AI-103: Developing AI Apps and Agents on AzureDemo

By Microsoft · Question Mode

QuestionQ71

Implement generative AI and agentic solutions

You have a Microsoft Foundry project containing a Retrieval Augmented Generation (RAG) solution. You need to run a pre-production evaluation using a labeled CSV dataset containing the query, context, response, and ground truth. The evaluation must measure:

  • Whether responses address the user query
  • Whether responses are supported by the provided context
  • Whether responses contain sensitive or proprietary information

Which AI quality evaluation metrics should you use?

Select
To measure whether the responses are supported by the provided context and address the user query:
To measure whether responses contain sensitive or proprietary information:
Explanation

In Azure AI Foundry (formerly Azure AI Studio) evaluation SDK, the Groundedness metric measures whether a response is factually consistent with and supported by the provided context, while the Relevance metric measures how well the response addresses the user's query. Together these directly satisfy the requirement to check both context support and query relevance. For detecting sensitive or proprietary content, the Protected material evaluator/metric checks generated content against known copyrighted or proprietary text, songs, or other protected material. The other options (Hateful and unfair content, Indirect attack, Violent content) are content-safety risk evaluators unrelated to proprietary/copyrighted material detection, and Coherence/Fluency, GPT similarity/F1 score, and ROUGE score measure linguistic quality or similarity to ground truth rather than context grounding or query relevance.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!