exam questions

Exam Certified Generative AI Engineer Associate All Questions

View all questions & answers for the Certified Generative AI Engineer Associate exam

Exam Certified Generative AI Engineer Associate topic 1 question 59 discussion

Actual exam question from Databricks's Certified Generative AI Engineer Associate
Question #: 59
Topic #: 1
[All Certified Generative AI Engineer Associate Questions]

A Generative AI Engineer is developing a RAG system for their company to perform internal document Q&A for structured HR policies, but the answers returned are frequently incomplete and unstructured. It seems that the retriever is not returning all relevant context. The Generative AI Engineer has experimented with different embedding and response generating LLMs but that did not improve results.

Which TWO options could be used to improve the response quality? (Choose two.)

  • A. Add the section header as a prefix to chunks
  • B. Split the document by sentence
  • C. Use a larger embedding model
  • D. Increase the document chunk size
  • E. Fine tune the response generation model
Show Suggested Answer Hide Answer
Suggested Answer: AD 🗳️

Comments

Chosen Answer:
This is a voting comment (?). It is better to Upvote an existing comment if you don't have anything to add.
Switch to a voting comment New
Hifly_AA
1 month, 2 weeks ago
Selected Answer: AD
A. Add the section header as a prefix to chunks By tagging each chunk with its section heading (e.g. “Vacation Policy,” “Leave of Absence,” etc.), you give the retriever additional semantic signals so it can better match queries to the right policy area. This often dramatically boosts retrieval relevance for structured docs. D. Increase the document chunk size If your chunks are too small, key details may be split across multiple chunks and never surface together in the top-k. Enlarging the chunk window (while still staying under your model’s context limit) lets each chunk carry more contiguous context, so the retriever can return more complete sections. Options B and E address text splitting or generation tuning but won’t solve retrieval gaps; option C (bigger embedding model) was already tried indirectly when swapping embedding LLMs and didn’t fix the core issue of missing context.
upvoted 3 times
...
Community vote distribution
A (35%)
C (25%)
B (20%)
Other
Most Voted
A voting comment increases the vote count for the chosen answer by one.

Upvoting a comment with a selected answer will also increase the vote count towards that answer by one. So if you see a comment that you already agree with, you can upvote it instead of posting a new comment.

SaveCancel
Loading ...