Task 1.2: Select and Configure FMs. AIP-C01 Practice Questions (Page 3)

Questions 11–15

1. 11application · medium

   A global e-commerce company uses Amazon Bedrock to generate product descriptions. They have deployed their application in the us-east-1 Region and want to ensure high availability. They are considering using Amazon Bedrock Cross-Region Inference to automatically route requests to a different Region if the primary Region experiences an outage. Which additional measure should they implement to maintain service continuity?

   • AEnable Cross-Region Inference and also configure a circuit breaker in AWS Step Functions to fail over to a different model or Region if the primary fails.
   • BUse Amazon SageMaker to deploy a custom model in a second Region and route traffic to it using Route 53.
   • CRely solely on Amazon Bedrock's built-in retry logic to handle regional outages.
   • DImplement a fallback to a different foundation model in the same Region using AWS AppConfig.
   Reveal Solution Discussion Select an answer first
2. 12application · medium

   A financial services company uses Amazon Bedrock to power a customer-facing Q&A bot. They want to be able to switch between different foundation models based on cost and latency, and also want to automatically fall back to a different model if the primary model's endpoint becomes unavailable. Which architecture best meets these requirements?

   • AUse Amazon SageMaker Model Registry to manage model versions and deploy a new endpoint for each model, updating the Lambda function to point to the new endpoint.
   • BUse AWS AppConfig to store the active model ID and have the Lambda function read it on each invocation, with a circuit breaker in Step Functions to fail over to a secondary model.
   • CDeploy a separate Lambda function for each model and use Amazon Route 53 to route traffic based on latency.
   • DHard-code the primary model ID in the Lambda function and rely on Amazon Bedrock's built-in retry logic to handle failures.
   Reveal Solution Discussion Select an answer first
4. 13application · medium

   A research organization wants to fine-tune a large language model on a specialized corpus of scientific papers. They have a limited budget and need to deploy the model for inference. They are considering using LoRA for fine-tuning. Which additional step is essential to ensure they can manage the model lifecycle effectively?

   • ARegister the fine-tuned model in Amazon SageMaker Model Registry and set up an automated deployment pipeline.
   • BUse Amazon Bedrock's fine-tuning capability and deploy the model using a SageMaker endpoint.
   • CUse Amazon SageMaker AI with full fine-tuning to ensure the best quality, and register the model in Model Registry.
   • DStore the fine-tuned model artifacts in Amazon S3 and manually deploy them to a SageMaker endpoint.
   Reveal Solution Discussion Select an answer first
5. 14foundation · easy

   Which AWS service is used to catalog, version, and manage machine learning models for deployment?

   • AAmazon S3
   • BAWS AppConfig
   • CAmazon SageMaker Model Registry
   • DAmazon Bedrock
   Reveal Solution Discussion Select an answer first
6. 15foundation · easy

   A company needs to select a foundation model for a customer-support chatbot. Which factor is most important to assess during model selection?

   • AThe model's popularity on social media platforms.
   • BThe model's training data size, assuming larger data always means better performance.
   • CThe model's ability to handle multi-turn conversations and follow instructions accurately.
   • DThe model's maximum token output length, which is the only factor that matters.
   Reveal Solution Discussion Select an answer first