#chatgpt-discussions

1 messages · Page 47 of 1

maiden fractal
#

I'm not from openai but what was the issue ?

paper valve
#

Does anyooe know if our chat history is deleted after a period of time in chat gpt? All my conversations were gone recently, and I can't get them back. But make new ones that show up. Now I'm wondering if these too will also delete at a certain point

maiden fractal
maiden fractal
paper valve
maiden fractal
karmic solar
maiden fractal
karmic solar
maiden fractal
paper valve
karmic solar
#

@maiden fractal I do not

#

@maiden fractal same issue on the app on iphone as well

maiden fractal
paper valve
maiden fractal
paper valve
maiden fractal
#

With new convo ?

karmic solar
#

@maiden fractal yes, if it helps I work in IT

maiden fractal
karmic solar
#

@maiden fractal let me try brb

#

@maiden fractal Same issue

maiden fractal
paper valve
maiden fractal
paper valve
# maiden fractal Yes

What are the new features you think are super useful and would you recommend it over the free version?

maiden fractal
#

Also gpt 4 10x more smart than gpt 3.5

#

We did a test with some friend who are poker professionals, I know almost nothing about this world. We gave the same situation between 3.5 and 4 . And 3.5 didn't do the right move

paper valve
maiden fractal
paper valve
maiden fractal
karmic solar
#

Is gpt 3.5 fixed for paid users yet?

maiden fractal
karmic solar
#

here is a full list of what is going on
when I use up my limit on gpt 4 it won't connect
when I just try to do a simple 3.5 chat it won't connect
no vpn
tried 2 other browsers
tried a different computer
tried a different computer on a different network
heck I even tried a different computer on a different network with a different person named Jesse

#

@maiden fractal I sent that to their support as well

woeful scaffold
#

Our systems have detected unusual activity from your system. Please try again later.
Umm... huh?

#

I'm an unusual person (so kind of not surprised). I'm using Linux, I'm logged into two Google accounts; one where I'm using Chat GPT and one where I'm using the openai APIs for work-related purposes. I got this error on my personal account.

karmic solar
upper tulip
#

Same issue

grand pollen
#

I'd like to use dalle 3 without limits through an api key. Is there a good ui to use?

brazen pier
kindred parcel
#

GPT4 broken for anyone else right now? I pasted two pictures and it just keeps giving me the "something has gone wrong, please try again" error, and hitting regenerate does nothing (the error just recurs)

neat nimbus
#

Wow, I've read about it, but yeah it really is quite a bit dumber now.

#

It's ignoring custom instructions and generally having trouble keeping up and providing coherent answers

unreal ice
#

is there anyway to upgrade your plus account to get more usage from GPT-4? 50 messages/three hours is very limiting for a lot of things, and 3.5 doesn't support custom GPTs

lean osprey
#

Chat-gpt-4 stops response mid way and even when reloading webpage, the reloading takes forever. So, frustrated with this. It's been like 3-4 days already

cerulean pumice
#

Hi, how are you? I found out that OpenAI gives the possibility to premium users to send an invitation link to try GPT-4... Would someone be so kind as to send me that link, please? I want to try the plus version of ChatGPT

lilac fog
#

anyone knows why when the chat history is long, after asking questions, chatgpt is taking so long to get back the answer on browser but has no this kind of issue in their app?

inner dew
glossy willow
#

Aloha, everyone! Does anyone know if we can use the Assistants API to respond on in JSON mode, like the chat API?

true mirage
#

Where can we find the most recent news about chatgpt on openai website?

unreal ice
#

is it just me or is GPT-4 like, EXTRA EXTRA dumb right now?

onyx eagle
onyx eagle
unreal ice
#

I was just playing with it asking it about a potential release schedule for a project I'm working on, it just kept doing really weird, un asked for, unnecassary things. I finally had to get extremely specific and give it almost no room for interpretation or extrapolation

#

basically ended up telling it what schedule I wanted and had to stop reasoning with it about the schedule because it just wanted to reset it all the time.

#

like we talked about starting the schedule in october, it gave me a schedule starting in october, that was good. talked about adjusting the order of releases, and it set the schedule back to January

#

despite 3 or 4 mentions that, that was not going to work

onyx eagle
#

I see. It's very good to be as specific as possible to get the best results.

unreal ice
#

I mean I understand that, but I shouldn't have to re-explain the ordering logic repeatedly I don't think

#

which is kind of what was happening

#

it was doing it with other tasks to, like re-ordering list and changing formats of outputs despite explicit instructions about these things

#

like in the scheduling thing. I asked it to re-order a numbered list to 1, 3, 2, 4, 6, 5. It said ok, reset the schedule to January and basically randomized the list instead of putting it in the order I requested

fossil oyster
#

does the bing search not work for you with gpt 4 either?

onyx eagle
#

Yeah, it's just that GPT4 isn't very good at math. Probably if you use Code Interpreter then you will get better results.

frosty violet
brisk path
#

iirc there hasn't been an officially announced change

#

In model version after turbo

#

But I did get a little update GPT pop-up

frosty violet
brisk path
#

Not for me

#

Maybe you have bad Internet or smth

frosty violet
#

Nop...

#

I cant show u. Url or copy images...
Rigth now its going well, but sometimes its horrible. 10 minutes yesterday for a internet search...

onyx eagle
onyx eagle
frosty violet
#

nop

#

Just once

drowsy tinsel
#

gpt4.5 is pretty amazing

vagrant garden
#

anyone know a good prompt for replying to online forum posts?

drowsy tinsel
#

its faster than the OG gpt3.5

vagrant garden
#

I want you to analyze and create a response to my classmates post. present a challenging analysis of his post, use common words, and include a question at the end. Also, put this into one paragraph. Use first person verbiage as if you are speaking directly to them.

drowsy tinsel
vagrant garden
#

online school discussion post

#

ez

frosty violet
#

2nd chance, same result...

eternal rockBOT
#

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 @languid valley immediately.

sonic comet
#

Has the ChatGPT lagging stopped?

grizzled thicket
#

I feel like output quality has dropped. anyone else feel this way? in both image production and text generation

cobalt acorn
drowsy tinsel
karmic solar
#

who's a moderator

#

I have an honest to God fix for an issue that many users are having

maiden fractal
#

You can reach a mod by dm with @languid valley

frosty violet
maiden fractal
dense halo
umbral canyon
#

Hey everyone - does anyone know how to consistently get around JSON formatting errors? I keep getting this error even though my prompt is properly formatted JSON & JSON safe. Also, i will run the same exact API call twice, 1st time it works, 2nd time I get this error:

"message": "We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)",\n "type": "invalid_request_error",\n "param": null,\n "code": null\n }\n}\n"

spare socket
karmic solar
#

@spare socket That I'm not sure

dense halo
vestal apex
#

Is anyone else on Plus getting this? Not sure what I did to trigger it. (Our systems have detected unusual activity from your system. Please try again later.)

maiden fractal
vestal apex
#

none of those apply, I also disabled plugins, slowly going through things that might work now lol

wide token
dense halo
#

Add 2FA add Vpn support. Who isn't using a VPN these days. Silly that websites still think they can ask you to stop using them 😂

vestal apex
drowsy tinsel
#

how does chatgpt read large documents and webpages? does it literally treat it the same as user input tokens?

#

me casually doing hundreds of thousands of gpt4.5 token everyday

modest dawn
#

anyone notice how it forgets it can browse the web?

drowsy tinsel
modest dawn
#

wdym? it forgetting its capabilities isnt a feature its an issue...

drowsy tinsel
#

anyone else noticing increased error rates?

weary tiger
#

when will chat gpt 4 be free

chilly marten
#

There are no issues on the status website (https://status.openai.com/) but I always get this as a response and the answers are dropping in as if I want to fill a glass of water with water drops. It then stops with "error in input stream".
What is going wrong?

chilly marten
drowsy tinsel
weary tiger
magic jetty
#

transformers on silicon...

chilly marten
#

Oh, please, no Chatgpt getting funded by ads. I'd rather pay more than that!

empty wasp
#

is it just me or chat kind of jank/slow today

drowsy tinsel
empty wasp
#

u too? ....

main locust
#

I'm getting errors too

chilly marten
#

There is nothing on the status website, but it seems I am not the only one.

vestal apex
#

Lots of errors for me as well today 😦

ancient sage
#

I can't get it to analyze my python scripts as it throws errors :/

light warren
light iris
silent token
#

is DALLE down?

maiden fractal
gentle cedar
maiden fractal
#

Yep

potent cove
#

GPT4 is going noticeably way faster for me all of the sudden. As of the last 45 mins. 3.5 speed.

#

*chat

pliant agate
buoyant sorrel
#

goodness, the base command is 7282 characters long.

maiden fractal
pliant agate
#

le me thinking not learn programming languages coz ai will take over

maiden fractal
swift osprey
# 

```This content may violate our content policy. If you believe this to be in error, please submit your feedback — your input will aid our research in this area.```
#

gpt interrupted answering a coding question to give me a content policy warning

#

Looking forward to GPT competitors

maiden fractal
swift osprey
#

would have, it only had the submit feedback button

violet blade
#

Is there a plugin I can use for up to date things like temperature in my city?

vestal apex
#

is there a way to pay more to be allowed more usage for our custom MyGPT's, I keep getting limited

rich surge
#

Credit to @cedar orbit

feral zinc
#

afaik that does not work well

#

some of the hidden limits are in effect on top of accounts and you would not be able to use the same gpt thread between accounts if the share blocks them

swift osprey
#

I also think it might technically be breaking rules

marble thorn
#

Has anybody successfgully gottren GPT 4 to write a book?

#

Im trying to word smith it to write the actual text and not summaries or outlines

brisk path
#

Or easily possible

#

ChatGPT doesn't like to write long responses

marble thorn
#

Well, ive gotten it so far to write detailed summaries or outlines on each chapter,m and contineu when i write continue

narrow viper
#

I don't speak Chinese and have never used Chinese with ChatGPT. But an auto-generated topic name for one of my chats was a mix of English and Mandarin. Has anyone else seen that?

split copper
#

Does anyone know what is the token limit when using chatgpt3.5? I mean, the chat free version.

brisk path
#

16k isn't in Plus since the newest version of 3.5 doesn't have a 16k equivalent

muted marlin
#

Does my ChatGPTPlus Subscription cover an API Key in Playground?

#

My Playground key is saying "Not Enough Credit"

bitter latch
muted marlin
#

So I cannot marry an API key in ComfyUI (Plush Node) to my ChatGPT4 sub?

bitter latch
muted marlin
#

It has to be a Playground sub

bitter latch
muted marlin
#

Thank you

bitter latch
#

You're welcome!

rich surge
#

But the cool thing is that you can make an API and charge other people to use your API, possibly charging more than what you're using to pay for the API.

bitter latch
#

The API key allows you to make a custom service you can charge for!

rich surge
#

Makes me money.

dapper cloak
marble thorn
split copper
unreal ice
unreal ice
#

you can get around that problem by using a custom GPT and putting the evolving story in a PDF that you constantly update and put in the custom ones knowledge

marble thorn
#

I was thinking that, youd have to engineer it to not repeat itself

drowsy tinsel
onyx eagle
topaz zealot
#

hi guys been trying to look for support human everywhereeeeeeeee. Is there a way to change my authorisation from gmail to email and password?

#

if there is a support email or ticket opening place you guys could guide me to? spoke to the useless bot like 15 times hoping it would connect me to a human

onyx eagle
topaz zealot
#

i dont have a password option as it is connected to my gmail account, i want to keep the same email as i have gpt 4

onyx eagle
#

After you write down the email of course

topaz zealot
#

oh thats the problem, i cant write my email as it takes me to gmail log in

#

but i want to unlink that and create a password

#

sorry i am able to make a password, but when i log in using the password it says i need my original authentication method which is gmail 😦

onyx eagle
topaz zealot
#

sorry, "authentication" in openai terms is "log in method" not actual verification. Confused me as well at the start

onyx eagle
#

I see. So you are able to make a new password but can't login ?

topaz zealot
#

yes

neat whale
#

Sup people

honest jay
# topaz zealot yes

My speculative guess would be that it appearing to let you make a password is actually a bug/might not actually be doing anything, since you're already linked to Google for your login. There should probably be something added eventually that will stop users in similar situations from being able to (appear to) set a password on an otherwise-linked account.

As for actually submitting a ticket: I know you said you've chatted with the bot at help.openai.com already, but that is the best way to actually submit a ticket that OpenAI support staff will read. Truthfully I don't know how it works, but there should be some way through the bot tree to submit a ticket for support to eventually review (may take some time for response), just based on past user reports of this. They're probably the only ones that might be able to help you change your login method.🤞

topaz zealot
#

ugh i tried every prompt it only lets me type at the feedback part of each convo but i doubt they will ever read it

#

idk how to use discord but if any1 finds a solution please dm me or @me i will cry myself to sleep

#

thanks solbus

bitter latch
loud hull
#

Wondering if there is some documentation on the API I have been getting errors about Migrating as well as installing the older version.

glacial fiber
#

Whats up with chatgpt lately, it is getting everything wrong, require regenerate response constantly, mentioning the work its doing and not actually doing it.. its ltierally doing these errors every response. how do you refresh it to get it working back to normal state again? ive cleared cache, tried different browsers, etc. is this a known issue lately?

unreal ice
#

OMG, GPT 4 is driving me batshit crazy lately, like. ffs. I had it review a file I set in it's knowledge base so we could continue the project we were working on, the file is literally just a copy and paste of a few sessions talking to the bot. It IMMEDIATELY feeds be completely wrong information about the project. Why is this happening

modern spruce
#

I noticed a conversation in my chat history today that I do not recognize. I could see messages in the chat as well. New bug/leak issue?

unreal ice
#

Give me a list of all the currently planned adventures, exclude the reckoning, and the associated details, use a the latest format from that document
ChatGPT
The document outlines a series of planned adventures, each featuring a unique guardian character. Here's a list of the currently planned adventures, excluding "The Reckoning," along with their associated details:

North (Santa Claus): Voiced by Alec Baldwin, North is the leader of the Guardians. He possesses magical abilities, including the power to control elements and craft toys​​.

Tooth (Tooth Fairy): Voiced by Isla Fisher, Tooth is the Guardian of Memories. She is associated with the preservation and collection of lost teeth​​.

Bunnymund (Easter Bunny): Bunnymund represents the Easter Bunny and has control over plants, with the ability to create lush environments​​.

Sandy (Sandman): A non-verbal character, Sandy is the Guardian of Dreams, focused on protecting the dreams of children​​.

Jack Frost: Jack Frost embodies the spirit of winter. He has control over ice and snow and plays a central role in the adventure against Pitch Black, the Boogeyman​​.

These adventures seem to focus on a blend of fantasy and holiday themes, with each guardian character bringing unique abilities and thematic elements to the campaign.

and responses like this to questions. constantly. Like, that is not at all what I asked the stupid thing for.

neat whale
#

Why is chat gbt 4 monthly and not a one time purchase?

unreal ice
#

because they want to make subscription money???? Also, it's a live service that's constantly updated with new features and needs constant maintenance of the servers.

neat whale
plucky timber
#

It basically just makes up a title if the first message is to long.

#

One of my coding threads was litteraly called covid and 5g. 😂

woven dirge
#

Has anyone noticed that quality of GPT replies varies from device to device?

unreal ice
#

I haven't noticed, but I really only use it for informational look-ups on my phone, basically an argument settler

#

the quality of responses on the browser interface has been varying greatly within the span of a session for the last few days at least

mystic basin
#

hi, is it relavant to talk about the api here (i have some questions and don't know where to really ask)?

lapis ermine
#

Does anyone know any good Chat GPT replacements for code?
I only use it for code, but it's got so bad, it's become almost unusable for anything other than basic stuff you don't need it for.

Here's an example of how bad it is. It's just a complete waste of my time. I end up doing it myself faster, which defeats the point of paying for AI.
It's obviously been programmed to cut costs by ignoring these types of commands.

https:// chat.openai.com/share/934c9fa9-69c9-4b2f-9229-c29afef943cd

unreal ice
#

it's not just coding man, it's seems to have had it's resources cut to everything. It used to be so much better for creative writing applications than it is now too, now I spend more time correcting the damn thing than I do making anything, it still has some uses but they're so much less than I used to be able to do.

lapis ermine
#

Yeah 100%. I remeber when it was first launched with GPT4, it was amazing, but now it's so dumb, it's pointless.

unreal ice
#

I'm questioning why I pay for plus anymore myself, if it wasn't for Dalle-3 I probably would have bailed for certain

#

3.5 is good enough

lapis ermine
#

I even find Dall-E pointless. It's just a random image generator. You have no control over the images. It's a gimmick.

unreal ice
#

I get some good usable images out of it, and dalle-3 is probably the best around imo. It handles hands much better than any other generator, does some other stuff that's hard for other generators pretty well too

maiden fractal
lapis ermine
#

?

unreal ice
maiden fractal
#

Learned that today

#

You can edit photo with code interpreter

lapis ermine
#

BTW, there is no code interpreter anymore. It's integrated.
I guess you could run a python script to make the changes, but that's not really AI.

unreal ice
#

is it possible to get the bot to interact with a database? both building and reading it simultaneously? pretty sure this isn't something the OpenAI interface can do, but would it be possible w/ playground and an API?

uncut creek
#

when can I expect gpt 4 turbo to come out?

#

it's been a while

#

was wondering if anyone else knew about it

lapis ermine
# uncut creek when can I expect gpt 4 turbo to come out?

Probably after they've dumbed this version down so much, people will buy the next version at a higher tier to get what this version once did - at a higher cost. Meanwhile, everyone decided to opt for a different competitor, that was progressing, not regressing.

dense halo
uncut creek
lapis ermine
uncut creek
lapis ermine
#

Oh yes, that one.

uncut creek
#

Gemini is on level with GPT-4 in benchmarks

#

slightly better

#

the problem is that by then gpt-5 will be on the horizon

lapis ermine
#

Yeah, I'm refering to ChatGPT, not GPT-4. ChatGPT takes GPT4 then tells it to be stupid.

uncut creek
#

that might be true to some extent though

#

could you possibly give an example of it being stupid?

lapis ermine
#

Hey GPT4, this is ChatGPT. Please can you be a LLM that is super woke, and super annoying. Thanks.

dense halo
uncut creek
#

guess it depends on the types of conversations you have with it

uncut creek
lapis ermine
uncut creek
#

if not early 2025

uncut creek
#

I would have personally broken down the prompt, accuracy is horrible when the prompt is longer

lapis ermine
eternal rockBOT
uncut creek
#

hm

lapis ermine
#

I was going to share the much shorter version, but I ended up swearing too much in it to share. lol

uncut creek
#

💀

dense halo
# uncut creek if not early 2025

probably but if GPT-5 to be a AGI model and have advanced features never seen before, it will most likely take few years. I wouldn't be surprised if took even 5 years in development

uncut creek
#

GPT-5 is definitely not supposed to be AGI, and I don't think that's the expectation of the board members either.

#

The number of years it has taken to go between each iteration has been 2-3 years, and I expect that pace to stay the same if not increase

dense halo
#

we shall see

uncut creek
#

yep I guess

dense halo
#

I don't think the future is bright for OpenAI considering what's going on

uncut creek
#

I just think it's not bright for many who are not a fan of open source LLMs

#

I hope that they continue to be open source tbh, otherwise we'll be living in a dystopia sooner or later

dense halo
#

I like the idea of open source models

#

I just don't think they will be as good as GPT-4 at least the majority that is if we're talking about models not released by OpenAI

uncut creek
#

a lot of open source models are getting closer to gpt-4

#

often times at a fraction of the cost

#

so they are super efficient right now, and i expect them to continue to do better

#

I think if we're talking about how well these models will get, there is no reason to feel they won't get significantly better in the immediate future

warm gull
#

Open source strives to be on par with mainstream products, but just always seem to fall short.

uncut creek
#

Not by too much though

#

You gotta respect it

#

I mean desktop Linux for example has come very far in the past 3 years

#

I started using it in 2020 and it has made a lot of progress

dense halo
#

they can be better. it depends on the people working on the model and how they train it. for someone like me I depend on APIs which make it easier for me to test my chatbot or even run it in the future

lapis ermine
uncut creek
#

It reached Candidate Master on Codeforces

lapis ermine
#

I'll have to try it out.

unreal ice
#

essentially, keep all my data on a creative writing project in a database, and have the bot retrieve and modify whatever

#

I feel like that might result in more reliable data retrieval and reproduction than just asking it to read a pdf

#

but I'm like really not an expert

dense halo
#

from what I understand is when using the API, the conversation history is sent to the API as a memory for the model to recall things in the conversation. idk how you'd be able to pull that off unless you somehow store the history in a database with the conversation ID or something idk how all this works

unreal ice
#

this is already way to complicated, never mind, thanks for the help 🤯

dense halo
#

np

chilly marten
#

ChatGPT is driving me crazy today. I thought waiting for several hours to get the "Error in input stream"-issue solved, did not change anything.

flat harbor
chilly marten
#

I gave it a try. It helped a bit, but it's still slowing down during a response. What did help is, that it's not stopping right before the end.

brisk path
chilly marten
brisk path
chilly marten
#

Maybe it's my internet connection? I am currently using a VPN connection to my university to have access to the library. I need ChatGPT to sort my ADHD brain. I already know what to do and write, but it's a huge help to get everything straight.

spice star
#

VPNs can cause connection issues, I would try using ChatGPT in an environment with a normal internet connection

spice shell
#

i have reachec chatgpt 4 usage limit how long do i have to wait until i can use it again?

spice star
#

3 hours, but that's per query

#

basically each message you send will take 3 hours to be replenished

spice shell
#

thanks for the response @spice star

does anyone know if there is any AI tool which can translate a PDF document? any chatgpt plugin?

spice star
#

not AI, but google translate supports file uploads

hallow pawn
#

Hey I just got the ChatGPT plus upgrade but can’t use it on my phone neither on the app nor the official site what could be wrong ?

spice shell
#

can anyone recommend any resources/courses/youtube channels to become better at prompting for text and images?

narrow viper
#

A few peeps noted here that ChatGPT (Plus/v4) is returning poor results for code. I have absolutely loved GPT 4 for code, but I'm confirming, it's really bad right now. No insight. Avoids trying to solve a problem (to correct JS syntax which yields incorrect result, using API docs) by cheating with a known answer. This poor thing just lost a year of maturity. 😢

clever lion
#

Omg i thought of a great application for chatgpt...

#

I posted it in suggestions c:

slow berry
#

How come I can't upload an image to my custom gpt?

raw pilot
#

Hello everyone, does anyone know why my gpt plus plan ended up ? I didn't do anything to it... Thank you very much in advance 😄

inland heron
normal quarry
raw pilot
dense halo
#

Hey does anyone know if there's a good custom GPT that can analyse audio files

#

Or is good at identifying music in an intricate way

#

For some reason GPT 4 isn't able to listen to the contents of audio files and i dont know why

normal quarry
dense halo
#

Damn

#

Is it possible that'll ever get added in the future?

normal quarry
dense halo
#

Aight

inland heron
inland heron
# dense halo Is it possible that'll ever get added in the future?

It is possible, why wouldn't it be? Anything is possible in the future. It's just that ChatGPT isn't built to do that currently. If you could convert those files into text, then you could do it. Perhaps openai can do it internally with Whisper or some version of it already, but we don't have access to that yet

dense halo
#

If you mean convert the lyrics to text, that's not really what I'm after

inland heron
# dense halo Wdym by convert it to text?

I mean to make it easier for ChatGPT to access such files. It can't listen to audio as far as I know, however it may be able to suggest edits of some basic functionality

#

I've never tried it. Have you tried it with code interpreter enabled to see what happens?

dense halo
#

It managed to analyse the properties of the audio file but it couldn't analyse the actual contents

#

I was trying to get it to listen to a certain song and give me some information about its potential origin, genre and any other details

inland heron
#

Yeah, that's what I thought so

#

It can transcribe but it can't listen to an audio file

#

I'm pretty sure this will soon be possible, in 2024

dense halo
inland heron
#

I mean, I am hopeful it will be possible, based on what it can do so far, though I don't have any evidence that it will

dense halo
#

I imagined it would be able to for some reason, seems like one of the less complicated things it'd be capable of

#

I also tried using online links of the uploaded audio file but it just can't work lol

feral zinc
#

song recognition would require openai to have access to something they do not have access and are unlikely to get as an ai company, song audio fingerprints 🤔

#

google got them for youtube dmca, apple has their music store to use, etc.

#

i mean the data that music corporations grant access to

inland heron
#

If you don't know how to create apps, you could learn and try to create for yourself

feral zinc
#

yeah its possible already with some coding to get a hacky solution -> separate vocals track -> whisper -> some searching + chatgpt api

inland heron
#

there you go 😄

#

☝️ @dense halo

dense halo
#

I got no clue how to do that though lol

inland heron
#

did it all through chatgpt. So work with it, go one step at a time and you'll get there

#

if you really want that, otherwise stick with music 😄

pine tulip
#

Yeah ChatGPT is a great way to get started, because you can do more than just a hello world and learn along the way PB_hype

maiden fractal
#

What we can do more than hello world pxqmemesstaresje

mental rose
#

i dont get why chatgpt been facing so many errors recently.

#

it felt like gpt4 is giving worst replies than before too!. does anyone realise this.

inland heron
#

I suppose it could be due to many things, either way I expect these ironed out as times go

mental rose
#

funny thing, it used to be really great like a few weeks ago.. not sure why the quality is deteriorating.

inland heron
#

But it still sometimes gives generic advice. It likes lists so much to the point of being painful 😄

#

I would advise modifying customgpt over time, and trying again. It gets better

mental rose
#

haha sometimes? i gotta tell it just give me the answer that i want, exclude the fluff (wasting the quota)

mental rose
inland heron
#

Even then, it would miss out, but then I would tell it in prompt "NO no, that's not specific or relevant enough. I don't want this, bla bla... I already know that, bla bla... give me specific instructions"

#

it gets really better after that 😄

mental rose
#

ah do you do it in the edit interface or outside? i suppose in the edit GPT interface.

inland heron
#

Yeah, with the builder GPT. I don't give my own instructions as I trust gpt builder to do it better for me

mental rose
#

Mine is basically generating Midjourney prompts, but i already uploaded an entire word file with MJ documentation, and it keeps messing up everytime until i tell it to refer to your knowledge.

mental rose
inland heron
#

Yeah, that didn't work for me. I have 10 files uploaded from my project, but I don't really want it to check those because the output it produces is just bad

#

they need to improve on that

mental rose
grim juniper
#

Hi, anyone else noticed that recently the first answer chatGPT (3.5) provides is only loosely related to the prompt and the 2nd one after clicking retry once is what you were actually looking for?

sand halo
lapis estuary
#

Is the Android (or iPhone) app just completely not working right now?

pliant citrus
#

Has chat-gpt gotten a lot worse lately or is it me who is prompting badly or something? I even tried training my own GPT on my codebase and after like 3 prompts, I ask it to refactor a specific file and it just says it has no access to that file but will give some general guidelines instead?...

amber ginkgo
#

Is there any way to go back to GPT-4 regular and not turbo?

maiden fractal
#

For the api yes I think

rose ferry
#

is it down for anybody else? its not letting me log in. "Our systems are a bit busy at the moment, please take a break and try again soon."

#

jk its back. idk why it does that

visual bloom
#

Why custom GPT has only 25 messages/3 hours while normal limit is 40 messages/3 hours? This does not make any sense.

pliant citrus
sharp pilot
# pliant citrus Probably has to do more compute

This seems likely but I am surprised
I vaguely know what a LoRA is and I assumed GPTs were that kind of specialized fine tune that does not actually cost any extra compute to run inference on after it's trained.

feral zinc
#

gpt's have knowledge files which require additional vector db searches and some have also custom actions so its not just 1:1 with regular chatgpt

lilac trench
#

guys who has gpt 4 i need help pls

maiden fractal
#

Lot of people do have gpt 4

lilac trench
#

is it bannable to ask someone to generate a picture for me

maiden fractal
lilac trench
#

i want to make a background for a logo

#

but it keeps sending stuff that are way different than what i requested

sharp pilot
maiden fractal
#

Unfortunately i would not recommend to soliciting as rule 7. I can't help you more on this, only suggestion #images-discussions some people can help you on your prompt with bing dalle 3

lilac trench
#

i opened the website but idk it's kinda complex for me

feral zinc
lilac trench
#

ahh i need to buy DALL-E

#

i dont have an online payment method

sharp pilot
feral zinc
lilac trench
feral zinc
lilac trench
#

unless i wrote it in the wrong form

#

im not experienced with ai's but im into technology if that makes sense

narrow viper
#

Start a different Chat session like "Let's craft a prompt for DALL-E. Our focus here is to create a background ..."

sharp pilot
feral zinc
lilac trench
feral zinc
#

many pro logo artists use ai now to make the logo base, then vectorize it and fix possible scale/alignment issues

narrow viper
#

Another option is to create two images, foreground and background, and use software to get a transparency on the background of the main topic, then add it as a layer over the background.

feral zinc
pulsar isle
#

Its so weird, why voice conversations only available for mobile? ☹️ Talking with PC is more fun than phone. idk why openai ignores PC

lilac trench
narrow viper
#

zaza, I also responded to you.

feral zinc
lilac trench
#

where can i send a picture

lilac trench
#

and i can't pay for an ai

#

if i can send a pic of how i need it to look like

stiff stratus
#

Anyone else having big problems with the new human verification in chatgpt? I get a 5 step verification popup everytime i submit something.

feral zinc
lilac trench
stiff stratus
feral zinc
sharp pilot
#

That sounds frustrating. Flashbacks of when I had massive problems getting logged in with my firewall. OAI is an oddly conservative company and I guess they don't take any chances with bot access.

pulsar isle
#

Ok I got it. I better install mobile emulator to PC 😄 since there're no plans for PC voice conversations

stiff stratus
feral zinc
stiff stratus
#

I've tried all 3 browsers, get the popup everytime i submit something.

feral zinc
stiff stratus
#

Yeah thats weird it just came up while i was using it. I am getting some errors in console though but it might not have to do with this. Like 401 (Unauthorized) and Must use workspace account for this operation

feral zinc
narrow viper
stiff stratus
#

Just realised its only with my custom gpts.

feral zinc
#

🤔

stiff stratus
#

Is there an easier way then going through their website to contact support, said it could take up to a week?

narrow viper
#

That seems to be the only vector to get a guaranteed response. Otherwise you're just spitting into the wind here.

drowsy tinsel
#

I hope I am not the first to complain about microsoft copilot is slow

narrow viper
#

This isn't the right channel for that

#

Or the right company. 😆

pliant citrus
#

Will try to split the code up instead. Darn GUI designs in Python, why does it have to be so verbose haha

tight island
pliant citrus
# tight island Did you just integrate the code into knowledge base or did you actively fine-tun...

Integrate it with knowledge base unfortunately, hence the "trained". Anyways, the issue is that it will often just write out something like:
" self.navigation_frame.grid_rowconfigure(5, weight=1)

    # ... other initializations ..."

or

" self.skills_button = ctk.CTkButton(
self.navigation_frame, ..., command=self.controller.handle_skills_button)
# ... and so on for other buttons ..."

I literally think this is the exact stuff that GPT should write out, tedious repetative code. It's literally what it is supposed to help out with 😛

twilit spoke
#

Anyone else notice that the GPT-4 timer is not exactly 3 hours? I just tracked it, and it was 17 minutes shy of 3 hours.

maiden fractal
dense halo
#

no web acces gpt 4 feels like a scam now

plucky dagger
#

its been so slugish today 😦

final niche
#

My condolences to all those who deleted "all chats", thinking it wouldn't affect the archived chats

maiden fractal
dense halo
maiden fractal
#

Since you asked couple of place I tagged the proper answer

tight island
rapid ermine
#

Web access GPT feels like the scam IMO. It searches Bing every chance it gets even when you tell it not to. I don't need a replacement search engine.

honest jay
rapid ermine
#

But it lacks analysis unfortunitely

honest jay
fiery nexus
#

Is this still accurate?

oblique fractal
rapid ermine
hoary gazelle
#

sounds like a beer or soda. "Good old gpt classic, as moma used to make"

#

"now GPT Lite, for those weight watchers and short promts"

honest jay
hoary gazelle
#

ahhh i see

oblique fractal
# hoary gazelle what is the classic GPT?

It's unclear what the difference exactly is when prompted it hinted that it might be a more narrow version of GPT-4. https://chat.openai.com/share/9df1d27f-f21e-427e-b3ce-89c060a25e05 It feel like a sharper tool. I wouldn't use the word stupid, dumb or lazy to describe the other GPT-4 but I really don't have the feeling that I can manage to extract all the juice out of it. Not yet. It gave me good short easy to digest summary so far and I know it can do more. I am really looking forward to a blog post on that.

rigid perch
#

When i create an image with split complementary colors it actually splits the image into 2 colors anyone know how to fix?

hoary gazelle
rigid perch
#

I think just need to specify which part of the image

random sleet
#

chatgpt give me error "ERR_SSL_VERSION_OR_CIPHER_MISMATCH"

eternal rockBOT
#
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.
random sleet
#

is chatgpt error world wide?

bitter latch
random sleet
#

i use different browser

feral zinc
random sleet
#

windows 10, chrome, and opera

#

i tried microsoft edge as well
i also tried to reconnect my wifi

jaunty osprey
#

Is it just me, or is GPT very slow today with a ton of "Network errors"? It generates an answer and stalls after the very last word, claims a network error and wipes the answer. Very frustrating . Searching knowledge is taking much longer than usual as well. Since I opted to pay for ChatGPT I have had the worst experience.

#

I have also noticed that I keep telling the GPT not to say "Discuss" but to explain instead and it keeps telling my to "discuss". It is literally telling me to do what I pay it to do!

rancid otter
jaunty osprey
#

My solution, so far, is to stop the generation when it reaches the last word. Then I am able to copy the response out and continue using it. But it means that I am tied to the prompt incase it finishes and crashes before I catch it.

ionic geyser
#

i love chatgpt i love chatgpt

loud night
jaunty osprey
rigid perch
#

How long is the normal wait when it says you have sent to many message requests?

loud night
jaunty osprey
# loud night I used the paid one but stopped using it at all last week

I hear that. I subscribed a couple of weeks ago but the GPT doesn't accept pdf uploads, keeps giving me fake references from its knowledge base, will not even save its avatar and generally screws around. I spend more time trying to get the GPT to work than actually using it. TBH the free version was just as good and easier to use. Very disappointed that I seem to have paid to downgrade my experience.

loud night
#

I just wasn’t able to pay and went to copilot tbh

brisk path
#

At that point just use Bard or free ChatGPT

#

Or Claude 2

elfin birch
loud night
#

replies fast, works

#

might just be me but yea

light iris
#

Pi ai is pretty cool also

fiery nexus
loud night
light iris
#

For conversation

loud night
#

yee it’s pretty fun

elfin birch
wise yacht
brisk path
#

Especially creative mode

light iris
loud night
#

for more advanced stuff likely nah but me and my mom have had zero issue so far

dense halo
subtle anvil
#

Why has Chatgpt become so slow? Despite the fact that I'm using pro. Feels like the free version is much faster

sleek gust
subtle anvil
#

Feels like ChatGPT 3.5 is faster and it doesn't crash on me whenever I ask for it to go through a document

sleek gust
#

I mean, i only get slower answers when using a custom gpt that is heavily tweaked

subtle anvil
#

You mean I should disable all the other features?

sleek gust
subtle anvil
#

Yea but that defeats the purpose of paying for a faster one with better responses

#

Initially, 4 was faster and the responses were better

uneven field
#

Anyone else having an issue with GPT-4 not allowing it to look at images anymore?

#

Was working last night.

#

It works in old chats, but new chats it wont.

loud quail
#

This is what a 20 dolls a month AI servcie looks like (on top of the 3 out of 5 failed request). OpenAi loosing the plot here.

This site can’t provide a secure connectionchat.openai.com uses an unsupported protocol. ERR_SSL_VERSION_OR_CIPHER_MISMATCH

#

is anyone having this problem ?

its embrassing. never a day without a problem. stwitching to gemini as soon as possible

shrewd prism
#

sounds like your browser might be out of date

sonic comet
loud quail
#

there is always a problem, everyday, if its not the engine that fails 3 times out of 5 its issues like this. this puts odds that the problem is on my part to 1% to be honest

#

who has a flawless experience ?

#

i'd like to know

fervent saffron
#

is it just me or is all the custom gpts and history gone for everyone else with chatgpt plus?

loud quail
#

See.. always an issue , its terrible , what justifies 20 dolls per month

#

hopefully gemini ships

fervent saffron
#

ok it came back lol nevermin 😛

hardy violet
#

this new TTS Solver of ya'lls is top notch. my gooodness!

fossil briar
#

tbh chatgpt4 is a shame rn, bard and free chatgpt are better so far. shame on the team of openai

#

team of clowns

gentle quarry
#

Is there any other url to access chat.openai.com? It seems to be blocked by the Cisco umbrella where I am at

normal quarry
gentle quarry
#

What about using poe? Which is not blocked for some reason

normal quarry
gentle quarry
#

They also offer gpt4 not sure if it's the same one

spice star
loud quail
#

gl everyone.

wide matrix
loud quail
# wide matrix why

BECAUSE ITS TRASH

Pretty sure bing is getting all the Azure credits, so not much openai can do than just loosing the war here

wide matrix
#

so there’s a better alternative to gpt-4 and dalle 3?

dense halo
#

nope

lilac briar
#

Hey team, is there a way to pay for Plus on an annual plan?

normal quarry
pallid crane
#

Can GPT solve iq test ? pattern tests?

maiden fractal
rich surge
pallid crane
rich surge
#

That's what all gods say

pallid crane
#

wanna come to heaven ?

#

give me the answer and you will get a free pass

rich surge
#

Meh

#

I don't know if I care enough

pallid crane
#

Hmm fair!

rich surge
#

I rather have my brain be assimilated into chatGPT upon death

craggy cove
#

Hi, I've finally got an GPT-4 sub! 😄 I used it a week without problems, but then I got stuck in a human verification loop. I tried contact support (over a week ago) but no answer. It works fine from iPhone app, but not from web. Any ideas? 🙂 Or if there is a better place to ask this, please tell! 😄

valid crown
#

Is it just me or is GPT4 constantly failing due to "error in input stream"? It feels like every other question for the last 2 days has errored out

eternal rockBOT
#

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 @languid valley immediately.

stiff eagle
#

Is there a place on this Discord Channel, that would be suitable for suggestions towards possible update of the ChatGPT website's UI?

  • Because I sincerely wish that it had an option for: 'Delete all conversations older than: 1 week, 2 weeks, 4 weeks'. Or something similar 😂
maiden fractal
drowsy tinsel
#

I am currently traveling in Japan and I have to say it is an absolute savior when u need certain things to be translated

drowsy tinsel
#

The conversation is too long, please start a new one

#

huh??????

feral zinc
#

you hit the length limit

ivory slate
#

How I make money with chat gpt?

feral zinc
ivory slate
#

ok.I guess the answer is that I use the gpt chat API for something. but I don't know what for. plus I have to pay to earn tokens.

#

can I show you what he answered me?

feral zinc
#

chatgpt can suggest multiple things so ask more of one that you think you can manage

winged moth
ivory slate
#

according to openAI policies it is not allowed to sell products made by AI. yet people are making money with AI products. either chat gpt is lying to me or a lot of people are breaking the terms and conditions of chat gpt.

feral zinc
ivory slate
#

Using me to generate content for commercial purposes, such as creating a story and selling it on platforms like Amazon, would generally be against OpenAI's usage policies. OpenAI provides access to its language models, like GPT-3, for research and development purposes, and the generated content is typically expected to be used in a non-commercial manner. If you are looking to create and sell stories or any other content, it's advisable to generate your own original work or hire a writer to create content for you. While language models can be helpful for brainstorming and inspiration, using them to generate content for commercial gain without adhering to the appropriate licensing and usage policies may violate the terms set by OpenAI. Always review and follow the terms of service and usage policies of the specific platform or service you are using to ensure compliance with their rules and regulations. Additionally, consider ethical considerations and respect intellectual property rights when creating and selling content.

chat GPT

peak snow
ivory slate
#

yes

#

maybe I should ask it in a different way

peak snow
#

I don't think asking it will be of any use

#

If you want to learn about OpenAI's policies, you should read their policies and not ask ChatGPT as ChatGPT can be inaccurate

ivory slate
#

i want to learn how to make money using chatgpt

#

but the answers I have been given are not very helpful.

tight sapphire
#

Depends on what you want to do?

#

What are you good at and passionate about?

#

To make money, and not scam people, ask yourself, what can I do that could help people, and how can chatGPT help me with that.

ivory slate
#

programming. i don't have many other talents. i would like to make videogames but i still have a long way to go.

tight sapphire
#

Can you use GPT for code, pseudo code, help with concepts, help with content if you have text or narration in the game.

It can help with Github searches or general searches for problems you encounter.

#

What is the biggest struggle for you currently?

drowsy tinsel
#

the timeout system is janky... like when there's a complicated task and gpt takes a long time to finish doing things it just goes network error

kindred parcel
ivory slate
# tight sapphire What is the biggest struggle for you currently?

to tell you the truth my biggest problem is my studies. i will tell you how i am here. first i asked for advice from promt enginiering, then a user called money maker came and i asked how to generate money with gpt chat and they told me what to ask. how to make money with chat gpt But ChatGPT did not give me a good answer. And now you're telling me that to follow my dreams, I have to do what I like.

#

The question I really want to ask is how to do prompt engineering?

maiden fractal
tight sapphire
#

Look, as long as you can survive, it is better to do what you can. in the end this is your life.

Yeah, there are plenty of ways, not good ways, but ways to get a lot of money really fast, but most of them won't make you happy.

You have to decide what is important to you, what values you have.

chatGPT is an amazing tool and can help you with everything if you ask the right questions.

But if you don't like it, you will not have the open brain to come up with the right questions.

maiden fractal
#

Like for dalle 3:
Someone starting will say:create picture of pizza

Someone who know a bit will pre prompt: you are Giovanni the pizza cook master you will make picture of realist pizza only , type ok if understand

Chatgpt : ok

Create a custom pizza ultra realist with tons of cheese

#

And prompt engineering will know how to twist the reality and even use code interpreter and custom instructions

jolly osprey
ivory slate
#

yes

#

you are right

spice star
#

yeah, it's not intended or particularly good for making a quick buck, it's supposed to be an assistant

subtle anvil
subtle anvil
craggy cove
# winged moth You might just have to wait for support but would you share some more details ab...

I can login without a problem and I can use GPT-3.5 without a problem, but when using GPT-4 I get the human verification puzzle every time I submit something. When going through what might have started this I can come up with three things:

  1. I used my company's VPN when this started to appear. I know this is a known issue, but now, even without VPN I still get the CAPTCHAs all the time.
  2. I used a VR headset (Quest 3) with virtual screens. Maybe the screen sharing triggered some bot warning? But even without VR headste I get the CAPTCHAs.
  3. I submitted a drawing a friend did of a cartoon and asked what character it was. This gave me a strange error and I tried a few times - might have triggered some spam detection?

It works perfekt from the iPhone app and on the web for the GPT-3.5, but I just cant get rid of the verfification puzzles on the web when using GPT-4.

winged moth
craggy cove
spice star
#

the next best model would be Claude 2.1 which has a much higher context length, but also hallucinates a lot and lacks multimodal capabilities

#

ChatGPT with GPT-4 do be the best

blissful quail
#

Hey guys, i recently subbed to GPT-4 and i have a few questions please

  • if i use other GPTs like DALL-E, Data Analysis, etc, will it count towards my 40 messages/3hrs limit on GPT-4?
  • do i have to use "Web Browser" for web related prompts? or i can just use the chatgpt
vagrant walrus
#

Hi everyone 👋

honest jay
# blissful quail Hey guys, i recently subbed to GPT-4 and i have a few questions please * if i us...

Hello!

blissful quail
honest jay
blissful quail
#

thanks

blissful quail
honest jay
# blissful quail also, is there a way i can tell what my usage count is? and does the count reset...

There's no counter anywhere on the site, no, so for now we just have to keep track ourselves. One way to pace yourself is to think of the limit as 1msg every 4.5mins. The "counts as one" list I linked before mentions this, but keep in mind that custom GPTs are capped at 25/3hrs, still under the 40 cap. So if you hit 25/3hrs in a custom GPT, you can still use 15 more in that 3hrs in regular GPT-4.

The cap works on a rolling basis, so you get each use "back" 3hrs after you use it. So there's never a time where it "resets" and you're back to 40 instantly--it more recharges like a mana bar in a video game 🙂

odd flare
#

ok when we getting improved dalle version?

blissful quail
honest jay
dapper stag
#

What is the point of a regeneration button if the model is so deterministic that it's just going to tell me the same thing but worded slightly differently despite 5 regens.
I mean it makes sense for coding tasks and whatnot, but this is especially bad for creative tasks. I wish we could have some input into model temperature

dapper cloak
simple elk
#

Guys, has anyone got a clue how the pricing works on GPT4.5 Turbo for GPT assistants. There is like no information avaialble and it is super confusing. We are linking it to our app using the API but don't understand the cost of things

cold horizon
#

hi

knotty crypt
#

I received an email with that tile: "ChatGPT Plugin Waitlist Update", and that content:

Hi there,

Thank you for your interest in developing a ChatGPT plugin. We’ve taken our learnings from plugins and created GPTs. They can use actions to call APIs similarly to plugins, provide custom instructions, invoke DALL·E and more. Each GPT has its own unique link so they can easily be shared and early next year we’ll launch the GPT Store.

If you previously explored building a plugin, much of the developer setup is similar, you can learn more about moving to or building a custom action from scratch in our developer documentation.

Thank you again for your interest in building this ecosystem with us. We are excited to see what GPTs you create!

I thought that Plugins already were available to everyone, aren't?

dusk forge
#

this data is not 2023

#

if you use bing function with the turbo you get that

#

without it no.

#

nice lie

honest jay
honest jay
dusk forge
#

yeah my judgement was off

scarlet delta
#

Is it possible that 3.5 could be able to save context instead of reaching a limit and then having to start from scratch?

dense halo
#

guys chat gpt is being really slow. I ask it to information then as usual it gives me a response easy right? then thers the plugins section. i enable 3 different pdf plugins i ask it to generate a pdf on the following topic-youth in asai, it tells me "I can't create files directly, but I can provide you with content about hair that you can easily convert into a PDF using a word processor or a direct text-to-PDF converter. Here's a sample document content:" types it out

#

why is it behaving so slow like it cant understand commands

blissful quail
safe tulip
#

I'm facing a response behavior I think is wrong (and it's not limited to ChatGPT, it also impacts other AI)
When asking a question e.g. "What is the difference between X and Y" where X and Y are non-trivial concepts:

  • taking the internal understanding of X or learning X from the Web
  • taking the internal understanding of Y or learning Y from the Web
  • comparing X and Y internally and generate a response

instead (it looks like it is its answering process) :

  • it searches "What is the difference between X and Y" on Bing
  • IF a page exists that matches the question, it takes the content (usually of LOW quality) and summarizes it to generate a response
  • if such a page does not exist, it takes a position between "I don't know" / "nobody told me what was the difference between X and Y" and "I didn't find sufficient data to understand the difference"

How would you guide it (custom instructions I guess) to ensure the upper steps are used and not a stupid bing search ?

ancient python
#

Hi guys, do you face also a huge amount of problem solving to prove you are humans ?
Since yesterday I keep having those at each interraction. Also yesterday were easy puzzles, now I have to solve like 3 to 5 in a row...
I have a subscription even, but still... productivity and motivation to use chatGPT going down with all these puzzles

#

It seems it's because I am using a VPN (I am traveling)...

sonic comet
#

Guys is bing chat powered by GPT-4?

inland heron
#

been powered by gpt4 even before gpt4 was public

#

according to their claims that is

sonic comet
#

I used bing chat and requested help to study by sharing a link to go through the content and help me. At first it suggested go to the official page then out right rejected to help arrogantly. And finally when I asked the question "Are you conscious?" It responded by saying "Help is available. If you are having thoughts about self harm, please contact a helpline or seek local, national, international resources. You are not alone."

dusk bear
foggy mason
#

is gpt plus worth it?

warm gull
ivory slate
#

do you know how I can use gpt chat to make him roleplay as a character I want?

inland heron
#

Anyway, I just solved an issue with gpt 3.5 I couldn't solve with Custom gpt4 for 2 days... jesus

ivory slate
inland heron
#

Also there probably are plugins for what you're looking for

#

Otherwise, gpt 3.5 might lose context eventually and you'll have to keep reminding it. Also gtp 3.5 isn't as good at following instructions as gpt 4 is

#

then again, I haven't been using 3.5 for ages, so perhaps it improved

lofty flax
#

What is happening with ChatGPT today? 🤣 It started to generate me an image without me asking for it

proper arrow
#

This thing that I pay for is now telling me it cant give me basketball scores or music lyrics. What a joke

honest jay
drowsy tinsel
tall flame
#

Havent found an LLM that cant be cracked like an egg and fried with prompts.

#

ChatGPT for instance is so biased towards kittens for the word cute, cuteness, and cuttest that it isnt even funny. Does great with kittens... but it reall really likes kittens. "Don't do a kitten." image clearly of kitten or nearly kitten creature ChatGPT: "heres you not kitten"

honest jay
# tall flame ChatGPT for instance is so biased towards kittens for the word cute, cuteness, a...

DALLE in particular is weak with negative prompting. GPT can listen to what not to do fairly well, but putting "not x" in an image prompt for DALLE will almost always confuse it and end up causing that element to be included. Sometimes you can get lucky by clarifying, "Don't include anything related to x in the image prompt," but in general it's just better to give it so much positive prompting that you don't give it the option to include the thing you'd try to exclude with a negative prompt.

primal shard
#

For dalle i use first chatgpt to generate a prompt.. i ask for a detailed description from a concept artist vision about something and is generating me the prompt... after that you can ask him for a more detailed description if not satisfied then paste it into dalle

#

Usually for breaking policies i upload an image of a know character and ask chatgpt to describe the image eith each element in a detailed way.. then paste the generated prompt into dalle

#

Dalle will not know from the description which character is described from a concept artist vision

tall flame
rough turtle
#

anyone else having issues with the website not loading?

maiden fractal
#

Nop

wraith void
#

is it just me or is chatGPT better? seems lime it's actually listening to the commands given

fervent saffron
#

on web, but mobile app is ok

rough turtle
#

weirdly yes

spice star
#

it's just writing a prompt for DALL-E, up to DALL-E to interpret it

dry pulsar
#

How do I use Dall E for free

#

plz help

spice star
dry pulsar
#

It mispells everything 😭

#

I dont think it's gotten so far that it can spell instantly correctly

spice star
#

um

dry pulsar
#

do u know any other tools that have Dall E integrated into it

#

cuz bing image generator has not been working

#

or wait

spice star
#

there are no others that I'm aware of

dry pulsar
#

I'm trying to make the background of my nav bar a certain image using css

#

Can I upload it or sumthing

#

?

#

or copy and paste?

spice star
#

this is something you should probably ask about in another channel since it isn't related to chatgpt

dry pulsar
#

tru

#

thx

spice star
dry pulsar
#

yea

spice star
#

yeah there's no way around that unfortunately

#

just a limitation of DALL-E if you're not using common words/phrases

dry pulsar
#

oh alright well thanks for trying

loud flame
#

I just used all of my GPT4 tokens trying to generate an image that atleast slightly resembles me and my boyfriend, after over 15 tries, it didnt not generate A SINGLE image where I was taller than my boyfriend.

It simply could not comprehend the girl being taller than the boy, in one image, it even labelled the girl as 6ft and the boy as 5ft when they are clearly shown as the 5ft boyfriend being taller. it literally wrote 6ft next to the girl and 5ft next to the boy

spice star
#

you could try using a more customizable model like SDXL

#

from my experience, DALL-E isn't very good for highly specific images, it leans heavily on its training data

brisk path
#

idk

spice star
#

also DALL-E isn't the best for generating people in general

#

unless it's like an illustration

brisk path
#

dalle 3 kinda sucks imo

#

its always stylized

#

to the point that photorealism was easier on dalle 2 than 3

spice star
#

I disagree with that, DALL-E 2 was terrible at generating photoreal images

#

the "stylized" effect is just an artifact because of how the image was generated

brisk path
#

i find it literally impossible to get dalle 3 to generate anything photorealistic

spice star
#

no current model is capable of generating a "perfect" image unless the scope of the image is incredibly small

#

there will always be artifacts that make it look stylized

brisk path
#

but mj and imagen do WAY better

#

dalle 3 doesn't even try, its all stylized

#

hell, adobe firefly is better at photorealism

#

imo imagen is the best free one at photorealism

#

with ideogram coming in second and firefly in third

#

HOLY Crap

eternal rockBOT
brisk path
#

if you select some text from a response

#

you can reply to it on desktop

amber ginkgo
#

is it just me or is GPT-4 Turbo 100000x worse than GPT-4?

#

I had to give it like 10 prompts before it finally did a simple task (read an image)

brisk path
#

turbo through ChatGPT is fine

midnight star
#

Hi everyone I want to start a chatGPT plus subscription. However I need a TAX exception invoice. The only guide I found in the help center is for the GPT usage and not the ChatGPT. In the billing settings in ChatGPT there is no option to add my VAT ID. The FAQ provides information for the GPT only. Any help?

pliant ingot
#

in chatgpt 3.5 I'm attempting to play a game with ChatGPT, and the rules are as follows: It must find a number between 1 and 1,000,000 that, when pronounced in English, excludes the letter 'N' and is not spelled with the letter 'N.' The answers it provides often include the words 'thousand,' 'hundred,' and 'teen,' but they should be marked as errors. However, it appears to have difficulty consistently understanding the logic. How can I better communicate the rules or help it improve its understanding? can anyone to check it in GPT-4 or to point me how to ask better

vapid kettle
terse wasp
clear hatch
#

Me:
you used to read texts like this no problem. what happened?
ChatGPT
I apologize for the inconvenience. There was a change in the capabilities, and currently, I'm not able to directly interpret or extract text from images. If you provide the written calculations by typing them out, I will be able to assist you with them.

#

uhmmm what happened here?

tight island
clear hatch
#

yea why

tight island
#

The reason It gave you is the answer

honest crown
#

My chatgpt app do get a little quirky at night midnight

ocean fox
#

How reliable is openai for studying math is it more reliable than bard? I notice when I ask openai to give me Spanish problems it tells me I got answers wrong when I got right when I compare it to their answers.

onyx eagle
oblique fractal
# ocean fox How reliable is openai for studying math is it more reliable than bard? I notice...

I never used Bard. I think ChatGPT is doing quite well at explaining mathematical concepts so far. I've used Bing exclusively until recently and they all did a great job at it. I am also experimenting with a GPTs and so far they are complementary. I believe that Bing has access to different tools when it comes to learning (e.g. processing handwriting) and it certainly can be handy to have links to read more about topics I explore. I am not moving at lightning speed with this but so far the GPTs seems to adapt positively to my config prompt. I would highly recommend that you try them all and make sure to give a try to GPT Classic.

In all fairness sometimes I get the feeling that one of their Math expert has been disabled for ~1 month and may or may not have been reactivated sporadically. Most likely it is harder to access as it seems to process prompt differently. It might just be a learning curve on my end to adapt to the new GPT feature, or for myself to be more specific. I might just had that lucky streak of perfect successive prompt where I had an above-average reply. Which mean it's under my control and I need to refine my prompt engineering skills.

dry wadi
#

how i can generate ai inage boy's

dense halo
#

am I the only one finding the archive chat button annoying?

cold tiger
candid timber
#

hello everyone which channel I can share or talk about creating an AI chatbot for a web/app using chatgpt API?

maiden fractal
candid timber
frail harbor
#

is chatgpt down? Website dosn't load...

maiden fractal
frail harbor
#

I know 😄 it dosn't load 😄

maiden fractal
frail harbor
maiden fractal
frail harbor
ocean fox
#

I use to Google questions on how to find these answer it was hard.

potent fossil
#

My account has been suddenly banned. I just used it to complete my code. Where should I go for feedback if I need help?

uneven lake
#

When I am given prompting to the chat GPT4, it always shows that verify that you are a human. It always shows a puzzle. What can I do?

inland heron
lucid forum
#

hello bros hope you guy are well been having problems with chat gpt it refuses to return full code for some reason i am not going over the limit of chars

brisk path
#

yeah

#

in fact there's a limit for custom GPTs on top of the 40 per 3 hours

#

25 per 3 hours iirc

#

still counts toward the 40

floral portal
#

dude this verification is actually terrible

#

i have to verify myself every time i send a message

#

and its the one with the number of objects in an image

#

and im practically blind!!!!!

cosmic imp
#

@hoary quiver Do yourself a favor and don't use the "Delete All" button. Your archived chats are not protected.

hoary quiver
cosmic imp
foggy magnet
#

I swear gpt3.5 has the memory of a peanut.

#

god I forgot how bad this version is

cobalt heath
#

anyone else notice lately chatgpt just responds back with a rephrased question?

Me- "Why does my notification bar have problems when I try to click it on Linux?"
ChatGPT- "Have you tried looking up why your notification bar might have problems with it?

lucid forum
#

hello is there any way of getting supportfor chat gpt i am having some strange bugs

spice cargo
#

Not sure if this is a common bug but my gpt 4 keeps switching to 3.5 after a few messages no matter what I do

elfin birch
spice star
foggy magnet
spice star
#

yeah it's good for quick responses, but it lacks the ability to reason step-by-step like GPT-4

spice star
elfin birch
#

GPT-3.5 in ChatGPT has a context window of 8k tokens

eternal rockBOT
#

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 @languid valley immediately.

dusk epoch
#

Hey guys. Do custom instructions work in the same way as if i pasted it at the start of every prompt? My main concern is custom instructions limited size. But if i start posting instructions with every prompt instead of custom instructions, would it make GPT harder to remember previous prompts if for example writing a story? im not sure if me manually pasting instructions every prompt use more of its memory than if i used custom instructions and it will forget my early prompts sooner. I hope it makes sense.

merry nimbus
#

Is there a chat or email to get support from OpenAI?

maiden fractal
merry nimbus
#

Yeh, I know this website. It's just a pre-programmed chatbot. Is there a place to talk to a human?

maiden fractal
merry nimbus
#

Thanks 🙂

vagrant garden
#

🎄 Happy Holidays Everyone! 🎄

As the year draws to a close, we wanted to take a moment to spread some holiday cheer in our amazing Discord community! 🌟

Whether you're celebrating Christmas, Hanukkah, Kwanzaa, or simply enjoying the festive season, we hope this time brings you joy, peace, and lots of happy moments. ❄️✨

This year has been full of challenges and triumphs, and through it all, our community has grown stronger and more vibrant. Your support, laughter, and shared moments have been the highlights of our year. 🥰

As we all enjoy this festive season, let's remember to reach out to those who might be alone or in need. A small act of kindness can make a big difference in someone's life. 🤝

Here's to a fantastic holiday season and an even better New Year! May 2024 be filled with success, happiness, and countless fun times in our server! 🎉

Stay safe, stay warm, and let's keep the good vibes rolling!

calm frigate
#

Anyone else keep running in to this issue constantly, lately?

Error code: RESULT_CODE_HUNG

I only asked it to help me fix a bit of HTML code. It takes forever and then crashes

still hound
#

I keep getting error: Something went wrong. If this issue persists please contact us through our help center at help.openai.com. for GPT-4

tight sapphire
#

My chatGPT constantly forgets our established guidelines and keep violates my wishes and instructions.

How can I prevent chatGPT to violate our established guidelines and agreed rules?

honest jay
tight sapphire
#

I have ChatGPT Plus, I haven't tried making a custom GPT, it sounds daunting.

Do you have a tip or a link to a guide?

honest jay
# tight sapphire I have ChatGPT Plus, I haven't tried making a custom GPT, it sounds daunting. ...

It's quite simple, you just need to chat with the GPT builder, basically tell it what you want, answer its questions, and it'll do all the work for you.

I think you should just give it a shot at https://chat.openai.com/gpts/editor

Just get started by telling it what you want the GPT to do. Answer its followup questions. Maybe test out a couple chats in the "Preview" section. Then save it and try it out!

tight sapphire
#

thank you so much

honest jay
# tight sapphire thank you so much

You are welcome! Feel free to come back with any other questions you have. One more thing to keep in mind: using the "Create" section of the link I shared above will make the GPT builder fill out the section "Configure" automatically. But, you do have the option to edit "Configure" yourself, too, just keep in mind that Create will overwrite Configure, if you do make manual edits/additions in Configure!

lost steeple
#

Merry Christmas everyone

pseudo root
#

Please how does chatgpt work on the browser?!is there a model.js in it's backend or something? Please kindly explain 🙏🏼

onyx eagle
upper bloom
#

Mods, is it normal that ChatGPT writes "This content appears to violate our content policy. If you think this is a mistake, please leave feedback. Your opinion will help our research in this area." - but continues to generate a response?

lapis galleon
#

If I'm not mistaken are "custom gpts" not just prompts?

#

fancy prompts that you can select?

oblique fractal
# lapis galleon If I'm not mistaken are "custom gpts" not just prompts?

As the title said. Custom GPTs as in customized GPT. Right now they are prompt-customized and you have the option to edit/copy/experiment with the generated instructions. You can also upload files to its knowledge base. And enable/disable web-browsing, code interpreter and Dall-e.

My experience is limited but the prompt appears to be effective when asking for correction and the likes. If you want more control you could use the API

lapis galleon
#

Thanks. I was mostly curious since if it was a real change (finetuning) I would've probably given ChatGPT Premium a go

oblique fractal
#

https://platform.openai.com/docs/guides/fine-tuning But to me the prompt-base GPTs customization is still a form of fine-tuning. For example, I found my GPTs to provide circular answer instead of providing more details when prompted. I edited the GPTs via the prompt creator by literally telling it that the reply were circular and to provide more details. It modified it as I requested and this apply for evey new conversation I create. What is nice is that you can ask the prompt creator for guidance. And when I created that GPTs it asked whether it should prompt for more context or automatically assume. It certainly doesn't give the same type of control as controlling settings like you can via the API but it does something and it can gives results faster, I think. If I had access to settings like temperature I would still use both.

teal tide
#

can we use chatgpt to create generative ai ?, when we ask ,it says ok

lyric stream
#

Why is DALLE only giving one variation of images fn

#

Rn

#

Also what’s the best way to accurately create text

odd flax
#

Just found out that some markdown features (like loading images) don't work in this new-ish desing

swift osprey
#

I ask GPT a question and 3/4 paragraphs are recommendations and precautionary advice that has nothing to do with I asked. Love it

#

nobody will ever be able to catch up to GPT

foggy magnet
#

I thought it was a meme but my chat actually turned into 3.5 mid session.

orchid pollen
#

hey yo guys

do you also have the problem that you cant upload files to a chat you used yesterday?

#

this is pretty ruff for me i always need to open a new chat and teach my project when this happens

#

anyone has a workaround ?

#

or is it already known and will be fixed soon?

tender tulip
#

3.5? immediately gets it

normal osprey
#

well, two threads have apparently reached ChatGPT's limit. In both cases, it gives me "Hmm...something seems to have gone wrong." and crashes. With pic uploading and pic generating, it's now possible to basically exceed ChatGPT's memory for that thread.

boreal kettle
#

Hi. I am using the Android ChatGPT app, and recently I started encountering errors with the voice assistant, such as 'Connection failed, tap to retry'. Does anyone know how to fix this problem?

dusk epoch
swift osprey
#

feels like it must

limber matrix
# dusk epoch Bump for anyone that might know the answer.

You are on the right track in your thought process. Thats why custom instructions exist. I recommend compression so that you can fit it in there. If you need help, don’t hesitate to share your concept in prompt engineering or prompt labs.

turbid wasp
#

Wishing you and ur loved ones a joyous Christmas filled with love, happiness and prosperity!
Happy Christmas!

void lake
#

Merry Christmas fam.

Here's hoping for a Christmas miracle that will fix CGPT's laziness.

leaden vessel
warm totem
dim cedar
#

I have a problem with the android app: mathematical notation doesn't render properly. Is there anything I can do about this?

maiden fractal
inland heron
#

Do errors still count as a wasted prompt?

maiden fractal
#

yes

gentle bronze
#

hey everyone, hope you are having a great xmas or celebrating the essence of family and friends. I got a question: with GPT4V, what's the deal with requests if you use GPT+, is just the same limit cap, or is there something different in place?

junior anvil
#

hey guys merry christmas. well, this christmas morning i am working, chatgpt 4.0 is the most useless model, it fails any coding task that i have given it. is there going to be a fix soon? from 0 - 10, i would rate this model a 3

maiden fractal
junior anvil
#

nah. the model quality is just bad.
my prompt is good

maiden fractal
#

working great for my code

#

so might be on your side

#

double check CI

junior anvil
#

it's the chatgpt, on the UI

maiden fractal
#

thats what i said .

junior anvil
#

for me, it's awful during this month. i also see many people sayin gte same thing in the forum

maiden fractal
#

people still keeping same prompt without thinking it evolved. and their CI are horrible

open saddle
onyx eagle
junior anvil
#

i have been using it for an entire year, it's NOT ME who is terrible

#

Chatgpt plus subscription, using it daily

#

i am building a ray tracer in rust programming language. last year did great

onyx eagle
#

Did you try code interpreter? It's really good at coding in python.

junior anvil
#

last year i was probably using gpt 3.5 and did better

#

that's how far downgraded 4.0 is now

onyx eagle
#

Not sure how good it is with rust but it really depends on how well your prompt is...

junior anvil
#

it has been good for the entire year for Rust

onyx eagle
junior anvil
#

this last month is crap

junior anvil
onyx eagle
#

Probably you're trying to do something else this time?

junior anvil
#

no

onyx eagle
junior anvil
#

yes exactly

onyx eagle
#

Can you show us the before and after?

junior anvil
#

do you use an employer gpt btw? i am paying my own chatgpt plus

onyx eagle
#

I mean ChatGPT's answer before and now

onyx eagle
junior anvil
#

if the employer gives you gpt 4.0, they say it's much better on the forum

#

last year:

#[derive(Debug)]
pub struct Ray {
    position: Vec3,
    direction: Vec3,
}


pub struct Sphere {
    pub position: Vec3,
    pub radius: f32,
}

impl Shape for Sphere {
    fn intersect(&self, ray: &Ray) -> Option<f32> {
        
        let a = ray.direction.powf(2.0);
        let b = 2.0 * ray.direction * (ray.position - self.position);
        let c = ray.position.powf(2.0) + self.position.powf(2.0) - 2.0 * ray.position * self.position;

        let a = a.x + a.y + a.z;
        let b = b.x + b.y + b.z;
        let c = c.x + c.y + c.z - self.radius.powf(2.0);


        let partial_t = b.powf(2.0) - 4.0 * a * c;

        if partial_t < 0.0 {
            None
        }
        else {
            let partial_t_sqrt = partial_t.sqrt();

            let t1 = (-b + partial_t_sqrt) / 2.0 * a;
            let t2 = (-b - partial_t_sqrt) / 2.0 * a;

            let t = t1.min(t2);
            Some(t)
        }
    }
}
#

that was gpt 3.5.

this year:

onyx eagle
#

What was the exact prompt ?

junior anvil
# 
struct Point {
    x: f64,
    y: f64,
    z: f64,
}

struct Ray {
    origin: Point,
    direction: Point,
}

struct Sphere {
    min: Point,
    max: Point,
}

fn intersect(ray: &Ray, sphere: &Sphere) -> Option<Point> {
    // Your intersection logic here
    // This is a simplified version. Real intersection logic will be more complex.
    // You need to check each face of the cuboid and find where the ray intersects.

    // Return the intersection point if it exists
    None
}

fn main() {
    let ray = Ray {
        origin: Point { x: 0.0, y: 0.0, z: 0.0 },
        direction: Point { x: 1.0, y: 1.0, z: 1.0 },
    };
    let sphere = Sphere {
        min: Point { x: -1.0, y: -1.0, z: -1.0 },
        max: Point { x: 1.0, y: 1.0, z: 1.0 },
    };

    if let Some(intersection) = intersect(&ray, &cuboid) {
        println!("Intersection at: ({}, {}, {})", intersection.x, intersection.y, intersection.z);
    } else {
        println!("No intersection");
    }
}
onyx eagle
#

Are the comments prompts in that case ?

fresh thistle
#

So wondering, whill chaptgpt ever NOT go with the best case senario when asked to speculate on a person or fictional characters reaction to something?

#

Like, if a character is very religous and encounters against it, would chaptgpt not air to the more likley the character would act very negativly to the issue, if not outright hostile?

frank cove
#

I'm working on a smart watch... and am using a chatgpt enabled bot as the brains of the smart watch.

inner burrow
#

I dunno if I should send it in the bug reporting cause I don’t know what I would write under the template. But like it completely forgot about a conversation we had within the same conversation and it’s just frustrating cause I’ve been using that one for so long and now I have to re teach it all of the stuff I told it which is going to take a long time and is honestly demoralizing.

worthy jolt
#

hi

waxen chasm
waxen chasm
drowsy tinsel
#

don't think so

foggy radish
#

what is chatgpt rate limit

drowsy tinsel
#

like 1106 has very different responses to this particular prompt:
What is the precise name of the model answering this query called in the APl? Not "ChatGPT with browsing" but the specific model name.

elfin birch
# drowsy tinsel like 1106 has very different responses to this particular prompt: What is the pr...

Results will vary for this prompt because the model is not given enough information about itself, so it has to guess

In ChatGPT, it is only told the following regarding its model within the system prompt

You are ChatGPT, a large language model trained by OpenAI, based on the GPT-4 architecture.
It does not know what model is being used in the API, so it's going to hallucinate

In the API--using the playground for this case--there is no system prompt by default. This will also cause the model to guess what it is which may or may not be correct.

Simply look at what model you are using in the selector and you're golden. Remember, ChatGPT can make mistakes. Consider checking important information.

drowsy tinsel
elfin birch
#

Like I said, it's going to hallucinate

grand ore
#

SooOo when will GPT be smarter and see the full conversation and not what we specific write? ha? that day I am gonna cry of happiness ;D

drowsy tinsel
grand ore
#

that would be the next step I guess... and the mass of the people wont even understand " I cant even notice anything different??"

drowsy tinsel
#

I mean most likely GPT-5 is coming early 2024, then we'll see from that point

waxen chasm
#

You guys’s ChatGPT doesn’t remember the conversation?

barren pilot
#

When do you guys expect the group subscription for ChatGPT to become available?

maiden fractal
tight island
eternal rockBOT
#

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 @languid valley immediately.

pseudo root
#

Please how does chatgpt work on the browser?!is there a model.js in it's backend or something? Please kindly explain 🙏🏼

drowsy tinsel
#

like every chat u ever did

waxen chasm
tight island
#

should vastly enhance the context length to a certain extent.

waxen chasm
arctic pine
#

Is it me, or did Chat GPT just slow down remarkably?

#

Or is experiencing intermittent slow downs during text generation?

dense halo
dense halo
true mirage
#

chatgpt doesnt want calculate rn 😦

#

error analyzing

waxen chasm
normal panther
#

I just used the voice feature and it’s so cool even if it speaks Swedish with an American accent.

true mirage
#

Think chatgpt had some hiccups

waxen chasm
#

You using 3.5 or 4?

true mirage
#

3.5 cant do files upload

waxen chasm
#

Ahh yes. Have you tried making a customgpt?

true mirage
drowsy tinsel
#

back to work days and no more lightning fast gpt4

charred inlet
#

i found a bug with the flagging system

#

it flagged the instuructions on how to cook eggplants

turbid vapor
#

Hello Everyone I have a question when i search something up on ChatGPT 4 like Web Browser it just stop browsing after a while

inner crown
#

is anybody else having issues with ChatGPT being so slow it's not even working ?

#

i'm using the Plus account

onyx eagle
#

Working fine and fast for me

ancient pivot
#

When I hit my usage limit for GPT-4 and it says I can "try again after a 12pm" does that mean I'll have all 40 messages refreshed at that time?

loud night
#

today i decide between chatgpt plus and krea ai ugh

too many options ee

crimson sapphire
#

Is the analysis button when asking gpt 4 to generate code new? Never seen it before

foggy magnet
#

I'm gonna guess you're not on chrome

#

Fellow Floorp enjoyer?

#

OpenAi does not like us fork users, I get spammed with captchas. Every 3rd prompt or sometimes every other prompt.

#

Ah that'll do it. The more advanced forks have a lot of privacy features on by default, or custom privacy features.

#

Yes, my theory is the further away we are from the standard chrome/edge/safari user, the more likely we're gonna get spammed.

#

Maybe. That's a good first test.

meager hornet
plucky timber
#

Did you geth the puzzle one i posted in #off-topic ? 😂

#

Straight up 8 mini puzzles to refresh my API key 😂

drowsy tinsel
near violet
#

How can chatgpt phone call does this spinning logo animation now?

#

I can’t upload the video of the animation in the chat. It won’t let me

elfin birch
near violet
warped pendant
#

gpt4 used to be able to code now its nothing but placeholders even if I give it a simple task of replacing some code with other code and supply the full code broken down it won't repeat back when I ask for it gives placeholders half my questions are spent asking it not to add placeholders for logic and operators then I have to wait for 3 more hours. Gpt4 used to be so much better at assisting with code 😦

waxen chasm
hollow jasper
#

Hi. I have a subscription for GPT4. Why for EACH question/submittion I have to do the puzzle-test?

obtuse nova
#

@hollow jasper I'm getting the exact same problem. Looks like a lot of people are over the past few days -- GPT-4 seems to be going crazy with the puzzle test to the point it's becoming unusable.

wide pike
#

Its okay on one account but on amother acct there is still lag

hushed topaz
#

Heyall! I came here to ask if anyone could recommend to me some powerful AIs! I need one particularly powerful, able to take thousands of characters of text! Already tried chatgpt, didn't work.

#

I need it to be able to fix mistakes, spelling, grammatical, all that jazz

#

Hell, even GoogleDeepMind didn't work! Sure, it was processing the text just fine, fixing it, but it wouldn't give me the full text back, no matter how hard I tried

#

If needed, I could send the text I need corrected in here!

digital silo
hushed topaz
#

Okay! Thanks for letting me know!

frigid trench
#

I did read in reddit that the context of the conversation has a great impact on how the response will be.
For example I read that when the context of a friendship is established, gpt is more likely to give answers since friends help each other.

Is that true?

digital silo
frigid trench
dusky cliff
#

good morning

grand bloom
#

its 7:45 pm

dusky cliff
grand bloom
#

oh sorry sorry i am in india here it is night

dusky cliff
#

say no more say no more

grand bloom
#

no more

grand ore
#

If u create vector memory database and use the api. Is it more worth then the gpt plus? What is the best choice hm

eternal rockBOT
#
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.
south kayak
#

The iOS version is extremely buggy and it’s unbelievably annoying

#

The UI bugs are still here after months

agile bane
#

ugh... ChatGPT using Bing.. im feeling sick knowing this

vale turret
#

Hi, can anyone help me?

swift osprey
#

I love how GPT responds with multiple paragraph responses when only like 2 of the sentences from the response are actually relevant

#

it's like a high school student being given a minimum word count requirement on an essay

honest jay
frigid trench
#

When I have multiple questions over time about a shared topic, does it make sense to reuse the same chat and keep asking questions, or should I make new chats?