Pytorch Interview Questions (MCQ): Test Your Knowledge!

Which PyTorch function computes the element-wise multiplication of tensors?

torch.mul()

torch.multiply()

torch.prod()

Both a and b

What is the primary function of torchvision in PyTorch?

To visualize tensors

To provide tools and datasets for computer vision

To optimize vision models

To create custom vision layers

What is the purpose of torch.nn.utils.parametrizations.orthogonal in PyTorch?

To create orthogonal matrices

For orthogonal parametrization of linear layers

To implement QR decomposition

To optimize matrix operations

How do you create a tensor with values drawn from a standard normal distribution?

torch.randn(size)

torch.normal(size)

torch.gaussian(size)

torch.standard_normal(size)

What does the torch.transpose() function do?

Flips a tensor along specified dimensions

Swaps two dimensions of a tensor

Rotates a tensor

Reverses the order of elements in a tensor

How can data shuffling be implemented in PyTorch's DataLoader?

By setting shuffle=True

By using torch.shuffle()

By implementing a custom Dataset

By modifying the model architecture

Which of the following is NOT a valid normalization technique in PyTorch?

InstanceNorm3d

BatchNorm2d

LayerNorm

GroupNorm

Which parameter in StepLR specifies the interval for learning rate decay?

step_size

interval

decay_step

lr_step

How does torch.sparse work in PyTorch?

It manages memory sparsely

For operations on sparse tensors

To implement sparse neural networks

To optimize storage for sparse data

What is the main advantage of PyTorch's dynamic computational graphs?

They allow for more flexible and intuitive model building

They are faster than static graphs

They use less memory

They are easier to deploy

Which PyTorch function applies instance normalization?

nn.InstanceNorm1d()

nn.InstanceNorm2d()

nn.InstanceNorm3d()

All of the above

How do you create a tensor with a specific data type?

torch.tensor([1, 2, 3], dtype=torch.float32)

torch.create([1, 2, 3], type=torch.float32)

torch.array([1, 2, 3], dtype=torch.float32)

torch.new([1, 2, 3], dtype=torch.float32)

What is transfer learning in PyTorch?

Transferring data between GPUs

Using pre-trained models for new tasks

Transferring gradients between layers

Moving tensors between devices

What is the primary function of torch.nn.ReLU in PyTorch?

To normalize inputs

To apply rectified linear activation

To compute loss

To optimize weights

What does the requires_grad attribute of a PyTorch tensor control?

Whether to compute gradients for the tensor

Whether the tensor can be modified

Whether the tensor can be moved to GPU

Whether the tensor can be saved to disk

How can you implement custom DataLoader in PyTorch?

By subclassing torch.utils.data.DataLoader

Using the @dataloader decorator

Modifying the iter() method of Dataset

PyTorch doesn't support custom DataLoaders

What is the main advantage of the Adam optimizer?

It's the fastest

It adapts the learning rate for each parameter

It doesn't require hyperparameter tuning

It's immune to vanishing gradients

What is the main purpose of the torch.clamp() function in PyTorch?

To compute gradients

To clip tensor values to a range

To normalize tensor values

To reshape tensors

Which PyTorch method is used to compute gradients during backpropagation?

backward()

backprop()

gradient()

differentiate()

How do you create a tensor with evenly spaced values on a log scale?

torch.logspace(start, end, steps)

torch.log_range(start, end, steps)

torch.logarithmic_space(start, end, steps)

torch.log_linspace(start, end, steps)

Score: 0/20