TensorFlow Saving and Loading Models 3

Question 1

What does setting include_optimizer=False in model.save() affect?

Accepted Answer

Excludes the optimizer state, making the file smaller but unable to resume training exactly

Answer

Removes the optimizer from the model permanently

Answer

Forces the model to use default optimizer settings on reload

Answer

Prevents the model from being compiled after loading

Question 2

What type of object does tf.saved_model.load() return?

Accepted Answer

A trackable object containing the model's saved functions and variables

Answer

A compiled tf.keras.Model

Answer

A TensorFlow 1.x Session

Answer

A plain Python dictionary of tensors

Question 3

Why should serving methods in a SavedModel be decorated with @tf.function?

Accepted Answer

It traces the Python function into a TensorFlow graph that can be serialized and run without Python

Answer

It prevents those methods from being serialized

Answer

It marks methods as non-trainable

Answer

It compresses the function bytecode for storage

Question 4

Which serialization format is required by TensorFlow Serving for model deployment?

Accepted Answer

SavedModel directory format

Answer

HDF5 (.h5)

Answer

Pickle (.pkl)

Answer

ONNX format

Question 5

What is the purpose of the signatures attribute on a loaded SavedModel object?

Accepted Answer

Providing named callable entry points used by serving infrastructure

Answer

Storing the model's cryptographic signature for authenticity

Answer

Defining which layers are frozen during fine-tuning

Answer

Recording the model version and creation timestamp

Question 6

Which Keras method serializes only the model architecture to a JSON string?

Accepted Answer

model.to_json()

Answer

model.get_config()

Answer

model.serialize()

Answer

model.export_architecture()

Question 7

Which single Keras function can load both HDF5 and SavedModel format files?

Accepted Answer

tf.keras.models.load_model()

Answer

tf.keras.models.from_saved_model()

Answer

tf.saved_model.load_keras()

Answer

tf.keras.models.import_model()