noahtren

@noahtren

Vrijeme pridruživanja: travanj 2019.

Medijski sadržaj

  1. 3. velj

    their autopilot is incredible, check out this talk I bet the hackathon would be fun even considering that they use pytorch 😉

  2. 27. sij
  3. 22. sij
    Odgovor korisniku/ci

    I think I see some early code for some adversarial networks -- can't wait to see the outcome. Neat code, I like the variables for what the neural network is "saying" vs. "thinking" -- a great way to describe the difference between a signal and a message 😁

  4. 4. sij

    ok, last visualization for Twitter, this is the best one by far. check out the code here:

    Prikaži ovu nit
  5. 3. sij

    first timelapse of evolving a visual language. interesting to see shapes go quickly from square grids to more organic patterns, almost like ink on a page. 🤔 will run again with some different parameters, hopefully to reduce checkerboard artifacts and get more distinct glyphs

    Prikaži ovu nit
  6. 2. sij

    evolving an artificial alphabet that is robust to noise. model is still converging, but this snapshot is super fun. looks like ancient runes, or an alien language 👽

    Prikaži ovu nit
  7. 30. pro 2019.

    interestingly, the generated images have different underlying patterns after each run. 🤔 this is probably because the task is too easy (no noise + few unique symbols) and the neural networks quickly converge to a good solution, depending on however they were initialized

    Prikaži ovu nit
  8. 30. pro 2019.

    implementing 's Dimension of Dialogue work in TensorFlow 2 and getting some good results after just a few hours of coding 😀 I think the real fun will begin once some noise is introduced to the communication channel. here's an example of some early gray-scale images:

    Prikaži ovu nit
  9. 29. pro 2019.

    this maybe seems petty, but using hotkeys is a pretty underrated game-changer for computer work. consider if you could save 2 seconds each minute by not having to reach for the mouse: ()

    Prikaži ovu nit
  10. 21. pro 2019.
    Odgovor korisnicima

    here's another fun one

  11. 21. pro 2019.
    Odgovor korisnicima

    definitely -- just tested GPT-2 for generating metaphors and got a pretty good result (cherry-picked from 3 candidates). this will only get better with language models that are trained on knowledge graphs (see "ERNIE" 😛)

  12. 14. stu 2019.
    Odgovor korisnicima

    the level of detail in this taxonomy is awesome... also looks like they have a couple more interactive demos on the site for anyone curious was surprised to see Photomath at the top of this list

  13. 12. stu 2019.
    Odgovor korisniku/ci

    agreed, my impression is that there is a whole lot of low-hanging fruit (ripe for picking 😉) if you are looking to get into it maybe check out this mildly entertaining vid () or hmu a month from now, I might have an alternative

  14. 12. stu 2019.
    Odgovor korisniku/ci
  15. 8. stu 2019.

    on understanding math/physics: "my rough feeling is that there's some threshold that one crosses in abstraction when one becomes kind of like a Turing machine" (timestamp link: )

    Prikaži ovu nit
  16. 29. lis 2019.
    Prikaži ovu nit
  17. 22. lis 2019.

    this video, A Brief History of Japan

    Prikaži ovu nit
  18. 22. lis 2019.

    I think this one had two parts originally. never took bio so idk what's going on

    Prikaži ovu nit
  19. 18. lis 2019.

    hopefully more people become aware of this issue and start implementing tools for interpretability. ex: Andrej Karpathy visualizing LSTM activations in natural language and source code:

  20. 16. lis 2019.

    this just in, I discovered that has changed their gray background color within the past month. did you notice???

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:

    ·