Matthias Fey

Sorry for late reply: Can you run with ``` pip install --verbose torch-scatter ``` This should give a more descriptive error message.

Try `torch-scatter==2.0.8` for PyTorch 1.8, see [here](https://data.pyg.org/whl/torch-1.8.1%2Bcu111.html).

You can find the new link here: https://github.com/rusty1s/deep-learning-on-graphs/tree/master/masterthesis (german only :( )

This is not the actual error message. Do you still need this package? It is actually deprecated since PyTorch provides its own `unique` function.

I think we should implement it on the table2excel side, cause it seems that it's working as intented for js-xlsx. I can look into it, but not before the weekend...

What do you mean?

No not yet, would be very thankful for a pull request.

Thanks for reporting. Looks like our wheels somehow require `GLIBC=2.29`. I need to investigate. In the meantime, you should have no problem to install `pyg-lib` from source directly: ``` pip...

The issue is that you don't have a CUDA toolkit installed: ``` Failed to find nvcc. Compiler requires the CUDA toolkit. Please set the CUDAToolkit_ROOT variable. ``` I will try...

> Also, looks like the latest pytorch version is 2.1.1, is it compatible with pyg_lib? thanks Yes. Regarding the other issue: Is it resolved by https://github.com/pyg-team/pyg-lib/issues/266#issuecomment-1794538047?