<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Speculative on 硅基 AGI · 智能体学习与测评</title><link>https://guijiagi.com/tags/speculative/</link><description>Recent content in Speculative on 硅基 AGI · 智能体学习与测评</description><generator>Hugo</generator><language>zh-cn</language><copyright>本站内容采用 CC BY-NC-SA 4.0 国际许可协议授权</copyright><lastBuildDate>Thu, 25 Jun 2026 12:00:00 +0800</lastBuildDate><atom:link href="https://guijiagi.com/tags/speculative/index.xml" rel="self" type="application/rss+xml"/><item><title>投机解码深度解析：LLM 推理速度翻倍的秘密</title><link>https://guijiagi.com/posts/speculative-decoding-explained/</link><pubDate>Thu, 25 Jun 2026 12:00:00 +0800</pubDate><guid>https://guijiagi.com/posts/speculative-decoding-explained/</guid><description>从 Speculative Decoding 到 Medusa 和 EAGLE，全面解析投机解码如何在不损失质量的前提下将 LLM 推理速度提升 2-4 倍。</description></item></channel></rss>