Batch transforms are operations that are performed on trial tensors after being accumulated into batches via the :meth:`__call__` method. Ideally this is implemented with pytorch operations for ease of execution graph integration.
(self, only_trial_data=True)