Verified project record

dmlc/dgl

dgl is a Python package designed to ease the implementation and scaling of deep learning models on graphs. It is a framework-agnostic, high-performance, and scalable library that works with PyTorch, TensorFlow, and Apache MXNet.

Project overview

dgl stands out for its ability to scale graph neural network training to billion-sized graphs across multiple GPUs or machines while remaining framework-agnostic, making it a strong choice for demanding graph-based machine learning workloads.

Project type: Model Development · Infrastructure
Deployment: Refer to project documentation
License: Apache-2.0

Best for

Researchers and developers who need a scalable, framework-agnostic Python package for implementing deep learning models on graphs.

Key capabilities

Provides a powerful graph object that can reside on either CPU or GPU, bundling structural data as well as features.
Provides efficient and customizable message passing primitives for Graph Neural Networks.
Provides many state-of-the-art GNN layers and modules for users to build new model architectures.
Conveniently train models on large-scale graphs across multiple GPUs or multiple machines, scaling to billion-sized graphs.
A command-line interface to get started with training, using and studying state-of-the-art GNNs.
Collects a rich set of example implementations of popular GNN models across a wide range of topics.

Getting started

Installation is straightforward via pip or conda. Users can get started by importing the library or using the DGL-Go CLI.

Evidence and sources

GitHub project description: Python package built to ease deep learning on graph, on top of existing DL frameworks.
README: DGL is an easy-to-use, high performance and scalable Python package for deep learning on graphs. DGL is framework agnostic, meaning if a deep graph model is a component of an end-…
README: DGL provides a powerful graph object that can reside on either CPU or GPU. It bundles structural data as well as features for better control. We provide a variety of functions for…
README: Moreover, DGL provides many state-of-the-art [GNN layers and modules](https://docs.dgl.ai/api/python/nn.html) for users to build new model architectures.
README: It is convenient to train models using DGL on large-scale graphs across **multiple GPUs** or **multiple machines**. DGL extensively optimizes the whole stack to reduce the overhea…

Reviewed at: 2026-07-17T23:02:54Z