I've noticed growing use of similar options in the UK already, like Payit and truelayer (often shown in checkout via "Pay by bank"). Ebay and Whetherspoons are examples of places providing these options.
The issue is in the UK we passed a law that "banned credit card charges", but what it actually did was ban discounts for alternate payment methods, meaning everyone has to pay the credit card fees, even if you pay cash. (this was the effect, you can still offer discounts, but the base CC fee legally needs to be baked in first.)
This will mean, its harder for competing payment services to function. And its easier for the big credit card companies to up the % they charge without the public noticing, or blaming them directly.
It was fascinating watching the media coverage, twisting themselves in knots to miss these points while cheering how we were 'standing up to the big cc companies', it was actually quite cringe-worthy at times.
RSI here is for "recursive self-improvement", instead of a harness being built to help users with repetitive strain injury like I first thought when reading the post title
Well for starters the use of the term "Recursive" is very dubious.
This is as far as the eye can see all very iterative, there is no tail-call or anything fancy, it is loops. Also "Recursive" I feel kind of tries to imply the LLM's weights are being pushed around in some feedback, because to recurse you have to invoke the thing you're recursing into at its very start right? That would be reinforcement probably, and there is none of that in any such RSI so far, at least not public. Please someone contradict me with examples.
So please: "ISI" for iterative self improvement is fine. Also "ISI" does not fit the (outdated!) Vernon Vinge "singularity" trope, and that is a good thing!
The example in the docs of improving nanochat is iterative. It's a looped process in one thing altering a second thing.
What would be recursive is raven updating raven to make it better at doing things. For what I picture as RSI the important part would be that it's able to make itself better at doing things and better at improving itself.
Now there's an "evolver" part that improves the harness over time but I don't know how far that goes or what scope it has to update things.
I guess it's that if you have a function called "optimise" that takes functions and makes them better, calling optimise(my_process) is iterative regardless of how many times you do it. Calling optimise(optimise) is inherently different.
A register is the obvious choice, but of course then you get a gold rush and people squatting on XPZ and stuff without even having invented an opaque term behind it.
A very popular yoga kata called sun salutations (available in various versions depending on your fitness/advancement level) helped me get rid of wrist and thumb RSI. It stretches and gently stresses these load bearing ;) joints, thus strengthening them.
Hope this helps the last few folk before searching for RSI will be impossible due to the term being taken over.
For those afraid of Satanism in yoga, there's also non satanic variants where you don't greet each other saying Namaste or say Shanti anywhere during the practice.
As long as we’re burying things for archaeologists to find, let it be known that AGI once stood for “adjusted gross income”, a measure used for income taxes.
The blueprint will be open source and you could technically build it yourself if you have the patience. They are quite proud of how few parts are actually involved in it
I respect that point of view. But to me, it being easily hackable and lacking the surveillance features of all other modern printers[0] is still a strong selling point and I hope for the project to succeed.
And that's entirely fine; it can a positive thing without being open-source. An institution can be a positive thing without being democratic. A food can be delicious without being a meat.
As much as this is nice protect, and physical one so different things any, still you cannot give/sell some software/design/media and restrict who, where and how someone can use it (noncommercial, not in this country, not at that phase of moon) while calling it open source, public domain, or similar.
As long as softwsre isn't going to be proprietary, it is a good idea just shouldn't misguide about being open source.
> Open Printer will use the Creative Commons BY-NC-SA 4.0 license for all of its files, including electronics and mechanical design files, firmware code, and the bill of materials.
> Many of the main authors of the Git source code are friends of mine and I have no intention to steal anything from anyone, only to make their great ideas more broadly useful.
By working-around/subverting the terms they provided their contributions under? While you claim to be doing this in good faith, and state "it's what's best for the broader Git ecosystem", that's all based on your own opinion which appears to ignore the benefits and intent of licenses such as the GPL.
Out of interest, Would you be happy for someone to do the same with the GitButler source code? (Feed it through an LLM and re-publish the result under an MIT license with different branding)
I've been using PyInfra for a while, albeit just for simple automation (Updating systems, checking certain stats) and I'm a big fan. Compared to Ansible, I found the docs, syntax and usage patterns much easier to get on with. Might just be a preference thing, but I always had trouble going through the Ansible docs.
Ran into some bugs, like one machine that seems to cause errors and mess up the output on restart, although that looks like it might have been addressed in this release.
Glad it clicked. The Ansible vs PyInfra docs gap isn't really preference, YAML plus Jinja plus a custom DSL is just more cognitive load than plain Python with type hints. Once you can grep the source and read it like normal code, going back feels rough.
On the restart bug: if it resurfaces, an issue on GitHub with the OS, connector (ssh/local/docker), and raw output would help a lot. The 3.x line cleaned up a bunch around connection handling and output buffering, so there's a decent chance it's already fixed.
Thanks for the video, will watch. Hands-on intro content is exactly what the project needs more of.
Really need to try PyInfra, the concept sounds nice.
You don't have to do crazy things with Ansible for that yaml DSL becoming the opposite of helpful. Things which would be quite straightforward to express in code become quite cumbersome, hard to understand and hard to debug. Also Jinja is often a horrible choice (you don't have in Ansible). Also Ansible excessively requires it in places where you want proper types and not just a string.
I can understand that viewpoint. Ultimately though, audience/growth is not a core success metric for me, and the values/points explained in the blog post are more important. Plus there's a factor of wanting to help de-centralise away from GitHub, and help provide momentum to alternatives.
Looking at those who have starred the new Codeberg repo, at least 15 people are new today, and thus form part of a bigger audience on Codeberg.
BookStack maintainer here. Just to clarify on that, the GitHub repo will continue to be updated and mirror the Codeberg repo (including release tags/code) for the foreseeable future, it's just that I might stop specifically publishing GitHub release entries (details on the release tag) at some point to avoid the duplication of work.
Just a warning, the license [1] specifically blocks EU use:
> 3. Conditions for License Grant. You represent and warrant that You will not, access, download, install, run, deploy, integrate, modify, or otherwise use the Model, directly or indirectly, within the European Union.
Even so, why would the licensor put it in and force it through a license. It's on the licensee to check the laws and regulations they themselves operate in.
The EU AI Act is supposed to affect all AI "providers", which includes any "natural or legal person, public authority, agency or other body that develops an AI system or a general-purpose AI model or that has an AI system or a general-purpose AI model developed and places it on the market or puts the AI system into service under its own name or trademark, whether for payment or free of charge" [0].
This would plausibly include anyone developing an LLM, even if they aren't selling access to it or building applications based on it. There are several exemptions, and the Act obstensibly avoids creating burdens for most general-purpose LLMs, but the point is that Huawei wants to avoid any worry by not "plac[ing] it on the market" in the first place.
I don’t agree. Tools like DeepL were and still are better than Google Translate long before chat bots became a thing. The French-made Mistral AI is pretty decent as well.
FWIW, I refactored 500+ Junit 4 to Junit 5 tests with locally running Mistral 8B on an M3 MBP. It worked flawlessly, but surely I cannot attest for other use cases.
And who thinks that, for even a second, that an European (in this case) will not download, install, and try to run this just because the LICENSE says you can't?
FYI, this is not intended to be offensive to Europeans, I am European myself. That is not the point. The point is, who gives a damn about the LICENSE in reality, on their PERSONAL computer? Serious question.
The licence is not there for enforcement from their side. It's a legal protection for Huawei. Essentially "We told you it's not for the EU. If you get sued don't try to put it on us."
Also any company of a serious size will have lawyers interested in licences of everything you're running.
they might license it to companies in the US, but don't want to have to deal with the changes and bureucracy needed to support individuals.
The statement's purpose is to say the equivalent "if you're a European and do run it, it's on you, this is not a product we release or support for the European market, don't expert support, liability, etc".
I'm really torn on the whole thing. I consider myself a patriotic American and would never do anything to undermine the security of my country or its allies (using the same definition of national security that the serious sworn oaths use, "all enemies foreign and domestic", which makes NSA backdoors that compromise American devices squarely a "domestic enemy").
But loyalties don't change facts and China is where serious hackers are rising on merit, doing a lot with limited resourves, giving zero fucks about empty slick talk.
If we wanted to hobble the PRC's technical rise we should have subsidized wasteful NVIDIA use and had Altman/YC be in charge: they'd still be gladhanding about how to pump their portfolio companies sticker price and avoid "systemic shocks" to the stock market anchored on NVDA.
What? I do not find anything confusing. You live in a Marvel world if you think a LICENSE is going to stop people from using a product. But like you said, it is not intended to be for enforcement purposes, but Huawei is trying to save its own ass.
So what is your answer? Mostly companies only? That is a fair answer, but you are the one who said this:
> You'll be both breaking their licence and potentially your local European data laws.
Again, who cares, dude? Companies might, but individuals probably give a rat's ass. So why leave that comment?
And just for the record, if you quote someone, quote them verbatim, otherwise it is not a quote.
For those that would not remember, this was a real thing in the late 80s and 90s relating the cryptography.
There were serious laws limiting the export a "modern" cryptography software from the USA.
Some of us had to face up to the serious challenge of connecting to an FTP server and downloading PGP and risking violating US law to download a software package.
A few years later we had to decide "Do you want the secure Netscape, or the insecure Netscape?".
A lot of companies and research institutes in the EU would like to be able to use a locally hosted LLM for their employees so they don't have to worry what data they give away.
Also it is not rational for any individual to buy the hardware for running a serious LLM and then let it idle 99.9% of the day.
I am not up to date with the models, but I have heard good stories about a couple of open source models. You should ask Simon Willis. I hope he will be summoned (@simonw).
A lot of companies and research institutes in the EU would like to be able to use a locally hosted LLM for their employees so they don't have to worry what data they give away.
They will certainly not violate EU laws and also probably not the licence.
It's plausible deniability. Someone at Huawei presumably thinks there's a chance that exporting this to Europe might be a legal problem at some point in the future. So they added a restriction, enough for plausible deniability.
It's not exactly "plausible deniability" in the common sense of the term.
It's not supposed to make them appear as plausibly denying that some European can download and use this.
It's role is to signal that if someone does, it's on him, not them, and he wont have any support, liability claims, etc as if they could if it was a product intended for their use.
GDPR is not the issue here, the new AI act is. Since this is an open-weight release it is not bound by the training data disclosure rules, but it probably didn't go through the evaluation that is required above a certain number of FLOPs. That's why many recent big player model releases had a staggered release in the EU.
Picture your PC as a cheery little planet in the EU’s cosmic backwater, sipping a digital Pan-Galactic Gargle Blaster. You download Pangu Pro MoE, hit “run,” and expect to chat with an AI wiser than Deep Thought. Instead, you’ve hailed a Vogon Demolition Fleet. Your machine starts moaning like Marvin with a hangover, your screen spews gibberish that could pass for Vogon poetry, and your poor rig might implode faster than Earth making way for a hyperspace bypass.
The fallout? This AI’s sneakier than a two-headed president—it could snitch to its creators quicker than you can say “Don’t Panic.” If they spot your EU coordinates, you’re in for a galactic stink-eye, with your setup potentially bricked or your data hitchhiking to a dodgy server at the edge of the galaxy. Worse, if the code’s got a nasty streak, your PC could end up a smoking crater, reciting bad poetry in binary.
To translate for those not familiar with the writings of Douglas Adams:
nord is suggesting it's possible that the physical computer running this model could be used as a "hub" for potential spyware, or be overloaded with workloads that are not related to the actual task of running the model (and instead may be some form of malware performing other computational tasks). It could potentially perform data exfiltration, or act discriminatorily based on your percieved location (such as if you're located within the EU). At worst, data loss or firmware corruption/infection may be of concern in case of license violation.
I'm not sure I would outright disagree that this as possible, but with some caveats. I would think the reason that the license stipulates that usage within the EU is forbidden due to the EU AI Act (here is a resource to read through it: https://artificialintelligenceact.eu/ai-act-explorer/).
how will the "open weights" know that the pc is running within EU?
again, you are not talking about software that actually runs in your pc but the file that the software reads and loads into memory for its own use.
No it's actually worse. Approximately three seconds after you install the model in offline mode on your computer, a small detector van will come and park outside your door with an antenna on the roof, and relay your position to a Chinese ICBM for immediate targeting.
Sorry, sounds like total bullshit. The weights aren't going to do anything. And if you are worried about the code, with current deployment practices of curl | sudo bash there are much more low-hanging fruits out there. That's not even mentioning the possibility of running the model on a PC without internet access (no matter how good the new Chinese AI is, it's still not good enough yet to convince you to let it out of the box).
Don't give it mcp then (and I struggle to understand why would anyone give a stochastic model such access even if it is trained on very American NSA-certified hardware approved by Sam Altman himself).
The same thing breaking any license does. If you do it in your basement, nothing by definition. If you incorporate it in a service or distribute it as part of a project, well then you're on the hook. (and that is what license holders tend to care about)
I called him out in another thread. It makes absolutely no sense. He is talking against himself, judging by his comments.
To answer your question, he modified my comment (see the parentheses):
"> The point is, who gives a damn about (doing an illegal thing) in reality, on their (private property where nobody is likely to see that)?"
So... at best what he said is purely theoretical. He admitted it himself: "nobody is likely to see that". Though I am not sure I agree with it, but then again, in reality, no one gives a fuck, at least not in Europe.
There are likely multiple potential issues here, but one specific example: Processing and storage of PII without consent/authorisation is not allowed, regardless of whether you do it yourself or for others. And you can't guarantee that this model does not contain private information hoovered up by accident.
There is not a single AI model that fully complies with GDPR. How can you inform everyone, even those not named by actual name but otherwise identifiable, that their data is being processed and give them the ability to object when the data they train on isn’t public.
Literally the same for all other open weights, this is just legal ass covering where most others don’t even do that.
It isn't acknowledging. It is just a legalese to wash their hands away from following whatever EU restrictions and requirements may be applicable here otherwise.
