Andreas Kirsch

@BlackHC

Trained code monkey. DPhil student at with . . Former RE , SWE . Fellow .

Oxford, England
blackhc.net
Vrijeme pridruživanja: kolovoz 2009.
69 fotografija ili videozapisa Fotografije i videozapisi

Tweetovi

Blokirali ste korisnika/cu @BlackHC

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

  1. Prikvačeni tweet
    25. lip 2019.

    Very happy & proud to share some research in Deep Bayesian Active Learning from , and me at 🎉🎉🎉🤗🤗🤗

    1 reply 75 proslijeđenih tweetova 257 korisnika označava da im se sviđa
    Poništi
  2. prije 24 minute

    ELBO is just unnormalized Bayes' theorem in information theory 🤓🤗

    Poništi
  3. 30. sij

    That time in the evening when you hate yourself for not installing TexLive earlier on your home workstation ⏳🤦

    3 korisnika označavaju da im se sviđa
    Poništi
  4. proslijedio/la je Tweet
    22. sij

    《バーコーダー》バーコードリーダーのスキャン信号をレジではなく、スピーカーに直接接続することで音を鳴らす。 いま渋谷で巨大レシート版、演奏できます┃┃┃┃_ρﾞ

    256 replies 27.076 proslijeđenih tweetova 81.782 korisnika označavaju da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  5. proslijedio/la je Tweet
    22. sij

    if you are wondering what dual numbers are useful for, i wrote up a couple things on em. here's the first: For hyperbolic numbers (also called split complex numbers, among other things), check this out:

    4 proslijeđena tweeta 13 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  6. 21. sij

    Things to love in the evening: and having swapped argument order (file, obj) vs (obj, file) 😭

    1 proslijeđeni tweet 9 korisnika označava da im se sviđa
    Poništi
  7. 20. sij

    When the quick take fails 🤷 also ouch at the subconscious bias 😬

    Tweet je nedostupan.
    1 korisnik označava da mu se sviđa
    Poništi
  8. proslijedio/la je Tweet
    25. pro 2019.

    Shameless end of year plug for: It's a ~15 page pdf with prompts you fill out to help you reflect on the past year, and plan for the next year. I did it last year and found it to be very helpful, and I will definitely be doing it again for 2019/20.

    2 proslijeđena tweeta 2 korisnika označavaju da im se sviđa
    Poništi
  9. proslijedio/la je Tweet
    18. sij

    There are errors in this post. (1) The likelihood will collapse onto the "good function" as we increase the data size if the data are from the distribution we want to fit, as increasingly fewer bad functions will be consistent with our observations.

    New blog post with -- "A Sober Look at Bayesian Neural Networks": Without a good prior, Bayesian uncertainties are meaningless. We argue that BNN priors are likely quite poor, and concretely characterize one specific failure mode.
    Prikaži ovu nit
    34 proslijeđena tweeta 139 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  10. proslijedio/la je Tweet
    18. sij

    Edo period cat meme

    18 replies 619 proslijeđenih tweetova 1.563 korisnika označavaju da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  11. proslijedio/la je Tweet
    15. sij

    How has helped us advance a new paradigm of agile software development: both experimentation and production ready code within a single workflow! 🌎

    61 proslijeđeni tweet 190 korisnika označava da im se sviđa
    Poništi
  12. 14. sij

    Reading Twitter and paper updates this morning and just wanna crawl into a hole. Too much imposter syndrome, too much overwhelm 🙈

    5 korisnika označava da im se sviđa
    Poništi
  13. proslijedio/la je Tweet
    14. sij

    The maddening saga of how an Alzheimer’s ‘cabal’ thwarted progress toward a cure for decades via

    14 replies 53 proslijeđena tweeta 114 korisnika označava da im se sviđa
    Poništi
  14. proslijedio/la je Tweet
    14. sij

    I wrote this self-contained, 150 line script that trains a ResNet-18 to ~94% accuracy on CIFAR-10. Useful for obtaining a strong baseline with minimal tricks. Uses / Torchvision: 👉

    1 reply 34 proslijeđena tweeta 226 korisnika označava da im se sviđa
    Poništi
  15. proslijedio/la je Tweet
    11. sij

    My machine learning education has progressed to the point where I lose sleep tossing around a “brilliant” idea, only to find the next day that it doesn’t actually work. This is great! I talked about this a few years ago:

    50 replies 254 proslijeđena tweeta 1.907 korisnika označava da im se sviđa
    Poništi
  16. 12. sij

    Kinda like BatchBALD 🤗

    In prior work (Doe et al, 2019) has considered the problem of parrot walking, however, the proposed method had severe limitations. The approach presented in this paper is novel and versatile. To our knowledge it is the first work considering multiple parrots simultaneously.
    Poništi
  17. 12. sij

    A good blog post overall. Especially that it is important to communicate with team members before rewriting someone's code 😇

    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  18. 12. sij

    Choose your interfaces wisely but decouple them from your implementations. This is a good pattern that gives you slack in both dimensions to tweak 🎉

    1 reply 1 korisnik označava da mu se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  19. 12. sij

    Deduplication of the implemention makes sense though! It allows for better testing and reduces the chance of bugs 💪

    1 reply
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  20. 12. sij

    My proposed take is not to change the interface. Keep it as flexible as needed, that is few assumptions, to allow for easy specialization when necessary. Keep your code hackable to be agile 🙏

    1 reply
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  21. 12. sij

    The example replaces repetitive code with a neat (👈literally) deduplicated abstraction. The problem with that is that it also tied down the interface to adhering to the chosen abstraction.

    1 reply
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi

@BlackHC 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:

    ·