tornikeo

@rthadur Any updates on this? Could you replicate this bug?

I'm not sure - I think you have to do the following: - add bear body type armors - add their recipes - make sure the dino has bear bodytype...

Here's an idea. We should add a joke .exe file that immediately opens a rickroll in the browser. That'd be fun.

Ran into this issue when doing a docker build from `pytorch/pytorch:1.13.0-cuda11.6-cudnn8-runtime`. This should be included within the `requirements.txt`.

@Wavelet303 You'd have to use a docker container. I've written a sample Dockerfile which allows Mask3D to work independently of the host OS. You should try to do that as...

The smallest 2 examples necessary for train+val sets are 20MB in total. @JonasSchult should I make a PR for adding those? Here's an image of the folder: 

Yeah, #61007 is on halt right now. Don't see myself workin on it anytime soon.

Sure, @merrygin that'd be amazing. :heart: Is there anything I can to help you with that?

@JannikSchneider12 For greedy cosine, I solved this issue by writing a custom CUDA kernel with numba. It's around x100-x500 faster than matchms. You can use it online on [colab](https://colab.research.google.com/drive/15RpnFVeiJmpR7i3VGF8qtBpA_Fsdb2EQ?usp=sharing), just...

@niekdejonge It is possible to discard pytorch, it will just require me to add a new numba kernel that does sparse reduction. Pytorch is only really used in "sparse" mode,...