Improving gnn model performance

[Favourj-bit]

GATConv

GraphNorm

Dropout for all layers

[Favourj-bit]

Try different number of layers Increase the number of hidden channels Use GraphNorm with GCN Try different dropout rate Different hidden channels for each layer Different GNN layer: Graphtransformer, GAT, GCN Different estimation function **Try different optimizer, learning rate and number of epochs