// the find
mrdbourke/tensorflow-deep-learning
All course materials for the Zero to Mastery Deep Learning with TensorFlow course.
This is the full notebook set for Daniel Bourke's paid 'Zero to Mastery Deep Learning with TensorFlow' course, covering fundamentals through CNNs, transfer learning, NLP, and time series forecasting. It's for someone with 6+ months of Python and a prior ML course who wants a code-first, from-scratch walkthrough of TensorFlow/Keras rather than a reference library.
The README's fix log is unusually honest about API churn - it documents exact breaking changes (EfficientNetB0 swapped for EfficientNetV2B0, lr renamed to learning_rate, keras.layers namespace moves) with dated entries and discussion links, so you're not left guessing why your run doesn't match the video. Exercise solutions for every notebook are community-contributed and public, which most paid-course repos don't bother open-sourcing. The progression (regression -> classification -> CNN -> transfer learning -> milestone projects) is genuinely well-sequenced for building intuition incrementally.
Notebooks 03-10 are gated behind the paid course - the public repo gives you code but not the explanations, so as a free resource it only really covers the first two notebooks. Last pushed August 2024, and the TF-specific framing (EfficientNet feature extractors, Universal Sentence Encoder from TF Hub) already reads dated next to where NLP/CV teaching has moved since. There's no CI beyond a docs build workflow, so there's no signal whether these notebooks still execute cleanly against current TensorFlow releases - the manual fix log is the only maintenance mechanism. This is pedagogical code meant to be read and retyped, not something you'd import or extend in a real project.