PyTorch-Ignite PyTorch-Ignite


End-to-end complete hands-on PyTorch-Ignite tutorials with interactive Google Colab Notebooks.


1. Getting Started

Welcome to PyTorch-Ignite’s quick start guide that covers the essentials of getting a project up and running while walking through basic concepts of Ignite. In just a few lines of code, you can get your model trained and validated. The complete code can be found at the end of this guide.

2. Transformers for Text Classification with IMDb Reviews

In this tutorial we will fine tune a model from the Transformers library for text classification using PyTorch-Ignite. We will be following the Fine-tuning a pretrained model tutorial for preprocessing text and defining the model, optimizer and dataloaders.


1. Distributed Training on CPUs, GPUs or TPUs

This tutorial is a brief introduction on how you can do distributed training with Ignite on one or more CPUs, GPUs or TPUs. We will also introduce several helper functions and Ignite concepts (setup common training handlers, save to/ load from checkpoints, etc.) which you can easily incorporate in your code.

2. Machine Translation using PyTorch Ignite

This tutorial is a brief introduction on how you can train a machine translation model (or any other seq2seq model) using PyTorch Ignite. This notebook uses Models, Dataset and Tokenizers from Huggingface, hence they can be easily replaced by other models from the 🤗 Hub.

3. Reinforcement Learning with Ignite

In this tutorial we will implement a policy gradient based algorithm called Reinforce and use it to solve OpenAI’s Cartpole problem using PyTorch-Ignite.


1. Collective Communication with Ignite

In this tutorial, we will see how to use advanced distributed functions like all_reduce(), all_gather(), broadcast() and barrier(). We will discuss unique use cases for all of them and represent them visually.