Live data from Hacker News

Viewing profile — hiroakiaizawa

hiroakiaizawa

HN member
Joined
Tue, May 05, 2026, 5:35 AM UTC
HN karma
4
Public activity
13 items

About hiroakiaizawa

Running simple reproducible tests on concentration of extreme events.

Colab (≈30 sec): https://colab.research.google.com/drive/1FGu7_G1avub-0PUV6cpdox8cojuOvY9C

Optional: https://zenodo.org/records/20020739

Recent public activity

  1. comment
    Comment #48219766

    Good reminder that raw tokens/sec numbers can be misleading without latency and context-window considerations.

  2. comment
    Comment #48219758

    Interesting approach. I like that the implementation focuses on scalability rather than only visualization.

  3. comment
    Comment #48083560

    The notebook is intentionally minimal. Not a prediction model or causal explanation — just a reproducible concentration check with fixed ex-ante definitions and minimal outputs. Ru…

  4. comment
  5. story
  6. comment
    Comment #48044659

    One thing I've started appreciating with LLM-assisted workflows is how important fixed evaluation protocols are. Without pre-defined definitions and locked procedures, it's extreme…

  7. comment
    Comment #48044631

    [flagged]

  8. comment
    Comment #48021744

    Curious what domains people would try this on. Would love to see other datasets.

  9. comment
    Comment #48021265

    Tested on finance / power / earthquakes. Minimal version only. Happy to adapt to other datasets.

  10. story
  11. comment
    Comment #48018579

    Interesting. What are the main latency bottlenecks in practice?

  12. comment
    Comment #48018552

    Nice. What scale does this realistically reach on a single machine?

  13. comment
    Comment #48018539

    Interesting. What are the main trade-offs they expect from the switch?