How edge_index have to be built in order to fed the Dataset class for GCNcon net?

I’ve implemented a Neural Net that uses 2 GCNConv layers, besides other 3 Linear layers.
The problem is that I don’t understand how the Dataset object should be fed with edge_index’s for several graphs, since I’m considering a graphs binary classification task.
Since the graphs have different numbers of edges, the edge_index cannot be an array of fixed dimensions. So I don’t understand how to build the batch diagonal matrix described in the official PyG docs: Batching .
Is it possible to have a numerical example?
Thanks for any help!