k__

27 ms·

301.
by k__ 1y ago
1) Fair. I did some experiments with recommendation systems 15 years ago and we basically stopped using dimensions generated by the system, because nobody could make anything of them. The human-made dimensions were much easier to create use
302.
by k__ 1y ago
I mean LLMs reduce the "vector space" that describes reality into a vector space with fewer dimensions (e.g. 300 in the article I was replying to.)
303.
by k__ 1y ago
Awesome, thanks! If I understand this correctly, there are three major problems with LLMs right now. 1. LLMs reduce a very high-dimensional vector space into a very low-dimensional vector space. Since we don't know what the dimensions
304.
by k__ 1y ago
More like the Mythical Machine-Minute.
305.
by k__ 1y ago
Photorealism gives the impression of capturing the truth, but any reasonably skilled photographer knows this isn't the case. Text generated by LLMs has similar properties.
306.
by k__ 1y ago
Pretty cool! WebAssembly planned?
307.
by k__ 1y ago
Maybe Tailwind has better dead code removal.
308.
by k__ 1y ago
"it had no sense of symmetry, beauty, anything. It was just a mess." Reminds me of the square packing problem, with the absurdly looking solution for packing the 17 squares. It also reminds me of edge cases in software engineeri
309.
by k__ 1y ago
Not a contradiction. For some people this is inspiring.
310.
by k__ 1y ago
Windows 10 support is ending. What Linux would you recommend for a notebook in 2025?
311.
by k__ 1y ago
It came unwound when we allowed enterprises to destroy the wages of the lower end and stopped building to respond to the housing demand. That can be easily undone, if the powers to be allowed it.
312.
by k__ 1y ago
Isn't this a question of how many people you select?
313.
by k__ 1y ago
I knew it as demarchy, which the author also avoided.
314.
by k__ 1y ago
Haha, mine too. It would probably make sense to start with a new new "house" or something. Might even make sense to have some quotas (at least 50% women etc.), so the whole things doesn't have to get to Chinese government siz
315.
by k__ 1y ago
"As I got older, I've leaned more and more into meritocracy" Sad thing is, that it's impossible.
316.
by k__ 1y ago
Half-OT: Is there any info out there how the major models differ?
317.
by k__ 1y ago
Neurotypicals, am I right? :')
318.
by k__ 1y ago
How much does it cost right now? Are there any alternatives coming out soon or generics?
319.
by k__ 1y ago
But people were also hating about media piracy, video games, and the internet in general. The dotcom bubble popped, but the general consensus didn't become negative.
320.
by k__ 1y ago
I hope DeepSeek R2, but I fear Llama 4.
321.
by k__ 1y ago
"respecting web crawling opt-outs during data acquisition produces virtually no performance degradation" Great to read that!
322.
by k__ 1y ago
I hope this will end better than in Outer Wilds.
323.
by k__ 1y ago
Especially since AI grammar tools automated B for years now.
324.
by k__ 1y ago
Yeah, I was thinking the same. They got lost in the details.
325.
by k__ 1y ago
Fair. I just thought, maybe the decline is just temporary.
326.
by k__ 1y ago
I had the impression, Grok wasn't on Elon's side when it answered my questions or explained tweets.
327.
by k__ 1y ago
Could it be a hole that was created by improvements in women's rights? My sister got three kids, but she got them after 30. My mom and my grandma got their first child with 20.
328.
by k__ 1y ago
Hot take: HATEOAS only works when humans are navigating.
329.
by k__ 1y ago
I have to admit, that using slow models is unbearable when I used fast one before. I don't know if the quality and speed are linearly related, though.
330.
by k__ 1y ago
So, how should we as a society handle this? Ensure the models are open source, so everyone can use them, as everyones data is in there? Close those companies and force them to delete the models, as they used copyright material?

More ›