Pre-trained models (ResNet, VGG, EfficientNet) in Computer Vision - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load a pre-trained ResNet50 model from torchvision.

Computer Vision

import torchvision.models as models
model = models.[1](pretrained=True)

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

Aefficientnet_b0

Bresnet50

Cvgg16

Dalexnet

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing a model name that does not match ResNet50, like vgg16 or alexnet.

Forgetting to set pretrained=True to load pre-trained weights.

The resnet50 function loads the ResNet50 pre-trained model from torchvision.

2fill in blank

medium

Complete the code to replace the final classification layer of a VGG16 model for 10 classes.

Computer Vision

import torch.nn as nn
import torchvision.models as models
model = models.vgg16(pretrained=True)
num_ftrs = model.classifier[6].in_features
model.classifier[6] = nn.[1](num_ftrs, 10)

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

ADropout

BConv2d

CReLU

DLinear

Attempts:

3 left

💡 Hint

Common Mistakes

Using Conv2d or ReLU instead of Linear for the final layer.

Not matching the output features to the number of classes.

The final layer of VGG16 classifier is a Linear layer. We replace it with a new Linear layer with 10 outputs for 10 classes.

3fill in blank

hard

Fix the error in the code to load EfficientNet-B0 pre-trained model from torchvision.

Computer Vision

import torchvision.models as models
model = models.[1](weights=models.EfficientNet_B0_Weights.IMAGENET1K_V1)

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

Aefficientnet_b0

Befficientnet_b1

Cresnet50

Dvgg16

Attempts:

3 left

💡 Hint

Common Mistakes

Using resnet50 or vgg16 instead of efficientnet_b0.

Incorrect weights enum for the model.

The correct function to load EfficientNet-B0 is efficientnet_b0. The weights argument uses the correct weights enum.

4fill in blank

hard

Fill both blanks to freeze all parameters of a pre-trained ResNet18 model.

Computer Vision

import torchvision.models as models
model = models.resnet18(pretrained=True)
for param in model.[1]():
    param.[2] = False

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

Aparameters

Bchildren

Crequires_grad

Dgrad

Attempts:

3 left

💡 Hint

Common Mistakes

Using children() instead of parameters() to freeze layers.

Setting grad instead of requires_grad.

We loop over all parameters using parameters() and set requires_grad = False to freeze them.

5fill in blank

hard

Fill all three blanks to create a dictionary that maps model names to their pre-trained loading functions.

Computer Vision

import torchvision.models as models
model_loaders = {
    '[1]': models.[2],
    '[3]': models.vgg16,
}

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

Aresnet50

Cvgg16

Defficientnet_b0

Attempts:

3 left

💡 Hint

Common Mistakes

Using function names as keys instead of strings.

Mismatching keys and function names.

The dictionary keys are model names as strings. The values are the corresponding functions from torchvision.models.