Free NVIDIA NCA-GENL Actual Exam Questions - Question 13 Discussion

Option D makes the most sense since chunking is about dividing text so retrieval steps handle it better, not rewriting or generating anything new like A or B suggest.

It’s D because chunking helps break info into parts that retrieval models can handle better.

I think we can rule out B and C since chunking isn’t about generating random text or overall LLM training. Between A and D, chunking feels more like splitting the input into pieces rather than rewriting it to fit a window. So if chunking is about dividing text into manageable parts before retrieval, D fits better. But does the question mean those chunks have to be meaningful segments or just any smaller fixed-size bits?

D imo, because chunking helps the model focus on relevant pieces during retrieval rather than rewriting or generating text. A and B don’t really match what RAG uses chunking for.

Maybe D. Chunking fits as it’s about dividing text into smaller pieces so the retriever can handle it better, which makes sense for RAG. It’s definitely not about generating random text (B) or training models (C). Option A feels off since rewriting to fill context seems unrelated to chunking itself—chunking is more about splitting, not rewriting.

Probably D. Chunking usually means breaking text into smaller, manageable parts for better retrieval, not just random splits or rewriting text like in A or B.

Is there any info on how big or what type of text is being chunked here? Just want to be sure if D means splitting by sentences, paragraphs, or something else.