prompt-engineering.md

Prompt Engineering Principles

Tactic: Write clear and specific instructions

• Use delimiters to clearly indicate distinct parts of the input/output
• Ask for structured output like JSON or HTML
• Ask the model to check whether conditions are satisfied
• "Few-shot Prompting": Giving the model example of successful attempts
• "Chain-of-Thought Prompting: with demonstrations, not just examples of answers

Tactic: Give the model time to "think"

• Specify the steps required to complete a task
• Ask for output in a specified format
• Instruct the model to work out its own solution before rushing to a conclusion

Other tactics

• Limit the number of words/sentences/characters
• Ask the model to focus on the aspects that are relevant to the intended audience
• Ask it to extract info and organize it in a structured format

Tactics for summarization

• Summarize with a word/sentence/character limit
• Summarize with a focus on a specific aspect
• Summarize multiple entries

Tactics for inferring

• Infer sentiment (positive/negative)
• Identify types of emotions (e.g. satisfaction or anger)
• Do multiple tasks at once
• Infer topics

Tactics for transforming

• Translation
• Tone transformation
• Format conversion
• Spellcheck
• Grammar check
• Expand on text