Note 3— Prompt Engineering for Developer

Topic: Iterative Prompt Development

Photo by Tim Johnson on Unsplash

Why

There probably isn’t a perfect prompt for everything under the sun. It is more important that you have a process for development a good prompt for your specific application — Andrew Ng

What

Iterative Prompt Development: Find a prompt that works more reliably every single time

How

Iterative Process:

• Try something
• Analyze where the result does not give what you want
• Clarify instructions, give more time to think
• Refine prompts with a batch of examples

Cycle of iterative prompt development

Example: How to iterate and refine your prompt
Ask LLM for a technical fact sheet

info of the fact sheet

Case 1: write product description

the prompt that needs to generate technical fact sheet

the response → too prolix, hope to shorten the response

After finding the response doesn’t meet the expectation (prolixity), you need to optimize the response (concise):

Phase of optimizing response: utilizing “Use at most 50 words” to shorten the response

Case 2: Rewrite the description that aims to different retailers who specialize in different domain
→ the response needs to emphasize on technical details and material-related words

describe the TA (furniture retailers), and what they care about.

the response

Case 3: Following the previous situation (Case 1, 2), the sheet needs to be added Product ID:

GitHub: MarsWangyang (Mars Wang) (github.com)

LinkedIn: Meng-Yang (Mars) Wang | LinkedIn