Jakob Foerster

@j_foerst

Research Scientist at FAIR, ML PhD from , incoming assistant Prof at . Focused on Multi-Agent Learning. Views and opinions are mine.

Vrijeme pridruživanja: studeni 2012.

Tweetovi

Blokirali ste korisnika/cu @j_foerst

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

  1. Prikvačeni tweet
    11. ruj 2019.

    Excited to be starting as an Assistant Prof (👨‍🎓!!) at the (Scarborough Campus) w/ appointment at the in September of 2020. I am looking for exceptional Master/PhD students and Postdocs to be starting with me next fall. Till then, ..

    Prikaži ovu nit
    Poništi
  2. 17. sij

    . : It's time to build (and open-source) the real "Android-Auto". Electric cars are supersized smartphones on wheels with extremely large batteries. Helping et al get started in the electric car business seems like a great way to help slow down climate change.

    Poništi
  3. proslijedio/la je Tweet
    8. sij

    Great kick-off to the new year: presents his fundamental work on Multi-Agent Learning from DIAL & Differentiable Communication Protocols to Multi-Agent Search in Hanabi ! 🦾🧠🧑‍💻

    Poništi
  4. proslijedio/la je Tweet
    1. sij

    Over the last few years, different communities (multi-agent, optimization & control) have worked on learning in differentiable games. Now you can try a firework of methods (🎇SOS, LOLA, LA, SGA, CO, EG, CGD, LSS🎇) in a single colab notebook: . Happy 2020!

    Poništi
  5. 31. pro 2019.

    Alistair Letcher is now on Twitter! Welcome to the dark side :)

    Poništi
  6. 22. pro 2019.

    Our "Simplified Action Decoder" (, w/ Hengyuan Hu), current SOTA for RL(w/o search) on 2-5 player Hanabi🎇 will be spotlight! The code, , includes trained agents and fast Pytorch version of R2D2&Ape-x 🔥🔥

    Poništi
  7. 14. pro 2019.

    Sushi ? I thought we were past peak-hype in ML. What is this, 2018??

    Poništi
  8. 14. pro 2019.

    "Arrival" can manage to communicate with this alien species of expert Hanabi players??

    Poništi
  9. 13. pro 2019.

    On a remote planet 🪐, SAD and SPARTA have been playing billions of games of Hanabi🎇, perfecting their skills. They just arrived on planet earth🌏! Come and be amongst the 1st humans to sit down and play cards with this alien species on Sat . Can you understand them?

    Poništi
  10. proslijedio/la je Tweet
    10. pro 2019.

    I wrote a blog post explaining the basics of emergent communication and why it's a fascinating field. It's a great read if you're interested in the workshop on emergent communication at NeurIPS this Saturday 😀

    Poništi
  11. 10. pro 2019.

    - Safety and Robustness in Decision Making (Friday): "Can I Trust the Explainer?.." () - Deep RL and Emergent Communication workshops: "Improving Policies via Search in Cooperative Partially Observable Games" () Enjoy the conference!

    Prikaži ovu nit
    Poništi
  12. 10. pro 2019.

    I'll be all week. Find me at the two posters on Wednesday 5-7 PM: #201 ("Multi-Agent Common Knowledge RL", ), and #207 ("Loaded DiCE:.." ). I'll also be presenting recent collaborations at the following workshops:

    Prikaži ovu nit
    Poništi
  13. proslijedio/la je Tweet
    10. pro 2019.

    Can We Trust the Explainers? Excited to share our work on Verifying Post-hoc Explanatory Methods If you are at NeurIPS, come by our poster at the Workshop on Safety and Robustness in Decision Making on Friday!

    Poništi
  14. 6. pro 2019.

    The research on Hanabi just got a lot more exciting - today we are adding search to the mix, vastly improving upon the previous SOTA 🎆🎇🤖 We are open-sourcing all code, incl. a new RL method and trained agents. A cooperative effort with , Hengyuan Hu,

    Poništi
  15. 18. stu 2019.

    ...If you *really* find this work interesting, then you may want to apply to for Masters/phd/post-doc so we can help write the next chapters of this story!

    Prikaži ovu nit
    Poništi
  16. 18. stu 2019.

    Nice review of some of the work that's been happening in the vibrant field of emergent communication! If you find this interesting, also come check out the third version of our workshop on the topic :) 🤖🗨️🤖🗨️🤖 & 🤖🗨️🤖🗨️🙎‍♀️

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

    We did a research - ! It's about emergent languages and asks the Q: "How does the capacity of the agents bias the resulting language?" The result was surprising!

    Prikaži ovu nit
    Poništi
  18. proslijedio/la je Tweet
    26. lis 2019.

    Happy to announce that WikiCREM, our new and large dataset for coreference resolution, is available online at . More details in our paper , accepted at EMNLP 2019. Come by our poster if you attend the conference!

    Poništi
  19. 24. lis 2019.

    Thanks to for pointing this out today at the office when he saw me working on the CRC :)

    Prikaži ovu nit
    Poništi
  20. 24. lis 2019.

    In case you are finalizing your camera ready copy and (like me) you didn't know the below, you can now stop trying to fit it into 8 pages.. "If your submission is accepted, you will be allowed a ninth content page for the camera-ready version." from:

    Prikaži ovu nit
    Poništi
  21. proslijedio/la je Tweet

    ., just announced the creation of 3 faculty positions, including one shared between + 🤖🧠:

    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:

    ·