Florian Laurent

@MasterScrat

Deep Reinforcement Learning 🤖 - Timelapse Photography 📷 - Running 🏃 - Alumni - Freelance, current job: writing a DRL course

Lausanne, Switzerland
Vrijeme pridruživanja: svibanj 2009.

Tweetovi

Blokirali ste korisnika/cu @MasterScrat

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

  1. proslijedio/la je Tweet
    30. sij
    Prikaži ovu nit
    Poništi
  2. 3. velj
    Poništi
  3. proslijedio/la je Tweet
    31. sij

    Dear RL Twitter, we are having a discussion with the Stable Baselines maintainers about its future, and we need your opinion. As a big change is required anyway, what framework would you prefer to be the base for V3.0 ?

    Poništi
  4. proslijedio/la je Tweet
    2. velj

    Wow: Google's "Meena" chatbot was trained on a full TPUv3 pod (2048 TPU cores) for **30 full days** - That's more than $1,400,000 of compute time to train this chatbot model. (! 100+ petaflops of sustained compute !)

    Poništi
  5. 3. velj

    TensorFlow → TensorFlow 2.0 reminds me of Angular → Angular 2.0: in both cases, a cooler contender came out in the interval (PyTorch/React), so when Google breaks backward compatibility people jump to the contender, since in any case they need to do a full rewrite

    Poništi
  6. 3. velj

    Didn't know was also located in France (Paris)! Nice to see some top-notch research from there :D

    Poništi
  7. proslijedio/la je Tweet
    2. velj

    Performance artist generates virtual traffic jams in Google Maps by pulling a wagon full of smartphones

    Poništi
  8. 30. sij

    OpenAI "Spinning Up in RL" now available in PyTorch 🔥

    Poništi
  9. proslijedio/la je Tweet
    11. sij

    “We’re turning AI development into sports.” founded crowdsourcing platform to help further and to tap into talent from around the globe. Full interview:

    Poništi
  10. proslijedio/la je Tweet
    30. sij

    Humans learn from curriculum since birth. We can learn complicated math problems because we have accumulated enough prior knowledge. This could be true for training a ML/RL model as well. Let see how curriculum can help an RL agent learn:

    Poništi
  11. proslijedio/la je Tweet
    29. sij

    Our paper 'A distributional code for value in dopamine-based reinforcement learning' on the cover of ! Read it here: Shout out to the amazing artists/designers at who make this possible, while we get to focus on the research.

    Poništi
  12. proslijedio/la je Tweet
    29. sij

    It’s happening! Congratulations to @_denormalized They are killing it deploying RL in the real world at major businesses. So impressive & hard to apply research to real robotic solutions at scale already!

    Prikaži ovu nit
    Poništi
  13. proslijedio/la je Tweet
    29. sij

    I propose we rename Google messaging services with the year of release, just like a copy of Madden or FIFA. The provides for easier tracking and better communicates the ~yearly lifecycle of these apps.

    Prikaži ovu nit
    Poništi
  14. proslijedio/la je Tweet
    28. sij

    Come join me for a talk on how to solve simulation to real transfer with at 2:35pm cc .. ps there are some fun videos

    Poništi
  15. proslijedio/la je Tweet
    28. sij

    What if we see level generation as a game, where each action changes the level in some way? Well, we could use RL to learn to "play" level generation! We introduce Procedural Content Generation via Reinforcement Learning (PCGRL), a new paradigm for PCG.

    Prikaži ovu nit
    Poništi
  16. proslijedio/la je Tweet
    26. sij

    Pixels know which way they roll and are completely customizable, letting you control how they light up based on your rolls. More info / sign-up at

    Prikaži ovu nit
    Poništi
  17. 27. sij
    Poništi
  18. 26. sij
    Poništi
  19. 26. sij

    And that's a wrap! Congratulations to team "drone_crash" from who were the first to reach a positive score - and who firmly defended their top spot until the very end 🥇 Thanks to and for the prize!

    Prikaži ovu nit
    Poništi
  20. 26. sij

    The fight for the top places on the leaderboard rages on, with participants slowly getting to a score of 0 - a challenging task 💪

    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:

    ·