Task 2.2: Implement Model Deployment Strategies. AIP-C01 Practice Questions (Page 5)

Questions 21–25

1. 21foundation · easy

   Which model loading strategy loads a model into memory only when it is first needed, rather than at application startup?

   • AQuantization
   • BLazy loading
   • CModel sharding
   • DEager loading
   Reveal Solution Discussion Select an answer first
2. 22application · medium

   A company is building a customer support chatbot that handles a high volume of routine queries and a smaller number of complex queries. They want to minimize costs while maintaining high accuracy for complex queries. Which deployment strategy should they use?

   • AImplement API-based model cascading, routing routine queries to a smaller model and complex queries to a larger model.
   • BUse a smaller model for all queries to minimize cost.
   • CDeploy the large model to a SageMaker endpoint and the smaller model to Lambda.
   • DUse a single large foundation model for all queries to ensure accuracy.
   Reveal Solution Discussion Select an answer first
4. 23expert · hard

   A company is building a document summarization service that handles a mix of short and long documents. They want to minimize costs while maintaining high-quality summaries for complex documents. The team is considering API-based model cascading. Which routing strategy should they implement?

   • ARoute short documents to a smaller model and long documents to a larger model.
   • BUse a single model with a dynamic context window to handle all documents.
   • CRoute all documents to a smaller model to minimize cost.
   • DRoute all documents to a large model to ensure high-quality summaries.
   Reveal Solution Discussion Select an answer first
5. 24application · medium

   A media company has an image captioning service that is triggered by new images uploaded to an S3 bucket. The service uses a foundation model in Amazon Bedrock. The workload is sporadic, with bursts of activity when images are uploaded. Which deployment strategy should they use?

   • AUse an AWS Lambda function triggered by S3 events to invoke the Bedrock model.
   • BProvision Bedrock provisioned throughput to handle the peak upload volume.
   • CDeploy the model to a SageMaker endpoint with auto scaling.
   • DUse a container-based deployment on ECS with a scheduled scaling policy.
   Reveal Solution Discussion Select an answer first
6. 25foundation · easy

   What is a key consideration when deploying large language models (LLMs) in containers?

   • AThe container must use a specific operating system to run LLMs
   • BThe container must be deployed on a serverless platform to avoid GPU costs
   • CThe container must have access to sufficient GPU memory to hold the model weights
   • DThe container must be deployed in a VPC to access the internet
   Reveal Solution Discussion Select an answer first