torch-geometric

To train a graph neural network for node classification, this Skill provides ready-to-run PyG templates, preloaded datasets, and evaluation routines. You simply install PyTorch and PyG, then execute the generated templates on a sample dataset to train and benchmark models.

The best way to benchmark graph neural networks is by using the provided benchmarking scripts to compare models across standard datasets. These scripts evaluate node classification, graph classification, and link prediction workloads efficiently.

Yes, you need PyTorch and PyG installed in your Python environment to use the graph neural network templates. The Skill explicitly requires PyTorch, numpy, and torch-geometric dependencies to run the templates, benchmarking scripts, and visualization components.

Yes, you can visualize graph neural network models and datasets during experimentation using the built-in visualization scripts. The Skill includes references and visualization components that guide model design and experimentation using matplotlib and networkx.

Yes, this Skill works with link prediction and graph classification workloads by providing ready-made GNN templates. It satisfies requirements for Python environments with PyTorch and PyG, offering ready-to-run models and evaluation routines across these specific tasks.