The deep-learning textbook that runs in Colab

What it does

This repository holds the Jupyter notebooks for Deep Learning for Coders with fastai and PyTorch, the O’Reilly book by Jeremy Howard and Sylvain Gugger. Twenty chapters walk through image classification, NLP, tabular data, and model internals using the fastai layered API and PyTorch, and they double as the curriculum for the fast.ai MOOC.

The interesting bit

This is a published textbook rendered as executable code, split by an unusual license: the Python code is GPL v3, but the prose and markdown cells are under strict copyright that forbids redistribution, reformatting, or commercial use. It treats the words as a book and the code as software.

Key highlights

• Twenty chapters spanning MNIST, pet breeds, NLP, tabular data, convolutions, ResNet, and the Learner class
• One-click Google Colab links for every chapter; no local environment required
• README translations in nine languages including Spanish, Korean, Chinese, and Bengali
• Serves as the basis for both the fast.ai online course and the printed O’Reilly edition
• Includes dedicated chapters on ethics, production deployment, and mid-level API usage

Caveats

• Prose and markdown cells are not open source: private forks are allowed, but redistribution or commercial use of the text is explicitly prohibited
• Contributors must assign copyright of any pull requests to the authors

Verdict

A solid choice if you want to learn deep learning by running a structured, book-quality curriculum inside a notebook. Look elsewhere if you need freely reusable documentation or a conventional library to import.