I would like to know what is the data representation used for final traning of submissions entries? Is it the same bag of cards as shown in examplary python code? I am asking as different data representations (tranformations) tends to have very different optimal hyperparameters settings (epsilon, gamma, C). If it is not bag of cards can we know on what representation the final traning is performed?
The bag-of-cards data representation used during the evaluation is the same as in the starter notebooks.