Why does tg.transforms.to_sparse_tensor switch row and col?

[zhiqiangzhongddu]

Hey,

While reading the source code of tg.transforms.to_sparse_tensor, I found it switches the place of row and col when creating the final adj. It is confusing and could anyone explain to me why we need such an operation.
Link to the code: https://github.com/rusty1s/pytorch_geometric/blob/a719b0419f89c9d19535dee01ef9e6d42dbb5a35/torch_geometric/transforms/to_sparse_tensor.py#L51

Thanks.

[zhiqiangzhongddu]

@rusty1s has answered this question in the Issues #2446 (comment).