PyTorch Geometric

Health v2 · maintenance only

Adoption not part of the score

24098 stars · 4062 forks observed · 2026-09-26

What it is AI-extracted, prompt v1, taxonomy v1, 2026-10-12, confidence 0.99

PyTorch Geometric (PyG) is a Python library built on top of PyTorch for building and training Graph Neural Networks. It provides a unified API, data loaders, and a collection of GNN layers and benchmark datasets.

Use cases

• train a node classification model on citation network data
• perform link prediction on knowledge graphs
• apply GNNs for molecular property prediction in drug discovery
• process 3D point clouds or meshes with graph convolutions
• benchmark graph algorithms using built‑in datasets
• run multi‑GPU training of large graph models

When to choose

• You are already using PyTorch and need a seamless GNN extension.
• You require fast prototyping with a high‑level, tensor‑centric API.
• You need built‑in loaders for common graph datasets and multi‑GPU support.
• You want to integrate recent GNN research papers with minimal effort.

When to avoid

• Your project is based on TensorFlow or JAX and you cannot add PyTorch as a dependency.
• You need a production‑grade serving solution rather than a research‑oriented library.
• You require a lightweight library with no compiled extensions.

Facets

library · maturity active

deep-learning machine-learning data-science parser artificial-intelligence machine-learning deep-learning python windows graph-neural-networks geometric-deep-learning pytorch gnn torch-geometric graph-processing linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="pyg-team/pytorch_geometric")