Manage your prompts
Managing your prompts in the cloud gives you multiple advantages like working more efficient in a team, faster iterations for improvements, validating different versions and testing new ideas and LLMs.
Create a Prompt
Your cloud prompts can easily be created under the menu "Prompts" on the left side navigation. Press the button "Create Prompt" on the top right side. Manage your cloud prompts.
Give the prompt a name and write a description of what you are trying to achieve with the prompt.
Tip: When you create different versions of s prompt, the descriptions field should include what differentiates it from other versions of your prompt.
Prompt Data
The "Prompt Data" area is the section where you define what text will be sent to the selected LLM. Simply type your prompt instructions into the field with the headline "Prompt Text".
Variable Replacement
You can define variables, that will dynamically be replaced with the values you are sending via the API. Press the button "+ Add Variable" and give it a name and description. The use case for the default value is to test the prompt result with different values. This way, the prompt engineer can easily test and work towards the best possible result. You integrate the variable in the prompt text by surrounding it with curly braces. E.G. {my_variable_name}
Test Run the Prompt
Run a test of the prompt by clicking the button "Run this prompt to test the result" in the tab "Test Run"on the right side of the page.
Prompt Versioning
In the tab "Versions", you can clone any version of your prompt to create a new version.
Tip: When calling a prompt via the api, setting the version name allows to execute that version of the prompt. If you don‘t set a version in the api call, Nter.ai always selects your default version. This allows the team to easily switch prompt versions in your own apps by making a new version the default.
System (Role) Description
This field primarily defines the system description field when communicating with ChatGPT. Find some more information about the system role in the OpenAI community here.
