selective attributes for multiple predictions

[ImranA10]

I've currently been using the Ludwig model wanted to know if it was possible to assign set attributes for individual predictions.
Lets consider an example dataset:
Input:
-Sales History: sequence of n previous days
-Day Type today+1: is this day a weekend/weekday
-Day Type today+2: as above
-Day Type today+3: as above

Output: prediction of Sales for the next 3 days (3 predictions)

Now when training I model I would like the following set up:
-Prediction of today+1 uses "Sales History" and "Day Type today+1" feature
-Prediction of today+2 uses "Sales History" and "Day Type today+2" feature
-Prediction of today+3 uses "Sales History" and "Day Type today+3" feature

So I ideally want the model to be able to use different features as some are only relevant for specific predictions, where as some are always relevant
I know I could train 3 different models but given Ludwig model can predict model outputs I'd like to try and see if I can manage this in a single model