From the course: AWS Certified AI Practitioner (AIF-C01) Cert Prep

Unlock this course with a free trial

Join today to access over 24,700 courses taught by industry experts.

Prompt workflow

Prompt workflow

- In order to learn about prompt engineering, it is first important to understand how the prompt itself works and all of the different elements that can be used to influence that prompt, as well as their relative priority compared to each other. So first, the prompt itself consists of several elements, and those are going to include a system message, and this is where you define the role to be assumed by the LLM before generating a response. The system message might be a default that is actually configured more on the server side and not always necessarily passed along with every single prompt in a conversation. Next, we have the user message, and this is the, the classic definition of the prompt, whatever has been typed in or the question to be asked, the task to be performed, the follow up statement and so forth. We can also include an assistant message, and this is the tone and format of the response. So if you wanted to have the response returned in JSON format, that would go in…

Contents