wongarsu

15 ms·

361.
by wongarsu 4mo ago
I know of multiple businesses in Europe that have been doing that for a while with 70B models, and are upgrading hardware to run the new crop of 700B-1T models (really started around Kimi K2, but buying and hosting that kind of hardware tak
362.
by wongarsu 4mo ago
It's also third best overall on "AA-Omniscience Non-Hallucination Rate", far higher than DeepSeek, GPT 5.5 or Fable. That's the one benchmark that allows LLMs to answer "I don't know" and punishes them for
363.
by wongarsu 4mo ago
It does really well on "AA-Omniscience Non-Hallucination Rate", far higher than DeepSeek, GPT 5.5 or Fable. I really like that benchmark because it's one of the few benchmarks that allows LLMs to elect not to answer if they a
364.
by wongarsu 4mo ago
Because they are not static images. That's the whole gimmick
365.
by wongarsu 4mo ago
Peak free information flow was in 2010. When "social" media had its big break through, and nobody had learned how to control it yet. That time gave us the occupy movement, the Arab spring, and lots of hacktivism for the social goo
366.
by wongarsu 4mo ago
Which itself is named after a military term, and has been described in the terms of a military campaign, just for peace. Wikipedia begins the history of the Peace Corps with an article titled "A Proposal for a Total Peace Offensive&quo
367.
by wongarsu 4mo ago
Holy slop. Of the three "root causes" the third isn't a cause at all, the first is a questionable mitigation, and the second is "oops, when we wired up our cool anti-bot product to the signup page we forgot to send it an
368.
by wongarsu 4mo ago
If you want to use a rope to get a straight line, your best bet is to turn the rope itself into the pencil. Coat it in chalk or other powder, then put it under tension and snap it on to the desired surface
369.
by wongarsu 4mo ago
> Curiously, fusing a model with itself also boosted performance Back in the GPT2 to GPT3 era this was a pretty common thing to do. You are effectively taking more samples from the space of likely outputs. If your model can do the task 6
370.
by wongarsu 5mo ago
A full sized European electric kettle is about 2000 watts, but if you limit capacity to a single cup you can get acceptable performance on 200 watt. A USB-C coffee pot or kettle scaled to 0.2 liters (7oz) could work. Would be a great option
371.
by wongarsu 5mo ago
A lot of SFF PCs already come with a power brick. Just with a 12V barrel jack instead of a USB-C port. Compared to that design you really wouldn't safe much space. Though I admit that USB-C would be convenient. Maybe with a tiny batter
372.
by wongarsu 5mo ago
Here in Germany, founding a company creates a public record. There are a number of companies who then send all newly formed companies an invoice that looks like a legitimate invoice for expenses related to creating said company, but on clos
373.
by wongarsu 5mo ago
That won't create justice, only judges that forever encode current prejudices
374.
by wongarsu 5mo ago
Until you get to the second order effects. If we want to find a cure for some disease, but we don't have enough people to do experiments on (maybe a lot of eligable prisoners died in previous trials) judges now have an incentive to han
375.
by wongarsu 5mo ago
But this knowledge is readily accessible today. At least for manhattan-project level bombs. For later developments you mostly get simplified overviews with important details left out. But even there you have communities speculating about th
376.
by wongarsu 5mo ago
Building frontier models to do safety research on them is what Anthropic was all about in the early years. That included building the best model, but only releasing it once it became the second best. Precisely to avoid an AI arms race where
377.
by wongarsu 5mo ago
That's why I added "Not zero, and trust towards AI companies is limited". Reaching the decision that adding one single US-based LLM provider had more benefits than risks took months. And we were selective about who that would
378.
by wongarsu 5mo ago
In other words: it rusts, but it doesn't rust like iron. It rusts in a much less destructive way because the aluminum oxide protects the rest of the aluminum from oxygen
379.
by wongarsu 5mo ago
ICE does a lot of objectionable things no matter what you think about deporting illegals
380.
by wongarsu 5mo ago
As a private citizen, yes. But at work the calculus is entirely different. There is already lots of exposure to US companies (guess where our emails and tickets life), so the increase in espionage risk from adding another American company i
381.
by wongarsu 5mo ago
As a whatsapp user, I am quite happy that marketing messages are expensive. Puts a price on spam, which means I get less of it. If they want to send me something actually useful that usually falls into one of the other much cheaper categori
382.
by wongarsu 5mo ago
Europe is in a weird middle-ground where WhatsApp is still dominant for person-to-person messaging. Not only was SMS costly back when WhatsApp emerged, users are also split 50/50 between Apple and Android, and SMS between them wasn&#x2
383.
by wongarsu 5mo ago
Plugin architectures are a niche where WASM really shines. Before WASM most plugins were either high performance (loading dynamic libraries) or sandboxed and safe for untrusted plugins (LUA etc). WASM allows you to have your cake and eat it
384.
by wongarsu 5mo ago
You still need a competent developer for the prompting, planning, etc. But once it's running, I want to avoid mental context switches and just have it run Giving it access to a cheap human who is just there to take screenshots, do QA,
385.
by wongarsu 5mo ago
To be fair, that is described as an 8 year old example. Their current UX is much more clear. Their current pattern is more about playing into the fear of what happens without insurance, without selecting your seat, when you don't pay f
386.
by wongarsu 5mo ago
Last time I booked I was actually impressed with how well Ryanair was explaining everything. There is not much to clarify. You can see the same in the video in the post (which also features much clearer UX than the 8 year old example of the
387.
by wongarsu 5mo ago
GP's (or rather GGGP's) statement was "I did not write that code but I paid money". So they claim no authorship. Anthropic is a company, they can't be the author either. So the only one left as author in that reason
388.
by wongarsu 5mo ago
Because you are human, and can thus create copyrighted works, and can assign that copyright when you do work for hire. Monkeys and AIs can't create copyrighted works, so there is nothing to assign when doing it for hire The other line
389.
by wongarsu 5mo ago
"I paid money to someone to write that code" is exactly the line of argument that would lead to no copyright. There was the famous monkey selfie court case in 2015 that ruled that at least in the US, monkeys can't hold copyri
390.
by wongarsu 5mo ago
No, the fire bombing of Japan did far more damage than the nuclear bombs. What made early nuclear bombs scary is that instead of needing hundreds of bombers you now only needed one for the same amount of damage

More ›