Did you remove Wirecard from your LinkedIn?
2,352 karma · joined April 23, 2011
https://github.com/19h
kenan [at] sly.mn ksulayman [at] hex-rays.com
They’re obviously trying to prevent companies from bypassing their extortion proposal in response to DMA by simply offering a PWA to users that can work around the „core tech fee“..
MacPaw lists Russian-developed software as a risk because the government can access your data at any time — this is self-hosted open-source software though.
The FSB can’t just access your local server with an arbitrary court order.
Therefore this doesn’t feel like a legitimate concern but more like Russophobia, which I understand but also think is utterly unasked for as I know first hand how much Russian developers are suffering from the stupidity of their government.
This is especially good to know when you’re exfiltrating .git folders that are unprotected and publicly accessible on websites :-)
Seriously though, make sure you’re deleting .git folders from deployed websites - or at least prevent them from being accessed. In at 300+ of the 1000+ website .git folders I downloaded from random websites there were private access credentials to AWS, GitHub, OpenAI, etc.
Iceland is such a beautiful place.. although I was angry and sad for most part of the trip seeing how tourism changed Iceland over the last 20 years. As recently as 2014 we had to ask the locals in mostly sign language and paper maps how to navigate the Westfjords; there's 5G now in even the last corner of the island.. guess I'll need to resort to Svalbard and Greenland to go truly off-grid in the future :-)
Essentially, if an operator was to "do a good thing", they'd legally become liable for all traffic. That was quite a learning experience for me.
THAT SAID, I've been using this for months and it's been performing extremely well.
Essentially, they’ll bill you 50€ per person and you’ll get it back (aka the authorization is cancelled) when you show up, but the transaction is executed when you don’t.
A few years back there were many empty tables during prime time because of the reservations, and now there’s barely any table unused.
It seems to work and I’m absolutely happy to reassure the restaurant with money that I’ll be there.
[1] https://www.windy.com/-Satellite-satellite?satellite,18.937,...
H2O is absolutely battle proof -- its h2 and h3 implementation perform amazingly well.
One interesting find is that I wrote an integration with GPT4 for binaryninja and funnily enough when asking the LLM to rewrite a function into “its idiomatic equivalent, refactored and simplified without detail removal” and then asking it to find vulnerabilities, it cracked most of our joke-hack-me’s in a matter of minutes.
Interesting learning: nearly all LLMs can’t really properly work with disassembled Rust binaries, I guess that’s because the output doesn’t exactly resemble the rust code like it’d do in C and C++.
Absolutely can't wait to test drive this one -- although I'm pretty sure my 96GB M2 MacBook is unable to run it.. time for M2 Ultra? :-))
Edit:
> You will need at least 400GB of memory to swiftly run inference with Falcon-180B.
What the ...
We dedicated an entire team of 6 for two months on evaluating LLMs and while Claude 2 was the final choice, we found Llama 2 70b to be absolutely great (for summaries and structured data generation from free text).
We chose Claude because running Llama on an A100 or H100 comes with a baseline cost that doesn’t go to zero when you don’t need it (you could spawn a new instance but right now GPUs are so rare everywhere except for expensive cloud providers that it’s possible you don’t get one).
That said, we found the smaller Llama models to be so hilariously bad we have an internal slack channel where Llama 7b writes jokes and the funny part isn’t the jokes but how utterly stupid and random they are.