Live data from Hacker News

Viewing profile — marviel

marviel

HN member
Joined
Tue, Nov 03, 2015, 4:31 PM UTC
HN karma
947
Public activity
354 items

About marviel

lukebechtel.com

Say hi : luke (at) lukebechtel.com

Current: - Founder @ Reasonote (https://reasonote.com) - AI/ML Consulting @ Positive Sum Products (https://positivesum products.com)

Former: - Principal AI/ML Eng. @ Regscale (https://regscale.com) - Dir. of Eng. @ Revaly (now Cadchat) (Acq. in 2023) - Cofounder @ Collider Inc. (Acq. in 2022)

Recent public activity

  1. comment
    Comment #44576361

    It's meant to replace the BPE tokenizer piece, so it isn't a full Language Model by itself. In fact in Gu's blog post (linked in a post below) it's mentioned that they created a Ma…

  2. comment
    Comment #44576316

    Thanks for sharing this blog post is a great speculative deep-dive.

  3. comment
    Comment #44575636

    it mentions native multimodality somewhere in either the Arxiv or post -- seems like it might handle it well?

  4. comment
    Comment #44575588

    No problem! I'm still parsing it myself, but it seems promising in theory, and the result curves are impressive.

  5. comment
    Comment #44575574

    https://arxiv.org/pdf/2507.07955 paper

  6. comment
    Comment #44575248

    > H-Net demonstrates three important results on language modeling: > 1. H-Nets scale better with data than state-of-the-art Transformers with BPE tokenization, while learning direc…

  7. story
  8. comment
    Comment #44563988

    nice! Can't agree more on Vibe Speccing. I wrote more about Spec Driven AI development here: https://lukebechtel.com/blog/vibe-speccing

  9. comment
    Comment #44494981

    how do you account for the success of reasoning models? I agree these things don't think like we do, and that they have weird gaps, but to claim they can't reason at all doesn't fe…

  10. comment
    Comment #44487078

    have often wanted something like "liquid types", but didn't know what it was called! Thanks for this

  11. comment
    Comment #44474189

    there is now I suppose! ;) Yeah it isn't a panacea but it has afforded me less frustration than the alternative of jumping straight in. > Since what I've been working on is just a …

  12. comment
    Comment #44473473

    I'm almost to this one in my read through! I'm excited to get to the "information age" arc

  13. comment
    Comment #44471677

    I totally agree! this is the idea behind my recent post actually[1] where I recommend people use AI to write specs before they code. If all you have to do is a human is edit the sp…

  14. comment
    Comment #44471634

    Identifying a problem is the first step to solving it. Changing these buttons to "Rent" would force people to acknowledge what they're getting for their $60 bucks -- effectively a …

  15. comment
    Comment #44451582

    def cleanup inactive specs over time. it's only useful if it'll be used!

  16. comment
    Comment #44445179

    Yeah just learned about this! The md files are actually pretty great for shareability, versioning, and picking up where you left off.

  17. comment
    Comment #44440706

    I need to try Claude Code's plan mode! Definitely feels useful.

  18. comment
  19. comment
    Comment #44439642

    Totally. Yeah I think your approach is a solid take!

  20. comment
    Comment #44438820

    Thanks for the writeup! I talked about a similar, but slightly simpler workflow in my post on "Vibe Specs". https://lukebechtel.com/blog/vibe-speccing I use these rules in all my c…

  21. story
  22. story
  23. comment
    Comment #44397709

    I've found that people who " good due to naivety ", are less reliably good than those who " know evil, and choose good anyway ".

  24. comment
  25. comment
    Comment #44344437

    don't mind me I'm just moloch posting https://slatestarcodex.com/2014/07/30/meditations-on-moloch/