MultiLayerNetwork has a clone() method

but this method always returns an object with a score of 0.0, not the score of the object the method was called on

this means that if EarlyStoppingTrainer is used with the default InMemoryModelSaver

InMemoryModelSaver.saveBestModel uses clone(), and then when EarlyStoppingTrainer is finished it gets the model with getBestModel()

this means that EarlyStoppingResult.getBestModelScore() always return 0.0