ChatGPT Prompt Engineering for Developers

https://learn.deeplearning.ai/chatgpt-prompt-eng/

Introduction

• Two types of LLMs
  • Base LLM - predicts next word, does not “understand” questions
  • Instruction Tuned LLM
    • Fine-tuned on instructions
    • RLHF = Reinforcement Learning with Human Feedback
    • Helpful, Honest, Harmless

Guidelines

• Write clear and specific instructions

  • Use delimiters to clearly indicate distinct parts of the input. This helps with avoiding prompt injection.
  • Ask for a structured output: JSON, HTML, etc.
  • Ask the model to check whether conditions are satisfied
  • “Few-shot” prompting: Give successful examples.

• Give the model time to think

  • Specify the steps required to complete a task
  • Instruct the model to work out its own solution before rushing to a conclusion

• Avoid hallucinations

  • First find relevant information
  • Then answer the question based on the relevant info