Thank you for making the code and paper publicly available. I encountered two issues while processing the TUD datasets regarding the lifting procedure:
- For the PROTEINS dataset:
When setting max_ring to 15 or higher, the process of converting graphs to complexes gets stuck and does not proceed. Specifically, the file PROTEINS_complex_list.pt becomes excessively large, reaching 7GB when k=15. For k=18, the conversion process for the PROTEINS dataset completely fails, and the progress bar remains stuck indefinitely.
- For other TUD datasets, such as COLLAB:
When dealing with a new dataset, I encountered issues during the lifting procedure. The step of converting graph data to complexes takes an extremely long time. Even with the smallest setting of k=4, the model fails to run on the resulting complex data.
I would appreciate any suggestions or solutions to these problems!
Thank you for making the code and paper publicly available. I encountered two issues while processing the TUD datasets regarding the lifting procedure:
When setting
max_ringto 15 or higher, the process of converting graphs to complexes gets stuck and does not proceed. Specifically, the filePROTEINS_complex_list.ptbecomes excessively large, reaching 7GB whenk=15. Fork=18, the conversion process for the PROTEINS dataset completely fails, and the progress bar remains stuck indefinitely.When dealing with a new dataset, I encountered issues during the
lifting procedure. The step of converting graph data to complexes takes an extremely long time. Even with the smallest setting ofk=4, the model fails to run on the resulting complex data.I would appreciate any suggestions or solutions to these problems!