acbull/pyHGT
A graph neural network architecture for learning on large-scale heterogeneous and dynamic graphs, published at WWW 2020.
Not currently ranked — collecting fresh signals.
star history
This repository implements the Heterogeneous Graph Transformer (HGT), a GNN architecture designed to handle heterogeneous graphs with multiple node and edge types. The model adapts the transformer mechanism to graph-structured data, using type-specific parameters and attention to capture relations across different entity types. It provides sampling algorithms for scalable subgraph extraction and training scripts for downstream tasks like node classification and link prediction.
Frequently asked
- What is acbull/pyHGT?
- A graph neural network architecture for learning on large-scale heterogeneous and dynamic graphs, published at WWW 2020.
- Is pyHGT open source?
- Yes — acbull/pyHGT is open source, released under the MIT license.
- What language is pyHGT written in?
- acbull/pyHGT is primarily written in Python.
- How popular is pyHGT?
- acbull/pyHGT has 927 stars on GitHub.
- Where can I find pyHGT?
- acbull/pyHGT is on GitHub at https://github.com/acbull/pyHGT.