Linux Conda Xformers Install Issue

[Pathos14489]

Everything goes fine installing, except xformers. When I run pip install -e ., this error occurs:

The detected CUDA version (11.4) mismatches the version that was used to compile
    PyTorch (10.2). Please make sure to use the same CUDA versions.

[0xdevalias]

Potentially related:

• windows xformers install breaking #33
• Xformer performance decrease #23

[0xdevalias]

@Pathos14489 are you following the xformers installation instructions here?

• https://github.com/facebookresearch/xformers#installing-xformers

---

This is probably more an issue you should raise on the xformers repo, as it seems to be unrelated to SDv2 itself?

[xiankgx]

If you are using Linux, modify the environment to you python 3.9 instead and use prebuilt binaries.

name: ldm
channels:
  - pytorch
  - defaults
dependencies:
  # Changed to Python 3.9 to make use of precompiled binaries for xformers
  - python=3.9
  - pip
  - cudatoolkit=11.3
  - pytorch=1.12.1
  - torchvision=0.13.1
  - numpy=1.23.1
  - pip:
    - albumentations==1.3.0
    - opencv-python==4.6.0.66
    - imageio==2.9.0
    - imageio-ffmpeg==0.4.2
    - pytorch-lightning==1.4.2
    - omegaconf==2.1.1
    - test-tube>=0.7.5
    - streamlit==1.12.1
    - einops==0.3.0
    - transformers==4.19.2
    - webdataset==0.2.5
    - kornia==0.6
    - open_clip_torch==2.0.2
    - invisible-watermark>=0.1.5
    - streamlit-drawable-canvas==0.8.0
    - torchmetrics==0.6.0
    - -e .

Then, install xformers using as recommended in the official repo.
conda install xformers -c xformers/label/dev

[hangxiaotian]

you can use pip to install. For example,
pip install xformers