Image Generation

Question 1

What is the primary goal of a generator in a GAN?

Answer 1

To generate images that can deceive the discriminator

Question 2

Which architecture is commonly used as the generator in Pix2Pix GAN?

Answer 2

U-Net

Question 3

In Conditional GAN (CGAN), how is the label information used?

Answer 3

Added to both the generator and discriminator

Question 4

Which problem in GAN training is characterized by the generator producing a limited variety of outputs?

Answer 4

Mode collapse

Question 5

What is the purpose of progressive growing of GANs?

Answer 5

To start with low-resolution images and gradually grow to high resolution

Question 6

In the diffusion model, what happens during the forward process?

Answer 6

Noise is added gradually to the image

Question 7

What is the role of CLIP in DALL-E 2’s architecture?

Answer 7

To encode images and texts into a shared embedding space

Question 8

In Stable Diffusion, which model is responsible for turning compressed latent codes back into images?

Answer 8

Variational Autoencoder (Decoder part)

Question 9

In the training of the prior in DALL-E 2, the model learns to map:

Answer 9

Text embeddings to image embeddings

Question 10

What does cross-attention enable in diffusion models like Stable Diffusion?

Answer 10

Controlling generation based on text prompts

Question 11

Which problems are commonly encountered when training GANs?
A) Mode collapse
B) Overfitting discriminator
C) Perfect convergence
D) Non-convergence

Answer 11

Mode collapse; Overfitting discriminator; Non-convergence

Question 12

In the Diffusion Process, the reverse process involves:
A) Removing noise gradually
B) Using a U-Net
C) Adding more noise each step
D) Predicting either clean images or noise

Answer 12

Removing noise gradually; Using a U-Net; Predicting either clean images or noise

Question 13

Which of the following techniques use cross-attention?
A) CLIP text-image matching
B) Stable Diffusion conditioning
C) GAN training without labels
D) DALL-E 2 conditioning

Answer 13

Stable Diffusion conditioning; DALL-E 2 conditioning

Question 14

Stable Diffusion components include:
A) Autoencoder (VAE)
B) U-Net
C) Transformer Decoder
D) Text Encoder

Answer 14

Autoencoder (VAE); U-Net; Text Encoder

Question 15

In GANs, the discriminator network is trained to:
A) Generate realistic images
B) Distinguish real from fake images
C) Provide gradients to the generator
D) Upsample noise

Answer 15

Distinguish real from fake images; Provide gradients to the generator

Question 16

Progressive growing of GANs helps by:
A) Speeding up training
B) Stabilizing the generator early
C) Reducing resolution at the end
D) Gradually increasing output resolution

Answer 16

Speeding up training; Stabilizing the generator early; Gradually increasing output resolution

Question 17

Which components are part of DALL-E 2’s generation pipeline?
A) Prior model
B) Diffusion model (Decoder)
C) LSTM text encoder
D) CLIP encoder

Answer 17

Prior model; Diffusion model (Decoder); CLIP encoder

Question 18

Benefits of latent space diffusion (as in Stable Diffusion) include:
A) Higher memory usage
B) Faster generation
C) Lower computation cost
D) High-resolution outputs

Answer 18

Faster generation; Lower computation cost; High-resolution outputs

Question 19

In Super-Resolution GAN (SRGAN), during training:
A) Low-resolution images are upsampled
B) GAN loss is used
C) Noise is directly added to high-res images
D) Discriminator distinguishes between HR and SR images

Answer 19

Low-resolution images are upsampled; GAN loss is used; Discriminator distinguishes between HR and SR images

Question 20

In the diffusion model reverse process, each denoising step depends on:
A) Previous denoising result
B) Original input image
C) Text embeddings (if conditioned)
D) Random noise injection

Answer 20

Previous denoising result; Text embeddings (if conditioned)