<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>HRM on AI内参</title>
    <link>https://www.neican.ai/tags/hrm/</link>
    <description>Recent content in HRM on AI内参</description>
    <generator>Hugo</generator>
    <language>zh-cn</language>
    <lastBuildDate>Mon, 30 Jun 2025 20:10:04 +0800</lastBuildDate>
    <atom:link href="https://www.neican.ai/tags/hrm/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>打破参数桎梏：一种仿生学模型如何重塑AI推理的未来</title>
      <link>https://www.neican.ai/insights/article-20250630201004540-0/</link>
      <pubDate>Mon, 30 Jun 2025 20:10:04 +0800</pubDate>
      <guid>https://www.neican.ai/insights/article-20250630201004540-0/</guid>
      <description>Sapient Intelligence研发的HRM模型，通过模仿人脑的分层与多时间尺度处理机制，仅用2700万参数和1000个训练样本，便在复杂推理任务上显著超越了DeepSeek和Claude等大型语言模型。这项突破性研究不仅挑战了AI领域对模型规模的依赖，还通过近似梯度、深度监督和自适应计算时间等创新，为构建更高效、可解释且具图灵完备性的AI推理系统开辟了新路径。</description>
    </item>
  </channel>
</rss>
