Indexing and slicing tensors in TensorFlow - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is a tensor in TensorFlow?

A tensor is a multi-dimensional array used to store data in TensorFlow. It can have any number of dimensions, like scalars (0D), vectors (1D), matrices (2D), or higher.

beginner

How do you get the first element of a 1D tensor named t?

You use t[0] to get the first element, just like accessing the first item in a list.

beginner

What does the slice t[1:4] do on a 1D tensor?

It extracts elements starting from index 1 up to, but not including, index 4. So it gets elements at positions 1, 2, and 3.

intermediate

How do you select the second row and third column element from a 2D tensor t?

Use t[1, 2]. The first index is the row (starting at 0), the second is the column.

intermediate

What does t[:, 1] mean for a 2D tensor t?

It selects all rows (the :) but only the second column (index 1). This extracts a 1D tensor of that column.

What does t[2:5] return for a 1D tensor t?

A Elements at indices 3, 4, and 5

B Elements at indices 2, 3, and 4

C Elements at indices 2, 3, 4, and 5

D Elements at indices 0 to 5

Slicing includes the start index (2) but excludes the end index (5), so it returns elements at 2, 3, and 4.

How do you select the last element of a 1D tensor t?

A t[-1]

B t[0]

C t[len(t)]

D t[1]

Using -1 as the index selects the last element in the tensor.

What does t[0, :] select in a 2D tensor t?

A Last row, all columns

B All rows, first column

C First element only

D First row, all columns

The first index 0 selects the first row; the colon : selects all columns.

If t is a 3D tensor, what does t[:, 1, :] select?

A All elements in the second slice of the second dimension

B Only the first element

C All elements in the first slice of the first dimension

D Only the last element

The colon : selects all in the first dimension, 1 selects the second index in the second dimension, and colon selects all in the third dimension.

What is the shape of the tensor after slicing t[:, 2:5] if t is shape (4, 6)?

A (4, 6)

B (3, 6)

C (4, 3)

D (3, 4)

Slicing columns from index 2 to 5 (3 columns) keeps all 4 rows, so shape is (4, 3).

Explain how to use indexing and slicing to extract a sub-tensor from a 2D tensor in TensorFlow.

Think about how you pick rows and columns like selecting cells in a spreadsheet.

Describe how negative indices work when indexing tensors in TensorFlow.

Imagine counting backwards from the last item.