Shane Gu 顾世翔

@shaneguML

日本🇯🇵-born 华裔🇨🇳 Canadian🇨🇦. RS Brain🇺🇸, PhD 🇬🇧/🇩🇪, EngSci . Secular Deep Reinforced Bayesian 🤖. Views my own.

Mountain View, CA
sites.google.com/view/gugurus
Vrijeme pridruživanja: lipanj 2016.
42 fotografije ili videozapisa Fotografije i videozapisi

Tweetovi

Blokirali ste korisnika/cu @shaneguML

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

  1. Prikvačeni tweet
    5. stu 2019.

    (first tweet!) Our paper got Best Paper Award at CoRL 2019! A summary and extension of imitation learning methods and their application to state marginal matching.

    8 replies 22 proslijeđena tweeta 193 korisnika označavaju da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  2. 2. velj

    Thanks to 机器之心 for discussing our ICLR paper! Impressed by the speed/quality/coverage of their articles on ML/AI. " 2020 9 Perfect Scorers; best paper may be among them?" Glad we got a pretty pic

    2 proslijeđena tweeta 12 korisnika označava da im se sviđa
    Poništi
  3. 2. velj

    50 cognitive biases, quite a few because of energy-intense human brains being so efficient at being more (energy-)efficient, e.g. Google effect, lazy causal inference, self-exceptionalism.

    140 proslijeđenih tweetova 552 korisnika označavaju da im se sviđa
    Poništi
  4. 31. sij

    We should have locomotion challenge for the AI community, maybe using physics models from

    3:12
    This is why has the 100m World Record
    1 proslijeđeni tweet 5 korisnika označava da im se sviđa
    Poništi
  5. 31. sij

    Nice advice from one of the researchers who led me into RL+robotics. I esp. like "honing your taste". In my PhD I actively talked to experts from different fields (DL, RL, Bayes, Kernel, Robot), and I learned a lot. Keep a collection of perspectives

    6 proslijeđenih tweetova 65 korisnika označava da im se sviđa
    Poništi
  6. 30. sij

    I'd say the important thing is both good execution and good storytelling/narration. Without one of high-quality execution and coherent storytelling, the ideas won't be recognized. Good execution includes good narration (paper writing and presentations).

    So glad you’ve internalized this Andrey :) Every PhD student should :)
    2 proslijeđena tweeta 31 korisnik označava da mu se sviđa
    Poništi
  7. 29. sij

    Congratulations to on coming out of stealth and delivering product-level results. It's about time to bring ML revolution to robotics and physical world, as it did for digital world.

    9 proslijeđenih tweetova 70 korisnika označava da im se sviđa
    Poništi
  8. 29. sij

    さんも確か似てる方面の結果も前々から出してました。両方ともこのまま頑張って欲しいですね。 　

    1 korisnik označava da mu se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  9. proslijedio/la je Tweet
    28. sij

    New paper: Towards a Human-like Open-Domain Chatbot. Key takeaways: 1. "Perplexity is all a chatbot needs" ;) 2. We're getting closer to a high-quality chatbot that can chat about anything Paper: Blog:

    i
    27 replies 621 proslijeđeni tweet 1.768 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  10. proslijedio/la je Tweet
    23. sij

    I agree. Jax is next generation framework for

    Have you ever wondered what will be the ML frameworks of the '20s? In this essay, I examine the directions AI research might take and the requirements they impose, concluding with an overview of what I believe to be two strong candidates: JAX and S4TF.
    1 reply 20 proslijeđenih tweetova 125 korisnika označava da im se sviđa
    Poništi
  11. proslijedio/la je Tweet
    19. sij

    Help me Twitter. Best (reasonably technical) math/science podcasts?

    41 reply 48 proslijeđenih tweetova 362 korisnika označavaju da im se sviđa
    Poništi
  12. proslijedio/la je Tweet
    18. sij

    Tried mixed precision yet? Took 10 min to set up and my model runs almost 2x faster with same results. Vars and grads are still 32 bits so it usually doesn't affect predictive performance. E.g. in TF2, set option and make all input to your layers float16 (data, RNN states, ..):

    1 reply 38 proslijeđenih tweetova 241 korisnik označava da mu se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  13. 16. sij

    Encourage students, esp early in ML research, to apply. You will befriend people that you will keeping bumping into through conferences, internships, and arxivs. Besides, Tuebigen is the birthplace of style transfer, and I enjoyed my one year there.

    1 reply 2 proslijeđena tweeta 20 korisnika označava da im se sviđa
    Poništi
  14. proslijedio/la je Tweet
    15. sij

    Really excited to see this work coming to fruition! Jess Sullivan and suggested doing a longitudinal head cam study of their own kids seven years ago and now these super-rich data are in their way out into the world!

    21 proslijeđeni tweet 122 korisnika označavaju da im se sviđa
    Poništi
  15. proslijedio/la je Tweet
    8. sij

    Many of our group's recent papers - DualDICE, ValueDICE, GenDICE, AlgaeDICE - can be framed as applications of this duality. Still, lots of potential remaining applications for others to explore, and lingering questions of how these formulations interplay with stoch. opt methods.

    Fenchel-Rockafellar duality is a powerful tool that more people should be aware of, especially for RL! Straightforward applications of it enable offpolicy evaluation, offpolicy policy gradient/imitation learning, among others
    4 proslijeđena tweeta 14 korisnika označava da im se sviđa
    Poništi
  16. 10. sij

    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  17. 10. sij

    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  18. 10. sij

    exciting demo by . Non-invasive wearable that reliably can sense spatial attention. I was able to spend reliable spatial command per 1-2s, after only 45s of calibration phase.

    1 reply 1 proslijeđeni tweet 9 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  19. proslijedio/la je Tweet
    8. sij

    Fenchel-Rockafellar duality is a powerful tool that more people should be aware of, especially for RL! Straightforward applications of it enable offpolicy evaluation, offpolicy policy gradient/imitation learning, among others

    31 proslijeđeni tweet 108 korisnika označava da im se sviđa
    Poništi
  20. 7. sij

    Finally, we can start putting ads on people

    8 proslijeđenih tweetova 14 korisnika označava da im se sviđa
    Poništi
  21. proslijedio/la je Tweet
    2. sij

    🔥NLP Year in Review -- 2019🔥 I have put together a report that contains a list of the top NLP and ML highlights that I came across in 2019.

    90 proslijeđenih tweetova 226 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi

@shaneguML još nije objavio nijedan Tweet.

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:

    ·