fastbook

Why we included this project

Jeremy Howard and Sylvain Gugger wrote this book as a set of Jupyter notebooks, and it doubles as the companion to the fastai library they created. If you are learning deep learning with PyTorch, it is hard to find a more direct route: every chapter pairs plain prose with working code that runs in Google Colab, so you never have to set up a local Python environment. The chapters move from image classification basics through tabular data and NLP into the internals of the fastai and PyTorch APIs, which makes the book work both as a structured course and as something you can jump into at any point. Since the notebooks are the teaching material for the free fast.ai MOOC, they track how the library is actually used rather than drifting into theory. Teams bringing new ML engineers up to speed will also get useful grounding from the chapters on production deployment and ethics before anyone touches a real system.