requires_grad flag in PyTorch - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to create a tensor that tracks gradients.

PyTorch

import torch
x = torch.tensor([1.0, 2.0, 3.0], requires_grad=[1])

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

ANone

BFalse

CTrue

D0

Attempts:

3 left

💡 Hint

Common Mistakes

Using False disables gradient tracking, so gradients won't be computed.

Using None or 0 are invalid values for requires_grad.

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

2fill in blank

medium

Complete the code to stop gradient tracking on a tensor.

PyTorch

import torch
x = torch.tensor([1.0, 2.0, 3.0], requires_grad=True)
x = x.detach()
x.requires_grad = [1]

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

ATrue

BNone

C0

DFalse

Attempts:

3 left

💡 Hint

Common Mistakes

Setting requires_grad to True keeps tracking gradients.

Using None or 0 are not valid boolean values for requires_grad.

Setting requires_grad = False disables gradient tracking for the tensor.

3fill in blank

hard

Fix the error in the code to enable gradient tracking on a new tensor copied from x.

PyTorch

import torch
x = torch.tensor([1.0, 2.0, 3.0])
y = x.clone()
y.requires_grad = [1]

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

ATrue

BNone

CFalse

D0

Attempts:

3 left

💡 Hint

Common Mistakes

Leaving requires_grad as False means no gradients will be tracked.

Using None or 0 are invalid for requires_grad.

After cloning, set requires_grad=True to track gradients on the new tensor.

4fill in blank

hard

Fill both blanks to create a tensor with gradient tracking and then disable it.

PyTorch

import torch
x = torch.tensor([4.0, 5.0], requires_grad=[1])
x.requires_grad = [2]

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

ATrue

BFalse

CNone

D0

Attempts:

3 left

💡 Hint

Common Mistakes

Setting requires_grad to None or 0 causes errors.

Not disabling gradient tracking when needed.

First, create the tensor with requires_grad=True to track gradients, then disable it by setting requires_grad=False.

5fill in blank

hard

Fill all three blanks to create a tensor, clone it with gradients, and then disable gradients on the clone.

PyTorch

import torch
x = torch.tensor([7.0, 8.0], requires_grad=[1])
y = x.clone()
y.requires_grad = [2]
z = y.detach()
z.requires_grad = [3]

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

ATrue

BFalse

CNone

D0

Attempts:

3 left

💡 Hint

Common Mistakes

Not enabling gradients on the clone.

Not disabling gradients after detaching.

Create x with gradient tracking enabled (True), clone it to y which also tracks gradients (True), then detach z from the graph and disable gradients (False).