Thread 22994928

>yandex ai
it's ovaaa

i just use yandex to search for free movie streaming sites, honestly
was it good for other things?

you ain't never gonna get these results on google, but the sites work fine

>>22994931
I just don't want t use google & bing & ddg & brave so I figured yandex was the best alternative
>>22994938
I guess I can still use it but I just used it for a search and it defaulted to the ai tab which put me off

>>22994948
i do hate how everything's being nerfed besides AI and short-format videos
>>22994948
>I just don't want t use google & bing & ddg & brave so I figured yandex was the best alternative
you can always use one of those aggregator sites like 4get.ca
it adds some privacy as well
they still rely on the big goofy search engines though

>>22994948
>>22994958
Stop being poor and get kagi

>>22994958
>AI
I dislike how people use it over critical thinking which puts me off from using it. I had an embarrassing interaction on /mu/ a while ago >>>/mu/126863330 I'm sure it has it's few minor uses but I know I can exist without it.
>short-format videos
I don't really notice these as much. I think it's because I treat them like ads so they aren't as important in my mind as they should be considering I prefer not to see them.
>4get.ca
this is the "4chan" one right? with minecraft captchas? I used to use it, not sure why I abandoned it.
>>22994965
>kagi
never heard of it, thanks for the suggestion.

>>22995001
>I dislike how people use it over critical thinking which puts me off from using it. I had an embarrassing interaction on /mu/ a while ago >>>/mu/126863330 I'm sure it has it's few minor uses but I know I can exist without it.
it's depressing, because they could actually be extremely versatile and useful
it's just matrix math they're built on
these bigass models the companies are using are dubious because of the model size and too much training data, you get overfitting
and that's before talking about whatever "filter layers" and other bullshit they apply to them
"neural networks from scratch" is a good book, from like 5 years ago
but yeah

you can start to get more interesting and unpredictable answers, which can actuallly STIMULATE your thought process, when you use smaller datasets (the hundreds of gigabytes of reddit posts and stackexchange posts the models are based on) and/or a smaller model (number of "neurons" and layers)

who the hell has time for that though...

it's being forced on people, and it's a retarded blunt tool which does make people dumber as it is

>>22995001
>I don't really notice these as much. I think it's because I treat them like ads so they aren't as important in my mind as they should be considering I prefer not to see them.
i see
>this is the "4chan" one right? with minecraft captchas? I used to use it, not sure why I abandoned it.
yes, that is the one
the captcha is kind of annoying and inconvenient, but it's not that bad, and you only have to do it once a day

>>22995007
>these bigass models the companies are using are dubious because of the model size and too much training data, you get overfitting
Interesting. So if you were to train a LLM on a math textbook, you'd get better results than if you trained it on the entire internet where it has the chance of taking in bad info?
>>22995052
>the captcha is kind of annoying and inconvenient, but it's not that bad, and you only have to do it once a day
See, I all my browsers set to delete cookies upon closing so I had to do them a bunch. Maybe that's why I stopped using it. Yandex had a few too but there wasn't a set pattern to it.

>>22995094
>Interesting. So if you were to train a LLM on a math textbook, you'd get better results than if you trained it on the entire internet where it has the chance of taking in bad info?
sort of
aaah i wish i had the words to explain it

the more data a model has, the more it tries to fit to the data it's already seen
this goes for neural networks in general, not just LLMs

if you make a neural network which tries to recognize the number from 0 to 9 which is drawn in an image, and you train it on 10,000 sample images, you should be able to feed it a completely new hand-drawn image & it'll guess it correctly
but if you train it on 1,000,000 different sample images, it become less and less able to recognize the digits in a new image
the more data it's trained on, the more it panics/fumbles when you feed it something which doesn't look like its training data

i don't know if you can see the implication for LLMs, but yeah

my guess is, if you trained your own LLM on like a hundred math textbooks, it would atleast produce interesting answers

i wish i had the words

>>22995094
>See, I all my browsers set to delete cookies upon closing so I had to do them a bunch. Maybe that's why I stopped using it. Yandex had a few too but there wasn't a set pattern to it.
ah, that makes sense
4get.ca is supposed to give you 100 free searches when you solve the captcha once
i would hate having to do it each time

>>22995119
>>22995124
It's ok, I understand what you're trying to say,
thanks

>>22995160
ye
good luck

>>22995119
>>22995094
It's just a new branding of Socialism.

>The internet is so diverse that it taught your ai ruler everything it knows.

Is more palatable than and better (AI) marketing than private public information.

Fuck I feel like an ai for using the word than wrong.

>>22995189
i think it's just some niggas who want to make alot of money
the more inefficient their shit is, the more graphics cards they "need", and the more they can charge consumers