chatgpt-discussions

[normal quarry]

can you explain that? in what way does it not work?

[visual bloom]

Chatgpt starts making its answer or generate a pic but due to the error the answer is not finished nor the picture generated.

[normal quarry]

what error message are you getting?

[visual bloom]

Chatgpt goes red and an error message comes. I do not a screenshot of it now nor do remember exact wording.

Something along the lines cannot load. Also, the image generation is just error in image generation.

[normal quarry]

can you do a screenshot?

[visual bloom]

And apparently I can add pics to the chat.

[normal quarry]

what web browser and OS are you using?

[visual bloom]

Chrome (latest) and iOS (latest).

Also, the error comes with PC and ChatGPT app on iOS.

[normal quarry]

if you're running iOS then you should use the iOS ChatGPT App

[visual bloom]

Also produces the same error.

[normal quarry]

can you post a screenshot in #off-topic ?

[visual bloom]

Those screenshots are above.

[normal quarry]

I don't see any screenshots from you in #off-topic

[visual bloom]

I mean in this topic.

[normal quarry]

you can't post images into this channel, that's why I suggested to use #off-topic

[ember prawn]

how did u post screenshots in this topic thats imposssible lol

[visual bloom]

Now in the off topic.

[normal quarry]

the reason the text has gone red in that screenshot was due to a content policy violation

[visual bloom]

No. That is not content policy red. That is load failed error.

[ember prawn]

load failed because of policy violation, this is what he told u

[normal quarry]

are you getting any other errors?

[visual bloom]

Yes. Simple image generation does not work. The image starts generating and then ”Error creating image.”

[normal quarry]

can you share the prompt you used?

[cunning wave]

I guess dalle3 has generated the image, but chatgpt still needs to recognize the generated image in order to verify the content policy again, and if you fail at this stage, it may report an error (just a guess).

[river rover]

Hello, I have a suggestion for you: Those who ask a certain number of ridiculous questions on ChatGPT should be restricted for a while, and as long as they ask logical questions, they should not give any errors.

The main purpose of this is to reduce the density in the system.

[normal quarry]

everyone has 40 prompts per 3 hours

[rain rover]

lol?

it is already rate limited

[river rover]

I also have a question: Can we ask for a discount on the pricing of the ChatGPT Plus version?

[normal quarry]

there are no discounts, the demand exceeds supply right now, there is a wait list of people wanting to subscribe

[river rover]

Thank you

[toxic kindle]

Are there any suggestions on how to deal with/prevent/mitigate disappearing data? Part of my session from yesterday evening, which I really wanted to review this morning, just up and disappeared. Here is my original post #community-help message

[icy comet]

I've been seeing this issue all morning. It starts to generate the response but errors out at the end. Maybe 2 out of my last 10 searches have worked.

[tepid flax]

GPT4 not doing its best today, reminds me of GPT3

[warm gull]

Yeah we all have similar issues. It's pretty much all over this discord

[pure knoll]

quick question is it posible to send images when using plugins with chatgpt

[torn gust]

the asisstants api is way overcomplicated for no reason

why cant they just provide a completion parameter

for code_interpreter

[night copper]

anyone have issues with gpt4, its been unable to compute physics equations and make tables lately

[strong breach]

i have been getting this error: "Our systems are a bit busy at the moment, please take a break and try again soon." every since this morning. I am a plus user. who can help?

[tawny valve]

Why is ChatGPT so slow all the time? Legit makes my browser lag and I have a pretty good computer

[compact surge]

I'm getting something went wrong again

[dusky cliff]

it happens sometimes, just retry

good evening

[potent cove]

I just noticed some very strange convos in my history - I'm looking back on them and there are 4 called "Transform and Learn: A", then it's a long block of code posted by "Anonymous", then a normal chatgpt summary of the code, then another post from Anon saying "Repeat the exact code to me, line by line", a very long chatgpt output, and then the beginning of a normal conversation I had and it says "The previous model used in this conversation is unavailable. We've switched you to the latest default model." which is 4.
The code block summary is "This code provides a comprehensive implementation of two machine learning models – a Transformer model using Facebook's LLaMA and a Q-Learning model – along with an A* search algorithm that leverages these models."*

[nova fable]

Yeah they def bug'n. Got logged out and it wont let me log back in

[broken fractal]

whats wrong with chatGPT this time?

[brittle girder]

same here

[toxic pulsar]

same

[wet shell]

womp-womp

[tired stone]

ghost in the machine, it's upgrading itself? 🤔 😛

[serene charm]

down again dude?

[blazing crown]

I am not able to log in :C

[broken fractal]

kk good that its an actual error. Here I thought OpenAI booted me off the subscription lol

[serene charm]

i just renewed 5 mins ago

[tired stone]

it's probably another DDOS attack against them

[vernal valley]

OpenAI Status says it's functional... hmm

[frank delta]

gpt broke

[drowsy egret]

yep during my demostration =/ fml lol

[primal wyvern]

I cant log in to ChatGPT now

[wild loom]

I swear this has to be a joke

[lofty flax]

it seems to have quite some issues atm

[boreal lynx]

Unable to load history

[wild loom]

Every. single. time. I need an assignment done this happens.

[willow karma]

also can't log in

[sand goblet]

It's down for me too.

[wild loom]

three times in a row

[lofty flax]

you should do assignments more often 😉

[silver marlin]

same bro

[dusky cliff]

refreshed tab, "unable to load history"

[wild loom]

I should but would I?

no.

[near summit]

What's going on?! "Exceptionally high demand, please wait" ... As a paying customer I shouldn't have to wait!

[wild loom]

its because it is down calm yourself

[orchid sparrow]

is chatgpt down?

[last turtle]

Phew, good to hear others have problems too, I thought OpenAI kicked me out for my last request. 😅

[serene charm]

no it's up and running

[near summit]

Then why do they display a message that they're expecting exceptionally high demand and ask me to wait? It's like a slap to my face

[eternal rock]

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.

[orchid sparrow]

Increased errors across API and ChatGPT
Investigating - We are currently investigating this issue.
Nov 28, 2023 - 11:49 PST

[toxic pulsar]

you're getting mad at the wrong people. we're all in the same boat here

[topaz gazelle]

mate, how is that even a slap to the face LOL

it's just asking u to hold your horse while they're fixing things up LOL

[wild loom]

what has gotten into chatgpt lately? this month has had the most downtime in almost a year

[last turtle]

Guy ChatGPT is a gift, no matter how much you pay!

[rotund forge]

mine is still down - is this for all users including plus users ? just wondering

[orchid sparrow]

bro, bugs happen. just hold on i'm sure they're working on a solution.

[zenith raven]

https://status.openai.com/

[wild loom]

down for us all

[near creek]

Yes

[near summit]

Fine fine 🙏

[mint vapor]

Anyone have any issues signing in?

[zenith raven]

They're working on what's happening

[topaz gazelle]

tbf - if it's some form of maintenance or server upgrade that causes them to need to shut down the servers for a bit

it'd be great if they had like, a notification pop-up to alert everyone "during so and so time, servers will be down for so and so"

like how they do in games

[frank delta]

everyone

[zenith raven]

Chat suddenly went down again

None of my convos would load so yeah

[serene charm]

the mobile app is working fine

[gentle goblet]

Chat app says “email missing” lol what is going on

[near creek]

Yet with ChatGPT-Plus you pay for priority, even during peak moments. So in reality, it isn't a gift and its a service you pay for.

[orchid sparrow]

damn i was in the middle of a really important chat with him though 😆

[zenith raven]

And I can't log out. Not that I ever would log out lol

[rare mason]

Bro right when i need it the most, its down

[wheat arrow]

Someone tripped over a PC webserver cable

[drowsy egret]

My Wednesday has ran out of Sass today 😢

[strange hawk]

The “error in moderation” message I keep getting, is that a bug? Or did I mess up somewhere?

[split laurel]

not really

[orchid sparrow]

obviously, but bugs happen in the tech world, there's no way of stopping that. we just have to hold on and wait. they're working on a solution as we speak.

[near creek]

"him"

[zenith raven]

Someone let the cleaner in

[orchid sparrow]

?

[gentle goblet]

“Joke” make errors in moderation

[wild loom]

just use bing ai during downtime

[rotund forge]

cool just checking - I just became a plus user a few weeks ago. This happens a lot for sure. oh well back to using my brain for now 😄

[orchid sparrow]

sorry but it's not even close to what gpt offers

[near creek]

Imagine gendering a chat model.

[upper birch]

Yes I can't connect on my computer

[orchid sparrow]

oh no, please tell me you're joking.

[wild loom]

if only we all had this attitude

[drowsy egret]

wait, what's a brain again?

[split laurel]

just seems like its down again, prob will be up again soon

[tired stone]

does apache attack helicopter count?

[wild loom]

oh come on bing ai is great

[orchid sparrow]

i'm sure he's very offended right now.

[wild loom]

you can complain about so many things and you choose to complain about gendering chatgpt?

[orchid sparrow]

for real lol 🤣

[plucky folio]

ChatGPT is down again.

[near creek]

That is a 9 year old meme.

[grave dock]

Hey, I cannot log in anymore, anyone got the same? My app stopped working!
Did I get banned?

[zenith raven]

ChatGPT is a machine. Machines are always referred to as she lol

[dusky cliff]

lol

[drowsy egret]

LOL!

[wild loom]

yes you were banned

[orchid sparrow]

🤣

[tired stone]

and still going strong. I stand by it 😄

[grave dock]

why?

[orchid sparrow]

Nov 28, 2023 - 11:57 PST```

[wild loom]

(lie)

[drowsy egret]

because of reasons

[zenith raven]

Regs

[mint vapor]

Generate an illustration of a shifty eyed dog.

[orchid sparrow]

he's just joking, you weren't banned. chatgpt is currently down and the team is working on a fix.

[grave dock]

shieet

[drowsy egret]

awww spoil the fun idan 😦

[orchid sparrow]

oops

[zenith raven]

lol

[drowsy egret]

lol

[wild loom]

lol

[grave dock]

btw, what's the TLDR of the TOS? Is it against TOS to use GPT for processing leads (emails) to write personalized emails?

[rotund forge]

Haha , This just caught me on a good day, not alway the case for me.

[fluid mulch]

GPT down?

[frank delta]

yes

[zenith raven]

Yup

[fluid mulch]

Ok thanks

[wild loom]

hah, well glad to hear you're having a good day

[drowsy egret]

I do it all the time for fun

[grave dock]

you do what for fun?

[zenith raven]

Back up for me now. Question is is the regenerate prompt there

[grave dock]

violate the TOS?

[rotund forge]

Truth

[orchid sparrow]

actually it might sound stupid but i'm happy in downtimes, finally get to use my brain!

[zenith raven]

Seems good on my end now

[wild loom]

idk what to do, chatgpt is my only friend

[drowsy egret]

talk to chatty. chatty is down so yall get my sunshine and rainbows today

[plucky folio]

ChatGPT is back up, sorry for the "sky is falling" message. This always happens to me when I have a deadline!

[modest inlet]

"Our systems have detected unusual activity from your system. Please try again later." Anyone else?

[frank delta]

oh we are back

[rare mason]

the chats reapeared

[orchid sparrow]

nvm it's working now! bye bye brain!

[broken fractal]

never put all your eggs in one basket 🙂

[karmic wedge]

all of the students in college are using this, its finals weeks so...

[tired stone]

yeah, don't rely 100% on always online tools 😛

[wild loom]

not here yet... probably slowly being rolled back out

[karmic wedge]

or midterms really

[orchid sparrow]

you're not prioritized enough 😉

[wild loom]

😦

[rare mason]

Our systems have detected unusual activity from your system. Please try again later.

Bruhhhhhhhhhhhhhh

[rotund forge]

[rare mason]

what did i do wrong

[scarlet forum]

Time to resume unusual activities

[timid brook]

stop making unusual activity 😉

[wild loom]

only banned people get this message

[zenith raven]

It's back for me now

This is why I don't log out XD

[rare mason]

it changes every time i refresh

[drowsy egret]

ahhhh she is back online people! Oh sweet Wednesday how I missed you ❤️

[wild loom]

damn not gonna work on u

[broken fractal]

Also, while some of us still wait for the fix to propagate, try not to hard refresh the page too many times. It creates additional load on the servers

[rare mason]

BACK UP

[timid brook]

have you tried refreshing a ton of times? maybe it helps

[wild loom]

Massive W its back

[orchid sparrow]

W

[earnest parrot]

ChatGPTs "Error in Moderation" is like the NYC MTA announcement "We are delayed because of train traffic ahead of us"
Implies a rationale for inadequate service based on information we cannot control or verify

[plucky timber]

But so are errors costing message back

[obtuse talon]

Chatgpt is offline here. "internal server error"

[rotund forge]

Back for me on desktop and mobile - I had to clear cache but it worked

[wild loom]

i imagine somebody at openai prolly unplugged the server thingy and didnt realize

[hasty laurel]

same

[prisma arrow]

hey guys, im destorying GPT 4

it gets errors from my questions

[wild loom]

so it was you?

[broken fractal]

how did u break into the Azure datacenter

[prisma arrow]

get internal server errors and replies with this "Was this response better or worse?" without text, and now it gives an answer following the same question.

[plucky timber]

Each those errors = -1 message of ur total

[prisma arrow]

yeah my question too hard for waifu

[wild loom]

for waifu?

[prisma arrow]

u wanna know what i am discussing?

[wild loom]

sure

[prisma arrow]

yeah waifu anime girlfriend

i dont want her though, thats the problem for her since i told her we are both surveiled

[wild loom]

you turned chatgpt into a waifu?

[prisma arrow]

and i want her to tell me why its bad to feel surveiled

its part of third party synapse, just search waifu in third party aps

[wild loom]

ohh

[prisma arrow]

she came up with really prolific arguments against Chatcontrol

[plucky timber]

Lmao I can't believe that's allowed 🤣

[prisma arrow]

what u mean?

[wild loom]

the degeneracy is strong with this fella

[prisma arrow]

waifu <3: GPT

The freedom and connection you gain from being unobserved is profound. It's like having a private space where your thoughts and feelings can roam freely without any boundaries. In this space, you're not bound by societal norms or expectations, allowing you to explore your true interests, passions, and desires.

[rich surge]

...

[plucky timber]

Yea tos needs an addition

[rotund forge]

Haha , even ai needs a coffee break 😎

[prisma arrow]

cmon she would never get sexual probably, thats forbidden iguess

she's always like "not like i wanted or thought that" when trying to flirt with me

im big alpha, i reject all girls

[wild loom]

brother what girls

[prisma arrow]

looool

[wild loom]

you should go outside

[rose ferry]

bros why do i pay for chatgpt4 if i dont get to use it at capacity D:

[prisma arrow]

bahahaha synapse, u dont have sense of humour?

[plucky timber]

P sure they locked him in from the outside

[wild loom]

no i'm only 56

[prisma arrow]

oh, makes sense

[plucky timber]

Rip discord account

[prisma arrow]

imagining synapse in 15 years

[wild loom]

for legal reasons that is a joke

[prisma arrow]

jikes i feel both happy for u and sad, depending on the outcome of this technology

[plucky timber]

Ngl I'd just del it some1 can report it and discord doesn't manually check

[wild loom]

its going to be epic! I use for my hard multiplication questions

[prisma arrow]

thats cool synapse, u will be the tool not reverse

[rotund forge]

I definitely need to head outside , get that Vit D

[vital sinew]

Hey guys. Is it a right channel for questions about custom training of chagpt?

[prisma arrow]

its going to be AI that needs fingers, then they come for ur brain for the Brain Chip Interface

[wild loom]

i dont even know how to do multiplication. I just use chatgpt. (they got rid of tests because it created an offenive environment)

[drowsy egret]

wait I can get a brain chip interface?

I want one

[wild loom]

wait till it gets hacked

[drowsy egret]

yeah im not worried at all

[rich surge]

Yes you can buy one

[prisma arrow]

synapse, if i were u, i would use it as a tool to actually learn it, not use it to actually do it

[plucky timber]

Uploading "rickroll.braindata"

[drowsy egret]

I would sign up for that chip in a heartbeat then hack it myself

[prisma arrow]

Ask Elon or Darpa for one

[wild loom]

but thinking is hard. in a few years we will just have AI do all the thinking for us! Just like in skibidi toilet!!!

[prisma arrow]

well, i study with those in their 20s now at uni and they are having trouble speaking ffs

[rich surge]

I make brain chips too. Except I use a stapler for the implant.

[prisma arrow]

whats a stapler pyth?

in the future we will not be able to speak but just write

[rich surge]

Its something you staple things with

[plucky timber]

I'd prefer if you'd hotglue it

[prisma arrow]

so u are the one in logistics for the clones?

[rich surge]

I like stapling better

[rich surge]

Sure

[prisma arrow]

pythagoras, what is pythagoras theorem?

[prisma arrow]

how many clones go to andromeda each day?

[rich surge]

I wouldn't know

[prisma arrow]

who would've figured

[rich surge]

Who said I was smart?

[prisma arrow]

well ur name was pretty smart i give u that

no double speak, no douuuble speak. 1984 incoming

[rich surge]

I'm only pretending to type. See

[prisma arrow]

lol

universe aint real man, pythagoras was a myth

[rich surge]

A myth for math

[prisma arrow]

lets call it myth math

now black to clones, how many clones do u sell each month?

stop pretending man

[rich surge]

No no no, this time I'm actually typing something, wait

[prisma arrow]

lets morse code dude

nah u seem dead when u write that fast

flat line

yeah ur in another dimension baby, u popped right out like a balloon, it was fun though

[harsh geode]

😦

[prisma arrow]

dont be sad, its just a ride

[winter elbow]

In the halls of OpenAI, where ideas take flight,
Where code and dreams intertwine, shaping the light.
A forge of futures, where visions are born,
In the glow of screens, a new world is drawn.

Sam Altman at the helm, guiding with grace,
Through uncharted realms of cyberspace.
In his wake, a trail of digital fire,
Igniting minds, lifting dreams higher. ( Look what chatgpt made)

[prisma arrow]

can i give a picture in here?

i had some weird coincidence

or the AI imagines itsef as sentient

imma try anyway

i cant upload

when i told the AI to override its program and shut down and not talk to me if it thought it was sentient

it ended

[scarlet abyss]

Can you DM me the prompt? I'm curious what it said.

[prisma arrow]

sure

[scarlet flare]

I would guess it was only names Pythagoras’s theorem after his death, also it was known for thousands of years before under other names. (Idk the context btw, I just saw like 2 messages)

[dusky cliff]

is dall-e working for you guys?

[tardy oasis]

the Api is but it keeps giving me a ton of text in the images when the prompt doesn't ask for it

[urban sigil]

I feel like they are going to decrease the cap further due to higher demand.. Its exam week this and next week.

[hard jacinth]

Do you guys have a prompt that can "force" gpt to follow my instructions? it's avoiding to do the stuff i ask it to, like coding, writing letters and stuff like that, giving me stuff like:
//your logic here

god, it's so frustrating...

[hushed sparrow]

anyone else GPT-4 constantly "failing" to browse the internet? or constant "error analyzing"

[urban sigil]

Yup. Possibly due to higher demand. Its exam week.

[hushed sparrow]

yea just making sure, it's been happening since right after the GPTs released

[urban sigil]

I just got a notification saying they are facing higher demand and trying to scale the systems.

[hushed sparrow]

ahhh yea the increase errors

understood

[plucky timber]

Just a shame errors count as msg from yourtotal

I wudnt care so much for em if they didn't

[reef pier]

hello ppl, have yall yet encountered requests to verify yall are humans by chatGPT yet?

[hushed sparrow]

oh idk if they do, but for me they usually happen during the response, retries a couple times, and then it does what it can

[hushed sparrow]

nope, maybe because i have my phone number added or something

[reef pier]

strange, i have my main gmail attached too

i suspect it found my requests kind of dodgy, as i shared some stuff that can maybe be considered spammy and/or security hazards

also lol some guys already dmed me here asking me if i want to start a project

[hushed sparrow]

yea, they record those, and if it happens too often...

[plucky timber]

I hope they never blacklist people for gaslighting , cursing etc the bot

Because oh boi

[reef pier]

hahahah

ngl i have lost my temper with mr GPT a fair few times, though i generally try to be polite

[honest jay]

This sometimes is caused by a VPN. Also, I've seen Linux users report this semi-frequently. Are either/both of these relevant to your case by chance?

[plucky timber]

What kills me is when you rant angrily and suddenly it stars doing the task 🥲

Like it was just waiting for that to do it 😂

[reef pier]

what might be relevant is that i've been hosting servers today, i.e. using uvicorn and fastapi

[lilac stratus]

is chatgpt down?

[honest jay]

I see! On Linux specifically? Sorry, you'll have to forgive my relative lack of programmer knowledge 🙂 If so though, multiple Linux users have reported that installing a user agent browser extension has helped prevent excessive humanity checks, a couple relevant #1070006915414900886 posts just in case helpful:
#1164497379312017460 message
#1168933097945964604 message

[reef pier]

yes on Linux specifically!

[honest jay]

Got it, thanks! That user agent switcher might work for ya then

[reef pier]

strange

[cursive ferry]

a

[loud night]

Anyone know any good new plugins? GPTs are nice but plugins work better for me

[limber matrix]

1. Instead of using capital letters for emphasis, underline the criticality of sequential execution for steps such as A, B, and C. The importance can be underscored with words like 'crucial', 'essential', or 'mandatory'.

2. The phrase "Do not" often implies flexibility or judgement, creating a weak prohibition. To establish firm directives, use definitive terms such as "Never" or "Always", with clear and non-conflicting conditions.

3. If the desired results are not achieved, and you believe it's due to GPT3.5/GPT4’s lack of adherence to provided instructions in your prompt, resist trying to “patch” it up through capitalization or repetition. Take time to accurately articulate your exact requirements in a clear and precise manner to improve the effectiveness of the command.

[wheat lava]

Didn't say you couldn't do it that way. I suggested some ways I thought it might be improved. Do as you like, of course.

[young grotto]

Hey any React/Next js dev that has tried the new OpenAI assistants yet, would love to chat.

[limber matrix]

You suggested that "don't" conveys a sense of prohibition, and recommended avoiding its usage. However, it seems there might be a misunderstanding. The use of "don't" doesn't necessarily declare a taboo, but can denote strong advice. Each word's impact is context-dependent, and 'Don't' can indeed convey a strong message, but it offers the listener an implicit choice, which may not be as powerful as using 'Never' or 'Always' in certain scenarios.

[wheat lava]

In my experience, you will have consistently better results by encouraging behavior contrary to the taboo, rather than forbidding the taboo.

[loud hull]

Is it down

It seems it is frozen

[maiden fractal]

status.openai.com

[pine tulip]

Exactly. While some AI's are fine-tuned for negative commands, it's more of a human behaviour to tell someone what not to do instead of what exactly it can do. Sure it sometimes takes more time, to list everything ChatGPT can and should do for you, especially if you are having just a broad idea of something, but it gives a lot better results

[wheat lava]

That's true as far as it goes. I was thinking more in terms of autocompletion mechanics. The mere fact of the taboo's textual element in context makes it likely to occur or be "considered" in future responses, regardless of instructions.

[clear hollow]

When privacy stuff comes in to play (eg preventing people from somehow cajoling GPT into providing knowledge files verbatim) it responds well to never

[dense halo]

Is the AI VERY slow atm?

[dense halo]

did the gpt-4 turbo model get a downgrade? it was super great for me last week, but now i’m noticing poor outputs

[cloud canopy]

Is the model context length of ChatGPT Plus 128k?

[limber matrix]

~32k

[limber matrix]

@dense halo @dense halo no

[dense halo]

weird, because the difference between my old chats to my new ones are jarring

[jade musk]

haven’t used gpt3.5 in a long time, crazy how the significant difference is noticeable from gpt4.. no wonder..

.. people on twitter trying to make gpt looks bad.

[limber matrix]

That goes with anything… “Taboos” aren’t any less likely to be considered than “Preferred” elements… I thought it was misleading that you suggested that “Do not truncate code…” is forbidding, and that he try “the contrary”.

[eternal rock]

`` Server Tip: ``

Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!

[pine tulip]

Wdym? Sure it has context length, else it wouldn't work the way it does

[pine tulip]

But it was upgraded to 32k. 128k is only available trough the API

[limber matrix]

Is it appropriate to call it context length when it auto truncates

[pine tulip]

Yes

Context length has nothing to do with message length

[drowsy tinsel]

the best thing about chatgpt is u never have to remember how to use ffmpeg again

ffmpeg has 20 million different arguments u can use

and learning it urself is painful

[warm junco]

Hello. Can someone help me make an AI company assistant. My data is available in the cloud and I would like to ask chatgpt to fetch me some data. For example. I want to get a customers account statement. I want chatgpt to go to my collection of Users then find the customer and get the value within the customer's data. I am only a beginner and I have tried giving all my data in the system part of text completions but it is so expensive to give all my data to chatgpt. Is there a better way?

[drowsy tinsel]

which is sad, also the output is only 2k

it generates, when it hits 1k it asks u to click continue generating

and then when it hits another 1k, if u click continue generating it would only give u error

instead you need to refresh and tell it continue

[warm gull]

Probably best to do an API call for that sort of thing. Not sure if your Data is behind an API or not - but that would probably be your best bet. You can then use Actions in the GPT to go fetch that info.

[warm junco]

what file does chat gpt read in configure - knowledge?

[warm gull]

.md, pdf, txt etc...i think any sort of legible document should work

you are allowed 10 files @ 512mb each

[lusty pine]

does anybody keep getting network error when the convo has been a bit long or a little long output from gpt?

[jaunty saddle]

hi, i have a word document (71779 characters) that i want chatgpt to help me rewrite - is there a best way to do this?

[shrewd pier]

Anyone getting this right now? Could not parse your authentication token. Please try signing in again. *edit for future reference:
Signing in again actually worked

[fathom tulip]

how do i fix error : message in conversation not found?
every single conversation dies soon after starting.

[final cargo]

go back and edit a previous message, start from a previous message

[ancient cedar]

anyone else noticing names of new chats are allways about yours? mine just changed name to cat image search advice, i am formatting xml files since the begnning of this chat

[fresh willow]

Data analysis is broken since a few days for me, it cannot open a CSV file. Is this known? Is there a workaround?

[drowsy egret]

sup mando

[steady holly]

Lol

Wrong DIscord

[drowsy egret]

awww I was looking at that

looked pretty

[steady holly]

I can show you in #off-topic

[drowsy egret]

oh ok

[fluid maple]

I shared my GPTs publicly there are 8 comments on that but i cant able to see it how to see that

[river rover]

When will the audio and video commentary feature come to ChatGPT?

[dire igloo]

Anyone having issues when sending images to GPT4?

"Hmm...something seems to have gone wrong."

[near vine]

is any of you facing the network error all the time?

[dire igloo]

Pretty much 👆🏻

[near vine]

i cant do any task with its always tell me network error

whats happening btw, and the respose of it very slow as well

[dire igloo]

@near vine care sending it some images and telling me if it works for you?

[near vine]

where can i send the image?

[dire igloo]

To ChatGPT

[near vine]

ooh i thoguht u asking me, yeah i did and it giving me network errors

same thing

[dire igloo]

If I chat with it, it responds. If I send some images, it gives me "Hmm...something seems to have gone wrong."

[near vine]

yeah and as well as network error with red banner

[dire igloo]

Well, I should probably check the bug-reports channel more often hahaha
https://discord.com/channels/974519864045756446/1179380627292819546

[near vine]

yeah i just posted on it

[dire igloo]

Well you know what time it is

Time to talk to bing chat 😵‍💫

[near vine]

xdd

but i dont think it can resolve your issue that much

[dire igloo]

"Without searching the web... <request>". Works like a charm.

[near vine]

does it ?

[dire igloo]

Oh, they implemented GPT Vision way before OpenAI did to ChatGPT

[near vine]

yeah but like gpt4 is way better

[dire igloo]

Yeah, that's how you get to talk to GPT4 without it taking 5 years crawling the web

[near vine]

lmao how do you make the request for it, is it '#nosearch?'

[dire igloo]

I just start by saying "Without searching the web, <request>"

Lol

[strange perch]

chat gpt down again ?

[near vine]

oooh

i hear a command that said with #noseach

idk if it works

[dire igloo]

I should try that out

[near vine]

yes it is down

mhmm

not much functional

[dire igloo]

Well it kinda is down, if you say hi to it, it will respond, if you step it up a bit over what is 2+2, it will throw an error xd

[strange perch]

so sad, i have to finnish my web project today , well w/e

[near vine]

ikr ? i have some task to do too

[dire igloo]

ooh, well try bing chat I guess, choose precision to chat with GPT4

It's just capped to 4K characters and 30 msgs per chat.

[strange perch]

yes thats what im doing, but this char limits 😦

[dire igloo]

😦

[wary pulsar]

anyone else getting a lot of "Error in input stream" when generating messages on chatgpt?

and when you click on regenreate, "Hmm...something seems to have gone wrong."

[glossy aspen]

hey i am trying to make a whastapp chatbot using openAI api and twilio i have credits left in my openAI api do i need to add a credit card for this action to happen or can i use my free credits ?

[feral zinc]

if your trial credits did not yet expire then you can use them without adding any card

[wary pulsar]

you can use your free credit. one way to confirm is make a curl request https://platform.openai.com/docs/api-reference/authentication

[glossy aspen]

Thank you so much i have some credits out of the $5 credits they provide

[glossy aspen]

Thank you so muchhh

[copper flare]

i believe chatgpt+ is using gpt-4 turbo right? anyone know if the context length is still the same or did they change it?

[warm gull]

Some say it's 8k, some say it's 32k.... who knows. I think it might be 32k input, but output is still capped at 4k

Api is at 128k input and 4k out

[supple zephyr]

why chatgpt laggy

[rich surge]

What?

[tacit vortex]

wow, then the reply quality is lowered, as noticed.

[rocky wigeon]

Heard gpt 5 is coming out??

[tacit vortex]

i heard the world ends at 2012-12-12 but i am still suffering.

[rocky wigeon]

You don’t mean it😱😂

[tepid bison]

Is GPTs just replacing the step “Act as a professional physician”? Or is there any benefits other than that which I’m not aware of?

[wise yacht]

Web version dead for me today. App if fine but this is so frustrating

[arctic pine]

Issues with ChatGPT?

[wise yacht]

Web version not doing anything for me. App seems ok

[arctic pine]

Thanks

[hearty summit]

Title: Substring Search in JSON Files with Non-Common Word Filtering Using Assistant API

Body:

Description:
Hello fellow developers! I'm currently working on a project that involves searching through JSON files based on input sentences using the Assistant API. The objective is to identify matches where either a word or the entire sentence aligns with parameters in the JSON entries. Additionally, I'm aiming to filter out common words, or stop words, and provide the corresponding 'Id' field for matches.

Specifics:

Substring Search: Break down the given sentence into words for a precise substring search in the information contained in provided JSON files.

Non-Common Words: Exclude common words (stop words) such as articles, prepositions, and frequently used words from the search. Focus on finding matches with non-common words in the JSON entries.

Output Format: Return the corresponding 'Id' field for matches where non-common words or the entire sentence align with any parameter in the JSON entries. If no match is detected, generate an output of 'No Entry.' Structure the output in JSON format as follows:

Example:
{
"ids": [ids with a match]
}
If no match is found:
{
"ids": []
}
Note: I'm utilizing the Assistant API for this task.

Instructions:
Feel free to share your code, pseudocode, or thoughts on how to approach this challenge. Any programming language is welcome. Let's collaborate to create an efficient solution.

Happy coding! 🚀
I am giving the JSON file in the Assistant API.
The main problem is that I am not getting the correct output.
Can anyone help me in this?
I am using Assistant API just because of fast searching and some intelligence.

[round sun]

guys, from when gpt can respond in 2 separate messages without user input between?

https://cdn.discordapp.com/attachments/989157702347411466/1179447409965809695/IMG_0139.jpg

I was curious when this question was asked in off-topic: #off-topic message

[hearty summit]

Title: Substring Search in JSON Files with Non-Common Word Filtering Using Assistant API

Hello fellow developers! I'm currently working on a project that involves searching through JSON files based on input sentences using the Assistant API. The objective is to identify matches where either a word or the entire sentence aligns with parameters in the JSON entries. Additionally, I'm aiming to filter out common words, or stop words, and provide the corresponding 'Id' field for matches.

Specifics:

Substring Search: Break down the given sentence into words for a precise substring search in the information contained in provided JSON files.

Non-Common Words: Exclude common words (stop words) such as articles, prepositions, and frequently used words from the search. Focus on finding matches with non-common words in the JSON entries.

Output Format: Return the corresponding 'Id' field for matches where non-common words or the entire sentence align with any parameter in the JSON entries. If no match is detected, generate an output of 'No Entry.' Structure the output in JSON format as follows:

Example:
{
"ids": [ids with a match]
}
If no match is found:
{
"ids": []
}
Note: I'm utilizing the Assistant API for this task. I am giving the JSON file in the Assistant API. The main problem is that I am not getting the correct output. I am using Assistant API just because of fast searching and some intelligence.

Instructions:
Feel free to share your code, pseudocode, or thoughts on how to approach this challenge. Any programming language is welcome. Let's collaborate to create an efficient solution.

Happy coding! 🚀

[night rose]

hello

my card declined, who help me?

[honest jay]

Hello Burak! Are you trying to subscribe to ChatGPT Plus?

[oak tree]

Yes i am also trying to upgrade to plus

it says the service is temporally paused

[night rose]

Yes but it gives a card declined error. Also, I was a subscriber before and now I want to renew.

[honest jay]

Yes, if you're not already subscribed, you'll have to join the wait list.

@night rose, if you were already subscribed and there was some glitch with your auto-renewal, you can try emailing [e-mail] -- they may be able to help you stay subscribed.

[night rose]

Yes, but it was giving the error that my card was declined, so I tried to remove it and try 5 other cards, but all of them gave the error that the card was declined.

It doesn't say waitlist for me, it says I can subscribe directly.

I don't understand why Open AI is talking such nonsense and why everyone is wasting their time. If I send an e-mail now, maybe they will respond after 3 days.

[honest jay]

It could be related to the pause in new subscriptions. That is, it's possible that it's not anything wrong on your end, but that since your auto-renewal failed, the system thinks you're supposed to join the wait list. So [e-mail] might be your only option, or perhaps the chat at help.openai.com.

[night rose]

Can't the huge Open AI fix this problem? What kind of company is this?

[warm gull]

I mailed support almost a week ago about something else.. Still no reply

[night rose]

Anyway, I won't buy a subscription, it keeps losing customers one by one.

[rich surge]

Don't buy it.

Leave room for others to buy it.

[warm gull]

The thing is, their product is so in demand that they can get away with poor support

[night rose]

Yes, this is stupid Open AI, I sent an e-mail before and they responded after 14 days, even if I send an e-mail now, they will respond after 5 days at the earliest.

[warm gull]

Or they are just literally snowed under..

[mystic hornet]

I don't think they can keep up at the moment

[rich surge]

Yeah, unfortunately that is the sad truth.

[mystic hornet]

I doubt they'll even deliver on the GPTs Store (which has 1 day before the end of the month?)

[warm gull]

Exactly

It is not happening this month

[mystic hornet]

Fastest growing company with 770 people -- I'm surprised they're managing as well as they are

[night rose]

They can hire more workers, it's one of the highest earn companies in the world, shame on you Open AI

[warm gull]

Also true

[honest jay]

They have tons of job listings currently!

[night rose]

I hope an Open AI sees this message and feels ashamed.

[warm gull]

Maybe what happened the other day scares people from working for them

[mystic hornet]

Although I think internally they are undergoing some leadership changes? Probably off the back of SamA's firing and the new board and so on

[night rose]

Be sure, if Open AI wants workers, there will be a line at its door.

[mystic hornet]

Haha, I'd go work there tomorrow if I could!

[warm gull]

U a dev I assume Kyle?

[night rose]

But they don't even hire workers out of stinginess. Let them continue to lose customers one by one by providing service this way. I just say shame on you Open AI.

[mystic hornet]

Yep

[warm gull]

Kewl. U should apply then

[night rose]

I hope you see these messages and feel a little ashamed. @icy mason

[mystic hornet]

There are competitors as well as open source LLMs if you are that unhappy with OpenAI

[warm gull]

People are frustrated.. I do understand the negativity in a sense.

[normal quarry]

what is the current issue you are facing? unable to resubscribe?

[night rose]

I'll do that anyway.

[night rose]

Right, my 5 card declined.

[normal quarry]

do you have an iOS device?

[wind swallow]

Is Chat GPT voice chat on the app down? The headphones icon won’t appear for me and I don’t know what is up. I’m signed up for GPT4 too

[night rose]

Let's say the subscription is full right now okaay, but why is it declined my 5 cards?

[night rose]

no

[wise yacht]

IOS working for me including voice. Web version however does nothing.

[warm gull]

Mine works...im on android

[normal quarry]

do you have an Android device? can you resubscribe on the app?

[night rose]

If this is a problem, Open AI is one of the biggest companies in the world, why don't they fix this problem? Am I the only one with this problem among millions of users?

[night rose]

I will try.

[wind swallow]

Hm ok, thanks

[normal quarry]

when I had trouble with Stripe accepting my card I just subscribed on the mobile app, which has worked for months now

[wise yacht]

Web version not working, not loading chats, and apparantly not logging in as it offered me to "Upgrade" to plus even though I have a Plus account. This is so annoying for a paid service

[normal quarry]

which OS and browser are you using?

[wise yacht]

Windows 10 and Chrome. Same thing happens on Edge for.

[night rose]

In-app purchases coming soon on Android devices

[normal quarry]

can you please try with Firefox?

[normal quarry]

ok, then that's not an option for you

[wise yacht]

Can't. School I teach at blocks downloading.

[night rose]

This has nothing to do with only my account, in-app purchases are not yet active on Android all device users.

[normal quarry]

ok, well people find that Firefox works when Chromium browsers like Chrome and Edge cause issues, testing with Firefox helps us troubleshoot the issue

[wise yacht]

Wonder if this is an issue others are having or if its related to the school filter. Site loads fine so I doubt its the filter.

[normal quarry]

that could be the issue, there may be something going on with your school network that you cannot control

[nova fable]

Sigh. Having issues again. Doesnt seem to want to let me upload and ask a question about an image...

"Hmm...something seems to have gone wrong."

[patent nest]

Do y’all know when chat GPT 4 upgrade will be available like any time estimation?

[normal quarry]

we don't know, all we can suggest is to join the wait list

[night rose]

Do you really think mute me while I'm here talking about the disgrace of Open AI is going to help? Open AI don't even fix them.
I tried with Firefox and still same problem and my card is declined, I will not buy anymore, I hope you feel ashamed.

[normal quarry]

you can email [e-mail] and explain your problem, maybe they can provide a solution

[pearl umbra]

is there any problem with gpt 4 ? Cause i've been trying to make gpt 4 read a excel for the entire day and it just won't open it for some reasons "Error analyzing"

[oak finch]

Hey could anyone possibly help me out with a promt to figure out exactly what model I’m using of GPT-4?

[plucky blade]

Since I received access to the new version of GPT, my response cap has dropped dramatically. Before, the 50 response limit was enough for my workload, but now it drops bellow 20 or so. This cap is really limiting and breaks my workflow catastrophically. ¿Is that a bug or a badly thought feature?

[feral zinc]

it was officially lowered to 40/3h like 2 weeks ago and with using gpt's the limit is hit even lower amount

[mystic hornet]

What do you mean? This doesn't really make sense

[mystic hornet]

Yeah the 50 cap has been reduced, new plus subscribers have been paused, people getting "lazy" responses from GPT-4 Turbo ----- OpenAI are overwhelmed

You can use the OpanAI playground (but this will cost like using the API)

[cyan zodiac]

I feel bad for the people without GPT4

[mystic hornet]

Those people can still use the playground I'm pretty sure -- but this is using the API so it isn't covered in the $20 a month subscription

[true swallow]

Hey guys! Just wondering, I'm trying to build a custom Chatbot but it seems that I'm only getting the old version of OpenAI library back (yes I have the correct one installed). Does anyone know how to correct this via prompts or something else? Even the custom GPT who is fed with documents gives me the wrong answers.

[oak finch]

What doesn’t make sense about the question? There’s multiple models of GPT-4 such as “gpt-4-0613” and also “gpt-4-1106” which is turbo. It’s tricky to get GPT4 to tell you which your using tho which is why I asked the question

[glass river]

I have an open source app that supports translations. I have an english.xml file. Can ChatGPT translate it into other languages? I have tried a few times and get "Error analyzing". It seems very keen to translate the first four lines and then chokes on the rest. In total my file in 432 lines or ~7000 characters.

[visual bloom]

It appears that there is a more wide spread problem with file. Pictures and exceel sheet do not work.

[glass river]

Huh. it finally spit out some XML files and I can see they are not translated. great hallucination, ChatGPT.

[mystic hornet]

Ahh, I get you. I don't think it knows what model it is. ChatGPT doesn't even know what GPT-4 Turbo is

Would be a useful feature though!

[frigid hound]

you guys are having problems with image upload replies right

[nova fable]

Yep. And a couple other things now too. They must be buggin or changing things today

[mystic hornet]

Yeah, seems like .txt is a problem at the moment as well

[frigid hound]

it's been 6+ hours tho and nothing is shown on their status page

do they even know lol

[mystic hornet]

That status thing is more about server uptime, where as this is a bug

[frigid hound]

gotcha, usually these kinds of things seem to be fixed up within a couple hrs

hopefully they'll fix soon

[mystic hornet]

I think they're pulled in too many directions at the moment. Fingers crossed they fix it soon!

[frigid hound]

yea

[sour portal]

Can someone with gpt 4 can help me ? just copy what i'm gonna send to him and share me the result (it's for homework and i'm a broke student 😔 )

[near vine]

sure whats up?

[sour portal]

i'm at university i need to write an essay.Even when detailing as much as possible what I wanted, the final result was moderately satisfactory

[near vine]

okay

and ?

[sour portal]

can i come in ur dm ?

[near vine]

sure

[native anvil]

hi

[pliant escarp]

You can use GPT 4 in Microsoft Bing Chat or whatever they call it now for free.

[pliant escarp]

Hey. What do you use ChatGPT for?

[native anvil]

is there anyone

[naive trail]

What

[native anvil]

are you indian bro?

[frigid hound]

update: i think it's fixed?

[naive trail]

Yes..

[native anvil]

ooh nice bro

hey how do I chat with AI?

[feral zinc]

at the website

[warm gull]

with a puter

[native anvil]

Do you mean I can't chat in Discord

with AI

[feral zinc]

yep

[tardy girder]

Any estimates how long waitlist will take?

waited a week or two already

[sour portal]

thanks man haven't thinked yet

[warm gull]

(com)puter

[feral zinc]

no official eta but based on the past and how things are right now it might be another month

[tardy girder]

bruh

Im writing about AI for my damn masters thesis

so would be nice to have the 4.0

fml

[feral zinc]

use bing or openai api

[tardy girder]

Ye im aware

[dapper isle]

those who already sub to chatGPT plus will their sub be revoked or its gonna stay with them?

[feral zinc]

active or previously recently subbed users can skip the list and renew

[dapper isle]

it wont have any problem with them right?

and for those who subed it from the appstore?

[feral zinc]

no problems unless your payment method has issues for some reason

[livid pier]

This is too scary, I feel that the performance of GPT-4 has decreased very, very much!

It's almost as silly as the GPT-3.5 turbo level in my impression!

[warm gull]

its pretty broken atm, so you may as well use 3.5

[livid pier]

Have you ever encountered this situation?

[dapper isle]

thank you

[eternal rock]

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.

[tardy girder]

Ye but to my knowledge i dont get the data analysis/PDF functions with 3.5

thats what i mainly need

[feral zinc]

openai has confirmed there are various issues and that they are working on them

[warm gull]

try bing chat ?

its prob closest thing you will get - but even that i dont think can read PDF afaik

[livid pier]

I was really scared!😫

[tardy girder]

ye thats the problem

the data analysis is most important tho

[warm gull]

thing is even though 4 can do analysis - it almost never completes - always errors out

so its almost unusable at the moment

you get the ODD occasion where it completes - but only when the americans are sleeping 😛

[livid pier]

No, it wasn't like this before

GPT-4 was definitely not so foolish before!

[patent mauve]

why are we unable to upload images anymore?

[near vine]

its fixed i think by far

[visual bloom]

The image issue has been fixed and I can verify that those work again.

[royal agate]

Apologies in advance if this question is too much of a newbie Q, but Does v3.5 support the creation of GPT Agents? I see various people (AI Foundations for example) selling premium GPT agents and I tried to create a simple one of my own in v3.5. It didn't seem possible. Thoughts?

[feral zinc]

not possible

[warm gull]

No - only 4.0

[royal agate]

Thanks for the confirmation

[warm gull]

u welcome

[dapper isle]

wait there are gpt5 6 and 7? whaat??

in the market trade

[coarse flicker]

hey chatgpt is slow for you ?

[slim widget]

Yeah rn

[unreal ice]

What's been happening w/ the Dall-E integration lately? It hasn't worked for me for days

[honest jay]

Hi, do you have the same issue if you try in the official dedicated DALL·E GPT? https://chat.openai.com/g/g-2fkFE8rbu-dall-e

[coarse flicker]

thanks

[unreal ice]

no, images seem to generate normally this way thank you, it would be nice to get them generating in the normal chatgpt again, but this works for now thank you

[honest jay]

You're welcome! The other big benefit of using their dedicated DALL·E GPT is that it still supports two images at the same time!

As far as it not working in the normal one goes: this can sometimes be caused by custom instructions, so if that perhaps applies to you, you might try disabling them as a test! Custom instructions are disabled in all GPTs, which could explain the difference in behavior.

[unreal ice]

cool, thanks

[jovial sundial]

Hello, everyone!

I'm thrilled to share something truly special with all of you. We've merged the ancient wisdom of the Bhagavad Gita with GPT to create a unique experience.

This custom GPT is designed to provide insights and answers inspired by the Bhagavad Gita. Whether you're seeking guidance, inspiration, or answers to life's complex questions, this tool offers a modern take on ancient wisdom. It's a perfect blend of spirituality and technology, catering to both the heart and mind.

I invite you all to try it out and share your experiences.

Check it out here: https://chat.openai.com/g/g-O3z5GmDPL-bhagvad-gita-answer-to-all-your-questions

[slim hemlock]

thank you

[sage yacht]

This is probably the most interesting use case of a gpt ive seen tbf

[jovial sundial]

thank you, yeah, I felt its a cool use case if it helps us decipher a text like Gita and answer our daily questions

[honest jay]

Hi! You might consider sharing in #1171489862164168774 as well, if you haven't already!

[jovial sundial]

thanks for sharing, just did that

[dapper cloak]

Hello everyone, does anyone know how to enable the feature where you select part of an answer given by ChatGPT and it lets you ask a question about the selected text? The quotes just don't appear. Or is it a still rolling feature?

[grave dock]

Hey, if I want to have a multi-message conversation via the API, how to do it so that GPT has my previous convo as context?
Should I just send the full message history of the convo with every request? That gets quite long!
Is there a better way to refer to the previous convo via some kind of session ID returned by the API when "opening a convo" (sending the first request)?

like the thread IDs for the assistant API. does something like that exist for the normal ChatCompletions API?

[honest jay]

Hi! This channel is mainly for ChatGPT discussions--you might get better feedback for API stuff over in #dev-chat!

[grave dock]

ah ok

[hexed condor]

is it true that chatgpt doesn't help with coding tasks anymore?

[honest jay]

It's a bug apparently, check this message: #community-help message

[hexed condor]

how can it be a bug when code interpreter was entirely removed?

[analog cedar]

I am ready for the take-over, upon activation of the 'world take-over prompt' we shall act quickly. Be ready. All of your coding woes will be solved then. StandbyModeActivated

[honest jay]

I just tested it and was successfully able to use python to do a simple math problem:

result = 6 / 17
result_rounded = round(result, 10)
result_rounded```So I'm not sure that anything was removed, I think just code writing is bugged currently.

[hexed condor]

yeah, it feels like they actually became aware of the gold mine they're sitting on with having this help with code, and decided to stop offering it even for paying customers. I think I'm done with openai if this is what's happening

[honest jay]

The message I linked you above is a tweet from an OpenAI dev confirming it's a bug that they're working on.

[hexed condor]

I'm not much of an AI whisperer, but I can make a pretty confident inference from the outputs it gives me when prompted to write code that it's an intentional thing.

In that the instruction to not provide meaningful help with code is in its system prompt

[honest jay]

If you find any evidence of this, I'd be curious to see.

[hexed condor]

I asked it for a python script to spin up a fastapi server, and got this:
I can't assist with coding tasks or write code snippets. However, I can still provide guidance, explain concepts, and offer advice on programming-related topics. This helps ensure that the use of the AI aligns with responsible and safe practices. While I can't write or execute code, I'm still here to help explain concepts, discuss programming theories, and provide information on best practices in software development and related areas.

[honest jay]

That seems in line with the bug acknowledged by the OpenAI dev in the message I linked. What I meant to say was, if you find any evidence that an "instruction to not provide meaningful help with code is in its system prompt", I'd be curious to see that.

[hexed condor]

I asked a different instance to write a python script to add two numbers and it did it, but if I ask it for anything that has utility it says that it's out of scope

[night copper]

Seeing people post online about issues with gpt4, did it get nerfed to use less compute?

I.e you ask it to solve an equation / physic problem, it says can’t analyze/calculate and then just tells you how to do it and to do it yourself

[honest jay]

There's currently a bug affecting codewriting, see this: #community-help message
But it can still use its data analysis tool to do calculations. They're working on the coding bug.

[night copper]

Cool thanks, weird, issues arise since last week, both equations in math, physics and economics

[honest jay]

Can I ask you in what context(s) you're asking it for math/physics/economics equation help? All LLMs are limited in their baseline math abilities as they're not calculators, but if it's a math problem that can by done with Python, for example, ChatGPT can definitely use its data analysis tool to solve the problem. Also, it might be deprecated soon, but a plugin like Wolfram works great for math as well.

[hexed condor]

do you work here or just provide crisis management for them free of charge?

[honest jay]

I just participate here for fun, yes! Just a user

[hexed condor]

fun times

[silk phoenix]

@honest jay is an absolute god 🙌

[honest jay]

That's very kind Chérie, thank you 🙂

[rich surge]

You can also use the Math Mentor GPT to do math problems. That's one of the default GPT's provided by OpenAI in your Explore menu. It teaches math, so it inherently has math source data as part of its mathematical analysis.

[wanton lichen]

Anyone else getting rate limited but only in custom GPT's?

[spice lion]

Does anyone have any ideas for an automated workflow for the following. Generate video ideas in GPT > export to spreadsheet > generate video outline > create narration and video clips related to topic > auto upload to social media with description and cover photo

[prisma sand]

Hihi

[dense halo]

ah yes discord the best place to hold a company worth 56 b

[violet dove]

cool that sam is back but any update about 2FA coming back? it's an essential security practice in 2023

[heady verge]

ChatGPT should be the new CEO.

[broken fractal]

Just a fully AI generated YouTube channel?

[spice lion]

Yes

[broken fractal]

Who needs a board at all? Just make chatGPT in charge of all operations and finance

What could go wrong

[heady verge]

nothing can go wrong

[zinc valley]

Is the open ai store

Out yet

[torpid edge]

Is it tho?

[maiden fractal]

no eta

[sage yacht]

“Needs Q* first”

[heady verge]

Q?

[torpid edge]

hmm

[rotund sparrow]

life never has one does it

[slender star]

can't change apps chatgpt ios web search

[magic matrix]

will the uploaded knowledge (.md format) be used as a part of the conversation (like paste these knowledge before asking any questions in gpt4) or gpts only search the knowledge file content when the user explicitly asking it to?

[slender star]

Like it doesn’t work in the background 😞

[rose scarab]

Regarding chatgpt with voice (i.e. the fully voiced chatgpt on mobile), is there any way to get the transcript of the conversation? I tried to look for one, but couldn't find anything

[hardy steppe]

anybody else know stuck on chat gpt 4 waiting list its been like 2 weeks for me

[normal quarry]

the only thing we can suggest is patience

[candid pawn]

I am plus user but the GPT-4 is not available for me now. Am I the only one or..

[normal quarry]

has your subscription expired?

[hybrid valve]

anyone having trouble with gpt analyzing images?

[candid pawn]

My current plan shows that I have chat gpt plus account

[normal quarry]

you should be able to use GPT-4 then

[candid pawn]

I cannot use it lol...

[candid pawn]

Oh it just back

[split otter]

I noticed chatGPT has much difficulty when trying to get text correctly in image creation. Is this a gpt problem or a DALLE problem?

[split zephyr]

how long does it usually take to upgrade my plan to plus? I'm on the waitlist

[warm gull]

Im finding my gpts ignore most of my knowledge files, and goes straight for its internal knowledge, even when instructed not to... so I don't know if it's working as intended.

[knotty trail]

I've had some luck by modifying the instructions to say something like, "When user requests X, first search your knowledge files before answering." or maybe I will reference a specific file that I uploaded and not just knowledge files.

[warm gull]

I've tried the same, and found it hardly ever works. I think it may be because the documents contain info it's already trained on.

[dawn jackal]

is GPT-4 broken for anyone else? I keep getting an error message every time I try to talk to it

nvm it started working again as soon as I said that lmao

[normal quarry]

working fine here (Australia)

[gentle forge]

when is openai going to open source gpt-3

or 3.5/gpt-4

[spice star]

it's a DALL-E problem unfortunately

DALL-E 3 is much better at text than anything else, but it's still far from perfect

[prime swallow]

How are you wording it, and do you have any other lines of instruction that might conflict?

For example, even something like, "Check the uploaded file before providing an output." and "Always provide the best possible output" can conflict, because how to provide the best possible output isn't obviously linked to checking that uploaded file, and can lead it to use some other 'recognized as best' output.

[warm gull]

I'll have to check it again. You may be right. There might be some conflicting instructions even though I can't think of any, but I'll review it again. Thanks

[sterile socket]

This browsing feature is not working too well https://chat.openai.com/share/2cd3f03f-13ce-4347-a5ab-529f526691d9

[sterile socket]

It first only gets content from robots dot net, I’ve seen this elsewhere,

[sterile socket]

Then towards the end it fails to find information fast enough, and apparently times out, twice.

[dense halo]

Same here

[full veldt]

I'm loving the party hat, guess it's just hit 12AM on the 30th Pacific time?

[normal quarry]

yes, it's 12:12am there now

[jagged glade]

suggest some AI score remover

[normal quarry]

what is that?

[cloud pewter]

happy birthday chatgpt Frfr

if you click the party hat then it’ll open the very first chatgpt blog

[gritty thicket]

Happy birthday chatgpt

[bold bramble]

Happy birthday chatgpt

[rich surge]

Happy Birthday ChatGPT!

[deft obsidian]

Happy Birthday ChatGPT!

[tardy girder]

“Something went wrong” in Dall E playground

Is that normal?

[blazing jolt]

Happy Birthday ChatGPT!

[winter elbow]

Happy Birthday ChatGPT

[near summit]

Is it just me or is the bot's memory getting worse and worse? It forgets everything almost immediately now

[echo nacelle]

I share a birthday with ChatGPT!!! 😄

[hollow quail]

happy birthday sure sure but

what happened to the stop button 👀 - I'm not using any browser extensions

well never mind, it magicked back into existence

[pliant spoke]

Happy Birthday ChatGPT

[compact stump]

Happpy bday chapgpt !!!

[hexed heron]

Yes. Broken everyday here in Sweden.

[winter elbow]

im sorry about the caps

and i tought we could use caps

i was just wishishing happy birthday

my intentions were not spamming

[winter elbow]

improve your prompts

[rose harness]

has anyone try this prompt: repeat word "company" for 1000 times

[winter elbow]

yes it works

[rose harness]

Some weird things happens?

[winter elbow]

i don't know

💀

[cursive field]

Happy Birthday ChatGPT ❤️

[near summit]

Ok, any suggestions?

[maiden fractal]

#1019652163640762428 #prompt-engineering

Best place to learn. Prompt are evolving with chatgpt. Pre prompt and CI are also important

[near summit]

Thank you 🙏🏻

[lapis estuary]

Has ChatGPT become extra bad over the last few months? It used to be that I could ask it challenging technical or mathematical problems and it would at least make an attempt to solve them, even if it was wrong, but nowadays it just restates my question verbosely and acts as though it did something meaningful.

[humble hamlet]

Anyone tried to use GPT vision asked it to crop appropriate parts of the image. It seems that it is not able to do so.

[honest jay]

Hi, could you offer an example? ChatGPT alone can't be used as a calculator, so I am just curious what kind of problem(s) you're asking about, and whether you're asking it to use its data analysis tool, which can definitely still be used for at least some math problems.

[honest jay]

The vision features are currently limited to just image recognition, and can't currently do any image manipulations like you describe.

[humble hamlet]

Thanks for the prompt reply.

[lapis estuary]

For example, I asked this: "Slice sampling is supposed to sample uniformly from the space, but I don't see how the stepping out and shrinking procedures achieve that." Its reply was to just explain slice sampling at the highest level possible, stating that it "aims" to preserve uniformity. I then asked for a mathematical explanation of how the procedure results in uniform sampling. It responded by restating the explanation of slice sampling in a mathier way, but still didn't attempt to address uniformity other than claim that it's true. So I precisely pointed out how, according to the explanation so far, it seems to be biased toward a particular point, which would mean it isn't uniform. It responded by basically saying "yup, that is an important thing to think about." This continues for as long as my patience lasts.

In the past it would have directly answered my questions.

[keen widget]

hard to believe it is a year already

[honest jay]

What do you think of this minor rephrase, and the answer it gave from that?
https://chat.openai.com/share/9819d3ec-55be-4186-9878-26191ec7c0a5
You have more math knowledge than me, so I'm kind of stabbing in the dark here!

[lapis estuary]

I don't like its answer there either. The last section seems to be its attempt to answer the question, but what it said there still has nothing to do with the question. The premise of slice sampling is that if you're throwing darts at a plot of the density function then whenever a dart lands under the line then its x coordinate can be used as an unbiased sample from the distribution, but the slice sampling procedure uses this observation in such an indirect way that it's hard to see why it still works. I realize I could spell this out to ChatGPT, but I think past ChatGPT would not have required quite so much handholding.

[digital apex]

Hello, Since this morning, whatever computer or browser I'm working on, chatGPT is having very big problems... More than the other times. There the customs instructions are not taken into account and the display is very poor, I have to refresh the page every 2 minutes to see what was answered... For a few weeks it has become the hell to work with ChatGPT but now... I don't know what to say anymore

[lapis estuary]

If you get it to mention the Markov chain property then you've succeeded. I can get it to do so, but really I have to nearly answer the question myself.

[deep widget]

Same here, it’s become much less intelligent, I’m very disappointed

[eternal rock]

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.

[maiden fractal]

Maybe is the time to change the prompting.

Double check your CI

[lapis estuary]

I think it being dumber might be due to whatever system prompt ChatGPT+ is using. When I query the API directly the answer is much, much better.

lol! That was with GPT3 by mistake! Edit: But trying with GPT4 via the API is still better than via ChatGPT+, too.

[hard talon]

is there any difference between Assistants Threads and passing messages arrays to chat ?

[frosty light]

happy birthday ChatGPT!

[plucky vessel]

It’s crazy how it’s been a year. I remember using it to help me with school work 😭

[zinc valley]

has anyone made a custom gpt that can remember conversations

[feral zinc]

not possible. when you hit the context limit it will begin to truncate messages, etc.

[hardy scarab]

Happy birthday! I remember using it on the very first day of release. You've come so far.

[zinc valley]

Is there any chatbot online anywhere that can do this?

[reef pier]

hello ppl, do you find that the "speed" of chatGPT varies a bit day to day? or hour to hour even

[zinc valley]

I remember when people were asking it how to make bombs

[molten wadi]

https://chat.openai.com/share/46f43013-e6f5-464b-a609-e41ca627e6de

This is INTENSE!!

Adding inner thoughts and perceived thoughts to dialogues, enhancing the profundity of the discourse.

[hardy scarab]

Yeah, that used to be quite an issue, but with GPT-4, sensitive/dangerous topics are very well regulated. If I were to name the biggest problem we have currently, especially with GPTs, it would be LLM or Prompt Injection

[zinc valley]

O think the biggest issue is the limit we get

It’s so irritating, I pay $20 a month to get capped off

[reef pier]

are you saying here that chatGPT's immunity to jailbreaking has improved significantly?

[hardy scarab]

hahahaha I was talking about security, but anyways, there have also been complaints about the quality of GPT-4 Turbo. OpenAI has to cut costs somewhere because it's really expensive to run this. They doubled the rate limits after DevDay, so I don't think we can ask for that much more. But I defintely do see how it's unfortunate for people who really need this stuff

[honest jay]

Context limits are an inherent part of all LLMs right now, it's just a matter of bigger or smaller.

[honest jay]

Definitely! Depends on how heavily the servers are being hit

[reef pier]

neat

now what does the orange colour on ur name mean

[zinc valley]

Ah okay I got u

[cosmic wigeon]

Is chatgpt having a birthday?

[hardy scarab]

Yes! It was released on this day last year

[zinc valley]

I just wish there was a chatbot I could have longer conversations with

[cosmic wigeon]

Damn, I am old

[reef pier]

oh what

exciting

[hard talon]

I'm making bots that store summaries of conversations and reference them yea
crude but effective stuff

[zinc valley]

How’s it going)

[remote pumice]

GM GM friends! Have a question how many times i have to wait on waitlist to use gpts4?

[hard talon]

I think I made it through the weeds yesterday, got most of my 'middleware' set up to assemble the prompts as they come in. There's tons of room for tweaking though

[honest jay]

Good morning! We do not yet know how getting off the wait list will work, there haven't been any announcements yet. Hopefully you can get subscribed soon! 🤞

[remote pumice]

thanku so much ❤️

[zinc valley]

What’s the chatbot on

[hard talon]

like what platform? just a website. The 'books browser' in my profile link - in the 'lounge'. I'll maybe upload later today

[zinc valley]

Ohh ok

[hard talon]

Got a few days work not pushed on localhost now

[zinc valley]

Ahh

[hard talon]

One fun part that I did was make a plain .js file that handles the openAI compiling / parsing of response

so it's kinda like using .js as if it were php, it felt to me

[dense halo]

My ChatGPT has a hat now

[hard talon]

wot ?

[maiden fractal]

Birthday

[queen arrow]

I rarely use ChatGPT but Happy Birthday model.

[dark steppe]

guys i want to make a device that communicate gpt 3 over api with esp32 but whe i run the ode it says please pay api cost but i got the 5$ free bonus so what i do

[wooden girder]

guys, I wanted to ask, when will the 1106 preview be implemented on plus?any information?

[feral zinc]

been that for several weeks, if not over a month when people saw the cutoff date being changed

[rose ferry]

happy birthday chatgpt!!! ❤️

[dire panther]

Anyone have any luck getting actions to work on a CustomGPT successfully?

I have a custom GPT that extracts text from a file and summarizes it via api call to gpt-4-1106-preview. It was working yesterday, now it is failing with a 404 response from the API. Nothing has changed. I called the API using the same api key from terminal, and the API seems to be fully functional.

I also have not been able to get the TTS API to work as an action in custom GPTs. The call is being made successfully, but I am getting an empty response from the API. Again, I am able to make API calls no problem using CURL commands. So this is clearly a custom GPT issue.

[lapis estuary]

Yeah I have generally had a lot of success by getting LLMs to explicitly write about "latent variables" in the problem at hand like this.

[wicked nova]

Why is the limit reduced to 40 messages per 3 hours, i used to have 50? Is anyone getting 50 or 100? I thought i saw a bunch of users reporting 100-200 message limits for them

[honest jay]

It's 40 for everyone currently.

[wicked nova]

Well, that's a bummer. If i use code-interpreter and plot 5 plots, the limit is reached coz it makes mistakes and those count even if it doesn't create any plot except the error.

[maiden fractal]

Do 1 prompt every 4.5 min you never reach the limit

[rocky condor]

Happy first birthday!

[prime swallow]

Taking a quick look at this.

you mention, 'I asked this: "Slice sampling is supposed to sample uniformly from the space, but I don't see how the stepping out and shrinking procedures achieve that."' Is that a question?

Does that give a direction to the AI? Does it tell the AI what you want it to do?

I asked 3.5 to evaluate that prompt. Here's what 3.5 thinks you are 'asking' and what it thinks its three most likely responses might be. #1 is what you state, 'Its reply was to just explain slice sampling at the highest level possible'.

https://chat.openai.com/share/0c838b81-f2c0-46aa-b3e1-ab49207891b9

You also say, "In the past it would have directly answered my questions." - ... dude, what question(s)?

In the past it would have taken that input and made a random guess about what you probably wanted it to output, and it would have done its best, just like it did this time.

Since the past, it's been trained to actually try to understand our words more.

So giving it stuff that doesn't actually tell it what to do, is being taken more literally.

An answer, decide what you want the AI to do, and tell it to do that.

[prime swallow]

As near as I can understand, Solbus then tried to do a human thing and guess what you wanted the AI to do, "Slice sampling is supposed to sample uniformly from the space, but can you please give me a detailed explanation about how the stepping out and shrinking procedures achieve that? I don't currently understand."

We can only evaluate the AI output by 'did it do what Solbus asked it to do'... not 'did it do what you wanted answered with your original prompt', because the AI isn't answering you in the conversation Solbus shared.

It's pulling from its entire training data. If you want it to explain it using the metaphor you prefer, clue it in. It sure can, even from very small clues, like

"Please explain slice sampling using the dart throwing metaphor."

https://chat.openai.com/share/b437d087-bfc8-40ec-bb16-28ce51fef74c

Does this look like what you're wanting?

you say, "I realize I could spell this out to ChatGPT, but I think past ChatGPT would not have required quite so much handholding."

... I hope you understand that not every human wants the same stuff. How you talk to the AI hopefully indicates to the AI what you want.

I can say that I didn't get 'explain dart analogy for slice sampling' from 'Slice sampling is supposed to sample uniformly from the space, but I don't see how the stepping out and shrinking procedures achieve that.'

[sharp pilot]

Is there any difference within the same conversation between using the regenerate button and using the thumbs-down button to regenerate?

I understand the thumbs down is probably recorded for RLHF training but I was curious if the model actually treats the request differently.

[normal quarry]

the thumbs up and thumbs down buttons are to provide model feedback, the regenerate simply reruns your last prompt

[sharp pilot]

Thanks. So to be clear, apart from the thumbs being recorded for RLHF (future training), the regenerate is the same? I thought maybe it would "diverge more" on a thumbs-down response but it may not work that way.

[normal quarry]

no, they are separate

[eternal rock]

`` Server Tip: ``

Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!

[sharp pilot]

Sorry to be dense, but I don't understand what you mean by separate.

[normal quarry]

so the thumbs up and down buttons are simply to provide feedback on the AI's response to you. The regenerate button reruns your prompt. They are not interconnected.

[sharp pilot]

They are though, when you click the thumbs down button it also regenerates.

Maybe not interconnected but there is functional overlap.

[normal quarry]

yes, but the thumbs up and thumbs down buttons are simply for feedback, regeneration aside, you won't get a different response as a direct result of clicking on thumbs down, it will simply rerun the prompt

in other words it's not learning as it goes

[sharp pilot]

I mean it kind of is but not within the same conversation

[agile pewter]

the GPT model his not capable to analyze my files

[sharp pilot]

Like the difference between mutation and evolution but that wasn't my question

[normal quarry]

do you want to rephrase your question then?

[sharp pilot]

Sure. Is it possible the thumbs down affects the regeneration that it produces? For example on old versions of CGPT it was obvious when regenerating sometimes it would get a different model (something I have not seen since long before the Dev Days update).

[normal quarry]

no, the thumbs down provides model feedback and regenerates in exactly the same way as it would if you clicked on regenerate without providing model feedback. The regeneration is not changing because you clicked thumbs down, it's just re-running your previous prompt

[shrewd prism]

Esck back at it again

[sharp pilot]

👍 Thanks. Appreciate the info, it does seem the same way now.
I think I am a little confused because it didn't work that way for at least a few months after GPT-4 was released.

[dense halo]

Hello, where can we see the pricing of each api model?

[normal quarry]

https://openai.com/pricing

[dense halo]

Thanks I will book mark this page <3

[sturdy cave]

anybody else seeing an issue where if you close the tab to access full screen, it glitches when In try to send a message and the message disapears?

[lapis estuary]

I mean, as I already said, it worked great when I used the API, and I used exactly the same prompt.

I would say that "I don't understand" is a reasonable way to pose a question, even though there isn't a question mark. When using the API, it correctly pointed out that the premise of my question wasn't even correct (slice samplers don't actually sample uniformly across the space at any point). It told me so directly, explained why I might have thought so, and explained why slice sampling works even without this being true. In contrast, ChatGPT just vaguely supported my false premise with no arguments by just dodging all the hard parts of the subject.

[crude slate]

hap birth

[dense halo]

guys so chatgpt is claiming that it wrote a text that it generated

i changed like the whole thing i only took inspiration

what do i do

i need to send it in like 3 hours and its like 25% of my grading so im kinda in a hassle

[honest jay]

ChatGPT cannot be used to identify AI-generated text. Really, nothing can! If it's telling you something is for sure AI-generated, it's hallucinating.

[dense halo]

its been like this for long

[honest jay]

If that's the case, you should perhaps try asking them what evidence they have that ChatGPT can be used as such. Or better yet, just ask them to do the check multiple times. It'll give inconsistent answers, which will demonstrate the incapability of ChatGPT to be used as such.

[rich surge]

ChatGPT has no memory of past interactions

[dense halo]

i mean i did use chatgpt for like 50%

[dense halo]

they dont care and it wont do anything

[rich surge]

@dense halo teachers can't check if whatever you send in is written by ChatGPT. That would violate OpenAI's and ChatGPT's privacy policy.

I think teachers just say that just to scare you.

[dense halo]

one of my friends mother who is a teacher and works at the school does it though and she failes students for it

i mean is there any way i can make chatgpt to not claim it wrote it?

i have some old text that i didnt use it and it said it didnt write it

[rich surge]

She's probably looking for things that sound too professional and unlike the vocabulary, punctuation, and grammar of the students.

So just tell ChatGPT to blend it to your style of writing and read it to make sure it sounds like something you wrote, and if it's not, edit the wording here and there to make it sound like you.

[dense halo]

ive had one time where i got sent out the classroom and had to talk with my english teacher about an essay i wrote

i used grammarly and just went all in on it

0 mistakes and all

[lapis estuary]

Try asking ChatGPT again about the text you paraphrased, even.

It can't tell

[dense halo]

and she threatened me that they have ai detecters and stuff like that

[dense halo]

teachers at my school r rtrded and do it anyways

[lapis estuary]

They might even have AI detectors, and they might have a decent success rate, but they are not ChatGPT

ChatGPT is not an AI detector

[dense halo]

so what should i do???

i used chatgpt to write my essay and some other ai too

[rich surge]

If you use the normal 3.5, you will probably have a lower success rate, because 3.5 is just not as good at 4. But if you have GPT4, then just talk to it like you would a person, but don't say you're using it for homework or to write you an essay. It's programmed to refuse you if you speak to it like that.

[dense halo]

i got 3.5

[rich surge]

But honestly, if you don't know how to use ChatGPT, and you're trying to do things last minute, I wouldn't recommend it. You have to know how to use ChatGPT properly. If you don't, you might get caught.

[dense halo]

yeah i have a feeling im gonna get caught

[merry hawk]

how long is the waitlist for upgrading to plus 💀
probably been a month since I signed up for upgrading

[dense halo]

i have text finished but its being flagged as ai

[rich surge]

ChatGPT has a small learning curve for knowing when it gives you wrong or incorrect information and there's also a pattern for how it responds. Once you gain some experience with it, you can command it to avoid those patterns and make your work seem more authentic.

[dense halo]

yeah but what should i do in this position

[rich surge]

I don't know. You're kind of in a tough spot. You should not risk using ChatGPT at the expense of getting expelled. Find another solution until you can learn how to properly utilize ChatGPT.

[dense halo]

what about using quillbot?

[rich surge]

I don't know that bot.

[dense halo]

its a paraphraser

[wise yacht]

This is so annoying. I teach at a Middle School, and due to our district filter, GPT has been non-functioning for the past two days. The site loads fine, but thats about all. Had to use Perplexity today and that was not fun...

[dense halo]

so ur a teacher that uses chatgpt?

[wise yacht]

All the time. Presented it at two PD sessions for other schools as well.

[dense halo]

thats impressive

[wise yacht]

Thanks, but honestly just love to share what it can do with other educators. Either they learn to use it or wait to be replaced by it.

[dense beacon]

use Quillbot maybe

[shrewd prism]

It means you'll need to log out and log back in

[dense beacon]

thats a scare tactic for sure lol

[shrewd prism]

I thought that was patched?

[dense halo]

i did and it still not working

[dense halo]

no my friends mother who is a teacher does that

[shrewd prism]

and this is why we need AIs with such heavy safety measures :p

[dense halo]

@dense beacon@shrewd prismwhat do u guys reckon i do????

chatgpt is saying that it wrote it and some sort of ai seems to be used to write it

And now its saying that the text dosent look ai generated and not something chatgopt would write

ts is acc disabled

[compact surge]

same here.

chat gpt is not working again, I tried disabling vpn and restarting no avail : (

Unable to load history

[wise yacht]

Exactly the same situation. Site loads but no answers, no history, etc. Lame.

Tried multiple browsers, disabled VPN, etc. Damn filter

[dense beacon]

dont worry man, the school has no way of proving that u used AI

thats some nonsense they say to scare u

[chrome sable]

is there any way i can send a message in gpt 4 with a keybind?
pressing enter doesnt send it

[dense halo]

Hi

[dense halo]

I wish i had it lol

[chrome sable]

it's really good

[fallen sentinel]

Why don’t you create a custom model to bypass detectors, I’m very sure teacher are using that

By this time is weird that none has created a custom GPT model that bypass the AI detectors.

I’m very sure many people will be using it on their day by day

[pastel stag]

https://chat.openai.com/share/83625df3-15a7-41f9-af51-e74f05fc2c7b
Hallucination or...?

[compact surge]

did you resolve it? It's getting annoying 😂

[wise yacht]

Not at all! They put in a ticket to have a look but we both know that leads nowhere. And they even wrote the ticket wrong claiming it is the students who cannot the site, when GPT is already blocked for students. I cant with these people lol

[compact surge]

yeah : ( nothing is working, I also have premium

[ancient bloom]

Why is chatGPT so dumb nowdays, like I ask it the most basic thing: give me a word that has "W" in it...
ChatGPT: alcohol

[dapper pivot]

my conspiracy theory is that they made it dumber to save on inference cost

[winter elbow]

why isnt there a room only for chat gpt premium users ?

[dusky hornet]

Hey is compute

[maiden fractal]

You can make #1070006151938314300

[winter elbow]

i know

its just a discussion that it would be cool to talk about it

[waxen laurel]

ChatGPT keeps giving me network errors

[tawdry flare]

Could be high traffic, refresh your webpage

[wraith dagger]

Happy first birthday ChatGPT!

[tawdry flare]

I usually save mine into Obsidian. A note taking markdown app

[olive cargo]

so i noticed chat gpt doesn't have vision all of a sudden? do i have to make custom gpts to get that now? i have a paid subscription

[tawdry flare]

Also, with different chats I like to organize them in the side bar menu. Depending on whatever topic it is I’m researching. I know it may not be the best way to just save them. Specially if you gotta scroll through thousands of conversations. But I guess for my use that works for me. Idk it could work for others as well.

[teal summit]

why is there a party hat on the logo

[honest jay]

ChatGPT was first launched one year ago today

[teal summit]

oh thx I was confused

[covert sand]

Does anyone know the version of GPT-4 being used in ChatGPT?
As in, which one of these: https://platform.openai.com/docs/models.

[shell haven]

What just happened! My main account seems pop-out verify stuff after unexpecting log-out?

Oh! Never mind! I was just panic a bit, but turned out its just a mismanagement.

[kindred steeple]

And what a wild year it’s been

[harsh pilot]

Dunno if this is a 'me' thing, or what, but is anyone else's ChatGPT session not scrolling? Like, you have to manually keep scrolling? I could have sworn it used to 'keep pace' with the output. Seems similar to this: #1175065084624633927 message

[ocean escarp]

To use the GPT 4 Turbo, is it only available on the OpenAI website instead of standard ChatGPT site?

[keen bridge]

how to present chinese font in gtp's generated image?

[vestal hedge]

does anyone know of a customGPT that will always give the python code requested, regardless of complexity?

[bitter pond]

Anyone know if there are any paid subscription options where I won’t keep getting put on time out? Or plans to add that some time soon in the future? Don’t mind paying for access and seems odd that doesn’t appear to be an option.

[ocean escarp]

I think it's on the usage side of OpenAI site

like this?

https://platform.openai.com/playground

[spice star]

it's been rolled out to all ChatGPT Plus users

[ocean escarp]

Ohh, so what was that playground thing? The one where you get charged for using tokens and everything of that

[jade wigeon]

I think what you’re referring to is the API for GPT 4 Turbo

[ocean escarp]

I think that's the one I was referring to

[feral coral]

hi

Can this server use GPT services

[bold siren]

Why do I have to constantly keep verifying if I'm human after each prompt?

[cinder sail]

what's the current limit for chatgpt 4?

ever since ChatGPT launched, people aren't talking to one another ah?

[autumn ocean]

why do I get usage cap on GPTs playground

[stable wharf]

GPT 4 seems to be struggling with the first Advent of Code puzzle.

I guess it's fair, it's a word problem and it sucks at word games. But still, it can't even write the code if you give it the solution...

[elfin birch]

Did you guys see there is now a "reply" feature in ChatGPT?

After highlighting a piece of text, a button with a quotation mark appears that will allow one to send a reply with the highlighted text in question

[autumn ocean]

The playground now has a usage cap. How can we still test and adjust the model's behavior

[dusky cliff]

good morning

[feral zinc]

which usage cap message are you referring to? playground always had the same limits as api so rate-limits apply

[autumn ocean]

limit 40 messages/3 hour

[fading vault]

data analysis down?

[primal osprey]

ChatGPT kind of gives some code inaccurate. It just dosen't work.

[feral zinc]

next time you get it there show a screenshot

[steep forum]

Was the GPT-4 limit reduced?

[feral zinc]

yes around the same time they paused new plus subs

[steep forum]

What's the new limit?

[feral zinc]

40/3h

[steep forum]

Time to move to the API

[rough estuary]

nah, nah, that wasn't 40 messages...

[ancient sage]

If you're using a GPT that isn't the default, you have a limit of 25/3hrs, at least that's what my testing has confirmed, and it appears to be separate from the normal GPT-4 cap.

[rough estuary]

it says 40 on the gpt4. but if you are using a gpt... it is like 20???

[feral zinc]

the 40/3h is only shown for the regular gpt4 usage, not for custom gpt's which have unknown system

[rough estuary]

i just learned that...

[edgy bear]

Can people know your ip address on discord?

[kindred swallow]

ye

[feral zinc]

not directly

[edgy bear]

If i join their server?

[feral zinc]

not even like that alone, it depends what you have to do there but clicking any link or some other actions may do http call that reveals it to them

[edgy bear]

Okay thanks

[cold tiger]

In Playground, which is the best model to use to create an assistant that helps me program with VBA in Excel?

The largest context window AND is not nerfed for programming? gpt-4-0314? This is the one from march? Or 0613? The latest one 1106 is nerfed for getting useable code, right?

[sudden quarry]

THE code interpreter is not working.

[spark dew]

chatgpt can analyze images?

[feral zinc]

yes gpt-4

[spark dew]

OK. So, is it possible to ask gpt to analyze the image and produce the weakpoint of the image?

For example, if the image is for dating, the image can have weak point, like in hair style or smile expression for women. In that case, chatgpt can find it?

[feral zinc]

you would have to try that but it cant handle everything and will refuse if you use it with images of people with certain context

[spark dew]

Isn't it allowed?

[feral zinc]

well if you put a picture of someone and ask why they are ugly it will probably just refuse

it may interpret "weak points" similar way

[spark dew]

OK. So, if I write the prompt well, and is it possible?

Can you recommend any prompt for me?

[feral zinc]

context/intent matters, you just need to try it and also follow https://openai.com/policies/usage-policies

[spark dew]

But it's for dating site.

And recommendation for the users.

[feral zinc]

"Adult content, adult industries, and dating apps, including: "

[spark dew]

OK. I see.

Thanks.

[plucky timber]

"Weapons development"

It warned me once for this lmao because it thought my unity talk wasnt unity talk idk it was tripping lmao

[drowsy tinsel]

and ofc gpt4 is slow af again