A full Keras curriculum, archived but open

What it does

This repository holds the complete Spring 2023 edition of Washington University’s T81-558 course, “Applications of Deep Neural Networks.” It is structured as thirteen modules of Jupyter notebooks that walk through Python preliminaries, Pandas, TensorFlow and Keras fundamentals, then advance through CNNs, GANs, transfer learning, time series, NLP with Hugging Face, reinforcement learning, and finally Flask deployment. Each module includes reading assignments, programming exercises, and links to external datasets, effectively functioning as an open-source textbook.

The interesting bit

The repository is explicitly the archived Keras/TensorFlow track; the university now teaches the course in PyTorch, and the README redirects current students to the new version. That makes this a rare, fully intact snapshot of a university-level Keras curriculum rather than an active project. The instructor also notes that the content evolves with technology, yet this branch remains frozen as a comprehensive historical reference.

Key highlights

• Thirteen modules covering the full stack from Python lists to TPUs and model retraining
• Practical applications including StyleGAN3, YOLOv5, Deoldify, Atari game playing, and Kaggle competitions
• Dedicated sections on Bayesian hyperparameter optimization, ensemble methods, and transfer learning for vision, NLP, and style transfer
• Complete course textbook published in-repo with an arXiv citation and ISBN
• Datasets and a hybrid on-campus/online syllabus structure with dated assignments

Caveats

• The README clearly states this is the previous Keras version; active WUSTL students should use the PyTorch repository instead
• Assignment due dates and campus meeting schedules are pinned to Spring 2023, so the logistics are stale even if the concepts hold up
• Some specific library versions (e.g., YOLOv5, StyleGAN3) may drift from current releases since the course snapshot is fixed

Verdict

Self-taught developers looking for a rigorous, end-to-end Keras curriculum with homework and datasets will find a full semester here. If you need the current PyTorch course or a quick cheat sheet, look at the instructor’s newer repo instead.