Storing min, max and mean of impute method for inference

[nils-fl]

Hi,

I am not sure how to use the impute function on my test data or during inference.
When I use it on my train data, I would like to store the transformer somehow, so that the imputation on my validation and test data is using the mean, min, max of the training data to avoid leakage.
I also need the (fitted?) imputation to use during inference to handle infs and nans.
Is there a way to store the imputation transformer in some kind of way?

EDIT:

OK, so I guess imputing is not working on the full data set with all ids, but inside one subset with a specific id. Which means, that it can be used in test and inference.