Build Generative AI Solutions with Microsoft Foundry AI-102 Practice Questions (Page 2)

Questions 6–10

1. 6expert · hard

   A team has built a RAG-based prompt flow that answers questions from internal policy documents. During evaluation, they find that the flow's answers are often too verbose and sometimes include information not present in the retrieved documents. They want to improve both conciseness and groundedness. Which combination of changes is most likely to help?

   • AIncrease the number of retrieved chunks to give the model more context.
   • BRemove the grounding data and rely on the model's internal knowledge.
   • CSwitch to a larger model to improve reasoning.
   • DAdd a prompt instruction to answer concisely and only use the provided context, then re-evaluate with groundedness metrics.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   In Microsoft Foundry, what is the purpose of a hub?

   • AIt serves as a central management unit for projects and shared resources
   • BIt is a storage account for model artifacts
   • CIt is a model catalog for deploying models
   • DIt is a runtime environment for prompt flows
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   What is the purpose of evaluating a prompt flow in Microsoft Foundry?

   • ATo deploy the flow to production
   • BTo assess the quality and effectiveness of the flow's outputs
   • CTo debug syntax errors in the flow
   • DTo compile the flow into executable code
   Reveal Solution Discussion Select an answer first
5. 9expert · hard

   A team is developing a prompt flow that generates product descriptions. They have two versions of the flow: one uses a detailed prompt template with many examples, and the other uses a shorter prompt. They want to decide which version to deploy based on quality and cost. What is the best approach?

   • ADeploy the version with the shorter prompt to save cost.
   • BDeploy the version with the detailed prompt because more examples always improve quality.
   • CDeploy both versions and randomly route traffic between them.
   • DEvaluate both versions using a representative dataset and compare quality metrics and token usage.
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A marketing agency uses a prompt flow to generate ad copy. They want to ensure that all outputs follow a consistent brand voice and include a specific call-to-action. What is the most effective way to achieve this?

   • AInclude the brand voice and call-to-action in the system prompt template.
   • BUse a different model for each campaign.
   • CAsk the model to infer the brand voice from previous outputs.
   • DPost-process the output to replace the call-to-action.
   Reveal Solution Discussion Select an answer first