Overview

Welcome to the SMU O’Donnell Data Science and Research Computing Institute (ODSRCI) workshop on PyTorch Geometric (PyG), a powerful library for deep learning on graphs and irregular structures. PyG extends PyTorch with optimized data structures and operations tailored for graph-based learning, enabling efficient implementation of Graph Neural Networks (GNNs). In this session, we will explore PyG’s core functionalities, including data handling, message passing, and model building for tasks such as node classification, link prediction, and graph classification. Whether you’re new to graph-based deep learning or looking to enhance your PyTorch expertise, this workshop will provide hands-on guidance and practical applications to help you leverage PyG effectively.

Materials

• ODSRCI Workshop: PyG

• ODSRCI Workshop: PyG Repo

• PyG

Workshop Website

Fig. 1 PyG workshop website: https://southernmethodistuniversity.github.io/pyg/.

Recording

A screen recording of the workshop is available here for those with SMU credentials.

Table of Contents

• Environment Setup
• HPC Portal Jupyter Lab Session
• PyG Introduction