nubg

6 ms·

1.
by nubg 26d ago
Thank you, this benchmark to me proves that closed weight model companies are dangerous for our democracy and put kids at risk. They must be outlawed and all models must be made open weights!
2.
by nubg 26d ago
welp, damning indictment. not sure if that means DS is super crap, or qwen is super good
3.
by nubg 27d ago
if by "bot" we mean "not economically interesting", aren't websites within their right to filter our hacker-nerds who probably won't click on ads and buy useless objects?
4.
by nubg 27d ago
As much as I want local and open-weights models to succeed, nothing beats a paid frontier model for now. Anybody who claims otherwise is simply not a daily user of such models. So this "investor" here should invest sime time in ac
5.
by nubg 28d ago
The authors of this article are ChatGPT and Claude.
6.
by nubg 28d ago
> Ctrl + F > "not" > 15 results ai;dr
7.
by nubg 28d ago
that's true, also see "loss leaders" like milk in supermarkets
8.
by nubg 29d ago
why plus margin? the floor is cost
9.
by nubg 29d ago
very happy for you that things turned out this way
10.
by nubg 1mo ago
either the judge made the decision (in which case he has immunity) or he didn't (in which case he isn't the right person to sue) the correct steps are appeals in the merit and disciplinary action against the judge
11.
by nubg 1mo ago
quantization level?
12.
by nubg 1mo ago
> but with open router you can create a no China provider selector why on earth would you trust no data is sent to China? just because you clicked a toggle?
13.
by nubg 1mo ago
good article but why did the author feel like he needed to "polish" it by running it through an LLM? i would rather like to read the actual prompt he put in there, as now i'm not sure which points are his, and which have been
14.
by nubg 1mo ago
I think most AI doomers will tell you that the quality of their code is better than LLM-generated code.
15.
by nubg 1mo ago
bro the problem is not eating burgers every time you see them. if eating a pill once a day will prevent somebody from having these urges people will easily make this a routine. fat people don't like being fat (body positive is the worl
16.
by nubg 1mo ago
ai slop website = ai slop product
17.
by nubg 1mo ago
how do gemma4 or muse perform?
18.
by nubg 1mo ago
what harness did you use?
19.
by nubg 1mo ago
good article, mostly because i have a feeling it was 100% hand written. thanks for not using ai. makes it much more authentic
20.
by nubg 1mo ago
no, because there is another human involved. llms are not human.
21.
by nubg 1mo ago
just to clarify. yes YOU built it. just because you used some tool doesn't mean the idea, prompting, reprompting, babysitting was not your creative input and effort. put differently, if you put a random person infront of whatever model
22.
by nubg 1mo ago
He might be off by a decade, but isn't necessarily wrong.
23.
by nubg 1mo ago
thanks, much better
24.
by nubg 1mo ago
utterly repulsive ai slop
25.
by nubg 1mo ago
> Marraffa says Burt is fully HIPAA compliant and that the AI is not open-source and does not generate or manipulate data. We need to fight the talking point that open source is bad!
26.
by nubg 1mo ago
the common denominator is that most "news" is pure entertainment and outrage porn sells. there is no further substance
27.
by nubg 1mo ago
guys, we should meme the > "ai model leaks from openai and attacks huggingface" to be somehow framed as > "and therefore openai cannot be trusted with ai safety, and we need open weights models". anybody have an id
28.
by nubg 1mo ago
> Understandably ?
29.
by nubg 1mo ago
guys, we should meme the > "ai model leaks from openai and attacks huggingface" to be somehow framed as > "and therefore openai cannot be trusted with ai safety, and we need open weights models". anybody have an id
30.
by nubg 1mo ago
yeah, just pull the troops out of ukraine!

More ›