Do We Need Anisotropic Graph Neural Networks?

Benchmark Model Rank Results
graph-property-prediction-on-ogbg-code2EGC-M (No Edge Features)#8Test F1 score: 0.1595 ± 0.0019Ext. data: No
graph-property-prediction-on-ogbg-code2PNA (No Edge Features)#10Test F1 score: 0.1570 ± 0.0032Ext. data: No
graph-property-prediction-on-ogbg-code2MPNN-Max (No Edge Features)#12Test F1 score: 0.1552 ± 0.0022Ext. data: No
graph-property-prediction-on-ogbg-code2EGC-S (No Edge Features)#13Test F1 score: 0.1528 ± 0.0025Ext. data: No
graph-property-prediction-on-ogbg-molhivEGC-M (No Edge Features)#24Test ROC-AUC: 0.7818 ± 0.0153Ext. data: No
graph-property-prediction-on-ogbg-molhivEGC-S (No Edge Features)#28Test ROC-AUC: 0.7721 ± 0.0110Ext. data: No
node-property-prediction-on-ogbn-arxivEGC-S (100k)#57Test Accuracy: 0.7219 ± 0.0016Ext. data: No
node-property-prediction-on-ogbn-arxivEGC-M (100k)#62Test Accuracy: 0.7196 ± 0.0023Ext. data: No