AI-102 Question 485
Single answerYou are implementing an Azure OpenAI model to generate responses based on proprietary company documents. You want to ensure the model uses your custom data effectively without altering the underlying OpenAI model. Which approach should you take?
- A
Fine-tune the Azure OpenAI model with your proprietary data.
- B
Embed your proprietary data into a vector database and use it with the Azure OpenAI model via prompt engineering.
- C
Replace the base OpenAI model with your proprietary data to create a custom AI model.
- D
Use Cognitive Search to index your proprietary data and integrate it with the Azure OpenAI model for retrieval-augmented generation.
Show answer and explanation
Correct answer: B
Explanation
The best approach to use your own data with an Azure OpenAI model is to embed the proprietary data into a vector database and integrate it with prompt engineering. This allows the model to retrieve relevant, contextual information from your data without the need for fine-tuning or altering the base model. Retrieval-augmented generation (RAG) is a common pattern for this scenario.
- A. Incorrect.
Fine-tuning allows you to train the model with new data, but it is resource-intensive and not always necessary when working with retrieval-augmented generation (RAG) or prompt engineering.
- B. Correct.
Embedding your proprietary data into a vector database and using it with prompt engineering enables retrieval-augmented generation, where the model retrieves relevant data from your custom dataset to generate more accurate and contextual responses.
- C. Incorrect.
Replacing the base OpenAI model with proprietary data is not a valid approach and is not supported by Azure OpenAI services.
- D. Incorrect.
While Cognitive Search can index and retrieve proprietary data, it needs to be combined with a vector database or prompt engineering for effective integration with the Azure OpenAI model.