Cover of Inside a Transformer, slowly
The library

Foundation Models · Book 01 of 04

Inside a Transformer, slowly

An illustrated, math-and-code-friendly guide from raw web pages to token batches, attention, residual streams, training runs, inference, and Kimi K3.

For ML-literate readers who know basic neural networks and want stronger Transformer intuition.

Reading time
136 minutes
Contents
19 chapters + self-test
Verified
August 4, 2026
Read online Download EPUB 8.9 MB Markdown source ↘

A guided route through the book

Inside the book.

  1. 01 See the whole model
  2. 02 Move from raw text to tokens
  3. 03 Turn tokens into vectors
  4. 04 Build attention and the residual stream
  5. 05 Assemble a Transformer block
  6. 06 Train, infer, and cache
  7. 07 Bridge from GPT-2 to Kimi K3

The complete learning path

In this series.

You’re at the beginning. Book 02 · NextInside nanoGPT