I am wondering whether it is possible in rasa to perform cross validation when producing the tensorboard plots for DIETClassifier. In other words, calculating the average validation accuracy across x folds instead of calculating that on one fixed subset, as I assume is happening by default. The reason I am looking for that is to increase the validity of the model performance report in the tensorboard given number of epochs. Any ideas? Would the argument “max_cross_validation_folds”, which is used for the SklearnIntentClassifier in the documentation, help in this case?