Is your feature request related to a problem? Please describe.
Yes. For transductive tasks I need to se the node features of val and test by taking them from the train ones.
The method should also check that all nodes in the incoming dataset appear in the features of the dataset that is providing the features.
Describe the solution you would like
I can do:
train_dataset.enrich_node_features(
enricher=LaplacianPositionalEncodingEnricher(num_features=32),
enrichment_mode="replace",
)
val_dataset.hdata.x = train_dataset.hdata.x[: val_dataset.hdata.num_nodes]
test_dataset.hdata.x = train_dataset.hdata.x[:, : test_dataset.hdata.num_nodes]
but want something like:
train_dataset.enrich_node_features(
enricher=LaplacianPositionalEncodingEnricher(num_features=32),
enrichment_mode="replace",
)
val_dataset.enrich_node_features_from(dataset_with_features=train_dataset)
val_dataset.enrich_node_features_from(dataset_with_features=train_dataset)
Describe alternatives you've considered
Add a node_features getter on Dataset and use like this:
train_dataset.enrich_node_features(
enricher=LaplacianPositionalEncodingEnricher(num_features=32),
enrichment_mode="replace",
)
val_dataset.enrich_node_features_from(node_features=train_dataset.node_features)
val_dataset.enrich_node_features_from(node_features=train_dataset.node_features)
Additional context
NA.
Is your feature request related to a problem? Please describe.
Yes. For transductive tasks I need to se the node features of val and test by taking them from the train ones.
The method should also check that all nodes in the incoming dataset appear in the features of the dataset that is providing the features.
Describe the solution you would like
I can do:
but want something like:
Describe alternatives you've considered
Add a
node_featuresgetter on Dataset and use like this:Additional context
NA.