// the find
ageron/handson-mlp
A series of Jupyter notebooks that walk you through the fundamentals of Machine Learning and Deep Learning in Python using Scikit-Learn, PyTorch, and Hugging Face libraries.
Companion notebooks for Aurélien Géron's O'Reilly book, now updated to use PyTorch and Hugging Face instead of TensorFlow/Keras. Covers classical ML through modern transformers, diffusion models, and RL. Best suited for developers learning ML with a textbook in hand — the notebooks are supplements, not standalone tutorials.
Coverage is genuinely deep: goes from linear regression all the way to SSMs, quantization, and diffusion models. The PyTorch pivot is the right call in 2025 — the TF version (handson-ml3) is aging faster. Colab support is first-class, so there's no local setup friction to get started. Math appendices on autodiff and linear algebra are worth having separately.
Heavily coupled to the book — without it, many notebooks lack the narrative context that makes the code make sense. 1,600 stars for a book companion repo by a well-known author is surprisingly modest, suggesting limited community velocity. No tests or CI on the notebooks themselves, so cells that silently break on library updates are a recurring issue with repos like this. The appendix on state space models (Mamba etc.) feels bolted on rather than integrated into the main curriculum.