Prompt Engineering

this is whole guide is based developer on prospective that you are developing applications using OpenAI's chatGPT API.


there are two types of LLMs

  • Base LLM
    • predict the next word based on the text training data. its typical ML model, its not giving any generative answers.
  • Instructions tuned LLM
    • this model will give a generative answer based on the large amount of data that he has been trained on.

Prompting principles

Model limitations

  • model doesn't know the boundary of its knowledge so some times it might make up the things.
  • this is called as hallucination.
To avoid hallucination
  • give instructions to model to first find the relevant information
  • then answer the question.
  • based on the relevant information.