TensorFlow Model Training and Optimization 2

Question 1

What does the tf.function decorator do in TensorFlow 2?

Accepted Answer

Compiles a Python function into a TensorFlow graph for faster execution

Answer

tf.function traces the Python function and compiles it into a static computation graph, enabling performance optimizations like XLA.

Question 2

What is the purpose of model.evaluate() in Keras?

Accepted Answer

Computes loss and metrics on test data without updating weights

Answer

model.evaluate() runs forward passes on provided data and returns the configured loss and metric values.

Question 3

Which regularization technique randomly drops neurons during forward pass?

Accepted Answer

Dropout

Answer

Dropout randomly sets a fraction of neuron outputs to zero during training, forcing the network to learn redundant representations.

Question 4

What is the purpose of the ModelCheckpoint callback?

Accepted Answer

Saves the model at specified intervals or when performance improves

Answer

ModelCheckpoint saves model weights (or the full model) periodically, allowing recovery from the best or most recent state.

Question 5

What does class_weight parameter in model.fit() do?

Accepted Answer

Assigns higher loss weight to underrepresented classes

Answer

class_weight applies a multiplier to the loss for each class, helping the model pay more attention to minority classes in imbalanced datasets.

TensorFlow Practice Test