• jacksilver@lemmy.world
    link
    fedilink
    English
    arrow-up
    10
    ·
    2 months ago

    LLMs do suck at math, if you look into it, the o1 models actually escape the LLM output and write a python function to calculate the output, I’ve been able to break their math functions by asking for functions that use math not in the standard Python library.

    I know someone also wrote a wolfram integration to help solve LLMs math problems.

      • Smokeydope@lemmy.world
        link
        fedilink
        English
        arrow-up
        1
        arrow-down
        1
        ·
        edit-2
        2 months ago

        Thanks for sharing, knew him from some numberphile vids cool to see they have a mastadon account. Good to know that LLMs are crawling from “incompentent graduate” to “mediocre graduate”. Which basically means its already smarter than most people for many kinds of reasoning task.

        I’m not a big fan of the way the guy speaks though, as is common for super intelligent academic types they have to use overly complicated wording to formally describe even the most basic opinions while mixing in hints of inflated ego and intellectual superiority. He should start experimenting with having o-1 as his editor and summarize his toots.

    • 0ops@lemm.ee
      link
      fedilink
      English
      arrow-up
      3
      ·
      2 months ago

      Wow that’s really clever actually. Basically using the library as digital scratch paper