Aws ai official test Q

A company is deploying a solution on AWS to enhance its knowledge base with semantic search
capabilities. The company plans to integrate the solution with Amazon Bedrock. Which AWS service
can the company use to secure access to Amazon Bedrock?

A Amazon Macie

B Amazon Rekognition

C AWS Identity and Access Management (IAM)

D AWS Config

A company wants to increase the consistency and quality of large language model (LLM) responses
by providing the model with access to external sources of knowledge.

Which technique will meet the requirement with the LEAST development effort?

Fine-tuning

Incorrect. Fine-tuning is the process to further train and refine a pre-trained LLM on a smaller,
targeted dataset. The purpose of fine-tuning a pre-trained LLM is to maintain the original capability
of the model and adapt to more specialized use cases. Fine-tuning requires additional development
effort to train the model.

Learn more about fine-tuning.

Retrieval augmented generation (RAG)

Correct. RAG is the process of improving the quality and consistency of LLMs by referencing an
external knowledge base that is outside of the LLM's training data sources. RAG references the
external knowledge base before generating a response. You can use RAG to provide the model with
access to external sources of knowledge with minimal development effort.

Learn more about RAG.

C

In-context learning

Incorrect. In-context learning is the process of providing a few examples to help an LLM better align
responses to an expected format or output. In-context learning is also referred to as few-shot
prompting. In-context learning does not increase the consistency and quality of an LLM by providing
the model with access to external sources of knowledge.

Learn more about few-shot prompting.

Prompt engineering

Incorrect. Prompt engineering is the process of designing and refining the input prompts for an LLM
to generate a specific type of output. Prompt engineering involves selecting appropriate keywords
and shaping the input so that the model can produce your desired outcomes. Prompt engineering
does not increase the consistency and quality of an LLM by providing the model with access to
external sources of knowledge.

A company wants to record API calls that are made to Amazon Bedrock. For compliance purposes,
the company wants these logs to include the API call, the user who made the call, and the time that
the call was made.

Which AWS service will meet these requirements?

Report Content Errors

Amazon Inspector

Amazon CloudWatch
C

AWS Trusted Advisor

AWS CloudTrail

A data science team wants to improve a model's performance. The data science
team wants to increase the number of variables in the training dataset and modify
the behavior of the algorithm.

Which combination of ML pipeline steps will meet these requirements? (Select

TWO.)
Report Content Errors
A
Hyperparameter tuning
B
Model evaluation
C
Feature engineering
D
Model monitoring
E
Data collection
A company wants to assess the performance of a foundation model (FM) for text
generation.

Which technique or metric will meet these requirements?

Report Content Errors
A
Reinforcement learning
B
F1 score
C
Recall-Oriented Understudy for Gisting Evaluation (ROUGE)
D
Fine-tuning
A company has a containerized frontend application for its AI application. The
company must implement a solution to assess its AWS environment's security
posture. The solution must identify potential security vulnerabilities across Amazon
EC2 instances and Amazon Elastic Container Registry (Amazon ECR) repositories
for the application. The solution should provide recommendations for remediation.

Which AWS service will meet these requirements?

Report Content Errors
A
AWS CloudTrail
B
AWS Config
C
Amazon Inspector
D
AWS Artifact
Select the correct AWS service or feature from the following list for each task. Each
AWS service or feature should be selected one or more times. (Select FIVE.)aggag

 Guardrails for Amazon Bedrock

 AWS Identity and Access Management (IAM)
Implement identity verification and resource-level access
AWS Identity and Access Management (IAM)

control.

Set policies to avoid specific topics in a generative AI

Guardrails for Amazon Bedrock

application.

Filter harmful content based on defined thresholds for

Guardrails for Amazon Bedrock

categories.

Define user roles and permissions to access Amazon

AWS Identity and Access Management (IAM)

Bedrock.

Monitor and analyze user inputs to ensure compliance with safety

Guardrails for Amazon Bedrock
policies.
Report Content Errors
A company wants to use generative AI to create product descriptions on its website.

What is a limitation of generative AI that the company should be aware of?

Report Content Errors
A
Generative AI models might produce biased or inappropriate content that
requires human review and editing.
B
Generative AI cannot handle the large volumes of data that is required for
product descriptions.
C
Generative AI cannot generate text in the multiple languages that is
required for an ecommerce website.
D
Generative AI models lack the ability to understand and incorporate
product specifications and details.
Select and order the Amazon SageMaker inference options from the following list
from LOWEST latency to HIGHEST latency. Each SageMaker inference option
should be selected one time. (Select and order THREE.)bca

A: Batch transform

B: Real-time inference

C: Asynchronous inference
B: Real-time inference

1.
C: Asynchronous inference

2.
A: Batch transform

3.

What is a valid data format for instruction-based fine-tuning?

Report Content Errors
A
Images that are labeled with categories
B
Playlists that are curated with recommended music
C
Prompt-response text pairs
D
Audio files with transcriptions
A company uses Amazon SageMaker for its ML models. The company wants to
implement a solution for model owners to create a record of model information. The
model information should include intended uses, risk ratings, training details, and
evaluation results.

Which SageMaker feature will meet these requirements?

Report Content Errors
A
SageMaker Role Manager
B
SageMaker Model Cards
C
SageMaker Model Dashboard
D SageMaker Model Monitor
A company wants to gain insights from diverse data sources and formats to improve
business operations. The data sources include audio from call centers, text feedback
from customers, product images, and scanned documents.
Report Content Errors
Question
The company wants to identify custom labels to categorize the product images
based on historic product images.

Which combination of steps will meet this requirement? (Select TWO.)

A
Create a training model project in Amazon Textract.
B
Create a training model project in Amazon Rekognition.
C
Provide the unlabeled historic images to the model training.
D
Label the historic images by category and provide the labeled images to
the model training.
E
Create a custom model in Amazon Comprehend.
A company wants to gain insights from diverse data sources and formats to improve
business operations. The data sources include audio from call centers, text feedback
from customers, product images, and scanned documents.
Report Content Errors
Question
Which AWS service can detect text and handwriting from invoices that are stored in
PNG format?

A
Amazon Polly
B
Amazon Textract
C
Amazon Kendra
D
Amazon Comprehend
Case Study - 3 Questions
A company wants to gain insights from diverse data sources and formats to improve
business operations. The data sources include audio from call centers, text feedback
from customers, product images, and scanned documents.
Report Content Errors
Question
Which solution will improve transcription accuracy for domain-specific speech?

A
Use a custom bot in Amazon Lex.
B
Use a custom language model in Amazon Translate.
C
Use batch language identification in Amazon Transcribe.
D
Use a custom language model in Amazon Transcribe.
A data scientist notices that a model has high accuracy on training data, but has low
accuracy on testing data.

What is causing these results?

Report Content Errors
A
Not enough training time
B
Underfitting
C
Too much training data
D
Overfitting
A financial company wants to use an open source foundation model (FM) to evaluate
if credit contracts adhere to compliance rules. The company wants to reduce the
human efforts that are performed on audits.

Which AWS service will meet these requirements?

Report Content Errors
A
Amazon SageMaker JumpStart
B
Amazon Textract
C
Amazon Kendra
D
Amazon Q Business
A travel company wants to use a pre-trained generative AI model to generate
background images for marketing materials. The company does not have ML
expertise. Additionally, the company does not want to customize and host the ML
model.

Which AWS service will meet these requirements?

Report Content Errors
A
Amazon Bedrock
B
Amazon SageMaker JumpStart
C
Amazon Rekognition
D
Amazon Personalize
A company is building a generative AI application by using a foundation model (FM).
The company decides to customize its own FM by using proprietary datasets instead
of using a pre-trained FM.

What are the tradeoffs of customizing the FM? (Select TWO.)

Report Content Errors
A
Increased risk of hallucination
B
Reduced accuracy
C
Higher latency
D
Higher cost
E
Higher implementation complexity
A marketing company wants to generate personalized product descriptions for an
ecommerce client's website.

Which prompt engineering technique will meet these requirements with the LEAST
operational effort?
Report Content Errors
A
Few-shot prompting with examples of well-written product descriptions
B
Zero-shot prompting without any examples
C
Fine-tuning to optimize the descriptions based on customer engagement
metrics
D
Continued pre-training on a different domain
What is a foundation model (FM) in the context of generative AI?
Report Content Errors
A
A task-specific model that is trained on a narrow domain, such as finance
or medicine, to serve as a foundation in that area.
B
A large, general-purpose model that is pre-trained on diverse datasets
that can be fine-tuned for downstream tasks.
C
A theoretical framework to understand how different types of models learn
representations.
D
A basic architecture that serves as a starting point to design more
complex neural networks.