LIME explanations with categorical features have different dimensions than the model parameters

[amir-rahnama]

I already read and follow this tutorial for cases when we have numerical and categorical features in the dataset. The problem is that in this way, LIME explanations and the model parameters will have different representations (and dimensions for example).

Is there a way to bypass this problem?

Also, what happens if I make dummy variables of my categorical features and then get LIME explanations after that? Is this somehow wrong and makes the explanations faulty?