TensorFlow tf.data Pipelines

Question 1

Which tf.data method applies a function to each element of a dataset?

Accepted Answer

dataset.map()

Answer

dataset.map() applies a given function to every element in the dataset, enabling preprocessing transformations.

Question 2

What does dataset.batch(32) do in tf.data?

Accepted Answer

Groups consecutive elements into batches of 32

Answer

dataset.batch(n) combines n consecutive dataset elements into a single batched element.

Question 3

Which tf.data method shuffles dataset elements?

Accepted Answer

dataset.shuffle(buffer_size)

Answer

dataset.shuffle() randomly shuffles elements within a buffer of the specified size before drawing the next element.

Question 4

What does dataset.prefetch(tf.data.AUTOTUNE) do?

Accepted Answer

Prepares the next batch while the current one is being processed

Answer

prefetch() overlaps the data preprocessing and model execution to reduce training bottlenecks.

Question 5

Which function creates a tf.data.Dataset from a list of NumPy arrays?

Accepted Answer

tf.data.Dataset.from_tensor_slices()

Answer

from_tensor_slices() slices the first dimension of each array to create individual dataset elements.

TensorFlow Practice Test