snap-stanford/ogb resource

Health v2 · maintenance only

Adoption not part of the score

2094 stars · 406 forks observed · 2026-08-28

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

The Open Graph Benchmark (OGB) is a collection of benchmark datasets, data loaders, and evaluators for graph machine learning, covering node-, link-, and graph-level prediction tasks across diverse domains and scales. It integrates with PyTorch Geometric and DGL, providing automatic downloading, standardized splits, and unified evaluation with leaderboards.

Use cases

• benchmark graph neural network models on standard datasets
• download and preprocess graph datasets for node classification
• evaluate link prediction models with standardized splits
• train GNNs on large-scale graphs like MAG240M
• predict molecular properties from molecular graphs
• compare my model against the state-of-the-art graph ML leaderboard
• load knowledge graph datasets for triple imputation

When to choose

• you need realistic, standardized graph ML benchmarks with unified evaluation
• you want datasets compatible with PyTorch Geometric or DGL out of the box
• you need datasets spanning small to web-scale graphs across many domains
• you want to submit results to community leaderboards

When to avoid

• you need a general-purpose graph processing library rather than benchmarks
• your task requires custom datasets not covered by OGB's collection
• you work outside the PyTorch/DGL ecosystem and cannot use the loaders

Facets

dataset · maturity stable

machine-learning deep-learning benchmarking data-science machine-learning deep-learning data-science python cross-platform graph-neural-networks graph-machine-learning benchmark-datasets pytorch-geometric dgl leaderboards data-loaders algorithms

Member repositories

For agents

markdown · JSON · MCP: product_card(name="snap-stanford/ogb")