Tweet activity

Feb 1 – Feb 22, 2024

Your posts earned 132.5K impressions over this 22 day period

20.0K10.0K30.0K5Feb 4Feb 11Feb 18
Your posts
During this 22 day period, you earned 6.8K impressions per day.
  • Impressions
    Engagements
    Engagement rate
    • 𝔊𝔴𝔢𝔯𝔫 @gwern 19h (What's really hilarious about all this is that even though we've spent a painful month on this, to a reader, this all looks 𝘦𝘹𝘢𝘤𝘵𝘭𝘺 the same—images already displayed and could be popped up! PDFs & web pages already popped up! Videos already popped up! What's new here?)
      2,164
      73
      3.4%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern 22h (Also, LibreOffice by default always dumps the convert doc into ./... along with all files inside the document. Apparently they *used* to do the sensible thing of inlining them into the HTML, as I expected it to, but they changed, for unclear reasons. At least there's a option.)
      2,674
      12
      0.4%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern 22h A nightmare feature cross-cutting across all systems, while hitting a bunch of random bugs. 180MB HTML files. Linux Chromium PDF view ignores Adobe commands but only in iframes. LibreOffice crashes on one specific spreadsheet. No mobile PDF. Safari event bugs. Nitter error lies.
      1,619
      24
      1.5%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern 22h We also finally fixed the <video> poster problems & made them clickable *and* lazy (believe it or not, they implemented posters without lazy image options); provide HTML versions of doc/docx/csv/xlsx to render in addition to PDF (but disabling PDF on mobile bc lol mobile); etc.
      1,208
      14
      1.2%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern 22h eg is basically ~180MB of animated GIFs. This has 3 benefits: we can transclude the .html, which is usually <1MB, and the images/videos won't immediately download; no text encoding overhead (−18MB); & we can optimize the files eg w/gifsicle (−15MB?).
      1,391
      47
      3.4%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern 23h One side-effect of expanding file-transcludes to documents is that the HTML snapshots can be 182MB. A user should not download 200MB just by scrolling a little out of curiosity! So, we have a script to unpack SingleFile snapshots & make them lazy:
      776
      38
      4.9%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Feb 15 𝘔𝘢𝘥 𝘔𝘢𝘹 3: 𝘉𝘦𝘺𝘰𝘯𝘥 𝘛𝘩𝘶𝘯𝘥𝘦𝘳𝘥𝘰𝘮𝘦 is our 𝘓𝘰𝘩𝘦𝘯𝘨𝘳𝘪𝘯. I am taking no questions on this fact.
      3,918
      87
      2.2%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Feb 13 Another useful script: just asking GPT-4-V if an image would look better with some more whitespace margin: (I often upload copied figures or screenshots or generated graphs where they really need another 50px around the edges to look good.)
      1,584
      106
      6.7%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Feb 13 That is, a LZMA compressor may not be good enough because it is still weak on small variations of boilerplate, while a LLM would be able to see through that and compress it all away, revealing the real redundancy due to poor design/language/ecosystem.
      1,762
      11
      0.6%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Feb 3 Proposal: 'novelty nets'. Generative models generate too many 'same-y' samples, but doing some sort of explicit nearest-neighbor lookup is too slow. So: approximate the function of distance-calculation with another NN. (It's NNs all the way down... 😉)
      2,598
      43
      1.7%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Feb 1 (It might seem obvious that it would have such things, since it seems to have everything, but it didn't occur to me that it would until GPT-4 suggested using SSIM. My install didn't actually have that, but PSNR works well enough.)
      1,784
      18
      1.0%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Feb 1 I finally got around to figuring out a heuristic for when to convert PNG→JPG images. Turns out ImageMagick ships with perceptual similarity/losses (eg PSNR), so you can just write a script to compare everything! Saves ~540MB so far. Also makes 'gallery' pages more feasible.
      4,690
      55
      1.2%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Feb 1 (Instead, we've mostly seen complaints from the Chinese side that they are data-impoverished and it's unfair how Western DL can scrape all this English text data so easily. Apparently 1.4b terminally-online Chinese... don't... write... enough... text? That seems interesting?)
      2,495
      91
      3.6%
      View post activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Feb 1 One consequence of pretending these falsified predictions about Chinese DL never happened is undertheorizing of *why*. eg. *every* China bull case claimed that China bigtech's DL would be supercharged by privacy-invasion/huge private datasets. But.. that didn't happen? at all?
      2,622
      88
      3.4%
      View post activity
You've reached the end of posts for the selected date range. Change date selection to view more.
Engagements
Showing 22 days with daily frequency
Engagement rate
4.6%
Feb 22
3.1% engagement rate
Link clicks
1.6K
Feb 22
198 link clicks
On average, you earned 71 link clicks per day
Retweets without comments
0
Feb 22
0 Retweets without comments
On average, you earned 0 Retweets without comments per day
Likes
1.1K
Feb 22
97 likes
On average, you earned 51 likes per day
Replies
106
Feb 22
22 replies
On average, you earned 5 replies per day