Epoch-based training in PyTorch - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to set the number of epochs for training.

PyTorch

num_epochs = [1]

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

A10

B'10'

Crange(10)

DNone

Attempts:

3 left

💡 Hint

Common Mistakes

Using a string instead of an integer for epochs.

Using a range object instead of an integer.

The number of epochs should be an integer value, like 10, not a string or range.

2fill in blank

medium

Complete the code to start the training loop over epochs.

PyTorch

for epoch in [1](num_epochs):
    print(f"Epoch {epoch+1}")

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

Alen

Blist

Cenumerate

Drange

Attempts:

3 left

💡 Hint

Common Mistakes

Using len which returns an integer, not iterable.

Using list without an argument.

Using enumerate without an iterable.

We use range(num_epochs) to loop from 0 to num_epochs-1.

3fill in blank

hard

Fix the error in the training loop to correctly zero the gradients.

PyTorch

for epoch in range(num_epochs):
    optimizer.[1]()
    outputs = model(inputs)
    loss = criterion(outputs, labels)
    loss.backward()
    optimizer.step()

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

Azero_gradients

Bclear_grad

Czero_grad

Dreset_grad

Attempts:

3 left

💡 Hint

Common Mistakes

Using non-existent methods like zero_gradients().

Forgetting to clear gradients causing accumulation.

In PyTorch, the correct method to clear gradients before backpropagation is zero_grad().

4fill in blank

hard

Fill both blanks to compute the loss and update the model parameters inside the epoch loop.

PyTorch

for epoch in range(num_epochs):
    optimizer.zero_grad()
    outputs = model(inputs)
    loss = [1](outputs, labels)
    loss.[2]()
    optimizer.step()

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

Acriterion

Bbackward

Cstep

Dzero_grad

Attempts:

3 left

💡 Hint

Common Mistakes

Calling step() on loss instead of optimizer.

Not calling backward() on loss.

The loss is computed by calling the criterion function, then loss.backward() computes gradients.

5fill in blank

hard

Fill all three blanks to print the epoch number, loss value, and accuracy after each epoch.

PyTorch

for epoch in range(num_epochs):
    optimizer.zero_grad()
    outputs = model(inputs)
    loss = criterion(outputs, labels)
    loss.backward()
    optimizer.step()

    _, predicted = torch.max(outputs, [1])
    correct = (predicted == labels).sum().item()
    accuracy = correct / labels.size([2])
    print(f"Epoch [3]: Loss={loss.item():.4f}, Accuracy={accuracy:.2%}")

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

A1

B0

Cepoch + 1

D2

Attempts:

3 left

💡 Hint

Common Mistakes

Using wrong dimension for torch.max causing errors.

Using wrong dimension for batch size causing wrong accuracy.

Printing epoch without adding 1, confusing learners.

We use dimension 1 for torch.max to get class predictions, dimension 0 for batch size, and print epoch number as epoch + 1.