<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Why Gated DeltaNet Survives 4-Bit Quantization: NVFP4 W4A4 for the Recurrent Half of a Hybrid 27B LLM · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2609.04098/</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/2609.04098/feed.xml" rel="self" type="application/rss+xml" /><item><title>Why Gated DeltaNet Survives 4-Bit Quantization: NVFP4 W4A4 for the Recurrent Half of a Hybrid 27B LLM — mentioned by skolos</title><link>https://news.ycombinator.com/item?id=49615483</link><guid isPermaLink="false">urn:hackerpapers:mention:2609.04098:49615483</guid><pubDate>Tue, 08 Sep 2026 19:22:10 GMT</pubDate><description>&lt;p&gt;There was a study specifically related to Qwen3.8 27B that showed that kv cache quantization has almost no impact on this model all the way to q4: https://arxiv.org/html/2609.04098&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2609.04098/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.AI</category></item></channel></rss>
