Diffusion model concept in Prompt Engineering / GenAI - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding the forward diffusion process

In a diffusion model, what best describes the forward diffusion process?

A Training a neural network to classify images

B Removing noise from data to recover the original image

C Gradually adding noise to data until it becomes pure noise

D Compressing data into a smaller representation

💡 Hint

Think about what happens to the data before the model learns to reverse it.

The forward diffusion process gradually adds noise to the original data, turning it into pure noise step by step. The model learns to reverse this process.

❓ Model Choice

intermediate

Choosing the right model for denoising

Which type of neural network architecture is commonly used in diffusion models to predict noise during the reverse process?

A Transformer

B Recurrent Neural Network (RNN)

C Feedforward Neural Network

D Convolutional Neural Network (CNN)

💡 Hint

Think about models good at capturing spatial patterns in images.

CNNs are widely used in diffusion models because they effectively capture spatial features needed to predict noise in images.

❓ Metrics

advanced

Evaluating diffusion model output quality

Which metric is best suited to measure how close the generated images from a diffusion model are to real images?

A Inception Score (IS)

B Mean Squared Error (MSE)

C Accuracy

D Cross-Entropy Loss

💡 Hint

This metric uses a pretrained image classifier to assess image quality and diversity.

Inception Score evaluates both the quality and diversity of generated images by using a pretrained classifier's predictions.

🔧 Debug

advanced

Identifying the cause of blurry generated images

A diffusion model generates images that look blurry and lack details. Which of the following is the most likely cause?

A The noise schedule adds too much noise too quickly

B The model was trained with too many training steps

C The batch size during training was too large

D The learning rate was set too high

💡 Hint

Consider how noise affects the model's ability to learn fine details.

If the noise schedule adds noise too aggressively, the model struggles to recover fine details, resulting in blurry images.

❓ Hyperparameter

expert

Optimizing the number of diffusion steps

What is the effect of increasing the number of diffusion steps in a diffusion model during training and sampling?

A Has no effect on image quality or computation time

B Improves image quality but increases computation time

C Improves image quality and decreases computation time

D Decreases image quality and reduces computation time

💡 Hint

More steps mean more gradual noise addition and removal.

Increasing diffusion steps allows the model to learn finer noise transitions, improving quality but requiring more computation.