AI-102 Question 484
Select 3You are designing a solution that uses Azure OpenAI Service to generate responses specific to your organization’s proprietary data. Which of the following steps are required to ensure the model incorporates your data without retraining it?
- A
Use Azure Cognitive Search to index your proprietary data and connect it to the Azure OpenAI model.
- B
Upload your proprietary data directly into the Azure OpenAI model for fine-tuning.
- C
Implement an Azure Function to preprocess your proprietary data into a prompt engineering format.
- D
Develop a system to extract relevant data at query time using an external knowledge base.
- E
Enable semantic search within Azure Cognitive Search for better data relevance.
Show answer and explanation
Correct answers: A, D, E
Explanation
To use your own data with an Azure OpenAI model, you can use a retrieval-augmented generation (RAG) approach. This involves indexing your proprietary data using Azure Cognitive Search and enabling semantic search for relevance. At query time, the system retrieves relevant data from the knowledge base and incorporates it into the prompt sent to the Azure OpenAI model, ensuring responses are specific to your data without retraining the model.
- A. Correct.
This is correct because Azure Cognitive Search can index your proprietary data, allowing the Azure OpenAI model to retrieve relevant information without retraining.
- B. Incorrect.
This is incorrect because Azure OpenAI models currently do not support direct uploading of proprietary data for fine-tuning. Fine-tuning is a separate process and not required for retrieval-augmented generation (RAG).
- C. Incorrect.
This is incorrect because preprocessing data for prompt engineering is not a required step when using Azure Cognitive Search with Azure OpenAI for retrieval-augmented generation.
- D. Correct.
This is correct because a retrieval-augmented generation (RAG) approach involves querying an external knowledge base (such as Cognitive Search) to extract relevant data at query time.
- E. Correct.
This is correct because enabling semantic search in Azure Cognitive Search enhances the model's ability to retrieve and rank the most relevant data for user queries.