Live data from Hacker News

Viewing profile — mattyyeung

mattyyeung

HN member
Joined
Wed, Sep 28, 2011, 3:05 PM UTC
HN karma
98
Public activity
20 items

About mattyyeung

mattyyeung.github.io

Recent public activity

  1. comment
    Comment #45779570

    oooh fantastic. Thanks! Exciting My initial test looks pretty good, seems I will need to muck around a bit to get to "one click runs the simulation and displays the trace and stats…

  2. comment
    Comment #45779436

    Does anybody have advice on how to embed perfetto UI into local tool? I'm currently using the old catapult chrome trace viewer to visualise and inspect scheduler outputs but there …

  3. comment
    Comment #40553834

    You may be interested "Deterministic Quoting"[1]. This doesn't completely "solve" hallucinations, but I would argue that we do get "good enough" in several applications Disclosure:…

  4. comment
    Comment #40388613

    I think you could get a pretty good solution for that using RAG and some tricks with prompt engineering and semantic chunking. With google's very-long-context models (Gemini) you m…

  5. comment
    Comment #40388575

    I would love to learn more, where would you recommend I look?

  6. comment
    Comment #40334918

    Yes, extractive QA is one of the improvements beyond the "minimalist implementation" from the article. In our lingo, we'd say that's another way to create a deterministic quotation…

  7. comment
    Comment #40333469

    Can quotations be hallucinated? Or are you using something like "deterministic quoting"[1]? Disclosure: author on that work. [1] https://mattyyeung.github.io/deterministic-quoting

  8. comment
    Comment #40292841

    Thanks for the thought-provoking comment. It's all grey isn't it? Vanilla RAG is a big step along the spectrum from LLM towards search, DQ is perhaps another small step. I'm no exp…

  9. comment
    Comment #40292627

    I'd put it like this: RAG = search engine, but sometimes hallucinates RAG + deterministic quoting = search engine that displays real excerpts from pages.

  10. comment
    Comment #40292582

    Unfortunately I don't believe that accuracy will scale "multiplicitively". You'll typically only marginally improve beyond 95%... and how much is enough? Even with such a system, w…

  11. comment
    Comment #40292538

    Two possibilities: (1) if the contents (unique reference string) doesn't match, then it's trivially detected. Typically the query is re-run (non-determinism comes in handy sometime…

  12. comment
    Comment #40292459

    Author here, thanks for your interest! Surprising way to wake up in the morning. Happy to answer questions

  13. story
  14. story
  15. story
  16. story
  17. comment
    Comment #35452488

    I didn't really track... apparently there's 30-ish hours of video, but lectures are just the beginning. The real learning happens when you play and build. The first lesson was rele…

  18. comment
    Comment #35452434

    Depends. Some would benefit from simultaneous, others sequential. I did them in "chunks" starting with fastai but that was more driven by the release schedule. Personally I'd recom…

  19. comment
    Comment #35450742

    I just finished this! My thoughts: I recommend it. I feel I can now read an arbitrary paper, frown a lot, and eventually understand what it's talking about - to the point where I c…

  20. story