Update guides for Keras V3 Saving Compatibility

guides/customizing_what_happens_in_fit.py

@@ -45,7 +45,7 @@
 """
 
 import tensorflow as tf
-from tensorflow import keras
+import keras
 
 """
 ## A first simple example

guides/distributed_training.py

@@ -45,7 +45,7 @@
 """
 
 import tensorflow as tf
-from tensorflow import keras
+import keras
 
 """
 ## Single-host, multi-device synchronous training

guides/functional_api.py

@@ -2,7 +2,7 @@
 Title: The Functional API
 Author: [fchollet](https://twitter.com/fchollet)
 Date created: 2019/03/01
-Last modified: 2020/04/12
+Last modified: 2020/07/10
 Description: Complete guide to the functional API.
 Accelerator: GPU
 """
@@ -12,8 +12,8 @@
 
 import numpy as np
 import tensorflow as tf
-from tensorflow import keras
-from tensorflow.keras import layers
+import keras
+from keras import layers
 
 """
 ## Introduction
@@ -607,6 +607,7 @@ def call(self, inputs):
 """
 
 
+@keras.saving.register_keras_serializable()
 class CustomDense(layers.Layer):
     def __init__(self, units=32):
         super().__init__()
@@ -635,7 +636,7 @@ def get_config(self):
 model = keras.Model(inputs, outputs)
 config = model.get_config()
 
-new_model = keras.Model.from_config(config, custom_objects={"CustomDense": CustomDense})
+new_model = keras.Model.from_config(config)
 
 """
 Optionally, implement the class method `from_config(cls, config)` which is used
@@ -776,6 +777,7 @@ def call(self, inputs):
 model = keras.Model(inputs, outputs)
 
 
+@keras.saving.register_keras_serializable()
 class CustomRNN(layers.Layer):
     def __init__(self):
         super().__init__()
@@ -830,6 +832,7 @@ def call(self, inputs):
 batch_size = 16
 
 
+@keras.saving.register_keras_serializable()
 class CustomRNN(layers.Layer):
     def __init__(self):
         super().__init__()

guides/intro_to_keras_for_engineers.py

@@ -13,7 +13,7 @@
 
 import numpy as np
 import tensorflow as tf
-from tensorflow import keras
+import keras
 
 """
 ## Introduction

guides/intro_to_keras_for_researchers.py

@@ -11,7 +11,7 @@
 """
 
 import tensorflow as tf
-from tensorflow import keras
+import keras
 
 """
 ## Introduction

guides/ipynb/customizing_what_happens_in_fit.ipynb

@@ -70,7 +70,7 @@
    "outputs": [],
    "source": [
     "import tensorflow as tf\n",
-    "from tensorflow import keras"
+    "import keras"
    ]
   },
   {

guides/ipynb/distributed_training.ipynb

@@ -70,7 +70,7 @@
    "outputs": [],
    "source": [
     "import tensorflow as tf\n",
-    "from tensorflow import keras"
+    "import keras"
    ]
   },
   {

guides/ipynb/functional_api.ipynb

@@ -10,7 +10,7 @@
     "\n",
     "**Author:** [fchollet](https://twitter.com/fchollet)<br>\n",
     "**Date created:** 2019/03/01<br>\n",
-    "**Last modified:** 2020/04/12<br>\n",
+    "**Last modified:** 2020/07/10<br>\n",
     "**Description:** Complete guide to the functional API."
    ]
   },
@@ -33,8 +33,8 @@
    "source": [
     "import numpy as np\n",
     "import tensorflow as tf\n",
-    "from tensorflow import keras\n",
-    "from tensorflow.keras import layers"
+    "import keras\n",
+    "from keras import layers"
    ]
   },
   {
@@ -1064,6 +1064,7 @@
    "outputs": [],
    "source": [
     "\n",
+    "@keras.saving.register_keras_serializable()\n",
     "class CustomDense(layers.Layer):\n",
     "    def __init__(self, units=32):\n",
     "        super().__init__()\n",
@@ -1092,7 +1093,7 @@
     "model = keras.Model(inputs, outputs)\n",
     "config = model.get_config()\n",
     "\n",
-    "new_model = keras.Model.from_config(config, custom_objects={\"CustomDense\": CustomDense})"
+    "new_model = keras.Model.from_config(config)"
    ]
   },
   {
@@ -1256,6 +1257,7 @@
     "model = keras.Model(inputs, outputs)\n",
     "\n",
     "\n",
+    "@keras.saving.register_keras_serializable()\n",
     "class CustomRNN(layers.Layer):\n",
     "    def __init__(self):\n",
     "        super().__init__()\n",
@@ -1324,6 +1326,7 @@
     "batch_size = 16\n",
     "\n",
     "\n",
+    "@keras.saving.register_keras_serializable()\n",
     "class CustomRNN(layers.Layer):\n",
     "    def __init__(self):\n",
     "        super().__init__()\n",

guides/ipynb/intro_to_keras_for_engineers.ipynb

@@ -33,7 +33,7 @@
    "source": [
     "import numpy as np\n",
     "import tensorflow as tf\n",
-    "from tensorflow import keras"
+    "import keras"
    ]
   },
   {

guides/ipynb/intro_to_keras_for_researchers.ipynb

@@ -32,7 +32,7 @@
    "outputs": [],
    "source": [
     "import tensorflow as tf\n",
-    "from tensorflow import keras"
+    "import keras"
    ]
   },
   {

guides/ipynb/making_new_layers_and_models_via_subclassing.ipynb

@@ -10,7 +10,7 @@
     "\n",
     "**Author:** [fchollet](https://twitter.com/fchollet)<br>\n",
     "**Date created:** 2019/03/01<br>\n",
-    "**Last modified:** 2023/03/16<br>\n",
+    "**Last modified:** 2023/07/10<br>\n",
     "**Description:** Complete guide to writing `Layer` and `Model` objects from scratch."
    ]
   },
@@ -32,7 +32,7 @@
    "outputs": [],
    "source": [
     "import tensorflow as tf\n",
-    "from tensorflow import keras"
+    "import keras"
    ]
   },
   {
@@ -820,9 +820,10 @@
    },
    "outputs": [],
    "source": [
-    "from tensorflow.keras import layers\n",
+    "from keras import layers\n",
     "\n",
     "\n",
+    "@keras.saving.register_keras_serializable()\n",
     "class Sampling(layers.Layer):\n",
     "    \"\"\"Uses (z_mean, z_log_var) to sample z, the vector encoding a digit.\"\"\"\n",
     "\n",
@@ -834,6 +835,7 @@
     "        return z_mean + tf.exp(0.5 * z_log_var) * epsilon\n",
     "\n",
     "\n",
+    "@keras.saving.register_keras_serializable()\n",
     "class Encoder(layers.Layer):\n",
     "    \"\"\"Maps MNIST digits to a triplet (z_mean, z_log_var, z).\"\"\"\n",
     "\n",
@@ -852,6 +854,7 @@
     "        return z_mean, z_log_var, z\n",
     "\n",
     "\n",
+    "@keras.saving.register_keras_serializable()\n",
     "class Decoder(layers.Layer):\n",
     "    \"\"\"Converts z, the encoded digit vector, back into a readable digit.\"\"\"\n",
     "\n",
@@ -865,6 +868,7 @@
     "        return self.dense_output(x)\n",
     "\n",
     "\n",
+    "@keras.saving.register_keras_serializable()\n",
     "class VariationalAutoEncoder(keras.Model):\n",
     "    \"\"\"Combines the encoder and decoder into an end-to-end model for training.\"\"\"\n",
     "\n",

guides/ipynb/preprocessing_layers.ipynb

@@ -123,7 +123,8 @@
    "source": [
     "import numpy as np\n",
     "import tensorflow as tf\n",
-    "from tensorflow.keras import layers\n",
+    "import keras\n",
+    "from keras import layers\n",
     "\n",
     "data = np.array(\n",
     "    [\n",

guides/ipynb/sequential_model.ipynb

@@ -32,8 +32,8 @@
    "outputs": [],
    "source": [
     "import tensorflow as tf\n",
-    "from tensorflow import keras\n",
-    "from tensorflow.keras import layers"
+    "import keras\n",
+    "from keras import layers"
    ]
   },
   {

guides/ipynb/training_with_built_in_methods.ipynb

@@ -32,8 +32,8 @@
    "outputs": [],
    "source": [
     "import tensorflow as tf\n",
-    "from tensorflow import keras\n",
-    "from tensorflow.keras import layers"
+    "import keras\n",
+    "from keras import layers"
    ]
   },
   {
@@ -454,6 +454,7 @@
    "outputs": [],
    "source": [
     "\n",
+    "@keras.saving.register_keras_serializable()\n",
     "class CustomMSE(keras.losses.Loss):\n",
     "    def __init__(self, regularization_factor=0.1, name=\"custom_mse\"):\n",
     "        super().__init__(name=name)\n",
@@ -464,6 +465,12 @@
     "        reg = tf.math.reduce_mean(tf.square(0.5 - y_pred), axis=-1)\n",
     "        return mse + reg * self.regularization_factor\n",
     "\n",
+    "    def get_config(self):\n",
+    "        return {\n",
+    "            \"regularization_factor\": self.regularization_factor,\n",
+    "            \"name\": self.name,\n",
+    "        }\n",
+    "\n",
     "\n",
     "model = get_uncompiled_model()\n",
     "model.compile(optimizer=keras.optimizers.Adam(), loss=CustomMSE())\n",
@@ -507,6 +514,7 @@
    "outputs": [],
    "source": [
     "\n",
+    "@keras.saving.register_keras_serializable()\n",
     "class CategoricalTruePositives(keras.metrics.Metric):\n",
     "    def __init__(self, name=\"categorical_true_positives\", **kwargs):\n",
     "        super().__init__(name=name, **kwargs)\n",
@@ -567,6 +575,7 @@
    "outputs": [],
    "source": [
     "\n",
+    "@keras.saving.register_keras_serializable()\n",
     "class ActivityRegularizationLayer(layers.Layer):\n",
     "    def call(self, inputs):\n",
     "        self.add_loss(tf.reduce_sum(inputs) * 0.1)\n",
@@ -615,6 +624,7 @@
    "outputs": [],
    "source": [
     "\n",
+    "@keras.saving.register_keras_serializable()\n",
     "class LogisticEndpoint(keras.layers.Layer):\n",
     "    def __init__(self, name=None):\n",
     "        super().__init__(name=name)\n",
@@ -1550,7 +1560,7 @@
     "        # the current checkpoint if and only if\n",
     "        # the `val_loss` score has improved.\n",
     "        # The saved model name will include the current epoch.\n",
-    "        filepath=\"mymodel_{epoch}.keras\",\n",
+    "        filepath=\"mymodel_{epoch}\",\n",
     "        save_best_only=True,  # Only save a model if `val_loss` has improved.\n",
     "        monitor=\"val_loss\",\n",
     "        verbose=1,\n",
@@ -1605,7 +1615,7 @@
     "    # This callback saves the model every 100 batches.\n",
     "    # We include the training loss in the saved model name.\n",
     "    keras.callbacks.ModelCheckpoint(\n",
-    "        filepath=checkpoint_dir + \"/model-loss={loss:.2f}.keras\", save_freq=100\n",
+    "        filepath=checkpoint_dir + \"/model-loss={loss:.2f}\", save_freq=100\n",
     "    )\n",
     "]\n",
     "model.fit(x_train, y_train, epochs=1, callbacks=callbacks)"

guides/ipynb/transfer_learning.ipynb

@@ -33,7 +33,7 @@
    "source": [
     "import numpy as np\n",
     "import tensorflow as tf\n",
-    "from tensorflow import keras"
+    "import keras"
    ]
   },
   {