• hperrin@lemmy.ca
      link
      fedilink
      English
      arrow-up
      0
      ·
      5 days ago

      LLMs won’t spit out the same word-for-word response for the same prompt

      You can give an LLM the same seed and it will spit out the same word-for-word response. That’s how they work. It’s just a bunch of math.

      • anothermember@feddit.uk
        link
        fedilink
        English
        arrow-up
        0
        ·
        5 days ago

        You can give an LLM the same seed and it will spit out the same word-for-word response. That’s how they work. It’s just a bunch of math.

        You’re assuming that because I missed out that detail I must be ignorant of it, that’s not very charitable, I could well have been ignorant of it but you could have made your otherwise useful clarification without telling me I was wrong.

        • hperrin@lemmy.ca
          link
          fedilink
          English
          arrow-up
          1
          ·
          edit-2
          5 days ago

          You said “that’s not how they work”. But that is how they work. Same prompt = same output. Throw some random data in there to jumble things around and you get a little variance. That’s the seed, and we only need to do that because LLMs are inherently deterministic.

          Same reason Minecraft has a random seed for world generation, and block cipher algorithms use an initialization vector and/or feedback loop. We don’t want the same thing every time.

          I did say that you’re right, because the tooling we use around the LLM itself does exactly what you’re talking about. So, in practice, you’re right.