跳到正文
原文
arXiv · Multiagent Systems· Yuhai Long (School of Computer Science and Engineering, Sun Yat-sen University), Yuanxin Wei (School of Computer Science and Engineering, Sun Yat-sen University), Kai Wu (China Mobile Internet Company Ltd), Jinhui Wei (School of Computer Science and Engineering, Sun Yat-sen University), Dan Huang (School of Computer Science and Engineering, Sun Yat-sen University), Jiangsu Du (School of Computer Science and Engineering, Sun Yat-sen University)·· 16 小时前AI 评分35

EdgeAgent:在 CPU-GPU 统一内存架构上编排端侧多智能体系统的 LLM 推理

EdgeAgent: Orchestrating On-Device LLM inference for End-User Multi-Agent Systems on CPU-GPU Unified Memory Architectures

AI 导读

EdgeAgent 是面向端侧统一内存架构与多智能体工作负载的跨层推理系统,用零拷贝 UMA 感知张量并行、非对称内存布局、按序列可预测性动态分配 draft 预算,以及异步挂起让出机制缓解工具调用停顿。

来源:arXiv · Multiagent Systems · arxiv.org