Yaroslav Ganin

@yaroslav_ganin

Research Scientist . Previously: Mila (Montréal, Canada), Skoltech (Moscow, Russia). Views are my own.

London, UK
Vrijeme pridruživanja: listopad 2016.

Tweetovi

Blokirali ste korisnika/cu @yaroslav_ganin

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

  1. Prikvačeni tweet
    8. kol 2019.

    You know how they say it's better late than never? Remember SPIRAL () from the last year? Well, I put on my software engineering hat and now we have some code publicly available: . Go ahead and generate your own paintings!

    Prikaži ovu nit
    Poništi
  2. proslijedio/la je Tweet
    15. sij

    We have 2 papers published in today! 🎉 One describes AlphaFold, which uses deep neural networks to predict protein structures with high accuracy. AlphaFold made the most accurate predictions at the 2018 scientific community assessment CASP13. 1/4

    Prikaži ovu nit
    Poništi
  3. proslijedio/la je Tweet
    21. stu 2019.

    This feels like a real breakthrough: Take the same basic algorithm as AlphaZero, but now *learning* its own simulator. Beautiful, elegant approach to model-based RL. ... AND ALSO STATE OF THE ART RESULTS! Well done to the team at

    Poništi
  4. proslijedio/la je Tweet
    30. lis 2019.

    Our new paper: AlphaStar is the first learning system to reach the top tier of a major esport without any game restrictions, achieving Grandmaster status in StarCraft II.  Researchers have been working on the StarCraft series for over 15 years. 

    Prikaži ovu nit
    Poništi
  5. proslijedio/la je Tweet
    30. lis 2019.

    Proud to share our latest results on AlphaStar: ranking higher than 99.8% of players on multiple maps and all races under professionally approved conditions. Full paper soon to come out in Nature, open access version available here

    Prikaži ovu nit
    Poništi
  6. proslijedio/la je Tweet
    24. lis 2019.

    We have released the code for transporter — a neural network architecture for unsupervised learning of object keypoints (now a NeurIPS paper):

    Poništi
  7. proslijedio/la je Tweet
    21. lis 2019.

    Our new work on memory uses a neural network's weights as fast and compressive associative storage. Reading from the memory is performed by approximate minimisation of the energy modelled by the network.

    Prikaži ovu nit
    Poništi
  8. 10. lis 2019.
    Prikaži ovu nit
    Poništi
  9. 10. lis 2019.

    Wanna play around with SPIRAL but the installation seems complicated? I've just built a Docker image to make the experience as hassle-free as possible. To get the agent up and running on your machine follow the instructions here: Have fun!

    Prikaži ovu nit
    Poništi
  10. 4. lis 2019.

    The 19-stroke SPIRAL agent for the oil paint simulator is now available on ! Enjoy 🎨🖌🤖 For usage instructions see The pre-trained snapshot is here:

    Poništi
  11. 3. lis 2019.

    A new paper on tweaking SPIRAL (). What's new: • Spectral normalization of discriminator (Miyato, 18) ⇒ sharper images • Reward shaping by (Ng, 99) ⇒ longer episodes • In-painting instead of stacking ⇒ better reconstructions Lots of nice samples :)

    Poništi
  12. 30. ruj 2019.

    Comes with python bindings!

    Prikaži ovu nit
    Poništi
  13. 30. ruj 2019.

    I just added a new environment to our SPIRAL package. It's a C++ rewrite of the oil paint simulator by . Feel free to use it in your own research (or just for fun). A pre-trained agent is coming soon 🤖 Github: Original:

    Prikaži ovu nit
    Poništi
  14. 20. kol 2019.

    So, , finally. The hardest question predictably came from the audience: "Are those Meeseeks on your t-shirt?"

    Poništi
  15. 17. kol 2019.

    Hey folks, does anyone have a snapshot of the famous DL debate on Yann LeCun’s Google+ page (Oct, 2012)?

    Poništi
  16. proslijedio/la je Tweet

    SPIRAL is an RL agent that generates images by interacting with off-the-shelf graphics software. We've released 9 agents trained to paint portraits of imaginary people in just 19 brush strokes on ! For more, see: →

    Poništi
  17. 9. kol 2019.
    Prikaži ovu nit
    Poništi
  18. 9. kol 2019.
    Prikaži ovu nit
    Poništi
  19. 9. kol 2019.
    Prikaži ovu nit
    Poništi
  20. 8. kol 2019.

    Somehow twitter doesn't like the video of the agent drawing those faces. Had to replace with a static image. For the most adventurous here is the original (uncensored and uncut)🤖

    Ovo je potencijalno osjetljiv multimedijski sadržaj. Saznajte više
    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:

    ·