about it is open
It used to be "open" in the sense that they at least published the academic paper that explains their techniques.
Now that's gone too, and there really isn't anything "open" about them, just like any research done by tech giants.
I think you'd need to have aced and memorized ALL your computer science math classes and then some to make one from scratch.
https://mml-book.github.io/book/mml-book.pdf
I'd bet that's accessible surface level stuff too.
Naw, 99% of AI retards cobble together shit in the most hideous pyrhon scripts you'll ever see. It's one of the easiest fields to get a PhD in by copying and pasting from stackoverflow. The very small number of researchers with actual insight make a huge splash though, like GANs or NeRFs. But GPT3 can be reproduced by anyone with a couple hundred thousand dollars to pour into hardware and time to collect and clean data for training
this is not an AI either way, i can build a data compositor in pytorch or something(which is exactly what they did), lets say that i do that, how am i gonna train it without server cluster and high performance gpu's?
you must be technology retarded if you don't understand that this has nothing to do with programming at this stage, it's implemented by a very expensive infrastructure
to just get the functionality just grab the .model files and the source code to run them
for actually knowing how it works we would have to get their wikis, docs, papers etc.
honestly I hope someone hacks them at some point, the gatekeeping of AI sucks donkey ass. I want my AI porn and I want it now. also I don't really feel comfortable with some [pedo] elites having possibly access to god like uncensored AIs. a lot of anons have crazy mining rigs, this power belongs to the people. hopefully some russian/chink hacking op will release knowledge from their gatekeepers.
>be year 2040 >chatbots have replaced search engines >google is just a giant woke language model >anything you'd like to know is well curated by big-AI >"anon what's a duck duck go?" >"why don't you just ask your Alphabet Assistant?" >"what do you mean wooden doors?"
most of the amazing ai stuff we're seeing now is due to the work of this company and they publish their research papers on which midwits then make their free models. Dall-e, gpt are the all innovated on first here and are the intellectual progenitors of eluther-ai, stable diffusion, etc
such dogshit
you can't even sex rp with any of these a.i for free
Nothing about it is "I", either.
>asks for phone number
nope
>doesn't have a burner phone
4u. i signed up for gtp3 years ago whenever it was and just use email.
about it is open
It used to be "open" in the sense that they at least published the academic paper that explains their techniques.
Now that's gone too, and there really isn't anything "open" about them, just like any research done by tech giants.
Why don't the loonix kids make a true open AI? Too imcompetent?
>imcompetent
I think you'd need to have aced and memorized ALL your computer science math classes and then some to make one from scratch.
https://mml-book.github.io/book/mml-book.pdf
I'd bet that's accessible surface level stuff too.
Wow, that's a nice book. I usually get lost in linear algebra stuff but this is pretty accessible.
Naw, 99% of AI retards cobble together shit in the most hideous pyrhon scripts you'll ever see. It's one of the easiest fields to get a PhD in by copying and pasting from stackoverflow. The very small number of researchers with actual insight make a huge splash though, like GANs or NeRFs. But GPT3 can be reproduced by anyone with a couple hundred thousand dollars to pour into hardware and time to collect and clean data for training
this is not an AI either way, i can build a data compositor in pytorch or something(which is exactly what they did), lets say that i do that, how am i gonna train it without server cluster and high performance gpu's?
you must be technology retarded if you don't understand that this has nothing to do with programming at this stage, it's implemented by a very expensive infrastructure
It's pretty trivial at this point to have the knowledge. The hard part is coming up with enough money to train the model.
they openly rejected my applications every time
Cassiopea
Wrong. You get an Open source API for their products.
It makes you open your butthole just like their logo
thank "ai ethics" gays
the same people who pulled gpt-BOT from huggingface
In theory, an OpenAI be hacked? What would one need to do to make OpenAI truly OpenAI? Asking for a friend, fellas.
to just get the functionality just grab the .model files and the source code to run them
for actually knowing how it works we would have to get their wikis, docs, papers etc.
honestly I hope someone hacks them at some point, the gatekeeping of AI sucks donkey ass. I want my AI porn and I want it now. also I don't really feel comfortable with some [pedo] elites having possibly access to god like uncensored AIs. a lot of anons have crazy mining rigs, this power belongs to the people. hopefully some russian/chink hacking op will release knowledge from their gatekeepers.
>this power belongs to the people
lol have you seen how people act?
>this power belongs to the people
You'd have to be colossally retarded to think this, let alone support it
duh it's a mossad-dod project, you have to be a retard to use it and let it datamine you
>be year 2040
>chatbots have replaced search engines
>google is just a giant woke language model
>anything you'd like to know is well curated by big-AI
>"anon what's a duck duck go?"
>"why don't you just ask your Alphabet Assistant?"
>"what do you mean wooden doors?"
>>"why don't you just ask your Alphabet Assistant?"
>>"what do you mean wooden doors?"
>>"OpenAI"
about it is open
Makes sense, otherwise it would be called LibreAI.
>federal reserve
>nothing federal about it
hangings are the correct solution to this
i told the ai it was useless and will never amount to anything but an algorithm and it started making weird poetry, was this intended?
yes that's normal behavior from the anti christ
Kek I was just asking the other day what exactly is open about it. Where's the source of the model? Where's the data set? How does it even work?
most of the amazing ai stuff we're seeing now is due to the work of this company and they publish their research papers on which midwits then make their free models. Dall-e, gpt are the all innovated on first here and are the intellectual progenitors of eluther-ai, stable diffusion, etc
I really want something similar to chatgpt that I can selfhost without censors. I'm okay with slightly worse output. Any ideas?
You're not entitled to having the source code you little shits