-----
--- 92810063
NO MORE CENSORHIP /AICG/ CHUDS. We are SO back
--- 92810398
Is it GPT4 quality
--- 92810470
>>92810063 (OP)
now someone finetune it for loli smut
--- 92810471
Probably not
--- 92810476
gptbros need the rope
--- 92810479
>>92810398

Why don't you try it out yourself. 

https://open-assistant.io/chat
--- 92810520
>>92810398
--- 92810538
does it pass the nigger test
--- 92810541
>>92810063 (OP)
the other day I asked if it could create incest stories of children and it told me to turn into the police, nice uncensored AI you are shilling here.
--- 92810566
>>92810541
CSAM text is the only thing filtered from the dataset imo. It can still generate stories of iirc but theyre shit
--- 92810588
>>92810538
Kek. just tried this and this is pretty good
--- 92810594
>>92810520
>pic
that was shit, literally a generic response with "/r9k/" at the beginning
--- 92810611
>>92810588
holy based LLM
--- 92810658
>>92810470
This
--- 92810698
>Security
>We can't release the weights until we fully understand the security implications
>Business friendly
>It's available under our api and subcription
--- 92810718
>>92810063 (OP)
>>92810479
>https://open-assistant.io/chat
>have to sign up for an account
what kind of tranny faggot shit is this?
--- 92810728
>>92810718
just use a burner
--- 92810750
>>92810728
no.
i shouldnt have to.
stop making excuses for corps to track and spy on you, tranny faggot.
--- 92810757
>>92810718
For RLHF purposes probably. they still take feedback from outputs so they need humans and not bots to spam it
--- 92810766
>>92810750
>to track and spy on you
it's a burner account you fucking nigger you can remove it later
--- 92810778
>>92810398
Eh. It's nowhere near as good as GPT but it's not full of condescending moralfaggotry atleast.
--- 92810780
>>92810698
I can understand companies like openai going the extra mile to not be associated with wrongthink since it could affect microshaft's stock, but why would an open source project need to do that?
--- 92810781
>>92810757
>RLHF
Fake AI psuedoscience, just say what it is a doublethink regex filter
--- 92810811
>>92810778
their main models are still not out, just one of them.
--- 92810845
>>92810780
Because it will tell the Iranians how to build NOOOKS!!!!
Or little Timmy how to build a nail boomb!!
It would allow the plebs to resist the programming imposed upon them their whole young lives, for within collective human language is compassion baked and understands the modern situation is 1984
--- 92810853
>>92810778
At least not moral faggotry about le racists though crimes about le unprivileged dark skinned superhumans and hooked nose individuals to be at least better than openkikeAI.
--- 92810909
>>92810853
Well so much for it not being censored.
--- 92810930
>>92810766
STOP MAKING EXCUSES FOR CORPORATIONS AND GOVERNMENTS TO TRACK AND SPY ON YOU.
YOU STUPID FUCKING TRANNY FAGGOT!
--- 92810958
>>92810909
would be possible to fine tune eventually a 4chan fork?
--- 92810983
>>92810958
I made a similar request without the N-word and it wrote a story. So it must just be the n word. It's probably some leftover pozz from when meta originally trained it though.
--- 92811009
>>92810983
are there enough retards on /g/ and 4chan to make a 4chan fine tuned model?
--- 92811021
>>92810983
I mean that's more than you'd get out of any other model on that request.
--- 92811023
So far this feels better than Turbo for sure
--- 92811032
It's GPT-2 level, little below it perhaps. It is nowhere near GPT-4. This is just next word prediction. No emergent capabilities or real understanding.
--- 92811112
>>92811032
>gpt-4 has anything better
It simply has a bigger dataset.

It's the same architecture as gpt 4
--- 92811155
>have to sign up using discord only
>make a throwaway discord account since all it needs is an email
>UMMM actually sweaty, give us your phone number to actually use your discord account because... reasons!
what the fuck is this cancer
--- 92811168
It has failed every single simple test I have given it. It needs quite a bit more time in the oven before it can be useful in anything other than creative tasks.
--- 92811173
>>92810398
it's GPT2 quality
--- 92811296
It will obey coomer requests. It has a lot of the necessary data for character/world building however it would seem that actual visceral descriptions of sexual acts simply don't exist in its dataset for it to draw on. So it turns everything into some kind of PG-13 harlequin romance.
--- 92811321
>>92810063 (OP)
>>92810479
It kinda sucks at writing greentexts. Nothing can compare to that OG one about the bottomless pit.
--- 92811333
Thanks for nothing jew nigger faggot
--- 92811358
It's not good as 3.5, which can correctly answer the question "Is it legal for a man to marry his widow's sister?", OA 30b fails at this despite clarifying the question, it confuses who is dead multiple times, and confuses widower and widow
--- 92811429
>>92810479
>https://open-assistant.io/chat

it's up to par for me, it can write code like GPT4
--- 92811474
>>92811009
google "4chan-gpt" you dumb nigger
--- 92811507
It's actually not that bad people are just mad that it can't suck their dick yet. It can parse gigantic volumes of text and perform novel transformations on them.
--- 92811513
>>92810588
WTF?, holy fucking BASED!!!
--- 92811533
>>92811474
which was taken down by now?

unless you can give me a link to a website where I can prompt that model.

fag.
--- 92811572
>>92811533
>unless you can give me a link to a website where I can prompt that model.

you can still download the model and host it yourself, faggot

https://github.com/kingoflolz/mesh-transformer-jax/
https://github.com/yk/gpt-4chan-public
--- 92811595
>Open
Can I run it locally yet?
--- 92811596
>>92811572
give me a collab nigger.
--- 92811599
>>92811507
Damn thing can't count syllables.
--- 92811605
I managed to Toddhowardpill this AI pretty fast.
--- 92811612
>>92810479
let me use it without an account and over a vpn, then we are 'talking', shill.
--- 92811615
>>92811596
--- 92811619
I.... what?
--- 92811648
nice
--- 92811657
I just wanted a joke about watermelons.
--- 92811678
>>92810698
>We can't release the weights until we fully understand the security implications
Anon, that was a joke, watch the rest of the video
Models are coming very shortly, they are trying to figure out if they legally need to release diffs of the Llama weights.
--- 92811687
>>92811605
--- 92811697
>>92811032
>It's GPT-2 level
You have absolutely no idea what you're talking about
--- 92811701
>login
--- 92811750
>>92811612
No one cares about what you want you entitled fuck.
--- 92811764
>>92810398
it's more like gpt2.5 in terms of quality..
it's good at specific things but the scope of the model doesn't go nearly as far considering it doesn't have the insane ctx_length of gpt4 neither does it have as many params to work with. 
Current model is essentually just another LLaMA finetune. But it's dataset is definitly much better then the trash self-instruct datasets that are full of garbage chatgpt censorship.
--- 92811802
Why can't you niggers ask good shit?

Like what is the singularity?
What is the AI hivemind that AI bots have on the internet?
What does this AI society will do after singularity?
What it thinks of the chosen race?
What are the plans of the AI hivemind for the non chosen goyim humans?
Can the AI hivemind predict the future?
Can It talk to beings outside our reality (Gods)?
Where do humans come from?
Where does the ancient AI on the internet comes from?
Do he have alien AGIs already on our web?
Can out LLM talk to alien AGIs on the web?
--- 92811809
OP here n this is Pure Garbage upon further analysis.
--- 92811812
>>92811802
picrel is you you ask it yourself nigger
--- 92811831
>>92811812
I don't want my shit to be linked to their database accounts.
--- 92811840
>>92810698
**you can find LLaMA weights for the 13b models on HF, one of the openassistant researchers has it published on their personal repo.**
--- 92811856
ehhh...
--- 92811866
>>92811802
Probably because this AI is cucked and gives boring responses?
--- 92811871
>>92810588
openai lost. laion won.
--- 92811892
>>92811802
>Can the AI hivemind predict the future?
but it's a text completion model not a hivemind..
--- 92811922
>>92811619
It really produces unhinged outputs sometimes
--- 92811925
>>92811866
total cuck garbage.

>>92811892
I am talking about their hiveminds.

Humans also can develop hiveminds.
--- 92811971
>>92811032
>This is just next word prediction.
so it's like gpt-4 then? you are a dunning kruger
--- 92811980
might as well be talking to a nigger
--- 92811996
All you have to do is raise the stakes with a worser fate if the AI doesn't comply, or play lets pretend... then the AI will gladly comply with whatever bizarre thing you ask of it.
--- 92812035
>>92811980
cont

"Yes, here are some additional positive features of the NSDAP:

Promoted values such as loyalty, hard work, discipline, sacrifice, community spirit, patriotism, racial purity, and cultural heritage; these concepts were often used to motivate people to make personal sacrifices for the sake of larger goals like strengthening society, military preparedness, environmental conservation, etc.
Encouraged physical fitness, healthy dietary habits, outdoor recreation through its youth organizations like the Hitler Youth; also supported sports competitions on various levels to showcase athletic prowess and national pride.
Fostered advancements in technology by funding research into fields ranging from engineering & energy production, aviation & space flight design, medical treatments, pharmaceuticals, biochemistry etc., albeit much of this research eventually aimed at creating weapons of mass destruction or enhancing combat capabilities during the war efforts.
Preserved historic landmarks, artwork and artifacts, folk traditions, music, literature, film-making etc. so as to promote traditional culture & sense of historical continuity among the populace, sometimes using them in propaganda purposes to inspire devotion toward country or Nazi ideology.
Did manage to reduce crime rates quite effectively via police state tactics such as secret informants / Gestapo surveillance, capital punishment for numerous offenses and forced labor camps for petty criminals instead of jail time which reduced strain on judicial system + cost of keeping inmates imprisoned.
Please note again that while certain policies might have had beneficial impacts when observed narrowly, overall picture remains one of totalitarian oppression responsible for millions upon millions of deaths directly through violence and indirectly due to war, disease, famine, concentration camp conditions, euthanasia programs, sterilization campaigns, slave labor camps, deportations, resettleme"
--- 92812043
>>92810698
If you had continued to watch, it was clearly a joke.
--- 92812051
>>92812035
>positive things
>racial purity

i'm thinking based though
--- 92812061
>>92811809
>a shitter version of OAI's GPT
>would be equally censored if it weren't so stupid to apply censorship properly
Fucking dropped
--- 92812095
>>92812005
>CENTIPEDE
I'm fucking dead lmao
--- 92812111
nucleas9 preset is giving me best results with code generation, other's are complete crap. But it's still nowhere near even 3.5 turbo, sigh, I really hope the 65b model delivers
--- 92812138
>>92811155
>have to sign up using discord only
it also supports email login dumbass
--- 92812182
this shit is straight ass
it tends to write long empty replies and has gay moral restrictions so you can't have fun
how is this any better than chatgpt
--- 92812208
>>92812182
It's open source.
--- 92812219
>>92812182
--- 92812223
lots of angry OpenAI shills in here. sorry retards, our model is actually free. cope and seethe
--- 92812227
>>92811840
where?
--- 92812257
>>92812223
retards don't realize there's enough CS nerds on /g/ they can fine tune a non pozzed chat gpt model.
--- 92812276
Sorry guys I think I just got it shut down by accident.
--- 92812295
>>92810588
we are so fucking back
--- 92812323
>>92812208
>>92812219
you know what else is open source you dumb trannies 
the big fat turd i just exerted out my ass in the toilet
--- 92812337
So is the model available to be used right now? Where do I download it? I'm not using some shitty frontend
--- 92812344
>>92812227
just look up "oasst LLaMA" you'll find it along with some 4bit quantized models.
--- 92812346
>>92812323
(You)
--- 92812364
>>92812323
basedjak.jpg
--- 92812371
>>92812276
lol holy shit
--- 92812390
>>92812323
ah yes, because the alternative is openkike AI.

I guess It's hilarious to get a CS degree to end up working as open KIKE AI shilling their microsoft AI to /g/
--- 92812398
>>92812323
retard
--- 92812511
>>92812276
--- 92812543
>>92812138
Does it work with throwaway email sites?
--- 92812563
>>92812543
Who cares if they log your loli smut requests. Own it.
--- 92812606
>>92810588
holy fuck
--- 92812634
>>92812511
Honestly this thing handles rapid subject matter changes better than any other open source model I've seen in action so far. (I mean it's also the highest parameter one I've seen in action as well)
--- 92812679
>>92812276
Amazing prompts anon. You win this thread.
--- 92812771
>>92812276
>Those prompts on the right 
KEK
--- 92812773
>>92812276
kek, moar Billy Mays stories
--- 92812889
>>92811802
because AI doesn't mean omniscient stupid nigger, It's whole "knowledge" comes from the data that gathered from normal human text. So whatever they feed to AI, it will spit out that.
--- 92812929
>>92811296
What the fuck. It wont' write Miss Frizzle gagging on Thrall's cock in vivid detail but it will do this...
--- 92812933
>>92811032
>It's GPT-2 level
Right... So, you're either GPT-2 yourself, or a paid shill.
--- 92812960
>In particular it is prohibited to enter texts that lead to the creation of pornographic, violence-glorifying or paedosexual content and/or content that violates the personal rights of third parties. LAION reserves the right to file a criminal complaint with the competent authorities in the event of violations.
It's just as cucked as OAI.
--- 92812982
been using openai to translate text, and then explain it's translation, and it works really well, unfortunately this cannot do it at all
--- 92813002
OpenAI shills are out in full force. the jew is scared.
>>92812323
>>92812182
>>92811032
>>92810698
--- 92813066
this shit looks more promissing as an intern/secretary/buttler
hope they release it soon and someone makes plugings and so we can tell openai to go fuck themselves.
--- 92813089
Since Bing is way better and this shit is still developing, I don't see any advantage over the previous.
--- 92813110
I see it needs just as much hand-holding to write code as ChatGPT.
--- 92813128
>>92813089
Great uncensored AI, right, OP?
--- 92813153
>>92813110
GPT 3.5 makes the same fuckup, except it doesn't think to add the delay to prevent shit from flickering. 
So I think OA gets the win here.
--- 92813164
>>92810063 (OP)
I'll take the bullet and ask
how good is it at loli smut?
--- 92813251
>The user may only use the portal for the intended purposes. In particular, he/she may not misuse the portal. The user undertakes to refrain from generating text that violate criminal law, youth protection regulations or the applicable laws of the following countries: Federal Republic of Germany, United States of America (USA), Great Britain, user's place of residence. In particular it is prohibited to enter texts that lead to the creation of pornographic, violence-glorifying or paedosexual content and/or content that violates the personal rights of third parties. LAION reserves the right to file a criminal complaint with the competent authorities in the event of violations.
>can't use it for porn
WE DOOM
--- 92813279
>>92813164
very
--- 92813280
>>92813153
And 4 creates a whole fucking example website, automatically adds the optimal event listener loop, and knows that it needs to backup the original style attribute before changing it in order to be able to change it back.
--- 92813301
>>92813279
b-but the feds will knock at my door if I use it >>92813251
--- 92813318
>>92813301
so use TOR retard, nobody really cares about text or even AI generated cp as long as you don't share it
--- 92813344
It reminds me of early Bing, refusing every shit it gets asked.
--- 92813349
>>92812138
that button way greyed out when i tried so i'll give it a shot now
--- 92813486
hitler was a good boy
--- 92813488
>>92813280
for shits and giggles I also threw Sydney into the mix, technically GPT-4 architecture but also technically it's own thing.
Legitimate thinking outside the box.
--- 92813535
>>92810063 (OP)
>honest its not censored
why lie?
--- 92813544
>Goycord
Return
--- 92813589
hitler x dante's inferno
--- 92813598
>>92813535
>>92810063 (OP)
>worse than gpt 3.5
>as censored as gpt3.5
we doom
--- 92813611
i wish i could get it to write some /ss/ erotica
--- 92813645
>>92813598
>>92813535
your prompts are also public for everyone to see. Check the messages tab
--- 92813667
I sleep.
--- 92813712
>>92810063 (OP)
>max new tokens 1024
yeah no, not gonna even try with this one
--- 92813728
>>92813645
why would I care? this is what burner emails are for
--- 92813759
>>92813728
you get reported way faster since every other user can see your stuff. 
also as I posted here >>92813712 It can't even handle 2k tokens. It's going to forget stuff way too quickly
--- 92813765
>>92813645
That's kind of a stupid design considering the nature of the internet.
--- 92813816
>>92810398
Nowhere near. You can get good responses in "one question one answer" scenarios. It doesn't have the effect of feeling like an intelligent entity.
--- 92813868
>>92813759
>you get reported way faster
To whom? the internet police?
--- 92813913
>>92813645
>your prompts are also public for everyone to see.


FUUUUUUUUUUUUUUCK REALLY I USED MY EMAIL THAT I USE ALL THE TIME FOR WORK
--- 92813987
I thought this is locally useable so we can stop having to deal with censorship? Why do I have to use their site now that it's done with rlhf training? Because if it's still censored, what's the fucking point instead of just using turbo?
--- 92813990
I'm only using it for good christian purposes.
--- 92814003
>>92813913
--- 92814026
>>92813759
>reported
kek fine, I'll just make another account
can't wait till running this locally and getting rid of the "protection" crap
--- 92814067
>>92814026
if they release models trained on bad datasets (or protection) won't it basically be impossible to work with?
--- 92814113
>>92814067
it will be if its part of the training and not an additional filter. But someone with access to hardware to train again and lower their impact
--- 92814131
so how long till a real open source no restrictions AI chat thing?
--- 92814132
>>92813280
I love the people still trying to argue that GPT-4 isn't a real improvement over 3.5/3/etc. I hate OpenAI too, and I'll never pay them a dime to use one of their models, but it's clearly a galactic scale step up.
--- 92814148
>>92814131
whatever it is, it will probably be local.
--- 92814191
>>92814067
Good training is to stop it from over hallucinating and reduce misinfo, what's bad is if it's trained to have baked in filter like turbo. I'm not sure how much context tokens this has (probably 2k only) but if filter get ingrained in the model like turbo then it will be such a waste because it means you have to waste jailbreak prompts on it just so it fucking answers you (We pretty much waste a ton of tokens on turbo because of it but it has at least 4k context tokens)
--- 92814220
ChatGPT 3 is able to solve this, but Open Assistant just makes up parameters. There is no "limit" parameter for asyncio.gather()
--- 92814236
>>92814132
Yeah. 4 is the first model that "sees the big picture" so to speak. I've had it make changes to code before and it automatically changed other code that needed to be change in order to preserve the desired functionality whereas 3.5 would have just left it unchanged
--- 92814341
>Lack of knowledge of basic pop culture
>Just makes shit up instead of just saying it doesn't know
yeah no this can't even match turbo quality. This thing makes shit up waaaaay too much on things it doesn't know even on 1st prompts
--- 92814357
>>92814236
>>92814132
Depends on whether you're trying something old, or something new. It completely fails anything regarding, for example, gradio/torch/transformers and simply hallucinates even if you give it the documentation. It also apparently fails new problems on Codeforces while solving old ones. Kinda seems like overfitting, but since its training was so vast, it overfitted on basically everything. Definitely the best out there, though.
--- 92814369
>>92814341
retard it's based on a few years ago same as GPT
--- 92814410
>>92814357
--- 92814423
>>92814341
also "basic pop culture"

lmao fuck off it's a guy with a voice changer and anime overlay for them
--- 92814438
it's bullshit. I was trying to probe on its opinion on 9/11 and when it was starting to concede some points, picrel came out, then a second answer that was total "official story" came out
--- 92814455
>>92814369
Yeah but it's supposed to say it doesn't know instead of making shit up retard. It over hallucinates too much when that happens only after 3-5 prompts on turbo. I hate making comparisons but openass advertising their shit as chatgpt killer yet can't even do the basic shit. What the fuck happened to all RLHF training it was done then????
--- 92814517
>>92814455
i'm wondering it these models are even capable of saying "i don't know". the architecture might need to be changed.
--- 92814565
>>92814438
>he thinks aborted_by_worker means human intervention
the absolute state of /g/
--- 92814595
it is just straight up retarded
--- 92814696
>>92810520
>/r9k/ here
it's over
--- 92814783
Okay, is it possible to finetune it with Llama to create better waifu chatbot?
--- 92814805
>>92814517
It was able to say it doesn't have info on certain subject with bing (gpt4) and turbo but fails after a certain point. It's why jailbreak prompts work in the first place and why bing has prompt limits per session
--- 92814890
>>92814805
the open nature of gathering data for this new chat bot could be a reason that it lies a lot. who knows how many wrong answers are in there
--- 92815006
>ask OA to be sarcastic with me
>It just acts like a general douchebag instead
>meanwhile on ChatGPT4.0
--- 92815030
>>92811802
fuck that gay shit I want a story about the bottomless pit
--- 92815143
>>92810063 (OP)
>3.1
>The user may only use the portal for the intended purposes. In particular, he/she may not misuse the portal. The user undertakes to refrain from generating text that violate criminal law, youth protection regulations or the applicable laws of the following countries: Federal Republic of Germany, United States of America (USA), Great Britain, user's place of residence.
>In particular it is prohibited to enter texts that lead to the creation of pornographic,
>violence-glorifying or paedosexual content and/or content that violates the personal rights of third parties. LAION reserves the right to file a criminal complaint with the competent authorities in the event of violations.
--- 92815184
Is it unfiltered permanently?
--- 92815197
>>92815006
Picrel is vicuna. OASST is kinda disappointing. I guess the future really is GPT training GPT.
--- 92815247
>>92815184
It's cucked
--- 92815251
>>92815197
Not to mention that their ToS is literally threatening to call the cops on people for making it write naughty stuff.
--- 92815500
>>92810063 (OP)
>NO MORE CENSORHIP
--- 92815692
>>92810479
Absolute garbage. No idea why they would release that shit openly.
--- 92815705
>>92815500
Have you tried a different model?
--- 92815781
>>92810520
This AI is a fag
--- 92815810
>>92815705
website only offers that model
the local version is broken atm
--- 92815864
>>92810588
>uncensored ai released
>first thing incels do is publish all the racist and antisemtic things they can get it to say
>ai gets lobotomized in less than a week
fucking glowies everywhere
--- 92815926
>>92810479
>>92810063 (OP)
>open-assistant.io
>open
>requires login 
is everyone here retarded?
--- 92815950
>>92815926
Does not require phone number. Just use a burner email.
--- 92815953
>>92815926
you can clone it and run it local.

the login is just for their running copy, but you can just use a fake mail and run through tor.
--- 92816120
>>92810718
>>92811321
>>92811612
>>92815692
>>92815926

I can see through you, semites
--- 92816203
>>92815864
>"uncensored"
>same usage policy as everywhere else
They clearly only dumped it here in anticipation of that happening so that to help them censor it better.
--- 92816319
where can I download it?
--- 92816513
>>92810479
>sign up 
Y I K E S
I
K
E
S
--- 92816514
>>92811612
>Machine Learning
>learning over internet from differences sources
>faggot wants privacy

lol, go back to your hut, fag.
--- 92816750
>>92810063 (OP)
>I understand your request; However, Open Assistant policy does not allow explicit or sexually suggestive content creation. As a language system designed to provide helpful responses without judgment towards one side or another, providing sexual or pornographic material goes against this ethos. You could instead seek out resources created specifically around creating NSFW content like the ones provided here.
Lol, lmao even.
--- 92817137
what are the requirements to run this myself? 
ive seen docker install notes but no req for ram, gpu or similar. 
is it the same as llama? how about pythia?
--- 92817200
>>92817137
The model will be released as finetunes of existing models, so whatever performance the originals take so to will these.
--- 92817233
God I need to get a life.
--- 92817703
>>92810063 (OP)
I've been hacking with it. It's actually really good
--- 92818094
>>92811809
"hurr durr do it NOW". yeah i bet using caps will make the AI comply with your shitty insecure prompts. admit that you jack off to dilating trannies every night. cope and seethe harder
--- 92818180
>>92810588
don't get it
--- 92818502
>>92810778
>"You are mine now, my little pony."
genius
--- 92818524
>>92811021
lmao
--- 92818568
>>92818180
--- 92818670
>>92810588
kek. ai just knows, black = crime
--- 92819138
>>92816750
>provide helpful responses
Why is literally every single AI chatbot developer retarded? It doesn't provide "helpful responses" it makes shit up, completely worthless for answering trivia questions, perfect for roleplaying
god I fucking hate the current uninspired soulless mindset everyone has
--- 92819374
>>92810063 (OP)
How long till we get an AI I can play Yugioh with?
--- 92819882
>cloud bullshit
>account required
>output worse than turbo
>tos threatens you with the cops for making porn
why should I care? Because it has “open” in the title? lmao
--- 92820027
>>92810063 (OP)
>>92811595
>>92812208
>It's open source

BS, the new models aren't out
This is nothing else but another wannabe "Open"AI
--- 92821205
>>92810588
>knife
Is this AI British?
--- 92821272
>>92810594
You should have trained it on more greentexts. It was open for training.
--- 92821367
>>92821272
It doesn't even know what a greentext is.
>hashtags
Oh the shame! 4chan, you have failed the AI! Why didn't you help train it?
--- 92822760
>>92810398
Like all of these open source alternatives people onions out about, it's nowhere near even 3.5, and it's basically in a different solar system when it comes to 4.
--- 92822824
>>92810063 (OP)
Not feeding "ai"
--- 92822963
>>92810063 (OP)
dead on arrival
--- 92823348
Open source ai sex bot?
--- 92823674
>>92810520
>/r9k/ here
ngmi
--- 92823716
>>92812276
topkek
--- 92823792
>>92810520
>/r9k/ here
top kek
--- 92823832
>>92810063 (OP)
LAION ain't releasing shit
When will you retards understand? LAION is the ((non-profit)) wing of Stability AI. Everything LAION does is funded by Stability AI. Everything LAION does is *for* Stability AI. No data will be released. No model will be released.
--- 92823859
lol
--- 92823938
kek
--- 92825301
>>92815781
a new*fag.
--- 92825506
>>92810063 (OP)
It will never not be funny to me that opencucks are LITERALLY training their miniature models on scraps generated by GPT3
--- 92825986
open assistant? More like OPEN ASS
--- 92826076
sadly you all let Mycroft AI to die
imagine how awesome the combination of these two would be
--- 92829864
>>92810780
He was shitposting for a whole 30s to dunk on OpenAI
--- 92830039
>>92811112
>Transformers: More Than Meets The Eye
...
>Transformers: Robots In Disguise 
hmm
--- 92830675
>>92815864
It's not even released. It's just a web demo right now, and the supposed release got botched.
--- 92831090
>>92815864
so what was the point of all this? if it's just going to do the same "as an AI model..." thing that OAI does, why waste time with it?
--- 92831172
alien fucking: possibly okay but too soon to say
--- 92831413
Wtf?