<?xml version="1.0" encoding="UTF-8"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>이호준 — 글</title><description>CS·ML·AI 연구 노트와 논문 리뷰.</description><link>https://hojunlee.me/ko/writing/</link><language>ko</language><atom:link href="https://hojunlee.me/ko/rss.xml" rel="self" type="application/rss+xml"/><item><title>LLM Context</title><link>https://hojunlee.me/ko/writing/llm-context/</link><guid isPermaLink="true">https://hojunlee.me/ko/writing/llm-context/</guid><description>컨텍스트는 role이 붙은 메시지 리스트로 관리된다. Qwen3-1.7B로 직접 확인해 본 메모.</description><pubDate>Sun, 19 Jul 2026 00:00:00 GMT</pubDate><category>llm</category><category>notes</category></item><item><title>LLMs Get Lost In Multi-Turn Conversation</title><link>https://hojunlee.me/ko/writing/llms-get-lost-in-multi-turn-conversation/</link><guid isPermaLink="true">https://hojunlee.me/ko/writing/llms-get-lost-in-multi-turn-conversation/</guid><description>단일 턴 평가와 실제 멀티턴 대화 사이의 간극을 샤딩으로 재는 논문 메모.</description><pubDate>Sun, 19 Jul 2026 00:00:00 GMT</pubDate><category>paper-review</category><category>llm</category><category>evaluation</category></item></channel></rss>