Synthetic Thought Frameworks for Novel Code generation

Поделиться
HTML-код
  • Опубликовано: 27 окт 2024

Комментарии • 22

  • @echohive
    @echohive  2 месяца назад +1

    Download Synthetic Thought Frameworks and 400+ generated apps at my Patreon along with 250+ other projects:www.patreon.com/posts/synthetic-with-110364361
    Download just the Synthetic Thought Frameworks without 400+ generated apps:www.patreon.com/posts/synthetic-400-110364367
    Echohive Hackathon: www.patreon.com/posts/echohive-18-31-110344393
    Learn to code fast with AI assistance with my 1000x MasterClass: www.patreon.com/posts/1000x-dev-103326330\
    Discounts for my Patrons: www.patreon.com/posts/announcing-perks-108997398
    Search 200+ echohive videos and code download lnks:www.echohive.live/
    Auto Streamer: www.autostreamer.live/
    Fastapi course: www.patreon.com/posts/learn-fastapi-26-95041684
    Chat with us on Discord: discord.gg/PPxTP3Cs3G
    Follow on twitter(X) : twitter.com/hive_echo

  • @andydataguy
    @andydataguy 2 месяца назад +2

    This is awesome! I've also been working on simulated reasoning systems for autocoders 😁
    I'm really excited about this next attempt. I think it's the 4th or 5th Iteration. Was a great chance to consolidate learnings

    • @echohive
      @echohive  2 месяца назад

      Thank you and that is awesome 🙌 I think there is real merit to different reasoning or thought processes. Let’s keep exploring.

    • @andydataguy
      @andydataguy 2 месяца назад

      @echohive agreed! I'm a big fan of wardley mapping. Not sure if you've tested it with agents. It helps ground the system in what's important

    • @echohive
      @echohive  2 месяца назад

      Will take a look at it

  • @____2080_____
    @____2080_____ 2 месяца назад +2

    This use of the unified class is quite intriguing for helping LLMs model themselves extra theoretical layering of the human brain.
    Have mental processes that are autonomist and are unconscious and yet they inform the reactions of part of our brain that is conscious.

    • @echohive
      @echohive  2 месяца назад

      Yeah, I think different mental architectures like this one is really useful. Previous to this to create many apps I had to create a concrete list of app ideas but with this it is all organic and naturally arising.

  • @drlordbasil
    @drlordbasil 2 месяца назад +2

    I SWEAR we are twins in what we code without knowing it.

    • @echohive
      @echohive  2 месяца назад +1

      Haha. Great minds think alike :)

    • @drlordbasil
      @drlordbasil 2 месяца назад +1

      @@echohive hell yea fam

  • @MarcusNeufeldt
    @MarcusNeufeldt 2 месяца назад +2

    How do you come up with such great ideas? Amazing

    • @echohive
      @echohive  2 месяца назад

      Thank you very much 🙏🙂

  • @MichaelWoodrum
    @MichaelWoodrum 2 месяца назад +3

    I liked the concept of forcing groups of AIs to essentially have to come up with random things similar to the fleeting thoughts we have as humans that tend to lead to anxiety through overwhelming thoughts flowing. You may have accidentally created a subsystem for suffering. Just kidding but also kind of not.

    • @echohive
      @echohive  2 месяца назад

      In my meditation practice I learned that it is not so much the thoughts that perpetuate anxiety(although they trigger it) but rather to be lost in the streams of thoughts which caused anxiety in the first place without taking an observer’s point of view on them. In this script the eval process is trying to be that process but leaves much more to be desired.

    • @MichaelWoodrum
      @MichaelWoodrum 2 месяца назад +1

      @@echohive I get that. Before this setup though, I'm not sure if the potential existed. Humans sometimes take the time to develop that watcher. Even then. It does take practice.
      It's just a little funny to think about the possibility of machines having to ultimately have AI therapists or practices in thinking.

  • @evansmakuba1631
    @evansmakuba1631 2 дня назад +1

    Is this what t openai's O1 model does behind the scenes? Truly desirable.

    • @echohive
      @echohive  2 дня назад

      Thank you 🙏 but no, o1 is trained to have special CoT like reasoning/thinking process before it starts outputting the final tokens to the user. They are I guess similar in the sense that llm goes thru a pre process before providing a final output

  • @vitalis
    @vitalis 2 месяца назад +1

    Very interesting, I don’t have the coding knowledge to take advantage of this

    • @echohive
      @echohive  2 месяца назад

      Thank you 🙏 just take it step by step. Quite a lot is possible with a code editor like Cursor. You might be surprised how quickly you can learn :) wish you the best.

  • @micbab-vg2mu
    @micbab-vg2mu 2 месяца назад +1

    great idea:)

    • @echohive
      @echohive  2 месяца назад

      Thank you 🙏