InMemoryDataset.copy() keep original indicies

[anthonysirico]

Is there a way to keep the original index when using copy?

dataset = Dataset('test')
train = dataset.copy(idx=range(split))
test = dataset.copy(idx=range(split+1, len(dataset))

[rusty1s]

I guess the easiest way is to add a graph_id to each data object. This way, the initial indices are maintained. Would that work for you?

[anthonysirico]

Yes thank you.