Google Professional Machine Learning Engineer Question 49
Select 3Google Cloud PlatformYou are tasked with building a retrieval-augmented generation (RAG) application for a customer support chatbot using Vertex AI. The application must retrieve relevant documents from a knowledge base stored in BigQuery and use a fine-tuned generative AI model to generate responses. Which combination of steps should you take to implement this solution?
- A
Use BigQuery ML to train a retrieval model and deploy it to Vertex AI.
- B
Use Vertex AI Matching Engine to index and retrieve relevant documents from the BigQuery knowledge base.
- C
Build a custom API to connect the generative AI model with the retrieval system.
- D
Use a pre-trained generative AI model in Vertex AI Model Garden and fine-tune it with your domain-specific data.
- E
Combine the document retrieval system with the fine-tuned generative model in a custom Vertex AI pipeline.
Show answer and explanation
Correct answers: B, D, E
Explanation
To build a RAG application in Vertex AI, you should use Vertex AI Matching Engine for efficient document retrieval from the BigQuery knowledge base, fine-tune a pre-trained generative AI model for accurate response generation, and combine these components in a custom Vertex AI pipeline. This approach ensures scalability, efficiency, and ease of integration without the need for custom APIs.
- A. Incorrect.
Incorrect. While BigQuery ML can be used to train models, it is not the optimal choice for document retrieval in a RAG workflow. Vertex AI Matching Engine is better suited for this task.
- B. Correct.
Correct. Vertex AI Matching Engine is specifically designed for fast and efficient retrieval of relevant documents and can be used to query your BigQuery knowledge base.
- C. Incorrect.
Incorrect. There is no need to build a custom API as Vertex AI provides tools to integrate retrieval and generation components efficiently.
- D. Correct.
Correct. Fine-tuning a pre-trained generative AI model from Vertex AI Model Garden with domain-specific data ensures the chatbot generates accurate and contextually relevant responses.
- E. Correct.
Correct. Combining the retrieval system and the generative model in a Vertex AI pipeline ensures the RAG workflow is automated and operationalized effectively.