7 comments

  • Art9681 55 minutes ago
    The absolute best way to prove this works is by releasing a model that was fine-tuned with this method and then showing benchmarks depicting the improvement delta between the base model and the fine tuned one.

    The work is not done. Then release it to the masses and wait a few days for the actual real world anecdotes.

    Until then, this is noise.

    • SilenN 49 minutes ago
      Valid criticism. Happy to answer any qs. We're still working on solidfying results.
      • dang 47 minutes ago
        Ok, I think it is in your interest to wait until you have more to show, and we'll be happy to help you with reposting it once it's ready.

        Waitlists are against the Show HN rules (https://news.ycombinator.com/showhn.html), and you're likely to get a lot of community pushback if you post before there's enough substance for users to sink their teeth into.

        • SilenN 41 minutes ago
          Thanks for the heads up, removed mention!
    • Reubend 51 minutes ago
      Yeah, this is just slop. No benchmarks, no concrete case studies, just some vibecoded "platform" to finetune models on your own traces.

      Which is an idea that has some value, but also some weaknesses. And this implementation of it isn't forthcoming with that concept. You have to really dig in to understand what they're even talking about.

      • SilenN 46 minutes ago
        Happy to answer any qs.
    • irishcoffee 51 minutes ago
      Benchmarks are the ultimate consolidation of halnons razor.
    • teravor 49 minutes ago
      [flagged]
  • adrianco 1 hour ago
    Local models need to be tuned to work well so this looks useful. Seems to be for general purpose model serving. I’ve been using https://github.com/adrianco/retort to run experiments for coding models across 13 different programming languages to see which frontier and local models work.
    • SilenN 56 minutes ago
      That's cool, thanks for sharing!
  • jack_pp 1 hour ago
    Not sure I get it. The model you're improving is local? If so how do you even calculate cost compared to an API
    • SilenN 50 minutes ago
      Open source models.

      wmo routes requests between frontier models and open source models that continuously train using Tinker. As the smaller models improve, more traffic gets routed to them.

      Calculating cost is just tokens in/out.

  • surround 44 minutes ago
    The title is misleading. This is model routing, not distillation.
    • SilenN 39 minutes ago
      Fixed formatting which will help with readability. We do routing, distillation, and token compaction.
  • digitaltrees 1 hour ago
    Cool project
    • SilenN 40 minutes ago
      Thanks :)
  • yiyingzhang 1 hour ago
    Cool idea! How do you guarantee privacy?
    • SilenN 49 minutes ago
      It's open source!

      We do have a platform we'll be launching as well to manage training + serving for you which will require more diligent privacy guarantees.

  • rglover 1 hour ago
    Excited to play with this.
    • SilenN 40 minutes ago
      Let me know if you have any questions!