<?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>技术深题 on AI 实战派 · 从技术到赚钱</title><link>https://guijiagi.com/tags/%E6%8A%80%E6%9C%AF%E6%B7%B1%E9%A2%98/</link><description>Recent content in 技术深题 on AI 实战派 · 从技术到赚钱</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>Wed, 07 Oct 2026 18:54:00 +0800</lastBuildDate><atom:link href="https://guijiagi.com/tags/%E6%8A%80%E6%9C%AF%E6%B7%B1%E9%A2%98/index.xml" rel="self" type="application/rss+xml"/><item><title>GPTQ 与 AWQ：权重量化两派，到底该选哪个</title><link>https://guijiagi.com/posts/2026-10-07-gptq-awq-quantization-deep-dive/</link><pubDate>Wed, 07 Oct 2026 18:54:00 +0800</pubDate><guid>https://guijiagi.com/posts/2026-10-07-gptq-awq-quantization-deep-dive/</guid><description>同样是把权重压到 4bit，GPTQ 用二阶误差补偿，AWQ 保护显著权重。拆解两派原理、质量差异与选型建议。</description></item><item><title>FlashAttention 与 KV Cache：长上下文为什么这么吃显存</title><link>https://guijiagi.com/posts/2026-10-07-flashattention-kv-cache-deep-dive/</link><pubDate>Wed, 07 Oct 2026 18:48:00 +0800</pubDate><guid>https://guijiagi.com/posts/2026-10-07-flashattention-kv-cache-deep-dive/</guid><description>注意力是 O(N²) 的显存怪兽，KV Cache 又随并发与长度膨胀。拆解 FlashAttention 怎么省显存，以及长上下文的成本账。</description></item></channel></rss>