ACorrect
Output indicator
Correct. An output indicator is a prompt component that, like 'Assistant:', indicates where and in what format the output should begin. It helps shape the format of the response.
A developer is organizing the role of each component of a prompt. What is the name of the component that, like 'Human:' or 'Assistant:', indicates where the model's response begins and its format, that is, the start position and format of the output?
1 / 1
Select an answer
CorrectA
Explanation
Choosing the prompt component that indicates the response start position and format.
- 1「like 'Human:' or 'Assistant:'」Indicates the delimiter/label of the response
- 2「indicates where the model's response begins and its format」Specifying the response format = output indicator
BIncorrect
Instruction
An instruction is a component that tells the model what you want it to do.
It conveys the content of the task, not the start position and format of the output, so it is incorrect.
CIncorrect
Context
Context is a component that provides background, situation, and reference information.
It is not the specification of the output start position or format, so it is incorrect.
DIncorrect
System prompt
A system prompt is a component that defines tone, role, and policy in advance.
It defines the overall policy, not a marker that indicates where the response begins, so it is incorrect.
Key Takeaway
Identify the prompt components with a concrete example. Consider a prompt like this:
・'Classify the following review as positive/negative.' ← instruction (what you want done)
・'Review: This movie was the best' ← context (background/reference information)
・'Classification:' or 'Assistant:' ← output indicator (a marker showing where and in what format the output should begin)
The output indicator clarifies the response start position and format to shape the output. Learning rate, epochs, and checkpoints are 'training' terms, not prompt components.