<?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>推测解码 on 硅基 AGI · 智能体学习与测评</title><link>https://guijiagi.com/tags/%E6%8E%A8%E6%B5%8B%E8%A7%A3%E7%A0%81/</link><description>Recent content in 推测解码 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, 02 Jul 2026 10:55:00 +0800</lastBuildDate><atom:link href="https://guijiagi.com/tags/%E6%8E%A8%E6%B5%8B%E8%A7%A3%E7%A0%81/index.xml" rel="self" type="application/rss+xml"/><item><title>推测解码2026：加速2-3倍</title><link>https://guijiagi.com/posts/specul-ive-decoding-2026/</link><pubDate>Thu, 02 Jul 2026 10:55:00 +0800</pubDate><guid>https://guijiagi.com/posts/specul-ive-decoding-2026/</guid><description>从Draft-Verify到Medusa头，系统梳理推测解码的演进与2026年最新方案</description></item><item><title>推测解码加速原理：Draft Model 验证范式</title><link>https://guijiagi.com/posts/speculative-decoding/</link><pubDate>Wed, 24 Jun 2026 15:00:00 +0800</pubDate><guid>https://guijiagi.com/posts/speculative-decoding/</guid><description>深入解析推测解码如何通过 Draft Model 并行验证加速 LLM 自回归推理</description></item></channel></rss>