<?xml version="1.0" encoding="UTF-8"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Yestino - The Signal · CUDA Add</title><link>https://yestino.com/zh-CN/entities/cuda-add-bcf945</link><description>与 CUDA Add 相关的全部事件</description><language>zh-CN</language><atom:link href="https://yestino.com/zh-CN/entities/cuda-add-bcf945/feed.xml" rel="self" type="application/rss+xml"/><item><title>trunk/7e1c35ebdb7015a89ecdb99030893a767550eef1: [CUDA] Add non-overlapping fast path for avg_pool2d backward (#191086)</title><link>https://yestino.com/zh-CN/events/trunk-7e1c35ebdb7015a89ecdb99030893a767550eef1-cuda-add-non-d43e4a</link><guid isPermaLink="true">https://yestino.com/zh-CN/events/trunk-7e1c35ebdb7015a89ecdb99030893a767550eef1-cuda-add-non-d43e4a</guid><pubDate>Sun, 30 Aug 2026 01:20:31 GMT</pubDate><description>Summary Adds a dedicated CUDA backward kernel for avg_pool2d when the pooling windows tile the input exactly ( stride == kernel_size , no padding, no ceil_mode…
来源：PyTorch Releases</description></item></channel></rss>