I would guess it’s pretty hard to reconstruct data with a different LLM

I think the idea is to have compressor and decompressor use the exact same neural network. Looks like arithmetic coding with a learned function.

But yes model size is probably going to be an issue.

So… crystal ball, I don’t have access to the paper either. Think arithmetic coders as neural nets are function approximators. You send an initial token and the NN will start to generate deterministically, once you detect a divergence from the lossless ideal you send another token to put it on track again. Make it a sliding window so things don’t become too computationally expensive. You architect the model not to be smart but to need little guidance following “external reasoning” so to speak.

The actual disadvantage of this kind of thing will be the model size, yes you might be able to transmit a book in a kilobyte (100x or more compression) but both encoder and decoder will need access to gigabytes of neural weights, and that’s just for text. It’s also not going to be computationalliy cheap, though probably cheaper than PAQ.

Not the customer’s problem. Also, fraud.

But probably failure of due diligence because any seller who’s not a complete idiot would rather let the sale fail and let the company go bankrupt than risk committing fraud.

I was talking about the foundation itself, not foundation+subsidiaries. And yes ever since the writing was on the wall wrt. google funds they’ve been putting more and more money in investments to make sure they can survive, as opposed to grants. Still keeping with the foundation’s mandate, though, e.g. all their VC investments into AI are the polar opposite of what the likes of OpenAI are doing. Kinda sceptical e.g. huggingface will ever turn a profit, much less a significant one, but it’s important to have them.

Firefox is a way for them to make money for their charitable projects. Always has been.

They’ve been opening secondary streams of income – also via Firefox, like pocket and stuff, but it’s unclear whether that would even allow them to keep developing Firefox. It certainly would mean that Mozilla doesn’t have any money left over to give to others which is their main purpose of existing.

If you actually had a look at their actual numbers, they’re a charity you know, they’re public, you’d see that the bulk of money is spent on charity. Mozilla has never been a charity to develop Firefox, Firefox has always been the breadwinner for Mozilla’s charity operations.

The Commission already has an instance and the ECJ also has an instance. I didn’t find one for the parliament, they might not even want to have one but instead leave it to the parties. A parliament instance would amount to the EP president policing parliamentarian’s speech not just inside parliament but also outside of it, kinda iffy.

A crawler is a data processing machine, nothing more. therefor you are disrupting dataprocessing through data. If you think its not thats ok too.

Nah it’s definitely disrupting data processing, even though at a very low-key level – you’re not causing any data to become invalid or such. It’s the intent to harm the operator that’s the linchpin: “Jemandem einen Nachteil zufügen”. “Jemand” needs to be a person, natural or legal. And by stopping a crawler you don’t want to inflict a disadvantage on the operator you want to, at most, stop them from gaining an advantage. “Inflict disadvantage” and “prevent advantage” are two different things.

I would still advise to contact your lawyer in germany if you are thinking about hosting a zipbomb

Good idea, but as already said before: First, you should contact a sysadmin. Who will tell you it’s a stupid idea.

The intent is to get rid of crawlers which are disrupting the operation of your servers. That’s not intent of doing harm to the crawler’s operator, or their business. It’s analogous to telling a hawker to fuck off: Polite, no, but them being able to profit off you is not your responsibly, you do not have to accede to that. And intent to harm the ISP is even less reasonable to assume.

cant find the primary source anymore https://netzpolitik.org/2012/berec-studie-dpi-bei-vielen-providern-bereits-im-einsatz/

That’s out of date anyway. How about this one. DPI is limited to OSI level 5 and only allowed to resolve network issues – and a crawler crashing is not a network issue.

Countering Animator with Blender, that’s brutal. For at least some stuff Blender is also the better Illustrator.

Casting concrete requires building formwork to cast the concrete into. For any standardised shape constructing the formwork is easy: Just assemble it from the parts you have. Straight sections? The most common case. Rounded corners? As long as you’re fine with “should be round” and don’t require some very specific radius, those are probably also at hand. A gargoyle? Well that’s not an easy form to produce but once you have it, you can cast 1000 gargoyles.

Where 3d printing comes in is places where you have a shape that’s literally or nearly unique, where building the formwork would be a PITA. In all other cases, the traditional method is quicker and cheaper.

Also interesting is stuff like solar sintering plain sand.

It also needs crowd sourced legal assistance

Basically any union in Germany has access to an army of lawyers, your dues double as legal insurance covering anything related to work and welfare.

Severely disrupting other people’s data processing of significant import to them. By submitting malicious data requires intent to cause harm, physical destruction, deletion, etc, doesn’t. This is about crashing people’s payroll systems, ddosing, etc. Not burning some cpu cycles and having a crawler subprocess crash with OOM.

Why the hell would an ISP have a look at this. And even if, they’re professional enough to detect zip bombs. Which btw is why this whole thing is pointless anyway: If you class requests as malicious, just don’t serve them. If that’s not enough it’s much more sensible to go the anubis route and demand proof of work as that catches crawlers which come from a gazillion IPs with different user agents etc.

I don’t even know whether I understand I just hear MacOS users griping about fullscreen, and a quick google gave quite recent results. Especially with fullscreen being incompatible with other windows on top (each fullscreen window necessary is on its own workspace) which would be highly annoying in Blender. You can configure blender to have file open dialogues, render results etc. in its main window, but certain stuff like preferences always open a second one.

They don’t (usually) display the temperature but they definitely sense it, and react to it. When the sensed temperature is at or higher than the set temperature, the valve will be closed, if it’s lower it will be opened. Mere valves can’t do that.

That’s what a thermostat is: A negative feedback control system regulating sensed temperature towards a setpoint, and keeping it there. They’re simple, inexpensive, reliable. Yes having the temperature sensor right next to the radiator isn’t ideal but unless the room is quite large that’s not an issue. Also with large rooms you probably have more than one heater and thus thermostat. And you could, in principle, put the thermostat far from the heater but I’ve never seen that done.

Those “valves” are, in fact, thermostats. They use thermal expansion of wax to open/close the valve to get to their set temperature. Settings 1-5 are 12, 16, 20, 24, 28 Celsius.

Can you fullscreen windows under MacOS yet?

That’s not deciding anything in the information-theoretical sense. We rely a lot on approximations and heuristics when it comes to day to day functioning.

You can’t decide the halting problem by saying “I’ll have a glance at it and go with whatever I think after thinking about it for half a second”. That’s not deciding the problem that’s giving up on it and computers are perfectly capable of doing that.

But humans can solve undecidables.

No, we can’t. Or, more precisely said: There is no version of your assertion which would be compatible with cause and effect, would be compatible with physics as we understand it.

Don’t blame me I didn’t do it. The universe just is that way.

What are you trying to argue, that humans aren’t Turing-complete? Which would be an insane self-own. That we can decide the undecidable? That would prove you don’t know what you’re talking about, it’s called undecidable for a reason. Deciding an undecidable problem makes as much sense as a barber who shaves everyone who doesn’t shave themselves.

Aside from that why would you assume that checking results would, in general, involve solving the halting problem.