AWS AI Practitioner Foundation Model Concepts

1. What is a token in the context of foundation models?

A security credential for API access

A basic unit of text (word, subword, or character) processed by the model

A database transaction identifier

A storage container for model weights

2. Amazon Bedrock requires you to manage and provision underlying compute infrastructure.

True

False

3. Hallucinations in foundation models are completely eliminated through fine-tuning.

True

False

4. Foundation models trained on diverse data can perform well across many tasks without task-specific retraining.

True

False

5. What does 'retrieval-augmented generation' (RAG) allow a foundation model to do?

Generate images from text descriptions

Retrieve and incorporate external knowledge into generated responses

Train models faster without any data

Eliminate the need for model evaluation

6. Which technique helps reduce model bias in foundation models?

Training on larger datasets without filtering

Using diverse training data and implementing fairness evaluations

Removing all evaluation metrics

Training models only on historical data

7. What is the primary advantage of using Amazon Bedrock for foundation models?

It requires you to build and train models from scratch

It provides serverless access to multiple foundation models via API without managing infrastructure

It only supports text-based models

It eliminates the need for prompt engineering

8. Which approach involves using a foundation model to generate synthetic training data for another model?

Distillation

Data augmentation

Transfer learning

Ensemble learning

9. What does 'context window' refer to in foundation models?

The maximum number of tokens the model can process in a single input

The GPU memory allocation for training

The time taken to process one inference request

The number of model parameters in billions

10. Which AWS service helps detect and mitigate harmful content in AI model outputs?

Amazon Comprehend

Amazon Macie

Amazon Bedrock Guardrails

AWS WAF

11. What is a foundation model?

A large-scale model trained on broad data, adaptable to many downstream tasks

A small model designed for a single specific task

A model that requires manual feature engineering for each use case

A traditional rule-based system without machine learning

12. Which of the following is a benefit of using foundation models over traditional machine learning?

They always require more computational resources

They can leverage transfer learning to reduce training time and data requirements

They cannot be customized for specific tasks

They eliminate the need for any data preprocessing

13. What does the term 'hallucination' mean in the context of foundation models?

When a model generates plausible but factually incorrect information

When a model processes visual data

When a model trains faster than expected

When a model requires human feedback

14. Which AWS service is primarily used for building, training, and deploying machine learning models?

Amazon Bedrock

Amazon SageMaker

AWS Glue

Amazon Redshift

15. What is responsible AI in the context of foundation models?

Using AI only for profit maximization

Ensuring AI systems are fair, transparent, and aligned with ethical principles

Removing all safeguards from model outputs

Training models exclusively on proprietary data

16. Which technique allows you to provide examples within the prompt to improve model performance?

Zero-shot learning

Few-shot learning

Batch processing

Distributed training

17. What is the primary purpose of fine-tuning a foundation model?

To increase the model's file size

To adapt a pre-trained model to specific downstream tasks or domains

To remove all learned knowledge from the base model

To replace the model's original training data

18. Which of the following is a key characteristic of transformer-based foundation models?

They process sequential data one token at a time without parallelization

They use attention mechanisms to capture relationships across tokens

They require labeled data for every training example

They cannot handle variable-length inputs

19. What does prompt engineering involve?

Writing code to build machine learning models from scratch

Crafting input instructions to guide model outputs effectively

Designing database schemas for data storage

Creating infrastructure for model deployment

20. Which AWS service provides access to third-party foundation models through a unified API?

Amazon SageMaker

Amazon Bedrock

AWS Lambda

Amazon EC2

AWS AI Practitioner Foundation Model Concepts Quiz

1. What is a token in the context of foundation models?

2.

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

2. Amazon Bedrock requires you to manage and provision underlying compute infrastructure.

3. Hallucinations in foundation models are completely eliminated through fine-tuning.

4. Foundation models trained on diverse data can perform well across many tasks without task-specific retraining.

5. What does 'retrieval-augmented generation' (RAG) allow a foundation model to do?

6. Which technique helps reduce model bias in foundation models?

7. What is the primary advantage of using Amazon Bedrock for foundation models?

8. Which approach involves using a foundation model to generate synthetic training data for another model?

9. What does 'context window' refer to in foundation models?

10. Which AWS service helps detect and mitigate harmful content in AI model outputs?

11. What is a foundation model?

12. Which of the following is a benefit of using foundation models over traditional machine learning?

13. What does the term 'hallucination' mean in the context of foundation models?

14. Which AWS service is primarily used for building, training, and deploying machine learning models?

15. What is responsible AI in the context of foundation models?

16. Which technique allows you to provide examples within the prompt to improve model performance?

17. What is the primary purpose of fine-tuning a foundation model?

18. Which of the following is a key characteristic of transformer-based foundation models?

19. What does prompt engineering involve?

20. Which AWS service provides access to third-party foundation models through a unified API?