cshimmin

10 ms·

91.
by cshimmin 3y ago
Well the article does show a photo of the wrong flag flying at (or very near to) the vatican. TBH I think the incorrect design looks better! But I suppose that's beside the point...
92.
by cshimmin 3y ago
That's not what an Oxford comma is, FYI. I think GGPs interpretation is grammatically correct but I agree that the phrasing is ambiguous. But given Tesla's documented unreliability one can assume what they meant...
93.
by cshimmin 3y ago
This hilariously proves OPs point. If satire, bravo, if not, well, we were all teenagers (or 22yos, in this case) once. In any case: most new startup founders are over the age of forty. Assuming that's what a 22yo on hn considers an &q
94.
by cshimmin 3y ago
I have a PhD in particle physics (and work in the field). So more than a layman's familiarity with quantum mechanics/field theory. I have no idea who this guy is; I know the name and for some reason I know he looks like a white-ha
95.
by cshimmin 3y ago
I have been using it to write bedtime stories and nursery rhymes for my 1.5yo son. Currently, his favorite things are butterflies, firetrucks, and the moon. ChatGPT writes some great stories about them all going on adventures together :)
96.
by cshimmin 3y ago
The dot product of two d_k-dimensional vectors is a scalar. So the entire NxN matrix qk_ij is just a bunch of scalar numbers. The rows are used to create a weighted sum (i.e. a linear combination) of the d_v-dimensional vectors.
97.
by cshimmin 3y ago
No offense taken, I have twice un-deaded your comments because you seem to be posting in good faith :) Again, you get used to it. Note, the underscore would represent a subscript (in the standard math-writing tool of LaTeX, but also in many
98.
by cshimmin 3y ago
Correct. In the context of LLM's, the "items" I refer to would be tokens. As a particle physicist, in the transformers I work with the "items" often instead represent either particles, or detector measurements.
99.
by cshimmin 3y ago
Apparently, you're missing the part where they're all defined :) It does take some getting used to, communicating math in English (and visa-versa). For example: "two learnable matrices Q and K ... are d_k by d_f". What m
100.
by cshimmin 3y ago
TBH I don't find this exposition enlightening. If you don't already know the pytorch API, the code will be pretty obscure. And if you do understand the API, it's still not very clarifying. Instead, it's a lot easier IMO
101.
by cshimmin 3y ago
Pretty much this. But the point is that a bitflip would be like, your CCD was supposed to read out value 0b00000 = 0 but due to cosmic ray interference, it read out 0b10000 = 16 instead. Instead, what's happening in GGP's experime
102.
by cshimmin 3y ago
mais ceci n'est pas une pipe!
103.
by cshimmin 3y ago
Well said re: gradient optimization vs. "getting slapped". However, note that since NN optimization is almost always nonconvex, we are NOT guaranteed to arrive an a optimal (or even close-enough) solution. A major limitation of gr
104.
by cshimmin 3y ago
Well, technically a circle can't be said to be a function but not for the reason you mean. A circle is a collection or a set of points, for example in a 2d plane, that are equidistant from a center point. Probably what you are trying
105.
by cshimmin 3y ago
Yeah but even in that case, "reward" is just the thing a NN is trying to predict. The NN itself is not receiving the reward (or any punishment). Instead, it's following gradient signals to improve that estimate of reward, whi
106.
by cshimmin 3y ago
True, but the hot pixels you see in the CCD isn't from bit flips. Rather the CCD is properly doing its job of collecting electrons. It's just that spurious electrons are being produced by the cosmic rays' ionization trail (as
107.
by cshimmin 3y ago
I would be surprised if the majority of bilingual speakers in the US aren't first/second gen immigrants. Not exactly known to be "a sign of coming from" wealth
108.
by cshimmin 3y ago
Not a dumb question! It definitely has nonzero vapor pressure (quick search says 2e-6 atm at room temp), so by definition it would boil at any pressure lower than that. The moon's trace atmosphere is something like 1e-15 atm. I guess y
109.
by cshimmin 3y ago
Thanks for the link, that's is so cool! Wikipedia says it was decommissioned "due to the test site's bad weather" (PNW). I guess one benefit of doing this on the moon is no weather, or atmosphere at all to contend with.
110.
by cshimmin 3y ago
Many years ago a colleague once told me about how one could take an arbitrarily large vat of mercury on the moon, and set it spinning. Gravity plus centrifugal force causes the surface to become a more or less perfect paraboloid, and boom,
111.
by cshimmin 3y ago
provenance*
112.
by cshimmin 4y ago
And how many are given away for free?
113.
by cshimmin 4y ago
And even if you are in horizon worlds, image segmentation is useful for room/object tracking for 6dof. And also so you don't stub your toe on the coffee table while you're in VR.
114.
by cshimmin 4y ago
I think it's a pedantic attempt to bring up the whole "initialism vs acronym" thing. You know, kinda thing that makes you real fun at parties.
115.
by cshimmin 4y ago
I mean, it's not a terrible convention. On the website they have a badge ("docker official image"), but devs aren't usually looking at the website, they're looking at their Dockerfile in vim or whatever. This is a s
116.
by cshimmin 4y ago
The conclusion section is amazing, highly recommend at least skimming to there if not RTFA.
117.
by cshimmin 4y ago
Who are you, who are so wise in the ways of clay?
118.
by cshimmin 4y ago
Reminds me of the hilarious aphorism from Italian (I believe): "If my grandmother had wheels she'd be a bicycle!"
119.
by cshimmin 4y ago
I think this comment is referring to the govt blocking the rail workers from taking collective action against the private companies.
120.
by cshimmin 4y ago
Lol yes, you might as well just say "in this example, simply solving the integral does the trick". The whole point of this article is that you can use dimensional analysis to get the form of the answer up to a dimensionless consta

More ›