Live data from Hacker News

Viewing profile — bytesandbits

bytesandbits

HN member
Joined
Sat, Jan 14, 2023, 11:28 PM UTC
HN karma
175
Public activity
72 items

About bytesandbits

No profile information was provided.

Recent public activity

  1. comment
    Comment #48995477

    I imagine it is based off World Labs

  2. comment
    Comment #48986133

    absolutely crazy level of detail... in awe

  3. comment
    Comment #47994435

    I thank Microsoft deeply for the forced copilot crap, almost impossible to remove, that they have put into vs code. Finally after 5 years I have deleted vs code in my Mac! that was…

  4. comment
    Comment #47880275

    I know a 10x engineer when i see one.

  5. comment
    Comment #47696312

    fine. I will port it myself. Real-time, sub 100ms latency. Here https://testflight.apple.com/join/myNP5XvU

  6. comment
    Comment #47670119

    Handy with parakeet is pretty awesome by the way!

  7. comment
    Comment #47635047

    here the original NASA photos at high resolution without unnecessary ads. https://www.nasa.gov/gallery/journey-to-the-moon/

  8. comment
    Comment #47623946

    why is a mail client needed in an onboard space computer at all ?

  9. comment
    Comment #47540055

    anything that doesn't touch the model parameters at all once it has been compiled. for example, in streaming ASR of an encoder-decoder you can get gains in accuracy just by enhanci…

  10. comment
    Comment #47539554

    we constantly underestimate the power of inference scaffolding. I have seen it in all domains: coding, ASR, ARC-AGI benchmarks you name it. Scaffolding can do a lot! And post-train…

  11. comment
    Comment #47304602

    https://machinelearning.apple.com/research/exploring-llms-ml...

  12. comment
  13. comment
    Comment #47245157

    4x faster PREFILL not decode. Decode is bandwidth-bounded. Prefill is flops-constrained.

  14. comment
    Comment #47228624

    do you run two eSIMs when traveling and if so how is stability / battery life?

  15. comment
    Comment #47228554

    incredible work

  16. comment
    Comment #47205386

    I am not sure what "edge" device you want to run this on, but you can compress parakeet to under 500MB on RAM / disk with dynamic quants on-the-fly dequantization (GGUF or CoreML c…

  17. comment
    Comment #47205340

    sensei karpathy has done it again

  18. comment
    Comment #47148107

    parakeet v3 has a much better RTFx than moonshine, it's not just about parameter numbers. Runs faster. https://huggingface.co/spaces/hf-audio/open_asr_leaderboard

  19. comment
    Comment #47056677

    maybe a deepseek v4 distill. give it a few days

  20. comment
    Comment #47001548

    its cause of a chain of events. Next week Chinese New year -> Chinese labs release all the models at once before it starts -> US labs respond with what they have already prepared a…

  21. comment
    Comment #46983061

    Not trained in Ascend that is BS. Hopper GPU cluster. Please remove that.

  22. comment
    Comment #46895611

    wow Mistral really cooked

  23. comment
    Comment #46881659

    how are you so fast man

  24. comment
    Comment #46881603

    John Giannandrea leaves and things start to work well in the Apple AI department. Maybe just a coincidence :)

  25. comment
    Comment #46881590

    I will never understand why when you expect to sell a few hundred units of a product at most, you would decide to go and build something in china. There are instances where buildin…