I’m not aware of any public frontier LLM provider that uses a static seed for inference. Meaning, even with an identical prompt and identical model you will not get the same output. Seeds should absolutely come back with the streaming metadata on requests imho, but they don’t in any api/harness I’m aware of.
Good news friend! Those machines are probably only 200 million each, because ram doesnt require quite the latest machines, way cheaper than you thought!
Bad news though friend! It’s not very economical to just buy one, because they’re so sensitive you need a cleanroom to house one. Then you need some really smart people to tune and maintain them, so you may as well buy a few, and now that cleanroom needs to be the size of 6 Costco’s.
Oops where did my 5 billion dollars go?
my guess is something like this
https://330k.github.io/misc_tools/unicode_steganography.html
A model could hypothetically be trained to insert zero-width characters (I doubt any have though). But any other layer could also very trivially insert these codes. The inference engine could be designed to delay output streaming by however many tokens is required to embed their coding and ninja-insert them during the decode stream. A proxy between the inference engine could insert them. A harness could insert them. Hell, even the rendering javascript frontend in your browser could insert them.
Either the inference engine or proxy would be the prime target if they want to enable this on api responses as well as copy/paste from a chat interface. They could also do a combination of the above depending on final output mode.
You are definitely correct though that it’d be trivial to detect and strip by someone aware of it.
Not necessarily, there are valid “characters” that are not rendered in most text parsers, or are rendered as whitespace. A great example of this is the byte-order-mark (BOM). You can embed a BOM code in the whitespaces of a text string and it looks exactly the same to a human as one without.
I assume that’s what they’re talking about here
Here’s an example of a steganography encoding technique using this method. https://330k.github.io/misc_tools/unicode_steganography.html
I don’t think his argument is that we should restrict anything, more so that circumvention requires learning, and is a great motivator. I definitely can +1 this from my own youth. Learning how to bypass well-intentioned internet protections on school computers so I could play flash games sparked an interest in technology that I may never have discovered otherwise.
It’s a weird internal conflict to know that my core is shaped by an experience I otherwise believe shouldn’t exist in a reasonable society.
Was the goal a sleeper build? I’m trying to figure out how you ended up in this position lol
You can actually disable the auto-download-via-windows-update mechanism in windows via a gpo. Would be nice if there was a toggle in settings but it does exist at least.
Computer Configuration > Administrative Templates > System > Device Installation > Prevent automatic download of applications associated with device metadata
Even with your comment it took me three lol
GOTTEEEEM!
There is precedent for anti-feature labels
It would be a really nice addition
Not to defend spectrum, but there is an actual reason for this besides malice.
Coax cable internet uses analog switching and routing between you and your ISPs first Ethernet hop. It’s basically a series of unconfigurable hardware low-pass and high-pass filters to ensure that upload traffic and download traffic flow cleanly.
When cable internet started and these systems were installed in your neighborhood, a range of low-frequency (low-bandwidth) channels were selected for uploads, and a range of high-frequency (high-bandwidth) channels were selected for downloads, going off the assumption that more people wanted downloads than anything else.
Fast-forward to today, and the ISPs don’t really have any choice but to sell you the asymmetry that they installed 20-30 years ago. The only way to redirect the flow and assignment of channels would be to rip out and replace all of those analog filters, and they’re not going to do that because grandma isn’t going to go without cable for 3 weeks. Grandma will call, get angry, and pick a new cable provider. Grandma is by far their most consistent and profitable customer.
I don’t understand, is suggesting someone write a 7 line bash script in a Linux forum post about using a cli downloader, unreasonable?
It would be more accurate to say they try and block the vpn providers and Tor exit nodes. In practice, they do have ~95% of Tor exit nodes blacklisted but major vpn providers it’s closer to 40%. Vpn providers go through a lot of effort to hide their exit traffic and it’s pretty simple to use a provider that has a reasonably robust cli to just rotate nodes until you find one that works
Connect
Test known-good video using yt-dlp
If fail, use vpn cli to connect to a new node
Reeat until success
If your provider let’s you connect to specific nodes you can even store the last-known-good and use it on first connect until it gets blacklisted, extending uptime considerably
I like the stable, but newer packages that Ubuntu offers, plus the first-party support you get from a lot of projects. I wish debian was the standard dpkg assumption for first-party support but it unfortunately is not. Ubuntu makes it pretty easy to build automation on top of that is good for 5+ years without touching.
Yes I have, admittedly it’s been about a decade but at the time I switched because getting hardware acceleration in chrome and Firefox was difficult on debian, trivial on Ubuntu. I like the stable, but newer packages that Ubuntu offers, plus the first-party support you get from a lot of projects. I wish debian was the standard dpkg assumption for first-party support but it unfortunately is not.
Ubuntu server (headless)
Install pulse, x11, awesomewm.
Building the gui stack yourself on top of Ubuntu server gives you all of the debian/Ubuntu stability and issue searchability with none of the gnome/canonical cruft.
Because the philosophy of the Linux kernel is
We don’t break userland
They have no way of knowing who is using what parts of the Linux kernel, where, and why. Generally if it’s not hurting anyone it just stays until it causes a problem.
Is this hopium or did they release a statement suggesting an 3.8 35b a3b was in development for release?