@mayhewsw

𝘱𝘭𝘦𝘢𝘴𝘦 𝘤𝘰𝘯𝘴𝘪𝘥𝘦𝘳 𝘵𝘩𝘦 𝘦𝘯𝘷𝘪𝘳𝘰𝘯𝘮𝘦𝘯𝘵 𝘣𝘦𝘧𝘰𝘳𝘦 𝘱𝘳𝘪𝘯𝘵𝘪𝘯𝘨 𝘵𝘩𝘦𝘴𝘦 𝘵𝘸𝘦𝘦𝘵𝘴

Joined December 2011
I’m excited to announce an ambitious project: ✨Universal NER ✨. Multilingual NLP is missing a gold standard multilingual NER dataset, and this project aims to fill that gap. This is a community effort: I need your help! See: universalner.org Or read more in the 🧵
12
56
4
291
shawty had them apple bottom jeans (jeans!) boots with the curse (w̴̘̓ì̴͇t̴̟̋h̸̤̿ t̵̤͋h̸̭̅e̴̖̅ c̷͍̉u̶̥̐ř̸̤ś̶͚e̶̠̐!)
32
5,198
92
39,910
444,435
fine, I’ll say it, I don’t think you do save when you bundle.
86
Stephen Mayhew retweeted
Et tu, Piggie?
1
6
111
Stephen Mayhew retweeted
My face hinge fell open and I couldn’t believe my looking blobs when I read that.
221
1,534
68
23,898
853,077
Stephen Mayhew retweeted
Is no one at Apple left-handed
But does it work for lefties
3
1
1
4
1,489
I love Rothko. lil anecdote: I visited MoMA recently, and loved it, of course, but as we worked our way from Van Gogh and Cezanne to later periods, a certain despair and nihilism started to set in, that seemed to coincide with the world wars. 1/2
Mark Rothko No. 10 1950 This isn’t rage bait. I genuinely love Rothko’s paintings—they are deeply spiritual to me—and I was lucky enough to see this one in person at MoMA.
1
2
148
I felt increasingly agitated and disturbed (not necessarily a bad thing, but it wasn't pleasant), and then we walked into a room full of Rothkos and I felt immediately overwhelmed by peace and calm. Such a relief. 2/2
1
22
it couldn’t possibly be a coincidence that the zoomer numerical meme (6-7) factorizes the gen x numerical meme (42)
1
4
173
the millennial numerical meme (>9000) is completely unrelated
41
I’ve been saying this!! jokes don’t have to be original to be funny, they just have to be original *to a particular audience*. kids are a blank slate. I’ve been killing it on the Dad joke circuit for years
Babies are great because simple stuff like this is *literally* the funniest shit they’ve ever seen in their life and you get to hit that bar like a million times in the first few years.
1
564
never heard of this LLM benchmark, but I like it
1
3
234
This is good, and will avoid most of the communication slop. side note: I’m delighted to see the Grice maxim rule here. I can’t remember where I saw it (@nsaphra’s blog maybe?) but the idea that LLMs constantly break Gricean maxims has been on my mind recently.
As agents perform increasingly complex tasks for researchers, it's important to develop norms for communicating with other humans. Here's my lab's policy on AI in writing and communication, focusing on helping us think deeply and communicate precisely. cs.columbia.edu/~johnhew/lab…
1
5
610
yes -- this paragraph. so good. nsaphra.net/post/uncanny/
1
3
86
big deal ChatGPT, I've been asking this question for YEARS
129
while I wait for my new search workflow (ChatGPT) to complete, I'm reminded of Jaime Teevan's ideas on "Slow Search" back in the day. cacm.acm.org/opinion/slow-se…
1
137
Stephen Mayhew retweeted
144 and 441 are not just mirror images of each other the same is true of their square roots, 12 and 21
58
275
37
2,170
86,097
this isn’t real, right? right??
Ngl I appreciate the strategy of hallucinating back at the AI interviewer
1
361