<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Data Generation on WorldSense Tech Blog</title><link>https://worldsensetech.com/en/tags/data-generation/</link><description>Recent content in Data Generation on WorldSense Tech Blog</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Thu, 06 Aug 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://worldsensetech.com/en/tags/data-generation/index.xml" rel="self" type="application/rss+xml"/><item><title>World Models as Synthetic Data Engines for VLA Training</title><link>https://worldsensetech.com/en/articles/world-model-synthetic-data-for-vla/</link><pubDate>Thu, 06 Aug 2026 00:00:00 +0000</pubDate><guid>https://worldsensetech.com/en/articles/world-model-synthetic-data-for-vla/</guid><description>&lt;p&gt;In the &lt;a href="https://worldsensetech.com/en/articles/world-model-lab-setup/"&gt;previous post&lt;/a&gt;, we set up and ran DreamerV3 from scratch. A reader asked: what can a trained world model actually do?&lt;/p&gt;
&lt;p&gt;Today we discuss a more cutting-edge topic: how to use world models to generate synthetic data that enhances the training of VLA (Vision-Language-Action) models. This has become one of the widely researched directions in robot foundation models in recent years.&lt;/p&gt;
&lt;h2 id="why-vla-needs-synthetic-data"&gt;Why VLA Needs Synthetic Data&lt;/h2&gt;
&lt;p&gt;The core capability of a VLA is enabling a robot to &amp;ldquo;understand human language&amp;rdquo; — you point at a cup on the table and say &amp;ldquo;hand me the red one,&amp;rdquo; and it can understand the language instruction and execute the corresponding action.&lt;/p&gt;</description></item></channel></rss>