Graph Neural Network Support

cuGraph-DGL has been removed from cuGraph GNN as of release 25.06. We recommend migrating to cuGraph-PyG, which offers the same functionality along with additional features like support for heterogeneous sampling and a unified API. The cuGraph team is not planning any further work in the DGL ecosystem going forward.

---

The cuGraph GNN stack

---

Here is a talk that explains Training GNNs at Internet Scale using cuGraph and WholeGraph

• cugraph_pyg
• WholeGraph

Blogs to explain how RAPIDS cuGraph supports GNN’S

• Optimizing Memory and Retrieval for Graph Neural Networks with WholeGraph, Part 1

• Getting Started with Large-Scale GNNs using cuGraph Packages for DGL and PyG