On the Connection Between MPNN and Graph Transformer

Benchmark Model Rank Results
graph-classification-on-peptides-funcGatedGCN+RWSE+virtual node#24AP: 0.6685±0.0062
graph-regression-on-peptides-structGCN+virtual node#19MAE: 0.2488±0.0021
node-classification-on-pascalvoc-sp-1GatedGCN+LapPE+virtual node#12macro F1: 0.3103±0.0068