NCA-GENL Practice Test 17 — 15 Questions

Question 1

A financial institution is using an LLM to generate investment summaries. To comply with regulations, they must ensure that the model does not produce discriminatory language based on protected attributes. Which Trustworthy AI principle does this requirement primarily address?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Fairness. The requirement to avoid discriminatory language based on protected attributes is a core aspect of fairness in Trustworthy AI. Fairness ensures equitable treatment and non-discrimination, making it the correct principle. Other principles like explainability, robustness, and privacy address different aspects of trustworthiness.

Answer

Robustness

Answer

Privacy

Answer

Explainability

Question 2

A retail company wants to let its support chatbot answer questions using internal policy documents, but executives fear the model will invent policies that do not exist. Which approach most directly reduces fabricated policy answers while keeping responses grounded in the approved documents?

Accepted Answer

Retrieve relevant passages from the approved policy corpus and require the model to answer only from those passages, citing them.. Grounding responses in an approved corpus through retrieval and citation is the most direct control against fabricated policies. Supplying the authoritative passages at inference time limits the model to what the documents actually say, and citations let reviewers verify each claim. Temperature, token limits, and fine-tuning change style, length, or memorized content but do not bind answers to an auditable source of truth.

Answer

Fine-tune the model on the entire policy corpus so the knowledge is baked into the weights.

Answer

Lower the temperature to zero so the model always selects the single most probable token.

Answer

Increase the max_tokens parameter so the model has room to explain its reasoning in full.

Question 3

A healthcare organization is preparing to deploy an LLM-based clinical documentation assistant. The Trustworthy AI review board requires evidence that the model's outputs are safe and reliable before go-live. Which two practices should the team implement to provide this evidence? (Choose two.)

Accepted Answer

Establish a continuous evaluation harness that scores model outputs against a held-out clinical benchmark and tracks metrics over time.. Adversarial red-teaming with clinicians and a continuous evaluation harness together provide both qualitative and quantitative evidence of safety and reliability. Red-teaming uncovers edge-case failures, while the harness tracks performance against a clinical benchmark over time. Combined, they give the review board documented, repeatable assurance that the assistant behaves safely across diverse patient scenarios.

Answer

Increase the model's temperature setting to encourage more creative and varied clinical documentation.

Answer

Fine-tune the model on a small curated dataset of ideal clinical notes and assume the fine-tuning eliminates all unsafe outputs.

Answer

Deploy the model in shadow mode for a single day and rely on informal developer feedback as the sole safety assessment.

Question 4

A team is preparing a stakeholder report on an LLM evaluation run. They must show how the model's accuracy on a question-answering benchmark changes as the temperature parameter is swept from 0.0 to 1.0 in steps of 0.1. Which visualization is MOST appropriate for this single-variable sweep?

Accepted Answer

A line chart with temperature on the x-axis and accuracy on the y-axis.. When a single ordered parameter such as temperature is swept and one metric is recorded, a line chart communicates the trend most directly. It shows monotonic decline, plateaus, or a peak without extra encoding. Stakeholders unfamiliar with the evaluation can read the x-axis as the sampling setting and the y-axis as accuracy, which is exactly the comparison the report requires.

Answer

A stacked area chart of accuracy by temperature.

Answer

A scatter plot of temperature versus accuracy with no connecting line.

Answer

A heatmap of temperature by benchmark category.

Question 5

A retail company is deploying an LLM-based chatbot that answers customer questions about product warranties. The legal team requires that the chatbot never provides legally binding interpretations of warranty terms. Which Trustworthy AI principle is primarily addressed by implementing a content filter that blocks responses containing definitive legal advice?

Accepted Answer

Safety. Safety is the Trustworthy AI principle that ensures AI systems avoid causing harm. By filtering out legally binding warranty interpretations, the chatbot avoids potential legal and financial harm to customers and the company. The other principles address different aspects: accountability is about responsibility, robustness about resilience, and explainability about understanding model decisions.

Answer

Accountability

Answer

Robustness

Answer

Explainability

Question 6

A data scientist is preparing a transformer-based language model for a text summarization task. She notices that the input sequences in her dataset vary widely in length, from a few tokens to several thousand. She decides to set a fixed maximum sequence length and pad shorter sequences with a special token. Which component of the transformer architecture is primarily responsible for handling the positional information of tokens in these sequences?

Accepted Answer

Positional encodings. Positional encodings are essential in transformer models because the self-attention mechanism is permutation-invariant. By adding positional encodings to token embeddings, the model gains awareness of token order, enabling it to handle sequences of varying lengths and padding effectively. This is critical for tasks like summarization where the meaning depends on word order.

Answer

Self-attention mechanism

Answer

Layer normalization

Answer

Feed-forward network

Question 7

An enterprise is deploying an NVIDIA NIM-hosted LLM for internal knowledge management. The security team wants to harden the deployment against prompt injection and jailbreak attempts before go-live. Which two measures should be implemented? (Choose two.)

Accepted Answer

Deploy NeMo Guardrails with input rails that detect and block known jailbreak patterns and instruction-override attempts.. Defending against prompt injection and jailbreaks requires both a runtime detection layer and sound prompt construction. NeMo Guardrails input rails catch known malicious patterns before generation, while strict separation of system instructions from user content prevents injected text from being interpreted as authoritative. The remaining options either weaken security posture, harm reliability, or remove the observability needed to improve defenses over time.

Answer

Increase the model temperature to make outputs less predictable to attackers.

Answer

Disable logging of prompts and responses to reduce the attack surface of the logging system.

Answer

Store API keys in the client-side application to simplify authentication for internal users.

Question 8

A machine learning engineer is training a large language model and notices that the model performs exceptionally well on the training data but poorly on a held-out test set. Which technique is most appropriate to mitigate this issue?

Accepted Answer

Apply dropout regularization. Overfitting is characterized by high training performance and poor test performance. Dropout regularization mitigates this by preventing the network from relying on specific neurons, encouraging more robust features. Increasing model size, reducing data, or training longer would all aggravate the problem. Therefore, dropout is the appropriate technique.

Answer

Reduce the size of the training dataset

Answer

Train for more epochs

Answer

Increase the model's parameter count

Question 9

A healthcare AI team is using NVIDIA NeMo to fine-tune a clinical summarization model. They want to ensure that the model does not inadvertently learn to associate certain demographic groups with negative health outcomes present in the training data. Which technique should they apply during fine-tuning to mitigate this bias?

Accepted Answer

Adversarial debiasing by adding a bias classifier that penalizes demographic predictability. Adversarial debiasing is a targeted method to reduce unwanted correlations between model representations and sensitive attributes. By training an adversary to predict demographics and simultaneously optimizing the main model to fool it, the model learns fairer representations. In clinical summarization, this helps prevent the model from associating certain groups with negative outcomes, directly supporting Trustworthy AI.

Answer

Regularization by adding L2 weight decay to all layers during fine-tuning

Answer

Post-processing calibration by adjusting predicted probabilities per demographic group

Answer

Data augmentation by oversampling examples from underrepresented demographic groups

Question 10

A data scientist is building a dashboard to detect data drift in the input distribution of a production LLM endpoint. They have access to daily embedding vectors of incoming prompts and to the model's output token statistics. Which two visualizations are MOST appropriate for surfacing prompt-distribution drift over time? (Choose two.)

Accepted Answer

A 2D UMAP projection of the day's prompt embeddings colored by day, updated each day.. Drift detection needs a quantitative distance from a reference and a view of where new data lands. The PSI time series gives a thresholded, alertable number computed on embeddings, while the UMAP scatter colored by day reveals the direction and shape of the shift. Surface token counts, finish reasons, and latency all monitor the wrong signal and would either miss semantic drift or fire on unrelated changes.

Answer

A candlestick chart of daily p50 output latency.

Answer

A bar chart of the top-50 most frequent tokens in the day's prompts.

Answer

A pie chart of the model's output token counts by finish reason.

Question 11

A software company is using an LLM to generate code snippets for developers. During testing, they discover that the model sometimes produces code with security vulnerabilities, such as SQL injection flaws. Which Trustworthy AI principle is most directly violated by this behavior?

Accepted Answer

Security. The generation of code with SQL injection vulnerabilities directly violates the security principle of Trustworthy AI, which requires that AI systems do not introduce exploitable weaknesses. Other principles like transparency, accountability, and privacy address different aspects and are not the primary violation in this scenario.

Answer

Privacy

Answer

Transparency

Answer

Accountability

Question 12

A research team is pretraining a transformer on a corpus of 200 billion tokens. They want the model to learn bidirectional context so each token attends to both left and right neighbors during pretraining. Which pretraining objective fits this requirement?

Accepted Answer

Masked language modeling. Bidirectional context means each token's representation is informed by both preceding and following tokens. Masked language modeling achieves this by hiding random tokens and requiring the model to reconstruct them from the full unmasked context, so attention spans the entire sequence. Autoregressive and causal-decoder objectives enforce left-to-right masking, which precludes bidirectional attention.

Answer

Contrastive next-sentence prediction

Answer

Sequence-to-sequence denoising with a causal decoder

Answer

Autoregressive next-token prediction

Question 13

A healthcare analytics team wants to fine-tune an NVIDIA-hosted LLM on patient records. Before training begins, the privacy officer asks what technical measure will prevent the model from memorizing and later reproducing individual patient identifiers. Which measure best addresses this concern?

Accepted Answer

Apply differential privacy during fine-tuning to bound the influence of any single training record.. Memorization of sensitive identifiers is a training-time phenomenon, so the mitigation must operate during fine-tuning. Differential privacy injects noise that mathematically limits how much any individual record can influence the resulting weights, providing a quantifiable privacy guarantee. The other options either worsen memorization or address unrelated concerns such as serving throughput or inference context length.

Answer

Deploy the fine-tuned model behind an API gateway that enforces rate limiting per client.

Answer

Use a larger context window during inference to reduce the need for retrieval of patient data.

Answer

Increase the number of training epochs so the model learns the data distribution more thoroughly.

Question 14

An ML engineer trains a sentiment classifier on 10,000 movie reviews but only 300 are negative. The model predicts positive for nearly every review, including obvious negative ones. Which technique best addresses this class imbalance during training?

Accepted Answer

Apply class weighting in the loss function to penalize errors on the minority class more heavily. With only 300 negative examples against 9,700 positive ones, an unweighted loss is minimized by predicting the majority class. Applying class weights in the loss raises the cost of minority-class errors, forcing the model to learn features that separate negative reviews. This is the most direct and least invasive fix for the skewed decision boundary.

Answer

Normalize the input text by lowercasing and removing stopwords

Answer

Add L2 regularization to all model weights

Answer

Increase the learning rate and train for more epochs

Question 15

A hospital's AI governance team is reviewing an LLM that drafts discharge summaries from patient notes. Clinicians report the model occasionally invents medication dosages that were never prescribed. The team wants a mitigation that constrains generated output to an approved formulary before any text reaches the clinician. Which approach best satisfies this requirement while keeping the LLM in place?

Accepted Answer

Configure NVIDIA NeMo Guardrails with a retrieval-augmented output rail that validates every generated dosage against the approved formulary and blocks non-matching entries.. The requirement is a runtime constraint that prevents unsanctioned dosages from appearing in generated text. NeMo Guardrails output rails can inspect model responses and validate them against an authoritative formulary retrieved at generation time, blocking or correcting mismatches before display. This keeps the existing LLM while adding a deterministic verification layer. Training changes and interface warnings do not provide that pre-delivery gate.

Answer

Retrain the base LLM from scratch on the hospital's historical discharge summaries to eliminate the fabrication behavior.

Answer

Add a disclaimer banner to the user interface stating that all generated dosages must be independently verified by the clinician.

Answer

Increase the model's temperature setting so it explores more phrasing variations when drafting dosages.