Why Conda virtual environment in Ray Docker container?

I have been using the rayproject/ray-ml Docker image and noticed that it has a Conda base virtual environment installed that runs by default. I’m curious what the reason is for including and running a virtual environment inside this image?

1 Like

To extend your question, why was conda chosen as the virtual environment manager over the default python venv module? It’s quite difficult to extend the dependencies of a ray docker image when certain packages are unavailable for conda installation.