Leonid Boytsov

@srchvrs

Research engineer previously ph-D student : IR, NLP, and a bit of speech recognition.

Pittsburgh, PA
Vrijeme pridruživanja: studeni 2009.

Tweetovi

Blokirali ste korisnika/cu @srchvrs

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

  1. Prikvačeni tweet
    13. srp 2014.

    Non-Metric Space Library: A cross platform similarity search library & toolkit to evaluate similarity search methods.

    Poništi
  2. prije 8 sati

    Do people freaking about about rare diseases know how many people die from flu & pneumonia every year? About 50K in US alone: h/t

    Poništi
  3. proslijedio/la je Tweet
    3. velj

    We are still looking for more interns for 2020!

    Poništi
  4. prije 14 sati

    Installing CUDA 10 and CUDA toolkit on Lubuntu (Ubuntu 18) is nearly an impossible mission. I succeeded at about 3rd attempt (attempts were 3-4 months spaced in time). Lubuntu comes with CUDA 9 drivers and toolkit, it's nearly impossible to unsit them.

    Poništi
  5. proslijedio/la je Tweet
    prije 21 sat

    Thus, I also strongly disagree with the PAMI-TC decision on no new results on rebuttal. The reviews provide new perspectives, which sometimes cannot be addressed without new experiments. I agree that this should be discouraged, but it should not be prohibited.

    Prikaži ovu nit
    Poništi
  6. proslijedio/la je Tweet
    3. velj

    Congratulations to (and to advisers and ) on successfully defending his thesis: “Transductive Semantic Parsing” Dr. Zhang will be joining

    Poništi
  7. 2. velj
    Poništi
  8. 1. velj

    Mitchell's post focuses on training sparse/compressed or in any other sense succinct neural network without training an over parametrized models first. I really like the table in the end, which explains why certain compression techniques are viable!

    Poništi
  9. 1. velj

    On problems and solutions of using Transformers in speech recognition.

    Poništi
  10. 1. velj

    A question to physicists and chemists, does the solar panel conversion rate is comparable to that in plants in the sense that do these conversion rate numbers use the same baseline? For example, plants seem to access only 45% of available wavelengths

    Poništi
  11. 31. sij

    "You'll want to jump to that idea. If your idea is quick to try and the potential upside is large, then go ahead and do it. But more commonly, your initial results on the new idea will be disappointing, and it’ll take a more sustained effort to yield significant results."

    Prikaži ovu nit
    Poništi
  12. 31. sij

    👇"In my experience, switching problems too frequently (and giving up on promising ideas) is a more common failure mode than not switching enough. Often, while you’re engaged in the long slog towards getting your current idea to work, another promising idea will come along ..."

    Prikaži ovu nit
    Poništi
  13. proslijedio/la je Tweet
    31. sij

    Reduce Bias in Text? Find it out in our podcast with 📢

    Poništi
  14. proslijedio/la je Tweet
    31. sij

    Finally it's out! A blog post giving an overview and some background on our work on text generation with LaserTagger.

    Poništi
  15. proslijedio/la je Tweet
    31. sij

    Supervised multimodal bitransformers now available in the awesome HuggingFace Transformers library!

    Poništi
  16. 31. sij

    There're quite a few reasons why SIMD's not used more often: 1) it's hard & u need C++ (or similar low-level lang) 2) Concurrency's difficult but arcane CPU/GPU instructions're horror movie 3) It's also far from being stand. We need smth like SIMDanywhere

    Poništi
  17. proslijedio/la je Tweet
    30. sij

    Non-compete agreements are unethical and enable abuse. Read this thread, and the FTC is inviting comments from the public here:

    Prikaži ovu nit
    Poništi
  18. 30. sij

    A summary of recent summarization techniques

    Poništi
  19. 30. sij

    Also trying to reproduce basically anything. h/t

    Poništi
  20. proslijedio/la je Tweet
    29. sij

    Microsoft shifted at least $39 billion in U.S. profits to Puerto Rico, where the company’s tax rate is nearly 0%

    Poništi
  21. proslijedio/la je Tweet
    30. sij

    "Bayesian approach to incorporating different types of biomedical knowledge bases into information retrieval systems for clinical decision support in precision medicine" w published by the J Biomed Inform

    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:

    ·