serialization_and_saving.ipynb

Kernel: Python 3

Save, serialize, and export models

Authors: Neel Kovelamudi, Francois Chollet
Date created: 2023/06/14
Last modified: 2023/06/30
Description: Complete guide to saving, serializing, and exporting models.

Introduction

A Keras model consists of multiple components:

• The architecture, or configuration, which specifies what layers the model contain, and how they're connected.

• A set of weights values (the "state of the model").

• An optimizer (defined by compiling the model).

• A set of losses and metrics (defined by compiling the model).

The Keras API saves all of these pieces together in a unified format, marked by the .keras extension. This is a zip archive consisting of the following:

• A JSON-based configuration file (config.json): Records of model, layer, and other trackables' configuration.

• A H5-based state file, such as model.weights.h5 (for the whole model), with directory keys for layers and their weights.

• A metadata file in JSON, storing things such as the current Keras version.

Let's take a look at how this works.

How to save and load a model

If you only have 10 seconds to read this guide, here's what you need to know.

Saving a Keras model:

model = ...  # Get model (Sequential, Functional Model, or Model subclass)
model.save('path/to/location.keras')  # The file needs to end with the .keras extension

Loading the model back:

model = keras.models.load_model('path/to/location.keras')

Now, let's look at the details.

Setup

import numpy as np
import keras
from keras import ops

Saving

This section is about saving an entire model to a single file. The file will include:

• The model's architecture/config

• The model's weight values (which were learned during training)

• The model's compilation information (if compile() was called)

• The optimizer and its state, if any (this enables you to restart training where you left)

APIs

You can save a model with model.save() or keras.models.save_model() (which is equivalent). You can load it back with keras.models.load_model().

The only supported format in Keras 3 is the "Keras v3" format, which uses the .keras extension.

Example:

def get_model():
    # Create a simple model.
    inputs = keras.Input(shape=(32,))
    outputs = keras.layers.Dense(1)(inputs)
    model = keras.Model(inputs, outputs)
    model.compile(optimizer=keras.optimizers.Adam(), loss="mean_squared_error")
    return model


model = get_model()

# Train the model.
test_input = np.random.random((128, 32))
test_target = np.random.random((128, 1))
model.fit(test_input, test_target)

# Calling `save('my_model.keras')` creates a zip archive `my_model.keras`.
model.save("my_model.keras")

# It can be used to reconstruct the model identically.
reconstructed_model = keras.models.load_model("my_model.keras")

# Let's check:
np.testing.assert_allclose(
    model.predict(test_input), reconstructed_model.predict(test_input)
)

Custom objects

This section covers the basic workflows for handling custom layers, functions, and models in Keras saving and reloading.

When saving a model that includes custom objects, such as a subclassed Layer, you must define a get_config() method on the object class. If the arguments passed to the constructor (__init__() method) of the custom object aren't Python objects (anything other than base types like ints, strings, etc.), then you must also explicitly deserialize these arguments in the from_config() class method.

Like this:

class CustomLayer(keras.layers.Layer):
    def __init__(self, sublayer, **kwargs):
        super().__init__(**kwargs)
        self.sublayer = sublayer

    def call(self, x):
        return self.sublayer(x)

    def get_config(self):
        base_config = super().get_config()
        config = {
            "sublayer": keras.saving.serialize_keras_object(self.sublayer),
        }
        return {**base_config, **config}

    @classmethod
    def from_config(cls, config):
        sublayer_config = config.pop("sublayer")
        sublayer = keras.saving.deserialize_keras_object(sublayer_config)
        return cls(sublayer, **config)

Please see the Defining the config methods section for more details and examples.

The saved .keras file is lightweight and does not store the Python code for custom objects. Therefore, to reload the model, load_model requires access to the definition of any custom objects used through one of the following methods:

1. Registering custom objects (preferred),

2. Passing custom objects directly when loading, or

3. Using a custom object scope

Below are examples of each workflow:

Registering custom objects (preferred)

This is the preferred method, as custom object registration greatly simplifies saving and loading code. Adding the @keras.saving.register_keras_serializable decorator to the class definition of a custom object registers the object globally in a master list, allowing Keras to recognize the object when loading the model.

Let's create a custom model involving both a custom layer and a custom activation function to demonstrate this.

Example:

# Clear all previously registered custom objects
keras.saving.get_custom_objects().clear()


# Upon registration, you can optionally specify a package or a name.
# If left blank, the package defaults to `Custom` and the name defaults to
# the class name.
@keras.saving.register_keras_serializable(package="MyLayers")
class CustomLayer(keras.layers.Layer):
    def __init__(self, factor):
        super().__init__()
        self.factor = factor

    def call(self, x):
        return x * self.factor

    def get_config(self):
        return {"factor": self.factor}


@keras.saving.register_keras_serializable(package="my_package", name="custom_fn")
def custom_fn(x):
    return x**2


# Create the model.
def get_model():
    inputs = keras.Input(shape=(4,))
    mid = CustomLayer(0.5)(inputs)
    outputs = keras.layers.Dense(1, activation=custom_fn)(mid)
    model = keras.Model(inputs, outputs)
    model.compile(optimizer="rmsprop", loss="mean_squared_error")
    return model


# Train the model.
def train_model(model):
    input = np.random.random((4, 4))
    target = np.random.random((4, 1))
    model.fit(input, target)
    return model


test_input = np.random.random((4, 4))
test_target = np.random.random((4, 1))

model = get_model()
model = train_model(model)
model.save("custom_model.keras")

# Now, we can simply load without worrying about our custom objects.
reconstructed_model = keras.models.load_model("custom_model.keras")

# Let's check:
np.testing.assert_allclose(
    model.predict(test_input), reconstructed_model.predict(test_input)
)

Passing custom objects to load_model()

model = get_model()
model = train_model(model)

# Calling `save('my_model.keras')` creates a zip archive `my_model.keras`.
model.save("custom_model.keras")

# Upon loading, pass a dict containing the custom objects used in the
# `custom_objects` argument of `keras.models.load_model()`.
reconstructed_model = keras.models.load_model(
    "custom_model.keras",
    custom_objects={"CustomLayer": CustomLayer, "custom_fn": custom_fn},
)

# Let's check:
np.testing.assert_allclose(
    model.predict(test_input), reconstructed_model.predict(test_input)
)

Using a custom object scope

Any code within the custom object scope will be able to recognize the custom objects passed to the scope argument. Therefore, loading the model within the scope will allow the loading of our custom objects.

Example:

model = get_model()
model = train_model(model)
model.save("custom_model.keras")

# Pass the custom objects dictionary to a custom object scope and place
# the `keras.models.load_model()` call within the scope.
custom_objects = {"CustomLayer": CustomLayer, "custom_fn": custom_fn}

with keras.saving.custom_object_scope(custom_objects):
    reconstructed_model = keras.models.load_model("custom_model.keras")

# Let's check:
np.testing.assert_allclose(
    model.predict(test_input), reconstructed_model.predict(test_input)
)

Model serialization

This section is about saving only the model's configuration, without its state. The model's configuration (or architecture) specifies what layers the model contains, and how these layers are connected. If you have the configuration of a model, then the model can be created with a freshly initialized state (no weights or compilation information).

APIs

The following serialization APIs are available:

• keras.models.clone_model(model): make a (randomly initialized) copy of a model.

• get_config() and cls.from_config(): retrieve the configuration of a layer or model, and recreate a model instance from its config, respectively.

• keras.models.model_to_json() and keras.models.model_from_json(): similar, but as JSON strings.

• keras.saving.serialize_keras_object(): retrieve the configuration any arbitrary Keras object.

• keras.saving.deserialize_keras_object(): recreate an object instance from its configuration.

In-memory model cloning

You can do in-memory cloning of a model via keras.models.clone_model(). This is equivalent to getting the config then recreating the model from its config (so it does not preserve compilation information or layer weights values).

Example:

new_model = keras.models.clone_model(model)

get_config() and from_config()

Calling model.get_config() or layer.get_config() will return a Python dict containing the configuration of the model or layer, respectively. You should define get_config() to contain arguments needed for the __init__() method of the model or layer. At loading time, the from_config(config) method will then call __init__() with these arguments to reconstruct the model or layer.

Layer example:

layer = keras.layers.Dense(3, activation="relu")
layer_config = layer.get_config()
print(layer_config)

Now let's reconstruct the layer using the from_config() method:

new_layer = keras.layers.Dense.from_config(layer_config)

Sequential model example:

model = keras.Sequential([keras.Input((32,)), keras.layers.Dense(1)])
config = model.get_config()
new_model = keras.Sequential.from_config(config)

Functional model example:

inputs = keras.Input((32,))
outputs = keras.layers.Dense(1)(inputs)
model = keras.Model(inputs, outputs)
config = model.get_config()
new_model = keras.Model.from_config(config)

to_json() and keras.models.model_from_json()

This is similar to get_config / from_config, except it turns the model into a JSON string, which can then be loaded without the original model class. It is also specific to models, it isn't meant for layers.

Example:

model = keras.Sequential([keras.Input((32,)), keras.layers.Dense(1)])
json_config = model.to_json()
new_model = keras.models.model_from_json(json_config)

Arbitrary object serialization and deserialization

The keras.saving.serialize_keras_object() and keras.saving.deserialize_keras_object() APIs are general-purpose APIs that can be used to serialize or deserialize any Keras object and any custom object. It is at the foundation of saving model architecture and is behind all serialize()/deserialize() calls in keras.

Example:

my_reg = keras.regularizers.L1(0.005)
config = keras.saving.serialize_keras_object(my_reg)
print(config)

Note the serialization format containing all the necessary information for proper reconstruction:

• module containing the name of the Keras module or other identifying module the object comes from

• class_name containing the name of the object's class.

• config with all the information needed to reconstruct the object

• registered_name for custom objects. See here.

Now we can reconstruct the regularizer.

new_reg = keras.saving.deserialize_keras_object(config)

Model weights saving

You can choose to only save & load a model's weights. This can be useful if:

• You only need the model for inference: in this case you won't need to restart training, so you don't need the compilation information or optimizer state.

• You are doing transfer learning: in this case you will be training a new model reusing the state of a prior model, so you don't need the compilation information of the prior model.

APIs for in-memory weight transfer

Weights can be copied between different objects by using get_weights() and set_weights():

• keras.layers.Layer.get_weights(): Returns a list of NumPy arrays of weight values.

• keras.layers.Layer.set_weights(weights): Sets the model weights to the values provided (as NumPy arrays).

Examples:

Transfering weights from one layer to another, in memory

def create_layer():
    layer = keras.layers.Dense(64, activation="relu", name="dense_2")
    layer.build((None, 784))
    return layer


layer_1 = create_layer()
layer_2 = create_layer()

# Copy weights from layer 1 to layer 2
layer_2.set_weights(layer_1.get_weights())

Transfering weights from one model to another model with a compatible architecture, in memory

# Create a simple functional model
inputs = keras.Input(shape=(784,), name="digits")
x = keras.layers.Dense(64, activation="relu", name="dense_1")(inputs)
x = keras.layers.Dense(64, activation="relu", name="dense_2")(x)
outputs = keras.layers.Dense(10, name="predictions")(x)
functional_model = keras.Model(inputs=inputs, outputs=outputs, name="3_layer_mlp")


# Define a subclassed model with the same architecture
class SubclassedModel(keras.Model):
    def __init__(self, output_dim, name=None):
        super().__init__(name=name)
        self.output_dim = output_dim
        self.dense_1 = keras.layers.Dense(64, activation="relu", name="dense_1")
        self.dense_2 = keras.layers.Dense(64, activation="relu", name="dense_2")
        self.dense_3 = keras.layers.Dense(output_dim, name="predictions")

    def call(self, inputs):
        x = self.dense_1(inputs)
        x = self.dense_2(x)
        x = self.dense_3(x)
        return x

    def get_config(self):
        return {"output_dim": self.output_dim, "name": self.name}


subclassed_model = SubclassedModel(10)
# Call the subclassed model once to create the weights.
subclassed_model(np.ones((1, 784)))

# Copy weights from functional_model to subclassed_model.
subclassed_model.set_weights(functional_model.get_weights())

assert len(functional_model.weights) == len(subclassed_model.weights)
for a, b in zip(functional_model.weights, subclassed_model.weights):
    np.testing.assert_allclose(a.numpy(), b.numpy())

The case of stateless layers

Because stateless layers do not change the order or number of weights, models can have compatible architectures even if there are extra/missing stateless layers.

inputs = keras.Input(shape=(784,), name="digits")
x = keras.layers.Dense(64, activation="relu", name="dense_1")(inputs)
x = keras.layers.Dense(64, activation="relu", name="dense_2")(x)
outputs = keras.layers.Dense(10, name="predictions")(x)
functional_model = keras.Model(inputs=inputs, outputs=outputs, name="3_layer_mlp")

inputs = keras.Input(shape=(784,), name="digits")
x = keras.layers.Dense(64, activation="relu", name="dense_1")(inputs)
x = keras.layers.Dense(64, activation="relu", name="dense_2")(x)

# Add a dropout layer, which does not contain any weights.
x = keras.layers.Dropout(0.5)(x)
outputs = keras.layers.Dense(10, name="predictions")(x)
functional_model_with_dropout = keras.Model(
    inputs=inputs, outputs=outputs, name="3_layer_mlp"
)

functional_model_with_dropout.set_weights(functional_model.get_weights())

APIs for saving weights to disk & loading them back

Weights can be saved to disk by calling model.save_weights(filepath). The filename should end in .weights.h5.

Example:

# Runnable example
sequential_model = keras.Sequential(
    [
        keras.Input(shape=(784,), name="digits"),
        keras.layers.Dense(64, activation="relu", name="dense_1"),
        keras.layers.Dense(64, activation="relu", name="dense_2"),
        keras.layers.Dense(10, name="predictions"),
    ]
)
sequential_model.save_weights("my_model.weights.h5")
sequential_model.load_weights("my_model.weights.h5")

Note that changing layer.trainable may result in a different layer.weights ordering when the model contains nested layers.

class NestedDenseLayer(keras.layers.Layer):
    def __init__(self, units, name=None):
        super().__init__(name=name)
        self.dense_1 = keras.layers.Dense(units, name="dense_1")
        self.dense_2 = keras.layers.Dense(units, name="dense_2")

    def call(self, inputs):
        return self.dense_2(self.dense_1(inputs))


nested_model = keras.Sequential([keras.Input((784,)), NestedDenseLayer(10, "nested")])
variable_names = [v.name for v in nested_model.weights]
print("variables: {}".format(variable_names))

print("\nChanging trainable status of one of the nested layers...")
nested_model.get_layer("nested").dense_1.trainable = False

variable_names_2 = [v.name for v in nested_model.weights]
print("\nvariables: {}".format(variable_names_2))
print("variable ordering changed:", variable_names != variable_names_2)

Transfer learning example

When loading pretrained weights from a weights file, it is recommended to load the weights into the original checkpointed model, and then extract the desired weights/layers into a new model.

Example:

def create_functional_model():
    inputs = keras.Input(shape=(784,), name="digits")
    x = keras.layers.Dense(64, activation="relu", name="dense_1")(inputs)
    x = keras.layers.Dense(64, activation="relu", name="dense_2")(x)
    outputs = keras.layers.Dense(10, name="predictions")(x)
    return keras.Model(inputs=inputs, outputs=outputs, name="3_layer_mlp")


functional_model = create_functional_model()
functional_model.save_weights("pretrained.weights.h5")

# In a separate program:
pretrained_model = create_functional_model()
pretrained_model.load_weights("pretrained.weights.h5")

# Create a new model by extracting layers from the original model:
extracted_layers = pretrained_model.layers[:-1]
extracted_layers.append(keras.layers.Dense(5, name="dense_3"))
model = keras.Sequential(extracted_layers)
model.summary()

Appendix: Handling custom objects

Defining the config methods

Specifications:

• get_config() should return a JSON-serializable dictionary in order to be compatible with the Keras architecture- and model-saving APIs.

• from_config(config) (a classmethod) should return a new layer or model object that is created from the config. The default implementation returns cls(**config).

NOTE: If all your constructor arguments are already serializable, e.g. strings and ints, or non-custom Keras objects, overriding from_config is not necessary. However, for more complex objects such as layers or models passed to __init__, deserialization must be handled explicitly either in __init__ itself or overriding the from_config() method.

Example:

@keras.saving.register_keras_serializable(package="MyLayers", name="KernelMult")
class MyDense(keras.layers.Layer):
    def __init__(
        self,
        units,
        *,
        kernel_regularizer=None,
        kernel_initializer=None,
        nested_model=None,
        **kwargs
    ):
        super().__init__(**kwargs)
        self.hidden_units = units
        self.kernel_regularizer = kernel_regularizer
        self.kernel_initializer = kernel_initializer
        self.nested_model = nested_model

    def get_config(self):
        config = super().get_config()
        # Update the config with the custom layer's parameters
        config.update(
            {
                "units": self.hidden_units,
                "kernel_regularizer": self.kernel_regularizer,
                "kernel_initializer": self.kernel_initializer,
                "nested_model": self.nested_model,
            }
        )
        return config

    def build(self, input_shape):
        input_units = input_shape[-1]
        self.kernel = self.add_weight(
            name="kernel",
            shape=(input_units, self.hidden_units),
            regularizer=self.kernel_regularizer,
            initializer=self.kernel_initializer,
        )

    def call(self, inputs):
        return ops.matmul(inputs, self.kernel)


layer = MyDense(units=16, kernel_regularizer="l1", kernel_initializer="ones")
layer3 = MyDense(units=64, nested_model=layer)

config = keras.layers.serialize(layer3)

print(config)

new_layer = keras.layers.deserialize(config)

print(new_layer)

Note that overriding from_config is unnecessary above for MyDense because hidden_units, kernel_initializer, and kernel_regularizer are ints, strings, and a built-in Keras object, respectively. This means that the default from_config implementation of cls(**config) will work as intended.

For more complex objects, such as layers and models passed to __init__, for example, you must explicitly deserialize these objects. Let's take a look at an example of a model where a from_config override is necessary.

Example:

@keras.saving.register_keras_serializable(package="ComplexModels")
class CustomModel(keras.layers.Layer):
    def __init__(self, first_layer, second_layer=None, **kwargs):
        super().__init__(**kwargs)
        self.first_layer = first_layer
        if second_layer is not None:
            self.second_layer = second_layer
        else:
            self.second_layer = keras.layers.Dense(8)

    def get_config(self):
        config = super().get_config()
        config.update(
            {
                "first_layer": self.first_layer,
                "second_layer": self.second_layer,
            }
        )
        return config

    @classmethod
    def from_config(cls, config):
        # Note that you can also use `keras.saving.deserialize_keras_object` here
        config["first_layer"] = keras.layers.deserialize(config["first_layer"])
        config["second_layer"] = keras.layers.deserialize(config["second_layer"])
        return cls(**config)

    def call(self, inputs):
        return self.first_layer(self.second_layer(inputs))


# Let's make our first layer the custom layer from the previous example (MyDense)
inputs = keras.Input((32,))
outputs = CustomModel(first_layer=layer)(inputs)
model = keras.Model(inputs, outputs)

config = model.get_config()
new_model = keras.Model.from_config(config)

How custom objects are serialized

The serialization format has a special key for custom objects registered via @keras.saving.register_keras_serializable. This registered_name key allows for easy retrieval at loading/deserialization time while also allowing users to add custom naming.

Let's take a look at the config from serializing the custom layer MyDense we defined above.

Example:

layer = MyDense(
    units=16,
    kernel_regularizer=keras.regularizers.L1L2(l1=1e-5, l2=1e-4),
    kernel_initializer="ones",
)
config = keras.layers.serialize(layer)
print(config)

As shown, the registered_name key contains the lookup information for the Keras master list, including the package MyLayers and the custom name KernelMult that we gave in the @keras.saving.register_keras_serializable decorator. Take a look again at the custom class definition/registration here.

Note that the class_name key contains the original name of the class, allowing for proper re-initialization in from_config.

Additionally, note that the module key is None since this is a custom object.