AIF-C01 ยท Prompt engineering
Prompt versioning with Amazon Bedrock Prompt Management
Storing, versioning, comparing and reusing prompts and variants as managed assets.
Key points
- Prompt management lets you save prompts with variables, choose a model and inference settings, and test variants side by side.
- Saving keeps a working draft; creating a version snapshots it for deployment.
- Applications can reference a prompt, and flows can use it in a prompt node.
Source: https://docs.aws.amazon.com/bedrock/latest/userguide/prompt-management-deploy.html
Practice questions
In Prompt management, what is a version of a prompt?
- A snapshot of the prompt taken from the working draft
- A placeholder filled in at run time
- An alternative model configuration being tested
- The visual editor in the console
Show answer
Answer: A. A snapshot of the prompt taken from the working draft
A version is a point-in-time snapshot made from the working draft. A variant is an alternative configuration, and a variable is a placeholder.
Which things can a prompt variant differ in?
- The prompt message text
- The model that runs it
- The inference configuration
- The AWS account that owns it
- The billing plan
Show answer
Answer: A. The prompt message text, B. The model that runs it, C. The inference configuration
A variant can differ in its message and in the model or inference configuration used. The AWS account and the billing plan are not part of a prompt variant.
True or false: saving a prompt in Prompt management automatically creates a new numbered version.
Show answer
Answer: False
False. Saving updates the working draft. You create a numbered version explicitly when you are ready to deploy.
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.
