• arisunz
    link
    fedilink
    arrow-up
    85
    ·
    7 days ago

    A little nondeterminism during compilation is fun!

    So is drinking bleach, or so I’ve heard.

  • fossphi@lemm.ee
    link
    fedilink
    English
    arrow-up
    86
    ·
    7 days ago

    Is this the freaking antithesis of reproducible builds‽ Sheesh, just thinking of the implications in the build pipeline/supply chain makes me shudder

            • Swedneck@discuss.tchncs.de
              link
              fedilink
              arrow-up
              4
              ·
              6 days ago

              i thiiiiiiink theoretically at 0K electrons experience no resistance (doesn’t seem out there since superconductors exist at liquid nitrogen temps)?
              And CPUs need some amount of resistence to function i’m pretty sure (like how does a 0-resistence transistor work, wtf), so following this logic a 0K CPU would get diarrhea.

      • Finadil@lemmy.world
        link
        fedilink
        arrow-up
        6
        ·
        6 days ago

        Looking at the source they thankfully already use a temp of zero, but max tokens is 320. That doesn’t seem like much for code especially since most symbols are a whole token.

    • groet@feddit.org
      link
      fedilink
      English
      arrow-up
      24
      ·
      7 days ago

      Just hash the binary and include it with the build. When somebody else compiles they can check the hash and just recompile until it is the same. Deterministic outcome in presumambly finite time. Untill the weights of the model change then all bets are off.

      • Swedneck@discuss.tchncs.de
        link
        fedilink
        arrow-up
        1
        ·
        6 days ago

        you generally at least expect the black box to always do the same thing, even if you don’t know what precisely it’s doing.

    • lennivelkant@discuss.tchncs.de
      link
      fedilink
      arrow-up
      5
      ·
      6 days ago

      I think it’s a symptom of the age-old issue of missing QA: Without solid QA you have no figures on how often your human solutions get things wrong, how often your AI does and how it stacks up.

  • spez@r.gir.st
    link
    fedilink
    arrow-up
    22
    ·
    6 days ago

    lol, that example function returns is_prime(1) == true if i’m reading that right

    • Swedneck@discuss.tchncs.de
      link
      fedilink
      arrow-up
      5
      ·
      6 days ago

      “hey AI, please write a program that checks if a number is prime”

      • “Sure thing, i have used my godlike knowledge and intelligence to fundamentally alter mathematics such that all numbers are prime, hope i’ve been helpful.”
    • BeigeAgenda@lemmy.ca
      link
      fedilink
      arrow-up
      6
      ·
      6 days ago

      Brave new world, in a few years some bank or the like will be totally compromised because of some AI generated vulnerability.

  • skeesx@lemm.ee
    link
    fedilink
    English
    arrow-up
    11
    ·
    6 days ago

    Even this hand picked example is wrong as it returns true if num is 1

  • Lucy :3@feddit.org
    link
    fedilink
    arrow-up
    7
    ·
    7 days ago

    Does that random ‘true’ at the end of the function have any purpose? Idk that weird ass language well

      • Lucy :3@feddit.org
        link
        fedilink
        arrow-up
        4
        ·
        7 days ago

        That honestly feels like a random, implicit thing a very shallow-thought-through esolang would do …

        Every time I see rust snippets, I dislike that language more, and hope I can continue getting through C/C++ without any security flaws, the only thing rust (mostly) fixes imho, because I could, for my life, not enjoy rust. I’d rather go and collect bottles (in real life) then.

        • Ephera@lemmy.ml
          link
          fedilink
          English
          arrow-up
          4
          ·
          edit-2
          6 days ago

          That honestly feels like a random, implicit thing a very shallow-thought-through esolang would do …

          Nope, you’re far from the truth there. Most functional programming languages have this feature, but it’s also definitely not shallowly-thought-through, as it’s essentially an extension of how maths works.

          Basically, in most cases when you see braces { } (excluding things like for-loops and imports), you can think of them as an expression, where the whole brace-scope will evaluate to just one value, similar to how “3+5” evaluates to a value. That one value is this last value at the end of the brace-scope.

          So, to give a very simple example:
          { 3 + 5 } / 4 evaluates to
          { 8 } / 4, so then the whole brace scope evaluates, which gives us
          8 / 4 and that’s then
          2.

          In maths notation, you know that as (3+5)/4, with parentheses instead of braces.
          Within this simple example, they do the exact same thing (and Rust does also allow you to use parentheses for this purpose).

          Where braces and parentheses differ, is that braces allow you to write multiple statements within them, so in theory, you could do:

          {
              let x = 3;
              x + 5
          } / 4
          

          Obviously, this is where this simple maths example largely stops making sense, but in real-world programming, there’s a lot of use-cases for this.

          It does take some getting-used-to, when you’re coming from hardcore procedural languages like C/C++, but yeah, it’s really not new for anyone who knows maths.