A team wants a model to consistently follow its house style for product descriptions. Which statement correctly contrasts prompt engineering with fine-tuning?
Show answer and explanation
Answer: D
Prompt engineering changes what is sent to a fixed model, such as instructions, examples, or context. Fine-tuning changes the model's weights using additional examples so the house style becomes built in. The two approaches are complementary and differ in what they change. Fine-tuning does require training data, while prompting alone does not.
Why the other options are wrong
- A. Prompting does not require a labeled corpus, since examples can be written inline in the request. Fine-tuning usually needs training data, so this reverses the truth.
- B. The two techniques operate at different points, with prompts changing inputs and tuning changing weights. Calling them identical hides the key distinction.
- C. Fine-tuning changes model weights through training, not the request schema. Prompting does not retrain the model, so this reverses both roles.