Evaluation scripts for various methods on the Cora, CiteSeer and PubMed citation networks. Each experiment is repeated 100 times on either a fixed train/val/test split or on multiple random splits:
- GCN:
python gcn.py - GAT:
python gat.py - Cheby:
python cheb.py - SGC:
python sgc.py - ARMA:
python arma.py - APPNP:
python appnp.py
Run the whole test suite via
$ ./run.sh