Key models overview (GPT, DALL-E, Stable Diffusion) in Prompt Engineering / GenAI - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to create a GPT model instance.

Prompt Engineering / GenAI

model = GPT([1])

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

Alatent_dim=512

Bimage_size=256

Cnum_steps=1000

Dnum_layers=12

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing image_size or latent_dim which are for image models.

GPT models are built with layers, so specifying num_layers is correct.

2fill in blank

medium

Complete the code to generate an image using DALL-E.

Prompt Engineering / GenAI

image = dalle.generate(prompt, [1]=256)

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

Anum_layers

Bimage_size

Cnum_steps

Dlatent_dim

Attempts:

3 left

💡 Hint

Common Mistakes

Using num_layers which is for model architecture, not output size.

DALL-E generates images, so image_size sets the output size.

3fill in blank

hard

Fix the error in the Stable Diffusion sampling code.

Prompt Engineering / GenAI

sample = diffusion.sample(num_steps=[1])

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

A1000

B512

C256

D12

Attempts:

3 left

💡 Hint

Common Mistakes

Using layer counts or image sizes as step counts.

Stable Diffusion typically uses many steps like 1000 for good quality sampling.

4fill in blank

hard

Fill both blanks to define a GPT model with 12 layers and 768 hidden size.

Prompt Engineering / GenAI

model = GPT(num_layers=[1], hidden_size=[2])

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

A12

B768

C256

D1000

Attempts:

3 left

💡 Hint

Common Mistakes

Mixing up hidden size with image size or step count.

GPT models use num_layers and hidden_size to define architecture.

5fill in blank

hard

Fill all three blanks to generate an image with Stable Diffusion using 1000 steps and 512 latent dimension.

Prompt Engineering / GenAI

output = diffusion.generate(steps=[1], latent_dim=[2], image_size=[3])

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

A1000

B512

C256

D12

Attempts:

3 left

💡 Hint

Common Mistakes

Confusing latent_dim with number of layers or image size.

Stable Diffusion uses many steps (1000), a latent dimension (512), and image size (256) for generation.