NCA-GENL Experimentation Practice Question
Exhibit
{
"layer_norm": "RMSNorm",
"activation": "SwiGLU",
"dropout": 0.1,
"weight_decay": 0.01,
"optimizer": "AdamW"
}Refer to the exhibit. You are experimenting with a model and find the validation loss is increasing while training loss decreases. Which parameter should you adjust first?
⚠ Common exam trap
Candidates frequently try to fix overfitting by adjusting the learning rate or adding more training epochs, rather than directly applying regularization techniques like dropout.
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
✓
Increase dropout
The divergence between training loss and validation loss is a classic sign of overfitting. Increasing the 'dropout' value is a highly effective way to introduce noise during training, preventing the model from relying on specific neuron activations. This forces more robust feature learning and is a standard first-line defense in the experimentation process to bridge the gap between training and validation 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.
- ✗
Change activation function to ReLU
Why it's wrong here
Changing the activation function is a major architectural change that would likely degrade the model's performance on modern architectures like Llama. ReLU is generally less effective than SwiGLU for deep models. The current overfitting problem should be addressed through regularization, not by changing the fundamental activation layer design.
- ✓
Increase dropout
Why this is correct
Increasing dropout is a direct way to regularize the model. By randomly dropping neurons during training, you ensure the model doesn't over-rely on any single path, helping it generalize better to unseen data. This is the most appropriate first step when validation loss shows signs of training-time overfitting.
- ✗
Change optimizer to SGD
Why it's wrong here
AdamW is the standard optimizer for LLMs. Switching to SGD, which is less efficient for large-scale models, would likely lead to slower, less effective convergence. It would not address the generalization issue between the training and validation sets, making it an ineffective strategy for fixing the observed overfitting behavior.
- ✗
Decrease weight decay
Why it's wrong here
Decreasing weight decay would reduce regularization, which is the opposite of what is needed when the model is overfitting. If validation loss is rising, you need to increase, not decrease, the constraints on the model's weight growth. This action would only worsen the observed overfitting gap during the training.
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.