PyTorch Geometric for Graph Neural Networks
Overview
PyTorch Geometric (PyG) is the standard library for geometric deep learning with PyTorch. It provides optimized implementations of Graph Neural Networks (GNNs), efficient data structures for graph data, and scalable solutions for large-scale graph learning.
Key Capabilities:
- Efficient sparse tensor operations for graphs
- 100+ pre-implemented GNN layers (GCN, GAT, GraphSAGE, etc.)
- Scalable data loaders with neighbor sampling
[Description truncada. Veja o README completo no GitHub.]