JanuaryDesk 

@yiyuezhuo

A naive student. Focus on uncertainly modeling of policy system and wargame using modern Bayesian statistic technique. Half anarchist, half libertarian.

A sad land
github.com/yiyuezhuo
Vrijeme pridruživanja: kolovoz 2018.
64 fotografije ili videozapisa Fotografije i videozapisi

Tweetovi

Blokirali ste korisnika/cu @yiyuezhuo

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

  1. proslijedio/la je Tweet
    prije 11 sati

    The results are in:

    13 replies 39 proslijeđenih tweetova 201 korisnik označava da mu se sviđa
    Poništi
  2. proslijedio/la je Tweet
    prije 12 sati

    What has drawn you to wargames?

    15 replies 4 proslijeđena tweeta 10 korisnika označava da im se sviđa
    Poništi
  3. 31. sij

    It's very annoying that you can't delete an accepted answer in Stack Exchange. So I must deleted my account as a response to an irrational downvote in Cross Validated.

    Poništi
  4. 29. sij

    Loading time comparsion: resolutio load speedup 1x mp4 1.00 1x jpg 1.01 0.5x mp4 2.00 0.5x jpg 4.09 0.25x mp4 5.80 0.25x jpg 12.48 Well, jpg is not that effective as I expect...

    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  5. 29. sij

    It may be strange to see that the distance between two clips is not that large. It's due to attributions of my video (furnace). Most area have exactly same apperance. Example (origin vs jpg and 0.25x scale): (4/4

    1 reply
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  6. 29. sij

    On the other hand, since T.Resize((128, 171)) is used in the very beginning, reduced scale video should not get a very different result. I used 0.5x and 0.25x here. (3/4

    1 reply
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  7. 29. sij

    Interestingly, TorchVision uses ffmpeg (through PyAV) to decode video dynamically, with those weird force GC code. So we may save time by splitting frames in advance. But lossless format such as png is too large to save, we can use loss format like jpg and tolerate loss. (2/4

    1 reply
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  8. 29. sij

    Tried to speed up video dataset loading. Here's the distance matrix between D_{clip, resolution, split}. Different clips (100 frames apart) show a bigger distance truly. "Resolution" means scaling down video in advance. "Split" means splitting video into frames earlier. (1/4

    1 reply
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  9. 29. sij

    Selecting Hatate Himekaidou with her conservative party is a big mistake. Without Aya Shameimaru's reform party, Allies of New Gensokyo doesn't have enough relationship bonuses to invite others to the faction.

    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  10. 28. sij

    Remove Moriya Shrine! This Touhou Mod would be more fun if commanders in surrendered countries will join the conqueror. Missing Kochiya Sanae and Mrs. Elder Toad is very annoying.

    1 reply
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  11. proslijedio/la je Tweet
    24. sij

    Absolute Chad

    11 replies 210 proslijeđenih tweetova 975 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  12. proslijedio/la je Tweet
    19. sij

    Here's an illustration of this... ...wait, where did that mule go?

    7 replies 23 proslijeđena tweeta 139 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  13. proslijedio/la je Tweet
    19. sij

    Ok, I got a pretty good port of these Maxwell equation figures into Python. Heavy use of JAX and . I'll tidy up the code post it here later.

    7 replies 47 proslijeđenih tweetova 312 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  14. 19. sij

    I personally noticed this project when the 2019.6.4 GFW blocking wave. But since Trojan doesn't support CDN, I turned to V2ray to save my blocked VPS.

    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  15. 18. sij

    Wow, Trojan appeared on HN with decent number of comments. The project became very popular due to a YouTube video or somewhat reason I guess.  I wonder if new users force GreatFire to leave her Telegram group again...

    1 reply
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  16. proslijedio/la je Tweet
    18. sij

    [1/2] The red dot accelerates at a constant rate and direction. The black dots can accelerate faster. How can they match the red’s position *and* velocity as soon as possible? The calculus of variations shows the optimal trajectories obey a simple rule:

    11 replies 65 proslijeđenih tweetova 307 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  17. 15. sij

    OpenCV is unable to read video from memory (buffer)🤔. So the server must save it to disk and read it back after receiving a 10s (6mb) video clip? It's so dumb imo.

    1 reply
    Poništi
  18. 14. sij

    "99.8% a teddy bear" That's why we need BNN, I guess...

    Odgovor korisnicima @martinRenou @jeremyphoward @ProjectJupyter
    99.8% a teddy bear ... 😂
    Poništi
  19. proslijedio/la je Tweet
    11. sij

    Plot of a p-value, calculated for a simple t-test, for sample sizes of 10 to 2010, when the null hypothesis is true. Because p-values are uniformly distributed, the p-value just wanders randomly between 0 and 1. This is why repeatedly analyzing data as it comes in is not good!

    31 reply 765 proslijeđenih tweetova 2.267 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  20. 13. sij

    Looks pretty. Would try to use ipywidget to implement my next API illustration.

    Amazing! This real working deep learning image classifier is running, for free, on , is written entirely in notebooks with ipywidgets, & deployed with Voila. 13 lines of code :) Try it here:
    Prikaži ovu nit
    Poništi

@yiyuezhuo 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:

    ·