Tensor creation (constant, variable, zeros, ones) in TensorFlow - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is a constant tensor in TensorFlow?

A constant tensor is a fixed value tensor that cannot be changed after creation. It holds data that stays the same during the program run.

beginner

How does a variable tensor differ from a constant tensor?

A variable tensor can change its values during program execution, useful for model parameters that update during training.

beginner

What does tf.zeros(shape) do?

It creates a tensor filled with zeros of the specified shape. For example, tf.zeros([2,3]) creates a 2x3 tensor filled with zeros.

beginner

What is the use of tf.ones(shape)?

It creates a tensor filled with ones of the given shape. Useful for initializing tensors where all values start as one.

intermediate

How do you create a variable tensor initialized with ones in TensorFlow?

Use tf.Variable(tf.ones(shape)) to create a variable tensor filled with ones that can be updated later.

Which TensorFlow function creates a tensor that cannot be changed after creation?

A tf.zeros()

B tf.Variable()

C tf.constant()

D tf.ones()

tf.constant() creates a tensor with fixed values that cannot be modified.

What does tf.zeros([3,2]) produce?

A A 3x2 tensor filled with ones

B A 3x2 tensor filled with zeros

C A 2x3 tensor filled with zeros

D A variable tensor with zeros

tf.zeros([3,2]) creates a tensor with 3 rows and 2 columns filled with zeros.

Which tensor type allows changing its values during training?

A Constant tensor

B Ones tensor

C Zeros tensor

D Variable tensor

Variable tensors can be updated, which is essential for training models.

How do you create a tensor filled with ones in TensorFlow?

A tf.ones(shape)

B tf.zeros(shape)

C tf.constant(shape)

D tf.Variable(shape)

tf.ones(shape) creates a tensor filled with ones of the specified shape.

Which code creates a variable tensor initialized with zeros?

A tf.Variable(tf.zeros(shape))

B tf.zeros(shape)

C tf.constant(tf.zeros(shape))

D tf.ones(shape)

tf.Variable(tf.zeros(shape)) creates a variable tensor initialized with zeros.

Explain the difference between tf.constant and tf.Variable tensors and when to use each.

Think about whether the tensor values need to change during training.

Describe how to create tensors filled with zeros and ones and why these might be useful.

Consider initialization in neural networks.