Live data from Hacker News

Viewing profile — i386

i386

HN member
Joined
Mon, Apr 20, 2009, 11:25 AM UTC
HN karma
2,797
Public activity
778 items

About i386

No profile information was provided.

Recent public activity

  1. comment
    Comment #49180877

    Its an entire misunderstanding of how the layout system works.

  2. comment
    Comment #49138082

    Most people working at the top of these large organizations don’t know how to make anything net new anymore. They pushed all the people that do out when padding their nests.

  3. comment
    Comment #48884710

    exo is Mac only.

  4. comment
    Comment #48879661

    I’d love a bug report - we can get it working for you!

  5. comment
    Comment #48879654

    Our skippy library is a patch queue on top of llama that allows us to access internal information, such as activations, and filter tensors on model load.

  6. comment
    Comment #48877850

    To be honest, both are very tough problems we don't have a good answer for yet. If that is something that concerns you, look into building a private mesh with trusted peers.

  7. comment
    Comment #48877841

    Each stage has its own KV for the layers it hosts. You are on the money there, when one stage is waiting it's free for more parallelism. I am planning on exploiting this for more t…

  8. comment
    Comment #48877785

    It can dynamically route. If a machine drops out of split, the topology is recalculated and the request is automatically retried.

  9. comment
    Comment #48877438

    Thank you! AMD is a weak spot in our testing right now. If you’re willing to contribute or let us borrow some compute time, drop in on the Discord.

  10. comment
    Comment #48877390

    The lab features two Mac Studios: an Apple M3 Ultra (32 CPU cores, 80 GPU cores, 256 GB unified memory) and an Apple M1 Ultra (20 CPU cores, 48 GPU cores, 128 GB unified memory), b…

  11. comment
    Comment #48877215

    That isn’t true. llama RPC is incredibly slow but staged splits in skippy are orders of magnitude faster.

  12. comment
    Comment #48877199

    This was done on my home lab simulating 5ms latency and jitter between machines. Splits work quite well if you your nodes are over WAN at metro latency’s but not super fast on glob…

  13. comment
    Comment #48877119

    I’m one of the contributors to Mesh LLM and happy to answer any questions. I authored the skippy engine that allows you to split large models across nodes.

  14. comment
    Comment #47991852

    PMs at Microsoft have incredibly bad taste

  15. comment
    Comment #47921191

    I kinda fucking love it. He’s doing something. It’s interesting. There’s no scam. Let live!

  16. comment
    Comment #47622554

    You can try this new model live using mesh-llm right now: https://www.anarchai.org/dashboard

  17. comment
    Comment #47614048

    switch on version, then messageId…

  18. comment
    Comment #47614019

    > I didn't know that Protobuf wasn't canonical but even without this knowledge, there are many other factors which make it an inferior format to JSON. If strict typing and versioni…

  19. story
  20. comment
    Comment #46995187

    yeah this product does not exist

  21. comment
    Comment #46545229

    I work for YouTube. You’re hired.

  22. comment
    Comment #46487010

    At the WiseTech office in 2006 we had a small shrine in the corner dedicated to Tridge, the patron saint of hackers.

  23. comment
    Comment #45978194

    This is very revisionist. While they have been catching up quickly there was no master 4D chess strategy here. Google was incredibly late to this game - Sergey had to come back fro…

  24. comment
    Comment #45575033

    Android supremacy at its finest. I would never recommend a family member buying one. The history of this kind of thing is long and keeps continuing to happen.

  25. comment
    Comment #45112256

    > What makes a group of coworkers into "an org"? Reporting lines that ladder up to a line in the P&L.