3 comments

  • delichon6 hours ago
    Some good stuff here from Dwarkesh around mashing up training and inference:<p><a href="https:&#x2F;&#x2F;youtu.be&#x2F;20p5-kQXF_Q?is=72ImTNxkOEKmOXQ9" rel="nofollow">https:&#x2F;&#x2F;youtu.be&#x2F;20p5-kQXF_Q?is=72ImTNxkOEKmOXQ9</a><p>He predicts this kind of model factory will become central to organizational learning and operations. Updating and upgrading the model stack becomes the core staff function.
    • typs14 minutes ago
      Some interesting ideas in this, but I think his argument is somewhat undermined by the fact that his main example of computer use has actually gotten much better recently because of RLVR
    • faangguyindia4 hours ago
      Interesting points made in the video.<p>But models did not become good at coding just because coding is replayable. It’s because there are countless repos, issues, Stack Overflow threads, and Reddit posts&#x2F;comments&#x2F;questions where a solution is clearly marked as “solved” or “that helped,” and AI can learn from that feedback.<p>Being replayable does play a role because a solution can be tested against a compiler, and the resulting errors or lack of errors&#x2F;warnings can reveal whether it worked.<p>This becomes much harder in fields like fitness, where changes take much longer and cause and effect relationships are not straightforward to establish.<p>Your muscle gain increased but was it because you increased protein intake? Or was it because you started eating more carbs, which added more energy to the system?<p>Once protein needs are already met, calories may become the limiting factor. In that case, the additional gains may come primarily from increased calorie intake rather than the higher protein intake itself.<p>AI is bad at fitness, evidently.<p>Many people forget, conversation with a model also generates training data. This is how your problems, algorithms, solutions end up in training data and end up right at your competitors without your competitor trying to actively steal your code.<p>I simply do not expose core algorithms which improve my product to AI agents.
      • NitpickLawyer23 minutes ago
        &gt; But models did not become good at coding just because coding is replayable. It’s because there are countless repos, issues, Stack Overflow threads, and Reddit posts&#x2F;comments&#x2F;questions where a solution is clearly marked as “solved” or “that helped,” and AI can learn from that feedback.<p>That&#x27;s at least 2yo take. Today&#x27;s gains for SotA (either closed or open models) come from RLVR 100%. The model unrolls many iterations, those iterations get verified w&#x2F; tests&#x2F;known tests&#x2F;rubrics and the model learns from that (grpo or similar).<p>And what&#x27;s cool about this (and why scale <i>really</i> matters now) is that you can mostly get this process automated (i.e. take a known good repo, ask one agent to remove one feature, keep the tests, ask another model to add that feature back, verify that old tests work on new implementation, repeat). This is why top labs are pulling away in the <i>breadth</i> of their capabilities, compared to open models. It&#x27;s scale, pure and simple. And the better their models become, the larger the gap due to automating better cases.
    • jaggederest6 hours ago
      I think this is an interesting thing that will happen once the rate of change slows down a little bit - imagine a world where there&#x27;s more or less a couple base models and everyone trains on top of them, and the bitter lesson is defunct just via sheer physics (maybe we have the best models we can physically run in reasonable energy density substrates, or something), then it becomes &quot;your personal model&quot; with your overlay, training, or feedback on top.
  • SpyCoder776 hours ago
    What is this &quot;aleph&quot; thing in names now? First aleph neuro, and now aleph alpha.
    • fxwin18 minutes ago
      fwiw aleph alpha have been around since 2019
    • verelo6 hours ago
      I&#x27;m glad you&#x27;re asking because I&#x27;ve seen it too and don&#x27;t get it either. I assumed initially it was alpha as a typo, then I Googled it and got even more confused.
      • boothby6 hours ago
        First letter of the Hebrew alphabet, used by mathematicians to denote infinities.
        • verelo6 hours ago
          That&#x27;s what Google told me, but i still don&#x27;t see how it links to this?
          • UltraSane3 hours ago
            It is just vibes man. It sounds cool, nothing more.
          • akoboldfrying4 hours ago
            It doesn&#x27;t -- it&#x27;s marketing, much like adding &quot;Labs&quot; to the end of your company&#x27;s name. Its association with infinity makes the company sound cooler to potential customers, many of whom are software engineers who consciously or unconsciously view pure mathematics as a prestige &quot;final form&quot; of their own logic-focused mental ability.
  • random36 hours ago
    &gt; TL;DR: Model training has grown complex<p>So they’ve built Savanah - a workflow engine because the existing zoo of hundreds of workflow engines didn’t cut it :)
    • usernametaken2927 minutes ago
      I was thinking the same. Airflow does exactly the same thing. The only benefit here is that it’s their little workflow engine so they can get all their little edge case accounted for…