Dive into Deep Learning, from Scratch
An interactive pure-Python adaptation of the open textbook 'Dive into Deep Learning' (d2l.ai) by Zhang, Lipton, Li, and Smola, used under CC BY-SA 4.0. You rebuild the book's core ideas from scratch — tensors as lists, gradients by hand, and tiny trainable models — with no numpy or torch required.
- Python
- 45 lessons
- ~9 hours
- Free
- by Aston Zhang, Zachary C. Lipton, Mu Li, Alexander J. Smola
Learn Python for free on Libre Academy with the Dive into Deep Learning, from Scratch course: 45 interactive lessons you complete by writing real code in a built-in editor, graded instantly by hidden tests. No paywall, no sign-up. It's part of the free, open-source desktop app for macOS, Windows and Linux.
What you get
- Hands-on lessons with a real editor and instant, test-graded feedback.
- Free and open source, with no account required to start.
- Works offline in the free desktop app for macOS, Windows and Linux.
- Python runs instantly in the app.
What's inside
- Introduction: Learning from Data
- Data Manipulation: Tensors in Pure Python
- Linear Algebra
- Calculus and Automatic Differentiation
- Probability and Statistics
- Linear Regression from Scratch
- Softmax Classification
- Multilayer Perceptrons
- Generalization and Weight Decay
- Convolutional Neural Networks (Concepts)
- Sequences and Recurrent Networks (Concepts)
- Attention, Transformers, and Where to Go Next
Download to start Dive into Deep Learning, from Scratch → More free Python courses
Helpful reading
Why passive video doesn't make you a programmer · Bring Your Own Book: turning any PDF into a course