AlbigensianGhoul

Just to clear it up a bit, these are three different operations in three different states with three different intents and even different targets. While sad, the framing makes it seem like it was all one big coordinated effort by the "Brazilian Police," which is actually divided in command on a state level (and are very independent from the government). For the Yankees, it'd be like reporting people killed in USA police operations in California and New York in a single breath. This is a common trick of Gringo news to make it seem like third world countries are more dangerous than they actually are, coupled with only reporting on them when tragedies happen.

I was almost agreeing with you until this one. If you think this is an example of "good discussion" you might want to go try it out with the libs over there to see how effective it is. I can assure you you'll be met with more "aggressive" responses.

Certainly those who have patience and enjoy kindly educating others should be encouraged to do so and maybe they'll even convince some. But at the end of the day this is an internet forum for internet randos, not a political party, and we should not police other users from shitposting and poking fun at obvious wreckers and trolls except when they go overboard or break rules. If you want to go talk with others, go for it. There's even !educatinglibs@lemmygrad.ml if you want to help organise how to properly communicate there and share experiences. But don't come here with your bloated pseudoscientifc "examinations" like you're writing a contrapoints script and expect to be taken seriously.

And the irony of wanting good discourse and ending with:

Thank you for coming to my TED Talk

As if the FMI loans weren't basically Nazi gold, but with interest and forced government changes to serve those Nazis.

Work visas/permits are just proletarian bans. If you derive your survival from labour and are not allowed to work somewhere legally, your only survival options are to move elsewhere or work illegally. In Northwestern countries like the USA this is by design, because then immigrants/refugees without visas are coerced to work without their deserved rights which then becomes a justification to jail them and force them into prison slavery.

Well yes, but the one it talked about was the SMO. Ukraine didn't exist to libs before then. I tried replicating it and they seem to have fixed it now, but the other day I also managed to get it to talk about the Wagner "rebellion" so there's definitely some data leakage in there.

How long is it? Just so you know, I’m a total newbie at LLM creation/AI/Neural Networking.

I have never done a full run on an LLM training, but on the lab we used to have language models training for like 1-2 weeks making full use of 4 2080s IIRC. Fine tuning is generally faster than training so it'd be around a day or two on that hardware, but I don't have access to it anymore (and don't think it'd be ethical to use it for personal projects either way). On personal hardware I think it would again be back at the week mark. Since it's an iterative process sometimes one might want to either have multiple training runs with different parameters in parallel or repeatedly train to try and solve issues. There are some cloud options with on-demand GPUs but then we'd need to be spending money.

The bulk of the work is actually on making sure the data is appropriate and then validating if the model works correctly, which a lot of researchers tend to skimp on in their papers, and in practice is usually done by low paid interns or MTurk contractors.

So… if they’re inherently unreliable, why make them? Genuine question.

Cynical answer, stock exchange hype. Investors get really interested whenever we get human-looking stuff like text, voice, walking robots and synthetic flesh, even if those things have very little ROI on the technical side. Just look at people talking about society being managed by AI in the future despite most investment going into human-facing systems rather than logistics optimisations.

The main issue (incredibly oversimplified) with ChatGPT is that due to it working on a text probability level, it can sometimes create some really convincing and human-sounding text, that is either completely false or contains subtle misrepresentations. It also has a lot of trouble providing accurate sources for what it says. Or it can mimic what appears like "human memory" by referring to previously said things, but that's just emergent behaviour and you can easily "convince" it that it has said things it had not, for instance. Also the data can get so large that some stuff that shouldn't be there can get in there as well, like how ChatGPT 3 is supposed to have a knowledge cut-off in September 2021, but it can sometimes answer questions about the war in Ukraine.

ChatGPT can still be useful for bouncing ideas around, getting some broad overviews, text recommendations or creative writing experimentation. They're also fun to dunk on if you're bored on the bus. I think this would be a fun project, but if do it, we should always have a big red disclaimer that goes "this bot is dumb sometimes, actually read a book."

Here's an example of how bad chatGPT is at sources. Bing has direct access to the internet and can sometimes fetch sources, but I'm not sure how that works and if it is feasible with our non-Microsoft-level resources.

CW libshit

I started crawling the Portuguese version of the MIA and there was definitely enough data there alone to fine tune one (about 30 GB but a lot of it was pdfs). Just be aware that it requires a lot of data preparation and each trial takes a while on civilian hardware, which is why I postponed it. If there's interest here I'd be happy to collaborate.

Also seems like a fun way to read a lot of diverse texts.

Edit: but be aware that LLMs are inherently unreliable and we shouldn't trust them blindly like libs trust chatGPT.

I wouldn't be surprised if dotworld specifically was administered by people related to Reddit Inc. The admins are generally very technically competent and it grew so fast with barely any technical difficulties that I wouldn't put it past them from being at least supported by some larger tech company.

What a spez move. Pre-emptively de-federating is just a bad move, no other way to look at it. They're a very diverse group and generally much kinder than most lemmy users. At the very least you should've tested federation for a day or two to see how the interactions play out. But anybody here can go over there and see for themself how nice they can be even when disagreeing, which they do a lot among themselves.

Also where in the Code of Conduct does it say the only ideology allowed is liberalism? Going the way of Reddit with vague justifications and arbitrary decisions will make the administration a lot of profit some day, but there's a reason people left that one.

Breaking News: Anarchists are now tankies! No one is safe unless they're inside a 50 ton Ironclad on wheels.

Those aren't "Actual Americans" to them though, they are "African-Americans" or "American Indians," and the other Americans from the rest of the continent are just "Latinx". Can't have them thinking they belong in the continent, let alone the country.

They're the ~~Immigrant Control Exterminators~~ Immigration Customs Enforcement, a Nazi-like government police with a different name, that go around kidnapping visa-less immigrants and putting them in the privatised detention centers, separating families, which rather than deporting them, force them into prison slavery. BTW, since Joe Biden got elected, the number of prisoners has only gone up, to near a million. Just writing that out actually hurt my brain, it's weird to think we already know all these things about the USA but don't connect the dots unless specifically addressing it.