GistNoesis

12 ms·

91.
by GistNoesis 1y ago
I think there is something wrong with your face filter settings on all your videos.
92.
by GistNoesis 1y ago
The problem with this visualization is that the pattern we see are meaningless. It's making me thing of sieving, like in Sieve of Eratosthenes. But we have progressed a lot since. What's nice about primes are the abstract ideas th
93.
by GistNoesis 1y ago
Some times it's useful to see the elemental transformations, but often striving for a higher level view makes understanding easier. It's particularly true when you try to apply it to physics. Often where you are introduced to vect
94.
by GistNoesis 1y ago
The way that really made me understand gradients and derivative was when visualizing them as Arrow Maps. I even made a small tool https://github.com/GistNoesis/VisualizeGradient . This visualization helps understand op
95.
by GistNoesis 1y ago
How would you go to make the llm generate object easier to 3d print (or manufacture)? Things like not using too much material, reducing the need for support, and maybe whether or not the produced part would be tough enough. Are there some d
96.
by GistNoesis 1y ago
As an experiment, I just spray painted 66 magnet spheres in half, to make a physical display in 5 minutes. I manually rotated the sphere into position and it holds the image. https://gist.github.com/unrealwill/b8f585758
97.
by GistNoesis 1y ago
> I think Eliezer’s take here is extremely bad, ie the AI doesn’t “know it’s making people insane”. The situation is more complex but interesting. Let's enter the details of how LLM are trained. The core of a LLM is all instinctive
98.
by GistNoesis 1y ago
The stated problem is getting Off-policy RL to work, aka discover a policy smarter than the one it was shown in its dataset. If I understand correctly, they show random play, and expect perfect play to emerge from the naive Q-learning train
99.
by GistNoesis 1y ago
In robotics the URDF format can be used to specify and simulate rigid-body systems. You can also check various physics engine input file formats like MuJoCo (MJCF format). (It's just some variant of XML). If you want soft-body dynamic
100.
by GistNoesis 1y ago
If only... When you first encounter diffusion models, you usually see a well formed picture emerge from noise. And then you realize, there is no reason it shouldn't work for anything where you can add noise to. Which means everything.
101.
by GistNoesis 1y ago
Fast, you gotta go fast : Let me draw the roadmap of this line of thinking. - Let's start by the traditional autoregressive LLM, where one token at a time is generated. It's a fundamentally sequential process which maps well to th
102.
by GistNoesis 1y ago
See "Deformations and distortions" paragraph in https://iquilezles.org/articles/distfunctions/ where it suggests playing with the ray-marching step size.
103.
by GistNoesis 1y ago
Time is irrelevant. What matters are units of computations. When things are predictable they can be simulated fast : A spinning ball in the void can be simulated for 10^78 years in O(1). When things are fuzzy, they can be simulated fast : A
104.
by GistNoesis 1y ago
From the mathematical point of view the literature is about the distinction between a "filtering" distribution and a "smoothing" distribution. The smoothing distribution is strictly more powerful. In theory intuitively t
105.
by GistNoesis 1y ago
I have not tried it, but lichess is open-source : so you can probably customize it for your specific use case. More specifically modify line that allow to pick the theme https://github.com/lichess-org/lila/blob&#x
106.
by GistNoesis 1y ago
https://lichess.org/training/defensiveMove For training more related to this article : https://lichess.org/training/xRayAttack You can find the list of specific training themes here : https:&#x2
107.
by GistNoesis 2y ago
124 light years seems quite close, it seems to be some sort phytoplankton for now, what can we send there quickly to eradicate them before they become a problem. How much mass of galactic "roundup" should we send ? Do we have the
108.
by GistNoesis 2y ago
I am wondering about hardware keyrings like Yubikeys : From a security point of view is it safe to cross international borders with them ? How do you connect back to your accounts without one when all your devices are based on two-factor au
109.
by GistNoesis 2y ago
Yeah you're right, I should have said nested (it is a valid Anglicism from French "imbriqué(es)" )
110.
by GistNoesis 2y ago
Point 8, the fast dismissal of Monte-Carlo method is a huge blunder. The point of Monte-Carlo method is that you can tradeoff accuracy for speed. The longer you run your algorithm the more accurate you get. But what's even more interes
111.
by GistNoesis 2y ago
What's the recommended way to handle being in such a Kafkaesque situation?
112.
by GistNoesis 2y ago
Will there be some diffraction optic in the future, can we just add some phase somewhere or will it need a complete rewrite ? I'd like to experiment with photon sieves and holograms.
113.
by GistNoesis 2y ago
Unfortunately bullying doesn't stop with adulthood. Bullies only get more power. Victims are ignored and blamed. How could they have children when everyone in power rather look the other way than help. Why would they bring a significan
114.
by GistNoesis 2y ago
I can't find it in Stellarium yet even though I've updated the satellites list now. I can find CSO-2 and CSO-1 but still not the shiny new one.
115.
by GistNoesis 2y ago
I wonder if some escrow protocol would be a solution to the ghosting problem. Something like, when a company make a job posting they must put some cash in escrow for each candidate they are currently interacting with, and if they ghost you
116.
by GistNoesis 2y ago
Is the recursion relation you use https://en.wikipedia.org/wiki/Hockey-stick_identity ? Are you sure you can't use https://en.wikipedia.org/wiki/Pascal%27s_rule instead for a simpler recursio
117.
by GistNoesis 2y ago
My code is just iterating over the relevant points. Line 186:187 is the trick to group negative probabilities. You can't apply this if you want to take into account a payout structure which depends on the result. The more flexible vers
118.
by GistNoesis 2y ago
I've edited my code to make the payoff structure dependent on the result of the events, (and therefore it can be dependent on the number of winners for the bet. It runs slower for now 1900s for 14 matches, of 3 different solutions. But
119.
by GistNoesis 2y ago
I have written some single threaded non-simd c++ code for this : https://gist.github.com/unrealwill/d1bc68d1f5c7ee6fe72b76dc5... When there are 14 matches, there are 3^14 = 4782969 different possible results and the s
120.
by GistNoesis 2y ago
Are you computing a single expectation value, or are you searching for the bet of maximum expected value among all possible bets ? How long does it take you to compute a single expectation value ?

More ›