Build a Large Language Model (From Scratch)

Sebastian Raschka's 'Build a Large Language Model (From Scratch)' from 2024 is unique — it's a practical guide that combines modern theory about transformers, attention, and token handling with actual code.

The book covers tokenization with BPE (exactly as mentioned in the tokens-and-ngrams article), attention mechanisms, and the practical choices that determine a working model. It's a book for those who want to not just understand but also implement.