<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
	<channel>
		<title>Qwen3.8 on Design Principia</title>
		<link>https://designprincipia.com/categories/qwen3.8/</link>
		<description>Recent content in Qwen3.8 on Design Principia</description>
		<generator>Hugo</generator>
		<language>en-us</language>
		
		
		
		
			<lastBuildDate>Tue, 08 Sep 2026 00:00:00 +0000</lastBuildDate>
		
			<atom:link href="https://designprincipia.com/categories/qwen3.8/index.xml" rel="self" type="application/rss+xml" />
			<item>
				<title>Measuring impact of KV quantization on Qwen3.8</title>
				<link>https://designprincipia.com/post/qwen3.8-kv-quantization-analysis/</link>
				<pubDate>Tue, 08 Sep 2026 00:00:00 +0000</pubDate>
				<guid>https://designprincipia.com/post/qwen3.8-kv-quantization-analysis/</guid>
				<description>I measure how much KV cache quantization (F16/Q8/Q4) hurts Qwen3.8-27B output quality via KL divergence, and how that compares to the impact of quantizing the model weights themselves.</description>
			</item>
	</channel>
</rss>
