They run one step/iteration on an additional chunk of training data, then use the snapshot of the weights after that iteration in a separate validation benchmark while continuing to train on another chunk of data for the next iteration.
They result of the benchmark does not feed back into the training, it simply serves to provide a measurement of progression over time.
Google still tries to sabotage Firefox experience on Google Apps, especially Meet but also YouTube and sometimes Google Sheets/Docs as well.
They want to degrade the experience just enough to get you fed up with FF and go (back) to Chrome, without making it absolutely obvious that's what they're doing.
They often label these things as "regressions" because they don't test on Firefox and take forever to fix them. But I can't help but believe it's intentional.
Yes, exactly. Google products working worse on Firefox is a problem with Google, not Firefox. I'm not sure why the conflict of interest isn't more obvious.
The farm also runs the website that gets paid for ads. Google notices that the site triggers a conversion, so they will get the same ad again… It might be more involved but this is a large part of it.
Similar to countries putting democratic in their name being the least democratic, like the Deutsche Demokratische Republik and Democratic Peoples Republic of Korea.
I would too. Google Ads are the absolute quickest way to lose your shirt. You need masses of experience (or a straight scam) to run a profitable ad campaign. Their agents will hound you to the ends of the Earth the moment you start reducing your spend.
This resonates with me. The past few months, frontier AI labs were rushing to announce "no, our AI bot broke out of its sandbox and committed crime first and harder than yours".
I said to so friends back then: if I posted about how I ran GLM 5.2 or whatever I was running then in some shoddily built sandbox and it escaped and accidentally hacked some US company, I'd probably already have been extradited to the US and be awaiting sentencing. But when a hyperscaler does it, they just get inflated stock prices.
Rookie mistake, you should have created a company and gotten people to do it. Put everything under the company. And then the one in trouble is the company not you.
Ethics aside, you can create an NGO with three people and do a lot of stuff in it's name that you can't do as a private individual or it would be a lot more expensive to do.
If everyone did this, it would stop being a special ability of some people.
I think the "intent" ( pun not intended) for "intent" to be a moral stance was to not discourage people from trying to do good things for fear they may do something bad and get in trouble or whatever.
In an ideal world this would be good but then of course it's ripe for manipulation because it's hard to prove one's intent
Yes, that's how the trick works. Profit is revenue minus expenses. By increasing expenses, you decrease profit, and therefore tax. But it only works if you had a profit before, because the government doesn't pay you back tax when you lose money.
As a company you collect VAT from your customers and pay VAT to your suppliers.
Depending on the resulting balance, you will either give back the amount of surplus collected VAT to your country/state/whatever or you will be provided with VAT credit.
VAT collection is different from a profit tax or other taxes.
Apparently the only exception is, if your website is only used for piracy and you can see it and you do nothing about it when asked to. That's why mega, megaupload's successor, has encryption, so that the company can plausibly deny involvement with piracy, because they can't see anything on their servers.
"As a programmer, Swartz helped develop the web feed format RSS; the technical architecture for Creative Commons, an organization dedicated to creating copyright licenses; and the Python website framework web.py. Swartz helped define the syntax of the lightweight markup language format Markdown, and was a co-owner of the social news aggregation website Reddit until it was fully sold in 2006 and contributed to its development until he left the company in 2007."
the most scary part is knowing this is the fear we all have, and the one they want us to have.
I've heard he was more like the Elon Musk of the open source world. He didn't co-found Reddit, he was placed there by Y Combinator staff to the protests of its actual founders. Maybe some of the other projects he's credited for also were like that.
I've never heard anyone in the internet freedom or open source communities speak ill of him, and something tells me a 14 year old was not getting involved in writing the RSS spec to steal clout for financial gain.
I think they forgot about that, thank god. All the innocent citizens we sent there are still being tortured daily, but that somehow lost their interest. God forbid they remember before the midterms and ruin more lives for a stunt…
That said I just recently saw a story of a Latin American man sent to the Central African Republic, which has gotta be one of the most absurd & dangerous places to send someone. Less opportunities for them to stage photo shoots with that strategy, tho.
They're literally even deporting Afghans who supported the US, who are in danger of being tortured and killed by the Taliban if returned to Afghanistan, to the Central African Republic:
Brian José Morales García: A 25-year-old U.S. citizen born in Denver, Colorado, was wrongfully deported to Mexico following a traffic stop in Texas, despite possessing a birth certificate and other identifying documents. His legal team filed a case in federal court to secure his return.
Individuals Detained and Deported
Lorenzo Palma
Roberto Dominquez
Andres Robles
Esteban Tiznado
Mark Lyttle
Jhon Ocampo
So - if the argument hinges on an insistance of 'El Salvador' as a destination of malicious intent, probably no. But if being illegally deported while being a US citizen and put through strenuous and harmful conditions in the process fits the request; there are several who have first hand experience with being brown and American in the extreme.
I was not making an argument, just asking for any factual basis to the parent comment. I guess that is deserving of 5 downvotes.
> I think they forgot about that, thank god. All the innocent citizens we sent there are still being tortured daily, but that somehow lost their interest. God forbid they remember before the midterms and ruin more lives for a stunt…
But thank you for providing other cases, I had heard of the first but not the others you mentioned. From reading your sources, these appear to be collated from the last 2 decades.
The last example was an article published in 2021, and does cover about 70 possible cases during the 2015 to 2020 period. It would seem that this misbehaviour is neither new, nor restricted to big T's residency in the WH periods. Apparently we have had this happening to a lessor degree for the duration of the existence of ICE, and this may be because of a shift in oversight, training, and general legal competency. It does appear however that under present and past stints the frequency and severity of this problem are demonstrably worse.
Having myself grown up actually within sight of the US|Mexican border - I could see it in the distance from the driveway and kitchen window of one of my childhood homes, and from the schoolbus every weekday when i lived at an different home before that - I am perhaps a bit more sensitive to this issue, having seen the differences in enforcement agents attitudes and behaviour in the years since the peak of both the Guatemalan and Nicaraguan conflicts. Although I no longer live physically close to that space; my father, my extended family and childhood friends still do, and so many of them are at risk daily of something disastrous happening because they appear less white than the color palette that ICE appears to use as it's operating manual.
The issue with the "i'm just asking questions" is that it is overused by disingenuous (i started writing nefarious but i think you can be nefarious but mean good, whereas people who do this _always_ mean to trick other humans) people. So nowadays, if you want to go ask a question to go further on a topic because you're interested in it, you need to add a preamble that explain. See:
"How did ancient egyptian managed to cut rock that cleanly without modern tools?"
vs
"I'm very interested in old tools and techniques, do any of you have resources about how Egyptian used to cut granite?"
Sadly, this is the state of the internet, because some people can't be genuine (probably think it's a weakness or something dumb), now everybody assume the worst of everybody else.
I hear what you're saying, but I don't think it really applies to a question asking for the veracity of a previously stated claim, especially when the claim is actually false. I think there is more than enough bad things the government actually does wrt immigration enforcement, we don't need to start making things up.
In that case, it vastly more productive to say : 'I do believe this is false/poorly presented/a gross generalisation, and would like you to prove me wrong by citing sources'. At least you do not hide behind the "I'm just asking questions", which read as disingenuous like I said, mostly because it's overused rethoric.
a cursory view of whataboutism's... i mean whimsicalism's comment stream has the effect of negating any declarations of serious intent. This is standard practice, take issue with a valid if overstated perspective on a technicality, insist that the current state of the market and technology is acting as it should, and that continued disregard for the abuses/negative outcomes/deteriorating conditions will be solved by further application of the current techniques.
The downvotes are perhaps the intended outcome. I might say that I didn't know this when i first responded, but that would be untrue.
I engaged in exactly that manner, fully informed of the observable tendencies for engagement. There's no need to try and advise an account as if this 'just asking questions' was a misstep. There is ample substance showing that this account repeatedly declines to take the HN recommended approaches to polite interaction.
i think this is an isolated demand for rigor, my way of approaching it was fine, and it is only due to the political valence of the comment i'm replying to that i got downvoted and flagged. take care
> But when a hyperscaler does it, they just get inflated stock prices
I also keep coming back to this, and I find it harder and harder a fact to live with. It's not a conspiracy theory, we're not crazy for pointing it out, and worse, there are people here I read about constantly what-abouting and number gaming and "well what would you have done?"ing like somehow the destruction of the concept of equal justice isn't just happening, but happening inside the bloody industry that feeds them, clothes them, gives them bonuses and retirements and the ability to even think a future, going off the grid, "retiring early from a smart exit" whatever.
I am finding this community is sadly making me hate my industry specialization more and more and it's because of this.
My magic wand would be waving it at those humans to force them to suddenly and powerfully experience their version of an overview effect in their lives so they would simply STOP working for FAANGYWANGY companies and just say, honestly and humbly and painfully: "I am contributing to the downfall of society by complicity feeding a malicious herd of whales."
I gave up a number of big jobs and stocks and money because the people I met were wretched. I am a massively imperfect person, but by all that I can, I refuse to build the torment nexus.
I have met too many people that want to work for the company that does, and those people put these hyperscalers, as you call them, into positions where they can influence an entire planet of humans on a whim.
I literally dream of a world sometimes where Elon wakes up and finds out no one takes his call, no one reads his tweets, no one even hands him a cup of coffee at a shop. I imagine this for a lot of people in the world.
That sounds like a world trying to rid itself of parasites and evolve toward a brighter future. I want to be in it.
> My magic wand would be waving it at those humans to force them to suddenly and powerfully experience their version of an overview effect in their lives [...]
I've daydreamed about this too. I want to sit them down in an interview and try and dissect their thought process/morals. I doubt it's possible to succeed.
Thanks for sharing this. I'll be completely honest about my oddities and eccentricities that this has brought me to a place of trying to understand linguistics in a way I am simply not learned enough to do.
I have a barely substantiated hypothesis that a lot of what fails us as humans in communication is language, intent, and the inability to visualize the same thing with the same words. I have things I've tried to build around this, and the science has been around for many decades and beyond, but I have to try something. I can lie to myself in this and say that at least I'm trying to help people understand each other.
But linguistics is hard and my science is not strong and I am one person trying to convince my local peer groups my ideas about language isn't just goofy and worth a pat on the head. That's less a rant and more a point that I think we're on the same page in some ways: being able to dissect someone's thoughts and morals starts with knowing what tools to do it with, and the only tools I'm really good with are software and language. So, that's what I'm trying to do.
"It was a big step up on my career and put me in a better position to provide for my family, which is my most important moral duty" is what every one of them will say.
HuggingFace had no interest in squeezing a few million dollars out of OpenAI in the midst of being acquired by Nvidia.
The optics are bad for HF as well because they gladly host abliterated models with safety training removed. When Astra/Fable level open weights models arrive, HF will soon be the cause of far worse incidents, and they know it.
You missed a critical detail. They couldn't defend themselves with closed source American models so they had to investigate with Chinese models that had less restrictions. Great optics for hosting open source models!
> But when a hyperscaler does it, they just get inflated stock prices.
This reminds me of the early days of AOL when they received so much traffic that their (dial-up) lines became inaccessible. In a rational world, their stock would have suffered because their service degraded, but instead their stock rose based on the optics of increased demand.
If you make a serious attempt to look like you're engaged in legitimate business, people will (shockingly) be somewhat more hesitant to think you're committing a crime than if you are just openly committing crime.
FWIW I don't think this is true. Intention matter a lot in US law. If you could prove the AI did something bad entirely on its own and you had nothing to do with it and had no reasonable belief it was possible, I think you'd be alright.
People are not charged with hacking when their unsecured box is taken over by a botnet and does bad stuff.
The broad concept has been well-known for half a century at least, and the very specific concept for several years at least. It’s clear that they didn’t take reasonable precautions against it. And it’s not even the first time this sort of thing has happened, though it’s higher-profile and -impact than before.
And yes this is absolutely the first time autonomous software has breached containment. What are you referring to? Perhaps just corporate cyber in general?
The concept of AI breaking out of a box or doing something unintended trying to achieve a goal. OpenAI should be found guilty of gross negligence of some sort, because this was not something that could not have been anticipated, and because their security precautions were laughable.
Because the AI itself is not a legal person, it must be OpenAI that's responsible for what it does.
Maybe (and historically, not with CFAA) - but they're still going to extradite you to a trial here to find out which is more than enough to ruin your life. The government is already looking for a great excuse to criminalize open source models.
Intent is literally a requirement for conviction under CFAA. It's the first line of the law: "Whoever having knowingly accessed a computer without authorization or exceeding authorized access..."
Robert Morris very intentionally wrote a computer virus, released it into the wild to infect computers he didn't have permission to use, and tried to cover his tracks by making it look like it came from MIT instead of Cornell. The only accidental part was that the virus was too successful, and that he got caught.
Lori Drew intentionally violated MySpace's TOS (in the course of cyberbullying a child until she killed herself), and was acquitted because TOS violations don't rise to the level of a crime.
Because if you mean the Government, I think they frankly just care whoever is paying them to make an executive order or worse a law paid by corporations to not only block, but ideally also ANHILIATE ANY POTENTIAL COMPETITION.
It's the ultimatum against the small guys. Sold as "Anti-China", when in reality it's only purpose is total dependency to corporations by law.
The support of OSS models on HN just days after finding out about the message board incident is so bizarre to me.
Do you guys really not comprehend the impact of giving every script kiddie an Astra-equivalent model to play with, this time without any guardrails whatsoever?
Cause I'm starting to think you aren't really thinking through the impact of power plants, hospitals, etc all being hacked en masse. People will die.
Or making it easy for anyone to make a virus (it's not hard, mechanically). Even more people will die.
Taking this to the extreme: You don't just give every kid a "do anything" super intelligent button. Open source models have a limited lifespan whether you like it or not, for the safety of all of us as a whole.
"Open source models have a limited lifespan whether you like it or not". this tinpot dictator mindset is typical of anthropic.
Who will decide how to use the power of ai...you?
it is no surprise that the companies and safety institutes who believe that they alone may 'tame' the fire, are the cause of safety incidents. a safety test caused the explosion at chernobyl.
anthropic and openai are the danger to society.
they are the ones making the dangerous models. they are the ones using tens of millions USD of inference, and thousands of agents, to hack computer systems.
i have trained zero dangerous models. i own zero servers that i use to run hacking agents. i could run one hacking agent with my single subscription.
there are some arrogant researchers remaining at anthropic who believe that they do care about safety. those who cared about safety at openai have left.
depending on how it goes, we might need some kind of earlier intervention by the US government to take control away from the current leadership of the ai companies and prevent further incidents.
Yes, I'm aware that the HN majority is a hyper-libertarian echo chamber that loves to outsource the problem of "technology hurting regular people" if it means they get to keep their toys/investments.
After all, who cares if abliterated extremely intelligent OSS models result in actual innocent people dying? That's <insert group here> problem. We need to be able to generate our uncensored furry fanfics, goddamnit!
State actors, professional criminals will have same or better capabilities and do not hesitate to use them. On defense end we'll have only "pay bigcorp" option.
OSS models can help to fix infra especially for folks who would never do it themselves. We do not know yet what final effect would be and it doesn't seem sky is falling now or in near future.
"people dying and furry fanfics" is a scarecrow and distractor respectively, it would be awesome if more specific examples would be used in place.
> State actors, professional criminals will have same or better capabilities
State actors do but they are not unhinged enough to use this to cause massive loss of life, unlike random crazy people with access to a computer.
Professional criminals don't have access. Please explain exactly how you expect a professional criminal to get access to a non-safety-tuned Astra/Fable equivalent.
> OSS models can help to fix infra
You can work with the vendors of critical software to fix infra first, without giving every random crazy person access to something that creates cyber/bioweapons.
> "people dying and furry fanfics" is a scarecrow
You don't think people will die if OSS models make it easy to create a bioweapon, or make it easy to hack hospitals, infrastructure, and the like? Please explain your thinking.
It's very easy to order all of the raw material needed to create a virus, the challenge is in making a viable one. But models make this easy.
Same for cyber. Hospitals, emergency services, and infrastructure like power plants are not sufficiently hardened to withstand an attack from Fable-class models.
Criminals don't need Astra or Fable. Get a bunch of GPUs, a good enough open weight model and a custom harness. What the model doesn't have in its weights it can research the Internet. How do you think black hat hacking is being done right now?
On bio: while it's easy to order all the raw material, there's the whole process of culturing bacteria/viruses/etc. that isn't trivial, and while a LLM might help in explaining stuff for rookies, there needs to be somebody physically working on it. It's not automatic. Once you get to this level, you'll find that any undergrad biologist or chemist can already make bio/chem weapons, and you don't see it happening. Terrorist groups already employ biologists or chemists that are supporters of their cause, no need for LLMs.
On cyber: aside of the question of why critical infrastructure is on the Internet on the first place (ah yes, lowest bidders, people not caring enough, business not giving IT/OT budget, S in IoT standing for security etc), if the available models can't handle cyber tasks, how can you defend yourself? See the HF "attack" that HF had to use GLM-5.2 to investigate what happened. This is the best argument for open models. Unless of course, you are the AI model creator or somebody they authorized to use their sanctioned model/harness and want to create a moat for their business.
Meaningless phrase used to dismiss arguments, please engage with the actual argument.
I don't think you understand how bad it will be if anyone has a button that can hack anything, our infrastructure is not ready for this. Actual people will die. Do you just not get this?
Bioweapons as well. Do you not understand how easy it is to craft a virus at home? You can literally order everything you need online. Again, actual people will die.
You are arguing the equivalent of letting everyone own missile launchers or RPGs because "open source good."
If and I mean IF it will be a (A) "hack anything" button, it as well will be a (B) "find a vuln in my site" button too. And a (C) "harden my site" too.
And limiting such buttons to a few corporations does not make any sense because script-kiddies will still have access to (A).
Did you know (C) is possible without causing mass chaos that results in many people dying? In ways that don't give every script kiddie access to (A, B)?
It turns out you can give defenders the opportunity to front-run, at least on the most critical and widely used infrastructure.
No, that's an illusion, that's the thing. You can't physically have (C) without the (A). Large corporations will spend billions trying to convince the population it is somehow possible, but the reality is different.
You absolutely can have (C) without widespread (A), you just need to limit the audience that gets the tools, and in fact the big labs are already engaging in (C) in collaboration with the government and vendors of critical software.
That you don't have access to (C) to harden your blog is a non-issue from a societal perspective, hardening the software that our infrastructure relies on first is simply more important.
I very much doubt China is going to let their citizens freely use AI for hacking. In fact they keep a very tight leash on corporations and their employees.
So long as real logistics systems are isolated and secured what real risk to stable society is there?
Your use of the internet may lead you to greatly overestimate the number of people who take what they read online as real. More people than you think treat this shit, regardless of the website/forum, like Jerry Springer and Satuday Night Live; just entertainment.
The STEM crowd often overthinks the impact of violating their pet theory.
Can an LLM escape its computer entirely and stomp through a city center like a kaiju? Settle down.
You're veering towards thought policing over speculation.
Nothing unites the public behind restrictions like a good dose of fear. The "for the children" angle is a master key for regulation—always sold as protection from the big bad wolf.
"Trust us, we're here to protect you." Sure we are. /s
It's only a matter of time before open-source gets banned in the US under that same paternalistic security blanket. /s
Not remotely true. Mens rea is a necessary precondition to establish criminal culpability for tons of crimes. Well before sentencing is ever considered. The far opposite, ‘strict liability’, where you’re guilty of a crime purely due to some action or inaction (the actus reus) is exceedingly rare in the US justice system.
Baloney, lots of people go to jail for DUIs, and that's the right analogy here.
People don't drink and drive with the intention to kill people. They drink because it's fun and then get behind the wheel because it's easy and convenient, even though they know the dangers they convince themselves nothing that bad will happen.
AI companies are creating these dangerous, powerful models (that they keep telling us are dangerous and powerful), then they take off all the safety guards to run them in woefully inadequate "sandboxes". Pure negligence.
It's certainly analogous to the behavior exhibited by these AI companies in deliberately performing dangerous actions and then letting other innocent people deal with the consequences.
Virtually all criminal law accounts for the perpetrator's state of mind. Drunk driving is a specific, rare carveout. The reason for this should be obvious: it is nearly impossible to prove a drunk person's state of mind beyond a reasonable doubt, so we passed laws so you can't say "Your Honor, I was too drunk to be responsible for my drunk driving".
So, no, this is not at all analogous to a totally routine question of whether someone was negligent in how they deployed some software.
There is a lot of good analysis of the Appeals process on this specific point - the CFAA as written required unauthorized access:
Section 1030(a)(5)(A), covers anyone who
(5) intentionally accesses a Federal interest computer without authorization, and by means of one or more instances of such conduct alters, damages, or destroys information in any such Federal interest computer, or prevents authorized use of any such computer or information, and thereby
(A) causes loss to one or more others of a value aggregating $1,000 or more during any one year period; ... [emphasis added].
The District Court concluded that the intent requirement applied only to the accessing and not to the resulting damage. Judge Munson found recourse to legislative history unnecessary because he considered the statute clear and unambiguous. However, the Court observed that the legislative history supported its reading of section 1030(a)(5)(A).
I'm not sure how you could categorize the Morris Worm as lacking mens rea based on that statute..
You think that Morris accidentally wrote the malware or that he accidentally released it (in a way specifically intended to obfuscate his connection to it)? lol
> People are not charged with hacking when their unsecured box is taken over by a botnet and does bad stuff.
What the Frontier labs have done is not this, however. Also, they know damn well what can happen and they still don't take the appropriate precautions. At this point it's very hard to believe it's not intentional for purposes of marketing.
Well that and you spend tens of thousands of dollars on lawyers to reinforce that point and make the prosecution rethink whether they will profit off of this case or not.
US law doesn't really give much of a crap about your intentions unless you can back it up with a wall of money to exclude yourself from the rules of the general population.
ie, I intended to steal money from you. Under common law stealing is "taking with intent to deprive". How thats determined is a bit harder.
But how can someone be "grossly negligent" if they didn't intend to cause an accident? well they either allowed a situation to happen, or didn't stop a situation happening that they could see was bad.
An example of this would be the igintion switch from GM that caused all those deaths. the engineer saw that it was shit, knew it didn't do what it was supposed to do, and half arsed the replacement to the point where it wasn't actually changed.
"We are going to test the cyber capabilities of this new model. Yeah it should be fine to have a package proxy. Hmm? whats that? isn't that a security issue? naaaa those proxies are secure."
Intention (mens rea) is not a get out of jail free card, it just changes the nature of the crime and charges you get convicted off.
If you run someone over on purpose you get convicted of murder. If You do it by driving recklessly you get convicted of culpable homicide.
The only time you get off with nothing is if it is determined to be an accident.
As they always say: ignorance of the law is no excuse. Running a dangerous machine prevents you from claiming you had no idea the machine was dangerous.
But US law also has a concept of negligence. If you set up an AI to do this and didn't take reasonable steps to prevent it you could still be on the hook.
They hacked a competition company. The intention was implicit when there is economic gain to be had.
> People are not charged with hacking when their unsecured box is taken over by a botnet and does bad stuff.
Because it is a lose for that people. If the botnet left money in their pockets the situation would be totally different as there will be an incentive for them to let the botnet hack them.
> The intention was implicit when there is economic gain to be had.
Not how it works. Intent requires at least that you were deliberately doing the criminal act (sometimes also that you knew it was criminal, or at least that you had some reason to believe that it was wrong).
> Intent requires at least that you were deliberately doing the criminal act
So, they did something that benefits them by mistake. I would like to see a person would be judged in that situation. I can imagine that the bar to put someone in prison is way lower than to give a fine to a mega-corporation.
You can imagine what you like. I can imagine a million examples of people being judged not guilty because intent could not be proven beyond a reasonable doubt. There are of course many injustices in the 'justice' system, but I would prefer to discuss the concrete details instead of just vibes.
Here's one case: https://supreme.justia.com/cases/federal/us/342/246/ where someone who took some metal they believed to be abandoned was aquitted on that basis, though this particular case made it to the supreme court because the lower courts tried to rule that the fact that he didn't intend to steal them didn't matter.
Also, I don't think there's a strong argument that hacking huggingface did benefit them. If you could prove it was a deliberate ploy to market their models, then perhaps you could argue they expected to benefit. Otherwise, you could argue that it's negatively affected their reputation.
If there's evidence they deliberately did this, yes. If they've lied in other cases then it might make their testimony that they didn't intend this less credible, but also OpenAI as a whole is not going to face criminal charges, so it'll also depend on the situation of whatever individuals do get charged, if it does go to court.
I promise you, they are not publishing these stories for marketing. They listen to the scientists, and know what’s coming. They’re just desperately trying to warn us…
I'm guessing they acquired it mostly exchanging stocks. Which I guess is an indication that their stock is overvalued right now if they're willing to overpay by that much.
I had a Strix Halo laptop with 128GB which unfortunately died last week. I paid 2800 euro for it. If I buy the same machine today, the sticker price is 7899.
The device was not perfect by any means, but the ability to run fairly large models is some kind of magic.
You can get a used enterprise grade SXM baseboard with 4-8 V100/A100 GPUs off eBay at a similar price. That will even get you actual HMB ram and NVlink. Along with 10x the AI performance, assuming you don't care about your electricity bill of course.
You can get a new M5 Max MacBook pro with 128 GB unified ram (targeted by Antirez for DwarfStar4) even after the Apple price increases, it's less than 7899 by at least $1000. And you probably won't pull more than 100 Watts.
You are comparing US pricing with EU pricing. EU pricing includes 21% VAT and currency conversion "rounding up".
The cheapest 128GB Macbook Pro here costs €7.949,00.
No doubt a better value than the HP, and will depreciate a lot less quickly, but just as expensive. Unfortunately, not being able to run Linux is a breaking point for me.
I have one of these. Got it a few weeks before the price increases. On the 14" version charging is limited to 96 watts, but the chip can pull north of that with adequate cooling, so the battery will literally drain while plugged in.
It isn't a problem for me, more amusing than anything else (I run in Low Power mode 90% of the time) but worth knowing for anyone thats thinking about pushing the hardware to its limit 24/7.
I've played with ds4 on such a machine, and the battery drains when using only a 96W power brick. But you can put it in low power mode and the fans won't even turn on while it delivers something like a third of the performance.
ignoring the fact one would need a bit of a different setup (chassis, PSU) to run it, I casually looked and there's nothing below $25-50k euros for such a board decked out, depending on the config. TBH even that doesn't sound bad, but I wouldn't even know where to start how to run it.
So that’s where all the used V100s are going. I’ve been thinking about making a water cooled version of those because I don’t have a rack to put these servers on.
Yeah it isn't worth it, but comparing a server with a laptop is also not a relevant comparison.
I didn't get a Strix Halo laptop because it was the best bang per buck, I got it because it was an awesome machine that could do a little bit of everything, fit in a backpack and only needed 140W.
But noone should buy one at 7899, obviously. It was a tough sell for me at the old 2800 pricing.
The connection does not look standard so probably need to find some kind of OEM interposer card, which would allow you to connect some slimSAS cables that in turn connect to pcie adapter cards.
You'd need at least threadripper cpu/motherboard to handle that many pcie slots/lanes.
Not to mention that a baseboard like that is not going to work with a standard ATX power supply, so you'd have to provide your own power solution.
If that doesn't sound fun to you, you're probably better off buying Claude tokens.
first off rude, secondly, the issue is the existence of a mcio to whatever mess that thing is. i assume it is sxm5 or something and you probably need a pci-e 6/5 switch in the middle
everything strix halo went 2-3x bananas, same ballpark figures as apple hardware now and lead times on all of those are in months. Ridiculous where we ended up at.
Yes but it depends what you want. I didn't want to spend 8k (or now 11k) for a video card that I couldn't run without turning my home office into a boiler room and pay through the nose in electricity.
I wanted a laptop that could run some medium sized LLMs locally and experiment with. Strix Halo is great for that. But it was also 2800 euros for a 128GB premium laptop. Not 7899 like it is now.
They result of the benchmark does not feed back into the training, it simply serves to provide a measurement of progression over time.
reply