<?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>Auto-Research on Dibi8 - AI工具目录</title><link>https://dibi8.com/tags/auto-research/</link><description>Recent content in Auto-Research on Dibi8 - AI工具目录</description><generator>Hugo</generator><language>zh</language><lastBuildDate>Thu, 24 Sep 2026 00:00:00 +0800</lastBuildDate><atom:link href="https://dibi8.com/tags/auto-research/index.xml" rel="self" type="application/rss+xml"/><item><title>SoL-Pi: Scaling Auto-Research Loops for Efficient Agent Harnesses — NVlabs Open-Source Optimization Framework 2026</title><link>https://dibi8.com/llm-frameworks/nvlabs-sol-pi-agent-harness-efficiency-2026/</link><pubDate>Thu, 24 Sep 2026 00:00:00 +0800</pubDate><guid>https://dibi8.com/llm-frameworks/nvlabs-sol-pi-agent-harness-efficiency-2026/</guid><description>SoL-Pi by NVIDIA saves 13.50/hour vs Claude Code and 5.71/hour vs native Pi. Four mechanisms: Action Fusion, ObservationPack, Evidence-Preserving Reducer, Online Context Compact. Free, opt-in, no patches required.</description></item></channel></rss>