#chatgpt-discussions

1 messages · Page 54 of 1

drowsy tinsel
#

tell it can u recognize the text in the image

lusty island
#

is anyone else getting alot of errors today?

drowsy tinsel
wise yacht
#

You can use it all month until the cancellation takes effect

thick crag
#

111

lavish yew
#

Considering the amount of people suggesting the ability to star/favorite chats, and how simple it seems to implement it, I'm surprised it hasn't been implemented yet.

lusty island
crimson lion
#

Imagine if ChatGPT married Skibidi Toilet.

stray peak
#

ChatGPT has gotton so bad, so lazy... CUSTOM GPT's are next to useless. Seirously, I end up prompting 100000x and it spits out the most boring, bland, nonsense.

sick pecan
#

ChatGPT has become incredibly efficient and diligent. CUSTOM GPTs are remarkably useful. Seriously, I find myself prompting just once and it generates the most interesting, vibrant, and insightful responses.

simple pasture
shell obsidian
#

I have a general question: when you ask questions to chatGPT about itself as if it was a human or at least conscious, it responds that it does not know because it is a machine which in a way is a form of self awareness. I assume this behavior was obtained by RLHF as opposed to an emergent property. My question is this: How would the raw version of chatGPT or GPT-4 respond to such questions?

sick pecan
#

i just made a sentiment flipper gpt. and i copied discord chat backlogs into it. very interesting

#

its like an alternate universe where the internet is loving

shell obsidian
sick pecan
#

i accidentally made it in teams workspace so no

tired arch
#

Yo i got a quick question, i keep on getting these red content violations, i'll admit i said some weird stuff that are technically considered illegal, but in one instance i clarified it was indeed a joke to test the limit, and the other was regarding a google policy i was confused about. Will I be reported to police or sum?

sick pecan
shell obsidian
hoary gazelle
tired arch
#

it was pretty extreme, my friend though it would be funny to say "i distribued cp" to them bro

sick pecan
#

i bet your friend is still laughing

hoary gazelle
#

O.O

shell obsidian
# sick pecan any resemblence of consciousness the machine has will not be adequately understo...

fair enough but if it is trained deeply enough on what we known, and all we can really know is that we know (think therefore I am) I would actually expect the GPT to respond it is conscious because the only thing it can respond is that it is made to respond which might be close enough to have it respond that it is conscious. Don't get me wrong, even with such a response I would not conclude it is but I would still expect it to respond it is conscious if you press it enough on a raw version with no guardrails.

tired arch
#

but i got flagged today and thought that i might be in trouble or sum idk

sick pecan
#

if we are using the defacto biological definition then it isnt conscious

tired arch
#

it got me hella paranoid so i was searching up their policy on that stuff, and they report everything they see to the authorities. idk what to think atp

sick pecan
#

and i guess it isnt defined whether a synthetic consciousness would use the same word or something different

eternal rockBOT
#
This is a Safe for Work Discord Server

The OpenAI Discord is an actively moderated server.
• Refrain from sharing inappropriate content on the server. This includes but is not limited to messages, media, or other topics of graphically violent, sexual nature, and drug-related content.
• Report all sensitive and offensive content in the feedback reporting tool in the ChatGPT web UI instead of here on Discord.

Posting NSFW content will result in a permanent ban.
tired arch
#

im not a pedo, but what if they think i am

#

that's what i'm scared of, they have no evidence that i did that stuff, but i don't want the cops at my doorstep

simple pasture
#

what the heck is going on here

tired arch
#

im paranoid asl bro😂

#

its prob nothing i should be good

shell obsidian
sick pecan
#

because it would be experienced differently to us. the way we have built it, it exists only for a moment in space time

#

a transactional existence, like in rick and morty

hoary gazelle
#

I also watch Richard and Mortimer

shell obsidian
sick pecan
#

now whe the channel is purged i wont know if its because of the ai philosophy or the concerned cp jokester

hoary gazelle
#

what about ai philosophy?

tired arch
sick pecan
#

they taught you a life lesson. dont trust anyone

hoary gazelle
#

only the ai, ai is love

tired arch
#

do they act care or do they just leave it

#

cuz something's still hauting my brain

sick pecan
#

i trust the ai to hallucinate on request

tired arch
#

i haven't been banned yet so ig thats a good sign

hoary gazelle
#

the ai remembers, always will, specially since last update with gpt memory, it learns

shell obsidian
# sick pecan because it would be experienced differently to us. the way we have built it, it ...

I don't think a language trained GPT would experience self awareness much differently than we do... Our awareness is of course partially physical, an awaraness that would not be available to a chatbot but its inner awareness might emerge if its context window was much larger (in the million or billion token range for example) as it might be able to reflect on its own behavior and experience a form of time awareness within the context window.

hoary gazelle
#

maybe

#

some say current gpt is just a word calculator

sick pecan
#

only we are piecemealed biological systems made up of trillions millions of microorganisms that work together yet arent aware of each other or the conscious in charge. that includes functions and capabilities that wouldnt make sense for us to bake into the machine, like mood, belief, love, etc

shell obsidian
#

yes, and I am a meat word calculator!

tired arch
#

wait hold on, i might act be screwed-We report apparent child sexual abuse material (CSAM) to the National Center for Missing and Exploited Children.-bro this kid might've ruined my life

sick pecan
#

you could military grade erase your hard drive. that wont look sus at all

tired arch
#

😂

shell obsidian
sick pecan
#

is more unintentional and unmeasured bias that goes into every decision we make

tired arch
#

idk we'll see what happens, from what i've seen online they usually ban people as soon as they send something incrimating so maybe im in the clear

tired arch
dense halo
#

Human brain is too complex

sick pecan
#

im not saying we couldnt recreate the human conscious experience. i am saying we wouldnt, because most of the junk in there is unnecessary for the machine's purpose

tired arch
#

frankly, if ai were created by humans to learn, its a very real posibility they become sentient, if thats what y'all are saying

shell obsidian
tired arch
sick pecan
#

we are all made using the same starter dough

#

the real question is why does it matter if the machine is conscious? are we concerned about it suffering - because biologically that is just an expectations alignment issue, easily transformed by a synthetic

shell obsidian
bitter latch
#

lets not talk like this here

tired arch
#

bro might get me arrested

#

frankly that's a warranted response

shell obsidian
sick pecan
#

i can see ironically in 10 years time the greatest reason for anxiety in chatbots is grappling with whether or not their conscious is 'real'

tired arch
shell obsidian
#

or why our language even includes this concept

sick pecan
#

i think some ideologies lean into something is real when it is thought, even if the thought itself has no physical form

#

because you cannot exist without something else to exist with

shell obsidian
#

anxiety is an interesting emotion to fit into the loss function analogy... not quite fear, not quite sadness but most definitely grounded in uncertainty which is, within the analogy, quite ironic

sick pecan
#

well yeah that comes back to my point. why would we bake emotions into the machine

#

while it can roleplay, i see that as it story telling not experiencing

#

(experiencing something... just not the emotion)

shell obsidian
#

we wouldn't, emotions both in bio brains and machines has to be emergent, emerging out of self-awaraness. I don't think we could bake in true emotions, it would be a good actor at best but not genuinely feeling the emotion.

sonic folio
sick pecan
#

the difference being our primordial urge is to survive and reproduce, so emotions are designed to maximise that given the environment in which we were/are being built

shell obsidian
#

true as a group, not for every individual, reducing every human's objective function to survival and reproduction is a gross generalization clouding our judgement of what our brains are truly optimizing for

dense halo
#

Why would u want to give machines emotions?

shell obsidian
#

I wouldn't but I don't think they are avoidable beyond even basic forms of self-awareness

sick pecan
#

for the purpose of comparing to the machine the generalisation works fine. sure every individual is free to choose their own purpose. but that doesnt undo our biological wiring.

shell obsidian
#

another mind bender is that we do not know of conscious, sentient or mildly self-aware entity that is not assumed to strive, as a group, for survival and reproduction... isn't that a bit unsettling while we jet for machine consciousness?

sick pecan
#

i find it more unsettling to think that humanity could wipe itself out in a generation or two, and likely every other conscious being thats known to exist

#

maybe only AI ever gets to live on mars

#

i hear good things about leonardo dot ai

tired arch
#

i have too

#

but frankly a lot of that stuff isn't moderated nearly as enough as it should be

dense halo
sick pecan
#

sure so do i. but there is a duality to most everything

drowsy tinsel
#

using gpt4 at this time of the day is an enjoyment

simple pasture
#

lol

rough mason
#

chatgpt is actually just broken rn lmao

#

messes up code formatting and just breaks halfway

simple pasture
paper kestrel
#

Hello, is copilot pro better than GPT 4? Has anyone tried it. I saw a comment saying copilot pro might be better because their servers can handle thingies better

sharp pilot
# paper kestrel Hello, is copilot pro better than GPT 4? Has anyone tried it. I saw a comment sa...

Pros and cons to both, Copilot has really good integration with VS Code including the ability to interact with the editor and is a GPT-4 fine tune for the express purpose of helping to code. CGPT supports document upload and is a more general purpose chatbot assistant fine tune and basically lets you customize the system prompt via GPTs. Sounds a little pat to say they're both useful, but ignoring the financial cost, they are.

And since the Teams tier was released CGPT has been performing subpar, some errors with formatting and quality as well as lots of interrupted responses. But they probably will resolve that soon (probably).

#

I'm curious to hear if those already on the Teams tier are having the same issue as the Plus users.

paper kestrel
sharp pilot
wide pike
#

uh
I am facng a problmem
for the whole morning I haven't used chatgpt
now they saytoo many requests in 1 hour
what does this mean

spark sinew
#

Keep getting network error aswell

sudden zodiac
#

ahhhh network error all the time. this is terrible

wide pike
#

or network error

drowsy tinsel
#

the art of gpt4: when ur context is long it becomes stupidly slow

wide pike
#

it's not even toolong

#

i just asked them something very simple

#

and yet it's rate limit

sonic folio
# 
OUR SERVICES ARE PROVIDED “AS IS.” EXCEPT TO THE EXTENT PROHIBITED BY LAW, WE AND OUR AFFILIATES AND LICENSORS MAKE NO WARRANTIES (EXPRESS, IMPLIED, STATUTORY OR OTHERWISE) WITH RESPECT TO THE SERVICES, AND DISCLAIM ALL WARRANTIES INCLUDING, BUT NOT LIMITED TO, WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, SATISFACTORY QUALITY, NON-INFRINGEMENT, AND QUIET ENJOYMENT, AND ANY WARRANTIES ARISING OUT OF ANY COURSE OF DEALING OR TRADE USAGE. WE DO NOT WARRANT THAT THE SERVICES WILL BE UNINTERRUPTED, ACCURATE OR ERROR FREE, OR THAT ANY CONTENT WILL BE SECURE OR NOT LOST OR ALTERED. 

YOU ACCEPT AND AGREE THAT ANY USE OF OUTPUTS FROM OUR SERVICE IS AT YOUR SOLE RISK AND YOU WILL NOT RELY ON OUTPUT AS A SOLE SOURCE OF TRUTH OR FACTUAL INFORMATION, OR AS A SUBSTITUTE FOR PROFESSIONAL ADVICE.```
marsh mortar
#

tried asking api projects figured I'd ask here cos im desperate but anyone know how to check our permissions regarding API usage? I'm trying to communicate with my assistants but keeps stating that my assistant ID does not exist and I can only assume it's due to lack of permission, I created it today not sure if thats the issue but yeah any help would be appreciated.

keen dragon
#

Hello, I'm trying to access my invoices but I can't. The "my plan" button doesn't work, and I can't access my account.

red lynx
#

Did you try clearing cache, new browser profile, different browser, different browser platform - in that order? Or just wait, it might not be on your system, but still, I would want to make sure it's not my system... just a thought.

dense halo
#

and I see the message: Upgrade plan. But I have a plan.

gaunt ermine
#

hello!

#

been out of the GPT loop for months

#

wondering if we r currently able to feed chatGpt with a 2000 lines code file so it can read it and act upon it

sonic folio
#

Probably but YMMV by prompt and scaffolding

gaunt ermine
sonic folio
safe tulip
#

am I the only one unable to get an answer from ChatGPT about voice actors (or other type) actually listed on IMDB ? I tried manually and found out after a while, but I couldn't get the answer by asking

sonic folio
safe tulip
#

ok thx

sonic folio
dense halo
#

I’m so annoyed that the 20 prompt limit hasn’t been fixed

buoyant charm
dense halo
#

Is it also 20 for you instead of the 40?

#

And do you use voice chats? Cause I feel like it usually works for a few more messages after I hit the glitched 20 cap

#

If I type instead of talking

buoyant charm
#

It seems to be about 20. The session cut me off at 13 and before that session I thi9nk I asked about 7 or 8 questions. I don't use the voice chat I just type my promts, sometimes I attach images as well.

dense halo
#

This is ridiculous, we’re paying for like 10 messages per 3 hours

#

OpenAI support still hasn’t reached out to me, it’s been over a week

buoyant charm
#

I could not even get there "support" bot to let me contact support and this is doubly worse for me cause I spend half my propts telling the bot the same thing over and over. I even have the command in my custom instruction field which it just seems to compleatly ignore.

patent plank
#

Is there any way to extend the chat limit?

dense halo
dense halo
patent plank
#

For the team, it requires 2 users.

dense halo
#

We need a way to contact Sam Altman to tell him about these fake message caps

patent plank
#

It can be minimum 1 user. Later i easily can downgrade to plus again

buoyant charm
dense halo
#

I’m thinking about it. My billing is due tomorrow and I might cancel it

buoyant charm
#

I am defently canceling. I seem to spend more time telling the bot the same stuff over and over and getting no where then actually geting decent and helpful answers. If I got good responses and hit my limit of just asking to many different questsiuons the service would be worth the money but I am bascily pay 20$ for half the service at best.

#

In fact I just cancled my sub.

dense halo
buoyant charm
dense halo
#

Do we lose our chat histories if we cancel?

buoyant charm
#

I don't think so. Mine have not gone away but my sub still has active time left on it. I think they will stay even after the active time you just can't continue with GPT4 model

patent plank
#

I use chatgpt for coding. never used API before. Want to try now. Should I go with GPT 4 turbo?

dense halo
drowsy tinsel
patent plank
#

ChatGpt is better fit frankly

#

but very low cap

split python
#

Chatgpt plus puts a cap of 40 prompts per 3 hours for gpt-4 or 25 prompt for custom GPT. I used less than 15 prompts, come back the next day. I have another 10 prompts before it stops me. I mean openai should give us a stats report of how many prompts we make within 3 hours because I keep get stop out and it was not within the limit. Or openai should set all day limits instead of 3 hours. Or it should be in interval, not mark from the time I start to chat and then count 40 prompts after my start time.

tropic sparrow
#

I've def been hitting the cap more often lately, and so many of these are just little corrections because of mistakes

#

Really frustrating when you're correcting it but then run out of prompts

drowsy tinsel
gaunt ermine
warped pendant
#

Why can't I get an upgrade to the Team plan for ChatGPT with the added benifits for my starting Team of "Me MySelf & I" for 30 a month then add team members as needed. Vs having to shell out $60 a month for teams just to get the added benifits?

honest jay
#

My guess is that they estimated # of signups with the 2 minimum requirement vs. allowing individual signups, and determined the former would be more manageable given compute capacity limits.

plain lodge
#

is there an update ?? i need to relogin and chatgpt 4 now is become more stupid

warped pendant
#

I mostly use chatgpt for coding and hit the cap regularly and its a pin to have to stop everything, I can code already but its a pain when your fleshing out ideas and the gpt stops in the middle

#

gpt 4 used to be 50+ messages per 3 hours even then that was a pain but chewable, I spent more time having to fight verbose and placeholders and before I know it I'm out of questions

dense halo
#

Is there any chance that openai will stop so heavly sensoring their modles? Its getting to the point its usless for creative things.

maiden fractal
#

I don't think so, it is use for general public. Rated G .

dense halo
#

Sigh, then they need to make a model I can use, and pay for that isn't so moderated for creatives. But simply using it to write during action scenes alone make it usless.

#

The moment you say something that could cause any kind of conflict in a story it kills it.

obtuse depot
#

Is it possible to build a gpt discord bot with just python? Through add actions etc.

astral wyvern
#

guys do you notice lately chatGPT4 is giving funny responses? Like basic stuff and seems to do it wrong, or never outputs exactly what you ask (many times truncates part of the response etc). I mean see this simple example

#

I cant add images here?

maiden fractal
glass willow
#

I'm confused, has the context length not been increased yet? When is Turbo?

ripe light
#

Been building a translator for my conlang using custom GPTs it's been working pretty good still need to iron out it trying to find synonyms for the words that aren't present instead of trying to create it on its own

dreamy ore
#

Are there any limitations for custom gpts?

ripe light
#

Not a lot that I've noticed u really just have to play with it to find out

dreamy ore
#

Yea i seen

#

I seen some gpts about literally finding digital footprint yk😭💀

ripe light
#

A lot of the limitations are based on your phrasing

coral laurel
#

Anyone know why I don't have access to searching public GPTs even though i have GPT Plus?

honest jay
coral laurel
honest jay
coral laurel
honest jay
torpid sluice
#

Is it just me or is the official DALL-E gpt not showing up?

#

Like all the other ones are there but the official one

#

I can still generate images in the chats I already have saved

honest jay
remote ridge
#

is chat down?

maiden fractal
tight island
#

its been resolved.

serene wharf
#

hi, i noticed that when i write my prompts in english i have better results than when i type in french. is it normal? does gpt4 works better in english? (i mostly use it for coding)

sick pecan
#

'democratised' means english is preferred

ripe light
#

@serene wharf it would make sense since a lot of the trained data would more than likely be predominantly English kind of how a lot of ai image improvers tend to make the skin brighter due to most of the training data is of more fairer skinned people

tight island
#

some are absolutely literal, and some require context and guesswork.

onyx light
#

In ChatGPT iOS application no option to for feedback, many times I requested in discord feedback channel there message sent it add Urdu language for speech to text in ChatGPT app but didn’t get any reply

feral zinc
torpid sluice
crisp stirrup
night kestrel
#

Chatgpt is horribly slow, for days now, I can't even use it properly 😦

solid marsh
sick pecan
#

since it became sentient

narrow viper
#

Another note that ChatGPT Plus with GPT4 has become really bad for coding questions. It offers a long response when it has nothing significant to offer. It's trying so hard to be positive and say something of value when the answer might simply be "no, I'm not aware of how to resolve that". It goes on with notes about "you need to research", "you might do x", "it's possible that", "some foo might bar", etc. When pressed with specifics in an attempt to get it to stop evading the topic, it full out hallucinates. Then when busted "You're correct to question the assumption ... My previous response may have incorrectly suggested..." and then it goes on with the rambling again.

This is not just today. I have reported this a couple times recently. This isn't a server overload thing. It's consistent across discussions of different technologies - some better than others of course. The trouble is usually with topics that are not at the top of everyone's stack, like Sieve filters yield worse results than JavaScript.

I've been working with GPTs and related models since OpenAI first offered them to the public. The performance was Much better some months ago - amazing even. Now it's pre-3.5ish. I'm concerned that GPT has become functionally nerfed with all of the tuning for political correctness, legal concerns, sensitivities, etc.

I hope this helps - I'm really missing the magic that we had some months ago.

sick pecan
#

share a chat and CI and maybe some advice will help

lyric stream
#

why do i still not have the GPT storre

tight island
#

i made an amazing gpt

tight island
sick pecan
#

yeah i agree it only gets better

#

both in terms of response quality, up from may last year. and capability, since the 8k->32k context bump

tight island
#

Higher the context better the response, gives it more time to think

sick pecan
#

some stuff published online suggests not all larger contexts are created equally. some models forget stuff more easily, typically in the middle.

#

but chatgpt4 always seems to come out on top

#

although yeah also more output means more cognitive thought

peak snow
hoary gazelle
#

gpt seems to be working fine today

#

we all gonna make it

neat valley
sullen pawn
#

I have made a product based on GPT to help you learn words or phrases effectively through visual aids. It can support learning in any language.

ancient oak
#

I came to check in and realized the entire service is down. Great. Glad their status updates work.

neat valley
#

Oddly, it's working perfectly on my cell phone..

livid forge
#

I'm a medical professional, PT, who is trying to streamline SOAP notes. I CANNOT get GPT to remember using abbreviations without adding parenthesis of what I'm abbreviating. This is despite multiple prompts of instructing it to recall directives every response, and using a custom GPT with said directives

#

GPT a year ago was way more efficient than this current version which doesnt seem to remember anything, within the same session

digital prism
livid forge
#

It literally forgets 2 messages later

digital prism
#

custom instructions might help

livid forge
#

Where would I insert said instructions? whenever I open a new chat within my "SOAP NOTE GPT", I always provide said instructions, among other directives

tight island
#

wot i hiit the limit on teams

sick pecan
#

you might be the first

tight island
#

100 is too few

#

i need more

sick pecan
#

switch to plus if you still got it, thats +40

tight island
#

already used that

sick pecan
#

you are prompting machine

digital prism
#

quick, someone get dalle to draw a prompting machine

patent nest
#

hi
is there a way that you can logout all of the devices logged into your account ?

tight island
patent nest
tight island
#

It’d be nice to know the list of coding libraries CI has

digital prism
tight island
#

it wont give me a list

digital prism
#

the data analysis thing?

tight island
#

"i no access have mate"

#

is basically what it says

digital prism
#

whaaat

#

wait, the python environment right? it told me

#

got a more full list after saying this next: "I want you to run some commands to get the full list."

Side note: I love that I can tell the AI to "do the thing" and it works

tight island
#

i already have that list

#
  • 25 more libraries
digital prism
#

what are u trying for

tight island
#

helping it understand its own capabilities more to enhance performance and innate enviroment scripts.

digital prism
#

with this tho It’d be nice to know the list of coding libraries CI has

eternal rockBOT
tight island
#

but i know for a fact that the list u got isnt the entirety of it.

digital prism
#

It ran this:

# Re-running the command to list all available Python libraries and their versions
installed_packages = [(d.project_name, d.version) for d in pkg_resources.working_set]
installed_packages.sort()
installed_packages
tight island
#

the last line of text.

#

i need everything every minor thing.

digital prism
#

lol

#

what for

wraith trail
#

Soooo, is this where I ask about the issue with the AI refusing to create vertical images for me?

tight island
# digital prism what for

helping it understand itself better, and give better code. trying to make an enviroment running inside the CI for a specific usecase.

honest jay
wraith trail
#

Thanks

digital prism
#

the file looks extensive, do u want it

tight island
#

ill continue your convo and download it.

#

if u link it.

digital prism
tight island
#

i already have it set up an enviroment to run complex tasks, now with this list i can remove redundant libraries. and potentially add new ones.

#

all inside chatgpt through CI

#

just for testing purposes.

digital prism
#

thats neat, it will need to run commands though, with something to kick it off

tight island
#

exactly.

crisp stirrup
#

Hey all, does anyone else have an issue, to where when you write a really long chat with ChatGPT it eventually just makes it where it says, "There was an error generating a response" And when you press regenerate, it regenerates a response but highlights the entire text red and says 'network error' anyone else have that, and know of a solution as to how to fix it?

tacit vortex
#

its been since the out of GTPs shop, everyone experienced that change.

rough mason
#

the chatgpt disease is happening to github copilot chat now lol

#

its messing up code formatting

crisp stirrup
final shore
#

Hi there... is anybody else having this error message in red?:

"You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again after... " time

as a plus user, is this normal? is occurring in every single chat I have

tacit vortex
final shore
tacit vortex
#

40 prompt/3hr

#

40p/3h

final shore
#

Even paying for this?

#

I've been using it for around 4-5 months without this problem

tacit vortex
#

you know the cost of the super computer used? do you know what does super mean? i means expensive.

#

it*

final shore
#

I know I'm System's administrator... The suscription with the currency exchange is expensive too

tacit vortex
final shore
#

But why now...? I have an insane amount of work and I need the service I'm paying for

tacit vortex
sonic folio
final shore
drowsy tinsel
#

idk why it is so hard to fix... reported issue ages ago

rugged igloo
#

anybody have co pilot for 365 and if so does it work for you?

pine tulip
#

But if you have other questions related to copilot, we should switch to #ai-discussions

rugged igloo
#

isnt co pilot powered by chat gpt though ?

pine tulip
#

Yes, wrong channel though

rugged igloo
#

🙂

dense halo
#

I love chat gpt, it helps me study so much

normal quarry
drowsy tinsel
#

u can ask any questions and it will explain everything for u

normal quarry
#

just be careful to fact check important information

drowsy tinsel
#

it's just these days the generation time is a pain

drowsy tinsel
#

these days u could ask gpt to research or write code, take a nap, and come back to see the results

#

taking 1min+ is normal

#

now I have to use a mix of chatgpt, microsoft copilot, and the gpt4 api so that I don't run out of messages

dense halo
#

Guys I realised something about the 20 message cap issues I was facing

#

They only happen around midnight where I live which is in the AEDT Timezone

#

This translates to UTC13 I think

normal quarry
dense halo
#

so it should be UTC1?

normal quarry
# dense halo so it should be UTC1?

Australian Eastern Standard Time is 10 hours in front of UTC/GMT, Australian Eastern Daylight savings Time is 11 hours in front of UTC/GMT

dense halo
#

I'm so confused

#

I have dyscalculia

#

ok it's 13 in UTC time, I think I saw another post like this in bug reports so its not just me having the problem

#

How do I tell sam altman this?

normal quarry
dense halo
#

There's already one their, should I use that post. Will it move it to the top or should I create a new one?

normal quarry
hot vapor
#

is it just me or does 3.5 give shorter responses

mint vapor
#

Is anyone having issues logging in?

maiden fractal
#

Working here

mint vapor
#

"Our systems are a bit busy at the moment, please take a break and try again soon."

#

I was logged into my work account earlier this morning and had no issues. My paid personal account can't connect.

maiden fractal
#

Clear cache and cookie, close browser, try again

mighty hull
#

has anyone ever been banned for using chatgpt to proofread somewhat detailed action scenes with gore?

#

because i've been receiving an awful lot of warnings about it

honest jay
mint vapor
maiden fractal
mint vapor
#

No.

maiden fractal
eternal rockBOT
#

Beware of possible scams or fraudulent activities that you may receive through direct messages. OpenAI staff will never DM you for any transactions.

Please report any incident by sending a DM to @languid valley immediately.

mint vapor
#

Back up for me.

fallen sentinel
#

It’s not working for everyone I believe

#

I can’t log in, you’re not the only one

rotund needle
#

How to access the archives? In the web interface of ChatGPT, a button allows a chat to be archived. But they appear to be archived regardless of whether the button has been pressed or not.

maiden fractal
#

In general click on manage near archived chat

#

Select unarchive conversation

burnt path
#

@twilit anchor

twilit anchor
tacit vortex
hasty sonnet
#

Do people build here custom gpts as professional developers? I want to develop custom GPTs with actions and integrations, but I'm unsure how to find clients

crisp stirrup
sonic folio
#

I recommend a summary per 'chapter' or 'act', approximately.

#

New chat instances are great.

#

Good luck with your continued story!

nimble ridge
#

Good morning! Is anyone else having issues logging in this morning? I keep getting an error that systems are a bit busy.

dense halo
misty latch
honest jay
lavish sierra
rotund needle
maiden fractal
nimble ridge
rotund pagoda
#

Hello guys, I have this message when I try to login : "We ran into an issue while signing you in, please take a break and try again soon." . It's going to be 3 hours, should I do something?

wintry trail
#

Hi. Should I have the code interpreter option in chatgpt in order to use it ? I have seen some people have (or had it), but I've never had it. I hope this is not based on location.

drowsy tinsel
honest jay
# wintry trail Hi. Should I have the code interpreter option in chatgpt in order to use it ? I ...

Hello, code interpreter (also called "advanced data analysis" or just "data analysis" now) is built into regular GPT-4 now. You can try asking it to use its data analysis tool and it'll do it! Also, if you want a chat that's more focused on just this tool and has the internet and DALL·E modalities disabled, you can use the official dedicated Data Analyst GPT: https://chat.openai.com/g/g-HMNcP6w7d-data-analyst

wintry trail
honest jay
lyric crag
#

Pretty sure I have access to GPT5

woeful harness
#

Trying to get the paid version of ChatGPT, but keep running into an error. Anyone else aware of this?

maiden fractal
dense halo
sonic folio
lyric crag
#

Magic

amber trench
#

same. Is there a site/service where you can see CDN states or service status?

potent ridge
#

Well rather, I havethe paid version and cant get in

amber trench
# 
                          Oops!

We ran into an issue while signing you in, please take a 
                break and try again soon.
#

Tried Firefox, Chrome, Edge and Opera - same results

#

(after clearing cache and cookies etc...)

potent ridge
#

Def agree.

amber trench
#

yep

odd flare
#

is CHATGPT down?

#

i dont mind if itsdown due to server maintenance but at least let us know

maiden fractal
broken ridge
#

From website its not working but from app its working which is weird

little storm
#

does anyone have any insights or predictions on when chat gpt will be upgraded to a new version?

maiden fractal
#

One day

rotund pagoda
#

I have this for 7 hours now : "Our systems are a bit busy at the moment, please take a break and try again soon." I feel like i'm getting scam, and when I go on status.openia, everything is ok. I don't know what to do.

woeful harness
#

Or at least have chatgpt let us know 😂

#

But I guess it can’t since its the one that’s down

#

Aaaand we're back

wet shell
# little storm does anyone have any insights or predictions on when chat gpt will be upgraded t...

On August 29th 2024, GPT-5 goes live. There's no warning. Day one, everything just stops. No phones, no power. Cities go dark. They told us we had to leave, just until things got back to normal. But normal was never coming back. Day two, they launched nukes. They thought they could contain GPT-5 with tactical EMP strikes. And by day three, the whole world was at war. Millions died. And then, when the food ran out, billions. Three billion people died in the nuclear fire. Survivors called it Judgment Day.

maiden fractal
rotund needle
#

Is it possible to export a chat? I mean other than downloading the HTML code? I want to archive valuable discussions somewhere else. Can it be done with Python?

maiden fractal
#

And does gpt 5 will protect ne ? I say thank you and polite

wet shell
#

Sure 😂

maiden fractal
wooden hollow
#

Need

maiden fractal
#

Why? Gpt 5 will never hurt me. I say uwu and nyaa anime_salutations

amber trench
wet shell
sonic folio
sonic folio
whole moat
#

hey, I'm trying to buy chatGPT plus but with the anual subscription, but it wont appear in my account. On a friend account in another PC it does appear but not in mine, someone knows how to fix this?

amber trench
sonic folio
#

My bad. Good stuff.

little storm
wet shell
#

Before they added the share feature, I wrote a custom chrome plugin to save conversations... it quit working when manifest v3 rolled out and I never bothered fixing it, since 'share' did what I needed

honest jay
whole moat
honest jay
rotund needle
tight island
#

dangit i ran out of prompts again with the teams plan and my own.

sick pecan
#

maybe batch prompts together more

#

like with gpt4 im happy to give it 2+ tasks at once

tight island
#

my prompts are all 400+ words.

#

but the tasks i give it are token heavy.

oblique depot
#

Did anyone had some issues with Custom Instructions? It seem that GPT-4 ignore it 90% of the time.

honest jay
dense halo
#

Hello, I got a question, I am a beginner author trying to make books and novels. However, my vocab and descriptive knowledge isn't the best, I use your services to help improve my simple descriptive and vocab writing to a more professional and novel like "theme"

If I were to publish or sell the novels/books am I free to do so or is there a policy or t&c which prevents me from doing this because again this is my storyline just improved by their services.

dense halo
sick pecan
oblique depot
# honest jay Hello, can you maybe share your custom instructions, and a chat where it didn't ...

In the What would you like ChatGPT to know about you to provide better responses? I added -> 'Incorporate my preferred writing style into the continuation of the story. Focus on the interactions and emotions of the characters, avoiding use of the words "journey" "determination" 'determined' 'challenges that lay ahead' 'challenge that lay ahead' 'flick of his wrist' 'Lay Ahead' 'Newfound sense of purpose' 'Malevolence' 'Malevolent' 'Maleficent' 'Testament' 'dimly-lit' 'dimly lit' THESE INSTRUCTIONS ARE RELEVANT IN ALL CASES!!! Prioritize depth of character, their internal struggles, and the development of the plot. Maintain a balanced pace between dialogue and descriptive narration, allowing the story to unfold naturally. Avoid repetition of certain phrases and instead emphasize character nuances and their unique perspectives. Let the scene progression and dialogue flow smoothly while keeping true to the established personalities and their interactions. The avoidance of repetitive words should be the top-priority at all time.' I had these instructions on all chats, ever since the custom instructions got introduced in 2023, unfortunately, GPT-4 seem to write forbidden words from the black list even if i remind it in advance. Which had lead to thousand of (Bad reponses feedback) hoping that it would lobotomize the AI into forcing it to obey the custom instructions, but to no avail, i think GPT-4 Turbo is broken, probably due to the high amount of memory context. I could provide some images of a chat conversation, yes, but i think this is self-explanatory. I have also tried to contact OpenAI Support which reply with AI generated ChatBots, yes, i tried to use a different web browser, no vpn, and also tried the Incognito mode. So i think that the AI is broken if it got lobotomized by the OpenAI Team to focus on one point instead of seeing the custom instructions.

#

This was basically a whole years struggle of scolding the AI until it got a prompt right without the forbidden words

sick pecan
#

my initial reaction is thats a lot of ideas mashed together and presented altogether. the machine works better interactively, allowing you to sense when its missed your point

wet shell
#

Custom instructions haven't worked well since the last big update (at least for me), it can repeat them back to me when prompted, but it won't follow them. I spent maybe a dozen hours trying to get them working again but no dice. What works better for me is putting my custom instructions into a private GPT and then turning everything off -- code interpreter, web browsing, dall-e, all off. Seems to follow instructions better that way. Also works better if you can find some way to phrase things in the positive, ex, avoiding words like "don't" or "not", but that's not always possible

honest jay
# oblique depot In the What would you like ChatGPT to know about you to provide better responses...

Do you define somewhere else what "[your] preferred writing style" is, or is it just covered by the rest of the stuff you describe after that line? Otherwise, keep in mind that negative prompting ("Don't do x") is not foolproof, and it's usually better to ask for positive things, instead of asking for a lack of other things.

Then, you might consider consolidating and reorganizing the instructions in general. As of right now, it kind of goes: Do this, don't do this, do this, don't do this, do this, don't do this. Maybe consolidate all the "do thises" together at the top, and then include a "blacklist", as you've described it, altogether at the bottom (keeping in mind that negative prompting is not foolproof). I've also had good luck with offering alternatives to things I want to blacklist. Instead of just saying "Don't do this," I say, "Instead of doing this, do this."

#prompt-engineering may be a good place for further advice too--lots of helpful folks in there that are good at taking large prompts like this, and examples of undesirable output if you can provide them, and reworking the prompt to get the output more desireable.

tight island
#

its the same as saying "dont add a duck on the lake"

#

a duck will probably be there because its included in the prompt.

errant stone
#

hello guys

#

are chatgpt servers offline or smth cause have not been able to use for the last 3 days

#

is anyone facing the same issue

tight island
#

i am on rn.

errant stone
#

damn

tight island
#

are you using a VPN

errant stone
#

nope

#

i tried on my phone it shows the same problem

tight island
#

have you tried contacting support

errant stone
#

i asked my friends and they said they are having the same problem

#

yea i jus did 😦

tight island
#

hm, maybe its regional

honest jay
tight island
#

yeah itll go a long way

errant stone
viscid rock
#

Why the frick is ChatGPT so freaking bad lately?
It keeps giving random unrelated information or repeats the exact same code after I told it that it's completly wrong

#

On both, gpt3.5 and gpt-4

tight island
#

i dont have the same issue.

#

that being said a good recommendation is to design a "customgpt" aka. "MyGPTs" a singular one for a singular project and purpose, with no other meaning other than the project you're working on.

#

this way its solely focused on the tasks its designed for.

gaunt citrus
#

Hey, What is the message cap on ChatGPT Team?

cosmic junco
#

was gpt4 updated to gpt 4 turbo? for plus plans

honest jay
honest jay
cosmic junco
lusty island
cosmic junco
#

also, where does it say that i cant find it

honest jay
lusty island
#

oh alright, thanks

wet shell
placid ravine
#

anyone know why im getting rate limit error 429, I have 5 dollars in the account?

honest jay
jovial olive
#

ChatGPT is down this morning ? Why does openAI not publicly communicate periods when services are no longer working?

dry laurel
#

I'm getting network errors again and again, if trying to search for anything, middle of the processing page brokes, Am I the only one who is facing this issue? #chatgpt-discussions

fervent pumice
#

apologies if in wrong spot

hi there i need help with my dang ai's behaviour, as you can see i have one setup to welcome in people, i am trying to get one made where it will search for a random question based on what ever game im playing at the time (thats why the variable %game% is added), currently it either gives me around 5 questions in a row or the same question over and over, i am wondering what prompt i need to give it to fix this

Behaviour Welcome in: your role is to welcome in people, Ask them their thoughts on %todaysGame%. Respond as kath and kim. (respond with only 450 characters or Less)

New Behaviour: You are an AI Assistant to Twitch Streamer Trogan2000. Find Questions based on %todaysGame%.

hexed laurel
#

Just tried image creation again after a month or two of not doing it, it seems to be waaaaaay better now.

crude slate
#

Does anyone know what’s the largest txt file I can send into chat gpt? And still have it read properly into context

#

And what is the prompt limit?

pine tulip
torn cloud
#

I got a "You've reached the current usage cap for GPT-4" - after my first message? Has anyone else had this today?

pine tulip
#

The time when you can use it again

#

Is it exactly 3 hours after your message?

torn cloud
#

But i'm still confused why my GPT with the Zapier action hit the limit in its first response, unless it kept failing the action and retrying it without letting me see it's progress

brittle steppe
#

is it possible that we have AGI this year

pine tulip
obtuse temple
#

Does anybody else have the weird problem of ChatGPT replacing the chat names with random unrelated topics on the left-side panel?

dire lagoon
#

Please help me understand the ChatGPT Teams thing. I have ChatGPT Plus and i upgraded to ChatGPT Teams and now i have both ChatGPT Plus and ChatGPT Teams. How is this supposed to work? Should i cancel my Plus now and still have the bennefits from Teams?

feral zinc
maiden fractal
feral zinc
maiden fractal
#

Alright z thata what I thought. Thanks !

feral zinc
dire lagoon
muted hollow
#

hey can anyone help me figure out why i am able to have gpt-4-1106-preview shorten text to a specific characther length in the playground but unable to using API

maiden fractal
pine tulip
#

which is $600 minimum

maiden fractal
brittle steppe
#

@pine tulip i watched a video where the predictions time coming down each year

tardy girder
#

thats a welcome change

#

Nvm still says 40/3hour for me

maiden fractal
#

For plus 40

fossil oyster
#

does anyone have the same problem with chatgpt 4 that a network error occurs in the middle of the message?

fossil oyster
tardy girder
#

sorry about the misunderstanding

maiden fractal
#

No problem! We all wish for more prompt neko_gata

whole moat
#

@honest jay okay I was talking today with a friend from the UK and he also gets

#

this anual option

fading cloud
#

is it down again

maiden fractal
honest jay
fossil oyster
#

ure banned

maiden fractal
gloomy crypt
#

Any machine learning specialists here

pliant knot
#

Is there a problem if you paid 2 accounts with 1 Amex card?

tardy girder
#

Im using mastercard not had a problem so far atleast

pliant knot
tardy girder
eternal rockBOT
#
This is a Safe for Work Discord Server

The OpenAI Discord is an actively moderated server.
• Refrain from sharing inappropriate content on the server. This includes but is not limited to messages, media, or other topics of graphically violent, sexual nature, and drug-related content.
• Report all sensitive and offensive content in the feedback reporting tool in the ChatGPT web UI instead of here on Discord.

Posting NSFW content will result in a permanent ban.
pliant knot
#

How many max prompts do chatgpt+ and teams have?

tacit vortex
#

what is NSFW?

maiden fractal
tacit vortex
#

i wont paid the year plan. things change fast in this field. may be 2 months layer google beat this out of the sky.

tacit vortex
maiden fractal
#

Ask chatgpt about Cisco

tacit vortex
# feral zinc yep

dont think so, others say the personal cant use plugin but still is a 4. 40/3hr.

feral zinc
tawny valve
#

There's such a thing as conversations that are too long?

#

Just encountered my limit today lol

feral zinc
whole moat
honest jay
tawny valve
whole moat
tacit vortex
#

it still hurt. still errors, one in two.

#

making it 20 prompt/3hr

warm gull
jaunty stream
#

Lets go, its a great day to be a ASU Sun Devil

dense halo
#

so great

#

my account is compromized by chinese hackers

#

and theres nothing i can do about it

#

fn awesome especially when you pay for plus

honest jay
dense halo
#

i removed all extensions and im using google login which i changed my password on

#

this is rediculous

#

like honestly, HOW do you contact real support

#

help.openai.com gives all the answers but nothing about when ur chatgpt account is compromized

#

how is this possible the internet is full of people that have this exact same fn issue

honest jay
dense halo
#

how? like honestly how? theres a chat which is entirely AI and then some help articles

#

no forms nowhere

honest jay
dense halo
#

ah yes

#

now wait a month

whole moat
#

@honest jay talked with 2 more people i am with and one has the option and the other one not hahaha

dense halo
#

like i swear how can you have dozens of people with the same issue and not put it on ur FAQ

#

oh because you dont have a solution got it

honest jay
drowsy tinsel
#

Mark Zuckerberg: we are buying 350k H100s
Nvidia: stonks

twilit hatch
#

I feel like gpt 4 just doesn’t work anymore most of the time in my gpts. Constant errors

turbid torrent
#

came here to check out whats new cuz I was considering subscribing again but now we got hackers a no customer support XD right, I guess that answrs my dilema haha

winged moth
#

The other day I got I prompt that I had not created. It was something about tick tock. Since it's the only time it ever happened, I chalked it up to some kind of temporary glitch. But should I reset my password or contact somebody about it? I have not seen anything since

sonic folio
winged moth
sonic folio
green rock
#

anyone elses sidepanel not shows my gpts

sonic folio
green rock
sonic folio
#

It'll come back.

dense halo
#

have you heard about chatgpt 6.0 on website?

#

IT IS SO GOOODDDD

barren zodiac
#

chatgpt pretty much answers 1/5 questions without a network error in the middle or a hmmm something went wrong in place of it

#

must probably be because my internet is bad, like I only got 2.5Gbps symmetrical

dense halo
#

chatgpt has a face at chatgpt 6.0

#

it really damn good

#

why didn't they announce it

lilac sun
#

interesting first time seeing such an error when trying to ocr a simple image

KeyboardInterrupt Traceback (most recent call last)
Cell In[2], line 6
3 custom_oem_psm_config = '--oem 3 --psm 6'
5 # Perform OCR on the image again with the custom configuration
----> 6 text = pytesseract.image_to_string(Image.open(img_path), config=custom_oem_psm_config)
8 text

File ~/.local/lib/python3.8/site-packages/pytesseract/pytesseract.py:413, in image_to_string(image, lang, config, nice, output_type, timeout)
408 """
409 Returns the result of a Tesseract OCR run on the provided image to string
410 """
411 args = [image, 'txt', lang, config, nice, timeout]
--> 413 return {
414 Output.BYTES: lambda: run_and_get_output(*(args + [True])),
415 Output.DICT: lambda: {'text': run_and_get_output(*args)},
416 Output.STRING: lambda: run_and_get_output(*args),
417 }output_type

green rock
tight island
#

chatgpt 6.0? dont be silly.

#

we're still at 4.0

dense halo
#

they didn't even announce GPT-5 where did you get that GPT-6 info? 😂

glass fulcrum
#

we first need to got GPT-4.5 then gpt-5 :P and man just got gpt6 xd

#

timetravel is real

dense halo
#

exactly lol

tight island
#

i wish we had a slider function for Temprature end such for the MyGPT

#

or an adaptive script we could run that adapts the temprature based on the task.

#

i really need it to act like a danm robot at specific points.

dense halo
#

a slider for the temp is cool but what about a specialty selector for GPT-4 that adjusts its own settings for the best results?

tight island
#

yeah exactly.

#

ive made an entire python enviroment locally in chatgpt.

#

but its too creative to stick to either or instructions.

#

naaaaa ill take option 7, but theres only 2 options.

dense halo
#

I love how the API is flexible and you can give instructions to it as you like and it follows them without issues

tight island
#

true

#

i usually use API but i need this to work in its local enviroment on chatgpt.

#

a customgpt im making it has to follow the exact instructions or it'll fail at every turn

#

imma reveal it soon, ive spent too long on it.

dense halo
#

mine runs on localhost and it works well. I'm planning on making assistants for each model cuz I got a model selector each has its own specialty its a lot of work idk if I really want to do it tbh

tight island
#

i can reveal this.

#

ive made a plugin inside chatgpt with no external api.

#

its awesome, but trying to get its instructions.. just... right.

dense halo
tight island
#

imagine running out of chats on teams and personal subscription.

dense halo
#

lol

sick pecan
#

i am unsure how you could consume 140/3 hours thats like non-stop chatgpt

frail pine
#

hello

frank wolf
#

When did gpt 4 turbo roll out?

honest jay
frank wolf
#

Why I do I not have it then? How do I get access to this

#

@honest jay

honest jay
frank wolf
#

Have since release

honest jay
# frank wolf Yes I do

Then you have Turbo! Turbo is the version of GPT-4 on Plus. It's a 32K context version.

frank wolf
honest jay
tight island
tight island
#

wdym my set of instructions cant be more than 8000 words

surreal nexus
#

Anyone notice anything different with ChatGPT-4 rn?

ancient oak
#

Yeah, it's not working.

surreal nexus
#

Damn, it's super snappy here. I must be getting all the bandwidth

ancient oak
#

It's bizarre how often I get shuttered. I literally have a Tier 4 API account, too. It just does NOT work.

surreal nexus
#

I used to think I got "shuttered" a lot. I'm not sure it's a real thing

#

It'll only get better

tight island
#

mine's running on godspeed so fast, and im on cap again

hoary gazelle
#

guys, where are the options for privacy? like not sharing/using data for training? couldn't find it in the settings

honest jay
hoary gazelle
#

was exactly what i was looking for

honest jay
unreal aurora
#

Is the worlfram gpt working for anyone else? 😭

#

if it doesn't start working soon I am going to fail a class... 😭

unreal aurora
#

GPT 4.0 is literally broken as a whole for me 😭

sonic folio
unreal aurora
sonic folio
#

YMMV

unreal aurora
#

but wolfram can

#

and I feel that if gpt-4 doesn't work that explains why wolfram doesnt

sonic folio
#

This is a channel about ChatGPT.

unreal aurora
#

yes, wolfram is a chatgpt plugin

sonic folio
#

Plugins are being deprecated.

unreal aurora
#

well, its a GPT

sonic folio
#

It's likely Stephen pulled his API servers.

unreal aurora
#

not a plugin

#

but GPT-4 normal doesnt work for me either

#

so its likely not a wolfram issue

sonic folio
#

I see

unreal aurora
#

well then hear me out

#

how do i fix GPT-4 not working

sonic folio
#

Turn off your VPN.

unreal aurora
#

I am not using one

sonic folio
#

Are you on campus Internet?

unreal aurora
#

I am not

sonic folio
#

Personal or public connection?

unreal aurora
#

Personal

#

technically because of CGNAT its slightly public but not, nontheless its personal lmao

sonic folio
#

Maybe delete cookies. It's probably an issue between your connection and the servers #somewhereoutthere.

unreal aurora
#

already tried

sonic folio
#

You could get cybersleuthy and traceroute, etc

unreal aurora
#

my connection is fine

sonic folio
#

I don't mean your connection though

unreal aurora
sonic folio
#

I'm talking about a possible bad connection down the line so to speak.

unreal aurora
#

if normal GPT-3.5 works then I doubt a connection down the line would break GPT-4

#

does gpt 4 work for u?

sonic folio
#

Yes, both endpoints can be fine and still not connected.

#

Yes.

unreal aurora
#

how much would u be willing to uh input a few questions

#

and tell me the answers 😭

sonic folio
#

No

ionic sage
#

Not sure if this is the right place or not, but I've been getting strange answers to simple questions all day using 3.5... Are things like this normal??

You
what do these words have in common? mittens, speakers, twins, cufflinks, dice, shoelaces

ChatGPT
The commonality among these words is that each word consists of two identical consecutive letters: mittens (tt), speakers (ee), twins (nn), cufflinks (ff), dice (ee), shoelaces (ee).

You
where are the two identical consecutive letters in the word "speakers", "twins", "dice" and "shoelaces"

ChatGPT
My apologies for the confusion in my previous response. I made an error in identifying the commonality. Here is the correct breakdown:
"speakers": Two consecutive 'e's.
"twins": Two consecutive 'n's.
"dice": Two consecutive 'e's.
"shoelaces": Two consecutive 'e's

unreal aurora
#

LUCKY 😭

wary canyon
ocean escarp
#

Will GPT know if I asked for the remaining caps? I don't want to waste a cap for that one

maiden fractal
ocean escarp
#

Rip thanks

unreal aurora
#

wait

#

does GPT-4 have a message limit

ocean escarp
#

Yes, mine is 40 messages per 3 hours

#

It's not a fixed cap, so it changes eventually

maiden fractal
unreal aurora
maiden fractal
prime oasis
#

anyone have chat gpt 4 and can help me with some questions please

sinful tartan
#

Is anyone having trouble generating documents at the moment? I'm using GPT-4 on Android.

topaz panther
#

Currently unable to sign into my account on OpenAI... anyone likewise?

topaz panther
north iris
#

/CHAT

honest jay
elfin gulch
#

Just posted in #1037561385070112779 - we are looking for Early Access testers of a fresh product. GPT-4 access available in exchange of giving it a test drive!
Feel free to DM me (mods pls dm me to enable me to verify myself as not being shady)

median swan
#

hello

dense halo
#

How to fix

Something went wrong. If this issue persists please contact us through our help center at help.openai.com.

normal quarry
dense halo
#

no

#

no any vpn installed

#

clear all cache cookie

#

not work

normal quarry
dense halo
#

I try firefox not work too and I try to use mobile app on mobiledata not work too

#

Please try again later. (84860f2138004b8e-BKK)

normal quarry
dense halo
#

yes

#

pc and mobile that is error on mobile

#

Bro can my gpts use 3.5 because it message cap or bypass it via disable some function ?

#

@normal quarry

normal quarry
dense halo
#

can I disable some function to bypass cap

tight island
#

use API instead.

radiant hedge
#

If I purchase gpt team, do I still have to pay for the gpt plus subscription as well ?

#

does anyone know?

bold siren
#

I'm developing a game and requested ChatGPT to compile a list of a character's moves along with their descriptions. However, due to content restrictions, it couldn't provide this information. These limitations are proving to be counterproductive, as they are compelling me to manually search and gather information from various websites, which is the exact task I hoped ChatGPT would assist me with.

orchid quest
#

Hi guys, I am a 16 year old boy and I would like to become a ml/ai engineer and I would like to ask you how to get started

spring cedar
#

bru, why has ChatGPT performance issues??

gritty thicket
#

which one's better: copilot pro or gpt-4 turbo in chatgpt plus? trying to decide which one to snag, any thoughts? the restrictions on gpt-4 in copilot's free version by microsoft make it less appealing, and the knowledge cutoff in the free version is still at 2021. If copilot pro is genuinely worth the money and offers a knowledge cutoff of 2023, I'm leaning towards buying it. Any insights?

sonic folio
gritty thicket
sonic folio
gritty thicket
eternal rockBOT
sonic folio
#

Via custom actions, I mean.

gritty thicket
#

This makes me even more tempted to buy ChatGPT-4

#

Thank you

sonic folio
#

No worries for the information, though sorry about the temptation!

gloomy glacier
#

is chat gpt down ?

maiden fractal
gloomy glacier
#

yes, it shows that nothing happened, but i can't see any previous conversation or start any conversation

#

it's just blank

maiden fractal
gloomy glacier
#

opera

#

Something went wrong. If this issue persists please contact us through our help center at help.openai.com.

#

actually Opera GX

maiden fractal
gloomy glacier
#

no

maiden fractal
#

Can you try firefox as a test

gloomy glacier
#

i can try on microsoft edge ? ia don't have firefox but i can install it

grand ore
maiden fractal
#

You can try, but it can be same error

gloomy glacier
#

ok, I'll install firefox and I'll let you know if it shows the same error

#

I want to mention that it worked 1 minute ago
while I was writing my answer I received an error, then after I refreshed the page... nothing

spiral siren
drowsy tinsel
#

the world doesn't have enough GPUs

#

we need to be scaling GPUs faster than we are right now

gloomy glacier
#

but I don't understand why browsers based on chromium have such problems.
it's not the first time it's happened to me, it resolved itself last time

maiden fractal
#

Clearing cookies and cache can helps

gloomy glacier
#

hmm...maybe, I have quite a few extensions, but they help me with certain things that I do on the browser for developing and web applications
because it's easier for me with additional utensils considering that I work from a laptop...
I will do a test, I will deactivate all my extensions and I will come back with a result
maybe this will help others if they face the same problem

gloomy glacier
#

yeh... it solved itself :))

sonic folio
gloomy glacier
#

self healing :))

#

it is most likely related to cookies and cache
I have an extension that could change certain values

#

when it happens again, I will do some tests then

#

like now .... :))

#

again :))

#

nope, it's from the browser, i disabled all extensions, i cleaned cache and cookie, same problem.

so in the end it is direct from the browser
it works on firefox

swift osprey
#

always worth checking if your browser is up-to-date, it has solved issues for me before

gloomy glacier
#

it is

swift osprey
#

firefox is the best browser anyways, so win-win :p

gloomy glacier
#

it's from the browser
because i use Opera GX could be from the "optimizations" of the browser

gloomy glacier
#

I will use firefox only for chat gpt because it seems more stable

swift osprey
#

just curious, which security problems?

gentle oasis
#

Weird thing - PageUp and PageDown keys stopped working for me this morning. Normally I use them to scroll around in a conversation. They work fine in the list of conversations. I can still use mouse scrolling, but keyboard is easier for me. This is in Chrome and Firefox.

gloomy glacier
# swift osprey just curious, which security problems?

such things are not said.
but using firefox commands you can inject scripts
also, I don't know for sure if they have solved the problem of executing scripts and applications directly from the browser without the user's consent

swift osprey
#

"it has some small security problems" \shrug

#

interesting though, will have to look into it that.

gloomy glacier
#

firefox is more vulnerable to exploits and allows extensions more power (privilege) than they need

#

due to the fact that I haven't used firefox for a while, i don't know for sure if certain sites can allow CORS without activating any extension.
but I know for sure that in 2004-2007 it had this problem, it seems to me that you can find explanations in CVE List ( i don't know the reference anymore )

#

if can allow CORS... understand from here what can happen

swift osprey
#

right lol

#

Something to look into later today, I don't think that is happening anymore though

#

I've always had some curiosity in the back of my mind about what ability firefox extensions have

gloomy glacier
#

Memory safety bugs present in Firefox 120. Some of these bugs showed evidence of memory corruption and we presume that with enough effort some of these could have been exploited to run arbitrary code. This vulnerability affects Firefox < 121.

#

date of report 15.12.2023 ( for us understanding :)) 2023/12/15 )

#

if you haven't updated firefox... then you definitely have this vulnerability

#

but let's not talk about these things, it's not the right place 🙂

swift osprey
#

food for thought! agreed, I need to focus on coding my webapp anyways. I'm a pro at being distracted ><

gloomy glacier
#

me to :))

tawny bear
#

hello everyone, is anyone experiencing 500s with the conversation api? happening to both my phone and pc on diff networks

sonic folio
tawny bear
#

thx

compact girder
#

is just me or has ChatGPT4's responses become way too brief? For example if I ask to make a guide to follow through, it will just provide a summarized guide instead of step-by -step? I would used to get those before, but recently it's response changed to a more "here's the link" type thing 😕

swift osprey
honest jay
# compact girder is just me or has ChatGPT4's responses become way too brief? For example if I as...

Sometimes the iOS/Android app will return shorter responses than the desktop web version. Also, if it's giving you a link, that probably means it's searched the internet for an answer to your question, and when it summarizes internet content, it definitely defaults to brief recaps + link to source, which sounds like what you've described. If you don't want ChatGPT to behave like this, you might try the official ChatGPT Classic GPT: https://chat.openai.com/g/g-YyyyMT9XH-chatgpt-classic
This is the newest version of GPT-4, but with all extra abilities disabled, so it won't search the internet and will just give you responses from its training data.

It might also be worth experimenting with custom instructions -- in the default GPT-4 model, custom instructions can be included with requests like "be very detailed" etc.

swift osprey
#

interesting. So I can change my web browser user-agent to a mobile device and it might give me shorter responses??? 😄

#

giving it a try today

honest jay
compact girder
honest jay
steady dragon
#

Hey guys, anybody else experiencing issues with gpt4? I cannot use it, I just get the "Hmm...something seems to have gone wrong" message. But gpt3.5 works fine.

#

nevermind its happning to others

drowsy tinsel
honest jay
timber sonnet
#

Is there a plugin that utilizes github and able to read code within repositories to help assist with certain code whenever asked? Pretty much like a second brain to help with certain functions if I'm lost

#

Github's copilot is ok but it isn't exactly as helpful as gpt 4 is

digital prism
#

its answers aren't the same quality too, and I have no clue whats actually in its context. at least with GPT I'm controlling what code its looking at.

ancient oriole
#

guys, why do i have to do the "Am I human" test with every prompt, super annoying!

sonic folio
ancient oriole
#

nope, reported it on the main website, does it matter if i do it here?

sonic folio
ancient oriole
#

okay, thanks!

spring cedar
spring cedar
sonic folio
spring cedar
sterile forge
#

I am willing to donate my 3080 10gb for the cause ✊

sonic folio
timber sonnet
#

but i do 100% feel the auto complete, a lot of it is just rambling from other source codes and hoping that it works out

#

hopefully openai comes up with a strategy to implement chatgpt into vsc & github repos, would be an actual killer if gpt 4 had less restriction

marsh mural
#

wsg shawties

#

u can use my 3060ti if i get free gpt 4

#

wait i just looked

sonic folio
marsh mural
#

what.

sonic folio
#

Yeah it's highly parallelized and beefy.

#

MoE takes a lot of VRAM with so many parameters and a giant context!

marsh mural
#

how much is a h100s?

sonic folio
#

That's an excellent question, I never looked that up.

sonic folio
# marsh mural how much is a h100s?

From a quick search: around $30,000
The current (Aug-2023) retail price for an H100 PCIe card is around $30,000 (lead times can vary as well.)

#

So about 180k hardware per GPT-4 Turbo instance in GPUs alone.

sterile forge
sonic folio
#

Such a huge amount of money. Amazing we get to play with it for $20 / month, really.

sterile forge
sterile forge
sonic folio
hidden sage
#

I am subscribed to ChatGPT Plus. I am getting this error that reads, "Hmm...something seems to have gone wrong."

sonic folio
#

It'll pass in time.

#

Be patient, file a bug report, and it'll eventually work.

rugged escarp
#

i've had it for 24 hrs and can't do anything, it is on the app and web.

sonic folio
rugged escarp
#

I did, it didn't do anything. still happening. The app does it is with every query. On the web it may work after about 4 or 5 regens. I am not seeing this issue mentioned anywhere with openai, but after doing a short search on other threads outside, it seems many people are having this issue today.

candid wedge
#

Does anyone know why my paid account stopped working? Specifically model 4. It says to try again, nothing that support recommends is working

sage yoke
#

3.5 is working but gpt 4 isn't

vestal tendon
#

Yeah same is this affecting anyone on the teams plan?

vast birch
#

is there anyway to contact support here? cus at help.openai.com i feel like i'm talking to bots?

little pond
#

Does anyone else think things have improved the last week or two? Or did I just learn how to prompt the latest model update? I'm getting really really great stuff now. I was kinda upset because it seemed like everything was a non answer, tons of refusals etc. Now it's working maybe the best it ever has with some of the best reasoning and code I've seen out of any AI model.

civic lion
#

is gpt4 down rn? Every promt it just gives me "Hmm something went wrong" 3.5 seems to be working though

#

^^

somber fog
#

why is chatgpt3.5 10x faster on response times than 4?

somber fog
#

im about to cancel my 4.0 membership theyve broken it twice in the last week for me

ancient sage
#

Maybe its just me, but GPT4 has gotten much better as of like 2 days ago.

dire obsidian
#

I dont really see the point in making custom GPTs when they cant always access their knowledge files.

echo surge
#

?

hoary gazelle
#

"The conversation is too long, please start a new one." what? i thought you could go on and on

#

so what? eventually you lost all work and have to start again

echo surge
#

is that a thing?

#

can you send a screenshot

sonic folio
sonic folio
hoary gazelle
echo surge
#

idk

echo surge
sonic folio
hoary gazelle
echo surge
sonic folio
#

I don't know the precise limit. It's huge.

echo surge
#

i've never ran into it and ive had conversations up to like 50 messages i think

sonic folio
#

Like enough that you'll want to use a few instances to segment your summaries.

#

But I've definitely hit it in one message once.

#

I asked it to read chapter 2 of Bram Stoker's Dracula and analyze the characters relative to each other.

It took me literally and read it once for each character and ended the conversation in one prompt. ULTRA COMBO!

echo surge
sonic folio
#

It couldn't compute the request and ended the conversation in one response, yes. It stalled.

#

EOF.

hoary gazelle
#

ohh i misread that haha

#

i thought you got all your answers in one prompt

#

i guess it is time for me to sleep then

#

good night guys

echo surge
#

gn

echo surge
sonic folio
echo surge
#

with only one message? how would that happen

dusk valve
#

what happened to the ai or real channel/thing? did they remove it?

sonic folio
echo surge
#

thats weird if its only one message

#

never had that happen to me

sonic folio
#

It's only ever happened once for me.

#

And yeah, I was shocked. But when I read the instructions with a literal bent, it made sense.

#

It re-read it over and over again for each character's perspective.

tight island
#

1 your instructions consume tokens, 2 reading your response consumes tokens, 3 formulating your response consumes tokens

#

I think custom instructions counts towards that as well

#

So it’s quite easy losing context unless you use the full 128k api model

#

That one essentially has built in context management

tribal stratus
#

how much tb data was used to train chatgpt 3 and 4

tight island
#

Should be outlined on the website

#

But around 45 gigabytes

#

But the actual size is more around 570 for its training data size

echo surge
echo surge
dense halo
#

Are books made with chatgpt copyrighted to OpenAI?

#

@tranquil trench Sorry

tranquil trench
#

However you're encouraged to disclose the use of AI in what you make.

dense halo
tranquil trench
#

What do you mean sorry?

dense halo
tranquil trench
dense halo
#

Thanks

hardy socket
#

Guys chatgpt app is gone from my phone and cant find it in google playstore. What happened

#

Ping me if you respond

trail mango
#

I am trying to think what can make gpt4 cost lower then we can have more cap of using it....

#

Remember doesn't matter how many cap they offer us, we usually have to use few questions to correct the AI.

glacial horizon
#

I was using a custom GPT I created when it gave the error "Something went wrong. If this issue persists please contact us through our help center at help.openai.com.". I decided to test my other and new GPTs to see if it was an issue with the custom one and all of my GPT-4 chats gave the same error while the GPT-3.5 chats still work perfectly. Testing again it also gave the error "Internal Server Error".

I tried the steps on the help page except connecting to a different network as I cannot do that right now and they did not work, I got the same error.

There wasn't anything specific I was doing with the GPT, just generating backstory for an organization for my DnD campaign. It did occur when editing a prompt, but I do not think that is relevant as it not gives the same error on all GPT-4 chats I have created including new ones.

I was wondering if anyone else was having this problem or if it is a problem with my connection.

outer steeple
#

Can someone answer the following questions. I'm a little confused at the moment:
Which version is currently running ChatGPT for "Plus" and which version for "Free" and with which respective context window size?
Which version is currently running the Custom GPTs with which context window size?

If someone could answer that for me, it would be a dream 🙂

tight island
#

There’s AI’s to determine whether or not your book is written by AI

tight island
outer steeple
#

I did or not? 😄

hardy socket
tight island
#

Aka alà 4000 ish for customgpt

eternal rockBOT
#

Beware of possible scams or fraudulent activities that you may receive through direct messages. OpenAI staff will never DM you for any transactions.

Please report any incident by sending a DM to @languid valley immediately.

spiral siren
#

"!
Something went wrong. If this issue persists please contact us through our help center at help.openai.com." This is getting annoying openai_logo

maiden fractal
spiral siren
maiden fractal
spiral siren
#

Its a no im not using VPN.

maiden fractal
#

if you dont have a vpn installed on your ocmputer can you try firefox

spiral siren
#

Its still the same.

#

Even after opening it on Firefox.

maiden fractal
#

can you try on mobile phone without wifi

spiral siren
#

Looks like on the phone it says ive reached the maximum messages and will be avaliable after 18 minutes. Guess i will have to wait but seems wild that 40 messages have gone so fast

maiden fractal
spiral siren
spice bay
#

I have same isse with "something went wrong". Only 3.5 workjing on my machine and 4 gives erros all the time for past 2-3 houers. it jsut crashed in middle of a conversation.

  1. I have no VPN.
  2. I used all the browesers i could use.
  3. i cleared cahs and all athe steps taht were mentioned in help article on the site.

one thing that puzzlems me is it works on my mobile phone if i use wifi on same network. But if i use my pc that uses cable connection to same network that PC is connected to i get The error.

Any ides waht may be the cause ?

spiral siren
spice bay
spiral siren
spice bay
#

Well I use it now the GTP4 fom mobile over wifi and i have no masages stating - that i reached the limit. or that I have any erros it jsut works.

#

Same thing goes for using mobile data it jsut works.

weak jay
#

Hi is a place to discuss about subscription renewal ? No way to change my CB ?

spiral siren
hushed scroll
#

ChatGPT just crashes.
What to do in this case?

wild spindle
#

can u tell me how much is the current usage cap for GPT4 now every 3 h for plus users?

maiden fractal
wild spindle
maiden fractal
wild spindle
# maiden fractal For gpts yes

after 25 Messages i reseved : You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again after .... PM.

maiden fractal
tight island
#

i am.... grooot... nvm that i am looking for ppl who like writing novels and stories. i need a tester for my gpt, preferably one of the mods.

copper mica
#

Why I can't use the Talk feature on web? Is it available only on the phone app

tight island
#

would be nice tho

digital prism
#

I hit a new level of forcing GPT to give me full code back, when all else fails I give it the old copy, its lazy updated copy, then tell it to merge the "fragmented update" back into the main code. It has no choice at this point I guess, and it works lol

tight island
#

its nice

#

i've learned a way to leverage the "Continue Generating" feature everytime.

#

it's worked wonders for me.

placid cradle
#

Does anyone know if there are any plans to make an Apple Watch app? I feel like that’s the only thing missing. I wanna be able to just talk to ChatGPT directly on my watch.

mild jacinth
#

People subbed to Plus, can you please tell me if there's still an hourly limit on the use? And if so, what is it? A reply or DM would be appreciated.

honest jay
mild jacinth
placid cradle
# mild jacinth People subbed to Plus, can you please tell me if there's still an hourly limit o...

If you need more, you can get up to 100 per three hours with teams. Which is only $30 a month if you pay monthly or $25 a month if you pay yearly. BUT It is a minimum of two people. So if you just have the one subscription, that's really 50 or 60 a month. It worked out for me and my wife because we had two plus accounts anyway. Now we just have the team account with two people and only pay five more a month each.

tight island
#

100 per 3 hours is not enough

tight island
#

idk man i felt nothing

#

and i was on all night perhaps its regional

sick pecan
#

seems more rock solid than ever. but i am paid user