Krisztián Szabó

trying my best
@[email protected]

[Krisztián Szabó @V4K0ND]

Can't even express how much I miss my old (pre-Elon) cybersec Twitter feed.
I can find some condoleance in reading these blogs: https://t.co/6Zt2nE54aA

[nyxgeek @nyxgeek]

google sucks, so I'm putting together a web directory of my favorite h/p/v/a/c sites.

only personal blogs/projects, no company stuff, just what people are doing for fun in infosec and related.

if anybody has suggestions, would love to hear them.

[Krisztián Szabó @V4K0ND]

Had a great time this year too at #FOSDEM2025

[Krisztián Szabó @V4K0ND]

TikTok seems to have been doing it: https://t.co/7vx8pXZfq4

[Drew Pavlou 🇦🇺🇺🇸🇫🇷🇺🇦🇹🇼 @DrewPavlou]

When you open a link to another website from within the TikTok app, TikTok injects tracking code that can monitor all keystrokes, including passwords, and all taps.

Please explain to us why the Chinese government have a First Amendment right to do this to 170 million Americans

[Krisztián Szabó @V4K0ND]

Checking if X/Twitter in app browser injects JS code: https://t.co/dWNLB2e2DB

[Krisztián Szabó @V4K0ND]

Nope, checked chrome / Firefox / Firefox focus

[Drew Pavlou 🇦🇺🇺🇸🇫🇷🇺🇦🇹🇼 @DrewPavlou]

When you open a link to another website from within the TikTok app, TikTok injects tracking code that can monitor all keystrokes, including passwords, and all taps.

Please explain to us why the Chinese government have a First Amendment right to do this to 170 million Americans

[Ali Alwashali @ali_alwashali]

From Pegasus to Predator - The evolution of Commercial Spyware on iOS

https://t.co/z9BMCp0ISC

[Bas Westerbaan @bwesterb]

Post-quantum key agreement (X25519MLKEM768) seems to be rolling out on Chrome for Android right now.

[Krisztián Szabó @V4K0ND]

Probably meant "most open to threat actors"

[Paul Thurrott @thurrott]

You cannot be serious.

Linux: "Hold my beer. The beer that I didn't just eject through my nose."

[swyx @swyx]

your annual reminder that .gitignore_global exists

please clean up your OS junk files, we live in a society

[Saagar Jha @_saagarjha]

Finally got around to reading about Private Cloud Compute and let’s just say I’m not super impressed

[Arda Büyükkaya @WhichbufferArda]

The Caffeine Phishing-as-a-Service (PhaaS) platform has undergone rebranding and is now known as ONNX Store. Key details include:

- Targeting Method: Cybercriminals use the service to send PDF attachments with embedded QR codes to financial institutions.
- Phishing Mechanism: The QR codes redirect victims to phishing sites that are set up to steal Microsoft email credentials and 2FA tokens.
- Data Collection: Stolen information is captured and transmitted via the WebSockets protocol.
- Evasion Techniques: ONNX Store uses Cloudflare's CAPTCHA to avoid detection by phishing website scanners and URL sandboxes.
- Anonymity Measures: Cloudflare’s IP proxying obscures the original hosting provider, making it more difficult to shut down the phishing domains created through the ONNX Store.
https://t.co/jbjZtUGIkK

[Krisztián Szabó @V4K0ND]

🥞 Another year, another @PancakesCon

[Soumith Chintala @soumithchintala]

PyTorch's design origins, its connection to Lua, its intertwined deep connection to JAX, its symbiotic connection to Chainer

The groundwork for PyTorch originally started in early 2016, online, among a band of Torch7's contributors.

Torch7 (~2010-2017)
These days, we also commonly refer to Torch7 as LuaTorch, as it was used via Lua. Torch7 was written by Ronan Collobert, @clmt and @koraykv in ~2010. I was deeply involved in Torch7 since 2012, with official "maintainer" status, joining these three original authors in April 2014.

Refactoring LuaTorch to be language agonstic (late 2015 to mid 2016)
LuaTorch's C backend with all the CPU and CUDA code for Linear Algebra and Neural Networks was deeply intertwined with Lua. So, a bunch of us lead by @lantiga @neurosp1ke @szagoruyko5 me @apaszke @fvsmassa refactored these backends to be agnostic of Lua, and usable independently. We did this after discussing online that we should move LuaTorch to a new, modern design, but hadn't quite framed what that design should be.

Writing a new Python based Torch (mid 2016)
@apaszke reached out to me early 2016 looking for internships. At that time, the entire LuaTorch team at @AIatMeta was ~3 people (@GregoryChanan @TrevorKilleen and me). I asked Adam to come do an internship to build the next version of LuaTorch, with modern design. @colesbury was in-between projects, so he joined in full-time as well.

We started from a fork of the LuaTorch, LuaTorch-nn codebases specifically for two things:
1. the TH/THC and THNN/THCUNN C backends
2. Building a compatibility with LuaTorch's checkpoints, so that LuaTorch users could smoothly continue into PyTorch. We did this by transpiling LuaTorch's `nn` code to Python. We called this package in PyTorch `torch.legacy.nn`.

Then, coming to the design itself, we debated a lot of designs. The strong inspirations were:
1. torch-autograd (written by @awiltschko and @clmt ) 2. Chainer (written by the team at @PreferredNet ).

@ebetica who loved Chainer would obsessively tell us its the best thing ever, so he came on board to build this together with us. Quite a few others such as Natalia Gimelshein and @adamlerer part-time got involved in various ways.

We wrote the code for the new design of PyTorch from scratch.

The connection to JAX: inspiration of HIPS/autograd

@awiltschko's torch-autograd (which was a big inspiration for PyTorch's design) was directly inspired by @SingularMattrix @DougalMaclaurin @DavidDuvenaud and @ryan_p_adams 's HIPS/autograd library, so in that indirect sense, we had strong inspiration from Ryan's library. In fact, we were so oblivious to certain origins that we named our Autodiff engine `torch.autograd` because we thought it was the norm within the autodiff community to call things "autograd". We later had to apologize to @SingularMattrix and team about the name of our subpackage conflicting with their `autograd` package.

Later, @SingularMattrix @DougalMaclaurin and others went on to create JAX, continuing down their design exploration of HIPS/autograd.

The inspiration from Chainer -> PyTorch and the inspiration for PyTorch -> Chainer v2

Chainer was a strong inspiration, we really liked the concept of Chains and stuff. The Chainer devs were friends of us, and we interacted with them a lot as well. I visited them in Japan in 2017.
Chainer's design is in my opinion a revolutionary design -- very original for that time and pretty awesome. We are proud to have been inspired from it.

However, unlike people commonly misunderstand and misattribute, we didn't simply replicate Chainer's design as-is. People have posted online on how PyTorch's design looks exactly like Chainer's and hence its origins are just copy-paste -- and that's because they don't understand the co-evolution. After PyTorch's release, Chainer evolved to include some of PyTorch's good ideas, and eventually they converged to look the same. For example, Chainer's nn Chains required you to pass in all the modules to the constructor (or use an add_link). The concept of self-assignment (i.e.) `self.conv = nn.Conv2d(...)`, the concept of `Parameter` was something we introduced as an evolved upgrade from Chainer v1. We also innovatively changed the way the autodiff engine was implemented -- things like "variable versioning" to detect correctness issues with inplace operations, and a few other new ideas, ideas that eventually went back into Chainer in their v2.

When Chainer's community wanted to stop development, @PreferredNet amicably and proactively joined the PyTorch community (link in references).

Post-launch evolution (2017 to present)
This post doesn't have the space to cover PyTorch's:
* evolution to add in ideas from Caffe2 (@jiayq @dzhulgakov et. al)
* its 5 compiler designs before we landed on what seems great (Zach DeVito, @ezyang @apaszke @jamesr66a Jason Ansel Christian Sarofeen et. al.)
* our inspirations from JAX and designing functorch (Richard Zou, @cHHillee @vfdev_5 Animesh Jain)
* our entire distributed design and evolution
* the origins of the sparse package (@braizh ) and its evolution (@cpuhrsch et. al.)
* PyTorch's domain libraries
* data loading (@colesbury @TongzhouWang )
* community design, community growth, innovation in design of incentives (@ptrblck_de Alban Desmaison, me)
* Several innovations in GPU code (several key folks from NVIDIA and Meta)

Many other parts of PyTorch that I didn't include -- its become somewhat of a monolith at this point.

Attributing ideas is healthy, awesome and should be done more often
Since PyTorch has launched, several new libraries have used the designs and ideas from PyTorch -- the particular new ideas that we introduced eventually propagated to many other libraries -- and this is awesome.
We are proud to have been inspired by work before us, and we are proud to have inspired work after us.
We also take pride in always attributing our inspirations clearly -- torch-autograd, chainer and many other projects that have inspired us in lesser ways.
I think people don't do this enough, attribute their origins clearly -- either ego or corporate controls come into play to erase history -- and people should do more here. In that sense, I'm really proud of my JAX friends who see framework design as a scientific endeavor, openly discussing ideas and evolutions, and proudly attributing their origins and inspiration.

References:
1. My reply in March'17 on the origins of PyTorch: https://t.co/zjViIvFm2M
2. Chainer's v1 design: https://t.co/gMdDC1ZMKR
3. https://t.co/tra6N8bJiT
4. PyTorch's autodiff innovations in a short paper: https://t.co/ke1bbkzv5o
5. The PyTorch paper: https://t.co/tbcGibPqgR

[Krisztián Szabó @V4K0ND]

It is on the back of the monitor!

[Krisztián Szabó @V4K0ND]

🥞 Another year, another #pancakescon

Awsome topic from @thomasareed: Mac Incident response, going to be useful since I finally have a Mac!

Can you spot it in the image?

[Krisztián Szabó @V4K0ND]

@karpathy - 😄it explains memes
https://t.co/IL4lp6Eaue

[Krisztián Szabó @V4K0ND]

@SamanthaZirkin You're welcome, just ask!

[Krisztián Szabó @V4K0ND]

So you want to backup your social graphs?

Here's how I did it using a python library:

[Krisztián Szabó @V4K0ND]

@IanColdwater I did a writeup of how I exported my social graph, this method is probably not the easiest but faster than waiting for the "request an archive of your data" and safer than using a third party service.

You can check it out here:
https://t.co/Cce6WxREpA

[Krisztián Szabó @V4K0ND]

So you want to backup your social graphs?

Here's how I did it using a python library:

[Krisztián Szabó @V4K0ND]

@Celeste_pewter I wrote about another way to export it, will be faster I think than waiting for the request for the archive: https://t.co/Cce6WxR6A2

[Krisztián Szabó @V4K0ND]

So you want to backup your social graphs?

Here's how I did it using a python library: