Is adj normalization essential? about pygat

[Diego999]

Hi @Andy5250

I think this is how it's done in the original implementation. But if both are the same, shouldn't change anything.

Best,

[bousejin]

Hi Diego,

Thank you so much for the code! In layers.py this line (edge = adj.nonzero().t()) says AttributeError: 'Tensor' object has no attribute 'sparse'. How can I get rid of it?

[Diego999]

Did you check your version of pytorch ?

[bousejin]

It is '1.0.0'.

[Diego999]

Ok I think the API might have changed since. I don't have time to fix it but a PR is welcome ;-)