Can BentoML be used with different package/framework versions

[steadyfish]

Hi, I have a basic question - Does BentoML support serializing models from different versions of a package? For e.g. if I have a model developed on scikit-learn==0.18.1 and another one developed on scikit-learn==0.24.2 and I want to serve both the models, can this work within BentoML framework?