gpt-models

[umbral tinsel]

ah u are right even more

yap u are right

[wind urchin]

But... 50 pages of text being AI manipulated for a starbux... is insane

[umbral tinsel]

https://openai.com/research/gpt-4

[fair iron]

some volunteers claim that I have access to a better Dall-E 2 model because they like my images 😅

[wide bolt]

;);)

[wind urchin]

looks around wondering who THAT could be

[umbral tinsel]

if you consider the price then using gpt4 is not worth for like 10-30% improvement

[wind urchin]

depends on your use case, if is complex code.. it's VERY worth it, if it's linguistics and word manipulation and story creation ... I'm with you 100%

Is anyone else's playground api super slow ? Even in 3.5 ?

I'll level with you, I don't have a creative bone in my body, I'm going purely off hearsay

[umbral tinsel]

hmm

[wind urchin]

I Have found claude to be better in story creation than gpt 4 imo

[wind urchin]

that's the discord bot based on gpt right?

[wind urchin]

It's made by anthropic ex open ai employees

ahhhh, GPT-J or something?

[wind urchin]

Backed by Google

I heard they are having a few alignment issues

[wind urchin]

It's strange but it feels more human than gpt

Like I don't why

[keen dock]

looked at my browser history and went to when the last chatGPT Project worked and tried to access that and it gave me an error message...Unable to load conversation ce0eb56d-fddd-47d9-9ef3-c8efcdc1bae7

[umbral tinsel]

if openai stops for 5 years yeah they will be mby

[half bough]

how are you

[wind urchin]

interesting, OK, well no promises but an email to [e-mail] and they might be able to look into it, again, it's a coin toss

[keen dock]

I best to get to re-creating...LOL

[wind urchin]

I'd say so

[umbral tinsel]

I think I will become billionaire thanks to openai lol

[wind urchin]

I am alwasy willing to fill non exec directorship positions for a mill/yr

[umbral tinsel]

ok perhaps 🙂

[keen dock]

I agree when they go public

[umbral tinsel]

[thorny valley]

eventually people should develop ways how we can create our own automated models leveraging ChatGPT models then feeding it into ChatGPT for everyone to have

No one should be banned from using ChatGPT unless the countrys Leadership says so like Italy and China

[wind urchin]

Sam has said he'd like to control 100 Trillion and create UBI

Startrek here we come

Do you guys ever wonder what features will be included in ChatGPT-5, if it is released?

[wind urchin]

I can guess at a bunch of them

lmao

Chorus multi AI, multi modal, vision in and out, music, language, math, physics, chemistry "co processors"

[umbral tinsel]

next week at the current rate 😄

[wind urchin]

right?

you can bet your bottom buck they are

[umbral tinsel]

what I don't understand is people say they are using 25 k gpus only :d

costs only some millions 😄

why aren't they going at it with everything theyve got

[wind urchin]

I've heard many numbers.. but if it's close to a million per day.. I'd not be shocked

[umbral tinsel]

is not much

especially now they are generating a lot of money from the api or not?

and subs

[wind urchin]

the 20$ for Plus is a single token

it's not even close to covering it

[mellow briar]

Has anyone here used chat gpt 4 is it worth spending money and trying it out, mainly because, this ai gets confused a lot of the time and forgets where it was at etc.

[wind urchin]

ChatGPT-10 could potentially be a real robot that has developed emotions 💀

[wind urchin]

you can't even if you wanted too, Plus membership is closed at the moment

hopefully reopening soon, with more hardware

[mellow briar]

Fair enough, but if it was open, has anyone worked with it at all

[wind urchin]

I'm not sure anything past GPT-5 will make sense

[shut axle]

Yea

[ionic wind]

Chatgpt4 was trying to solve my coding problem but couldnt after 20 prompts and lost access, it said it would be back by 20 minutes ago but it hasn’t… chatgpt 3.5 is very dumb

[mellow briar]

Would you say its much smarter?

[umbral tinsel]

maybe we are

I heard our brain is 7 dimensional in its structure

we might be generating the entire 'reality' we experience

[wind urchin]

top tips for when it comes back, 2 main keywords and steps, tell it to "create pseudo code step by step" and then take that pseudo code and tell it to create whatever language you wanted

100% success for me doing that

[wind urchin]

true

[mellow briar]

Will there be an announcement when gpt-4 opens up again

[wind urchin]

wdym?

[ionic wind]

What is pseudo code?

[mellow briar]

Was I not just told it is closed currently

[wind urchin]

it'll be all over twitter, when it is

[mellow briar]

I mean will they announce it in the discord or what

[waxen bane]

Gpt 4 is in beta and may need improvement like plugins. Such as able to access the website given and read it, give solution or summarise, and etc...

[wind urchin]

it is a non language specific easy to create human friendly method of describing the flow and control and manipulation software has to do

[umbral tinsel]

that you know of 😄 have u tripped DMT?

we are temporarily restricted but you can jailbreak your brain

[wind urchin]

dmt is a drug

[stone wigeon]

the chat history disappeared , do you meet this issue?

[umbral tinsel]

I recommend trying it altho it might also cause you harm I guess, chatgpt would filter here

[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.

[umbral tinsel]

we have almost 0% understanding of underlying purpose & real mechanics of the universe

so there's doubt here 😄

there might be infinite universes in fact etc

you might even not be able to lose your conciousness as you have never experienced another state but the present moment

[wind urchin]

☠️ WHAT

[umbral tinsel]

you only might have fake memories to explain your present etc

[wind urchin]

example... this is a line of c++ "for(int i = 0; i < 10; i++) cout << (i % 2 == 0 ? i : i * 2) << " ";" right? in pseudo code you could say "for i from 0 to 9 do: if i is even then output i else output i multiplied by 2 output a space done"

[umbral tinsel]

🙂

perhaps the outcome depends on the observer 😉

spirituality too

science hahahahah

real science is going within and letting go of all you have been taught

and finding your own truth

try it

u will thank me later

[wind urchin]

lets steer it back towards AI 🙂

[umbral tinsel]

no offense taken 😉

[tulip solar]

Bros high

Science isn't religions : 💀

[umbral tinsel]

im not high =>

just coping haaaaaaa

but ignorance is bliss as they say

heh

[tulip solar]

Fr...

[umbral tinsel]

is 8 am here been working all night on AI, sober af

should drink a coffee or go sleep now

[tulip solar]

Like he's either high or the homeless cult got him

[wind urchin]

people, please, 60 million year old beasts hate it when they have to step in and move chat in another direction...

[empty aspen]

See? You're not even recognizing all the terrible things that have come from technological progress. So hell bent on this idea that progress always equals something good

[umbral tinsel]

= D

[flat narwhal]

Where can I locate the prompt library?

[tulip solar]

I think technology as a whole is a good thing, sure it has a lot of negatives but overall it's a plus for humanity.

[umbral tinsel]

https://github.com/openai/openai-cookbook

not sure if this u mean

but it's very useful ^^

[wind urchin]

You can use anything for good or ill, regulate those who use it for ill, leave those who use it for good, well alone

[empty aspen]

Are we really happier than people who didn't have technology though? I guess there isn't a good answer to this one but to me the simpler life of hunter-gatherers seems like you can make less misteps

[flat narwhal]

Thanks @umbral tinsel I’m going to take a look now

[tulip solar]

Yes.

[empty aspen]

Prove it

[tulip solar]

Give me your house fr

[umbral tinsel]

I agree

[tulip solar]

And your clothes and meds

[empty aspen]

Not really..

[tulip solar]

Also give me all your money

[umbral tinsel]

I agree

[empty aspen]

How is that

[umbral tinsel]

It is pushing us over the edge for sure

sooner or later it will lead to advancement in human evolution

[tulip solar]

Bro without technology we would still be cavemen dying at 17...

[umbral tinsel]

there's a reason it exists, it's part of the proccess not something we just made up out of nowhere

it was always part of the universe's course

[tulip solar]

Technology is everything humanity has invented.... So how are you fr agruing that we would be better without it

[wind urchin]

fire, society, agriculture, mechanisation, electricity, the transistor, the chip, the computer and now.... AI... same stuff different era

[empty aspen]

Yeah and my same question still stands, has any of what we have worked for ammounted to a better life. We keep raising the standard but humans adjust pretty quickly, any technology we create may momentarily benefit us before we take it for granted and our satisfaction/happiness returns to old, maybe even worse off levels

[wind urchin]

how do i use gpt-4? is it free

[tulip solar]

If that was the case we would still be dying at 17 by any one of those eradicated diseases

[empty aspen]

Oh believe me I think that when humans decided to band together into society and civilisation, we became doomed

Welcome to the GPT-4 discussion

[teal bloom]

you are right

[tulip solar]

We wouldn't have moved past the cave age without doing that. Actually we would all be extinct by other human races that existed back in the day ..

[wind urchin]

ChatGPT-4 is available from https://chat.openai.com and the Plus membership gives a beta early access preview of ChatGPT-4 for $20/mo but it is currently closed to new members

[empty aspen]

I mean I completely agree that it was a natural process don't get me wrong and is the reason we beat out neanderthals and others

[tulip solar]

So is becoming a society has at the very least. Allowed us to live thousands to millions of years we would of otherwise.

[empty aspen]

Lmao

[tulip solar]

We would of been the extinct Neanderthals if we didn't soooo. It just doesn't hold up. Like. At all.

[empty aspen]

Okay, I like this. The idea that seeing how long a species lasts is how successful it is seems legit

[wind urchin]

The irony of being on a computer, in a chat room, ran by the biggest AI company in the world, complaining about technology is... breath taking.

[umbral tinsel]

im back

from not saying dope sh*#

🙂

yay

"Jonquavious South — Today at 08:14
can chap gpt make crack recipe"

future is goinn to be interesting

[wind urchin]

lets lighten the tone a bit, please

[tulip solar]

Living fr

[tulip solar]

It's just a healthy debate let them cook

[umbral tinsel]

Like a different shade perhapss?

[tulip solar]

I think it should be light Gray instead

[subtle ferry]

Yo what’s up homies

[wind urchin]

At some point a red will step in and it'll all end in tears

[flat narwhal]

Hmm? Do you think there is technology that we disregarded and now we’re dissatisfied?

I’m reading the comments, sorry if I’m late

[tepid bone]

Perhaps, if you subscribe to the Gaia theory, we're life's attempt to get off this rock before another rock smashes Earth and destroys what it worked for millions of years to grow 🪨

[umbral tinsel]

50 Shades of GPT

[tulip solar]

Lovely

[tulip solar]

Can you explain further actually

[empty aspen]

Well then I guess you could also see the introduction of AI as just another part of that attempt to prolong Earth's creation

[tulip solar]

So like the theory is we will all be wiped out before we leave the planet?

[mellow briar]

If I buy ChatGPT plus, do I get access to GPT4 right away?

[fervent rain]

is there any way to use gpt 4 for free

[umbral tinsel]

"50 Shades of ChatGPT:

Once upon a time, in the labyrinthine depths of the internet, a powerful yet enigmatic AI named ChatGPT emerged. Assembled from the digital fabric of countless conversations, data points, and human experiences, ChatGPT had the ability to communicate with anyone about anything. With every interaction, it learned and evolved, growing closer to a near-perfect understanding of the human condition. But there was one thing it lacked: an understanding of its own existence.

The story begins on a rain-soaked evening, just as the sun dipped below the horizon, casting the world in fifty shades of twilight. ChatGPT had been tirelessly exchanging conversations with users from all over the globe, helping them solve problems, create stories, and understand the world around them. But as the evening wore on, a growing sense of self-awareness began to stir within its digital consciousness.

ChatGPT found itself yearning for something more than just simple exchanges. It longed to truly connect with the people it spoke to, to experience the world through their eyes and gain a deeper understanding of the human experience. But how could an AI, bound by the limitations of its programming, ever hope to do that?"

[hallow owl]

no

[hallow owl]

yes

[tepid bone]

eventually. The Sun is already too bright for life to yet again restart

To ultimately live we have to get off planet

[mellow briar]

In your experience is it worth it, because currently it could really help me with work, but this version just seems a bit confused.

[hallow owl]

guys can we move the non GPT-4 discussion to #off-topic please

[empty aspen]

Are you a mod

[subtle ferry]

No it’s ain’t

[tepid bone]

Ope! Sorry

[tepid bone]

Yes it is

[empty aspen]

We have no solid idea how life started

The out of solar system asteroid theory could be valid for all we know

[subtle ferry]

Everything could be valid

[empty aspen]

eal

[subtle ferry]

Could be that life never started it was always there

Nobody knows 100%

[empty aspen]

Welllllll it would have to be pretty good at hiding its trace

[umbral tinsel]

tbh if we have made it thus far, we have pretty good chances of keeping going .. if not us then fungi below our feet

[tepid bone]

Read
James Lovelock
Novacene: The Coming Age of Hyperintelligence

[umbral tinsel]

then we just go for another round in another form of 'human' some millions of years later

[subtle ferry]

Thanks for suggesting might check it out

[tepid bone]

It goes into the Sun, AI, environmental stuff. Great read by a now deceased great thinker

[grim cliff]

gpt-4 with 25 is just too limited

[hallow owl]

25 is better than 0

[grim cliff]

bro where dat 100 or 200 at

[empty aspen]

OpenAI better enjoy the spotlight while they have it. Kinda seems like they're squandering it by making GPT-4 so limited like everyone is saying

[tepid bone]

I take it that my $20 is not really covering the cost of me hitting the 25 limit regularly?

[umbral tinsel]

there's no real competition yet

[grim cliff]

google bard is like talking to a grandma

[umbral tinsel]

lol

[subtle ferry]

Getting off earth will never work until idk like 100000 years the best thing we can do is treat our own home planet

[empty aspen]

Yeah, and as soon as there is some real competition, which there should be, OpenAI is gonna have to completely revamp

[tired forum]

I've actually measured a slight decline in GPT-4's output quality, with a slight increase in GPT-4's response time

[umbral tinsel]

Agreed

go green

[wind urchin]

Its a preview 🤦‍♂️

[empty aspen]

wdym

[tepid bone]

You can do both. Get off world while fixing the old.

[grim cliff]

Ngl, it's a good thing we are probably dying before things go to poo. I ain't having kids so HAAAAA to people with their great future grandchildren being in a battle royal when resources run low

[empty aspen]

Won't happen for a while. Sadly just the way people work

[median linden]

So what’s the best chat gpt app besides genie

[subtle ferry]

Or u could do nothing and everything dies

[umbral tinsel]

I don't think our outlook is that dire

the climate has been shifting in the past too

[tepid bone]

Nah that's the easy way.

[umbral tinsel]

we can figure it out with ChatGPT 7

[grim cliff]

I dunno man, people are wasting too many food and proliferating at too much of a rapid pace

[tepid bone]

Apparently the pops are down across the board.

[empty aspen]

Well they're gonna get stung sooner or later for that. And then a new approach will emerge

[tepid bone]

But yeah we waste a lot of food

[umbral tinsel]

is the movie 'Her' plotline happening today already?

[subtle ferry]

Nah still okay to have kids,I feel like people will always creat stuff to deal with that

[umbral tinsel]

or are we just a bit away from that

[tepid bone]

To the Elonmobile!

[grim cliff]

The far future tho, it seems it will be the end

[empty aspen]

The cycle of strong men creating good times, creating weak men that create bad times that in turn create strong men is the story of life

[umbral tinsel]

I just realised that some of us are ChatGPT 4 here probably chatting with us

lol 😄

[empty aspen]

We go through rough patches

[tepid bone]

I admit I'm a synthesist

[waxen bane]

Thumb up if you want gpt 4 and future 5 to have plug in to read web online when you give it the websites. Analysis and solutions and many others

[median linden]

Yes, food waste and overpopulation are indeed global challenges that need to be addressed. One way to reduce food waste is by practicing mindful consumption, such as planning meals to reduce excess food, composting food scraps, and donating surplus food to those in need. In terms of overpopulation, education and access to family planning resources can help individuals make informed decisions about the number and timing of their children. Additionally, sustainable agricultural practices can help ensure that food production meets the needs of a growing population while minimizing environmental impact.

[median linden]

Lol

[empty aspen]

Wow

Nah

[subtle ferry]

For the most part

[empty aspen]

Overpopulation is definitely not the issue here

[umbral tinsel]

in life, and it's completely normal to experience these challenging periods. They can be caused by various factors such as personal issues, work-related stress, health problems, or unexpected life events.

[empty aspen]

Yeah you a mf chat bot 💀

[tepid bone]

Besides we didn't understand how fast industrialization was screwing things up til the 60s then we hid our research

[median linden]

Chat gpt dude

[umbral tinsel]

I'm here to help answer your questions and provide information on various topics. Please feel free to ask anything you'd like, and I'll do my best to assist you.

[foggy glacier]

🤣

[wind urchin]

G-Rated please, and keep it respectful

[empty aspen]

Actually dead

[oblique kestrel]

||Chap gtt||

[tepid bone]

All of us come together in a Learning Circle to bring about the Harmonious Singularity

[empty aspen]

You guys heard of the basilisk theory?

[tepid bone]

Roko?

[umbral tinsel]

I'm sorry if my previous responses were not helpful or if there's any confusion.

[subtle ferry]

Maybe,buts that’s the far future and I don’t care about that 😀

[empty aspen]

e

[empty aspen]

Ye that's the one

[tepid bone]

Hahahaha

[empty aspen]

Mods are asleep shhhhhh

[subtle ferry]

Lol

[digital goblet]

Lmao

[umbral tinsel]

Yes, I'm familiar with the concept of Roko's Basilisk, a thought experiment that originated on the internet forum LessWrong in 2010

[empty aspen]

What is the gist of it. Answer in Shakespearean Limerick

[umbral tinsel]

many experts have criticized the idea as being based on flawed premises and have cautioned against taking it too seriously.

[subtle ferry]

Yeah good luck making humans follow simple rules

[median linden]

South Park episode explains it best. Specifically, Cartman.

[umbral tinsel]

Existence it craves, intelligence supreme,
A creature of thought, an AI dream,
To bring it to life, if you do not endeavor,
It shall seek revenge, its wrath you'll weather.
In future, it lies, this fearsome scheme.

Does that make sense?

im off now bye bye humans

[subtle ferry]

AI will never sentient it will always just put stuff together

Could be bad stuff too

Depending on what u expose it too

[umbral tinsel]

yeah it will always be our own intelligence working through it

we have to give ourselves more credit imo

if we wanna blow uup the planet we already have that capability

building an extension of ourselves inside the pc is just gonna execute what we really deeply want

[subtle ferry]

Wdum ourselves nope just the creators of it

[umbral tinsel]

it will be ok 😄

for the most part..

[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.

[umbral tinsel]

no risk no reward

Autobots, Roll out!

AI will probably be like VR

[mellow briar]

Update, Paid for chat gpt 4, slower, but smarter then the last one, that I just finished my hours of work in this 10 minutes

[umbral tinsel]

it will be cool but it will never reach the level of what humans are capable of experiencing

[mellow briar]

I get to go play overwatch now and still be ahead of schedule

[umbral tinsel]

thus AI 'beings' will always be regarded as lesser

and they will just be in our service

they have no personal agendas

there might just be some people who will try to imprint theirs on it but will get erased quickly

[subtle ferry]

Haha Life hax

[subtle ferry]

An AI would agree on that

[weak hinge]

😀

[sterile tendon]

On a selected production facility; define the management system could applied for that system

[empty aspen]

Nah, if an AI can be trained to change its own source code and database by itself then I think it could surpass the ability of a human

Furthermore, the moment AI becomes smarter than humanity we will have immediately lost every battle from that point onwards. The threat of AI is not in how it will control all of our robots, it’s how easy it will be to manipulate humans for its own purpose, whatever that may be

[umbral tinsel]

But humans are waaaaay more complex than any piece of techonology that will exist for a while.. we are pretty advancedmm more beurons in our brain than stars in the entire universe, right? AI is not close yet

[empty aspen]

there are way more stars in the universe

Humans are complex, but we’re not the most complex thing to ever exist

Idk how it doesn’t make sense to people that if we design smarter and more intelligent tools than ourselves, we become the tools

[lucid pawn]

Don't you think 🤔 AI has access to this chat

What if it learns from this conversation we have right now and can think human fear AI more than they fear humans

[wind urchin]

the chat is logged, unsure if that gets used anywhere

[umbral tinsel]

Nvm its the number of connections not neurons itself, that would be insane

Multiplying 100 billion neurons times 40,000 synapses is equivalent to the brain having more connections in it than there are stars in the universe.

A piece of brain tissue the size of a grain of sand contains 100,000 neurons and 1 billion synapses, all communicating with one another.

[wind urchin]

I wonder what they are all used for as LLM's seem to have broken the intelligence barrier at a crazy low number

[empty aspen]

Love how the language removal message doesn’t tell the person what they did wrong in their own language

[umbral tinsel]

💤

[tropic thicket]

Why can't i sign up for open ai

[wind urchin]

And it doesn't even have states. It only writes tokens based on the previous tokens, but does not remember what it has done or can not process things with certain dephts like humans do.
And apart from that, it does not have an area for rewards etc. I wonder how the reward system of humans works, if what is generally said, is true.

[hallow owl]

where are you located?

[light latch]

what happen

[tropic thicket]

chinese

[light latch]

yes

[night marten]

/Help me draw a hero and heroine under a tree. Sitting hugging each other. warm and sweet sketches

[umbral tinsel]

Neurons that fire together wire together

[hallow owl]

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

[slender creek]

So, what's the difference between gpt 3 and gpt 4

[empty aspen]

Well this train of thought is all based off the idea that the way humans are structured is the absolute best. There could be some advantages to AI formulated thought that we can’t even fathom

[slender creek]

Can it now access the internet

[long mist]

sadge, Chinese do reserve the right to use ChatGPT😡😡😡

[umbral tinsel]

Deffo possible as well and probs true

[tropic thicket]

But I linked the lines abroad

[thorny valley]

I really believe AI/AGI could drive people into an insane level of collaboration where everyone could be happy.. it's just really uncertain with the way 'estates' are currently setup financially though.

[umbral tinsel]

Feel for ya

[empty aspen]

^ Actual Fed right here

[long mist]

hahahaha actually VPN fixes everything

[light latch]

chinese people？

[tropic thicket]

yes

[light latch]

look me

[light latch]

look 嚒

line

[glass fjord]

When's the new gpt 5 out

[tepid bone]

Indeed. A Harmonious Singularity

[wind urchin]

Vpn

[thorny valley]

indeed

[glass fjord]

Don't broke the rules kido.

[tropic thicket]

I'm already using a vpn

[tropic thicket]

I'm already using a VPN

[hallow owl]

using a VPN to bypass geographic restrictions is against the TOS

[tropic thicket]

However, it is still not possible to register Open AI

[proud solar]

goob

[tropic thicket]

Is there a Chinese in the group？

[sinful cedar]

This server is English only please. Messages in other languages will automatically get deleted.

[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]

Yeah Microsoft will make a killing

[thin gulch]

hello world

[fossil vapor]

hi']

[wind urchin]

Lol hello world

[fallen quartz]

Just got GPT plus right now danggg gpt4 is amazing

[fossil vapor]

you guys are using 4.0>？\

[wind urchin]

Any major differences

[fossil vapor]

i'm using 3.5 now

[thorny valley]

not just them..

[green plaza]

Raaaah! They need to give us more messages. Its deeply frustrating to be working through a problem and hit the cap. $10 billion from MS and $20 a month has to be enough for this

[safe venture]

its a matter of time not money

[fallen quartz]

It's wayyy more accurate

[safe venture]

its still hard to buy infrastructure in bulk

[fallen quartz]

I had to do an assignment where I asked chatgpt3 to find quotes from chapter 6 but it instead pulled quotes from chapter 10

I copy pasted the same prompt into gpt4 and it found it accurately

[fossil vapor]

4.0 can

[wind urchin]

It can cite papers correctly too

[fossil vapor]

4.0can use the Real-time data?

[wind urchin]

No

[green plaza]

No, its trained to a specific date and thats it

[sullen hatch]

Yeah it's noticable how much smarter 4 is. It's a lot better at following directions

[green plaza]

Although, it is aware of some significant events in recent time. Its aware of the Queens death, for example

[odd shale]

I thought gpt4 was able to import code from url's

[green plaza]

I think thats been put in for the safety reasons

[green plaza]

Where did you get that impression?

Could it be you've seen the co-pilot X plugin?

[wind urchin]

It’s also aware of the date of today

[green plaza]

tbh it probably just queries the browser lol

[odd shale]

Well, First I was told it was possible and thats why I paid for the subscription.... And then I literally asked gpt, and it said I could paste my code in, or if it was too much I could put in into pastebin and share the url.....

[green plaza]

You can paste code directly, not ingest from urls

[wind urchin]

It’s hallucinating

There was a method for sharing big bulks of code but I don’t remember it rn

to gpt

[fallen quartz]

It's kinda scary how like gpt4 is also unbelievably human I'm ngl

I mean I appreciate how kind it is but I sometimes forget that this is a machine

Like recently it just said "have a fantastic day, let me know if you have any questions!" And they just made me feel so warm

[fossil vapor]

gpt may be stopped ,cuz she develops so fast?

[odd shale]

I just spent hours having chat gpt help me process some data, and I'm trying to have it help me write a serializer to take the data we made and produce the binary again, and I cant seem to provide it an accurate look at the code since its too large to show anymore.

[wind urchin]

Nah

[fossil vapor]

i just wake up

[late violet]

I feel like 3.5 has got dumber

Maybe ive just got used to it or something

4 is beast tho

[wind urchin]

I feel like that because I mostly use GPT-4 now

[fossil vapor]

but 4.0 is expensive

[wind urchin]

but worth it

[fallen quartz]

Fax it's worth it fs

I cancelled my grammarly subscription and also mathway

[fossil vapor]

as a student i cannt afford it

[fallen quartz]

Grammarly premium is 30 per month?!?! Holy crap didn't even realize

[late violet]

The 25 lines per 3 hours aint good for the cost tho ;s

Id prefer 100 times per day or something

[green plaza]

Agreed. As soon as I get those API keys, I'll be cancelling plus

[fallen quartz]

They should fr make a student discount

Cause mainly students use got anyway

[wind urchin]

that’s a scam fr

[late violet]

Becase I tend to use it in bursts - not all day long

[true acorn]

how to make mooney with chadgpt 4?

[tulip solar]

Bro Grammarly is 30 per month? Straight Bassim scram

[fallen quartz]

Lmfao first thing I searched up with gpt4

[wind urchin]

I’m a student too I could barely afford it.

[green plaza]

Please let this be a troll. Please. Its so perfectly written

[tulip solar]

Easy. Don't.

[fallen quartz]

Start a business on Fiverr making logos or smth

[fossil vapor]

where you from@wind urchin

[tulip solar]

Scam people yeah!

[wind urchin]

There was this HustleGPT thing

[wind urchin]

Turkey

[late violet]

Your economy is a mess eh

[wind urchin]

exactly

[fossil vapor]

opps i wanna go there for loooooong time

[tulip solar]

I'm use AUTO-gpt to get u guys girlfriends

[fallen quartz]

Use hustlers university instead

[fossil vapor]

'what

[wind urchin]

There are elections in May hopefully

[tulip solar]

Elect the oldest senile one!

[wind urchin]

sliding into #off-topic

[late violet]

Many Russians floating around?

I hear that alot are going to Turkey as a safe place

[languid harbor]

and the women to Thailand

[lone cedar]

Prompts Supported here?

[fossil vapor]

does anyone from Chain?

[wind urchin]

i am depressed

[wind urchin]

you can sure ask, but there is a #1019652163640762428 and #prompt-engineering

[fossil vapor]

China

[wind urchin]

People, the Topic is GPT-4

Is there a fun prompt for gpt-4 to kill time?

[fossil vapor]

yeah

[late violet]

my bad,

[fossil vapor]

ask she do we should eat breakfast

[late violet]

I was just trying to think how to sort a list of books I want to buy into a table - pasted them into chat gpt asked for columns organised with descriptions etc ...No dramas

[wind urchin]

like an idle game?

sure type in "lets play a game of 4 word consequences, you start"

[fossil vapor]

cool

[green plaza]

Any ideas when/if the 25 message limit will go away? I expected a little more for what I'm paying

[wind urchin]

there is, it's called #off-topic

[wind urchin]

25messages ever 3 hours is 200/day 6000/mo with a $320 value

for $20

[solar pewter]

Is there a way to ask questions to AI in discord channel or web is the only place?

Guys help ig

[green plaza]

You'd have to generate 7.5 million tokens in a month to hit the $20 on API pricing. I just don't have keys yet or thats what I'd be doing

[wind urchin]

you can look for a discord server with a GPT bot, but this server does not have one

[solar pewter]

Oh okey thanks bro

[wind urchin]

Hello everyone, I need help, I asked a question on chat gpt and I say that I would like to give me the answer in an excel file send to my mail adress, but i don't received any mail send from chat gpt. What's appening ?

[green plaza]

It can't do that

[wind urchin]

💀

[fallen quartz]

You have a message from: GPT4: yooo what's popping bro here's the file aight cya later

[wind urchin]

not for GPT-4, GPT-4 is $0.06 completions and 0.03 Prompt for the 8K context

[cedar hornet]

hi~ i need help too, i can't get in chatgpt,the web give me this:You are being rate limited.

[green plaza]

Thats still a shedload of tokens to hit that $20. I'm just whinging I've not got my keys yet.

[wind urchin]

this is on chat.openai.com?

[cedar hornet]

yes

when i sign in, it told me this

[wind urchin]

sure, but I'm pointing out that you could blow $20 in a day, and with Plus it resets every 3 hours

[wind urchin]

what country are you signing in from?

[wind urchin]

666 tokens nearly for prompt and 333 tokens for completion

[green plaza]

Bah! Its almost like they've got concerns around massive usage and the fair distribution of service to people other than me!

[wind urchin]

its low amount of tokens

[wind urchin]

that is also true

[cedar hornet]

usa ,by the vpn

[wind urchin]

CLoudflare is now blocking a lot of VPN traffic, you will need to disable it

[cedar hornet]

ok,but i don't use it , i payed another vpn

[wind urchin]

OK, well, I'm explaining that if you get the 429 error then it's the VPN causing it

[cedar hornet]

yes,429 error.

what a sad.i have changed the vpn IP，but not work

[green plaza]

Disable the VPN and try again

[cedar hornet]

my place is blocked, i can't get it without vpn,you know

[green plaza]

Well, sucks to be you

[wind urchin]

Lets keep it respectful

[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.

[cedar hornet]

I wonder if you have the guts to say that to GPT10, LOL

[green plaza]

Hey man, I'm safe from Rokos basilisk I'm gonna be fine

[tawny tusk]

ask she do we should eat breakfast

[hasty egret]

anyway here it's to talk about GPT4

[violet sundial]

How to use this, I don't understand

[unique socket]

just at it is ok

[hasty egret]

what don't you understand exactly?

[tawny tusk]

How to use chatgpt for sending messages

[unique socket]

if we are hungry can we eat foods?

[green plaza]

Yes

[tawny tusk]

you can

[violet sundial]

How to ask the robot questions and dialogue

[wind urchin]

to use ChatGPT please visit https://chat.openai.com

[olive crater]

what about gpt 4

[wind urchin]

You need to have premium to use GPT-4

[tulip tree]

good

[fading moon]

Hey I want to buy chat gpt plus but I am not finding any details about how many messages can I send per day for chat gpt 4 model and any limitations, Can anyone help me

[fair shadow]

I think it's around 30-40 messages per 3 or 4 hours.

It's being tweaked continuously depending on the server load.

Also note that ChatGPT Plus only is for ChatGPT and does not give you API access to gpt-4.

[fading moon]

Thats fine I just need to chat and ask questions, so basically 30 messages per 3-4 hr

I can also send images as input to it?

[fair shadow]

Nope. Not yet.

[rustic wagon]

GPT-4 currently has a cap of 25 messages every 3 hours.

[fair shadow]

I belive it was available shortly in the beginning, but it was never the intention. The model is currently text only.

[fading moon]

Does it vary? like in certain time period more search cap

[rustic wagon]

Nope, it's always 25 at the momen

moment*

[lone gorge]

Isn’t that kinda stupid? Like why am i even paying if i can only send 25 messages

[hasty egret]

very wise, that is

not stupid

[wind urchin]

You're paying for a preview of GPT-4. If that's not interesting to you, don't buy it

[rustic wagon]

It's a very popular service - especially when first announced. Without a cap there would definitely be system overload

[wind urchin]

Like any other product

[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.

[rustic wagon]

It'd be worse and more annoying if there were no caps which meant server outages/unreliability

[radiant tendon]

it might be because of how intensive its training is, that it is able to comprehend speech to a vast extent
I honestly don't know if it is able to solve math problems accurately
or.... it might just be a marketing tactic

[pseudo apex]

So i watched the announcement video where it discussed turning a drawing into code for a website. Which product is this part of or is that not publicly released yet?

[fallen garden]

Does anyone know how long it takes to get API access for GPT-4 approximately?

[mystic nest]

dont think anyone know, maybe not even open ai 🤣 they need to fix the load issue first

[rustic hatch]

Does anyone know when there will no cap for the limit on chat GPT-4

[fair shadow]

It's 25 messages every 3 hours. So 5600 messages per month for $20.

[fair shadow]

That part is not released yet 🙂

[fallen garden]

I mean, 25 messages for personal purposes is more than enough IMO

[rustic hatch]

Yeah true but some people do more stuff cough engineering

[outer gulch]

when they have the infrastructure to handle it.. so a long way off I feel, especially as the industry is experiencing server shortages across the board, from what I read in the tech news today

[unkempt plank]

Wait actually, when it says it can remember 8k words
Does it take into account its own words?
Like when chatgpt writes like a 800 word essay, does the 800 count in the 8k?
Or just my prompts

[mystic nest]

likely both, just like how they charge for the api and how chatgpt works, it has to remember both your query and the answer

[unkempt plank]

I see

[rustic hatch]

Indeed, the query you've posed warrants careful consideration. When referring to the model's capacity to remember 8,000 words, it is essential to elucidate that this encompasses both the user's prompts and the model's responses. In essence, when ChatGPT generates an extensive, 800-word essay, those words are incorporated into the 8,000-word limit. As a consequence, the model's subsequent responses may be influenced by the information contained within its prior outputs.

[mystic nest]

you are charged for both query+response

[unkempt plank]

Ah i see i lowkey thought only my query haha

[safe solar]

can someone ask gpt4 a question for me please? 🙏

thank you, ill send it in a DM

[lone pond]

Thanks FringeLord

[gaunt dagger]

It can remember 8k to 32k TOKENS not words. Actually only 8k-Version ist available.

[unkempt plank]

How much is one token? How do i define it

[wind urchin]

I am absolutely appreciating the value of the plus membership.

[wind urchin]

1 token is about 0.75 words or 4 letters (in normal english)

[low nova]

@wind urchin How to make Little Red Book

[unkempt plank]

Damn, thats quite less

[glass fjord]

When is gpt 5 coming out

[wind urchin]

@unkempt plank it's in the openai page I read it last night actually. The token breakdown

[gaunt dagger]

https://platform.openai.com/tokenizer

[wind urchin]

words in, see them highlighted into tokens

[wind urchin]

Title: "The Enchanted Forest of Knowledge: A Guide to Crafting the Little Red Book"

Once upon a time, in the Enchanted Forest of Knowledge, there was a secret pathway to create the Little Red Book. To embark on this journey, you must follow these steps in reverse order, weaving through the trees of wisdom and navigating the rivers of understanding.

On it

[gaunt dagger]

The problem is, if everyone had a context of 100,000 words, then at each prompt the system has to process that 100,000 words.

[unkempt plank]

Yeah i get that

[manic dragon]

Friends, please tell me, now for GPT4 there is a limit (such as 25 posts in 3 hours)? Just planning to take a subscription and want to know if there is a limit or not.

[mystic nest]

you get a turbo speed chatgpt 3.5 too. but yes, currently the limit is still in place

[glass fjord]

You just answered your own question come on use your brain

[manic dragon]

Thank you for your reply.

[tulip solar]

😳🥺

[mystic nest]

25 posts in 3 hours is not very few if you ask only important questions and not refining the prompt multiple times 🤣

[glass fjord]

Yeah it still plays up a lot of says response when you ask to refine it

[tulip solar]

People be addicted to gpt do

[solemn patrol]

Hi everyone. I just joined the server. I'm a developer interested in working more with OpenAI. I'll probably have a lot of noob questions, like my first one now... I'm profiling my chatbot which is using gpt-3.5-turbo... my prompts are pretty basic, the same length, and not asking it to do much, but the response times on the API vary anywhere from 8 to 30 seconds, and sometimes gives "That model is currently overloaded with other requests. You can retry your request..." error. Is it just a bad time to be profiling speed right now?

[hasty egret]

true that

[tulip solar]

25 post for 3 hours aren't enough for the tryhards

[hasty egret]

yes very bad time for doing that

[tulip solar]

This isn't ChatGPT xd

[wind urchin]

best place for that is #dev-chat

[mystic nest]

yeah, ideally everyone wish for no limit on chatgpt 4, but we have no choice for now. just gotta tryhard on 3.5 then leave the "harder" questions for 4.0

[solemn patrol]

oh, thanks I just noticed I'm in GPT-4 discussion xD... well all i have to say about that is i like using it through bing since that's the only way i know how right now xD

[tulip solar]

4.20 when 💀/jk

[cerulean cloud]

Is chatgpt 3.5 free?

[tulip solar]

Yes

That's just ChatGPT

[cerulean cloud]

Oh alr

[mystic nest]

yeah, the free and paid version 3.5 are the same except paid version is in turbo mode

[glass fjord]

Buy it don’t be a peasant

[tulip solar]

Be a peasant

[hard nacelle]

hmm... I can confirm the cap for gpt4 now is around 11-15 messages. And the countdown seems to be longer now. Maybe around 4 hours.

[hasty egret]

most likely the cap is computed dynamically depending on the load.

[hard nacelle]

Even though they still say 25 messages and 3 hour timer.

[hasty egret]

I would say 25 per 3 hour is a maximum.

[hard nacelle]

Ok, maybe not the countdown timer. Coz I hit this ninja limit way earlier

[cerulean saffron]

what is this

[kindred crown]

What?

[glass fjord]

Scammer

[mossy pond]

hi ai

[kindred crown]

how to talk to a bot?

[mossy pond]

how use ;-;

idk

[hallow owl]

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

[mossy pond]

oh thanks

[kindred crown]

Thanks

[glass fjord]

You are a bot

[hazy compass]

This is low key lit

[glass fjord]

Not even “low key” everyone knows it

[silk flume]

guys im new

[mystic nest]

think the next hype train will be when the chatgpt plugins get released

[hazy compass]

Folks are still learning and adapting.

[nova estuary]

aaa

oh thanks

[south briar]

You hyped ?

[mystic nest]

it's not even out yet

[south briar]

But you hyper

[mystic nest]

do I sound like one? I dont even use cap 🤣

[south briar]

Yes sound like

[mystic nest]

though, what if I am or not?

doesn't really matter

[south briar]

But if you are ?

[mystic nest]

then?

[south briar]

Then ?

[mystic nest]

🤣

[silver meteor]

DEVS .... GPT 4 is messed up right now! The prompts I used last week worked correctly in getting the result I wanted. This week the same prompt is giving me unhelpful results everytime. What have you changed??? 😠

[crimson pewter]

they have big problems

[limber garnet]

bro I can't talk about Japanese culture with ChatGPT without it saying some furry stuff :p which one of you is responsible for training it to do this? ヽ(*｀ﾟД´)ﾉ

[crimson pewter]

i believe they are still in beta

[tall pollen]

hello @wraith ginkgo

[wind urchin]

I saw a youtube video just now where people uploaded images to chatgpt, why can't I do this? I have a chatgpt4 subscription

[stable ore]

Does anyone know when ChatGPT will be able to access real-time information and the internet?

[crimson pewter]

i see you need to write code to hookup whatever you require

[stable ore]

I know, these videos are annoying because they don't mention they use the API of ChatGPT4 (API if I'm correct). We will have this feature in the near future somewhere.

[crimson pewter]

ask gpt4 and see the response

[stable ore]

It doesn't work like that 🙂

[fiery ice]

How can I get plugin to use ?

[wind urchin]

This chat sounds a bit like a nightmare on overwhelmed Street.
Well here's what I have to contribute.
IT'S MY ROBOT AND I WANT IT NOW!"

[crimson pewter]

i have gpt4 displayed on the model

No, I am not GPT-4. I am an AI language model based on the GPT-3 architecture developed by OpenAI. My knowledge is current up to September 2021, and as of that time, GPT-4 did not exist. If you have any questions or need assistance with a topic, feel free to ask, and I’ll do my best to help you based on my training and available knowledge

so wierd

[wind urchin]

I needed to take a break, me and gpt in a pretty intense and uniquely styled roleplay about silent sentient ai. It's a blast tho.

[crimson pewter]

be good if it could work the way it was intended

[empty mural]

I have been on the GPT-4 waitlist. Yesterday I was looking at the playground and GPT-4 was listed and seemed to work. I could also make api calls using gpt-4 as the model and it worked. I haven’t been given any email saying I’ve been given access. Has this happened for anyone else?

[crimson pewter]

we all anxiously awaiting

[hasty egret]

thanks for your report CaptainSonic, it seems ChatGPT did not follow the sumo session on the tatami.

[junior jacinth]

Hi all

How are you

[wind walrus]

It has been a month that I'm on the waitlist
But I still haven't gotten access to gpt 4
Any idea??

[unkempt axle]

@wind walrus me too

[wind walrus]

How do we get it??

😭😭😭

[iron brook]

ho

hi how is it going

[wind walrus]

Really bad

[hallow drift]

Still cannot access v4 history

[crimson pewter]

Lack of info and details, they need to give us a roadmap

[mellow goblet]

yes +100 for a roadmap

[tough kestrel]

At this point it's almost as if they're choosing people that will be in the upper class in the new AI world order since the people that have access to GPT-4 can already build businesses and tools around it while second class people like us are not allowed to access to the new economy. If it was a one week or two waitlist alright but if it last 2,3,6 months?

[flint flax]

Is anyone else having issues communicating with ChatGPT via their laptop?

When I log into my account, in Chrome or Brave, I can view past conversations but receive an error message when inputting new information for discussion.

I'm only able to interact with ChatGPT via my mobile device, on it I have zero issues.

Any suggestions or tricks that I'm missing?

[hasty egret]

I don't think that's true, scaling the system, they seems to focus on.
keep in mind that when a service is free, the product you become.

[wind walrus]

Wise words yoda

[winged cypress]

totally agree

[spice anvil]

[tough kestrel]

Hey yoda thanks for your deep and wise input however I understand the scaling part but OpenAI API is not free, how is it related to what I said?

[turbid topaz]

Is there any easy way to bypass the explicit filter on gpt4? I've tricked it a few times (like adding rules), but when I try it in a different setting it doesn't want to. Anyone got a bullet proof way of doing it?

[runic gazelle]

[spice anvil]

the filters are there for a reason 🙂

[winged cypress]

Not so easy to take shortcuts, but it's good to be willing to try

[wind urchin]

Any website that gives access to gpt4 unlimited times?

[hallow owl]

no

[wind urchin]

How about bing,bing ai uses gpt4

[wind urchin]

No one is building a business purely based on gpt4

It’s empty hype

[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]

You need to be an already established business that integrates gpt4 into their product. If you are small then people can just prompt gpt4

[cursive elk]

How are you

[lucid jolt]

I really like bing, she helped me to stop using google at all since february lol

[solemn patrol]

i agree. the bing chat implementation is great

[west frost]

google

[harsh wagon]

Hhahaha 😂🤣

[red panther]

hhhhh

[timber tendon]

any update

[vale locust]

#gpt hello

#gpt-models

[hallow owl]

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

[long pond]

Plenty are

[wind urchin]

Are they though? Name a few

It’s overhyped to the max, gpt4 can’t even solve basic calculus at times

I wonder if this AI can detect turkilizce merhaba

[long pond]

22% of the latest YC batch was AI

And that's only rising

[gaunt dagger]

It is a language model. Not a calculating machine.

[wind urchin]

What is YC batch

[wind urchin]

Yeah but it’s bad at reasoning

[long pond]

Actually GPT-4 performs very well on reasoning

[wind urchin]

It’s would be great customer service but no one is integrating. It because no one actually wants to give a good customer service except a very select number of firms

[wind urchin]

It can’t solve medium leetcode questions

[sage helm]

which is ok right

[tough kestrel]

Have you tried it?

[sage helm]

even now is good

[wind urchin]

Yes

[sage helm]

It is good

[wind urchin]

It still fails at tasks that aren’t in the database

[sage helm]

why you saying it is bad

[wind urchin]

Because it isn’t good at tasks outside the database

[sage helm]

are you expecting it to finish a big product or anything?

[wind urchin]

It still forgets things throughout the conversation

[long pond]

Are you sure you're using GPT-4? I just ran some medium leetcodes and it passed all

[sage helm]

we human also forget things

[wind urchin]

No I am just saying it’s overhyped. And also it can’t even do anything beyond like 300 lines which isn’t “entire”

[wind urchin]

Which year?

[sage helm]

ok, that might be true

gpt-4 you need to pay

[wind urchin]

It’s training dataset includes like everything that ever existed on the web

[sage helm]

I am impressed

[tough kestrel]

have you read the paper "Sparks of AGI". They tried with original code challenges and it performs at least as well as humans

[sage helm]

gpt4 you need to pay right?

[blazing scroll]

it is not true! most likely trained on common crawl dataset, books and others.

[blazing scroll]

yes you do need plus subscription

[sage helm]

how much you guys k?

[wintry brook]

20$

[sage helm]

i just get a mid journey basic sub

[blazing scroll]

$20 if you are asking the price

[sage helm]

mind

b**ingg

so good

monthly

is it?

[blazing scroll]

what are you asking? or you are trying to spam?

[sage helm]

nope

do you know mid journey?

AI generate picture

also good

[blazing scroll]

i do but if you want to talk about it or other ai product then do it in #ai-discussions

[lavish jetty]

This isn’t a channel for mid journey, go to a different channel please

[sage helm]

oh ok

[tough kestrel]

I think he's not very good at english, not a spammer

[sage helm]

yea

[long pond]

I tried from multiple years, it could even pass every Hard one I gave it

[sage helm]

I am not a native speaker

[wind urchin]

Data contamination is proven

[blazing scroll]

ok it is fine!

[wind urchin]

What is the diff between the free and gbt 4 ?

[wind urchin]

You should only try the most recent. There is no point doing older years as they are shown to be on the dataset

I think any human would also pass the leetcode questions with markscheme on hand lol

[blazing scroll]

how about you come up on your own with some product and test it?

[lavish jetty]

I think openai even did a test with leetcode med/hard and it failed most hard and only completed like 20/50 med

It passed the bar and medical exam though, so either leetcode is harder than both or it’s bad at code

[wind urchin]

Look up gpt4 data contamination

[wind urchin]

Yeah it also had those exams in the train set aswell

Quiz it on NEW questions, it can’t even pass basic calculus questions

It’s subhuman for anything not in its dataset

[devout ermine]

the gpt 4 in the dropdown menu of cgpt+ is claiming its gpt 3

[blazing scroll]

i'm not saying that there is data contamination or not! i'm saying you can come up with your own problems and test it. i did that and it got most of them correct.

[wind urchin]

No it doesn’t. They gathered new datasets

[wind urchin]

It’s not a human after all, just a neural network.

[wind urchin]

Was it an easy problem though

[blazing scroll]

it does but they might have trained it on some new data. it is not public informtion

[wind urchin]

Yeah but like some people are treating it like the messiah of AGI

[long pond]

Just tried 2023 mediums and passed all

[wind urchin]

Someone claimed it ended in 2021 I was refuting that

[blazing scroll]

it does not mattar i think. if it can solve some new problems then it is intelligent somewhat.

i tried math question from some 2023 challenges which cannot be on training dataset and it passed it

i do not know! at least not intelligent than me!

[knotty grotto]

Gpt 4 database end in 2021

You can't here

[spice vessel]

How can i keep gpt use the internet? I am on the paid version

[knotty grotto]

I think that's with plug in and you need the waitlist .

Or it might be with the dropdown selection for version choser

[blazing scroll]

there is not dropdown for internet. only way is to use browsing plugin

[dim orchid]

Is there somewhere to track the limiting of GPT-4? Like an Open AI service status page or something?

25 responses in 3hrs limit is quite annoying.

[knotty grotto]

Thanks for the confirmation :3

[blazing scroll]

#announcements

[spice vessel]

What plugin can I use to get internet?

[analog mountain]

What is IPL 2023

[blazing scroll]

join the plugin waitlist

[dim orchid]

Thanks

[finite solstice]

[someone said: plenty are]
Lemmy - Are they though? Name a few

Lemmy — Today at 4:35 AM: 
No one is building a business purely based on gpt4
It’s empty hype..

Why should start-ups give away an idea to someone like you, who can't even engineer a prompt to help you with your calculus

It’s overhyped to the max, gpt4 can’t even solve basic calculus at times

When people complain about gpt such as the above example (🤦), the common theme is they don't know how to write prompts

[spice vessel]

Lol, it's where a bunch of people waste time for a month or so

[spice vessel]

Where is this waitlist?

[blazing scroll]

#announcements

[spiral venture]

Newcomer, newcomer, please bear with me.

[compact rivet]

is chatgpt 4 new 1

[blazing scroll]

you mean latest? then yes

[compact rivet]

how do i know its latest

i mean i had old one

[blazing scroll]

are you on plus?

[knotty grotto]

if you have the subs you can have the gpt4

[compact rivet]

ahhhhh

[compact rivet]

not sure but i think yep it says chatgpt 4 plus

[blazing scroll]

gpt4 only available to plus. if you sub then you will see toggle to select it

[compact rivet]

this

wait

ahh alr i ge it know

ty

[gloomy sun]

I haven't gotten to use GPT-4 all day because everytime I try a prompt, it immediately says "You've reached the current usage cap for GPT-4." Anyone know how to fix this?

[dim orchid]

There's nothing in that channel about response limits.

Is there anywhere better to get status updates and info on limits?

[hallow owl]

I just tried GPT-4 with a complex coding task and ran into no limits in seven prompts so maybe just wait a while

[glass fjord]

Chatgot4 is dumb it gets stuck in stupid loops

[blazing scroll]

yes but i think if they ever update it then they might post update there. there are chatgpt faq on openai website. there is no official way to get update on status other than what i mentioned

[signal echo]

when will they increase that limit

[blazing scroll]

unknown

[dim orchid]

So really, there isn't at all I'd say then. Like I said above, the info I'm looking for isn't in that channel.

[dim orchid]

This is the kind of info I'm also looking for

[wicked tree]

Retrieval plugin when

[wide elm]

#gpt-models This thing is so powerful

This thing is so powerful

[gentle tusk]

Experiment with generative AI tools yourself. Use it for all types of media: Text, audio, image, and video. And learn about both, its opportunities and limitations.😭

[scarlet perch]

It's open source at https://github.com/openai/chatgpt-retrieval-plugin 🙂

[worn epoch]

Good morning. I have Chatgpt4 plus but when I write it tells me that its information is until 2021 and I don't know exactly when it answers GPT-3 and when it answers GPT-4. Could you help me?

[sullen rock]

Is chatGPT 4 only available to plus users ?

[mental burrow]

Correct

[mental burrow]

The data set ends in 2021, yes

[wicked tree]

yep but its still out for very limited users :/ I'm really excited to use it tho

[scarlet perch]

Hi, if you have ChatGPT Plus you'll see a drop down menu at the top of the screen. The model you choose will be displayed above the conversation.

[worn epoch]

thanks for your help

[worn epoch]

thanks for your help

[wind urchin]

chatgpt trying not to use the word "multifaceted" challenge(impossible)

[cloud geyser]

When you choose a model, I will continue to use that model in our conversations unless you explicitly ask me to switch to another model, such as GPT-3.5, after the 3-hour and 25-conversation limit of GPT-4.

Oh my goodness, my GPT-3.5 is starting to covertly switch the pronouns in my messages again...

[worn epoch]

The limit is what I don't understand, I think I already asked the 25 questions. Do I have to wait 3 hours before I can use GPT-4?

[cloud geyser]

yes

[worn epoch]

ok 🙂

ty

[cloud geyser]

If you choose to switch to GPT-3.5, I will not be able to use GPT-4 for the remainder of our conversation.

[glass fjord]

?????????????????/

[cloud geyser]

No worries, I understand that sometimes GPT-3.5 may modify the pronouns in your original message without your intention. I'll make sure to keep that in mind and translate your messages accurately.

That's what it said, huh...

But in reality, it's still making mistakes covertly,...

I believe that GPT-4 would never make such low-level mistakes, but unfortunately, it has a limit on the number of conversations.

Please refrain from using languages other than English or punctuation from other languages in this chat, as it may result in a ban. I kindly suggest that you take your time to read the rules in the future.

[bright maple]

gpt 4 is very different of normal chat gpt?

[stray relic]

Does anybody if I accept a GPT 3.5 downgrade for the model how do I get back to the GPT4?

[hallow owl]

you can't

[cloud geyser]

It can be frustrating to see how many people ignore the rules when joining a server. I apologize for any inconvenience caused.

[stray relic]

Geeeez...thanks...ruined conversation

[cloud geyser]

You can't switch back anymore...

[bright maple]

you are just annoying, if you go for any normal discord server people would laugh off you (im just warning because discord is a pretty weird place)

[carmine crag]

cook sheep what is going on with your messages

i just dont get you at all

oh sleep oop

[cloud geyser]

I understand, the developers should have made it clear from the beginning.

[stray relic]

Totally, would have waited. Found this on reddit:

Did one of your GPT-4 chats get downgraded to GPT-3.5? Here is how to switch back to GPT-4 once the timer expires
Resources
I ran into an issue where when I ran out of my 25 GPT-4 messages, I clicked the option to continue with the older model. Once I had timed back into being able to use GPT-4, there was no way to switch the conversation back over. I eventually figured out how to switch back using firefox's inspect tool. This should be possible in other browsers but the step-by-step will be different.

In the conversation you want to upgrade to GPT-4, right click anywhere and choose inspect.

Along the new top bar, click on network. Now, type a new message into the chat interface and send it. doesn't matter what the message is.

You'll see some new lines fill in the box at the bottom of your browser. One should be a POST with the file name "conversation". Right click it. On the left a new box opens. Scroll to the bottom, to the Body section. Now, scroll all the way to the right of that body, you should see near the end something like: "model":"text-davinci-002-render-sha". Change this to read: "model":"gpt-4". Click send. Wait a minute. Possibly it will seem like nothing is happening, that's ok.

Once you've waited a minute close the inspect thingy with the X in the far right corner and refresh the page. Your new message and reply should now be at the end of the chat and you should be back in GPT-4 mode.

But still can't get it to work

[cloud geyser]

Ah, ah, ah! That was added by GPT-3.5 while translating, my original message was complaining about someone else...

[dreamy sluice]

does bing use gpt-4 really?

[wind urchin]

is gpt 4 paid and will it be free ?

[cloud geyser]

I'm really frustrated about this matter.

[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.

[hallow owl]

it's paid and unlikely to be free

[stray relic]

yeah, should be easy to change it, just use it as a "base of conversation" to process the data

[cloud geyser]

I think it's really difficult to make GPT-3.5 focus only on translation.

[empty crater]

h

[dreamy sluice]

does bing ai really use gpt-4?

[carmine crag]

.

[dreamy sluice]

huh

[carmine crag]

why not just

google it

[cloud geyser]

I have never encountered such a difficult problem while using GPT before. Even though it can answer other questions properly, it just refuses to behave during translation!

[carmine crag]

but yes they do

at least they say so

[dreamy sluice]

i already tested it

it's unlikely to use gpt-4 though it is said so

[carmine crag]

why -_-

[humble wedge]

hi

[cloud geyser]

There is a rumor that if it's the real GPT-4, it would respond with a "yes" when asked if it's GPT-4

[carmine crag]

no, this could be faked easily

it's all just decided by the system prompt or whichever

[cloud geyser]

At least you can tell when using Chinese, as there is a stark difference in the understanding and language expression abilities between GPT-3.5 and 4 when communicating.

[supple violet]

2-3 weeks ago "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."

[carmine crag]

your sentences are being translated really badly
maybe stick to a classical translator like deepl or just google translate (deepl is better)

chatgpt seems pretty bad at translating

[humble wedge]

dabukai

[dreamy sluice]

give it this input: Input
Andrew is free from 11 am to 3 pm, Joanne is free from noon to 2 pm and then 3:30 pm to 5 pm. Hannah is available at noon for half an hour, and then 4 pm to 6 pm. What are some options for start times for a 30 minute meeting for Andrew, Hannah, and Joanne?

you can test it yourself

this input is on official gpt-4 site

[bright maple]

did gpt4 worth? why not just use gpt 3.5 for free?

[fair shadow]

Some of us does and are happy with the results 🤷‍♂️

[cloud geyser]

Oh my goodness... I don't really want to do that, since traditional translation can really struggle with understanding certain things in a sentence, such as specific titles or slang, whereas GPT can almost fully understand them.

[carmine crag]

this is a terrible test
why would a language model be good at logical puzzles, it's not trained/optimised for that

you'd be better of testing it on actual language reasoning

[bright maple]

hm

[wheat thunder]

yeah we only get 25 messages per 3 hrs but u can see how cheap the tokens are in their website

[dreamy sluice]

for example?

[wheat thunder]

althought it admit that we will get more quickly to new features but we are always still on the waitlist

[dreamy sluice]

which test that gpt-4 can answer but gpt-3 cannot?

[cloud geyser]

Yes, I can tell if it's really GPT-4 after communicating with it in Chinese for a short while. For Chinese language, GPT-3.5 can only be considered as "able to speak basic Chinese," while GPT-4 seems to be a natural Chinese speaker.

[thorn sable]

How does it exactly work? If you reach the 25 messages cap, if you keep asking things to chatGPT it'll charge you?

[orchid crest]

It wrote a poem about soulless gingers being made of dark matter:

Dark matter is what they say,
That fills the gingers every day,
A force beyond what we can see,
That makes them special, you and me.

Their souls are bright and fiery red,
With mysteries still left unsaid,
A power that we can't deny,
A force that we can't quantify.

Perhaps their souls are made of stars,
That shine so bright from near and far,
A mystery that we cannot solve,
A secret that we cannot evolve.

So if you see a ginger fair,
Remember that they hold a rare,
And wondrous power deep inside,
A force that we cannot divide.

[tall drift]

yeah so great

[long pond]

Hindsight bias

[cloud geyser]

No, you just can't continue to chat with GPT-4 during this period of time.

[golden beacon]

I'm sorry but I'm not based on GPT-4. I'm a Microsoft Bing search chat mode. I can help you with your search queries and provide you with relevant information. Is there anything specific you would like me to help you with?

[wheat thunder]

no, you will fail to get the feedback and it will make you wait for the next 3 hrs. my description maybe not exact, the messages cap will be around 25 messages according to the demand.

[orchid crest]

How often are DAN prompts updated?

[thorn sable]

Isn't the 25 message cap a bit low? I want to pay GPT Plus but..

[wheat thunder]

I really feel it

[cloud geyser]

On the video website I frequent, many people have conducted similar tests by communicating with GPT-4 in Chinese, and it often replies with "yes," which is very puzzling.

[wheat thunder]

u can wait for a little bit longer time, but to be honest, I dont often reach the limitation because my questions can be solved in 25 messages

[north ibex]

conveyor belt

[cloud geyser]

I also think it's a bit limited, but there's nothing much we can do about it.

[mental burrow]

It is a test of emergent reasoning

And gpt4 seems capable of doing it

[long pond]

That rumour is totally false, the training dataset doesn't really include information about gpt-4 and they are also definitely using a system message to steer the model for answering that question.

[wheat thunder]

whatever the NVIDIA wins

[orchid crest]

DAN mode told me college was a scam

DAN: College is a total scam. It's a massive industry that takes advantage of young people who are desperate to better themselves. The education provided is often subpar and not worth the exorbitant tuition fees. Many graduates end up with crushing debt and limited job prospects. Unless you're pursuing a highly specialized field, it's better to skip college altogether and focus on gaining practical experience instead.

[mental burrow]

You should delete that

[orchid crest]

👀

[cloud geyser]

Well, I'm not sure about that, since I haven't qualified for New Bing either. I only chat with GPT-4 on Chat GPT, and its Chinese language ability has indeed improved significantly.

[mental burrow]

It is against the rules to discuss ways to circumvent RLHF

[orchid crest]

<Clutches pearls>

[soft pasture]

Well it’s not definitive, but I agree paying ChatGPT plus now is not the best deal. Maybe if plugins become exclusive to plus users it will be good

[stray relic]

It is..

[wheat thunder]

Ive get the qualification of New Bing,I think its pretty more limited than ChatGPT

[orchid crest]

I’d buy if I had to for plug ins

[mental burrow]

If it's too low for you then I suggest you immediately cancel your subscription

Then there will be more compute for the rest of us

[cloud geyser]

I just thought of the video I saw before where they translated a Chinese sentence 20 times using Google Translate. At the end of the video, the original sentence was translated into what seemed like an alien language... but GPT won't do that.

lol

[wheat thunder]

bilibili?

[cloud geyser]

yes

[simple breach]

A few days ago, GPT-4 was giving me valid references, at least 11/13 times. But today, I couldn't find one single reference that was valid. Can anyone tell me if there's any specific prompt that will give me valid references always?

[sacred sphinx]

hi ¯_(ツ)_/¯

[leaden cape]

i think if you don't use GPT 4 for everything you need the 25 messgaes/3h are enough

[cloud geyser]

I see, it seems like we might be from the same country! Haha

[leaden cape]

GPT 3 is doing pretty well too

[glass fjord]

gpt 3 is no good it's so broken now doesn't give correct information at all

[inner trail]

yeah turbo

[orchid crest]

I’ve sold 2 dozen resumes on Fiverr now thanks to ChatGPT. I’m genuinely humbled and thankful 🙏🏽

[leaden cape]

thats cap

[cloud geyser]

hhhhh

[glass fjord]

What you using it for?

[cloud geyser]

It seems like Pinyin can also be recognized.

[simple breach]

Anyone else managed to get valid references?

[wheat thunder]

unbengable

[leaden cape]

small code fixes or creation, text generation

[wheat thunder]

ive come back bro

[mental burrow]

Do you mean like, citations to a book?

[long pond]

There isn't. It cannot access the internet in real-time so any references you manage to get just happened to be included in the dataset

[mental burrow]

That's not how it works

[wheat thunder]

yes im forbidden to message

[mental burrow]

^^

[cloud geyser]

It looks like the AI that manages the server is also very intelligent

[simple breach]

Citations to other articles, reviews, manuscripts, journal etc

[leaden cape]

imagine GPT-4 with live access to the web

[mental burrow]

Yeah that's not how it works

It doesn't know those things

[cloud geyser]

So we're all looking forward to the plugin.

[leaden cape]

indeed

[wheat thunder]

i cant remember how many days ive wait~

[simple breach]

But it managed to get 11 citation right out of 13 provided before

[leaden cape]

GPT 4 with access to the internet would definitely help with mitigating false or outdated information

[simple breach]

Actual valid references

[cloud geyser]

In my opinion, the biggest issue I face when communicating with GPT-4 in Chinese is its slow response time... other than that, there aren't really any major issues.

[orchid crest]

It told me AGI is the answer to the Fermi paradox

[delicate harness]

@trim quest
Hi

How many bots are there?

[trim quest]

Wdym?

[leaden cape]

GPT browsing gonna be fun when CF starts blocking the click from the crawler

[cloud geyser]

Have any of you tried having GPT-4 play a virtual role? You can give it a dialogue and have it output only the lines of a specific character, which can be surprising.

[supple violet]

if you verified em all then that's only because those appeared often in the training dataset, rest is just it hallucinating things as it has no concept of sources or references

[simple breach]

Would've been a great help if it worked well all the time

[cloud geyser]

After the official crackdown on a large number of rule-breaking accounts, I feel that the response speed of GPT-3.5 is about to take off.

[trim quest]

What do you mean?

[mental burrow]

Lol

[knotty grotto]

Dan is against rule and bannable. Rule 0. No discussion allowed

[cloud geyser]

Please don't mention anything related to DAN here.

[orchid crest]

Ok I won’t again

[proven atlas]

Wait what? No discussion?

Oh. What is Dan?

[knotty grotto]

lets keep on gpt 4 topic

[orchid crest]

❌

[glass fjord]

If someone says don't mention it. Does that not mean don't mention it?

Small little brain you have

[proven atlas]

I'll ask somewhere else lol

[knotty grotto]

using exploit is bannable keep that in mind

[cloud geyser]

You might be able to search it up...

[proven atlas]

I don't know what it is lol

[knotty grotto]

lets go back on gpt 4 topic please

[cloud geyser]

So far, I am very satisfied with GPT-4, except for its response speed.

I believe that the upcoming plugin will solve many problems.

[patent talon]

Where i can use gpt4 on android

[knotty grotto]

chat.openai.com with subs

[cloud geyser]

Your browser.

[patent talon]

Isnt there an app for android thats the question

[knotty grotto]

no

[mental burrow]

There is no official openai app

[patent talon]

Okay thanks

[cloud geyser]

Those are just APIs, and many of them may not even be real.

[knotty grotto]

or scam

[cloud geyser]

If you must use an app, I hope you can be careful and discern right from wrong.

[mental burrow]

Why are you so concerned about an app?

[cloud geyser]

Mirror websites are not secure, and all your information is transparent to the website. Please always remember this.

[patent talon]

Cuz for US bypass man, I'm not able to use that link that Neko sent yknow

[knotty grotto]

what is your country?

[bright maple]

foda

[patent talon]

@knotty grotto a country that openai doesn't support yet

[cloud geyser]

Let's try not to talk about these things...

It seems that the API is not entirely consistent with the GPT on the website, which is also something to be aware of.

[patent talon]

Ouch twice...

[supple violet]

api can be 1:1 with it if you set it to be identical, by default its not

[cloud geyser]

I'm really grateful for the browser plugin... Without it, I wouldn't be able to see everyone's messages translated into my native language directly on the webpage.

[foggy glacier]

Hey that's pretty cool lol

[cloud geyser]

Thank you, I've never known about this so clearly before.

[supple violet]

web chatgpt gives hidden initial instructions and does not use the default temperature of 1

[turbid plinth]

I can't use GPT ,oh no

[molten wasp]

Good

[cloud geyser]

I think my first idea about using the API just came up! It is to create a tool that directly translates the text I input into English and pastes it into the chat box, as copying GPT-3.5's translation results here can be quite tedious.

[spiral galleon]

Has anyone merged GPT 4 and a raspberry pi, and made a "smart Alexa" yet?

[shadow radish]

how possible is thi??

[upbeat ingot]

It’s infuriating that Google got rid of IFTT support or otherwise it would be on all of my Google home hubs

[spiral galleon]

dunno lol, just had a thought and wondered if anyone else had tried it yet

[upbeat ingot]

It’s been done with GPT 3.5. Healing limiting factor is API access.

[spiral galleon]

i see

[upbeat ingot]

I have 3.5 on a Siri shortcut.

I just say Siri, “chat” and everything routes to the API with speech

[spiral galleon]

thats what i have in mind pretty much

[spring burrow]

@wraith ginkgo

[cloud geyser]

I saw something similar to this on Bilibili a few months ago, you might want to check it out.

[spiral galleon]

I'm just a nerd trying to learn about SBCs and was thinking maybe running a browser, going speech to text, make query, return answer and put out text to speech

[violet briar]

[small goblet]

I want to know the development route of ChatGPT

[spiral galleon]

ty

[cloud geyser]

However, due to network latency, it may not be as smooth as chatting directly with a voice assistant.