<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>谷歌研究 on AI 早报</title><link>https://ai-news.example.com/categories/%E8%B0%B7%E6%AD%8C%E7%A0%94%E7%A9%B6/</link><description>Recent content in 谷歌研究 on AI 早报</description><generator>Hugo</generator><language>zh-CN</language><lastBuildDate>Sat, 31 Oct 2026 01:32:18 +0800</lastBuildDate><atom:link href="https://ai-news.example.com/categories/%E8%B0%B7%E6%AD%8C%E7%A0%94%E7%A9%B6/feed.xml" rel="self" type="application/rss+xml"/><item><title>[EN] 谷歌发布ScreenAI：5B参数视觉语言模型，刷新UI与信息图理解多项基准</title><link>https://ai-news.example.com/articles/en-screenai5bui/</link><pubDate>Sat, 31 Oct 2026 01:32:18 +0800</pubDate><guid>https://ai-news.example.com/articles/en-screenai5bui/</guid><description>谷歌研究院推出ScreenAI，一个基于PaLI架构并采用pix2struct灵活分块策略的视觉语言模型，专门用于理解用户界面（UI）和信息图（图表、示意图、表格）。该模型仅5B参数，却在WebSRC、MoTIF、ChartQA、DocVQA和InfographicVQA等任务上达到同级最佳或业界领先水平，并同步开源三个新数据集。</description></item></channel></rss>