NCA-GENL Data Analysis and Visualization Practice Question
A data scientist observes that the model's loss plateaus early during fine-tuning. Which visualization would best help diagnose if the model is suffering from 'catastrophic forgetting'?
⚠ Common exam trap
Candidates often choose loss curves or confusion matrices, which track training progress or classification accuracy, but fail to explicitly compare performance across two distinct datasets to detect relative skill degradation.
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
✓
A line chart comparing original and new task performance.
Catastrophic forgetting occurs when a model loses the ability to perform tasks it previously mastered while learning new ones. A line chart comparing the model's performance on the original evaluation set versus the new training set over time is the best visualization. Seeing performance on the original tasks plummet while the new task performance improves confirms the issue, allowing developers to adjust training parameters like lower learning rates or replay buffers to maintain overall performance.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
A histogram of activation values.
Why it's wrong here
Histograms of activations are useful for identifying vanishing or exploding gradients, not for monitoring performance across multiple tasks. They offer information on the internal signal propagation within the network but provide no data on the functional correctness or the retention of previously learned skills in the model.
- ✓
A line chart comparing original and new task performance.
Why this is correct
Tracking performance on both tasks simultaneously is the only way to detect forgetting. By plotting two lines on a single chart, one for the original baseline and one for the new fine-tuning task, you can visually observe when the model begins to sacrifice its previous knowledge to accommodate new information.
- ✗
A pie chart showing weight distribution.
Why it's wrong here
Weight distribution is unrelated to the functional task performance of the model. Pie charts are ineffective for displaying numerical trends and cannot capture the complex relationship between task accuracy and time, making them a poor choice for diagnosing catastrophic forgetting or any other form of model degradation during training.
- ✗
A heat map of the training loss per sample.
Why it's wrong here
Heat maps showing loss per sample are useful for detecting outliers or noisy data but are not designed to compare model behavior across different task categories. While they might show which samples the model struggles with, they do not provide the longitudinal performance comparison needed to detect catastrophic forgetting.
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.