PostOnce

14 ms·

151.
by PostOnce 2y ago
Looks down to me. https://archive.is/F78tf
152.
by PostOnce 2y ago
I usually just go Lutris -> install windows game from exe -> done & working. Lutris handles the wine config for me, works 99% of the time.
153.
by PostOnce 2y ago
Gerontocracy will remain even if age makes you unelectable, i.e. no one elected Murdoch. Wealth and power are both subject to a sort of reverse entropy, they tend to concentrate, and that means the older someone is the more they are likely
154.
by PostOnce 2y ago
The government is also the reason I don't rob you, competitors don't cut/jam your phoneline at busy times, the reason you can buy from wholesalers and trust their product not to be poison or counterfeit, and you don't ha
155.
by PostOnce 2y ago
I don't think this is the kind of advice a good lawyer would give. BrandShield, Funko, and iwantmyname all caused serious financial harm through, at a minimum, tortious negligence. I'm not a lawyer, but even a yokel like me knows
156.
by PostOnce 2y ago
Not only that, I'm moving a domain to a local registrar so I can get legal redress here if I need to. Not from iwantmyname, never heard of them, but of course now that I have, I couldn't do business with them in light of the situ
157.
by PostOnce 2y ago
Dying in a car accident is normal, but having your (robot) maid kill your children or burn your house down is not. I think that will weigh on people's opinions of domestic general purpose robots when compared to robotic cars.
158.
by PostOnce 2y ago
For factories and closed environments, stuff is getting good fast, but for the rest of the "real world", no robot or AI is practical without human supervision. I automate physical things for a living and have thus become convinced
159.
by PostOnce 2y ago
Software doesn't require a degree to get a job, even an advanced high paying one, but just about all the other STEM fields do.
160.
by PostOnce 2y ago
IA stores lots of redundant stuff in 5 file formats and none of them are particularly well-compressed, I think. There are (big) savings to be had, but maybe figuring that out (software dev and compute time) isn't worth it?
161.
by PostOnce 2y ago
I used to know a guy who would literally roll around on the floor and growl and not be able to communicate when off his meds, but otherwise when on them, lucid, interesting, funny, intelligent, kind, and generally a good contributor to soci
162.
by PostOnce 2y ago
No need to pay someone, you can make one. What do you need? A secondhand manual pantograph mill someone is getting rid of for a few hundred dollars and a little jewellers lathe? And of course a book on watchmaking. We've been making wa
163.
by PostOnce 2y ago
A client worth north of a million dollars to me once said point blank that my competition can't write an email, and that's a big part of the reason we're doing business.
164.
by PostOnce 2y ago
They're having quite a bit of trouble convincing people to work in the office again, nevermind 6 day weeks or 10 hour days. We raised a generation who learned companies have no loyalty and no consideration for their wellbeing, or the w
165.
by PostOnce 2y ago
At some point, someone is going to realize that people are not going to slave away for the privilege of living in a rental and drinking meal replacement shakes because they're too busy to eat. People who worked at Google early on were
166.
by PostOnce 2y ago
Torch was originally a Lua project, hence why pytorch is called pytorch and not just torch. In another timeline AI would have made Lua popular. The best part is it trampled TensorFlow which I personally find obtuse.
167.
by PostOnce 2y ago
Too many trillion dollar companies, not enough million dollar companies, and consumer reluctance to pay for consumer software. We know how to build it, but not how to pay 200k salaries on it. Edit: and developer unwillingness to live on 75k
168.
by PostOnce 2y ago
If you're not "allowed" to watch a movie you paid for on a TV you paid for, something is deeply wrong with the law.
169.
by PostOnce 2y ago
In my experience, the water pistol was still used to denote violence, it was just dressed up to somehow make big tech feel less culpable for violent(ly emotional) social media.
170.
by PostOnce 2y ago
The DoE subsidized the development of GPUs, but so did Bitcoin. But before that, it was video games, like quake. Nvidia wouldn't be viable if not for games. But before that, graphics research was subsidized by the DoD, back when visual
171.
by PostOnce 2y ago
This assumes that "AI content" and "human content" are of equal value, which is still up for debate and I'd argue not true. That means this will devalue AI content and ad impressions, but not necessarily human conte
172.
by PostOnce 2y ago
Suppose you live outside of America and the supermajority of LLM companies are American. You want to ask a question about whisky distillation or abortion or anything else that's legal in your jurisdiction but not in the US, but the LLM
173.
by PostOnce 2y ago
Maybe you want to conduct experiments that the cloud API doesn't allow for. Perhaps you'd like to plug it into a toolchain that runs faster than API calls can be passed over the network? -- eventually your edge hardware is going t
174.
by PostOnce 2y ago
Defense contractors employ a lot of (real, not javascript) engineers with good salaries and then those salaries get taxed again and then they pay for government meteorologists and then those salaries get taxed again and then they pay for co
175.
by PostOnce 2y ago
I explained above that they don't have to care, but if they're going to want to be included on major industry benchmarks like those created by Google, they should probably go to the effort of ignoring the half dozen or so notable
176.
by PostOnce 2y ago
Even so, they can drop from their training set a few dozen pages from the internet and it won't be a huge loss among the trillions of documents. The LLM will still know how LLMs work without having trained on the handful of documents c
177.
by PostOnce 2y ago
The actual valid counterargument is maybe Anthropic does't care about bigbench and nor should they have to, google can make up any benchmark it wants, anthropic does't have to use or care about it, or ignore it, or anything else.
178.
by PostOnce 2y ago
The string is to indicate that it was found alongside the benchmark data, that's the whole point. If that string shows up, then it's extremely likely the test / benchmark data was there too, indicating it's contaminate
179.
by PostOnce 2y ago
No, not "untraining" after the fact, but scrubbing the benchmark data from the dataset prior to training. That's absolutely trivial. Take the set of benchmarks you care about, and then as you build a training dataset (by sc
180.
by PostOnce 2y ago
Either way, shouldn't a company the size of Anthropic be scrubbing benchmarks from their training set regardless of how the benchmark data got into the training set?

More ›