Documentation needed for model_config

[cluhmann]

For both MMM and CLV, the model constructors take a config dict called model_config. I can find examples this dict in various notebooks, but I cannot find any specification about what keys/values are valid/invalid. Searching for "model_config" in the docs returns the model_config class, which contains nothing but parse_model_config which provides more examples (and a rather unhelpful invalid example) , but no specification. Is this documented anywhere? If so, it should be pointed to in the documentation more thoroughly. If not, it should be documented.

I also see that there are other (potentially nested) config dicts like intercept_tvp_config. Are the specifications of those documented anywhere?

[wd60622]

How about here: https://www.pymc-marketing.io/en/stable/notebooks/general/model_configuration.html#use-in-pymc-marketing

[cluhmann]

I guess that is some additional info. But where does one figure out that keys such as "alpha" and "beta" are valid? Or keys like "intercept", "saturation_beta", "saturation_lam", and "likelihood"? And/or when are such key valid and when are they invalid? Are they model-dependent? Are they tied to choices about the adstock and/or saturation functions?