Suppose I have a graph named G1, and a graph G2 with a self-loop added:
G2 = dgl.add_self_loop(G1)
When I use the explain_node function in dgl to explain a vertex Vi in G2, I found that the length of edge_mask returned by the explain_node method will be larger than the edge corresponding to the Vi in the original graph G1.
If I use the dgl.remofe_self_loop(G2), it will throw errors.
How to filter out the edges belonging to Vi in G1 from the G2 edge_mask?