Predict edge features with given node features
Hi developers,
I am working on crystal structure generation. I hope to find a suitable graph generative model. So I wonder if this model can generate edge features conditioned on given node features (they represent type of elements).
Moreover, crystal graphs have additional invariance. The edge features will change depending on the node ordering. But they still represent the same structure. Can such property be considered in this model?
Thanks.
Hi,
I am also workig on generating the edge features for each graph. Have you found a suitable graph generative model for this task? Thanks.
best wishes
Hi,
I wanna know whether your generated graphs by edp-gnn have node features. I use ENZYMES as training dataset and edp-gnn as score model. ENZYMES has node_attributions while generated graphs doesn't have. I wonder whether it's correctly.
Thanks.