// the find
fchollet/deep-learning-with-python-notebooks
Jupyter notebooks for the code samples of the book "Deep Learning with Python"
Companion notebooks for François Chollet's Deep Learning with Python, now on its third edition covering Keras 3 with JAX, TensorFlow, or PyTorch backends. This is the canonical hands-on material for the book — code only, no prose. It's for people working through the book, not standalone learners.
Keras 3 backend-agnostic design is genuinely useful — you can run the same notebook against JAX, PyTorch, or TensorFlow without touching anything but one environment variable. The progression from math fundamentals through transformers to image generation is well-sequenced. All three editions are preserved in the same repo, so you can track how the field and Chollet's own thinking have evolved over eight years. Colab-first setup means zero local environment pain for most readers.
The notebooks are intentionally stripped of explanation — code blocks and section headers only. Without the book, they're close to useless; you're buying a $50+ textbook to make a free GitHub repo functional. No standalone exercises or self-assessment; it's pure transcription of book code. Kaggle dependency for several chapters adds friction and a second account to manage. The repo structure has three editions flat in one directory with no changelog — comparing what changed between editions requires manual diffing.