community-help

[mild valley]

Even for the Plus account users, they paid for the service , BUT do not get proper service yet

[near pulsar]

#gpt-4-faq answers 99.99% of questions

[tacit stone]

Yeah I dont know whats going on, they just dont want my money

[urban depot]

Image input is currently not available for the public right

[tacit stone]

Support won't answer as well

[urban depot]

What do you mean by this

[near pulsar]

Point proven

[tacit stone]

People who pay or want to pay dont get the proper support

[mild valley]

I think MircoSoft is their TRUE customer, and paid the bills.

[tacit stone]

They haven't answered my in 2 days

[urban depot]

Right, perhaps it's not ready for the masses yet

[near pulsar]

I mean point proven that 99% of questions are answered by #gpt-4-faq

[rigid hare]

I'm new her

can anyone tell me how to use this

[fossil adder]

So I have a session open with ChatGPT. Is there a specific reason that it is unable to recall messages from the same chat tab after closing out of that specific tab?

[near pulsar]

Recall information from a chat after closing it?

[fossil adder]

yea, like if I have a chat open with the GPT and close the tab and open it up later and ask about something from earlier in the conversation it seems to not be able to recall that info

[tight pine]

Hi Administrator, can you add checking option to the payment method in the future?

[untold perch]

This could be the cause if it was a long conversation https://help.openai.com/en/articles/6787051-does-chatgpt-remember-what-happened-earlier-in-the-conversation

[surreal grove]

Can you me give some lead of plumber in usa

[tepid oak]

Ask gpt

[bitter flume]

Maybe u live in a country that is not supported?

[rugged wing]

GPT-4 about using robots with AI for colonizing Mars (or any other planet). Send this to Musk and Bezos, the will understand...

[hazy yarrow]

Hi, I live in Hong Kong and openAI isn’t available there. What is the best way to access GPT-4? VPN? Willing to pay

[oak fiber]

I can't log in with my pro account. I want to cancel him because the credit card is bound, what should I do?

[near pulsar]

don't try to bypass set restrictions

[rustic ingot]

Why the 32k version and visual version of GPT-4 are not available ?

[near pulsar]

what error do you get?

[near pulsar]

to bully you

[covert chasm]

how can I upgrade to chatgpt+ for 20$ a month without having a credit card? why not PAyPal available? 😦

I want to give u money, but you only accept Credit CARD 😦

[oak fiber]

@near pulsar

[near pulsar]

people will chargeback if they buy through paypal

[covert chasm]

But Im stupid german and don't own an credit card 😦

[near pulsar]

try to clear cookies/cache, if that doesn't work do as instructed in the message

[hazy yarrow]

Are there non-bypass ways to use it then? Like available services that are powered by gpt-4, like you.com’s youchat (I don’t really know what it is lol)

[oak fiber]

cleared cookies/cache, this didn't work

[near pulsar]

yeah maybe, you can look into that

[near pulsar]

tried on another device or browser?

[oak fiber]

Yes, it won't work if you change 3 devices

[oak fiber]

I changed the browser, still the same error

The pro account is bound to a credit card. If I can't log in, I have no way to unbind it.

[hazy yarrow]

Surely there should be international third party web services that connect to gpt-4 that charges money, no?

[covert chasm]

But i dont want to use 3rd party apps

[neon stump]

Hi !
I'm trying to log in ChatGPT on the extension serveice on IDE. But I sign uped with Google OAuth, So I can't log in Open ai on the extension. Then I tryed to change log in method from OAuth to password.
I sighn up once again same Email and new password. but error occur.

How do I change log in method from OAuth to password? Anyone knows?
Error message is fallowing
"You tried signing in as "My Email Address" using a password, which is not the authentication method you used during sign up. Try again using the authentication method you used during sign up. (error=identity_provider_mismatch)"

[viscid mural]

Hello, How can I send photos to chatgpt4, this was a new feature right ? Or did I misunderstood?

[pastel field]

It’s not out to the public yet as of now

[viscid mural]

ah ok understand, Nay new son when this will happen ?

*any news

[near pulsar]

what extension?

just login through google on https://chat.openai.com/chat

[pastel field]

No update as of now unfortunately

Hi, does anyone know how to get chat GPT to consistently generate content into a copy code

[lament apex]

Yo guys
I was thinking on buying gpt4 for a month
U think that is a good idea?

[near pulsar]

how can we decide that for you

[lament apex]

If you have used it before

[near pulsar]

idk what you will use it for

[lament apex]

Then you can tell

[near pulsar]

if you use it for my purposes, then buy it I guess

[lament apex]

I was thinking of using it for making some small apps and websites

[near pulsar]

you have much expierence with html, css, javascript?

[lament apex]

Yes intermediate

[near pulsar]

then it can be helpful

[neon stump]

@near pulsar
It is an extension for Rider! I can't paste the URL, so I'll write the URL as a string. "h ttps://plugins.jetbrains.com/plugin/20603-chatgpt"

[lament apex]

Just out of curiosity, what was your purpose?

[pastel field]

@near pulsar do u know how to get chat GPT to write in html/JavaScript without all the fuzzy and irrelevant stuff like the “<>” and “/“

[near pulsar]

Generating essays and specific questions about research papers

[pastel field]

I want it to generate copy codes so I can just instantly copy and paste the generated content but it seems to struggle to either remove the jargon or make the copy code

[reef moat]

Is there a support channel

[lament apex]

I want to buy gpt4 to either break even or make profit 😂😂

[oak fiber]

@near pulsarWhat should I do in this situation?

[stiff pumice]

Any workout yet for India payment issues?

[upper pebble]

so i cant use 3.5 turbo engine with my trial credit ?

[oak fiber]

You can not let me use it, but you can't let me untie it

[near pulsar]

Idk brother

If it's on 3 devices it sounds like it has to do with your ip

[oak fiber]

My IP has also changed. U.S.

[tidal nimbus]

any solution to this? persisting for a few days on the PC...
already removed browser history and all that stuff...tried opening across three browsers and still this issue

[near pulsar]

Hopefully it will fix itself

[oak fiber]

There is no other way?

[near pulsar]

Sounds like you've tried everything else

You can mail them but their response time is slow

[oak fiber]

What is their mailing address?

[near pulsar]

Unless you are using a VPN

[oak fiber]

I am using a VPN

[near pulsar]

Well... Obviously that's why

[rigid musk]

Disable proxy, disable VPN. Try incognito browser

[tidal nimbus]

not using VPN...
will try disabling proxy and incognito browser

or does using the laptop via wifi could also cause this issue???

[near pulsar]

Did disabling the vpn work?

[oak fiber]

I'm in China, if I disable VPN, I'm afraid I won't be able to get on

[near pulsar]

Then it's quite obvious why you don't have access no? 🤣

[rigid musk]

If you using wifi from work or schoop they can block access

[rigid musk]

We can't help you. Btw using a vpn for restricted country is against ToS and bannable

[tidal nimbus]

It's home wifi and sometimes the mobile hotspot

[near pulsar]

I guess he doesn't care about a ban if he can't use it either way

[rigid musk]

Problem is later if the country isn't restricted his account still ban and probably futur account if they find it is alt. Just saying cause people already banned for that

[rigid musk]

If you disable wifi it is working ?

[oak fiber]

🤣

[near pulsar]

Yeah that's a big IF

[rigid musk]

Anyways we can't do nothing let's move on

[tidal nimbus]

it's not working from both the home wifi and the mobile hotspot

[rigid musk]

On your cellphone does it work with data or every tech device you have is block ?

[tidal nimbus]

yeah, it actually works on the cellphone tho it isn't maintaining the history for the time being

[rigid musk]

If this is working on your cellphone something blocking the connection on your pc.

I suggest try with a fresh new browser like Firefox.
Make sure every extension is disable on current browser some of.them block connection

Disable anti-virus like avast or avg they also block connection

[tidal nimbus]

thanks...
I will try this approach again today...

[rigid musk]

Sorry if I can't help more. If everything I said doesn't work you can wait till another user have another solution or contact [e-mail]

[tidal nimbus]

I hope it works...
and I appreciate all the help...I had asked about it yesterday too but you were the first person to help...
thanks again

[rigid musk]

Let me know if something worked :3

[tidal nimbus]

definitely

[ivory bolt]

Yes thank you I understand the at. The issue is the character limit ( chat GPT 3.5 only ). So I rather it write one chapter instead of a whole book still in one go lol. It would still write all of those chapters and end 🥹

[rigid musk]

Tell him you don't like

[hazy yarrow]

If Bing Chat is based on gpt-4… is it better or worse?

[dense hound]

How about stop talking about bing in a open ai channel?

😂

[rugged wing]

if you use GPT-4 do u only have 100 messages?

I thought I saw that the limit was 100 messages, but that is now gone.

[dense hound]

For plus

@rugged wing check dms

[rugged wing]

Ah sorry, I see it now correctly. You have 100 messages every 4 hours.

[dense hound]

Nvm I can sent images here

[rigid musk]

#gpt-4-faq 100 message and then cool down for 4h

[rugged wing]

yes! 😄 Thank you @dense hound

[rugged wing]

haha, I did not see that. I do not use discord very often so I am still learning. Telegram has just 1 chat 😂

[rigid musk]

No problem :3

[muted sphinx]

oh wow theres a limit

[ornate patio]

Can i put photo of math problem and expect to solve it?

[near pulsar]

#gpt-4-faq

[ornate patio]

Ty

[severe sand]

Is there any group in Spanish?

[dense hound]

Not in chat gpt

He’s with the api when it’s released

[tropic wind]

just wonder but it is possible that chatgpt messup the output and give anwser that not related at all?. like i just ask chat gpt to help me find a bug in my program. all work well until now. that somehow half of the code is turn into something i never ask at all. like im working with microphone. in new chatroom. but suddenly it now show me a complete code for motor speed that i even never asking that in my life before. it not even have my code inside it. but it does 1 one thing in common. and that is i use "dummynetral" as variable and this code that complete also have that var inside once
im very confuse right now?

[vivid shale]

GPT4-32K: Have you tried translating computer programs in one programming language into idiomatic code of a differing programming language. If this works well this would be an in demand use case.

[tropic wind]

what?

[zenith kiln]

Does anyone have the invite to Midjourney?

[rigid musk]

#ai-discussions

[dense hound]

No

[tropic wind]

is invite you mean the discord invite?

[dense hound]

That’s what he wants

People need to learn to figure stuff out themselves

[woven hull]

Lmao

[zenith kiln]

Thanks👍

[peak lark]

Does ChatGPT not take context into consideration anymore? It's been replying "I'm sorry, but I'm not sure what you're referring to since there is no context provided" and then "I'm sorry, but I don't have access to any previous messages or conversations as my memory is reset after each interaction"

[worn lynx]

Anyone else seeing response codes of 500 for https://chat.openai.com/backend-api/accounts/check and https://chat.openai.com/backend-api/models and https://chat.openai.com/backend-api/conversations?offset=0&limit=20?

[oak fiber]

Will PRO automatically deduct money every month?

Or after confirming the deduction

[remote spruce]

Guess its broken now for plus users- cant even get personalized link to work.

Oauth doesnt work anymore on login

[drowsy violet]

Is there anyway I can contribute to OpenAI, particularly by helping them with POS tagging/processing/cleaning data in Tamil?

Particularly literary and Old Tamil as well, since I am familiar with it

[hearty sinew]

Hey everyone - question regarding GPT-4: I'm trying to figure out if there is a good way to use GPT-4 (besides on a web based demo). For example, can it be used commercially some how to write articles, marketing etc.??

[steel walrus]

Why always out of work?

[hazy robin]

Has anyone made an easy to use version of Whisper yet? And how would you say it compares to machine translating that was done earlier for media?

[mellow walrus]

Am I understanding requests per minute correctly? If I wanted to make a game where NPCs are powered with GPT-3.5 (like the many that already exist) and it somehow got 1k concurrent players, all talking to the NPCs, does this mean they’d only be able to send 3.5 messages per minute? Meaning it’d be completely impractical for a larger number of people? (5k, 10k)

[lilac gorge]

So how can I upload pictures to gpt 4? I even asked it if I can do it and he said je doesn't have that option ATM. I don't see where I can do it

[exotic field]

It's very busy rn

[umbral hare]

Buy plus

[rugged wing]

Hello there!

I hope everyone is doing well. I was wondering if anyone else has experienced issues with accessing ChatGPT Plus after making a purchase.

I recently bought access and everything was working perfectly for a couple of hours. However, the next day, after signing in, I noticed that the "Try GPT Plus" badge was back and it was showing that I have a free plan. I have been trying to contact support for the past three days, but unfortunately, I have not received any response yet.

Could anyone please share any tips or advice on how to resolve this issue? Any help would be greatly appreciated. Thank you!

[plucky sandal]

Any mods here? I am unable to subscribe to gpt plus

[rigid musk]

Mod couldn't help you on that one. You need to contact [e-mail] or you can try to ping @burnt galleon

[plucky sandal]

Thank you!

[desert citrus]

when is GPT4 coming to the playground? what was the point of the update?!

[rigid musk]

Plus subs can use it. Free have to wait

[desert citrus]

I pay like 40-150 a month

the playground is anything but free

[rigid musk]

The subs is 20$

[desert citrus]

why? I dont pay to talk to a chat bot, I pay to have the versatility of the playground

I dont get why they havent update it when it was out a long time before the silly chat was

anyone that uses the playground pays way more than a plus user

so I dont get why there is no gpt4 yet for it

[mellow walrus]

So, on the limits page, OpenAI lists a limit of 3,500 requests per minute for pay-as-you-go users for their GPT API. I wanted to try making a game with highly dynamic NPCs (like many have done) as a fun project, but I realized that if the project ever blew up, rate limits/costs would be an issue. If one had 1,000 players messing around with NPCs, for example, they’d be able to send no more than 3.5 messsges/minute, correct? If you scale that up to 10k or 100k, the idea for the game wouldn’t work at all. Is there a way around this? All I can think of is having multiple accounts. How much would it cost to support 10k concurrent users making, say, 3 requests per minute? Are there any alternatives that’d be cheaper/better suited for this kind of system? I’d be really surprised if there’s no solution for large-scale use of the API

[placid inlet]

Hi there! I don't know if I am at the right place to ask this question, so please excuse me if no...
I have a question concerning API requests and their price.

Today, I have been charged for example 0.02$ for these 3 tasks (these were all text-davinci-003 completion API calls):
12:35
text-davinci, 1 request
28 prompt + 300 completion = 328 tokens
12:45
text-davinci, 1 request
29 prompt + 300 completion = 329 tokens
12:55
text-davinci, 1 request
25 prompt + 300 completion = 325 tokens

I have read that a completion containing less than 1K tokens is charged as it was 1K tokens.

So I thought I would be charged in this example 3 times 1000 tokens. So, $0.002*3=$0.006

Can you please tell me what I'm missing?
Thanks a lot

[exotic nymph]

Hi all. Has anyone had any luck with OpenAI account support? I'm stuck with two personal accounts, it seems, attached to the same mobile number, and even after deleting one of them, my mobile is stuck with it and I am not able to register for my company account. I couldn't get any reply from OpenAI support either in over a month 🥺

[shrewd hatch]

/hi

[tender pebble]

When you take a look at chatGPT network traffic in the network tab, you can see that the requests only include the newest question of the user. It doesn't work the way the chatGPT API works for us (which is great for openAI as we pay by token). Instead the chatGPT interface uses conversation_id as a field in the request, apparently giving the backend hints to where it should continue from.

Why is this not included in the chatGPT API? It would make request way smaller, improve performance and - let's be fair - cost us less to request things? I'd honestly not even be mad, if they'd still include all the tokens from before. Handling it with a conversation_id would still be greater than sending all those messages with every request.

[rugged wing]

Hi , I have a question about usage rights .

Can i use Chat gpt created art for monetary purposes?

[abstract remnant]

Getting network error. Anybody else has this issue ?

With gpt4

[abstract remnant]

Thanks i'll try but my internet is fine though

It's doing it again, although i cleared cache. I reported the issue to the team

[scenic vector]

Hello! Where is suppport channel? My account have gptplus plan, but in ChatGPT site it shows Free Plan. Thanks in advance

[abstract remnant]

seems like others reported this earlier in the discord forum

[ashen quartz]

Account Access Help?

My account has been suspended for no reason. Please help me solve this problem. Thank you. GPT is a brand new AI application, and I suspect that it was mistakenly judged because of various tests. I will study the OpenAI related user guidelines carefully and try to avoid such troubles from happening again. My account is [e-mail].

[rugged wing]

do website

[neon prism]

Is there anyone here from openAI? I would like to volunteer my pro bono services for the front end. I think I can help fix a lot of the quirks.

[neon prism]

If code sharing is an issue, I think I could also just provide some modular code for things like front end state management that you can plug-in. I don't care I just want to use your web interface and be less annoyed frankly.

[silent sphinx]

is an openai (chat gpt ) app on IOS exist ?

[glossy shale]

how can i ask question of chat gpt?

[vestal fog]

I keep getting network error, i know that it would happen again so I was going to cancel a prompt. But how do you cancel it? It only has an option to retry

[rotund olive]

can it run a code of an app or chrome extension that when I read about a match game or a competition schedule it can identify it and give me the option to add it in my calendar ?

[rough badge]

I just got the gpt-plus subscription, and me and my fellow grad student buddies wanted to see some multi modal output. 🙂
Unfortunately I think we are using it wrong, the model told us it is gpt-3 despite being listed as gpt-4, is gpt-4 only available through the API atm?

[abstract remnant]

This specific feature is not available yet

[paper cave]

g

[crude musk]

Does anyone else keep get a network error everytime they try to get a response to the bot?

[rough badge]

got it not multi-model atm, but to confirm this is the correct model selection for it?

[abstract remnant]

yes.. it's annoying

[crude musk]

So its not just me then?

[abstract remnant]

exactly

[crude musk]

Something wrong with the bot then

[abstract remnant]

yeah.. submitted a report to the team

[inland urchin]

Hi, Why is it that I try to subscribe to chatgpt plus but the page does not load

[crude musk]

Alright thanks for your answer

[gray cedar]

How to make an assignment

[crude musk]

?

[hardy kiln]

any1 able to get indian cards working for plus?

[late kiln]

any1 like to answer to me that if chatgpt is open source or not?

Thx

[crude musk]

Can you ping me when it gets fixed

[abstract remnant]

Sure, if someone else ping's me when it's fixed 😅

[broken depot]

Asked this earlier on discussions channel, but nothing seems to be working. This is getting extremely frustrating, if you experienced the same, please let me know what fixed it for you:
ChatGPT hasn’t been working for me for the past 2 days:
-it is “unable to load history”
-if i click on submit (the little paper plane icon) nothing happens, and if I refresh, it sometimes gives me one chance to submit and then gives me the classic error "Something went wrong. If this issue persists please contact us through our help center at help.openai.com." -I've tried on different devices, browsers, clearing cache etc...

[late kiln]

so this whole channel is just for openai project beta test?

[scenic vector]

So, I assume, there is no support stuff for chatgpt?

[late kiln]

Its very impressive that 3000000 people are here just for test

[crude musk]

Alright thanks.

[late kiln]

Chatgpt is literally just a person that knows much so what a support stands for😂

[flint cosmos]

I signed up for ChatGPT plus a day ago, is it not the same as platform.openai.com?

Is the ChatGPT API included in the ChatGPT Plus subscription?
No, the ChatGPT API and ChatGPT Plus subscription are billed separately. The API has its own pricing, which can be found at https://openai.com/pricing. The ChatGPT Plus subscription covers usage on chat.openai.com only and costs $20/month.

[sacred mulch]

i asked chat gpt polinomial questions twice it gave the wrong answer for both why is that?
(i asked it using my language and wrote the whole question.)

[untold perch]

Do you mean polynomials? It is a language model so not made for math, although gpt-4 should be better at it

[sacred mulch]

uh yeah i made a typo. Oh ok then thanks

[untold perch]

I can ask it to gpt-4 if you want and send you the result

[sacred mulch]

thank you, i got the answer for it tho

[thorn laurel]

can someone explain me pls why doesn't it work?

[verbal quiver]

Because ChatGPT isn't able to write 5000 words at once. Try doing it in smaller pieces instead.

[rugged wing]

Does chatgpt learn from answering queries? If so, how does it know when it answered correctly or not

[verbal quiver]

It does not.

[thorn laurel]

it's gpt-4

[verbal quiver]

GPT-4-8K can't output 5000 words at once either.
GPT-4-32K might, but ChatGPT doesn't use that one. Too expensive.
Besides, ChatGPT's cap is much lower than 8K tokens simply to keep the cost down

[thorn laurel]

don't you know when gpt-4-32k will be available?

[verbal quiver]

It's available (I think), join the waitlist

[thorn laurel]

I have already bought plus version

[verbal quiver]

GPT-4-32K isn't ChatGPT, it's separate (it's available via an API)

[abstract spade]

Hello, why can't I leave a request on the waitlist, I get an error "There was a problem submitting your application."

help please

[crude musk]

@hollow narwhal

[hollow narwhal]

wow lol I see that

[orchid plaza]

Hi! Having some issues posting to api-projects. Wanted to show off my latest ChatGPT project. Does anyone know why that may be the case?

Trying to message /modmail also doesn't work

[dusk smelt]

Could you explain your issue further?

[ripe pike]

Is it DNS?

[crude musk]

What?

[orchid plaza]

I tried sharing our project and got an error about auto mod filter. I tried reaching out to modmail, and this is what I saw:

[dusk smelt]

Are you sure that you are dming @burnt galleon ? Also, could you please try on another device?

[orchid plaza]

Sure! Thank you.

I was actually able to get it through to modmail! Thanks 🙂 Not sure what it was.

[ripe pike]

How do you make such a high speed?

[crude musk]

I have a fast internet plan

[oak hollow]

what is wooden gray marble uses ?

[rugged wing]

Will GPT4 be available to non plus users

[verbal quiver]

Nothing announced yet, maybe at some point in the future when it's been optimized a lot more

[rugged wing]

Does bing use gpt4?

[white steppe]

I wonder if I can ask GPT-4 to describe a picture like people did in the demo? Which channel should I fo

[verbal quiver]

No, image input is not released yet

[white steppe]

Sad. But thanks @verbal quiver

[shrewd mountain]

hello everyone chat.openapi mostly does not complete more multi file questions for code is there a way I could prevent it to stop or use my own api in an alternative official UI so i get better results?

[verbal quiver]

Yes, it does, but it's... A bit dystopian and unhinged, to be quite honest

[shrewd mountain]

or perhaps there is even a way to generate files so I say " write me a simple application in react that does X " and it generates all the files?

[verbal quiver]

I guess you'd want to use Codex or Copilot...?

[shrewd mountain]

i have 90% of the time that it won't complete using chat.openapi.com which is such a waste of resource having to ask multiple times to complete

no i want openapi to write full code examples for me im learning react and it is really helpfull but the chat.openapi.com website is no longer cutting it for me with all the incomplete answers. I do have an API dev account but I see no way of using the api with some opensource UI that works correctly I tried several ones they all have flaws

just having to generate 5/10 files for a simple code implementation fails almost always with timeouts or Something went wrong, please try reloading the conversation. so its basically useless that way

[rugged wing]

is the openai chat version better than bing's?

[shrewd mountain]

even platform.openapi using my api token does not complete my questions

[rugged wing]

is gpt4 on openai better than bings gpt4?

[gritty echo]

hello, is it possible to prevent chat gpt (free) to stop crashing when sending a code with the things i told him to change

he stops sending it when its finishing it

and will this problem be solved if i buy the premium?

[verbal quiver]

The difference is insane... Literally. Bing still has issues with the model going insane after a few sentences, at least last time I checked (a few days ago)
ChatGPT with GPT-4 doesn't have these problems at all from what I can tell

[rugged wing]

yea bing seems very restricted

[shrewd mountain]

opening the playground on openapi i see there is even a "codex" code-davinci-002. but my question won't get true. Can I use it with an opensource UI for generating files?

[cyan comet]

Hello, has anyone done enterprise procurement with OpenAI? Do they have staff you can talk to about getting a contract in place, legal review, infosec review? I realize the products are new so maybe not yet.

[lament apex]

What do you guys think would have been the first day release money made from new subscription of chatgpt plus for gpt4 access?

[regal rivet]

How can i input images in chatGPT 4?

[tender hemlock]

is quillbot good ?

[near pulsar]

#gpt-4-faq

[regal rivet]

When will it be available to public? This month or later?

[minor forge]

Are you human?

[proven kettle]

hello, just so i don't bug anyone too much...is there a link to anything describing when the current throttles on GPT 4 features (like token length) will be lifted for Plus subscribers interacting directly at chat.openai.com? Genuinely sorry to ask the question, just new here and can't figure out if this is said somewhere.

[near pulsar]

The extra token length is for the API

[proven kettle]

thank you. so they never intend to release that for those that just hit it straight at the browser?

[late ingot]

how to put the main thesis in text-davinci-003?

[near pulsar]

No you can at the playground (like in the live demonstration) in the future, but it costs money per request

I feel like the input/output is about equal to gpt-3 model

[rugged wing]

hi all! today i started exploring openAI API and i'm a bit lost what to use if i want for GPT certain dataset and generate certain output based on that dataset. e.g. i have some attributes that will be mostly static (maybe in one month i will need to update them) and from them i want to generate certain JSON. what is the best approach how to start developing this? any tips/help very much appreciated!

[proven kettle]

thank you @near pulsar , i appreciate it. yeah mine's telling me it's throttled at 4,096, which is a bummer.

[lime bluff]

You will need to utilize embeddings.

You run your dataset through an embeddings model such as ada002 which turns your documents into numerical vector stores. You can then use those embeddings to input extra context information into your prompt to help with Q/A, you can also put in hallucination safeguards depending on your dataset by telling it to only use the relevant context information.

[copper drift]

how to pay chat gpt plus with paypal?

[near pulsar]

Payment options are on the site when you try to buy

[rugged wing]

Thanks for your tips! Is there a good tutorial how to utilize those embedings and use in new prompts? Also I'm using NodeJS so would be great if it wouldn't be Python 🙂

[copper drift]

@near pulsari don t have it

[lime bluff]

I think this: https://github.com/openai/openai-cookbook/blob/main/examples/Question_answering_using_embeddings.ipynb

I only use Python though and if you want more API help its better to ask in #dev-chat

[copper drift]

am from morocco btw

[near pulsar]

You don't have what

[copper drift]

paypal on my payment options

[near pulsar]

If paypal is not in the payment options it means you can't pay with Paypal

[rugged wing]

Really appreciate it! Thanks!

[drowsy crypt]

Hi, I have a problem with the ChatGPT Plus payment, I have tried with three different debit cards and still my cards are declined.

[twilit sun]

Hello, does anyone know hoe can i from a picture generate code?

[near pulsar]

#gpt-4-faq

[lavish karma]

I paid for plus this morning and this afternoon is gone

The charge went through and everything, is there anywhere I can get support?

[pulsar ibex]

can anyone tell me why i'm stuck at the screen after loading screen

[rotund pecan]

give me a good isntagram caption and hashtags for this design post

[near pulsar]

#LoveMeSomeCheese

[hybrid stump]

Hi guys. I noticed some people have $18 credit limits in their account dashboards under "Usage" but mine says $0/0. Is it because they have research accounts? If so, how do I apply for one? Thanks.

[pulsar ibex]

can anyone tell me why i'm stuck at the screen after loading screen

can anyone help me pleaaaase

[near pulsar]

Ahh that screen

[bold fog]

How can I upload an image to GPT-4?

[pulsar ibex]

? what

[near pulsar]

#gpt-4-faq you can't

[near pulsar]

Saying you're stuck at "the screen" isn't too informational

[pulsar ibex]

this screen

[near pulsar]

You can't type in the textbar?

[pulsar ibex]

i can but it's not response

[ashen mason]

when can chat 3.5 model be fine tuned?

[frigid pewter]

same problem as @pulsar ibex

[worn lynx]

open the browser developer console (usually f12) go to network tab and reload the page, usually blatant errors show up. I reported this (response 500 - internal server errors) hours ago and it has been like this for 2 days. Others on the open ai forum reporting the same.

[pulsar ibex]

no errors

[worn lynx]

If you type in the box and press enter what happens?

[pulsar ibex]

no response

[pulsar ibex]

and my profile pic not showing

1s

[worn lynx]

Nothing like this?

[pulsar ibex]

@worn lynx

i got now an errors

[frigid pewter]

nop

[worn lynx]

I guess they are just utterly swamped with traffic though it would be nice if the status.openai.com page would reflect these problems.

[pulsar ibex]

idk anyway i wanna use it as soon as possiple

[worn lynx]

Me too xd

[pulsar ibex]

for my study it helps so much

[pulsar ibex]

1mm
it works on my phone and on my laptop not works @worn lynx

[worn lynx]

whats the url on the mobile?

[ionic tapir]

is chat gpt 4 gonna be able to read images?

[split hazel]

GPT-4 can take an image as input, but the feature is not available on chat.openai.com at this time

[ionic tapir]

oh ok ty

[sage mason]

How to make sure to get long pieces of code from Gpt once it stops midway?

[worn lynx]

in think you can ask it to continue

*I think

[pulsar ibex]

The same on the laptop

[worn lynx]

I tried on my phone still the same problem 😦

[frigid pewter]

debugger says this

here so you can actually read it

[toxic oracle]

How can I get my account reactivated? I did not mean to delete it.

[full thicket]

Dall-e don't understand what is mushrooms because I wait very long time

[untold isle]

Where is my side bar for chat history? I can't find it even I re-logged in. Can anyone please help me with that please?

[runic flame]

file:///C:/Users/chakib/Desktop/GMSI/S2/Automatique%20lin%C3%A9aire/3-TD%20N%C2%B02%20du%2007-03-2023%20.pdf corrigée ca

[hard karma]

keeps saying unable to load history

[woven kettle]

Hello everyone, is anybody here?)

I have a question connected with openai api, where can I ask it?

[formal crow]

I was getting the same. Looks like my history was just fixed. I'm sure with the new GPT4 update some functionalities were down

[proven kettle]

only change i see is taking the 100 message cap down to 50 now. no change in token length, etc.

[hard karma]

all fixed now?

still cant seem to send any msgs

[shell hawk]

hi, is there somewhere to try the image recognition module?

[worn lynx]

same here

[hard karma]

yuppp idk what to do tryna finish some work rn

[worn lynx]

its been like this for about 24hrs for me, hope it is resolved soon.

[hard karma]

damm was able to use it yesterday, trying to upgrade too but doesnt seem like its helping cant even pay for the upgrade

[worn lynx]

wait thinking about it more like 18hrs

[hard karma]

hope they fix soon and let me buy plus soon idk y I cant purchase

[worn lynx]

clicking the upgrade button does this for me "The payments page encountered an error. Please try again. If the problem continues, please email [e-mail]."

[hard karma]

yup been doing that too

[wind jetty]

Hello AI friends, i have a GPT4 sub and was wondering if we will we soon be able to upload a picture very simply by just clicking a button " UPLOAD PICTURES " and ask GPT4 to comment it or do whatever it allows us to... ? because right now it ask to install Pillow lib + to use a python script so it's kind of only for coders.

[mortal moon]

hey can u get free access to textbooks from chatgpt

[half canopy]

can anyone help me, i am trying to upgrade to chatplus but the number associated with my account is not my own phone number? how did this happen and how can i fix it.

idk how this would even happen, like i verified my account with my phone ofc and now its a diff number?!

[wind jetty]

I mean i can learn python, but a button to upload a pic will be much easier

[mortal moon]

are u able to create another account with your ideal phone number?

[balmy sentinel]

It seems GPT-4 use is now limited to 50 messages every 4 hours with ChatGPT Plus. This change went unannounced.

[exotic fiber]

It’s most likely due to the popularity.

[proven kettle]

also unannounced: whether people signing up for subscriptions today based on international hype realize that they won't get any new functionality, other than interacting with 4

[winged laurel]

Hello everyone, I created an OpenAI account to access the API and everything is working fine so far. However, upon logging back in, it is asking me to verify my phone number (which I cannot do because it is already linked to my account). I would like to know if anyone has a solution or if it is possible to contact OpenAI (I could not find an official email on their website). Thank you all and have a great day.

[modest crescent]

What are the new functionalities that are expected except for GPT-4?

[glossy vortex]

Hey team - does anyone know how to contact OpenAI regarding billing questions? Somehow when i set up my account, it set up 2 billing accounts (i.e. every month I would be billed for the same service twice, under the same one account). Ive just worked this out and cant seem to find anywhere to contact OpenAI to ask about it...?

[balmy sentinel]

I have no doubts about the popularity. But in the communication capacity department there is a room for improvement it looks like. Perhaps adding message counters to the interface is also due if the route of limitation is to be followed?

[slender perch]

Hey all, I had a question about GPT-4's closed source status. I just read a Verge article that tries to explain some of the fears and realizations OpenAI had with GPT-3 that influenced the decision to close source GPT-4. I was wondering if OpenAI ever explored hybrid models like the US patent system: make some of the research public at a high level of abstraction for safety and 3rd party research purposes, but still be able to benefit from legal protections against others outright stealing the work. I could see that working as a "have your cake and eat it too" thing if done right.

Or is the issue export controls? I can see AI being closed source to avoid hostile nations like China getting their hands on AI tech.

I just think the answer you guys gave to the public is super washy given the safety concerns around AI in general.

[green delta]

How can one pay from Germany?

[proven kettle]

4 itself, increased token length, image processing, etc. all i really wanted to know is when increased token length would be available. it could be next month, whatever. they could have been much more clear than they have been, and I'm underwhelmed that they are pushing vapor in a release.

the instability and the further truncating of messages they've done today all adds up to this not being ready and them being unable to admit that

[modest crescent]

I didn't have enough time to experiment with the token length. Thank you for informing me. As for the image processing, I must admit I was disappointed as well.

[ionic timber]

If I break Bing's implementation - as in get it to not give a response at all to a query - where should I report that?

it did five searches and then stopped

[proven kettle]

mostly annoyed that they've been throwing strikes hard since i've been using their products and this was simply stupid and unnecessary. transparency is the one thing AI companies need to prioritize right now. all Open AI did with this launch is further imply they're in it for profit taking. yes i want them to make money, no they don't have to mislead people to do that.

math is still an issue, apparently. lol.

[ionic timber]

at least your math got a response 😄

[proven kettle]

lol glass half full i guess

[stone abyss]

OpenAI's been at this for awhile and getting less and less open but more and more successful, I would expect this to be the norm

[near pulsar]

Are tokens even a thing it knows about?

[ionic timber]

Fully agree with that take

[proven kettle]

just frustrating because i was pleasantly surprised by the competency demonstrated so far

[stone abyss]

Yeah, I'm pretty sure they'll deliver though, the limits and other early launch issues will be a thing of the past

[near pulsar]

I really doubt token calculation is in the training data. They have a tool for this you can use

[proven kettle]

well it seems to. lol. it was close to word count the few times i tried it, just not consistent.

[wind creek]

isn't it a bit weird that you have all these people working in NLP being discouraged about GPT-4. Like I want a personal LLM that I won't have to obscure company information or anything like that when I ask it a question about my company's data

[proven kettle]

i feel like they could save some pain simply by letting the bot...or even release notes on the sidebar...be able to clarify things. this wou'dnt be hard.

[near pulsar]

Who knows how tokens were counted in September 2021

[proven kettle]

fair

[near pulsar]

Like the source it uses?

Or like what's new in the model

[deep coral]

Has anyone used InstructGPT?

[near pulsar]

I don't think they even know exactly what's improved, it's just better trained. It isn't like the typical development of implementing features X, Y and Z

[summer sapphire]

Will openai get to Saudi Arabia?

[proven kettle]

When they say things are launching, if those things are not there, clarify when and why they aren’t. That’s it.

If they get the amount of free advertising and momentum they are getting these days, I don’t feel like that’s much to ask.

[noble mountain]

how to import chat gpt bot to my own bot

[burnt blade]

@grave lark !gpt - comms check

[gritty vapor]

?

[exotic iris]

overloaded yet again it appears

[somber torrent]

is there someplace on this discord where i can play around with image and text inputs for GPT4?

[exotic iris]

I think that would likely need to be inside the GPT-4 interface and they haven't released that functionality yet (even for Plus members).

[lone saddle]

What do you mean? I signed up like 2 days ago. There's an option for Legacy model in addition to GPT4. It's not available in FREE version.

[jagged tree]

do we know when imaging feature will be live in chat gpt plus?

[civic fox]

#newHere: prompt create bot to summarize all questions here to now in five categories relevant to all here now 🦻 🤖 📣

[proven kettle]

Yes. You get 4. You do not get the additional features they announced. They also left out the part about capacity constraints you will now enjoy as a plus subscriber.

[somber arrow]

Is there a way to disable chat logging us out of the website? Im using it on my phone on Safari and when I go to another app and return, im always logged out.

[proven kettle]

Remember how they said pay to avoid timeouts? Well welcome to unannounced tiered pricing schemes.

20 bucks a month is turning into the new free

[lone saddle]

Personally, GPT4 was the biggest reason for me to upgrade, Legacy model a 2nd thing. As for capacity and limits, I wasn't complaining with FREE version

[royal sand]

How can I get chatgpt on my phone…

[amber wave]

Not possible yet if your talking about an app. You can add a website to your home screen on iOS though.

[proven kettle]

I love the product and no issues paying for it. I’m simply disappointed that they over promised, under delivered, and are taking steps to dilute the value of the money i give them.

They are obviously flipping to a tiered pricing model for Plus and they are counting on hooking enough people now to increase their conversion rate for the higher tier.

[lone saddle]

sorry but what features are you referring to exactly, that they supposedly promised and didn't deliver?

there were no actual numbers as far as the limits are concerned nor anything like it

it always depended on the workload among other things

you are given priority access, but that's polite way of saying "slightly faster"

[calm bolt]

I subscribed to the Plus plan yesterday and was charged for $20. Today when I logged in to ChatGPT, my account had reverted back to the Free plan for no apparent reason, while the $20 charge is still on my card. Anyone ever had the same issue?

[exotic iris]

@calm bolt yeah, I had that issue right before they actually brought out gpt-4. it should revert after not too long.

[calm bolt]

thanks for the reply. hopefully you're right because it seems to be a bit difficult to get in touch with OpenAI support

[lone fulcrum]

So it’s $20 for gpt4?

[cloud cedar]

can GPT-4 do some advanced math is it worth getting it for 20 bucks a month to send pictures of math 💀

[lone fulcrum]

Ye

[cloud cedar]

u have it?

[lone fulcrum]

Soon

[cloud cedar]

oh hmm but is it like better in math maybe 😅

[lone fulcrum]

How did Greg get the scribbled picture to do programming then?

[rugged wing]

I love capitalism

[lone fulcrum]

It’s decent

[rugged wing]

open

[exotic iris]

@lone fulcrum so it's $20 for GPT 3.5, 3 and 4. As it currently stands today, they've put a restriction of 50 queries every 4 hours for GPT-4. Yesterday it was 100 queries every 4 hours. 3.5 or 3 (Legacy), are unlimited on Plus.

[lone fulcrum]

Thank you 🙏

[lone fulcrum]

Open sesame

[zealous veldt]

I received an e-mail last night from OpenAI that is asking me to fill out a survey form for a "Private Preview." Has anyone else received this? I ask because I am a little worried it's a phishing attempt.

[unique quiver]

Spanish?

Spanish

[lone fulcrum]

Phishy

[zealous veldt]

Okay, thanks.

[tidal flicker]

When are y'all adding new payment methods? (like paypal)

[exotic iris]

@zealous veldt no

[zealous veldt]

Thank you.

[merry orchid]

Have there been any investigations on how using a language besides English for the prompt (in part or in whole) changes the output? I did some testing with codegen and the outputs are massively different between the languages I looked into - some possibly better.

I need to quantify it a bit better but it's such a basic thing I assume there must already be work out there.

[exotic iris]

@merry orchid I have tested a bit in Japanese and had pretty great results. For reference, my Japanese is well advanced so I can competently understand discrepancies.

[polar grove]

I purchased the premium version for unlimited requests but was timed out. Was I incorrect in this being a feature?

(for the normal one too)

[exotic iris]

@polar grove it depends on which version you are querying on. currently GPT-4 was reduced to 50 queries over a 4 hour period

[polar grove]

default told me I had too many requests and had to wait an hour

[merry orchid]

I got everything ranging from "nicer comments in German it seems" to welsh where it gave me the "as a language model.." response, in welsh

I'm an english native but I can speak German. I wonder for people with a formal and informal "you" if it matters and also which they would use with an AI

I will probably benchmark a few thousand generations just for my own purposes regardless

[calm bolt]

Could I ask how long it took for your issue to be resolved on its own?

[mild helm]

Are any of the OpenAI higher ups aware of the fake OpenAi GPT-4 stream crypto scam?

[magic needle]

Hello everyone.
Im having an issue with chat gpt-3.5-turbo model. Basically Im able to run it using curl but when I try to use the HTTP request from the docs I have an error. Can someone help me understand if thats a normal behaviour for beta features?

[edgy carbon]

anyone know how to fix this? I get this everytime i try to use chat gpt

[neon prism]

You could contact them through their help center at help.openai.com

[edgy carbon]

i go to the site and its just pre written responses that dont help, how do i actually talk directly with someone?

[random axle]

Basheer km narikkuni

[exotic iris]

@magic needle if you're asking it if can externally access the internet, the answer is no.

[exotic iris]

probably an hour. maybe 2. It probably wouldn't hurt to try to log out and back in. Again, when mine popped back up, I then had GPT-4 capability so in my case was likely an issue with a server update.

[exotic iris]

this typically resolves itself for me if I refresh the page

[edgy carbon]

i logged out, refreshed, got rid of cookies, and none of it worked

[exotic iris]

I would think perhaps a server issue then? I do see this error probably once or twice a day and it seems to always resolve with a refresh. Maybe try to close your browser completely?

[desert meadow]

How do i get chat gpt?

[magic needle]

Im just trying to make a request to this endpoint https://api.openai.com/v1/completions with this payload: {
"model": "gpt-3.5-turbo",
"messages": [{"role": "user", "content": "Say this is a teest!"}]
}

But I get this response:

{
"error": {
"message": "Unrecognized request argument supplied: messages",
"type": "invalid_request_error",
"param": null,
"code": null
}
}

[exotic iris]

@magic needle Based on the error message you received, it seems that the OpenAI API server is not recognizing the "messages" parameter that you included in your payload.

One possible explanation for this is that the "messages" parameter is not a valid parameter for the "completions" endpoint. According to the OpenAI API documentation, the valid parameters for the "completions" endpoint include "model", "prompt", "temperature", "max_tokens", and others, but "messages" is not listed as a valid parameter.

[neon prism]

@exotic iris You generated that response with ChatGPT

[exotic iris]

😉

@neon prism it's our secret!

[neon prism]

Oh @magic needle wrong endpoint

Should be https://api.openai.com/v1/chat/completions

[mental hazel]

Can I use chatgpt for audit smart contract?

[neon prism]

The one you're using is for the classic davinci, babbage, curie, ada stuff. GPT 3.0

[magic needle]

Thanks

[neon prism]

Naughty.

[obsidian patio]

as always

[neon prism]

You're clicking too hard.

[exotic iris]

@mental hazel I've done a few contracts with it that have turned out quite well so far so I'm sure you'd be able to do that.

[mental hazel]

Thanks for your message

[magic needle]

It worked fine. Thanks

[mental hazel]

Should I use gpt3.5?

[exotic iris]

@mental hazel it's faster than gpt3 and unlimited if you have plus although gpt-4 would likely give you a more comprehensive document

@mental hazel I am switching over to GPT-4 typically when I am looking for a specific tone or I want to teach it something advanced that I can reference later. I'm still learning the new capabilities though.

[mental hazel]

So it means you used gpt3 for audit so far and now you are going to switch to gpt-4, right?

Did you try to audit the solidity smart contract?

[exotic iris]

Since they've limited GPT-4 to 50 queries every 4 hours, I limit my queries to what I think need the most comprehensive answers. I've used both 3.5 and 4 for contracts and have had great results with both.

GPT-4's capabilities on legaleze have been documented to be much more robust than GPT-3.5.

[mental hazel]

I see

did you use openai api, right?

[exotic iris]

Correct.

[mental hazel]

Can you give me the sample code for that?

which language did you use?
JS or python?

[ripe kestrel]

wondering about using chatgpt for resume writing because i wanna know if ATS does detect the AI generated texts , i thought it would be easier to match the jobs with my resume from AI

[exotic iris]

@ripe kestrel ask it to write in a tone that is ats friendly

[ripe kestrel]

🤙🏼

[mental hazel]

Hi @exotic iris
Can you guide me more for use gpt3 and 4 to audit solidity smart contract?

[fair oar]

Greetings! Am new here. ^_^
For now, am mostly curious about the difference in coding ability between GPT3&4. Also curious about any prompt ideas that can be used for improving ChatGPT's ability to write useful code. Has anyone here played around with prompts such as asking ChatGPT to roleplay a character and act out the role of a programmer? Or perhaps even asking ChatGPT to mimic the style of a specific famous programmer?

If so, what sort of prompts seem to result in better code?

[prime blade]

e

[exotic iris]

@mental hazel so I'm definitely not a coding expert here and somewhat had to fumble my way through but here is what I can tell you about my journey. Any question that you have, ask chatgpt. It seems to have a good understanding of Solidity and Brownie (like I used using Python). The AI model could probably give you a more comprehensive answer than I could and give you better answers about your specific situation.

[mental hazel]

Thanks @exotic iris

[exotic iris]

@fair oar Ask the ai model itself! Being more verbose and letting it know particulars seems to help. Personally, I haven't noticed any difference between the two in capabilities although my attempts at coding with GPT-4 have been limited. It seems to have a relatively robust understanding of any language I've ever attempted (C, C++, js, Solidity, Python, Flutter). It does have a 4096 character limit so be aware of that.

[toxic terrace]

what symbol is this

[exotic iris]

Is it a hiragana mi?

[toxic terrace]

idk

[vital pike]

I subscribed plus account from last month, but why it shows I not having a plus account? Anyone face this issue also?

[exotic iris]

I mean if it's supposed to be Japanese, that is the Japanese hiragana character for "mi". @toxic terrace

[toxic terrace]

tnks

[exotic iris]

Yes, happened to me yesterday. Seemed to resolve itself and at least on my end was related to a server update.

[rugged wing]

Is it against content policy to ask for medical advice? Like just asking for it but not making a full diagnosis,

[exotic iris]

@toxic terrace wait, that's slanted the wrong direction for "mi". Sorry, it's been a few years. Depending on the context, and whether it's actually Japanese, someone could write "shi" like that although the top two strokes are a bit too long.

[grand magnet]

anyone has any good advices on how can i fix my usage update. it hasn’t update for me since 10 days ago and it really bothers me that I can’t tell how much money i have spent on it.😂

[fair oar]

Hmmm interesting.
When you say 4096 character limit, do you mean for the code it can output at once? Or is this in reference to the size of the prompt I can input? Or both?

[exotic iris]

@fair oar both actually

[exotic iris]

@fair oar you can break it into multiple instances but, as it's told me, it treats each response as a new input. From my personal experience it seems to combine multiple inputs so I don't know if its own response is completely accurate?

[stuck tusk]

Account appeal

[fair oar]

Hmm, not sure where, but I thought I heard some youtuber say that ChatGPT4 can handle larger prompts than ChatGPT3.

On the topic of limitations, I've also noticed that after the conversation drags out long enough, it's like one moment it's fine and then the next response ChatGPT3 suddenly has total amnesia.

Have experimented slightly with the idea of having ChatGPT create something like a "save file" every now and then by summing up the conversation so far in a way that a future incarnation of ChatGPT can easily understand.

It can work to some extent, but is a bit messy. Also doesn't help that I'm not sure exactly where the cutoff point is when it will get amnesia. =/

[stuck tusk]

How to restore an account

[vital pike]

I see, do you contact the OpenAI team on this issue?

[stuck tusk]

Who should I refer to for account questions? Can you help me solve them

[exotic iris]

@fair oar How does the idea of a save file work? I haven't tried that.Yeah, GPT-4 should have that capability I think it just hasn't been released to us plebs yet. Probably when the server issues get more ironed out they'll roll out more of the capabilities.

@vital pike I'm just a guy here on the sub but you can email the team at [e-mail]

[fair oar]

The simple version would be to simply ask ChatGPT to "sum up our entire conversation so far, in a way that a future incarnation of ChatGPT can easily understand."

That's the basic idea anyways.
Could either ask it manually when you want a "save file"
or could even include it in your initial prompt when you define how you want the AI to behave throughout the conversation. Maybe something like: "Every 10th response you give, include a summary of this entire conversation so far, written in a way that a future incarnation of ChatGPT can easily understand."

[kindred rock]

what do you know about application development

[fallen ether]

Maybe you’re logged into the wrong google account?

[vital pike]

Alright, thanks for the advice!

[vital pike]

I'm pretty sure I'm using the correct account

[cyan spruce]

Can anyone tell me...the difference of gtp1,2,3 and new one that is gpt 4?

[fallen ether]

Then you should check if you’re getting charged the monthly fee and get ahold of open AI support. I’m on plus but if I’m in my work email it obviously doesn’t show up

[vital pike]

Yes, the monthly fee was charged yesterday, but I can't use PLUS permissions on these few days

[exotic iris]

@fair oar I love that idea

[calm bolt]

I just subscribed yesterday, charges are still there, plan also got reverted for no apparent reason today.

[neon prism]

I've been doing exactly this in an SMS frontend for ChatGPT

[fair oar]

It needs refinement but I think it seems like a natural idea for working around the current memory limitations. Would be helpful if I could get it to only create the "save file" right before it's about to suffer from amnesia. Would also be helpful if I could think of a better way to compress the data, in a way that a new incarnation of ChatGPT freshly patched with whatever initial prompt, can actually unpack and understand.

Another thought would be to clean up the chat a bit to focus on saving at least the most important highlights of a conversation, while dropping redundant or less important info when creating the save point.

[fair oar]

What's your max size for a prompt?

[neon prism]

I'm not enforcing that but once GPT-4 is api available it will be quite large.

[fair oar]

I've been reading up on API's and find the tech to be quite fascinating. Am curious how you got ChatGPT working with an SMS frontend. @_@

So like, you can speak with ChatGPT via SMS using your smartphone as if you were simply texting someone?

[neon prism]

I can't share the external link that provides all the info

Twilio + Airtable

And I use Airtable automations to summarize the conversation and make the summary the first line of the conversation, deleting old lines

[fair oar]

Though I was really only looking for a basic general overview.

[fair oar]

Am not familiar with Airtable or Twilio, will have to look into it.
I've just been having ChatGPT summarize the chat, and saving the summary into a rich text document, then copy/pasting from there when generating an initial prompt to start a new ChatGPT conversation.

[neon prism]

You would have to develop something with the api like I did.

[hot echo]

why😭

This is a new device, why can't I use this IP to register a new account?😭😭

[fair oar]

Ah ok. ^_^
Experimenting with the API is on my todo list.
So far I've mostly just been poking and prodding the AI and trying to improve my ability to communicate with and leverage it for useful purposes.

As far as useful purposes go, am currently thinking of how I want to create a prompt to turn ChatGPT into a learning tool to help me learn GDScript by writing and explaining code to me.

For some reason, I've always found it easier to jump in and learn my way around an IDE when I have functional code that I can tinker with and reverse engineer. I haven't written any code on around 2 decades though, so I'm more than a bit rusty. xD

[fair oar]

IIRC, I think you need a non-VOIP phone for SMS verification to use ChatGPT

[hot echo]

Thanks,bro but I am not blocked in the phone verification section, but when I only entered my email and the desired password for registration, it shows that this IP has registered too many accounts.

🥹

[fair oar]

are you using VPN?

[hot echo]

Yes

[fair oar]

That's probably why.
specially if using a free VPN

[hot echo]

I’m gonna try it again

Thanks😘

[steel river]

How to use chat gpt 4 guys?im using chat gpt 3.5

[clever acorn]

You have plus?

[steel river]

No sir, how to have plus

[clever acorn]

There should be an option on the left side to upgrade. It’s $20/mo. for plus

[steel river]

We must have plus to use chat gpt 4 sir?

[clever acorn]

For now, unfortunately. There might be a wait list for others but I’m not sure.

[empty mango]

When accessing ChatGPT from our company's environment, an "Access denied" message is output and the service cannot be used. Is it possible to request permission to use it?

[muted spruce]

Does anyone know if there will be eventual rollout of GPT-4 on playground? Kind of like how we saw on the demo?

[rugged wing]

anyone from poland? do you had a message about "not allowed in your country bla-bla-bla"?

i start being mad about that

[lime bluff]

If you have API access you can already use it there

[fallen ether]

https://openai.com/waitlist/gpt-4-api

[muted spruce]

@lime bluff @fallen ether thanks

[calm bolt]

Anyone know how long it takes for support to reach back

My subscription is still gone for no reason

[calm bolt]

If they do at all, that is 😭

[neon prism]

Would anyone be interested in trying the API by using my Siri shortcut for your iPhone/iPad/Mac?

I heard this come up a few times that folks want to try out the API. With this you would just need to paste in your own key and it should work right away.

[neon prism]

Oh wow it's not just me.

[calm bolt]

Yeah I just subscribed yesterday and today my plan just went back to the free plan for no apparent reason. The $20 charge is still there though.

[runic basalt]

Guys
I can't create a account in OpenAI
''Signup is currently unavailable, please try again later.''

[raw imp]

I can't log into my Chat GPT from my computer with my gmail. But I can log in from my phone. Is this happening to anyone else? How do I fix it?

[bold iron]

Happened to me, it fixed a day later. Dont know why

[raw imp]

Wow okay, thank you for responding. It was working earlier this afternoon and then stopped. I was so confused as to why it happened and why it won't let me back on. I click the email associated with my account and it just sits there. It doesn't move from that screen.

[bold iron]

if you need a quick fix, try creating a new account with a new email

could prob fix it, dont know if there`s any rules against that

[raw imp]

I thought about that. Just didn't want to lose all of my history, you know? I'll see if it comes back by tomorrow otherwise I might have to do that.

[raw imp]

Thanks so much again for responding. I really appreciate it. I didn't know what to do to get help lol

[bold iron]

yeah that could be a problem lol

[bold iron]

hope you get it back

[raw imp]

Thank you!! Me too

[fringe swan]

Hey, does anyone know why i can't press on the submit button after writing question in chatGPT? Pressing "Enter" doesn't work either, it will just add a line break.

[spiral copper]

anyone here gone through the interview process at Open AI for software engineer?

[exotic iris]

I have had my history disappear a number of times but it always seems to come back eventually. I'm confident yours will too.

[raw imp]

If only I can get into my account lol. Thank you so much!

[gray cedar]

Happy birthday

[tribal ridge]

Maybe a network error? Copy your question and refresh the page.

[eager silo]

When I use the OpenAI API to generate an image from a prompt, a set of URLs are returned. How long can I expect these to stay present on the web? I.e. will the images be taken down at some point in future? How long will the URLs remain valid?

[wise elbow]

I have a question about the privacy of ChatGPT. According to ChatGPT itself, if someone generates content using ChatGPT and posts it somewhere, there is no way for OpenAI to know which user or account generated the output. Is this true, and if so, how is this balanced with enforcing the policies on outputs that may breach the Content Policy?

Another way to phrase: Can OpenAI identify a user based on the ChatGPT output, by examining usage logs or something similar?

[median cedar]

I have a question, a few month ago, i deleted m'y account. But now i would like to réactivite him but i don't now how i have to do that. How i can do this ? Can you réactivite m'y account ? Thanks you by advance for your answer

[dusk smelt]

If you deleted your account, it's deleted. You can create a new one, but a deleted account is deleted forever.

[pale vortex]

Hello. Here the other day I bought a ChatGPT plus membership, but when I logged in today it says my current membership is a free one. I have the mail verifying my membership and everything. How to fix?

[near pulsar]

They probably collect data yea

In their privacy policy they mention it pretty clearly

[pulsar ibex]

It works now

[dusk smelt]

Can you try in an incognito window (ctrl+maj+n)?

[pale vortex]

Just did, same thing. Got the popup window that said "This is a free research preview".

[frank remnant]

Hello, why can't I buy Premium with anything other than a credit card? Most of the world outside of the US doesn't use credit cards as main form of payment, which is why most germans I know don't even have one

[glossy charm]

I'm getting an error when trying to buy ChatGPT Plus, "The payments page encountered an error. Please try again. If the problem continues, please email [e-mail]." can anyone help me about this?

[dusk smelt]

That dosent mean that you aren't plus. This popup appears for everyone.

[dusk smelt]

You can try with a prepaid card.

[pale vortex]

I made it work now. It was some network-issue. Thanks for the help tho!

[frank remnant]

How so?

[fringe swan]

yeah tried that too, but didn't work. For some reason, the only thing that helped was creating a new account

[frank remnant]

I would just edit the question, then it works again

[wise elbow]

I was looking for a concrete response. There is a section in the Privacy Policy about de-identifying content:

Section 2 under Aggregated or De-Identified Information

[echo pivot]

Dumb question... that new Microsoft Office 365 GPT integration: is that available in home subscription or only business subscriptions of Office 365? How do I get access to it?

[near pulsar]

Yeah what about it

[regal kelp]

Does OpenAI offer any consultation of whether some use cases for projects could be supported or not? I'd love to enhance the app I'm building with OpenAI, but I'm not sure if the things I have in mind would be possible to build with OpenAI. Would appreciate some guidance!

[last bramble]

Hmm sure try contacting their sales

go to the pricings page on openai

[regal kelp]

true, good point! thanks

[last bramble]

curious about ur app, whats it about?

[wise elbow]

It goes back into the original question

[near pulsar]

I don't see how banning accounts that have flagged messages contradicts that policy

[wise elbow]

I don't think I mentioned banning accounts

My question was about the privacy of the prompts and replies

[near pulsar]

Or just usage of the service

[near pulsar]

They are not private no

[last bramble]

what was your question again?

[wise elbow]

^

I think it is important information to know for legal reasons, and for privacy advocates

[near pulsar]

They are not private and that doesn't contradict that policy

[wise elbow]

Here is a better way to phrase

[last bramble]

whats openai? their people?

[near pulsar]

They do know which user generated which content, it's in their policy

[last bramble]

Oh do they save logs that is tagged by user accounts? no i dont think so

[wise elbow]

Can you show me where it says that? I do not see that explicitly stated

I even had ChatGPT review the entire policy, and it was undecided. That is why I ask here

[near pulsar]

User Content: When you use our Services, we may collect Personal Information that is included in the input, file uploads, or feedback that you provide to our Services (“Content”).

Input you provide to our services

[wise elbow]

Hmmm. So if someone asks ChatGPT to make a poem praising democracy, for example, and the person is not from a democratic country, could OpenAI technically find out who generated that output, based on the output? I do not ask if they would, but if they are technically capable of doing so

[near pulsar]

Well they save the input, and I don't think their output is your privacy

[wise elbow]

Indeed, but the question is, is the input tied to that specific user, or is it de-identified?

I have heard mixed answers on this

[near pulsar]

It has to be tied to a user, how else could it save the chat history

[wise elbow]

I have thought about that. But if you clear history, then is it still tied?

[near pulsar]

I think that's section 8

"We’ll retain your Personal Information for only as long as we need in order to provide our Service to you, or for other legitimate business purposes such as resolving disputes, safety and security reasons, or complying with our legal obligations. "

They don't say "We'll retain it till you delete the chats"

[wise elbow]

Indeed, and:

"We may also anonymize or de-identify your Personal Information (so that it can no longer be associated with you) for research or statistical purposes, as described above, in which case we may use this information indefinitely without further notice to you."

[near pulsar]

Emphasis on may

[wise elbow]

In this context, it means "we are allowed to"

[wise elbow]

This was a good find, though. Thank you

[dusk kestrel]

/?

[median cedar]

Ok thanks

[steady path]

I assume that ChatGPT/GPT-3 was primarily trained with English texts. If I use ChatGPT/GPT-3 in another language than English, will ChatGPT/GPT-3 use its English training data, too, or will it rely only on its Dutch/German/France/whatever training data?

[near pulsar]

You won't get different information based on the language you ask it in

Maybe only in asking it to generate poems, since translations won't rhyme

[midnight tree]

Something went wrong. If this issue persists please contact us through our help center at help.openai.com. I always received above sentences,what is that mean?

[steady path]

Okay, that's interesting. Thanks. I wasn't sure if I'd get better results, if I asked in english instead of another language.

[midnight tree]

maybe I should go to another room

[near pulsar]

It means there's an issue, and if persists, contact them through their help center

[hazy stag]

tiananmen square

[midnight tree]

[weary tulip]

Hi, can GPT AI be built and integrated into company's sharepoint?

[native musk]

？

Uh....
Hello?

[near pulsar]

What?

[native musk]

How to fix it?

[near pulsar]

Try again later

[native musk]

K

[rigid musk]

Try in incognito

[sleek vault]

Donno

|⎸⎸ٳٰٰٰ⎸⎸⎹|ٳٰٰٰ|ٳٰٰٰ|⎸⎸⎹|⎸⎸⎹|ٳٰٰٰ|⎹

I don't |⎸⎸ٳٰٰٰ⎸⎸⎹|ٳٰٰٰ|ٳٰٰٰ|⎸⎸⎹|⎸⎸⎹|ٳٰٰٰ|⎹understand this

[abstract kestrel]

yesyes

[civic sky]

So will gpt4 be intergrated with chatgpt?

[rigid musk]

It is already

[sleek vault]

@rigid musk ur here also

😂✌️

[rigid musk]

Hewwo

[sleek vault]

😅

DiscordGhostss

[civic sky]

Was it rolled out?

[drifting shoal]

why is it sending an error message

[rigid musk]

yep

[oak moth]

refresh the page

[civic sky]

Alright so I can put images into it?

[rigid musk]

1. log off every device connected, refresh and log back again
2. make sure every vpn and proxy is disable
3. make sure to disable useless extension that can block. Please not some anti-virus can be problematic and desactive it
4. some works or school have firewall . Use data connection or cellhphone
5. try a new fresh browser or incognito
6. if this is working with your cellphone and not on your pc you still have something that block on pc. Check again step 1 to 5 .
7. if this isn't working on cellphone you might need to contact [e-mail]

[rigid musk]

#gpt-4-faq

[drifting shoal]

i have vpn on because my school intternet blocks a lot of things

[rigid musk]

90% chance this is the issue.

[drifting shoal]

oh ok

[tidal canyon]

I try to contact openAI with mail etc since more than 48 hours and i can't work because i paid 20$ for plus mod and ChatGPT considers me like a free users, i need help

I can't work with the free mod it's very slow and limited after a few times

[near pulsar]

Yeah that sucks but community support can't activate premium for you, only they can

Normally you're on the wrong account, but many reports like this have been made today

[coral shore]

the problem is how are you supposted to contact them

i opened a bug report nothing and they say go to discord so here I am

[tidal canyon]

I understand but i have no response by openAI. I am on the log with my account i can see the shipment, but it's still appear on "free mod" and ask me if i wanna join the plus mod.

[near pulsar]

Yeah they are super slow

[tidal canyon]

I have try to DM "Valerie", i have send email but nobody answer me after 48hours ^^

[near pulsar]

This is something only the openai team can solve unfortunately

[oak moth]

You can use email or
You may also try contact Support Team by the Help Chat in below article, by clicking the Help button in the right bottom of the page.
https://help.openai.com/en/articles/6783457-chatgpt-general-faq

Hope it helps.

[near pulsar]

All you can do is wait for a reply

[oak moth]

exactly

[tidal canyon]

Yes i have send at this adress 48hours ago, i wait i wait but in the meantime I can't work I have to do something else. Anyhow, thanks for your help 🙂

[rigid musk]

OpenAI is currently experiencing a high volume of messages and are working hard to respond to everyone in a timely manner.

[worn lynx]

Anyone know of any 3rd party ChatGPT providers, specifically with the chat.openai style interface? chat.openai.com is still unable to load chat history etc for me.

[dusk smelt]

try logout login 🙂

from chatgpt

[worn lynx]

Hey Paillat, tried that, tried a fresh browser install, even tried Edge xD always the same nextjs script errors and openai backend internal errors.

[dusk smelt]

Have you tried on cellphone data?

[worn lynx]

I did try on cellphone but it was on my wifi, that is an idea.

[worn lynx]

Same on cell data, unable to load chat history

[dusk smelt]

And after logout login on cellphone data?

[worn lynx]

Same, ah well I guess it will work eventually xdd

[winged spruce]

Where can I generate art in this server? I mean i can't find that channel here

[lone saddle]

I don't see it compatible with certain governments and their ideology to be completely honest

it would be banned anyway soon enough even if they tried

yes

[half vector]

how to insert photos?

[lone saddle]

As far as I know all undependable from government news reporting services and agencies are now banned there

not to mention even facebook has/had issues?

so how could chatgpt work? lol

nd chatgpt can tell you facts about the war among other things...

which don't go with their narrative

maybe, but certainly not Russia or China lol

Well I already explained at least some of them. For other dunno, maybe there are other reasons or they need some kind of approval to do it.

compliance with local laws etc

don't forget theoretically OpenAI could get sued too if they are careless

[cold elk]

write a rock song about the acceleration of gravity

[lone saddle]

many reasons, some non-sensical but they would probably have to go through the legal process in the country of origin

If the government sues it's obvious no? lol

"country x sues OpenAI for xyz in violation of local laws" - how do you stay 'unknown'?

that's the thing, I'm not a lawyer but I would think they need to abide by local laws however non-sensical they may be

and they are certainly not gonna inject propaganda into it to make it suitable for certain regions ...

It doesn't take sides explicitly, but it can still present facts without any bias

and those are different too from the certain narratives

ok, bye

[next locust]

What is the difference between Default and Legacy? Default seems to be better in Speed and Conciseness

[fallow trench]

how to access chatgpt-4?

do i need to upgrade plus/

how to access chatgpt-4?
do i need to upgrade plus?

[rigid musk]

yes

[severe tulip]

How do I buy Chat GPT Plus? Can’t find any place to sign up

[rigid musk]

https://openai.com/blog/chatgpt-plus

[crimson moss]

How do you input photos in GPT4?

[normal linden]

What's new in ChatGPT 4

[crimson moss]

General processing, accuracy, output and input limits are bigger, etc. and you can use photos as an input but unsure how to do that.

You cant copy and paste or insert an attachment...

[normal linden]

At first impression, I don't see much difference

[silver heart]

Hello everybody,
I'm seeking for a count down intro for my videos!
Anybody can help please ?

[rigid musk]

#off-topic

[rigid musk]

#gpt-4-faq

[frigid pewter]

how do i fix this

its constantly loading

[rigid musk]

1. Log out of all devices: Please ensure you have logged off from every device connected to your account. Once done, log back in.
2. Disable VPNs and proxies: Ensure that all VPN and proxy connections are disabled on your devices.
3. Disable problematic extensions: Check for any unnecessary browser extensions that could be causing issues and disable them. Please note that some antivirus software may interfere with your connection, so consider deactivating them temporarily.
4. Firewalls: If you are using a work or school network with a firewall in place, try switching to a mobile data connection or using your cellphone as a hotspot.
5. Browsers: Use a different browser or incognito mode: Test your connection in a new browser or by using your current browser's incognito mode. (Chorme/Edge are the same browser internally)
6. Troubleshoot device-specific issues: If the issue persists only on your PC and not on your cellphone, review steps 1 through 5 to identify any potential issues on your PC.
7. Contact support: If the problem remains unresolved even on your cellphone, please reach out to the support team at [e-mail] for further assistance

[late ingot]

how to reset dialog in gpt-3.5-turbo API? In Python

[fiery peak]

I really hope they will do something about the problem that chatGPT doesn't complete it's code when things become long as for example letting him making a basic template using tailwind

[heady night]

There are and will be specific models to overcome this. If OpenAI doesnt someone else will, and I dont think OpenAI wants to lose marketshare.

[fiery peak]

Lets hope so 😄

[heady night]

Its not a matter of if, its a question of when. Cats out of the bag on this one. Misaligned LLM systems are here to stay. The only question remaining is if we well get perfect alignment for various uses.

[severe tulip]

There is no call-to-action? I wonder if it’s because I’m not located in the US

[quick pine]

I get this message when I want to summarize the transcript from an interview. "The message you submitted was too long, please reload the conversation and submit something shorter." I thought it could summarize large blocks of text

[robust cradle]

Why is chat gpt don`t working in the Egypt?

[rigid musk]

law and restriction

[robust cradle]

why?

[rigid musk]

ask your country why they banned it

[robust cradle]

Can I ask Chat gpt on this server?

[rigid musk]

they are no chatgpt bot in this server

[full ore]

Either try refreshing your page, logging out and logging in, or just clearing your cookies. it depends on what you are trying to fix. The left pannel is just blank because the servers are probably trying to catch up with influx of use, not much that can be done there

[exotic iris]

@robust cradle from the source itself: "ChatGPT is generally available worldwide, and there is no inherent reason why it should not be available in Egypt or any other country. However, there may be regional or national regulations, restrictions, or policies in place that impact access to certain types of online content or services, including AI language models like ChatGPT."

[inland matrix]

i'm not sure this is a good place to post this, but I have a question about openai privacy. It regards my desire to delete something from the system so personnel don't see it?

is this a good place to pose it?

[exotic iris]

@inland matrix ChatGPT conversations are processed in real-time, and the model does not retain any memory of the conversation once it is complete.

@inland matrix this is of course only if using the ChatGPT interface directly and not using a chatbot.

[junior oxide]

I'm trying to create a new account using my business email, but when i go to enter in my phone number I can't continue because my phone number is already attached to two trial accounts. I need to buy ChatGPT Plus using my work email but need an exception for my phone number to use it one more time.

In the mean time I've submitted account deletion on the accounts tied to that phone number so it becomes available again but I need that expedited if possible.

[rigid musk]

dont forget , using several account for 1 person is bannable

[rigid musk]

you need to wait till [e-mail] help you

[junior oxide]

gotcha - essentially that's why I was hopping to reach openai support to figure out how to do this transition the right way

Thanks Neko

[hasty nova]

hi, how do i get chatgpt plus from turkey by turkish liras

[inland matrix]

@exotic iris Thanks for that, but I'm confused because today openai started booting chatgpt with a warning that personnel might read your stuff and not to type sensitive info, right??

[livid vapor]

Hello! How to create and train my own model by fine tuning using Davinci base model with Node.js? 🙌

[old pagoda]

Why can't I accsess OpenAI services from RU number and Email? I've moved abroad. I tried temp emails and numbers, but it resulted in BAN. Fix it please. My email is [e-mail]

[uneven pagoda]

Hey everyone does chat gpt 'remember' everything in a conversation? Say if the conversation is 200 messages long?

[proven kettle]

depends on length, but unlikely. they currently have it constrained to 4,096 tokens for Plus users, per my understanding. They overhyped the extended memory in this week's launch and that's the same length as earlier versions. (happy to be wrong on this if folks know better...i definitely wish it was longer)

that's inclusive of the prompt and the response lengths, all in

"token" is roughly equivalent to a simple word

[uneven pagoda]

So that means 8,000 tokens for prompt AND response? Or 4,096 tokens for prompt and response?

[proven kettle]

hah no, 8,000 would mean they actually delivered the features everybody signed up for this week. 4,096 total.

[uneven pagoda]

Right, I assumed they're "rolling it out" eventually no?

[proven kettle]

and get ready for a new pricing tier if you're interested in the audio and image features they also "delivered" this week

[proven kettle]

well you'd think so. but lacking the ability to build a simple calendar expressing that, they've chosen to be ambiguous about it. Why tell the truth and lose a few subscriptions in the process? heck no.

[uneven pagoda]

Via api I use audio with whispr. It is basically $6 for nearly two years of 24/7 speech to text transcription.

[proven kettle]

i'm dumb and non-technical so i like the good ole' native interface i paid for. I'm happy with 20 bucks a month for the tool, i would just like it to 1. work and 2. have features released when they say they will be.

[uneven pagoda]

I was just confused about the token contextual thing. So the 32K is prompt and response.

[dense hound]

If your talking about gpt4 that’s exclusive to plus

[proven kettle]

not through the traditional Plus interface. 4,096. congrats if your API connection is doing more

[dense hound]

And it’s been fully rolled out

[proven kettle]

good point, i'm talking about 4 stuff

[proven kettle]

that's not correct, unless you're able to import images and audio into Plus

[uneven pagoda]

That too, an inclusive "hardly need to fiddle with" interface is worth it imo. I actually WANT to give them more of the little money I have. haha. I think this is quite worth it.

[dense hound]

Every plus user has it, so it’s technically rolled out. It’s just not fully delovoped

[proven kettle]

i guess that's one way of putting it. i actually wanted the moon roof they said i would get, tho. Ultimately it's simple: send a statement to users saying when it will be available for them. Tell new subscribers they dont' get that yet. Transparency is everything and we should be very loud when they start acting like 1990s Microsoft, which is exactly the bs they pulled this week.

and everybody should be going ap*shit when they are acting that way. this stuff is far too important to just leave to the highest bidders and easiest ways to capture market share (i.e., dropping 4 when Google is launching their stuff). this was just cut and paste nonsense this week, and i'm really dissappointed in how they are handling it.

[uneven pagoda]

What do you mean by this? I just have the text interface. I have no image capabilities..

[proven kettle]

you can read that sarcastically or as an Open AI employee lol

technically, it's rolled out. ;-)\

[dense hound]

Because it’s not released yet

It’s a W.I.P

[uneven pagoda]

Well why would that be?

[proven kettle]

i hear you, but again...they're not being quite forthcoming about that

but lots and lots of influencers are doing their marketing work for them. simply put, stupid behaviors, easily fixed, but a company has decided to forego that.

[dense hound]

It’s just not public yet

Can’t say anything else about it lol

[exotic iris]

@proven kettle where are you getting your information that you are forming that opinion on? that the devs are not being forthcoming?

[proven kettle]

Again I understand. But if I had a nickel for every IT leader who said “it’s done but not available” I would have like 35 cents.

[proven kettle]

The ambiguity about feature availability and an inability to get clarity on it. Source: me.

[dense hound]

I’m in the alpha gang 😎 Can’t say anything about it tho

[uneven pagoda]

Spill the beans!

[proven kettle]

Lol they sent me one of those nice mugs but that’s the limit of my street cred

[uneven pagoda]

I had all these graph picture I could have fed the bot

[exotic iris]

@proven kettle I have seen this. Give me a few.

[dense hound]

I never got a mug xd

[muted trellis]

what

[proven kettle]

I absolutely would love to be proven wrong, I’m being serious. I’m chaffed about this.

[proven kettle]

I don’t know what this makes me but they sent it in December or something lol

[dense hound]

Lol nice

[proven kettle]

I carry it around to impress the ladies

[dense hound]

Why did they send it to you lol, any specific reason?

[uneven pagoda]

That is quite..

0.35 cents times 1 billion is 350 million

[proven kettle]

They said I was part of some community that provided a lot of feedback. I thought it was a cool gesture on their part, another reason why this week is bugging me

[dense hound]

Oh weird

[proven kettle]

Oh I meant 35 cents total. Seven times. Lol

[dense hound]

You weren’t invited to the private build?

[uneven pagoda]

Can't provide feedback on what you don't have access to. Where's image bot and 32K token context??

[proven kettle]

Nah no reason I should be. I’m not technical. I just use it a lot.

[proven kettle]

Part of their announcement and boilerplate for about 5,000 influencer posts this week

[dense hound]

Probably plus only in the future

[exotic iris]

@proven kettle ok. here is what is going to be the eventual pricing model for GPT-4 and its image search capabilities https://openai.com/pricing#image-models

[proven kettle]

Hit me dude, I’m starving for this.

[uneven pagoda]

This tech is great

[muted trellis]

image search is for dall-e

[proven kettle]

So you anticipate another pricing tier for people wanting to use image and audio, or do you reckon some of that will be embedded for Plus users? I just figured they would take advantage of people to kick off a 100 dollar tier or something.

[uneven pagoda]

I ask it to create Ethereum smart contracts in SOlidity instead of writing common legal contracts and it did just that...

[proven kettle]

I don’t even have a particular need for image or audio. I just really want longer tokens in Plus.

[muted trellis]

this is the pricing for gpt-4, image support is for their partners, for public it comes later

[uneven pagoda]

I had a "sh88-ton" of candle stick graphs I wanted to feed the image bot

[proven kettle]

I saw a demo of someone doing this, pretty sweet how it identified that particular one as part of a pyramid scheme.

[exotic iris]

I'm not an expert here but they are charging a separate tier for api access where it seems to have that capability.

[proven kettle]

But what does that mean for Plus users

[muted trellis]

wdym? plus and api is different, if you want access to gpt-4 on chat.openai.com you pay $20 a month

image support again coming later

[exotic iris]

@proven kettle Pretty sure it means the status quo.

[proven kettle]

Nah man I appreciate the discussion. Lacking official guidance they could easily resolve, we are left to speculate.

[exotic iris]

I'm pretty interested in InstructGPT

[proven kettle]

I hope so. I definitely think they could charge more, I would just like that clarity (and larger token length!).

Mostly amazed that a company sending out mugs to slobs like me can’t simply state what you get with Plus, and when you will get it.

[uneven pagoda]

The people that will pay are those that see the value

[proven kettle]

Yep, I pay. It’s an amazing product.

[uneven pagoda]

People that don't don't. Their loss

and it isn't even that much

anyone can afford it

[exotic iris]

@uneven pagoda I mean I think that most people in this Discord jumped right in. 😆 😆

[uneven pagoda]

Agreed.

Now if perhaps 1 year down the line I learn how to make a great app or something

[proven kettle]

Mostly I think all of us should hold AI companies to a higher standard than typical tech nonsense. These tools are so powerful. Launching preemptively is a big red flag to me. Especially when open AI has been throwing a no hitter to date. This week was unnecessary.

[uneven pagoda]

that's potentially more revenue for their pockets

from my side

[hazy stag]

Was gpt-4 released under openai LP or openai inc?

[proven kettle]

Again, totally open to paying more for Plus for the increased functionality. I’m not a everything for free guy.

[proven kettle]

No idea

But like I was saying…I’m happy to pay. This is the coolest technology I have seen in my 50 years. I’m a productivity consultant and this will burn out about 30-50 percent of the majority of white collar jobs. As such, I want to spend money on ethical companies.

[uneven pagoda]

You know how they say something or other about ideas and execution? Well, I have tons of ideas but couldn't execute them. That's changing thanks to this technology.

[hazy stag]

Ironically the trained model uses almost no computational power, though adjusting the model is very time and energy intensive

[proven kettle]

I’m not smart enough to appreciate the power consumption, but that’s another metric to watch. I don’t want another bitcoin sucking energy.

[hazy stag]

It uses 0.4kwh for 100 pages

A raspberry pi can run it

[exotic iris]

@proven kettle I 100% agree with you on this. The fact that it can make Solidity contracts with minimal input by itself is extraordinary but I've had it complete such a wide range of tasks that would have been extraordinarily daunting otherwise.

[uneven pagoda]

Here's what I think about this. Just because it is 'cheap to run' or perhaps 'cheap to deploy' doesn't mean they can earn some money.

It's like you go to see a doctor, he sees you for 5 minutes charges you $200 and you get pissed.

It isn't about earning money based on time or whatever. It is about value.

[hazy stag]

Anyway this channel is useless theres literally no reason for it to exist since the team behind the server has no association with openai

[proven kettle]

It’s amazing.

[uneven pagoda]

TYPO, I meant that it doesn't mean they CAN'T earn money.

[dense hound]

Just pay ur $20 and quit complaining all of you

[exotic iris]

@uneven pagoda I'm curious, you said you had it write a Solidity contract for you. We were debating this last night here whether you would get better results using GPT-4 or GPT-3.5. Do you know which you ended up using?

[uneven pagoda]

GPT 4 but I didn't test or deploy it. I just wanted to see if it could do that. I noticed that GPT 4 passed some BAR exam at the 90th percentile, correct?

Well the reasoning on my part was that if it is legally smart than perhaps that legal smartness can be transferred over to an Ethereum smart contract. It did like 10 different contracts in like a minute.

I don't think many BAR exam passers that are expert blockchain bros exist.

[hardy iron]

I can't buy ChatGPT Plus from Azerbaijan.
I tried through 4 different cards from different banks and different people. Does not work. What is the problem? I asked someone from Azerbaijan can not buy. Maybe not open for Azerbaijan?

[exotic iris]

@uneven pagoda same reasoning here. It's really going to make a lot of roles obsolete.

[uneven pagoda]

This thing does everything. It's awesome.

This tech even as just a "copilot" can really amplify and augment everyone. There's no excuse in my view.

[exotic iris]

@hardy iron I can't link out for you but it says it should be available in your country.

[hardy iron]

Speaking is available. But it's not. Our payment cards do not work.

[exotic iris]

@uneven pagoda That's exactly how I feel like I'm currently using it. When I first came onboard, I was telling my wife that it's one of those pinnacle shifts forward where people that don't embrace it will be left behind.

[uneven pagoda]

Anyways back to questions. I have all sorts of images I was wanting to feed the image GPT portion of the bot. Whatever it is called.

[exotic iris]

@hardy iron Sorry, I'm not an expert.

[uneven pagoda]

When is imageGPT being deployed??

[hardy iron]

Where to find the marketing department of OpeAI?

[exotic iris]

@uneven pagoda I think it will just be presented upon us one day like mana from heaven. 😆

[exotic iris]

I think your best bet is emailing [e-mail]

[hardy iron]

They don't answer

[ivory imp]

is there a way for him to remember something from another chat? or export the chat

It's just that a chat of mine has a lot of things in it and it's lagging and ChatGPT is also giving errors

[uneven pagoda]

That's what I'm saying there are literally no excuses that anyone can have. Even the free version is capable. None of this is unaffordable or beyond the reach of anyone in the USA.

[silver heart]

Hello everybody, is there anybody tried to use GPT4 ?

[uneven pagoda]

Well hopefully soon

[smoky tree]

chatgpt cant play chess anymore?

[uneven pagoda]

Anyone else see this, "GPT-4 currently has a cap of 50 messages every 4 hours. Expect lower cap next week, as we adjust for demand."

[exotic iris]

@uneven pagoda I still just see this "GPT-4 currently has a cap of 50 messages every 4 hours". I don't like the sound of your message though.

[uneven pagoda]

haha maybe i'm using too much

[untold perch]

Yes and even with that cap the model is 🐢

[charred shale]

wasn't it 100 messages?

[uneven pagoda]

True, and that in itself limits how much you can use it per month. Some people think that 'speed' doesn't matter. I think it does. I think even Steve Jobs thought it was important.

[untold perch]

Yes it used to be, I suppose the demand is higher now

[charred shale]

I see, fair enough

[uneven pagoda]

I don't know what their costs are say per individual user.. I assume they have it all figured out.

[untold perch]

Definitely when your reading speed is faster than the models speed

[uneven pagoda]

Good point

But I think its also relevent in api right? speed

[vapid bear]

Lol.

[uneven pagoda]

But maybe that is not necessarily true though. I don't use Bing chat because it was painfully slow.

Unlike chat gpt I don't always read sequentially. I skip forward perhaps even a paragraph.

[untold perch]

I forgot about all those "Although I'm not an expert in..." and "Always contact your..." paragraphs 🤣

And skimming is nice, so yeah preferably way faster than reading speed

[uneven pagoda]

Yeah, the juicy parts are like sentences into the response.

But I like the verbose answers overall

[untold perch]

Idk if I would ever contact my doctor for how much water I should drink daily, sometimes it's a bit much haha

[uneven pagoda]

At any rate, I'm not rich so I can't pay hundreds of thousands of dollars for my specific wants.

[untold perch]

But I see why they do it for sure

[uneven pagoda]

Unless it is for highly specific needs. Perhaps chat gpt can help in this regard too.

[uneven pagoda]

I imagine some sort of trade off between speed/cost?

[lilac tinsel]

how i can i generate images ?

how to start ?

[untold perch]

No clue actually, would make some sense

[untold perch]

You can try here
https://labs.openai.com/

[lilac tinsel]

tnx lemme try it

[empty dew]

So if I paid for an account (CHATGPT+) do I get access to the playground on openai.com?

I don't know why but my account is still showing the "Upgrade" button on Openai.com but on chatgpt when I go to the "Manage account" it shows that I'm on the plus account.

[prisma gate]

Same thing for me and therefore I cannot get the API Keys that I need.

I also am having trouble resetting my password on the Platform.openai.com

[viral acorn]

I must be missing something easy for others to see... but to my understanding I should be able to get a subscription to ChatGPT plus for $20/month... except when I try I only end up at this page which is for companies getting GPT services in prices designed for companies. I am not a company, I am just an old and getting slower in thinking retired guy. How do I get ChatGPT Plus for $20/month or is it really not available to me?

[proven kettle]

i'm a Plus subscriber, i think they may be limiting new subscribers since they launched GPT 4 and were unprepared to support it for their existing customers, as well as new ones. but yes i'm old and slow, and i have had an account. Others here might have a faster link for you, assuming you've created an ID there.

[proven kettle]

you can always use Bing, if you have been given access to chat there, while they stabilize things.

[viral acorn]

Sometimes I feel like my brain is running on low gas, but I want the ChatGPT subscription which should be able to be on 3.5, but I can't find the link to join... it does not give an option or say it is overloaded, it just gives options for Big Business as opposed to private individuals.

[proven kettle]

basic, reasonable, and effective communication has not been their strong suit this week. i feel you.

the links you are trying are likely the ones that should work. but their system seems pretty junked right now, and i do remember getting vapor links when i tried to sign up others a few weeks back. it's not just you.

[viral acorn]

Thanks, at least I know I am not the only one having trouble.

[rigid musk]

No one . And dont ask here. It is against tos

[rugged wing]

i rlly dont care

[viral acorn]

Neok Maid, what? I can't ask on OpenAI about how to get an OpenAI product they serve? That makes no sense.

[rigid musk]

Thr reply wasn't for you

[near badge]

look up concepts of prompt engineering. Understand what ethics teams use to break the chatbots and then use critical thinking to develop your own attacks

[viral acorn]

Oh

[proven kettle]

out of curiousity, what happens when you hit this link http://chat.openai.com/

[viral acorn]

tRYING

[spiral fossil]

Text updated.

[proven kettle]

gotcha. well, i think that link should take you to a login prompt. if it's not, that's their inability to plan well with new product releases and their network is stressed in general. and they opened up to international payments for added complexity to handle. I get the sense they are flipping switches as they watch performance, and you may eventually get through.

[proven kettle]

as dumb as it sounds i applaud them for actually telling us what will happen, that's different from the rest of this week

[viral acorn]

It gave me a different path, signing in

[spiral fossil]

Yes

[exotic iris]

well damn it's finally come for me too "GPT-4 currently has a cap of 50 messages every 4 hours. Expect lower cap next week, as we adjust for demand."

[proven kettle]

nice! hopefully when you sign in you see an option on the left hand side to manage your account

[umbral hornet]

How do I fix this error when using the GPT-4 API? This is a chat model and not supported in the v1/completions endpoint. Did you mean to use v1/chat/completions?

[viral acorn]

Now I need to go find where I wrote my password... never grow old if you can avoid it.

[proven kettle]

😀

[proven kettle]

i hear you dude, good luck. it's worth it once you get through, even if they just decided to let loose a pack of hyperactive giant weasels in a room this week to see what would happen

[exotic iris]

@proven kettle I guess I have to make sure I'm maxing out then until next week 😆 😆

[proven kettle]

lol i honestly don't do much back and forth with it, having used gpt for a little while now. i just want that extended token response so bad in Plus. I just wish they'd sell us our own DBs we could train with imported data to complement the general model. Just like they are with Foundry in snapchat, etc...plenty of folks would pay money to have a place to dump their work or personal stuff and have it part of the ecosystem

it did do a hell of a job in creating activity-based costing methods to take out various roles. I'm gonna write up an article on that for folks who feel secure in the jobs.

Here's a snapshot of how it breaks down a typical corporate lawyer's activities and their ability to be replaced by current AI tools

crazy

[oblique condor]

lol do you use premium?

[proven kettle]

this is hypothetical but it came up with 44% for a corporate lawyer. it's estimating, but i promise that number is not 0%

happy friday, future universal basic income supporters!

[uneven pagoda]

hahaha

[floral sail]

so my chatgpt is down in laptop browsers, working on mobile fine, just not on brave, chrome, firefox, when i go to chat.openai.com, this message comes,

[near pulsar]

Do you hate the world?