[{"data":1,"prerenderedAt":54},["ShallowReactive",2],{"community-tag-stats":3,"community-cats-tag":16,"community-tag-yolo模型-1-12":42},[4,7,10,12,14],{"name":5,"slug":5,"count":6},"实证分析",2,{"name":8,"slug":8,"count":9},"spss",1,{"name":11,"slug":11,"count":9},"stata",{"name":13,"slug":13,"count":9},"yolo模型",{"name":15,"slug":15,"count":9},"微调训练",[17,21,26,30,34,38],{"slug":18,"name":19,"description":20,"post_count":9},"uncategorized","未分类","投稿时未选择分类的文章",{"slug":22,"name":23,"description":24,"post_count":25},"dev-tech","开发与技术","编程语言、框架、调试与工程实践",0,{"slug":27,"name":28,"description":29,"post_count":6},"data-empirical","数据与实证","统计分析、实证方法、论文写作与 Stata 经验",{"slug":31,"name":32,"description":33,"post_count":25},"office-productivity","办公与效率","文档处理、PDF、表格与日常办公技巧",{"slug":35,"name":36,"description":37,"post_count":25},"study-notes","学习笔记","课程笔记、读书摘要与知识整理",{"slug":39,"name":40,"description":41,"post_count":25},"tools-tips","工具技巧","在线工具使用心得与场景分享",{"items":43,"total":9,"page":9,"page_size":53},[44],{"slug":45,"title":46,"summary":47,"category_slug":18,"tags":48,"author_name":49,"minutes":6,"view_count":50,"like_count":9,"comment_count":25,"share_count":9,"date":51,"published_at":52},"yolo","YOLO 模型微调训练：让通用视觉能力落地到你的业务场景","YOLO 微调的本质，是在预训练模型已学到的通用视觉特征之上，用少量自有数据\"教会\"它识别新类别，而非从零开始训练。 预训练模型已从数百万张图像中掌握边缘检测、纹理识别、形状理解等通用能力，微调只需复用这些知识，因此收敛更快、数据需求更少，这也是它在实际工程中被广泛采用的核心原因。",[13,15],"ShankZhang",14,"2026-09-02","2026-09-02T14:14:45+08:00",12,1788487608677]