[SOLVED] Can't change virtual environment within Azure ML notebook

Can't change virtual environment within Azure ML notebook

For some reason the jupyter notebooks on my VM are in the wrong environment (ie stuck in (base)). Furthermore, I can change the environment in the terminal but not in the notebook. Here is what happens when I attempt !conda activate desired_env in the notebook:

CommandNotFoundError: Your shell has not been properly configured to use 'conda activate'.
To initialize your shell, run

    $ conda init <SHELL_NAME>

Currently supported shells are:
  - bash
  - fish
  - tcsh
  - xonsh
  - zsh
  - powershell

See 'conda init --help' for more information and options.

IMPORTANT: You may need to close and restart your shell after running 'conda init'.


# conda environments:
#
base                  *  /anaconda
azureml_py36             /anaconda/envs/azureml_py36
azureml_py38             /anaconda/envs/azureml_py38
azureml_py38_pytorch     /anaconda/envs/azureml_py38_pytorch
azureml_py38_tensorflow     /anaconda/envs/azureml_py38_tensorflow

I tried the answers here (e.g., first running !source /anaconda/etc/profile.d/conda.sh).

I also tried activating the environment using source rather than 'conda activate': !source /anaconda/envs/desired_env/bin/activate. This runs but doesn't actually do anything when I see the current environment in conda env list

Edit: also adding that if I install a package in the (base) environment in the terminal, I still don't have access to it in jupyter notebook.

Solution

I'm the PM that released AzureML Notebooks, you can't activate a Conda env from a cell, you have to create a new kernel will the Conda Env. Here are the instructions: https://learn.microsoft.com/en-us/azure/machine-learning/how-to-access-terminal#add-new-kernels