Joseph Suarez

@jsuarez5341

Agents that scale to their environment + Environments that scale to the real world. MIT PhD Student and creator of Neural MMO.

Vrijeme pridruživanja: ožujak 2019.

Tweetovi

Blokirali ste korisnika/cu @jsuarez5341

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @jsuarez5341

  1. prije 13 sati

    The Neural MMO v1.3 whitepaper is now on arXiv:

    Poništi
  2. proslijedio/la je Tweet
    2. velj

    Got promoted to a full professor at CMU:) Thanks to my amazing students, collaborators, and my mentors. CMU is a remarkable place, but what makes it truly unique is its brilliant students! Hard to find a better place to do ML and AI.

    Poništi
  3. 30. sij

    In the v1.2 release, I mentioned needing to solve a few research problems before making additional environment updates. I was referring to the IO system, which now provides a much smoother API. If you found the platform a bit complex in v1.2, now is a great time to dive back in!

    Prikaži ovu nit
    Poništi
  4. 30. sij

    Additional details:

    Prikaži ovu nit
    Poništi
  5. 30. sij

    Single image files:

    Prikaži ovu nit
    Poništi
  6. 30. sij

    NMMO is an open-source massively multiagent game environment for AI research. v1.3 is the third major update. Highlights: Standardized, streamlined libraries for interfacing with the game’s rich hierarchical IO spaces, trained prebuilt models, tutorials and documentation

    Prikaži ovu nit
    Poništi
  7. 30. sij
    Prikaži ovu nit
    Poništi
  8. proslijedio/la je Tweet
    30. sij
    Poništi
  9. 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
  10. proslijedio/la je Tweet
    6. sij

    I am extremely excited to announce (1) I've joined OpenAI to lead a large-scale effort into AI-generating Algorithms research, & (2) I'll be an Associate CS Professor at U. British Columbia in 2021, where I will continue to lead the OpenAI project. Both are dreams come true! 1/2

    Prikaži ovu nit
    Poništi
  11. 13. pro 2019.

    Lots to learn from this work -- Thank you for making it public

    Poništi
  12. proslijedio/la je Tweet
    27. stu 2019.

    Our 1st seminar hosted entirely in VR :) watch live stream at 16:30 GMT. Speakers list > the creator of 's Neural MMO > co-founder of > will introduce one of OxAI’s most ambitious research projects - Vrai

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

    We're releasing Safety Gym, environments and tools to evaluate reinforcement learning with safety constraints: Aims to ultimately help agents satisfy real-world safety requirements while training (eg not driving off a cliff, not writing abusive content).

    Poništi
  14. 22. lis 2019.

    I do not want to give this undue attention, but it is important that public perception of the good research at be aligned with reality. As such, I've posted a correction to the slide that has been circulating over the past couple of days.

    Poništi
  15. 21. lis 2019.

    OpenAI's title is spot on. That is a robot hand. It solves a Rubik's cube. Details of what parts are learned are given in the blog post. A few of the criticisms here (one object, extra sensors) are factually incorrect, and the blog post presents proper discussion of the rest.

    Poništi
  16. proslijedio/la je Tweet
    15. lis 2019.

    We've trained an AI system to solve the Rubik's Cube with a human-like robot hand. This is an unprecedented level of dexterity for a robot, and is hard even for humans to do. The system trains in an imperfect simulation and quickly adapts to reality:

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

    We've observed AIs discovering complex tool use while competing in a simple game of hide-and-seek. They develop a series of six distinct strategies and counterstrategies, ultimately using tools in the environment to break our simulated physics:

    Prikaži ovu nit
    Poništi
  18. proslijedio/la je Tweet

    Facebook AI is open sourcing MiniRTSv2, a real-time strategy game designed to test and evaluate a range of AI techniques related to reinforcement learning, hierarchical decision-making and natural language processing.

    Poništi
  19. 29. kol 2019.

    v1.3 will continue expanding on the base created by v1.0-v1.2 -- More environment features, more infra upgrades, etc. However, I've also encountered a few research problems along the way that I may need to solve before some such plans are feasible. More about that on the Discord.

    Prikaži ovu nit
    Poništi
  20. 29. kol 2019.

    Bonus: A bit of perspective on progress thus far:

    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:

    ·