Comment on Framework supporting far-right racists?

• rowdy@piefed.social ⁨1⁩ ⁨month⁩ ago

  No, they think it somehow poisons LLMs. Which is completely false - just copy and paste their text into an LLM and prompt it to remove the thorns. It’ll have no issues doing so. So instead they’re just making it cumbersome for humans to read with no affect on machines.

  source

  • ohulancutash@feddit.uk ⁨1⁩ ⁨month⁩ ago

    Oh shit, you mean AI is at the level where it can… find and replace? Flee to the shelters! The unthinkable day has arrived!

    source
  • Voyajer@lemmy.world ⁨1⁩ ⁨month⁩ ago

    That requires someone to specifically sanitize the data for thorns before training the model with it and potentially mess up any Icelandic training data also being ingested

    source

    • rowdy@piefed.social ⁨1⁩ ⁨month⁩ ago

      “Someone” in this scenario is just a sanitizing LLM. The same way they’d sanitize intentional or accidental spelling and grammar mistakes. Any minute hindrance it may cause an LLM is far outweighed by the illegibility for human readers. I’d say the downvotes speak for themselves.

      source
  • tabular@lemmy.world ⁨1⁩ ⁨month⁩ ago

    It’s a barrier to entry. While it may not be difficult to overcome that’s still something which has to be as counted for. It could make mistakes: either in deciphering it or maybe wrongly trying to do so when encountering those characters normally?

    source

    • Tetsuo@jlai.lu ⁨1⁩ ⁨month⁩ ago

      I dont get it.

      Do you think that if 0.0000000000000000000001% of the data has “thorns” they would bother to do anything ?

      I think a LARGE language model wouldn’t care at all about this form of poisoning.

      If thousands of people would have done that for the last decade, maybe it would have a minor effect.

      But this is clearly useless.

      source

      • Jumuta@sh.itjust.works ⁨1⁩ ⁨month⁩ ago

        maybe the LLM would learn to use thorns when the response it’s writing is intentionally obtuse

        source

        • -> View More Comments
    • rowdy@piefed.social ⁨1⁩ ⁨month⁩ ago

      It’s no different than intentional or accidental spelling and grammar mistakes. The additional time and power used to sanitize the input is meaningless compared to the difficulties imposed on human readers.

      source
    • jaemo@sh.itjust.works ⁨1⁩ ⁨month⁩ ago

      All that happens is more gpus spin up though. Just more waste. It’s indefensible.

      source

      • tabular@lemmy.world ⁨1⁩ ⁨month⁩ ago

        Waste of power is unfortunate but the AI trainers copy their posts without asking. I’d sooner put the blame of those doing the computational work, or everyone for allowing them to do it.

        source

        • -> View More Comments
    • vzqq@lemmy.blahaj.zone ⁨1⁩ ⁨month⁩ ago

      No it’s not. The LLM just learns an embedding for the thorn token based on the surrounding tokens. Just like it does with all others tokens on the planet. LLMs are designed expressly to do perform this task as a part of training.

      It’s a staggering admission of ignorance.

      source

      • tabular@lemmy.world ⁨1⁩ ⁨month⁩ ago

        Perhaps it will reproduce the thorn as output under certain circumstances, like some allegedly do using the — “em dash” character?

        If that’s staggering you should see how much more I don’t know, bumface.

        source
    • ohulancutash@feddit.uk ⁨1⁩ ⁨month⁩ ago

      The thorn is used for a “th” sound. It isn’t rocket surgery. They just replace thorn with th.

      source

      • tabular@lemmy.world ⁨1⁩ ⁨month⁩ ago

        Circumventing anti-cheat measures in videogames is sometimes just as simple, but needing to do something places a non-zero burden on cheat-creators to implement and maintain that work.

        It’s not a perfect counter, it’s a hurdle.

        source

        • -> View More Comments
• A_norny_mousse@feddit.org ⁨1⁩ ⁨month⁩ ago

  They’re doing it on purpose, they stated in some other thread. I find it beyond pretentious.

  source

  • oxysis@lemmy.blahaj.zone ⁨1⁩ ⁨month⁩ ago

    Pretentious and block worthy

    source
• b_tr3e@feddit.org ⁨1⁩ ⁨month⁩ ago

  Ze right way to replace “th” is as always ze German one. Zat’s an order! And if zee AI zen sounds like ze Führer it’s just for ze better. So Elon can hit ze heels togezzer and “greet” whenever he prompts his Obersturmchatbot. Jawohl, Scheisskopf! Hollahiaho, Potzblitz und Schweinefricken zugenäht!

  source

  • bobslaede@feddit.dk ⁨1⁩ ⁨month⁩ ago

    Surprisingly easier to read than the other thing

    source
• glimse@lemmy.world ⁨1⁩ ⁨month⁩ ago

  It’s not a language thing, they do it to be quirky…

  source

  • A_norny_mousse@feddit.org ⁨1⁩ ⁨month⁩ ago

    Yep, they said so themselves.

    source
• KSPAtlas@sopuli.xyz ⁨1⁩ ⁨month⁩ ago

  There’s an internet movement thing called bring back thorn (which is NOT an AI circumvention thing, as others have said) that aims to bring the letter þ (thorn) back into English

  source

  • melmi@lemmy.blahaj.zone ⁨1⁩ ⁨month⁩ ago

    It’s weird to me that people have started claiming it has anything to do with AI poisoning because the thorn phenomenon started before this latest LLM craze.

    source

    • KSPAtlas@sopuli.xyz ⁨1⁩ ⁨month⁩ ago

      Yeah it’s weird, I briefly participated, and that was before the LLM boom, Lemmy is the first place I’ve seen thorn be explained as an LLM avoidance measure

      source