Why automatic differentiation enables training in PyTorch - Test Your Understanding

Practice - 5 Tasks

1fill in blank

easy

Complete the code to create a tensor that tracks gradients for training.

PyTorch

import torch
x = torch.tensor([2.0, 3.0], requires_grad=[1])

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

ATrue

BFalse

CNone

D0

Attempts:

3 left

💡 Hint

Common Mistakes

Setting requires_grad to False disables gradient tracking.

Setting requires_grad=True tells PyTorch to track operations on the tensor for gradient calculation.

2fill in blank

medium

Complete the code to compute the gradient of y with respect to x.

PyTorch

y = x.pow(2).sum()
y.backward([1])

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

Aretain_graph=True

Bcreate_graph=True

Cinputs=x

DNone

Attempts:

3 left

💡 Hint

Common Mistakes

Passing unnecessary arguments to backward() when y is scalar.

Calling backward() without arguments computes gradients for scalar outputs.

3fill in blank

hard

Fix the error in the code to correctly zero gradients before backward pass.

PyTorch

optimizer.zero_grad()
loss = model(input).sum()
loss.[1]()
optimizer.step()

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

Abackward

Bstep

Czero_grad

Ddetach

Attempts:

3 left

💡 Hint

Common Mistakes

Calling step() or zero_grad() instead of backward() on loss.

The backward() function computes gradients for the loss.

4fill in blank

hard

Fill both blanks to create a simple training loop that updates model parameters.

PyTorch

for data, target in dataloader:
    optimizer.[1]()
    output = model(data)
    loss = criterion(output, target)
    loss.[2]()
    optimizer.step()

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

Azero_grad

Bbackward

Cstep

Ddetach

Attempts:

3 left

💡 Hint

Common Mistakes

Forgetting to zero gradients before backward pass.

Before computing gradients, zero them with zero_grad(). Then compute gradients with backward().

5fill in blank

hard

Fill all three blanks to define a tensor, compute a function, and get its gradient.

PyTorch

x = torch.tensor([1.0, 2.0, 3.0], requires_grad=[1])
y = x.[2](2).sum()
y.[3]()

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

ATrue

Bpow

Cbackward

DFalse

Attempts:

3 left

💡 Hint

Common Mistakes

Not enabling gradient tracking or forgetting to call backward().

Enable gradient tracking with True, use pow(2) to square, then call backward() to compute gradients.