TensorFlow 2.x / Keras 15 questions 20 min

TensorFlow & Keras MCQ · test your framework knowledge

From Sequential models to custom training loops – 15 questions covering layers, compile/fit, callbacks, and best practices.

Easy: 5 Medium: 6 Hard: 4
Sequential
Functional API
Callbacks
Training

TensorFlow & Keras: the industry standard for deep learning

TensorFlow 2.x integrates Keras as its high-level API, making model building intuitive. This MCQ covers essential classes like tf.keras.Sequential, the Functional API for complex topologies, callbacks (EarlyStopping, ModelCheckpoint), and core concepts like eager execution and AutoGraph.

Why tf.keras?

It combines the flexibility of TensorFlow with the simplicity of Keras: you can go from quick prototyping (Sequential) to fully custom training loops while staying in the same ecosystem.

TensorFlow/Keras glossary – key concepts

Sequential model

Linear stack of layers. Ideal for most feedforward networks. tf.keras.Sequential([Dense(64, activation='relu'), Dense(10)])

Functional API

Build non‑linear topologies: multi‑input, multi‑output, residual connections. Uses layer call syntax.

Callbacks

Utilities like EarlyStopping, ModelCheckpoint, TensorBoard that hook into training.

compile & fit

model.compile(optimizer='adam', loss='mse') – configures learning process. fit() trains the model.

Eager execution

Default in TF2.x: operations run immediately, intuitive debugging. Can still build graphs via @tf.function.

tf.data.Dataset

Efficient input pipelines: batching, shuffling, prefetch. Integrates seamlessly with Keras.

Custom training loop

Use tf.GradientTape for fine‑grained control. Essential for research.

# Typical Keras pipeline
model = tf.keras.Sequential([
    layers.Dense(128, activation='relu'),
    layers.Dropout(0.2),
    layers.Dense(10, activation='softmax')
])
model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['acc'])
model.fit(train_ds, validation_data=val_ds, epochs=10, callbacks=[EarlyStopping()])
Interview tip: Know the difference between Sequential and Functional API, when to use model.fit() vs custom loops, and how callbacks work. This MCQ covers these core distinctions.

Common TensorFlow interview questions

  • What is the difference between TensorFlow 1.x and 2.x regarding execution?
  • How do you create a multi‑input model in Keras?
  • Explain the role of @tf.function and AutoGraph.
  • What is the purpose of the validation_split parameter in fit()?
  • How can you implement early stopping without a callback?
  • Describe how tf.GradientTape works.
AI Hub Next: PyTorch MCQ