snap-stanford/GraphGym
Stanford research platform for designing and evaluating Graph Neural Networks (GNNs) across configurable experiment spaces.
Not currently ranked — collecting fresh signals.
star history
GraphGym provides a highly modularized pipeline for GNN research covering data loading, model architecture, node/edge/graph-level tasks, and evaluation metrics. It enables reproducible experiments via configuration files and scalable management of thousands of parallel experiments with automated analysis across random seeds. The platform supports PyTorch Geometric and DeepSNAP backends, allowing researchers to register custom modules including data loaders, GNN layers, and loss functions.
Frequently asked
- What is snap-stanford/GraphGym?
- Stanford research platform for designing and evaluating Graph Neural Networks (GNNs) across configurable experiment spaces.
- Is GraphGym open source?
- Yes — snap-stanford/GraphGym is an open-source project tracked on heatdrop.
- What language is GraphGym written in?
- snap-stanford/GraphGym is primarily written in Python.
- How popular is GraphGym?
- snap-stanford/GraphGym has 1.9k stars on GitHub.
- Where can I find GraphGym?
- snap-stanford/GraphGym is on GitHub at https://github.com/snap-stanford/GraphGym.