Gianluca Truda

@QVagabond

📚 AI Student 💻 Data Scientist 🧠 Rationalist 🧪 Self-Experimenter 🎙 Co-Host:

Amsterdam, The Netherlands
Vrijeme pridruživanja: lipanj 2018.

Tweetovi

Blokirali ste korisnika/cu @QVagabond

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @QVagabond

  1. Prikvačeni tweet
    18. srp 2019.

    Thrilled to announce that my ML toolkit for dosing is finally open-sourced. It promotes reproducible and open research by evaluating models on the dataset in only a few lines of code. PRs welcome!

    Prikaži ovu nit
    Poništi
  2. 29. sij

    I’d be interested to hear from people who think AI risk is manageable or overblown. What trends/arguments make you optimistic? What directions look most promising to you?

    Poništi
  3. proslijedio/la je Tweet
    29. sij

    So, are brains just neural networks?

    Poništi
  4. proslijedio/la je Tweet
    29. sij
    Poništi
  5. 29. sij
    Poništi
  6. proslijedio/la je Tweet
    28. sij

    Seems like a good time to mention a preprint I wrote about the dynamics of how worst-case infectious diseases are slowed and stopped by public heath and population reaction as the disease spreads. There's even a toy model on to play with!

    Poništi
  7. 27. sij

    Which statement about the risks of Artificial Intelligence do you most agree with?

    Poništi
  8. 27. sij

    A thread, in which Moloch rears his ugly head of civilisational inadequacy once more.

    Poništi
  9. 25. sij

    Just saw this superb video on . Really great summary of how the brand has been so effective globally.

    Poništi
  10. 25. sij

    This seems pretty neat, but perhaps would be mangled by handwriting and look like square roots.

    Poništi
  11. 24. sij

    Perhaps, but coordination is much easier under authoritarianism—that's probably why China has become a powerhouse so rapidly. But it comes associated with many costs, just as with any simple function optimiser.

    Poništi
  12. 24. sij

    ... and I'd love to do a follow-up episode digging into Kotler's findings about contemporary Flow-hacking techniques like transcranial stimulation.

    Prikaži ovu nit
    Poništi
  13. 24. sij
    Prikaži ovu nit
    Poništi
  14. 24. sij

    Just saw 's awesome video on the Flow state! 's work got me onto this topic years ago and, honestly, learning to harness Flow has undoubtedly been one of the most valuable things I've done in the past decade.

    Prikaži ovu nit
    Poništi
  15. 24. sij

    Herbert’s Dune and ’s HPMOR are two books that do a superb job of integrating cognitive processes into the storyline. We’re privy to the scheming, compromises, and game theory going on within characters’ minds. What are other stories that do this well?

    Poništi
  16. proslijedio/la je Tweet
    22. sij

    people don't understand yet but the next 20 years will be dominated by kids who grew up stealing pdfs

    Prikaži ovu nit
    Poništi
  17. proslijedio/la je Tweet
    6. kol 2019.

    I'm collecting interactive ML/DL/Math tools. :) Things like: - GANs in the browser - ConvNet Playground - Activation Atlas - Embedding Projector - Initializing Neural Networks and more. Please add to the list if you know more great resources.

    Poništi
  18. 22. sij

    In other words, think from First Principles 🧠

    Poništi
  19. 22. sij

    I dunno guys, an Escher staircase would certainly be a good way to escape the other two 🤷🏻‍♂️

    Prikaži ovu nit
    Poništi
  20. 21. sij

    Watching Yoshi repeatedly attempt this trick felt like watching the avg. loss creep down on a training neural net 😅

    Poništi
  21. 21. sij

    Pick one to defend you. The other two will come after you. Show your working below...

    Prikaži ovu nit
    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·