snarkconjecture

7 ms·

1.
by snarkconjecture 20d ago
On the one hand it's much more realistic and detailed than an after-image; if I picture a red apple, and someone asks what kind of apple it was, I can retroactively say "oh, like a Red Delicious, really dark and shiny and with tha
2.
by snarkconjecture 2mo ago
I think the most important question here is whether Huggingface was "in on it" – whether they knew in advance that this was going to happen and agreed to it. Second most important is whether OAI caused this to happen on purpose, w
3.
by snarkconjecture 2mo ago
It's absolutely not a trivial combination of prior work, or mathematicians would've found it already. This was not an obscure conjecture.
4.
by snarkconjecture 2mo ago
Checking the counterexample is a Calc II problem. Finding it was beyond the efforts of the best algebraic geometers for almost a century.
5.
by snarkconjecture 2mo ago
Current LLMs can absolutely multiply floats without a tool call. In fact, that's a much more rote symbol-manipulation task than doing original math research.
6.
by snarkconjecture 3mo ago
No, it's ^J because `J` is 0x4A in ASCII and `\n` is 0x0A, just as `I` is 0x49 and `\t` (tab, ^I) is 0x09. In the stone age, pressing CTRL flipped that bit, so ^J is literally "ctrl-J".
7.
by snarkconjecture 4mo ago
Unless I'm mistaken, this uses "standard deviation" to refer to standard error throughout. They differ by a factor of sqrt(num_samples). This is actually much more commonly useful than the t distribution, in my experience.
8.
by snarkconjecture 4mo ago
The map contains a bunch of references to America, the West Indies, Guiana, and Mexico. (Often with a connotation of "faraway exotic place" or "exciting new international development".) He may not have written about the
9.
by snarkconjecture 5mo ago
> These reactors can be made safer, but they all still have a foundational design flaw which means the ultimate goal should be replacing rather than continually spending money reinforcing. This was about the Fukushima reactors that were
10.
by snarkconjecture 5mo ago
> the underlying working principles are the same as GPT-2 I don't think anyone was claiming otherwise. Sonnet is still better at writing code than GPT-2, and worse than Opus. Workflows that work with Opus won't always work with
11.
by snarkconjecture 5mo ago
I am much more likely to fault them for omitting important information specifically to hide a weak point of the product rather than out of laziness.
12.
Quine Game(adam.scherl.is)
1 points by snarkconjecture 6mo ago|1 comments
13.
by snarkconjecture 6mo ago
Write a Python quine with three or four hands tied behind your back
14.
by snarkconjecture 6mo ago
Computer screens have three-dimensional color spaces. Tetrachromacy doesn't change that.
15.
by snarkconjecture 6mo ago
Tetrachromacy wouldn't affect a test taken through a phone screen.
16.
by snarkconjecture 6mo ago
Deep neural networks can generalize well even when they're far into the overparametrized regime where classical statistical learning theory predicts overfitting. This is usually called "double descent" and there are many pape
17.
by snarkconjecture 1y ago
It's more like saying pi is approximately "3..14". Easily corrected syntax errors aren't as bad as semantic errors.
18.
Hex Runes(adamscherlis.github.io)
2 points by snarkconjecture 2y ago|0 comments
19.
by snarkconjecture 2y ago
Versions numbers for LLMs don't mean anything consistent. They don't even publicly announce at this point which models are built from new base models and which aren't. I'm pretty sure Claude 3.5 was a new set of base mod
20.
by snarkconjecture 2y ago
Not really. Dirac's trick works entirely at a depth of two logs, using sqrt like unary to increment the number. It requires O(n) symbols to represent the number n, i.e. O(2^n) symbols to represent n bits of precision. This thing has ar
21.
Iterated Log Coding(adamscherlis.github.io)
109 points by snarkconjecture 2y ago|36 comments
22.
Iterated Log Coding(adamscherlis.github.io)
3 points by snarkconjecture 2y ago|0 comments
23.
by snarkconjecture 2y ago
I think it's better phrased as "find the best rule", with a tacit understanding that people mostly agree on what makes a rule decent vs. terrible (maybe not on what makes one great) and a tacit promise that the sequence prese
24.
by snarkconjecture 2y ago
For the individuals shown in the graph, this buys about $6k per American (and after the first year you can't do it again).
25.
by snarkconjecture 2y ago
There are two kinds of naturalness principle in physics, sometimes called "technical naturalness" and "Dirac naturalness" respectively. Dirac naturalness is as you describe: skepticism towards extremely large numbers, en
26.
by snarkconjecture 2y ago
She's saying that a different model -- one of the three disagreeing methods for distance ladder measurements -- must be wrong, because they disagree with each other. But if one or more of those models are wrong, then there's not m
27.
by snarkconjecture 2y ago
It's the magnetic field that has the arbitrary sign convention. You can't determine the direction of a magnetic field from observations without using the right hand rule.
28.
by snarkconjecture 2y ago
That definition doesn't work well because you can have changes in entropy even if no energy is transferred, e.g. by exchanging some other conserved quantity. The side note is wrong in letter and spirit; turning potential energy into he
29.
by snarkconjecture 2y ago
It's not a dupe. "I commissioned a professional voice actor to give a full dramatic reading of that blog post."
30.
by snarkconjecture 2y ago
R squared has extremely nice properties. If you add a bunch of linearly uncorrelated variables together, then compute R² between each variable and the sum, the R² values will sum to 1.

More ›