Maxim Kochurov

@ferrine96

MSc , dev & Geoopt, member in Samsung AI. Also a great fan of Bayesian statistics and approximate inference. Love rollerscating.

Moscow, Russia
Vrijeme pridruživanja: prosinac 2016.

Tweetovi

Blokirali ste korisnika/cu @ferrine96

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @ferrine96

  1. 26. sij

    Spent the whole day looking for a font for st terminal that is zsh friendly. Finally, nerd-fonts-dejavu-complete, feel delighted😌

    Poništi
  2. proslijedio/la je Tweet
    21. sij

    We are organizing an Workshop on Geometric and Relational Deep Learning! Registration invites will be shared soon. Interested in participating? Consider submitting an abstract or get in touch: w/

    Prikaži ovu nit
    Poništi
  3. 25. pro 2019.

    Wow, Christmas/New Year in Moscow is so fancy

    Poništi
  4. proslijedio/la je Tweet
    14. pro 2019.
    Poništi
  5. 10. pro 2019.

    NeurIPS poster session is going fine

    Poništi
  6. 8. pro 2019.

    Hey! I'm at ! Whistler was great⛷️! Will be happy to chat about a lot of things: , , Bayesian Stat Models and Hyperbolic Deep Learning!

    Poništi
  7. proslijedio/la je Tweet
    8. pro 2019.

    Excited for , will present a poster in Sunday with . If you're at the conference and would like to chat, let me know!

    Prikaži ovu nit
    Poništi
  8. 29. stu 2019.
    Poništi
  9. proslijedio/la je Tweet
    23. stu 2019.

    Max is absolutely right. Lots of people in ML/DL know that causal inference is an important way to improve generalization. The question is how to do it. Prominent ML people like Bernhard Schölkopf, Léon Bottou and...

    Prikaži ovu nit
    Poništi
  10. 22. stu 2019.

    API simplifications for pyro to perform MCMC and VI 🤗

    Poništi
  11. 18. stu 2019.

    Literally what I do as a Bayesian adept: No. No. Do not use p-values, they may give you wrong intuition. No. You do not necessarily need flat priors. No. Point estimate is not the only thing we can have from posterior. No. Don't just stare at your posterior, use it in your loss!

    Poništi
  12. 15. stu 2019.

    Why don't people do their research in open github repos? What are the concerns or fears behind it? As I think, it adds more transparency and collaboration as well as reproducibility. However, this would require discipline, and plagiarism protection.

    Poništi
  13. proslijedio/la je Tweet
    13. stu 2019.

    Research efforts in computer vision and are on the rise. To accelerate 3D research, NVIDIA releases Kaolin as a PyTorch library. See how researchers use Kaolin to move 3D models into the realm of neural networks.

    Poništi
  14. 13. stu 2019.
    Poništi
  15. 3. stu 2019.

    Long way passed. We've uploaded v0.1.0 geoopt on pypi🎉 Now with Hyperbolic stuff and mixed curvature product manifolds that are properly treated in Riemannian Adam. $ pip install geoopt of visit

    Poništi
  16. 28. lis 2019.
    Poništi
  17. 19. lis 2019.
    Poništi
  18. 2. lis 2019.

    Special thanks to , , who helped with submission and implementation. Also thanks to , , , and Robert Goldman for discussions and paper review. You are so great!

    Prikaži ovu nit
    Poništi
  19. 2. lis 2019.

    I'm so excited to share that our () PyMC4 abstract will appear on NeurIPS workshop. We have positive experience with coroutines based PPL design on top of tf2.0 + tfp. We are eager to answer any questions on our design or future development.

    Prikaži ovu nit
    Poništi
  20. proslijedio/la je Tweet
    16. ruj 2019.
    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:

    ·