<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>재현성 on CodeBerry</title>
    <link>https://codeberry.work/tags/%EC%9E%AC%ED%98%84%EC%84%B1/</link>
    <description>Recent content in 재현성 on CodeBerry</description>
    <generator>Hugo -- gohugo.io</generator>
    <language>ko</language>
    <copyright>© 2026 성경재</copyright>
    <lastBuildDate>Fri, 08 May 2026 14:15:00 +0900</lastBuildDate><atom:link href="https://codeberry.work/tags/%EC%9E%AC%ED%98%84%EC%84%B1/index.xml" rel="self" type="application/rss+xml" />
    
    <item>
      <title>Anthropic 엔지니어링 블로그 #19: Infrastructure Noise in Agentic Coding Evals — 벤치마크 점수, 그거 진짜 모델 차이인가요?</title>
      <link>https://codeberry.work/ai-news/infrastructure-noise/</link>
      <pubDate>Fri, 08 May 2026 14:15:00 +0900</pubDate>
      
      <guid>https://codeberry.work/ai-news/infrastructure-noise/</guid>
      <description>에이전트 코딩 벤치마크 점수가 RAM·CPU·네트워크 설정만으로 6%p까지 흔들린다. 리더보드 상위 모델 간 격차보다 큰 인프라 노이즈.</description>
      
    </item>
    
  </channel>
</rss>
