understanding train_batches and val_batches #question

[ma7555]

Hello, I am afraid I didn't quite understand from the documentation clearly what should be passed as train_batches and val_batches...

is it the the predictions made by the model that would be used to train the metalearner?

From what I understand, I need to join all predictions made by the trained model for the validation data that was used, and use this data for training the meta-learner.