Capacity-Controlled Global Attention for Graph Transformers

Benchmark Model Rank Results
graph-property-prediction-on-ogbg-molhivSigGate-GT#2Test ROC-AUC: 0.8247
graph-regression-on-peptides-structSigGate-GT#4MAE: 0.2431
graph-regression-on-zinc-500kSigGate-GT#5MAE: 0.059
graph-property-prediction-on-ogbg-molpcbaSigGate-GT#9Test AP: 0.2984
graph-classification-on-peptides-funcSigGate-GT#17AP: 0.6947