understanding-oversquashing
understanding-oversquashing copied to clipboard
Dear contributors, I found the numba.py file performance differs from cuda.py under `understanding-oversquashing/gdl/src/gdl/curvature` folder. Shoudn't [this](https://github.com/jctops/understanding-oversquashing/blob/main/gdl/src/gdl/curvature/numba.py#L18) be `continue` instead of `break` ? Also line 29. best regards
Hello and thank you for your work. I can't find the results of the paper on the node classification task. Can I have an example reproducing the results please? Thanks
It'd be great if we could use your great findings to find nodes/edges prone to oversquashing and possibly fix those!
Hi, thanks for your inspiring work! Could you please release the full training and testing scripts with specific configurations?