dan_the_3rd

And we just have our first binary wheel uploaded! https://pypi.org/project/xformers/0.0.16rc377/#files You can test it with `pip install -U --pre xformers`

Hi @archywillhe and thanks for opening this issue. It looks like there is an issue with your xformer/pytorch setup. You mention gunicorn - have you tried running your code outside...

Is it possible to share a reproducible example? Like a minimal version of `script.py` if you can

Stale PR - closing. The files modified here will also be deprecated in a next version

Stale PR - plus we plan to deprecate these components

Hi, It's possible that the compilation might be OOMing. Can you try to reduce "MAX_JOBS", as written in the [troubleshooting section](https://github.com/facebookresearch/xformers#install-troubleshooting) ?

The recommended way of installing xformers is using the [conda packages](https://github.com/facebookresearch/xformers#installing-xformers) we provide. No build is necessary then. It just requires to have pytorch already installed with conda (see here:...

You can try "python -m xformers.info" Closing as this should be resolved now:)

I think I've seen this bug before ... may be due to wrong triton version. Can you make sure you have this one installed? https://github.com/facebookresearch/xformers/blob/main/requirements-test.txt#L30

@blefaudeux do you have any idea what's going on here? This is on RTX 3060, might be an issue of insufficient shared memory? (we have a "CUDA: invalid argument")