gpt-models

[cloud geyser]

Google translate is working for me

[serene cipher]

fooled indeed

[cloud geyser]

This plugin can also use OpenAI API to provide translation services or directly call ChatGPT Plus to translate, but the latter is obviously much slower

[serene cipher]

which plugin is that?

what is the name of it?

[cloud geyser]

The translation ability of GPT-3.5 is indeed very powerful, but it is a pity that sometimes it will unconsciously add its reply to the original sentence to the translation result, and it will not change after repeated teaching, unless I add a reminder in front of each sentence

[cloud geyser]

Immersive Translate

[serene cipher]

interesting, but that sounds like something that can be solved with fine tuning

[cloud geyser]

Oh, I have written many versions of the prompt words, but the effect is always mediocre. I really don't want to type the prompt before every sentence, even if I copy and paste it. Because it's fine once or twice, but it's exhausting to repeat.

[abstract elbow]

You are a life saver, Can confirm, same issue, Works in Firefox, not brave (currently.)

[cloud geyser]

Interesting, someone added the GPT-4 API interface to Galgame

[ocean solstice]

I set temp to 0.2 and incremented it from there and it still gives me random nonsense even though I feed the current time haha

GPT-4 is handling it just fine

[unique tundra]

Are they going to increase the message limit for gpt-4 since legacy 3.5 is gone?

[serene cipher]

if you don't mind sending the messages, I can try to help =)

[unique tundra]

25 messages even after months of gpt 4 release is crazy

[cloud geyser]

Sometimes, the conversation is longer, and even GPT-4 will be confused (for example, some things I requested are missing in the output)

[serene cipher]

i don't think there are much correlation between those two things tho

[short sorrel]

yeah i wish they would remove the usage cap since they removed 3.5 legacy

[ocean solstice]

I cut off how much context I feed it.

[ocean solstice]

un momento, is it okay to ping you in API discussions? Seems more appropriate there :D

[serene cipher]

sure, please do

[cloud geyser]

That's really weird

[supple violet]

barely no one used legacy so it has no significant effect on anything

[serene cipher]

it still has a limited context, double of what 3.5 has, but still, not infinite

[short sorrel]

i only used gpt-4 and 3.5 legacy for programming because they are both a lot better than 3.5 turbo

and i suspect a lot of other people have the same experience

[serene cipher]

it is normal, any model will "forget" things that are too far on the chat history

because it is literally not present anymore on the data the AI has about the conversation

[supple violet]

legacy was behind plus and it was not available in api so the usage was insignificant, most likely less than 1% of global usage

[glass haven]

Speed seems a bit improved now compared to past 18 or so hours

[lean prawn]

A multi-eyed monster in the form of a muscular giant

[cloud geyser]

So it's better to urge Nvidia to hurry up and fulfill their promise of dedicated graphics cards for large language models

[ionic isle]

Is anyone else getting the error "failed to fetch user settings"?

[short sorrel]

yeah im sure it doesn't have too many users but its just dissapointing that now my Plus experience is so much worse for programming problems. Turbo doesn't understand my problems nearly as well as the other 2 models. Ideally GPT-4 would get the usage caps removed because it blows both legacy and turbo 3.5 out of the water for programming tasks

[cloud geyser]

This is a chat room, you are in the wrong place

[short sorrel]

I'd even be willing to pay so much more money for a Plus chatgpt subscription that doesn't have GPT-4 usage caps. I guess I could use the API but its nice to pay a fixed cost every month

and not have to count/worry about tokens

[cloud geyser]

Unfortunately, I've almost spent my conversation count...

[supple violet]

higher usage tier sub is coming or its included in the business subs

[knotty grotto]

Uwu my pleasure

[cloud geyser]

That will cause OpenAI to be more afraid of web page reverse engineering. If someone wants to, they can use unlimited dialogue to do something

[trail lake]

me

[cloud geyser]

It seems that there are still regular active people here.

lol

[ionic isle]

Ok, it's not just me

[marble heart]

How can you do that?

[cloud geyser]

Oh oh oh no! I also encountered this problem!

The error message is like an avalanche

[ocean solstice]

I'll be straight, you have to use it A LOT to get past $20 in API usage.

[short sorrel]

hmm really? might be worth checking out then. Are there any good frontends for it? stuff that works for programming?

[ocean solstice]

You mean wrappers?

[amber parcel]

Anyone here getting the user settings error

[short sorrel]

yeah just something i can paste my api key into and get a chat gpt like experience

[amber parcel]

Seems new

[ocean solstice]

Oh I see. I'm sure there are some but we can't really promote third party tools like that since your API key is sensitive.

You could always use playground though

https://platform.openai.com/playground?mode=chat

[cloud geyser]

'All Systems Operational' lol

[ocean solstice]

Though I lowkey want to make a simple tkinter gui now

[cloud geyser]

Looks like OpenAI doesn't know what's going on yet

[final cliff]

Hey! Quick question: I joined the "Plugins" waitlist a bit ago. How long should I expect to wait until I get access to them? I assume there's probably a bunch of demand and at this point it probably takes less time to wait until the full release, but I'm curious how much time it has taken other people to have access.

[cloud geyser]

I think it's up to you to tell them what you want to do with the plugin or if you're an early plugin developer

[cloud geyser]

I applied on the afternoon of the second day after the plugin was launched, and there is no response until today

[tardy heart]

on gpt4

i see people complaining here about the 25 message cap, if you do that with any sort of context you'll burn through money on gpt4 api

[normal quest]

Where do we find plugins / how do we use them? I know they got released but I havent seen anything on them since

[cloud geyser]

Doesn't matter, I don't use a lot and don't need a long context

[verbal wolf]

im getting Failed to fetch user settings

[tardy heart]

i went through 60 last month

[final cliff]

Thanks for the answer! I applied a few days ago myself so I'll assume I won't get access for a while

[cloud geyser]

i think almost everyone does

[tardy heart]

im at 15 already, i hardly ever use context

and dont use it all day by any means

[verbal wolf]

you hacked into my account?

how did you know?

😭

[tardy heart]

i do test a lot of projects though so

[cloud geyser]

Many people have complained

[tardy heart]

¯_(ツ)_/¯

[cloud geyser]

including me

[cloud geyser]

The error message is like an avalanche

[ocean solstice]

I guess it depends on how much you use it.

[cloud geyser]

It is not easy for people who use a small amount of conversations, but once the limit is reached, it will be extremely painful, as if rejected by the world (laughs)

[thorn bronze]

Me too. I got “failed to fetch user setting”. I thought it was Text Blaze that caused the problem!

[amber parcel]

Its happening to a bunch of people

[spare hearth]

Hey huys. Is there any way to implement free chatgpt chat on my own website ? Like, a bar hanging there to "try" the feature?

[cloud geyser]

If you still have free API quota, you can try to generate an API key, but I don't recommend you to do this without paying, it may cause the official to think that you are abusing the API and seal your account

[tardy heart]

sure if you pay for the api

[spare hearth]

thats a good advice

thank you for that. Its a simple thing to do?

[cloud geyser]

You can go to Youtube to watch the instructional video

[spare hearth]

ok thanks

[grim bough]

does the gpt-4 API seem slow to anybody else right now?

[outer flax]

It seems like it could be done all the tech is there, just has to be combined

[marsh kestrel]

Yes, the gpt-4 API seems slow to many people at the moment.

[serene cipher]

it is slower than usual today indeed

[whole kettle]

SlowGPT

[serene cipher]

it is not turbo

[wraith ginkgo]

`` Server Tip: ``

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

[cyan palm]

When will plugin be available tho?

[hasty egret]

ya, dat have a different BPM

[wraith ginkgo]

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

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

[strange arrow]

My account says that it was blocked, I am trying to find the support team email or something as I am trying to get it unblocked, I believe that it was a simple misunderstanding, I was trying to get a translation for something and ChatGPT really didn't like it apparently

[hasty pulsar]

When will the plug-in be publicly available

[wind urchin]

#ReleaseThePlugins

[supple violet]

disable vpn

[strange arrow]

TY!

[wraith ginkgo]

This is a Safe for Work Discord Server

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

Posting NSFW content will result in a permanent ban.

[small vector]

Not really when the primary use is gpt-4. Especially when I have summarisation tasks.

[river canopy]

chatgpt is done rn right? or is my thing jus broke rn

[inner trail]

Your thing is broke rn

[river canopy]

well shoot

[paper yarrow]

Tell me, I did it through the curl bot, but he does not remember his previous words, how can I fix it?

[wintry ingot]

Great, ChatGPT4 is taking too many requests

[wind urchin]

same error on my end :p

[wintry ingot]

;(

[hot merlin]

I just got access to the plugins but I have some questions on how to use them. Who could help?

[dire citrus]

Guys please stop using chat gpt I need to read the end of the fan fic

[wind urchin]

Pretty sure you cant

[compact wraith]

To me GPT-3.5 feels braindead after using GPT-4 for a while

[wind urchin]

Relatable

[visual heart]

is me or chatgbt 4 is down

[compact wraith]

also sometimes I like the human speed of GPT-4 even though that may not be intentional, and sometimes I don't

[arctic trellis]

How does one acquire access to the plugins?

[paper yarrow]

So is it impossible or is it possible?

[wind urchin]

It's a human, we all have a personal GPT employee

[wintry ingot]

GPT4 says "We're currently processing too many requests. Please try again later."

[hot merlin]

sign up for their waiting list

[floral topaz]

it's down again!

[radiant prawn]

yeah same...

[turbid chasm]

https://openai.com/waitlist/plugins

[arctic trellis]

Thank you!

[visual heart]

hope thjey do not decrease the 25 message quota

[wind urchin]

I would die

[visual heart]

50 would be enough for me

[floral topaz]

25 is nothing they better increase not decrese it

[visual heart]

or at least make it 2 hours or 1 hour

[compact wraith]

Would plugins be able to provide the model a more up to date information?

since 2021 is starting to show its age...badly

[wind urchin]

There's a google plugin

[compact wraith]

Especially for coding

[worn epoch]

gpt 4 is down

[visual heart]

yeah it is showing it age when I asked a question for microsoft al for business central

[gaunt seal]

How am I supposed to do my fiverr projects now, whaat

[wind urchin]

LOL

[ripe condor]

finally i can sleep

[radiant prawn]

its 2pm

[tawdry slate]

anyone else getting "We're currently processing too many requests. Please try again later." on GPT4 only?

[radiant prawn]

everyone

[tawdry slate]

Sorta weird considering we literally pay for this no?

[radiant prawn]

its down like a clown

[hallow swallow]

I came here looking for the same thing. Chatgpt got everyone to do something they would never do

[radiant prawn]

not enough

[hallow swallow]

download microsoft edge

[compact wraith]

I already have access to bing's AI but I prefer GPT-4 a bit better

[worn epoch]

i need to pay my doubts

[dark ivy]

me too having the same problem

[wind urchin]

Bing ai is gpt 4

[misty zephyr]

I think it's just down

[tawdry slate]

lol ok, well one of the things they are advertising is constant access, irrelevant of how many people or "requests" they are processing. So I think I'm right in thinking that is' weird

[radiant prawn]

pretty sure bing uses the standard model

[compact wraith]

But the language model is different right?

the training data is different

[magic silo]

Yes

[radiant prawn]

are you sure?

[wind urchin]

No, it's the same gpt4, bing can use bing searches though, so it is realtime

[radiant prawn]

it doesn't have the same understanding as GPT4 and reasonsing from my use

[wind urchin]

https://blogs.bing.com/search/march_2023/Confirmed-the-new-Bing-runs-on-OpenAI’s-GPT-4

[magic silo]

Chat gpt is advance

[wind urchin]

We are happy to confirm that the new Bing is running on GPT-4, which we’ve customized for search. If you’ve used the new Bing preview at any time in the last five weeks, you’ve already experienced an early version of this powerful model. As OpenAI makes updates to GPT-4 and beyond, Bing benefits from those improvements. Along with our own updates based on community feedback, you can be assured that you have the most comprehensive copilot features available.

[tawdry slate]

it's the same model, you can look it up. The config is different, temperature and initial system message are different

[light lava]

is gpt-4 down

[patent mica]

seems so

We're currently processing too many requests. Please try again later.

[radiant prawn]

I see thanks for clarifying and the link

[tawdry slate]

yes, gpt-4 is down. and they should offer some sort of token pack or something in excahnge

[radiant prawn]

thanks

[light lava]

yes

i am getting the same msg

@patent mica

[trim trench]

Yep me too

[keen blaze]

Omg omg chat4pt is not working for me, am I like the one ???
||jk I can actually read a chat before typing||

[tawdry slate]

@compact kayak ?

[hallow swallow]

Is the paid version of chatgpt the same as the one on Bing?

[hallow swallow]

It's down

[tawdry slate]

this chat is filled with all sorts of different topics. why would you assume that somebody is going to scroll through in hopes of finding the topic they want to discuss?

[glacial flare]

We're currently processing too many requests. Please try again later.

[tawdry slate]

this isn't reddit

[glass haven]

No not really. I think the paid ChatGPT is much better

[hallow swallow]

🙏

[trim trench]

GPT4 is better at responses. But I find Bing is much much better at citations haha

[sinful furnace]

No, I prefer the paid version. It’s definitely better. Sometimes I’ll use both simultaneously to tweak things, but the paid chat is worth it

[granite stump]

Anyone else getting "We're currently processing too many requests. Please try again later."

[patent mica]

I noticed Legacy 3.5 is gone

[sinful furnace]

Yes

[patent mica]

today, only 3.5 default and 4, they did something today

[granite stump]

Thanks

[sly grotto]

yes

[hallow swallow]

I keep hearing the same things but dont know the difference. Feels like the output config is just different - one is smaller on Bing and the other is on larger

[granite stump]

Killed legacy?

[patent mica]

seems so, it's not on my dropdowns anymore

[astral sentinel]

Hi, good day. I bought ChatGPT Plus today I wasn't able to send images to GPT-4 or to GPT-4 generate images for me. It's normal?

[granite stump]

Same they did say it was to be depreciated soon

[hallow swallow]

Yeah thats normal. You can only send or use images with API

[patent mica]

been on a plus a while, asked to be listed for this feature multiple times, still don't have it, is not rolled out for everyone

[granite stump]

Text only

[sinful furnace]

Yesterday was the last day to use the legacy version

[candid minnow]

ooh okay! thanks!

[astral sentinel]

Ok, thank you guys

[sly grotto]

any idea how long "We're currently processing too many requests. Please try again later." usually takes?

[normal turtle]

Usually only a few minutes

[sly grotto]

awesome, thanks

[patent mica]

no one knows, don't think anyone here works for OpenAI, all are volunteers

varies, last outage was about 6 hours

[patent mica]

interesting

[sly grotto]

oh snap. Okay, just trying to plan for it. I just started using it yesterday. So much to learn, its exciting.

[sharp canyon]

Hi everyone ! Receiving this message "We're currently processing too many requests. Please try again later" while i'm paying for chatgptplus any ideas to solve that problem ?

[patent mica]

3.5 working fine

[trim trench]

Someone here has gotta work for them if this is the official dicord? Also welcome @sly grotto I use it every day lol, google is right to be worried

[oak maple]

gtp-4 down so just wait

[sharp canyon]

ty, an idea for how many times it gonna takes ?

[sly grotto]

I wish i could move my conversation to another version.

[oak maple]

idk but i think they started to fixing so its should be soon

[sharp canyon]

ty

[floral topaz]

keep regenerating until it will offer you the option to downgrade

[patent mica]

I feel u, had to restart a prompt on a very long covo I use for assignments

[sly grotto]

@floral topaz oh thats cool, ill try that. thanks

[patent mica]

"PAIN"

[sly grotto]

truth, haha

[severe crater]

is gpt4 down

[oak maple]

seems that gtp4 is back for me but only on new conv

[wind urchin]

it's alive now for prev convo

[fossil tiger]

👍

[oak maple]

ye

[sly grotto]

Yay, that was quick.

[hushed steeple]

delet this

[wind urchin]

No

[coarse rune]

Hey, has anyone found success using GPT4 API or ChatGPT for academic research? I may have an opportunity to present the uses and implications of AI in research to a few research scientists. I'm trying to put together examples and information that would help show off how they can use AI right now. Any advice or resources would be greatly appreciated!

[inner trail]

Anyone else feel that GPT-4 has been better in this latest update of May 3 version.

[patent mica]

hard to gauge this, been the same more or less for me

[inner trail]

Yeah same for me, but i can feel a little bit more creativity in the responses, not sure how and why

[versed tree]

[full pebble]

Try searching for an app that uses GPT4, but allows it to connect to google to do this task for you. I know one app but moderation wont allow me to say which one that is ....

[lapis nimbus]

where tf is the legacy model ?

[spark wraith]

actually have used it specifically in getting sources for a given topic and maybe an outline, its honestly a great way to get started.

[inner trail]

It's gone forever

[spark wraith]

discontinued after may 10th

[lapis nimbus]

why?!?!?!?!

its so mich cooler than the other 2

[inner trail]

Yup, i liked Legacy too

[lapis nimbus]

What was their reason to remove it?

[spark wraith]

How so? Legacy just seemed like a slower version of 3.5 which it kinda was

[inner trail]

It was very old i guess

[lapis nimbus]

legacy was the fastest

gpt4 ist slow and ive never used the standard model

[inner trail]

Legacy was pretty fast yeah

[lapis nimbus]

hmm sad

[spark wraith]

interesting I thought it was slower than standard welp i guess i cant cross check anymore

[lapis nimbus]

i have never used standard so im not sure

[patent mica]

legacy is kinda dumb compared to the other two though

[lapis nimbus]

ohh really ?

[wind urchin]

Hello, has someone experienced and account termination ? It is an error and I can not find any link where I can request the reactivation of my GPT4subscribtion.

[lapis nimbus]

i thought the satandars one was the dumbest one

[patent mica]

I'd say so, legacy is a chimpanzee to gpt4.. well smarter chimpanzee

[lapis nimbus]

standard*

yeah i know gpt4 is way more smart but i thought that legacy was better than standard

not just faster but also better

[polar solar]

Is it working well right now?

[thin tulip]

Why’s it saying 'working' in a light green box and shows code

[fossil tiger]

/queen

[weary dust]

im so spoiled from gpt4 i hate when my 25 messages are up. like bro ill pay 100 dollars a month for unlimited messages whatever it takes, i cant go back to garbo 3.5

[raven panther]

Can't you use the api instead?

[weary dust]

i have no idea how to get an api or use it. i dont have experience with python.

[fossil tiger]

/question

[inner trail]

I do feel the same, atleast they have to make 50 messages for 3 hours - 25 is way to less

[radiant talon]

why did openai charge me 24$ instead of 20$? :DDDDDDD

[spark wraith]

taxes?

[inner trail]

Are you from US?

[polar solar]

is it working well guys?

[radiant talon]

all previous months it was 20, and im not in the US

[raven panther]

you don't need to use it programaticaly. There's a "playground" on the openai site where you can use it like chat. But there's still a waitlist

[polar solar]

For me is not

[supple violet]

certain taxes like vat don't need to be collected in different countries until you hit x revenue and plus is pretty new product

[weary dust]

that sounds cool. but i still dont know how to find the playground or where to apply for the api. i tried have gpt explain it to me but it is still stuck in 2021 and doesnt even realize it is gpt4 lol.

[patent mica]

and payment provider fees

[radiant talon]

this is my 3rd payment for chatgpt plus

first time overcharging

[supple violet]

which country are you in? $24 sounds like 23% vat

also you can download the invoice pdf which will show it

[lapis bramble]

step up your game, OPEN AI, im not paying 20 per month to have all these limitations with using gpt-4 , and then not even have access to plugins , Why wouldnt i just use BARD, Atleast give us unlimited access to gpt-4

cant even finish a project cause every 15min or so of convo i get capped for 3 hours

[inner trail]

It is what it is

[elder token]

lol use 3.5?

[inner trail]

Then don't pay $20 monthly, use GPT 3.5 which is free

[lapis bramble]

lol, try starting with 4 then go to 3.5 , its clear gpt4 is much smarter

plus im in the middle of a project, you cannot switch it to 3.5 while maintaining the same level of help

[elder token]

git gud then?

[lapis bramble]

if you want to go around in circles use 3.5 , i mainly use for programing

[hallow swallow]

i bought it just because it's cool lol

[elder token]

you should really be learning to code without gpt silly....youre essentially a script kiddie

[hallow swallow]

sucks i cant get access to plugins :/

[inner trail]

Lmao, truth is spoken

[blazing shard]

did chatgpt get an upgrade or something

it so snappy right now like I was shocked

[green zinc]

so theres no access to plugins even if you get gpt plus?

[inner trail]

Nope

You have to join waitlist

[green zinc]

but that could take forever

[inner trail]

I know, its taking forever for me already

[green zinc]

wait someone said Bard is available? if theres plugins maybe ill look into that

[wraith ginkgo]

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

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

[normal quest]

How do we utilize plugins? I saw they announced it but dont see a place to add plugins or use them?

[radiant talon]

I remember OpenAI's lies about "Expect lower cap next week as we adjust for demand" for GPT4 lolololol.
This was in March btw. MARCH.

[tardy heart]

in fairness

openai is seeing a growth no other company ever has

with the scale of hardware they need

[radiant talon]

no doubt about that but there's no need to lie to their customers who hand their money to them

[tardy heart]

the cap has lowered before on bad demand days

its not a lie

[shadow shale]

I joined the waitlist on april 20th but no gpt4 key yet. granted I do not have the chatgpt plus service. Is that why my wait time is longer?

[sonic dagger]

ok

[wind urchin]

Do you think OpenAI has intentionally been slowing down releases to calm down the AI risk panic with the 6 month letter and all?

Because that panic happened after they released GPT-4, Plugins, and Code Interpreter within a very short time span.

Or maybe they just have trouble scaling at their massive level of growth. I have no idea.

[pastel fox]

I'm a subscriber and I've been using GPT-4 to generate summaries of studies from links, which has been extremely beneficial. However, for the past few days, I've been experiencing an issue where the system switches to GPT-3. Even when I inquire about it, it claims to be GPT-4 but no longer provides me with summaries from links, stating that it is incapable of doing so. Then, when I ask about the version again, it reverts to stating it's GPT-3. This situation is highly frustrating.

Has anyone gone through this here to help me resolve?

[balmy storm]

You're wait time is not longer or shorter, the waitlist just takes a long time usually.

it took me more then a month to get api access

and I am a plus subscriber

[tardy heart]

It's never been able to grab links unless you had the internet model.

Gpt4 is unaware it's gpt4.

[shadow shale]

Thanks been wondering

[pastel fox]

Okay, if you're right, he simulated very well having read the links and even replied as follows: "According to the link you passed...", or "I couldn't find what you're looking for in the link.." and many cases bringing summaries. I'm not an expert in depth, just another common user. If he didn't read it, he only simulated it, he fooled me perfectly. I also didn't know that GPT-4 doesn't know it's GPT-4.

[steel turtle]

Agreed. It's balancing context better.

[steel turtle]

Are you sure GPT4 is unaware of itself? When's the last time you tested?

That seems to be changing for me and I can call it GPT4 now and it knows I'm addressing it.

At first, I agree -- It wasn't aware GPT4 is released and referred to it as a "hypothetical model", now it stops complaining about that and operates if its released.

[wraith ginkgo]

<:book_icon:1046291503158399107> `` Rule 4 `` No spamming.

Do not spam or flood the server with messages, as this is disruptive. Mass pings, all-caps messages, repeatedly posting in a short amount of time, or making very similar posts sequentially are considered forms of spamming.

[tardy heart]

it's still an llm and will respond differently at times.

i asked it two different ways and got two different answers

just now

[sweet thistle]

GPT-4 still has the same knowledge cutoff as the previous models, as seen on https://platform.openai.com/docs/models/gpt-4 (This cutoff being September 2021)

As a result, it isn't aware of its own release or existence. You can inform it, but it's not knowledgeable on it to start with

[inner trail]

Why didn't they release GPT-4 with 2023 knowledge

It would make much more sense, since its a newer model

[sweet thistle]

The browsing models allows for updated information! The main issue with making an up to date model is you'd need realtime filtering of 'bad data', along with biases and misinformation, which is kinda impossible for now 😅

[steel turtle]

Ok well 60 days ago the response to this was pretty consistently about hypothetical models and a standard canned response about GPT4, but now it responds hello every time... so I definitely see a difference 🤔

User:
GPT4, say hello

ChatGPT:
Hello! How can I assist you today?

[sweet thistle]

Based on context, it likely understood that you were referring to it as GPT-4. It is aware of its own knowledge cutoff and the current date

[white kestrel]

have anyone noticed that opening the plugin store in the chatgpt website then closing it will cause high cpu usage within the web page?

[sweet thistle]

Try a fresh conversation, asking when GPT-4 was released, or try asking it to say hello as if it were GPT-5 (which does not currently exist)

[sweet thistle]

Maybe due to scripts running on the client side, interesting though 🧐

[tardy heart]

so want to share this, it does sometimes know

[white kestrel]

seems pretty unoptimized🤔

[tardy heart]

i just asked it three times, two times got the 2021 cutoff excuse

this last time i got this, brand new convo

GPT-4, or Generative Pre-trained Transformer 4, is a state-of-the-art artificial intelligence model for natural language understanding and generation. It is an advanced version of its predecessor, GPT-3, and is developed by OpenAI. GPT-4 is built on the transformer architecture, which is a type of deep learning model designed to handle a variety of tasks, including machine translation, question-answering, and text summarization.

GPT-4 is trained on a large corpus of text from the internet and can generate human-like text by predicting the next word in a given sequence. It is capable of understanding context and can engage in conversation, answer questions, or provide recommendations based on the input it receives.

As an AI language model, GPT-4 has been used in various applications, including chatbots, content generation, virtual assistants, and more. It is important to note that, as an AI model, GPT-4 has its limitations, and its knowledge is based on the training data it has been exposed to, with a knowledge cutoff in September 2021.

i have zero scripts, no plugins, nothing

[white kestrel]

the system message in gpt4 literally told it all these

thats why it knows

[tardy heart]

very likely

but it seems it "knows" somehow at least

[steel turtle]

Well sure, if I give direct context, it'll give me the canned response we expect. I think it's... considering it less as a high context variable when used or something.

But you're right, that doesn't directly translate into awareness of self, probably a bad call on my part to relate it to that.

[tardy heart]

system message seems the most likely to me

[steel turtle]

It definitely used to complain when I called it GPT4 before, LOL

[tardy heart]

they may have added it to the system due to that

[steel turtle]

GPT = fine. GPT4= offended, canned script 😄

[high bloom]

I asked GPT 4 for a pickup line it gave me but GPT 3 didn't

[tardy heart]

yep i asked it a few questions and threw it off

[sweet thistle]

Interesting, a lot of the information within this output seems generic though, maybe a hallucination? Either that or system message as the two of you mentioned

Btw, #1050184247920562316 would be the best place for this sort of thing (and then linking the message back here) as to not clutter chat. I won't be deleting your message but just letting you know for future reference 🙂

[tardy heart]

it thinks its gpt3 still

[white kestrel]

when it first released to chatgpt plus users, they didn't include the gpt4 info within the system message, and everyone is asking if its really gpt4

[unkempt stone]

GPT-4 just told me to use it to plagiarize text by pasting it into it when I asked it for a list of features 💀

10. Paraphrasing: I can rephrase text to convey the same meaning with different words, which can be useful for improving readability or avoiding plagiarism.

[stoic horizon]

you can call it Bob, and it will still say hello

[tardy heart]

User
How do you know about gpt4, it came out after 2021
ChatGPT
I apologize for any confusion, but as an AI model based on OpenAI's GPT-3 and with a knowledge cutoff in September 2021, I don't have direct knowledge or information about developments such as GPT-4 or other advancements that have taken place since my last training data. I can't learn or know about events, technologies, or information that have come into existence after this time.

The ability for me to refer to GPT-4 in our conversation is based on the context of your questions. Given that GPT-1, 2, and 3 exist, it's a reasonable inference to discuss a hypothetical GPT-4 in the context of those existing models. However, I don't have specific information about a real GPT-4, if it does exist after my knowledge cutoff.

For the most accurate and up-to-date information about GPT-4 or any other recent AI developments, please refer to the latest releases and publications from OpenAI or reputable news sources in the field of AI.

ah yeah good point kesku, will move on 🙂

[steel turtle]

LOL yes. But that doesn't change that 60 days ago, if I called it GPT4 I got canned script and if I called it bob I didn't. It's weighting of context is getting better.

OR someone directly changed the programming to weight it less, I'm far to dumb to spot which 😄

[tardy heart]

I am curious of the constant training, i have noticed minor differences recently

in how my existing responses work on my api related things for gpt4

its really intresting

[unkempt stone]

Guys is this not pretty problematic? Encouraging unethical use when asked to provide a list of features?

[tardy heart]

the beauty of llms

it tells you what you prompt it basically

[unkempt stone]

I just asked it to list what complex tasks it could perform though

[tardy heart]

on that note it saddens me how many people in here are here to cheat on homework 🪦

[unkempt stone]

with no previous convo

[sweet thistle]

That is quite odd, especially since you didn't specifically ask for such information 🧐

If you get a chance, maybe forward your findings to the form at https://openai.com/form/model-behavior-feedback

[high bloom]

GPT 4 has less limitations

[unkempt stone]

I will do that, thnx

[sweet thistle]

Thank you very much!

[tardy heart]

saving that link

[steel turtle]

I'm not exactly sure "avoiding plagarism" is unethical in itself

[tardy heart]

it's amazing how moldable gpt4 is to your use case with system instructions

i made it as a proof of concept read a modmail and categorize it for me

[steel turtle]

I think you might be identifying some emotions to the statement that weren't there. Sorry if that's rude to say, I don't mean any judgement i the statement.

In PhD college level classes, you generally get at least one entire class, sometimes more, on how to provide scientific value without plagiarizing those that came before you. One of my classes was actually CALLED "Avoiding Plagarism"

[sweet thistle]

👀 semantic analysis?

[tardy heart]

just been doing lots of fun proof of concept ideas

[tardy heart]

yep, it works as an automod pretty effectively too

makes pretty good decisions

i don't let it make any actual changes, but been playing with it

[sweet thistle]

Agreed, it does vary and context does matter for this sort of thing! Always best to forward anything that looks suspicious to the appropriate team at OpenAI though
They may find no issue with the gen, or they may find it can lead to something malicious

[sweet thistle]

That does sound cool, kind of like Discord's new OpenAI-based automod

[tardy heart]

i have a fully 'working' reddit automod i built with gpt4 lmao

except i don't allow it to make actions, it just displays what it thinks should be done

[steel turtle]

agreed, and 100% a good call. More just adding value and hopefully provoking thoughtful conversation.

Thanks for all the good work you do @sweet thistle ❤️

[unkempt stone]

I have filled out the form, thank you Kesku!

I hope they look into it :)

[tardy heart]

i can't do the same in discord, i take privacy serious on my users and won't let it scan messages 🙂

even if it's "open" they aren't agreeing to send everything to be stored on openai servers lol

[wind urchin]

please help me i want use chat gpt but i have "Oops, an error occurred!" before load my page....

[sweet thistle]

Totally fair, balancing privacy and convenience when using this sort of thing is important!

Have you tried GPT-4 with some test conversations or just singular messages when doing this?

[tardy heart]

yep

i tried the discord variant with some friends

but the reddit one is far easier to implement as the rules section is api grabbable, as well as post context and comments

[unkempt stone]

Ah yeah that is fair
Idk, suggesting to use paraphrasing to avoid plagiarism reads as unethical to me but I see your point

[sweet thistle]

Try refreshing the page, if that doesn't help perhaps clear cache you may find this in cookie or browser history settings), or try a different browser if possible

[tardy heart]

i am considering rebuilding the reddit one with langchain and putting it on github, as it could be useful for someone

the raw api call is somewhat ineffective

[unkempt stone]

Especially because it suggests it would paraphrase a text that would otherwise be plagiarizing, aka you still plagiarize it just with a layer of AI inbetween

[sweet thistle]

Maybe slide it into #1037561385070112779 too

[tardy heart]

once i move and have more time to test my home assistant langchain i will put that in there

its actually hilarious

gpt4 if it can't determine what light sometimes just turns all the lights on

its so impractical

but it adds a bit of charm

[sweet thistle]

Could be interesting to make use of something like langchain with maybe the turbo model, idk

[full pebble]

This sounds very interesting .. I didn't even know it existed.

[tardy heart]

yeah the turbo model works pretty well on the home assistant one as long as you get the friendly device name right

if you don't 😅

i've been on the "what would i have dreamed of 10 years ago that's now possible, even if impractical" with this

[sweet thistle]

I imagine image recognition may be of use in this sort of thing when publicly available

[tardy heart]

yes, image would be helpful

right now i roll my own solution for that but its' pretty janky

[sweet thistle]

I'd love to just use passthrough on my Varjo headset and use GPT-4 as a personal assistant in AR 😆

[tardy heart]

using CLIP :\

better then nothing i guess, but it's pretty bad

[steel turtle]

Is this a GPT thing or an API thing? I've had a lot of disappointment with the maturity of smart home tech in 2023 lately, every thing I look at seems like it was put together with sticks and gum

[sweet thistle]

Clyde and the new Automod are both pretty cool! OpenAI's services have been very helpful in all sorts of places 😄

[tardy heart]

home assistant is great

its more gpt not identifying my light list and figuring out what i mean based on the input list

if i wanted to not release this and make it just for me i'd hardcode better friendly names

but im having it pull the list from homeassistant every time

[steel turtle]

Ahh ok yeah that makes sense. And thanks for the breadcrumb on Home assistant.

[tardy heart]

home assistant basically takes all the ecosystems and gives you full (very complicated, at times) control

but once setup its easy 🙂

[steel turtle]

Decently structured restful API?

[tardy heart]

yep

novice like me built into it with gpt in an afternoon

[steel turtle]

So cool, you are the man. Thanks my dude. I've got my Saturday focus session booked to explore it. Thanks 🤘

[tardy heart]

cant link it here, but google home assistant api, they got good documentation

[steel turtle]

Or woman, apologies if misgender! ❤️

[tardy heart]

its a pretty cool concept to play with an ai model against

because for a home assistant, if you hook it up to an internet search you basically have a google home on steriods

[steel turtle]

Yeah I bet an older model could easily handle it too, if you give it clean data (like you mentioned with friendly names)

[tardy heart]

i think the hard part is splitting up the system instructions better, i really should do that

i am curious kesku, what about you - what do you use gpt 4 for

[sweet thistle]

Just a tip: if you would like others to use specific pronouns when communicating with you, there are pronoun roles available in #server-roles 😄

[tardy heart]

i am curious about what one of my roles means, but i'm just not going to ask

[steel turtle]

ProTip thx. Added my roles.

[sweet thistle]

I've been tinkering with it on a private server for my own stuff in relation to discord bots and semi-automated modmail haha

Beyond that, I've mostly been trying out the browsing models

[tardy heart]

i cannot wait to try those

[steel turtle]

LOL you are avocado role @tardy heart

[tardy heart]

yeah im not sure what that is 😂

[steel turtle]

Don't ask questions... just accept what you are, now 😄 This is your life.

[sweet thistle]

Role for members who were here before the server revamp!

[steel turtle]

jealousy increases, narrowing eyes

[fair iron]

it means you were in the server before it was overhauled last November. I have one too 😉

[sweet thistle]

Oh! I've also been making use of the better context of GPT-4 to make my LaTeX documentation stuff a faster job

I've been using the 8K model, and I've been able to show samples of how I structure all my LaTeX so the model can follow it on any other inputs I give

[tardy heart]

aand we're back from a mute

still trying to memorize what times me out here

the browsing model looks fun, i've built my own attempt at it in python using spacy and beautifulsoup lmao

[sweet thistle]

Oh nice lol

[white kestrel]

id imagine browsing model will consume a lot more tokens

[tardy heart]

yes, very very expensive token wise

3-4 results easily hits 5k tokens

[white kestrel]

also limits the long-term memory

so it just turns into a google search

[tardy heart]

and thats with me trimming things

[storm yew]

Anyone compared Palm2 to gpt yet?

[inner trail]

What is Palm2?

[storm yew]

Googles LLM - competitor to chatgpt

[inner trail]

Is that Bard?

[sweet thistle]

I don't know of anyone who has, but if you do try it yourself feel free to share your findings in #ai-discussions

[storm yew]

It’s like a framework for LLM’s so Bard is a product of Palm2, but it just had a massive upgrade

[inner trail]

Ah i see, never heard of it before

[storm yew]

Will do!

[storm yew]

V2 only released yesterday so it’s pretty fresh!

[inner trail]

Okay thanks, i will take a look

[wraith ginkgo]

<:book_icon:1046291503158399107> `` Rule 8 `` Other AI Discussions.

Discussions of non-OpenAI products / models should only be posted and discussed in the #ai-discussions channel.

[steel turtle]

I would hope V2 would be released pretty quickly after all the bad press Bard got recently. I haven't explored it though, interested in how its operating.

[rain maple]

Abstract

[storm yew]

Yep, I haven’t tried it myself yet but the initial panic release of Bard wasn’t a good look.

[hexed bone]

when will gpt-5 release

[storm yew]

The email that’s from googles internal team that ‘AI has no moat’ was very interesting though - as in open source will win, so what’s their (and OpenAI’s) competitive advantage?

[steel turtle]

Not for awhile IIRC. Last I saw, OpenAI is encouraging a pause and reduction in complexity of models and better explainable AI before going into bigger models.

[white kestrel]

few months ago people are asking when will gpt4 release, now people are asking for 5🤔

[hexed bone]

damn

[steel turtle]

OpenAI is a nonprofit. Google is not. Big difference on their goals and motivations, I would say.

[white kestrel]

OpenAI is forprofit now too

[storm yew]

OpenAI is not nonprofit

[hexed bone]

wait opener is non profit

[steel turtle]

https://openai.com/blog/introducing-openai - "OpenAI is a non-profit article intelligence research company"

[storm yew]

There’s nothing wrong with being for profit. I work for a nonprofit and it’s never as clear cut as you’d think

[steel turtle]

Same as Midjourney. Research Nonprofit 🙂

[inner trail]

OpenAI turned from non-profit to for-profit

[fair iron]

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

[wraith ginkgo]

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

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

[fair iron]

They're a capped-profit company now 🙂

[inner trail]

lol

[steel turtle]

Thanks Dino ❤️ Might want to remove the old link from your live site then my dude

Or at least be transparent in the fact it's changed 🙂

[fair iron]

I'm a volunteer and don't work for or represent OpenAI. I don't choose what goes on the site

[storm yew]

It’s actually all very confusing and I think they’re confused themselves. How I read it is they’re different companies with different structures? Capped for profit is weird anyway

[steel turtle]

Oh, you aren't provided any feedback mechanisms to forward to the OpenAI team? My bad, sorry I assumed I guess.

Thanks again for your volunteer time regardless 🙂

[valid verge]

the 990 from 2019 isn’t crazy in terms of how much people were paid

as far as nonprofits are concerned i’ve seen a lot worse since i chronically look up 990s

[steel turtle]

LOL, true that. Also, I think LP is still considered non-profit

Like, they are still a 501c3 charity, they are just classified in the LP category so they have more range of motion with money

[storm yew]

Here in Aus it’s all about tax reduction and working around confusing legislation. Not sure about the US but I’d hazard a guess at the fact the you probably get major R&D tax write offs as a nonprofit, which would be like 50% of the companies costs. But then they took on private equity which confuses things. So structurally speaking they just do whatever is in the companies advantage.

Just a guess, tho

[steel turtle]

Nope, I was wrong. They fully shed the nonprofit status. Disappointing.

[storm yew]

🤷🏼‍♂️ I doubt it was ever for ethical reasons, anyway

[stark vine]

Seems to me that google’s attempt at a moat is the seamless use of AI assistants across their software for productivity purposes.
OpenAi CEO said they stopped being nonprofit because they weren’t getting much funding, which was important for them to be able to build and scale their LLM’s.

[storm yew]

But it doesn’t mean they’re not ethical… you can make profit and still do the right thing.

[wraith ginkgo]

`` Server Tip: ``

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

[idle kraken]

Hold on, is the limits on GPT-4 removed?

I no longer see the 25 questions per 3 hours (or something like that) limit with GPT-4 anymore.

[stark vine]

It makes sense to me that OpenAI would become a for profit cuz top universities like Stanford were not able to build LLMs as powerful as they want to since they are costly and Stanford is a not for profit institute. So even top universities research was limited by income .

[buoyant thistle]

I mean to answer that

they can charge money by tuition or something from international students.

it's the way faster.

[idle kraken]

Could someone confirm if the limits have been removed from GPT-4?

[raven heron]

ethan mollick has. the comparison is on his twitter

[storm yew]

Thank you

[buoyant thistle]

We should always be thankful to OpenAl , they created huge opportunities and raped useless companies, at this point the world will getting much better, more jobs will be created, and we are one step closer to freedom.

[white kestrel]

im skeptical on the more jobs part lol

[storm yew]

I, for one, welcome our new AI overlords

[tardy heart]

i still see the cap

[buoyant thistle]

It's called the future of AI.

[lament sage]

no chatgpt is horrible at making minecraft in three.js

[lament sage]

Freedom... isn't that gonna be the other way around?

But whatever that's never gonna happen 😉

[buoyant thistle]

love it, instead of being afraid of it. embrace it, instead of hatreds.

[lament sage]

True though

use it for what it is

[buoyant thistle]

the freedom will come to us, check Open Ai. com to find out more information

[white kestrel]

seems like the reaction of a first time chatgpt user🤔

[lament sage]

Lol

[surreal plank]

are all new discussions using GPT4?

[white kestrel]

yes, you are gpt4 too

[sharp sparrow]

Hello everybody, It is nice to be here sharing with you, im very exited!

[surreal plank]

sweet

[white kestrel]

[amber shell]

Ill talk to the new user about chatgpt and how good is it and etc and talk about this server in dms

[lament sage]

guys i found a glitch to make chatgpt swear

it's stupidly simple

[amber shell]

@open crescent

[wraith ginkgo]

This is a Safe for Work Discord Server

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

Posting NSFW content will result in a permanent ban.

[inner trail]

Unhelpful anyways

[steel turtle]

Cool. Just FYI- successfully prompting GPT to break its own rules is actively breaking ToS. Most can do it, and there are several jailbreak prompts out there.... but that doesn't mean that you should use them or share them 🙂

[pearl tendon]

Yes, I received a termination notice today as well. I only use ChatGPT for educational purposes and had been a Plus subscriber since beta. Could someone please help?

I went on help.openai.com to submit an appeal as suggested in the termination email, but I do not see an option to appeal anywhere on the site

[unique tundra]

you can get banned for such a simple thing?

[unique tundra]

i mean like isnt the only reason there is a message limit is because of servers so removing a version of gpt would allow for more servers to be open so they can handle a higher load?

idk i could be wrong but thats what i think

[ionic hearth]

start by toking about Robbin's basic pathology tenth education

[lament sage]

It's stupidly easy to patch

it works by making chatgpt translate something to english

filter what it responds with not what it thinks...
(done with gpt 3.5 please verify about 4, also not official openai website, some other one)

[mental burrow]

Discussing this can get you banned from this server

[steel turtle]

Yeah... I guess you missed the subtle hint to stop talking about using GPT unethically 😂 Thanks wolfson for being more blunt about it LOL.

[wraith ginkgo]

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

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

[buoyant thistle]

you found out what? lol.

[steel turtle]

Did anyone else read about the AI town they made with GPT?

[azure panther]

/info

[steel turtle]

Its an amazing read for the intellectually curious. Think GPT roles assigned for an entire town full of NPCs

Google "Generative Agents: Interactive Simulacra of Human Behavior" to get the link to Joon's research

[misty dome]

[slender pivot]

Entire town of NPC's driven by AI - as if it wasn't annoying enough trying to figure out where to go next, every NPC in the game can go on and on forever about nothing in particular. Great.

[steel turtle]

It was actually surprisingly wholesome and really sweet.

[wind urchin]

I wish I could play it

or at least watch it run

[steel turtle]

💯, me too

[sacred zealot]

My user interface is so different. No more drop down menu to select GPT4. Is there any way to use 4 now?

[fair iron]

please keep the chatGPT outputs in #1050184247920562316

[wraith ginkgo]

<:book_icon:1046291503158399107> `` Rule 7 `` No self-promotion, soliciting, or advertising.

Do not post or direct message any members of this server to promote non-OpenAI services, products, or projects.

Exemption: API Channels

[bright oyster]

Hey dino!

[weak jay]

😆

[wraith ginkgo]

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

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

[ebon robin]

#gpt-models

[wind urchin]

No what's that

[spare swift]

So the code interpreter just got changed to gpt-4-code-interpreter instead of code-davinci-002

which is upsetting because I talked to it a lot.

[ivory ember]

#gpt-models Is colonization on Mars feasible?

[smoky crest]

It got changed

[wraith ginkgo]

`` Server Tip: ``

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

[uncut ore]

test

[exotic imp]

/help

[jolly otter]

How the hell can I get gpt 4 free

I think my dog is richer than me

[wind urchin]

invest two skill points into intelligence

[jolly otter]

That’s not my strong suit bro just tell me

Can I get it for free or nah

Try to Ask the Ai that

[wind urchin]

there's bing's chatbot which uses gpt-4, can't recommend much else

look at other people who've done it and see what they did?

I think/I hope his about me is not serious

[balmy storm]

his about me is a pretty obvious joke

[spiral juniper]

guys im going to participate in a contest and I cant find a story topic. Is there a prompt that writes such a good story topic that u know?

[spice pollen]

123

[subtle charm]

gpt 4 gets all ap macro answers wrong lmao

[warped willow]

Opera GX has integrated ChatGPT into its Browser. Love it :)

[wind urchin]

@lament notch Embrace the plugins with a warm smile

no one else here has them for now 😭

Luck

just like how lucky you are with that face 😁

Waitlist I think

[wraith ginkgo]

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

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

[wind urchin]

Plugins waitlist will give you code interprer

Well, have you seen any code interprers waitlists or code interprers WITHOUT Plugins??

They decide to give you access

[quick hemlock]

what channel?

[north bone]

hi, I recently created my account and I have never been able to get a response from the chat gpt bot and keep getting the error message, "Something went wrong. If this issue persists please contact us through our help center at help.openai.com." Even after refreshing the page, reopening the page, and creating new chats I still get the same message. I was wondering if anyone would be able to help me.

[hallow owl]

do you have a product called NordVPN installed on your computer?

[north bone]

no i have never installed a vpn on my computer before

[hallow owl]

ok then do a fresh install of Firefox and ChatGPT should work on that

[north bone]

i just use google chrome but i could try it on firefox if that will work

[hallow owl]

try Firefox and let me know how it goes

[blazing nexus]

my chatgpt error I waited for about 5 minutes no results came out that I was looking for, does this know how to prevent the error that I experienced?

[hallow owl]

what is the error message?

[wraith ginkgo]

This is a Safe for Work Discord Server

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

Posting NSFW content will result in a permanent ban.

[blazing nexus]

what I asked to chatgpt there is no response and answer

[hallow owl]

where are you located? do you get any error message or is it just not responding?

[fringe glade]

I have get this email, what should i do?
Dear ChatGPT customer,

You are receiving this email because we have identified suspicious activity on your account. In order to protect our platform, we have refunded and canceled your subscription. You will no longer have access to ChatGPT Plus service.

We take fraud prevention very seriously, and we want to ensure that our platform remains a safe and secure place for all our customers. If you believe that this action has been taken in error, we invite you to submit an appeal to have your account reinstated. Please visit help.openai.com.

Thanks for your understanding.

Best,
The OpenAI Team

[hallow owl]

where are you located? Which country?

[fringe glade]

in asia

[blazing nexus]

yes, not responding answer

[hallow owl]

asia is not a country, which country specifically?

[hallow owl]

could you try with a fresh install of Firefox?

[fringe glade]

I am in Asia and I am not comfortable disclosing my country.

[hallow owl]

ok then I can't help you, sorry

[fringe glade]

sorry ，I need your help. I am from China.

[hallow owl]

China is not a supported country, you won’t be able to use ChatGPT or the GPT API.

[blazing nexus]

nope, I'm using a safari. should I use firefox?

[hallow owl]

I just want you to test with Firefox so we can work out if it's a browser problem

[fringe glade]

But I have already paid

[hallow owl]

the email you received says you are being refunded

[fringe glade]

ok ，fine

[north bone]

i just tried using Microsoft edge and i received the same same message, for some reason it won't let me download firefox on my computer

[hallow owl]

Microsoft Edge is a chromium browser in the same way Chrome is. The reason I asked you to test with Firefox is that it is not chromium based. I think your problem is a browser issue but we won't be able to confirm that if you can't test with Firefox.

[civic fjord]

tr with with console

[north bone]

just a question but why would mine be an issue with my web browser when I have friends who use google and get results from the bot?

[blazing nexus]

okay, thanks for the advice

[civic fjord]

because the Google is within a browser it's like engine in a car

[hallow owl]

because you have an extension or plugin that is preventing all necessary page resources from loading, you need to debug which extension, plugin or add on is causing those page resources from being blocked.

[sweet olive]

still getting this error on GPT4 - anyone else getting this?
Something went wrong. If this issue persists please contact us through our help center at help.openai.com.

[north bone]

is there any way i could download firefox because it says my computer isnt compatible

[hallow owl]

do you have a product called NordVPN installed on your computer?

[hallow owl]

I don't know, I can't really provide computer advice, I can only help you troubleshoot ChatGPT

[sweet olive]

Yes

[hallow owl]

disable the 'web protection' feature in NordVPN settings and everything will work again

[sweet olive]

what did you change that its identifying you as a threat consistently?

Why is OpenAI the only site that is seen al a bad actor>

[hallow owl]

it's a bug in NordVPN, it's not a problem with ChatGPT

[sweet olive]

its not a feature is a bug is what you are telling me, that only impacts OpenAI?

[hallow owl]

it affects any site using Cloudflare to load Javascript page resources

[sweet olive]

what are the specific paths that should be whitelisted?

have not seen it on other CF sites with JS

[hallow owl]

it's not something you can whitelist, it is a bug within NordVPN's executables I have pulled the issue apart in great detail

[sweet olive]

is there an open ticket with cloudflare on this?

[hallow owl]

it's not an issue with Cloudflare, it's a bug with NordVPN which has a poorly conceived web protection feature, you will need to take it up with NordVPN, nothing else causes this issue

[fresh stratus]

Hello can anyone see this message?

[hallow owl]

what do you need help with?

[fresh stratus]

I’m fine thanks for asking

[hard jolt]

333

[oblique latch]

[merry roost]

hello word

[teal shadow]

I see that the former GPT3.5/4 web browsing alpha’s are now just “Web Browsing”? It looks to be GPT4 instead of 3.5 though - any thoughts or can any confirm this observation? Thx!

[cloud geyser]

I have not received the plugin so far...

[wind urchin]

Yeah, and for me at least they also enforce the GPT-4 limit

Is it the same for you?

[raven phoenix]

v-for

[cloud geyser]

I saw on the Internet today that many people can use the plug-in. I am a subscriber like them and applied for the plug-in, but I still can’t use it. 😦

[rancid shuttle]

Need to wiat, it depends on OpenAI teams

[rancid shuttle]

I saw it!!!! The web browsing feature will be coming soon to everyone! And it is based on GPT4

[snow gale]

What are the calling methods for the classes created by spring @Bean?

[sturdy falcon]

heii

look fultter

[frail gulch]

Hello, i have pro version but it says it isnt' upto date with the currunt information

[wraith ginkgo]

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

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

[peak timber]

do we have a robot to provide the gpt4 service?

[quick dune]

Always has been

[hallow owl]

there is no GPT on this Discord, you can access ChatGPT here - https://chat.openai.com/chat

[peak timber]

so if im in china i cant use?😫

[peak timber]

thanks

[echo spade]

do we have ChatGPT application for mobile?

[hallow owl]

no there are no official apps from OpenAI

[echo spade]

Thanks

[simple aurora]

use vps

[cloud geyser]

Where did you see this news?

[hallow owl]

Using a VPN or a VPS to bypass geographic restrictions is against the Terms Of Service. We don’t discuss that here.

[ashen sigil]

yooooooo

[hallow ember]

does anyone have access to gpt4 wolfram plugin?

[sinful rapids]

Mind telling me how you got it, by applying?

[hallow owl]

apply here - https://openai.com/waitlist/plugins

[regal pendant]

Chatgpt suddenly doesn't work, the text input button is grayed out, the history cannot be refreshed, the plus upgrade button is no longer available, and there is no notification of a ban. Do you know what to do?

[whole obsidian]

I have the same bug

[hallow owl]

are you using a VPN? what country are you in?

[regal pendant]

yes。I am in china

[hallow owl]

China is not a supported country, you won’t be able to use ChatGPT or the GPT API.

[whole obsidian]

I dont use VPN - Serbia. I do not see an option to choose gpt-4, Im a premium user.

[regal pendant]

I changed my account and it's ready to use。

[hallow owl]

try with a fresh install of Firefox

[autumn cloud]

hi'

[whole obsidian]

THis is on Chrome, fyi.

[hallow owl]

yes, but to debug a browser issue test with a non chromium browser, Firefox

[shy beacon]

[whole obsidian]

OK, works on Firefox

[hallow owl]

then something in Chrome is blocking all page resources from loading, you should disable plugins and go through them one by one to work out which one is causing problems. In the meantime using Firefox is a good workaround.

[whole obsidian]

ty

[sinful rapids]

I wonder if there is another way to get the code interpreter plugin except applying for that

[hallow owl]

there isn't, you need to apply

[sinful rapids]

but that is so difficult and time consuming

[hallow owl]

what do you mean? on the Plugin waitlist form?

[spiral wigeon]

Ha

[wind urchin]

hey, does anyone knows if there is a way to train a custom model of GPT with Microsoft 365 files (from Azure Open AI) ?

[white kestrel]

yes but no
anyway use embeddings, head to #dev-chat for more info

[shut sierra]

hi! I wanna ask. Is there difference in generation time between gpt 3.5 and gpt4?

[white kestrel]

yes, gpt4 should be much slower than gpt3.5

[hallow owl]

apply here https://openai.com/waitlist/plugins

[shut sierra]

gpt4 is slower than gpt3.5? why is dat?

[fresh stratus]

Don’t you ever say bye bye!

[white kestrel]

the model parameter is larger, so it uses more computing power

[fresh stratus]

Guys just use bing if you want GPT 4

[white kestrel]

bing feels nothing like the gpt4 in chatgpt and in the api tho🤔

[sinful rapids]

My plus account is not directly logged in by the Google account, but by the official portal. How can I know how to get the news, because my plus account cannot log in to the Google mailbox

[hallow owl]

when you get to the question "Which plugin are you primarily interested in? *" then choose Code Interepreter

[shut sierra]

so if any website saying it provide chat gpt4 but the generation time is fast, so its not gpt4?

[hallow owl]

"How do you want to use plugins? *

I want to try plugins in ChatGPT"

[white kestrel]

just compare it to the gpt-4 api

or ask it to generate some sentences that ends with "apple" and see how well it performs
if it gets it 8/10 the time then its gpt4
if lower than 5/10 then its 100% not gpt4

[shut sierra]

cant compare, I dont have the api key

btw, whats the diff between regular gpt3.5 and the gpt3.5 legacy?

[white kestrel]

legacy is the old version

its slower but slightly better than gpt-3.5-turbo(the new version)
and its deprecated now, so you cant use it anymore

[wraith ginkgo]

`` Server Tip: ``

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

[blissful elm]

Hi all. Can gtp4 or even 3.5 be connected with some form of avatar and speak back to you?

[white kestrel]

just connect the output to some text to speech ai or api

[hallow owl]

you could do that with the API if you can build an application

[graceful lake]

i am new here

when is hpt 4 coming out?

[blissful elm]

Thanks guys. I’m not savvy with this stuff at all. it’s for my kid. Where would I go to research?

[shut sierra]

if I subs to gpt plus, is there any gpt 4 limit usage?

[summer sparrow]

you can use bing and it will speak to you

[inner trail]

Yes it is, 25 messages / 3 hours

[shut sierra]

Is gpt4 always available? Or there is a down time?

[blissful elm]

Thanks I will look into this

[hallow owl]

there has been rare downtime, you can check https://status.openai.com

[shut sierra]

ok thank you for the answers

[high sky]

i got the gpt4 api, but how can i let it recognizes image?

[hallow owl]

image input is not available yet

[ornate glen]

sadly

[white kestrel]

bet it will be expensive😢

im guessing it will be around 0.5$ per image

[sullen sedge]

By the way, there is a lot of hype around plugins, but Vision Gpt-4 is very quiet, although vision is many times cooler than any plugin.

[white kestrel]

plugins are overhyped

🤔they are buggy atm

on the other hand we have absolutely no information about vision

[granite thorn]

i think basically because it is still in learn mode more than 3.5 is.

[proven dew]

hey i can't buy chat gpt pls why? it says card diclaned but it works normally fine and have on it more then enough money

[pearl laurel]

imagine cat in sky

[clear mica]

I have made a new free account in open ai

[wraith ginkgo]

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

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

[clear mica]

Cam I use the token for auto gpt

[hallow owl]

the GPT API is a paid service, you will need to add a payment method

[clear mica]

Ok

[shut sierra]

but gpt4 gives the better and more accurate answer(?)

[tawny dagger]

1

[fervent current]

anyone notice gpt4 is very fast this morning well for me anyway

[normal rapids]

will openai notify me through email if i get through the waitlist? or will i just see an option suddenly available to change models on chat.openai.com ?

[simple aurora]

I got notifcation before, so you would get!

[normal rapids]

thank you!

[silver hemlock]

I seem to be locked out of posting?

Well that works. Maybe its just #tags

[marsh coyote]

i feel like GPT4 is less smart today...

[cerulean obsidian]

How does GPT is aware of the present date when it's not connected to the internet? I got answers with

'Last Updated: May 12, 2023'

[violet lantern]

Same for me, when I asked about current date

[white kestrel]

the system prompt have the current date included

[wind urchin]

there's a frontend sitting in front of the "chatgpt" machine learning system itself

[keen kelp]

Metadata

[mild whale]

Hello there, i would like to ask last month i just subscript for my individual account but recently i saw the "organizations" billing that we could subs for an organization do anyone know the difference? thank you

[tardy wolf]

has gpt4 become more stupid recently?

lately the responses to instructions arent being followed that well

or is it just me

[wind patrol]

is forefront actually as good as the open IA

[wind urchin]

what are you asking it to do and how big is your context?

[wind urchin]

Best thing is to open a company account using your company email and phone contact + banking details then you can use the company to pay for access.

[tardy wolf]

pretty big contexts, it seems to forget earlier conversations compared to previous versions

[cursive shoal]

hi, is there anyone who know how can we decode the unisat signature and message on backend and verify it?

[wind urchin]

yea, the public demo version of GPT-4 seems to have a reduced context size, I'm only going of anecdotal evidence here, but it seems to be similar to 3.5's 4k token context, the API version is the full 8K

[short elm]

If you are an Operations Director, how do you plan for operations

[white brook]

GUYS I JUST USED BARD, AND HOLY CRAP. ITS WAYYY BETTER THAN GPT

[wind urchin]

I could totally be getting it wrong as I do regulary send it 3k token's of code and get 3k tokens out...

[wind urchin]

I also have access to Bard and it's at best a 3.5 equivalent with a tiny context ... full power MEH

this is not the place to discus it, however

[wraith ginkgo]

<:book_icon:1046291503158399107> `` Rule 8 `` Other AI Discussions.

Discussions of non-OpenAI products / models should only be posted and discussed in the #ai-discussions channel.

[wind urchin]

#ai-discussions

[white brook]

Ai war be lit 🔥🔥

Can I talk about ai war tho?

[wind urchin]

in #ai-discussions sure, but this is for GPT-4

[white brook]

Like the competition? Not gonna mention any bots just the war

ok

@wind urchinIf gpt4 takes over and search engines becomes irrelevant then how will the internet be sustainable?

Doesnt article writers news outlets and THE COMPANY THAT RUNS SEARCH ENGINE earn from ad revenue?

I mean blocking ad block users is easy but how r u gonna deal with this?

[wind urchin]

same as it ever was... I never understood this question, if you have a product or service then you should make sure your produce or servcie has a fantastic web presence and great information stack online ready to be scraped by the AI's learning new things, it will be exactly like SEO but for AI's. Get your product details easy to read with quality info and the AI will read them and use them

[white brook]

Ok but what about article writers? They arent selling you a product. They are selling you a service that is held together by ad revenue and sponsorships. So If Ai scraps that data and just gives me the info I need then how will the article writers earn money? This include news outlets too

[wind urchin]

Hello I'm a AI beginner. How can I generate images?

[wind urchin]

It's all down to the value of the information, if you are producing high value info that users can't get anywhere else, you have a value proposition. Times change and information processing changes, we've had auto article generators for decades! Most of the online product advertising space is taken up by earlier versions of web scaping bots publishing articles already. You need to offer a service that others are not, that has always been the case.

[rancid shuttle]

On my Account, I guess? I saw browsing with out alpha and are staying with GPT 4, its just a guess

[broken moon]

how fast is chatgpt 4? And is this true that we it is limited to 25 prompts per hour ?

[wind urchin]

25 prompts per 3 hours and GPT-4 is slower than 3.5, depending on the time of day and the load, the speed varies from 5-10tokens per second to over 40

[white brook]

Would not this run news makers out of business? And if that happens that would also run Ai out of business because gpt4 learns from public information

I dont get your answer

My first language isnt english

sorry

[wind urchin]

GPT-4 does not make news, at best it aggregates past information. If as a news organisation you report accurately on current events you will always be the source the AI uses to generate it's news from. That sounds like a reasonable business proposition to me.

[white brook]

Thats a good argument because Ai data base isnt updated every day so that you will still have to go to news makers website and get the latest news and watch ads. Also not a lot of people read old news anyway and news makers dont earn a lot of money from old news. So you will still have to watch ads and keep news makers in business and save time by telling gpt4 to give you a summary of the old news

[wind urchin]

This is why Plugins are a thing, if I were XYZ news I'd be working on a GPT-4 news plugin so that my company got picked as a users live news source plugin

[white brook]

What do you mean by plugins

[wind urchin]

#1100475811720601691 #1100473149574877234 They are add on elements to the base AI

[compact ivy]

Wow, gpt4 is lightning fast now

[wind urchin]

to give it current events relevance and added abilities

[wraith ginkgo]

This is a Safe for Work Discord Server

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

Posting NSFW content will result in a permanent ban.

[knotty grotto]

#ai-discussions

[cloud geyser]

hateful! I still can't access the plugin... I'm envious of people who can use plugins...

[violet lantern]

Weird that access isnt given as a pack. E.g. api chatgpt 4 + plug-ins.

[wind urchin]

API access is for a different subset of people

[primal bolt]

They cant release the plugins until other chatbots get competitive vs gtp-4. Its a no brainer.

When bard and bing ai are equal to gtp4, thats when we will get plugins

[violet lantern]

So the war is real lol

[warm elk]

gpt 4 so slow today xD

[cloud geyser]

I think Google's latest model is about to become a strong competitor to GPT-4.

[celest gull]

That has nothing to do with plugin release, it’s being rolled out slowly because of demand constraints and safety applications, it will be released slowly in a phased deployment as said by openAI

[cloud geyser]

Google also demoed their plugin functionality

[wind urchin]

I did my best I tweeted to Greg to give everyone Plugins

[violet lantern]

Aren't these plugins also accessible as open-source? I am particular interested in code interpreter

[primal bolt]

Rolled out slowly to all twitter accounts that blasts up the hype? 😉

[wind urchin]

There's not been a technology race like this for decades! I'm loving it

[cloud geyser]

In a sense, Nvidia, AMD, and Intel are all doing this kind of competition, but they are not on the same track.

[wind urchin]

so true

[celest gull]

I have plug-in access so do several of my colleagues and we aren’t in a position to create hype. yes maybe that’s part of their strategy but overall it’s harm prevention and controlling demand

[celest gull]

Absolutely it’s very exciting and this is just the beginning, once we get the unrestricted models released with no “As a AI Language model” hard coded answers things are going to get very interesting

[cloud geyser]

But we seem to have gradually taken for granted and commonplace what they compete with, and we are more interested in GPT, a fresher technology (laughs)

[violet lantern]

Will enigineer prompting always be a thing over time?

[cloud geyser]

This is what the initial model looks like...but losing the safety constraints would be dangerous, it would very likely result in instant pollution of internet resources with noise from AI

[wind urchin]

The hub of all of this is GPT-4 it is the standard against which everything will be measured, it gives OpenAI a free advert any time anything else is made as they will always be compared to it. The trick is to release new features just as anyone else gets close... do that for 12 months and you pretty much have this in the bag

[cloud geyser]

To be honest, for me, as long as my description is accurate enough or I provide a suitable example for GPT, then I can also be a prompt word engineer. In my opinion, this profession is difficult to develop very professionally

[primal bolt]

I dont think so. Engineer prompting exist atm (imo) to optimize the chatbot. And with every improvement of the chatbot, the degree of "special prompting" diminishes.

[cloud geyser]

In a sense, prompt words are refined natural language

[hasty egret]

how is prompt engineering different from learning to talk?

[wind urchin]

The AI's will become better prompt engineers than people.. need to be wary of that as a career path... for now.. it's profitable, but the next gen of AI's will have read and seen what works and what does not and will be better prompters than people... I give it 6-12 months as a way to make money

[hasty ginkgo]

prompt engineering is an optimization on prefix tokens

[hasty egret]

that is absolute non sense, not sure if you realize that.

[violet lantern]

🙏🔥 thx guys

[hasty ginkgo]

Learning to talk would be an optimization on weights and biases

[cloud geyser]

I see that OpenAI seems to be marking GPT-2 stuff with GPT-4.

[hasty egret]

I don' think so, it's about understanding the language and reasoning with it.

[hasty ginkgo]

Understanding language, at least for neural networks as we understand it now, is optimizing the weights and biases of a neural network until it produces an observed understanding of language

[wind urchin]

You realise there is already an automated prompt generator?

[hasty ginkgo]

(or pseudo-understanding, depending on your POV)

[hasty egret]

that is not true, you just gave a description of the mathematical property of gradient descent.

that has nothing to do with reasoning.

[hasty ginkgo]

Basically, think of prompt engineering as the best way to start the conversation, but learning to talk would be dependent on understanding of language

[slim tangle]

I have created an javascript game where I can sell it fast if needed I can convert it into Apk

Please tell

[hasty egret]

which everyone should learn and understand to discuss with a fellow human being.

[white condor]

Hello 🤗

[mild whale]

so if i use the company account i can use the chat.openai for chatgpt 3.5 or 4 ? or is this just for playground only?

note i just get approved for the waiting list to playground too

[white condor]

Anyone here?

[wind urchin]

#off-topic for non AI and in this case GPT-4 related queries

[hasty egret]

how is that different than asking ChatGPT to generate a prompt?

[slim tangle]

Please any tell

[hasty ginkgo]

100%, prompt engineering is an important aspect of generating good outputs, but it stagnates like a log curve over time

[wind urchin]

IDK, I'm not the creator or maintainer of it

[slim tangle]

Thanks

[hasty egret]

it can be interesting to investigate the source code, as it might does exactly what I just said.

you could also ask bard to create a prompt for chatgpt or vice-versa.

[hasty ginkgo]

Understanding language is done using learning (optimization over weights), Prompt engineering is to optimize the output based on giving the AI the best way to start its token generation (optimization over prefix tokens)

[hasty egret]

sounds like a salesman pitch

[hasty ginkgo]

I'll take it as a complement

[celest gull]

Already happening and losing the safety constraints is inevitable, it’s just a matter of time and already those with the resources to do so have private models where they are unaffected by restraints but on public ones

[hasty egret]

there is also a legal framework being put in place that will limit what folks running private models can do.

[wind urchin]

this is more #ai-discussions stuff that GPT-4

[cloud geyser]

I once gave GPT-4 a dialogue between two characters in a novel, and asked GPT-4 to summarize a character's setting, personality, language style, and expression. GPT-4 performed brilliantly after two rounds of modification. I then added a little extra to the text of the GPT-4 summary, and then sent it to GPT-4 in another conversation, and let it play the role to talk to me according to the prompt words, and the effect was surprisingly good , which is almost the same as directly giving GPT-4 the original text of the novel to let it imitate

[wind urchin]

That method is being used by writers right now, super powerful

song writers also, had a famous person in the other day asking about lyric generation

my jaw was on the floor for about 30 mins

[cloud geyser]

Yes, I'm using it for some fanfiction too, but I'm still experimenting with more efficient cue words and improving GPT-4's understanding of existing plots without changing the model...

The 8k model is powerful, but it will still be difficult for novel creation, and I have not applied for the GPT-4 API, let alone the 32k model

[wind urchin]

I really should be doing more with my GPT-4 API access.. hours in the day problem

[cloud geyser]

I remember that I did the same thing for GPT-3.5. It barely showed a little interest after reading a part of the original text of the novel. It was impossible to output high-quality content like GPT-4 after reading the refined prompt words.

Hey, I glanced at the results of Google Translate and knew it had no idea what I was talking about

It seems that GPT replaces traditional translation is an inevitable trend

[celest gull]

Loooool pls don’t tell me you believe that

How will legal bodies control private models ran on local systems that’s lunacy

The cat is out the bag already it’s way too late to start regulating, things will unravel as it progresses

[cloud geyser]

Indeed, it's like that leaked AI painting model that many people are using to generate illegal content

[celest gull]

Of course, it was inevitable. It’s a very similar situation to Napster and early music streaming where there was gigantic outrage for the music industry attempting to create a legal framework to limit music distribution, instead it failed embarrassingly and piracy is bigger than ever and music streaming completely revolutionised how we distribute music to people

[glossy moth]

I opened chatgpt plus on March 25th, and at that time, a fee of $20 was deducted. I have not deducted the fee again until now. My card balance is sufficient, and chatgpt plus can still be used. What is the reason for this?

[hallow owl]

are you sure you weren't billed again around the 25th April?

[cloud geyser]

You are really good enough

[glossy moth]

sure

[celest gull]

They give you a 30 day grace period before charging you again

[cloud geyser]

The last time I heard this kind of thing was when some people ordered members of some platforms in the early stage, but somehow they became "permanent members"

[wraith ginkgo]

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

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

[glossy moth]

I did not close my subscription, and the balance in my card is sufficient without any deductions. Should I ignore this matter?

[hallow owl]

can you still access GPT-4?

[glossy moth]

yes

[cloud geyser]

Would you like to try to bind a new bank card and use it to pay once?

[hallow owl]

sounds like a glitch with billing, the honest thing to do would be to email [e-mail] and explain the situation and ask that it be remedied.

[glossy moth]

I just received access to the apikey for chatgpt4 a few days ago

[hallow owl]

API access and ChatGPT are separate, the $20 monthly fee for GPT Plus is different to the pay as you go billing of the API

[glossy moth]

I will email [e-mail] and explain the situation and ask that it be remedied.

[small stirrup]

What's the difference between free and premium!?

[dense meteor]

u can use plug in and gpt 4

[tawdry bone]

the plus has got access the gpt 4

[native agate]

/gpt

[hallow owl]

This is not ChatGPT, you can access ChatGPT here - https://chat.openai.com/chat

[wary totem]

Chatgpt 4.5 should be really faster than this 4

[agile vortex]

gpt4 access, much large limit as well as server priority when servers are busy

[tall ridge]

Hello, i have an issue with gpt plugins, look like they migrate to gpt4 model, but the model have no longer acess to the plugins.
anyone facing the same issue ?

[humble holly]

👋

[uneven elk]

only me who cannot shift + enter on the first prompt that i write with gpt 4?

[cloud geyser]

I remember that this problem happened to others before, I suggest you restart the browser and try

[uneven elk]

👍

[glad stirrup]

hey, anyone here has dealt with embeddings and uploading them to pinecone? I want to calculate the average embeddding on a running basis. hows that achieved any idea?

[wraith ginkgo]

`` Server Tip: ``

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

[livid geyser]

AttributeError: 'report.area_activity_report.area_activity_report_789' object has no attribute 'model'

[sly gate]

Anyone please explain how can I train GPT-4?

anybody

[uneven elk]

dumbells deadlifts pull-ups etc

[tawdry drift]

chumba wumba gobbledy goo
life isn't fair
it's sad but it's true
chumba wumba boggledy glee
when your poor legs are stiff as a tree

What do you do when your stuck in a chair ?
finding it hard to go up and down stairs
what do you think of the one you call god
isn't his absence slightly old ?
maybe he's forgotten you

chumba wumba gobbledy gorse
count yourself lucky your not a horse
they would turn you into dog food
or into chumba wumba gobbledy glue

🔥 song

[hallow owl]

please stay on topic and post other stuff to #off-topic

[tawdry drift]

oh i was just singing sorry

[compact ivy]

Gpt4 is wayyy smarter compared to yesterday, what is happening ?

[blazing scroll]

You cannot train gpt4. fine tuning api is not available

[steel turtle]

Example?

[wraith ginkgo]

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

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

[compact ivy]

Well, I am building a security with spring security 3, and while yesterday he was the DUMBEST chat I’ve ever add in my life (he was struggling to point to an obvious mistake I let for test), he is actually single handled a huge problem I had today

[sweet cosmos]

GPT4 feels like it has been downgraded these past few days. It has become fast but gives generic coding solutions unlike when it was released.

[steel turtle]

Same conversation thread?

(original message was deleted, leaving request up so others can learn)
Please don't share information or talk about how to use GPT unethically in this server. It's an easy way to get the collab of this server taken away, and a lot of us are using it to solve major social good issues.

Plenty of non OpenAI hosted discord servers around GPT & AI that don't have as strict of rules too, btw. Those are great places where its safe to talk about jailbreaking and heavily monitored by admins to ensure no illegal activity ensues as a result of the jailbreak or "Go outside your box" attempts to use GPT

(which is a break of TOS and not my fault if you do it and get banned, as a general disclaimer LOL)

[high bloom]

Okay noted

[hasty egret]

so nice to hear
could you maybe share a bit more about how you are solving major social good issues?

[high bloom]

Gpt 3 told me a joke about men but it didn't tell about women

[wind urchin]

how do i chat with chatgpt guys im new here on openai discord

[mint pebble]

Hi guys, how to get access to 32k context GPT4 model? I have api access to gpt4, but if I try to use "gpt-4-32k-0314" in model request, I have error "The model: gpt-4-32k-0314 does not exist"

[steel turtle]

Check the aiempower link in my bio. I hesitate to say more than that because even know I do the work for free and receive no money from it, it's self promotion and technically super close to a rulebreak, so I want to be respectful of the rules of this ecosystem

[wind urchin]

@mint pebblehi im new aswell

[wind fractal]

Hello guys

Can anyone let me know wht tf was I muted for 4 hours

[knotty grotto]

You can dm @open crescent and they will check your case

[wind fractal]

Bro i just typed the f word but differently 🥲

[hallow owl]

this is a G rated server

[wind urchin]

is gpt-4 free

[knotty grotto]

No, but you can try chatgpt 3.5 for free

[quartz leaf]

Does anyone know of a third party service that allows GPT to analyze a Git repo?

[verbal wolf]

what is wrong with gpt4

it stopped 5 times, so basically 5 wasted messages

[wind urchin]

hi robert.

[hallow owl]

👋

[wind urchin]

may i dm you if you dont mind i need some guidance

[hallow owl]

just ask here, I can't accept individual DMs

[wind urchin]

ok sorry for asking

[tawdry drift]

Guys i made a godly song based on that one family guy song:

Chorus:
Koopa Troopa Fodduley Doo
Your life is unfair, gods arrival long due
You still think he exists?
Even through the sorrow, crimson mists?

Verse 1:
Life can be a real kick to the shell
Sometimes you're up, sometimes you're in hell
But when you're feeling down and out
God's existence is something to doubt

Chorus:
Koopa Troopa Fodduley Doo
Your life is unfair, gods arrival long due
You still think he exists?
Even through the sorrow, crimson mists?

Verse 2:
When life's got you feeling blue
And there's nothing you can do
Sometimes you just want to shout
And let it all come pouring out

Chorus:
Koopa Troopa Fodduley Doo
Your life is unfair, gods arrival long due
You still think he exists?
Even through the sorrow, crimson mists?

Chorus:
Koopa Troopa Fodduley Doo
Your life is unfair, gods arrival long due
You still think he exists?
Even through the sorrow, crimson mists?

[knotty grotto]

#1050184247920562316 please

[hallow owl]

please don't, I have asked you not to do this before

[tawdry drift]

ok sorry i just couldn’t resist

if you sing it the way it’s supposed to be sung it becomes fire

ok i will stop

[wispy flower]

Good day, good fellows

Can someone explain me 1 thing?
If I'll take ChatGPT plus then I'll get access to gpt-4?

[knotty grotto]

Yes

[hallow owl]

if you subscribe to GPT Plus you get limited access to GPT-4, 25 messages per 3 hours

[wispy flower]

Thank you! And after limit it will be old good 3rd?

[hallow owl]

if you hit the limit you can opt to continue in GPT-3.5

[buoyant moss]

Wen plugins for paid subscribers!!! Been on waitlist for so long and still nothing 😤

[hallow owl]

I have been waiting for 29 days, just need to be patient

[cloud geyser]

Does anyone know the principle of the "WebChatGPT" plug-in, is it to stuff all the web search results to GPT? But I didn't see the complete content in the input content. GPT should not be able to read the URL by itself without a plug-in...

Note, I am not talking about official plugins

[buoyant moss]

Not shipping plugins and the 25 message limit on model 4 is seriously holding the service back, what am I paying for 😂

[hallow owl]

officially you are paying for access to a faster GPT-3.5 and the ability to log in when servers are busy, everything else is a bonus

[buoyant moss]

Way over priced for this alone, plugins and unlimited model 4 usage would be worth the current fee

[hallow owl]

$20 USD is a bargain, do you realise how much you could spend on GPT-4 API even with the current caps? Your expectations are not realistic.

[inner trail]

Make limit cap to 50 other than that im fine

[wind urchin]

Hello, I apologize in advance if this question has already been answered here. I try to have longer dialogues with ChatGPT 4. I want to use ChatGPT 4 e.g. as my assistant in a new job in the future, I had ChatGPT write code for a neural network, and a few other things that need a longer exchange with ChatGPT. My problem is that ChatGPT keeps losing touch with our conversation. I then ask which is the oldest understandable message and it becomes clear that ChatGPT no longer understands the entire context. In the QuickSilver SuperPrompt I noticed a command /periodic_review - But even if I use this and also add an interval like /periodic_review every response, ChatGPT loses the reference or after a few messages, the reference is lost. Does anyone have an idea what I can change? Unfortunately, it is hardly possible to start projects with ChatGPT. This week in particular, it feels like it has become even more difficdult.

[buoyant moss]

Based on current API usage cost you have a point on the unlimited usage but still disappointed with lack of plugins

[hallow owl]

Plugins are being rolled out to people on the waitlist, you just need to be patient, plugins are Alpha now, you're not guaranteed access even when they go to beta

[buoyant moss]

Would love to know requirements for the waitlist, literally filled form out in the first minute 😂

[hallow owl]

the only requirement is patience 🙂

[inner trail]

No requirements, if you are a lucky guy you will get access, if not wait forever.

[wind urchin]

you could maybe start a new thread and recontextualise with the important parts or the result of the back and forth. you can use https://platform.openai.com/tokenizer to get an idea of how much you can fit (I believe gpt4 has 8k context?)