<?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>稳定性 on 硅基 AGI · 智能体学习与测评</title><link>https://guijiagi.com/tags/%E7%A8%B3%E5%AE%9A%E6%80%A7/</link><description>Recent content in 稳定性 on 硅基 AGI · 智能体学习与测评</description><generator>Hugo</generator><language>zh-cn</language><copyright>本站内容采用 CC BY-NC-SA 4.0 国际许可协议授权</copyright><lastBuildDate>Tue, 30 Jun 2026 12:00:00 +0800</lastBuildDate><atom:link href="https://guijiagi.com/tags/%E7%A8%B3%E5%AE%9A%E6%80%A7/index.xml" rel="self" type="application/rss+xml"/><item><title>主流大模型API完全对比：延迟、吞吐与稳定性</title><link>https://guijiagi.com/posts/api-performance-comparison-2026/</link><pubDate>Tue, 30 Jun 2026 12:00:00 +0800</pubDate><guid>https://guijiagi.com/posts/api-performance-comparison-2026/</guid><description>2026年主流大模型API的全面性能对比：延迟、吞吐量、稳定性和实际表现</description></item><item><title>大模型训练稳定性：梯度爆炸、Loss Spike 与恢复策略</title><link>https://guijiagi.com/posts/training-stability/</link><pubDate>Thu, 25 Jun 2026 12:00:00 +0800</pubDate><guid>https://guijiagi.com/posts/training-stability/</guid><description>深入分析大语言模型训练中的稳定性问题，涵盖梯度爆炸/消失、Loss Spike、数值不稳定等原因诊断与解决方案。</description></item></channel></rss>