// the find
rasbt/deeplearning-models
A collection of various deep learning architectures, models, and tips
A large collection of Jupyter notebooks implementing deep learning architectures in PyTorch (and some TensorFlow 1.x), spanning MLPs, CNNs, RNNs, VAEs, GANs, transformers, and ordinal regression. It's from Sebastian Raschka, who writes clearly and teaches well — this is a reference library, not a framework. Aimed at people learning deep learning who want working, runnable implementations alongside explanations.
The breadth is genuinely useful: you get AlexNet, VGG, ResNet, DenseNet, MobileNet, and more in one place with consistent style. The ordinal regression section is rare — CORAL and CORN implementations are hard to find elsewhere with clean PyTorch code. Notebooks include both raw PyTorch and PyTorch Lightning versions for most architectures, which makes it easy to see what Lightning is actually abstracting. The author's helper utilities (helper_train.py, helper_evaluate.py) are reusable and not buried inside notebooks.
Last push was February 2024 and the TensorFlow side is stuck on TF 1.x, which is dead. The description column in most README tables just says 'TBD' — so you get the code but no prose explanation of what makes each variant different or when you'd pick one over another. Transformer coverage is thin: only DistilBERT fine-tuning, nothing on modern architectures like vision transformers or anything post-2022. No diffusion models, no LoRA/PEFT examples — the collection stops well before where practitioners are working now.