zdt

computational physics, pure maths, ML, and CTF

[™³k @_tm3k]

A retard in motion is always better than a genius at rest

[Conks @conksresearch]

but they were, all of them, deceived, for another watch was made…

[zdt @ZeroDayTea]

It seems that the hillclimbing capabilities of the LLMs are particularly well suited for problems in analytic number theory. I wonder what the success rate is on attempted problems in less computationally tractible fields

[levent @__alpoge__]

3/2 - 1/\sqrt{2} cot 1/\sqrt{2} ~ 67%, well played universe

[ali @aliteracy]

*taps window*

[Alec Petridis @chop0_]

if you're at yc startup school you're ngmi

[levent @__alpoge__]

hello there the jacobian conjecture is false thanx to my close friend akhil for asking about it and my other close friend fable for working during the world cup final

((1+xy)^3 z + y^2 (1+xy) (4+3xy), y + 3 x (1+xy)^2 z + 3 x y^2 (4+3xy), 2 x - 3 x^2 y - x^3 z): \C^3\to \C^3, has jacobian determinant -2, and sends (0, 0, -1/4), (1, -3/2, 13/2), and (-1, 3/2, 13/2) to (-1/4, 0, 0)

[Psyho @FakePsyho]

the biggest cultural clash in tech

[LaurieWired @lauriewired]

Modern DRAM is based on a brilliant design from IBM.

But, we're still paying for a latency penalty that's existed since the 60s!

In this video, I'm introducing my research project (Tailslayer) that immensely reduces p99.99 latency on traditional RAM!

By implementing a hedged read strategy taking advantage of (undocumented!) channel scrambling offsets, I've gotten as much as 15x reductions in tail latency.

The technique works across Intel, AMD, Graviton, DDR4, DDR5, x86, ARM, you name it.

Check out the C++ lib I wrote, watch the video, and try it yourself!

[BSF 🇬🇧🐀 @BSF42069]

We’ve come a long way

[james hawkins @james406]

wow, today i learned

[َ @4NGELWING]

all quiet on the frontal lobe

[lcamtuf @lcamtuf]

If you ask AI to rewrite the entirety of an open-source program, do you still need to abide by the original license? In philosophy, this problem is known as the Slop of Theseus

[zdt @ZeroDayTea]

@tetsuoai @SpaceX What's the reasoning? It's my understanding that radiative cooling in space is way way way less effective than the convective cooling that happens with liquid cooling in on-earth data centers. On top of that having to build radiation resistant large scale electronics?

[zdt @ZeroDayTea]

@gf_256 awe baby bro is all grown up in the big city

[Anthony Bonato @Anthony_Bonato]

"Mom, the topologists are acting weird again"

[zdt @ZeroDayTea]

@dystopiabreaker inaccurate. the rat's not bloated enough

[zdt @ZeroDayTea]

The kernel exploit -> open safari webpage -> another kernel exploit flow of this chain is the most modern 0-day thing ever

[Andrej Karpathy @karpathy]

"Operation Triangulation"
https://t.co/DUBBQWPqTS

A newly discovered spyware campaign targeting Apple iPhone using a zero-click remote code execution via an attack chain of 4 zero-days, including highly mysterious, completely undocumented MMIO registers and hardware features that are not even ever used by the firmware.
TLDR the attack begins with an iMessage to an arbitrary phone that, without any user action and invisibly, gets it to collect and upload tons of private data (and much more, e.g. microphone recordings) from there on, and actively takes steps to hide all of this activity from the user and aspiring forensic researchers. Apple has patched the core vulnerability on Oct 25, 2023.

"This is definitely the most sophisticated attack chain we have ever seen"

The talk itself, a lot more wild information there:
https://t.co/eTEeltBMpD

The author of this attack is unknown, as is the method by which they gained knowledge of these unused, undocumented hardware features. Russia's intelligence service accused Apple of providing the NSA with a backdoor.

For a more general audience intro to this underworld I usually recommend the book "Countdown to Zero Day".

[zdt @ZeroDayTea]

meanwhile high energy physicists are wondering why this is gaining traction

[Matthias Niessner @MattNiessner]

Google's Gemini paper has about 950 co-authors :)

That's 10 full paper pages that list the author names. In comparison, the method description is only a single page (3 pages if training is included).

Don't use "𝑒𝑡 𝑎𝑙." when citing the paper!

https://t.co/XWO5bP5YY8

[LiveOverflow 🔴 @LiveOverflow]

So... where are the reverse engineering AI papers? I want to see models trained with many gigabytes of binary->code mappings

[View Source @vsctf]

Less than 2 weeks until vsCTF 2023, View Source's 2nd annual capture-the-flag competition with $1,000 in prizes!

Register at https://t.co/1izWXk6T5l