Prompt Management 101 - Full Guide for AI Engineers

Поделиться
HTML-код
  • Опубликовано: 14 ноя 2024

Комментарии • 21

  • @daveebbelaar
    @daveebbelaar  2 месяца назад +4

    How are you managing your prompts? Let me know down in the comments.
    Also make sure to share this video with your fellow AI Engineers!

    • @HerroEverynyan
      @HerroEverynyan 2 месяца назад

      Django Templates. Syntax like Jinja but has limitations so advanced logic is done in code to prevent complexity in the template. I'd prefer Jinja, since I don't think it gets _too_ complex with prompts, but the client's system uses Django Templates.

    • @antoniov845
      @antoniov845 2 месяца назад

      F-strings => Jinja and I agree that it seems to be the best 👍

    • @daveebbelaar
      @daveebbelaar  2 месяца назад

      @@antoniov845 Why F-strings > Jinja?

    • @antoniov845
      @antoniov845 2 месяца назад

      @@daveebbelaar sorry that’s my poor phrasing 😀 I meant that I managed prompts using f-strings first and later migrated to jinja

    • @daveebbelaar
      @daveebbelaar  2 месяца назад +1

      @@antoniov845 ah, got it!

  • @FeralFluxMusic
    @FeralFluxMusic 2 месяца назад

    Very intersting!
    I would love to see a deep dive into prompt management with Jinja, learning about some best practices you found using it, etc.

  • @NS_Miata
    @NS_Miata 23 дня назад

    I find Ell a cool little framework for prompt mangement. It makes an automatic database were you can view you whole history of prompts and automatically rapport the results of prompt edita

  • @IsaacMeisner-l2x
    @IsaacMeisner-l2x Месяц назад

    Great video!! Thank you for sharing your team's approach. Very insightful for me to implement with my team

  • @micbab-vg2mu
    @micbab-vg2mu 2 месяца назад

    Thanks for this - I have a library of prompts - I need to manage them more efficiently

  • @bdagnino
    @bdagnino 2 месяца назад

    Hi Dave! Awesome video, super interesting and and useful. Thanks for taking the time to put it together!
    Question: any extra tips on how to maintain different versions and do rollbacks / keep traceability of what you have used in the past?

    • @daveebbelaar
      @daveebbelaar  2 месяца назад

      It's tricky haha. CMS or custom database setup with evaluation datasets and changelogs, but you have to find something that works for your situation.

  • @motaseamyousef7668
    @motaseamyousef7668 2 месяца назад

    Thanks for sharing @Dave 😊

  • @harrydeboer921
    @harrydeboer921 2 месяца назад

    txt file in repository at the moment (with output struc statement as part of the prompt) together with langchain templates. Looking at prompt management frameworks, but cannot find one that let different departments edit different parts of the prompt. Also output struc management is difficult, uses a lot of nested dicts/lists. I think we will end up programming our own solution. Thanks for another great video Dave.

  • @motaseamyousef7668
    @motaseamyousef7668 2 месяца назад

    we need a prompt best practices, tips and tricks work with different LLM (as Local LLM as Llama or API from openai as example)
    Thanks ☺

  • @EmpreendedoresdoBEM
    @EmpreendedoresdoBEM 2 месяца назад

    Cool: it's the beginning of AI-end department to help BACK-end and FRONT-end...

  • @fredericherrera
    @fredericherrera 2 месяца назад

    tres bien la video!

  • @StanislavKozlovski
    @StanislavKozlovski 15 дней назад

    You know how you can use a Clipboard Manager to very efficiently scroll through your latest copied items and paste them?
    Why don't we have the same thing for AI prompts? Basically a pre-defined set of AI prompts you like, and you can paste them in the same speed as a clipboard manager.
    Does such a thing exist? It's pretty low-effort to do, but super useful.
    Someone should just fork Maccy (macos clipboard manager) and do it.

  • @aiplaygrounds
    @aiplaygrounds 2 месяца назад

    I Am messy haha this is helpful ❤ thanks