnT), University of Luxembourg(安全、可靠与信任跨学科中心(SnT),卢森堡大学) ; Telecommunication Networks group (TKN) at the School of Electrical Engineering and Computer Science, TU Berlin(电信网络组(TKN)位于电气工程与计算机科学学院,柏林技术大学) ; Center for neXt-Generation Communications (CXC), Department of Electrical and Electronics Engineering, Koç University(下一代通讯中心(CXC),电子与电气工程系,科克大学)
专题命中 指令微调 :LLM(summary_cn,abstract);large language model(abstract);language model(abstract);分类 cs.LG
AI总结 本文提出AugMP策略,通过图表示学习框架和迭代操纵算法,增强对抗性更新的有效性和隐蔽性,实验表明其能显著降低全局LLM准确率。