gpjanik

7 ms·

1.
by gpjanik 24d ago
"people had 1 year to protest this" this is terminal stage subjugation to bureaucrats
2.
by gpjanik 24d ago
There's a million more plausible things that can have a time-release backdoor, including a PDF, a JPEG, and about anything you don't closely inspect upon running. That's why you have an interpretable software layer running on
3.
by gpjanik 24d ago
And more cores means what exactly in terms of transistors count?
4.
by gpjanik 24d ago
"When Moore’s Law slowed in the mid-2000s" it did not, in fact, slow down in the mid 2000s, or at all. https://ourworldindata.org/data-insights/moores-law-has-accu...
5.
by gpjanik 2mo ago
TLDR: the experiment asks for in-distribution responses and gets those. The right answer here is to ask a LLM to create a scene similar in quality to those, but completely out of distribution. I asked GPT 5.6 Sol to give me a pelican playin
6.
by gpjanik 2mo ago
I am not expecting it, it's the Qwen team is claimingthey can do much harder tasks than this, like rendering a consistent page of a maths paper, or creating true to fact explainers. They can't.
7.
by gpjanik 2mo ago
The real performance is nowhere close to what is presented in the marketing materials, which is pretty annoying. Especially text rendering and accuracy. Try asking it for a plot of Polish GDP growth over the past 20 years. It's slop.
8.
by gpjanik 3mo ago
A lot (all?) VCs charge some form of fees (typically capped at 20% of the entire fund, split in various percentages through 4 years investing, 4 divesting period). These fees often are only paid out only based on the actively deployed capit
9.
by gpjanik 8mo ago
Other EVs are incomparably better though.
10.
by gpjanik 8mo ago
You can, Deel etc. make this pretty easy.
11.
by gpjanik 10mo ago
Thanks god we're getting rid of nuclear energy btw.
12.
by gpjanik 10mo ago
There have been numerous cases of sanctioning and wealth confisactions (Afrghanistan, Venezuela, Iraq, Iran, Libya), and "_now_ carries confiscation risk" is just factually incorrect. It has always carried such risk, this risk has
13.
by gpjanik 10mo ago
"Pozsar’s argument: the moment Western nations froze Russian foreign exchange reserves, the assumed risk-free nature of these dollar holdings changed fundamentally. What had been viewed as having negligible credit risk suddenly carried
14.
by gpjanik 10mo ago
If you're adding some computational/problem breakdown/heuristic steps on top/instead of mathematical concepts, then you're doing the opposite of what the author proposes. Scientific conensus in math is Occam's
15.
by gpjanik 10mo ago
"Think in math, write in code" is the possibly worst programming paradigm for most tasks. Math notations, conventions and concepts usually operate under the principles of minimum description lenght. Good programming actively fight
16.
by gpjanik 1y ago
Agreed, but that's not in the C dimension of a first-layer embedding of a single token though, it's across the whole model and that's what I said in the comment above.
17.
by gpjanik 1y ago
It's... really not what I meant. This requirement does not have to be relaxed, it doesn't exist at all. Semantic similarity in embedding space is a convenient accident, not a design constraint. The model's real "understa
18.
by gpjanik 1y ago
Language models don't "pack concepts" into the C dimension of one layer (I guess that's where the 12k number came from), neither do they have to be orthogonal to be viewed as distinct or separate. LLMs generally aren&#x2
19.
by gpjanik 1y ago
I somehow missed the "decode phase" paragraph and hence was confused - it's essentially that separation I meant, you're obviously correct.
20.
by gpjanik 1y ago
"Here's the key insight: each forward pass processes ALL tokens in ALL sequences simultaneously." This sounds incorrect, you only process all tokens once, and later incrementally. It's an auto-regressive model after all.
21.
by gpjanik 2y ago
This is finetuned to the benchmarks and nowhere close to O1-Preview in any other tasks. Not worth looking into unless you specifically want to solve these problems - however, still impressive.
22.
by gpjanik 2y ago
I mean again, even within a single sport, there are role differences, but the degree of fitness that you have to have to build 130kg muscle mass that JJ Watt has, and to run the 38.5 that Mbappe does... is just not the same level of fitness
23.
by gpjanik 2y ago
Ugh, NFL players need to run on average about half of what a third tier soccer players do. At the extreme, a goalkeeper runs about the same as most running NFL player. For sprinting, the peak speed is similar, but soccer players run 2-3km o
24.
by gpjanik 2y ago
Oh yeah, there are 4 slightly chubby guys in "elite" sports out of 10 000, therefore we got athleticism wrong. Also, sorry to say it so directly, but none of these guys would have a remote possibility to (athletically speaking - n
25.
by gpjanik 2y ago
What is "pirate" about this? A lot of big companies self-host npm to avoid chain of supply attacks.
26.
by gpjanik 2y ago
Where is EU when you need it? Subscriptions are a mess and it's one place in which EU could've forced something, but it won't. I also think they're mentally aligned with the idea of having to go through 20 forms to achie
27.
by gpjanik 2y ago
Regulation in this space works exclusively in favor of big tech, not against them. Almost all of that regulation was literally written for the benefit and with aid of the big tech.
28.
by gpjanik 2y ago
Hi from Germany. In case you were wondering, we regulated ourselves to the point where I can't even see the demo of SAM2 until some other service than Meta deploys it. Does anyone know if this already happened?
29.
by gpjanik 3y ago
I understand that all these new models are an attempt to catch up with GPT-4, but frankly speaking, in the current shape and form, they're almost entirely useless. I frantically tried anything available on Groq to improve performance o
30.
by gpjanik 3y ago
You can literally make that entire graphic a font glyph. I understand there's caching, but your point doesn't really apply to the above example.

More ›