Error when installing on MacOS Catalina

[danicuki]

when I run conda install pytorch=1.1 torchvision=0.3 cudatoolkit=10.0 -c pytorch

I get this error:

Collecting package metadata (current_repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.
Collecting package metadata (repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.

PackagesNotFoundError: The following packages are not available from current channels:

  - cudatoolkit=10.0

Current channels:

  - https://conda.anaconda.org/pytorch/osx-64
  - https://conda.anaconda.org/pytorch/noarch
  - https://repo.anaconda.com/pkgs/main/osx-64
  - https://repo.anaconda.com/pkgs/main/noarch
  - https://repo.anaconda.com/pkgs/r/osx-64
  - https://repo.anaconda.com/pkgs/r/noarch
  - https://conda.anaconda.org/conda-forge/osx-64
  - https://conda.anaconda.org/conda-forge/noarch

To search for alternate channels that may provide the conda package you're
looking for, navigate to

    https://anaconda.org

and use the search bar at the top of the page.

How to fix this?

[johndpope]

You're not going to get cuda working on a mac - except for High Sierra. Nvidia and apple were once in bed together - but things soured when they started doing stuff intel normally does - then they were ostracized - and you can't get drivers for modern day cards. I got myself a hp workstation /ssd / ubuntu 18.04 - it's working out - except you need a heap of 16gb VRAM to run this - so probably going to buy a RTX 5000 card - $2000 second hand. You can use docker with ubuntu / cudagl so you can fire up projects quickly.

[briancpark]

+1. I gave up on trying to run AI/ML programs on my MacBook. You are much better off SSHing into a cloud VM or using a windows/linux machine with NVIDIA and CUDA support.

Right now, I am using free 1 year credits to run jukebox on Google Cloud Platform/Google Colab.