Kaggle Support ?

[tunbmilk]

Is there anyway , i can get it installed on kaggle ?

Can u make one which can be easily installed on kaggle ?

[brian6091]

I no longer use this precompiled whl. I switched to the facebook uploads. Instructions here:
https://github.com/huggingface/diffusers/blob/75d53cc83966b4046e5a329ddf7baa6aa24f52e2/docs/source/en/optimization/xformers.mdx

I don't use kaggle, so I can't say whether that will work for you.

[Testertime]

Is there anyway , i can get it installed on kaggle ?

Can u make one which can be easily installed on kaggle ?

Found this through Google and I wondered the same. After further searching, I found this Kaggle notebook. Search for the line that says "# install xformers", there is all the code (or alternative here so that it's preserved):

# install xformers
!pip install torch==1.12.0+cu113 torchvision==0.13.0+cu113 --extra-index-url https://download.pytorch.org/whl/cu113

from subprocess import getoutput
s = getoutput('nvidia-smi')
if 'T4' in s:
    !pip install https://github.com/camenduru/stable-diffusion-webui-colab/releases/download/0.0.14/xformers-0.0.14.dev0-cp37-cp37m-linux_x86_64.whl
elif 'P100' in s:
    !pip install https://github.com/Isotr0py/xformers-prebuild-wheels/raw/main/P100/xformers-0.0.14.dev0-cp37-cp37m-linux_x86_64.whl

[Testertime]

I'm liking to tag you @tunbmilk because I just did something that you are going to like. You can check my compiled xformers repo for T4 and P100 👍 It's more up-to-date now than the xformers I found last time. Feel free to share this and reupload elsewhere, xformers must become less painful to install 😂