graphdeeplearning/benchmarking-gnns

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

2671 stars · 459 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

An open-source benchmarking framework for graph neural networks (GNNs) published as a JMLR 2023 paper. It provides reproducible pipelines, diverse mathematical and real-world graph datasets (ZINC, AQSOL, PATTERN, CLUSTER, etc.), and a common parameter budget for fair comparison of GNN architectures.

Use cases

• benchmark graph neural network architectures fairly
• reproduce published GNN results on standard datasets
• compare message-passing GNNs under the same parameter budget
• evaluate new GNN designs on molecular graph regression
• test GNNs on theoretical graph properties like cycle counting
• add a custom dataset or GNN architecture to a benchmark suite

When to choose

• you need standardized, reproducible comparisons of GNN models
• you are doing GNN research and want established baselines and leaderboards
• you want curated graph datasets like ZINC, AQSOL, PATTERN, or CLUSTER with ready pipelines

When to avoid

• you need a production GNN library for deploying models rather than benchmarking
• you work outside PyTorch/DGL ecosystems
• you need actively updated support for the latest GNN architectures, as the repo's last release was 2023

Facets

framework · maturity maintenance

benchmarking machine-learning deep-learning testing machine-learning deep-learning developer-tools python cross-platform graph-neural-networks gnn pytorch dgl benchmark-suite graph-representation-learning reproducibility jupyter-notebook algorithms research gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="graphdeeplearning/benchmarking-gnns")