A developer of an image generation application is organizing ways to control the output. In image and text generation, what is the name of the prompt specification that explicitly states elements to suppress from the output, such as specific words or styles, so they are not included?
A question that asks for the name of the specification that suppresses unwanted elements from the output.
System prompt
A system prompt is an instruction that sets the overall role and behavior policy of the model in advance (for example, "You are a courteous medical consultation assistant. Avoid technical jargon and always recommend seeing a doctor at the end," set as the premise for the whole conversation).
It directs the output as a whole and is not itself a specification for individually suppressing unwanted elements, so it is incorrect.
Token limit (maximum number of tokens)
The maximum number of tokens is a setting that determines the upper limit on the length of the output.
It is not a specification for suppressing specific elements, so it is incorrect.
Stop sequence
A stop sequence is a setting that halts further generation the moment a predefined string appears in the model output (for example, in a conversational format, specifying "User:", "\n\n", or "###" to stop the answer before it runs into the next speaker's turn or unwanted continuation).
It controls where generation stops and is not a specification for suppressing unwanted elements (words or styles), so it is incorrect.
Negative prompt
Correct. A negative prompt is a specification for explicitly stating elements that should not be included in the output and suppressing them (for example, in image generation, specifying "blurry, extra fingers, text" in the negative prompt produces an image that avoids them). It is used to avoid undesirable results and bring the output closer to the intent.
Remember the correct answer, the negative prompt.
・It explicitly states elements that should not be included in the output (specific words, styles, subjects, and so on) and suppresses them.
・It is used to avoid undesirable results and bring the output closer to the intent.
The system prompt (overall role and policy), maximum number of tokens (upper limit on length), and batch inference (execution method) are none of them the specification that suppresses unwanted elements.