![](/static/66c60d9f/assets/icons/icon-96x96.png)
![](https://lemmy.world/pictrs/image/c47230a8-134c-4dc9-89e8-75c6ea875d36.png)
There’s this map: https://liveuamap.com/
I haven’t been following the war very closely, but AFAIK, they’ve been in a stalemate for many months with neither side making any significant gains.
There’s this map: https://liveuamap.com/
I haven’t been following the war very closely, but AFAIK, they’ve been in a stalemate for many months with neither side making any significant gains.
We’re close to peak using current NN architectures and methods. All this started with the discovery of transformer architecture in 2017. Advances in architecture and methods have been fairly small and incremental since then. The advancements in performance has mostly just been throwing more data and compute at the models, and diminishing returns have been observed. GPT-3 costed something like $15 million to train. GPT-4 is a little better and costed something like $100 million to train. If the next model costs $1 billion to train, it will likely be a little better.
Rentals seem extremely expensive in my area. $100/day for a shitty 4" wood chipper, $300/day for 6" chipper. For some tools, it’s often about the same price or cheaper to buy a tool from Harbor Freight than to rent.
Libraries don’t make a profit, AFAIK. Non-profits and co-ops are things too.
I’ve been in cells without toilets or water. They’re designed for temporary holding (an hour or so), for things like waiting on transportation or intake, for example.
Server side rendering looks like it could be useful. I imagine SSR could be used for graceful degradation, so what would normally be a single page application could work without Javascript. Though, I’ve never tried SSR, and nobody seems to care about graceful degradation anymore.
Nah, the politicians and the populace would gain nothing by seceding. The Republicans are trying to push extreme “state’s rights” on all fronts. The goal of this particular fiasco is likely to get a favorable outcome from SCOTUS and to prevent Texas from getting more blue/brown from immigrants’ children. The goal of extreme “state’s rights” is to ensure Republican control over the federal government (there are more red states than blue states, and if states have enough rights to do things like overriding election results, voter suppression, extreme gerrymandering, and other various ways of “rigging” elections, Republicans can ensure control over both state and federal government).
Are the people who join these groups typically struggling workers? The Jan 6 mob seemed to be mostly relatively well-off small business owners and the self-employed. Who else has time to do all this stupid shit? And all that money to waste on guns and LARPy gear?
Lack of diversity, the higher degree of isolation in rural/suburban life, and higher amount of churchgoers getting preached hate probably has something to do with it too.
Quality. Yeah, using the extra compute to increase speed of development iterations would be a benefit. They could train a bunch of models in parallel and either pick the best model to use or use them all as an ensemble or something.
My guess is that the main reason for all the GPUs is they’re going to offer hosting and training infrastructure for everyone. That would align with the strategy of releasing models as “open” then trying to entice people into their cloud ecosystem. Or, maybe they really are trying to achieve AGI as they state in the article. I don’t really know of any ML architectures that would allow for AGI though (besides the theoretical, incomputable AIXI).
The equivalent of 600k H100s seems pretty extreme though. IDK how many OpenAI has access to, but it’s estimated they “only” used 25k to train GPT4. OpenAI has, in the past, claimed the diminishing returns on just scaling their model past GPT4s size probably isn’t worth it. So, maybe Meta is planning on experimenting with new ANN architectures, or planning on mass deployment of models?
In the U.S., the difference between average and median income is ~$25k/yr, so, if my logic is correct, it should theoretically be possible to have an UBI of $25k/yr (which would bring the average income on top of UBI down to around the median).
Wouldn’t it just be because it’s divided among all their great grand children and spouses? If everyone had 4 children, the wealth would be divided 4^3=64 times. So, $1 million becomes $15k (assuming none is spent by the first 2 generations).
Dunno your country’s specifics, but in the U.S., the eventual social security deficits could be completely resolved by removing the caps on social security contributions.
UBI could be payed for by a radically progressive tax structure similar to the U.S. tax structure in the 1950s.
Goes both ways. As someone who has tried bringing new products to market, it’s extremely annoying that nearly everything you can think of already has similar patent. I’ve also reverse engineered a few things (circuits and disassembled code), as a little guy, working for a small business . I don’t think people usually scan patents to learn things, and reverse engineering usually isn’t too hard.
If I were a capitalist, I’d argue that if a big business “steals” an idea, and implements it more effectively and efficiently than the small business, then the small business should probably fail.
I’ve checked it out, but was never a twitter user anyways, so I’m not the right audience. I think the lack of an algorithm really hurts the experience for many users. Seems like you have to put work in to get your feed right. Also, the network effect Twitter has is strong. Mastodon is probably the least toxic social media I’ve seen though.
Meh, patents are monopolies over ideas, do much more harm than good, and help big business much more than they help the little guy. Being able to own an idea seems crazy to me.
I marginally support copyright laws, just because they provide a legal framework to enforce copyleft licenses. Though, I think copyright is abused too much on places like YouTube. In regards to training generative AI, the goal is not to copy works, and that would make the model’s less useful. It’s very much fair use.
Trademarks are generally good, but sometimes abused as well.
He will likely be allowed to run because so much of the country supports him, there is some legal gray area (he has not been convicted), and the courts are “conservative.” I personally think he will win because Biden is getting even worse at speeches, much of the population doesn’t think their personal lives improved under Biden, and a lot of people are upset for how Biden has/is handling the Israel war.
A lot of things could happen before the election that would hurt Biden as well. A recession, expansion of Israel war, and losses in Ukraine are possibilities that could hurt Biden. I don’t think anything could hurt Trump. I think he could win the election from prison. Trump voters will eagerly buy any conspiracy theory to keep supporting Trump, and they don’t care about democracy or human rights. Democrat and Biden voters are much more critical and fickle.
I think wages have kept up with inflation, if I’m reading this chart correctly: https://fred.stlouisfed.org/series/LES1252881600Q
It’s normalized by CPI, so a flat line would be no wage growth.
Depends on the the sector, obviously. I’m guessing service sector wages have not kept up with inflation, since it seems like every place is understaffed (they’re not offering enough to attract workers). I also don’t think it accounts for part-time and gig worker wages.
I don’t understand the Scrum one. Scrum is also agile with short development cycles, and prioritizes communication with the product owners and stakeholders.
I’ve never heard of lean development, but not a fan of “lean manufacturing,” at least not the way it’s commonly implemented in the U.S. (using primarily temp workers so they can ramp up and down their workforce as needed; and it also exacerbates supply-chain problems).