Pytorch_sentdex

Question 1

what is F and nn. Is there a relation between them

Answer 1

Yes, F is function and nn is more of Oops way. Pytorch has this difference due to architectural differences

Question 2

what is the 1st and 2nd step in building the neural network

Answer 2

1st step is to define the structure of NN and next step is to define how the data is passed through the Network

Question 3

what do you understand by flatten

Answer 3

Flattened is that all the data in the 28*28 form is converted into one continuous row

Question 4

Basic code to form and check the layers

Answer 4

Roam -> Pytorch, #nn_layer_initialization

Question 5

why wont this work #nn_data_pass

Answer 5

Because activation function is not defined

Question 6

Name some standard optimizers and does the last layer needs activation function if no y if yes y

Answer 6

ReLu, LeakyRelu, Sigmoid, Tanh. Yes

Question 7

For a probability distribution output, what is the best loss function? What is the dimension that is needed

Answer 7

Log_soft_max always work with multiclass classifier. Dim=1 You can further note that dim = 0.

Question 8

Remember how to define the optimizer, loss

Answer 8

nn_init_forward_optim_loss. Can the loss be defined

Question 9

How can you send a random training data into a nn and how your dimension of input data should look.

Answer 9

X.view(-1,28*28).

Question 10

How to build a network where the 1st few layers are general layers and the rest of the later layers are specific layers

Answer 10

nn_logics_forward_cool.

Question 11

If your output is numbers can you use MSE ?

Answer 11

No better use nll_loss. MSE can be used only for one hot encoded output value [0,0,0,1,0,0,0,0,0,0 ]

Question 12

Why do we need to call zero_grad() in PyTorch?

Answer 12

In PyTorch, we need to set the gradients to zero before starting to do backpropragation because PyTorch accumulates the gradients on subsequent backward passes. This is convenient while training RNNs. So, the default action is to accumulate (i.e. sum) the gradients on every loss.backward() call.

Question 13

TORCH.ARGMAX

Answer 13

Returns the indices of the maximum value of all elements in the input tensor.

Question 14

COde to write one hot vector for multiclass classification

Answer 14

np.eye

Question 15

vol = torch.unsqueeze(vol, 0) # Whats teh meaning of 0

Answer 15

0 represents the lockation at which the new dimension should be added

Question 16

How to convert an array into tensor float in one line

Answer 16

vol = torch.from_numpy(vol_arr).float()

Question 17

How to plot an image data which is in an array form

Answer 17

import matplotlib.pyplot as plt

plt.imshow()

Question 18

We should be aware of three different kinds of numerical values

Answer 18

continuous values (strictly ordered, and a difference
between various values has a strict meaning) Further ratio scale and interval scale
Ordinal Value (small medium large)
categorical values

Question 19

Can you tell the order of the neural network

Answer 19

Epoch
    Forward Pass
    Loss
    Backward pass
    Update Params

Question 20

*params

model(t_un, *params) what is this equivalent to

Answer 20

to pass the elements of params as individual arguments

model(t_un, params[0], params[1])

Question 21

what comes under functional API

Answer 21

whatever doesnot have parameters and need not be defined in init function, which doesnot have parametes that require a gradient to carry. These things can be associated to funtional API

Question 22

through locality and translation invariance

Answer 22

locality

Translation invariace - neifhours