Another question: is

bentoml build

supposed to create a bento with dependencies already included? I thought it would have the dependencies since

bentofile.yaml

has a section for pip install dependencies, however I don’t think my bentos have any dependencies included. Is that expected? I’m hoping to use Spark batch inference on my bento, do I need to pip install all dependencies from my Spark job that invokes the bento server? Thanks!

Dependencies should be already there inside bento - this is the accepted behaviour. Make sure that you properly set path to your req. file: https://docs.bentoml.com/en/v1.1.11/concepts/bento.html#pip-install-options