Matthias Fey
Matthias Fey
You are correct that `TUDataset.load` is not applicable due to additonal information being stored. That's why we use [custom logic](https://github.com/pyg-team/pytorch_geometric/blob/master/torch_geometric/datasets/tu_dataset.py) in `TUDataset` to load the dataset. I personally think this...
Wow, this is pretty cool. Can we fix the tests so that we are Python 3.8 compatible (no Python 3.10 type hints such as `str | int`)?
Looks like I overlooked this issue. Really sorry. To which second for-loop are you referring to?
Got it, I think this depends a bit how your graph and its edge features look like (the bottleneck here would still be `networkx`). If you have a simple example,...
Thank you for this issue. We always wanted to clean-up the example folder, sorry that it looks confusing to you. I see what I can do. In the meantime, happy...
Happy to include this. Do you want to contribute it?
Does `conda install pyg -c pyg` work for you?
And what happens if you run `conda install pyg -c pyg -c nvidia -c pytorch`?
You are right that this part is not included in this repo, I am sorry. You should be able to run "serial" mode by setting `_async=False` in the `ScalableGNN` base...
It's not possible for me to merge 50 files and 9000 LOC. Do we need the model implementations for this PR?