Indexing and slicing tensors in TensorFlow - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of slicing a 3D tensor

What is the output shape of the tensor after slicing in the code below?

TensorFlow

import tensorflow as tf

x = tf.constant([[[1, 2], [3, 4]], [[5, 6], [7, 8]]])
sliced = x[:, 1, :]
print(sliced.shape)

A (2, 1)

B (1, 1)

C (1, 2)

D (2, 2)

💡 Hint

Remember that ':' means selecting all elements along that axis.

The original tensor x has shape (2, 2, 2). The slice x[:, 1, :] selects all elements in the first dimension, the second element (index 1) in the second dimension, and all elements in the third dimension. This results in a tensor of shape (2, 2).

❓ Model Choice

intermediate

Choosing correct slicing for extracting a column vector

Given a 2D tensor of shape (5, 4), which slicing extracts the third column as a 1D tensor?

A tensor[:, 2:3]

B tensor[2, :]

C tensor[:, 2]

D tensor[2]

💡 Hint

Remember that the first index is for rows and the second for columns.

tensor[:, 2] selects all rows and the third column (index 2), resulting in a 1D tensor of shape (5,). tensor[:, 2:3] would keep the dimension and return shape (5, 1).

❓ Hyperparameter

advanced

Effect of slicing on batch size in training data

If you have a tensor representing a batch of images with shape (64, 28, 28, 3), what is the shape after slicing with tensor[:32]?

A (64, 28, 28, 3)

B (32, 28, 28, 3)

C (32, 28, 28)

D (28, 28, 3)

💡 Hint

Slicing with :32 selects the first 32 elements along the first dimension.

The first dimension is the batch size. Slicing tensor[:32] selects the first 32 images, keeping the rest of the dimensions unchanged.

🔧 Debug

advanced

Identify the error in tensor slicing code

What error does the following code produce? import tensorflow as tf x = tf.constant([[1, 2, 3], [4, 5, 6]]) slice = x[:, 3] print(slice)

TensorFlow

import tensorflow as tf
x = tf.constant([[1, 2, 3], [4, 5, 6]])
slice = x[:, 3]
print(slice)

A IndexError: index 3 is out of bounds for axis 1 with size 3

B TypeError: unsupported operand type(s) for slicing

C ValueError: cannot slice tensor with ':'

D No error, prints a tensor

💡 Hint

Check the size of the second dimension and the index used.

The tensor has shape (2, 3), so valid column indices are 0, 1, 2. Index 3 is out of bounds, causing an IndexError.

🧠 Conceptual

expert

Understanding advanced tensor slicing with steps

Given a 1D tensor x = tf.constant([10, 20, 30, 40, 50, 60]), what is the output of x[1:5:2].numpy()?

TensorFlow

import tensorflow as tf
x = tf.constant([10, 20, 30, 40, 50, 60])
result = x[1:5:2].numpy()
print(result)

A [20 40]

B [20 30 40 50]

C [30 50]

D [10 30 50]

💡 Hint

Remember slicing syntax: start:stop:step selects elements from start up to but not including stop, stepping by step.

x[1:5:2] selects elements at indices 1 and 3 (20 and 40).