A data analyst at MetroVoyage tests a foundation model by writing the prompt "Translate the phrase 'good evening' into Italian." The instruction is given directly and no example translations are inclu

Question

Options

• AFew-shot prompting
• BRole prompting
• CZero-shot prompting
• DOne-shot prompting

• B
  4% (1)
• C
  88% (23)
• D
  8% (2)

Explanation

Zero-shot prompting (C) means giving a model a task instruction with no examples - the model relies entirely on its pre-trained knowledge to respond. The analyst's prompt 'Translate the phrase good evening into Italian' provides a direct instruction and nothing else, which is the definition of zero-shot. Few-shot prompting (A) would include multiple example input-output pairs before the actual task (e.g., 'good morning → buongiorno, good night → buonanotte, now translate good evening'). One-shot prompting (D) would include exactly one example. Role prompting (B) involves assigning the model a persona or role (e.g., 'You are a professional Italian translator') before the task. No examples and no role assignment means this is zero-shot.

No community discussion yet for this question.