<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
	<channel>
		<title>Machine-Learning on NicoLabs</title>
		<link>http://blog.hellonico.info/tags/machine-learning/</link>
		<description>Recent content in Machine-Learning on NicoLabs</description>
		<generator>Hugo</generator>
		<language>en</language>
		
		
		
		
			<lastBuildDate>Tue, 28 Jul 2026 10:00:00 +0900</lastBuildDate>
		
			<atom:link href="http://blog.hellonico.info/tags/machine-learning/index.xml" rel="self" type="application/rss+xml" />
			<item>
				<title>Running GGUF Models Natively with Coni and Apple MLX</title>
				<link>http://blog.hellonico.info/posts/coni/running-gguf-natively-in-coni/</link>
				<pubDate>Tue, 28 Jul 2026 10:00:00 +0900</pubDate>
				<guid>http://blog.hellonico.info/posts/coni/running-gguf-natively-in-coni/</guid>
				<description>&lt;p&gt;Picture this: You see a shiny new Open Source LLM drop on HuggingFace. It promises to be the next big thing in local reasoning. You want to try it out on your MacBook. So you spin up your terminal.&lt;/p&gt;&#xA;&lt;p&gt;You &lt;code&gt;pip install&lt;/code&gt; some packages. You get an error about &lt;code&gt;torch&lt;/code&gt; being compiled without Metal support. You switch to &lt;code&gt;conda&lt;/code&gt;. You update your environment. You wait 15 minutes for 4 gigabytes of dependencies to download. You run the Python script. You get a &lt;code&gt;ValueError: weight shape mismatch&lt;/code&gt; because your &lt;code&gt;transformers&lt;/code&gt; library is exactly 0.0.1 versions out of date. You update &lt;code&gt;transformers&lt;/code&gt;, which breaks &lt;code&gt;accelerate&lt;/code&gt;.&lt;/p&gt;</description>
			</item>
	</channel>
</rss>
