Live data from Hacker News

Viewing profile — wolfgangK

wolfgangK

HN member
Joined
Sat, Jun 04, 2011, 8:43 AM UTC
HN karma
201
Public activity
70 items

About wolfgangK

No profile information was provided.

Recent public activity

  1. comment
    Comment #49195950

    You understand that drones are used for precision strikes instead of indiscriminate bombing, right ? It's the opposite of "sponsoring a genocide" (if there actually ever was one ha…

  2. comment
    Comment #49174671

    Everything is food and air. There is no such thing a "human labor" without food and air to sustain it. I'm not sure what your point is.

  3. comment
    Comment #49129482

    BTW, I don't understand why having the driver converting AC to DC inside the LED spots is the default. For a new house, it seems to make sense to me to have at least one external d…

  4. comment
    Comment #48522047

    I presume this works (will work) also for JupyterLite that is based on Pyodide ? Would be great if it helped getting the latest OpenCV-python version [0] and it's dnn goodies being…

  5. comment
    Comment #48482933

    OpenCV being in the list of Pyodide modules [0] was the biggest boon for my online teaching experience because remotely dealing with install woes (corporate proxies & cie) was a sh…

  6. comment
    Comment #47514220

    Nice ! My most pressing request for VSS would be efficient binary vectors : is this on the table ?

  7. comment
    Comment #45078979

    Only those who don't care/know about prompt processing speed are buying Macs for LLM inference.

  8. comment
    Comment #44850527

    For LLM inference, I don't think the PCIe bandwidth matters much and a GPU could improve greatly the prompt processing speed.

  9. comment
    Comment #44850380

    Indeed, recent Flash Attention is a pain point for non CUDA.

  10. comment
    Comment #44834003

    The idea is presumably that you would "sell" at an artificially low price.

  11. comment
    Comment #44833951

    > The Soviets had the […]first woman,[…] That is quite the claim !

  12. comment
    Comment #44576863

    You forgot the "/s", or do you actually believe that it's capitalism's fault is a mother taking care of her children is "unpaid labor" ?

  13. comment
    Comment #44576849

    "is hard" ≠ "sucks"

  14. comment
    Comment #43936490

    Most interesting ! Would you mind sharing the prompt and the resulting CLAUDE.md file ? Thx !

  15. comment
    Comment #43275058

    IMO, it would be more interesting to have a 3-way comparison of price/performance between DeepSeek 671b running on : 1. M3 Ultra 512 2. AMD Epyc (which Gen ? AVX512 and DDR5 might …

  16. comment
    Comment #42973109

    DeepSeek is not a model.Which model did you use (v3 ? R1 ? a distillation ?) at which quantization ?

  17. comment
    Comment #42705273

    Nice ! Is it possible to connect to an in browser DB like WASM DuckDB https://duckdb.org/docs/api/wasm/overview.html or https://github.com/babycommando/entity-db ? That would be mo…

  18. comment
    Comment #42685165

    It seems that this aims to refutes claims for inaction with facts about spending money. However, the high speed rail project or homelessness management seem to show that in Califor…

  19. comment
    Comment #42593964

    How do we know that this extension can be trusted ?

  20. comment
    Comment #42578460

    «Unfortunately, I have only seen 3 models, 3B or over, handle RAG.» I would love to know which are these 3 models, especially if they can perform grounded RAG. If you have models (…

  21. comment
    Comment #42570246

    Just played a bit with it. Were you working with ASCII ? This example didn't work for you ? https://github.com/jfalcou/eve/blob/a141ba93048bb2916c2157a9...

  22. comment
    Comment #42563256

    Counterpoint : your message is not synthetic data and will contribute to lots of LLMs saying the same. Many such cases ? (It seems to me obvious that a fgrep would sanitize synthet…

  23. comment
    Comment #42563207

    DeepSeek v3 can run on CPU & RAM : https://www.reddit.com/r/LocalLLaMA/comments/1hqidbs/deepsee... Epyc Gen4 and 12 memory channels of DDR5 @4800 should give you 7 to 9 t/s.

  24. comment
    Comment #42563172

    I don't think that Python would be the right language for such low-level performance maxxing endeavor. I would have picked C++ but t was eye opening for me to see how rust enabled …

  25. comment
    Comment #42563161

    Amazingly thorough ! I love how the author leaves no stone unturned. I had no idea you could do the kind of low level efficiency shaving in Rust. I wonder how a C++ implementation …