Skip to content
Cook ML
GitHubRSS

About Cook ML

Cook ML is a place to learn machine learning the way a recipe teaches cooking: one step at a time, with the ingredients measured out.

Most explanations of language models either stay at the level of analogy or jump straight into notation. The tutorials here try to do both in order. Each idea starts with a familiar example, then the mechanism, then the math, and then a worked calculation with real numbers you can check by hand.

  • Tutorials. Multi-chapter series meant to be read in order. The first series, LLM Fundamentals, follows a language model from next-token prediction to deployment and evaluation.
  • Blog. Shorter posts on papers, tools, and lessons from building ML systems.

Cook ML is written by Zheng Zhang, an ML engineer working on LLM agents and applied ML. The material grew out of personal study notes that were rewritten for readers who are new to the field but want depth.

If you spot a mistake or an unclear explanation, please open an issue on GitHub.

Figures in the tutorials are either exact calculations under stated assumptions or clearly marked teaching sketches. They are not benchmark results.