Generative AI Application Basics Quiz

1. What is the primary advantage of using diffusion models over GANs for image generation?

Diffusion models are always smaller in file size

Diffusion models offer more stable training and better sample quality

Diffusion models do not require any labeled data

Diffusion models work exclusively with black-and-white images

2. Which technique helps prevent mode collapse in generative adversarial networks?

Using Wasserstein distance and gradient penalties

Increasing the batch size indefinitely

Training only the discriminator network

Removing all data augmentation techniques

3. In the context of generative AI, what does 'temperature' control in sampling?

The physical heat generated by the hardware running the model

The randomness or diversity of the generated output

The duration of model training in hours

The percentage of data used for validation

4. True or False: Generative models always require massive amounts of labeled data to perform well.

True

False

5. What is the main limitation of autoregressive language models?

They cannot understand semantic meaning

They generate text sequentially, making inference slow

They require labeled data for every training example

They are unable to perform any downstream tasks

Submit

7. Which of the following is a common application of generative models in data science?

Imputing missing values in datasets

Generating synthetic training data to address class imbalance

Creating realistic samples for model evaluation

All of the above

8. What ethical concern is most relevant when deploying generative AI models at scale?

Models consuming too much electricity during inference

Risk of generating biased, misleading, or harmful content

Difficulty in saving model checkpoints to disk

Inability to use open-source frameworks

9. True or False: Transfer learning in generative AI always requires retraining from scratch.

True

False

10. In NLP, what does 'tokenization' represent?

Breaking text into smaller units like words or subwords for model input

Converting numerical predictions back into human-readable text

Assigning fixed numerical IDs to each unique word in a corpus

Removing punctuation and special characters from text

11. What is the primary difference between discriminative and generative models in machine learning?

Discriminative models learn P(y|x); generative models learn P(x,y)

Generative models are always faster than discriminative models

Discriminative models can only classify binary outcomes

Generative models require less training data than discriminative models

12. True or False: Generative models can only be used for creating synthetic images.

True

False

13. Which loss function is typically used to train generative adversarial networks (GANs)?

Cross-entropy loss applied to the generator only

Adversarial loss between generator and discriminator networks

Mean squared error on unlabeled data

Hinge loss with L2 regularization

14. What is a variational autoencoder (VAE) primarily used for?

Learning interpretable latent representations and generating new samples

Classifying high-dimensional data into predefined categories

Compressing data without any information loss

Detecting anomalies using only labeled outlier examples

Submit

16. What does 'prompt engineering' involve in generative AI applications?

Designing and refining input text to elicit better model outputs

Writing the source code for training generative models

Optimizing the hardware used to run generative models

Creating new neural network architectures from scratch

17. Which technique is commonly used to reduce hallucinations in large language models?

Increasing model size indefinitely

Retrieval-augmented generation (RAG) with external knowledge sources

Using only rule-based systems instead of neural networks

Removing all training data containing subjective information

18. What is the primary purpose of the attention mechanism in generative models?

To reduce the number of parameters in the model

To weigh the importance of different input elements when generating output

To ensure the model only uses historical data

To prevent overfitting on training data

19. In the context of large language models, what does 'fine-tuning' refer to?

Adjusting hyperparameters like learning rate and batch size only

Further training a pre-trained model on task-specific data

Removing less important parameters from a trained model

Converting a model to a smaller format for deployment

20. Which of the following best describes a transformer architecture?

A neural network that uses only convolutional layers for sequence processing

A model using self-attention mechanisms to process sequences in parallel

A traditional recurrent neural network optimized for language tasks

An ensemble method combining multiple generative models

DataAI Generative AI Application Basics Quiz

1. What is the primary advantage of using diffusion models over GANs for image generation?

2.

What first name or nickname would you like us to use?

2. Which technique helps prevent mode collapse in generative adversarial networks?

3. In the context of generative AI, what does 'temperature' control in sampling?

4. True or False: Generative models always require massive amounts of labeled data to perform well.

5. What is the main limitation of autoregressive language models?

6. The process of reducing a generative model's size for deployment is called ____.

7. Which of the following is a common application of generative models in data science?

8. What ethical concern is most relevant when deploying generative AI models at scale?

9. True or False: Transfer learning in generative AI always requires retraining from scratch.

10. In NLP, what does 'tokenization' represent?

11. What is the primary difference between discriminative and generative models in machine learning?

12. True or False: Generative models can only be used for creating synthetic images.

13. Which loss function is typically used to train generative adversarial networks (GANs)?

14. What is a variational autoencoder (VAE) primarily used for?

15. In data science, generative models can be used for ____.

16. What does 'prompt engineering' involve in generative AI applications?

17. Which technique is commonly used to reduce hallucinations in large language models?

18. What is the primary purpose of the attention mechanism in generative models?

19. In the context of large language models, what does 'fine-tuning' refer to?

20. Which of the following best describes a transformer architecture?