jtoy

@jtoy

I study cognition, ponder consciousness, test intelligence models. I like building companies & computational art. Currently writing an AI book

Vrijeme pridruživanja: svibanj 2008.

Tweetovi

Blokirali ste korisnika/cu @jtoy

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

  1. 30. sij

    "Not everyone has an internal monologue" -interesting, but where are the research papers discussing this

    Poništi
  2. proslijedio/la je Tweet
    29. sij

    New blog post: Contrastive Self-Supervised Learning. Contrastive methods learn representations by encoding what makes two things similar or different. I find them very promising and go over some recent works such as DIM, CPC, AMDIM, CMC, MoCo etc.

    Poništi
  3. proslijedio/la je Tweet
    28. sij

    A rat with no brain? Not quite, but a rat equivalent of the famous Lorber "is your brain really necessary?" case has been discovered.

    Poništi
  4. 27. sij

    more proof is coming out that glial cells are doing a lot more brain processing that we previously imaged

    Poništi
  5. proslijedio/la je Tweet
    26. sij

    Nice high-level overview of recent progress: Deep Learning State of the Art (2020) - MIT Deep Learning Series - YouTube

    Poništi
  6. proslijedio/la je Tweet
    22. sij

    An abstract language of thought for spatial sequences in humans

    Poništi
  7. 17. sij

    Excellent paper reviewing how concepts may be stored in the brain in relation to grid cells and current machine learning techniques: "Concepts and Compositionality: In Search of the Brain’s Language of Thought"

    Poništi
  8. 17. sij
    Poništi
  9. proslijedio/la je Tweet

    My machine learning education has progressed to the point where I lose sleep tossing around a “brilliant” idea, only to find the next day that it doesn’t actually work. This is great! I talked about this a few years ago:

    Poništi
  10. proslijedio/la je Tweet
    1. sij

    GPT-2 chess is promising. After an hour of training, 1.5B is pretty good at opening theory. Longer sequences tend to fail due to invalid moves, but this shows it's possible in principle to make a GPT-2 chess engine. And maybe after more training it'll make fewer invalid moves.

    Prikaži ovu nit
    Poništi
  11. 30. pro 2019.
    Poništi
  12. 11. pro 2019.

    Excited for some light reading :) The Consciousness Prior

    Poništi
  13. proslijedio/la je Tweet
    9. pro 2019.

    A concise and beautiful summary of my tutorial! Thanks

    Poništi
  14. proslijedio/la je Tweet

    Happy 91st, Noam Chomsky!

    Poništi
  15. 7. pro 2019.

    nice article on evolution strategies instead of gradient descent for deep reinforcement learning

    Poništi
  16. proslijedio/la je Tweet

    Habits that have a high rate of return in life: - sleep 8+ hours each day - lift weights 3x week - go for a walk each day - save at least 10 percent of your income - read every day - drink more water and less of everything else - leave your phone in another room while you work

    Poništi
  17. 11. ruj 2019.
    Poništi
  18. 2. lip 2019.

    I've started writing a book about artificial intelligence, aiming to publish end of year. The book essentially tries to give a realistic view of where we are in our progress (while we are in the middle of a AI hype craze), and possible directions of research. Ping me for ideas

    Poništi
  19. 26. ožu 2019.
    Poništi
  20. proslijedio/la je Tweet
    21. ožu 2019.
    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:

    ·