AIF-C01 Applications of Foundation Models Practice Question
A data scientist is fine-tuning a foundation model on SageMaker. They want to prevent overfitting. Which THREE actions can help? (Select THREE.)
⚠ Common exam trap
AWS often tests the misconception that increasing epochs or using a smaller learning rate directly prevents overfitting, when in fact these are hyperparameter tuning strategies that can exacerbate or fail to address overfitting without explicit regularization.
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
✓
Apply dropout
Option A (Apply dropout) is correct because randomly deactivating neurons during training acts as a regularizer that reduces the model's reliance on specific weights, thereby curbing overfitting. Option B (Increase training data size) is correct because more diverse examples give the model a broader signal and reduce the chance it memorizes the limited training set. Option D (Use early stopping) is correct because monitoring validation loss and halting training when it stops improving prevents the model from continuing to fit noise in the training data. Option C (Increase the number of epochs) is not correct because training longer typically worsens overfitting rather than preventing it. Option E (Use a smaller learning rate) is not correct because a smaller learning rate mainly affects optimization stability and convergence speed, not the model's tendency to overfit.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Apply dropout
Why this is correct
Dropout randomly deactivates neurons during each training step, forcing the network to learn redundant representations rather than memorising training samples. This regularisation directly counteracts the overfitting the data scientist observes when fine-tuning the foundation model on SageMaker.
- ✓
Increase training data size
Why this is correct
Increasing the volume of training examples exposes the model to greater feature diversity, reducing its tendency to memorise noise in a small sample. This directly counteracts overfitting, the constraint named in the stem, by lowering variance between training and validation performance.
- ✗
Increase the number of epochs
Why it's wrong here
More epochs let the model memorise training data, worsening overfitting rather than preventing it. It is tempting because additional epochs improve training-set accuracy, and would be correct when the model is underfitting and validation loss is still falling.
- ✓
Use early stopping
Why this is correct
Early stopping halts training once validation loss stops improving, directly countering the overfitting the stem asks you to prevent. By monitoring a held-out validation set and restoring the best checkpoint, it stops the model memorising training noise before that degradation is locked in.
- ✗
Use a smaller learning rate
Why it's wrong here
A smaller learning rate slows convergence but does not itself regularise the model against overfitting. It is tempting because gentler updates can stabilise training, and would be correct when training loss oscillates or diverges due to too large a step size.
Go deeper
Related to this question
About these practice questions
One of 862 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.