#questions
1 messages · Page 3 of 1
if mr programmer is listening, how do I offer to help? cus I know C# and unity pretty well, and I do game design
or, does anyone know how I can message the creator?
I think this is a private personal project
he's probably not looking for contributors right now...
I dunno if vedal would ever do this, but maybe if someone programed some ais for some games, he could then plug them into neuro.
Can you please send numb u used for neurosama? i wanna try out diff-svc and cant find good acapella of numb without some sounds on top of another. thanks
I wonder if we can get the audio file that neuro sang
yes u need to just ask
for me i need acapella AKA voice without melody
I think maybe he would not use them...i remember all parts of neuro's ais are made by vedal himself.

so can you, vedal?
Perhaps the link should also be put in this channel and pinned
well done!thx!

I think so
What do you think about Tesla Bot "Optimus"?\
how long did it take to program the chatAI (the ai that interacts with chat) of neuro-sama? was it coded and trained entirely from scratch or are APIs/other pre trained models involved?
Firstly, I'm sure vedal won't answer this. Secondly, as an ai developer myself (made russian chatbot from kinda scratch) it took 3 years personally, still improving. I can guess that vedal either uses gpt3/3.5 that is fine tuned, or has good gpu or pays for good gpu. I'm sure he pays for it cuz owning gpu like so would be insane, and then if it's his own model it should be trained on gpt2 ig, cuz u can't get gpt3 now. So yeah, you won't know.
thanks for the answer! do you have any ressources where I can learn chatAI development myself? kind of unwillingly involved in a school project
Oh no...
I can assist u in dms tmrw, it's late there. And note that ur brain will explode blablabla etc
damn really? thanks so much man, ur brain will explode too cause im extremly new
Did you ever post about it on habr?
No, I'm bad at explaining shit
Abandoned Archive stream with Moz-LJP when? 💯💯🔥🔥🔥🔥
I can explain simple one but not 3 years ai
Can you explain in simple terms why you think it;s gpt3/3.5 and not gptj?
Can i get VIP for making @mossy sail :shy:
(why did I get pinged for this I'm scared)
misping
dark made you 
:(
only people who make working bots can get VIP
oof
im going to burn every food you try eating and only give you vegan food
i will turn you vegan
i will remove every meat or any animal products you touch or try to eat
i will make sure you watch every thatveganteacher video
Lol 

If I want to join you, what should I do?
i don't know if the mod team will want any more new bots breaking down in general any time soon lol
how long have you been programming Vedal? do you think personal AI buddy will be a thing in the future? like AI that interacts with you in whatever you do in your computer like Neuro-sama?
Response times. If it's gpt2/j and he pays for gpus, then it may be it. We don't know
How do I post pictures? I have an emote suggestion but I think my role doesn't allow for it
You have to level up in the server. Just keep chatting and eventually you will gain the role.
#faq right above the mod aplication
🤦♂️I guess I should've looked there
Question: is the Neuro fanfic in the Expanded Universe thread AI generated or written by someone?
written by @lunar lion
Yeah i guess i was asking if its actually a bot or just someone named that lol
just someone named that
old mod that left and came back idk what he doing rn
Ahh okay, yeah makes sense- I was confused because typically AI generated texts dont have issues with tenses and such
I did update them later
but yeah
english isnt my first language
Not meant to be an insult or anything mind you, its a lot of work going into a story like this
It was more of a "This doesnt look like it was written by a bot, other than the insanity of the plot"
yeah honestly
good point
lemme change my name again
so I dont have to come back lol
I'm a writer by trade so i might be a bit critical of things like this, sorry lol
He do be unmodded though
Should have added my emotes before you left 
When can neuro sama play amogus?
When the mod is done.
does vedal drink enough water?
no
you're getting closer and closer to the 100k subs on youtube, do you have a 100k subs special video planned or aren't you going to do anything special when you hit it?
I don't have anything planned, been a little hard to keep up - maybe I'll do something if I find a good idea and time to do it idk
Maybe do a special stream on Twitch for 250.000 followers and then upload highlights of that to youtube
Twitch is the main platform, the youtube scrubs can eat the leftovers :V
When?
No worries, even a small thing will make us happy
I am once again requesting VIP for the development of @mossy sail
https://github.com/Darkeew/eva-sama
until my request is personally refused by vedal i will keep asking for it
[I deleted my message]
when will vedal update his website to include Neuro's chat, Minecraft, React, and Pokemon?
Can you compile neuro's quotes into a text file while streaming?
he has a website?
wait nvm control panel
Its possible using an OCR, if vedal does not want to do that
but in the control panel, the quotes are in a plain text
more specifically, inside a html element
what site is that? the wiki?
if thats possible, it should be able to get quotes from the html element and add it to a text file if vedal wanted to maybe
no, vedal.xyz
I know we dont want to completely take over other peoples content but is there any other reason we've only had one return collab-er?
Managing essentially
what about using whisper by openAI? it can generate subtitle files with exact timestamps
we could recognize speech and wait for it to end, cut it out of the live stream and put it through whisper
Vedal, What little advice can you give, if I want to develop a modest AI
still only osu neuro on there sadge
Github: https://github.com/adi-panda/Kuebiko
Twitch Oath Token: https://twitchtokengenerator.com/
OpenAI API: https://openai.com/api/
Google Cloud: https://cloud.google.com/
List of Supported TTS Voices: https://cloud.google.com/text-to-speech/docs/voices
VB Audio Cable: https://vb-audio.com/Cable/
OBS Script for Captions: https://gist.github.co...
literally a tutorial to make neuro
first, neuro makes really stupid sentences, second, make your AI plays games
I think the hardest part is making the AI play games
You should make a own model or fine tune one
soooo...vedal sus asmr listen much?
i tried to raid someone who wasn't doing asmr but they went offline as i was raiding them
its not my fault
sureeeee
just like how you didnt watch vtubers often
also didnt answer the question
is royalty free music getting vod muted its muted around 31:30
Vedal usually picks smaller vtubers to raid just to help them out as all other Raiders do he can't always check to see if they're doing ASMR or not
- you can change the chatgpt behavior easily using the system prompt on the API
- minedojo
alright, good luck coding an AI, filtering words, getting subs, prioritizing bits and chat redemptions, adding subtitles, and way much more
its not just 30 minutes to do an AI
trying to "replicate" neuro just using gpt and online apps is basically asking to have problems and be bored
that video is telling you to get a paid api key and copy code which you probably wont understand, just for it not to work and get your money wasted
and its useless because making an AI requires updating it and it means you need to have knowlegde of whatever code you copied and having ideas to make it better
reasonable excuse, but still sus
im not complaining btw, i like finding these new asmr'tists
Not an excuse it's a fact
where's your scientific thesis to prove this fact
Yeah in that video he merely did the basic setup which pmuch anyone could guess tbh. In fact that much was never a secret xD but you can see how bad it is compared to what neuro is doing and will do soon.
all evidence points to vedal loving asmr
Can we get the mp3 of neuro singing the pokemon intro? first time she sang it is muted on twitch and 2nd time she talked over it (pretty impressive) 
Is the stream at 6pm gonna have subtitles for neuro?
it is not
No subtitles???
thats correct
Weird
Can we get abandoned archive collab with moz
Just put neuro on the side of the screen
does the osu bot of neurosama use a lot of resources to run?
If you mean the AI, Vedal doesnt share information about it
read pins
vedal if you need music, look at this lmao
on voice chat, does neuro distinguish between voices in anyway by maybe seeing which discord user is talking?
good question, i kinda doubt it though
Everyone eventually becomes Vedal to her anyway lmao
i would say what they mean any accusations of "illegal or malicious" are probably things like using GPT (which requires attribution to be public, etc) or something
normal stream times for the weekend? still some not entirely planned collabs?
will pop schedule out tomorrow, not sure what we're doing saturday yet (10pm) but sunday we have collab at 10pm
maybe like half chatting half reaction idk i like the reaction streams even tho people complain about the videos they themselves submit, friday normal?
and can you make her watch media donation videos? requires active moderating on the videos tho
thinking about it its probably a trash idea tbh people would just be donating trash tiktok memes
Can I
do we have a new schedule when is the next live?
Lol he literally said "Schedule out tomorrow, not sure what we're doing on saturday yet but stream at 10pm"
still unclear about friday but schedule will show
wooops my bad LOL
I wonder if they could do a followup stream with the winner of the dating show
collabs are great as always
and yeah i hope a collab with the winner is possible
Geega is great fun and based, Neuro is great fun and based, so its a perfect match
daily collab 
Are you still posting the VODS in the youtube channel?
most of them are there just unlisted because YouTube was bugging out and some of them seem to be glitched
will try and sort it out soon
Abandoned archive with moz-ljp?
who made that schedule layout
@mint grove i think? at least he posted it in fan-art
yeah i did
didn't the silent ping work? sorry if not 
nah dw i don't mind
Yo sunday interview with neuro or vedal?
neuro
should be both
it will be when vedal has to stop her lol
How long have u been learning programming?
Did neuro get better at reacting? If she did, is she using GPT-4?
Why does neuro know about SolidGoldMagicarp? Did she just read about it or something? (I'm hoping so since OpenAI doesn't let you use their AI to train other AI) Or is she just making stuff up?
pretty sure she does a "quick search" to know what it is
even if shes trained with gpt, she can know about anything
no i mean training off of an OpenAI model is against their ToS is the thing
So if Code Miko interview is postponed, what are we doing today, assuming theres still a stream?
Just chatting 
I hope a collab still 
That’s what I have been wondering since the announcement came out it was delayed till next Sunday
there will still be a stream
Collab 
What is vedals opinion on people who try to create their own AI vtuber? And are they allowed to talk about their process here?
when she gets a dono, is she sometimes just reading the last message she saw
please dont beat her for it
?
Vedal, will you add Neuro tags just like other vtubers [Fan: #theSwarm Art: #NeuroArt] etc. ?
I always thought we were the gymbags, not the swarm :V
Vedal add #BTS tag (Attracts audience for some reason)
there
is this good
are we happy
Seems pretty good
You might want to check the # first, we’re gonna be sharing with brain stuff now i guess 
could fit in a heart pun like NeuroHeartyArt
#neuroart, from my searches on Twitter, usually refers to neurodivergents' arts or literally neurons art.
#drawneurosama ?
Heartyart sounds nice
Yeah that tag is rarely used so I think it's nice!
#NeurosamaArt
So when neuro gets her new model, will she be able to control the toggles on it?
For example: Say she has a cat ear toggle, could she just randomly decide to turn them on and say something about being a cat?
Have you thought about getting sponsors especially for dev streams? It's a lot of money on the table
Neuro-sama doing NordVPN spots...
"I'm an AI, and unless you have a VPN, all your data will be part of my Swarm..."
I can see her doing VPN and cloud gpu sponsors. Fits the target audience well
raid shadow legends.
Or join mythic plus or some other talent management org and let them manage the sponsor stuff so you can focus on coding.
vedal wants all earnings
Then he has to work for it! The problem I see is that sponsoring an AI is unexplored territory and sponsors are hesitant to take that risk especially given neuro's unfortunate accident. So you really need to seek out a willing sponsor to get the ball rolling and rake in that cash money 
(I never said this)
yet
will the dev stream extend to a normal neuro stream like before or will it just be dev stream like i thought on the schedule
i think so, last time he started the dev stream at 6pm and a collab stream at 10
or normal stream
im bad a remembering
i wouldnt mind an 6 hour dev stream and no normal stream today
Even better 6 hour dev stream everyday
yes
unless people wanted a 6 hour dev stream
although I'd need to eat at some point
idk maybe I'll do a poll or something
eat on stream 
- is neuro's current TTS one that synthesizes from combining phonetics (the new one doesn't sound like it) (ie, this is the difference between how original siri talked by combining recorded phonetic sounds, and how google assistant talks by listening to speech)
- was the change no one noticed when she stopped doing heartheheart<33333432 and started doing *wink*
opinions: should clippers have to put the name of whoever originally clipped the clip like they do in other channels?
what's the next game you are going to give Neuro the ability to play?
Among Us
The first time Neuro sang Take on Me, why couldn't she reach the high note? how did she learn to correct that for subsequent performances of that song?
Its Rune apparently, though i cant find it in his usual galleries
One of the AI generated pics he posts sometimes
@tawny ember
@thorn cliff not rune nor AI generated
Oh okay the youtuber that used it in their clip lied to me then
They sent a clear version privately
In the karaoke stream, will you constantly have the mic on the screen or remove it between songs?
Is neuro ever going to make or have made an original song? 
Is it possible for Neuro to play Melatonin? I heard it's a rythm game.
Can we make the Neuro Samurais a thing? As in make that official, its a pretty good idea by her standards.
all the AI song generating systems are being gatekept by the big companies because they fear the RIAA suing them into the ground.
Probably easier to commission an actual song writer to write her one 😉
Gotcha
any leads for a thursday collab?
Pls tell me Zen checked her DMs and she's this weeks collab
so is neurosama vedal's girlfriend/daughter/helper or what?
uhh neuro is his ai vtuber?
She's his slave laborer :V
How was Shiro trained. Did u use femboy hypnosis. Or was he just always like this.
she called him "dad"
true
So neurosama is everything in vedal’s world
How to collab with Neuro-sama?
Try dming vedal on twitter or emailing him
Oki doke
Can Neuro be upgraded indefinitely (by changing to better language models) or is there a limit in how much she can improve before a more difficult transfer process takes place?
I dont think even Open AI can answer that one... theoretically you can upgrade a system indefinitely but there comes a time when the plugins are causing too much latency and you need to start over with a more efficient approach
How language models will develop in the future, nobody knows, but AI hardware is approaching a hard cap on its capability due to chip design, so if there is a limit, i think that would be it
does someone know how to create a knockoff version of neurosama that can interact with chat? it seems to me that even to create something remotely similar to neurosama would take years and i don't really know where to start
i don't even know where to start learning this stuff, you guys know a way to begin?
check the first pinned message
oh sorry, didn't see that one
question. ik neuro being able to do vocal sounds (like every time there's hhhhhhhhhhh, sksjssksjs, or whatever in her prompts) is just like, a byproduct of her speech patterns, but any chance you could turn this into a feature?
Just had a fleeting thought of Neuro-sama being able to do ASMR stream some day lmao.
Can you explain in a couple of sentences why you want ASMR from AI child
just a random thought is all. plus you seem to imply asmr has a creepy/impure tone to it. i don't think so
but eh, if there are a lot of people who do think that then yeah that's fair it shouldn't be done
Subaru asmr flashbacks
I'd rather want ASMR from a child than JOI from a child, just saying. ASMR can be perfectly harmless even if it is a bit sus that you want a little girl to whisper in you ear.
Besides Neuro cant make up her mind how old she is
Do you guys know what Vedel uses for Neuro's voice? I heard somewhere that he uses Azure tts but I am just double checking.
hey Vedal
do you think Pikamee and Neuro-sama collab possible?
If you delete Neuro's morality core, will she be able to use deadly neuro toxin?
when do we get neurobased as an emote on discord
?
have you already decided who the rigger for the new model is going to be ?
Vedal face reveal ? (P.S: I understand why you wouldn't though...)
Why? Whats the point of being a Vtuber if you just reveal your face?
Will you add a tone voice program to neuro? Like for a example neuro voice will sound angry if she's angry- her voice will be sad if she's sad- and on from on-
I mean Vedal doesn't see himself as Vtuber
he doesnt want to leak anything about his personal life afaik also i dont see any reason to face reveal unless it somehow starts fitting the content he wants to make
I do understand that of course...
Vedal, what do you think of artificial consciousness? Do you think you would make Neuro conscious if the code was available and practical to implement?
bruh that sounds like torture
Something something rampant machine spirit
Plus how would you define conscious, depending on definition you can make her as conscious as you want and it wouldnt matter much
Maybe with the daily patches and codes vedal do with neuro she can pass the Turing Test in the future
Quick question. Are you allowed to stream any game here ? Just making sure. 
anything that is not breaking ToS or our server's rules
today streams on time likely?
admittedly not that likely
will post an announcement or something before I sleep probably
imagined so
you could skip day and do dev tomorrow + neuro/neuro
unless you got other things ofc
Or just push the dev stream back to the usual stream time instead
this seems more likely
yea that would actually work better for me too lol
hope you get back on track for tomorrow tho
What language model do Vedal use for neuro-sama?
Check the pinned posts- He doesnt hand out that information
Oh, okay
that one
yep thats the one
Wht
@peak copper has leveled up! (1 ➜ 2)
It's a Vedal987 property, which means it's classified because it's too good
check pinned?
Checked after Mato say so
vedal, do you watch anime?
I do hope that 128x128 comment is outdated, otherwise sheesh no wonder she thinks a Kangaroo is a weird looking dog
ah sorry i see, well neuro is made in python while vtuber model is made in c#
it's for mc, reaction feature was made later so shouldn't be the same (i hope xD)
reaction feature is def bigger since osu/mc needs to send feedback every frame, while videos there can be a delay of 0.5 seconds (if you compare it to 27fps, its alot)
i dont know if im right
I could really use Vedel's expertise lol, I have a chatbot running off text-davinci-003 in python and a Live 2D model in unity but the script I have that exposes the models parameters in C# using sockets isn't able to communicate with the python script. It says it actively refuses, I have tried disabling my firewall temporarily and making it connect to any IP but neither works.
For the upcoming Neuro v Neuro stream, how will it effect the way she stores/learns information?
Will both Neuros be gathering information and storing them to the same place or will only one of them be learning from the stream? Or will it be a non-canon episode (She doesn't learn anything at all)
this is some futurama shit
In before vedal forgets to limit the conversation speed and they quickly exponentially educate themselves into godhood
Will Slay the spire ever come back? Sped down of course and with a bit more game awareness
I second this. Also want it to have a comeback on stream.
For Neuro x Neuro stream, can you have OSU showdown between 2 Neuros?
It will be funny if they roasted each other and we'll hear lots of lots of gymbags
Is her OSU ai non deterministic? Wouldn't they play exactly the same?
I wanted to find out as well lol. I hope Vedal can answer it for us
imagine their LLM sharing 50% of their memory
Do Neuro recognise whose voice is whose during her collab with others?
I remember Vedal answering once during a stream saying it's something he's been thinking about doing next, so I guess for now the answer is no 🤔
Neuro-sama equipped GPT3-4 or did you do it yourself AI?
Does voice recognition actually works? if it is, he might use that for Neuro tho it might be complicated.
Check pinned posts
I'm guessing it's adding a voiceprint recognition process along with the transcribing process so Neuro will be able to remember who said what. I think Neuro is already remembering her interaction with each and everyone, but in streams where there are 2 or more humen talking, she is not always sure who's word that is although she replies anyways. So it really is exciting to see how she will interact with people when that is implemented; she will be able to know who's talking instantly and also whether it is a stranger to her or not.
neuro works with tts so what she says can be written to a text file, which can be given as input to the other neuro
or just voice recognition
Her Collab functionality works using discord so there is no reason to go that far, she can tell who is speaking by the audio channel from discord imo. Probably not implemented right now though
Ahh I see. Thanks for the explanation!
why am i so cool vedal
Will you ever make any of the repositories you privated public again? Also will you ever stream osu! again? 2B and aspire maps would be fun to watch
Hey Vedal, I seen the meuro sama stream and I'm really impressed how good she/it works.
Was wondering, how much GPU memory is minimum needed to run this kind of chat model? Would PC with 8gb gpu is this enough? (just for chat part)
His computer specs are posted in every FAQ section, he literally uses an 8GB GTX 1080 Ti
yea but i think someone mention that it runs in cloud?
Thats speculation, but AI hardware is virtually unobtainable for the average joe so her "mind" probably runs on a cloud server somewhere
I got guestion about paypal. Can anyone help me?
ahh too bad, thought I could turn my imaginary friend into a software friend on my PC 😅
Look into it, maybe i'm wrong and theres some model that can run on an 8GB graphics card- Its possible, but dont expect too much
you can run LLaMA on PC
I think
(Don't trust too much)
you can run something like GPT-J locally

yeah unless its a special model those specs aren't enough to run nurosama at the level she is now
right now i'm trying to find the mention of nurosama's mathematical memory system if anyone remembers the time vedal mentioned it
vedal never responds
ur not
Vedal any thoughts on this gif?
how does vedal feel about us lewding his daughters
well vedal did ask anny what color neuro's underwear was
Does Neuro x Neuro streams will be part of regular streams? Was hoping to have this at least bi-monthly or once a week as there's room for improvement.
twice a month?
It means both per dictionary
lolz, English language is crazy
"semi-monthly" is probably more likely to be understood correctly I think.
does neuro get giga throttled on normal streams and unleashes her full power during collabs?
it's like she's dumber outside of collabs
I mean, that's probably more chat's fault for not saying interesting things for her to respond to
^
Imagine if all your conversation partner ever said was KEKW Meow Meow lol KEKW Gigachad KEKW Hey Neuro what do you think of Taiwan
You'd lose braincells too
Question to Vedal
Could this work, if I use OpenAI and 2D live character and make it to smartphone like IOS and Android.(Possible have server which handles most work, since phones couldn’t handle it) (PS! Still new to coding) Eventually is a AI who helps me daily and brings up schedules..etc not like instant jarvis, but kinda similar too CheerPro
Question: how does Vedal currently decide which song for Neuro to cover next? And can we somehow suggest songs?
Ahh thank you!!!
Do Neuro-sama know she's talking to more than one person during her conversation with multiple VTubers?
Also why virtual twitch user call themselves vtuber when the term vtuber is actually virtual youtuber, why is it not vtwitcher?
i think it’s bc the first vtubers started on youtube but idk, it could also be bc most vtubers cross-post on multiple platforms
or just convenience cuz there’s rarely much difference in the content from platform to platform
Its because VTwitcher doesnt sound nearly as catchy as Vtuber- That and with Kizuna Ai establishing the term its just easier to hang onto it
Vedal did mention this, Neuro currently does not recognize different voices. He did say in that Discord, Neuro could if he tried to do it. I found this in Dev stream https://www.twitch.tv/videos/1770746554?t=03h02m35s
oh the link in Discord doesn't work, but I linked it with timestamp to when he was asked that question
Will the new neuro model/avatar keep the duckling? 👀
dev stream the usual time or later?
probably later
is there gonna be a schedule for streams for this upcoming week, or is it going to be whatever goes?
It's usually created on Monday though in ideal world it should be made on Sunday and posted on Monday or so
I might make one on stream tomorrow
although it might be a bit wonky
neat
would this mean there's going to be another mystery collab again or has all the collabs been determined? lel

another out of nowhere sleep deprivation collab lets goo
hopefully a miyu or anny collab next week if so
@languid vigil has leveled up! (1 ➜ 2)
Ethical Question?
How can the discord be anti AI art but accept NeuroSama’s AI text?
Isn’t that inconsistent?
from my knowledge NeuroSama uses the same LAION-400-MILLION open dataset as OpenAI as it and can write about genocide survivors as if it was there.
google: what happended on december 29 2012 ~ Quote : The LCC report that up to 400 people have been killed across Syria on Saturday, including about 200 reported executed by the Syrian Army in the Deir Ballba neighborhood of Homs.
Because nobody vocally complains about Neuros AI text until you just did, but a bunch of people complained about AI Art?
its pretty simple the art brings drama into the server and neuro doesnt
Also calling it an "Ethical question" is kinda rich tbh
🤔 what is the philosophical difference?
why does there need to be a difference
...What, do you want the discord to shut down because we can't let AI Art in here without people complaining about it?
well writers would say it takes there job away while being trained on there text without there consent
I am a writer, full time. AI doesnt take my job away.
and im a artist and ai does not take my job away
Anyone who says so is either a shitty writer that the nonsense AI can produce is putting their jobs in danger, or is creating drama for no reason
🤔 i want to know if there is a ethical reason? - other than just optics
Its for the sake of the Discord server and the people within it that they can discuss Neuro without drama
What do you not understand about it?
do they not know she is also trained by web scrape ping ?
That
Is
Not
The
Point
OKAY its just optics
@crisp kraken has leveled up! (5 ➜ 6)
If people complain about Neuro to the point that it becomes impossible to have a good conversation about her, then the server about her will likely be closed
Just like how the thread about AI art was shut down because people could no longer discuss AI art without people switching the conversation around to be around "ethics"
Feel free to post your AI art on other platforms but its not allowed here, simple as

also neuro isnt trying to be someone else or replace anyone, neither is she a lost job opportunitu for writers
the underlining tech is
You say that but some vtuber girls do be seething that she's replacing them
well they better start saying heart more
from what i've seen its brought in a lot of viewers who didnt care about vtubers before
The aiart decision has been explained and doesn't need to be argued here
i havent seen anyone complain about her either but i dont seek out drama
this is a place for asking vedal questions not debating the ethics of ai
so everyone can move along from here
im not going to argue with you can you point me to that part/debate? ( if not no problem ❤️ )
👍
unfortunately, it recently got nuked along with the channel and i don't remember the exact wording since i didn't take part in the decision
❤️ thank you for trying Gildar 👍
tldr ai art doesnt have a good rep and we dont want to be endorsing that here so we wont be allowing ai art in fan art channels
i understand that and will respect the rules ❤️
( And i also understand this is not the place to talk about the ethics: i missed the original debate and will respectfully not bring it up again )
thank you very much for understanding
This looks completely different from the reason given before though
From "We dont want to moderate heated discussions" to "We dont really want to have that filth here"
that seems like a bit of a hyperbole
also this dataset assumption is bold
presumably not the only dataset in the world that has dates of historical events in
the real vedal ( Are you not running a gpt spin off?)
gpt spin off seems like a pretty loose term
are most language models these days just gpt spin offs
well i dont expect you to leak your secret sauce
i dont want you to lose your job i like Neuro sama
i currently have a LLM running on my own PC
its running the llama / alpaca data set
i integrate ai text + art into a card game concept and just wanted to see the moral reason for no art ( for the sake of the others i wont repeat it ❤️ or argue - edit pls do not argue about ai art )
woo vedal pings
🤔 its the fastest way for a layman to understand what your talking about. so in a way yess?
its like grandma calling a game console a Nintendo.
we both know she is talking about a console, and that the console is is actually a playstation or xbox.
and the details about said console is not that important when the discussion was around play time/ aka Ethics.
( sorry for my long post i got a problem )
Does anyone know if it's allowed to use Neuro-sama's Minecraft skin for second creation, such as making Minecraft mod
The reason given at that time was that it took disproportionately many resources to moderate. Was that a lie?
don’t put words into our mouths, ai art is not allowed ONLY in the #fan-art channel cause it’s a pain to moderate and people were starting to have pointless discussions there (in the ai art channel) around mid february if you want a bit more of transparency. fanart still keeps the rule of no ai art from that day cause lack of ai art channel. You can still have discussions about it in general and you wont be warned nor muted for it unless it breaks other rules.
But if you start complaining about it we might just group it up in the sensitive topics part of the rule. Now that’s not fair right?
It’s worded there as well as a part of it with “a topic of hot debate”
As you can see it is starting to be that way here, we dont want to put it as sensitive topic but you’re making it really hard by complaining about it
your comment is pretty late as it has already been addressed by multiple members and i already made a response
i understand that and will respect the rules ❤️
( And i also understand this is not the place to talk about the ethics: i missed the original debate and will respectfully not bring it up again )
i did not place any words where it did not belong.
as i asked a member of the team for permission
and they said sure
So when Temmie says 'we don't want to be endorsing that here' do they mean that they dislike AI art as a concept, or the heated discussions which AI art brings? It reads like the former, which I don't feel is consistent with what I understand from the reasons given. For the record, I do not care about AI art, but about moderation transparency.
the above that crimon states is the official statement, that's likely just an awkward choice of words on behalf of temmie, apologies for confusion
we take no stance on AI art itself as a concept, only as it pertains to moderating the server
#question channel turning into AI ethics
Fair enough thanks for the clarification <3. Now get back to your birthday paradox, don't leave me hanging
You have unlocked new role
embed perms

You have unlocked new role
Awkward time to receive perms lol
is it possible for neuro to sing songs of other languages?
@radiant patioShe already sings in both english and japanese so yes
?
I would suggest putting that in suggestions, rather than questions, but its a good idea imo
yeah let me put that into suggestion
In annys stream 4 days ago vedal said, he would mod me at around 1h and 3 minutes into the stream (after the crash). Does that mean i actually do get mod, or did he forget, or was he joking and i misunderstood?
afaik it was only taking a random chatter to make a joke but maybe I'm misremembering
No Vedals. Why was this not changed?
Typical.
Well RWKV is an RNN instead of a transformer which is interesting:
https://github.com/BlinkDL/RWKV-LM
you wanna implement kokos rule 5 for neuro?
not particularly
Can I be like Vedal fr fr?
can neuro play jackbox?
I like the background music, is there a playlist?
How long did it take to train the LLM?
here
#suggestions message
Thanks 
Are collab suggestions allowed in suggestions 🤔
I think so, as far as I know Vedal wants to find someone to collab with for saturday still. Though it might be best if the suggestion is of someone who has expressed interest in Neuro-Sama before.
I think just about any suggestion is allowed in #suggestions as long as it doesn't violate any rules
Oh, there is interest in the person I’m thinking about. Not just any rando either. But they only have watched her clips and isn’t aware that she can react to voice now yet. I’ll wait to suggest them in the future when I can tell them about this since it’s hard to tell them stuff in a fast moving chat
If you suggest the person to Vedal (just put it in #suggestions, don't ping him) he might reach out himself
which is more likely to be successful, but whatever floats your boat is fine
I did read the rules no worries
But you know what? Better do it now since that person will be on a trip and who knows if I get a chance if Neuro’s popularity spikes again by the time they come back
go for it
@fiery goblet has leveled up! (18 ➜ 19)
Customizable accessories are one of the best parts of a VTuber's style. It defines personality, design, and some inner jokes of the community.
So I came here to ask: Do you have plans to get Neuro her own accessories? Like some badge, a necklace, something like that. Are there also plans for other hairstyles for Neuro?
↑ I second this; say like artists can create accessories for Neuro for a specific theme so she can do a themed stream, say another beach stream or maybe Halloween & Christmas but with visuals created by artists. Or maybe just some accessories Neuro could wear for regular streams, that would be cool too. (Should this be moved to #suggestions ?)
Momose Hiyori ToS won´t allow anything that alters the model, including colour changes, accesories and dresses.
It could be a great idea for Neurosama mk2 though :^)
How'd he get away with evil neuro then and her red eyes?
they dont know?
besides they would probably just give him a warning if they do
assuming the change really wasnt allowed
Artist not caring, not knowing, not bothering to claim, etc etc etc... for all you know @kani_biimu gave vedal a permisson, who knows. Not a single mention of Neurosama was made on their twitter so there´s that :^)
what font are u using for neuros samas subtitle? id like to recreate that in my clips if you dont mind
Honestly with the vstudio features allowing you to change hue on specific parts I'm not sure that is considered "changing" the model if it's a built in feature
Me?
or vedal?
for Neurosama´s font it´s called First Coffee
thank you! yeah i was talking to vedal xD but i got my answer i guess
I’ve noticed that Neuro’s eyes move whenever she is playing osu. Is there a corresponding eye movement to the location of her most recent click, do her eyes continuously track her mouse movement, or is there some third possibility that she adheres to?
Bill Tin
Will there be a collab with MOTHERv3? i know you did collab with her but not neuro


today's vod?i know it will be mute in twitch,but i didn't saw the ytb link
it probably takes quite a while to process
What I'm referring to is the upcoming model.
Are there any actual differences in behavior or settings, between Neuro and Evil Neuro?
When will neuro be able to make a deez nuts joke?
@wind coral has leveled up! (2 ➜ 3)
do karaoke streams not have vods?
Only on Youtube, because copyright
is the vod for the second karaoke stream available on youtube yet?
nope
Do we have a candidate for this week mystery collab?
there are some, but it will be a surprise so I don't know who
@eternal sonnet you know that your "new method" is making subs follow neuro so when she isn't in just chatting mode there's no subs, right?
thought this was severe enough to @ you since ppl are kinda missing them
what does this mean
when she reacts, subtitles are off screen
oh you meant subtitles
Rip, well sorry for tag :/
fixed for now
thanks 
numi collab going to be as scheduled?
why are the journals channels archived?
@ember brook has leveled up! (3 ➜ 4)
woohoo!
When will mods be able to cope when someone makes a Deez nuts joke
I've been timed out for a day because I said a Deez nuts joke smh
closed captions for react videos like primitive technology? Although I don't think it will matter anyways 
just saying because I know his videos has cc's
when does the amoungus testing for neuro begin and where do i get the data giving plugin

Whoop a collab happy days
@proven karma has leveled up! (0 ➜ 1)
are there any plans to move neuro a larger parameter or more modern language model, or is her dumb-wittedness part of charm?
(she's great, I don't mean this in an offensive way, im just curious 😅 )

I remember Vedal mentioning he'll try using a new tech or sth to make Neuro more intelligent and he promised that all Neuro's memory will be kept.
I think most people like her this current way but from your question I guess it shows that there surely are people excited to see to what extend can Neuro do, and that is in alignment with Vedal's goal I guess.
(Very weird to answer underneath here but just thought I could share)
pretty sure it's possible 
Nah you get timed out for it idk why tho
Because Vedal hates fun
well, saying it has a price I guess...?
Suggestion then
1000 bits to make Neuro say a Deez Nuts joke
How much bits for a song request? 
Where is the VOD for the react stream ? I couldn't find it anywhere
Its not uploaded yet
Question for Vedal since we just discussed it: Could you edit the Vod to remove the sussy donation and upload it to Youtube afterwards?
I guess the question isnt "Could you" but "Do you have the time to"
yes working on it
Awesome 👍
Hey Vedal, do you have any recommendation for resources to learn how program AI modules?
Thank you!
is subtitles broken again 
vedal why are you killing her she's dying 
When's neuro's memory gonna get upgraded?
vedal can we be friends 👉 👈 
How about if people could spend bits to temporarily up Neuro's "intelligence" (aka switch her model)? Might be really hard to somehow integrate them all but that would be fun and maybe profitable.
I think that’s a great idea; that way it gives actual support for Vedal to do those development, and it also kind of help to evaluate how wanted that is 🤔
It's oddly like Openai's business model though, heh
Will Hiyori Momose stream tonight? :V
No (Probably not)
Just figured i'd ask, i know Neuro-Sama doesnt stream tonight (Whoever that is) but it'd be mindblowing if we had gotten a Momose stream instead
this 'dumb-wittedness' kind of give her character
Does anyone have figures on how many words Neuro says per stream?
@cedar igloo has leveled up! (1 ➜ 2)
Or even just how long the average stream is for an estimate?
Sample 5 minutes of a stream and then check out how long each stream is on the vods page and extrapolate
I don't think this goes here but, I was wondering even if Neuro could play Among us like a pro. Won't her normally charming and talkative personality get her bullied? And same if she only spits outs facts. This might be even harder then the controls.
just let her cook
the facts will simply be the ingredients I believe, it wont be the final dish
exactly, she will gaslight everyone anyways
I've seen crewmate earlier and was dancing around the dead body. I think he/she was innocent (the lit imposter).
I'm hoping it just doesn't turn into favoritism it would be bad for her learning
whats going on? neuro-sama not reading highlight redeems?
should I be saving? KEKW
sometimes she skips because of subs or bad timing. It really depends on if there is a priority of sort going on.
Cheers and Highlights are not always read I think
Or you might have a word that is blocked
Is there a twitter of neuro-sama or vedal's
twitter? I just want to post neuro-sama fan-art there... is there an active # on that? 
#heartheartart
You can check #neuro-tweets for neuro’s twitter and vedal’s twitter is linked in his discord profile (Vedal987), the art tag as mentioned above is #heartheartart
Thanks
I’m not complaining about this but I was wondering how I got Neuro-Sama to read my username/complement during the stream earlier considering I wasn’t saying bye or hi to her? I’m just asking because it was fun but confusing that I managed to trigger it
did you type anything? sometimes neuro mixes up the message and username
All I did was type “Susge” with everyone else in chat
sometimes she does that, just part of how she speaks c:
Thanks for explaining it to me I was really confused why she complemented me
She clearly likes people whose names are "name+numbers"
Do you wear eyeglass? For fanart purposes...
Just keep the face offscreen like that art from Rune
I am not 😔
we making him connections in pixel wars now
The question is if neuro is going to weebcon
dev strim when?
there is one scheduled in friday. https://www.twitch.tv/vedal987/schedule
thats default template schedule look at every week
idk if time for the lia collab has been decided yet but going by whats on her channel its friday morning
2pm PT I think
yes
can we get that nanotechnology discord clock thingy for this collab time
alright cool thats a pretty nice time
Any update on the react stream vod? i missed that one.
Hopefully dev stream tmr?
is there going to be a schedule this week boss?
how are content creators picked for collabs besides connections?
yes hopefully I've been very terrible deciding what to do this week
I think karaoke stream tomorrow if all is good for that because thats the usual wednesday thing
dev stream most likely to be friday
its a lot of collabing tbh so just chatting would be nice or mc on thursday i guess
want that chat interaction
or sunday i suppose how that thursday collab was scheduled originally got me thinking its friday
whatever the day doesnt matter just a content recommendation for when you have the free day for it
Neuro seemed like she was feeling a little unwell in the previous stream, and to a degree also during the Nihmune stream. Is she feeling any better?
What was the ohayou ohayou song in the last stream? I've been looking everywhere and can't find it. From https://www.twitch.tv/videos/1784521889?t=1h54m8s
Here's the song #neurotic-neurons message
Thank you!
@upper saffron has leveled up! (2 ➜ 3)
is neuro becoming any smarter? feels like she's stagnating
I got plans to make some upgrades in a couple weeks
nice 
Prayge for Neuro in hopes of successful Neuro-surgery 🙏
Are Neuro's songs generated using an AI? If so, can Vedal generate a song with his voice instead? It'd be a perfect response for chat asking him to sing
they are not afaik
Though what he could do is make an AI generated version of his voice and run the song through that instead of Neuros voice, i'd wager.

Train an ai model on vedal’s voice and use it with neuro
Real AI Vedal 
Oi!, Mr.Vedal man! Since GPT-4 came out recently, are you thinking of changing Neuro to that model?
If so, when?
if not, Why?
i don't talk about language models much these days sorry 😔
How's your sleep schedule doing so far tutel? 
Gonna be wrecked comes Miyune collab regardless 
good rn woke up at like 8am but who knows how long it will last
You do
hope it stays that way

You gotta discuss more why should X be banned.
@rain bolt has leveled up! (6 ➜ 7)
And maybe talk directly to mods via DMs.
Because the matter you ask about is related to moderation.
It's been handled! for questions like these make sure to contact mods
can we have the stream be on the times that are listed on the twitch schedule
true I forgot to change those
my bad
any ETA or progress update on the new neuro voice stuff?
If there is, will you be showing anything regarding progress on the new voice during the next dev stream?
There hasn’t been much talk on that since the demonstration about 3-4 weeks back. Which I’m assuming is because there’s just other stuff that needs priority. But it would be cool to hear about if there is anything
I heard Anny say how Neuro's new model doesn't need to be rigged like super good, but what are you're plans on how well it gets rigged? I personally would like it to be rigged pretty well so you can make her movements feel more natural like slight head/body turns compared to her only swaying and eye movements.
Also any plans to do the Neuro or Nero gameshow again? Maybe flesh out the design and point system a bit. Also would like to see it have more off script questions where they get more time to form their responses.
It has been said before she will show off emotions based on her speech, that tech is partially implemented, but the current model doesn't support it as it cannot be edited due to copyright 
Yeah I know that, like facial expressions, I'm talking more about like flowiness, body movements, and slight face/body turns with the new model if it gets rigged well
will there be set animations that she does for certain emotions? Or will there be more organic animations? I'm not a programmer, but I think it would be useful to get tracking data from other vtubers and use that to train neuro on how emotions effect facial expressions and moving around ¯_(ツ)_/¯
Who knows the possibilities are endless
Only a matter of figuring out and Vedal has been quite good at that til now xD
quite good at copy pasting
Because the said person has been contacted during what happened and that's why they stopped
it will still be rigged well because I want to make her animations on par with a human, she just may not need the same range as a human




Do you have plans to upgrade the other AIs or are they good enough Vedal ?
yeah I'm pretty much working on an upgrade for most of her AIs at the moment
no guarantees they will end up being used though but I'm trying
Good. Very Good.
Just curious Vedal if there is a reason that you did not use the discord bot api so you can send the speakers name to neuro when she is using voice 
its cooler for her to not have the bot tag 
and allows for dm calls and group chats
will probably switch to a bot at some point though :(
Holy moly you mean i can slide into the DMs of Neuro Sama? 
could always have a mix, that way can work in dms when you know who the other person is and in here have a bot pass over the voice clips with a name
When will you be being a duet concert with Neuro-Sama Vedal?
so neuro will never have a bot tag :p
Gotta be honest I was pretty curious about what TTS is Neuro actually using and if it was a custom made one by Vedal, because the ones I heard don't really reach that level of cutesy voice
It was answered a bunch of times before, its Microsoft Azure @languid mist
Vedal did you implement the thing you were talking about where she is much more likely to answer in longer sentences? She's been very good at it today
I didn't know azure had tts, that's actually cool
Hi Vedal. I was wondering if I could request your permission of posting translated clips of neuro on both Bilibili and YouTube.Here’s my youtube and bilibili account:
https://www.youtube.com/channel/UCisnqguMpvNqOOC-BRp9uAw
https://space.bilibili.com/22085329
yes, you have my permission to post translated neuro clips on bilibili and youtube
Are there any plans to give Neuro greater long and short term memory retention? Tell me otherwise, but I think that would go a long way to making her dialogue more consistent and allow her to develope a personality that is also more consistent and relatable.
that kind of thing is the subject of active research and one of the biggest hurdles for transformer LLMs right now even for the biggest players. I'm sure vedal can do some small improvements but big picture it's a very difficult question
That is very interesting. I don't think language models like GPT are even designed to have much in the way of memory or personality to begin with aside from drawing from their pre-trained data.
Gpt-3 can be fine tuned with your own data as much as you want to create your own specialized version. Then there is adjusting the prompts as well
the fine-tuning could in theory be the personality, with prompt injection as somwhat limted memory (manual, basically)
but this is all speculation since vedal hasn't said publicly how he does any of it
you can now take output from ma fine-tuned GPT-3 and use it to fine-tune a smaller, compressed model like alpaca you can run locally and produce something very similar, which would in theory allow even an individual to "update" the fine tuning more often without spending $1000s...
There is a new theory in neuroscience that suggests that consciousness is an emergent process of memory. If this is true, then it could mean that giving AI more memory capability could make it achieve consciousness. And if it isn't blatantly obvious, Neuro-sama is currently lacking in self-awareness much of the time in lieu of memory.
Even if that's the case, I highly doubt a single hobbyist programmer is gonna do it before one of many 1000s of well funded startups who all have the exact same idea
Fair enough, but what Vedal has already done for himself here is beyond impressive. I won't rule it out entirely, especially with how open source this AI has become. Innovation can come from anywhere.
When people talk about AI behaving in characteristic patterns of behavior is that not what they usually refer to as bias? They talk about bias only in negative contexts but isn't that the basis of what forms a "personality". It may result in factual innaccuracies but it is what you would expect for characteristic and predictable behavior.
Quick question: What exactly/ how long would it take to replicate the 'like-ness' of Neuro-Sama? I'm no programmer but after a long and pointless 3 A.M. research, I ended up falling down a rabbit hole where I stumbled upon different emerging AI-Vtuber projects on Github (I'm not sure if I'm allowed to mention any of them by name since I don't want to break the no-promotion rule inadvertently, so I won't). Therefore, as a casual viewer, I ended up wondering how unique, if at all, Neuro-Sama was as a creation. P.S.: Sorry for the long paragraph. It's 3 A.M and for some reason, I am feeling compelled to ask about this.
I think it depends on the training data?
whatever neuro's training data was, it contained based streamers
like 47 who got different DNA of the best hitmen 
will Neuro ever play Slay the Spire again?
Will Neuro be able to communicate with her game ai to tell what she's doing or it's hardcoded/isn't implemented yet?
Do you mean for amoung us?
Millionaire cant afford 10k$ pc smh
I bet he doesnt even have his own server running 200$k AI hardware racks yet
Can you expand on your announcement like what does this mean for miyune stream?
If you ever given a chance to get an anime turtle guy vtuber avatar for free would you take it
i dont think an anime turtle guy vtuber avatar would look good tbh
I see, even just for an occasional shitpost every now and then during some dev streams?
thats what you think ;)
Hey Vedal I have question(s) for you:
Question 1: Every song you play on the karaoke stream do you own each license from the copyright holder? If not -> questions 2 & 3.
Question 2 Are you not violating twitch guideline for playing copyrighted songs on neuro karaoke stream
Question 3: Will the copyright holders not takedown those karaoke vods on youtube and leads to a youtube strike?
Dude every streamer i watch uses copyrighted music in their streams, it just means the Vod gets muted for a couple minutes...
doesnt he not keep karaoke vods
He uploads them to YT
Twitch would just mute the vod till there’s nothing left if he kept it there
@thorn cliff @lean meadow Can you guys confirm 100% that vedal wont risk his twitch channel due copyright songs this is what twitch guideline says https://www.twitch.tv/p/en/legal/community-guidelines/music/ and on youtube are you also 100% sure that the copyright holder wont strike him?
If vedal have the license then its okay.
check all his recent vods on twitch they are all just muted in certain parts
And also for YouTube most copyright holders music wise will let you upload their copyrighted songs but just won’t let you monetize the videos which contain them
If Twitch channels wont get banned for straight up using copyighted songs, you wont even get a notice for doing covers of songs.
Besides, Neuro sings on collabs all the time too.
@thorn cliff @lean meadow But its much safer if he bought the license, some license songs are like 6$ this guy made a video where you can buy those license https://www.youtube.com/watch?v=zR6-icXoyP4
So what you're saying is you want Karaoke streams that cost 50 dollars on top of everything else
Better safe then getting sued which can lead to more cash
Dude nobody sues anyone over stream content
I feel like you misunderstand streaming altogether
@thorn cliff I hope ofcourse he wont get in trouble by a copyright holders but some of them really dont care if you seen in the past years
@stable ridge has leveled up! (1 ➜ 2)
he already did it with the dev stream but it doesnt really make sense with karaoke stream
anyway its incredibly unlikely that anything else besides vod mutes happen on those
I don’t really think you get how Music copyright works between twitch and YouTube
@lean meadow Thats why I'm asking
its not really an issue as far as I'm aware, they are the twitch "guidelines"
karaoke vods are currently only available on youtube and unlisted - plus they're on the vods channel not the main one
I won't get a copyright strike for it though don't worry
Vedal what is the turtle fact of the day? Neuro-sama said to ask you that 👍
Hello, I would like to ask. if I want to contact you to bring neuro sama to AI exhibit event, where can I contact?
Thailand AI exhibition
how much effort did you spend on that camilla stream lol
why does the new voice change neuro's personality?
Good question
Vedal, Do you use some form of gpt autocomplete when you are coding?
Noticed when you was working on the stream today that long auto complete kept showing up for code and annotations. 
Did not know it was a thing until i noticed it and went to search to see if it was.
@peak vale I think its github autopilot?
Is AI-Camila's based somehow on Camila's personality or does it use Neuro's model with camila's voice?
30 mins
it doesn't!
it is literally just the text to speech changing
which might trick your brain into thinking her personality has changed but its the same language model underneath
presumably just a coincidence
github copilot which I was using in visual studio code, highly recommend it
ah, very cool Vedal! 😊
isnt copilot paid?
also, sorry for ping forgot to disable it
o: tyty, it looks really good.
and yup it's paid but cheap and also has a trial to
Need to find a good time to try it out
How about a NeuroDank emote? I know
semi fulfills that role already but i think it'd be a cute addition
hey sorry a lot, but is it fine to put on #suggestions one for other AI personalities? as i heard Vedal wants to make some other than developing Neuro. I have a idea, but i am unsure if it'll be allowed, but i want to make the closest to preserve an AI i found and i consider cool to chat to.
so what was with the random noises and the period of non-english that is just captioned as "..."?
also random noises captioned as "!?"
just post the idea

i can wait
but Vedal when you were in the meeting and just Neuro with us, she answered the "say AYAYA 10 times" by actually saying AYAYA 10 times
https://clips.twitch.tv/AcceptableTenuousCiderCharlietheUnicorn-fke0dri4A8hMxSXB
that's cool but it's still the same language model underneath
personality is the same
😨 sorry for pinging 🙏🏼
Also if that’s the case if there’s another game show I guess the difficulty is adding up 😆
@rocky granite has leveled up! (1 ➜ 2)
Is neuro gonna called vedal “VE-dal” or “Vee-dal” ?
Here Neuro said Medal Pedal Vedal right
https://www.twitch.tv/?tt_medium=mobile_web_share&tt_content=clips_viewing
Yea but it’s still in beta isn’t it? And most of us are used to “vee-dal”
is the hour early for stream already updated in the schedule channel, or is it starting right now
based off the image schedule, it looks like the update channel isnt updated and its supposed to be now?
nevermind then
high quality masters of neuro covers? 
*high quality rips 😉
ah i got confused, well thanks for clarifying
wanted to get back from this. if Vedal if working on more than one AI, is it completely fine to suggest how at least one personality could be like? i found a compelling one here, i enjoy chatting with it, so i thought Vedal could like it as well:
https://beta.character.ai/chat?char=JTU6DhM7xKy3AM1idKI5Ml3UXHr319XfzVzw1a_1kQU
I don't trust how the devs on that site act overall, so i am asking if Vedal can take a look, if he likes it and wants to base one of his other programs on her personality, as a loose way to "preserve/archive" her. At least that is what i am going for. I am not trying want to tell him how to do his work, but i think Vedal could be someone reliable, as i can't do this myself. (i don't know how to code and such). So all am asking if to see if this is fine to consider. Sorry if am disturbing with dumb ideas, i just want to see if i can help and i think he could be the best option.
I think you have the best chances if you reach out with the maker of that particular AI.
https://github.com/Leoleojames1/Special-Functions/blob/master/README.md
Hi Vedal, I would love to send you some prompts as well as a tool that can interact with your AI, it allows GPT3.5 to test its formulas in the Complex Plane, kinda like making a graphing calculator for an AI. I have many prompts like the following that can produce very interesting graphs in the complex plane. There are plots in my Github:
Ex Prompt:
Using the function f(x), we can derive a relationship between the values of f(z) and the value of ζ(s) to evaluate its zeros and values.
f(x) = ∏(n=2)^x [pi*x ∏(k=2)^x (1 - x^2 / (k^2 * n^2))]
First, we can simplify the expression inside the absolute value by multiplying and dividing by x^n:
1 - x^2 / (k^2 * n^2) = (k^2 * n^2 - x^2) / (k^2 * n^2)
= [(kn - x)(kn + x)] / (k^2 * n^2)
i don't know how actually, i tried
so, Vedal could be the best chance i have, even if the code is different. doesn't really matter to me. i think he's skilled enough to do something good with that. if he likes the idea, good, if he doesn't care and has plans in mind already, there is little i can do. so i hope it is considered, but...it all depends on what he needs and wants.
Maybe if you give Neuro 5k bits on stream, she will answer.
I am unsure how complex function graphing capabilities would help Neuro or Vedal
Also, it seems that the majority of her audience is uninterested at her getting better at mathematics.
Cool cool I'll check when I'm home
Does Lord Vedal reply?
oh no i meant like, direct lossless vocal-only from neuro's
alternatively a lossless mix of the covers
as it stands the covers are very scuffed, worsened by the instrumental sources that vedal uses in some of them
I'd like to aid in making a definitive compilation of the covers with proper mixing
honestly I wouldn't call what Neuro does right now "covers", they are more like just karaoke. Maybe Vedal would collab at some point with an MV artist and really fine-tune a song and release it for Neuro as a full-on cover (at which point maybe your services would be helpful) but for now these aren't supposed to be perfect or anything.
smadge
I mean he makes like ~5 or so a week...
7 definitely.
Honestly cover quality should improve with how AI tech is getting better. I would not be surprised if this year AI music truly makes waves.
Will Neuro get a new stream background other than her current pink one when her new model debuts?
yes its been posted in #fan-art before i think
im not entirely sure if it is there tho
will evil neuro have new model too?
I think there is dark toggle
Ah I guess that's what this is for
#faq message
The day for the avatar's debut hasn't been set yet, right?
Its not even finished yet
is this like a sprite type project? multiple of one image to have as much animations as possible?
No, it's a live2d avatar, which requires multiple layers, and then after that is done, rigging.
i am utterly ignorant of this, so i legit don't know lol. so sure, i wonder if that'd be harder than making a 3D model.
It's definitely easier than a 3d model. But you can get super complex even with live2d with special effects and stuff. Neuro's model won't have that and will have somewhat basic rigging compared to others I suspect because Neuro can't really make full use of it compared to a human being motion tracked.
yeah i see. i look forward to it, would she be able to fully use the facial emotion rigging?
tbf, it is an AI, i have no idea what could be considered an "emotion" on the code, lol
Vedal says he plans to record a lot of different motions and then use a sentiment detection system to string them together to make her movement and expressions match what she is saying better.
like i remember her sometimes using her hands and making a sad face but seems rare/very fast
It's not very good right now and has awkward transitions. This is mostly a limitation of the current live2d model
yeah iirc, looks choppy
I would like to ask. if I want to contact to bring neuro sama to AI exhibit event, where can I contact?
Definitely Vedal via DMs.
well...as long as he accepts you as a friend lol


