Route
Beginner
Reads: L1 Core of every section, plus Appendices A and B as needed
Does: E1 at the end of each chapter
The companion to Machine Learning, All the Way Down
Read it one layer at a time.
Every section has a Core layer for the idea, a Deepen layer for the formal statement, and a Research layer for the frontier. Choose how far to go. Nothing is lost if you stop early.
Chapters loaded
24
of 24 planned
Sections
144
each in three layers
Labs available
58
of 58 planned
Glossary terms
552
with a symbol index
Start here
Begin at Core, then go deeper when you are ready
Core gives the idea with one concrete case. Deepen gives the formal statement. Research points at the frontier and the open questions. Each layer stands on its own.
Four ways to read
The routes are printed in the book's front matter. Each one names what to read and what to do, so you can choose before you start.
Route
Practitioner
Reads: L1 and L2 of the sections you choose
Does: E1 and E2 for chosen chapters
Route
Researcher
Reads: L1, L2 and L3 of every section
Does: E1, E2 and E3
Route
Refresher
Reads: Chapter diagnostics, then only what you get wrong
Does: Targeted E1
Diagnostic map
Each chapter opens with three questions. All three right means you may skip L1 and go to L2. One or two wrong means read L1, then E1. Three wrong means read L1 slowly, do E1, and return to the diagnostic.
Placeholder
The map will show every chapter as a node, coloured by your local diagnostic result, with a route through the chapters you still need. It is not wired to chapter diagnostics yet: the diagnostic JSON export lands with the first merged chapters.
The three Chapter 1 questions exist in src/data/chapters/c01.json; this build does not yet render them as an interactive check.
Chapter map
24 chapters are loaded. The outline plans 24.
Chapter 1
1. Rules That Turn Inputs Into Outputs
What does a program promise you?
Chapter 2
2. Learning a Rule From Examples
What must a table of examples carry for prediction to be possible?
Chapter 3
3. How to Know a Claim Is True
What does "for all" commit you to?
Chapter 4
4. Experiments That Do Not Lie
How much does one run tell you?
Chapter 5
5. How Long Does It Take?
How do we measure the work a program does?
Chapter 6
6. The Real Machine
How is a decimal number stored in 32 bits?
Chapter 7
7. Lists of Numbers That Do Things
What makes a list of numbers a vector?
Chapter 8
8. Projection and the Best Fit
How do you measure the angle between two lists?
Chapter 9
9. Directions That Stay Put
Which directions does a matrix leave alone?
Chapter 10
10. Change, Slopes and Approximation
What does it mean to get arbitrarily close?
Chapter 11
11. Probability as Reasoning
What rules must a degree of belief obey?
Chapter 12
12. Averages, Spread and Surprises
How do you attach a number to an outcome?
Chapter 13
13. Estimating From Data
What makes one guess better than another?
Chapter 14
14. Lines, Curves and Boundaries
What does the best straight line mean?
Chapter 15
15. Judging a Model Honestly
How optimistic is a model about itself?
Chapter 16
16. Information and Compression
How much information is in one symbol?
Chapter 17
17. Going Downhill on Purpose
How does a gradient point to a better answer?
Chapter 18
18. Getting Gradients From a Computer
When do you use symbols, numbers or a program?
Chapter 19
19. Neural Networks: Simple Parts, Stacked
How does one unit draw a line?
Chapter 20
20. Training Deep Networks Without Breaking Them
Where does the gradient go?
Chapter 21
21. Attention, Images and Sequences
Why does looking at neighbours help?
Chapter 22
22. Making Pictures Out of Noise
What does it mean to learn a distribution?
Chapter 23
23. Learning by Acting
What is different about learning from reward?
Chapter 24
24. Scale, Language Models and What They Learned
How does a model learn to predict the next token?