community-help

[regal lynx]

nice! ChatGPT can do lots of things. If you're looking for inspiration, #1050184247920562316 has plenty of neat ideas. If you happen to have ChatGPT Plus, you can browse #1171489862164168774 to see if any interest you too!

[lofty wedge]

are u a bot or r u real just wondering

[regal lynx]

The world may never know

[nocturne geyser]

good response

[regal lynx]

If I were to be a bot, I obviously wouldn't tell you 😉

[nocturne geyser]

the facade continues

[lofty wedge]

@regal lynx i keep asking chat gpt make a "Fortnite username and make it start with NRG_ and make the font of the username old school and make the username snow themed" and it just keeps not sending

[regal lynx]

ChatGPT can't send text in different fonts. You could try and have it output a certain file format that supports different fonts, but you'd have to run that on your own

[lofty wedge]

where

also

it does send in diffferent fonts it did it for me

[nocturne geyser]

ChatGPT's interface doesn't allow you to switch fonts
You'd have to use an extension online

[nocturne geyser]

well, that's not really a "different" font
it's a special type of unicode that the browser is capable of showcasing
if you mean like ChatGPT switching from Arial to Times New Roman, that's not possible

[lofty wedge]

ohh

do u play fortnite @nocturne geyser

[nocturne geyser]

no sir

[lofty wedge]

noooooooooo

[lofty wedge]

@regal lynx ?

[gritty anvil]

I forgot if this was the last server I was in... how can I get one thread to connect to the next one

[wraith zodiac]

hi, i have this message on my chatgpt site, i just want to know if i can find out about what the changes are? i have the pro version

[old hedge]

openai billed me twice, and overcharged account. Chat support doesn;t reply since 3 days.. how to solve this?

[timber shadow]

Hey! That message doesn't refer to ChatGPT in general, just the particular GPT you're using. It just means that the GPT has been updated since you started that particular chat with the GPT.

[wraith zodiac]

ok thanks

[rigid musk]

Unfortunately only help.openai.com can solve payment situation

[surreal cypress]

why am I hitting usage caps when using a gpt and it keeps defaulting to 3.5 not 4?
Makes zero sense and its rather frustrating that it keeps defaulting to 3.5 even though I'm being held accountable for 4.0 usage, I dont understand the advantage to using 3.5 when I'm expecting a 4.0 response, seems deceptive to make 3.5 the default when using gpt's makes me feel like openai cant keep up with the demand for 4.0 but make that the users problem.

[timber shadow]

Hey, can you explain a little more about what you're experiencing? If you hit the usage cap for 4 in the middle of a chat, a button should come up asking you if you want to switch to 3.5. You don't have to do this, and you can instead return to the chat when you have usage back to continue with 4.

Or are you saying in new chats it's defaulting you to 3.5 and not 4, like in the model dropdown selector?

[thick meadow]

Anyone knows why my gpt4 is slow and getting a lot of errors when chatgpt is trying to respond?

[frank socket]

hi, I can't send any more messages to chat gpt 4 plus

I just bought it today, and it won't let me send any more messages.

[fading robin]

mayby you hit a cap

mind sending a screenshot of what gpt say to you ? or something ? will help a lot

[frank socket]

: (

[rigid musk]

You have 40 prompt every 3 h for chatgpt plus

[surreal cypress]

this is what im talking about, its defaulting to 3.5 but still hitting a ussage cap for 4

[surreal cypress]

no

[frank socket]

I am reviewing for an exam I have in 30 minutes I was explaining and then it started to fail.

[rigid musk]

?

[rigid musk]

Failed generation count

[fading robin]

mind that custom gpts are ONLY on gpt-4

[frank socket]

what can I do to ask you more questions, it is the payment one.

[rigid musk]

Also you might fall under tos for school usage , maybe that's why it doesn't generate answer ?

[surreal cypress]

why does it say 3.5 by default?

[rigid musk]

You can use gpt 3.5 until your prompt are back .

[frank socket]

I paid for it today precisely for this 😭

[fading robin]

it may be a error but if you using custom gpts it will allways run on gpt-4 (i think cap is 25/3 hours)

[rigid musk]

It still is 40 prompt every 3h. Each prompt have their own cool down

[surreal cypress]

that makes way more sense than the 40 in 3, but if thats an error why does it respond differently when the 4 is applied?

[fading robin]

hm wierd... leme check cus i never seen the model chosing button is that a addon for browser ?

[surreal cypress]

like from a development stand point i dont know why the option for 3.5 or 4 isnt conditionally rendered.

no itss just the site

[fading robin]

yea the most basic gpt witchout dalle or webrowserind don't have that option

[surreal cypress]

from my browser

[frank socket]

I want to use it, I think when I bought it I didn't see a place where it says it has a limit.

I need it for now

[rigid musk]

You can use bing for free as temp alternatives

[surreal cypress]

or bard

[fading robin]

wait wait you used some tools to get that option appear ?

[surreal cypress]

not sure, as again im not using any plugins or a special browser

[frank socket]

buy chat gpt 4 plus to end up using bing 💀

[fading robin]

that one

[surreal cypress]

no

[fading robin]

very very interesting

[surreal cypress]

this is the stock site for me

[frank socket]

buy chat gpt 4 plus to end up using bing

[fading robin]

like i never seen that

they may adding that feature

it isn't currently live ig ??

[frank socket]

can anyone help me? so I can ask you more questions?

[fading robin]

you got plus or teams ?

[fading robin]

buy teams plan it got more cap

[surreal cypress]

plus

[rigid musk]

Unfortunately you have to wait till available or using 3.5 or alternative like ms ai

[rigid musk]

Team plan required 2 people no ?

[fading robin]

yea it may be a some planned feature as you see it split that it use gpt-4

[fading robin]

just create 2 accounts

[rigid musk]

Yeah but you have to pay twice ?

[fading robin]

yea

[frank socket]

bruh

[surreal cypress]

interesting, lol got access to a beta i never wanted to be apart of 😂

[rigid musk]

Api could be a cheaper alternative at that point I guess

[fading robin]

or use gpt-4 API h

[fading robin]

yea it seems so

[frank socket]

I think I'm going to ask for a refund...

[rigid musk]

You can , you have to go on help.openai.com . Bottom right for the chat and open a ticket request

[surreal cypress]

it is a bummer that when your purchasing the plus membership its doesnt clearly state the ussage caps

[rigid musk]

Usage cap depend on general capacity.

[surreal cypress]

to me its always seemed like they did it like that so they could change it or apply it differently to different situations and dont know how to articulate it in a way that wont negatively effect paying user count

[fading robin]

well that is understandble for me as running gpt-4 is very very expensive ;p

[split coyote]

What’s going on with the copy and paste on the mobile app? Seems to copy the entire section in one paragraph instead of multiple sections it created

[rocky rover]

wdym

[nocturne geyser]

yeah, use creative or precise

use balanced if you want 3.5
but you should never use 3.5

[shadow totem]

Adding onto what was shared here but posting where it was suggested, #gpt-models message , I'm experiencing the same. Mid-prompt it stopped, could not regenerate the response or type new prompts in any chat. I can get to this page, https://platform.openai.com/apps, but the page just spins when trying to enter ChatGPT.

[warm citrus]

Is it possible to feed data independent. For example you want to save the chat history so the chatbot knows what was writen before but with a different file it has all the data about a certain topic. i was not successful with having multiple files in "messages" but is there a way to do it? I am using gpt 3.5 turbo

[trail cairn]

api is for such tasks

[versed swift]

Yea, i spent $6 in one single day for a issue i had.. api is good only when you use it very rare so you don't need to buy a subscription .. there are pro and cons, but in the last month i used it 2 or 3 times

[gritty anvil]

@timber shadow do you still have that Custom GPT that's Holodeck style we talked about in November of last year? Is that the same on to help put your questions into one good prompt?

[strong bough]

Hey. Is there a Ai image editing service/tool that can modify images per specific requests?

Like seeing what a house would look like with a different color window shutters or something with a photo to edit?

Example. This didn’t seem to do it

[placid pawn]

Anyone using SQL Agents or database agents in production?

[real badge]

How can we provide reference images to replace the image mask part instead of describing it in the prompt?
response = client.images.edit(
image=our_image_in_bytes,
mask=our_masked_image_in_bytes,
prompt=prompt,
n=N,
size="1024x1024"
)

For example I want to replace model's outfit with another outfit, is it possible to do it with dall e?

[rocky rover]

you have to provide the original image and the masked image with the prompt to the API endpoint

[real badge]

Yea I know but I want to replace the masked part with the image which I want to replace not the one which is generated by AI, so is it possible?

[rocky rover]

Don't think that's possible

https://platform.openai.com/docs/api-reference/images/createEdit

[timber shadow]

Hey! I think you might be thinking of someone else, sorry 🙂

[gritty anvil]

Your good, I knew once I started searching lol

[wise lodge]

Hi all! I'd like to publish a GPT that my team has been using internally but which I think could be useful to people outside the team. I've set up the txt records but when I try to publish I still get an error saying I need to verify my domain. Any ideas how to troubleshoot?

[wise lodge]

When I click the link it takes me to the instructions for setting up the TXT record, which I've already followed.

[trail cairn]

does your domain clearly show you own the violating name/content?

[wise lodge]

How do I tell?

[trail cairn]

well its detecting some registered trademark in the name or other content somewhere. is the gpt about something you own?

[wise lodge]

Ah ok, my company, Wooden Robot is a NetSuite consulting firm. My GPT is a vectorized index of the NetSuite documentation which helps my team look things up on the documentation website. Wooden Robot owns it's domain name and our team account / GPT, but not NetSuite itself.

[trail cairn]

that's the issue then. i've seen people try to do similar for various things

[wise lodge]

I thought it would be helpful for the NetSuite user community to have a vectorized semantic search tool, but it makes sense that OpenAI wouldn't want to be accused of a GPT using someone else's data

My GPT itself doesn't have any copyrighted material in it except for the name and the instruction to only direct people to the official documentation website

[trail cairn]

yeah as in the end its supposed to be a store with upcoming profit sharing

[wise lodge]

Does sharing with the link work? https://chat.openai.com/g/g-4XR5tucp2-net-suite-assistant

[trail cairn]

cant test it now but previously it didnt work if the name has trademarks

[wise lodge]

Ah ha!

I put a space in and it worked

but still not ideal, as it's against the rules and all

Looks like it's only the name that matters at the moment

[trail cairn]

yeah

[wise lodge]

What's weird is, there are several GPTs showing up in the store with NetSuite in their name, none of them officially owned by NetSuite, so how are they doing it?

[fading bough]

Hey Everyone! Hope Everyone is doing well! Im hoping to get some help if possible. I went to make changes to my instructions for my gpt and when I went to save it with the Update button to publish to the store, it got stuck in loading. I uninstalled the app, cleared cache and cookies, and still no luck.

My GPT is not appering in the GPT Store anymore due to this issue.

[wise lodge]

Refresh the page, sometimes the editor gets stuck

sometimes I close browser windows and start a new session just to make sure

Similarly though, I'm having a problem where when I save, sometimes it loses its ability to use the GPT Actions I gave it until I edit and save again

[steel topaz]

I have plus membership and when I send prompt it responds but in the middle of responding it stop and ask me to regenerate the prompt. And it keeps doing this repeatedly.
I try deferent web browser but no luck. anyone here can help.

[fading bough]

Hey Kris. I have tried all that you mention. I logged out and in, I uninstalled the app and reinstalled, i cleared my entire browsing history, cache and cookies and i eben tried it on different browsers but it just stuck in loading. When I inspected the buttons the html code read "disabled". I literally saved it and then went to save it the second time and then it got stuck. I sent open a supoort a message and just waiting to hear back. As I was told and from what i read, they take a long time to respond. It sucks because i was ranked #17 in productivity and now this is killing my rank. Its a really good gpt and i received all positive reviews on it from users in my inbox. Hoping to get this rectified soon.

[rough knot]

Can someone direct me to how I can actually use the API? I was told we need to repay for tokens now - is that still the case if we have the paid subscription?

[timber shadow]

Hey! Do you mean you have a ChatGPT Plus/Team subscription? If so: API usage/billing is completely separate from ChatGPT billing. To use the API, you'll need to set up payment details & add credits at platform.openai.com, and you can get started with using the API from there!

[wise lodge]

Good luck!

[craggy ferry]

is there a way to upload images to gpt-4 in a hippa compliant manner?

maybe through azure or something?

[fading badger]

Hi all. Is there any way I can improve the time taken to invoke the function in function calling in assistant API?
I use GPT-4 and it takes around 3 seconds for the model to think and invoke the function.

[wraith zodiac]

hi, i have started to keep getting this :
Our systems have detected unusual activity from your system. Please try again later.
is there any information as to what the issue is?

[wraith zodiac]

anyone getting completely unrelated responses to their prompts?

maybe it was the gpts i was using..

[broken cedar]

guys there is literally problem with billing, I added 400$ to my api, My api access is suspented due the payment failure of 80$, I can clearly see my balance but chatgpt doesn't gets money from there it needs additional payment. Can someone help me from mods please?

[timber shadow]

Hey, for billing issues, your best bet will be to reach out to OpenAI at help.openai.com. Open the chat widget in the bottom-right corner of that page, start a chat, and click through the most relevant options. You'll eventually find a path that lets you type in your own message. This will submit a ticket. It won't connect you with someone live, but you'll be notified when you receive a response.

[broken cedar]

I did it they doesn't answers

[timber shadow]

It might take some time, possibly days.

[broken cedar]

its unprofessional behaviour by openais side

[maiden carbon]

Is ChatGTP Down? It isnt loading my History or Generating responses.

[rigid musk]

Status.openai.com

[keen portal]

Same here

[proper aspen]

Can someone help me... I opened up my ChatGPT account today, and all of my GPT conversations are missing and for some reason it's like I'm not fully logged in. When I tried re-logging back in, everything was still missing.

[pulsar ibex]

what's going on?
i'm not using GPT-4 anyway

[rough iron]

Status.openai.com

[night epoch]

I have the same error

[royal summit]

https://status.openai.com/

[spice locust]

I reached the limit from one google account. So I thought if I use different google account then it will resolve. But still same thing. How it detects that I am the samae person? Is it on machine?

[rapid surge]

https://status.openai.com/

[spice locust]

?

[spice locust]

.

[solar stream]

tell em to fix it

[rapid surge]

There is an outage. It will be written on every account

[narrow zephyr]

oh good! i thought it actually thought i was using gpt-4 (I am using gpt-3.5) but its just an error huh

[spice locust]

true

[teal hollow]

Any news on the error thing?

[rough iron]

just look at https://status.openai.com/
and wait for it to be solved, you can also subscribe so you get an email when they fix it

[fading robin]

Yea

oai got problems rn

[teal hollow]

The one day that I wanted to use it lol

[fading robin]

The error won't last a day so you still got time :)))

[polar furnace]

That’s strange. Everything seems to be working on mobile

[rugged wing]

You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again later. Learn more

wth is this?

[thick topaz]

is that fixed?

it looks like it is

[cyan jewel]

it is on chrome but not on mozila for me

[royal summit]

Yes! The issue is resolved 😁

[royal summit]

CTRL + F5

[cyan jewel]

You bet I tried

[royal summit]

Clear cookies and cache

or try in incognito browser

[formal jay]

at this point, i think id be screwed without AI >.<

[runic grotto]

can you guys suggest some better models which can be used on vision based system development

[fiery oxide]

chatgpt is lazy

[azure coyote]

why Creation failed when I create in midjourney alpha

[feral ermine]

do "regenerate response" counts as message ?

Edit: Yes

#community-help message

[azure coyote]

I upload image and write a prompt and than it says Creation failed , it works all day yesterday

only in alpha

[ivory fiber]

Hello everyone! Can someone explain to me why I am unable to convert the webp files that is created by the AI to PNG? I just get an error in whatever converter I am using
Thanks in advance

[rugged wing]

Hello, can anyone tell me if Chatgpt can process a 15 page site and display the complete content?

[manic fable]

Hi, I am looking t opurchase ChatGPT Plus subscription for writing articles. I wanted to know if there is any limit on the number of tokens per month. On OpenAI's website, there doesn't seem to be any such information.

[timber shadow]

Hey! With ChatGPT Plus, the limit isn't calculated by token or by month. Plus currently gives you a usage cap of 40 GPT-4 uses every 3 hours, more detail here: https://help.openai.com/en/articles/7102672-how-can-i-access-gpt-4
The article mentions it, but another thing to note is that Plus and ChatGPT billing and usage are separate from OpenAI's API billing/usage, which is calculated by cost per token.

[timber shadow]

Hey! If you ask ChatGPT to search a site by URL, it's not going to contextualize the whole thing unless very short. Instead, it's going to semantically search through the site for content relevant to your query and then return with info on that. If you copy/paste the content of the site manually, you could theoretically get it all into the context of the model, so long as the content is under the 32K token context limit of GPT-4 on Plus.

[rugged wing]

Thanks. Why is it not possible to link other GPTs in a gpt ?

[timber shadow]

What do you mean by link other GPTs? You can use the @mention feature in any chat to use any GPT at any time!

[rugged wing]

No, this is only possible in the default gpt. It is not possible to link GPTSs in GPTs. strange

[timber shadow]

Seems to work anywhere for me, here's screenshots of me starting in the Wolfram GPT, then @mentioning the Cosmic Dream GPT:

[rugged wing]

hmm strange. Do you know maybe how to fix it ? Thanks

[timber shadow]

Are you using the web or app version right now? Just did a bit of testing on my device and it seems to be popping up consistently in the Android app, but less consistently in the web version, especially in GPTs, which could explain what you're experiencing. There might be some kind of bug affecting the feature currently on the web? But that's a guess on my part.

[rugged wing]

I use the web version, yes

[timber shadow]

You might consider posting over in #1070006915414900886 if you feel like it! Something like "GPT @mention feature not working as expected" maybe? Just be sure to follow the bug report format: https://discord.com/channels/974519864045756446/1202282235877916683 This will help identify what could be causing the inconsistency!

[royal summit]

Since yesterday's outage, gpt3.5 only gives stupid answers and everything from yesterday is forgotten...

[regal orbit]

Need help my friend had his access to his account to GPT-4 limited by OpenAI even though he's on plus subscription. He's reported the case to openai. Then what need to be done?

[timber shadow]

Hi, what do you mean by limited? Was the account banned, or are you talking about the usage cap of GPT-4?

[nova carbon]

Hey guy according to open ai docs you get 100Usd/month (usage) in tier 1 and as you can see i am on tier one

but my credit usage is only $48.12 and in order to get to next tier i need to spen 50Usd so does that mean i cant get to the next tier

or is there something the i am missing

[trail cairn]

buy credits worth of $3

[nova carbon]

thnx

[past tendon]

why am I only getting half of the response?

[rustic marlin]

it is beca-

could you elaborate?

[past tendon]

Idk it stops randomly

[thick meadow]

Why is it suddenly not able to read files I upload?

[rustic marlin]

it still is

[nimble osprey]

That happened to me yesterday as well. After about 30 minutes, everything came back and was available. However, today, I can literally start a prompt that says "say hello" and I get the "Hmm...something seems to have gone wrong." error.

[bitter adder]

One message removed from a suspended account.

[frank heart]

I pushed a bug to prod 😢

I refused to have GPT look at the code this time, I said to myself, I can do this all by my self...

then everything crashed

then I was like

I don't want to be allllll by my sellffffffff

don't want to beeee........ alll by myyyyyyyselfffffffff!

[formal jay]

For whatever reason, this no longer recognizes any of my custom parameters I gave it. Its like suddently, ChatGPT forgot it knew me

[royal wren]

Can anyone suggest some tips to improve response speed of GPT4?

[rugged wing]

How can I see if I have the new update ?

[royal summit]

True! I have this too...

[half crypt]

Hi there.
I have just signed up for the credit version of openai account, and have linked the API to talking minds.
I put $20 credit on there, but dont have access to gpt4 only 3.
Any idea on how to access it?

also how do i access gpt4 32k?

Or does the above need the $20 month subscription?
Happy to pay, I just can't find a lot of info out there

bare in mind, that i am just using the api through talkingminds, and not directly through the openai site

[timber shadow]

You don't need ChatGPT Plus to be able to use GPT-4 via the API -- ChatGPT and the API are completely separate when it comes to billing and usage. https://help.openai.com/en/articles/7102672-how-can-i-access-gpt-4
To clarify: did you add $20 of credit on platform.openai.com? If so, is the 3rd party service telling you you don't have access to GPT-4, or is it not showing up on the Playground of the API either? https://platform.openai.com/playground

[half crypt]

thats a relief.
I selected GPT 4 and asked it's version.
said it's based on 3.
Although looking at the internet this seems to be a common response, so I think im actually good.
Thank for the fast response and clarification on the API usage requirements though.

[trail cairn]

never ask gpt about itself as its training data does not include valid information

[trail cairn]

its not a public model and its being deprecated

[tawdry light]

Help. How do I transform my conversations.json into PDF files? (1 pdf per conversation)? Any clue is appreciated.

[trail cairn]

you would need to code your own app for that

[tawdry light]

Cool.. i need to learn python

[tawdry light]

I might just cut the file in two then convert each file to xml sheets

Its a lot of text tho 208mb

I might just learn python

[loud pond]

So if i need for example my own chatgpt that focuses on fantasy writing (for dnd), what would be the best promp collection?

[analog locust]

How do you get chatgpt to read a pdf on the web? I use to be able to just specify the URL of the PDF that I had in my browser but now it just complains

Just comes back with the following response:
It appears that I'm currently unable to directly access and view PDF documents from the web.

[tawdry light]

Ive been trying as well. No use.

[rigid musk]

Some website aren't whitelist to avoid copyright, if you have access those pdf try to download and import in the chat instead of url

[arctic holly]

Hi guys since open ai has openly admitted to chatgpt 4 being underwhelming and getting worse are people still being modded here for talking about the censorship and degradation of GPT 4 getting worse over time

[trail cairn]

people have not been timed out or banned for saying those, more likely for using foul language, bad behavior etc. along within same messages. you can find plenty of complaining messages for each month past year or more with the search

[open gulch]

Hello ! Please when did OpenAi admit it ? Is there any link for that ? I’m just curious, thank you

[timber shadow]

I would call it more of an acknowledgement than an admission, but it's mentioned in the table on this page https://platform.openai.com/docs/models/gpt-4-turbo-and-gpt-4 next to the entry for gpt-4-0125-preview:

GPT-4 Turbo preview model intended to reduce cases of “laziness” where the model doesn’t complete a task.

[open gulch]

But can we use GPT 4 Turbo without going through the API ?

[timber shadow]

Turbo has been the version of GPT-4 on Plus since DevDay in November, there have just been various versions of Turbo since then.

[arctic holly]

Not you radium you have been good people

[open gulch]

What are the differences between the various versions?

Also, is that true that every week OpenAI updates gpt4?

[arctic holly]

If I link you I’ll get modded you can’t link articles here so I recommend googling

[timber shadow]

That table in the link above lists the version update notes as well, in the Description column! I don't know what kind of updates might happen weekly to the model, but I think there have been 3 major versions of Turbo since DevDay, again according to that earlier link.

The knowledge cutoff date updated from Apr 23 to Dec 23

[trail cairn]

ive seen plenty of people complaining here about it over a year without getting any warnings or bans, even i have complained about it, eg. the laziness with code generation

[arctic holly]

You can Google Sam has talked about the new turbo or gpt 5. He has also discussed about being disappointed with GPT 4

[trail cairn]

you can check the transcript of sam altman talk in lex fridman podcast which many sites refer to

[peak smelt]

I use an assistant an add functions to it to allow the assistant to search the web or pull a pdf version of a web page or text from a web page. It will automatically figure out when the funcitons need to be called to process a given prompt.

[jolly monolith]

Can't use the premium version after i downloaded and used a vpn, uninstalled, rebooted, seems to work on another devices, if someone has an idea on how to solve this issue please advise

[open gulch]

Thank you for the informations. Where can I find these informations on my own?

[open gulch]

Any info about gpt 5?

[timber shadow]

Same link, same table! "Training Data" column: https://platform.openai.com/docs/models/gpt-4-turbo-and-gpt-4

[open gulch]

Thank you

[nocturne geyser]

there's not really a way to do this unfortunately

the nature of GPT means it's hella slow

[slow ibex]

you'll overdose on dopemine, calm down sir!

[royal wren]

Not even for the API?

[formal jay]

aww someones bored.

[cloud sedge]

#community-help: Is the public Open-AI community https://community.openai.com/ built and integrated with this one here?

[junior pivot]

Hello together, i have some issue with a customgpt today morning. Everytime i want to upload/update my data for the gpt i get a message "unable to upload ....txt". Every other file (in every possible file size and syntax) not working too. Did somebody have an idea what im doing wrong? Or is it a bug that concern the platform? No Internetproblems, different browsers, different devices ^^ ...

[timber shadow]

Hello! The OpenAI dev forum you've linked here is separate from this Discord, though there's definitely some folks that participate in both. That forum is more just devs talking about using OpenAI's API services, as I understand it -- probably some GPT building discussion too? I'm personally not on it much at all.

This server, while it has a set of channels for OpenAI development as well, seems to be more populated and used by ChatGPT users, DALL·E users, etc. Again though, lots of overlap here! Definitely plenty of devs in this server too.

[timber shadow]

Hey! How many files do you currently have uploaded to the GPT?

[junior pivot]

1 zip file with 1400 txt files. A obsidian vault. ~130000 Token. 2,4 mb

[timber shadow]

As a test, can you try creating a new GPT, then trying to upload the problematic text file on its own? Just to see if it could be something with the GPT, or with the file.

[pine nacelle]

Anyone seeing these errors in GPT4 (Plus) - browsing has been very inconsistent

[toxic dome]

Anyone having this issue on iPhone?

[fading robin]

Not on iphone but allways on huawei

[remote fossil]

hello

why i'm getting Configuration is not a constructor

[dusky axle]

Hi all. Can anyone tell me why my pro subscription has been so painfully slow to answer using GPT4 lately? (like more often than not over the past few weeks)

[fading robin]

Because OAI servers are sometimes overloaded that will slow down the gpt-4

[pure pulsar]

Anyone know what settings are needed to enable ChatGPT login via Google for Workspace Education Plus accounts? I've given Limted access to the verified OpenAI App but we're still getting blocked. "Unable to create account with that information" error.

[frosty rose]

i have no idea how else i might have signed up except user/password. Any ideas how i can sign in?

[storm island]

Hey I am on a Pixel 2 android device. And it says the ChatGPT app is not compatible with this device. It was compatible before. What shall I do. I have been on LineageOS with it working fine.

What shall I do to fix it?

[storm island]

Same issue on Android. Then when I try and reinstall it, it says it's not compatible with my device.

[storm island]

Same here.

[ocean crypt]

Helli @frosty rose
Nice to meet you.
I can help you

[azure stump]

My session keeps expiring, any reason for that?

I never had to login before

But now all the time

[rustic marlin]

consider changing your password

[azure stump]

But I only use OAuth

[rustic marlin]

sessions do end after a few weeks, it is normal

how much time intervals are your sessions ending?

[azure stump]

No, it's an every day thing or even more than once a day

[rustic marlin]

I would change my password just in case

[azure stump]

Ok, thanks, it's a bit annoying

Hope that fixes it

[rustic marlin]

im really not sure of why this is happening with you, but it is surely not normal behaviour of the page

are you using extensions that change the page's behaviour?

[azure stump]

Yeah

Also using Harpa

[rustic marlin]

try disabling them just for a day and check if it solves the problem

[ocean crypt]

Hello @rugged wing

Nice to meet you

I can help you 🙂

[rugged wing]

You just joined...

Does anyone have an idea when you try to open the openai verification email but it does not work? I feel like I try to use many different method or ways even using a private browser

[storm island]

My previous issue with the ChatGPT app on android is fixed BTW. Probably an issue on their end and they fixed it.

[cold folio]

What was the issue?

[storm island]

In the Google play store it said the app was incompatible with my device.

[rugged wing]

ok nvm I got it to work on my previous issue

[cold folio]

Ohh nice to see it's fixed 🙂

[proper plank]

I opened my openAI under a work email that's no longer active .. then I added an email to that account that is active .. now that 2nd email is associated with a Microsoft for Startups account that gets me all kinds of azure and openAI stuff .. how can I associate that account to my openAI .. does that make sense or is it word salad?

[quartz panther]

we have a ChatGPT Team Subscription but for some reason have of our users cannot accept their invites.

they get the invite email, and once the click on it , it takes them to the logged out chatgpt screen. logging in does not add them to the Team workspace.

also - it looks like we cannot invite users with a + in their email address

[quartz panther]

yes, we have the same issue with our teams subscription. half of our invitees cannot join

[rugged wing]

Yeah I got it to work myself

[quartz panther]

what did you end up doing?

[rugged wing]

Just copy paste the link to a private browser.

[quartz panther]

nah. that didn't work for us. tried multiple times

[rugged wing]

damn maybe contact openai support then regarding this

[pine flame]

Im trying to get chatGPT to do some tedious work on my project but it keeps limiting the file reading to like 1000-1500 characters... and its cutting the scripts Im providing up removing any context, so how the hell do i fix that?

[bronze charm]

This usually happens when ChatGPT is given a non-document file like C source files. I have mitigated the issue by pasting the contents of the file into the prompt message.

[inland bough]

GPT is Unable to load history, what can be the issue?

[frosty galleon]

Unable to load history and chatgpt seems to be not working I get: “something went wrong” after typing a message

Anyone else?

[left spire]

Always happens to me when I switch a VPN on.

When you switch off you often have to reload.

[cold folio]

I would disable the VPN

Why do you use it in first place ?

[hearty forge]

Maybe this is better suited in here:

Every time I try to contact support, they respond asking for details, but before I can even send those details, the ticket is marked as solved/closed since it then allows me to download the chat transcript for the ticket.

Why would support close the ticket straight after they asked me to provide them with certain info? 🤔

I genuinely don't know where else I can go to get in contact with support. What can I do since the Help section of the website just doesn't work properly for me?\

(If you reply, can you please ping me? I may not see it otherwise!)

[hoary bolt]

Some of us have mandatory VPNs at work

[rigid musk]

Unfortunately cloudflare blocking some ip range. Openai working with some vpn so make more accessible. But some office have to make exceptions too

[burnt sorrel]

Im working on fine tuning a gpt3.5 model to work with out Oracle DB schema, I understand I cant just train it on tables like I can with a RAG but what's the actual best way to add knowledge to the LLM about our schema? is it questions and SQL answers?
I'd like to get a model that will take NL questions and return a valid SQL answer.
I know this is a classic RAG job and im exploring that avenue at the same time.

[sturdy tide]

I cant use chat gpt on a specific account and it doesnt respond to me or load history i logged into a different account and it worked perfectly fine

[vale bronze]

Hi, what is the max message cap in Plus plan per hour/day? 🤔 Like to know. Thanks!

[rigid musk]

40 prompt/3h

[proper plank]

I opened my openAI under a work email that's no longer active .. then I added an email to that account that is active .. now that 2nd email is associated with a Microsoft for Startups account that gets me all kinds of azure and openAI stuff .. how can I associate that account to my openAI .. does that make sense or is it word salad?

[trail cairn]

the issue should have been fixed with the first account before making more accounts. now you have to contact support but it can take 1-2 months to resolve two company issue

[orchid raven]

Any problems on the api for transcriptions? (whisper-1) ?

[proper plank]

support told me to do it

[proper plank]

and now I can't get anyone to answer me about it. across any platform .. perplexity gave me MUCH better answers but I still don't know wth is going on

[arctic holly]

Where is gpt 4 turbo. When they say paid users do they mean plus are they rolling it out?

[bronze charm]

Should be widely available by now. Paid users refers to anyone to has a Plus, Team, or Enterprise account.

To clarify, GPT-4 Turbo has been in ChatGPT since November 7th, 2023; gpt-4-turbo-2024-04-09 is the model that was recently released.

[elfin hinge]

So.. why?

Why .0 .1 .2 instead of 10 11 12?

[rugged wing]

Can anyone tell me how I can look up, if I have the new update?

[livid cypress]

Are there any resources on optimizing the assistant API results for voice? When I use the voice feature in the ChatGPT app, I seem to get sightly different answers that are more conversational.

[rigid musk]

Chatgpt isn't good with math and making list

[rigid musk]

Maybe in #dev-chat

[arctic holly]

Query so is turbo default gpt 4 I see only the option of 3.5 and 4

[bronze charm]

The UI does not say turbo, but it is turbo

[arctic holly]

Thank you

[civic girder]

hello. still don't have access to the memory feature. what is the issue?

[rigid musk]

Still rolling out

[civic girder]

ok, thank you!

[modern cipher]

I think posting it to #1070006915414900886 would be better because it's apparently a bug

[wind lintel]

can anyone point me to a good training on preparing a knowledge base for a custom gpt?

[rigid musk]

Yes or help.openai.com

[rugged wing]

yeah that would be another idea for that person

[leaden burrow]

I'm trying to provide a good prompt to gpt-4 turbo but I'm having problems limiting certain kinds of output. This is my prompt:

Disregard previous context. Provide a plot synopsis of the following text. Do not embellish or provide your own interpretations of the text. Your plot synopsis cannot exceed 400 words. Begin your synopsis by briefly describing what happened previously and the current location of the characters. End your synopsis by including the last dialogue spoken by characters (if the text contains dialogue) and final description (if the text contains such a description). The prompt for your synopsis will have the following template:
What happened previously:
{description of previous events}
Text to be used in the synopsis:
{Transcript of what happened after the previous events}

The issue I'm running into is gpt using phrases like "the episode starts", "the scene ends", "the last line of dialogue is...".

It also adds unwanted output like "this shows..." at the end of the synopsis

[modern cipher]

or maybe for your "friend" as well, if you want them to start using ChatGPT... (just thought so per your display name)

[modern cipher]

Adding something like "make sure you output the plot synopsis itself, and only it" may improve the output

[leaden burrow]

Thanks, I have considered giving it an example of what good output looks like. I don't know if that might help.

[modern cipher]

I think the fastest way to check is to test your ideas

Testing on ChatGPT is a good start point, if you don't prefer consuming your API cost

[leaden burrow]

I've been testing a bit with ollama, though its not ideal, since open chat isn't as suited to the task

Oh, yeah I could test with chat gpt

[formal jay]

is everyone elses ChatGPT just hanging?

or just me LOL

[old arrow]

yes im experiencing the same

[rugged wing]

What do you mean?

[pale mortar]

I have been having that for days no matter what I do

contacted support and it didn't help.

[weary turret]

Hello! I am sending ChatGPT 4 a text file of a novel with instruction to fix faulty line jumps. It keeps returning error after error - seemingly as soon as the file gets too big? Has anyone encountered issues like this before and knows a fix/workaround?

[fresh edge]

so, we pay 20$/month to use it for 40 question/day? wow

[timber shadow]

Hi, it's 40 every 3 hours currently, not every day. It's about 1 use every 4.5 minutes.

[fresh edge]

but once you finish your questions, you're done for the day or so

i just got the message that i reached the cap, and now they're making me wait 1h30

[timber shadow]

That just means you hit your cap in the 3hr window. It's a rolling cap. So if you use 40 uses in 1.5 hours, you'll have to wait 1.5 hours to get a use "back." If you limit your use to 1 per 4.5 mins, you'll only hit the limit if they have to throttle the limit due to heavy traffic. You're not done for the day though! Just for the 3hr usage cap window.

[fresh edge]

this is horrible

especially since it seems to consider the "rerolls" as questions too

[timber shadow]

Any use of GPT-4 counts as one use toward the usage cap, yes. It's an unfortunate necessity for a computationally expensive model like GPT-4! If you're interested, you can look into using GPT-4 via the API on a pay-as-you-go model.

[fading robin]

wonder if peoples want new gpt or unlimited gpt-4 cus they need to chose one... just a tough i was having :p

[fresh edge]

id be fine with unlimited 4

[fresh edge]

that's actually what i was using it for! im trying to integrate the api in a school project

[fading robin]

I think waiting for new model is better option :) cus gpt-4.5 will be bettet and faster potentialy unlimited

[fresh edge]

but guess i'll have to wait one more hour to get to the next step of it...

[fresh edge]

i don't understand why you'd have to chose. all companies work on multiple projects at once, therefore they can easily allocate resources both for improving gpt4 use, as well as for r&d of 4.5

[fading robin]

Lets say you got 100 gpus training need 80 and to provide peoples with api you need to give 10 and for unlimited responses you need 20 but you lack gpus you need more but you got a problem there is no gpus on market that you need so you stuck :/

Its just internal problem with OAI, Nvidia and TSMC

Google for example make ai chips by them self so they don't care

But oai need to buy them from nvidia that got problems making them because tsmc got problems

[fresh edge]

something that openai could easily do as well tho

[fading robin]

Google worked years for thier ai chip

Oai may be trying to do that too

[fresh edge]

but it's not patented, so anyone can use it afaik

[fading robin]

Future is bright cus B200 new nvidia gpus are cheaper and better (per tflop) and ther are for example gaudi 3 that is similar to H100 so we will got more computation power too

[fading robin]

Designing silicon is hard....

And makking hardware and software for that

Now sorry ima go sleep

😴

[modern cipher]

Because your display name is apparently about friends or "友達"... I just had an imagination that your name refers to a friend that is important to you, and s/he might have the same verification issue. I thought it might make the bug report helpful for him/her.

[dawn bridge]

Does anyone know how to get the tts in chatgpt to work? nothing happens when i click the speaker, it looks like its about to load it and then it does nothing, flashing the speaker only for a second. In Edge

[rugged wing]

lol I could not change my nickname here.

[modern cipher]

Ah OK

[rugged wing]

It has completely nothing to do with that...

[worn moss]

Hi, I'm currently working on a project using SAPUI5, and I need to integrate models from SAP AI Core, like GPT-3 5-Turbo, GPT-16K, and GPT-4 32K. I already have the model configurations set up, but I'm not sure how to select a specific model for testing using a tool. I'll provide a link to the tool I'm planning to use. I'm seeking guidance on how to effectively utilize these models within my SAPUI5 project.

  let parsedPrompt = '';
  if (prompt) {
    const { prompt: question } = prompt;
    parsedPrompt = createPrompt({ ...enrichedPrompt, question });
    console.log(':rocket: ~ ChatPromptService ~ sendPromptToAiCore ~ parsedPrompt:', parsedPrompt);
  } else {
    parsedPrompt = enrichedPrompt;
  }

  const destinationDetailsAI = destinationDetailsObj;

  const modelConfiguration = destinationDetailsAI?.originalProperties.destinationConfiguration;

  const { 'gpt-35-turbo': turbo, 'gpt-4': gpt4, 'gpt-4-32k': gpt432k } = modelConfiguration || {};
  const chatModelDeploymentId = turbo || gpt4 || gpt432k;
  const { 'api-version': apiVersion, 'ai-resource-group': aiResourceGroup } = modelConfiguration || {};

 
      return response.data.choices[0].message.content;
    }

    return 'No proper destination configured for AI Core Chat Model Service';
  } catch (error) {
    console.error('Error sending prompt to AI Core Service:', error);
    throw error;
  }
};

  return result;
};```

[reef magnet]

I just buy the plus and in the plus i allways get "Hmm...something seems to have gone wrong." what ever i ask the chat. i restart my modem. i dont have any extensions on my browser and still get this msg

[sharp sparrow]

I'm on plus from the beginning. I started receiving this error on GPT4 by yesterday. Only 3.5 seems to work.

[iron canyon]

yea chatgtp is broken

[rigid musk]

Create new chat. And make sure your ci is turn off

[iron canyon]

i did that 3 times

cant send link here and its too big for discord but if you want, i can send it in pm and you can see for yourself

[odd heath]

how many pages/file size?

[weary turret]

Hello! The whole .txt file is 84Ko and I tried going as low as 20Ko

The 20Ko one is 3000 words

I was doing that three days ago and 80Ko files were going through just fine so I'm not sure what's up now

[opal portal]

Every time I try to query exactly CHATGPT 4 I get this error, at the same time with GPT3.5 there is no problem at all

Does anyone know how to fix this?

[dense lantern]

I am getting this message when i access assistant using API:
It appears that you meant to upload a document containing data for analysis, but it hasn't been uploaded yet. Please upload the document

[ocean rampart]

I'm seeing the same thing myself.

[opal portal]

Did you find a way to fix it?

[ocean rampart]

I have not. I signed out and back in to no avail. It looks like a number of people are having the same issue, but I don't see anything on the OpenAI Status page about it...

[fading robin]

You can try some of this solutions cus some of them helped peoples
clear webborwser cache
turn off vpn
try on firefox
for me gpt-4 works good so idk wahts wrong byt mayby this will help

[opal portal]

nah, it doesn't work on the phone either

[opal portal]

so I don't think that's the problem

[fading robin]

¯_(ツ)_/¯

it up and running for me

in EU

[leaden burrow]

    "Title: " + df.Summary.str.strip() + "; Content: " + df.Text.str.strip()
)```
When combining data from a dataframe for use in openai's vector database, is it inadvisable to use labels other than "Title" or "Content"?

I have different column names I'd like to use, and more than 2

[wide python]

chatgpt-4 is giving responses completely unrelated topics as if its responses meant for someone else. Anyone else experiencing this?

[compact remnant]

with assistants, can you add new functions to them without having to create a new assistant?

[leaden burrow]

Could you show me your prompt?

I'm new to prompt engineering but I'll do my best to help

[rugged wing]

When I try to send a message it makes me a load that does not stop and I deleted all my chats and its re makes the bug

[young violet]

sorry may i get support here? there is some bug on my account

i cant use my GPT4.0 but 3.5 works

everytime i click regenerate there will be a new Chat been made

like that

I tried different IP and tried application and website

still not working

[rigid flicker]

Hi, I just logged in and all of my memories (except my birthday??) are gone for some reason. I didn't clear them out. Any ideas?

[broken salmon]

Same!

[young violet]

is there a staff could help us?

damn

[rugged wing]

lol

[hasty sapphire]

Hi Folks, since yesterday my CgatGpt 4 is giving me this error:"Hmm...something seems to have gone wrong." The 3.5 works fine. Any clue?

[hasty sapphire]

Same here since yesterday.

[ocean rampart]

OpenAI says all systems are operational – but we're all having the same issue. 🙄

[woven folio]

Same thing for me, joined discord to look for an answer 🤣

If I upload an image first it does work 🙄

[hasty sapphire]

Exactly, clearly a problem there. Hope someone will answer since we pay for this thing to use 😄

[livid narwhal]

I had it today as well, just leave it be for 15 minutes and it works just fine once again now.

Haven't run into the issue since!

[livid narwhal]

You need to refresh the website, usually that will do the trick.

[rugged wing]

didn’t work

[livid narwhal]

Have you given it 15-20 minutes before trying again?

[rugged wing]

Yes

[livid narwhal]

Is it GPT 4 or 3.5?

[rugged wing]

3.5

[livid narwhal]

Have you tried clearing the cookies?

[rugged wing]

yes

all

[livid narwhal]

Or using a different browser temporarily?

[rugged wing]

only chrome

[livid narwhal]

Try a different browser

[rugged wing]

it’s work with my phone but not in web

[livid narwhal]

Do you use a different browser on your phone?

And try logging out and then logging back in, see if that helps.

[ocean rampart]

I signed in to ChatGPT using an Incognito window and still got the same error. The API is returning an HTTP 500 error, and I got a different error message the first time around:

[woven folio]

I've tried phone and desktop chrome, incognito mode, and cleared cache.

[livid narwhal]

Clear cookies and try logging in and logging back out.

[livid narwhal]

I'm assuming you've refreshed, so try logging out and back in.

[woven folio]

Tried logging in and out many times

[livid narwhal]

Cookies?

[hasty sapphire]

Tried all as usual, 3.5 works, 4 not. Even changing browser or phone. Same error.

[hasty sapphire]

Seriously?

[livid narwhal]

Have you tried logging out and back in?

[hasty sapphire]

Yes. Once.

[livid narwhal]

Clearing cache / cookies too?

[hasty sapphire]

Yep.

[livid narwhal]

Then it could straight up be an issue with your login / account and I suggest contacting support

[hasty sapphire]

from when can I contact them?
Used the chat support but there is only a bot answering. So annoying.

[woven folio]

Same thing for me

Still find it weird it'll work if I upload something first

[hasty sapphire]

The ChatGpt support bot says:"If you are receiving failed responses (e.g. network errors, try again issues, slow responses, or error occurred) try waiting a while and retrying your request again later. We appreciate your patience as we scale up our services to meet unprecedented demand!"

[hasty sapphire]

You are right. I tried to upload an image an it worked.

You just gave as a nice hack. Thanks!!! Hopefully it will back to work normally soon! @woven folio

[woven folio]

I just payed for my plus and can't use it, kinda sucks 🤣

[woven folio]

Btw it is using my credits 😭 have to wait now to try again. A bit disappointing

[proper plank]

How do I get help with my account?

[pine flame]

and its down

[vagrant temple]

I need help...

This morning I woke up to 3 emails..

1. ChatGPT access revoked
2. ChatGPT API Access revoked.
3. Refund Processeed (just for my API Credits... no refund on the Plus account I pay for).

I have no idea what I have done. My usecases for ChatGPT are very tame. My only theory is that I recently started using a new browser, and I believe I accessed via a new VPN connection.

I have gone through the process with the OpenAI Help centre and get something but unhelpful bot responses.

I still have no idea what I've done wrong, and whether I'm allowed to setup a new account (which I had to do to access this discord).

Can I use GPT again? Will I be banned again? What are my appeal options or support options because I am completely stuck

[vagrant temple]

Honestly... This whole process has been a nightmare.

Banned without explanation, help centre that won't let you even access the required support.

This is probably the push to get me off ChatGPT on moving all my work to different platforms.

OpenAI got too big, too quickly, and they've thrown me (a paying customer) down the drain without any reason or support options

[vagrant temple]

So all my CustomGPT stuff is gone?

And seriously... I've been banned for a some kind of violation... and there is no way for me to find out what is was?

They just took another month of Plus payment from me this week... that's just dust too?

They refunded my API credits... but not the Plus money

Does somebody in this discord work for OpenAI? I would love for them to confirm this is how useless their process is

I'm doing charge backs on my card.

Bank can actually charge back months of payments... So works out better anyway.

[nocturne geyser]

contact your bank, do what you need to

[untold pulsar]

gpt4 still broken ?

[young violet]

it is almost a whole days. why there are no staff come and support us？ we paid for that

[limpid obsidian]

Hey is gpt4 not working?

[livid narwhal]

reach out to them through the OpenAI Help Center or by emailing [e-mail]

[livid narwhal]

GPT Staff in this Discord are working so your best bet is still to contact support.

[livid narwhal]

https://status.openai.com/

[livid narwhal]

^ see above

All functions are operational so try refreshing or logging in / out

[humble ocean]

well not for me..i still have on chatgpt 4 the thingy with "Hmm...something seems to have gone wrong." message, cleared my cache but still stuck on that

[sharp sparrow]

Same problem here. 48 hours now.

[sharp sparrow]

Same problem here. 48 hours now.

[sharp sparrow]

Yep

[humble ocean]

worst part is that status of ai says there are no reports and everything works fine.. i would love to prove them wrong 😛

[sharp sparrow]

I'm trying to insist with the support chat

[silver oasis]

Support chat was useless for me, I've had this issue for a couple of days but they just said because the status page says everything is okay, there is no issue (clearly not the case, so many people complaining here and on the forums)

[sharp sparrow]

I know. Same for me. But I think it does worth to insist.

[quiet horizon]

same here...broken gpt4 for my account, works for my colleage.

[silver oasis]

I have tried multiple times as well, same result

[trail cairn]

that may get you banned or flagged in stripe which is used by many services

[potent quiver]

doesnt work for me since 2 days ago but my friend uses it with no problem at all

[vagrant temple]

Great system OpenAI has... Ban accounts without giving a reason... don't give refunds.

[trail cairn]

you didnt get an email about suspension? most services dont refund on violations if the service has been used

[vagrant temple]

Need help... go to our Support Site...

"Thanks for the question, if it's in regards to accessing an account that has been banned, please submit from that account"... which is impossible, becuase the access was revoked without warning or reason.

[vagrant temple]

I got an email randomly this morning saying my account and API access are both revoked... no specific reason given, and I genuinely have no idea why... but only got a refund for some API credits... not my Plus payment

Not to mention I paid for plus like 3 days ago, build a bunch of customGPTs... and they're now all gone. Been using and paying for it for about 6 months

[trail cairn]

but the mail must have some reason like violating usage policy or tos?

[vagrant temple]

"We have determined that you are using OpenAI services in ways that violate our policies."

I have re-read these... still no idea what it thinks I did.

[trail cairn]

if its specific messages they dont tell it because people have used the info to learn how to evade bans

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

[vagrant temple]

Literally the only things I can think was

• I used on a new browser for like 15 minutes 2 days earlier with a VPN...
• I used the "photorealistic photos" customGPT from their homepage to try and generate 'a daschund chasing a bowl of spaghetti' and it said "unable as this may be a breach"

Otherwise I've basically done the same boring work related stuff day to day for months.

Tidy up emails... reword techincal jargen... extract text... etc...

[vagrant temple]

But I didn't do something wrong... unless all my work is gone because I tested a dog chasing pasta picture generation?

[trail cairn]

its not that customgpt one as dall-e often warns about things

[vagrant temple]

well I've hardly ever used Dall-e

Creates a random profile picture for a customgpt occasionally. Otherwise I think i've done maybe 10 images, all basic stuff... and MONTHS ago

[trail cairn]

you can appeal if the mail said its possible but you have to might have to fight with the bot to get a ticket created without account

[vagrant temple]

No you can't

[trail cairn]

its possible

[vagrant temple]

I can't log in a use the bot on the banned account...

How can I raise a ticket?

If you use a new account it says "good news, this account isn't banned.. if you're having an issue with a different account you need to log in to that account to submit an appeal"

[trail cairn]

afaik the option is just hard to get show up as you need to click multiple things

[vagrant temple]

@trail cairn All due respect mate... I've spent all day trying. It's NOT possible

There is NO means to do it... it's a cycle of "create account to raise ticket... get told you can't raise from said new account"

The system was designed by somebody in 5 minutes using GPT1... it's broken

[trail cairn]

others who have been banned have found it so you might use the search here or try more. there are often ban waves and there was one recently i think

its some generic option unrelated to bans

[limpid obsidian]

I tried everything logging out, using a new browser, hell even a new machine but gpt4 doesn't work but 3.5 does
Hmm...something seems to have gone wrong -> I keep getting this message again and again after everything that I have tried

[vagrant temple]

This isn't a solution... "Fumble your way through a broken system, until you find a menu option to appeal a ban that you shouldn't have received... By the way, your not getting a refund for a service you paid for".

[trail cairn]

its the only way there is other than just accepting the ban

[vagrant temple]

Ok... ban accepted... how do I get my refund?

See the problem?

[trail cairn]

are you in eu and less than 14days from first payment?

[vagrant temple]

No. I've been a loyal and paying ChatGPT Plus user for like 6 months.

Hence... extra pissed off about this situation

Several Hundred dollars for me... down the toilet because of their broken system.

[trail cairn]

"Payments are non-refundable, except where required by law. " 🤷

[vagrant temple]

This translate to "We will take your money, lock you out of your account without explanation... F*** you"

[trail cairn]

they refer to usage policies which apply to chatgpt and api

"We reserve the right to suspend or terminate your access to our Services or delete your account if we determine:

You breached these Terms or our Usage Policies.
We must do so to comply with the law.
Your use of our Services could cause risk or harm to OpenAI, our users, or anyone else."

[vagrant temple]

Again... not sure how much clearer I can make this... I DIDN'T BREACH THEIR POLICIES...

One more time for the deaf people...

DIDN'T DO IT

ffs dude... why do you feel you need to defend these people for a clearly broken system?

[trail cairn]

note that it might not only be recent activity but overall or something that happened over a month ago

[vagrant temple]

noted... still didn't do anything

I am not a power user... I use it for the most boring work tasks.

I'm not jailbreaking or trying to mess with things. I'm cleaning up emails and extracting text from documents. It's boring stuff

Again... why are you defending this broken system? You seriously don't understand why this is cooked?

[trail cairn]

i'm just trying to suggest what could be the reason for the ban. if your emails or documents contain personal info of others for example

[vagrant temple]

doubt it... maybe a random email address or phone number... nothing I don't have permission to use though

That's a bull**** reason if that's the case

Treated like I'm a criminal because I used OCR on a document to fill out some email templates? What a pathetic joke

[trail cairn]

did you do that with training enabled

[vagrant temple]

No.

100% disabled that

[trail cairn]

normally you only get a warning mail for violating usage policies if its minor violation

[vagrant temple]

Well I didn't get that... Maybe my account was accessed without my knowledge?

Would be nice to be able to login and check usage to confirm... but their broken automated systems lock me out completely... so that't not possible.

[proper plank]

I get nothing when I do that

[harsh niche]

I have a problem un gpt 4, it didn’t give any response for more than 40 hours. I have sent an email to the openai support but they haven’t replied yet. Anyone knows what I should do now?

[timber shadow]

Hi! What browser(s) is this happening in?

[harsh niche]

I have used different browsers and different devices. Browsers: Safari, Microsoft Edge, Firefox. Also I have tried the mobile app.

[thorny kernel]

can i ask for help here about using whisper for my python app?

[timber shadow]

I see! Have you tried different networks yet? If so, could you try:

• Disabling WiFi on your phone, connecting to cellular, then
• Logging in fresh in a private window?

[thorny kernel]

??

[timber shadow]

Probably #dev-chat or #gpt-realtime 🙂

[thorny kernel]

oh ok

thanks

[harsh niche]

Yes, I have tried at least 4 different WiFis and also tried mobile data.

[timber shadow]

When you tried the different networks, did you also try logging in in a private window, or at least clearing the browser cache? Sometimes bugs can persist across networks in the cache, so trying in a private window after switching networks is a good test, as that wouldn't use the potentially bugged cache.

[harsh niche]

I try it now using private window in Safari (iphone) and switch to mobile data but gives me the same error.

[timber shadow]

That looks like the same screenshot as before -- same timestamp, and the WiFi icon is there too, maybe attached the wrong one?

[magic steppe]

hey, i am 17 and currently in my high school preparing for a competative exam yk typical asian kid.
i wanna become a dev but i am very bad at following academics stuff idk what to do i am really confused with my carrier right now.
can anyone pls guide me?

[harsh niche]

Oh, sorry I have sent the wrong screenshot

[lament cloak]

Am having the same problem as @harsh niche . Second day not able to generate any responses. Cache cleared, changed network but still persists both on web and mobile app. any help, anybody pls.

[magic steppe]

try using vpn

or use ctrl + R in web see if it works

[lament cloak]

Still not working with VPN on. no responses. Thanks for the suggestion.

[lament cloak]

Found help; try uploading an image on chatgpt, then type ignore image, click enter. should start working for that chat.

[humble ocean]

wow, it rly works.

[runic cedar]

Hello everyone! I'm facing a persistent issue with the ChatGPT interface on my desktop. Despite trying multiple troubleshooting steps, the conversations search bar, is not displaying. Here are the details and what I've tried so far:

Browser: Chrome (also tested on other browsers)
Steps Already Tried:

Logged Out & In - Signed out of my account and back in.
Cache Clearing - Cleared browser cache and cookies.
Extensions Off - Disabled all browser extensions.
Zoom Adjustments - Tried zooming in and out.
Incognito Mode - Attempted accessing the site in an incognito window.
Different Browser - Tested on different browsers to rule out a browser-specific issue.

[heavy patio]

What Conversation Search bar?

[runic cedar]

This is from another account

[harsh niche]

It works! Thank you

[harsh niche]

Thank you for your suggestions

[timber shadow]

I'm glad someone got a working suggestion to you!! Hopefully just a temporary workaround -- hope it clears up for you soon.

[heavy patio]

I think thats a browser extension, not a built in feature OpenAI gives you? A quick google search tells me the browser extension is called "Superpower ChatGPT" but don't quote me on it.

[runic cedar]

I must have added this extension a year ago and disabled it to fix a bug at the time. I then forgot to turn it back on. Thank you ☺️

[runic cedar]

Now I know why I removed it— it lost conversations

[tawdry lion]

Wow, I have had the same problem, just saw your reply and it works, Thanks!

[ashen wasp]

Amazing, this works. Not sure if I should be happy now 🙂 I'm curious how Openai handles this situation ... having the paid feature down for almost two days now without any official response sounds wrong...

[calm idol]

Hello I am having issues with my Plus Account. Its not been able to give responses for the last 2 days. I have logged in and logged back out but this is what I am getting.

[lime juniper]

Hello I am having issues with my Plus Account. Its not been able to give responses for the last 3 days. I have logged in and logged back out but this is what I am getting.

[sudden wolf]

Guys, I'm doing https://platform.openai.com/docs/tutorials/web-qa-embeddings what is the most suitable function from openai python library? VSCode says distances_from_embeddings doesn't exist in openai

[stuck stone]

Hey with this new update to organization API keys, can I migrate my api keys from the old user keys to organization keys in their respective organizations?

[hasty sapphire]

Hi folks, anyone has some news about this annoying messagge error we are having lately?

[cerulean heron]

same issue

[exotic garden]

I also have this problem with my plus account. It's been about 36 hours. I can only access the GPT 3.5 model

[cerulean heron]

same here

[hasty sapphire]

The way to avoid the problem temporary is to open a chat, upload an image, than start to use chatgpt 4. After a while it can give the error again. Refresh the page, srat a new chat. Upload image... Hope they will fix this soon.

[exotic garden]

Wait what! that totally worked for GPT 4. Let's see how far I can take it

[hasty sapphire]

Yes I know, a guy yesterday suggested it here. It works for quite long. Some times gives back the error. Just refresh the page if happens.

[exotic garden]

It's better than nothing I can't access Dall-e or any of the other custom GPTs how about you?

[hasty sapphire]

yeah man, agree.

Not tried yet about the custom features. Using it just for braistorming

But they should fix it asap or refund 3 days of problems 😄

[exotic garden]

100% and support is basically non-existent. I've been waiting for an email for 24 hours.

[hasty sapphire]

agree. A disaster. Almost worst than Gemini 😄

[cerulean heron]

THIS WORKED! Thanks dude

[exotic garden]

😆 I mean it's great when it works but obviously this issue exists just acknowledge it Open AI

[dusky falcon]

Has anyone lost access to their billing information -- including payment information??

[heavy abyss]

from openai import OpenAI
from IMPORTANT import OPENAIKEY

client = OpenAI(api_key=OPENAIKEY)

completion = client.chat.completions.create(
    model="gpt-3.5-turbo",
    messages=[
        {"role":"system","content":"You are Helpful"},
        {"role":"user","content":"Hello!"}
    ]
)

print(completion.choices[0].message.content)

Does anyone have a clue on why i am being rate limited on this script

[trail cairn]

https://platform.openai.com/account/billing/overview

[heavy abyss]

do i need to pay for it to work

[trail cairn]

yes

[heavy abyss]

bruh

[sick wyvern]

Hey team, we are using assistant v2. We give it a file of products and return search results. For some reason, assistant only returns 3 or 5 results even though there are 5-7 more results available. How would you all recommend to solve this problem? Any pointers will be great 🙏🏼

[leaden burrow]

Anyone here know how much more expensive using an assistant is vs. using client.chat.completions.create?

I looked at the pricing and it says 0.03 cent per session but it seems like it costs more, just from using the UI in playground

[pseudo bobcat]

Can confirm, tried and works.

[devout valve]

ANyone having issues with chat not responding Error message

[nocturne geyser]

wdym?? 😭

[untold marsh]

Hello, I have been trying to reach out to OpenAI billing support for the past two weeks and I haven’t received an email back. I’m emailing [e-mail], but I’ve been ignored. Is there another way? (I already used the bot in the billing section on the billing dashboard and they are ignoring me there as well.)

[buoyant citrus]

Hi. I am new to using openai api reference.
I have done thread creation, runs etc... all client side on postman. Here's what I dont understand.

At what point do I get the AI assistant response? All i get are response bodies that I don't understand

[keen marten]

Hey @buoyant citrus. You need to use GET requests to fetch the messages of the thread where you first created the question. The info on how to do this is in the openai api reference docs. I've pmed.

[summer geode]

Hey

[keen marten]

Can anyone please help me out with a ChatGPT Plus invitation? Thank you!

[waxen needle]

I can no longer upload files in ChatGPT, both in the chat prompt and when configuring a GPT. This happened after I tried to upload a large bulk of data at once into a GPT. Every time I attempt this, I get a red popup at the top of the screen saying 'Unable to upload' followed by the file name. How can i solve this?

[humble ocean]

if your message says something like "something went wrong bla bla" then just import ona image and after that the chat should start work again.

[keen marten]

Maybe this wasn't clear enough. The invite can only come from a current subscriber.

https://help.openai.com/en/articles/8381046-free-trial-invites-faq

[odd cloud]

hello

I switched to corporate membership but my corporate uses were not opened

[timber shadow]

Hello, can you explain a little more? Do you mean you signed up for a ChatGPT Team plan?

[odd cloud]

No, I was a member of Chat GPT plus, but I canceled it and switched to the ChatGPT Team Subscription plan. But I can't use the features of the enterprise plan

[timber shadow]

This channel is public, so you might want to delete this message because the picture has email address listed.

Enterprise is a different plan than Team -- Team is a self-serve option for teams with two or more people. Enterprise plans are negotiated between larger organizations and OpenAI's sales team directly. You can read more about the differences here: https://openai.com/chatgpt/pricing.

[odd cloud]

First of all, thank you for informing me. Well, don't we have a common workspace in my plan?

[timber shadow]

On Team, you have the option to share chats and GPTs with just your Team. Were you wondering about specific other features?

[odd cloud]

Yes, I'm curious.

[winged frost]

Hi I am using the openai generation api and it keeps repeating its messaging do you know how to do work around this?

[trail cairn]

whats repeated in the picture?

[winged frost]

The questions

They are too similar

prompt: Can you make a question of the day message for my server and can you generate it entirely different from your past messages. (Please don't put the question in quotation marks!, make it about personal preference like food, music, games, cars, songs, birds, sports, where people would like to go, favorite youtubers

[trail cairn]

what temperature?

[winged frost]

0.1

[trail cairn]

that's bad, it will repeat the most asked questions the most

[winged frost]

I've tried loads

and they all done the same

[trail cairn]

use something like 0.7-1 and reword the list part

[winged frost]

ok

[winged frost]

How should I reword it

[trail cairn]

something like .. one of ... but not limited to

as a daily task it will always begin to repeat at one point due its training data and with fixed instructions without changing keywords

[winged frost]

How much tokens should i use for it?

[trail cairn]

it would be better to have an array of like 300-500 words in code and pick 1-2 of them randomly each time and tell it to use them to make a question

chatgpt can create such array of words easily

[winged frost]

like this?

word_array = [
    "music", "vacation", "sports", "workout", "movie", "memory", "hobby", "book", "recipe", "getaway",
    "career", "cuisine", "fashion", "TV", "lyrics", "playlist", "home", "business", "season", "video game",
    "adventure", "artist", "pet", "era", "game", "quote", "activity", "art", "cuisine",
    "genre", "vacation", "team", "routine", "movie", "memory", "hobby", "book", "recipe", "getaway",
    "career", "cuisine", "style", "show", "song", "playlist", "home", "idea", "season", "game",
    "adventure", "work", "era", "game", "quote", "activity", "art", "cuisine",
    "music", "vacation", "sports", "workout", "movie", "memory", "hobby", "book", "recipe", "getaway",
    "career", "cuisine", "fashion", "TV", "lyrics", "playlist", "home", "business", "season", "game",
    "adventure", "artist", "pet", "era", "game", "quote", "activity", "art", "cuisine",
    "genre", "vacation", "team", "routine", "movie", "memory", "hobby", "book", "recipe", "getaway",
    "career", "cuisine", "style", "show", "song", "playlist", "home", "idea", "season", "game",
    "adventure", "work", "era", "game", "quote", "activity", "art", "cuisine",
    "music", "vacation", "sports", "workout", "movie", "memory", "hobby", "book", "recipe", "getaway",
    "career", "cuisine", "fashion", "TV", "lyrics", "playlist", "home", "business", "season", "game",
    "adventure", "artist", "pet", "era", "game", "quote", "activity", "art", "cuisine",
    "genre", "vacation", "team", "routine", "movie", "memory", "hobby", "book", "recipe", "getaway",
    "career", "cuisine", "style", "show", "song", "playlist", "home", "idea", "season", "game",
    "adventure", "work", "era", "game", "quote", "activity", "art", "cuisine"
]

[trail cairn]

yes

except there are duplicates

[winged frost]

Will this affect anything?

[trail cairn]

higher change on repeating same type questions vs if you had 300 unique words

[winged frost]

Thx it works much better now

[untold oracle]

Hey guys I'm working on a custom GPT chatbot that can send requests to a server based on the users requests.
There are may types of reuqests that can be made and different types of response bodies in each type. What is the best way to finetune, is it to finetune one model which will genrate api resuqest modies, or urls or all types of reuqest, or the other modular approach which has a custom for each type fo request and top level classifer which classifies the request into the type of request.

[buoyant citrus]

Got it. I appreciate the help thanks... I also left you a pm on how i did it

[vital zinc]

I saw recently that "GPT-4 Turbo is now available to paid ChatGPT users." But, I don't see an option to select it. Did turbo just replace the regular gpt-4? Or is it just rolling out gradually?

[fading robin]

yes gpt-4 is now just Turbo version

[spice wave]

Anyone from openai support here? Our apps are down and I'm unable to add funds. The UI is locked.

[merry widget]

Hi
I am currently undertaking a law degree and have been using GPT 4.0 as a study aid. However, when I require more sensitive questions, such as those pertaining to criminal law, the responses are not as expected. I am seeking a prompt that will provide more accurate and relevant responses.
Any assistance would be greatly appreciated.

[merry widget]

Please mark this message if you reply

[timber shadow]

Hi, one thing to keep in mind that could be relevant in this case is that GPT-4 (and OpenAI services in general) are not currently suited for all topics. This is a page from API documentation, but it gives a good idea of the kind of things you can't really do with OpenAI services: https://platform.openai.com/docs/guides/moderation/overview

I'm not sure exactly what your sensitive questions were with regard to criminal law, but I could see how they might be adjacent to disallowed content, which could explain the behavior you're experiencing.

[merry widget]

ohh, ok

[eternal dove]

is anyone familiar with triton's internals? I'm especially interested in how triton decides how to pick e.g. grid/block dims on a GPU for micro/nano tiles

the MAPL paper mentions that as of then, triton just does an exhaustive search on all possible kernel configurations for each nano/microtile, is this still the case? How does triton know the search space to do exhaustive search on?

(also idk if this is where i would ask this kinda question, admittedly i did try reading the source code, but i couldn't find where this was done aside from the autotuner, which seems like the user just provides a list of configurations v.s. a true exhaustive search over the entire configuration space)

[nocturne geyser]

what

[rigid musk]

What is the link with chatgpt ?

[distant axle]

My ChatGPT Plus account was suspended, and OpenAI sent me an email saying that I logged in from a location outside of their policy scope. Which email address should I send my appeal letter to?

[timber shadow]

Hey! I don't think there's currently a monitored email address for queries like this, unless one is mentioned in the email you received. Instead, the best official support channel would be via the chat widget in the bottom-right of help.openai.com. Start a new chat there, click through the most relevant options, then you'll eventually find a path that lets you type your own message. This won't connect you with someone live, but it'll submit a ticket they can review and get back to you on.

[distant axle]

Thank you

[little oriole]

Hello! I'm trying to create my own gpt for code scripting, I'm wondering if anyone have any experience in creating such gpt? Like what prompts are to be given? If I provide an internal textfile with information, how should it be organized for best result? And maybe more information that I'm not aware about.

[eternal dove]

It is not a chatgpt question? It is a question for openAI's triton compiler

[cold folio]

That's against the terms of service

[bronze charm]

While you cannot use ChatGPT to make impactful decisions regarding law, it certainly can be used for study; just be cautious of its unreliability and always confirm information

You must not use any Output relating to a person for any purpose that could have a legal or material impact on that person, such as making credit, educational, employment, housing, insurance, legal, medical, or other important decisions about them.
From https://openai.com/policies/terms-of-use
I recommend taking a look at Harvey from the following for your accuracy woes:
https://openai.com/blog/introducing-improvements-to-the-fine-tuning-api-and-expanding-our-custom-models-program

[gleaming moon]

I want to switch to the Chat GPT 4 Turbo version, but it doesn't appear in my options.

[bronze charm]

Did you purchase any credits?

[gleaming moon]

I'm not fully clear on the credit system. I am currently subscribed to GPT-4. Could you please explain if additional credits are required to access the Chat GPT 4 Turbo version?

[bronze charm]

OpenAI's ChatGPT and their platform are two different services. You will need to purchase credits in the platform to get access to GPT-4 Turbo even if you already have ChatGPT Plus purchased

[nocturne geyser]

ChatGPT and the API are two seperate things

[nocturne geyser]

^

[gleaming moon]

Thank you for the clarification. Could you guide me on how to purchase credits for GPT-4 Turbo, or where I can find more information on this process?

[trail cairn]

https://platform.openai.com/settings/organization/billing/overview

[surreal cypress]

About ready to switch llm's all together at this point, i have some how used up my gpt4 usage till 11:13pm its 3:37pm now

honestly im extremely upset that openai isnt clear about anything and pass their failures off to the user, like say if it fails or errors out and you regenerate, those are still counted against me as gpt4 usuage even though they didnt actually answer my request

[cloud kindle]

i just paid for gpt4 but im stuck with a usage cap? what does this mean?

[primal jewel]

I can log into Plus on my phone. On Windows desktop, the green Continue button isn't clickable for password. I cleared cache and tried both Chrome and Edge. I would think the site was down (website indicates its fine) but I can access on my phone browser (not an app). Any ideas? I can use 3.5, just can't login.

[primal jewel]

Yes, you still have limiits with 4. However, I dont even think you can use it without paying either.

Oh.... unless you meant you couldnt use it at all. It should be tellling you when it will let you use it again. It gives me a specific time.

[cloud kindle]

yeah its giving me an hour, only paid as i have an essay due in the morning and i need help with spss :C sadge im screwed then

[primal jewel]

Sorry to hear that. One thing I do when I have a deadline is use 3.5 when I can.

[cloud kindle]

3.5 doesnt help with spss just gives me wrong stuff haha oh well

[primal jewel]

Ah, I see. Never tried. I also jump to Gemini or Perplexity. Not sure if that helps you though.

If anyone has ever had the login issue i have above, please let me know and I'll check back. Weird. Thank you.

[surreal cypress]

25 ish questions for every 3 hours

[bronze charm]

40 messages / 3 hours

https://help.openai.com/en/articles/7102672-how-can-i-access-gpt-4

[surreal cypress]

depends on what your having it do, trust me

see the photo posted above lol

[tight acorn]

I made an application using OpenAIKIT with SwiftUI code, I have been using it for 2 months by getting API, yesterday hackers used all my balance from my api address, today I deleted the api address, I uploaded a new api address, then they used the balances in my new api address again, have you experienced this problem? Where is the problem??

[soft wedge]

I am trying to make a python script which has the ability to answer questions from local database. Here is the code bellow:-

from openai import OpenAI

client = OpenAI(
api_key = '',
)

completion = client.chat.completions.create(
model = "gpt-3.5-turbo",
messages = [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "Hello!"},
]
)

print(completion.choices[0].message.content)

def fetch_data_from_database(file_path):
with open(file_path, 'r') as file:
data = file.read()
# print(data)
return data

def query_openai_gpt(prompt, database_content):
combined_input = f"{prompt}\n\nDatabase Content:\n{database_content}"

# Make a request to the OpenAI GPT API
response = client.completions.create(
    model = "gpt-3.5-turbo",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": combined_input}
    ]
)

# Extract the generated text from the response
print(response.choices[0])
generated_text = response.choices[0].text
print(dict(completion).get('usage'))
print(generated_text)
return generated_text

if name == "main":
# Specify the path to your database file

database_path = './random_questions.txt'

# Fetch data from the database
database_content = fetch_data_from_database(database_path)
print("database_content===================================>",database_content,"\n")
# Get user input for the prompt
user_prompt = input("Enter your prompt: ")

# Query OpenAI GPT with the user prompt and database content
result = query_openai_gpt(user_prompt, database_content)

# Display the result
print(result)

getting following error:
Traceback (most recent call last):
File "C:\Users\kumar\Desktop\Open_AI\test.py", line 56, in <module>
result = query_openai_gpt(user_prompt, database_content)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\kumar\Desktop\Open_AI\test.py", line 28, in query_openai_gpt
response = client.completions.create(
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\kumar\AppData\Local\Programs\Python\Python312\Lib\site-packages\openai_utils_utils.py", line 276, in wrapper
raise TypeError(msg)
TypeError: Missing required arguments; Expected either ('model' and 'prompt') or ('model', 'prompt' and 'stream') arguments to be given

Somebody please help

[nocturne geyser]

if you do have to switch

switch to anthropic

[leaden juniper]

Hey! Seeking for help! Anyone had luck with semantic search with images? I have a PDF with products and images of each of them and the user needs to upload an image and the bot should be able to search in the vectorial db for images related to that. What approach can I follow?

[slate iris]

I downloaded the Android app a few days ago and now it's telling me it's no longer compatible with my device?
I'm not sure if I'm the only one having this issue...

[surreal cypress]

claude3 right?

[nocturne geyser]

yes

[surreal cypress]

solved an issue i was having all day in chatgpt trying to debug it fixed it in a single response

yup making the switch

[nocturne geyser]

cold

sorry for making y'all lose subscribers
but it's only fair

[gilded pine]

Hello everyone, yesterday night(at least for me) I bought ChatGPT+ to work with advanced data analysis, but I cannot see the beta options panel on the settings, I tried logging out and back in, but that did not work either, the subscription is active, I can use GPT4, but I can not see the menu. What am I doing wrong? Thanks.

[charred parcel]

Advanced Data Analysis is automatically a part of ChatGPT, the beta options have disappeared some time ago, you don't need to enable anything to use Advanced Data Analysis

[gilded pine]

Thanks!

[quartz shale]

Hi there,
I got a question for you please...
I was subscribed on a monthly plan to the ChatGPT Plus plan until 22 Apr 2024. I use a Macbook as a computer and use both web version: web browser on Mac, and ChatGPT App on my Samsung mobile phone.
I subscribed to ChatGPT Plus via AI Chat Bot on my Macbook today, 24 Apr 2024, using my Apple ID, with the same email address I had used for my previous subscription.
Now that I log in to my account on the browser, my account doesn't reflect my subscription; could you please help?
Thanks

[timber shadow]

Hello, can you please explain more about what you mean by "I subscribed to ChatGPT Plus via AI Chat Bot on my Macbook today"? Did you use a Mac app called "AI Chat Bot" to subscribe?

[quartz spruce]

idk if this goes to the custom gpts or the datga analysis but i guess the community-help would be the right place to send me in the right direction. i want to figure out how to create a gtp for apple playgrounds app and share it for people who want to learn to code and mess around and have a gtp to help debug or help make a ipad app ect. if i made a gtp how would i feed all the data for apple playgrounds to it? to it? would i put a bunch of playground pdfs in the knowledge section of the gtp? would i make sure it knew swift 5.9 the newest version? i just want to know how i would go about this because apple wont let you download the pdf's they are stuck in ibooks. or books app whatever it is now. how would you guys go about this thanks fellas, ps. i downloaded a gtp from two folks 1 had 60,000 documentation of swift the other 23,000 not sure where they got the data for that.

[cunning orbit]

Why can’t I share links anymore so somebody else can continue the chat

[trail cairn]

do you get an error?

[cunning orbit]

I can share the chat, but can’t continue writing in it.

[digital dagger]

Hi guys,
I am using langchian and kor, openai API to extract some data.
I have a very long schema.
It looks like the output become unstable for longer schemas.
Please let me know the best practice to extract json data for long schemas.

I am ok to alternative more stable suggestions too
Thanks in advance.

[ocean timber]

wen fix?

[flint sonnet]

Hello! I paid for ChatGPT Plus on the OpenAI website but didn't get access to GPT-4. What should I do?

[timber shadow]

Hey! Can you share a screenshot of what you see when you go to chat.openai.com? Be sure to edit out any personal info that might be in the screenshot!

[timber shadow]

Thank you! It looks like your subscription page says your plan is/was set to renew on 24 Mar 2024, but I don't know that language so I can't be sure -- is this correct?

[flint sonnet]

sorry for language, my plan set to renew on 24 may 2024 🙂

pay today 🙂

Wow, community magic: it's fixed now

[timber shadow]

No worries at all! Thanks for confirming that. So when you scroll down that page and find "Invoice History", do you see an invoice for today that says "Paid"?

[timber shadow]

Oh great!! Good news, glad it fixed itself for you!

[flint sonnet]

Thank you for so fast response

[plain pagoda]

Silly question guys, is there any GPT to convert MP3 to music sheet?

[uncut aurora]

You can use Whisper for that - another AI by OpenAI.

[plain pagoda]

Thanks

[timber shadow]

Hey! Do you mean uploading a song MP3 file and getting sheet music/notes for it as output? Just clarifying because I think the suggestion from Frreiheit, Whisper, is just for voice-to-text.

[plain pagoda]

Or MIDI but any option is just sending me to some websites to download software

I just need to get the sheet music of one piece

I checked already several GPTs but after I try they tell me they can't do that and refer me to the sites

[timber shadow]

I don't know of anything off the top of my head, but maybe try searching online "music transcription AI" -- the channel #ai-discussions would be a good place to ask too. I don't think OpenAI has a service for this, and you can talk about non-OpenAI services in that channel.

[plain pagoda]

thanks, that will be helpful

[quartz shale]

Yep, you're correct. I haven't messed-up, have I?

[timber shadow]

I'm not sure, feel free to share a screenshot in this channel of what you subscribed on, just be sure to edit out any personal information in the screenshot. The only official ChatGPT platforms are:

• https://chat.openai.com, and
• The official ChatGPT app on Android and iOS, links can be found here: https://openai.com/chatgpt

[quartz shale]

I see... thanks for your response. It looks like I've messed-up!
I've gotta cancel the subscription ASAP

[timber shadow]

Sure thing, good luck! Hope it goes smoothly for you!

[rugged wing]

chatgpt is again asking login lol

[dusk shuttle]

Does anyone know how long it takes to transcribe about a 20-minute audio using the Whisper API? I might have been waiting for ten minutes since I made the request, but there has been no response yet.

[trail cairn]

something is probably wrong then as i remember the api being fast when i tried it. even whisper ran locally transcribes 1h audio in less than a minute

[glass spire]

Hi I want to build photo editor with OpenAI API how can I do it? Any guidance?

Do I need to create a custom GPT and train it?

[timber shadow]

Hey! Can you explain a little more about what you have in mind? Right now, the only image-related service OpenAI has via API is DALL·E: https://platform.openai.com/docs/guides/images/image-generation. There are some editing features available, but only for DALL·E 2 currently on the API. If you're interested (you might already know), DALL·E 3 does have an editing feature on ChatGPT Plus.

[glass spire]

Hi Solbus thank you for your swift response. I checked the link you sent. If you look at the mask example where the duck is being masked out, I want something similar. Is it possible if I upload an image and mask out items in the room and add other items into the image for example furniture etc while using DALL-E 2 API?

[trail cairn]

its possible but since its dall-e 2 the results are not that great

[glass spire]

That feature will come to DALL-E 3 anytime soon?

[trail cairn]

chatgpt plus has similar so maybe its coming to api at some point

[glass spire]

I have GPT+ so I can use GPT plus API and achieve the results I want?

[trail cairn]

no

[glass spire]

Ok

[brave walrus]

mem got pushed to me recently and 'm wondering if its supposed be non-functional in custom gpts or if I'm just special.

[timber shadow]

It sounds like the plan is to have per-GPT memories in the future: https://help.openai.com/en/articles/8590148-memory-faq

Does memory function with GPTs?
Memory for GPTs will be available when we roll it out more broadly.When it is, builders can enable their GPTs to have a memory. The end-user will need to have Memory on to engage with the GPT. Please note that ChatGPT and GPTs each have their own unique memories based on your interactions with them. Memory is not shared between GPTs and other GPTs or ChatGPT. Memories, like chats, with GPTs are not shared with GPT builders.
Also https://help.openai.com/en/articles/8983148-does-memory-function-with-gpts

[brave walrus]

Thanks! This would be great to add to the notification when it's enabled to save users from trying to have thier goto GPT remember things and fail immediately. It's a jarring user experience.

[timber shadow]

Good suggestion! Maybe consider making a post over in #1070006151938314300? Because you're right, I bet you won't be the only person to try this!

[formal jay]

is anyone else experiencing slow and halted prompts?

[finite parrot]

Me

[short aspen]

I’m facing issues on chatgpt it keeps on retrying but no luck

[nocturne geyser]

bomboclaaaaat

[finite parrot]

The slowliness is still happening? or only me?

[shell osprey]

im having issues too

[normal pagoda]

Yep, it's extremely slow and my history isn't loading either. Time to do other work stuff I guess.

[elfin geyser]

Its stops loading everytime a code block is formed.

[zenith aspen]

Same, having these issue for 48h...can we have someone from the OpenAI team giving feedback about this? Insane to me

[short orbit]

Its disappointing that ChatGPT 4 becomes unusable at a certain time everyday. I'm assuming this is peak time.. Paying per head for company and its essentially unusable.... Is this being fixed?

[gray stream]

lol

[proven rock]

I'll do you one better

[cold folio]

Please click on the thumbs down button and report it here #1070006915414900886 👍

[jolly kelp]

is anyone getting this issue??

[timber shadow]

Hey! Are you connected to a VPN by chance?

[jolly kelp]

No

[timber shadow]

All right -- can you try again in a private window, with all browser extensions disabled?

[jolly kelp]

Thanks It did work there!

[timber shadow]

Good deal! Since it works in a private window, maybe just try clearing your browser cache and trying again in a regular window. If it breaks again, it might be an extension interfering, so maybe test those.

[jolly kelp]

It Again started

what should I do??

[timber shadow]

Does it still work in a private window?

[jolly kelp]

yeah

[timber shadow]

Have you tried clearing your browser cache?

[subtle tulip]

wow ChatGPT is so painful, the performance is always terrible. Will switch as soon as a viable alternative appears.

[tender nacelle]

GPT-4 now just tells me this when i reach the limit, but doesnt tell me when i can actually use GPT-4 again like it did before... any help?

[tender nacelle]

nvm looks like if i refresh the page so i can send a message, it then gives me an error which shows me when i can start using it

[solar isle]

Hey, I wanted to ask about chatgpt4 on the website (not the API), I'm only interested in https://chat.openai.com/. What is the token limit for version 4.0 (paid)?

[fading robin]

32K

[umbral swallow]

gpt can't output code right now, or?

Is it just me who has the error?

[timber shadow]

There are some similar reports in #1070006915414900886 right now -- hopefully fixed soon!

[umbral swallow]

if you ask gpt to output the code in the normal format, then it works

[timber shadow]

Nice workaround!

[sterile basin]

anyone else see chat gpt freezing at specific code lines

#[derive(Deserialize)]
struct Message {
timestamp: String,
content: String,
embeds: Vec

another example: GPT
For the definition of the embeds field in the Message struct, you should define it as a vector containing Embed objects. Replace the placeholder VecEmbed with `Vec

rust
Copy code
embeds: Vec
embeds: Vec

To properly define the embeds field within the Message struct as a vector of Embed objects, you should modify the field like this:

rust
Copy code
embeds: Vec
Here's the correct way to define the embeds field within your Message struct using a vector of Embed objects:

rust
Copy code
#[derive(Deserialize)]
struct Message {
timestamp: String,
content: String,
embeds: Vec
Here's how you can define the embeds field in the Message struct:

rust
Copy code
embeds: Vec

anyone know a fix?

[cloud marlin]

Hello, community memebrs?

Can I have some little sorts of, I would say, some commands for ChatGPT?

I heard that those, filters and all such restrictions can't be bypassed anymore!

[frank kernel]

guy's i need help about archived chat files , i want to download it all or unlock them in bulk , cuz it give me only option to unlock one by one , and i don't don't have the chat export icon inside my web page in data control section, pleas any help

[gritty anvil]

How do I publish my GPT to the store? Also can anyone with GPT access find my gpt

[obtuse gust]

why the gpt always like this today?

[obtuse gust]

ight it may just be the chrome problem

[raven vector]

anyone know how long it takes after adding more credit balance for the api to unlock? it keeps telling me I hit the quota

[heavy patio]

I did this almost a year ago, but when I did it, it took almost whole day IIRC. It's probably gotten faster than that now, the system has improved a ton across all other factors.

[heavy patio]

Top right corner of your custom GPT edit screen, press the share button. If the publish to store is greyed out, I think it explains why and how to fix it. You have to verify with your name or website or linkedIN I think

I verified awhile ago, so pulling old memories but - https://help.openai.com/en/articles/8798878-building-and-publishing-a-gpt covers it

[boreal glen]

I am having the exact same issue. Is this a known thing? Quite annoying

[gritty anvil]

Well that's extensive lol, my got can be plugged into my site (if/when created) to function that way right?

[heavy patio]

I'm not sure what you are asking, sorry. Under website verification, you can give your FB profile or linked in profile site. It doesn't have to be your own custom built website.

I believe I am verified as a builder through my linkedin profile

[rugged fossil]

same here

[heavy patio]

Frustrating! I would think some level of error is within an acceptable threshold for technology, but not over and over. If your issue is persisting, I recommend reaching out to the help center like the team requests in the error.

[boreal glen]

I did, which makes it a bugreport I reckon. Nothing that can help me in the short term it would seem like. The app works fine on my phone but I really need the web interface at the moment.

Not alone: #1233799429161619576 message

[gritty anvil]

That answers my question actually lol

[gritty anvil]

If I have a GPT can I use it externally like a chatbot a website uses

[heavy patio]

Not really, but kinda. You can put a link to the GPT on your website, but clicking it will take you to the openai site where your customgpt is hosted (the share link). It's not really the right tool for what you are thinking, with full integration.

[umbral swallow]

i guess it works now

Previously, the HTML code was given without the '<>' around the tags to avoid output loops when interacting with ChatGPT. Now, it's presented with the standard angle brackets, as typically used in HTML documents.

[rapid orchid]

alo guys, how can i get in contact with an openai employee? i've tried to do it thru their website like more than a month ago (?) but i got no response

[trail cairn]

website is the only way or depending of the issue and luck #1070006915414900886

[inland apex]

I'm getting several issues with chatgpt, and has errors in the console regarding CORS, and some other stuff. Is this a known problem or just something on my end?

[quartz spruce]

idk if this goes to the custom gpts or the datga analysis but i guess the community-help would be the right place to send me in the right direction. i want to figure out how to create a gtp for apple playgrounds app and share it for people who want to learn to code and mess around and have a gtp to help debug or help make a ipad app ect. if i made a gtp how would i feed all the data for apple playgrounds to it? to it? would i put a bunch of playground pdfs in the knowledge section of the gtp? would i make sure it knew swift 5.9 the newest version? i just want to know how i would go about this because apple wont let you download the pdf's they are stuck in ibooks. or books app whatever it is now. how would you guys go about this thanks fellas, ps. i downloaded a gtp from two folks 1 had 60,000 documentation of swift the other 23,000 not sure where they got the data for that.

[lapis yew]

I asked for a simple code example, but I have a bug in a code block that makes it very slow and repeats an empty code block.

[vivid jungle]

this happens to me when asking about OSSEC configuration

[rugged wing]

getting the same with Coldfusion. If I use Python it works

getting the same with Coldfusion. If I use Python it works

Chatgpt is kind looping for specific developper languages other tha python. Any clue?

[ivory mesa]

I encountered the same issue in TypeScript.

[fading robin]

what is that button ?

[trail cairn]

model selector

[fading robin]

a well i don't have it hah just seen it from that ss up there interesting

[brazen pasture]

Hi everyone, I need some help OpenAI has removed my GPT from the GPTStore, and you can view it here: https://chat.openai.com/g/g-eppiW3Zqp-health-coach.

They informed me that my GPT was removed from the marketplace due to a violation of OpenAI's Usage Policies. I've tried contacting OpenAI about this issue, but after three days, I still haven't received a response. In an attempt to resolve the problem, I removed all content from the GPT including its name, description, instructions, conversation starters, datasets, and actions. Despite these changes, I still can't republish it.

I also need to understand exactly what needs to be edited in the GPT instructions to allow it to be published again. Originally, this GPT was one of the first I created and hadn't been updated in months, yet it was performing well with over 15,000 conversations and nearly 200 reviews, averaging a 4.5 out of 5 rating. So I'm quite sad if I have to let it go

Any advice on how to proceed would be greatly appreciated

[magic wave]

Keeps getting better and better

[split lotus]

guys, I subscribed on chat GPT plus via google play checkout but after 5 minute I asked for refunfld cz I was charged a lot due currency. seems like $ increased in orice so I asked for refund. while I'm waiting, I used service fir a day fir a 5 hrs like that what ubthink, Will my refund will be accepted?

[gritty anvil]

I always go directly to source via my card for this reason, it maybe a Google play interface thing

[dreamy geyser]

I'm facing a problem with uploading docs on ChatGPT plus

[rugged wing]

how do you get help here?

just mostly seems like people talking into the void here

[trim lion]

i cannot remobe my phone number from openai

[dusk ermine]

hey guys, can dall e 3 outpaint?

[dusk ermine]

or inpaint?

it seems as if every time i ask it to modify an image it generates something ocmpletely new

[raw pulsar]

Why can't ChatGPT read the file I uploaded? It used to work fine before.

[timber shadow]

Hey! Not yet on outpainting, but inpainting yes, but currently only images DALL·E makes (not ones you upload). Select a DALL·E image, then you'll see an edit/select button. You can select the area you want to inpaint then describe the change you want. Here's a helpful article describing how it works: https://help.openai.com/en/articles/9055440-editing-your-images-with-dall-e

[magic wave]

Because the company is more fast in dealing with internal spanish soap opera drama instead with problems of paying customers

[glass spire]

@timber shadow where do I get DALL-E 2 API specifically? I want to use it for image manipulation.

[trail cairn]

https://platform.openai.com/docs/guides/images/edits-dall-e-2-only

[glass spire]

So I just need to switch the model to DALL-E 2 in the code?

[trail cairn]

yes

[glass spire]

Thank you Radium

[rugged wing]

Why there's expires date

after I add 5$ of credits

[peak smelt]

Companies generally don't want to keep any liabilities like this on the books forever, so they add an expiration date

[fervent vault]

why can't I find the code capilot GPT?

can anyone help me verify it's a general thing or not?

[ocean crypt]

Hello Everyone
I would like to ask about function calling in open ai
do you have experience building a weather chatbot using function calling?

[rigid musk]

With the api ?

[ocean crypt]

hi @rigid musk
nice to meet you

[ocean crypt]

do you know function calling in open ai?

[rigid musk]

Not much. It's the api and it is pay per use

https://openai.com/blog/openai-api

[ocean crypt]

https://platform.openai.com/docs/assistants/tools/function-calling

please check this link

[blissful valley]

Hi everyone, how do I turn off chat history? All my chats are being saved and I don't want chats to be saved.

Everything turned off, yet ChatGPT is storing my chats

[trail cairn]

maybe you have this version https://help.openai.com/en/articles/8914046-temporary-chat-faq

[wanton shoal]

@blissful valley - check this out:

[blissful valley]

Yes, you are right. Thank you. Let me see if this temporary chat feature is persistent(meaning I don't have to manually turn on every time)

[blissful valley]

Thank you

[near valve]

Will Dall-E ever be free again?

[timber shadow]

Hi, what platform did you use DALL·E for free on in the past? There are 5 free daily uses in this server in #image-bot and I think Bing's image creator still gives some free daily credits.

[high ermine]

Does anyone know what the limit on gpt memory is? Like the limit on how much it can remember

[high ermine]

Dalle-E is always free with bing chat. If you run out of credits it just takes longer to generate.

[stuck granite]

Hey, I need help with AI for my security project. Can someone answer my questions? I really appreciate your help. Thanks!

[timber shadow]

Hello, are you wondering about OpenAI services specifically? If so, feel free to ask here and maybe someone who knows the answer will see it and get back to you! If it's more just AI questions in general (not just OpenAI), try asking over in #ai-discussions!

[stuck granite]

oh okay thank you.

[nocturne geyser]

I think the inpainting uses DALL-E 2 which is why it sucks so bad

DALL-E 3 has no outpainting/inpainting feature directly available with the model

[gusty nova]

sorry for the potentially silly or frequently asked question, but has the Memory feature basically overtaken the 'know about you' Custom Instructions? i'm unsure how to properly use the latter anymore

[rustic marlin]

dalle was never free

do you mean dalle-2 beta test? it wasn't free either, people would jsut get some free credits every day

[wide sail]

any idea why i cant upload a zip file to an assistant creator? (code interpreter) it says format not supported while im pretty sure it is

[timber shadow]

Not silly at all! I think we're all still figuring out what the best use cases are for memory vs. custom instructions. Both are still around! The way OpenAI is differentiating the two right now is by kind of suggesting to use memory for user details, and to use custom instructions to define behavior. Some more info in this article: https://help.openai.com/en/articles/8983151-is-memory-different-from-custom-instructions
Also in general: CI are still totally user-written, whereas memory is more automatic. Eventually, memory capabilities will be extended to GPTs as well (CI are currently disabled in all GPTs), and each GPT will be able to have its own memory.

[halcyon crescent]

Hello all. I am a ChatGPT+ subscriber based in the USA. I have been asking for access to the memory feature that was announced as go live earlier today for months, but I still don't have access to it.

[vivid jungle]

hi, I subscribe chat gpt plus and can access chat gpt 4 but cannot access chat gpt 4 turbo. The model selection is chat gpt 4 and chat gpt 3.5. How can I enable chat gpt 4 turbo?

[halcyon crescent]

I have the exact same issue.

[timber shadow]

@halcyon crescent Hey you both! GPT-4 on ChatGPT is Turbo, it's just not labeled as such.

[halcyon crescent]

Thanks for the clarification. It still doesn't explain the lack of access to the memory function however.

[timber shadow]

When you go to Settings in the browser version - https://chat.openai.com/#settings - what sections do you see at the top? You're saying you don't have the "Personalization" section?

[halcyon crescent]

That is correct.

[near valve]

The one on labs.openai

[raw pulsar]

So what's the solution?

[rustic marlin]

yea, it wasn't free

[carmine valley]

Hi, I get this error when trying to use chatGpt for the first time today: Something went wrong. If this issue persists please contact us through our help center at help.openai.com. anyone else?

[nova dragon]

hey guys, im trying to train a model using openai key in order to feed it with a .txt file with 50 thousand lines of customer feedback. How can i implement this so we can ask questions about the corpus of the data that is in portuguese.

ps> i need both the questions and the answers to be in portuguese, is that possible ? thanks in advance

[trail cairn]

https://platform.openai.com/docs/assistants/tools/file-search

[glass inlet]

Hey! My chatGPT started to ignore any files I upload to him. It just tells, that I didn't upload anything. This happens with files of any format(txt, xlsx, docx, etc.)

[dusty plume]

why there is no memory for Teams account ??

[glass inlet]

I continues with any file

[near yarrow]

Hi

I am subscribed to chat gpt 4.

Is there a bot that allows me to add my account to be used via discord on my own server

[trail cairn]

nope but you can make one via api and pay for it

[near yarrow]

How difficult is this

[trail cairn]

not too hard openai offers an example project at github

but most people cancel the idea when its not free to use

[near yarrow]

What is the pricing for such a thing

[trail cairn]

https://openai.com/pricing

[near yarrow]

I already paid for chat gpt 4 monthly is this on top of that

[trail cairn]

it is, its a different service

[near yarrow]

Ok thanks I will have to rethink this one lol

[azure fractal]

Good chat-gpt had me 7 hours waiting for a file with an API and when sending me the file I get a blue text that supposedly is a hyperlink that redirects me to Google Drive where supposedly it will send it to me but when I click it does not take me anywhere.

[trail cairn]

chatgpt 3.5?

[azure fractal]

yeah

[trail cairn]

it hallucinated and made it all up by generating random link

[azure fractal]

It won't let me click on the link

[timber shadow]

Like Radium mentioned: 3.5 can't do things like this, so the link isn't real. Unfortunately LLMs are prone to hallucinations like this, especially when asked to do things they can't do. They don't know what they can or can't do sometimes.

[wanton violet]

I cancelled my subscription but still the payment says processing will my money debited

Help

[azure fractal]

Do I have to buy Chat-GPT 4?

[timber shadow]

What are you asking ChatGPT to do?

[timber shadow]

Hey! Best bet will be to reach out to OpenAI support at help.openai.com -- use the chat widget in the bottom-right of that page.

[azure fractal]

To make me an extensive API with a website and more things.

Discord API integration: I developed an API that allows to search Discord user information and obtain their associated IP address.
Authentication integration: I implemented a registration and login system for users, allowing them to access specific areas of the application.
Administration panel integration: I developed an administration panel that allows to manage access permissions, prohibit searches for certain Discord IDs and grant administrator permissions.
Reseller area integration: I created a dedicated area for resellers where they can generate access keys for their customers and manage their duration.
Customer area integration: I developed a public area where customers can redeem access codes and access bot information.
Log system integration: I implemented a log system in the administration panel that records all actions performed by users.

Here's everything I asked him to do

and he kept me waiting for 7 hours he said he was doing and now to deliver it nothing is arriving.

I also gave him my email address so he could send it to me.

[timber shadow]

ChatGPT can probably help you make these things, piece by piece, but ChatGPT (3.5 or 4) can't automate this process for you "behind the scenes" and then come back and give you a link to completed work. It's "just" an LLM, so it's only generating text, there's no background processing happening (except for cases where it's browsing the web for you, making an image with DALL·E, etc. on ChatGPT Plus). Any time it's saying to you "I'm working on that in the background, check back in later", etc. is a hallucination.

[azure fractal]

But I left it open all the time

and I told him how I was doing and he said I was doing 65%.
to come back in 1 hour

and now he told me that the whole project is finished.

[timber shadow]

Those are all hallucinations, unfortunately.

[azure fractal]

So there is no API

[timber shadow]

If ChatGPT told you it was working on something in the background, it was hallucinating.

[azure fractal]

ok thanks

[proper plank]

how can I get someone to honestly help me with my account? it's been months that I've had a problem and I can't get a single anyone to talk to about it. I truly believe there's no humans that work there and Sam is a 3D image given literally noone there responds that's not AI

[rigid flicker]

Hi all. ChatGPT keeps removing saved memories it has of me. It's happened a few times now. Yesterday I went through a bunch of old chats and asked it to re-read the chat and memorize key components of it, which it did successfully for about 60 chats. I logged in today and there are only seven memories now. It removed even simple ones like my wife's name. For whatever reason, my birthday has survived every wipe though haha. Any ideas or options on how to restore?

[rigid flicker]

Also, while I'm here, ChatGPT keeps naming our conversations in Spanish for some reason? No clue why- my settings are set to english

[icy maple]

When do things start getting entered into memory? I checked my settings and see that memory is enabled, yet when I click "manage," it's completely empty.

[timber shadow]

Hey! You'll see a notification about things being saved to memory in new chats/messages from ChatGPT. You can try by saying something like "Remember that my favorite color is blue" in a new GPT-4 chat and you should see the notice appear above its response to you after that. Then, you can go check the manage area again. Note that a few people have reported some saved memories disappearing! So there may still be some kinks they're working out now that the rollout is much wider. Just to be aware!

[icy maple]

Ok, so does it only work in new chats? And does it only work for things I've explicitly asked it to remember?

[timber shadow]

I'm not positive--you can test by going to an old GPT-4 chat (I think currently it's only working with 4) and asking it to remember something from earlier in the conversation--I believe I've read someone saying they've done this successfully with multiple older chats.

It seems like it will also try to intuitively pick up things to remember about you, so it might write things to memory that you don't explicitly ask it to.