What is PyTorch Geometric (PyG)?

PyTorch Geometric is a specialized library built on PyTorch specifically designed for deep learning on graphs and other irregular data structures.

Does this skill support large-scale graph training?

Yes, the skill includes patterns for neighbor sampling, mini-batch processing, and distributed training to handle datasets that exceed memory limits.

Which benchmark datasets are supported?

The skill provides guidance for loading common benchmarks like Planetoid (Cora, CiteSeer), TUDataset (ENZYMES), and molecular datasets like QM9.

Can I create custom graph convolution layers?

Absolutely. This skill provides implementation patterns for inheriting from the MessagePassing class, allowing you to define custom message, aggregate, and update functions.

PyTorch Geometric (PyG)

Name: PyTorch Geometric (PyG)
Author: Sologa

bySologa

0•

データサイエンスとML

Develops and trains Graph Neural Networks (GNNs) using the PyTorch Geometric library for irregular data structures and geometric deep learning.

The PyTorch Geometric (PyG) skill empowers Claude to architect, implement, and optimize Graph Neural Networks for complex data structures. It provides specialized patterns for node and graph classification, link prediction, and molecular property prediction. By leveraging this skill, developers can efficiently handle irregular data like social networks, citation graphs, and 3D point clouds using industry-standard architectures like GCN, GAT, and GraphSAGE, complete with best practices for mini-batch processing and large-scale graph learning.

主な機能

01Support for heterogeneous graphs with multi-type nodes and edges

020 GitHub stars

03Implementation of state-of-the-art GNN layers including GCN, GAT, and GraphSAGE

04Efficient mini-batch processing and handling of large-scale graph datasets

05Standardized patterns for molecular property prediction and chemical informatics

06Custom MessagePassing interface for developing bespoke graph convolution layers

ユースケース

01Building recommendation systems and social network analysis tools

02Developing 3D geometric deep learning models for point clouds and meshes

03Predicting molecular properties for drug discovery and chemical research

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add sologa/codex-pipeline torch_geometric

For use in Claude.ai and ChatGPT

Download Skill