<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>The Power of Noise: Redefining Retrieval for RAG Systems · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2401.14887/</link><description>Collected HN comments and submissions linking this paper.</description><language>en</language><lastBuildDate>Mon, 02 Nov 2026 11:27:26 GMT</lastBuildDate><atom:link href="https://hackerpaper.kaipfister.de/p/2401.14887/feed.xml" rel="self" type="application/rss+xml" /><item><title>The Power of Noise: Redefining Retrieval for RAG Systems — HN submission</title><link>https://news.ycombinator.com/item?id=39335365</link><guid isPermaLink="false">urn:hackerpapers:submission:2401.14887:39335365</guid><pubDate>Sun, 11 Feb 2024 14:55:14 GMT</pubDate><description>&lt;p&gt;Ask HN: Including irrelevant documents improve RAG accuracy by 30%?&lt;/p&gt;&lt;p&gt;I was very intrigued after reading &amp;quot;The Power of Noise: Redefining Retrieval for RAG Systems&amp;quot; paper (https://arxiv.org/pdf/2401.14887.pdf), where they explored the impact of relevance, position and the number of retrieved documents on the accuracy of a RAG pipeline. The authors defined 3 sets of documents: 1. Golden document - One which contains the answer to the given question. 2. Relevant documents - A set of documents that talk about the same topic but don&amp;#x27;t contain the answer to the question. 3. Irrelevant documents - A set of documents that talk about different unrelated topics and…&lt;/p&gt;&lt;p&gt;3 points; 0 discussion replies on HN. &lt;a href="https://hackerpaper.kaipfister.de/p/2401.14887/"&gt;Paper and source history&lt;/a&gt;. A submission is not an endorsement.&lt;/p&gt;</description><category>cs.IR</category></item></channel></rss>
