I have a stack ensemble of bagged LightGBM models fit with AutoGluon in Python.
Is there a way to perform the same inference as AutoGluon's predict() on this model, except entirely/mostly outside of AutoGluon, for example, with LightGBM's C API?
I'm aware of predictor._trainer.load_model() which returns a autogluon.tabular.models.lgb.lgb_model.LGBModel object, but am unclear how I would make predictions with that model in LightGBM directly.
