Fine-tuning strategy in PyTorch - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load a pre-trained model for fine-tuning.

PyTorch

import torch
from torchvision import models

model = models.resnet18(pretrained=[1])

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

A0

BTrue

CNone

DFalse

Attempts:

3 left

💡 Hint

Common Mistakes

Setting pretrained to False loads a model with random weights, not suitable for fine-tuning.

Using None or 0 as pretrained argument causes errors.

Setting pretrained=True loads the model with pre-trained weights, which is essential for fine-tuning.

2fill in blank

medium

Complete the code to freeze all layers except the final fully connected layer for fine-tuning.

PyTorch

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

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

Adetach

Bgrad

Ctrain

Drequires_grad

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'grad' or 'train' attributes which do not control gradient computation.

Calling detach() on parameters instead of setting requires_grad.

Setting requires_grad = False freezes the parameters so they are not updated during training.

3fill in blank

hard

Fix the error in replacing the final layer to match 10 output classes.

PyTorch

import torch.nn as nn

model.fc = nn.Linear(model.fc.in_features, [1])

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

A10

B5

C100

D1

Attempts:

3 left

💡 Hint

Common Mistakes

Setting output features to a wrong number causes shape mismatch errors.

Using 1 or 5 instead of the correct number of classes.

The final layer's output features must match the number of classes, which is 10 here.

4fill in blank

hard

Fill both blanks to create an optimizer that only updates the final layer parameters with a learning rate of 0.001.

PyTorch

import torch.optim as optim

optimizer = optim.SGD([1], lr=[2])

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

Amodel.fc.parameters()

Bmodel.parameters()

C0.01

D0.001

Attempts:

3 left

💡 Hint

Common Mistakes

Using all model parameters causes the whole model to train, not just the final layer.

Using a too high learning rate can cause unstable training.

Only the final layer's parameters should be updated, and the learning rate is set to 0.001 for fine-tuning.

5fill in blank

hard

Fill all three blanks to write a training loop that computes loss, backpropagates, and updates parameters.

PyTorch

for inputs, labels in dataloader:
    optimizer.zero_grad()
    outputs = model([1])
    loss = criterion(outputs, [2])
    loss.[3]()  # backpropagation
    optimizer.step()

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

Ainputs

Blabels

Cbackward

Dforward

Attempts:

3 left

💡 Hint

Common Mistakes

Passing labels to the model instead of inputs.

Calling loss.forward() instead of loss.backward().

Forgetting to zero gradients before backpropagation.

The model takes inputs to produce outputs. The loss compares outputs to labels. Calling loss.backward() computes gradients.