Feature map visualization in PyTorch - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to import the PyTorch library.

PyTorch

import [1]

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

Atorch

Btensorflow

Cnumpy

Dmatplotlib

Attempts:

3 left

💡 Hint

Common Mistakes

Importing TensorFlow instead of PyTorch.

Importing numpy or matplotlib which are not PyTorch.

PyTorch is imported using import torch.

2fill in blank

medium

Complete the code to create a simple convolutional layer in PyTorch.

PyTorch

conv_layer = torch.nn.Conv2d(in_channels=1, out_channels=6, kernel_size=[1])

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

A7

B3

C1

D5

Attempts:

3 left

💡 Hint

Common Mistakes

Using kernel size 1 which is too small for feature extraction.

Using kernel size 3 or 7 which are valid but not the intended answer.

The kernel size is set to 5 to define the filter size in the convolutional layer.

3fill in blank

hard

Fix the error in the code to get the feature maps from the convolutional layer output.

PyTorch

feature_maps = output.[1]()

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

Adetach

Bnumpy

Ctolist

Dcpu

Attempts:

3 left

💡 Hint

Common Mistakes

Using numpy() directly without detaching causes errors.

Using tolist() which is not suitable for tensor operations.

Use detach() to get a tensor that is not tracked for gradients, which is needed before converting to numpy or visualizing.

4fill in blank

hard

Fill both blanks to convert the feature maps to a NumPy array and select the first channel for visualization.

PyTorch

feature_maps_np = feature_maps.[1]().[2][0]

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

Acpu

Bnumpy

Cdetach

Dtolist

Attempts:

3 left

💡 Hint

Common Mistakes

Trying to convert tensor on GPU directly to NumPy causes errors.

Using tolist() instead of numpy().

First move the tensor to CPU with cpu(), then convert to NumPy array with numpy(). Selecting [0] gets the first channel.

5fill in blank

hard

Fill all three blanks to plot the first feature map using matplotlib.

PyTorch

import matplotlib.pyplot as plt
plt.imshow(feature_maps_np[1][2]) 
plt.[3]()

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

A[0]

B.squeeze()

Cshow

Dplot

Attempts:

3 left

💡 Hint

Common Mistakes

Using plot() instead of show() to display the image.

Not squeezing the tensor before plotting causes shape errors.

Use squeeze() to remove extra dimensions, then select the first map with [0], and finally call show() to display the image.