Is there a detailed process for creating a shared conda env for all users in TLJH or Jupyterhub

JupyterHub The Littlest JupyterHub
1

I would like to create a conda virtual environment in jupyterhub to use unsloth for multiple users. but I have used as many methods as I could find on the web and none of them work. There are two directions if possible, one is to successfully read the env from anaconda3 in home to jupyterhub in opt (jupyterhub single user mode works, but I don’t know how I can do it for multiple users). The other direction was to create the environment directly from the command line in jupyterhub and successfully install the pytorch-cuda package.

Is there anyone who can help me? Implementation code would be great!

2

The TLJH user environment is already a shared conda environment for all users:

What errors are you running into?

4

Thank you this time I tried to create a virtual environment under the admin user in TLJH, this time it worked, but there is a problem with installing the dependencies needed for the virtual environment, is there a similar solution for this?

(llamafactory) jupyter-admin@KI-smw:~$ conda install torch torch_npu metrics deepspeed bitsandbytes vllm galore badam gptq awq aqlm qwen modelscope quality auto_gptq optimum
Collecting package metadata (current_repodata.json): failed

NoWritablePkgsDirError: No writeable pkgs directories configured.
  - /opt/tljh/user/pkgs
  - /home/jupyter-admin/.conda/pkgs

(llamafactory) jupyter-admin@KI-smw:~$ sudo -E conda install torch torch_npu metrics deepspeed bitsandbytes vllm galore badam gptq awq aqlm qwen modelscope quality auto_gptq optimum
Collecting package metadata (current_repodata.json): / WARNING conda.models.version:get_matcher(546): Using .* with relational operator is superfluous and deprecated and will be removed in a future version of conda. Your spec was 1.7.1.*, but conda is ignoring the .* and treating it as 1.7.1
done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.
Collecting package metadata (repodata.json): / WARNING conda.models.version:get_matcher(546): Using .* with relational operator is superfluous and deprecated and will be removed in a future version of conda. Your spec was 1.8.0.*, but conda is ignoring the .* and treating it as 1.8.0
WARNING conda.models.version:get_matcher(546): Using .* with relational operator is superfluous and deprecated and will be removed in a future version of conda. Your spec was 1.6.0.*, but conda is ignoring the .* and treating it as 1.6.0
WARNING conda.models.version:get_matcher(546): Using .* with relational operator is superfluous and deprecated and will be removed in a future version of conda. Your spec was 1.9.0.*, but conda is ignoring the .* and treating it as 1.9.0
done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.

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

  - galore
  - quality
  - modelscope
  - awq
  - badam
  - aqlm
  - vllm
  - qwen
  - gptq
  - metrics
  - torch
  - torch_npu

Current channels:

  - https://conda.anaconda.org/conda-forge/linux-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.
5

Already solved, I decided to use transformers as env