Reviewing Hands on Large Language Models

Hands on Large Language Models is effectively an introduction to the architectures of large language models (LLMs). A clear and concise book on the components of LLMs. I think it does a really good job of expressing the modularity of the models in a practical way, and for what the book strives to do, its a really good read.

The only concern is that it does skip some details, namely in the math (discussed below) but also misses some discussions on the emergent abilities of LLMs, for example.

A pro and con of this book is that offers low mathematical rigor. This obviously makes the book more accessible and a lot of that isn't needed to understand the content of this book for its purposes. However, we also skim a lot of the technical details on topics like cosine similarity and even vector embeddings. I do think it makes sense to leave this as out of scope for the book though, as the math required encompasses multiple subjects anyways (you probably should do calculus 1-3, linear algebra 1, and some probability (arguably not enough) before learning the math, which obviously won't be covered effectively in one book).

A very minor gripe is with the flow as well. Arguably, the prompt engineering section should follow the fine tuning section instead.

If you found this helpful or enjoyable, please consider subscribing to our newsletter for more.  

You can reach out to us here for consulting services.