<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Thoughts on Sams Personal Site</title><link>https://alpha.samuelv.dev/thoughts/</link><description>Recent content in Thoughts on Sams Personal Site</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Tue, 24 Mar 2026 13:42:39 -0400</lastBuildDate><atom:link href="https://alpha.samuelv.dev/thoughts/index.xml" rel="self" type="application/rss+xml"/><item><title>Setting MMCV with PyTorch 2.11 and Cuda 13.1</title><link>https://alpha.samuelv.dev/thoughts/mmcv-with-cuda13/</link><pubDate>Tue, 24 Mar 2026 13:42:39 -0400</pubDate><guid>https://alpha.samuelv.dev/thoughts/mmcv-with-cuda13/</guid><description>&lt;h2 id="introduction"&gt;Introduction&lt;/h2&gt;
&lt;p&gt;MMCV is a widly used computer vision framework with a large collection of tools and I need it to experiment with a realtime instance segmentation vision model called solov2 on my local machine.&lt;/p&gt;
&lt;p&gt;My computer (called the space station) is equipped with a RTX 5070ti which is only supported by cuda 13. Which leaves me in quite the predictiment since MMCV doesn&amp;rsquo;t have a precomplied binary for CUDA 13.&lt;/p&gt;</description></item></channel></rss>