How do I train two XGBoost models at the same time?

Hi, I am trying to learn two independent variables, but with a loss function that takes both of them as input.

My idea is vaguely in the line of

   alpha = xgb.XGBRegressor()
   beta = xgb.XGBRegressor()
def obj_func(preds_alpha, preds_beta, y_train):

How can I achieve this?

You can wait for 1.6 release, which supports multi-output model.