<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Sinkhorn-Knopp on danilchenko.dev</title><link>https://danilchenko.dev/tags/sinkhorn-knopp/</link><description>Recent content in Sinkhorn-Knopp on danilchenko.dev</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Fri, 03 Apr 2026 06:00:00 +0000</lastBuildDate><atom:link href="https://danilchenko.dev/tags/sinkhorn-knopp/index.xml" rel="self" type="application/rss+xml"/><item><title>DeepSeek's mHC: How a 1967 Algorithm Fixed the Biggest Problem in Scaling LLMs</title><link>https://danilchenko.dev/posts/2026-04-03-deepseek-mhc-manifold-constrained-hyper-connections/</link><pubDate>Fri, 03 Apr 2026 06:00:00 +0000</pubDate><guid>https://danilchenko.dev/posts/2026-04-03-deepseek-mhc-manifold-constrained-hyper-connections/</guid><description>DeepSeek&amp;#39;s mHC uses the Sinkhorn-Knopp algorithm to fix training instability in hyper-connections. Here&amp;#39;s how doubly stochastic matrices stabilize LLM scaling.</description></item></channel></rss>