<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>RF-DETR: Neural Architecture Search for Real-Time Detection Transformers · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2511.09554/</link><description>Collected HN comments and submissions linking this paper.</description><language>en</language><lastBuildDate>Mon, 02 Nov 2026 11:16:22 GMT</lastBuildDate><atom:link href="https://hackerpaper.kaipfister.de/p/2511.09554/feed.xml" rel="self" type="application/rss+xml" /><item><title>RF-DETR: Neural Architecture Search for Real-Time Detection Transformers — mentioned by yeldarb</title><link>https://news.ycombinator.com/item?id=48663168</link><guid isPermaLink="false">urn:hackerpapers:mention:2511.09554:48663168</guid><pubDate>Wed, 24 Jun 2026 17:35:13 GMT</pubDate><description>&lt;p&gt;&amp;gt; See the url in my comment (search for the term rfdetr-2xlarge). 2XL does indeed go up to 800x800 and has PML1.0 license instead of apache 2.0. All of the models, including the Apache 2.0 ones, can be configured to go higher than 800x800. The difference between the ones with the PML license and the Apache 2.0 ones is the backbone, not the resolution. I&amp;#x27;d suggest you read the ICLR paper[1] which shows clearly the difference between the backbones at various latencies in Figure 1. &amp;gt; For many domain-specific (often less common and odd dimensioned) objects, downscaling will severely reduce…&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2511.09554/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.CV</category></item></channel></rss>
