Chris Olah

@ch402

I want to understand things clearly and explain them well. Clarity Team,

Vrijeme pridruživanja: lipanj 2010.

Tweetovi

Blokirali ste korisnika/cu @ch402

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @ch402

  1. proslijedio/la je Tweet
    28. sij

    1/2 Last year I wrote an article with . Distill represents a promising new paradigm in academic publishing. I wrote a blog post about my Distill experience:

    Prikaži ovu nit
    Poništi
  2. proslijedio/la je Tweet
    21. sij

    The market price for trained listening seems to be ~$200/hr. (therapists) Seems worth cultivating listening skills, as most people value listening highly and will never seek out therapists.

    Prikaži ovu nit
    Poništi
  3. proslijedio/la je Tweet
    19. sij

    The aspect of humility worth cultivating is avoiding making others feel small.

    Prikaži ovu nit
    Poništi
  4. proslijedio/la je Tweet
    20. sij

    Remco ('s in-house talent expert) explains why headline-grabbing stats on the "exodus" of Chinese scientists may not be what they seem at first glance:

    Poništi
  5. proslijedio/la je Tweet
    15. sij

    I sometimes hear that looks a lot like a blog rather than a journal—fair enough—so we added author names, as well as badges linking to peer review results. Do you think this helps? Or is it pure bike-shedding? All opinions welcome! See:

    Distill.pub's landing page, now listing author names under paper titles.
    Poništi
  6. proslijedio/la je Tweet
    15. sij

    Visualizing the Impact of Feature Attribution Baselines -- A new Distill article by Pascal Sturmfels, Scott Lundberg, and Su-In Lee.

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

    In 1944, the CIA distributed a Simple Sabotage Field Manual with instructions for discretely making organizations less effective. Useful lens for noticing where organizations can become unintentionally self-sabotaging. HT

    Ovo je potencijalno osjetljiv multimedijski sadržaj. Saznajte više
    Prikaži ovu nit
    Poništi
  8. proslijedio/la je Tweet
    24. pro 2019.

    me: i wonder what would happen if i trained the neural net gpt-2 on christmas carols? me: ... me: oh NO

    Rudolph the Red Nosed Reindeer, with its red belly
The All-gracious King of all the earth,
Had a baby at Christmas time,
On Christmas Day,
A true and holy Deity,
Went down to earth,
With human flesh for sacrifice.

For sinful men such a Deity doth appear,
And wink and nod in reply,
As he winked and nod in reply,
As he winked and nod in reply

The wretched world is run by ox and ass
The wretched world is run by ox and
    Prikaži ovu nit
    Poništi
  9. proslijedio/la je Tweet
    24. pro 2019.

    A little holiday present from me: rendering 512 feature visualisations in 8 seconds on a TPU!

    Prikaži ovu nit
    Poništi
  10. 20. pro 2019.
    Prikaži ovu nit
    Poništi
  11. 20. pro 2019.

    I guess I'm frustrated by what can sometimes feel like a kind of pseudo-rigor where we measure statistics I'm not sure are meaningful and prove theorems I'm not sure relate to the problem and feel like that makes us more genuine.

    Prikaži ovu nit
    Poništi
  12. 20. pro 2019.

    To be clear, the right summary statistic can be beautifully clarifying. But you need to use them very carefully, because once you adopt a summary statistic it blinds you to the world beneath it's abstraction. Classic example is Anscombe's quartet:

    Prikaži ovu nit
    Poništi
  13. 20. pro 2019.

    *Grumpy Chris:* One of my biggest research gripes is people boiling the world down into a couple summary statistic numbers and feeling like that somehow made things more rigorous.

    Prikaži ovu nit
    Poništi
  14. 20. pro 2019.

    A lot of important historical science (eg. discovery of cells) is qualitative observational science. But now "qualitative" seems to almost be a dirty word. How did this change?

    Prikaži ovu nit
    Poništi
  15. proslijedio/la je Tweet
    20. pro 2019.

    For the last 10 months, I have been working on Program Synthesis. Our (me and ) first paper was just accepted () and I think it's my favorite research I have done?: (1/N)

    Prikaži ovu nit
    Poništi
  16. 17. pro 2019.

    If deep learning is summoning demons, interpretability is demon biology. 🔬👹

    Poništi
  17. proslijedio/la je Tweet
    16. pro 2019.
    Poništi
  18. proslijedio/la je Tweet
    7. pro 2019.

    For anyone with opinions about comms, Josh here is starting a listening conversation about it! (Also, IME, Twitter doesn’t have to suck for this if you’re willing to use multiple threaded tweets for longer thoughts)

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

    AI research Twitter: what concrete norms/checklists do you think research orgs should use to communicate about AI science to mixed audiences that include some AI researchers but also lots of other people (many nontechnical)?

    Prikaži ovu nit
    Poništi
  20. 3. pro 2019.

    Secret hideaways are an under-appreciated essential part of library design. While I'm being nostalgic, I also miss UofT's annual booksales. Sooo many books.

    Prikaži ovu nit
    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:

    ·