Freezing layers in PyTorch - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to freeze all parameters in the model.

PyTorch

for param in model.parameters():
    param.[1] = False

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

Arequires_grad

Bgrad

Cfreeze

Ddetach

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'grad' instead of 'requires_grad'.

Trying to call a method like 'freeze()' which doesn't exist.

Setting requires_grad = False tells PyTorch not to update these parameters during training, effectively freezing them.

2fill in blank

medium

Complete the code to freeze only the first layer of the model.

PyTorch

for param in model.[1].parameters():
    param.requires_grad = False

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

Alayers

Bmodules

Cchildren

Dlayer1

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'children' or 'modules' which return iterables, not a single layer.

Using 'layers' which is not a standard attribute.

Accessing model.layer1 targets the first layer to freeze its parameters.

3fill in blank

hard

Fix the error in the code to freeze the model's convolutional layers only.

PyTorch

for name, param in model.named_parameters():
    if 'conv' in name:
        param.[1] = False

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

Agrad

Brequires_grad

Cdetach

Dfreeze

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'detach' which returns a new tensor but does not freeze parameters.

Trying to set 'grad' or 'freeze' which are invalid.

To freeze parameters, set requires_grad = False. Other options are incorrect or invalid.

4fill in blank

hard

Fill both blanks to freeze all layers except the last fully connected layer.

PyTorch

for name, param in model.named_parameters():
    if not [1] in name:
        param.[2] = False

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

A'fc'

Brequires_grad

C'conv'

Dgrad

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'conv' instead of 'fc' to exclude the last layer.

Trying to set 'grad' instead of 'requires_grad'.

We skip freezing parameters whose name contains 'fc' (fully connected layer). For others, set requires_grad = False to freeze.

5fill in blank

hard

Fill all three blanks to freeze all layers except those containing 'bn' or 'fc' in their names.

PyTorch

for name, param in model.named_parameters():
    if not ([1] in name or [2] in name):
        param.[3] = False

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

A'bn'

B'fc'

Crequires_grad

D'conv'

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'conv' instead of 'bn' or 'fc' to exclude layers.

Trying to set 'grad' or other invalid attributes.

We exclude layers with 'bn' (batch norm) and 'fc' (fully connected) from freezing. For others, set requires_grad = False.