<?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>Grpo on TPOW Lab</title><link>https://tpow-001.netlify.app/tags/grpo/</link><description>Recent content in Grpo on TPOW Lab</description><generator>Hugo</generator><language>en</language><copyright>Copyright &amp;copy; 2025-2026 TPOW-001. All Rights Reserved.</copyright><lastBuildDate>Tue, 02 Jun 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://tpow-001.netlify.app/tags/grpo/index.xml" rel="self" type="application/rss+xml"/><item><title>Tutorial: NVIDIA-NeMo/Nemotron — 完整解讀（Nemotron 3 Super/Nano/Ultra/Omni 訓練 recipes、15 個 steps、9 個 cookbook、7 個 use-case、`nemotron-customize` Claude 插件）</title><link>https://tpow-001.netlify.app/post/2026-06-02-nemotron-tutorial/</link><pubDate>Tue, 02 Jun 2026 00:00:00 +0000</pubDate><guid>https://tpow-001.netlify.app/post/2026-06-02-nemotron-tutorial/</guid><description><![CDATA[<h1 id="nvidia-nemotron-完整教學" data-numberify>NVIDIA Nemotron 完整教學<a class="anchor ms-1" href="#nvidia-nemotron-完整教學"></a></h1>
<blockquote>
<p><strong>一句話定位</strong>：NVIDIA 官方為 <strong>Nemotron 模型家族</strong>（<strong>Nano / Super / Ultra</strong> 三層 + <strong>Nano Omni</strong> 多模態變體）建的 <strong>Developer Asset Hub</strong> — 一站式包含 (1) 可重現的訓練 recipes（pretrain → SFT → RL）、(2) 部署 cookbook、(3) RAG / Agent / SQL LoRA / Voice 等完整 use-case、(4) <strong><code>nemotron-customize</code> Claude Code 插件</strong>讓自然語言組合 pipeline，整合 NVIDIA AI 全家桶（NeMo-Curator / NeMo-RL / Megatron-Bridge / Automodel / Evaluator / DataDesigner）。</p>]]></description></item><item><title>PyRAG 完整教學 — 把多跳 RAG 當 Python 程式來執行</title><link>https://tpow-001.netlify.app/post/2026-05-18-pyrag-tutorial/</link><pubDate>Mon, 18 May 2026 00:00:00 +0000</pubDate><guid>https://tpow-001.netlify.app/post/2026-05-18-pyrag-tutorial/</guid><description><![CDATA[<blockquote>
<p>⚠️ <strong>重要警示</strong>：PyRAG 的核心機制是用 <code>exec()</code> 執行 LLM 產生的 Python code。<strong>研究 / lab 環境完全可用，但任何 non-research 部署必須先沙箱化</strong> — README 作者已主動聲明，本教學會在 §6 詳細說明風險與緩解。</p>]]></description></item></channel></rss>