Applying ClassificationInterpretation methods on the training set

[lesscomfortable (Francisco Ingham)]

@bartp88

To do that, it seems we need a way to apply the top_losses() method to the training set.

You can find a workaround for this here