separate train and test datasets for sinkhorn eval

 logger.info(f"Creating data loader for the dataset")
        # FIXME separate train and test datasets
        self.train_dataloader = DataLoader(dataset=dataset, batch_size=batch_size, shuffle=True)
        self.test_dataloader = DataLoader(dataset=dataset, batch_size=batch_size, shuffle=True)
        if self.debug_flag:
            self.__validate_dataset()  # FIXME remove later

mbaddar1 / genmodel

separate train and test datasets for sinkhorn eval #23