PyTorch

Question 1

What are Tensors in PyTorch?

Tensors are data __ similar to a__ and m__.

Answer 1

Tensors are specialized data structure similar to arrays and matrices.

Question 2

In PyTorch, what are Tensors normally used for?

Give 2 uses.

1. Encode the in… and ou…
2. Encode the model’s pa…

Answer 2

1. Encode the inputs and outputs of a model.
2. Encode the model’s parameters.

Question 3

What does Datasets do?

It __ the __ and their __.

Answer 3

It stores the samples and their corresponding labels

Question 4

What does DataLoader do?

It wraps an iterable around the __ to enable __ to the samples.

Answer 4

It wraps an iterable around the Dataset to enable easy access to the samples.

Question 5

What does ‘transforms’ do?

To perform some __ of the __ and make it suitable for __ the model.

Answer 5

To perform some manipulation of the data and make it suitable for training the model.

Question 6

What is torch.autograd?

It enables __ differentiation, making it possible to __ how changes in __ affect the __, essential for __ __ __.

Answer 6

It enables automatic differentiation, making it possible to compute how changes in inputs affect the outputs, essential for optimizing neural networks.

Question 7

What is the typical training procedure for a NN?

• Define the __ that has some learnable __ (or
  weights)
• __ over a dataset of __
• Process __ through the __
• Compute the __
• Propagate __ back into the network’s parameters
• Update the __ of the network, typically using a simple update
  rule: weight = weight - learning_rate * gradient

Answer 7

• Define the neural network that has some learnable parameters (or
  weights)
• Iterate over a dataset of inputs
• Process input through the network
• Compute the loss (how far is the output from being correct)
• Propagate gradients back into the network’s parameters
• Update the weights of the network, typically using a simple update
  rule: weight = weight - learning_rate * gradient