Fix imports for {text,image}_dataset_from_directory (#1181)

jbischof · web-flow · commit e40fab4a194a · 2022-12-31T11:30:19.000+01:00
* Fix typos in "Getting Started Guide"

* Fix import for preprocessing.{text,image}_dataset_from_directory

* Remove changes to "KerasNLP getting started" from other PR
diff --git a/examples/generative/dcgan_overriding_train_step.py b/examples/generative/dcgan_overriding_train_step.py
@@ -38,7 +38,7 @@
 Create a dataset from our folder, and rescale the images to the [0-1] range:
 """
 
-dataset = keras.preprocessing.image_dataset_from_directory(
+dataset = keras.utils.image_dataset_from_directory(
     "celeba_gan", label_mode=None, image_size=(64, 64), batch_size=32
 )
 dataset = dataset.map(lambda x: x / 255.0)
diff --git a/examples/generative/ipynb/dcgan_overriding_train_step.ipynb b/examples/generative/ipynb/dcgan_overriding_train_step.ipynb
@@ -87,7 +87,7 @@
    },
    "outputs": [],
    "source": [
-    "dataset = keras.preprocessing.image_dataset_from_directory(\n",
+    "dataset = keras.utils.image_dataset_from_directory(\n",
     "    \"celeba_gan\", label_mode=None, image_size=(64, 64), batch_size=32\n",
     ")\n",
     "dataset = dataset.map(lambda x: x / 255.0)\n",
diff --git a/examples/generative/ipynb/stylegan.ipynb b/examples/generative/ipynb/stylegan.ipynb
@@ -166,7 +166,7 @@
         "\n",
         "# Create a dataset from our folder, and rescale the images to the [0-1] range:\n",
         "\n",
-        "ds_train = keras.preprocessing.image_dataset_from_directory(\n",
+        "ds_train = keras.utils.image_dataset_from_directory(\n",
         "    \"celeba_gan\", label_mode=None, image_size=(64, 64), batch_size=32\n",
         ")\n",
         "ds_train = ds_train.map(lambda x: x / 255.0)\n",
diff --git a/examples/generative/md/dcgan_overriding_train_step.md b/examples/generative/md/dcgan_overriding_train_step.md
@@ -46,7 +46,7 @@ Create a dataset from our folder, and rescale the images to the [0-1] range:
 
 
 ```python
-dataset = keras.preprocessing.image_dataset_from_directory(
+dataset = keras.utils.image_dataset_from_directory(
     "celeba_gan", label_mode=None, image_size=(64, 64), batch_size=32
 )
 dataset = dataset.map(lambda x: x / 255.0)
diff --git a/examples/generative/md/stylegan.md b/examples/generative/md/stylegan.md
@@ -82,7 +82,7 @@ with ZipFile("celeba_gan/data.zip", "r") as zipobj:
 
 # Create a dataset from our folder, and rescale the images to the [0-1] range:
 
-ds_train = keras.preprocessing.image_dataset_from_directory(
+ds_train = keras.utils.image_dataset_from_directory(
     "celeba_gan", label_mode=None, image_size=(64, 64), batch_size=32
 )
 ds_train = ds_train.map(lambda x: x / 255.0)
diff --git a/examples/generative/stylegan.py b/examples/generative/stylegan.py
@@ -78,7 +78,7 @@ def log2(x):
 
 # Create a dataset from our folder, and rescale the images to the [0-1] range:
 
-ds_train = keras.preprocessing.image_dataset_from_directory(
+ds_train = keras.utils.image_dataset_from_directory(
     "celeba_gan", label_mode=None, image_size=(64, 64), batch_size=32
 )
 ds_train = ds_train.map(lambda x: x / 255.0)
diff --git a/examples/nlp/ipynb/text_classification_from_scratch.ipynb b/examples/nlp/ipynb/text_classification_from_scratch.ipynb
@@ -161,7 +161,7 @@
     "colab_type": "text"
    },
    "source": [
-    "You can use the utility `tf.keras.preprocessing.text_dataset_from_directory` to\n",
+    "You can use the utility `tf.keras.utils.text_dataset_from_directory` to\n",
     "generate a labeled `tf.data.Dataset` object from a set of text files on disk filed\n",
     " into class-specific folders.\n",
     "\n",
@@ -190,21 +190,21 @@
    "outputs": [],
    "source": [
     "batch_size = 32\n",
-    "raw_train_ds = tf.keras.preprocessing.text_dataset_from_directory(\n",
+    "raw_train_ds = tf.keras.utils.text_dataset_from_directory(\n",
     "    \"aclImdb/train\",\n",
     "    batch_size=batch_size,\n",
     "    validation_split=0.2,\n",
     "    subset=\"training\",\n",
     "    seed=1337,\n",
     ")\n",
-    "raw_val_ds = tf.keras.preprocessing.text_dataset_from_directory(\n",
+    "raw_val_ds = tf.keras.utils.text_dataset_from_directory(\n",
     "    \"aclImdb/train\",\n",
     "    batch_size=batch_size,\n",
     "    validation_split=0.2,\n",
     "    subset=\"validation\",\n",
     "    seed=1337,\n",
     ")\n",
-    "raw_test_ds = tf.keras.preprocessing.text_dataset_from_directory(\n",
+    "raw_test_ds = tf.keras.utils.text_dataset_from_directory(\n",
     "    \"aclImdb/test\", batch_size=batch_size\n",
     ")\n",
     "\n",
diff --git a/examples/nlp/md/text_classification_from_scratch.md b/examples/nlp/md/text_classification_from_scratch.md
@@ -91,7 +91,7 @@ We are only interested in the `pos` and `neg` subfolders, so let's delete the re
 !rm -r aclImdb/train/unsup
 ```
 
-You can use the utility `tf.keras.preprocessing.text_dataset_from_directory` to
+You can use the utility `tf.keras.utils.text_dataset_from_directory` to
 generate a labeled `tf.data.Dataset` object from a set of text files on disk filed
  into class-specific folders.
 
@@ -113,21 +113,21 @@ get have no overlap.
 
 ```python
 batch_size = 32
-raw_train_ds = tf.keras.preprocessing.text_dataset_from_directory(
+raw_train_ds = tf.keras.utils.text_dataset_from_directory(
     "aclImdb/train",
     batch_size=batch_size,
     validation_split=0.2,
     subset="training",
     seed=1337,
 )
-raw_val_ds = tf.keras.preprocessing.text_dataset_from_directory(
+raw_val_ds = tf.keras.utils.text_dataset_from_directory(
     "aclImdb/train",
     batch_size=batch_size,
     validation_split=0.2,
     subset="validation",
     seed=1337,
 )
-raw_test_ds = tf.keras.preprocessing.text_dataset_from_directory(
+raw_test_ds = tf.keras.utils.text_dataset_from_directory(
     "aclImdb/test", batch_size=batch_size
 )
 
diff --git a/examples/nlp/text_classification_from_scratch.py b/examples/nlp/text_classification_from_scratch.py
@@ -67,7 +67,7 @@
 """
 
 """
-You can use the utility `tf.keras.preprocessing.text_dataset_from_directory` to
+You can use the utility `tf.keras.utils.text_dataset_from_directory` to
 generate a labeled `tf.data.Dataset` object from a set of text files on disk filed
  into class-specific folders.
 
@@ -89,21 +89,21 @@
 """
 
 batch_size = 32
-raw_train_ds = tf.keras.preprocessing.text_dataset_from_directory(
+raw_train_ds = tf.keras.utils.text_dataset_from_directory(
     "aclImdb/train",
     batch_size=batch_size,
     validation_split=0.2,
     subset="training",
     seed=1337,
 )
-raw_val_ds = tf.keras.preprocessing.text_dataset_from_directory(
+raw_val_ds = tf.keras.utils.text_dataset_from_directory(
     "aclImdb/train",
     batch_size=batch_size,
     validation_split=0.2,
     subset="validation",
     seed=1337,
 )
-raw_test_ds = tf.keras.preprocessing.text_dataset_from_directory(
+raw_test_ds = tf.keras.utils.text_dataset_from_directory(
     "aclImdb/test", batch_size=batch_size
 )
 
diff --git a/examples/vision/image_classification_from_scratch.py b/examples/vision/image_classification_from_scratch.py
@@ -87,7 +87,7 @@
 image_size = (180, 180)
 batch_size = 128
 
-train_ds, val_ds = tf.keras.preprocessing.image_dataset_from_directory(
+train_ds, val_ds = tf.keras.utils.image_dataset_from_directory(
     "PetImages",
     validation_split=0.2,
     subset="both",
diff --git a/examples/vision/ipynb/image_classification_from_scratch.ipynb b/examples/vision/ipynb/image_classification_from_scratch.ipynb
@@ -172,7 +172,7 @@
     "image_size = (180, 180)\n",
     "batch_size = 128\n",
     "\n",
-    "train_ds, val_ds = tf.keras.preprocessing.image_dataset_from_directory(\n",
+    "train_ds, val_ds = tf.keras.utils.image_dataset_from_directory(\n",
     "    \"PetImages\",\n",
     "    validation_split=0.2,\n",
     "    subset=\"both\",\n",
diff --git a/examples/vision/md/image_classification_from_scratch.md b/examples/vision/md/image_classification_from_scratch.md
@@ -116,7 +116,7 @@ Deleted 1590 images
 image_size = (180, 180)
 batch_size = 128
 
-train_ds, val_ds = tf.keras.preprocessing.image_dataset_from_directory(
+train_ds, val_ds = tf.keras.utils.image_dataset_from_directory(
     "PetImages",
     validation_split=0.2,
     subset="both",
diff --git a/guides/intro_to_keras_for_engineers.py b/guides/intro_to_keras_for_engineers.py
@@ -88,9 +88,9 @@
 
 Keras features a range of utilities to help you turn raw data on disk into a `Dataset`:
 
-- `tf.keras.preprocessing.image_dataset_from_directory` turns image files sorted into
+- `tf.keras.utils.image_dataset_from_directory` turns image files sorted into
  class-specific folders into a labeled dataset of image tensors.
-- `tf.keras.preprocessing.text_dataset_from_directory` does the same for text files.
+- `tf.keras.utils.text_dataset_from_directory` does the same for text files.
 
 In addition, the TensorFlow `tf.data` includes other similar utilities, such as
 `tf.data.experimental.make_csv_dataset` to load structured data from CSV files.
@@ -113,7 +113,7 @@
 
 ```python
 # Create a dataset.
-dataset = keras.preprocessing.image_dataset_from_directory(
+dataset = keras.utils.image_dataset_from_directory(
   'path/to/main_directory', batch_size=64, image_size=(200, 200))
 
 # For demonstration, iterate over the batches yielded by the dataset.
@@ -135,7 +135,7 @@
  you can do:
 
 ```python
-dataset = keras.preprocessing.text_dataset_from_directory(
+dataset = keras.utils.text_dataset_from_directory(
   'path/to/main_directory', batch_size=64)
 
 # For demonstration, iterate over the batches yielded by the dataset.
diff --git a/guides/ipynb/intro_to_keras_for_engineers.ipynb b/guides/ipynb/intro_to_keras_for_engineers.ipynb
@@ -118,9 +118,9 @@
     "\n",
     "Keras features a range of utilities to help you turn raw data on disk into a `Dataset`:\n",
     "\n",
-    "- `tf.keras.preprocessing.image_dataset_from_directory` turns image files sorted into\n",
+    "- `tf.keras.utils.image_dataset_from_directory` turns image files sorted into\n",
     " class-specific folders into a labeled dataset of image tensors.\n",
-    "- `tf.keras.preprocessing.text_dataset_from_directory` does the same for text files.\n",
+    "- `tf.keras.utils.text_dataset_from_directory` does the same for text files.\n",
     "\n",
     "In addition, the TensorFlow `tf.data` includes other similar utilities, such as\n",
     "`tf.data.experimental.make_csv_dataset` to load structured data from CSV files.\n",
@@ -143,7 +143,7 @@
     "\n",
     "```python\n",
     "# Create a dataset.\n",
-    "dataset = keras.preprocessing.image_dataset_from_directory(\n",
+    "dataset = keras.utils.image_dataset_from_directory(\n",
     "  'path/to/main_directory', batch_size=64, image_size=(200, 200))\n",
     "\n",
     "# For demonstration, iterate over the batches yielded by the dataset.\n",
@@ -165,7 +165,7 @@
     " you can do:\n",
     "\n",
     "```python\n",
-    "dataset = keras.preprocessing.text_dataset_from_directory(\n",
+    "dataset = keras.utils.text_dataset_from_directory(\n",
     "  'path/to/main_directory', batch_size=64)\n",
     "\n",
     "# For demonstration, iterate over the batches yielded by the dataset.\n",
diff --git a/guides/ipynb/keras_nlp/getting_started.ipynb b/guides/ipynb/keras_nlp/getting_started.ipynb
@@ -169,11 +169,11 @@
    "outputs": [],
    "source": [
     "BATCH_SIZE = 16\n",
-    "imdb_train = tf.keras.preprocessing.text_dataset_from_directory(\n",
+    "imdb_train = tf.keras.utils.text_dataset_from_directory(\n",
     "    \"aclImdb/train\",\n",
     "    batch_size=BATCH_SIZE,\n",
     ")\n",
-    "imdb_test = tf.keras.preprocessing.text_dataset_from_directory(\n",
+    "imdb_test = tf.keras.utils.text_dataset_from_directory(\n",
     "    \"aclImdb/test\",\n",
     "    batch_size=BATCH_SIZE,\n",
     ")\n",
diff --git a/guides/ipynb/transfer_learning.ipynb b/guides/ipynb/transfer_learning.ipynb
@@ -506,7 +506,7 @@
     "\n",
     "First, let's fetch the cats vs. dogs dataset using TFDS. If you have your own dataset,\n",
     "you'll probably want to use the utility\n",
-    "`tf.keras.preprocessing.image_dataset_from_directory` to generate similar labeled\n",
+    "`tf.keras.utils.image_dataset_from_directory` to generate similar labeled\n",
     " dataset objects from a set of images on disk filed into class-specific folders.\n",
     "\n",
     "Transfer learning is most useful when working with very small datasets. To keep our\n",
diff --git a/guides/keras_nlp/getting_started.py b/guides/keras_nlp/getting_started.py
@@ -115,11 +115,11 @@
 """
 
 BATCH_SIZE = 16
-imdb_train = tf.keras.preprocessing.text_dataset_from_directory(
+imdb_train = tf.keras.utils.text_dataset_from_directory(
     "aclImdb/train",
     batch_size=BATCH_SIZE,
 )
-imdb_test = tf.keras.preprocessing.text_dataset_from_directory(
+imdb_test = tf.keras.utils.text_dataset_from_directory(
     "aclImdb/test",
     batch_size=BATCH_SIZE,
 )
diff --git a/guides/md/intro_to_keras_for_engineers.md b/guides/md/intro_to_keras_for_engineers.md
@@ -92,9 +92,9 @@ using `Dataset` objects, since they will take care of performance-critical detai
 
 Keras features a range of utilities to help you turn raw data on disk into a `Dataset`:
 
-- `tf.keras.preprocessing.image_dataset_from_directory` turns image files sorted into
+- `tf.keras.utils.image_dataset_from_directory` turns image files sorted into
  class-specific folders into a labeled dataset of image tensors.
-- `tf.keras.preprocessing.text_dataset_from_directory` does the same for text files.
+- `tf.keras.utils.text_dataset_from_directory` does the same for text files.
 
 In addition, the TensorFlow `tf.data` includes other similar utilities, such as
 `tf.data.experimental.make_csv_dataset` to load structured data from CSV files.
@@ -117,7 +117,7 @@ Then you can do:
 
 ```python
 # Create a dataset.
-dataset = keras.preprocessing.image_dataset_from_directory(
+dataset = keras.utils.image_dataset_from_directory(
   'path/to/main_directory', batch_size=64, image_size=(200, 200))
 
 # For demonstration, iterate over the batches yielded by the dataset.
@@ -139,7 +139,7 @@ Likewise for text: if you have `.txt` documents sorted by class in different fol
  you can do:
 
 ```python
-dataset = keras.preprocessing.text_dataset_from_directory(
+dataset = keras.utils.text_dataset_from_directory(
   'path/to/main_directory', batch_size=64)
 
 # For demonstration, iterate over the batches yielded by the dataset.
diff --git a/guides/md/keras_nlp/getting_started.md b/guides/md/keras_nlp/getting_started.md
@@ -127,11 +127,11 @@ powerful `tf.data.Dataset` format for examples.
 
 ```python
 BATCH_SIZE = 16
-imdb_train = tf.keras.preprocessing.text_dataset_from_directory(
+imdb_train = tf.keras.utils.text_dataset_from_directory(
     "aclImdb/train",
     batch_size=BATCH_SIZE,
 )
-imdb_test = tf.keras.preprocessing.text_dataset_from_directory(
+imdb_test = tf.keras.utils.text_dataset_from_directory(
     "aclImdb/test",
     batch_size=BATCH_SIZE,
 )
diff --git a/guides/md/transfer_learning.md b/guides/md/transfer_learning.md
@@ -407,7 +407,7 @@ learning & fine-tuning example. We will load the Xception model, pre-trained on
 
 First, let's fetch the cats vs. dogs dataset using TFDS. If you have your own dataset,
 you'll probably want to use the utility
-`tf.keras.preprocessing.image_dataset_from_directory` to generate similar labeled
+`tf.keras.utils.image_dataset_from_directory` to generate similar labeled
  dataset objects from a set of images on disk filed into class-specific folders.
 
 Transfer learning is most useful when working with very small datasets. To keep our
diff --git a/guides/transfer_learning.py b/guides/transfer_learning.py
@@ -381,7 +381,7 @@
 
 First, let's fetch the cats vs. dogs dataset using TFDS. If you have your own dataset,
 you'll probably want to use the utility
-`tf.keras.preprocessing.image_dataset_from_directory` to generate similar labeled
+`tf.keras.utils.image_dataset_from_directory` to generate similar labeled
  dataset objects from a set of images on disk filed into class-specific folders.
 
 Transfer learning is most useful when working with very small datasets. To keep our
diff --git a/templates/getting_started/faq.md b/templates/getting_started/faq.md
@@ -571,9 +571,9 @@ On the other hand, the testing loss for an epoch is computed using the model as
 You should use the [`tf.data` API](https://www.tensorflow.org/guide/data) to create `tf.data.Dataset` objects -- an abstraction over a data pipeline
 that can pull data from local disk, from a distributed file system, from GCS, etc., as well as efficiently apply various data transformations.
 
-For instance, the utility [`tf.keras.preprocessing.image_dataset_from_directory`](https://keras.io/api/preprocessing/image/#imagedatasetfromdirectory-function)
+For instance, the utility [`tf.keras.utils.image_dataset_from_directory`](https://keras.io/api/data_loading/image/)
 will create a dataset that reads image data from a local directory.
-Likewise, the utility [`tf.keras.preprocessing.text_dataset_from_directory`](https://keras.io/api/preprocessing/text/#textdatasetfromdirectory-function)
+Likewise, the utility [`tf.keras.utils.text_dataset_from_directory`](https://keras.io/api/data_loading/text/#text_dataset_from_directory-function)
 will create a dataset that reads text files from a local directory.
 
 Dataset objects can be directly passed to `fit()`, or can be iterated over in a custom low-level training loop.
diff --git a/templates/getting_started/intro_to_keras_for_engineers.md b/templates/getting_started/intro_to_keras_for_engineers.md
@@ -92,9 +92,9 @@ using `Dataset` objects, since they will take care of performance-critical detai
 
 Keras features a range of utilities to help you turn raw data on disk into a `Dataset`:
 
-- `tf.keras.preprocessing.image_dataset_from_directory` turns image files sorted into
+- `tf.keras.utils.image_dataset_from_directory` turns image files sorted into
  class-specific folders into a labeled dataset of image tensors.
-- `tf.keras.preprocessing.text_dataset_from_directory` does the same for text files.
+- `tf.keras.utils.text_dataset_from_directory` does the same for text files.
 
 In addition, the TensorFlow `tf.data` includes other similar utilities, such as
 `tf.data.experimental.make_csv_dataset` to load structured data from CSV files.
@@ -117,7 +117,7 @@ Then you can do:
 
 ```python
 # Create a dataset.
-dataset = keras.preprocessing.image_dataset_from_directory(
+dataset = keras.utils.image_dataset_from_directory(
   'path/to/main_directory', batch_size=64, image_size=(200, 200))
 
 # For demonstration, iterate over the batches yielded by the dataset.
@@ -139,7 +139,7 @@ Likewise for text: if you have `.txt` documents sorted by class in different fol
  you can do:
 
 ```python
-dataset = keras.preprocessing.text_dataset_from_directory(
+dataset = keras.utils.text_dataset_from_directory(
   'path/to/main_directory', batch_size=64)
 
 # For demonstration, iterate over the batches yielded by the dataset.

Original file line number	Diff line number	Diff line change
`@@ -38,7 +38,7 @@`
`38`	`38`	`Create a dataset from our folder, and rescale the images to the [0-1] range:`
`39`	`39`	`"""`
`40`	`40`
`41`		`-dataset = keras.preprocessing.image_dataset_from_directory(`
	`41`	`+dataset = keras.utils.image_dataset_from_directory(`
`42`	`42`	`"celeba_gan", label_mode=None, image_size=(64, 64), batch_size=32`
`43`	`43`	`)`
`44`	`44`	`dataset = dataset.map(lambda x: x / 255.0)`
Original file line number	Diff line number	Diff line change
`@@ -46,7 +46,7 @@ Create a dataset from our folder, and rescale the images to the [0-1] range:`
`46`	`46`
`47`	`47`
`48`	`48`	```python
`49`		`-dataset = keras.preprocessing.image_dataset_from_directory(`
	`49`	`+dataset = keras.utils.image_dataset_from_directory(`
`50`	`50`	`"celeba_gan", label_mode=None, image_size=(64, 64), batch_size=32`
`51`	`51`	`)`
`52`	`52`	`dataset = dataset.map(lambda x: x / 255.0)`
Original file line number	Diff line number	Diff line change
`@@ -82,7 +82,7 @@ with ZipFile("celeba_gan/data.zip", "r") as zipobj:`
`82`	`82`
`83`	`83`	`# Create a dataset from our folder, and rescale the images to the [0-1] range:`
`84`	`84`
`85`		`-ds_train = keras.preprocessing.image_dataset_from_directory(`
	`85`	`+ds_train = keras.utils.image_dataset_from_directory(`
`86`	`86`	`"celeba_gan", label_mode=None, image_size=(64, 64), batch_size=32`
`87`	`87`	`)`
`88`	`88`	`ds_train = ds_train.map(lambda x: x / 255.0)`
Original file line number	Diff line number	Diff line change
`@@ -78,7 +78,7 @@ def log2(x):`
`78`	`78`
`79`	`79`	`# Create a dataset from our folder, and rescale the images to the [0-1] range:`
`80`	`80`
`81`		`-ds_train = keras.preprocessing.image_dataset_from_directory(`
	`81`	`+ds_train = keras.utils.image_dataset_from_directory(`
`82`	`82`	`"celeba_gan", label_mode=None, image_size=(64, 64), batch_size=32`
`83`	`83`	`)`
`84`	`84`	`ds_train = ds_train.map(lambda x: x / 255.0)`