Help me understand the zinc dataset

[tonydavis629]

In the docstring, it describes Zinc as having 249,456 graphs, 23.2 nodes, 49.8 edges, 1 feature and 1 class. When I load the dataset, I get a very different shape.

from torch_geometric.datasets import ZINC
dataset = ZINC(path)
print(dataset.data)
>>> Data(x=[5094322, 1], edge_index=[2, 10959434], edge_attr=[10959434], y=[220011])

What is going on here? It looks like every sample is getting concatenated into the Data object. I would expect Data.x to be of the type List[Data] of length 249,456, with each Data object representing a molecular graph. Is this expected behavior, or is this a bug?

[rusty1s]

dataset.data is our internal storage format. You can access individual elements via dataset[i].