yes, this one is running in venv and not docker, because I had some issues with cuda and docker. The virtualenv==20.4.6 in the requirements.txt. I think it broke after installing  clearml-serving   in the same env.