wayground logo

Free Printable Worksheets

NEW

Font size

S
M
L
XL
Worksheets

Essentials of GenAI

Total questions: 15

Worksheet time: 8mins

Name
Class
Date
1.

What is the primary goal of Generative AI?

a)

Classifying data into categories

b)

Generating new data similar to training data

c)

Reducing computational costs

d)

Optimizing reinforcement learning

2.

Which of these is a generative model?

a)

SVM (Support Vector Machine)

b)

Random Forest

c)

GAN (Generative Adversarial Network)

d)

Logistic Regression

3.

In a GAN, the Generator’s role is to

a)

Discriminate between real and fake data

b)

Generate synthetic data to fool the Discriminator

c)

Compress input data

d)

Classify images

4.

The Discriminator in a GAN is responsible for

a)

Creating fake samples

b)

Distinguishing real data from generated data

c)

Reducing data dimensions

d)

Training the Generator directly

5.

Autoencoders are mainly used for:

a)

Supervised learning

b)

Unsupervised representation learning

c)

Reinforcement learning

d)

Time-series prediction

6.

Which part of an Autoencoder reduces input dimensionality?

a)

Decoder

b)

Encoder

c)

Latent space

d)

Loss function

7.

How does a VAE differ from a standard Autoencoder?

a)

VAEs have no decoder

b)

VAEs use probabilistic latent variables

c)

VAEs are only for classification

d)

VAEs don’t use neural networks

8.

The latent space in a VAE is

a)

Fixed and non-learnable

b)

Deterministic

c)

Probabilistic (e.g., Gaussian distribution)

d)

Discrete

9.

Which loss function is used in VAEs?

a)

Only Mean Squared Error (MSE)

b)

Cross-Entropy Loss

c)

KL Divergence + Reconstruction Loss

d)

Hinge Loss

10.

"Mode collapse" in GANs refers to

a)

The generator producing diverse outputs

b)

The generator producing limited varieties of samples

c)

The discriminator failing completely

d)

The model overfitting

11.

Which of these is NOT a generative model?

a)

VAE

b)

GAN

c)

Autoencoder

d)

k-Nearest Neighbors (k-NN)

12.

What does the "Adversarial" in GAN refer to?

a)

Cooperation between Generator and Discriminator

b)

Competition between Generator and Discriminator

c)

Reinforcement learning

d)

Data augmentation

13.

In VAEs, the "reparameterization trick" is used to:

a)

Generate higher-resolution images

b)

Enable backpropagation through random sampling

c)

Reduce training time

d)

Remove the need for a decoder

14.

Which architecture is best for generating realistic images?

a)

Autoencoder

b)

VAE

c)

GAN

d)

Decision Tree

15.

Which model explicitly learns a probability distribution?

a)

Standard Autoencoder

b)

VAE

c)

GAN

d)

CNN