Tweet activity

October 2020

Your Tweets earned 865.2K impressions over this 31 day period

50.0K100.0K10Oct 4Oct 11Oct 18Oct 25
Your Tweets
During this 31 day period, you earned 27.9K impressions per day.
  • Impressions
    Engagements
    Engagement rate
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 17 Many people write much more than they think they write, but they often write the same thing each time—because they don't save it for future use. If you've explained or ranted about something at length 3 times, it's time to write it up!
      56,310
      668
      1.2%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 23 Tip for programmers, researchers, and hobbyists: don't thoughtlessly assume you have to use the cloud for everything. Dedicated servers are underappreciated— easy, often far cheaper, flexible, and free you from billing anxiety; the more you use them, the more useful they are.
      44,400
      910
      2.0%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 18 "You nag—like I give a fig for gingerly treating horses according to the race rules!", said Tom puckishly.
      29,926
      95
      0.3%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 6 Isochronic history: if we taught history as it happened (1 year at a time), we'd have a different perspective on progress—nothing happened, then everything started happening in the 1800s. Yet even then... "The light bulb is next week? But they had all the pieces 𝘭𝘢𝘴𝘵 week!"
      29,202
      465
      1.6%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 10 I keep wanting to fiddle with the metadata... "The web developer of the future will be a single designer with a smol cat, a medium-sized laptop, and a large neural network: the neural network optimizes for traffic, and the laptop is sat on by the cat to distract the designer."
      23,984
      547
      2.3%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 2 Pleased at latest dark mode fixes. How to do images? Inverting photos/color looks crazy. But not-inverting graphs is blinding white, defeating the point. My solution: ImageMagick detects monochrome images, I manually tag misses, dark mode inverts only those; +original on hover.
      22,080
      1,021
      4.6%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 25 Now, why did several of us fail in May to get compare_gan BigGAN interpolations working at all, when training samples were? Because /sighs heavily ...the Batch Norm statistics weren't being done right at sample time, and no one noticed. 𝘋𝘦𝘢𝘵𝘩 𝘵𝘰 𝘉𝘢𝘵𝘤𝘩 𝘕𝘰𝘳𝘮!
      15,853
      111
      0.7%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 4 The GAN transmutation/hybridizing seems to have a lot of potential. People have tried hard to train various kinds of CycleGAN and other tricks to do anime ↔ real, and it's never really worked too well. Blend some StyleGAN models, though...
      15,386
      100
      0.6%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 25 Problems caused by (hidden, global, mutable, unrecorded) state of normalizations are up there with BPEs for frustratingly subtle fatal yet unnecessary problems in DL engineering. If only they failed outright! But as Karpathy threatens researchers, "Neural nets 𝘸𝘢𝘯𝘵 to work."
      15,304
      112
      0.7%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 3 Worth mentioning l4rz's continued work on S2 nudes (although it doesn't use Tensorfork resources). His latest trick: as images get bigger, 'high-resolution layers' are responsible for increasingly big details in eyes/faces. So spend the parameters *there*:
      15,144
      286
      1.9%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 6 (People don't love the human touch. They do love to read in big lessons to cross-country variance and Poisson clusters which are noise. Look at videoconferencing. Obviously, we could've been doing all this years ago! Why not? Inertia. From the view of history, a minor delay.)
      14,904
      207
      1.4%
      View Tweet activity
Engagements
Showing 31 days with daily frequency
Engagement rate
2.2%
Oct 31
1.4% engagement rate
Link clicks
4.4K
Oct 31
63 link clicks
On average, you earned 142 link clicks per day
Retweets without comments
211
Oct 31
2 Retweets without comments
On average, you earned 7 Retweets without comments per day
Likes
2.8K
Oct 31
16 likes
On average, you earned 91 likes per day
Replies
213
Oct 31
0 replies
On average, you earned 7 replies per day