diff --git a/use-cases/zebra2horse/dataloader.py b/use-cases/zebra2horse/dataloader.py
index 25d2914c..ff414495 100644
--- a/use-cases/zebra2horse/dataloader.py
+++ b/use-cases/zebra2horse/dataloader.py
@@ -44,25 +44,25 @@ def preproc_test_fn(img, label):
         # Apply the preprocessing operations to the training data
         train_horses = (
             train_horses.map(preproc_train_fn, num_parallel_calls=tf.data.AUTOTUNE)
-            .cache()
             .shuffle(self.buffer_size)
+            .cache()
         )
         train_zebras = (
             train_zebras.map(preproc_train_fn, num_parallel_calls=tf.data.AUTOTUNE)
-            .cache()
             .shuffle(self.buffer_size)
+            .cache()
         )
 
         # Apply the preprocessing operations to the test data
         test_horses = (
             test_horses.map(preproc_test_fn, num_parallel_calls=tf.data.AUTOTUNE)
-            .cache()
             .shuffle(self.buffer_size)
+            .cache()
         )
         test_zebras = (
             test_zebras.map(preproc_test_fn, num_parallel_calls=tf.data.AUTOTUNE)
-            .cache()
             .shuffle(self.buffer_size)
+            .cache()
         )
 
         return tf.data.Dataset.zip((train_horses, train_zebras)), tf.data.Dataset.zip((test_horses, test_zebras))