yunshen-blogv1

Read a theory markdown file (model/algorithm explanation), then generate a companion Jupyter notebook (.ipynb) via a four-phase pipeline: Research → Plan → Implement → Review. Every notebook contains TWO parallel paths in a SINGLE file: - Learning path: understand the model by building/running key components - Engineering path: use mature toolchains for production-grade implementation Path depth and form adapt independently based on feasibility decisions. All implementations target free Colab (free T4 GPU available; must also work on CPU as fallback). Final notebook includes interview-oriented extension and project expression content.

Convert raw markdown notes, lecture materials, or fragmented technical drafts into a well-structured, publication-ready technical blog post. The transformation must: - Preserve all technical content, formulas, and code blocks exactly. - Improve logical structure and section hierarchy. - Introduce meaningful headings and subheadings where missing. - Clarify complex concepts without oversimplifying them. - Eliminate redundancy, incomplete thoughts, and fragmented sentences. - Ensure consistent terminology and notation throughout. - Improve readability while maintaining technical depth. - Avoid adding unsupported claims or fabricating information. - Avoid marketing language or exaggerated tone. - Maintain a professional, educational writing style. When applicable: - Add a concise contextual introduction (no fluff). - Add a structured summary or key takeaways section. - Highlight core insights, design motivations, and trade-offs. - Make implicit reasoning explicit. The result should read like a carefully edited t

将论文精读笔记（.md + .html + .ipynb）合并为高质量 MDX 博客文章，自动处理公式、图片、代码等 MDX 兼容性问题。触发词：论文精读、paper reading、MDX 博客、论文笔记。

将普通技术博客 Markdown 源稿整理为高质量 MDX 博客文章，自动处理公式、图片、代码块、裸花括号、移动端公式滚动等 MDX 兼容性问题。适用于算法原理讲解、工程实践总结、框架机制解析、normal 技术博客、公式排版、手机端适配等常规技术博客场景，不适用于带 .html / .ipynb 联动的论文精读场景。