‹ BackHN Continuity

Thread

Ollaya – Ollama for open-source, Jev-style decision models

618 points · 145 comments · Ardakilic

  1. solaire_oa · · focus · HN ↗
    I installed it, I tried the examples, it works.... But forgive my lack of imagination... what is this useful for?

    Like, their example is of classification for a support interface.... `refund_requested`. Pretty convenient bool given the example is about a refund- what if 99% of submissions don't ask about a refund? Also, is that user not a `churn_risk`? What could possibly qualify as a churn risk if not a user asking for a refund?

    <a href="https:&#x2F;&#x2F;ollaya.dev&#x2F;library&#x2F;laya" rel="nofollow">https:&#x2F;&#x2F;ollaya.dev&#x2F;library&#x2F;laya The examples suffer the same problem of why I&#x27;d prefer to use a string column vs an enum. Changing an enum means you need to update the db, using a string you can do whatever.

    I&#x27;m not trying to be negative, I genuinely want to know about some practical examples (that don&#x27;t require tons of backwards maintenance).

    1. devttyeu · · focus · HN ↗
      I have a lot of semi-practical examples of how you can use this model wrapped in unix-ish tools - <a href="https:&#x2F;&#x2F;github.com&#x2F;aurorainfra&#x2F;grev" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;aurorainfra&#x2F;grev (readme links to docs of each tool with some more or less practical examples)

      Really I think &quot;smart grep&quot; is a pretty good one (&#x27;look for an error looking vaguely like this&#x27;). Also I think sql-based shell history + decision model is quite good to make the last &#x27;which one of those choices is best fit given users past few commands&#x27; etc.

      1. spaniard89277 · · focus · HN ↗
        Isn&#x27;t it better to use an LLM to train modernbert or xgboost et al?
        1. devttyeu · · focus · HN ↗
          It is &#x2F;possible&#x2F; to use an LLM.

          But with Jev you&#x27;re just paying for input (prefill) which is really fast, and in case of Jev specifically costs 50% of Deepseek V4.1 Flash (which has famously really cheap input token pricing).

          I put 250MB &#x2F; 1M lines of logs through Grev and it cost ~$10USD, DSv4.1 would be at least 10x that and much, much, much slower. With Jev&#x2F;Grev that 1M requests took 10 mins

          Edit: completely misread your question - yeah you could finetune specialized models to do that, probably based on some decent pretrained llm base, that is true for roughly any Jev-shaped problem. Do you want to bother doing that, also having to deal with having to host a zoo of specialized models?

        2. motoboi · · focus · HN ↗
          Laya, the open source model, it’s actually modernBERT based.
Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.