brokensegue

3 ms·

1.
by brokensegue 8d ago
you can ask structured output for probabilities...not that they necessarily mean anything.
2.
by brokensegue 10d ago
Multiple places
3.
by brokensegue 10d ago
Aren't the new models thinking in neuralese?
4.
by brokensegue 16d ago
surely you'd have to also control for the kind of driving rideshares do (e.g. local v. highway) to have any real sense. maybe we could look for insurance rates?
5.
by brokensegue 19d ago
Isn't storj a crypto thing? I don't think they'd want to be involved. I get the sense they are idealists
6.
by brokensegue 21d ago
Also it includes insurance and such that doesn't appear in TC Disclaimer: I work for the wmf
7.
by brokensegue 23d ago
If you're doing a job you aren't unemployed. Underemployment is measured separately
8.
by brokensegue 23d ago
IQ tests? Very good. But most are in the dataset so it's not very meaningful
9.
by brokensegue 27d ago
Do you want your phone to allocate 4GB to read a mailing list?
10.
Ask Me Twice – A longitudinal archive of AI chatbot responses(wayback-api.archive.org)
2 points by brokensegue 1mo ago|0 comments
11.
by brokensegue 1mo ago
yeah that was my understanding. for some reason corn syrup is vilified. the real problem there (as I understand it) is corn subsidies.
12.
by brokensegue 1mo ago
Is there evidence supporting the idea that sugar is more satiating? My understanding was that they were equally unhealthy
13.
by brokensegue 2mo ago
Asianometry is underrated. So much in depth content
14.
by brokensegue 2mo ago
it's cool it lasted so long
15.
by brokensegue 2mo ago
Per hour?
16.
by brokensegue 2mo ago
But surely people's willingness to listen to an AI is contingent on their past experience with this model
17.
by brokensegue 2mo ago
Why didn't they use a model people actually use?
18.
Open Sauce 2026 – Exhibits(opensauce.com)
3 points by brokensegue 2mo ago|0 comments
19.
by brokensegue 3mo ago
Fable told me > Verdict: I checked every step and found no error. The argument appears to be a correct proof of the Cycle Double Cover conjecture, modulo two standard cited results (the reduction to loopless cubic graphs and the Jaeger–K
20.
by brokensegue 3mo ago
personally, i've never had good luck with MT's quality.
21.
by brokensegue 3mo ago
Is this at all related to kicanvas.org
22.
by brokensegue 3mo ago
Brainfuck is probably the most famous esoteric programming language. It's mentioned (and used) in published papers e.g. https://link.springer.com/article/10.1007/s10994-019-05821-2 . Probably not a good compar
23.
by brokensegue 3mo ago
Websites can request to be removed. The data is still kept though
24.
by brokensegue 3mo ago
it's really not much to manage assuming you are already doing backups. you ~ pay the energy cost either way.
25.
by brokensegue 3mo ago
You mean neither?
26.
by brokensegue 3mo ago
there are models between 2-grams and 600m param models that would be good options. i don't expect a 2-gram to do very well here. also i'm not sure why this model isn't a fine choice if it solves their problem
27.
by brokensegue 3mo ago
you should link your data to wikidata which will get you free connection back to crunchbase and other sources e.g. https://www.wikidata.org/wiki/Q97041185 You could even back some of the data from there
28.
by brokensegue 3mo ago
EU tried to ban tracking cookies. Anyone could've seen the result coming (every site now disrupts your reading). The defeatist attitude would've been right there. I don't see how games won't just charge you a $0.01 &quot
29.
by brokensegue 3mo ago
anyone have a free photo for wikipedia?
30.
by brokensegue 4mo ago
Using mtp takes away some of the magic for me

More ›