Conda problem in WSL 2

Hey guys, I have another question, if I have many projects on my WSL 2 (e.x. Ubuntu 22.04 LTS), and I use Anaconda or Miniconda to manage the package and environment for each project, which means I need to respectively install different version of torch, CUDA (cudatoolkit), cuDNN…traditionally I’ll directly use commands from Pytorch official website (e.x. conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch), while I found that if I download from NVIDIA offical website, there is an option saying “WSL-Ubuntu” (maybe more compatible?). While it seems cannot be used for different project (according to the commands showed, only “sudo apt-get …”), so in this case, how can I deal with this dilemma?

1 Like


How did you solve this? We are currently having the same thoughts / problems.