NCP-GENL Prompt Engineering Practice Question
A team is designing prompts for an NVIDIA NIM-hosted LLM that must produce concise, citation-backed answers from retrieved documents. They want to improve factual grounding and reduce unsupported claims. Which two prompt engineering practices best support this goal? (Choose two.)
⚠ Common exam trap
The trap here is treating creativity or general knowledge as helpful, when citation-backed grounding requires restricting answers to retrieved evidence and permitting refusal.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Instruct the model to cite the specific document ID or snippet for each claim it makes.
Factual grounding in retrieval-augmented generation improves when the model must cite sources and when it is allowed to refuse when evidence is missing. Citations create traceability, and a refusal fallback prevents guessing. Allowing general knowledge, raising temperature, or stripping identifiers weakens provenance and increases unsupported claims.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Remove document identifiers from the context to simplify the prompt and reduce tokens.
Why it's wrong here
Removing document identifiers makes citations impossible and weakens traceability, since the model cannot reference which source supports a claim. Simplifying the prompt at the cost of provenance undermines the citation requirement. Identifiers are needed so the model and reviewers can map statements back to specific retrieved passages.
- ✓
Instruct the model to cite the specific document ID or snippet for each claim it makes.
Why this is correct
Requiring citations forces the model to tie statements to retrieved sources, making unsupported claims easier to detect and reducing free-form invention. It also gives reviewers a way to verify answers. This practice directly supports factual grounding because the model must reference evidence rather than rely on parametric memory.
- ✗
Increase temperature to encourage the model to synthesize multiple documents creatively.
Why it's wrong here
Higher temperature increases variability and the likelihood of unsupported or fabricated statements, which works against factual grounding. Creative synthesis is not desirable when answers must be citation-backed. This setting would make hallucinations more frequent and harder to trace to a source.
- ✓
Instruct the model to say 'Insufficient evidence' when the retrieved documents do not contain the answer.
Why this is correct
A refusal fallback prevents the model from filling gaps with invented content when retrieval fails. It makes the boundary of available evidence explicit and keeps answers within the supported set. This practice is essential for citation-backed responses because it stops the model from guessing when no source exists.
- ✗
Allow the model to answer from general knowledge when retrieved documents are incomplete.
Why it's wrong here
Permitting general knowledge when retrieval is incomplete reintroduces the risk of unsupported claims and contradicts the goal of citation-backed answers. It gives the model an escape hatch to fabricate plausible details. Grounding requires restricting answers to retrieved evidence, not expanding the knowledge sources.
About these practice questions
One of 352 original NCP-GENL practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official NVIDIA exam blueprint
This NCP-GENL practice question is part of Courseiva's free NVIDIA certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the NCP-GENL exam.