some module are lock and those are important for learning

The course is not well maintained, and rather superficial

need live coding labs to test out the scripts

I followed the course with interest, but I believe it suffers from a major identity crisis: it lacks enough substance to be a standalone course. It’s unclear what its target audience is. While it seems intended as an introductory course, it takes too much prior knowledge for granted. Furthermore, it rushes through advanced topics like tokenization and data loaders without providing enough detail or a logical framework to help build mental connections. The methodology is inconsistent—it oscillates between trivial details and significant omissions of key concepts. It lacks a proper introduction and is certainly not self-contained. Ultimately, it feels a bit like clickbait.

It basically just consists of Jupyter Notebooks without much explanation. I tried to understand the content and implement it on my own because the exercises did not help me understand it at all. As a result, it took me much longer than expected to work through the content, which disrupted my schedule. I would not recommend this course. Unfortunately, I am not aware of any better alternatives for learning about this topic.

For a learning resource it has some jumps that leave the learner with missing context. The heavy use of functional programming style of code made it really hard to read and understand

Tokenizer & dataloader are quite important parts but I am surprised by how shallow they are touched and how easy are the quiz questions.

I would not recommend this course. It is basically llm-slop being used to explain llms.

very superficial