Depends what your real tasks are. If they involve writing, researching, watching videos, filling out web exams, etc., yeah, very capable. If you’re compiling massive codebases, editing composited video, or recording several audio channels, obviously not. But that’s the pitch. If you’re a student and want a well built machine at a competitive price with a long battery life, they’re tough to beat.
It bums me out that the corporate misdeeds were so damaging with LLM’s that people say stuff like this. The technology of using backpropagation to tune networks and using high dimensional matrices to predict new vectors itself isn’t toxic. It is fascinating and can be useful. But without focus I can see how it all looks like just ChatGPT hallucinating wrong answers to lazy college students or what have you. That’s painting with pretty broad strokes though.
Fascinating how uttering the term “AI” can cause so many downvotes even when the system is just a harness that runs self hosted and can use completely openweight, even self trained models. Or maybe many on Lemmy just hate pewdiepie.
Fair. I meant proportions/silhouette, not details like lights. I just mistakenly thought those were also part of design.
I seem to be unclear I guess. But to me there is a difference between, “can’t use the same design language” and “can’t look sleek”.
My point is simply that electric vs combustion power distribution and volume / mass differences make it, I assume, difficult to maintain the same design language.
Difficult to use the same design language with massive batteries. I actually think it looks okay. For me the issue is the 700k price tag.
That nvidia one is probably impossible to pin down, since it is really just a big ouroboros investment human centipede eating its own revenue out of all the other asses.
You’re right that hypersonic cruise missiles are probably a greater threat and maybe there is a propaganda side to this, but the Oreshnik is no slouch and not particularly easy to intercept. Yes it is ballistic, but it carries MIRV’s which are notoriously challenging. Not arguing at all, just explaining that these are a substantial threat too.
Our first few batches were not backsweetened. They were a bit dry but I still love the flavor. But the backsweetening is actually a fun bit of alchemy itself. Some people think it is heresy for purity reasons, and that is a-okay, but to the worries some have that it is unsafe, that can be mitigated. We actually pasteurize, so it is no bottle bomb issue.
The maple is really good. Maple and vanilla are 2 of my favorite flavors, but my wife talked me into the butterscotch extract and it actually turned a corner. Like, it leveled it up way further than I expected.
The caramelization idea is interesting. Maybe we should experiment.
Yes. Which is why he got involved in this. Hoping to get some information on the process then pull out before shipping. Then the next day they finally decided to ship, or something like that, so he got stuck.
And coffeezilla.
We haven’t tried any melomel yet. Our backsweetening is based around warm flavors like butterscotch, maple, and vanilla so those profiles would clash.
The honey we prefer is Honning AS Ukrainian Sommer. Might be best to get from a local beekeeper, but this stuff is extremely consistent and delicious. Doesn’t really matter too much though, candidly. The fermentation strips a lot of the character.
I run docker with about 10 services on a DS923+ with no issues.
Safe in ed.
Not due to cost as much as just aging and having some time and space, my wife and I have started making wine and mead. Actually what prompted it was moving to a place with a plum tree.
I highly recommend it; very rewarding. Pretty easy, not a huge startup cost. Delicious!
We are now rotating 6 x 5 L carboys, and making lot’s of mead. We have got the process down to a very precise set of steps with careful measurement and controls. If you amortize cost for non-consumables like carboys, siphons, hygrometers, etc. and calculate only consumable cost, each batch is way cheaper than anything in a storefront, and made with love. Everything is better when made by one’s own hand.
There is definitely some truth to what you’re saying, but my point is that those aren’t conflicting with the technology working. There are many scholarly refereed papers on transformer performance and generational improvement on standardized metrics. I don’t see the value in conflating something working with it being good or ethical. There is a gap between utility and hype, yes. That doesn’t mean it doesn’t work, and the inexorable negativity that comes invariably to comments recognizing this simple truth undercuts actual critical feedback.
I think it is so strange people say stuff like this as though there aren’t objective metrics showing it does. We don’t have to like the billionaires using it to subjugate people, or the energy and water consumption, or the theft of copyrighted materials to be honest about the technology.
It does work. As far as ml models go, since backpropagation was implemented in training, transformers have become extremely capable.
Good luck using a website without a browser or at least curl.
Pretty simple solution to that.