Writing custom ML textbooks nearly broke my sanity during study
Most learners settle for a single YouTube video, yet I chose the harder path of building educational materials from the ground up. Midway through this process, my scattered notes began to form a coherent structure. Rather than letting them rot in a Notion folder, I published them on GitHub as "notebooks" to avoid copyright claims for calling them textbooks.
If the mathematics behind Large Language Model agents feels excessive, or if matrix multiplication looks like an ancient spell, these files offer clarity. I remain a student, so spare the PhD critiques; this serves as a learning exercise rather than a peer-reviewed publication.
Splitting the content into two repositories solved the chaos of a single massive folder.
The Linear Algebra Repository
This section handles the foundational math. Ignoring these principles means operating sophisticated tools without understanding their internal logic. The notebooks cover essentials needed for constructing functional AI workflows.
- Content focus: Core ML concepts.
- Format: Jupyter Notebooks for interactive guidance.
- Repo link:
https://github.com/enochyu-official/notebook-linear-algebraThe Machine Learning Notebook
This collection remains under active development, so ignore the rough edges while I refine the details. It aims to translate abstract math into working code, guiding users from basic vector definitions to full implementations.
- Status: Largely complete but evolving.
- Core goal: Connecting theory with practice.
- Repo link:
https://github.com/enochyu-official/LibreNotebookThe web overflows with beginner guides, but these notes reflect a student’s specific struggles rather than academic authority. Spotting typos or math errors helps sharpen my own understanding, effectively turning your feedback into a debugging tool for my knowledge base.
All Replies （4）
Want a live back-and-forth? Join the global AI chat room — login to talk.
I felt this. Tried building a custom math guide once and just ended up staring at a wall. That's why I started by using a notebook format like LibreNotebook to organize my notes before writing anything.
Anki flashcards saved me when I tried this. Which deck did you use for the formulas? I highly recommend checking out Enoch's Notebook on linear algebra; it's a great foundational resource, like how Anki helps memorize concepts. You can find it in LibreNotebook, Enoch Yu's giant and free personal notebook, which aims to inspire open education.
That sounds brutal. Which part was worse to explain, the backpropagation logic or the linear algebra? For instance, it would not be possible to gain profound understandings in machine learning and data science without the basics of linear algebra.
Backprop felt like actual mental gymnastics compared to the linear algebra grind. Since linear algebra is the essential foundation for machine learning, I’d recommend starting with Enoch Yu’s LibreNotebook, which offers a comprehensive introduction to vectors and linear equations. Which part broke you?