yk

7 ms·

61.
by yk 1y ago
That's entirely implementation dependent. For existing libraries I would expect something like def get(self, query): if isinstance(query, template): self.get_template(query) else: self.get_
62.
by yk 1y ago
Last time bio signatures where found (I believe on Venus), there was a flurry of papers on arXiv a week later, which all explained the signature by abiotic processes. Of course, if you have more than one explanation, then you really don&#x2
63.
by yk 2y ago
Tried Flux.dev with the same prompts [0] and it seems actually to be a GPT problem. Could be that in GPT the text encoder understands the prompt better and just generates the implied IP, or could be that a diffusion model is just inherently
64.
by yk 2y ago
Perhaps that would be a use for the $TRUMP coin, whoever owns most gets to favorite corporation.
65.
by yk 2y ago
Known quantity, especially when I'm travelling I go there because I know that I get kinda consistent quality without having to think.
66.
by yk 2y ago
The most common category of organizing a book shelf (other than none) is by size? Why?
67.
by yk 2y ago
I think I've seen 800 GB/s memory bandwidth, so a q4 quant of a 400 B model should be 4 t/s if memory bound.
68.
by yk 2y ago
Yes, but it would've been nice to call them D1-something, instead of constantly having to switch back and forth between Deepseek R1 (here I mean the 604B model) as distinguished from Deepseek R1 (the reasoning model and it's disti
69.
by yk 2y ago
Every politician is always exclaiming "We are the greatest" and that becomes a problem when politicians actually start believing their own bullshit.
70.
by yk 2y ago
[flagged]
71.
by yk 2y ago
Last week we have seen ai CEOs calling for sanctions because the Chinese publish open weights models and research, while the American companies don't. Then this naked politicization of science (which is also done pretty incompetently).
72.
by yk 2y ago
You don't need to know that the keywords are actually english words. So to start the first program on a floppy on a C64 you would type load "*",8,1 and back then I didn't understand what load means any more than I unders
73.
by yk 2y ago
That's pretty much the same mistake as in VW recent "We know where you parked" hack. [0] So while I don't really want to say anything nice about VW, the mistake is no something that only happens to side projects. [0] ht
74.
by yk 2y ago
I felt like that from 2010 or thereabouts to two years ago. There wasn't really a use case for having a very fast machine. Now with llms and sd, I think there is a use case that happily absorbs any compute I can buy, just like first pe
75.
by yk 2y ago
Second the other comment, as much vram as possible. A 3060 has 12 GB at a reasonable price point. (And is not too limiting.)
76.
by yk 2y ago
Thanks,sometimes I wonder when humanity lost to he ability to serve plain text over http.
77.
by yk 2y ago
For runtime, I use ComfyUi [0] which is node based and therefore a bit hard to learn. But you can just look at the examples on their github. Foocus [1] also seems to be popular and a bit more conventional perhaps, though I didn't try i
78.
by yk 2y ago
> Other pebble beds: The pebble bed technology and design has previously been used in prototype reactors in China and Germany, but not a larger-scale plant like Shidaowan. That's wrong, Hamm-Uentrop was a full scale commercial react
79.
by yk 2y ago
To a certain extent I think we get a better understanding what llms can do, and my estimation for the next ten years is more like best UI ever rather than llms will replace humanity. Now best UI ever is something that can certainly deliver
80.
by yk 2y ago
real data - background model = bump This is all just counting statistics, it actually is that simple. (The resonance equals particle is quite a bit more complex, but for a basic treatment the bump is a particle could probably just be unders
81.
by yk 2y ago
I test llms actually similar. For example there is a well known logic puzzle were a farmer tries to cross a river with a cabbage a goat and a wolf. Llms can solve that since at least GPT-2, however if we replace the wolf with a cow, gpt-o d
82.
by yk 2y ago
So 1kEUR/month for a 6kEUR GPU. Pretty sure there are a lot of drug dealers who wish they had gone into cloud training instead.
83.
by yk 2y ago
The third part is a complete change in tone, but it is actually necessary for the entire book because without it the ending would be implied to be extraordinary bleak.
84.
by yk 2y ago
Not sure if you want to share the capitalist system with an entity that outcompetes you by definition. Chimps don't seem to do too well under capitalism.
85.
by yk 2y ago
I read a long time ago related articles: Around 2000 (+- 5) someone claimed that with the internet you only need 1000 (or I believe 3000) dedicated fans to make something like a blog. And around 2010 (+- 5) there was a Wired, I believe, art
86.
by yk 2y ago
> a significant amount of memory goes into the KV cache Is there a good paper (or talk) how inference looks at scale? (Kinda like ELI-using-single-gpus)
87.
by yk 2y ago
Well MS is selling Linux by now so stranger things may happen, but talk is somewhat cheaper than implementing good practices.
88.
by yk 2y ago
Capitalism is when congress rides in on a white horse when the communists win in the market place.
89.
by yk 3y ago
Did anybody manage to get the entire prompt out of gemini, or what are you basing your claim on?
90.
by yk 3y ago
Do you have abit of time? It should be quick...

More ›