Matthias Fey

Thanks for your interest and I am glad that you like it. I will try to add multi-GPU support if I have some free-time, but it shouldn't be too hard...

`device` refers to the device your model is on, while `emb_device` refers to the device where the historical embeddings are stored. In general, `device=cuda` and `emb_device=cpu`. Note that the `device`...

Yes, this looks correct to me. Histories will be on CPU while model parameters are on `cuda:3`. If there is a process running on `cuda:0`, that is definitely a bug...

Thanks for your interest. It is correct that we do not treat `Reddit` as inductive (I found out that some works do while others do not). If you want to...

The usage of CUDA streams can parallelize memory device transfers (via `*.to(device, non_blocking=True`) and actual GPU kernel execution. That's exactly what we are using it for. From what I know,...

Thanks for reporting. This is indeed wrong and I fixed it. Luckily, I confirmed that it has been working successfully anyway.

Thanks for the updates. It looks like with Pyroma 4, `description_content_type` rather than `long_description_content_type` is set in `get_build_data`, leading to `content_type` being `None` [here](https://github.com/regebro/pyroma/blob/master/pyroma/ratings.py#L347).

It just happened for me here: https://github.com/pyg-team/pytorch_geometric/pull/4368 (needed to revert back to "3.3"). I made the following change [here](https://github.com/regebro/pyroma/blob/master/pyroma/ratings.py#L347): ```python content_type = data.get("long_description_content_type", None) if content_type is None: content_type =...

:P Thanks a lot for the quick fix :)

Except for the pre-processing part (which is pure numpy), this model can be easily implemented with PyTorch Geometric (e.g. see `mnist_graclus.py`). The pre-processing segmentation/adjacency/feature computation can be found [here](https://github.com/rusty1s/embedded_gcnn/tree/master/lib/segmentation). After...