I am planning to do a hyperparameter tuning using hydra’s optuna plugin and mlflow logging. Instead of using hydra I generally use Hydra-zen to define and manipulate the config file within the python file as a string. Do i need to rebuild the docker container to install these packages?
If you want to use another package with Modulus, I see a couple options one could try:
- Install the new package via pip inside a running docker container. This won’t put the package into the image we provide, but you can keep and reuse your stopped container which you could start again.
- Build a new development Docker image using the Modulus image a the parent image. Basically your new docker file would have
FROM modulus:22.09then pip install whatever. - Build a Modulus Docker image from scratch (not recommended since this will remove some items) adding the new package into the Dockerfile.
Hopefully one of these works for you.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.