Tweet activity

Last 28 Days

Your Tweets earned 657.3K impressions over this 28 day period

20.0K40.0K60.0K1020Sep 18Sep 25Oct 2Oct 9
Your Tweets
During this 28 day period, you earned 24.5K impressions per day.
  • Impressions
    Engagements
    Engagement rate
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Sep 26 Is Pearl really claiming to have coined the term 'Bayesian inference' in the 1980s? Because that seems obviously wrong (eg _Bayesian Inference_, Wu 1966 - when it's the title of a whole book on the topic, generally suggests it's not new...)
      8,096
      211
      2.6%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Sep 28 That would be a reasonable analogy only if, among many other things, anyone had ever, even once, seen an impressive Cyc demo (of anything at all).
      7,406
      223
      3.0%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Sep 28 The anti-Hans people were just as earnestly asseverating a few weeks ago that Hans *of course* was only cheating on a few crucial moves, and that's why the analyses could prove little; and now it's "100% agreement"? My conclusion: their arguments are all bullshit.
      7,270
      195
      2.7%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Sep 23 Some playing with textual-inversion: can we generate dropcaps, eg yinit? Well, we're not far from it! SD 1.4 is very weak on text but inversion still captures the yinit style. I couldn't get clear sharp images, but I suspect that was the very crummy jaggy yinit samples I used.
      7,190
      157
      2.2%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Sep 30 Ha, I was right back in 2020! (again) You 𝘤𝘢𝘯 do creative long-form writing with GPT-3-style models using recursive expansion of hierarchical summaries→full sections! "Sampling can show the presence of knowledge, but not the absence..."
      6,961
      196
      2.8%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Sep 17 The best part is, the Tay thing may not have been real and just a boring trivially-disabled 'echo' function like 95% of chatbots & IRC bots have, a scandal manufactured by social media & selective screenshots - while this is both real & interesting as a semantic attack.
      6,695
      100
      1.5%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 1 If there's one thing we should've learned by now, it's that everyone sounds dumb in their texts, which is why Warzel conspicuously fails to mention any comparisons, merely comparing against some imaginary 'actually-impressive' moving-goalpost set of texts.
      6,163
      230
      3.7%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Sep 19 Ah yes, the famously healthy slim non-obese Mexican population/cuisine. Sure is plausible it's... [checks notes] protective rather than harmful effects in obese mothers but mostly only some subgroups, w/no effects in 'foreign-born' of all ethnicities or general fast food effects?
      5,985
      157
      2.6%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Sep 20 This sounds more like a comment on how incredibly crippled and impoverished smartphone UI/designs are. (A *second* calculator - just to display history? - needs to be *an entire bespoke paid app*?!)
      5,973
      162
      2.7%
      View Tweet activity
    • 𝔊𝔴𝔢𝔯𝔫 @gwern Oct 5 Another subtle cognitive distortion I've been noticing: date intervals or durations. It seems like people are increasingly losing their grip on intuitive temporal durations, thinking in a Perpetual Now, and are increasingly surprised when I tuck in an explicit 'X years ago'.
      5,510
      72
      1.3%
      View Tweet activity
Engagements
Showing 28 days with daily frequency
Engagement rate
2.9%
Oct 11
2.2% engagement rate
Link clicks
3.1K
Oct 11
156 link clicks
On average, you earned 109 link clicks per day
Retweets without comments
0
Oct 11
0 Retweets without comments
On average, you earned 0 Retweets without comments per day
Likes
6.5K
Oct 11
243 likes
On average, you earned 234 likes per day
Replies
405
Oct 11
10 replies
On average, you earned 14 replies per day