How to save the ExplainerDataset?

[CXX1113]

Such as
dataset = ExplainerDataset( graph_generator=BalancedBinaryTree(num_levels=8), motif_generator=CycleMotif(num_nodes=6), num_motifs=80, transform=MultiConstantFeat(value=1., num_dim=10, norm=True),)
The purpose of saving the data set is to keep the generated data set unchanged the next time you run the program.

[akihironitta]

AFAIK, there's no interface that PyG provides specifically to save generated data, so you have to implement it on your side. I think you can utilise self.save(...) to save data like other InMemoryDatasets, e.g., https://github.com/pyg-team/pytorch_geometric/blob/fd7ff50287e6bd251247804f15e0ae96f866ddac/torch_geometric/datasets/movie_lens_1m.py

[rusty1s]

You can also just set a seed so that data is equvialent across runs. Otherwise, you can just do

torch.save(dataset, 'dataset.pt')

dataset = torch.load('dataset.pt')