Bitzu

AIF-C01 ยท Prompt engineering

Prompt constructs

Instruction, context, input data, output indicator and negative prompts, and what each contributes to a response.

Key points

  • A prompt is built from parts.
  • The instruction states the task.
  • Context supplies background and role.
  • Input data is the material to work on.
  • The output indicator signals the expected format or where the answer starts.
  • A negative prompt says what to avoid.

Source: https://docs.aws.amazon.com/bedrock/latest/userguide/prompt-management.html

Practice questions

Single choice

Which prompt construct states the task the model should perform?

  1. Instruction
  2. Context
  3. Input data
  4. Output indicator
Show answer

Answer: A. Instruction

The instruction is the task statement. Context only adds background, and input data is the material the task is applied to.

Select all that apply

Which of these are constructs you can put inside a prompt?

  1. An instruction
  2. An output indicator
  3. A negative prompt
  4. A model checkpoint
  5. A temperature value
Show answer

Answer: A. An instruction, B. An output indicator, C. A negative prompt

Instruction, output indicator and negative prompt are all text parts of a prompt. A model checkpoint and a temperature value are model or inference settings, not prompt text.

True or false

True or false: a negative prompt lists the content and style the model should aim for.

Show answer

Answer: False

False. A negative prompt describes what the model should avoid or exclude. Desired content belongs in the instruction and context.

Practice 8 questions on this topic in Bitzu

Short lessons, instant explanations and spaced review of what you get wrong. Free on iOS and Android.

Download on the App StoreGet it on Google Play

More AIF-C01 topics in this module