‹ BackHN Continuity

Thread

OpenJev

722 points · 296 comments · ilreb

  1. tantalor · · focus · HN ↗
    What's a "Jev"?
    1. techjamie · · focus · HN ↗
      A model that was introduced a few days ago that's LLM-based, but instead of producing text, you can ask it questions and it will return decisions. The key thing being that it responds pretty quickly and predictably.

      Site: <a href="https:&#x2F;&#x2F;typesafe.ai&#x2F;blog&#x2F;introducing-system-one-models-and-jev" rel="nofollow">https:&#x2F;&#x2F;typesafe.ai&#x2F;blog&#x2F;introducing-system-one-models-and-j...

      1. baobabKoodaa · · focus · HN ↗
        It&#x27;s a closed model and they claim that it&#x27;s not LLM-based, so I&#x27;m not sure why you are claiming that it is LLM-based.
        1. brazukadev · · focus · HN ↗
          where is the claim it is not LLM-based? The claim I saw is that it is not chat-based but still text-based (JSON).
          1. baobabKoodaa · · focus · HN ↗
            In the link of the comment I was responding to, they explicitly say that Jev is not LLM-based:

            &gt; Is Jev just a smaller LLM?

            &gt; Jev is neither small nor an LLM, hence being off the intelligence Pareto curve.

            1. brazukadev · · focus · HN ↗
              Sounds like BS for me. Jev is transformed-based, trained on language&#x2F;texts, it receives text inputs and output text (json).

              Jev is a language model. It doesn&#x27;t matter if it is not a &quot;smaller LLM&quot; or not LLM by some weird definition.

              1. baobabKoodaa · · focus · HN ↗
                &gt; output text

                according to the people who made Jev, it does NOT output text. it&#x27;s a closed model, so we can&#x27;t inspect the internals, but i would just take their word for it.

                just because the API responds with JSON text, does not mean that the underlying model is generating JSON text.

                1. FergusArgyll · · focus · HN ↗
                  Yeah, it outputs probabilities over the options you give it. It&#x27;s basically a generalizable classifier.
    2. timnetworks · · focus · HN ↗
      <a href="https:&#x2F;&#x2F;typesafe.ai&#x2F;" rel="nofollow">https:&#x2F;&#x2F;typesafe.ai&#x2F;
Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.