AI0-001 AI Concepts and Techniques Practice Question
A team is fine-tuning a BERT model for a document classification task. They notice the model achieves high F1 scores on the training set but low F1 on the validation set. Which regularization technique would be MOST effective?
⚠ Common exam trap
AI0-001 often tests the misconception that L2 regularization is the go-to fix for overfitting in deep learning — candidates overlook that dropout is already embedded in transformer architectures and is the more targeted regularization technique for BERT fine-tuning.
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
✓
Dropout
Dropout is the most effective regularization for transformer-based models like BERT because it randomly deactivates neurons during training, forcing the network to learn redundant representations and preventing co-adaptation. BERT already includes dropout layers in its architecture (attention dropout, hidden dropout), and increasing or tuning the dropout rate directly addresses overfitting on the training set.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
L1 regularization
Why it's wrong here
L1 regularization drives some weights to zero, yielding sparsity rather than controlling the capacity that lets BERT memorise the training set. It suits feature selection with linear models. Dropout directly limits co-adaptation across transformer layers, reducing the train-validation gap.
- ✗
L2 regularization
Why it's wrong here
L2 regularization shrinks weights toward zero but does not stop a fine-tuned BERT from memorising the small training set, so the train-validation gap persists. It suits mild overfitting in smaller models. Dropout on the classifier and transformer layers directly reduces memorisation here.
- ✓
Dropout
Why this is correct
Dropout randomly deactivates neurons during each training pass, forcing the network to learn redundant, distributed representations rather than memorising training samples. This directly targets the overfitting gap between high training F1 and low validation F1 described in the stem, making it the most effective regulariser for fine-tuned BERT classification.
- ✗
Reduce batch size
Why it's wrong here
Reducing batch size alters gradient noise and optimisation dynamics; it does not penalise memorisation of training samples, so the F1 gap remains. Smaller batches suit memory-constrained training or generalisation tuning. Dropout is the regularisation technique that directly combats overfitting.
About these practice questions
One of 962 original AI0-001 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 and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official CompTIA exam blueprint
This AI0-001 practice question is part of Courseiva's free CompTIA 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 AI0-001 exam.