This benchmark suite provides evaluation scripts for semi-supervised node classification, graph classification, and point cloud classification and runtimes in order to compare various methods in homogeneous evaluation scenarios. In particular, we take care to avoid to perform hyperparameter and model selection on the test set and instead use an additional validation set.
$ pip install -e .