mediumMultiple Select
Generative AI Leader Practice Question: Fine‑tuning a large language model for a…
A company is fine‑tuning a large language model for a domain‑specific task. They have a limited budget and want to minimize the cost of fine‑tuning. Which TWO approaches are most cost‑effective?
⚠ Common exam trap
Candidates often think that larger models or full fine-tuning always yield better results, but the trap here is that cost-effectiveness prioritizes resource efficiency over raw quality, and adapter methods like LoRA provide a practical trade-off that they overlook.
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
✓
Use a smaller base model like Gemini Flash
Option A is correct because using a smaller base model such as Gemini Flash reduces compute and memory requirements, which directly lowers the cost of fine-tuning while still being adequate for many domain-specific tasks. Option D is correct because adapter-based fine-tuning with LoRA freezes the base model weights and trains only small low-rank adapter matrices, dramatically reducing the number of trainable parameters, GPU memory, and training time compared to full fine-tuning. Option B is not cost-effective because full fine-tuning updates all model parameters, requiring substantially more compute, memory, and storage. Option C is not cost-effective because increasing training epochs raises compute time and cost without guaranteeing better results, and can even cause overfitting. Option E is not cost-effective because a larger base model like Gemini Ultra increases training and inference costs significantly, which conflicts with the limited-budget goal.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Use a smaller base model like Gemini Flash
Why this is correct
Choosing a smaller base model such as Gemini Flash directly reduces fine-tuning cost, since compute and memory scale with parameter count. It satisfies the stem's limited-budget constraint by training fewer weights, while still adapting adequately to a narrow domain-specific task where a larger model's extra capacity adds little value.
- ✗
Use full fine‑tuning for better quality
Why it's wrong here
Full fine-tuning updates every parameter, demanding the largest GPU memory and compute spend of any method. Tempting because it can yield the highest quality, but it would be correct when budget is unconstrained and maximum task performance outweighs cost, unlike parameter-efficient alternatives.
- ✗
Increase the number of training epochs
Why it's wrong here
Adding training epochs raises compute time roughly linearly without improving convergence, so cost grows while returns diminish. Tempting because extra passes can lift accuracy on underfit models, but it would be the right lever when a model is clearly undertrained rather than when budget is the binding constraint.
- ✓
Use adapter‑based fine‑tuning (LoRA)
Why this is correct
Adapter-based fine-tuning (LoRA) freezes the base model weights and trains only small low-rank matrices, so gradient computation and optimiser state shrink dramatically. This directly satisfies the stem's limited-budget constraint, cutting GPU memory and compute costs versus full fine-tuning while retaining domain-specific accuracy.
- ✗
Use a larger base model like Gemini Ultra
Why it's wrong here
A larger base model such as Gemini Ultra multiplies training and inference cost, directly contradicting the budget constraint. Tempting because scale often improves quality, but selecting a bigger model would be right when accuracy is the priority and cost is secondary, not when minimising fine-tuning spend.
Go deeper
Related to this question
About these practice questions
Courseiva writes every Generative AI Leader question from scratch — 1,008 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 by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This Generative AI Leader practice question is part of Courseiva's free Google Cloud 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 Generative AI Leader exam.