S02 - Deep Learning

The following activities are required to build a neural network, except ...

• calculating loss

• applying the model

• data preparation

• creating learnable parameters

PyTorch provides two kinds of data abstractions. They are ...

• optimizer

• tensors

• variables

• loss

A tensor containing only one element is called ...

• batch

• matrix

• scalar

• vector

Given data = [5, 7, 9, 11], the code data[:3] gives you which output?

• [5, 7, 9]

• [5, 7, 9, 11]

• [11]

• [9]

Batch of images will give you tensor with how many dimensions?

• 1D

• 2D

• 3D

• 4D

What is the name of the function provided by PyTorch to copy a tensor on the CPU to the GPU?

• numpy()

• cuda()

• grad()

• matmul()

The followings are the components of the Variable class, except ...

• creator

• data

• loss

• grad

For learnable parameters, the 'requires_grad' is set to false.

• True

• False

In Neural Networks, 'weight' is a parameter and 'learning rate' is a hyperparameter.

• True

• False

What is the name of the package in PyTorch that provides various optimizers, such as Adam, RMSprop, and SGD?

• torch.variables

• torch.opt

• torch.rand

• torch.optim