Common Issues and Fixes

1. Do not install packages on the login nodes or inside a Jupyter Notebook. It has to be done in a terminal/shell:

   $ interactive -t 30 -p htc
   $ module load mamba/latest

2. Do not install any packages in the base environment. If you see base in front of your username in the command line, please deactivate it:

   (base) [example_user@sol-login01:~]$
   (base) [example_user@sol-login01:~]$ source deactivate
   [example_user@sol-login01:~]$

3. Improper use of the conda commands will inject some code into your ~/.bashrc file, which will cause problems. To remove these lines:

   $ remove_conda_from_bashrc
   $ source ~/.bashrc

4. Improper use of the pip install commands will create a hidden cache folder, which usually breaks Python-related functions. To check and remove such folders:

   $ cd ~/.local/lib
   $ ls
   $ rm -rf python3*
   $ mamba clean --all

Please note that pip install can only be used after activating a mamba env:

$ module load mamba/latest
$ source activate myENV
$ pip install something

5. How to install a. Pytorch, according to their official document:

   $ interactive -t 60 -p htc
   $ module load cuda-12.6.1-gcc-12.1.0
   $ module load mamba/latest
   $ mamba create -n myENV_pytorch -c conda-forge python=3.12
   $ source activate myENV_pytorch
   $ pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu126
   $ mamba install -c conda-forge other_packages ...

   b. TensorFlow, according to their official document:

   $ interactive -t 60 -p htc
   $ module load cuda-12.6.1-gcc-12.1.0
   $ module load mamba/latest
   $ mamba create -n myENV_tensorflow -c conda-forge python=3.12
   $ source activate myENV_tensorflow
   $ pip install tensorflow[and-cuda]
   $ mamba install -c conda-forge other_packages ...

6. How to remove an env:

   $ interactive -t 60 -p htc
   $ module load mamba/latest
   $ mamba remove -n ENV_NAME --all
   $ mamba clean --all

Additional Help

If you require further assistance, contact the Research Computing Team:

• Ticket-based support via RTO Request Help Portal.
• Slack support via the #rc-support channel in the ASU Research Computing workspace.
• Weekly office hours for one-on-one assistance.

We also offer Educational Opportunities and Workshops.