Twiitit

Olet estänyt käyttäjän @subramen

Haluatko varmasti nähdä nämä twiitit? Twiittien näyttäminen ei poista käyttäjän @subramen estoa.

  1. 1. jouluk.
    Kumoa
  2. 23. marrask.

    In the 90s, ML researchers invented multi-task learning to use the fact that different interrelated tasks often use the same underlying concepts. Today et. al have a 107(!) NLP tasks model. If we end up with AGI, it's going to be a hyperscale MTL model.

    Kumoa
  3. uudelleentwiittasi
    17. marrask.

    🤔 Explain Like I’m Five 🤔 Watch explain , Meta's tool that helps anybody make high-quality forecasts over weeks or years and allows custom hand-tuning to factor in specialized prior knowledge. Check it out:

    Kumoa
  4. uudelleentwiittasi
    16. marrask.

    Register for the live event this Thursday with Developer Advocates and

    Kumoa
  5. uudelleentwiittasi
    25. lokak.

    ICYMI: PyTorch 1.10 was released last Thursday. Here are some highlights of the release. Stay tuned for tweet threads in the next couple weeks delving deeper into these cool new features! 1/8

    Näytä tämä ketju
    Kumoa
  6. uudelleentwiittasi
    21. lokak.

    PyTorch 1.10 is here! Highlights include updates for: - CUDA Graphs APIs updates - Several frontend APIs moved to Stable - Automatic fusion in JIT Compiler support for CPU/GPUs - Android NNAPI now in beta Blog: Release:

    Kumoa
  7. 19. lokak.

    New generic AMP API in torch >= 1.10! Check it out by installing the pytorch nightly (conda install pytorch -c pytorch-nightly) Read the entire thread for a quick lowdown on lower precision in

    Kumoa
  8. uudelleentwiittasi
    19. lokak.

    ✨ Low Numerical Precision in PyTorch ✨ Most DL models are single-precision floats by default. Lower numerical precision - while reasonably maintaining accuracy - reduces: a) model size b) memory required c) power consumed Thread about lower precision DL in PyTorch -> 1/11

    Näytä tämä ketju
    Kumoa
  9. 7. lokak.

    We loved hosting members of the awesome community this year, and meeting so many of you on the discussions! Excited for what the next season will bring - tell us what you'd like to see more of!

    Kumoa
  10. uudelleentwiittasi
    6. lokak.

    How can we improve ConvNetsto better recognize 👀 images in ways that 🐰mammals 🐨 *actually* do? Jump into our livestream to hear the answer from and Abraham now:

    Kumoa
  11. uudelleentwiittasi
    1. lokak.

    prompt: trees communicating by Erin Hanson

    4 purple swirling abstract trees in a landscape
    Näytä tämä ketju
    Kumoa
  12. 22. syysk.

    Learn from the edge device experts how to make your models a lean machine. Starting in 15 minutes!

    Kumoa
  13. uudelleentwiittasi
    21. syysk.

    “A drawing of a cat” CLIP + ES + Triangles

    Näytä tämä ketju
    Kumoa
  14. 14. syysk.

    Implementing new DL architectures is the best way to learn about engg. Watching someone do it is probably the next best (and def way easier)! Shout out to 's for his excellent videos using to implement DL research 🙌

    Kumoa
  15. 14. syysk.

    Wrote a small thread on `inference_mode` in , a nifty feature in v1.9 that can speed up your . It works great for tiny ops like Loss functions (I got a 30% speedups) but less so on large ops. Read on for details, and tell me what you'd like to read next!

    Kumoa
  16. uudelleentwiittasi
    13. syysk.
    Kumoa
  17. uudelleentwiittasi
    2. syysk.

    Imaginary numbers help AIs solve the very real problem of adversarial imagery

    Kumoa
  18. uudelleentwiittasi
    1. syysk.

    How do we combine knowledge from multiple labeled and unlabeled datasets to train a great general model? Multi-Task Self-Training (MuST) trains specialized teachers on labeled data, which then label unlabeled data to train a single general model.

    Näytä tämä ketju
    Kumoa
  19. 25. elok.

    Folks, join us in 3 hours from now on where we'll be learning about QUANTUM ML (no, I didn't pull out buzzwords out of a hat) and . We'll be speaking with and Thomas at 12pm ET Watch live:

    Kumoa
  20. uudelleentwiittasi
    24. elok.

    If you're like me, you've written a lot of PyTorch code without ever being entirely sure what's _really_ happening under the hood. Over the last few weeks, I've been dissecting some training runs using 's trace viewer in . Read on to learn what I learned!

    Näytä tämä ketju
    Kumoa

Lataaminen näyttää kestävän hetken.

Twitter saattaa olla ruuhkautunut tai ongelma on muuten hetkellinen. Yritä uudelleen tai käy Twitterin tilasivulla saadaksesi lisätietoja.

    Saatat pitää myös

    ·