torch-geometric

To perform node classification with graph neural networks, this Skill provides end-to-end PyTorch Geometric templates including model architectures like GCN, dataset loading scripts, and training loops to quickly establish baselines on citation networks.

Yes, graph classification and link prediction workflows are supported through coordinated PyTorch Geometric templates. These workflows apply to various domains including social graphs, knowledge graphs, and programming graphs for comprehensive GNN evaluation.

You need PyTorch, PyTorch Geometric, and NumPy installed to run graph neural network experiments. These libraries provide the foundational tensor computations and graph data structures required for the provided model templates and benchmarking scripts.

You can benchmark graph neural network models using the provided scripts and dataset references. The Skill includes benchmarking utilities and visualization tools to evaluate model performance across citation networks and other graph structures.

The quickest way to start training a GCN model on the Cora dataset is using the minimal two-layer GCN template provided. This quick start approach allows you to establish a baseline rapidly using PyTorch Geometric.

Yes, comprehensive references for GNN layers, datasets, and transforms are included. These references support the end-to-end graph neural network workflows by detailing the available PyTorch Geometric components for model customization.