This is a PyTorch implementation of the Graph Convolutional Network (Kipf and Welling, 2016).
In order to install the required libraries it is recommended that you create a conda environment by running
conda env create -f environment.yaml
and before running any experiments activate the environment:
source activate gcn
The file gcn.py accepts several arguments. For instance, to train a
model with the default configuration on GPU 0 you may run the command:
python gcn.py --config_path default.yaml --gpu 0
When the --gpu argument is not used training is performed on the CPU.