Thank you note example

Replace static text in the Thank you note generation built-in sample prompt with variables to make the prompt reusable.

To add versatility to a built-in prompt, complete the following steps:

1. From the Prompt Lab, click Sample prompts to list the built-in sample prompts. From the Generation section, click Thank you note generation.

   The input for the built-in sample prompt is added to the prompt editor and the flan-ul2-20b model is selected.

   Write a thank you note for attending a workshop.
   
   Attendees: interns
   Topic: codefest, AI
   Tone: energetic
   

2. Review the text for words or phrases that make good variable candidates.

   In this example, if the following words are replaced, the prompt meaning will change:

   • workshop
   • interns
   • codefest
   • AI
   • energetic

3. Click the Prompt variables icon and create a variable to represent each word in the list. Add the current value as the default value for the variable.

   Table 2. Variables for the thank you note sample

   Variable name	Value
   event	workshop
   attendees	interns
   topic1	codefest
   topic2	AI
   tone	energetic

4. Click Preview to review the variables that you added.

5. Update the static prompt text to use variables in place of words.

   Write a thank you note for attending a {event}.
   
   Attendees: {attendees}
   Topic: {topic1}, {topic2}
   Tone: {tone}
   

   

   The original meaning of the prompt is maintained.

6. Now, change the values of the variables to change the meaning of the prompt.

   From the Fill in prompt variables view of the prompt variables panel, add values for the variables.

   Table 3. New values for the variables

   Variable name	Value
   event	human resources presentation
   attendees	expecting parents
   topic1	resources for new parents
   topic2	parental leave
   tone	supportive

   You effectively converted the original prompt into the following prompt:

   Write a thank you note for attending a human resources presentation.
   
   Attendees: expecting parents
   Topic: resources for new parents, parental leave
   Tone: supportive
   

   Click Generate to see how the model responds.

7. Swap the values for the variables to reuse the same prompt again to generate thank you notes for usability test attendees.

   Table 4. Another set of values for the variables

   Variable name	Value
   event	usability test
   attendees	user volunteers
   topic1	testing out new features
   topic2	sharing early feedback
   tone	appreciative

   Click Generate to see how the model responds.

Devil's advocate example

Use prompt variables to reuse effective examples that you devise for a prompt.

You can guide a foundation model to answer in an expected way by adding a few examples that establish a pattern for the model to follow. This kind of prompt is called a few-shot prompt. Inventing good examples for a prompt requires imagination and testing and can be time-consuming. If you successfully create a few-shot prompt that proves to be effective, you can make it reusable by adding prompt variables.

Maybe you want to use the granite-13b-instruct-v1 model to help you consider risks or problems that might arise from an action or plan under consideration.

For example, the prompt might have the following instruction and examples:

You are playing the role of devil's advocate. Argue against the proposed plans. List 3 detailed, unique, compelling reasons why moving forward with the plan would be a bad choice. Consider all types of risks.

Plan we are considering:
Extend our store hours.
Three problems with this plan are:
1. We'll have to pay more for staffing.
2. Risk of theft increases late at night.
3. Clerks might not want to work later hours.

Plan we are considering:
Open a second location for our business.
Three problems with this plan are:
1. Managing two locations will be more than twice as time-consuming than managed just one.
2. Creating a new location doesn't guarantee twice as many customers.
3. A new location means added real estate, utility, and personnel expenses.

Plan we are considering:
Refreshing our brand image by creating a new logo.
Three problems with this plan are:

You can reuse the prompt by completing the following steps:

1. Replace the text that describes the action that you are considering with a variable.

   For example, you can add the following variable:

   Table 5. Variable that represents the possible action

   Variable name	Default value
   plan	Refreshing our brand image by creating a new logo.

2. Replace the static text that defines the plan with the {plan} variable.

   You are playing the role of devil's advocate. Argue against the proposed plans. List 3 detailed, unique, compelling reasons why moving forward with the plan would be a bad choice. Consider all types of risks.
   
   Plan we are considering:
   Extend our store hours.
   Three problems with this plan are:
   1. We'll have to pay more for staffing.
   2. Risk of theft increases late at night.
   3. Clerks might not want to work later hours.
   
   Plan we are considering:
   Open a second location for our business.
   Three problems with this plan are:
   1. Managing two locations will be more than twice as time-consuming than managed just one.
   2. Creating a new location doesn't guarantee twice as many customers.
   3. A new location means added real estate, utility, and personnel expenses.
   
   Plan we are considering:
   {plan}
   Three problems with this plan are:
   
   

   Now you can use the same prompt to prompt the model to brainstorm about other actions.

3. Change the text in the {plan} variable to describe a different plan, and then click Generate to send the new input to the model.

Parent topic: Prompt Lab