<![CDATA[gitblog_09746的博客]]>

<![CDATA[gitblog_09746的博客]]>https://blog.csdn.net/gitblog_09746zh-cnhttps://blog.csdn.net/<![CDATA[DLSS Swapper DLL 替换内核：哈希校验、签名验证与回滚机制解析]]>https://blog.csdn.net/gitblog_09746/article/details/142948036https://blog.csdn.net/gitblog_09746/article/details/142948036gitblog_09746Sun, 20 Sep 2026 15:36:34 +0800<![CDATA[Agentic Awesome Skills 之 NestJS 后端模式：用 Screaming Architecture 生成工程级 Backend Rules]]>https://blog.csdn.net/gitblog_09746/article/details/143297786https://blog.csdn.net/gitblog_09746/article/details/143297786gitblog_09746Sat, 19 Sep 2026 20:31:04 +0800 **导读**：`backend.md` 是 `agents-generator` 技能（位于 `plugins/agentic-awesome-skills-claude/skills/agents-generator/`）在检测到目标项目使用 NestJS 时使用的后端规则模板。它定义了面向 NestJS 项目的模块化目录规范（Screaming Architecture）、DTO 校验与依]]><![CDATA[Wox 单进程 Go UI 内存泄漏排查实战：从自动化语义树复现、同进程采样到堆剖析归因的完整方法]]>https://blog.csdn.net/gitblog_09746/article/details/141946609https://blog.csdn.net/gitblog_09746/article/details/141946609gitblog_09746Sat, 19 Sep 2026 12:11:37 +0800<![CDATA[Jekyll 3.4.1 发布深度解读：一次由“意外功能“引发的文件名日期正则修复]]>https://blog.csdn.net/gitblog_09746/article/details/143144557https://blog.csdn.net/gitblog_09746/article/details/143144557gitblog_09746Fri, 18 Sep 2026 10:01:53 +0800<![CDATA[anarlog 说话人分离：如何把多人会议转写快速拆成“谁说了什么”]]>https://blog.csdn.net/gitblog_09746/article/details/142950799https://blog.csdn.net/gitblog_09746/article/details/142950799gitblog_09746Thu, 17 Sep 2026 20:37:28 +0800<![CDATA[eslint-plugin-unicorn 的 no-useless-continue 规则：消除循环末尾冗余的 `continue` 语句]]>https://blog.csdn.net/gitblog_09746/article/details/143289049https://blog.csdn.net/gitblog_09746/article/details/143289049gitblog_09746Thu, 17 Sep 2026 11:54:43 +0800<![CDATA[swarms 单智能体领域化实战：保险理赔处理与 VC 法律文档生成双 Demo 深度解析]]>https://blog.csdn.net/gitblog_09746/article/details/143412732https://blog.csdn.net/gitblog_09746/article/details/143412732gitblog_09746Wed, 16 Sep 2026 14:53:23 +0800<![CDATA[A2UI Express 推理格式迭代优化实战：run_017 选项对象自动归一化的变更与评测复盘]]>https://blog.csdn.net/gitblog_09746/article/details/143154198https://blog.csdn.net/gitblog_09746/article/details/143154198gitblog_09746Sun, 13 Sep 2026 17:20:02 +0800<![CDATA[darktable 新增 DC-TZ99 支持：.RW2 文件从导入到验证的快速指南]]>https://blog.csdn.net/gitblog_09746/article/details/143009530https://blog.csdn.net/gitblog_09746/article/details/143009530gitblog_09746Sat, 12 Sep 2026 18:48:06 +0800<![CDATA[如何在 claude-code-templates 仪表盘上通过构建开关启用 WebMCP 只读目录工具？]]>https://blog.csdn.net/gitblog_09746/article/details/141975813https://blog.csdn.net/gitblog_09746/article/details/141975813gitblog_09746Fri, 11 Sep 2026 21:44:19 +0800<![CDATA[Fiddler中文版安全测试指南：如何发现和修复Web应用漏洞]]>https://blog.csdn.net/gitblog_09746/article/details/143120438https://blog.csdn.net/gitblog_09746/article/details/143120438gitblog_09746Wed, 15 Jul 2026 08:58:02 +0800<![CDATA[Qwen-Image-Layered核心功能解析：3层图像分解让创作更自由]]>https://blog.csdn.net/gitblog_09746/article/details/141946746https://blog.csdn.net/gitblog_09746/article/details/141946746gitblog_09746Fri, 29 May 2026 07:58:05 +0800<![CDATA[Yolov8n-pose进阶教程：如何快速实现自定义数据集训练与模型微调的完整指南]]>https://blog.csdn.net/gitblog_09746/article/details/141945087https://blog.csdn.net/gitblog_09746/article/details/141945087gitblog_09746Fri, 29 May 2026 07:51:04 +0800<![CDATA[KeyPhraseTransformer：基于T5架构的终极关键词提取工具，50万样本训练的文本主题提取神器]]>https://blog.csdn.net/gitblog_09746/article/details/141971954https://blog.csdn.net/gitblog_09746/article/details/141971954gitblog_09746Fri, 29 May 2026 07:46:04 +0800<![CDATA[从7B到12B性能飞跃：TeleChat系列模型架构演进与技术突破分析]]>https://blog.csdn.net/gitblog_09746/article/details/141978695https://blog.csdn.net/gitblog_09746/article/details/141978695gitblog_09746Fri, 29 May 2026 07:37:22 +0800<![CDATA[CANN/catlass CopyL1ToL0ATla API]]>https://blog.csdn.net/gitblog_09746/article/details/141947141https://blog.csdn.net/gitblog_09746/article/details/141947141gitblog_09746Fri, 29 May 2026 07:32:38 +0800 [代码位置](https://link.gitcode.com/i/edb9516802145d3ccab4ae471d360439) [TOC] ## 功能说明 `CopyL1ToL0ATla` 实现 Conv 场景下将 Fmap 数据从 L1 搬运到 L0A 的 TLA 风格版本，同时完成 im2col 操作。 - AtlasA2：`LoadData` + `Conv2dFilte]]><![CDATA[GPT2_PMC模型推理指南：深入理解inference.py与医学文本生成机制]]>https://blog.csdn.net/gitblog_09746/article/details/141971367https://blog.csdn.net/gitblog_09746/article/details/141971367gitblog_09746Fri, 29 May 2026 07:32:28 +0800<![CDATA[E5-base-4k完整指南：如何参与开源贡献与获取专业支持]]>https://blog.csdn.net/gitblog_09746/article/details/141973053https://blog.csdn.net/gitblog_09746/article/details/141973053gitblog_09746Fri, 29 May 2026 07:26:35 +0800<![CDATA[对比分析：Granite-Embedding-97M-Multilingual-R2 vs E5-small vs BGE-M3，谁更适合你的应用？]]>https://blog.csdn.net/gitblog_09746/article/details/141977334https://blog.csdn.net/gitblog_09746/article/details/141977334gitblog_09746Fri, 29 May 2026 07:25:29 +0800<![CDATA[CANN/catlass稀疏GEMM搬运模板]]>https://blog.csdn.net/gitblog_09746/article/details/141977601https://blog.csdn.net/gitblog_09746/article/details/141977601gitblog_09746Fri, 29 May 2026 07:22:21 +0800 [代码位置](https://link.gitcode.com/i/7f40ad11d4aaca7675d316a3b6300771) [TOC] ## 功能说明 `TileCopySparseTla` 是稀疏 GEMM 专用的 TLA 搬运模板，与 [TileCopyTla](https://link.gitcode.com/i/808e8d84ebd7000309be84486100]]>