rasbt/deeplearning-models

A collection of various deep learning architectures, models, and tips

What it solves

This repository provides a comprehensive collection of deep learning architectures and model implementations to help developers and students understand how to build and train various neural networks using popular frameworks.

How it works

The project is organized as a series of Jupyter Notebooks that demonstrate the implementation of different model types. These notebooks provide code for both PyTorch (including PyTorch Lightning) and TensorFlow, covering a wide range of architectures from basic machine learning to advanced deep learning models.

Who it’s for

It is designed for practitioners and learners of deep learning who want concrete, runnable examples of neural network architectures implemented in PyTorch and TensorFlow.

Highlights

  • Diverse Architectures: Covers traditional ML (Perceptrons, Logistic Regression), MLPs, CNNs (AlexNet, DenseNet, VGG, ResNet, MobileNet), Transformers (DistilBERT), and Autoencoders.
  • Evaluating Modalities: Includes implementations for computer vision, natural language processing (sentiment classification), and ordinal regression.
  • Framework Flexibility: Provides parallel implementations in PyTorch, PyTorch Lightning, and TensorFlow.
  • Specialized Techniques: Includes examples of normalization layers, metric learning (Siamese Networks), and various autoencoder types (Fully-connected, Convolutional, and Variational).

Related

  • Project
    AI4Finance-Foundation/FinRL

    An open-source framework for financial reinforcement learning that enables users to automate quantitative trading through a train-test-trade pipeline.

  • Project
    datawhalechina/leedl-tutorial

    LeeDL-Tutorial (also called the "Apple Book") is a Chinese-language deep learning tutorial based on Hung-yi Lee's popular course. It reorganizes the course into a clear, book-like structure, provides detailed derivations for every formula, and adds extra topics and code, making deep learning easier to learn for Chinese speakers.

  • Project
    NVIDIA/Megatron-LM

    Megatron-LM is a GPU-optimized library and reference framework designed to scale the training of massive transformer models across large clusters using advanced parallelism and precision techniques.

  • Project
    MathFoundationRL/Book-Mathematical-Foundation-of-Reinforcement-Learning

    The project is a repository for the LaTeX slides of a mathematical RL book, offering a friendly yet rigorous introduction, with extensive supplementary videos and community code.