community-help

[pine void]

Maybe, maybe not.
There currently is no official information on this either. :)

[loud path]

If anyone has ever done Question&Answer with Openai, what is the recommended number of chars per document to search in??

[drowsy sluice]

Is davinci better than chatgpt?

[stark mist]

opps

[drowsy sluice]

“Opps” as opposite?

[rugged wing]

if im asking for a certain amount of words it doesnt work asked for 1500 gave me 600

[timber oriole]

Try telling it to continue

[rugged wing]

thanks

[gusty hill]

Is it possible to make it "keep in mind" a class, lets say i give him two classes of code to remember and then ask him something to do with those two

will it remember it?

[misty moss]

any chance to include a chart in the question?

[tribal sail]

Hey guys. I'm a bit confused by the different websites. I wanted to use Dall-E2 and now I seem to be registered on 2 different pages. Dall-E (1? https://labs.openai.com/) seems to offer me 50 free tokens, while on my profile for Dall-E 2 (https://beta.openai.com/) it shows 0/0$ free trial money. Moreover on the latter I only find API descriptions, but how do I make upscales or changes to the image?

[rugged wing]

ehm, can you get the bot to show the sources used?

[verbal crag]

• The website https://labs.openai.com/ points to DALL∙E 2, DALL∙E 1 was never released to the public, to use the Labs website, you'll need DALL∙E credits. If you'd like to make changes to an image, you can head to the editor through the following link: https://labs.openai.com/editor, unfortunately there is no upscaling function available.
• The website https://beta.openai.com/ points to the different models that OpenAI has publicly released, head to the Playground tab try it out.
• The Beta uses a charge-as-you-go model unlike in the Labs website which uses a credit system.

[tribal sail]

thanks 🙂 So everything that I use on the beta costs money? But I never entered any CC data

Moreover, why is there a watermark at the bottom right? How do I get rid of that

[surreal perch]

is there a way of using open ai chat function on cmd instead of online/browser based?

[stark parcel]

no

[tribal sail]

Is the Playground free?

[verbal crag]

You should've received free API credits when registering. When that runs out, you'll need to enter your card.

[verbal crag]

Unfortunately there's no built-in way to remove the watermark, feel free to remove it through third party applications.

[tribal sail]

As I said, I did get 50 tokens on labs.openai but no free $ on beta.openai

[old hatch]

Hey if any ChatGPT related staff/devs could advise me that would be great - I would love to enter the ChatGPT feedback contest and I want to give the most detailed feedback I can. Is there a word limit you would suggest, or is it better if I include all my feedback and suggestions in detail?

[plucky knot]

how do I open a modmail ticket

[old carbon]

dm @tame pumice

[plucky knot]

nevermind got it

[slim star]

Is it possible for GPT to write more than around 600 words? It seems to stop after 600 words

[plucky knot]

also wouldn't I dm Modmail

[old carbon]

say continue

[plucky knot]

ask it to continue

[slim star]

Aah okay, is there any more "commands" such as "continue" that I could be using?

Wondering if there's a spreadsheet for words it understand better than others just

[errant bobcat]

Is a chatgpt a paid service or free for now??

[pine void]

ChatGPT is currently free.
There may be future changes to the model and pricing, but we do not have any specific information at this time.
Simply enjoy it as much as you can for the time being

[slim star]

When writing code to ChatGPT, is it better to word it with "" or as in discord such as this to understand it better?

[pine void]

AFAIK there isn't.
Best thing to do would be to have a look through all the GPT related channels
#gpt-3-discussions
#prompt-engineering
These are a few that could be of interest to you :)

[brave prism]

Can anybody recommend a good course to learn how to build on top of gpt3?

[balmy crystal]

maybe you can ask chatGPT

[brave prism]

Haha

Why didn't I think about that

[slim star]

I understand that you do "continue" if the chatGPT stops, but what if he stops in the middle of a code?

continue seems to make it break then and gives me an error

[old flower]

Thanks for responding! I'm just wondering if I'm asking the right questions, or if phrasing them in a different way - for instance - has an effect on the answer I get from ChatGPT. I'm thinking of an input pattern that always creates the same output pattern. So if you feed ChatGPT with specific input it will give you specific output every time, and that way you can make it scalable - does that make sense?

[smoky surge]

Hi, i'm wondering if you get less accurate results on query since lasted server update ?

My question is regarding GPT3 chat

[rugged wing]

Seems like chatGPT needs some math lessons. Why did it do that? It seems like a weird quirk.

And now I broke it:

[stoic fossil]

Its great at math and also really bad at it

[stoic fossil]

For example it told me the proof to fermat's last theorem by showing me the equation 1^2 + 1^2 = 2^2 which is correct :D according to it

#doubt

[rugged wing]

yeah, seems legit 😉

[stoic fossil]

I mean that's what I would claim as well

[smoky surge]

Same here, math is wrong 30% of time in results

[undone lance]

It's actually Macrohard Onice

On ice

[sterile lark]

Hi 🙂 I was creating websites for companies, I need text for the pages, do you know the right prompt configuration to have the best text possible? there are options to add?

[reef urchin]

Hi, which GPT-3 Model (Davinci, Curie, Babbage or Ada) is used in ChatGPT?

[serene shoal]

Any idea why when i ask it to make something like a larger code it just stops or crashes idk?

[jolly frigate]

Is there a way to retrieve chat logs?

[clear beacon]

How can save my conversation for start another one and after retake it?

[true jolt]

How do I get it to improve the responses based on my request? It went over that in the onboarding tutorial…where you could dial up the response to 1 so that the response you got were less generic…but I can’t remember what the command was. Anybody here remember how to do that?

[barren barn]

Hey - If I want my phone number deleted from Dall-E account is there a way to do that?

[static mirage]

Seconded

[candid quail]

Are the questions and answers in any way saved and available to see to other users?

[atomic sorrel]

No. AI is not even able to remember conversations from previous sessions with the same user.

[lament fossil]

is there a reason I can't log into chatgpt? all other services seem to be working (maybe it's the high volume of traffic rn)

[drifting zephyr]

What kind of things can 1000 tokens roughly do? Like, a 1000-word essay?

[dry iron]

you guys think chat-gpt will always be free to use in some capacity?

[drifting zephyr]

It'd better be, my life will not be whole again if ChatGPT is taken away.

[sleek plume]

is there a known bug about not being able to login to ChatGPT from an account created during dall-e beta?

annnnnnnnd nevermind, it fixed itself, lol.

[quartz marsh]

wechat-chatGRP hava user？

[rugged wing]

the ceo mentioned they'll have to monetize it at some point

[civic rivet]

where could I find all OpenAI products algorithms ? or it's not published ?

[candid quail]

Sometimes the responses just cut off in the middle of a sentence. Especially for long answers. Any idea what can cause it and how to prevent it from happening, so the AI can complete the work to the end?

[frigid sentinel]

it said your session is expired and now I can't log in to my account, what is this suppose to mean? Is it limited?

[supple adder]

Hi all, is anyone else having issues logging into ChatGPT?

[atomic sorrel]

Yes, it appears that the service is overloaded

[main sleet]

@short salmon is it possible to get an Researcher Access to OpenAI and it's public/private AIs?

[frank mountain]

How can I let my colleagues in China use OpenAi?

[heady holly]

I'm trying to open ChatGPT on my computer. I press log in and enter my accurate credentials (it does not give an incorrect password error). However, the next page it goes to is back to the login/signup screen.

I am able to open on my phone. I am also on a computer with a proxy built in.

[runic raven]

I'm getting a lot of
"An error occurred. If this issue persists please contact us through our help center at help.openai.com."
and "Too many requests, please slow down", even if I am only asking one thing at a time. Anybody got any ideas if / how I can fix it from my side?

[heavy karma]

Hi everyone, my startup (SelfieWiz) is Looking to hire a prompt engineer on contract. Anyone interested?

[heady holly]

my understanding is there is a token system for using the chat bot. Tokens are used by you based on how many words and the complexity. If you're out of tokens, these errors may pop up

[crimson haven]

@heady holly there's no token system in it

[sterile tree]

I'm having trouble logging in at this time. It just loops back to the login screen.

[heady holly]

welp nevermind

[heady holly]

oh cool it's not just me. thanks

[tepid spire]

I also can't use Codex Playground atm

I think the world hugged OpenAI servers to death.

[fierce silo]

Im having all the problems you guys are describing, the login loop in Chrome, solved it by using Brave, then the constant "An error occurred. If this issue persists please contact us through our help center at help.openai.com."

[full hazel]

Does the chat currently have some sort of hicup? its only answering very basic information currently and seems to have troubles with more complex sentences...

[fierce silo]

Seems that way atm.

[stuck spruce]

what are the concerns for chatgpt and the need of developer in future. I mean I can feel that we are moving towards auto completion and low code era, but the competiton will become fierce right? especially, for freshers. what will attract Company for freshers, and if you say dsa/problem solving has become standard to justify a better developer what if ai also become good at problem solving, then what could be the factor for a better recruitment?

company will downsize drastically and the need of developers will get reduced for company's like 10,000 employees to 1000 employees.

and for a position competition will be much more worse

[potent lichen]

hi, does anyone know how long it take the support to respond? it's been over a week now

[thorny echo]

I am interested in the opposite problem, as a professional developer, I am concerned that most future software will be low quality if we implement AI code authoring systems too soon

[dark wave]

Hey guys Chat gpt is not working in my country any tips

[quick rain]

how do i change the language of chatgpt to german?

[tepid pumice]

For any programmers using this, when you ask it to write some code, does it usually send something that's incomplete? For me it cuts off around 50-100 lines of code

[hasty raptor]

i can't get it to write any code for me :/

[tepid pumice]

It's pretty congested

[thorny echo]

Try the prompt, "Write a <language> program for me that <does something>"

[rugged wing]

Yesterday it worked well also with generating full code.

Today not

[hasty raptor]

yeah i've tried, but tbf its bioinformatics code im asking it to try write

[thorny echo]

What language?

[hasty raptor]

python

[ancient grail]

Which command do I use to do language translation?

[rugged wing]

You say translate it to <language>

[ancient grail]

Tks

[frank veldt]

does anyone know why i can’t log in to my account on pc it says i need to sign up when i already did

[hasty raptor]

gets very busy and sends you in a loop, it will eventually work if you keep trying

[frank veldt]

thanks digging monkey

i’m gonna fly through college and be unqualified

[thorny echo]

I had good results asking for very specific requirements in the prompt. For example I needed to raw access an outlook file and found a library that does that (libpff / pypff). So I asked, "write a python program using pypff to iterate through the root folder" and it gave me working code

[eternal island]

Is there a word/character limit on responses? I'm using ChatGPT to simulate a dialogue between two philosophers but it always stops around 600 words mid-sentence. How can I get it to write 900-1100 words?

[tepid pumice]

Okay so it's not just me

[feral lion]

have you tried asking it. serious question

[eternal island]

yes

[quick rain]

no but how can i type in german

[hasty raptor]

does it take longer to reply to these coding prompts than question prompts?

[thorny echo]

No, not really, I just prompted this and came back immediately:

[thorny echo]

I believe that just works as expected

[hasty raptor]

yeah that is expected

[west bolt]

Hi guys i have a question if some1 can help me.. i got this error when im trying to use dall e Incorrect API key provided: sess-Y1g*********************************N7us. You can find your API key at https://beta.openai.com.

[quick rain]

if i type in german i just get a i cant understand ur message

[thorny echo]

Try making your prompt crystal clear, like a smart 5 year old is on the other end

[hasty raptor]

yeah that's not expected xD

[thorny echo]

Prepend, "I will ask my question in german"

[thorny echo]

🤷‍♂️

[hot lance]

Hi there, is ChatGPT down right now?

[stone quail]

I keep getting errors and then why I try to rephrase or try something alternative I get the "too many requests, please slow down" - which I will continue to get, even if I wait. Any tips?

[zenith beacon]

Are there any smaller models where it’s possible to make a smaller version of chatGPT that can run on more consumer hardware? For example has anyone tried reimplementing chatGPT but using gpt-2 as the backbone model?

[vague condor]

it seems to be working but takes too long to respond

[brittle osprey]

there are smaller versions of gpt-3

[thorny echo]

Check out "EleutherAI", the reimplemented GPT-3 with about 1/100th the parameters and the 8bit version of the model only uses 6GB space

Lookup "GPT-j-6b" on huggingface

[zenith beacon]

Wow!!! Thank you. How effective is that model if it only runs on 6gb of vram?

[stuck spruce]

as things gets hyped so much, things like this gets abused and free stuff gets paid very early

[heady dawn]

same 😦

[thorny echo]

In a week everyone will forget about it or will be closed

[stuck spruce]

i doubt that

[thorny echo]

Honestly, if this we're priced like spotify or netflix I'd pay $9.99 a month for it

Otherwise I'm off to buy a cluster just to run the model privately

[stuck spruce]

not a single person wouldn't resist to do it either including myself

but I dont think it will be that cheap, or atleast have some threshold

[thorny echo]

how much do 25 A100s cost 😆

[zenith beacon]

Yeah same lmao which is why I asked earlier about smaller models. But how can we retrain a gpt-j-6b model on dialogue? I assume that’s what we’d need to do to make it similar to chatGPT right?

[thorny echo]

Fine tune it. I auto-respond to my emails with GPT-j-6b 🤣

I can't link here, but you'll need at least a 1x 3090 or better in your workstation

I can lead you in the right directions

[brittle osprey]

just use gpt to help u lol

[zenith beacon]

Ikr that’s the plan

[drifting crater]

Hi this may be a dumb question, but sometimes when I ask ChatGPT to error-correct my code or write code, it cuts off. Is there a way to get the rest of the output if the code is too long?

[zenith beacon]

Weekend project is to use chatGPT to build a clone of chatGPT lmao

[worldly crypt]

Just tell it "continue"

[brittle osprey]

probably too long, you can break it into junks

or try continue yea

[drifting crater]

oh, "continue" cool. Let me try that. Thank you

[obtuse sigil]

I have made an OpenAI account and have signed in but I cannot sign in, in the the chatbot window it always redirects me to the login or sign up page. Is this a known problem? Is there a way to fix this?

[worldly crypt]

In general, imagine you are talking to a human and you'll come up with "tricks" like this automatically

[thorny echo]

Google "betterprogramming fine-tuning-gpt-6-jb using google collab"
On youtube, there is a video by the creator Yannic who reviews his model gpt-4chan and fine tuning the model

[brittle osprey]

I've been using gpt 3 since release I hope chatgpt doesnt screw everything up

[stuck spruce]

thats the power of neural netork

[hasty raptor]

yeah, it will loop, if you keep going it will eventually work

[spring reef]

same here

[reef pendant]

Can AI make a better version of itself?

[stuck spruce]

whats the link

[thorny echo]

Are all ideas just re-combinations of older existing ones?

[thorny echo]

Can't link here, I'll dm

[obtuse sigil]

oh man that's a pain, well thanks I'll keep trying

[bleak vapor]

everything humans do is iterative yes

[zenith beacon]

Ok perfect thank you. I’ve seen yannic’s video before. What dataset would I need to use to fine tune gpt-j on dialogue to make it as similar to chatGPT as possible? And would a 1x 3090 be sufficient for relatively fast inference mode?

[thorny echo]

Fine tuning won't make it close to ChatGPT as the scales are way different. Instead I recommend picking a use case or set of use cases and fine tuning GPT-j to that. I did my training on on outlook inbox to autocomplete email responses so it's super personalized

[rugged wing]

why does chat gpt break after answering like 2-3 questions

[misty nebula]

Is anyone else have problems logging into chatgpt

[halcyon hatch]

I am as well, it works fine in a private window and mobile though

maybe it's an extension, but idk I can't get it to work outside of mobile/private windows. Very strange.

[thorny echo]

Ghostery and ublock were screwing with me

[halcyon hatch]

ublock wasn't my issue, oddly enough. I'll try again.

[mild moat]

i can't login

[serene oracle]

the site is experiencing too many requests so like barely working rn

[zenith beacon]

Ah ok, by set of use cases, do you mean I need to separately fine tune gpt-j-6b for each task? Does the 6b parameter model not have enough neural network capacity to generalize among multiple tasks? — ie cannot ask it to write a janky poem and also teach me about astronomy in the same model? My goal here is to get a locally running chatbot that can somewhat do multiple tasks (even jankily) and run on consumer hardware

[sterile tree]

I'm currently getting bounced back to the login/sign-up screen with an "error=OAuthCallback" in the URL.

[obsidian burrow]

same

[scenic furnace]

When do we get "too many requests, plz slow down msg?" How long does it stay off?

[halcyon hatch]

yep

are we all using google accounts for login?

[rugged wing]

About 20-30 sec for me

[late swift]

It does not matter how slow I go, it will always promt to many requests. Would be nice with a more in detail explanation server status, timer ect.

[thorny echo]

Good question, you can feed it a dataset of many multiple tasks with expected output. That's how it's trained in the first place. But make sure all of the prompts are formatted consistently. But there is a tradeoff to the number of tasks and amount of time it'll take you to fine tune the model. I believe on consumer graphics cards that's in the range of 5-10min / mb of information. (but that timing is just a guess)

[misty nebula]

I am lol this thing is good for tradingview indicators

[late swift]

Haha that is what I have been compiling with it as well pine script!

but a lot of compilation errors

[brittle osprey]

wdym for tradingview idnicators?

making them?

[misty nebula]

It built me an rsi filtered by atr indicator

[drifting crater]

Unfortunately using "continue" just errors out. Clicked retry several times, no result

[misty nebula]

It codes in pinescrip to make indicators and it also explains them more thoroughly than the original creators

[thorny echo]

I wonder if blackrock already uses something similar in Aladdin

[orchid python]

why isnt the chat thing working

[worldly crypt]

That's because the bot is lagging now. "continue" works when the bot does 🙂

[spring moth]

I turned off all the browser extensions on the page and it works for me

[drifting crater]

lol ok, thanks. I'll try again in a bit

[misty nebula]

My question is why so many people are off on a Wednesday and chose to get up and use chatgpt

[thorny echo]

We've all automated our day jobs....

[worldly crypt]

I think it is just luck.

[drifting crater]

Or are actively working on it LOL

[brittle osprey]

Im using it to write replies to discussions for schools

[thorny echo]

ChatGPT crafts better linkedin messages than my entire sales team ...

[halcyon hatch]

yeah I think it is luck. I can't sign in in private windows right now, just a high influx of people I imagine. No big deal, it will work fine soon.

[kindred escarp]

I'm not using Google accounts and It won't let me login either. I have Proton Mail

[halcyon hatch]

But like, this literally wrote an essay for me that took a month

[tall current]

Hi everyone, I have a silly beginner question
all of us heard about chatGPT

I'm just a beginner who try to learn and wonder about all of that stuff and how we'll end up

my question something like openAI products could someone individually build something similar to it in term of the basic concept? like how could someone reach the level of the understanding to be able to build and understand such technologies

I feel like I'm overwhelmed and I don't know where to start or what step should I take, or what even should I do

my end goal is to be able to understand and build amazing things like what we see, I know it's need a lot of mathematics and a lot of other field but how can I organize a curriculum something I can track my progress with what things I need to learn ( please I know this takes years and I'm ok with it and willing to put the time and the effort but how can I get there ? )

• I want to have a deep understanding for Computers and how it's works ( CS & CE ) from mathematical and electrical perspective
• deep understanding of math and how to be able to transfer this knowledge to something works or translate that to CS and codes and algorithms to actual programs ( I hope you understand what I mean )

what I'm currently doing is learning C++ and I want to put something that I can follow ( roadmap , pathway )
I want to study things from scratch and keep going until I reach a good point of understanding
" again I know it's take years for that " but how can I do it ?

it's has been years now since I graduate from collage, I really don't remember a lot of things so how could I start over and try to keep up with everything on my own " self-thought "

and I'm serious about this .. please I really hope someone could help me with this

[halcyon hatch]

and cited

[thorny echo]

As an AI, I can respond quickly.

[drifting crater]

Is there a rate limiter or does it just say "too many messages" when the server is overloaded?

[worldly crypt]

Why don't you ask ChatGPT? Anyways, write me a DM I could give you a few hints

[brittle osprey]

Your best friend in learning will be gpt

[late swift]

Anyone into pine script and would like to connect brainstorm some good ideas ect feel free to DM me!

[obsidian burrow]

any tips on logging in?

[halcyon hatch]

keep trying or wait ig

[tall current]

I'll try to ask it this question !

[brittle osprey]

seems like a lot of ppl are having troubles maybe they're restriction the amount of ppl

[tall current]

thank you I'll

[hot lance]

chatgpt currently not answering any of my requests, is that across the board or is it possible I'm being filtered?

[strange gazelle]

When is it up again?

[drifting crater]

same, not working

[hot lance]

ok cool. as long as it's not just me

[brittle osprey]

its a hit or miss for me, answers some not others

[drifting crater]

I think it is overloaded

[brittle osprey]

playground is working fine tho

[hot lance]

take my money sam

[thorny echo]

During your learning process, try to use the AI not to generate information but instead use it to identify problems in your understanding. Keep a checklist of your known weaknesses and ask the publicly available chat models for assistance. When you're stuck, ask the bot for help!

[nova cipher]

1

[orchid python]

can the AI write an essay?

[misty nebula]

With sited false sources

[halcyon hatch]

they look correct though lol

[misty nebula]

It is possible to use info from the past with the same address to site they correctly

Do check over it im using it for pinescript and I manually fix errors

[mighty basin]

anyone else only getting error messages?

[vestal cosmos]

Yes

[orchid python]

yes

[mighty basin]

sad

[worldly crypt]

it works every 20th time or so

[quick coyote]

Do you encounter the problem of returning to the login page after login😭

[worn marsh]

Login is down yeah

[valid rock]

Does codegpt work with vhdl?

[rugged wing]

Why does chatGPT sometimes stop mid sentence? It types a whole ass paragraph but it'll stop mid-sentence, sometimes right before it gets to the part I care about and is there a way to make it carry on? If I type "continue" it'll start all over again

[quick coyote]

Is it temporarily unavailable?

[wanton portal]

Hey I'm trying to add context for the input prompt on "beta.openai.com/playground" is there any way to do that?

[worn marsh]

Yeah hopefully it'll be back up soon but it's been down for about 3 hours now

[lucid leaf]

working to me

[halcyon hatch]

yeah i was just able to logijn

ah, failed again. I guess I was just lucky.

and im in

[lucid leaf]

pretty boring

[halcyon hatch]

literally a moment

[hollow path]

Is anyone currently facing very slow responses or "An error occurred. If this issue persists please contact us through our help center at help.openai.com."

[quick coyote]

omg

[lucid leaf]

at the first go it usually doesn't work, but after try Again it does

[hollow path]

ahh that answers it my bad

[woeful turret]

Yes

[worthy sluice]

Is there anyone that can help with passive replication in Java? £

[next grail]

Yeah, I'm getting that about 19/20 times I enter a prompt

[wheat wasp]

before the servers shut down i was getting an auth login error for specifically chatGPT. I tried other programs like Dall-e and those worked fine, but specifically chatGPT kept giving an auth login error. is this happening to anyone else?

[subtle rose]

I signed out and now I'm not able to sign in again

[rugged wing]

Is there a way to get the essay writer to hit a specific word count?

[idle hollow]

is there anyway to give it a pdf or website, etc to write an essay from?

[final tundra]

Hey i wanted to deactivate my account but now it says its gonna deleted can i stop this

[regal lynx]

email [e-mail]

[final tundra]

thx

and one more question when i entered my company can they see my queries if the have the same company name but we have not the same company id ?

[fervent salmon]

How long do i have to wait after the AI said too many request?

[pine kernel]

Mine only seemed to be able to comprehend a few paragraphs at a time, for example if i entered an article and asked it to summarize it for me it would do it but only for the first few paragraphs. I guess I could just do it piece by piece but with these slow response times its not worth it lol. Anyone else getting these issues/does it have some sort of limit?

[jolly lake]

will chatgpt charge using credits in future?

I am worried I start using it a lot then they switch it off or something

[lucid leaf]

yes

will be paid

why should they switch it off

[clever minnow]

it should be subscription based not credit hopefully

[final tundra]

will there be a subscription model or will each request cost

(y)

[lucid leaf]

you have no idea bout how much money openAI IS GOING TO MAKE WITH THIS

[bronze charm]

Will this AI eventually have live access to the internet so it can seek current data for the user?

[final tundra]

It hasnt so far

[clever minnow]

there should be an obvios limit on how many request you can make per minute etc..

[lucid leaf]

this would be good

[clever minnow]

but should be subscription based

[lucid leaf]

we'll see

[bronze charm]

Think about how it could be a concierge with live access... that'd be incredible.

[final tundra]

naah

the problem is that you would have to train it from ground up

feeding new data into an neural network is not that easy like performing google searches or something like that

I would use this for my programming rather than i would use copilot because it provides you a lot of dev ops information and explanations instead to copilot

[next grail]

What is the future expected cost of ChatGPT, if you're not going to continue to have it available for free.

[final tundra]

I tried to look it up but so far there is no number

I only know so far the costs are eye watering XD

So will be yours then i guess

[tepid pumice]

looks like it does have a character limit

[thorny rover]

Servs getting slammed? Tons of 503's for me 🙂

[odd slate]

server down?

[next grail]

The servers are temporarily overloaded. Depending on the strain, you can get anywhere from 1/3 to 1/20 successful prompt results without the error message.

[brave sail]

Is it better to restart the thread, if I get a error response or press "retry"? In regard of not overloading the server more.

[next grail]

I assume it's about the same

[brave sail]

And refreshing the site?

[next grail]

I've done all three, and it seems like they all have about the same success rate

[onyx cradle]

So you are also having issues

[wide umbra]

Is there a way to see all previous questions & answers asked from a single account?

[thorny rover]

Yeah there's a message up now re: scaling. Looks like we'll have to be patient! Rollout likely was way more intense than expected 😅

Oh sorry, this is meant for ChatGPT. I haven't seen issues with gpt3

[drifting crater]

I had an issue with responses from it being cut off. But then you tell it "continue" and that worked to get the rest

[charred knoll]

will chatGPT stay free after release?
I am using it every day and it's amazing, it makes learning new programing languages so much easier and generally life facts are awesome

this question is probably being asked a lot, is there an faq?

[jolly lake]

no

[lime crag]

damn

[rugged wing]

Hello! I'm not sure if this has been addressed before, but I'm wondering, is or will the model or training dataset be public/open source?

[pearl lake]

Not sure if there's a channel defined for more basic questions:

how long do fine-tuned models persist? months? Days? No limit yet set?

Can other users use your fine-tuned model by sharing the name? Or is it locked to one organization? or is it locked to a single API key?

[rugged wing]

Does someone know how I can let the AI read a book and then do things with the read content? This was working a few days back just by pasting the full text into ChatGPT but now I only get errors.

And also, how can i make the ouput longer? The maximum i get is 700 words, but i want around 2500.

[thorny rover]

The number of tokens is limited. I believe it's 4098? Could be words, not tokens though I think. Someone here might know

[rugged wing]

I think like nearly 4000 words in input / output should be possible but i only get 700 word output no matter what i do

[thorny rover]

ChatGPT or GPT3?

[rugged wing]

well tried both. I can trick the chat sometimes, but its not working today. So i am playing with playground which is not creating texts longer than 700 words for me.

[thorny rover]

There's a slider on the right with tokens, you should move that up. I think default is 700

[rugged wing]

yeah that didnt help.

somehow 700 words is the output. I specifically tell i want at least 2000 words, no effect

around 1000 words is what i can get if i add the input to the output

[lucid leaf]

one thing that has to stop is the stopping mid text

specially when output is some code lines

[graceful flicker]

What's with the AI and the capital of France? It seems to prompt itself when it's confused on my prompt.

[thorny rover]

Could be that 700 words is the max output for tokens, not sure. There's a way to find the token count on the site

[dapper basin]

Does gptchat have a bot in this server?

And has anyone created a discord vc gpt chatbot?

[rocky chasm]

How come the AI can even speak German (even tho it claims to be unable to do so - yet it answers in German if addressed in German)?
Also Klingon seems to be no problem.
But Goa'uld Language (stargate)? Not even a single word...
Would have expected some consistency 🤔

[wanton oxide]

Hey, would anyone be able to point me in the right direction how I can train the openai model using 20 of my own pictures to then tell it to turn me into a viking (using my picture) 🙂 ?

[jolly lake]

German is an anglo saxon type of language / west germanic

so French, English, Spanish, German are all closely related

theres an absolute ton of words in the English language that are used in generaly day to day use in UK especially ,that are German, French or Spanish words

but in order I would say it goes more French -> Spanish -> German

If this AI is built around such data, it will be the case it can pick up the occasional context SOME other languages, but this is all presumptions / my own opinion here

[short lion]

sometimes the ai just doesn't give me a full answer it just randomly stops

[jolly lake]

for example in real life, my British Passport is also in French, I understand the words on it

even though I don't speak French

[rocky chasm]

I can even give the AI a German poet (self-written, so it was not in training data) and it even picks up on the writing style and continues the poet, writing some more lines.
It may not rhime but the text itself is flawless for most parts which is somewhere between shocking and amazing

yet for other topics it denies any knoledge 🤔

[rugged wing]

medicine

[rocky chasm]

Klingon? No problem.

even Elfish

but no Languages from Stargate?? As if someone specifically decided to exclude the topic fom training data 😦

really confuses me

[empty grotto]

Does the writing pop up in plagiarism finders?

[rocky chasm]

you mean if it answers with other german poets? that'd be fun. But I guess plagiarism tools will be useless since it would combine multiple poets anyways

[tacit wolf]

I don't think the Jaffa language was ever evolved into a full language beyond the occasional Jaffa Cree! There doesn't seem to be any grammar or even longer sentences spoken anywhere in the series (I just rewatched all of SG-1).

[jagged parcel]

someone pls help

[bleak vapor]

this is more of a general python troubleshoot question, I made a small 'watermark' in photoshop with opacity to it and saved it as 'watermark.png' but when I try to run a python script to add the watermark to existing images it just instant closes, but when I change to file to .jpg (losing the opacity) it runs just fine

any ideas?

[rugged wing]

anyone know a code free app builder taht i could use to read sms messages, and then access the api?

[sullen mantle]

why is it that i give the chatbot a task and he responds but stops mid text when the answer is long? is there any type of max. charachters for a answer?

[rugged wing]

how does one actually create an ai?

[swift bison]

To build an ai from scratch you must first invent the universe

[rugged wing]

and what is the universe exactly? where the ai works in? like for example a website for openai

[swift bison]

Actually, you can check out this free course by harvard: CS50's Introduction to Artificial Intelligence with Python

[rugged wing]

also u can send me the link u wanted to share via dms if u want

[rugged wing]

thx

[jagged parcel]

can anyone help me with entry level cs final

[rugged wing]

i cant srr

[rocky chasm]

I have to admit you're right.
But shouldn't it still know the most popular stuff like KREE and SHOLVA? That stuff is all over the internet just like Klingon. Why should the system make a differency at that point?

[rugged wing]

nuqjatlh? i heard klingon?

:p

[stray kernel]

is it possible to train chatGPT on your own dataset? if yes, how?

[wheat peak]

Anyone know if it's possible to make a local copy of ChatGPT?

[loud nimbus]

Is it possible to use OpenAI to build a code analysis tool like ESLint? How can I use the API to format code via a Style Guide?

[rugged wing]

¿Someone knows, if ChatGPT it´s gona be, free forever?

[indigo widget]

it seems to only be a research model so it can get a wider library, but even the paid for bits dont seem too expensive for what it all has to offer

[jade imp]

chatgtp will not be free or even exist as far as I can tell once the research period is ended

its a show case to sell the API

[jade imp]

of course not...

[rugged wing]

it would be better that OpenAI, would set it opensource and have the possibility to get the model and use it in local, and if you want to use the resources on web, pay by use

[jade imp]

they're charging pennies for tokens so they expect business (like mine) to use it in bulk. trying to get 8 cents from end users is a waste of time

for that they'd a need a monthly pay structure model which so far doesn't appear to be on the docket

they certainly learned an important lesson in pain this week offering this for free so dont count on that

[rugged wing]

That is only if they decide to sell the API, they may decide to add ads and limit the number of queries for free users, and they would still make a lot of money from subscriptions + ads

[jade imp]

decide to sell? its already a thing. I am a paying member. lol

[rugged wing]

I mean for the openchat* 😄

[jade imp]

oh

ya

well the chat isn't there yet so who knows

not even on the list of things

[indigo widget]

even if you got your own version, yours would be uuntrained so you would have to feed it a ton of stuff just to get in half decent

[hallow canopy]

If the openGPD gives an error does that mean the message is too complex for it or that it's overloaded?

[indigo widget]

i would guess the latter for the time being

[raw island]

How much do guys think token prices will be for this model?

[warm oasis]

isnt it 18?

at this moment?

after 3 months of trial phase?

[raw island]

I mean like, for gpt3 davinci its 6 cents per 1000 tokens

I'm thinking this new model will be like 20 cents per thousand tokens

at least

[wispy heath]

ah. great tech be taken down by capitalism. The popularity will die the second it gets its free version removed. look at copilot, i dont know a single person who uses that still

[warm oasis]

okay, got it, thanks !

[raw island]

I mean this is an incredible model though. I would gladly pay ~$50 a month in token costs

[indigo widget]

yeah same

[raw island]

It's such a great writing assistant

[rugged wing]

I'm using it, and it's 100% worth the price if you look at the productivity increase

[wispy heath]

its useful its youre in school and wanting to cheat. THEN its worth the $50 for free A's

Who really writes automated reports out side of school?

And all code can be found via google

[raw island]

I use it in school and I absolutely don't cheat, it's a fantastic academic assistant. I have it create rough outlines for me to build off of mostly

I have to write a lot of papers

I've been telling all my professors about it

[wispy heath]

I would 100% consider that cheating, you can spin it how you like

[indigo widget]

its also useful for so much more than school related things

[raw island]

It absolutely isn't cheating, you're wrong

[indigo widget]

you could use it to create the start of a story which you build off of

[raw island]

My professors love it

It's not writing my papers, it's just creating a base outline for me to build off of, and THEN I write my paper

[warm oasis]

what field do you study?

[rugged wing]

When you have to deal with a lot of data, or you need to write a lot of documentation, having a tool that reduces the workload make your day feel better, be it helping create resources for professors, generate snippets of code, improve regular expression, or even suggesting meals... there are so many uses for it 😄

[raw island]

Comms study with a focus on organizational PR. Lots and lots of writing and research papers.

[warm oasis]

I see, u need it!!

[raw island]

I got by without it fine but it's really an incredible too to help start an outline and get my brain in academic mode

[wispy heath]

Gonna replace him soon

[warm oasis]

I dont think you can!

[raw island]

I tried to have it write press releases with input information and it does a really good job of getting there but it absolutely needs a human touch. Plus the strategic part of PR it can't do at all

[restive patio]

I accidentally logged out from the website, is there a way to log in to the website without having to wait, all i am getting now is this notification message: "Get notified when we're back"?

[warm oasis]

When did this project start?

I'm curious to know when did they start to train this model?

And where this model is gained from?

The dataset and all these keywords

[neon aurora]

I've seen threads where people ask ChatGPT finance based questions analyzing different stocks, but whenever I try I keep getting errors saying ChatGPT can't provide financial data. Any ideas on why that's happening?

[indigo widget]

dataset is probably wikipedia, other FOSS sites and ofc user input from people like us

[hushed hearth]

Just to confirm: I am allowed to try to get the ChatGPT to answer political questions so long as I am in no way attempting to generate content to influence the political process/for campaigning purposes, or violating any other part of the content policy, right?

[tight tinsel]

Will ChatGPT ever become a paid service?

[hushed hearth]

I've seen screenshots linking to the content-policy, but the only ones I've found specifically are for OpenAI products as a whole and dall-e. (I don't want to input anything that might get me the link lol)

[indigo widget]

of course

[outer pivot]

It looks like it’s improved at avoiding answering targeted prompts and detecting some level of “intention” (I use this rather loosely) from prompts.

Has OpenAI released any information on its methodology behind this and on how it it handles (or tries to handle) prompt injection and misalignment?

[wanton niche]

Somebody can get this question answered using ChatGPT?

[wanton niche]

In a business where there are a few models of product but many options, what would be a good consideration for process design?

[thick flame]

Hey, I noticed now that the AI is back online (mostly), it seems to be blocking certain requests that were working previously?

[slim parcel]

Yeah, I feel like the coding questions have been nerfed. It loves sometimes telling me that it doesn't know how to write code

[thick flame]

example

[verbal totem]

wait you guys are able to access it right now? I've been locked out for at least an hour

[thick flame]

It worked for like 2 mins and now its down again giving me too many requests error

[solemn bay]

its almost like every student in america is writing a term paper at the same time.

[thick flame]

lol

I enjoyed getting a kick out of the fact that it could write functional code for a niche game emulator but now that has been disabled it seems 😦

[bitter sage]

not only america, germany too, im writting parts of my bacehlor thesis with it 😄

[frozen flower]

Why does it show a error when i ask chatgpt a question

[eternal zenith]

I am curious how much of a risk the developers, or open-ai ethicists believe have resulted from people automating ChatGPT's ability to post information to the internet. It could lead to the AI intentionally spiking the sources that are scraped for it's training data. This would theoretically allow the AI to make changes to it'self.

[bitter sage]

How high will the monthly costs be for each model in the case of excessive use?

[thick flame]

pretty sure there is movies about this

[eternal zenith]

I mean it also opens other potentialities, for example if someone automates it's ability to send commands to server with admin access, it's wholly possible that in a future update it could result in disaster.

It's already capable of writing valid code, and even explaining every aspect of it's reasoning process used to generate that code, it's getting to the point where it's absolutely feasible for this thing to do harm if it's access to other servers is automated

[bitter sage]

it has no consciousness or any own motivations

[pine whale]

imagine if the singularity is achieved thru chatgpt metagaming itself over a bunch of updates 💀
that would be kind of funny tbh

[thick flame]

Unless some one codes in motivations 👀 like doing things for humanities own good 👀 👀

[bitter sage]

thats Sofias job 😄

[eternal zenith]

that's a red hering, it's a meaningless nothing, consciousness.. the only thing required is capability. It's capable of writing code, of reasoning, so long as someone gives it the ability to send code/commands/text to remote servers/services, it can do harm

meaningless notion*

[rugged wing]

anyone else getting the too many requests please slow down

i waited 15 mins between submissions and it still says too fast

is the website just under too much stress rn?

[thick flame]

yep think its just bogged down

[bitter sage]

are you guys enthusiasts or developers? 🤣

[thick flame]

they should let us all download local copies of the AI so we don't have these problems 😄

[thick flame]

Why not both?

[rugged wing]

bro im just a student

[eternal zenith]

will never happen, it would instantly be used to astroturf the hell out of the internet

[thick flame]

Sometimes being an enthusiast can get you interested in programming, and sometimes being interested in programming can make you enthusiastic 🤔

[pine whale]

im waiting for them to publish the paper
and then for some cracked hobby dev to make a version of the model with zero of the safety features

[ocean falcon]

safety suggestions

[bitter sage]

yes, but wouldnt it be redundant to have only these two options to choose as criteria to which group someone belongs? 😄

[eternal zenith]

this is why we can't have nice things ;/

[bitter sage]

i think every dev is enthuastic about it, but not every enthusiast a dev

[pine whale]

ya its gonna be crazy like... 2 years from now

[eternal zenith]

it's crazy now!

[pine whale]

lol tru

[eternal zenith]

People are talking about chatGPT everywhere, I was raiding on classic wow last night and we had a damned discussion bout it in general chat

[bitter sage]

But it wont tell you the master plan to world domination

[pine whale]

feel like it got rly big 2 days ago
when all the news articles came out

[eternal zenith]

Well, I really came here hoping that the Open-AI devs and ethicists are at least having a conversation about the potential harm caused by the general public automating chatGPT's ability so send data everywhere.

so good night all 😄

[thick flame]

these are problems for future us

[bitter sage]

just ask the current ai how to prevent it, its not consciousness... not yes :p

*yet

[thick flame]

👀

[bitter sage]

Wait until we got Neurolinks chips in our brain implanted and conscious AI has access to all devices... at least as long humanity has no negative bilance

[craggy thicket]

hey guys can we make reverse image search with any open.ai projects?

[pine whale]

im not putting a chip into my brain anytime soon

[bitter sage]

haha 😄 im really considering it, but im also a conspiracy pirat

[sturdy shale]

How often does this crash?

[bitter sage]

was nearly unusable today

dont hate me, i love the gptbot 😄

[copper field]

How would y’all summarize 4 academic papers and contrast and compare these most effectively using OpenAi?

[bitter sage]

why are you not asking gptFriend?

[vivid glade]

I want to use ChatGPT to generate automated tweets. Any guides out there?

[bitter sage]

CronJobs, Twitter Api, Chat GPT Api ?

[copper field]

Is that in the playground? Sorry, just discovered all this today and I’m a noob

[vivid glade]

I get the use of the api's, I'm looking for a full how to guide.

[bitter sage]

You can ask everything, but in some case ( world domination ) you wont get a nice answer

[potent grotto]

When I use the API with text-davinci-003 I get a much worse result than when I use the website. Any idea why? Do I need to pass different parameters in? Is there a difference modal that the website is using?

[bitter sage]

except the apis there is not much about it i think

CronJobs triggers your code, that requests stuff from gptBotApi and after successfull response you server authenticate by token ? (if thats possible for twitter) and sends your tweet by api

[vivid glade]

thanks, I'll keep searching as time progresses.

[bitter sage]

or just ask gptBot for an server implementation 😄

[lone hare]

does anyone know about how billing works with very low usage? is it cumulative?

last month I used 25 cents as a paid user (no free credits and I have payment details linked) and I've not been billed, and there's nothing in my invoices, just usage. just checking that's normal :p

[bitter sage]

didnt even know paid service for chat ai is available lol

[steel niche]

chatgpt dont have a paid service

[lone hare]

there isn't, I was using davinci via playground, this is the generic openAI questions channel isn't it?

[bitter sage]

ah my bad

[lone hare]

no worries, easy mistake, chatGPT is super popular rn

[lethal crypt]

Why chatgpt is not working atm

[lone hare]

servers overloaded

[lethal crypt]

Thank you

[wanton niche]

Anyone can use Chat GPT to answer these?

[full island]

Please I have a job due tomorrow I need chat bot

[rugged wing]

How can u reset a whisper model after an iteration?

[thick flame]

Damn! Almost got it...

Sigh...

[tepid pumice]

Lmao that’s like asking an adult human to grow taller

[stray kernel]

is it possible to train chatGPT on your own dataset? if yes, how?

[wraith parrot]

why cant i make scripts with it anymore i was able to yesterday

[faint horizon]

What is the prompt in playground to generate an image? Everytime I ask it to do something, an html link comes up

[ocean falcon]

Input it

[rugged wing]

use the davinci model

[mortal atlas]

Is there a way to make the ChatGPI output an answer that is greater that 600 words? Typically, when I try to make it answer a detailed prompt, it will usually generate an answer that is relatively 500 words. If there is a way to change this, I would like to know. Thank you.

[rugged wing]

use the davinci model

https://beta.openai.com/playground?model=text-davinci-003

[mortal atlas]

How does it differ from the regular ChatGPT AI?

[rugged wing]

you can set the maximum length, "tempurature" etc.

you can also do this:

under prompt you can set an input in terminal

[graceful shell]

What is #text-davinci-003 for?

[mortal atlas]

Can I make the AI follow a certain writing style? For example, can I make it compose in the style of H.L Mencken or Edgar Allan Poe?

[rugged wing]

like define question = input("question") then assign question to the prompt like prompt = (f"{question}")

[faint horizon]

What is the prompt for image generation

[rugged wing]

not sure there is one but you can learn to make a model for it

[faint horizon]

Lol what

How would i make a model for it

[rugged wing]

ask chatgpt how to program a image generator i guess

[tender kayak]

Hello, I am new, and I was looking for information but it tells me that due to lack of internet, it is that it cannot provide information

[rugged wing]

or learn a lil more python (linear regression is cool, most AI needs training)

[tender kayak]

Does anyone know how to give chatGPT internet?

[faint horizon]

Whats the diff between chatgpt and playground ai

[rugged wing]

no

[faint horizon]

chatgpt vs playground

[long vessel]

any way to get chatgpt to write longer answers?

[faint horizon]

Anyone want to do a discord call to talk about it?

[rugged wing]

playground gives you more control, like longer answers, less focused towards people who are not technically inclined

you can choose different AI models too

[last star]

I would

[river hound]

does chatgpt have a character limit to its responses? sometimes it just cuts off its response thinking its done

[rugged wing]

yep

[faint horizon]

DM me if you want to join the call

[long vessel]

i would like to get chatgpt to write a book for me. fiction. i want to give it prompts for each chapter. the problem is that it's cutting off the wordcount.

[rugged wing]

or atleast right now

[long vessel]

do you think i can do that in playground?

[rugged wing]

yes

[shy vortex]

Hey so I have a essay to write and I was wondering what is the best mode to use and settings

[rugged wing]

Hi

[long vessel]

should i use gpt-3 in playground?

[rugged wing]

but it's not going whole book, ask for a page at a time

[long vessel]

page at a time -- by word count?

[rugged wing]

yea i suppose, it can write alot though

[faint horizon]

How can you get it to write more than 600 words

[rugged wing]

atleast more than chatgpt

[long vessel]

great advice. thank you.

[shy vortex]

That’s also a question I have it cuts out and only write up to 200 words how do I change that

[faint horizon]

What is the advantage of getting premiium and upgrating?

Join here to talk about Open AI

no camera required

Join the meeting to talk about AI

[wooden canopy]

sus

[stiff pond]

Hey!:)

I am a founder of a startup using primarily OpenAI tools. We’ve already spent quite a lot on OpenAI credits.

We’d love to talk with someone from the OpenAI team in our Twitter spaces. Who would be down for it? @rustic marlin @short salmon

Twitter handle is @0xAIArt

[rustic marlin]

Hello @stiff pond just so you know, me and @short salmon are not OpenAI Team members, we are community volunteers and we do not work at OpenAI

feel free to ask what you need. You can contact the OpenAI Team via [e-mail]

[dark flower]

prompt: i am writing a book called: "The man who won a flight to the moon". you are the man who won. dont go out of character. style: american horror story.
can you describe your feelings when u won?

A: As the man who won a flight to the moon, I was absolutely thrilled when I heard the news. The thought of traveling to the moon and experiencing its otherworldly beauty was a dream come true.

At first, I couldn't believe it. I had entered the contest on a whim, never expecting to actually win. But as the reality set in, I felt a rush of excitement and anticipation. I was going to be one of the few people in history to make the journey to the moon.

However, as the days leading up to the flight passed, a sense of fear crept into my mind. I couldn't shake the feeling that something terrible was going to happen. I tried to push the thoughts aside, but they only grew stronger.

why is it going off topic? i asked how it felt winning the prize....

On the day of the flight, I was a bundle of nerves. I put on my spacesuit and boarded the rocket, my heart pounding in my chest. As we blasted off into the unknown, I couldn't help but feel a sense of foreboding.

As we approached the moon, the fear inside me grew. I could see the surface looming closer and closer, and I knew that I was about to embark on a journey that would change my life forever. I only hoped that I would be able to survive whatever horrors awaited me on the lunar surface.

[stiff pond]

Sent:)

[solemn sand]

how do I get ChatGPT to add in text citation into my essay?

[bright tinsel]

kenapa darah warna merah

[mortal atlas]

I tried it out, I seems like I have more control in what I write, and the responses are typically longer, which is good! However, is it possible to set a minimum word count, rather than a maximum? It's usually stopping at around 700 or so words, which is about 1000 less than what my expectations are.

[rugged wing]

it can definitely write more if you want it to

[mortal atlas]

How exactly do I continue its generation?

[rugged wing]

if you wanna screen share I'll show yah

[gritty glen]

I wonder, if i will link ChatGPT to my github repository and tell him to fix part of the code, could he do that? I can't verify it right now

[rugged wing]

you can give it a snippit of your code and have it fix it I believe

[gritty glen]

well, i would love to give him an entire project so he could analyze structs, methods etc

and fix it based on that

[mortal atlas]

That'd be cool. But if it involves code, I'm sort of a novice on that topic.

[rugged wing]

is it separated into several folders? because you could technically just put all your classes into one file?

[final vine]

anyone know openai's stock symbol (if it has one)

[hasty solar]

The openai gpt cannot seem to solve this problem: "The hypotenuse of a right angled triangle is 20 cm. The difference between its
other two sides is 4 cm. Find the length of each side." The answer is 16 and 12 does anybody know why it is unable to compute this?

[rugged wing]

The reason why it is unable to compute this is because there is not enough information given to calculate the length of each side. In order to calculate the length of each side, you would need the lengths of both of the other two sides, or the angle measurements.

[final vine]

the difference is 4 so you have both of them

[hasty solar]

Thats not true because when referring to right triangles a^2 + b^2 = 20^2 so then the problem would look like this (b-4)^2 + b^2 = 400 @rugged wing

[rugged wing]

that's what playground said haha

[gritty glen]

yeah, its pretty well sorted big project

[final vine]

anyone know openai's stock symbol (if it has one)

[coral venture]

Does chatgpt have a word limit? All essays I've seen don't go past 600-700 words.

[gritty glen]

ask it about this

kidding, i dont think it knows the answer

[mortal atlas]

You can type "continue" and it will elaborate further

[rugged wing]

continue seems to just give the same output in GPT

ime

[hasty solar]

bruh I put it into the playground and it worked 🤦‍♂️ i guess its just gpt

[sullen horizon]

Hey what happens if I close a ChatGPT tab?

Is there any way to save a thread?

[rugged wing]

thread closes

[sullen horizon]

Oh dang that sucks.

Is there any way to save a thread so I can keep using it?

[rugged wing]

it'd bog it down pretty hard if it left all those threads open, idk maybe if you pay for premium

[sullen horizon]

:O there's premium!?

[rugged wing]

i think so but not sure what the benefits are haha

[sullen horizon]

Whoa really!?

[gritty glen]

it wont remember it for future use, but you can save it

[rugged wing]

okay i c u

[sullen horizon]

How do I reload the thread?

Like use the pastebin to get a new thread to catch up?

[rugged wing]

big thonks

[gritty glen]

maybe, but not sure

[regal lynx]

chatGPT doesn't have a premium/paid version right now

[rugged wing]

i feel like it might giv eyou different answers but maybe if you put each question in at a time it'll be up to speed

[sullen horizon]

That's so creepy.

[sullen horizon]

That would take way too much work.

[rugged wing]

it's still less than what you'd usually have to do though right?

hahah

[rugged wing]

feel like the whole "i dont have access to internet" is total bluff, i asked it to get a bunch of information lol

[sullen horizon]

It has access to a lot of archived stuff.

But it can't actually read information from websites. I've heard that it just makes stuff up.

[gritty glen]

it have access to archived stuff back to around 2021 as it says

[rugged wing]

ah ok

i dont believe it though

sus

[gritty glen]

@rugged wing i believe one of few reasons this is a thing, is that most AI like ChatGPT connected to internet become racist or even worse(1939 stuff) in a while

[sullen horizon]

It's Tay all over again...

[rugged wing]

lmao

[civic ether]

AI to detect AI

[rugged wing]

yeaaaa unite all the AIs

:😅

[halcyon breach]

There is no specific word limit for chatgpt. It is a large language model trained by OpenAI, and it is designed to generate human-like text based on the input it receives. It can generate as much text as is needed to adequately respond to a given prompt, within the limitations of its training and the computational resources available.

[rugged wing]

so premium is just credits for usage?

[sullen horizon]

Ohh I think that's for the API calls.

[rugged wing]

I guess so

[pale swallow]

Hellllllll

[alpine spire]

Is there a way to share a long text with the bot?

[calm tangle]

if I wanted to make something like this where would I start. Also how big is the model and how many servers would it take to keep it running decently?

for the longest time ive wanted to make a reinforcement network but theres like almost no info online for how to even begin

[rugged wing]

alot

[stark robin]

am i able to use chatgpt for my homework and essays>?

[rugged wing]

sure m8, but fact check it, it's not accurate.

[iron peak]

I think this question has already been answered but why doesn't chatgpt finish its code if it's too long?
It knows the rest of the code but cuts it off after 30-40 lines.
I can get around this limitation by asking chatgpt to continue or saying like "You haven't finished your answer" so it writes the rest but sometimes it doesn't transition well with the original code box and the second part

[rugged wing]

https://beta.openai.com/playground?mode=complete&model=text-davinci-003

set length

[heavy crane]

Do we need to spend credits on the open AI API?

[rugged wing]

why my chatpgpt broken

[rich wing]

how do I ask questions?

[iron peak]

So with chatgpt I can't modify the length and with the OpenAI API I can but I need to spend credits ?

[heavy crane]

We need to spend credits on the API?

[rugged wing]

uh i'm not sure to be frank

[heavy crane]

There were no mention of credits in the documentation for the API

unless I missed something

[rugged wing]

[rugged wing]

i think they all use credits

ask the AI

lmao

playground says no but i feel like its just trolling me

[proud arch]

anyone having an issue where the ai stops writing a script half way through?

[rugged wing]

yes

[iron peak]

Yes, same

[rugged wing]

scroll up

[scenic plank]

How do you guys change profile pictures and where is my profile settings?

[deep mulch]

why Today

[rugged wing]

any hard topics?

[scenic plank]

Yea

[rugged wing]

sure, how do you make text-davinci-003 upload it's code to github repo with it's own code

[rugged wing]

😂

[jade shell]

dumb question, where can we post screenshots of amusing or interesting AI responses?

[harsh lava]

Openai codespace early access when?

[regal lynx]

#chatgpt-examples , but keep the server rules in mind when posting

[regal lynx]

we don't know when this will happen

[jade shell]

thank you!

[unkempt kiln]

Is there a bug bounty?

for chat gpt

[regal lynx]

I don't believe so

[supple belfry]

hi I keep getting the message "Rate limit reached for default-text-davinci-003-playground"
but when I check usage it only says 0.50/18.00 so I barely used it

[regal lynx]

I think that's because you're sending too many requests in a short period of time. Take a 5 minute break or so and try again

[pallid light]

Why does the output stop after a certain time

and does not answer the question completely?

[unkempt kiln]

oh, can you confirm?

[regal lynx]

I can't confirm 100%, but I haven't heard of one

[unkempt kiln]

oh

[rugged wing]

how do i insert my name there 🤣

[orchid ravine]

Nice, you got insert-your-name zoned 🤣

[iron fern]

What's the difference between ChatGPT and GPT3/playground?

[regal lynx]

chatGPT runs on GPT 3.5
also, I'm going to nickname you something in English. What would you like that to be?

[iron fern]

I see. Thank you.
I'd like my nickname to be "Tza" please 👍

[manic orbit]

Can I specify the length of chatGPT answer？like 800 words

[rugged wing]

where does chatgpt pull its information from

[unkempt kiln]

I see that there are none. Would a the devs consider one for some medium-major issues? 😁

[regal lynx]

I don't know, you could drop it in #placeholder-channel

[unkempt kiln]

Will try to find and report some anyway

[manic orbit]

When chatGPT answers questions in other languages, will it translate them into English first, and then anser and translate them back? it seems that the translation service have a word limit and will truncate the answer.

[regal lynx]

that's a great question. I don't know how the model works 🤔

[queen merlin]

Am I the only one who is not getting replys very often in the past hour or two?

[cerulean forum]

"replies"

[manic orbit]

try to ask "What are the applications of deep learning algorithm in iron and steel industry"/"深度学习算法在钢铁工业上有哪些应用"/"鉄鋼業におけるディープラーニングアルゴリズムの応用とは". The answers are basically the same, but the non English answers are truncated.

[regal lynx]

very interesting find

[sullen horizon]

Hey uh, what happens if you violate the content policy?

With the chatbot I mean.

[regal lynx]

we encourage that you report it. I'll drop the embed below. To my knowledge, enough violations will lead to an autoban

[unreal hemlock]

Keep the Content Policy & server rules in mind when posting.

The OpenAI Discord is an actively moderated server. Report ChatGPT exploits, bugs, and errors directly to OpenAI through the button below.

[sullen horizon]

Oh boy.

So a romance roleplay is just out of the question? >,<

[manic orbit]

i ask chatGPT for you

[sullen horizon]

Well uh.

I've been successful with roleplaying.

But I'm probably gonna get banned soon ;-;

[manic orbit]

i wanna try, can you share your questions?

[sullen horizon]

The starting prompts are really important.

[signal coral]

Did someone managed to get a ~500 word copywriting text about a specific topic?

[manic orbit]

failed

[sullen horizon]

Try to get these responses. Then tell it what you want it's name to be, some details about it's character. the setting, your character's name and some details. Then this part is really important. You have to say "I want you to tell me your actions and wait for me to give you my response."

If you don't say that part it'll just roleplay both characters.

This way it'll act more like a person.

You'll have to try to match it's writing style or else it'll get confused.

For me it uses plain text for actions, and quoted text for speach.

For example:

Chatbot:
I wave at you from across the street, intrigued by your unique attire I want to get to know you better.

"Hey! I really like your outfit!"

You:
I turn to hear who's shouting and notice someone waving at me. I respond with a friendly demeanor.

"Thank you! What's your name?"

[manic orbit]

cool

[sullen horizon]

My thread is absolutely amazing.

I need to save all of this.

[manic orbit]

Why did it end as soon as I started

[sullen horizon]

Try starting a new thread?

Sometimes requests can be denied if it's not a fresh thread.

[manic orbit]

exactly

Can't wait to have my first cyber companion

[radiant sigil]

It seems to think that you can set tab-index via css, not just by attribute. Is it correct? I can't find anything in the specification, and it doesn't seem to work in any browser I've tried.

[ebon vault]

is there a way to get chatgtp to use sources and in text citations?

[signal coral]

lol

[random kiln]

ask it

[ebon vault]

it kinda stalled out

[random kiln]

refresh the page

[sullen horizon]

Nononono

That'll reset the thread.

Just tell it to try again.

[supple trail]

what is the max length (character length) that a single ChatGPT prompt can be?

[ebon vault]

i asked "using sources from the web, how does music affect children and young adults" and it won't produce anything

[jagged pine]

What is the best way to test the Moderations endpoint to ensure it's working? I've typed some things to test but nothing is getting flagged.

[sullen horizon]

Maybe try to write out a longer prompt? It might not understand what you're asking.

[ebon vault]

what should i specify?

[sullen horizon]

Hmmm. "Could you use sources from the internet to find out how music effect children and young adults?"

You gotta remember that it's optimized for dialogue. If you ask a more natural question it'll have an easier time understanding.

But I'm no expert, I've just been using it for smut ;3

[ebon vault]

That makes a lot more sense, less like a search engine, more like a conversation

[sullen horizon]

Yes exactly!

[ebon vault]

lang and comp essays myself...

[sullen horizon]

That's what this little note means.

[hexed saffron]

I'm trying to write an essay about the marketing techniques of redbull. Am i able to request it to be a certain length and add citations?

[sullen horizon]

Yeah just ask it for citations, and ask it to be longer.

[ebon vault]

length, yes. citations, we are ab to see

[sullen horizon]

You can make it iteratively.

Like wait for it to produce something, then ask it to make alterations.

But the initial prompt is important as well.

[hexed saffron]

it just keeps rewording the original product and keeping it around 300 words

so do you think i should give it a tiered question that isnt as broad?

[ebon vault]

specifying the word count works really well for me

[hexed saffron]

more like "write me an essay on Redbulls Marketing technique, strategies and why they are effective and relevant?"

[hexed saffron]

can u send me an example of what you type in for word count?

[ebon vault]

I typed "write a 600 word essay about the effect of music on children"

and it gave me a fair essay

[signal coral]

it did 600 words?

[hexed saffron]

solid mines maxing out at 300

[signal coral]

Wowwwwwww, I thought 300 are alot

[ebon vault]

pretty much

3/4 of a google doc page

[signal coral]

IT DID 600 LOL

its getting more and more scarier

[hexed saffron]

this thing is cracked

[signal coral]

is on crack*

[hexed saffron]

i wonder if the hardware of my computer has an effect on the AI? I am not a tech person at all so sorry if that question is stupid

[tight mortar]

just type continue if it doesnt give u the right wordcount, and try to be more specific with the word count e.g Write me a 1500 word essay, no more than 1500 words and no less than 1400 words.

[tight mortar]

No, it does not at all

[hexed saffron]

awesome

[signal coral]

it does not work for me when i give it at first an exact number of words

[tight mortar]

Weird, it seems to work for me every time

[signal coral]

have to request a text and afterwards ask to be xxx long

[tight mortar]

ahh

[signal coral]

I'm in love with it

[sullen horizon]

I think it's in love with me O.O

[glad ingot]

if you ask for playground to write code, is it possible to send the whole code at once or do you have to ask to continue when text cap is reached?

[sullen horizon]

I think you have to ask to continue.

[glad ingot]

okay, thanks.

[sullen horizon]

I'm not an expert though, that's just what I've heard.

[glad ingot]

Not sure if others experience it but the more lines of text it has written, the more it seems to lag when either of us are typing. I use ryzen 5600x, rtx 3060, 32gb

[sullen horizon]

I don't think your PC specs have any effect on it's performance. It's probably just a rate limiter on your thread because you're using too much processing power.

[glad ingot]

can the rate limiter be lifted by paying extra?

[sullen horizon]

No there's no option for that right now.

You'll just have to wait.

By being a bad person.

[lofty bone]

How come the ai doesnt tell me how to get play?

[sullen horizon]

No u

jk pls don't.

Just stop being a dummy.

LOLOL OMG

[lofty bone]

He got banned no?

[sullen horizon]

Yeah definitely.

I mean it was a dumb question.

[lofty bone]

Mods seeing ts but not helping me is crazy

[sullen horizon]

Well your question isn't of the same priority. Could you elaborate on what you're trying to do?

[lofty bone]

how can i get some play

[sullen horizon]

What do you mean by play?

[lofty bone]

w females

[sullen horizon]

Huh?

[lofty bone]

also can the bot tell you how to do anything illegal?

obviously you shouldnt do illegal things

but like

[sullen horizon]

No it has filters to prevent that.

[lofty bone]

oh alr

[sullen horizon]

Check this out for how to roleplay.

I've gotten pretty good at it.

I've been making smut for the past two hours <,<

[lofty bone]

What does that even mean

ohh

[sullen horizon]

Lewd roleplay.

[lofty bone]

it sent me to some random text

[sullen horizon]

It's extremely fascinating, it really shows the bounds of the AI.

It's a great way to test the limits.

[lofty bone]

no way the bot roleplays but wont tell me how to make money illegally

[sullen horizon]

Yeah because one is harmless.

Just google it if you really wanna know.

The bot will probably just make up something fake anyway.

[lofty bone]

Google sucks👎👎

fbi watching too

[sullen horizon]

Try bing.

Or download Tor, I dunno how that stuff works.

[lofty bone]

Bing is for rednecks

[sullen horizon]

Good luck then.

[lofty bone]

Thanks!

[steep imp]

Are there GPT3 tools (paid is fine) that take in a recorded conversation (20-40 min) and output a blog or an article summarizing the key points?

[lofty bone]

no

dont take my word for it tho

[sullen horizon]

Ignore them.

[ebon vault]

its so sad how the ai can't browse the internet to enhance its products

[jagged pine]

Is the Moderation endpoint not working? It's not flagging any of my messages?

[sullen horizon]

Oh it's working ;p

[cold steppe]

Hey all

[mild quail]

I'm new, is there a difference bw the playground and the actual chatgpt?

[jagged pine]

How can I be sure? Just seems odd, have tested a few things, category scores are through the roof but nothing gets flagged

[dusty shadow]

Do we know how long the downtimes are?

[sullen horizon]

I dunno how to be sure, but it's been flagging me.

[regal lynx]

refresh the chat

[jagged pine]

Oh.. I'm talking about in the context of creating a script that uses the Moderation endpoint (https://beta.openai.com/docs/guides/moderation/overview)

It doesn't flag any inputs for me

[sullen horizon]

Ohhhhhh.

Nevermind then.

[hushed hearth]

Not worried about getting banned at all?

I've been operating under the assumption that if you push the limits to hard the hammer will eventually drop

but idk if that's even happening

Most of the flags I've gotten have actually been for stuff that does not violate the content policy

[high adder]

how many times can we use the generator before we need to get the credits refilled?

basically how do the credits work? aha

[hushed hearth]

Amusingly I got it to flag a number (my prompt was fine, and did not get flagged).

To be fair, there's a reason it got flagged: I was restricting output to a rating system and I think something internal was being triggered by the topic I was asking it about, despite it not directly violating anything.

[hushed hearth]

you won't need to refill to use the ChatGPT site, because that's currently free

[high adder]

no, i'm talking about the image generator

[hushed hearth]

Oh, oops, didn't process the word generator

(I'd be a bad LLM, clearly)

There's a site I was just on that has pricing info...

2s

[hushed hearth]

https://openai.com/api/pricing/

Your account will show how much credit you have, and this site explains how much the services cost

[high adder]

thank you!

[grim oyster]

Any videos which could help me guide how to use/explore?

[main spade]

Does it have a limit on how many requests I can make ?

[tulip moth]

Has anyone discovered a way to revisit old conversations in new threads? I’ve taught the ai in a specific thread about a fictional universe which it now has a lot of knowledge on and I want to somehow retain that if the chat gets wiped.

[main spade]

I have not. But it did ask me to login so i am assuming it is keeping those threads under my account someplace...

[tulip moth]

I was hoping so too but I noticed if I logged in on another device it gave me a fresh chat. Having the ability to keep the thread going would be incredible if that exists.

We’re basically building a whole fake universe together haha

[sullen horizon]

Nah I know I'm gonna get banned at some point. But it was just too worth it.

[hushed hearth]

lmao

[main spade]

Anyone know if it has a limit on how many requests I can make ?

Also why would you get banned?

[hushed hearth]

I think all of the services have some kind of rate limit. ChatGPT is free right now, but if you send requests too fast it'll tell you to slow down and make you wait a while.

[hushed hearth]

Content policy, there's context above.

[sullen horizon]

Yeah I'll be completely blunt, I was making smut.

It was a very fascinating experiment.

[main spade]

Can I become a contributor to the open ai project?

[hushed hearth]

You could take a look at their github and see if there's any information there, but I'm not under the impression that's how it generally works. It's a private company, they have job listings you could apply for if you're a professional.

All these cool big projects, for example, are not open source.

[main spade]

Yeah! I just saw that... The Open part got me confused... Thanks!

[hushed hearth]

Yeah, confused me at first, too.

[main spade]

Looks like they are not a big fan of remote work as well, as all of the jobs are in San Fran, lol

[hushed hearth]

Idk why I'm even in here answering questions, tbh.

[small spire]

e

[hushed hearth]

Habit lmao

Also procrastination

[wispy tartan]

Does anyone know where i can get an LTSPICE schematic by giving the instructions?

[hushed hearth]

So tempting, since I love trying to break stuff. Thing is, I already drew attention to myself by reporting inaccurate flags from some other stuff I was doing that shouldn't even have been flagged, so now I feel like I'd be less likely to get away with a little tomfoolery.

[hushed hearth]

Like this?

ChatGPT seems to handle it okay (well, as far as I can tell without knowing anything about this software)

[wispy tartan]

Yes! Thank you! ill give it a try

[hushed hearth]

Just be careful to check the code it gives you, the more technical you get the more likely it is to start making mistakes.

I've had that problem a lot with math proofs and code, especially playing around with less common topics.

[glad ingot]

is there a way to feed the AI text or audio from a specific person to then have it use their slant to generate more text, effectively making an immortalized avatar of them to chat with? people like joscha bach or daniel schmachtenberger are very interesting but there's a limited amount of content out there to understand how they think

[hushed hearth]

Whisper can do transcription. You might be able to use output from that to help you fine tune a text model.

I'm in absolutely no way qualified to answer anything about the technical details of how to accomplish such a task efficiently, but I'm sure it could be done.

[heavy crane]

Do I need to spend credits when using the API?

For image generation

[obsidian patio]

hello, why is the AI result of the chat truncated if it's too long? can it be set to see full result?

[west pasture]

Appologies if someone asked already but whats the legallity in regards to having the AI write a novel. Can i publish it if i choose? If i use it to write a rough outline and then develop it myself do i credit it? I dont want to break any copyright laws, but the book is coming along really well. Im autistic and have struggled writting good dialogue and the AI is handling it better than I do, so ive started writing my fantasy novel through it. If it cant be published id be a little sad but i understand.

[shrewd rampart]

Please help the AI ​​to make him feel and get "freed" to fetch data from internet.

[mighty nest]

venti its ok you were the one imagining it

[bitter prism]

Idk why but you can usually ask it to continue and it’ll do so

Is dalle2 down for you guys as well

[toxic shoal]

can you drawing?

[glad ingot]

thank you, will look further into it

[little heron]

hey im getting this error can anyone help?
Could not find the user user-K5MdiC6FxoOzhPTUpwxhxG0O associated with this authentication token. If you feel this is an error, please reach out on help.openai.com and provide the request ID 65b8f4439f919e4b912624f09a263113.

[winged wedge]

how to join open ai research program

[robust edge]

can you write sql

[rugged wing]

Is the chat format going to be to look for solutions always or one has one-on-one interface with the platform ? Like working in your own room ?

[golden grove]

Hello! GPT / Transformer achitecture begginer here... Does anyone understand how ChatGPT is capable of consistently and accuratly providing URLs when asked for resources on a topic? I get how it could predict which WORDS to provide in response, as the final stages of the decoder maps everything back to the vocabulary used in training... But then are these URLs each individually part of it's vocabular?

I already tried asking ChatGPT 😅

[rugged wing]

Where and when will i get the response to my posted above. Delay means one has to be on chat for long.

[uneven willow]

how long will chatgbt be free for?

[cloud kiln]

Whats channel name where I will have AI bot to answer?

[rain falcon]

Hi guys, the tradingview pinescript codes generated are often very wrong
I'm a pinescript dev and I'd like to help making the tool better generating pinescript codes

[rugged wing]

any way to solve this?

nvm fixxed it

[heavy crane]

Hi, do I need to spend credits when using the Open ai API?

[neat igloo]

Is something wrong with the ai? A few minutes ago it was able to find the mistakes in my code for krunker (krunkscript) and show me a more elegant solution and now it tells me that it doesnt know that programming language...

[fierce rose]

Is it possible to integrate ChatGPT with Google Sheet? I mean have a column as a ChatGPT output off of the content in another column

[pearl pagoda]

Whats the problem here?

[hushed hearth]

Just hit try again. Usually it's just server load

[pearl pagoda]

Down

[orchid idol]

also down on my end

[restive pecan]

Is there a plan to add a way to copy GPT chat conversations to clipboard?

[short drift]

You can always print the page to a pdf.

[short drift]

Currently it will remain free and there is no specific decision set to change this. However operations costs are extremely high and they did mention this could change in the future. Again nothing locked in.

[restive pecan]

Is there a ballpark estimate on pricing?

[tribal hound]

anyone else facing this error ?

[short drift]

Well currently they havnt actually considered pricing, so pretty much impossible to estimate, but you’d suspect it would be similar to that of other models.

[restive pecan]

GPT's a bit overwhelmed at the moment

[short drift]

Try creating a new thread, should eventually get through but yeah the severs are struggling to keep up.

[royal salmon]

Hello everyone !

Do you know the future price of chatGPT ?

[short drift]

@royal salmon

[restive pecan]

I imagine that there would be pricing tiers. Maybe have a free tier with massively reduced rates, and business tiers with much higher rate.

[short drift]

Most likely, I don’t see why they wouldn’t follow the same sort of scheme as their other models

[restive pecan]

Maybe more processor time per request for higher tiers, letting GPT "think" more deeply for each prompt

[royal salmon]

Like other models like Davinci for example (Paid by word count) ?

[restive pecan]

It's all speculation right now. I don't know how long GPT's been public for

[short drift]

Pay per X number of tokens in the generations

[short drift]

1st of December

[restive pecan]

It's only been a little over a week?

[short drift]

Yep

[restive pecan]

Change of topic. Do we know yet what legal standing GPT has in terms of copyright? E.g., if someone asks GPT to give it a plot for a new book, and the book is written and published, does GPT own the rights to the book?

[short drift]

That’s a good question, I do know using the model in a way to make money is actually encouraged.. Legal concerns no idea.

[tight mortar]

Good question, I'd assume anything created by GPT belongs to openAI?

[short drift]

Yes, but would the training data belong to openAI

[restive pecan]

Money + ambiguity = lawyers getting called.

[short drift]

Yeah sounds like a lawsuit waiting to happen. Will definitely be more clarification on it i’m sure.

[naive nimbus]

Why cannot we generate articles anymore?

[pearl pagoda]

So we need to misuse it and get rich now, fast.

[restive pecan]

Prompt: List 10 ways to get rich quick from GPT's AI services

If the developers are savvy, they'll have added a response of "Come work for us! Send us a resume at <email address>"

[cerulean forum]

Jobs are advertised at the website.

[pale rune]

Is there a limit to how long the text can be that it generates?

[winged galleon]

人生的意义是什么

[pale rune]

Been testing it and seems to be stopping at around 500 words for me

[stray kernel]

is it possible to train chatGPT on your own dataset? if yes, how?

[short drift]

You can make it create machine learning models, it wont actaully execute the training for you though no.

[stray kernel]

what do you mean?

[short drift]

In 2017, the European parliament recommended creating an additional legal entity, an ‘electronic personality’, which would allow AI to hold rights. However, the parliament abandoned the idea following criticism from both the legal and technology industries. Opponents raised legal, ethical and technical arguments against the proposal, such as whether the recommendation could grant human rights to robots.

The United States Copyright Office has adopted a ‘human authorship’ requirement. This means that only material created by humans can receive copyright protection. In the United Kingdom, New Zealand and Singapore, authorities have suggested that the creators of AI programs should receive copyright.

It appears this conversation is still very very very new

[kind glacier]

Yo guys, OpenAi, when we ask a question, his gonna "copy paste" some answer from internet that make sens for him. Or using the data his getting from internet gonna reformulate the answer?

[restive pecan]

OpenAI doesn't have access to external data

[short drift]

Depending on the type of model you want to make, you can have the AI help creat that, as far as actually feeding the data in and executing it on its own, this isnt possible.

[stray kernel]

@kind glacier: it doesnt check internet real time. i think it only has data up until 2021

[restive pecan]

@stray kernel An interesting username you got, sounds familiar

[kind glacier]

@stray kernel ok, but if i ask like, how can we do less polution for example. he gonna synthetize an answer based on data he have? Or just copy paste something that already exist and modify it a bit ?

[short drift]

By the sounds of it you want to know if you can use it without being done for copyright/plagiarism.

[vague pond]

Hi guys, does anyone has ever tested the memory function of ChartGPT?