NCA-GENL Core Machine Learning and AI Knowledge Practice Question
A team deploys a retrieval-augmented generation pipeline and observes that answers frequently cite facts not present in the retrieved passages. They want to reduce this unsupported generation behavior. (Choose two.)
⚠ Common exam trap
The trap here is treating hallucination as a creativity problem and raising temperature, when unsupported claims usually come from weak retrieval or an unconstrained generator.
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
✓
Improve the retriever so that top-ranked passages are more relevant to the query
Unsupported generation in retrieval-augmented pipelines stems from two main sources: the generator lacking relevant evidence, and the generator ignoring the evidence it has. Improving retrieval precision supplies the needed facts, while explicit grounding instructions constrain the model to answer only from context and to abstain otherwise. Together they reduce fabricated claims without degrading answer quality.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Fine-tune the generator on additional unlabeled web text
Why it's wrong here
Continued pretraining on unlabeled web text strengthens the model's parametric memory but does not teach it to rely on retrieved evidence. This can actually increase confident fabrication, because the model has more internal knowledge to draw on when retrieved context is thin or ambiguous.
- ✓
Improve the retriever so that top-ranked passages are more relevant to the query
Why this is correct
Unsupported claims often arise when retrieved context is irrelevant or missing the needed evidence, leaving the model to fill gaps from parametric memory. Raising retrieval precision with better embeddings, hybrid search, or reranking ensures the generator receives passages that actually contain the answer, which measurably reduces hallucinated content.
- ✗
Enlarge the retriever's index to include every document in the enterprise
Why it's wrong here
Indiscriminately expanding the index raises the chance of retrieving loosely related passages that distract the generator, and it does not guarantee better precision. More documents without improved ranking can introduce contradictory or tangential context that the model blends into unsupported answers.
- ✗
Increase the generator's temperature to diversify its outputs
Why it's wrong here
Higher temperature flattens the next-token distribution and increases randomness, which makes the model more likely to invent plausible-sounding but unsupported details. In a factuality-sensitive retrieval pipeline, raising temperature works against grounding and typically worsens hallucination rather than reducing it.
- ✓
Instruct the generator to answer only from the provided context and to abstain otherwise
Why this is correct
Prompt-level grounding instructions explicitly constrain the model to cite or paraphrase the retrieved passages and to state when the answer is absent. This reduces the tendency to fabricate facts and encourages abstention, which is a direct, low-cost mitigation for unsupported generation in retrieval-augmented pipelines.
About these practice questions
Courseiva writes every NCA-GENL question from scratch — 367 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. 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 NCA-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 NCA-GENL exam.