Why the training loop is explicit in PyTorch - Test Your Understanding

Practice - 5 Tasks

1fill in blank

easy

Complete the code to start the training loop by iterating over the data loader.

PyTorch

for [1] in train_loader:
    inputs, labels = batch
    # training steps follow

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

Amodel

Bepoch

Cbatch

Doptimizer

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'epoch' instead of 'batch' as the loop variable.

Using 'model' or 'optimizer' which are not iterable.

The training loop iterates over batches of data from the train_loader. 'batch' is the variable holding each batch.

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'

Azero_grad

Bstep

Cbackward

Deval

Attempts:

3 left

💡 Hint

Common Mistakes

Calling optimizer.step() before zeroing gradients.

Using backward() or eval() which are unrelated here.

Before computing gradients, we clear old gradients with optimizer.zero_grad().

3fill in blank

hard

Fix the error in the code by completing the line that performs backpropagation.

PyTorch

loss.[1]()

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

Astep

Bzero_grad

Cforward

Dbackward

Attempts:

3 left

💡 Hint

Common Mistakes

Using step() which updates parameters, not computes gradients.

Using zero_grad() which clears gradients.

Using forward() which runs the model.

Calling loss.backward() computes gradients for backpropagation.

4fill in blank

hard

Fill both blanks to update model parameters and calculate the loss.

PyTorch

optimizer.[1]()
loss = criterion(outputs, [2])

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

Astep

Blabels

Cinputs

Dzero_grad

Attempts:

3 left

💡 Hint

Common Mistakes

Using zero_grad() instead of step() to update parameters.

Using inputs instead of labels for loss calculation.

optimizer.step() updates parameters; loss is calculated comparing outputs to labels.

5fill in blank

hard

Fill all three blanks to complete a simple explicit training loop in PyTorch.

PyTorch

for [1] in range(num_epochs):
    for batch in train_loader:
        inputs, labels = batch
        optimizer.[2]()
        outputs = model(inputs)
        loss = criterion(outputs, [3])
        loss.backward()
        optimizer.step()

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

Aepoch

Bzero_grad

Clabels

Dbatch

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'batch' as the outer loop variable instead of 'epoch'.

Forgetting to zero gradients before backward().

Using inputs instead of labels for loss calculation.

The outer loop uses 'epoch', gradients are zeroed with zero_grad(), and loss compares outputs to labels.