<?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>훈련비용 on TechLLM</title><link>https://techllm.github.io/tags/%ED%9B%88%EB%A0%A8%EB%B9%84%EC%9A%A9/</link><description>Recent content in 훈련비용 on TechLLM</description><generator>Hugo</generator><language>ko</language><lastBuildDate>Sat, 16 May 2026 12:00:00 +0900</lastBuildDate><atom:link href="https://techllm.github.io/tags/%ED%9B%88%EB%A0%A8%EB%B9%84%EC%9A%A9/index.xml" rel="self" type="application/rss+xml"/><item><title>학습 없이 GPQA Diamond 88.89% — Darwin Family, 가중치 재조합으로 프론티어 추론 달성</title><link>https://techllm.github.io/posts/darwin-training-free-reasoning/</link><pubDate>Sat, 16 May 2026 12:00:00 +0900</pubDate><guid>https://techllm.github.io/posts/darwin-training-free-reasoning/</guid><description>그래디언트 한 번 없이 오픈소스 LLM 가중치를 재조합해 GPQA Diamond 88.89%를 찍은 Darwin Family. 14차원 머지 게놈, MRI-Trust Fusion, 이종 아키텍처 교배로 훈련 비용 없이 SOTA 추론을 꺼냈다.</description></item></channel></rss>