It should still create a new venv, but inherit the packages from the system-wide (or specific venv) installed packages. Meaning it will not reinstalled packages you already installed, but it will ive you the option of just replacing a specific package (or install a new one) without reinstalling the entire venv
PompousParrot44 , so you mean like a base conda env?
Configuring trains-agent to use conda is done here:
Then for every experiment trains-agent will create a new conda environment based on the requirements of that experiment.
You can tell it to inherit the base conda env (or the one it is running from, I think) by setting