Scholarlink

This chapter focuses on SLMs in edge environments and explains how compression, distillation, quantization, and efficient decoding enable resource-aware models suitable for mobile and embedded platforms. It discusses how SLMs can inherit reasoning capability from larger teacher models while operating under strict constraints on memory, latency, and energy at the network edge. The chapter further examines enhancement strategies and parameter-efficient adaptation methods that allow SLMs to support personalized and privacy-preserving services on devices and edge nodes.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Small Language Models in the Edge Network

  • 章节

  • 发表日期：

    2026年

查看全文
  • Hongyang Du,
  • Xianhao Chen,
  • Yuanwei Liu,
  • Kaibin Huang

摘要

This chapter focuses on SLMs in edge environments and explains how compression, distillation, quantization, and efficient decoding enable resource-aware models suitable for mobile and embedded platforms. It discusses how SLMs can inherit reasoning capability from larger teacher models while operating under strict constraints on memory, latency, and energy at the network edge. The chapter further examines enhancement strategies and parameter-efficient adaptation methods that allow SLMs to support personalized and privacy-preserving services on devices and edge nodes.