Best practices on handling different data parametrizations for a model

I am using data dictionaries and referenced models a lot to enable code reuse and a productive team workflow. My use case is the following: I have developed an algorithm which is completly parametrized by variables contained in a data dictionary. Whats the best way to hold multiple variants of these variables? Model is exactly the same, only things that differ are the parameters. Intuitive use and version control compatibility is important for me.

Approaches I already reviewed:

• Init scripts --> do not comply with data dictionaries philosophy
• Variant subsystems --> require that a duplicate instance of the model file to allow attaching two different data dictionaries
• Input data via external port --> not very intuitive and results in complex submodels

Happy to hear your best practices on this and thank you very much!