Batch normalization (nn.BatchNorm) in PyTorch - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to create a batch normalization layer for 10 features.

PyTorch

import torch.nn as nn

batch_norm = nn.BatchNorm[1](10)

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

A3d

B4d

C2d

D1d

Attempts:

3 left

💡 Hint

Common Mistakes

Using BatchNorm2d for 1D features causes shape errors.

Forgetting to specify the number of features.

The nn.BatchNorm1d layer is used for batch normalization on 1-dimensional feature vectors, such as in fully connected layers or 1D conv outputs.

2fill in blank

medium

Complete the code to apply batch normalization to input tensor x.

PyTorch

output = batch_norm([1])

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

Ann.BatchNorm1d

Bx

Cbatch_norm

Dinput

Attempts:

3 left

💡 Hint

Common Mistakes

Passing the layer itself instead of the input tensor.

Using undefined variable names.

You apply the batch normalization layer to the input tensor x by calling it like a function.

3fill in blank

hard

Fix the error in the code to correctly initialize batch normalization for 3D input.

PyTorch

batch_norm = nn.BatchNorm[1](16)

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

A2d

B1d

C3d

D4d

Attempts:

3 left

💡 Hint

Common Mistakes

Using BatchNorm1d or BatchNorm2d for 3D data causes shape mismatch errors.

Passing 'channels' as a keyword argument instead of 'num_features'.

For 3D inputs like videos or volumetric data, nn.BatchNorm3d is the correct batch normalization layer.

4fill in blank

hard

Fill both blanks to create a batch normalization layer and apply it to input tensor x.

PyTorch

batch_norm = nn.BatchNorm[1]([2])
output = batch_norm(x)

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

A2d

B1d

C10

D16

Attempts:

3 left

💡 Hint

Common Mistakes

Mixing BatchNorm1d with 2D data.

Using wrong number of features.

BatchNorm2d is used for 2D convolutional outputs with 16 channels.

5fill in blank

hard

Fill all three blanks to create a batch normalization layer, apply it to x, and get the mean of the output.

PyTorch

batch_norm = nn.BatchNorm[1]([2])
output = batch_norm([3])
mean_val = output.mean()

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

A1d

Bx

C20

D3d

Attempts:

3 left

💡 Hint

Common Mistakes

Using BatchNorm3d for 1D data.

Passing wrong variable names.

BatchNorm1d is created for 20 features, applied to input tensor x, then the mean of the output is computed.