AIF-C01 Applications of Foundation Models Practice Question
A data science team is fine-tuning a Llama 2 7B model on Amazon SageMaker for a text classification task. After the first training run, they notice the loss is not decreasing and the model is overfitting to the small training set. What should the team change to mitigate overfitting?
⚠ Common exam trap
The trap here is that candidates may mistakenly think increasing epochs or batch size helps with overfitting, when in fact these changes often worsen it by allowing the model to memorize the training data more thoroughly.
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
✓
Add dropout layers and reduce the learning rate.
Adding dropout layers introduces regularization by randomly dropping neurons during training, which prevents the model from relying too heavily on specific features and reduces overfitting. Reducing the learning rate helps the model converge more smoothly and avoid oscillating around a suboptimal minimum, which is especially important when fine-tuning a large model like Llama 2 7B on a small dataset. Together, these changes address the core issues of overfitting and non-decreasing loss.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Add dropout layers and reduce the learning rate.
Why this is correct
Dropout randomly drops neurons to prevent co-adaptation, and a lower learning rate helps stabilize training, both reducing overfitting.
- ✗
Increase the number of epochs to allow the model to learn more patterns.
Why it's wrong here
More epochs often lead to more overfitting, especially with a small dataset.
- ✗
Increase the batch size and use gradient accumulation.
Why it's wrong here
Increasing batch size can help generalization but is not a direct anti-overfitting technique; overfitting may persist.
- ✗
Remove dropout layers from the model architecture.
Why it's wrong here
Dropout is a regularization technique; removing it increases risk of overfitting.
Go deeper
Related to this question
About these practice questions
One of 619 original AIF-C01 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 by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AIF-C01 practice question is part of Courseiva's free Amazon Web Services 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 AIF-C01 exam.