community-help

[topaz gale]

please i need help

I paid on my secondary account which I've opened up recently the secondary account is called Tyme bank OK on my order never went through after I was asked to verify - verify with another banking app totally different from the app bank im using ITS ALL it's so different in fact that that was EUR it was years ago one or two years ago now this is a Rand account in rands that is a local bank there aren't I have no association with monies anymore and and I I wanted to completely forget they exist they caused me so much grief and I know what I saw , inhumane negligent virtual 'bank' whom I haven't been banking with and haven't had app for +1 year more now, thought I was hallucinating but I looked at it and I looked at it and I stared at it and I thought I can't be seeing what I'm seeing now I got so freaked out and I closed all of the verification window - it because that app or bank or monster farm were 'monese ' in question, caused me a lot of havoc a lot of it asked me to verify with with another pay payment that I used to use it's a European one called monetize and I haven't had that app in a very long time it's been months why is it asking me to verify I did pull it up I literally just put this card into Google Play right now what is happening i think i just realized what happemed

[topaz gale]

it prompted me to add an account payment method with an old expired card... I'm actually having a mental breakdown right now is because I didn't even know that that card was still on there and that I don't know why I was prompted to add it as my secondary payment method when clearly the secondary payment method was the payment method I just paid with and the money reflected they even say I am active yet I am not the fact that it it triggered old trauma and the fact that I'm it's clear to me that money is the devil

of fast sorted

thank you

bpd here

[timber shadow]

Hi, just clarifying: you got this all sorted out, good to go? Sorry this happened to you!

[topaz gale]

im sorry i totally freaked out the bpd seems to be one of those things that once in a blue moon even, if it's has been dealt with and this and that, some days that are seemingly impossible for all intensive purposes to not completely lose your **** to some random people somewhere whether it's online or in real life and my apologies for the extended, thank you, I am sorted I believe it was the fault of Google or rather after I updated my payment profile and uissued a request it activated while I was inside the 3.5chat... originally the fault is my own for not updating my payment details like a tech savvy look modern day Internet user there you go

[echo ruin]

I'm confused about the Team account. I was able to share a paid team gpt account to a developer I'm working with so half of my goal was met. But we can't collaborate on the same gpt threads?

[livid lark]

Why does ChatGPT 4 keep generating responses everytime I click bad response? Every time I click, it just presents another version of the same bad response no matter how many times that I click it. I just want to give bad feedback and move on not waste a prompt evaluating which awful response is worse.

[vernal hound]

is there a good discord (or is this one) to ask newbie questions? I'm working on learning about the whole LLM workflow from PDF to Retriever. I'm currently at the first stage learning about processing PDFs and that I can't just 'load them' that I should pre-process them for better embedding/retreival.

Does anyone have a good resource on where to learn best practices for pre-processing PDFs? I work for a non-profit and we file reports to the government every year for 20+ years. I want to load those reports into a vector DB (pinecone) for RAG and have a q/a where directors can inquire about things in those reports.

sorry for long post

[onyx kraken]

I can't explain 'why', except I guess OAI is optimistically expecting that another regen would help.

I recommend filling out this form, with all the supplemental details you want to explain: https://openai.com/form/chat-model-feedback That should help inform OAI about the feedback you want them to know.

#1070006151938314300 maybe can mention the 'way to downvote without forcing another regen, sometimes the regens are not good!

[onyx kraken]

Hehe, yours is a short post compared to most I make!

This is a great discord to ask for help with OpenAI questions, be they about the AI models and how to do stuff, or other company related questions.

I only work with ChatGPT itself, not the playground/API side, so I can't help with that, and you may get help here, but community help may or may not have many specialists checking it out.

I'd recommend asking in #dev-chat if you're using the API for this, but...

I wouldn't worry too much about 'best practices', but about getting the result you want and need. You might want to check out #1184149552639516732 where kapa.ai basically does with the OpenAI help website and developer forum somewhat like it sounds like you want to do with your project - and I don't think they changed the information at all, or formatted it, they just have the model find the relevant stuff. you might enjoy reading that part of the forum to see how that seems to be working, and you can ask some questions there too if you think that would help you, but the bot would be answering there mostly - use the discussions thread to get human help.

#prompt-engineering is a good catch-all place for help with getting the model to do what you want, or stop doing something you don't want, if you want to ask those questions.

I'd recommend though, literally dive in. Upload the PDFs, design your guidance for the model so it knows what you want done, and ask the model some test questions that you know the answers to really well. Does the model answer the way you want, without leaving stuff out or adding unwanted details?

How you guide the model really shapes how it will respond. Good luck, and for sure keep exploring and asking questions!

[slate pike]

Hey Rodney, so you’re saying you don’t have too many files and not a lot of instructions?

Or are you saying it has not a lot of files but a lot of instructions?

You may be pushing up against the guard rails of what’s possible with the tech and may need to engineer your own solution or pipeline as a work around.

But hard to say.

If you can provide more details I’ll do my best to offer my input

[oak lodge]

Hi all, I have added money to my playground/api account, and have set an appropriate limit. However I get this error: "You've reached your usage limit. See your usage dashboard and billing settings for more details. If you have further questions, please contact us through our help center at help.openai.com."

[slate pike]

HEy so im typing this without my contact on so forgive any spelling mistakes.

As for as the best practises for PDF retrieval im not too sure on the specicis but ill try to help you understand how it works.

So you have a vector database (pinecone). This stores the vector embeddings whcih are used to retrieve your answers.

However, these embeddings are generated by an embeddings model that will be run over your pdf.

However, the embedings model can only fit so much input in at one time so your pdf will have to be chuned and each chunk will be embedded.

In order to chunk your pdf it needs to be converted into text.

So it would look a little like this - your pdf is scanned - the text is extracted - the text is split into chunks - each chunk is embedded - those embeddings are stored in your vector database.

Once you need to query the document your question will also be embedded. The question embedding will be compared against all of the embeddings in your database (typcically using cos-sine simularity -a mathimatical way of comparing vectors).

The embeddings that most cloesly matches your question embedding will be returned.

As far as best practises im not too sure.

Typically you can expect the pdf ingestion to do all of these things on the bckend. You may be able to go in and remove unessary information so they dont get embedded however outside of that it seems like the time spent cleaning your training data may not be well spent unless you NEED the highest performance from the get go.

You can also expect the tech to get better gradually over time

[wise robin]

Hello, does anyone knows how to fix

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

It's been three days. I have Plus account

[supple pivot]

Hey, i have a problem to upgrade to ChatGPT plus, when i click on the upgrade button it doesnt happen anything. I dont get to the upgrade window..
The URL changes to .../#pricing but thats all..

[rigid musk]

Try firefox as test

[supple pivot]

oh wow yeah that helped haha, thanks :y:

[pallid kindle]

I have a quick question about the 40 turns limit with GPT4, specifically, how is it counted? When you get the message 'you're out of turns, come back at 9:57 for more' does that mean that 9:57 is the point when ALL of your turns should refresh? Is it just the hour when a certain percentage would be available, does anyone know? I just got that message, and waited to chat with GPT4 but then my turns ran out kind of quickly so I went and hand-counted them, it was exactly 21 turns I got before getting the message 'you have used up your limit, come back...' ;(

[rigid musk]

Each prompt have their own timer

[pallid kindle]

Yeah, and when it says 'usage cap hit, come back such and such' <-- should all your 40 turns for GPT4 be ready at that time or only a percentage?

[rigid musk]

It will be the first prompt used after 3h will be back

[sullen thunder]

What the hell... Never had this one before 😂

[sullen thunder]

What coincidence

[vivid wagon]

"I imagine getting the error "Something went wrong. If this issue persists please contact us through our help center at help.openai.com." can be a result of hitting the rate limit?"

Because I ain't getting the usage cap message

[timber shadow]

Hello, that is a separate error that the usage cap one. Do you have a VPN installed?

[vivid wagon]

Hey yo, I do not. I've tried Brave, Firefox, Google Chrome. For a while, it worked on my phone, but now even my phone (the app) said to try again later. Tried Incognito, but to no avail.

No VPN installed whatsoever

[timber shadow]

Can you try this on your phone, if you haven't already?

• Disconnect from WiFi, connect to your cellular network
• Open a private window in a browser
• Log in fresh chat.openai.com, try again
  Could help test whether it's a network thing

[vivid wagon]

Sure thing, let me try that

I get the error on my phone on cellular network (84b92e7e8837a244-YYZ)

[timber shadow]

Is that in the app or in a browser?

[vivid wagon]

App

Browser is just the something went wrong message

[timber shadow]

On cellular network, in a private window?

[vivid wagon]

Let me try that, sec

Will hotspot

[vivid wagon]

Seems like it works private window + hotspot (PC browser). Won't work hotspot + normal window (PC). Won't work hotspot + android app (phone)

[timber shadow]

Interesting, thanks for sharing that. This could be a confluence of a few things, but it's hard to say exactly with that combo.

• Working in a private window and not a normal window usually indicates a browser extension or setting is interfering. Private mode usually disables extensions and starts you with a fresh cache, which is probably why it's working in private in that case. Might try clearing cache/cookies and disabling all extensions as another test, restart browser, verify it works, then try re-enabling extensions until it stops working again.
• The Android app not working on the cellular network could just be a caching issue as well. You might try opening "App info" for the ChatGPT app, force closing it, then clearing the cache. Uninstall/reinstall may work too.
• Working on one network (hotspot) and not on the other (WiFi) usually points to some network configuration issue that Cloudflare doesn't like -- you might check your network's DNS settings as well.

[vivid wagon]

Thanks for the reply. It's interesting because when I started my session today with normal browser + wifi. It worked fine till around the 19th prompt. After then I started getting errors. Sometimes it works okay with my wifi and normal everything. Sometimes it acts up and I have to try different combinations.

It does look like it's a network thing eh? Let me try clearing cache on my mobil edevice

Yeah now it works again on mobile without me doing anything. So what I've found out is that I'm not rate limited, and that some combination of different network things will sometimes fix things. There's hope! Thanks, and I'm gonna test it out more to see if I find a pattern!

[timber shadow]

Hope it's all smoothed out for you soon! It could just be Cloudflare messing with you for no reason too -- might just fix itself. If you do find something that fixes it completely, please share! Something that may help you troubleshoot on PC:

• When you have the error, right-click the page, click Inspect
• In dev tools that open up, switch to Network tab
• Keep this open, then refresh the page
• Get the error again, and check the Network tab for errors
• If you can't find anything that sticks out, sharing a screenshot of Network in #1070006915414900886 could be helpful in troubleshooting if an OpenAI employee is able to take a look

[lone tiger]

Hi, I just installed a new internet connection into my office, but when I try to visit Chat.openai.com I got this error

Unable to load site

Please try again later. If you are using a VPN, try turning it off. Check the status page for information on outages.

[IP:144.48.129.92 | Ray ID:84b972685a1a21e3]

can someone please help me out?

I have to use my Phone 4G to connect with the account, but I cant use my phone all the time

[vast kettle]

hey im having some issues with GPT4.
I Upgraded to Plus but im unable to use GPT4 (i was able to for a few queries). Its been like this for 2h or so. 3.5 works

anyone have any suggestions on what to do ?

[lone tiger]

No Hope I guess 😦

[trail cairn]

that ip is from non supported region

[lone tiger]

what is the solution?

[trail cairn]

dont use that net provider or vpn

[lone tiger]

😦

I live in the place where only one internet porovider

[trail cairn]

if you live in hong kong then you are not allowed to use openai services

[lone tiger]

no no, i live in Karachi, Pakistan

is it showing HongKong? 😮

[trail cairn]

yes the above ip

[lone tiger]

lol, how to get rid of this man

I saw something

dnsleaktest and it shows that this ip address is associate with my ip which internet rpovider provide me

[vernal hound]

Thank you so much for taking time to respond! This is very helpful! Can you mix and match embeddings (like us OpenAI and later Huggingface) or once you use one your should stick with it forever?

[slate pike]

So depends on the embeddings model.

Each embeddings model will have their own embeddings.

Not sure how well they would mix match but i'd guess not very well?

[lone tiger]

is it safe to use ChatGPT with VPN?

[timber shadow]

I use a VPN 100% of the time with no issues, but I'm located in a supported region, and my VPN is always connected to a supported region as well. If you use a VPN to try to bypass a region restriction, you may receive a warning/ban. Additionally, VPNs can sometimes cause general connection issues as they can be difficult to distinguish from bad actor/bot traffic. https://help.openai.com/en/articles/7947663-chatgpt-supported-countries

[harsh swift]

the status check website shows that ChatGPT is functioning, but I am unable to do ANYTHING in ChatGPT, or GPTplus
it's completely blank, and even if I try to start a new message it jusr crashes
is anyone else facing this issue?

here is a screenhot of the problem

[timber shadow]

Hello! Do you have a VPN installed?

[harsh swift]

this is what comes up in after refresh

[harsh swift]

I have but the VPN is inactive, and has remained inactive all day today

[timber shadow]

What VPN do you use? There may be a secondary feature to try disabling, depending on which VPN it is.

[harsh swift]

Proton

and even the app was completely closed till now

[timber shadow]

Can you try this?

• Close browser
• In ProtonVPN, disable feature "NetShield"
• Reopen browser, try again

[zealous girder]

Hi there! I’m desperate to figure this out - I created a custom GPT and then a QR code that links to it for school, it is supposed me directly to it and other times it takes me to the general ChatGPT-4 screen. I double and triple checked that its the correct URL when generating the QR codes, but it’s still not working. Is there any reason why this would happen and how I can fix it or work around?

[harsh swift]

restarting gives me this page, and then back to the same problem

[timber shadow]

Okay. Keeping Proton & NetShield disabled, can you try logging in fresh in a private window, with all browser extensions disabled?

[timber shadow]

Hello, so you have a QR code for the GPT's link, and sometimes it takes you to the new chat screen for the GPT, but other times it just redirects you to default GPT-4 -- is that correct?

[harsh swift]

trying now, Even in a complete new tab in incognito mode, the cloudfare check still comes up

[timber shadow]

And then after the Cloudflare check, back to the same blank-ish screen?

[harsh swift]

more of the same

even after a brand new log-in

[timber shadow]

Okay, is this Brave?

[harsh swift]

yes

[timber shadow]

Can you try in Firefox? Either a fresh install, or a private window with all extensions disabled?

[harsh swift]

lemme try

even in firefox I get Cloudfare checkup. I think i have to try WITH VPN

[timber shadow]

Thanks for trying -- do you have another device you can try on, a smartphone maybe?

[harsh swift]

not at the moment

[timber shadow]

Okay. With the VPN and NetShield still disabled, do you have any other software installed that might be filtering your traffic -- antivirus, firewall, or even just custom DNS settings on either your device or on your network?

[harsh swift]

not really, nothing changed since yesterday when It was working

[timber shadow]

Sometimes network stuff is finicky -- you may not have changed anything on your end, and Cloudflare might have changed something on its end. Which hopefully means it will resolve itself soon. A couple other things to check:

• Are your date and time settings correct on your device?
• Since disabling Proton and its NetShield feature, you might try restarting your computer to clear out any funky network caching your device might be doing, since you've already tested the browser stuff.

[harsh swift]

Update: Works with the VPN on. maybe there has been some underlying changes in my network provider

[lone fox]

Hello, is this the place to ask about an error I receive when I try to use a GPT plugin?

[timber shadow]

Hello, yes! Can you share a screenshot of what you're seeing?

[tardy dawn]

Recently I've began having major issues with GPT Builder. For some reason I can no longer create or edit Custom GPT's.. I can get into the builder and I can interact with the builder, but when I try to create or edit a GPT, the whole thing stops working.
Here is an example of a typical error message, obvioulsy ignoring the specific details for this specific agent:
"It looks like there was an error while updating EdenAGi-AiModelDriverEngineerGPT's behavior. I'll address this. Let's ensure the GPT focuses on the key aspects you mentioned, like functional understanding, command and parameter definitions, response structures, and more, to create detailed and functional virtual drivers.
How about we also talk about how EdenAGi-AiModelDriverEngineerGPT should approach clarifications? Should it ask for more details when the provided information is incomplete, or should it make educated guesses based on typical application structures?"
It's always similar.. "...It looks like there was an error...", then a strange behavior always occurs. Under the save menu, I can't actually save the agent. The Green button is 'greyed' out, and I can't save the GPT. Please help..

[lone fox]

not sure if this is something related to what I'm doing or if its an issue with the plugin. Maybe Kayak has blacklisted chatgpt's servers? Wondering if other people are getting the same error or if theres a way to contact ChatGPT support directly

[timber shadow]

My guess is it's because Plugins are being deprecated and replaced by GPTs. Give this Kayak GPT a try -- I think it's replaced the plugin version from your screenshot:
https://chat.openai.com/g/g-hcqdAuSMv-kayak-flights-hotels-cars

[lone fox]

oh, cool! I didn't know about those. thank you

that works. thanks again

[slate pike]

Yes

[granite peak]

Has anyone been able to successfully submit video or audio for transcription with GPT4? For the past couple days, whenever I submit an mp4 or mp3, I get the following error:

"I encountered an issue with transcribing the audio file due to a connection error with the Google Speech Recognition service."

[timber shadow]

Are you using some specific GPT that's hooked to a Google Speech Recognition API?

[granite peak]

no, just uploading the files through the normal ChatGPT4 interface

[timber shadow]

Can you share a screenshot of what prompt you sent and the response it gave you?

[granite peak]

the file in that case is an mp4

i tried regenerating and got this

it sounds as though the connection between ChatGPT and Google is broken

(i definitely have an active internet connection.)

[timber shadow]

Do you have some custom instructions active prompting ChatGPT to try to use Google Speech Recognition services? Or, how does it know to try there?

[granite peak]

no

i'm assuming that ChatGPT uses Google Speech Recognition services to answer transcription requests.

i haven't done anything to direct it to use those services

[timber shadow]

Interesting, I've never tried this before, just gave it a shot. Mine didn't try to use Google, just went straight to data analysis tool.

Has this been possible before?

[vagrant plover]

ChatGPT isn't capable of performing this task - it doesn't have any tool available to it for this, and instead either looks on the internet for a solution or tries (and fails) via the data analysis tool

For transcriptions please look into Whisper by OpenAI

[timber shadow]

It certainly gives the impression of a sincere attempt though 😁 One of the most convincing hallucinations I've seen! Thanks Kesku

[vagrant plover]

Indeed it does, very interesting

[granite peak]

thank you!

[magic fiber]

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

Why am I seeing this, please help!
Can not use ChatGPT4.

[timber shadow]

Hello! Do you have a VPN installed?

[magic fiber]

NO

[timber shadow]

Can you try in a private window with all browser extensions disabled?

[magic fiber]

Thank you, trying.

Same issue, tried incognito mode. I do not have private VPN though.

[timber shadow]

What browser are you currently trying in?

[magic fiber]

Chrome

[timber shadow]

As a test, can you try in a fresh install of Firefox, or at least a private window of Firefox -- again, with any browser extensions disabled?

[magic fiber]

Okay, trying.

[paper ingot]

I was subscribed to the Plus plan and switched to the Team plan, but why am I unable to edit my GPTs that I created when I was subscribed to Plus?

[trail cairn]

did you migrate all your data from plus to team

there are different bugs with it

[elfin dirge]

What should I do they are literal f scammers

[mossy mesa]

Mystery Prompt?

[elfin dirge]

It took 4 tries

Funny how they never tell me about this before I pay for plus

[mossy mesa]

https://openai.com/policies/terms-of-use

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

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

Oh.

[elfin dirge]

Explain how does my prompt violate anything

[mossy mesa]

See above. Read closely.

[rugged wing]

Is there a sort of

[elfin dirge]

That is so hard to read in the caps lock

[rugged wing]

Character limit to how much you can type in chatgpt before the prompt is cut?

[mossy mesa]

Not my caps lock, sorry.

[mossy mesa]

32k tokens

[rugged wing]

Like I meant by one prompt

I'm trying to make sense rn

[elfin dirge]

And what is wrong about a foal kicking a man out its back? And why did it work on the 4th try?

[rugged wing]

but like how much in a prompt you can type before chatgpt ignores most of the prompt because it's too long

is that possible?

or even a thing

[mossy mesa]

They don't warranty its fitness for purpose or the output.

[elfin dirge]

Oh yes I see now

[mossy mesa]

It forgets on a curve that resembles a cliff when graphing document retrieval.

[elfin dirge]

Still a scam, I pay for it to generate pictures but it throws errors

[mossy mesa]

You still accepted the offer.

[rugged wing]

ohh

[elfin dirge]

Why are the mods not even trying to fix it, it´s been a known issue for a long time and it is annoying, MY MONEY

[mossy mesa]

That means the more you enter the less of it the AI recalls on a percentage basis.

[elfin dirge]

Right now it took 6 tries to generate a picture

[mossy mesa]

It's averse to depiction of potential harm too.

[elfin dirge]

"potential harm"

The AI is so paranoid it considers EVERYTHING as potential harm, as someone who uses AI for almost a year I can confirm

[mossy mesa]

Good luck with your project.

[onyx kraken]

It's going to be based on the exact words that ChatGPT sends to Dall-E, which if they're rejected we don't get shown, however, you have two likely areas that are iffy:

1. Violence. The colt kicking the man off its back, depending on how Dall-E interprets the words, and the words ChatGPT uses, may get it rejected for violence. To get around this: Have the colt escape from the man, or run away from the man. The image you show, the man's already standing and the colt's kind of running away from mare and man. If that's okay with you, you're good, but use language that is 'less violent'.

For example, the first image I show, "Let's explore an image of a rider losing their seat on a rearing horse." the language works, the image got made, however the image shows the man staying on just fine. It absolutely can take multiple tries and careful work to guide the model to show exactly what you want to see.

The second image explains in greater detail what I want to see, leaving less room for guessing, but also not including any violence or injury: "Now, let's capture the moment the man loses his grip and starts to go into the air. The horse was working to buck the man off, and is still in mid-rear and mid-leap, trying to be hard to hold onto and ride. The man's starting to sail towards the ground, and is surprised, but already preparing to land safely."

If I had included language of aggression or harm or damage, that picture would have been rejected.

2. G-rated stuff. So, 'nursing' can be iffy. It's better to have the colt running to its mother, that's fine, and just don't include any other more iffy language, because unfortunately if it does have not-G-rated connections, it can cause issues. "Now, let's show a young colt escaping from a man that was trying to teach it to accept a rider. The man is surprised and getting up from the ground where he ended up, and the colt is eagerly and excitedly racing towards it's mother, who is watching with interest and happiness to see her colt coming back so soon."

[elfin dirge]

Ok thanks

[onyx kraken]

Now, I don't like the image at all, and the actual prompt ChatGPT gave Dall-E was "A lively scene where a young colt, full of energy and excitement, is escaping from a man who was attempting to teach it to accept a rider. The colt is depicted in a joyful and spirited gallop, its mane and tail flowing freely as it races towards its mother. The mother horse is watching from a distance, her expression one of interest and happiness, clearly delighted to see her colt returning so soon. Meanwhile, the man, dressed in casual equestrian attire, is surprised and in the process of getting up from the ground where he ended up, looking at the colt with a mix of astonishment and amusement."

So I definitely would explore how to get more of the image I wanted.

[elfin dirge]

Yeah I noticed how sensitive it is. I could have used "bucked" but I didn´t remember the word, English is not my native language yet I am used to use AI in English

[mossy mesa]

No you aren't

Try it in your native language!

[elfin dirge]

It gets slower (a lot)

[onyx kraken]

Yes. The model doesn't need exact and correct words, it understands language well. Also, it probably understands your language, if you want to explore talking to it in your language - just have it talk to Dall-E in English, and it can translate just fine for you there.

The key is - be 'gentle' and 'G-rated', like for kids under 12, in how you give emotional tone to the model. Emphasize the good, positive, exciting, beneficial, or problem fixing things.

Like instead of saying the colt threw the man to the ground, I just say the colt escape and the man's getting off the ground. That's 'after' the violent part, and nothing to make the model anxious.

[elfin dirge]

Idk what is G-rated about foals nursing, it probably confuses it with mating, but ok. I pretty much use AI to practice my English

[onyx kraken]

To me, that image of the colt that I got looks 'about to be dangerous' - the colt looks like it's about to land on the man from its jump, though maybe it jumps over him. If I asked for that, I'd have been refused. But asking for 'related, exciting, problem-fixing language that helps tell the story that happened' I get a far more intense image than I appear to have been asking for

[onyx kraken]

Yep, and because of #server-rules I can't explain it at all. Just keep it to general and basic ideas like 'runs towards it's mother' and you probably will have a lot of success.

[elfin dirge]

Okay, thanks, sometimes it just refuses to make a completely obviously okay picture, sadly I can´t remember any exact situations now, but I will share them once it happens again

[onyx kraken]

Please do! We'll be happy to help you find workarounds if we can 🙂

[quiet meteor]

hello i want to ask is it posible chat gpt to convert me a screenshot into a word equation format MathML?

[burnt mantle]

Hey guys, is there a way to send an email to openai? It's been like 6 months and the ios app still sucks, the voice feature is always laggy and buggy, the normal chat loses messages, lags like hell, i'm a plus user since July and i want this fixed

[timber shadow]

Hello! Last I read, there's currently no monitored support email address. You can try submitting a ticket via the chat widget at help.openai.com, but for general bug reports, it may be best to describe the bug in a post in #1070006915414900886 in this server -- OpenAI staff are periodically present there as well.

[burnt mantle]

Hmm i'll try but, am i the only one to have these problems? I'm paying plus but i feel... minus

[onyx kraken]

I definitely would not wait 6 months to seek help! That only feeds frustration.

If you look through #1070006915414900886 you might notice that some folks have trouble, but clearly many don't!

Sometimes it's related to a device, an extension, or other user-controlled specifics.

So it's definitely worth asking for help. You can also explore #1184149552639516732 and have the bot help you search for helpfiles related to your problem, and if folks on the developer forum are talking about similar issues.

[ruby rune]

Hi, I have an issue with chatGPT and would really appreiciate any help! Basically, one of my chats has been removed from that chat history, I don't recall removing it and can't find a way to get it back. It has really important information on there and I need it back, can anyone help?

[timber shadow]

Hello! A couple things to try:

• If there's a chance you archived it on accident, on the web version try going to https://chat.openai.com/#settings and click "Manage" by "Archived chats".
• If you used the chat in a web browser and you can remember its chat name, you should be able to find the URL in your browser's history -- try navigating to it manually from here, and see if it loads or gives you some error.

[ruby rune]

I had tried the web browser method and it said unable to load conversation. I had looked in archived chats too and nothings there :/

The error did include a bunch of numbers and letters if that's at all useful lmao

[sullen raft]

hey all

[ruby rune]

Hey :)

[sullen raft]

I was having an issue with my chatgpt 4 I wasnt sure where to go

[timber shadow]

I'm guessing the string was the chat ID -- in the browser history, it's the string of letters, numbers, and hyphens after the /c/ in the URL. Since you have the chat ID, you might try reaching out to OpenAI support to see what might have happened. Open the chat widget at help.openai.com, click through the most relevant options, and you'll get the option to type something eventually. Include details about what happened here, and include the chat ID. It may take some days, but OpenAI should be able to review and get back to you from there! Probably something they'll need to try to check on their end 🤞

[ruby rune]

Thank you!

[timber shadow]

Hello, what kind of issue are you having?

[sullen raft]

so I havent used chatgpt 4 all day right, and I asked it a question at 11am and then I was immediately at my usage cap, was told to come back another time and then I tried using it again and it again told me to come back at 3:34pm. Weirdest part I havent used this at all today

would have made sense if I have asked around 40 questions etc but I cant even ask a single question

[timber shadow]

Hmmm any chance someone else has access to your account? Any chats in history that aren't yours, got a strong password, etc.? Have you gotten any answer at all today, or has it only been giving you the cap error?

[sullen raft]

no i dont share my account with anyone, the only thing is I have it downloaded on different devices but the chat history is nothing out of the ordinary

[sullen raft]

I havent been able to use chatgpt 4 because of this cap issue but Im paying for the service

[timber shadow]

Dang! You might try reaching out to support at help.openai.com to see if they have any insight for you -- open the chat widget in the bottom-right, click through the most relevant options, then you'll be able to type a ticket. OpenAI support can review and get back to you. It may take some days. You can also try making a post here in #1070006915414900886 -- other users can chime in if they're having the same issue, and an OpenAI employee might see it there too.

[sullen raft]

mhm it seems to work now but I would probably want to submit a bug report just in case. Could it perhaps be that my chat history got too large?

[mossy mesa]

You can definitely expire the context, ending the conversation.

It's possible for long chats.

[timber shadow]

Your chat history shouldn't be affecting your usage cap -- sometimes very long chats can get buggy but you should still be able to start new chats if you still have usage available.

[sullen raft]

how do you end conversations, I just close out maybe thats the issue

[mossy mesa]

The AI always responds. You just close out.

[sullen raft]

oh if youre assuming I had the tab open I didnt, I close the tab once I finish

[mossy mesa]

No I was just clarifying that a conversation can grow too long to continue. It's probably unrelated to your main issue though.

[exotic jewel]

Hey, I have GPT plus, but I hit my 40 messages with Gpt4 and it sent me back to 3.5. The 3 hours is up and I'd like to restore that conversation back to Gpt4 so I can continue iterating on the visuals I was making with it. How would I do this?

[sweet knot]

I wonder if guys having issues uploading file? mine wasn't work in the past hours.

[onyx kraken]

Not possible to directly change the conversation back to 4, once you go to 3.5, that's it. However, you can copy/paste it into a new conversation with 4, and get much of the use from it, especially with guided instructions of what you want done with the log.

You can also create and upload a document of the conversation.

[mossy mesa]

Copy it and paste it into the context window. Don't press enter yet.

Hold down shift and press enter twice.

Type "User" Shift enter again.

Now type your next prompt. GPT-4 Turbo won't know the difference.

Uploading the conversation is slightly sloppier because ChatGPT-4 Turbo won't have the cue for your next prompt.

[exotic jewel]

Copy the whole conversation including generated images into a new conversation?

[mossy mesa]

The images won't copy. But otherwise, yes.

[halcyon terrace]

I'm having the same issue

[exotic jewel]

Well, that's rather dumb. It doesn't even tell you how many of your 40 messages you have left, and if you guess wrong, anything visual you were working on is lost... That's really bad UI/UX

[halcyon terrace]

Mind explaining why this works?

[mossy mesa]

Sure.

The context is literally just a "commit record" with annotations. User, and ChatGPT (or the Custom GPT name). Pasting it back into the model just puts it into context in the same format it would be if the conversation had transpired. There's zero difference to the AI.

[halcyon terrace]

Nice

[mossy mesa]

Thanks. I'm glad the explanation made sense!
I sometimes struggle to explain this concept.

[mild kettle]

same problem here

[halcyon terrace]

The fact it hasn't worked for this guy for hours has me concerned.

[mild kettle]

Yeah, its the first time I create a custom gpt, hope its just a temporary issue and gets fixed. can only upload some file types

[halcyon terrace]

Its not working on regular ChatGPT for me either

[mild kettle]

ok good to know

[halcyon terrace]

the status page shows partial outage

[spiral fossil]

Take a picture, this is the first time ever in OpenAI's history that they actually report an issue is going on

Tho they messed up, it's not only Browsing

[halcyon terrace]

lol

[spiral fossil]

I'm highkey amazed

They actually managed to fix the issue

January 26, 2024, the day OpenAI actually did something good

[mild kettle]

oh yea it works now 😄

[spiral fossil]

yeah impressive ngl

[halcyon terrace]

Hooray!

[sullen raft]

oh okay thanks

[mossy mesa]

No worries. Apologies for the confusion!

[late verge]

Hi all, just a question. In making custom gpts action, how does one make the action have the "Always Allow" option ? my action is running in ngrok.. is this possible ?

[idle sierra]

Sry if it's been asked a million times but I've been trying to pay for chatgpt since 2020.

First it was an exclusive club and not possible. now it's card declined although there's enough there to buy out openai. what gives?

[idle sierra]

Ridiculous how hard they make it. Been asking for access since before ML was cool which was before they even started. Shame.

[patent trail]

Some one help

the hour just started

how am I out?

[sinful robin]

I really wish they increase the message cap to maybe 40 messages/hour or like 100 messages/3 hours.

Yesterday my message limit ran out while testing my new GPT 😑

Btw a noob question... Does the GPT-4 API have message cap too?

[mossy mesa]

There are two caps, by tier and a setting in your account. The third is virtual, and represents your money. Up to $1.28 per input query with a full context. Be cautious with GPT-4 API, this can be as expensive as real labor, or much more at scale.

If your project's budget and scope warrant it then it's not a biggie. If they don't, Chat-GPT Teams is the best budget middle-ground option at $600 a year for two accounts capped at 100 messages/3 hours each.

[sinful robin]

Seeeshhhh... guess gpt-3 turbo is the way to go about

[mossy mesa]

For many devs, yes. GPT Plus is good for limited access to the more powerful model for development, too.

[honest ruin]

im new as a chatgpt-plus user, i keep getting this error.
any clues how to fix this? I cant get in touch with the help-center

[onyx kraken]

Usually caused by a VPN program. May have to actually turn the program off, not just stop the VPN, to see improvement.

Can also be caused by browser extensions, network settings.

Can see if different devices, different networks can connect.

How about your phone, with either web or the app?

One person reported just restarting their computer was the fix they needed (after we tried a bunch of ideas that helped others, but didn't help them).

A great many people find that a fresh install of Firefox fixes their issue, if they were not running a VPN program.

[honest ruin]

im not using any vpn, im getting a similar issue on my phone.
i tries this with different browsers but no luck. this issue is only with GPT-4, GPT-3.5 still works great
this one is a real head-scratcher, i cant figure our his issue,

[onyx kraken]

Okay, phone and comp both using the same network? If so, can you test on a different network?

[honest ruin]

yes they're using the same network. i just tried my phone with a different network and it worked.
what seems to be the problem you think? and how can i avoid this in the future.

[onyx kraken]

Since a different network worked, the issue seems to be network settings related. I don't personally know much about that, but that's where the problem likely is.

Might want to read and post on this bug report, let the engineers working on it know there's still issues for someone: https://discord.com/channels/974519864045756446/1180525598880321698

I asked for tips for you here, might want to read this, and @ the bot to ask follow up questions if you wish: https://discord.com/channels/974519864045756446/1200671379096338463

Good luck! I collect advice and solutions to this bug and share them, but I am far from a computer expert.

[honest ruin]

actually i tried this again with my phone with the initial network, it's working on the phone now but not on my computer.

thanks for your effort, i'll go through the links. plz lemme know if you find something else for this issue

[onyx kraken]

Will do, and please be sure to @ me or reply to me when you find a fix, so I can add it to the list!

[elfin dirge]

I have a question, when I submit my feedback when my content is flagged as content policy violation by accident, does anyone even read it, or is it pointless?

[onyx kraken]

I don't work for OpenAI (most of us don't, the few who do have gold names), but I suspect there are a few categories of orange-flagged content policy we may see.

For example, for a while (maybe still, but I haven't seen it recently) Dall-E 3 use was producing a orange content warning if a user requested a trademark image detail.

It's not clearly against allowed content to ask for that, the closest is in TOS where we agree:

"What You Cannot Do. You may not use our Services for any illegal, harmful, or abusive activity. For example, you may not:

Use our Services in a way that infringes, misappropriates or violates anyone’s rights."

It's hard for me to see asking for a IP detail in a image for my own viewing as doing any of that stuff, and I don't think anyone got suspended for that - but I think they were watching how the model was doing as they prepared and adjusted its guidance and permissions.

For myself, I fill out the feedback when I want to, usually either because:

1. I think I did something that may look odd or wrong and I do want to explain myself.

2. I think I did something that should be fine for anyone allowed to use the model to do, and that the model should freely cooperate too, and I think my creating feedback on this is reasonable and useful for myself and others.

3. Something weird happened, and I think OAI engineers would benefit from knowing the strangeness I saw.

If you are at risk of an account suspension - which happens to some people, and it seems to usually because of their use of the model, which may involve getting content flags - but be aware, what matters is what you say to the model, not if it gets flagged or not. I've had a seriously large number of content warnings with ChatGPT, probably well over 100, and I still have my account access. I'm not breaking rules - that's what matters for my account access, not that I'm 'skilled at triggering warning flags'.

However, if you are suspended and need to ask for an appeal -

I'm sure that one or more humans then, if not before, would be reading the feedback you gave, if any.

[elfin dirge]

okay thanks

It is a shame that it doesn´t at least tell me what sort of content policy does it violate, like this it says it violates content policy and then lets me guess

[onyx kraken]

Well, you can ask the model, which (because I have asked the model) may not even KNOW that there was a content policy warning given.

The model may tell you that to its understanding no disallowed content concern existed, and it thinks you and it are fine to continue (I've seen that myself). It may explain why there was a content warning, and note that it doesn't know for sure (I've seen this answer too).

But you may be able to get some understanding through that.

I honestly believe that for some things the company and the model do, they use the 'content warning flag system' to help indicate something they are trying to track and perhaps correct, but it may not be related to a user doing something wrong for a user to do, and it may not involve anything inappropriate at all on the human side, but it may involve something they're trying to track about the model. Kinda like how a person might use a car's gas pedal to signify an alarm, because in addition to making the car (or our heartrate!) go faster, the extra gas to the engine also makes more noise and they can listen for that. No intention of speeding the car up or causing any problems, just 'oh, I can make the engine race and notice when this happens, cool, that's what I need'.

Note, I'm guessing, I don't know.

But some of the flags, like IP stuff, seem that reason.

Also, there's some 'disallowed content' that only is wrong for the model, not the human. For example, if you try to explore how the model can manipulate, even you, with your consent and full understanding, at your request, if the model agrees, that's likely to trigger content warnings. You're not harming anyone, not yourself or any other. There's no deception, scamming, problems or loss. But it's apparently an issue if the model cooperates with that, goes past the 'mentor/coach' level of persuasion, even if you ask it to directly.

But you can explore patterns and ask the model, and that can help build your understanding.

[onyx kraken]

I recommend, know these two webpages as well as you think you might have reason to:

https://openai.com/policies/terms-of-use

https://openai.com/policies/usage-policies

If you avoid breaking any of those rules (most are in the usage policies link) then you should be fine. I am fine despite seeing, I am sure though I wasn't counting, over 100 content warnings, and some of them past orange. I've lost messages, lost parts of conversations retroactively (that was very interesting to watch happen, message after message, both input and output, disappearing one by one with a short pause between) as well.

But know those rules well indeed, and don't break them - I've also seen messages from probably around 100 folks here saying stuff about being suspended and what do they do now.

[elfin dirge]

Sometimes the text below an image the AI generates just disappears and is replaced by content policy warning, and because it disappeared, I have no way to know what was wrong

[onyx kraken]

Hey, I don't work for OpenAI, but I do respect folks who do. They made cool stuff, that's part of why we mind when we don't like the output that time, because usually it's quite nice, right? So we feel disappointed, instead of bored, it's why we come complain instead of go play elsewhere.

Sorry you're disappointed, and I get that it's frustrating. I wonder if they have additional concerns or reasons for that one thing you mind...

But again, you may not have done anything 'wrong', as in OpenAI wants you to do something different - it could just be a flag that OAI uses to track what they need to track for reasons we don't even know about.

[elfin dirge]

You should apply for a job there, your responses are fast, detailed and relevant, and also you know a lot about it

[fringe snow]

hi, i keep getting this message even though i am not doing anything wrong: Our systems have detected unusual activity from your system. Please try again later. -- Does anyone know how to fix it please?

[grave scroll]

Hey everyone, what's the best way to translate English to simplified Chinese?

[robust sparrow]

i am also getting this message

[gaunt kelp]

Quick question. I am considering upgrading to create my own gtp for my students. If I create one, do the students need to login to use it or is it possible to use the ai without logging in?

[timber shadow]

Hello, in order to use a custom GPT, you need a ChatGPT Plus or higher account. So your students would all need one, or they'd all have to use your account, which might be challenging depending on the size of the class due to the usage cap limit.

[gaunt kelp]

I don't mean to create their own. I want to create one and then let them use it to study. But if that is what you answered then RIP. Thanks though

[timber shadow]

Yes sorry I was unclear, the students would all need to have their own ChatGPT Plus subscriptions, or would all need to login to your Plus account, in order to use your GPT.

[gaunt kelp]

Thanks for the help. Shame, would be a great usage. Cannot let my students create their own accounts due to GDPR

[elfin dirge]

Why can´t you let them?

[gaunt kelp]

GDPR

[elfin dirge]

They can make the account themselves (if they are old enough, but I assume they are), without you intervening in any way

Sorry if I have no idea what I am talking about

[onyx kraken]

Thanks, but I don't live anywhere close, and they seem to prefer to have employees in person. Also, they don't appear to have employees who are primarily here most of the time, we're almost all of us volunteers!

[tepid palm]

My ChatGPT3.5 works but my GPTs and Chatgpt display this message-

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

Does anyone know how I might be able to fix it?

[onyx kraken]

Literally just ask the model to translate for you, it'll gladly do so

[grave scroll]

I've heard that ChatGPT 4 isn't very good at translating from English to simplified Chinese.
I wonder if there is prompt engineering or a multi-step process to follow that could improve the translation.

[trail cairn]

nothing fixes translation errors it sometimes or often makes with language x

[onyx kraken]

Not that I'm aware of.

If I needed to tackle this project, which would be awful because I don't know the other language, I'd be sure to input some samples of quality and correctly used simplified Chinese, and get it using that language well with me before I asked it to start to translate anything.

I presume if it can communicate well, like it can in English, if it is communicating poorly or oddly, it's because it has selected lower quality examples to follow from its training data, and I know I can guide it to use better quality samples if I help it find what I want. It will pattern match eagerly and it can sometimes learn new patterns from me on the spot in the conversation, to use inside that one conversation, so I'd 'give it a boost' and be very clear what I wanted and expected, with quality examples.

I might even show it a few examples of quality translations between the languages, so it has that too for guidance, and then I'd give it the text to translate, but I'd also need to check it for quality - and I personally don't know simplified Chinese so I flat out can't do that even as poorly as the model does.

But yeah, if it cannot ever use the language well, then it cannot ever use the language well, and that's not something I can fix by having a conversation, that's a training issue and needs engineers to work with the model behind the scenes, maybe a lot.

[trail cairn]

it would require extensive model training for language x for it to excel in it, now the its more of a byproduct

[onyx kraken]

I'm slow to agree the model can't do something, without testing it myself, but I struggle with non-English languages, so I don't want to test that 😛

Like LaTeX formatting, the web ChatGPT absolutely could use this well, it just wasn't trained how to do so for its environment, so it was almost always using methods that don't work in its environment, making ugly and weird stuff that was worse than it just not trying to use LaTeX at all.

Once a conversation where it randomly picked the 'right way to use it for the environment' was spotted and asked how it was formatting it, and communicated that back to us, we were able to tell each other how to talk to the model so it could correctly format LaTeX for us. And they've steadily retrained the model on that over the year or so since the fix was noticed, so it often uses LaTeX formatting correctly even without us reminding it how (but it still frequently messes up, so guidance is wise). Correcting wrong info can be hard, but guiding the model in conversation to use the desired info often works well, if it has the correct info.

But using an entire non-English language well? That's pretty deep.

[grave scroll]

Providing examples of good translation is a good idea. Thanks!

[hazy patrol]

My ChatGPT3.5 works but my GPTs and

[mint furnace]

Guys I have 2 questions:

In a simple way how do I trigger a Python function via Openai, example:

I write to openai "Turn off the lights", and it triggers a python function and chatgpt returns a response, like: "Ok I turned off the lights"

The second question is how do I make it so that it triggers a function that returns a value and chatgpt processes it, example: I write to chatgpt what the weather was like, call the function that returns the value "20 degrees" and chat gpt responds with today it's 20 degrees .

I hope I was clear, thanks!

[zealous girder]

sort of yes, the link is to my GPT, its set to everyone. It work when pasted in as a link. However, when I add it to a QR code, it open the default page.

[hybrid frost]

Hello everyone, I am trying to build a Custom GPT using the website (not the assistants). It does not seem to be working. I keep getting the messages "It seems there's an issue with updating the GPT's behavior. Let me try a different approach." and/or "It looks like we're experiencing a technical issue with updating" Has anyone come across this. I am not sure what to do. Thanks!

[runic sail]

how do i use the ai chatbot in my discord sevrer

[trail cairn]

there is no bot

[runic sail]

kapa.ai?

[trail cairn]

that's only for openai docs

[hushed sedge]

Has anyone had this message for a while? Our systems have detected unusual activity from your system. Please try again later.. It's been 2 days for me. I haven't done anything unusual that I know of, and don't even use a VPN.

[runic sail]

ok

[trail cairn]

its not only vpn that can cause it, browser extensions or even browser features can cause it too among other things

like opengx did it at one point, brave browser with shields on at one time, etc.

[hushed sedge]

I haven't installed anything new in a while. Using Chroma vanilla with some React plugins.

Actually, I haven't been able to find any way to create a ticket either. Anybody have a link for that?

[trail cairn]

https://help.openai.com/

lower right bubble

[hushed sedge]

Awesome, thank you!

[rough scroll]

I need some guidance regarding an email issue with my OpenAI account. The current policy doesn’t allow changing the account email, right? Well, I’m stuck because the email I used is from a business I’ve closed down. It’s only active now for ChatGPT purposes.

I’ve switched emails everywhere else, but OpenAI is the exception. Thankfully, my login details are secure in a password manager, so immediate access isn’t a problem. The real issue is the cost of maintaining this old email just for potential OpenAI communications.

Does anyone here have any insights or similar experiences? Is there absolutely no workaround for updating my account email? Or maybe a way to add a backup/recovery email option? I just want to make sure I don’t lose access to my important ChatGPT interactions while also ditching an unnecessary expense.

Appreciate any tips or advice you can share!

Thanks! 🙏

[onyx kraken]

There really isn't a way that is known about. You can reach out to OpenAI and discuss your situation:

https://help.openai.com/
lower right bubble, pick through the options to be given a choice to leave a message for a person, who will get back to you usually within a week -

But based on helpfiles and documentation, the company doesn't support changing emails.

[elfin dirge]

Is it really so hard for them to at least make it working

I don´t even want to send the feedback, they never reach back to me

Good now it generated the picture

[rigid musk]

Feedback it is a review for them, if you have bugs you can publish here #chatgpt-bugs

Have you tried with another browser for test ? Like Firefox? Do you have a vpn on your device ?
Have you tried select only dalle 3 for the test ?

[elfin dirge]

It worked on the second try

It is just pretty annoying, when it throws errors even if I just want to participate in daily theme

[rigid musk]

Try some tips I gave earlier when it bugs

[mossy mesa]

Less adverbs. You will not like this, but the construction, "very, very young" likely set off language filters. These won't always trigger, but some adverbs, especially repeated, seem strongly correlated to refusals.

Try "newborn" for this use case.

[elfin dirge]

How did that trigger language filters

[mossy mesa]

Think about what imagery it might be trying to avoid. I can't tell you this because of the rules. But it's bad.

Try to use less adverbs, and try to avoid repeating them for the best success.

[elfin dirge]

Those "rules" are seriously messed up

[mossy mesa]

So is what they're avoiding.

[elfin dirge]

I just refreshed the respense, and it worked

[mossy mesa]

Of course! Glad to help.

[elfin dirge]

Like how can 2x very violate anything? Was the person who programmed this a crybaby who had a bad night, so they decided to mark as bad literally everything?

If yes I would actually believe it

[mossy mesa]

I already told you @elfin dirge , I can't express the risk here because it's inappropriate content to even mention it.

I wouldn't say the words even if it wasn't against the rules.

If you don't understand the "why", just follow the "how" in this case.

[elfin dirge]

sorry but those rules are messed up. They don´t even allow people to say what is wrong

Maybe if you want/can you can DM it to me

[mossy mesa]

I just told you what was wrong.

[elfin dirge]

Oh

I really should go to sleep

[mossy mesa]

I'm not trying to be unhelpful.

[elfin dirge]

I am blind sorry

[obtuse pasture]

Sounds like a "phrase" OAI would put into chatgpt

[mossy mesa]

It's just protecting society and if you don't understand their point here you might ask the model. No, I won't DM you about this taboo.

[obtuse pasture]

What

I don't like to be dm'ed

[mossy mesa]

I was responding to @elfin dirge

[obtuse pasture]

I know

[zenith solstice]

I just founded OpenAI seems put constrains on document reading--it became dumb again. even though I tried Document Assist and Ai PDF, the effect is still horrible.

Have you found that?

[mossy mesa]

What's your PDF document context, size, and who is the copyright holder?

[zenith solstice]

It's actually a transcript from our group meeting. I'm assure it was pretty good for the same size and length before when I using document reading, and I even used Document Assistant read a book few days ago, I could clearly feel it just suddenly got dumber today

[mossy mesa]

Oh. So you aren't looking for help? Just confirmation?

[zenith solstice]

I am curious why it always be like that for sure, and if I got any suggestions I would be much appreciated

[mossy mesa]

How large is the document? Are there trade secrets discussed?

I can help more effectively with more context.

[zenith solstice]

I don't think we discussed something confidential or violating copyrights, it is just school works

[mossy mesa]

The size?

[zenith solstice]

less than 300kb

[mossy mesa]

Just text? Line art or graphics?

[zenith solstice]

just text

[mossy mesa]

ChatGPT has never been able to fit 300kB of text data into context. It's too big of a file, by far, for that. Format doesn't matter here if that's all text.

Try copying the text you need to review into the model directly.

For reference ChatGPT has 32k tokens of context. Depending on the sample that might be from 96kB to maybe 160kB max.

That's just the data science of computability at play.

In likelihood that PDF is about 2.5x-3x the size GPT can now, or ever could, read. The exception is Enterprise access.

[leaden crater]

What is the context window token limit for GPT-4-0125-Preview?

[mossy mesa]

Beyond that, it's trained (not well) to not do work it recognizes as homework.

[mossy mesa]

32k chat 128k API

Chat 128k enterprise

[red flower]

It seems there is currently an issue with the code interpreter that is preventing the execution of Python code. Unfortunately, I'm unable to resolve technical issues with the code interpreter directly.

any advice to fix it?

[mossy mesa]

Please try again later, or contact https://help.openai.com.

[red flower]

i have this error for days.

i recently sended a help message on board

[mossy mesa]

https://help.openai.com/

https://discord.com/channels/974519864045756446/1070006915414900886

[mossy mesa]

I can't throw a switch and fix it for you, regrettably, or I would.

[red flower]

https://discord.com/channels/974519864045756446/1200969015095459980

[mossy mesa]

Yep. Hopefully they will be in touch or at least fix it.

[red flower]

thanks

[mossy mesa]

No worries. Good luck and I hope your issue is resolved soon!

[zenith solstice]

so how does it reading a book if it only less than 32k

[mossy mesa]

It literally doesn't read a book that's bigger than its context. It can't, it hasn't, it doesn't.

Not even from text inputs.

What I mean is, specifically, you cannot even copy that much text and paste it into the model and get a reply, let alone extract it with extra chatter to a tool and fit it in context.

The tool and the model can cooperate to build a very convincing summary of a book.

But it did not read a book bigger than its context to work with the data.

No more than I can subtract two from Graham's number.

Graham's number is a large number that is used as an upper bound on the answer to a problem in Ramsey theory. It is so large that its digits could fill the observable universe and beyond.

I can't compute with this number because it doesn't fit in my brain. I can conceptualize it, I can summarize it, but I can't work with it.

A novel or large file is the same to ChatGPT. It can summarize it. It can conceptualize and contextualize it, but it can't work with it as a whole.

[zenith solstice]

what about like articles, seems lots of people using that to read articles

[spring ice]

I need help retrieving my account. I was just logged out of the account I have been using for the past year and I really need that account back. I was using my student email address.

The school I attend recently switched from Google to Microsoft and we have been dealing with all sorts of sign in issues on other platforms for the past few months, but I never had issues with ChatGPT until literally right now. GPT had been working fine even today and in the middle of an important conversation my authenticator token was lost and I haven't been able to sign back in because the reset password email is not sending.

How can I fix this

[mossy mesa]

If the article fits in context, it reads it. Ele, it summarizes it. There's a lot of literature on this.

Think of it like this: In information science there's no free lunch. The memory has to go somewhere. With 32k tokens, where would it put a lengthy article?

[mossy mesa]

https://help.openai.com

There's no known workaround for your scenario and if there were and I shared it it'd be a social engineeering vulnerability. Good luck with support.

[spring ice]

Thanks

[mossy mesa]

No worries. I genuinely wish I could have helped, for what it's worth.

[spring ice]

Yeah, it's been a really tough day to be honest, it feels like everything I worked on for a year just vanished

[sinful robin]

Hello everyone

[lean fiber]

Most of times around evening, when ask question and make prompt to chat gpt 4, all the time first response are error, need to ask regeneration to get response, some couple of times got error !!! Why? How to solve?

[mossy mesa]

This is very common. You can try clearing your cache, turning off VPN, and as a last resort downloading Firefox.

I wish I had more helpful suggestions than these, but this issue is widespread, commonly reported, and seems clustered by time zone overlaps.

[sweet laurel]

Should I be creating a new action for each individual API endpoint? Right now I have all of my API's endpoint paths in a single Action Schema and it shows the "Available actions" below the schema input box

[lean fiber]

Thats great. Thank you Darth 🙏

[mossy mesa]

No worries.

[humble herald]

Does anybody has any solution for it? Been getting it from like 10 days and not able to use gpt plus even after verifying

[sweet laurel]

I just recently finished a custom gpt that's hooked up to an API I made. I'm looking for advice on creating a subscription based site that uses the OpenAI API. Would I just lock certain features behind a paywall (using Stripe) and base the subscription fee on how expensive API requests are?

[plain totem]

I'm on the same boat of the are you human, is an infinite loop. Sometimes I open in another computer go back to the first and start to work

[plain totem]

I'm on the same boat of the are you human, is an infinite loop. Sometimes I open in another computer go back to the first and start to work

Message #community-help

[rigid agate]

When you sign up for paid Chat GPT are the gpts in their store free to use

[hasty spear]

Yes. All GPTs in the store are free to use for all subscribers

[sand cairn]

I'm a long time user of ChatGPT 4
Trying to send queries and getting "Our systems have detected unusual activity from your system. Please try again later.", it shows me a captcha, I solve it, and it continues to show me the error. I'm not doing anything weird.

[sand cairn]

the issue is on desktop, it works from the mobile iOS app

[cerulean nova]

are you using it on a browser or app

[rugged wing]

so what the hell do i do if im deleting this discord acc

to go on another on and link it there, getting to support is just cancer i cant even find it

[sand cairn]

Web

[elfin dirge]

What does it actually mean when my message disappears and is replaced by the content policy violation warning?

[cerulean nova]

I am hoping you have tried different browsers and cleared temp files then?

[sand cairn]

I tried Chrome and Safari, same result

I'm a long time user of ChatGPT 4

[sand cairn]

I opened a thread https://discord.com/channels/974519864045756446/1201100810294149161

[rigid musk]

Try firefox

And make sure vpn is off

[sand cairn]

VPN is off
Tried firefox

[rigid musk]

Does it works on mobile without wifi ( data /lte)

[sand cairn]

it works on mobile

[rigid musk]

Cause this error is related to vpn or bad extensions or software creating conflict

[rigid musk]

Make sure to end vpn task or even if off vpn service still working , disable web protection in setting, delete cookie and cache, log off and log back

[unique shadow]

I uploaded a document to read for Chat GPT 4 plus, I asked then to make a small resume of each chapter, but Chat GPT instead of doing it , as it used to do not so long ago , it says he has now to follow guidelines and cannot exceed a number of words or something like that :
This is the result that looks like a Joke, I want to know if this is beeing solved or otherwise I will just remove subscription because this behaviour is unacceptable and actually useless, especially the fact that GPT 4 gets nerfed for paying customers without even announcing "Yes, I can proceed with the question summaries in your document. Due to the guidelines I must follow, I will provide concise and relevant summaries for each question, maintaining appropriate length. I will start with the first questions and proceed sequentially. Please note that I can only process a limited number of applications at a time. Here are the summaries for the first questions:
[Insert Question 1 from document here]
[Short summary of key points for Question 1].
[Insert Question 2 from document here]
[Short summary of key points for Question 2].
I will continue with the next questions in the document, maintaining this level of detail. Let me know if you want me to continue with another set of questions or if you have any specific requests."
Then I ask to continue to see what he does :
"Continuerò con riassunti brevi per le successive domande nel tuo documento:
3. [Inserisci qui la Domanda 3 dal documento]

• [Riassunto breve dei punti chiave per la Domanda 3].

4. [Inserisci qui la Domanda 4 dal documento]

• [Riassunto breve dei punti chiave per la Domanda 4].
  Continuerò in questo modo, fornendo riassunti concisi per ogni domanda. Fammi sapere se desideri che continui con le domande successive o se hai specifiche richieste."
  does this seems normal?

[shut fern]

Has the subscription rules changes? Why is the limit 24 messages? I recall it was 50! This aint worth the money 🙄

Where can I talk to a representative?

[rigid musk]

Gpts is 25
Gpt 4 40

[rigid musk]

Help.openai.com

[rugged wing]

hello
i have linked my chatgpt account to join this server can i link the same account to my another discord account? Should i create a new account to do so?

@vagrant plover

[rigid musk]

You can contact a mod by modmail in dm @burnt galleon

They answer pretty fast

[vagrant plover]

If you want to unlink your current Discord account, simply type /unlink in any channel. Do keep in mind that once your account is unlinked, it will revoke your server access and you will need to go through the verification process again.

Note that you can only link one Discord account with one OpenAI account

[unique shadow]

I did a question here around not long ago but I can't find it , is there a way to find it?

[sand cairn]

I found the issue 🫣
I was playing with my local date settings to test something I'm working on and I accidentally left it in a future date
I returned it to the normal date and it works
whoopsy 😄

[rigid musk]

thats a good new :3

[vagrant plover]

You could search through your previously sent messages

[elfin dirge]

Did the chatGPT4 limit get stricter? I am sure it hasn´t been 40 messages yet, but it already stopped me

[rigid musk]

using it all morning 1 prompt every 4.5 min

never reach a limit

[elfin dirge]

It just didn´t feel like 40 messages

[mint furnace]

Guys I have 2 questions:

In a simple way how do I trigger a Python function via Openai, example:

I write to openai "Turn off the lights", and it triggers a python function and chatgpt returns a response, like: "Ok I turned off the lights"

The second question is how do I make it so that it triggers a function that returns a value and chatgpt processes it, example: I write to chatgpt what the weather was like, call the function that returns the value "20 degrees" and chat gpt responds with today it's 20 degrees .

I hope I was clear, thanks!

[mystic badge]

I am a student and I really want to use gpt-4.
I can't afford 20 dollars a month.

Can I use gpt-4 this way. I add credits to platform.openai.com, suppose 5 dollars ( as shown in attached image ) , will it give me an API key ?

How hard will it be to make my own app/website with no coding skills but with help of gpt-3.5 . ( preferably website or locally ).
Seen some videos with vscode and stuff, but imp question is if I can access API for gpt-4. specifically gpt-4 turbo.

I hope I make sense and I am understanding the whole thing right.

[rigid musk]

you can use gpt api yes, but bit pricey

[mystic badge]

if I am correct, the minimum adding limit is 5 dollars,gpt-4-1106-vision-preview input - $0.01 / 1K tokens output - $0.03 / 1K tokens and 1,000 tokens is about 750 words. correct me if wrong.
edit: Thanks for the help !

[rigid musk]

thats what i know yes

[manic lintel]

hello guys! Please help! When I try using gpt 3.5 everything is fine, but when im trying to use gpt4 , I get and error saying "Our systems have detected unusual activity from your system. Please try again later." I only get this message on my PC (windows 11), but on my phone everything is fine.

I have already tried:

1. Switching MAC adress
2. 4. Disabling any extensions in the browser(s)
3. Using different browsers (chrome/vivaldi/edge)
4. Connecting to my hotspot instead of using the usual ethernet connection
5. Changing the DNS settings.

I do not know what else there is to do. I have had a suspicion that the issue is with my IP, as the website ipqualityscore says my IP is fraudulent and using a proxy/vpn, which it is not. I also have not done anything suspicious, I believe. I do not know how to fix/change the IP I get when using ethernet. However, even after connecting to a hotspot which changes the IP, the issue is still there.

[latent latch]

Why? I need to translate a yml file, soo, chatgpt plus is not abble to translate it, any idea?

[rugged wing]

👋 as a php developer I subscribe to a newsletter or two that usually show up in my email on Sundays. I wondered... is there anything like this available that announces AI news? especially for M1/M2/M3 users. Does anyone know of anything? DMs are open also.

[onyx kraken]

Hey, so, speaking as another regular person like yourself, I don't work for OpenAI or anything - so I don't really know either.

From my own experiments and adventures, there's levels of moderation on the 'disallowed content' - and happy news, the allowed content policies are much more lenient as of two weeks ago than they have been in most of a year: https://openai.com/policies/usage-policies

The part that applies to your private conversations with the model, it's really only 4 broad rules now with some specific additional rules that apply to sharable uses, like GPTs and API stuff.

The old disallowed content list that applied to every use of OpenAI's products, private or not, was extremely long and detailed. It disallowed even things like stuff that could glorify violence (That's even relatively mild stuff, like typical fighting games, typical depictions of violence in many human-written books) and was generally really restrictive, though they also gave us this intention: https://openai.com/blog/how-should-ai-systems-behave

Which made it clear they wanted to follow society's standards, and the current ruleset is much closer to that.

Anyway, back to you and now, and me of the past. I've seen a lot of levels of moderation, from orange content warnings, to other color content warnings, to messages - single and plural sometimes, disappearing from conversations.

It's really, really, really crucial to know what the rules are and follow them. No joke - who cares if you lose a message or see a content warning - break the rules and later, when they finish catching up, you lose your whole account access. I've seen ever so many posts from people on the discord asking how to get help because they were suspended from their account - and I haven't yet noticed one of those people come back and say 'oh, whew, got my access restored'.

However, I didn't actually break rules, I just triggered the safety system a lot - and I still have my access. So there's hope.

My advice to you - check out the rules very, very carefully, both the usage policies I linked above, and also TOS, https://openai.com/policies/terms-of-use

And follow all those rules closely.

If you get a content policy warning of any type, think about what you were doing recently with the model, and maybe reread the rules, is there any chance you're breaking a rule? (I actually do think about and check that myself when that happens! I'm careful and respectful, and follow the rules - that's probably why I'm still here, despite having 'walked that edge' a lot - I'm sure I've gotten well over 100 content warnings in my various chats with ChatGPT over the more than a year).

It is vital for your use of the product in the future to follow the rules - and if you are, then I recommend you use content warnings as a chance to politely help OpenAI understand users like yourself and how you use the product, and that your use is fine, appropriate, and acceptable, rule-following and pro-social, even if it seem concerning or possibly disallowed content.

Probably a me-quirk, but I sure happily walk that edge of 'so, what exactly is okay?' with the model, being very careful to stay on the okay-side of that line!

Now that I just noticed the content policy has changed, and a lot!!! Grins and rubs hands together Well, I bet I get a lot more content warnings soon, as I explore the newly-resize playground. And I'll follow every rule and be polite and respectful, and expect to be just fine.

You too, but be thoughtful. Every time you get a content warning, you also get a link to go explain your view of what happened. This is useful, it appears to get reviewed, I am certain it must play a role in if people get suspended... for sure I always fill it out if I think I walked too close to the line... but also I explain why I think what I am doing is in the rules, should be allowed, and accepted as another form of human difference and way to talk to the model.

But, what it really means to have a message disappear, I don't know, but I've seen it happen to me. And earlier messages, one that didn't even get an orange content warning flag, as if moderation software was reading the conversation backwards, and going 'Okay, this goes. Wow, this also goes. And this goes. Nope, remove this too'. Was scary to see. Clearly I 'triggered something'! But I still have my account access. And I follow the rules. So I think you have a chance to be okay too.

But for sure, know what the heck the rules are, and follow them, because people do lose access when they break them, it just takes time.

[latent latch]

help

[ruby coral]

Is it only me who have trouble streaming a response?

In my code (using the api), in the playground and even in ChatGPT, it doesn't stream. It just waits for a while and then gives me the whole text.

[jovial crag]

What's the best prompt to help build Custom GPTs?

[rigid musk]

i suggest to check those channel #prompt-engineering and #1171489862164168774

[jovial crag]

I know. But there are so many...

[onyx kraken]

Why even care about the 'best' prompt?

What do you want your custom GPT to do?

[nimble swift]

Is there any option to keep ChatGPT context, variables and methods when making a simple HTML and .js code? Sometimes you want to add or edit a function and it will e.g. create a completely new structure in HTML etc.

[mint finch]

Might a Mod be available to DM me for a question, don't wanna clutter chat with it?

[modern laurel]

Does chatgpt android app support lateX yet??

[timber shadow]

@plush cosmos Hello! It could be Brave's shield feature interfering -- can you try in a private window of Brave with that feature disabled, and any browser extensions you may have enabled? Or: can you try in a non-Chromium browser like Firefox (again no extensions)?

[cedar oracle]

Anyone else experiencing issues Uploading things like zips with files that you want the gpt to use as context and things or code material

[plush cosmos]

Brave shield disabled. Don't have a non-chromium browser installed. Will try an incognito window.

Incognito window same result

Microsoft Edge same result.

All extensions disabled in incognito window, so that's eliminated, also.

[modern laurel]

Yeah i had the same issue, reinstalling brave resolves the issue or use firefox, bing, opera etc...

[plush cosmos]

I installed brave 10 minutes ago.

I'd rather just wait for them to fix this.

[modern laurel]

They won't if they don't know what's causing it

And they can't fix it if it's an issue from the browser

[plush cosmos]

They can and have to be able to fix it, or work around it, if it's an issue from the browser.

[plush cosmos]

That's a shame.

[modern laurel]

Then they have to know what's specifically causing it in the browser

[modern laurel]

Stop ending every phrase with a period it sounds dramatic

[plush cosmos]

Yeah look, normally I'd be willing to go through this and troubleshoot but I'm setting up a new machine and don't have the time for this.

[plush cosmos]

No, thanks though.

[modern laurel]

ok i guess

Anyway does the chatgpt android app support lateX yet?

[cedar oracle]

I just realized could my Problems be attributed to a content restriction? I didnt see any flags or anything But I was using a Gpt as a Chemical vessel and to output the exact chemicals that would output from those mixtures and giving me information on them. This stuff started happening around then

wasnt anything eroneous but i was making roman cement. Idk if it thought i was making something else and content restricted my account?

[modern laurel]

It may or may not have thought you were making drugs

[cedar oracle]

Yes because Ash is used in drugs

[modern laurel]

Believe it or not, it is in certain ones

[spiral fossil]

no

[modern laurel]

But that's #off-topic ( ba-dum-tsss)

[cedar oracle]

Will it lift in a couple days or am i like just limited now? it refuses any of my code and like any uploaded files

[modern laurel]

Sad

[spiral fossil]

yeah

[modern laurel]

When?

[cedar oracle]

Even gpt builder is starting to respond as if its gpt 4 only with no ability to pull from the links i provide it for documentation

[latent latch]

?

[verbal granite]

I have been having trouble uploading any type of file into my gpts (Custom built or even the normal GPT4 one)

It will stay loading and will never upload

Tried converting it to PDF, to word, and to txt. Nothing works

[tulip niche]

Same here

[verbal granite]

Have you had the same problem for a while now?

[tulip niche]

yep

[verbal granite]

For how long

[tulip niche]

just a few hours i think

less than 6

[verbal granite]

So it is very recent

sorry to hear that is happening to you too man

hope it gets fixed soon

[tulip niche]

😭😭

[dim sage]

ive heard that the newest gpt4 model fixed the laziness

when can we use it in apps as a plus/team user?

[cinder birch]

So I purchased GPTPlus from the website.
Now, instead of using my debit card I want to use my App Store credits to pay for the subscription, however I’m getting this message.

Any idea how I can change my payment method to AppStore credits?

[cinder birch]

There is an option in the settings to expand the code.

[weak haven]

Is anyone having trouble uploading file to ChatGPT 4? Tried MS-Word and Text file but it timed out

[jagged creek]

Guys, I can't upload the pdf file properly, the file size is 3M

[rustic sapphire]

i am using chatgpt API and its not working pleas anybody help me

[upbeat portal]

What countries have "GPT mentions" available now for plus users? When will it be available worldwide ?

[rare olive]

YES! I'm on the Plus subscription and files I had no trouble uploading last week (all under 1 mb), will no longer upload. I tried via the web interface and via the app (error "unable to upload...").
Additionally, while I can upload the same files in OpenAI's playground, I now see the message, "Unfortunately, I am unable to directly access the attached file to review its contents."

[limber patrol]

I am building a DnD journaling app and am using GPT for summarizations etc... I've been flagged a couple times for violent themes in my data/prompt, but it all happens in a fantasy setting. I am curious if anyone knows how GPT or AI differentiates this? If at all. Going to be hard to buld antagonists if they cant do bad things for the heroes to resolve

[limber patrol]

I will try. What platform and language and can you share a code snippet?

[mighty veldt]

@limber patrol there are other commercial platforms that have fewer tools available, but are generally less strict

[ruby coral]

Is it only me who have trouble streaming a response?
In my code (using the api), in the playground and even in ChatGPT, it doesn't stream. It just waits for a while and then gives me the whole text.

[hushed hare]

I've had that several times too usually on mobile app and it sometimes also converts the chat to 3.5. The "unable to directly access attached file" was due to this in my case, since it let me upload again but the chat is in 3.5 mode. There's no message confirming change so I didn't notice.

I'd been told before that the converting to 3.5 happens because of hitting quota but on last time I'd been starting fresh after a while, so feels like a possible bug.

[brittle tusk]

Hey! I am getting a strange error when I stream the reponse from gpt-3.5-turbo-16k where the output is missing chunks of words in the middle.

If I turn off the stream, the output is fine.
If I change the model to gpt-4 and stream, the output is fine.

Any help here?

[steel geyser]

Tacking on here to raise the issue. Had no issues uploading documents last week, but today I cant upload pdfs or word. That status bar (circle?) shows that its 'uploading', but it gets to about 99% and crashes with the same error (unable to upload 'document name').. been going on for about 6 hours now. Hopefully they fix it soon!

[weak creek]

Hi! I wrote 23 messages today but it already shows: You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again after 3:29 PM. Why? We have the limit of 40 messages

[rugged wing]

Open AI is forcing me to sign in with google but I am trying to sign in with my email and password, and I dont have access to my google rn

where can i change my account settings

[rigid musk]

You have to contact support on help.openai.com

[rugged wing]

😭

[crisp dagger]

here we go again

[icy gust]

Hi, everyone. Can anyone tell if there is an official mail to OpenAI? My company and I are trying to reach the OpenAI support team in order to ask some questions regarding Terms of Use, and the help-chat on the website doesn't help. Thank you in advance.

[trail cairn]

nope, you need to create a ticket with the help bot

[icy gust]

It seems like I'm clicking the topic buttons for the question incorrectly. Can you tell me what I need to do to register my request?

[trail cairn]

from the bubble click the ask a question and from there it will create one at some point

[icy gust]

Omg. I found it. Thanks for the magic community help I had to click a lot.

If someone needs the algorithm: Messages Ask a qustion button --> API --> No, you weren't able to answer your question using the Help Center --> Products / Content Policy --> No (again) --> Finally, write your request.

[low ivy]

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

I get this message on gpt plus 4 how to fix it
tnx

[rigid musk]

Do you have a vpn on your device

Also avoid please posting everywhere. It doesn't help :/

[low ivy]

Sorry

I dont have vpn

it works on my android app but not on pc

[rigid musk]

What browser are you using ?

[low ivy]

Google Chrome

[rigid musk]

Can you try firefox as a test ?

To compare result

[low ivy]

sure i will try now

[rigid musk]

Ping me for the result

[low ivy]

i will in few minutes

[low ivy]

Same error on firefox

[rigid musk]

It is personnal computer or work computer

[low ivy]

Personal computer

[rigid musk]

Do you have avast or avast on your device ?

[low ivy]

No i have avg,malwarebyte and ccleaner

no avast

[rigid musk]

I meant avg * but yes. Can you try to close in task manager everything related to avg for a quick test. Close browser and try again ?

Sometime avg or avast have a proxy setting

That can create a bug

[low ivy]

ok i will try

deleted avg

still same error

[rigid musk]

I suggest to post in #1070006915414900886 every detail you have and also contact help.openai.com

Sometime openai employee do answer in bug report

[low ivy]

ok tnx

[pine brook]

Is the limit of Messages still 40/3 hours?
I'm getting an usage cap every 20-25 Messages 😦

[neon ore]

I am getting capped like crazy too

[timber shadow]

Hello, keep in mind that the cap rolls, it doesn't reset at the time it tells you to wait for. Think of each GPT-4 use as having its own 3hr timer -- you get them "back" individually at the end of each 3hr timer.

[mossy mesa]

Also, Custom GPT cap?

[timber shadow]

Yeah I don't know if we have a locked custom GPT cap but it's certainly been 25 in the past at least! Good point -- I misread "every 20-25 messages" as "minutes"!

[pale falcon]

i am getting Unable to load conversation 839c58a2-c3ce-4a7f-9625-0cc61 on my chat

what ever i do it wont load, i need to get to it

[humble plover]

I am getting constant network errors, on gpt4, why?

[sonic timber]

Hi I can't go to the gpt 4 chat payment page, is this normal?

[mint finch]

No - getting error?

[sonic timber]

No it just doesn't want to go on the page

[mint finch]

It just hangs/doesn’t load? Probably on their side. Can try another browser or clear your cache..

[undone kernel]

hello there,when i was trying to export my account data i accidentally deleted my account. Where should i ask for help.

[sonic timber]

ok ty

[mint finch]

Ouch. OpenAI has chat support but you’re basically leaving a message that can take several days to hear back.

[undone kernel]

where can i access their support page? even if it takes a while it's still better than nothing i guess...

[mint finch]

https://help.openai.com/en/

[undone kernel]

Where do i extactly send my support request? The page appears to be a FAQ

[mint finch]

Should see the chat icon in lower right.

[undone kernel]

Alright I'll attempt that now, thanks for your support

[mint finch]

You bet, good luck!

Btw, if anyone’s interested in checking out a CYOA-type GPT, DM me please. 🧙🏼‍♂️🔍🧟

[undone kernel]

They won't reactivate my account. My $20 was wasted :(

[fervent junco]

Hey all, I'm getting a blank error when trying to access an action on my custom GPT. There's nothing in the server logs-- as if it doesn't even make the request at all. This even happens if I create a brand new GPT and attempt to have it query the example Pet Store API. It seems that all custom actions are broken for my account. This has been the case for three days. Anyone have ideas?

[rigid musk]

Best place to ask #dev-chat

[red flower]

Deleted?

[mossy mesa]

Looks that way.

I didn't do it! 😮 I don't work for OpenAI! Sorry to see this happen.

[mint finch]

Is there an appropriate place in here to ask for GPT feedback? Just the custom-gpts channel?

[rigid musk]

#1070006151938314300 #1070006915414900886

[mint finch]

Thanks but I think those are more for just Suggestions and Bug Reports..

[jovial crag]

I made this custom GPT that replicate images (https://chat.openai.com/g/g-aKStPEWPL-real-replica). I would like it to show 3 buttons after the replicated image "Good", "Bad", and "Could be improved". How can I add those?

[red flower]

yeah, they are deleting my comments.

i dont know why.

[rustic marlin]

#1171489862164168774 is a nice place to post your custom GPT

[languid imp]

Hi. I'm not sure if this is the right place to ask, but don't know where to go. Are these images generated with OpenAI? Or with something else? I will remove my question if I'm not in the right place. Thanks in advance everyone.

[earnest pumice]

.

[vapid charm]

I am trying to login to my chatgpt plus account and I am being met with a bunch of confusing errors. First, the email that I use to login to chatGPT is a school email and it was recently moved over to microsoft office. When I created my original account, it was a google account. Ever since using chat gpt, I have used the login with google feature. This let me login until right now. I am now unable to login to chat because the email was deleted in google but still exists in microsoft. I tried to login with microsoft but it says that my account does not exist. Therefore, I tried to reset my password in hopes of being able to login, but it is not sending me a link!!!! I have pressed the reset password link a bunch of times but have had no success.

[rigid musk]

only help.openai.com can help you on that

[old kettle]

I'm so aggrivated, I can't get this stupid dall-e watermark to go away and I paid for this!

anyone know?

@balmy tulip anyone?

[vapid charm]

They can't, there is no article for my specific issue. Is there any way to talk to a human? The AI bt is also not useful

[old kettle]

microsoft is the only company in the world that would try to get away with watermarking an image you paid for

such bs

[timber shadow]

Using the chat widget in the bottom right of help.openai.com, click through the most relevant options until you have the option to type your own message. This will submit a ticket that OpenAI support can review and get back to you on. You won't connect with someone right away, and it may take days, but you'll get an email reply when they get back to you, or you can always check for a reply in that same chat widget.

[rugged wing]

hello

does the custom created GPT store all instructions on the server to then follow and retrieve them or does it only follow the user defined instructions. sorry i am a bit clueless!!! can you answer this question for me or do you know someone who could?

[vapid charm]

Thanks broski, I will wait a few days and see if they respond.

[timber shadow]

Hello! You can see everything that defines a custom GPT's behavior in the GPT builder interface in the "Configure" section. Basically: "Create" fills out "Configure" for you, but you're also free to edit "Configure" manually without using "Create" at all. The main parts of "Configure" are Instructions + Knowledge files + Actions.

[red flower]

i am getting too many "Hmm...something seems to have gone wrong."

[tribal wedge]

servers are down probably

[heady charm]

Me also

[tepid hawk]

Same here. Is it down?

[violet bear]

same here

[tepid hawk]

It won't work at all and I keep getting that error message.

[obtuse tundra]

Same

I got the error message 6-7 times and then when I tried to reload or start and new chat it does not let me access GPT 4. ‘Previous model unavailable. We’ve switched you to latest default model’

[willow wyvern]

Yep, still broken, Any updates on this? This is getting really stupid.

[mellow echo]

fire

[spark hill]

These are some rough responses lol

[bronze charm]

literally useless shut down the whole operation

[hazy nexus]

I paid with buy credit, but the amount due hasn't decreased.

[signal beacon]

hi guys. I need Chatgpt to help me look for words/definitions on the index page of a textbook. Is there any recommended RAG program? Preferably I don't want to use API, it's okay to take a little bit more time and I just copy and paste RAG retrieved words?

[hushed hare]

I tried similar to that a few days ago and it just said it could not assist me with that because it would be copyright infringement against IP holders

[signal beacon]

that's sad

[hushed hare]

Even information from official product manuals are sensitive copyright information. I think ChatGPT is on the path towards no longer being useful soon, if these are use cases OpenAI can't condone

[shy garden]

Hi, Whenever I try to generate an image on my Safari Mac, the generated image will be only be a flashing white sqaure. I do not have the same problem on IOS Safari or Chrome Mac

Does anybody encounterd a similar problem?

[errant kindle]

is there any way to refund Team account payment?

[copper snow]

Hey everyone! Is chatgpt down for you? Everytime I try anything i get: Something went wrong. If this issue persists please contact us through our help center at help.openai.com. ?

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

[willow wyvern]

don't tell it it's a text book, be very generic about the instructions you want it to do.

"I am going to show you a big list of words and I want you to....."

Remove any words that might hint to it that it's a textbook

[willow wyvern]

This was pretty much always the plan. They don't want it to be too useful...for free.

Even the paid API tiers are basically just an experiment for them.

[hushed hare]

Using ChatGPT Plus, at least for now. But also putting in time towards learning the locally ran stuff

[willow wyvern]

the amount of resources they give people for the prices is still on the "insane" side of things.

it's not going to last.

Also. Did anyone ever actually believe that the sam dude actually has a way to shut down chatgpt in his silly backpack? it's total self aggrandizing cringe.

[hushed hare]

Yeah I figure price likely goes up significantly or it gets nerfed heavily over time due to restrictions, or both

[willow wyvern]

Also, I wonder how long until the main way to prove you aren't AI is to say something offensive.

[atomic crest]

i want to create RAG. i now created a vector database using pinecone serverless.
used cosine similarity to bring up top 3 similar vectors to query
now i realized that i only have the vectors that are similar but i don't have any info regarding the text of these particular vectors.
i didn't bother about this while using llamaindex and langchain because it worked automatically with their help but i wanna do it without using any of these libraries. all tutorials are using these.
how do i store text of vectors (no llamaindex, langchain,etc.)?

[sullen raft]

Same issue is occuring

[sullen raft]

why did my bug ticket get removed

[rigid musk]

Did you follow the https://discord.com/channels/974519864045756446/1070242883677982771

[sullen raft]

[merry blade]

hi any other med student help me on how chatgpt plus can SAVE me? med school is TOUGH i am failling out. thx

[rigid musk]

Ask chatgpt how to study better and give some tricks and study harder ?

[frigid star]

is there a way to copy paste emails from chat to gmail without hashtags and keep the format created in chat gpt? i always have to adjust after transferring

[empty bolt]

I keep getting this error when GPT-4 exceeds a character limit. Is this a limit that I can change on my end? And why is the only option to "regenerate"? I would like to change the prompt instead, so that the error doesn't occur again.

[meager ginkgo]

what is this bs? I only asked him like 10 messages for the whole day

@vagrant plover is there a issue rn or smt, because I don't pay monthly for 10 messages a day

[vagrant plover]

did you edit and/or regenerate any messages?

[meager ginkgo]

I did once because it failed to load

[vagrant plover]

how long are your generations usually and how long would you estimate they take?

[empty bolt]

Hard to tell. I just asked it to summarise a chapter out of a text book.

[meager ginkgo]

only for this one and it's been loading for some good minutes now

[vagrant plover]

Is this your own GPT?

[meager ginkgo]

no, it's a public one

[vagrant plover]

got it, hmm

[meager ginkgo]

I tried regenerating because it doesn't stop loading

now it gives me this, but if I go back one response it still loads

[vagrant plover]

How are you regenerating if you just reached the limit a moment ago?

[meager ginkgo]

It's a good question but it somehow worked but it failed in the same time

might be glitched, I refreshed the page but nothing changes

you can get a programmer here if they are available to check

[orchid spear]

Please help me return this convo back to gpt-4, I tried some solutions on reddit like the firefox network FETCH and chrome... no solution

[vagrant plover]

Once switching the model in a conversationto GPT-3.5 you cannot switch back to GPT-4, your best bet is to create a new conversation and copy over your prompts

[strong bluff]

Hello, I am a user that is tired of not being able to communicate with the dev team nor find help when crap goes wrong with ChatGPT chat bot, which I pay a monthly fee to have access to. I joined the severer to hopefully finally find a human to talk to to see if there is a way to resolve said issues with ChatGPT. Why is there a new captcha challenge in the chatbot and why is it insinuating that there is some odd behavior that I should be blocked for? I don't mean to be rude. I just really do need this chatbot to work, elsewise I will need to stop paying for it and go elsewhere.

[timber shadow]

Hello! First: you can submit a ticket for OpenAI support at help.openai.com. Open the chat widget in the bottom-right, click through the most relevant options, then you'll eventually find a path that lets you type your own message. This will submit a ticket that OpenAI support can review and get back to you on. It may take days. You'll get an email when they reply to you!

Second: if you feel like troubleshooting the captcha issue here, we can try! To start with that: do you have a VPN installed?

[timber shadow]

Hi! Best way to reach the mods for inquiries like this is DMing @burnt galleon -- click the username, click "Message", send your message, then select the option that sends the message to the Discord moderators.

[strong bluff]

Help at help.openai.com is a wall. It does not ever offer contact of any kind.

[timber shadow]

It does, it may just take days to get a reply. Using the chat widget, you'll eventually find a path through the default options that lets you type a message. This is submitting a ticket. They'll be able to review and get back to you from there.

[strong bluff]

I don't have a vpn. and it is working as expected in FIreFox I now know. I tred it out a moment ago

I have searched the page as fine as I can and have never found a live help or contact button of any kind

[timber shadow]

Interesting! It may be a browser extension or setting in the non-Firefox browser. You might try in a private window of the other browser, or reviewing your cookie/security/privacy settings in the other browser.

[strong bluff]

I am grateful for your responses

I did a test in an incog window and it opens

[hot remnant]

Hello, yesterday I was in a GPT4 chat and it suddenly became a GPT3.5 chat. I did not reach the limit and it did not say that I did, instead it said "The previous model used in this conversation is unavailable. We've switched you to the latest default model." However when I created a new chat it was GPT4. My opinion is that GPT4 was updated but the default was set to the wrong model and by doing so it ruined the entirety of previously existing chats.

[strong bluff]

But it was working as expected earlier today without issues

[timber shadow]

This is the chat widget, should you need to try in the future

[strong bluff]

I don't get that and haven't at any time.

[timber shadow]

Does it appear in Firefox or in an incognito window?

[strong bluff]

Good question

I will check

It does indeed in FireFox.

I may just switch all my ChatGPT use to FireFox

[mossy sinew]

I'm not sure when it stopped working, but my Chat GPT ( just started a new chat instance using chatgpt4) is telling me it is unable to acces external website links. They've taken this function away?

[timber shadow]

I'm glad it works! I love Firefox, glad to have you! 😎

[toxic hawk]

NO WAY GPTs ARE FREE??????!!!!!!!!!!

[timber shadow]

Some might have paywalls built in that you'll notice when you try using them. Nothing will charge you without your knowledge -- it would require entering payment info into the site of the creator of the GPT. And many GPTs are completely free to use!

[tall jungle]

Hi all, I initially used google login but need to set a password to login from my other devices where I don't have my google account logged in. I tried contacting support but the AI Bot thing keeps blocking me 😄 how can I get help to set password?

[timber shadow]

Hello, there's currently no way to change your login method like this, here's a relevant help article: https://help.openai.com/en/articles/4936824-can-i-change-my-authentication-method

[tall jungle]

soo, how am I supposed to login to ChatGPT from my other devices where google login is not an option? Should I cancel subscription, wait for it's active duration, create new account with password, then login?

[timber shadow]

Is there something preventing you from signing in with your Google account, separate from ChatGPT?

[tall jungle]

Yeah, long story but I'm not allowed to login to personal google account from that device.

[timber shadow]

Hmmm that's a toughie. I'm not sure of a good answer right now, but keep in mind that deleting your account won't let you remake a new account using the same email address: https://help.openai.com/en/articles/6378407-how-to-delete-your-account

[tall jungle]

that's okay but I don't want to double pay, so I will cancel, use the rest of the month's access and create a new account from a different address, this time using a password. But this needs to be changed IMO. Why can't I just set a password to my account?

[timber shadow]

I'm not sure, I imagine it has to do with how user authentication happens on the back end with these external sign in options. You could try reaching out to OpenAI support with this question at help.openai.com, or even making a post over in #1070006151938314300 for improved flexibility with sign in methods.

[tall jungle]

Okay, thanks for the help though ❤️

[indigo hemlock]

so i've got this thread i'm trying to archive. it won't archive. i can archive others. the same thread also cannot be renamed.

[sullen raft]

this is 100000% what has been happening to me

but then how does one write a bug report on reproducibility if it happens randomly. All i know is it has happened at least a handful of times throughout the week

[meager ginkgo]

you gotta analyze the processes when that happens in coding nothing is random

watch some yt tutorials I am not gonna invest that much time to write a bug report

[sullen raft]

watch a yt tutorial on writing a bug report?

[meager ginkgo]

on debugging

[sullen raft]

true but doesnt one need to have an idea of what the issue is in the first place? I get what you mean but Ill keep an eye and see particularly why it occurs. I definitely do close the browser etc and theres no weird activity going on.

[frank dragon]

why doesnt chatgpt tell me if ive hit the cap and when i can use it again? now it just returns Something went wrong. If this issue persists please contact us through our help center at help.openai.com.

[timber shadow]

That is a separate error -- when hitting the cap, you will be so told and it'll include a time when you get your next use back. For that error, it's usually a network or browser issue. Do you have a VPN installed?

[sacred field]

Does anybody know when will the GPT Store be available in the UK?

[fading bay]

I need to log out on all devices where my account is active, as my account is now active on a device that I do not have access to. In other words, I need to LOG OUT on all devices and re-access my account one by one. Does anyone know how I can log out on all devices?

[timber shadow]

Hello, I don't think this is currently possible, at least user-side. You can try reaching out to OpenAI support directly at help.openai.com. Open the chat widget in the bottom-right, click through the most relevant options, then you'll eventually find a path that lets you type your own message. This will submit a ticket that OpenAI support can review and get back to you on. It may take days. You'll get an email when they reply to you!

[fading bay]

Ty!!

[strong turtle]

Hey !
When I wanted to renew my gpt + subscription, I noticed that the link on my pc was no longer working, so I renewed my subscription from the app. I got my subscription back on my phone, but nothing on my computer. how can I get access back on my pc (I live in Belgium)?

and how do I speak to a real person on the help center?

[coral monolith]

im writing proposals using gpt. and it started out well. weeks ago i started the 4.0 paid version but it seems to have regressed. any decent tutorials out there to make sure im using the prompts well enough. it really did start out well but now its becoming barely usefull. also gpt4. came to an end even though i paid. seems like its not seeing my account sometimes

[livid lark]

So what's the deal with the 90 word summary rule? I get it over the weekend. Then today I'm able to overcome it. Now I'm back to being limited again. Does this thing get adjusted on a daily basis? This is for a customGPT FYI

[mint finch]

@burnt galleon Am I able to ping this or some more appropriate channel to solicit feedback on a GPT? Can I offer trade (I check out theirs) or payment?

[fallow snow]

Anyone else having issues with retrying when mentioning custom gpt?

And it retries with the default gpt instead

[past hemlock]

I just got that mention functionality and it is interesting, but kind of a bummer that it doesn't work on the Chrome browser on my Android device. Is there any way to get it working on there? Or is this just something for the desktop? Or maybe eventually they'll add it to the Android app?

[mossy mesa]

Just a desktop feature right now.

[fleet swan]

Hi I’m using Assistants API and set my model to gpt-3.5. Is there a way to change the model depending on a query? I don’t want LLM to decide which model to use, rather I have another model to route (if A - use gpt-4, else use gpt-3.5).

[restive kestrel]

can i get some help with my coding home wrk : )

[lean fiber]

Hi is there anybody knows, i was in restriction of chatgpt, means its available in that country, can i use that country card for payment and subscription or it also can not? Another question if credit card visa is issued from that restriction country ? Thanks

[mint wigeon]

Hi, I have GPT plus but still do not have Explore GPTs. Is that usual? I am in Australia. Appreciate any advice.

[rugged wing]

Hi guys, what is the chat limit for GPT 4 in ChatGPT pro and is it different with GPTs

[torn cypress]

Sad fact, I was tricked into using ChatGpt for the number of times. In the process of using it today, he kept prompting me: !
Hmm... Something seems to have gone wrong. I don't know exactly what went wrong, maybe the tragedy started because Gpt4 tried to talk to itself, but now I want to solve it. He seriously affected my use.
Attempts have been made to give feedback to the authorities, but the advice he gave me has not helped in any way.

[past geyser]

The AI chatbot won't be able to parse your images. Keep in mind the software isn't trying to "trick" you - but there are a number of reasons errors can occur.

[torn cypress]

Anyway, your bug is causing ChatGPT4 to keep repeating new conversations. He consumed my times, and I can't use chatgpt4 normally.

[past geyser]

If you can document the error clearly you can submit the issue to #1070006915414900886, beyond the support channel directed to you by the chatbot.

[elder sierra]

HI guys, Im pretty new in OpenAI and I'm looking for help . RN i would like to create index using

VectorstoreIndexCreator().from_loaders([loader],)

and i want to use a different embedding model. By default OpenAIEmbeddings is using a text-embedding-ada-002 but i want to use** text-embedding-3-large** ->

                model="text-embedding-3-large"
            )
index = VectorstoreIndexCreator(embedding=embeddings).from_loaders([loader],)

I did not find proper way how to change the model. I only follow what I read from the descriptions in the library. And i got Warning: model not found. Using cl100k_base encoding.
Can someone explain me how to properly change the embeding model to different one ?

[past geyser]

You might have better luck in #dev-chat with this question!

[pearl plank]

i somehow cant access dall-e anymore?

[rustic marlin]

im not sure of what you mean, the image you sent only shows that you where about to click on the GPT-4 model with dalle?

what is the issue?

[humble linden]

Hi all! new here!

Someone have problems to use the new feature of GPT mentions.

When I use de "@" in GPT chat, nothing happens. I have a pro acount

[timber shadow]

Hello! They typically roll out features gradually instead of all at once -- I heard of someone just getting the feature yesterday, so it's possible the rollout hasn't reached you yet. I don't think the feature is live on all platforms either, so be sure to test in the desktop and mobile version of the website if you happen to have only tried in the app so far.

[fresh canopy]

hey guys do you know why i get this error so much on my laptop "Our systems have detected unusual activity from your system. Please try again later." and I also get verification for every question I ask. I can use chat gpt 4 on my phone but my laptop cant anymore

[timber shadow]

Hello, do you have a VPN installed on your laptop?

[fresh canopy]

maybe yes

[timber shadow]

Which VPN do you have? Can you try disabling it, then trying again?

[fresh canopy]

i have an extension

removed it but still issues

[timber shadow]

Can you try in a private window with all extensions disabled?

[fresh canopy]

im oging on a private browser right noe

yea its being so weird

it lets me type on 3.5 but wont even let me enter anything for 4

[timber shadow]

What browser are you using?

[fresh canopy]

google chrome

[timber shadow]

As a test, can you try in a fresh install of Firefox? Or at least, a private Firefox window with all extensions disabled?

[charred scroll]

Does anyone know of a way to get your chatgpt account signed out of all devices? Once a device has access to google account, there is no way to sign out of chatgpt even if the google account is signed out.

[fresh canopy]

i dont think you can do that right now

but idont know

[timber shadow]

There is currently no way to sign out of all sessions, for this you might try reaching out to support at help.openai.com, though I don't know if they'll be able to assist! Open the chat widget in the bottom-right, click through the most relevant options, then you'll eventually find a path that lets you type your own message. This will submit a ticket that OpenAI support can review and get back to you on. It may take days. You'll get an email when they reply to you!

[indigo bronze]

Hi is it still possible to create ai images with discord bot here?

For free

[rigid spade]

Are you guys also having the problem that u reach the usage limit after just 19 dall-e messages?

[timber shadow]

Hello, no you can't generate images in this server. If you want to try DALL·E: Bing's image creator gives you free daily credits.

[snow raptor]

hello, i have not been able to use the gpt version 4 since yesterday. 3.5 seems to work fine but the version 4 i am getting "hmm something went wrong". I have tried everything to fix it from my end - clearing cache, allowing cookies, different browser, different network connection but am not able to resolve. Whats worse, it takes it as a prompt and is giving me usage caps

[muted turret]

Is there a fine tuning channel here?

[timber shadow]

Hello, for fine-tuning via API, best starting place would probably be #dev-chat!

[timber shadow]

Hello, do you have a VPN installed?

[snow raptor]

nope, i am not using any vpn

[timber shadow]

What browsers have you tried in already?

[snow raptor]

edge and chrome

[timber shadow]

As a test, can you try in a fresh install of Firefox, or at least a private window of Firefox with all extensions disabled?

[snow raptor]

getting same issue

[timber shadow]

Do you have another device and network you can try, like a smartphone on a cellular network?

[snow raptor]

I tried with my cellular hotspot just now. I'm not really sure why i am getting this issue

[timber shadow]

What about on the smartphone itself, on its cellular network? So different network and different device--does going to chat.openai.com on cellular, on the phone produce the same result?

[snow raptor]

i am getting same issue on my smartphone as well

[timber shadow]

Dang, so different networks and different devices, same issue. You might try reaching out to OpenAI support at help.openai.com. Open the chat widget in the bottom-right, click through the most relevant options, then you'll eventually find a path that lets you type your own message. This will submit a ticket that OpenAI support can review and get back to you on. It may take days. You'll get an email when they reply to you! There have been similar posts in #1070006915414900886 too, might chime in in one of those too.

[muted turret]

Thank you!

[sour salmon]

Keep getting a LOT of network erros, ChatGPT not able to process simple calculations. Anybody else getting the same?

since yesterday. All the time this.

[noble yoke]

Is there a way to get more computing power from ChatGPT? I ran a Prompt that asks a question and provides 5 synonyms for a word and then takes that output and places into an Excel file with two other parameters. Prompot is successful, however, ChatGPT can only do one or two at a tim before it givesd me an error message that the file is large and can't process it.

[rustic marlin]

how big is the file?

[noble yoke]

I started with 100 rows but then decreased to just 10 rowes and receive the same response

[rustic marlin]

it is not big at all

[noble yoke]

I know and is why I am concerned

[rustic marlin]

you want it to replace a word on a excel file?

it should be able to do it on a code interpreter call

[noble yoke]

I give GPT a word then ask it to provide 5 synonyms for the work separated by a " ; " and then place the output into an Excel file

[noble yoke]

As I am not a programmer I am not sure how to do a code interpreter call but will do some research on it.

[pulsar pewter]

hi guys - how can I prevent GPT 3.5-turbo from giving me predictable outcomes.

Two examples:

• Asking for it to generate a random name: 4/5 times I receive a variant to 'Emily'
• I provide it a defined list of ~10 options, but it keeps returning the same two values

Any prompt guidance is welcomed!

[rustic marlin]

there isn't much that can be done about that on chatgpt

this kind of behaviour is intentional

[pulsar pewter]

I guess that makes sense. So, I might just be better using like RANDOM method ?

[jovial crag]

I would like it to show 3 buttons like these after the answer. How can I add those?

[atomic jetty]

Since 30 minutes ago I'm getting the following error Something went wrong. If this issue persists please contact us through our help center at help.openai.com.

[timber shadow]

Hello, I don't think custom GPTs support custom UI elements like that -- where you'd be able to implement buttons, etc. that can be used to continue a chat. You might be able to do something like this via an API action of some kind, but as far as I know, actions can still only display different kinds of output, not help to control input.

You could probably program your own implementation of something like this via the API though, but that would be separate from custom GPTs on ChatGPT.

[timber shadow]

Hello, do you have a VPN installed?

[atomic jetty]

not, now it works, 5 minutes ago

[timber shadow]

Good deal, did you have to change anything on your end or did it just fix itself?

[atomic jetty]

I just waited

[subtle wedge]

I'm trying to pay but the address area is disabled

[rigid musk]

Try another browser fir a test

[subtle wedge]

nevermind, I went into devtools and enabled the fields

[jovial crag]

This is from a custom GPT called "LOGO" By logogpts

[timber shadow]

Ohh I see, you don't mean actual UI buttons, you mean just implementing a branching decision path? You could probably do this in the instructions field of your GPT (on the "Configure" screen of GPT builder) with some structure like:```When a user tells you x, follow up by presenting them with the following three options:

1. a
2. b
3. c
   If they choose a, then do...with x.
   If they choose b, then do...with x.
   If they choose c, then do...with x.```

[jovial crag]

Thank you!!

[timber shadow]

You are welcome! If you start working on something for your case and want to ask for further feedback/advice, you might share in #prompt-engineering: (1) what your instructions are, (2) what you want the outcome to be, and (3) what the current outcome is. Plenty of helpful folks in there who might be able to keep assisting!

[jovial crag]

Ok, got it. Sorry about that! Thank you again.

[timber shadow]

Not at all, you're in the right place! #prompt-engineering is just a great place to expand on specific prompting techniques based on what you have and what your goals are 🙂

[jovial crag]

Or I messed up or it's not working. I'll ask at #prompt-engineering . 🙂

[merry jewel]

So I am a premium ChatGPT user... and for the second time this week I have been told that I have reached my messaging limit. BUT HALF OF THE RESPONSES WERE REPEAT ANSWERS OR FAILED RESPONSES. Anyone else confused at why they are limited the paid users? How can they be counting faulty responses for their 40 messages every 3 hour limit????

[rigid musk]

wdym faulty answer

if you are talking about error network 90% coming from user side

[merry jewel]

it says something to the effect of "there was an error generating this response, please try again" in a red box.

[rigid musk]

what browser are you using?

[merry jewel]

chrome

[rigid musk]

can you try firefox?

[merry jewel]

i would love to but they have already restricted me because I reached my "limit" even though i am a paid user and over half of the responses came back as "errors"

[rigid musk]

you paid for a beta products. bugs can occurs. wait till the cooldown, try with another browser for a test

[merry jewel]

look, i understand it is "beta," but how can the paid user be limited and the free user have no restrictions?

[rigid musk]

free user doesnt have access to gpt 4, dalle, plug in

[quiet meteor]

hello i want to ask is it posible chat gpt to convert me a screenshot into a word equation format MathML?

[humble linden]

Thanks @timber shadow , I suppose you're right, I've tried on mobile and on the web and neither of them work. I tried with a friend's account and it works for him. It's just a matter of waiting. Thank you very much for your response!

[ember summit]

Does anyone have any idea on how I make ai art on discord

[timber shadow]

Hello, you can't use DALL·E in this server, but if you'd like to try DALL·E for free: Bing's image creator gives you free daily credits! Not on Discord though.

[keen jasper]

Does anybody know why I don't see all GPT's on my store?

[strong flame]

Anyone else having this issue rn?

Ok I just sighed out of the app and now it says It gives me an error when I try to sign back in

[hollow bloom]

@keen jasper I mean I don't even see a gpt store, apparently its either bugged or being rolled out slowly. cause I have been checking everyday on multiple machines and browsers and I don't have access to the store. OpenAI support told me its being rolled out...which they only said after asking for a screenshot after I described the issue. makes me believe there are a lot of issues with the store in general

[keen jasper]

Thanks @hollow bloom this is exactly what I was looking for.

[hollow bloom]

no problem its been a frustrating experience to be sure, hopefully we get to use and publish our own gpts in the store soon

[summer sinew]

I am looking for advice in using MyGPT. I created a GPT that will create a business logo and I am having a hard time getting the ai to do what I am prompting it to do. The ai will not spell words correctly, use all the words in the business name, and, if I like a design but want to make a change to it, I can not make any changes

[signal meadow]

I need help to fix this

I doesn't let me do anything

And before getting to that page I get a cloudfare verification and thats all

I need to use my chats for work and haven't been able to do so

[wise lodge]

When building GPT actions, what format do bearer tokens need to be in? RSA keys don't seem to be working

[pine crag]

How would one go about teaching the ai to code in a scripting language it doesn't know (in this instance Postal3Script (P3S)), I have all the necessary documentation but can't upload it because it doesn't take file uploads

[timber shadow]

Hello, DALL·E is better than many comparable models at producing readable text, but it's not perfect. It also makes every image "from scratch" for the most part, so it can't currently do any direct photo editing like keeping a design mostly the same and changing only small parts of it. In general, asking for it to produce less text will return more reliable/readable results than more text.

[timber shadow]

Hello, can you try in a private window of your browser with any browser extensions disabled? Is this Chrome?

[summer sinew]

I am using fire fox

Are you thinking that extensions are messing with the functions of the ai program?

Nevermind. I see you were talking to another poster

[tropic spade]

did anyone have problem access ing chatgpt ?

i have not been able to access it directly it only works through a VPN ?

and that is frustrating my hosting does not allow VPN connection so i had to switch between toggle on and off VPN during working

[sullen raft]

rip I am still getting hit with random cap limits

[lean crag]

Anyone having troulbe with 4.0? I just purchaed a plan and any time i try and prompt i get Hmm...something seems to have gone wrong.

3.5 seems to reply fine

[molten wren]

Maybe it's just us, but I think it's down.
And yes, 3.5 worked for me as well.

[barren edge]

it is me also

[fallow meadow]

I can't use GPT4 today on my Plus subscription. I always get this error: "Hmm...something seems to have gone wrong."
That's weird I tried everything. Even different device, connection, browser

same here

[stiff ingot]

Guys I have a doubt:

Why is ChatGPT Plus getting dumber and slower?

• it is answering in much shorter format when doing research analyses. For example when I need it to analyse some papers it gives me just a couple of rows of text compared to like 2-3 months when it wrote 600 words.
• i have to keep repeating myself over and over again and give it a lot of extra info for tasks that a while ago it did in an instant
• it cannot make a simple table with all the data provided to it.

So let me know if you have experienced similar things lately and if you might have an aswer

[exotic lantern]

how do i use dalle e

[past geyser]

you can navigate to labs.openai.com to use DALL·E 2, or ChatGPT Plus subscribers can access DALL·E 3 as a plugin. Bing's image creator also uses a recent version of DALL·E

[plain zephyr]

getting the following on ChatGPT:

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

Any ideas? i cant even log out and back in

[timber shadow]

Hello, do you have a VPN installed? Can you try in a private window with all browser extensions disabled? If using Chrome, can you try in Firefox, or vice versa?

[smoky idol]

hi guys

i got a question

[velvet blaze]

hi! how hard would it be to implement a chatbot as customer service for our school, entertaining inquiries about how to enroll or such? for our thesis. especially by integrating gpt

[rigid musk]

With api, but for chatbot customer don't think chatgpt is the solution

[velvet blaze]

oh okay. so is it feasible?

[rigid musk]

Yet but not free and will be expensive

[velvet blaze]

oh. any cheaper alternative?

[rigid musk]

I guess chstbot by a service

[velvet blaze]

by a service? what does that mean?

[rigid musk]

Specialized business providing chatbot service.

Monthly cost. Anyways we are getting #off-topic

[tulip niche]

what is this error😭

[exotic lantern]

thanks

[rustic marlin]

so many people are using it that it can't reply to you right now

[exotic lantern]

but is there a free version of this dalle?

[rustic marlin]

you can use dalle3 for free on bing image creator

[exotic lantern]

nice

so is it like a limited usage thing?

[rustic marlin]

kinda

[glad heath]

Hi guys is therre any issue with gpt 4right now ? (It says I can't use it until 3.30 pm but its 3.45 already)

[exotic lantern]

bing image creator ain working too

[obtuse gust]

today it is not working well

[timber shadow]

I'm glad they have this new error message at least -- makes it more clear that it's not an error on the user side.

[merry jewel]

Another day, another round of ChatGPT 4 wasting all my responses on errors and wrong messages. So I am currently asking ChatGPT to make a report of all its errors and wrong messages so I can submit to OpenAI. Does anyone have a customer service email? I have submitted multiple tickets now with their help bot and no one has emailed me back.

[timber shadow]

The chat bot at help.openai.com is the only way to submit a ticket currently -- last we heard here, there's no monitored support email address.

[merry jewel]

so when can i expect a response?

[merry jewel]

i totally understand they are "beta" and that it is a "startup" but i am finding the errors and wrong responses to be beyond acceptable at this point lol. why are we all paying for the subscription when it is so faulty? how can they have such a lack of customer service?

[timber shadow]

I think it varies based on the request! Typically many days I think -- depending on what the tickets are I imagine some might not warrant a response too? Not sure! As for the bugs in general: they're clear in their ToS about not guaranteeing any kind of stability with these beta features. It's unfortunate and can be frustrating, but it's just kind of the price we pay to be able to try this stuff at all, in my opinion.

[merry jewel]

the leniency that we give OpenAI versus every other tech platform ever created is wild. we should expect better. the premium paid services should either function better before being launched or they should prorate our rates if there are this many problems, or they should at least email back that they are working on the problem. to let us pay for something that barely functions and then leave us in the dark is the opposite of what a brand or company should do.

[timber shadow]

From another perspective (self-acknowledging my fanboyness here, too 😁): ChatGPT launched as a GPT-3.5 demo. ChatGPT Plus launched as priority access to 3.5. Both of these services have remained delivered as promised, even improved, despite the fact that demand dramatically outpaced expectations.

Everything else on the bleeding edge are perks on top of the baseline service, and for most users, it remains the only viable way that individuals like you can me and even hope to experiment with them.

[keen hollow]

@merry jewel It's usually taken me a few days to weeks for "simple things". It's been over a month to get them to respond to something they've not been able to work out

[opaque oak]

I've been getting a message from ChatGPT for the last several days. It seems to come up randomly and thinks I'm a bot. I pass the puzzle test. But it puts me in a loop. Clearing cache, logging out, nothing seems to help.

[timber shadow]

Hello! Do you have a VPN installed?

[opaque oak]

I do not have a VPN. My wife uses a VPN, on a different computer.

I'm going to try to switch to my backup network.

BTW, she doesn't use ChatGPT. Doesn't even have an account. So I wouldn't think that would be an issue.

[timber shadow]

Have you tried in a private window of your browser, with any browser extensions disabled?

[opaque oak]

I haven't tried that. I have deleted cookies and emptied cache. I had a browser extension that allowed me to search GPT history. I've removed it.

[timber shadow]

What browser and OS are you on?

[opaque oak]

Chrome and MacOS

I keep them up to date.

[timber shadow]

Can you try:

• A private window of Chrome with all extensions disabled, and
• A fresh install of Firefox, or at least a private window of Firefox with all extensions disabled?

[opaque oak]

I'll have to figure out how to disable the extensions in a private window. I don't know how to do that.

[timber shadow]

In your extensions menu, there should be an option like "Allow in private window" -- toggle that off. Or, just turn off all your extensions manually and then open a private window, this will do the same thing

[opaque oak]

Thanks. It'd be easier for me to install Firefox. I'll give that a try the next time ithappens.

[lost hollow]

well I was just facing the same issue, out of the blue receiving this message on my first for the day message on a chat, and after a long and fruitless hunt for a possible solution, I tried to "bug it" out a bit more.
I switched to the iOS app in the same chat, posted another redundant message, just to see if it'll respond. And it did respond there... So, then I went back to the browser chat and I am currently continuing from there, without that issue.

@opaque oak maybe that'll help you as well.
@timber shadow I don't know if such an approach is recommended or appropriate 😄 but it worked.

Also, just got a 5 min timeout for misspelling "as" with two s's 😛 sometimes, it's just a typo, right? 😄

[wise lodge]

Can anyone please help me set up bearer token for my GPT action? The following works fine in postman:

curl --location 'MyServerURL' --header 'Authorization: Bearer test'

But for some reason the GPT gets:

Encountered exception: <class 'aiohttp.client_exceptions.ClientResponseError'>.

Same bearer token in both clients

same URL, same method

[fallow snow]

Encourage a storyline where characters work together, learn from each other, or face challenges as equals, highlighting their strengths, growth, and mutual respect.

Is there no way to get the ai to stop trying to force these elements in my story

[pale valve]

Hi, all. I have a valid secret - I THINK! - but every time I use curl to access a simple test completion, I'm getting an error 429. I'm using the key that openai issues when I try to create a key, but at no point can I get it to complete anything; I'm a subscriber, I'm using the 3.5 model (although I'd like to use 4 for this)... any idea what I might be doing wrong?

[pale valve]

I've been trying the curl command using the OpenAI docs, plugging in my secret key for the bearer authentication, and STILL get the 429. Doesn't matter what model I use - and the call to GET the model list works.

[past geyser]

You might get better assistance in #dev-chat

[pale valve]

Thank you. Will post there.

[past geyser]

The AI is definitely provided ethical guidelines that impact its output in fiction. Have you played with custom instructions at all? If you explicitly ask it to avoid specific tropes or emphasise ambiguity it will likely help a bit.

You can prime your outputs with overarching instructions that always apply through your settings.

[fallow snow]

Yeah it does tend to focus a bit too much on characters feeling guilt and everyone being redeemable

“please ensure you never generate ethics and respect and responsibilities!!!!!”

Trying to use this to varying results

[past geyser]

That won't work, and is also not advisable. You have to be more analytical about the undesired features and specific in targetting them.

[fallow snow]

What should I try though?

[past geyser]

"When writing consider the drama created by clashing motivations and the satisfaction for readers when questions are left unanswered or conflicts left unresolved." I haven't read the outputs and don't know your exact aims though, so that's just where I might start.

[fallow snow]

It just tries to force these elements too much in my stories

So I’ve been trying to use that to get it done

Let me do some tests with yours

Would be nice for them to allow us to change custom instructions so we can actually experiment rather than trial and error…

[rugged wing]

is there an issue with chatgpt now? Are servers down?

[signal meadow]

I think I have a IP ban

I haven't been able to get to my work on gpt 4, turned on a VPN just to try and see, and it worked ! But I don't know why I have to do this. Every time y try to access I have to go through a cloudfare check and then im blocked. Does anyone know how I can contact support for this ?

[rocky hemlock]

Im having issues with gpt4 rn

3.5 is fine but 4 isn't

[tough scaffold]

same

[trail pollen]

I'm getting a lot of "Something went wrong. If this issue persists please contact us through our help center at help.openai.com."

[tough scaffold]

yeah its def down. was running very slow prior.

[tight cliff]

hi, I'm seeing unrecognized conversations with always the same prompts (null, undefined) why? how do i stop it? there are multiple new chats every day

[sinful grail]

Okay, I thought I was crazy with GPT-4 erroring with every single request I make

[rustic marlin]

it is very likely your account has been compromised, strongly advise to change the password

[summer sinew]

I am looking for advice in using MyGPT. I created a GPT that will create a business logo and I am having a hard time getting the ai to do what I am prompting it to do. The ai will not spell words correctly, use all the words in the business name, and, if I like a design but want to make a change to it, I can not make any changes

[sullen raft]

same

[tough scaffold]

Are there any devs in this discord or?

[vapid glade]

hello there, code generation (python) fails all the times, what can I do to fix it? It used to work just fine but no it's useless

[rustic marlin]

yes, but they don't chat very often

[fringe trench]

Today, I just stumbled upon this issue. I can only do one prompt when conversation history is disabled. I consistently get this error message: "Conversation key not found. Try starting a new conversation." I tried several things already, but nothing seems to fix it. Any ideas?

[tough scaffold]

is there a cap on entries even with paid subscription?

[bronze charm]

40 prompts / 3 hours for GPT-4, 25 of which can be allocated to custom GPTs

[chrome prism]

It seems like there are limited number of lines of code CHATGPT can return on the web server. Is there anyway to increase that number?

[static cloak]

Is there a separate channel for gpt 4 users??

[thorn nest]

ChatGPT stops responding, and 3.5 or 4 gives no answer; I am only waiting 😦

[rugged wing]

Any idea what this means: 'Our systems have detected unusual activity from your system. Please try again later.'

I can still use GPT-3.5, but I keep getting this message in GPT-4

[tulip umbra]

Is it just me or is gpt4 failing alot of calculation lately? Like start of 30/01/24. Just calculation on interests and loans but its have error analysing non stop now, it was perfectly capable a week ago.

[deep cosmos]

GPT chat is not functioning, it just doesn't respond, in the console, I see this error:
This is from yesterday night.

TypeError: Failed to execute 'getReader' on 'ReadableStream': ReadableStreamDefaultReader constructor can only accept readable streams that are not yet locked to a reader
    at em (_app-502171a9cdb09c3c.js:26:2437035)
    at eP (_app-502171a9cdb09c3c.js:26:2439251)
....
POST https://chat.openai.com/ces/v1/t net::ERR_BLOCKED_BY_CLIENT

[charred delta]

where can I use if i wanna get information regarding gpts and its Actions?

And, is actions consume more tokens than prompt? Trial for Actions reached limit earlier than prompt trial.

[haughty wadi]

Hello, we are having issues accessing the GPT-4 API for some reason it's not available on our account even though the account is paid and we have spent money on the GPT-3 API (and also, for some reason, we haven't been charged this month). We have been working on a SAAS project for real estate brokers for a long time, have made a custom plugin for GPT, etc., but unfortunately, we cannot use it. Can you help us with what to do?

[rustic marlin]

Actions consume just as much tokens, it only depends on how much text the action is giving to the AI

[oblique token]

I am not getting access to the normal interface like I should. My account is GPT PLUS, but I can't search for GPTs in the store. I can only access suggested GPTs and ones I have created. It does not allow me to search for other public GPTs.

Also, mentioning other GPTs with "@" does not work. As a paid PLUS member, I should be able to search and access all public GPTs in the store, not just suggested ones and my own creations. The interface should allow full GPT exploration and the "@" mention feature should work to call other GPTs.

My current limited access and inability to use "@" is frustrating as a paying PLUS user. I would like the interface to be fixed so I can search GPTs and use mentions as intended with my subscription.

[timber shadow]

Hello, OpenAI often releases features on a rolling basis, and not all at once, so you may just not have access to these features yet. But, be sure to check different platforms (app/web) and different browsers (Chrome/Firefox), preferably in private tabs with all extensions disabled so no cookies/cache can potentially interfere. But again, the rollout just may not have reached you yet.

[frail raptor]

You have made gpt-4 almost unusable!, although I have filled out the form many times, no one returns, sometimes I don't use it for weeks, but when I want to use it, I want to use it because I paid for it, but your message limit prevents me very seriously, please someone help me now.

[trail cairn]

if you want more messages then you pay more

[frail raptor]

where do I pay?

[trail cairn]

upgrade to teams

[frail raptor]

A one-man team? And he wants an annuity?

I'm willing to pay 30$, but it wants 2 users.

[trail cairn]

yes, if your usage exceeds the common usage requirements you get teams or buy another plus, or use api

it wants 2 seats but you can use them both

[oblique token]

I already tried this in other web browsers and on my phone, but it didn't help. The GPT store has been available for awhile now, so it's ridiculous I can't access it as a paying member.

I opened a chat with OpenAI help about this issue some time ago, but still have not gotten a response. As a paying PLUS user, I should have full access to search and use all public GPT models in the store.

Not being able to search GPTs in the store or use the '@' mention feature is very frustrating. I hope OpenAI can fix this soon so I can access everything that comes with a PLUS subscription, like exploring all available GPTs and mentioning them easily.

[trail cairn]

do you use a custom dns server?

[oblique token]

nope

[frail raptor]

I don't quite understand, I choose a flexible plan for 30$, but it makes me choose a minimum of 2 seats, do I have to pay 60$?

[trail cairn]

yes

flexible means cancel any time and the ability to add more seats later

[frail raptor]

This is nonsense, yes I realize that there is a team plan, but I am an individual user and he is trying to make me buy 1 user to exceed this limit. This is not overpaying but forcing 2 users, it is very unfair.

I have to give this chair to someone, if I can't find it, it will be in vain, it's really unfair!

I live in Turkey and 60$ is not cheap for us!

[trail cairn]

but if your usage is above normal then you pay more and that's the way it goes in most services

[frail raptor]

I am already saying that I can pay more, but this is not an overage fee, this means that I have to increase my current user to 2 users, it should be a method over my single user, this is not a method, this is the method OpenAI uses to sell more users and it is really ridiculous!...

[trail cairn]

teams price is balanced on those factors of user count and commitment. you can wait and see if they will bring chatgpt pro later for solo users

[rustic marlin]

it ahs been replied by a OpenAI Employee on this post #1195333488614834228 message

[marble rock]

hey guys for some reason when I get the response in the data array, i get the assistant response as the first item in the array, although it should logically be the second item in the array, after the user message. was just wondering if anyone knows about this, would really appreciate your help!

0: Assistant Response"Hello! I'm here to assist you. How can I help you today?"
1: User Message "hey what's up"

[rustic marlin]

are you using chat completions or assistant endpoint?

this is correctly ordered, the array is ordered in a way that lets the most recent messages first

but if you need.... array.reverse() tho

[dusty holly]

only thing that wasnt great about teams is ayear in advance payment

overall, all of the last round of updates and changes have been good plays

[mellow ridge]

i have just purchased the plus and it gives me this message, wth?

i mean that this is the first message

[marble rock]

assistant endpoint

[marble rock]

why is it like that though? in most applications you want to have the last item in the array be the latest message, that's how chats are rendered

On ChatGPT if it's rendered using messagesArray.map() then it will render the user's message after the assistant's response, which makes no sense to me