/graph_regression_template

A small model that performs graph regression on QM9

Primary LanguagePythonMIT LicenseMIT

A template for graph-level regression, demonstrated on QM9.

The network implemented is a graph attention network.

The code uses the following tools:

  • pytorch_geometric
  • pytorch_lightning
  • hydra
  • weights and biases

Installation

  • First follow the installation guide of pytorch geometric.
  • Install the other packages with pip install -r requirements.txt

author: Clement Vignac