Multi-graph in One Sample.

[matcyr]

Hi, I am recently trying to implement a regression model based on multi-graph data interaction.
I would like to have multi-graphs saved in one sample and applied mini-batch. I found the example of having a pair of graphs here: https://pytorch-geometric.readthedocs.io/en/latest/notes/batching.html.
However, I am having 7 graphs together for one sample. Do I have a better way rather than using the 'follow_batch'?

Thanks,
Yurui

[wsad1]

Ah, i guess the problem you want to avoid is is you'll have to specify seven variabes for x and edge_index in your Data object and then pass 7 xs to follow_batch. Unfortunately at the moment follow_batch is the only way to do this.
You could use for loops and setattr in __init__ to make your code shorter

SevenGraphs(Data):
    def __init__(list_x, list_edge_index) # list of 7 node features and edge indicies
        for i in range(7):
            setattr(self, f'x_{i}', list_x[i]) 
``