Cluster GCN or Neighbour Loader for graph classification tasks

[shubhiambast]

Hi,
I am trying to do graph classification and the number of nodes in each graph vary between 2000 - 10,000. The node feature vector has 2000 features. Can anybody suggest if applying cluster-GCN or neighbor loader on each graph is efficient enough ? Or what kind of a model should be used to handle these many nodes in each graph? Thanks in advance.