<?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>AI on tt67wq的博客</title><link>https://blog.tt67wq.com/tags/ai/</link><description>Recent content in AI on tt67wq的博客</description><generator>Hugo</generator><language>zh-cn</language><lastBuildDate>Thu, 22 Oct 2026 12:00:00 +0800</lastBuildDate><atom:link href="https://blog.tt67wq.com/tags/ai/index.xml" rel="self" type="application/rss+xml"/><item><title>大模型一个字是怎么生成的？</title><link>https://blog.tt67wq.com/posts/how-an-llm-generates-one-token/</link><pubDate>Thu, 22 Oct 2026 12:00:00 +0800</pubDate><guid>https://blog.tt67wq.com/posts/how-an-llm-generates-one-token/</guid><description>用一个能在笔记本上跑起来的 1420 万参数语言模型 llama2.zig，讲清 LLM 推理全流程：分词、嵌入、注意力、FFN、采样。</description></item></channel></rss>