Graph Inductive Biases in Transformers without Message Passing

Benchmark Model Rank Results
graph-classification-on-cifar10-100kGRIT#4Accuracy (%): 76.468
graph-classification-on-mnistGRIT#10Accuracy: 98.108
graph-classification-on-peptides-funcGRIT#14AP: 0.6988±0.0082
graph-regression-on-pcqm4mv2-lscGRIT#11Validation MAE: 0.0859
graph-regression-on-peptides-structGRIT#10MAE: 0.2460±0.0012
graph-regression-on-zincGRIT#4MAE: 0.059
graph-regression-on-zinc-500kGRIT#3MAE: 0.059
graph-regression-on-zinc-fullGRIT#6Test MAE: 0.023
node-classification-on-clusterGRIT#1Accuracy: 80.026
node-classification-on-patternGRIT#2Accuracy: 87.196