-2

I sometimes use some language model assistance (e.g., GPT-4o) while coding. I'd like to save somewhere the prompts and language model versions I used. Are there good practices or even conventions to store LLM prompts I use while coding for a project?

Answering comments:

I use prompts that generate new code, e.g. a function to restructure some JSON object or to parse some text file. I'd like save these prompts and LLM versions to remember what kind of prompts and which LLMs worked well, as I may have similar needs in the future.

Improve this question
9
  • 1
    I don't think there are any general practices here. Fundamentally, this question is opinion-based, and will likely get closed fairly quickly. Although it feels similar in spirit to this question: Best practices for sharing tiny snippets of code across projects. That being said, it's an older question that would likely be closed as opinion-based today. Commented May 14, 2024 at 19:53
  • 2
    This has perked my interest because programmers have been dealing with code snippets for ages, but LLM prompts are a new thing. The LLM prompts can lead to code snippets, so this is a new perspective on an old problem. I would just like to see it rephrased so the community can respond with something more objective than personal opinions. Commented May 14, 2024 at 19:57
  • 3
    That is good information, but I just don't think it moves the question beyond opinion-based answers yet. Commented May 14, 2024 at 20:05
  • 1
    @GregBurghardt In what way is the proper use of LLM systems more opinion-based than the proper use of the fantastically complex systems that modern programming languages constitute? Software engineering is all about non-obvious useful knowledge gained from experienced practitioners. Commented May 15, 2024 at 6:40
  • 2
    @KilianFoth, I voted to reopen. I've gone back and forth about this question. I'm curious. Commented May 15, 2024 at 10:50

1 Answer 1

Reset to default
0

The comments on this question tend to indicate that there are no good practices or conventions for storing LLM prompts. Any places to store texts should suffice, and I'd recommending saving the LLM version (to make more reproducible) and if possible, the link to the LLM chat (e.g., to be able to continue the chat if needs be).

Improve this answer

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.