<?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>Discrete-Diffusion on Xu'Blog</title><link>https://xuquant.com/tags/discrete-diffusion/</link><description>Recent content in Discrete-Diffusion on Xu'Blog</description><image><title>Xu'Blog</title><url>https://xuquant.com/images/profile.jpg</url><link>https://xuquant.com/images/profile.jpg</link></image><generator>Hugo -- 0.152.2</generator><language>en</language><lastBuildDate>Fri, 08 May 2026 18:00:00 +0800</lastBuildDate><atom:link href="https://xuquant.com/tags/discrete-diffusion/index.xml" rel="self" type="application/rss+xml"/><item><title>ReflectDrive-2：理想汽车的离散扩散端到端驾驶与 RL 联合优化</title><link>https://xuquant.com/posts/reflectdrive-2-discrete-diffusion-end-to-end-driving/</link><pubDate>Fri, 08 May 2026 18:00:00 +0800</pubDate><guid>https://xuquant.com/posts/reflectdrive-2-discrete-diffusion-end-to-end-driving/</guid><description>深度解读理想汽车 ReflectDrive-2：首创离散扩散模型用于端到端自动驾驶规划，提出「决策-起草-反思」三阶段推理范式，通过强化学习联合优化起草+编辑实现 AutoEdit 增益放大 6 倍，纯相机输入达 NAVSIM SOTA 91.0 PDMS，Thor 芯片上 31.8ms/帧实时部署。</description></item></channel></rss>