Thank you! I checked it. From what I understand I should use a link like https://matrix.to/#/@[user]:[server.zzz]. Then from there they are redirected to use their own Matrix app, if they have one.
also @recreationalplacebos@midwest.social thank you! I had no idea about this possibility and these Firefox forks. Looks a little complicated but I'll try it. From what I gather, Firefox plans to bring back full extension support in the future?
I'd like to add one more layer to this great explanation.
Usually, this kind of predictions should be made in two steps:
calculate the conditional probability of the next word (given the data), for all possible candidate words;
choose one word among these candidates.
The choice in step 2. should be determined, in principle, by two factors: (a) the probability of a candidate, and (b) also a cost or gain for making the wrong or right choice if that candidate is chosen. There's a trade-off between these two factors. For example, a candidate might have low probability, but also be a safe choice, in the sense that if it's the wrong choice no big problems arise – so it's the best choice. Or a candidate might have high probability, but terrible consequences if it were the wrong choice – so it's better to discard it in favour of something less likely but also less risky.
This is all common sense! but it's at the foundation of the theory behind this (Decision Theory).
The proper calculation of steps 1. and 2. together, according to fundamental rules (probability calculus & decision theory) would be enormously expensive. So expensive that something like chatGPT would be impossible: we'd have to wait for centuries (just a guess: could be decades or millennia) to train it, and then to get an answer. This is why Large Language Models do two approximations, which obviously can have serious drawbacks:
they use extremely simplified cost/gain figures – in fact, from what I gather, the researchers don't have any clear idea of what they are;
they directly combine the simplified cost/gain figures with probabilities;
They search for the candidate with the highest gain+probability combination, but stopping as soon as they find a relatively high one – at the risk of missing the one that was actually the real maximum.
(Sorry if this comment has a lecturing tone – it's not meant to. But I think that the theory behind these algorithms can actually be explained in very common-sense term, without too much technobabble, as @TheChurn's comment showed.)
Unfortunately the original article is based on statistical methods that are today acknowledged, by a large number of statisticians, to be flawed (see this official statement and this editorial of the American Statistical Association). So the findings might be correct, and yet again they might not be.
Had never heard about Graphite, thank you! I'll try to stay updated about it. But please feel free to post important news about it in this community, whenever there'll be steps forward.
Absolutely amazing!! I suppose you've seen some renderings like this one.
However, these molecules don't really have a will or a scope, and in fact I don't like how they are deceivingly represented in some of these animations. These animations show, say, some aminoacid that goes almost straight towards some large molecule and does this and that. And one is left with the question: how does it get there and how does it "know" that it should get there? The answer is that it's just immersed in water and moved about by the unsystematic motion of the water molecules. Some aminoacids go here, some go there. In these animations they only show the ones that end up connecting with the large molecule. OK, this is done just to simplify the visualization, but it can also be misleading.
Similarly with molecules like kinesin, which seem to purposely walk around. Also in that case there's a lot of unsystematic motion, that after a while ends in a particular more stable configuration thanks to electromagnetic forces. Simulations such as this or this give a more realistic picture of these processes.
Don't get me wrong, I'm not saying that the whole thing isn't awe-inspiring or mind blowing. It is. Actually I think that the more realistic picture (without these "purposeful" motions) leads to even more awe, because of the structured complexity that comes out of these unsystematic motions.
Cool! Let's see what kind of material people bring out :)
I remember I did a search 5 or more years ago, and it was actually tricky because I only got something after searching for very specific terms, which in turn I had gotten from other searches. I'm trying to remember what they were...
Luckily "effectively the standard" is just a temporary thing. What browser was considered "standard" has changed many times in the past, and will continue to change in the future. Of course for this to happen everyone who cares must keep on pushing.
Not hostile at all, thank you :) Indeed I asked this question because I suspected I was misunderstanding how to use the site. Here's an example. A comment in this post:
If I click that link, I do see the Mastodon post, but it's on a Mastodon instance different from mine (https://c.im/). If I wanted to boost, favour, or reply, I have to open another tab/window, go to my Mastodon account on my instance, search for that post, and then I can boost etc. Let me know if this is unclear.
Maybe it's just because I use Mastodon via browser and not a 3rd-party app, as some comments here have suggested.
Math requires insight that a language model cannot posess
Amen to that! Good maths & science teachers have struggled for decades (if not centuries) so that students understand what they're doing and don't simply give answers based on some words or symbols they see in questions [there are also bad teachers who promote this instead]. Because on closer inspection such answers always collapse. And now comes chatGPT that does exactly that instead – and collapses in the same way – and gets glorified.
Amen to what you say on infographic content as well 😂
This image/report itself doesn't make much sense – probably it was generated by chatGPT itself.
"What makes your job exposed to GPT?" – OK I expect a list of possible answers:
"Low wages": OK, having a low wage makes my job exposed to GPT.
"Manufacturing": OK, manufacturing makes my job exposed to GPT. ...No wait, what does that mean?? You mean if my job is about manufacturing, then it's exposed to GPT? OK but then shouldn't this be listed under the next question, "What jobs are exposed to GPT?"?
...
"Jobs requiring low formal education": what?! The question was "what makes your job exposed to GPT?". From this answer I get that "jobs requiring low formal education make my job exposed to GPT". Or I get that who/whatever wrote this knows no syntax or semantics. OK, sorry, you meant "If your job requires low formal education, then it's exposed to GPT". But then shouldn't this answer also be listed under the next question??
"What jobs are exposed to GPT?"
"Athletes". Well, "athletes" semantically speaking is not a job; maybe "athletics" is a job. But who gives a shirt about semantics? there's chatGPT today after all.
The same with the rest. "Stonemasonry" is a job, "stonemasons" are the people who do that job. At least the question could have been "Which job categories are exposed to GPT?".
"High exposure: Mathematicians". Mmm... wait, wait. Didn't you say that "Science skills" and "Critical thinking skills" were "Low Exposure", in the previous question?
Icanhazcheezeburger? 🤣
(Just to be clear, I'm not making fun of people who do any of the specialized, difficult, and often risky jobs mentioned above. I'm making fun of the fact that the infographic is so randomly and unexplainably specific in some points)
stravanasu @ pglpm @lemmy.ca Posts 38Comments 279Joined 2 yr. ago 
Thank you! I checked it. From what I understand I should use a link like
https://matrix.to/#/@[user]:[server.zzz]. Then from there they are redirected to use their own Matrix app, if they have one.