GraphLearn-for-PyTorch (GLT) is a PyTorch-based graph learning library that aims to make distributed GNN training and inference easy and efficient.

GLT utilizes GPU-accelerated graph sampling and utilizes UVA to reduce the transformation and duplication of vertex and edge features. For large-scale graphs, it supports distributed training on multiple GPUs or multiple machines through efficient distributed sampling and feature lookup. In addition, GLT provides flexible distributed training deployment to meet different needs.

The GTL schema is as follows: