Added dataset_shuffle and removed commented out os.rm commands

vinayak19th · vinayak19th · commit f2a9792452fc · 2025-10-30T22:57:33.000-07:00
diff --git a/examples/vision/nl_image_search.py b/examples/vision/nl_image_search.py
@@ -84,7 +84,6 @@
     )
     with zipfile.ZipFile(annotation_zip, "r") as zip_ref:
         zip_ref.extractall(annotations_dir)
-    # os.remove(os.path.join(root_dir,"captions.zip"))
 
 print("\nDownloading the images.")
 
@@ -93,7 +92,6 @@
     print("Downloaded the images.\nunzipping")
     with zipfile.ZipFile(image_zip, "r") as zip_ref:
         zip_ref.extractall(images_dir)
-    # os.remove(os.path.join(root_dir,"train2014.zip"))
 
 print("\nDataset is downloaded and extracted successfully.")
 
@@ -251,6 +249,7 @@ def get_dataset(file_pattern, batch_size):
             num_parallel_calls=tf.data.AUTOTUNE,
             deterministic=False,
         )
+        .shuffle(batch_size * 10)
         .batch(batch_size)
         .prefetch(tf.data.AUTOTUNE)
     )

Original file line number	Diff line number	Diff line change
`@@ -84,7 +84,6 @@`
`84`	`84`	`)`
`85`	`85`	`with zipfile.ZipFile(annotation_zip, "r") as zip_ref:`
`86`	`86`	`zip_ref.extractall(annotations_dir)`
`87`		`- # os.remove(os.path.join(root_dir,"captions.zip"))`
`88`	`87`
`89`	`88`	`print("\nDownloading the images.")`
`90`	`89`
`@@ -93,7 +92,6 @@`
`93`	`92`	`print("Downloaded the images.\nunzipping")`
`94`	`93`	`with zipfile.ZipFile(image_zip, "r") as zip_ref:`
`95`	`94`	`zip_ref.extractall(images_dir)`
`96`		`- # os.remove(os.path.join(root_dir,"train2014.zip"))`
`97`	`95`
`98`	`96`	`print("\nDataset is downloaded and extracted successfully.")`
`99`	`97`
`@@ -251,6 +249,7 @@ def get_dataset(file_pattern, batch_size):`
`251`	`249`	`num_parallel_calls=tf.data.AUTOTUNE,`
`252`	`250`	`deterministic=False,`
`253`	`251`	`)`
	`252`	`+ .shuffle(batch_size * 10)`
`254`	`253`	`.batch(batch_size)`
`255`	`254`	`.prefetch(tf.data.AUTOTUNE)`
`256`	`255`	`)`