grungegun

In a sense, I spend most of my time on material after quals because it's so much harder to understand.

So there's no performance gain for quantization enabled by the transformer architecture? It seems very strange that quantization works so well since in most of my experiments, the internal model weights of mlps look random.

Does anyone know if this works on vanilla deep networks? These quantization articles always seem to target LLM's which leads me to wonder if there's something special about the LLM architecture vs a vanilla deep architecture.

I taught one of my younger brothers a simple sign language when I was younger. His speech was delayed by like 8 months compared to the rest of us. Can't say for sure, but there's a need to emphasize spoken communication with the sign language.

This is somewhat overblown, and only really an issue for some particular theorems and obscure corners of mathematics. And this isn't a recent problem, Hilbert's proof of Nullenschtatz in the early 20th century had logical errors, but is otherwise true.

If a proof is used by 3 people it won't have much scrutiny, but once major results start to be based on the proof, it'll get reviewed more carefully and either accepted or rejected in the long run. The ABC conjecture is probably the biggest example.

the infinite density is actually after a finite amount of time, since black hole collapse has already happened, which is why it matters that they are predicted to have infinitely dense points at their centers. The key is that this isn't a limit, GR seems to expect that there are infinitely dense objects just floating around breaking space-time.

would be interesting if you could used deepfloyd first for image composition, then apply stable diffusion after for purely stylistic modifications

What does this mean? Isn't the quality of a model determined by how easy it is to get a good picture?

Loved this book.

When I first started programing, I was convinced that text applications weren't real programs, and the visual aspect of the book pulled me in to be a life-time programmer :)

For diversity, Dalle 2 has a random chance of injecting "women" or "black" after a prompt. When this happens, at least for me, it generally destroyed the quality of the images. Probably "King" was identified as a gendered word. You can find some discussion of this on the subreddit r/dalle2. Sometimes the images are quite poor, but in this case, openAI is doing additional tampering.

A twitter user figured out which words they were using by generating a lot of images with the starting prompt "A sign being held that says "

I'd argue that the incredibly fast rise in quality over the last year alone is what most people are interested in. DALLE and GPT-2 were always able to make heaps of trash. The trajectory of the quality of DALLE-N and GPT-N is what interests me in terms of the AI internet...

yep, that sounds right to me. also I see you're the author of the Noether article from Ars Technica. I enjoyed the read.

Yes, the quantumness of a differential equation is not a property of the differential equation, but a statement about one possible taxonomy of differential equation. Then, whether quantum-type diff eq's have unique properties pertaining to chaos conditioned on our knowledge of them being labelled 'quantum' is an interesting mathematical question.

Note that there is no Lagrangian for the NS equations, by the way.

I don't know much about fluid dynamics, but I was under the impression that Bennett derives the Lagrangian form in the book Lagrangian Fluid Dynamics

Differential equations don’t give rise to physical systems.

Sure, but differential equations describe physical systems, and there is a canonical way to derive a quantum differential equation from a classical one by quantifying the classical Lagrangian using the Path Integral formulation. Giving a fairly natural distinction between the types of equations

Whether any particular DE is a useful model of a particular physical system is a matter for the imagination, and either backed up or refuted by experiment.

This doesn't make sense to me. The Navier-Stokes equations are known to describe the classical behavior of water and are experimentally confirmed to predict things like trajectory. Their effectiveness has nothing to do with my imagination. If I write x=x' for the position vector of atoms in a fluid that will completely fail to describe anything physical.

The differential equations which give rise to physical chaotic systems operate on quantum states interacting with each other at the atomic level, so there has to be a link of some kind.

It looks like in the results section eqs (8)-(13), they give the main thrust of their idea. To demonstrate that not all machines are reversible they have to take the limit of the number of qubits of their machine as n goes to infinity, so their result is still in the limit, same as the typical laws of thermodynamics.

So this is primarily an interesting mathematical restatement at the quantum level. I don't know how useful this is, not having much knowledge about existing quantum information theory/thermodynamics.

Actually, it doesn't increase much faster than linear. It's quasilinear, and so in O(n^(1+\epsilon) for all \epsilon > 0. There are a variety of such results, so most objections don't hold weight.

The paper you linked to names challenges, but it only states those in respect to reaching practical scaling, which is a result of the constant that is associated with the growth rate, but not the growth rate itself.

False, except for a naive implementation, see: https://en.wikipedia.org/wiki/Quantum_threshold_theorem

If you look at the theorem, it is definitely sub-exponential in the number of qubits needed to correct for error.

For reference, see the book High Dimensional Probability by Vershynin. It's free online. See Theorem 3.1.1. It proves that a sub-gaussian random vector is in some sense close in norm to sqrt(n) where n is the number of dimensions. Most of these results are true up to multiplying by some unknown constant.

It didn't have very many interesting properties I could find

Usually notation is defined for higher order functionals when the need arises - when a new result is found to have interesting properties. Topological proofs on function spaces can use third order functions.

Mathematics only works as well as it does because mathematicians are all working on pretty much the same thing

I don't think any person alive can understand all the major branches of mathematics well. By that measure mathematics is very broad.

The doctors don't believe the story. The woman does. The doctors believe that the woman's belief caused her early death, but that's independent of whether they accept that story as true - people (the woman) can convince themselves of a lot of things.

Is there a reference for this to verify that this isn't just some historical reenactment? I'm trying to be more careful after I was once pranked by a friend who showed me a black and white video, claiming it was real, of people playing ping pong with nunchucks. I believed him at the time, but it was an ad campaign that someone on youtube had reposted to make it seems legit.

For those interested, there is a fairly basic WYSIWYG editor for this: http://www.tikzedt.org/

It's useful to throw together basic 2d diagrams, which can then be refined later

Plus, when you double click a spot in a pdf you generated with latex, Sumatra opens up your default .txt application with the line highlighted.

You're right. I've looked at quantum codes before, but apparently my understanding was lacking - thanks for the info.

There's one big issue with QC currently. There is no mathematical demonstration that QC is scalable. All the speed-ups need some form of error correction to scale the way digital computation has. All hardware advances are just increases in computational fidelity so far to allow more useful qubits. Shor has an error correction algorithm for a single qubit, and there have been improvements to that, but nothing general enough to show that QC can even scale like digital computers can.