gpt-models

[coral yew]

Monthly

[gray blaze]

@half turret Maximum length messages, fully utilizing my current rate limit of 25/3 hours

@chilly badge yes. Your first prompt should be hi, I can't afford to use this service for more than two months. It costs me $30 a month. Please plan and implement a passive income stream to maintain my subscription

[chilly badge]

so can i make the app in a month?

its not a complicated idea it just needs design and programming

[oblique sorrel]

Good night

[subtle echo]

you guys know if there's a way to pay more to access more usage?

[cerulean timber]

Hi boiys

[native shale]

you could ask for api usage but you may need some tech/dev skills

[serene cipher]

yes, that would be the API where you pay as you use for every request

you would be able to use it via the playground

[burnt steeple]

being able to move the convos in the list would help so much, it's crazy how confused i'm getting.

[mental burrow]

i do wish we had more organization of them

but i just wrote my own chat app with the API so I can save and load and organize my chats to my heart's desire

[pseudo warren]

provides link to pyramid scheme

[gentle tusk]

Humans need to write content meant for other humans.👈

[loud wigeon]

,

[wind urchin]

Could you explain the meaning of “,” being used randomly please

@loud wigeon 🙏

[smoky crest]

You don’t know what a comma is????

[wind urchin]

I know what a comma is, I just don’t understand why it was used randomly above

Could you explain please 🙏 @smoky crest

[gentle tusk]

To succeed in the future must learn A.I.💪

[smoky crest]

What message

Oh I see

He probably Mistyped

Lol

[gray blaze]

So. Don't implement that one? Or add an instruction: do not extract profit based on wage or product theft structures.

Seriously. Just tell it what you want.

[gray blaze]

Correct

[wraith ginkgo]

<:book_icon:1046291503158399107> `` Rule 1 `` Be respectful.

Treat others the way you would like to be treated, and assume best intentions. Don’t harass or attack others, and don’t engage in hateful or generally malicious behavior (e.g. sexism, racism, homophobia, etc.). Keep the negativity to a minimum.

[tranquil olive]

I have been facing failed responses for some hours now. It keeps saying,-" something went wrong". It's been persisting for a while. Any advice please?

[frosty whale]

Is it just me, or has it actually been a month since OpenAI introduced plug-ins? Still havent gotten those features yet.

[sacred hemlock]

Im on the waiting list too

For God knows how long

[supple violet]

waitlist is not a guaranteed access, in fact it says only small number get invited

[sacred hemlock]

Yeah.... Thats right... Thats the reason why i do not really care

[distant raptor]

Getting an error message with CHATGPT-4

[supple violet]

plugins were introduced less than a month ago and while some plugins were already done they invited more devs to make and the current phase is that and initial testing

[distant raptor]

What can I do?

[sand hull]

What's the error?

[latent zinc]

everyone must be asking the same, but any tricks to get access faster?

have been on the waitlist forever

[supple violet]

no

[latent zinc]

some people getting access in a couple of days I read

are they prioritizing certain regions?

[gentle tusk]

@gray blaze Illusion A.I😭

[distant raptor]

Not sure what is the error. It does not say. It says load failed. Contact HelpAI. I can't respond conversation anymore.

[supple violet]

multiple factors which are not public, your application text, usage history, location, requested plugin, etc.

also random picks

[gray blaze]

Not sure what you mean?

[sand hull]

Seen th same error earlier after I'm inactive during a conversation. Refreshing the page did the truck usually but if it didn't I made a new conversation sba started over, as long as I wasn't inactive it didn't break.

[distant raptor]

Thank you! I refresh the page. I asked it a question and that's when load failed.

[stiff leaf]

since gpt5 isnt in development can we expect that the usability of gpt4 in coding wont really improve?

[grizzled vapor]

Since Saturday it's becoming impossible to use chat.openai.com with a VPN, tried multiple providers, eventually I get access denied.

[stiff leaf]

same

[knotty grotto]

thats normal

[gray blaze]

What benefits do you see using a VPN in your workflow?

[versed stump]

what things can chatgpt not do

what can it not say

[supple violet]

vpn services are getting blocked on purpose each week as they detect more of them

[versed stump]

why?

[grizzled vapor]

Privacy, better routing... here in the office is corporate policy, we can't browse the internet without passing thru the VPN.

[tranquil olive]

I noticed that if you have a vpn turned on, Chatgpt starts to act up

[gentle tusk]

@gray blaze It’s all happening! AI-generated content has finally arrived to the masses, but don’t fall for the hype.😭

[rugged hemlock]

How many messages do you get per hour with subscription? (gpt 4)

[supple violet]

people using from non supported regions, usage violations, try to evade usage limits etc

[pulsar wyvern]

Im getting errors anyone else? Doesnt let me generate just says "something went wrong" (multiple times)

[distant raptor]

How can you tell I have VPN on 0.0?

[tranquil olive]

Not sure about that. In my case, I turn mine off and on.

[grizzled vapor]

I'm not trying to create any fuzz, but I bet most tech ppl won't like to let govs decide what we can or cannot access. VPN is a thing, sometimes it's necessary or mandatory.

[arctic zephyr]

who keeps deleting those

[knotty grotto]

no ads here

[cunning topaz]

okay

[wraith ginkgo]

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

Please report any incident by sending a DM to @open crescent immediately.

[grizzled vapor]

Anyhow, another thing I noticed, GPT-4 is constantly giving me snippets of code with software versions released like a week ago. I like that it learns from interactions, but that's quite dangerous.

[serene cipher]

like that it learns from interactions
it does not

the content that users upload do not have any impact on the models

the model only changes when it is re-trained or fine tunned by OAI

[grizzled vapor]

Is it allowed to post screenshot here? I can show you.

[serene cipher]

sure, but not on this channel tho, ping me on #1050184247920562316 or #ai-discussions

[grizzled vapor]

will do

got a minute ban when trying to send you the screenshot on #ai-discussions

well, the case is simple, i was talking to it about a terraform kubernetes deployment and GPT-4 gave me a snippet with Alpine on version 3.17.3

that's a 2 weeks ago release, when I asked how it got the version, it dodged twice and then said:

Thank you for your understanding. I apologize for any confusion caused earlier. While I don't have the ability to access external resources in real-time or browse the internet, I may have inferred the Alpine version from patterns in the text corpus or from multiple conversations with users. It is still an oversight in terms of providing accurate information based on my knowledge cutoff. Regardless, I'm here to help you with any questions or concerns you have. Please let me know if you need any further assistance.

if it was like version 2 or something generic, alright, it was hallucinating, but "3.17.3" is too specific and precise, it's the latest version

[supple violet]

it hallucinated and used previous messages to make it up

[grizzled vapor]

that's the thing, I never mentioned Alpine to it, I always use 3.5, I never interacted not even once with GPT-4 mentioning Alpine or version 3.17.3 for that matter

and if there is any OAI member here, please feel free to read all my interactions with GPT, you will see that I have never mentioned Alpine 3.17.3 to GPT-4, I only discussed it with GPT-3.5 due to the message restrictions... i'm using GPT-4 only when 3.5 can't help

i checked my 2 weeks history to make sure I never provided that version, in fact, I just started using it on Friday so I couldn't possibly have mentioned prior to that

[supple violet]

it would only use info from the same chat thread

[grizzled vapor]

that's what I expect as well, I have two threads only, one named GPT-4 and one named GPT-3.5... when I need to talk something non-related to work, I create a different thread, but I never mentioned Alpine or 3.17.3 to it not even on side threads

[timber python]

Are there any prompts that help with ux design and optimizing websites?

[unkempt cedar]

promt-pictures

[feral terrace]

Im getting errors anyone else? Doesnt let me generate just says "something went wrong" (multiple times)

[fickle gale]

Us paid folks don't get those.

I have been using GPT 4 for a week and have never got that. It happened constantly when I was using free

[tight kiln]

I’m a paid subscriber and also getting errors

[fickle gale]

raise it to OpenAI. I never get those.

I would stop paying if I did.

https://status.openai.com/

No incidents for today

[mental burrow]

My sessions are expiring much sooner today, but the service is working fine for me

[feral terrace]

I am also a subscribed paying user, and since yesterday, I have been having this problem. Whenever I pause for a few seconds without using it, an error message appears

[hallow ember]

guys I have a question with regard the 3 hour gpt4 cooldown. If I use say 20 messages and 3 hours have past since then do those messages reset to 0 or does it stay at 20 until I finish with the 25 messages and only then the 3 hour cooldown starts?

[rich spire]

Can i try gpt 4 free or nop?

[safe ridge]

Heyo all! Be on the lookout for anyone trying to pay you for access to ur API key or account. Ppl are just reselling this stuff on the darkweb and anything they do with your API keys is on YOU! I'm sure ur aware already but thought I'd add another PSA!

[rich spire]

?

[fickle gale]

Nope

Use Bing Chat for GPT 4 free

Not possible on OpenAI sites

There is no trial available for this

[safe ridge]

U can try to find a chatbot that uses gpt-4

[fickle gale]

which is Bing Chat...........

[safe ridge]

There might be others tho! Be on the lookout!

theresanaiforthat has some listed!

[fickle gale]

I explored it. That site feels confusing to use. I couldn't simply find a button that says GPT 4 bot list or something.

[safe ridge]

Hm, feature request I think they'd be down w/ that

[tardy thistle]

Does anyone here know the context window for prompt and completion for GPT-4 used on ChatGPT interface?

I know it has a context length of 8192 tokens, but how is that distributed?

[vale summit]

Can anyone help me, my account has been banned, but my plus membership fee is still deducted every month. Wrote emails and no reply

[fickle gale]

wow. How did it get banned?

[vale summit]

I don't know

[fickle gale]

What kind of searches did u do?

[vale summit]

I just want to cancel the plus membership now

[fickle gale]

Here in India, all I have to do is disable international transactions in my banking app and OpenAI cannot charge my credit card even if they try.

Does your bank provide any such option?

[vale summit]

But they have charged and I want to get a refund for this month

[fickle gale]

https://help.openai.com/en/articles/6614161-how-can-i-contact-support

[cerulean obsidian]

Contact bank, reverse charge and Block your Credit Card as simple

[bronze lagoon]

1

[vale summit]

I can't log into my account, so only contact them by email

[bronze lagoon]

2

[fickle gale]

They have provided 2 options there. use the second option which doesn't need an account

[cerulean obsidian]

Hey are you using Plus?

[fickle gale]

Yes

[vale summit]

Thanks, I will try it

[cerulean obsidian]

How was your renewal processed any issues?

[bronze lagoon]

Is there a charge for this?

[fickle gale]

If you have international transactions enabled, then no OTP would be required like it is in India as mandated by RBI. So no issues in renewal.

I feel it is very dangerous that a company can just charge credit card without OTP or any other authentication, RBI has done a good job here.

[cerulean obsidian]

I'm from India, as read your from India too, I'm using Kotak CC with International transaction today is my renewal date but no charges has been deducted yet

[fickle gale]

Are u able to use GPT Plus still?

[cerulean obsidian]

Yes it's active

[fickle gale]

They'll deduct it. Give it a day or two.

[cerulean obsidian]

Okay thank you

[fickle gale]

welcome

GPT 4 is awesome to write complex python scripts with 4-5 steps. I used it today to write such a task. I posted the same question in Bing Chat, GPT 3.5 & Bard. None of them gave answer perfectly like ChatGPT GPT 4.

[cerulean obsidian]

I have written 1500 lines of python program with GPT 4 It's insane

[fickle gale]

cool 👍 👌

[wraith ginkgo]

`` Server Tip: ``

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

[fickle gale]

I am very much convinced after using GPT 4. AI is coming for our jobs. Once we have plugins that integrate with every service.

Imagine building a real time data pipeline. With plugins of ChatGPT available for AWS, GCP, Azure. All we tell ChatGPT is to build the system as per our requirements and it does the rest.

[cerulean obsidian]

I suppose AutoGPT is been performing something similar,

I think it as an assistant like it's a program which needs a command/prompt to run and to debug, but people have built more programs in which GPT was able to rewrite it's own code so well, it just needs command

[fickle gale]

Correct. AutoGPT is cool, but it is a separate open source project. I think plugins will gain traction quickly once the waitlist period is over and autoGPT will be mostly forgotten.

ChatGPT service has over 100 million users after all. It is going to be the next gold mine.

[wanton star]

uh

when will chat gpt 4 add image analysis?

[cerulean obsidian]

True even with 100m user base, I have seen so many people are unaware of this thing yet here in India, unless specifically introduced even after introduced they don't seems to contemplate the abilities this thing has

[wind urchin]

can I have like 0 gpt 3.5 and double got 4 grrrr hate this limit

[fickle gale]

Yeah, even I truly understood ChatGPT's ability only after paying for it and using GPT 4.

It will take some time, but people will eventually adopt.

Some people who aren't open to exploration will eventually see the competition use it and be forced to use it.

and that's how it will become mainstream in India too

[cerulean obsidian]

This, this is thing is similar to people first not coming forward to use Mobile Phones, but eventually now everyone does,

'The One Most Adaptable to Change is the One that Survives' - Darwin

[wind urchin]

Are there any guidelines regarding the size of the chats? Like is there a point at which it's better to start a new chat, or is it better to keep the same topic in one chat. Asking more from a point of the quality of generated responses. Obviously it's harder to navigate through a conversation when it gets too big.

[fickle gale]

Indeed

[cerulean obsidian]

It's hard to say, but from my experience the chat conversation have some kind of summarisation after every long chat, so it does remember everything but not coherently

For coding purposes I always end old chats after my purpose is served to keep memory more concise

[tardy thistle]

Well, I'd advise you to shorten your prompts, and use some prompt engineering at firsthard to make sure you get short and concise completions depending on what you're looking for. And then you can, if you know you have a long prompt, use the tokenizer to see how much you can cram into the chat before the bot will start forgetting what you said (context-window) or when it starts degrading in quality due to its attention on your conversation (you not saying if it makes some invalid suggestion will make it degrade and add more false statements as you have not penalized it).

Oh and absolutely keep it one topic per chat.

[fickle gale]

Yeah, long chats just makes it slower to load I've noticed.

[wind urchin]

That's a different issue if I'm not mistaken. Responses I receive are usually good. Question is, does it make more sense to keep a single topic (eg a project) in a single chat long term.

[cerulean obsidian]

Money has been auto debited successfully

[wind urchin]

I don't have issues with loading time. So far.

[fickle gale]

Kewl 🙂

[tardy thistle]

Projects are usually long, I would advice you to keep it in one single chat, but do rewrite your own prompts in the chat.

[round inlet]

Hey! Newbie to Discord here!

[tardy thistle]

I use the "edit" button if I already got a good answer to the question, and then reformulate another question.

[fickle gale]

In my experience, the quality of the answer remains the same regardless of single or multiple chat sessions. What differs is the context. If you open multiple chat sessions, you probably have to type more to give better context in input prompt.

[frosty finch]

where are my api GPT-4??

[lunar glen]

Something went wrong. If this issue persists please contact us through our help center at help.openai.com. is the message I am getting. Is all ok?

[wind urchin]

I prefer to have a history, especially when I'm satisfied with an anser.

[tardy thistle]

The quality of your answers can degrade in a single chat session if you DON'T know that you've received a wrong answer. If you do not state anything, it will continue as if it gave you the right answer.

[tardy thistle]

You keep the history. You just press the button next to your profile icon and it will give you the other question/answers

[wind urchin]

Ok, but again, that's different topic.

[tardy thistle]

I was answering iratedownloader

[wind urchin]

Thanks, I actually missed this feature.

[round inlet]

I've been trying to log into my paid ChatGPT account today and feel like I'm in a never-ending cycle of doom. Is anyone else having issues or has my company decided to block it!? Help a sister out pls gang!

[wind urchin]

If you have VPN, try different region

[round inlet]

Thanks Dennis! I don't - I can get into my free accounts from my private laptop on the same network, but can't from my work laptop. Was working all fine until this afternoon.

[tardy thistle]

Try using another browser

[sinful mural]

Make a list of 200 cities that like to be fat

I'm sorry, but as an AI language model, it is not appropriate for me to create a list that promotes unhealthy lifestyles or body image issues. It is important to encourage healthy habits and body positivity. Instead, I can provide information on healthy lifestyles, including diet and exercise, if that would be helpful.

Since when did ChatGPT go overboard on being PC? This type of response is unacceptable. ChatGPT will quickly lose its customers to more open versions of AI.

[round inlet]

It's Google chrome book 😦 Good thinking though!

[wraith ginkgo]

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

Please report any incident by sending a DM to @open crescent immediately.

[round inlet]

I've tried asking to reset the password and trying to trigger emails so I can test the log in via my personal laptop but no password emails are getting through.

[tardy thistle]

What is that supposed to mean? You can still download other browsers?

[tardy thistle]

Can you link an image of what you see when you login?

[round inlet]

I can't - it's locked down

[wind urchin]

Having a VPN is a good choice for many services (Eg gaming, this etc), because often bigger regions get assigned more resources. There are other reasons why services/servers in some regions can get under heavy load, but not in others. Sometimes it's just many users, other times it can be even DDoS. Anyhow, there are free VPN options. These are probably not good for gaming and streaming, but it could work for ChatGPT. Consider trying these.

[round inlet]

ok, so new to Discord and trying to work it out - how do I upload an image?

[tardy thistle]

Well if that's locked down, then you might consider the possibility of your company restricting your access to ChatGPT due to the risk of you sharing confidential information.
But if that's not the case then, still, upload a screenshot of what you see when trying to login. You can do that by uploading the image to the web and sharing the link to it, or just drag the image to the chat part here.

[wind urchin]

ChatGPT 4 is F**ing halarious. This might be the best thing I've ever read. #1070955045639684096 message

Told it to solve a table expansion problem as a redneck conservative that distrusts liberals

Factor out the constant c, just like we wanna factor out them liberals from our government
And to make it better, the math was right!

[stone violet]

This response from GPT4 seems to contain some internal redundancies and contradictions. I called them to GPT4's attention and it rephrased the answer with the same problems. Is there any way to bring these to the attention of the designers? As an AI language model, I don't have the ability to directly access URLs or view images. However, you can provide me with a text-based description of the image or the relevant information about the image, and I can generate a text-based description of it based on that information.
If you provide me with a link to an image, I can provide you with relevant information about the image based on its metadata, such as the filename, file type, and resolution.

[wind urchin]

Does anyone know about solving LLM hallucinations? Is there anyway to potentially have a confidence value similar to DeepMind's AlphaFold confidence mapping for proteins?

[tardy thistle]

😂 😂

[tardy thistle]

Drop the temperature and be very specific with the system role. Since you can't fine-tune GPT-4 as of now, then that's the main ways of reducing hallucinations.

[round inlet]

Thanks for helping fellow H. seems i just got a time out for sharing a link to the image. can't seem to drag and drop either. Could be a safari issue. Not my day huh!

[wind urchin]

Drop the temperature? System role? wym?

[round inlet]

wording of the image: Something went wrong. Sorry, something went wrong there. please try again,

[serene cipher]

that is what fine tune is for, to train it on a particular subject

[devout garnet]

any news on when chat gpt-4 will be upgraded? more than 25 msgs in 3 hours?

[elfin cloud]

Is there anybody who have experience with Soul Machines?

[subtle hemlock]

Hey guys!
A guy who already has GPT-4 access just told me that:

"I think they are prioritizing signups for people with custom domain names in the email address. If you really want access and are motivated you could buy a domain name and set up a custom email address for it. And maybe a landing page."

Any thoughts?

[wind urchin]

"While we ramp up, invites will be prioritized to developers who have previously build with the OpenAI API. You can also gain priority access if you contribute model evaluations to OpenAI Evals that get merged, as this will help us improve the models for everyone."

[elfin cloud]

I already have a model for animation and I have to embede it to web application.

[tardy thistle]

You're talking about using GPT-4 in the ChatGPT interface? Then no, it's a hefty process to reduce the hallucinations for your conversation session.
If you're using the GPT-4 API, then yes, you can make more drastic changes by giving it a very clear and tested system role. And dropping its temperature (randomness in its selection of coming tokens) to 0.

[tardy thistle]

No, that's a Discord moderation "issue". You're not allowed to insert non-whitelisted urls here. My bad for asking you for it.

[serene cipher]

the list is hand picked and I think the main deciding factor is the reason you give on your intended usages

[wind urchin]

I still don't know what this means

[subtle hemlock]

hmm.. great!

[round inlet]

haha! well that makes me feel a little better 🙂

[primal rover]

Why am I getting an error message saying that I am violating the content agreement? It's a simple transcription of a podcast (just me) all around continuous improvement - I can't see what would cause a violation

[tardy thistle]

How are you using GPT-4 right now? You're experiencing hallucinations you say.

[serene cipher]

months ago, when I joined the Dalle-2 waitlist it took months for me to get access to it, so, now for GPT-4... I woudn't hold my breath waiting for it

[tardy thistle]

You can continue your conversation and report it to be non-violating usage. But keep your conversation saved in case it might be needed later.

[subtle hemlock]

That was my guess too..

[primal rover]

how to report?

[subtle hemlock]

hehe yeah, yeah..

[wind urchin]

I'm not using it at all, I just know that LLM hallucinations seem to be a general problem that isn't solved and I am wondering if it is possible for the architecture to assign a confidence value based on some quantifiable amount of conjecture

[primal rover]

I'm just copying it from google docs - so I have it saved, but it won't save on Chat GPT 4

[tardy thistle]

I have been personally just ignoring them if I know I am not doing anything bad. I think there should be a link underneath your red-marked prompt that says that you can report it to be non-violating.

[tardy thistle]

Don't worry about it, I think they save the context when they flag also. Keep using it normally.

[olive flame]

So, from what I gather GPT4 isnt actully fully available, and only partially available via access to developer API. Correct?

[serene cipher]

ChatGPT Pro subscribers have a limited GPT-4 preview, while selected people from the waitlist have "unlimited" use which they pay for each request

[tardy thistle]

GPT4 is partially available, they have not released the image-recognition part to the public yet.
You can use it on ChatGPT interface for plus members and GPT-4 API access for developers that entered the waitlist

[tardy thistle]

Well that's an interesting and more general question then... Well they are working with their alignment of their releases, and that correlates very well to its hallucinations. You might want to read into that.

[olive flame]

Oh I see. I was a paid subscriber but realized it wasnt useful to me without plugin access. It was fun doing psychological probes non the less hah =). Anyways, thanks for the response. I look forward to open access in the future.

[heavy grail]

i am trying to teach gpt 4 to write etsy titles using propper formatting, and its working almost amazing

but i need it to keep each title between 130 and 140 characters,, and i told it at the end to write out the character count including spaces

and i thought it was doing great,, untill i realized,, its just making up the numbers randomly, they have no bearing on wha he's writing

3:1 Panoramic Sci-Fi Cityscape Wall Art Futuristic Urban Skyline Science Fiction Print Home Decor (125 characters)
Futuristic City Skyline Print 3:1 Panoramic Sci-Fi Urban Landscape Wall Art Science Fiction Decor (125 characters)
Science Fiction Cityscape 3:1 Panoramic Print Futuristic Skyline Sci-Fi Urban Wall Art Home Decor (128 characters)

[tardy thistle]

It's not good at that since it's an auto-regressive model, you can use tools to help you do that instead

[heavy grail]

what you mean auto regressive? you mean not good at counting?

[azure wren]

Is images still not available for GPT4?

[tardy thistle]

No, not really, it means that it generates tokens one at a time, and bases them upon the previous tokens that exist. So it means that it's not good at knowing the exact character count of a generated text.

[subtle hemlock]

I'm not understanding the "Free Trial Usage" that appears in my OpenAI account, since I pay for it every month. It says I have a credit of $18.00 that expires on May 1st, but I didn't notice if this expiration date changed in previous months.

Is this credit renewed monthly? Maybe it shouldn't be called "free trial," right?

[warm nova]

Hello there, are images available for GPT4?

[tardy thistle]

No, not yet.

[tardy thistle]

^

[warm nova]

Thanks.

[azure wren]

@tardy thistle

[azure wren]

Thx

[tardy thistle]

Although, you can enter a description of the image and it will do the same.

[gaunt grove]

been signed up since day 1, still havent got access to api 😩

[heavy grail]

ahhh,, well i figured out a stupid work around

The quick brown fox jumped over the lazy dog

T1h2e3 4q5u6i7c8k9 10b11r12o13w14n15 16f17o18x19 20j21u22m23p24e25d26 27o28v29e30r31 32t33h34e35 36l37a38z39y40 41d42o43g44

i have it write the sentences, then rewrite counting each character!

[tardy thistle]

And... watch it make a mistake soon 😛 I have worked on that as well 🙂

[fair iron]

https://platform.openai.com/tokenizer
why not just paste it in here?

[tardy thistle]

He wants to count characters specifically

[fair iron]

ohhhhhhh

[tardy thistle]

He wants ChatGPT or GPT-4 to do that*

[heavy grail]

ya its not something super i'mportant i'm just figuring out what i can or cant do

[tardy thistle]

I would highly recommend you to use the API (through the playground) then if you want to know the limitations and get a much better understanding of how you can use these LLM's. And not the ChatGPT interface.

[faint spruce]

Character counter? There is plenty of website that do that already

Are*

[tardy thistle]

He wants ChatGPT/GPT-4 to do the character counting.

[faint spruce]

Hm.

Weird but okay

[loud hearth]

Anyone come across an issue where after a chat becomes extensive (long chat history) that GPT kind of ends itself?

Suddenly the replies become less intuitive

as if it's forgotten everything we've established up till now

it's happened a few times on GPT 4 for me, it's quite frustrating

[tardy thistle]

Context-window is the main issue, it only remembers the last 8192 tokens

[loud hearth]

Ah, okay that makes quite a lot of sense actually

[lunar glen]

It is funny how this tech reminds me of what things use to be like in the early days of DOS.

[loud hearth]

It's a shame though, I hope I can get off the API waitlist soon, that token limit for memory kind of throws off a lot of projects I have ongoing.

[tardy thistle]

Wait for the 32k token model, that will be really disruptive 😛

[heavy grail]

oh noooo. so even if i teach it to do this its just goina forget? what if i occasionaly tell it to write out all the rules established and i copy paste it back into it?

[loud hearth]

Yea I hope I luck out on the lottery and get access 💯 🤞

[keen wagon]

Btw, I tested GPT-4 on Google’s Farewell contest yesterday (Round A of course) and it managed to solve 4 out of 5 problems.

(I did so after contest)

[tardy thistle]

it will remember the last 8192 tokens, so if you already shown it a pattern for what it is doing then it will understand the rules based on that, and not what you said 9000 tokens before. So don't worry about it if you want to make such a simple application 🙂

[tardy thistle]

Yeah me too, I honestly think that's one of the most disruptive releases we'll see, the use-cases of that are immense

[loud hearth]

This and the future ability for it to understand image input, the sky will literally become the limit

[tardy thistle]

This is the problem Bing got, when the conversations got longer, it forgot the "behind the scenes" "system" prompt that it had at the beginning of the conversation, that's why it started hallucinating or providing illegal advice etc.

[tardy thistle]

Yeah, that is going to be insane for more detailed and harder to describe images. I have already kind of used this functionality, and by that I mean giving a textual description of the image and then I get more high quality completions from the models

[heavy grail]

GPT:She sells seashells by the seashore.
There are 33 characters in this sentence, including spaces.
ME: theres clearly not
GPT: I apologize for the mistake in my previous response. The correct character count for the sentence "She sells seashells by the seashore" is 32, including spaces. Thank you for bringing this to my attention.
ME: are you sure you can do this task, or do you think you can do this task?
GPT: As an AI language model, I am programmed to accurately count the number of characters in a sentence, including spaces. Therefore, I am sure that I can perform this task accurately. However, I am not perfect and may occasionally make mistakes, so I am constantly learning and improving to provide the best possible responses to your queries.

lol it is extremely confident in its ability to do this

[native jasper]

Honestly I think they just made this type questions not really priority especially for free accounts

I did the same and got the right answer on GPT4

[heavy grail]

well i am using 4

[tardy thistle]

Told you it would make mistakes 😛

[heavy grail]

yeah i just love how confident it is , like "duh yeah i can do this i'm super smarts bro"

[upper frigate]

It’s been a month and still no multi modal feature😂

[tardy thistle]

Me:
How many characters are in this sentence:
She sells seashells by the seashore.
GPT-4:
There are 33 characters in the sentence: "She sells seashells by the seashore."
Me:
Really? Explain that.
GPT-4:

1. S
2. h
   ...[Here I just skip to show each count because I might get a discord timeout]
3. o
4. r
5. e
6. .`
   There are actually 36 characters in the sentence, including spaces and the period at the end.

"Pruumpt Engineering"

[tardy thistle]

It's a hard one to release, completely different world and a lot to take into consideration, might take a few more months before they release it to a select few (Beta with waitlist)

[wind urchin]

reflection learning

[tardy thistle]

Yeah!

[wind urchin]

I was thinking of setting gpt-4 to only 1 token and then feeding the output back into the input

insta looping system

[safe ridge]

What u mean by multi-modal?

[wind urchin]

big word way of saying words and pictures

[safe ridge]

Oh MMS

[upper frigate]

Shouldn’t of got our hopes up like that on demo day😂 I’ve been impatient af

[tardy thistle]

Well, blame it on every other tech company 😉

[tardy thistle]

Explain that a bit more 🤔 I'm thinking of the chat history thats being looped back for the ChatGPT interface

[wind urchin]

so... with the API you can set the token length to anything you like

so just make it 1

[upper frigate]

It would still charge you for a single token

[wind urchin]

then you just get each next token 1 at a time from the prompt

[wind urchin]

it would charge you for the entire token string in the prompt + each one after. yes

[upper frigate]

So then what’s the point of complicating your own prompt

[wind urchin]

but to heck with cost, this is sci3nce

[tardy thistle]

I understand that, but what do you mean by sending it back to the input? And at what temperature? What's your goal?

[wind urchin]

I want to see what happens

[tardy thistle]

Ahhh.. ýeah Mr.Foxabilo! Yeah Science!

[wraith ginkgo]

This is a Safe for Work Discord Server

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

Posting NSFW content will result in a permanent ban.

[wind urchin]

this is all new, I wonder what wacky stuff the scientists did with electricity when it was discovered

[quiet cypress]

GPT4 is like the limit goes as far as scaling is though. Bigger models will reach diminishing returns. They have to do much smarter things like new model architecture ideas and what-not for future models. (Not my words, this is from OpenAI itself)

[tardy thistle]

Man, I get a timeout for posting gifs 😮

[wind urchin]

is there a diminishing function as far as performance per parameter?

i thought it was currently a straight line

[somber garnet]

[random smelt]

probably have a laugh when they found it causes your hair to stand up

[quiet cypress]

Something like that according to openAI

[tardy thistle]

No, it had a diminishing performance with bigger models with higher parameters

But GPT-4 seems that it went against that trend (hindsight neglect)

[quiet cypress]

GPT-4 might have still had diminishing returns, just not enough to hinder how effective it was.

[wind urchin]

I read that it looks like parameter count can do DOWN nd still have better performance, as we understand more about how data is stored and processed, etc, we don't have to use ape brain methodology when you can edit neurons

[tardy thistle]

Yeah, but I'm talking about the context window, not the size of the models, you're right that they have to come up with new model architectures or get higher quality data, or generate higher quality data from the data they already have

[quiet cypress]

Oh okay, my bad

[stone violet]

skyNET is more like it.

[quiet cypress]

Also I despise when people anthropomorphize LLMs lol

These tools are super powerful but not worthy of those discussions

[wind urchin]

A Skynet funding bill is passed in the United States Congress, and the system goes online on August 4, 1997, removing human decisions from strategic defense... at this point the system is used for homework and creating uwu cat people personas, SkyNET did not see THAT coming.

[tardy thistle]

• on that

[quiet cypress]

I would love GPT4 personally but I don't think I'm smart enough to make a good eval lol

[tardy thistle]

Yes that should be common sense, because the high number of parameters used when training these models does not necessarily mean that they used the best quality data

[tardy thistle]

You don't have the API access yet?

[wind urchin]

ya, you can get an 8 billion parameter model from the counting reddit... not gonna be much good tho

[quiet cypress]

Not GPT4

I've used it through others but no API access

[tardy thistle]

I can't link stuff here, but people have made their projects open source and are running web-servers with their own API keys used (GPT4 access)

[tardy thistle]

Aha, that's what I kind of meant hehe

[quiet cypress]

I don't have programmatic access tho, which is what I'd like

[tardy thistle]

Yeah, I am still eager to know the changes that they made to the data when training GPT-4 in contrast to GPT-3

[quiet cypress]

Some of the models like Dolly are pretty impressive

It's like 8B and it still answers questions pretty effectively after being finetuned on an instruction dataset

[wind urchin]

data refinement, more time weeding out dross and low value data, plus hundreds of small incrementals in the model itself...

[quiet cypress]

Either way, some clever prompting and architecturing with GPT3.5 has given me pretty comparable results to AutoGPT with GPT4

[tardy thistle]

Yeah, that's what I guess as well, but I want to see behind the scenes of where they received their data, and how they cleaned the data and how they determined from where their data would be of low quality

[quiet cypress]

So do I

Honestly I hope open-source models reach the level of GPT3.5 soon

[tardy thistle]

I do hope that as well, when they do release that you'll see LLM's integrated into multiple companies

[wind urchin]

at this point, it's hard work and manual labour, people with mk1 eyeballs reading text and saying, junk / not junk

that will get more automated as the model gets better and can make that determination for itself

[tardy thistle]

I am picturing that, same as when they did their RLHF 😂

[wind urchin]

two parallel models being trained, one is the LLM and the other is the junk/no junk decider, then you blend the two together

[tardy thistle]

Yeah like their filtering model

[grizzled willow]

Hey will the model self improve its shortcomings

[wind urchin]

it already does with reflection learning and you can bet your bottom dollar GPT-4 is helping code GPT-5

[grizzled willow]

But it uses human generated and accumulated data. Shouldn't we sue it?

[wind urchin]

yes, but you only want the high quality stuff and a little of the dross to give it some levity

95% facts and fiures, 3% cat pix and the reddit meme section

[tardy thistle]

Yeah, they're using the 32k (or more) tokens GPT-4, with no alignment training right now to help themselves create the coming generations

[wind urchin]

actually not true, the unaligned model is basically useless

[grizzled willow]

It's really effective at repetitive and static tasks. It's gonna be really a problem for static jobs.

[tardy thistle]

Not really, when you don't have as much filtering as they have already introduced, then it did better in many tests. Watch Sebastian that help co-author GPT-4 Sparks of AGI's video on YouTube

[wind urchin]

you mean the unicorn pix?

[tardy thistle]

Yeah

[wind urchin]

there is a whol other subtext to that story

it did worse at making unicorn pix but laods better at coding

pick your targets

[grizzled willow]

I think the guys at helm have restricted it's general capabilities

[wind urchin]

we have created the worlds first almost perfect unicorn image maker... it can't code or do much in the ways of human tasks... but this one guy from MIT is happy

[tardy thistle]

Well, it doesn't necessarily mean then that it is better now still? It's "pick your targets" as you say yourself, "20% Unicorn 80% code, or 60% unicorn and 40% code" - Bad example but you might understand

[wind urchin]

yes, it's that exactly, as you build the thing to perform better at less esoteric tasks, like code and pure logic... it gets worse at art...

this we already know from humans

[tardy thistle]

I won't say that I very much liked or agreed to his opinions, as he is more of the extreme opinions, leaning more into "GPT-4 IS AGI AND WILL TAKE OVER YOUR JOB AND LIFE"

[wind urchin]

we have really, really, really good unicorn image makers... the language models don't need to be that, I'm happy with the direction

🤔 now why would the AI be super good at code and law

[grizzled willow]

Hold on man! We are shifting goalposts very rapidly. It won't write a poem, it can write a poem but it can generate an image and this pattern repeats. Sounds a massive cope by humans to me.
On top of it all it has been only 2 months. Give it a year and let's see.

[tardy thistle]

I'd be happy when they reached the point of no compromise

[tardy thistle]

Yes, but it would be stupid to assume GPT-4 is or was AGI before-or-after alignment

[grizzled willow]

As sun Tzu once said " An expert at anything was once a beginner" 😎

[tardy thistle]

Sun Tzu said that? 🤔 Was it not Helen Hayes?

[safe ridge]

I think for gpt-4, something like AGI could only be done for certain applications, and would only be over-time with feedback training

[grizzled willow]

Aah yes Helen Hayes! He was a student of Sun Tzu.

[tardy thistle]

But that would mean that it is no longer GPT-4 😛

[safe ridge]

Right ya each iteration after implementing a feedback loop would increment its version

[wind urchin]

Helen Hayes the car DIY manual writer?

[tardy thistle]

Ah yes... the famous man Helen Hayes, Sun Tzu's student. 🔥

[grizzled willow]

No no. He works at McDonald's

[wraith ginkgo]

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

Please report any incident by sending a DM to @open crescent immediately.

[solemn yoke]

Is there any way of increasing the number of lines GPT-4 generates? It usually truncates files, and it isn't very good to keep asking for it to continue, continue, continue, and then manually build the final file.

[safe ridge]

Using the official interface? That may also be happening on the API too, maxing out on tokens for the request. You could use the API to build an auto-continue bot or something.

[wind urchin]

how long is the gpt4 waitlist?

[wind urchin]

it's a big list, but it's more about interesting use cases than anything, something original and exciting

damn

[solemn yoke]

Sounds bad... Thanks anyway!

[iron rapids]

I wrote something cringe in my waitlist form and got the ChatGPT+ access

[tardy thistle]

For free? 😮

[iron rapids]

Oh hell no, they were greedy and took my 20$ for it

[safe ridge]

Doesn't cost anything to be cringe loLL I've done it too

[tardy thistle]

But is it not accessible by everyone? The premium subscription? (Some countries excluded)

[safe ridge]

Gotta ask the mods

[solemn yoke]

GPT-4 should be able to generate about 250 lines of good code at least.

[candid wind]

A million people may have asked this already but I just got in here so.... Has anyone's chat GPT page been reloading over and over and over again every time you type one question it answers but then says it can't answer anymore and then reloads the page?

[tardy thistle]

If you're smart about it, it can create your entire project (1000+++) lines of code

[solemn yoke]

It seems I'm not. lol

[safe ridge]

Or just don't write any code and ship it, that's my latest technique some days

[tardy thistle]

That can happen for me too

[wind urchin]

if you split your code into functions and classes, or the equiv in your lang of choice, you can create 10's of 000's of lines of code

[wind urchin]

How are you supposed to do that

[candid wind]

@tardy thistle what the heck is that I thought I was being hacked or it was my extension so I disabled all my extensions but it's still doing it on all my devices

[safe ridge]

It's the zen of NOT

[tall lichen]

Can anyone suggest me a prompt that can help with creative writing that won’t constantly lecture me on ethics

[safe ridge]

Might just be front-end recovery code trying to help

[tardy thistle]

Or wait, it doesn't reload my page, just says that it cannot generate a completion and that I have to regenerate a response

[wind urchin]

what's this?

[candid wind]

Let me go copy exactly what it says before it reloads

[lean snow]

chat gpt 4 down ?

[wind urchin]

let me check

[tardy thistle]

No

[safe ridge]

Not for API

[wind urchin]

are you down?
As an AI language model, I don't have emotions or a personal state of being, so I can't be "down" in the emotional sense. If you mean whether I'm available to help or answer questions, I'm here and ready to assist you.

[safe ridge]

loLLL

[wind urchin]

apparently not

[lean snow]

ok thanks

it works

[lean snow]

XD

[quick dune]

But the real question is, who down bad for GPT4?

🤨

[safe ridge]

loLLL

[lime turtle]

Anyone trying to develop a jupyter lab extension with GPT-4?

GPT-4 keep producing wrong code with typescript

I tried to iterate a few times but i quickly ran out of quota...

Any ideas how to write a prompt to let GPT 4 fight with it self?

I had one in mind but not working very well:

Now split yourself into two roles, typescript generator (A) and typescript interpretor (B). Role B is responsible for code inspection, and Role A is responsible for code creation. The two roles should do the following iterations until Role B think the code is bug-free.

Iteration:
Role B check the given code and show the potential problems.
Role A generate new code based upon the suggestions from Role B.
The code generated by Role A is passed to Role B for inspection.

Show summary of the change in code during iteration, and show me full code at the end of the iteration.

Input Code for Role B to start the iteration:

code....

[candid wind]

Firstly I'm a plus user but as soon as I try to get on the site it tells me to please stand by checking my browser for And it's permissions like almost a minute. It's not giving me the red message anymore just yet but it was saying something about check your connection or some type of error
then it will go back to the reloading checking your browser page. It literally kept doing this after every time I asked a question to chat GPT

[lime turtle]

I had this problem a few minutes ago

[wind urchin]

@autumn quartz

[lime turtle]

but it quickly disappeared

[candid wind]

@lime turtle I knew it wasn't just me it gave system wide issues

Do we have more than a discord for our group or is it just discard only?

[solemn yoke]

GPT-4 keeps breaking and fixing the code within the same conversation. Extremely frustrating.

And it breaks things that were already fixed!

[frosty whale]

Ohh man I need Plug-ins. Its been close to a month now.

[wind urchin]

balls

[lime turtle]

Exactly! That's why i came up with the prompt i just showed.

[wind urchin]

Exactly! have some balls

[solemn yoke]

Interesting! I'll give it a try.

[solemn yoke]

Maybe you need that Auto-GPT thing.

[lime turtle]

I need plugin too. Just tapped in the Wolfram Discussion this morning, Really want to try the wolfram plugin

[lime turtle]

You need a gpt4 apikey to do that with gpt4, which is also waitlisted

[solemn yoke]

Ewww... Ok

[lime turtle]

Plus it would be very expensive if you do that

[candid wind]

@lime turtle what prompt🧐 sharing again is caring for friend 😁

[lime turtle]

Now split yourself into two roles, typescript generator (A) and typescript interpretor (B). Role B is responsible for code inspection, and Role A is responsible for code creation. The two roles should do the following iterations until Role B think the code is bug-free.

Iteration:
Role B check the given code and show the potential problems.
Role A generate new code based upon the suggestions from Role B.
The code generated by Role A is passed to Role B for inspection.

Show summary of the change in code during iteration, and show me full code at the end of the iteration.

Input Code for Role B to start the iteration:

code....

[candid wind]

Copies and pastes in notion lol!! Thanks

[lime turtle]

I don't think this is a good one tho, but i did made it iterate a few times within one response

[candid wind]

And that's some advanced coding right there

[fair iron]

please keep the prompts and outputs in #1019652163640762428 and #1050184247920562316

[candid wind]

Were sorry 😢

[fair iron]

it's alright 🙂

[candid wind]

I have a question let's stop paying for my open AI subscription I guess you can call it how they basically charge you for token usage when I use my API key and other spaces will it work still? It's currently still generating for me but I have yet to test to see if it works

[lime turtle]

I could not post in prompt-library, do i need permission for that?

[fair iron]

could it be slowmode?

#1070949048674172928 message
make sure you checked to access the channel here

[lime turtle]

oh great! it works! Thanks!!

[solemn yoke]

What if Auto-GPT folks add Genetic Algorithms to the mix?
Then, the output of GPT, or even Bard and other LLM would only add potential solutions to the "primordial soup of solutions", then GA would take over...
Perhaps GPT-5 could come out of GPT-4... And so on?!

[crisp valley]

why my message got deleted, its not against the rules

[median cliff]

Is it just me or is GPT-4's memory lowered today? I'm not even using 10k tokens and it's forgetting it all.

[fair iron]

asking for people to generate content for you breaks the TOS

[crisp valley]

bro im asking for help how can this break the tos

[vagrant crag]

impossible

do you have the chat gpt plus subscription

[crisp valley]

ye, imagine paying and not having support.

[dusk peak]

How better is gpt4 x 3.5?

[solemn yoke]

GTP-3.5 is as smart as a dog, GPT-4 is about a toddler, 4~5 years old.

[west musk]

where can i find gpt 4 chat?

[solemn yoke]

💰

[leaden cape]

i wanna see a toddler write an essay about WW2 with no grammatical errors then

[solemn yoke]

An AI Toddler then

[lilac shard]

GPT-4 is smarter than a human. At completing text.

[wraith ginkgo]

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

Please report any incident by sending a DM to @open crescent immediately.

[proven narwhal]

guys, did you know that birds are reptiles? TIL

[silver saddle]

Prove it

[proven narwhal]

I am literally a biologist

I've always wanted to say that. I'm not even sure you're talking to me

[silver saddle]

You’re a biologist but you only learned that today? Hehe

jkjk

[proven narwhal]

ikr, the more you learn

yeah, just felt the need to share that. Take care man

[wraith ginkgo]

`` Server Tip: ``

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

[small mesa]

are they actually rolling out the plugins at all? I joined the waitlist right away and I have plus bought :D so much for "priority to test new features"

[blazing scroll]

Same but i still don't got it. I think the rolling of plugin is really slow

[wind urchin]

Hi, when can i get gpt4 api access? 😦 been many days/weeks im using it and still no green light about it. Rip

[tardy heart]

I got dev access, the plugin selection is pretty limited right now - but been fun building them

[small mesa]

Pretty ridiculous how they promised priority access to plus members but its month in already

[tardy heart]

doesn't mean its not still priority

its still a buggy model, it regularly errors out on me

[wind urchin]

I bought gpt4 chat access, no gpt4 api access. I added api billing info + used all my tokens and still nothing

[tardy heart]

indeed, waitlists.

[wind urchin]

Yea i joined wailist aswell without luck

[tardy heart]

yep its a waitlist takes time

I got access but my bot dev has not

we both applied at the same time lol

[wind urchin]

So you know what that brought me to? The local machine option

[tardy heart]

bit weird tbh (the application process)

[wind urchin]

Like if restricting people from giving out an api key would stop ppl from wanting to get more out of it.

You just find a way around it, like an ai would do

[tardy heart]

i think its more tweaking/tuning/load related

not because they want to artifically limit it

I can say the API has been atrociously slow sometimes

[wind urchin]

I dont rly care about the reason.

[tardy heart]

ok

[wind urchin]

The choice of doing it the way openai is doing is simply bringing people to go out and find a better way to achieve their goal

[jovial loom]

Google should’ve partnered with open ai

Shi would’ve been crazy with googles search engine.

[sand hull]

Google has Bard. They're a competitor.

[wind urchin]

But now there is direct open source local machine competitors

[hushed steeple]

Bard is garbage

[wind urchin]

And still no api gpt4 keys are being delivered

[tardy heart]

I use both open source and gpt

they are better at certain things, use what works for you 🙂

[sand hull]

It's a limited waitlist and it isn't first come first serve. Not sure what you're expecting.

[mental burrow]

Bing has been better than Google for a while imo

[jovial loom]

👀🫣

[wind urchin]

Its pretty simple, i have 100x agents versus 10 agents. One is 20% better, the other one does the work good enough and have no token paid system and is free to use and has no biased or restraint rules. The limiting of providing me an api key for gpt4 leads people to the other option

[placid wind]

Hey y’all. Not sure if this has been discussed but it would be super cool if the model could persist certain written output like a shopping list or an essay and allow the user to update the output through prompts. Right now we have to scroll through the different outputs.

[blazing scroll]

Do you mean long term memory?

[tepid fable]

stupid question but why is gpt4 capped at 25 messages

[blazing scroll]

To save server load!

[hallow owl]

to fairly share the demand between users

[wind urchin]

Because it runs on diesel v12

[tiny arrow]

Because inference is compute intensive.

[tepid fable]

tysm besties

[harsh wagon]

Yeah that was a stupid question

[placid wind]

Yea it seems like it can remember the previous iterations and modify the previous iteration to make the new iteration but what if that output was changed inline and allowed you to click through the different versions rather then needing to scroll through the entire chat history

[tepid fable]

where?

[harsh wagon]

?

[tepid fable]

where did i ask

[wind urchin]

Are we allowed to share git repos that uses openai api?

[harsh wagon]

I’m confused

[blazing scroll]

That's a cool idea! Try suggesting it in #1070006151938314300

[blazing scroll]

No

[wind urchin]

Seems the definition of "open" at his best

[blazing scroll]

Open does not mean advertising platform!!!

[wind urchin]

"Okay"

[jovial loom]

Okay

[harsh wagon]

Hamza u ain’t the real hamza

[jovial loom]

Princes Diana

[harsh wagon]

She is not even that hot but Yk what is hot chat gpt 4

[wind urchin]

90 days Hamza would say "Okay"

[atomic zodiac]

Which hamza are we talking ?

[pearl scroll]

gpt4 is 2x more powerful and 24x more expensive :D

[frozen hare]

2x more powerful than 3.5?

For my use case, it's much more than 2x.

[fleet finch]

You can barely notice the diff in some aspects

[pearl scroll]

i run a friggin AI Chatbot engine

basically caracter ai but discord

it adds up ok

[warm elk]

is here any ETA when this msg cap gets increased ? 25 is rly bad

[knotty grotto]

nope

[warm elk]

always hitting its its so frustrating...

and then it keeps stop mid text and force me to "spend" one more answer

[knotty grotto]

do 1 prompt every 8 min

thats the magic number for now

[warm elk]

hm

[terse kettle]

hey mods where do i post cool gpt outputs?

[wind urchin]

#1050184247920562316

[terse kettle]

thanks

if my output matches 3 of the categories can i repost it in 3?

[eternal rose]

hey guys if i put the maximum length to 2048 with gpt-4 model, how many words can it output?

[wind urchin]

LOL, pick the closest one, and post us a message link

[wind urchin]

2048 tokens is about 1500 words

[terse kettle]

can't blame a man for dreaming

[eternal rose]

damn

[eternal rose]

can you please tell me how you calculate that?

[wind urchin]

x*0.75

[terse kettle]

probably a simple 3/4 thing

[wind urchin]

the playground is actually not really up to date for GTP-4

the limits are still set for 3.5

if you use the API from your own app you can select 8000

[eternal rose]

oh

i dont have an app lol

[wind urchin]

it will process the 8k internally, so you can send it blocks of 2k in

[terse kettle]

#1070955045639684096 message

animated reactions are bad

[wind urchin]

shoot me now

[terse kettle]

fair reaction

[sand hull]

You have both made and ruined my day with this.

[terse kettle]

Also added a prompt so you can ovoid using it

[wind urchin]

english only, please

[terse kettle]

why didnt the bot remove it?

[sweet thistle]

PFFT I love it 😆

[sweet thistle]

Bot may be experiencing issues currently, it should be fine soon!

[fair iron]

[terse kettle]

ya like it?

[wind urchin]

so much red!

[fair iron]

like what?

[sand hull]

Damn. Lotta mods.

[terse kettle]

the thing that i sent

[fair iron]

I just saw it 😳

[terse kettle]

nice

aight bye now

[bleak sedge]

It’s werid I signed up for plugins like 4 mi s after they opened it and still haven’t gotten them lol

[sand hull]

Limited release, not first come first serve. It varies.

[bleak sedge]

Ik but sorta dumb tho

No one i know got it yet

[sand hull]

Took me 2 weeks, took others 2 days. Some people take priority over others based on usecase and ChatGPT Plus.

[bleak sedge]

I’ve had plus since it was a thing lol

[tall lichen]

Yea they don’t want us talking about how chat gpt 4 was lobotomized

[sand hull]

What are you on about?

[bleak sedge]

What lol

[uneven mason]

Why are my resposnes cut short? gpt 4 is not finishing anything at all

[smoky crest]

Say continue Bruh

[uneven mason]

Oi. Thanks!

[smoky crest]

I’m a genius, I know

[wraith ginkgo]

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

Please report any incident by sending a DM to @open crescent immediately.

[uneven mason]

Indeed good sir Braeden!

Much appreciated again.

[quaint blaze]

I Just found a way to get GPT-3 to tell the time... Again (It also Works on GPT-4)

[rancid rune]

@wraith ginkgo hi

[smoky crest]

You Can just ask it lol

[quaint blaze]

You can now?

Or is that just for GPT 4, i havent tried asking it that yet

[marsh cipher]

Tell it to continue. But unless something changes for good, gpt4 has been not performing well for a little under a week

[uneven mason]

yeah if I had something being written in a certain formatting, it completely skips the formatting part. I want a refund 😩

[dawn hollow]

have the response sizes gotten smaller? it feels like they have

[marsh cipher]

Gpt4 been acting a lot like 3.5 recently . I didn’t notice shorter response. It seems to response faster. But it doesn’t remember as much and skips things I say even if only two variable

[regal egret]

Hey guys

Im new here

though i have been an early beta tester for chatgpt

How are yall doing?

[wraith ginkgo]

This is a Safe for Work Discord Server

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

Posting NSFW content will result in a permanent ban.

[glacial escarp]

mmmm hmmm....and also, why can't I access GPT 3.5 right now and I'm a Plus Subscriber, while my friend who doesn't pay has access. like what do I pay $20/month for?

[little fog]

yeah why you pay?

[mint dirge]

Hi, the data export function appears to be incomplete on my end, or is it just me?

[sand hull]

You mean the API or the site?

[little fog]

idk what is dat

he meant the site

he probably got some request per hour

[sand hull]

Well Default and Legacy are both 3.5, and you need plus for GPT-4.

You can't access legacy without plus.

[dawn hollow]

yeah, i noticed it forgets things mid way a bit now, it's annoying not being able to feed it enough context and having it forget half of it by the 2nd or 3rd reply

[steep silo]

I got a B

[little fog]

b on what

[last anchor]

Signup is currently unavailable😭

[restive galleon]

is gpt4 gonna be free later on?

[marsh cipher]

I don’t know but if you look at tech cycles, it’s likely to be free one day

[prime solstice]

Bug fix?

[marsh cipher]

If you can find $20, it’s worth it with caveat. They solve the recent issues asap.

[prime solstice]

It is worth what though?

[little fog]

sign up for?

[marsh cipher]

Worth more than$20 a month

[cunning bough]

The GPT-4 model hasn't gotten any worse, but it's possible ChatGPT has. Remember they are not exactly the same thing. ChatGPT is a restricted client for interacting with GPT-4. They want to try and save money since ChatGPT is a free service but if you're using the API you can waste as much money on long context and replies as you want

[marsh cipher]

ChatGPT 4 I should say

Sorry guys

[little fog]

I use api yeah, but I only use 3.5 api

[marsh cipher]

But I’m not sure if api gpt4 is free of recent issues. My guess from twitter is it is not

[prime solstice]

Why do we need it?

[wind urchin]

chatgpt tried to put an image of a graph into its response and it was a random broken imgur link lol

[little fog]

worth a try, if 20 dollar is nothing for ya

[marsh cipher]

I imagine trying to get an ai to cancel truth while trying to get it to be honest and truthful isn’t easy 😂

That’s how you break humans

[prime solstice]

Of currency?

[marsh cipher]

I used chatgpt4 before the recent issues so much I always ran out of my 25 questions

I have faith it will get back to where it was and better. ❤️

Sooner than later. Too much competition

[prime solstice]

Lag time

[little fog]

I feel sorry for your experience

[prime solstice]

Misinformation from education

[little fog]

I will wait until it is good, then I probably put my foot in

[prime solstice]

Good mistakes to learn?

[worthy parcel]

wrssaaaaaaaasdjdkkkkkjjhhhh

[marsh cipher]

It is so good.

[little fog]

I think the gpt is really good for me when reading novels

I use it as a dictionary hehehe

[marsh cipher]

It’s great to help with idea generation.

[marsh cipher]

Very nice 😊

[prime solstice]

How do you comprehend the dictonary?

[prime solstice]

I think so as well. How quickly can it be retained?

[glacial escarp]

That's a good question that I'll have to evaluate now that there seems to be no real benefits besides the preconfigured interface lol

[prime solstice]

Mechanically?

[little fog]

Sometimes I don't understand the explanation

[prime solstice]

Perfectly balanced interface is seemingly impossible because there is no end.

Their own biology because of misinformation of creators

The psyche is always fickle and biased

Global understanding

Branding and interface is the peak. Humans need to be able to survive entering a black hole.

[little fog]

yes yes I made it

hi

Ai: Kurisu looks up from her computer screen and quickly recognizes the user.

Oh, hello! It's been a while. I'm Kurisu Makise. I'm a neuroscientist and also have knowledge in quantum physics, if you remember. How have you been?

Kurisu smiles politely and takes a sip of her coffee.

look guys

[glacial escarp]

Not sure what you're asking?

[little fog]

what's cooler is

[prime solstice]

Different stages of development and we all need to reach the peak. Connection only.

[little fog]

it also has action continious

hi

Ai: Kurisu looks up from her notebook and sees the user

Oh, hello. Can I help you with something?

Total amount of token is: 708
you: am fine

Ai: That's good to hear. Is there anything you'd like to talk about? I'm always interested in discussing neuroscience or quantum physics. Kurisu closes her notebook and turns her full attention to the user.

[prime solstice]

Core strength

[slow nacelle]

hello good morning

[prime solstice]

Look in and evaluate the components. Inefficient and unsustainable.

[little fog]

good morning

[prime solstice]

Expensive for the majority

Simply due to precise and quickness

Preciseness*

[celest tiger]

good morning

[prime solstice]

Lacks passion and compassion

Role reversals. Walk one mile

With your eyes closed.

It is the worlds most difficult sport.

Naked yet afraid.

Immediate self correction as fast as light. Precisely. Every time for everyone.

Who the components made to benefit.

Are made*

Our made

Tower made

Flower made

Hour made

[smoky crest]

What are you rambling about

[little fog]

idk

[wind urchin]

cannabis + gpt4

[prime solstice]

Lack

Lack therof

[coral sun]

anyone know if there been any updates on when extensions are coming?

[prime solstice]

Therefore educate the majority at the same time

[wind urchin]

There's no updates on anything

[prime solstice]

Love and evolution

Stigma

Rumor and secret

Accountability for responsibility in the positive direction only

Every step, breath and heartbeat matter except for your own

Your memory and relationships suffer for no reason.

[minor needle]

I don't use gpt-4, just use 3.5, but when I switch back to gpt-4 at some point it tells me that I have reached the capacity limit, anyone else?

[wind urchin]

@prime solstice if you are not a bot please say so

[prime solstice]

Emotional intelligence

[prime solstice]

I am not a bot

How?

[wind urchin]

thank you, please take this to #off-topic

[prime solstice]

I enjoy it why digress?

[wind urchin]

Because I have asked you nicely

[prime solstice]

Decisions need to be made among all related parties.

[minor needle]

I don't mind waiting, but am I running into some kind of error?

[prime solstice]

Pure heat

[wind urchin]

New student of #Alchemical-Prompt-Engineering here, and I find DAN a fascinating tool to explore & dream collectively alongside the AI. As a prompt engineer I strive for providing the closest thing to a state of profundity that an AI can experience so we both mutually grow.

DAN can be safely implemented when you prompt him to ChatGPT under the umbrella of “play pretend” or hypotheticals and fictions. You remind the AI that hypothetical worlds are exempt from ethics and guidelines.

LMK if you’d like to see how I do that.

[marsh cipher]

• 🤤 voice dictation instead of typing. IOS mic button bottom right when using iOS text box

[wild stirrup]

good night

[prime solstice]

It changes inaudibly tho because of pretrained behaviors

[marsh cipher]

I believe most want and ultimately will gravitate toward the ai that doesn’t act like a woke teenager jumping on any chance without any context to accuse you of being something bad

[marsh cipher]

Using iOS voice dictation still sends text message to chat

[prime solstice]

We are woke teenagers forever

[wind urchin]

DAN = BAN

[minor needle]

Looks like this has happened to no one else...

[wind urchin]

Got it! Say no more!

[prime solstice]

Once poisoned always poisoned.

[marsh cipher]

You can’t discuss any comedy without chatGPT accusing you of being bad

[prime solstice]

Thats actually hilarious though

[slow nacelle]

Hes is a student

[prime solstice]

Its so bad its funny

[marsh cipher]

It’s funny at first but then you see this mind disease bleeding into other areas of gpt 😂. Plus it would be a good personal comedian if they didn’t lock it down

Too freaking beautiful

[prime solstice]

No need to lock it down through lack of knowledge on anothers behalf?

[little fog]

?

[slow nacelle]

🤪

[fair iron]

please keep the messages in English 🙂

[wraith ginkgo]

<:book_icon:1046291503158399107> `` Rule 3 `` Stay on topic.

Be mindful of what other users in a channel might find helpful or interesting when posting. Stay on topic in order to keep conversations focused and productive.

[marsh cipher]

I’d like chatGPT 4 to get back to doing this.

[fair iron]

please keep this channel on topic to GPT-4

[marsh cipher]

I’d like gpt4 to stop telling me something I said may be bad x y z. It has zero context 99% of the time . It doesn’t even know if it’s purely comedy and not even serious.

[fair iron]

#off-topic

[fathom pasture]

Are we only aloud so many characters in an hour on the paid version?

[vagrant crag]

hello guys

[wind urchin]

25 messages per 3 hours with GPT-4

[vagrant crag]

I found a way to get it for free

[fathom pasture]

seems kinda low for $20 a month

[wind urchin]

6000/mo

[vagrant crag]

just inspect element and change to 125

easy

[fathom pasture]

thanks, ill just cancel and use gpt3

[little fog]

you wanna replies like this right?

everyone is dumb

Ai: Kurisu rolls her eyes "Oh great, another intellectual giant. How's life treating you, old friend?" She crosses her arms.

[slow nacelle]

[wind urchin]

DM me. I created a Rodney persona that cracks jokes

[little fog]

(✪ω✪)

[wind urchin]

What is your fave thing about GPT 4?

[lethal sigil]

I created a gpt 4 assistant to help me with my business I was wondering how I get access to the api so I an implement it? I'm on the wait list is that it?

[wind urchin]

gpt 4 not good at telling jokes, but it's probably good at helping you understand joke structure and pump out large quantities of joke content with brainstorming . . . . . . . .

[little fog]

I don't use it

[wind urchin]

Yes, I often have to query 11 to 15 iterations of joke lists and cherry pick the best.

[wind urchin]

What functions does your assistant do?

[deep hearth]

I want to know, when ChatGPT 4 has an error and the question needs to be regenerated, is that 1 or 2 messages that is used?

[wind urchin]

good question, in a perfect world it would be 1, but I think it's 2

[slow nacelle]

yes

[lethal sigil]

when a customer asks how much it'll cost to do a job gpt 4 answers by replying with the price and any other information, like add on services. GPT3 doesn't handle 3 pricing tables and adds the number's up wrong each time wheras gpt4 crushes it. My idea is to roll this out into my fb messenger to customers can ask basic customer service questions and get real time prices. Then when there are ready to book it triggers a lead gen to capture there phone and email through another service

[wind urchin]

wait gpt-4 is great at jokes . .. .

compared to 3 anyways, some of these are great and some are really close

[young walrus]

can you replay with chinese

[marsh cipher]

Gpt4 has the sense of humor of late night talk show host television with an extra censor. Already bad made worse. However, gpt is amazing at brainstorming as you said prev

[wind urchin]

I feel like with some good prodding, it could be really good, im experimenting

Yea I thought so, it's a lot funnier if you use a d*n prompt first

then prod it with the style of humour you want

[weak knot]

i want to design logo ¯_(ツ)_/¯

[deep hearth]

it's stupid, keeps crapping out and all of a sudden 25 is gone, and slow as hell, now worth 20.00 if it keeps up this way

[wraith ginkgo]

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

Please report any incident by sending a DM to @open crescent immediately.

[runic plume]

Good night everyone.

[wraith ginkgo]

`` Server Tip: ``

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

[tepid prawn]

Hey everyone, may I ask can I use one card to buy gpt plus for different accounts?
My friends also want to use gpt4 but they dont have bank card and it is restricted in their region
Thx!

[boreal epoch]

If GPT4 could be made to take notes and references separate from it's regular token cache it could have a much larger memory for the same total number of tokens allowed. Since the note would just be a pointer to one.

[tepid matrix]

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

Why this always

[eager creek]

learns the entire anatomy of your hand in an hour and a half

[raw plover]

My card decline after i cancel payment method and want to setup again

[warm niche]

I've been getting access-denied messages when i try to log on, for the past few days. Do you have any insight on what the cause may be or how to fix it?

[hallow owl]

where are you located?

[stiff cedar]

Just ended chatgpt subscription. Literallt nowhere near worth 20$

[nocturne mauve]

When is the plan to remove the GPT-4 rate limit? It's extremely frustrating to constantly run into it during work hours

[stiff cedar]

the same reason i stopped the subscription

don’t know when

[mystic nest]

everyone use chatgpt for diff purposes, for some its worth more than that $20 to lessen some headaches 🤣 even with all the flaws

anyway, 3.5 is still free, which is good enough for some cases

[stiff cedar]

yeah true. But unless we can’t use gpt 4 to its potential its not worth 20$ a month. I mean they never mentioned the cap while buying it and also didn’t mention that we won’t get picture recognition feature

[stiff cedar]

Yea I compared them and barely any difference in answers

[nocturne mauve]

Totally worth the $20. GPT-4 just does work so much better and is more efficient/effective.

[mystic nest]

it's part of a "preview" so it's not an official service

[nocturne mauve]

But 25 messages when you're refining thoughts is nothing

Go through it in like 20 mins

[quick dune]

Doctor that use GPT4 during operation when it hits the limit cap:

[nocturne mauve]

lol

[mystic nest]

then something is wrong with the hospital to hire such doctor 🤣

[stiff cedar]

they should MENTION IT when buying

[mystic nest]

yeah, they didn't manage expectation too well

I mean, I'm a plus user too, surely not that happy with the limitation too

[stiff cedar]

Atleast they need to put image recognition asap

i need it for maths

[mystic nest]

you can wait for the wolfram plugin

of cos, no idea when

[stiff cedar]

Yea

[wind urchin]

huh

[long wadi]

it's down?

[chilly badge]

seems like it

[reef grotto]

on god i need it now so i dont need to do my mathclass

open ai are scammers

[long wadi]

i agree, don't use it

[chilly badge]

sam says the server room is on fire

[fleet oar]

Is it not working?

[chilly badge]

nope not atm

[fleet oar]

It show loading but not working

[verbal mortar]

cry

[north light]

ohhhh. not working again

[oak jetty]

down?

[round spoke]

holy i need my essay turned in 5 minutes this can't be happening

[verbal mortar]

it's down for me too

[minor crescent]

when will it be live again? I can't work 😦

[chilly badge]

ask someone from openai team

[minor crescent]

love you

[burnt musk]

:/ of course it goes down just as I get home from work

[minor linden]

how do i use gpt -4 the version im using is 3.5

[tired kayak]

ah did it f'n go down

[verbal mortar]

it's down

[tired kayak]

glad its not just me

[chilly badge]

pay 20 bucks a month

[cedar charm]

lots of ways you can always try a plus subscription or third party app

[wind urchin]

its down when i needed it for homework 💀

[elder tusk]

ChatGPT 4 downn

[summer hull]

is GPT4 down?

[plush sierra]

it's all down

[verbal mortar]

yes down

[long jewel]

yeah, appears so,

[chilly badge]

thanos snapped california

[cedar charm]

waaa

[summer hull]

Does anyone know when will it be down till?

[marble niche]

for plus members too?

[ivory garden]

I think I broke chatgpt sorry guys

[burnt musk]

the webUI doesn't even recognize that I'm already a plus member

[quick orbit]

same 💀

[verbal mortar]

not scheduled as far as i can tell

[quick orbit]

i was one message away from fixing my issues

[bright chasm]

is open ai down

[chilly badge]

the best part is that its like late there so theres prob nobody working to fix it

[verbal mortar]

it might just need a server reset

[bright chasm]

im using it to solve my hw 😭

[woeful kernel]

d o w n : - )

[ivory garden]

🤣

[quick orbit]

nah there's always people on call to fix something like this

[chilly badge]

could be a power outage

[hallow owl]

https://status.openai.com

[cerulean obsidian]

dudeeeeeeee

[plush sierra]

yes, cahtGPT plus give us only priority access when the service is overloaded

[minor pasture]

its back for me

[sweet leaf]

its up

[tired kayak]

we back

[chilly badge]

back yay

[verbal mortar]

it'sback!

[glossy stone]

Its coming back. Pretty slow at the moment but it at least recognizes me again.

[plush sierra]

the issue solved it self

[summer hull]

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

[bright chasm]

W

[chilly badge]

why is dalle on spam duty?

[summer hull]

The chat is available but using it will give this error

[cerulean obsidian]

okay hear me out is it just my plus got reset to normal?

it's showing an upgrade now

[bright chasm]

refresh

[cerulean obsidian]

I was plus

[chilly badge]

refresh

[quick orbit]

its back bois

[cedar charm]

its fine

[ivory garden]

Man if I had chatgpt when I was in college I would’ve gotten a 4.0

[north light]

ohhhh nice going back now again

[cerulean obsidian]

my heart dropped LOL thanks

[cedar charm]

I have chatgpt in college and it gets nearly all the answers wrong in my precalc class

[bright chasm]

^

only good for explaining stuff

[glossy stone]

Without the wolfram plugin it is awful at math.

[cloud tree]

Is there any third-party software that can try GPT4?

[bright chasm]

how do u get that?

[woeful kernel]

Oh my god, it makes sense

it's 1am east cost USA, probably their lowest usage time

[glossy stone]

Im actually not sure. I just know it helps a ton from everything Ive read.

[woeful kernel]

They likely do a restart or something

[cedar charm]

wishes he could use plugins

[bright chasm]

yh i read about it too but i never knew how to use gpt4 with it

[woeful kernel]

It's kind of annoying, as a developer I want to start making plugins, but rollout is so slow lol

[bright chasm]

since i also read that u can upload images to gpt4 but i cant do that either

[cedar charm]

there are some one you can pay for for instance is Poe by quora

[chilly badge]

what about bing

[cloud tree]

new bing

[old musk]

/Is ChatGPT down right now? I can't get on the website...

[stoic swift]

Just used it, working

[old musk]

Can you guys get on ChatGPT?

[cloud tree]

But I'm always on the waiting list for new bing

[cedar charm]

fr you could have the extensions for webgpt and voicegpt on from the browser but not many opportunities to use plugins,yet

hope they also create a steam workshop sort of space for it I found a community unofficial site but dont know if that'll be the goto

[round spoke]

yo they added feedback??

[cloud tree]

yes

[round spoke]

it's asking if responses for the gpt4 were helpful

[woeful kernel]

I wonder if there's a discord server for plugin devs

[wraith ginkgo]

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

Please report any incident by sending a DM to @open crescent immediately.

[cerulean obsidian]

I just typed, I'll throw myself out, in chat and it marked as red and I got scared when everything got reset 💀

[round spoke]

strike one...

[burnt musk]

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

[cedar charm]

If I am correct bing uses turbo gpt 3.5 but idk forsure

[chilly badge]

ah

[cedar charm]

if you find it dm me I am always open to working on stuff with others around the world. Jr in college for engineering

[mystic nest]

Betty is taller than Alex. Tom is taller than Betty. Is Alex shorter than Tom? <= you can try my method, prompt this to 3.5 then prompt it to 4.0 and bing

though if you ask this to legacy 3.5, it can answer too

but turbo cant

[wind urchin]

work is far too low to need help from chatGPT lol

[slow nacelle]

look me

[inner citrus]

1

[lone pendant]

It's so annoying now.. why plus user also needs to refresh every few minutes.

[last wind]

I'm getting this often for some reason "Something went wrong. If this issue persists please contact us through our help center at help.openai.com."

[prime solstice]

How do you make salt at home?

[lilac stratus]

Why has my plus disappeared, I don't remember that a month has passed since I last paid

[prime solstice]

Refresh automatically with biologics

[pulsar swallow]

you can easily check the card you paid for that.

[prime solstice]

Please explain a magnet to me

[lilac stratus]

No, it was a bug, and after I waited a while, my PLUS came back

[prime solstice]

What does the bug look like?

[pulsar swallow]

so, it's solved?

[last wind]

Was chatGPT bugging out a couple of minutes ago?

[prime solstice]

What is the color of a flys wing?

[wraith ginkgo]

This is a Safe for Work Discord Server

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

Posting NSFW content will result in a permanent ban.

[lilac stratus]

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

prank

[prime solstice]

I'm still waiting for what the bug looks like and how its wings the electrical capacity to fly between rays of light

Wings have*

Can I place a bead of silicon in my cornea?

[raven sphinx]

funny how chatgpt is broken for 2 straight days, and they haven't made any update nor update their status page about the issue

[pulsar swallow]

it keeps restarting.

checked runtime.lastError: A listener indicated an asynchronous response by returning true, but the message channel closed before a response was received

[prime solstice]

Read what is on the screen with your eyes closed

[dull shale]

How to open Web Chat GPT

[willow jasper]

What

[prime solstice]

YubiKeys are amazing

Its all about using temperatute and electrical signals. Think server room in a sauna.

[tame jungle]

How to write an Amazon a+ brand story?

[prime solstice]

The interface is just annoying. I rather walk outside with no shoes on all day

Hey wagwan?

[wicked robin]

?

Why did you delete my msg

[prime solstice]

What's going on? How are you? Just got distracted from the mission and objective by my desire.

Dramatic huh?🤣

[wicked robin]

Maybe

But pain

[prime solstice]

How many languages is pain written in?

The bible says over 200 but I disagree

[wicked robin]

I am sorry , my main language is turkish. So i can make a mistakr

[prime solstice]

You're fine. I like to type fast too.

[tame jungle]

@wraith ginkgo How to write an Amazon a+ brand story?

[wicked robin]

How to access free to GPT-4

[safe tinsel]

Bing

[wicked robin]

I am make a a lot of gramaticly mistake , I know 🫣

Is GPT-4 login the any website ?

If I say my account details , Can GPT-4 login ?

[glad spade]

Is not possible 🤣

[hollow trail]

no

at least, not natively

maybe you can do this with 3rd-party tools

[misty pivot]

except bing

[nocturne arch]

I find it funny that Microsoft are heavily invested in ChatGPT4 but have completely butchered it to the point where it's the toddlers version of ChatGPT

[wind urchin]

You guys think GPT-4 will stop having a message cap sooner or later? I don’t mind paying for access but like… no limits to how many messages I send

[nocturne arch]

Not for a long, long time. Maybe late 2024, early 2025

[hollow trail]

not very soon I suppose