<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>从模型结构到真实系统的项目沉淀。 on 南巷</title><link>https://nx.wmlab.top/portfolio/</link><description>Recent content in 从模型结构到真实系统的项目沉淀。 on 南巷</description><generator>Hugo -- gohugo.io</generator><language>zh-CN</language><managingEditor>liuwenhao1968@163.com (南巷)</managingEditor><webMaster>liuwenhao1968@163.com (南巷)</webMaster><copyright>© 2026 南巷</copyright><lastBuildDate>Mon, 01 Jun 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://nx.wmlab.top/portfolio/index.xml" rel="self" type="application/rss+xml"/><item><title>LeRobot pi0.5 + SO-101：双任务机械臂操作</title><link>https://nx.wmlab.top/portfolio/lerobot-so101/</link><pubDate>Mon, 01 Jun 2026 00:00:00 +0000</pubDate><author>liuwenhao1968@163.com (南巷)</author><guid>https://nx.wmlab.top/portfolio/lerobot-so101/</guid><description>基于 LeRobot 与 pi0.5 策略框架，在 SO-101 机械臂上完成叠毛巾与清理桌面两个真实桌面操作任务，验证从数据采集、策略学习到实体执行的闭环。</description></item><item><title>CodeLab-LLaMA2 / 星语 MoE：大模型训练实践</title><link>https://nx.wmlab.top/portfolio/xingyu-moe/</link><pubDate>Wed, 15 Apr 2026 00:00:00 +0000</pubDate><author>liuwenhao1968@163.com (南巷)</author><guid>https://nx.wmlab.top/portfolio/xingyu-moe/</guid><description>围绕 LLaMA2 内部原理与工程实现，系统拆解架构、组件、预训练、SFT、LoRA、推理与 RAG/Agent，形成从理论到应用的大模型学习与实现闭环。</description></item><item><title>星语 Vision：自研多模态大模型</title><link>https://nx.wmlab.top/portfolio/xingyu-vision/</link><pubDate>Fri, 01 May 2026 00:00:00 +0000</pubDate><author>liuwenhao1968@163.com (南巷)</author><guid>https://nx.wmlab.top/portfolio/xingyu-vision/</guid><description>基于 CLIP 视觉编码器和文本 Transformer 的图像-文本联合推理系统，完成多模态最小可验证闭环。</description></item><item><title>星语 pi0：VLA 模型源码拆解与最小复现</title><link>https://nx.wmlab.top/portfolio/xy-pi0/</link><pubDate>Wed, 20 May 2026 00:00:00 +0000</pubDate><author>liuwenhao1968@163.com (南巷)</author><guid>https://nx.wmlab.top/portfolio/xy-pi0/</guid><description>围绕 pi0 源码构建最小可运行 VLA 系统，拆解视觉编码、语言建模和动作预测模块。</description></item><item><title>NxV2Net：裂缝分割科研代码</title><link>https://nx.wmlab.top/portfolio/nxv2net/</link><pubDate>Sat, 10 Jan 2026 00:00:00 +0000</pubDate><author>liuwenhao1968@163.com (南巷)</author><guid>https://nx.wmlab.top/portfolio/nxv2net/</guid><description>面向复杂真实场景的鲁棒裂缝分割开源项目，围绕嵌套多尺度结构、MCFA 融合注意力和 SUES-CRACK 数据集提升泛化能力。</description></item><item><title>HAFNet：轻量化语义分割网络</title><link>https://nx.wmlab.top/portfolio/hafnet/</link><pubDate>Tue, 10 Feb 2026 00:00:00 +0000</pubDate><author>liuwenhao1968@163.com (南巷)</author><guid>https://nx.wmlab.top/portfolio/hafnet/</guid><description>轻量级多层级可定制语义分割网络，围绕图像语义理解中的边界、结构与高层语义协同建模。</description></item><item><title>Classification：通用图像分类训练模板</title><link>https://nx.wmlab.top/portfolio/kaggle-leaf-classification/</link><pubDate>Sat, 15 Feb 2025 00:00:00 +0000</pubDate><author>liuwenhao1968@163.com (南巷)</author><guid>https://nx.wmlab.top/portfolio/kaggle-leaf-classification/</guid><description>面向研究和工程复用的 PyTorch 图像分类训练框架，支持替换主流网络、自定义数据集、SwanLab 日志监控和推理可视化，并以树叶分类作为完整示例。</description></item><item><title>2018 Data Science Bowl：细胞核分割</title><link>https://nx.wmlab.top/portfolio/data-science-bowl-nuclei-segmentation/</link><pubDate>Wed, 15 Jan 2025 00:00:00 +0000</pubDate><author>liuwenhao1968@163.com (南巷)</author><guid>https://nx.wmlab.top/portfolio/data-science-bowl-nuclei-segmentation/</guid><description>改进 U-Net 进行细胞核实例分割，结合 Deformable Convolution、密集跳跃连接和强数据增强，最终排名前 5%。</description></item><item><title>WebSocket_Test：群聊与私聊实时通信</title><link>https://nx.wmlab.top/portfolio/websocket-test/</link><pubDate>Mon, 01 Apr 2024 00:00:00 +0000</pubDate><author>liuwenhao1968@163.com (南巷)</author><guid>https://nx.wmlab.top/portfolio/websocket-test/</guid><description>基于 WebSocket 通信实现群聊与私聊功能，聚焦实时消息传递、会话区分和基础即时通信交互。</description></item><item><title>school_blog：前后端分离校园博客</title><link>https://nx.wmlab.top/portfolio/school-blog/</link><pubDate>Wed, 01 May 2024 00:00:00 +0000</pubDate><author>liuwenhao1968@163.com (南巷)</author><guid>https://nx.wmlab.top/portfolio/school-blog/</guid><description>基于 Spring Boot、Vue、MySQL 与 Redis 的校园博客系统，包含 PC 网页端和移动网页端，并接入云服务器与对象存储。</description></item></channel></rss>