search

PetarV- / GAT

Graph Attention Networks (https://arxiv.org/abs/1710.10903)
https://petar-v.com/GAT/
MIT License
3.18k stars 642 forks source link

ppi #6

Closed nnzhan closed 5 years ago

nnzhan commented 6 years ago

Hi, thanks for sharing the code, could you please provide codes for training ppi data? cuz I see it has a significant improvement over other methods on this dataset.

sidneyaraujomelo commented 6 years ago

Codes for ppi would be really great, I second that. Thanks a lot for sharing the code, anyway.

PetarV- commented 5 years ago

I realised that I left this issue unanswered - I'm sorry.

At this time, we have decided not to include the PPI training script, to avoid overcluttering the repository. However we have provided (in utils/process_ppi.py) the script file for preprocessing the data. With its outputs, you should have everything you need (along with appropriate edits to execute_cora.py). Should you have any issues with this, please contact me directly.

It should also be noted that our PPI results are no longer state-of-the-art. As far as I know, Gated Attention Networks (GaAN; UAI 2018) hold it at this time: https://arxiv.org/pdf/1803.07294.pdf