How to train multiple parameters with a GAN in PyTorch?

Hey :slight_smile:
I’m implementing actually a GAN, which should generate MNIST-Superpixel Graphs.
My model should learn how the nodes are connected and which node is important to represent the number in the graph.
So the connection problem does it solve very well, but how can my model the node features additionally?

Thank you in andvance :slight_smile: