Motoki Wu   

@plusepsilon

Wrangling neural networks .

Oakland, CA
medium.com/@plusepsilon
Vrijeme pridruživanja: listopad 2013.
91 fotografija ili videozapis Fotografije i videozapisi

Tweetovi

Blokirali ste korisnika/cu @plusepsilon

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

  1. Prikvačeni tweet
    18. tra 2018.

    Unsupervised learning is a trap.

    6 korisnika označava da im se sviđa
    Poništi
  2. 4. velj

    Don't settle for automating your scripts, make it shift-enterable.

    1 korisnik označava da mu se sviđa
    Poništi
  3. 3. velj

    🌊 Excited to announce our launch and Series A! 🚀 I've had the privilege to work with an incredibly strong product and engineering team that will hop on a plane any day for a customer. Stay tuned, we are just getting started.

    5 korisnika označava da im se sviđa
    Poništi
  4. 26. sij

    GPT-2 is trained on next word prediction. I also expect it to make me breakfast. 🍳

    3 korisnika označavaju da im se sviđa
    Poništi
  5. 15. sij

    Can we just have a sock pair detector and award the creator the Nobel Prize.

    Poništi
  6. 8. sij

    If Earth was back to Pangaea, we would all work remote from the same time zone!

    1 reply 1 korisnik označava da mu se sviđa
    Poništi
  7. 7. sij

    I was looking for a "Wirecutter for cat litter", and there is a Wirecutter for cat litter!

    2 korisnika označavaju da im se sviđa
    Poništi
  8. 3. sij

    Zero shot learning is so good now. Perfect for a baseline model and makes you think about the problem.

    Poništi
  9. 1. sij

    Next year I vow to type fewer "kubectl logs ...".

    1 reply 1 korisnik označava da mu se sviđa
    Poništi
  10. 20. pro 2019.

    I read Ben Horowitz's book and there's a passage that says the best leaders provide 4K monitors to their employees.

    2 proslijeđena tweeta 10 korisnika označava da im se sviđa
    Poništi
  11. 17. pro 2019.

    I proudly overestimate my engineering tasks by 2x, and writing tasks by 10x.

    1 reply 6 korisnika označava da im se sviđa
    Poništi
  12. 14. pro 2019.

    Oh my, tuna belly sandwich 🐟

    Poništi
  13. 6. pro 2019.

    Oh wow, Snorkel is so easy to use now. Props to the team. 🏊

    5 proslijeđenih tweetova 12 korisnika označava da im se sviđa
    Poništi
  14. 5. pro 2019.

    This package implements a bunch of augmentation methods for NLP. Insert a sentence, out comes an augmented sentence. Even includes methods with large language models.

    2 proslijeđena tweeta 3 korisnika označavaju da im se sviđa
    Poništi
  15. 30. stu 2019.

    Write rules like a linguistics PhD in , by my former colleague Jenny Lee. Syntax is underappreciated in NLP pipelines!

    1 reply 37 proslijeđenih tweetova 140 korisnika označava da im se sviđa
    Poništi
  16. 23. stu 2019.

    Replace every "just use X" and "why not X" with "what about X".

    1 korisnik označava da mu se sviđa
    Poništi
  17. 22. stu 2019.

    Interview question: "Why does this [DL technique] work well?" Correct answer: I don't know

    2 korisnika označavaju da im se sviđa
    Poništi
  18. 20. stu 2019.

    I am all aboard the BERT-mobile. 🚅

    Poništi
  19. 31. lis 2019.

    World's first AI ghost writing party. 😱

    1 proslijeđeni tweet 3 korisnika označavaju da im se sviđa
    Poništi
  20. 30. lis 2019.

    Who deploys this deep shit classifier? 💩

    1 reply 1 korisnik označava da mu se sviđa
    Poništi
  21. 25. lis 2019.

    Creating space for in-context annotations. I think I need this.

    3 korisnika označavaju da im se sviđa
    Poništi

@plusepsilon 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:

    ·