<?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>Inference-Infrastructure on Cognaptus</title>
    <link>https://cognaptus.com/tags/inference-infrastructure/</link>
    <description>Recent content in Inference-Infrastructure on Cognaptus</description>
    <generator>Hugo -- 0.145.0</generator>
    <language>en-us</language>
    <lastBuildDate>Wed, 09 Sep 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://cognaptus.com/tags/inference-infrastructure/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Spend the Next Byte Where It Repairs the Model</title>
      <link>https://cognaptus.com/blog/2026-09-09-spend-the-next-byte-where-it-repairs-the-model/</link>
      <pubDate>Wed, 09 Sep 2026 00:00:00 +0000</pubDate>
      <guid>https://cognaptus.com/blog/2026-09-09-spend-the-next-byte-where-it-repairs-the-model/</guid>
      <description>AWSRC reframes low-bit LLM quality recovery as a post-quantization allocation problem: preserve the existing checkpoint and spend a measured sidecar budget on the residual corrections with the highest estimated value.</description>
    </item>
  </channel>
</rss>
