Training loop structure in PyTorch - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to start the training loop for 5 epochs.

PyTorch

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

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

A10

B0

C5

D1

Attempts:

3 left

💡 Hint

Common Mistakes

Using range(10) which runs too many epochs.

Using range(0) which runs zero times.

The training loop runs for 5 epochs, so the range should be 5.

2fill in blank

medium

Complete the code to zero the gradients before backpropagation.

PyTorch

optimizer.[1]()

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

Aupdate

Bstep

Cbackward

Dzero_grad

Attempts:

3 left

💡 Hint

Common Mistakes

Calling optimizer.step() before zeroing gradients.

Using optimizer.backward() which does not exist.

Before backpropagation, gradients must be reset using optimizer.zero_grad().

3fill in blank

hard

Fix the error in the code to perform backpropagation on the loss.

PyTorch

loss.[1]()

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

Abackward

Bupdate

Czero_grad

Dstep

Attempts:

3 left

💡 Hint

Common Mistakes

Calling loss.step() which does not exist.

Calling loss.zero_grad() which is incorrect.

To compute gradients, call loss.backward() after computing the loss.

4fill in blank

hard

Fill both blanks to update model parameters and print the loss value.

PyTorch

optimizer.[1]()
print(f"Loss: {loss.[2]()}")

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

Astep

Bitem

Cbackward

Dzero_grad

Attempts:

3 left

💡 Hint

Common Mistakes

Calling loss.backward() instead of loss.item() in print.

Calling optimizer.zero_grad() instead of optimizer.step().

optimizer.step() updates parameters. loss.item() gets the scalar loss value for printing.

5fill in blank

hard

Fill all three blanks to complete the training loop: zero gradients, compute loss backward, and update parameters.

PyTorch

for epoch in range(3):
    optimizer.[1]()
    output = model(data)
    loss = loss_fn(output, target)
    loss.[2]()
    optimizer.[3]()

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

Azero_grad

Bbackward

Cstep

Deval

Attempts:

3 left

💡 Hint

Common Mistakes

Calling optimizer.eval() which is for evaluation mode.

Skipping zero_grad causing gradient accumulation.

In each epoch, zero gradients, compute gradients with backward(), then update parameters with step().