Kernel size, stride, padding in PyTorch - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to create a 2D convolutional layer with a kernel size of 3.

PyTorch

conv = torch.nn.Conv2d(in_channels=1, out_channels=1, kernel_size=[1])

Drag options to blanks, or click blank then click option'

A3

B5

C1

D7

Attempts:

3 left

💡 Hint

Common Mistakes

Using a kernel size too large for the input image.

Confusing kernel size with stride.

The kernel size defines the size of the filter that slides over the input image. Here, 3 means a 3x3 filter.

2fill in blank

medium

Complete the code to set the stride of the convolutional layer to 2.

PyTorch

conv = torch.nn.Conv2d(in_channels=1, out_channels=1, kernel_size=3, stride=[1])

Drag options to blanks, or click blank then click option'

A2

B1

C3

D4

Attempts:

3 left

💡 Hint

Common Mistakes

Setting stride to 0 which is invalid.

Confusing stride with padding.

Stride controls how much the filter moves at each step. A stride of 2 means the filter moves 2 pixels at a time.

3fill in blank

hard

Fix the error in the code by choosing the correct padding value to keep output size same as input size with kernel size 3 and stride 1.

PyTorch

conv = torch.nn.Conv2d(in_channels=1, out_channels=1, kernel_size=3, stride=1, padding=[1])

Drag options to blanks, or click blank then click option'

A0

B3

C1

D2

Attempts:

3 left

💡 Hint

Common Mistakes

Using padding 0 which shrinks output size.

Using padding too large which increases output size.

Padding of 1 adds a border of zeros around the input, keeping output size same when kernel size is 3 and stride is 1.

4fill in blank

hard

Fill both blanks to create a convolutional layer with kernel size 5 and stride 2.

PyTorch

conv = torch.nn.Conv2d(in_channels=3, out_channels=6, kernel_size=[1], stride=[2])

Drag options to blanks, or click blank then click option'

A5

B1

C2

D3

Attempts:

3 left

💡 Hint

Common Mistakes

Mixing up kernel size and stride values.

Using stride 1 when stride 2 is required.

Kernel size 5 means a 5x5 filter. Stride 2 means the filter moves 2 pixels at a time.

5fill in blank

hard

Fill all three blanks to create a convolutional layer with kernel size 7, stride 1, and padding 3.

PyTorch

conv = torch.nn.Conv2d(in_channels=[1], out_channels=[2], kernel_size=[3], stride=1, padding=3)

Drag options to blanks, or click blank then click option'

A3

B6

C7

D1

Attempts:

3 left

💡 Hint

Common Mistakes

Using wrong number of input or output channels.

Incorrect kernel size or padding causing size mismatch.

in_channels=3 for RGB images, out_channels=6 for 6 filters, kernel_size=7 for a 7x7 filter, stride=1 and padding=3 to keep output size.