Cryptocurrencies have uses beyond just currency.
FaceDeer
joined 1 year ago
It's actually more true for proof-of-work mining than it is for proof-of-stake. PoW mining has strong economies of scale, a professional miner with a warehouse full of mining rigs and a special deal with an industrial electricity supplier can churn out hashes more cheaply than a home miner can. Whereas the hardware needed for PoS is negligible so there's nowhere near that disparity between small and large miners.
Also, under Ethereum at least (the largest proof-of-stake chain and the one I'm most familiar with the workings of), stakers don't "dominate" the network. They have no decision-making power over what the consensus rules are. If the users decide to upgrade to a new version and the stakers refuse to go along with that or try to push an upgrade that the users don't want then those stakers lose their stake after the resulting fork.
It hasn't been quantized, then. I've run 70B models on my consumer graphics card at a reasonably good tokens-per-second rate.
My understanding is that the bottleneck for the GPU is moving data into and out of it, not the processing of the data once it's in there. So if you can get the whole model crammed into VRAM it's still faster even if you have to do some extra work unpacking and repacking it during processing time.
The paper was posted on /r/localLLaMA.
It's been discovered that you can reduce the bits per parameter down to 4 or 5 and still get good results. Just saw a paper this morning describing a technique to get down to 2.5 bits per parameter, even, and apparently it 's fine. We'll see if that works out in practice I guess
And at 72 billion parameters it's something you can run on a beefy but not special-purpose graphics card.
Social media loves to have targets that "everyone agrees" are terrible and worthy of hating. Being part of a righteously angry mob is fun, gives a nice safe dopamine hit and lets you show you are part of the in-group.
Which isn't to say that Meta is innocent and Zuckerberg is a poor victim, of course. But it does mean that it's very easy to go overboard and believe false accusations that also fit with the narrative that Meta is terrible and awful and evil. There was a post a couple of weeks ago about Threads and the narrative in that conversation was that Meta was dedicated to eradicating open source (Embrace Extend Extinguish, think of XMPP!). I tried pointing out Meta's many contributions to open source but redirecting the mob was hopeless. Sure, the motives of these contributions are selfish. They want to crush their enemies. But in this case their enemies are OpenAI and Microsoft, and the weapon they're using is to fling open the walled gardens OpenAI depends on for the common man to enjoy.
It's a useful cautionary tale to keep in mind whenever there's a target that everyone around you hates because they're "obviously" the devil. Maybe they are, but it's likely more complicated than it seems.
One of the size classes they mention in the abstract is called "Weaver Pro" so my initial assumption would be that it's not. However, I find that with this sort of thing the most important secret is that something is possible. If Weaver works as advertised we will now know that it's possible fir a 34B model to get better-than-GPT4 performance, which means lots of people will be willing to devote resources to recreating it since they now know those resources won't be wasted.
And if Weaver is meant to be "commercial" I wouldn't be surprised if there's a bunch of censorship baked into it, so the eventual open-source version will have an advantage.
Zuckerberg: Yeah so if you ever need info about anyone at Harvard just ask. I have over 4,000 emails, pictures, addresses, SNS
[Redacted Friend's Name]: What? How'd you manage that one?
Zuckerberg: People just submitted it. I don't know why. They "trust me" Dumb fucks.
This exchange was from 2004, when Zuckerberg first launched Facebook from his college dorm. Facebook has never pretended to be anything other than what it is, people keep giving it their information, and then they make a surprised-Pikachu face and complain when Facebook does exactly what they've always done with it. What Facebook said they would do in the TOS that they agreed to.
Dumb fucks.
A quick Googling of "phones without Facebook" gives lots of options.
Facebook (and one assumes Threads/Instagram) create ghost profiles of everyone mentioned or photographed in any posts on their platform if they can’t link that mined data to an actual account. So if your friends use the platform, then you do too, whether you know it or not.
This would be Facebook using your friends' data. It happens to be about you, but it's not yours.
I'm not saying people shouldn't care about privacy or take actions to protect it, but I think it's possible to go overboard in believing that nobody is allowed to know anything about you without your permission. There's a lot of public information that's available to anyone simply by being in public.
If you don't want your friends telling people about you, that's on you to tell your friends that. If your friends refuse to keep your existence secret then you'll have to decide whether that's more important to you than having friends.
Indeed. All "value" is ultimately something that is collectively decided upon by society. A chunk of rock could be worthless or worth billions depending on how much people want it.