hard20 pts · first answer countsragquery-transformationchunking

Generating questions for each chunk

Instead of rewriting queries at search time, a team uses an LLM at ingestion to write 3 likely questions per chunk and indexes them with the chunk. What is the main trade-off?

A.No ingestion cost, but every query now needs three LLM callsB.Higher ingestion cost and index size, but no extra latency at query timeC.It lowers recall, since questions never match user questionsD.It removes the need to store the original chunk text

Discussion 0

Join free to post in the discussion.

Generating questions for each chunk

Discussion 0

Related questions