Live data from Hacker News

Viewing profile — piecerough

piecerough

HN member
Joined
Sun, Jan 31, 2021, 10:12 PM UTC
HN karma
260
Public activity
51 items

About piecerough

No profile information was provided.

Recent public activity

  1. comment
    Comment #45380227

    Have you tried 2.5 Flash Lite to cut costs further?

  2. comment
    Comment #45152389

    "quantize enough" though at what quality?

  3. comment
    Comment #43393371

    What's a decent european enterprise?

  4. comment
    Comment #43062221

    What US government attacks?

  5. comment
    Comment #43057999

    [...] > But there is something fundamentally different about talking with a bot as opposed to a person. A person can be a friend. An AI cannot be a friend, despite how people might…

  6. story
  7. comment
    Comment #42829146

    SFT forces the model to output _that_ reasoning trace you have in data. RL allows whatever reasoning trace and only penalizes it if it does not reach the same answer

  8. comment
    Comment #42825480

    I think the reason why it works is also because chain-of-thought (CoT), in the original paper by Denny Zhou et. al, worked from "within". The observation was that if you do CoT, an…

  9. comment
    Comment #42526655

    Who's Tavi?

  10. comment
    Comment #42229612

    Have you had a common theme for these projects you navigated?

  11. comment
    Comment #42003786

    It's very related to LLMs. Though instead of text tokens you are working with audio tokens (e.g. from SoundStream). Then you go to audio corpus, instead of text corpus.

  12. comment
  13. story
    AI Startups Acquisition Models

    So it turns out that the AI boom introduced a new exit strategy compare to 10 years ago. Inflection, Adept, Character and many others they all went through founder drain. Big Tech …

  14. story
  15. comment
    Comment #40833817

    It's great!

  16. comment
    Comment #40590759

    > It would be very interesting if LLMs were no longer static. Little bit of a nightmare too. Instructions keep piling up for you that you no longer openly can access and remove

  17. comment
    Comment #40404750

    > I remember a French institution could not buy our product, because they had a contract with a local manufacturer. I doubt this is a EU thing. It's due to exclusive contracts/lice…

  18. comment
    Comment #40268742

    This is only going to get worse with Large Language Models. Let's imagine a somewhat knowledgeable individual, could craft both emails, messages and even commits with a bunch of pr…

  19. comment
    Comment #40109888

    Isn't this what we're all betting massive Transformer architectures are going to give us? Tools to explore and handle complex concepts. Reasoning may still be left to us, though.

  20. comment
    Comment #39982833

    "We are also releasing three new datasets: Screen Annotation to evaluate the layout understanding capability of the model, as well as ScreenQA Short and Complex ScreenQA for a more…

  21. comment
    Comment #39597614

    That seems brutal, indeed. Why did you move there in the first place?

  22. story
    Low-growth FAANG vs. High-growth Startups

    I'm a FAANG employee and I believe high growth days are over. Layoffs and cost cutting are taking the fun out of these big tech conglomerates, making it not just hard to grow caree…

  23. comment
    Comment #39436226

    So what's next?

  24. comment
    Comment #39339364

    As a FAANG employee, working with ML, what do you want to get from other companies, besides more money? It's hard to have more chips, for example. You run less experiments, you hav…

  25. comment
    Comment #39095552

    In today's market, if you are available for the intro call, recruiters go hunt for the next hard-to-get candidate. Bigger likelihood it'll be an actual conversion. Happened to me.