deeplearning-models

This repository is a collection of deep learning architectures, models, and tips packaged as Jupyter Notebooks. Jupyter Notebooks are interactive documents that mix code, charts, and explanations, which makes them well suited for learning how a model works step by step. The README presents implementations across two of the main deep learning frameworks: TensorFlow and PyTorch (with some additional examples using PyTorch Lightning), so the same idea is often shown in more than one style.

The notebooks are organised by topic. There is a section on traditional machine learning starters such as the Perceptron, Logistic Regression, and Softmax Regression, trained on simple datasets like 2D toy data, the Iris dataset, and MNIST (a classic dataset of handwritten digits). A multilayer perceptron section adds variations like dropout, batch normalization, and a from-scratch backpropagation implementation. The largest area covers convolutional neural networks, which are models specialised for image data: basic CNNs, weight-initialisation variants, and named architectures including AlexNet, DenseNet-121, an All-Convolutional Net, LeNet-5, MobileNet v2 and v3, and Network in Network, with examples trained on datasets such as MNIST, CIFAR-10, and QuickDraw.

Someone would use this when they want to study or reuse reference implementations of well-known deep learning models side by side, rather than reading papers or building from scratch. The tech stack stated in the README is Python notebooks running on TensorFlow, PyTorch, and PyTorch Lightning. The full README is longer than what was provided.