How This Blog Became What It Is Today

This English version was machine-translated from the Chinese original. Technical terms may require verification.

The article was really called An Example. It's half the opening statement, half the markdown writing notes. Looking back, most of the notes are out of date, but I still want to leave the portal, by the way, to record how the site has become a little bit like this.

The questions in this article can also be addressed2025 Annual summary: far and far、The Illustrated Guide to a Ph.D.How the concept of a relatively close read together is developed in different contexts.

Start with a useful blog.

From warehouse history, blogs were set up on December 26, 2025. At the time, the website was using Jekyll, and the page also contained a GitBook-style layout and resources. Target is direct: put Markdown in. _posts, let GitHub Pages generate accessible articles.

The article was then changed to an example. The text is written about how the document should be named, what fields Front Matter has, and how Claude Code, GLM 4.7 and Cursor are involved in the build-up process. At that time, it was more like an operating manual left to itself, and it could not speak of a complete personal station.

In the following months, there was an increase in the number of articles and repeated adjustments to the front page and classification. The old system can continue to work, but the site is no longer just a sporadic blog, and the original structure is beginning to look exciting.

Replace with Hexa and Redefine

On June 28, 2026, bloggers moved from Jekyll to Hego and used Redefine themes. Move article to source/_postsThe site configuration, thematic resources and independent pages are also being rebuilt. Previously published articles continue to be kept /blog/:year/:month/:day/:title/ in the form of a link to avoid the expiry of the old address.

The next day, a round of adjustments was made to the front page banners, navigation, the About menu and personal pages, and the processing of the Chinese-English interface was completed. From a blog with a personal introduction, the website slowly becomes a set of sites shared by the home page and technical blogs. Thank the company AI for paying for this.

It takes more time to organize than to change the subject.

When the relocation is completed, the real time is spent on the article itself. At the beginning of July 2026, the classification name was harmonized into English, and the classification level was subsequently redesigned and the label was compressed from over 200 to over 180. The interface is mainly in English, but the body is not limited to languages; Chinese notes and English metadata can be available simultaneously.

From 4 to 7 July, the learning notes accumulated in the past were concentrated in computer science, machine learning, interpretable machine learning, in-depth learning, mathematics, probability statistics and data analysis. These articles have made the archives much more complete and the front page too crowded at once. The site has since added the processing of hidden articles: old notes can be kept on the archive, classification and tab page, but need not all appear on the front page. A further import of research notes took place on July 11th in the same way.

There are now both more complete technical articles, as well as learning notes, project reviews and occasional whims. Classifications and labels will continue to be adjusted, although the basic principles are clear: the first page retains what is expected to be publicly displayed in the near future, and the archives are responsible for keeping the remaining records.

I don't write blogs just to record what I have learned, but I want to slowly develop my own Belief and make them public. Technology will change, and Belief may not be right from the start; but I would like to make it clear why it is so judged, then let the project, reality and time validate it, and amend it. Continuing to write these provisional conclusions is both training my judgment and hoping that my work will find someone who can continue my discussions.

Why do you keep this article?

It is no longer a writing template and will not appear on the front page. It was retained mainly because the old link still existed and that history had indeed happened. The article is still visible when you directly access the original address or enter it from the archive, classification and tab page.

I set the date of its article before the first undisclosed article of the day. When this is viewed by time, it remains the page before the beginning of the public writing, but is still in the archives in silence.