Databricks Generative AI Engineer Associate Question 49
Single answerYou are building a Retrieval-Augmented Generation (RAG) application that answers customer support questions using your company's knowledge base. During testing, you notice that irrelevant promotional content in the source documents is degrading the quality of the generated responses. What is the most effective step to address this issue?
- A
Use a preprocessing pipeline to filter out irrelevant sections of the documents based on predefined rules.
- B
Increase the size of your language model to handle more diverse types of input.
- C
Rely on the retrieval model to automatically ignore irrelevant content during document selection.
- D
Manually rewrite all source documents to remove any irrelevant content.
Show answer and explanation
Correct answer: A
Explanation
To ensure high-quality responses in a RAG application, it is crucial to filter out extraneous content from source documents during preprocessing. This step ensures that only relevant and high-quality information is fed into the RAG pipeline, improving the overall performance and accuracy of the application.
- A. Correct.
Preprocessing the documents to filter out irrelevant content is a scalable and effective way to improve the quality of the input to the RAG system, ensuring only useful information is included.
- B. Incorrect.
Increasing the size of the language model does not directly address the issue of irrelevant content degrading the response quality. It may also increase costs without solving the root issue.
- C. Incorrect.
The retrieval model may retrieve relevant documents, but it does not guarantee that the selected documents are free from extraneous content. Explicit filtering is still necessary.
- D. Incorrect.
Manually rewriting all source documents is time-consuming, impractical, and not scalable for large datasets.