Live data from Hacker News

Viewing profile — davidguetta

davidguetta

HN member
Joined
Wed, Apr 06, 2022, 8:38 AM UTC
HN karma
685
Public activity
590 items

About davidguetta

No profile information was provided.

Recent public activity

  1. comment
    Comment #49202177

    especially when 99% tasks don't need fable, and certainly not the next model

  2. comment
    Comment #49109530

    "good interference" vs "bad interference" this is a ludicrous argument in general. having people, even in power, give their opinion like musk or obama is not "interference" stop tr…

  3. comment
    Comment #49095895

    the beginning of the arguments was good but the "because of randomness" is wild

  4. comment
    Comment #49043509

    Rhoda Ai also does vidéo to robot models

  5. comment
    Comment #49004750

    copyright is bullshit

  6. comment
    Comment #48648532

    you only need to handle immigration if you have immigration. problem is kinda solved if you don't have immigration or at least "immigration that cause problems" i know many immigra…

  7. comment
    Comment #48632174

    Hopefully the successor will clean the mess still left realted to the Pakistani rape gangs and police collusion. The entire UK society seem to have been trying to ignore the issue …

  8. comment
    Comment #48559475

    Assuming best intentions, making a colony on mars is gonna require money

  9. comment
    Comment #48542725

    reducing unskilled and hateful immigration is the democratic thing do this has nothing to the other good idea which is to start building AI

  10. comment
    Comment #48360660

    your mind maybe shrugs

  11. comment
    Comment #48349863

    yeah but you dont pretend vim is better

  12. comment
    Comment #48310885

    It's symbolic for it's demographic voters

  13. comment
    Comment #47998660

    This is true unless it isn't basically. People claim that finetuning is good because no model can be 'that' general since gpt3 and at every generation it's becoming less true

  14. comment
    Comment #47902259

    "there will be no need for a scientific theory of deep learning"

  15. comment
    Comment #47900458

    > Creating an "urgency chain" is effectively the same thing as justifying behavior. What ? we SHOULD ABSOLUTELY create an urgency chain

  16. comment
    Comment #47896729

    Trilions were sent as aid and it did not help

  17. comment
    Comment #47883120

    People use their model otherwise they would not.

  18. comment
    Comment #47824478

    laugh in vibe coding

  19. comment
    Comment #47701097

    4 day delay kinda kills the point of the map no?

  20. comment
    Comment #47689520

    Same problem as before the LLM era tho. This is one more argument to the only LLM trained to find "truth"

  21. comment
    Comment #47688548

    What consequence would he suffer if he attack nuclearly vs everything it's done so far ? the marginal cost for him is unfortunately not that high at this point

  22. comment
    Comment #47648557

    still doesn't mean all tokens are useful. it's the point of benchmarks

  23. comment
    Comment #47572681

    Wow ! Only 2 years !

  24. comment
    Comment #47540378

    It's fine as long as the state aligns with you and is generally sane. History shows its not necessary always the case.

  25. comment
    Comment #47540372

    Im not the guy obviously. Its not reactionary to say you dont want the state to interfere too much in your child éducation. Whether you are left or right its fine as long as the st…