Specformer

Code of Specformer: Spectral Graph Neural Networks Meet Transformers

How to run

• For node-level task, e.g., signal regression and node classification, you should first run preprocess_node_data.py to generate .pt files for each dataset.
• For graph-level taks, you can direcly run dgl_main.py.

Q&A

Any suggestion/question is welcome.

Reference

If you make advantage of Specformer in your research, please cite the following in your manuscript:

@inproceedings{specformer2023,
  author={Deyu Bo and 
          Chuan Shi and
          Lele Wang and
          Renjie Liao},
  title={Specformer: Spectral Graph Neural Networks Meet Transformers},
  booktitle = {{ICLR}},
  publisher = {OpenReview.net},
  year      = {2023}
}