torch-geometric - Build GNNs with PyTorch Geometric
Guide users in representing graph data, building GNNs, and handling node classification, graph classification, link prediction, and heterogeneous graphs with PyG.
Tags
Updated: 2026-10-02PyTorch Geometricgraph neural networksGNNnode classificationlink predictiongraph classificationheterogeneous graphs
Typical Inputs
What this skill does
- Represent graph data
- Build GNN models
- Implement message passing
- Train node classifiers
- Perform graph classification
- Perform link prediction
- Process heterogeneous graphs
- Apply graph transforms
Inputs
- GNN task requirements
- Graph data
- Node features
- Edge features
- Labels and masks
Outputs
- Implementation guidance
- PyTorch Geometric code
- Model configurations
- Training patterns
- Evaluation patterns
Requirements
- PyTorch environment
- torch_geometric package