podcasts / Gaurav Sen / episode / question
Question

How many chunks should a RAG pipeline retrieve per query?

Answered from AI Interview #7 - Why is this chatbot failing? · Gaurav Sen · 2026-09-28

Gaurav, the candidate in Gaurav Sen's AI interview #7, says 20 to 25 chunks is too many to pass to the model because it pollutes the context. Retrieving 20 to 25 from the vector database is fine, but a re-ranking and filtering step should cut that down to the 5 to 10 most relevant chunks, with 10 as his working number, before they reach the LLM.

Source: our own transcription of the episode. The full conversation is on Gaurav Sen (2026-09-28). Read the whole episode summary →

▸More from this episode

Q1Why is a RAG chatbot giving inaccurate answers, according to Gaurav Sen's AI interview #7?Q3What does a cross-encoder re-ranker do in a RAG pipeline?Q4Why do vector embeddings miss negation and numeric ranges in search?Q5Is a re-ranker worth the extra latency in RAG?
Transcribe your links. Paste a YouTube, Google Drive or podcast link and WhipScribe gives you the transcript. Transcribe a link →