Is it correct to concat the edge_index tensors of graphs batch as GCNConv.forward() input?

[fportera]

I'm trying to make a GNN to work with several datasets (MUTAG, PROTEINS, etc...), using 2 GCNConv layers for a Neural Network.
While the 'x' input tensor can be batched to a batch-size first dimension, It's not clear to me how to stack/concatenate the relative batch edge_index tensors for batch-size graphs.
If it is possible to use multiple batch graphs with these GCNConv layers....
Thanks for your attention!

[EdisonLeeeee]

• torch_geometric.data.DataLoader will concatenate edge_index to a batched graph automatically.
• Yes, you can use GCNConv, that makes sense.

[fportera]

Many thanks!