divelab/DIG

Health v2 · maintenance only

Adoption not part of the score

2007 stars · 289 forks observed · 2026-08-28

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

DIG (Dive into Graphs) is a Python library built on PyTorch Geometric that provides a unified testbed for advanced graph deep learning research, covering graph generation, self-supervised learning, explainability, 3D graphs, and out-of-distribution generalization. It complements lower-level libraries like PyG and DGL by offering higher-level, research-oriented implementations and benchmarks.

Use cases

• research graph generation models
• self-supervised learning on graphs
• explainability methods for graph neural networks
• 3D graph deep learning experiments
• graph out-of-distribution generalization benchmarks
• develop new graph deep learning methods on a unified testbed

When to choose

• you need higher-level graph research tasks beyond what PyG or DGL provide
• you want reproducible benchmarks for graph generation, explainability, or self-supervised learning
• you are doing academic research in graph deep learning

When to avoid

• you only need basic GNN layers and data loading - use PyG or DGL directly
• you need a production system rather than a research testbed
• you require frequent updates or long-term support - development activity has slowed

Facets

library · maturity maintenance

deep-learning machine-learning deep-learning machine-learning python graph-neural-networks graph-generation explainability self-supervised-learning 3d-graphs research pytorch-geometric algorithms

Member repositories

For agents

markdown · JSON · MCP: product_card(name="divelab/DIG")